Ernst Tugendhat Traditional and Analytical Philosophy Lectures on the Philosophy of Language 1982

Traditional and analytical philosophyL E C T U R E S O N T H E P H I L O S O P H YO F L A N G U A G E

E R N S T T U G E N D H A T

Traditional and analytical philosophyL E C T U R E S O N T H E P H I L O S O P H Y OF L A N G U A G E

T R A N S L A T E D BY P. A. G O R N E R

C A M B R I D G E U N I V E R S I T Y P R E S S

C A M B R I D G E

L O N D O N N E W Y O R K N E W R O C H E L L E

M E L B O U R N E S Y D N E Y

Published by the Press Syndicate o f th e University o f C am bridge T h e Pitt B uilding, T ru rh p in g to n Street, C am bridge CB2 1RP 32 East 57 th Street, New York, NY 10022, USA 296 Beaconsfield Parade, M iddle Park, M elbourne 3206, Australia.

© Suhrkam p 1976English transla tion © C am bridge U niversity press 1982

First published 1982

Prin ted in T h e U nited States o f A m erica by Vail-Ballou Press, Inc., B ingham ton, N.Y.

T h is book was originally published in G erm an in 1976 by Suhrkam p u n d e r the title Vorlesungen zur Einführung in die sprachanalytische Philosophie.British Library Cataloguing in Publication Data T u g en d h a t, E rnstT rad itiona l and analytical philosophy.1. Languages— PhilosophyI. T itle II. V orlesungen zur E in fü h ru n gin die sprachanalytische Philosophie. English401 P I 06ISBN 0 521 22236 2

T O T H E M E M O R Y OF M A R T I N H E I D E G G E R

Contents

Preface ixT ra n s la to r’s preface xi

P art I Introduction: confrontation of analytical philosophy with traditional conceptions of philosophy

1 A question o f m ethod 32 A ph ilo sopher in search o f a conception o f philosophy 123 O ntology and sem antics 214 Has form al sem antics a fu n d am en ta l question? 355 Consciousness and speech 506 T h e a rg u m e n t with the philosophy o f consciousness

con tinued 657 A practical conception o f ph ilosophy 76

P art II A first step: analysis of the predicative sentence8 P relim inary reflections on m ethod and preview o f the course

of the investigation 939 H usserl’s theory o f m ean ing 107

10 Collapse o f the trad itional theo ry o f m ean ing 12111 Predicates: the first step in the developm en t o f an analytical

conception o f the m eaning o f sentences. T h e d ispu te be­tween nom inalists an d conceptualists 133

12 T h e basic princip le o f analytical philosophy. T h e d ispu te con tinued . Predicates and quasi-predicates 150

13 T h e m ean ing o f an expression and the circum stances o f its use. D ispute with a behaviouristic conception 163

14 T h e em ploym ent-ru le o f an assertoric sentence. A rgum en t with G rice and Searle 177

15 Positive account o f the em ploym ent-ru le o f assertoric sen­tences in term s o f the tru th -re la tio n 192

Contents V l l l

16 Supplem ents 20717 ‘A n d ’ and ‘o r’ 22718 G eneral sentences. Resum ption of the problem

of predicates 24319 T h e m ode of em ploym ent o f predicates. T ransition to sin­

gular term s 25720 W hat is it fo r a sign to stand for an object? T h e traditional

account 27021 T h e function o f singular term s 28422 Russell and Strawson 29723 W hat i s ‘identification’? 31024 Specification and identification. Specification and

tru th 32325 Spatio-tem poral identification and the constitution of the

object-relation 33726 Supplem ents 348

I T h e connection between object-relation, situation- independence and the truth-capacity o f assertoricspeech 348II Reciprocal dependence of the identification o f spatio-

tem poral objects and the identification o f spatio-tem poral positions 357

27 Results 372I T h e analytical concept o f an object 375II T h e m ode o f em ploym ent of predicative sentences and

the explanation of the w ord ‘tru e ’ 38128 T h e nex t steps 391

Notes 411Bibliography 429Index o f nam es 434Index o f subjects 436

Preface

In so-called analytical o r language-analytical philosophy there is little reflection on its own foundations, and today less than before. For the most part the problem s treated are inherited problem s which are not questioned. Partly this is due to a lack o f historical consciousness. A way o f philosophizing can only become a fundam ental philosophical posi­tion by confronting it with earlier conceptions of philosophy. This reflection on foundations is not ju s t an additional act o f self-clarifica- tion. It is a condition o f a philosophy’s ability to perceive the task that has always been the genuinely philosophical task: the exam ination o f existing questions, m ethods and basic concepts, and the developm ent o f new ones.

T hese lectures aim to provide an im petus in this direction. They therefo re have the character o f an in troduction . By m eans o f a con­fron ta tion with traditional philosophy’s fundam ental o rientation to the subject-object schema they attem pt to bring questions which already exist in analytical philosophy into the context o f a specifically language- analytical fundam ental question. As regards conten t they move in a field o f investigation that is by no m eans new; and even in this field they take only a first step.

T h e book is directed at th ree d iffe ren t groups o f readers. T h e reader whom it addresses directly in the form of lectures is the philosophical beginner, for whom it could serve as an in troduction to the philosoph­ical way of thinking. At the same time it is directed, if only in an oblique way, at the read er who is already well-versed in linguistic analysis. Above all, however, it is directed at those who, being m ore o r less fam il­iar with traditional philosophical m odes o f conception, miss in analyti­cal philosophy a fundam ental question which can be com pared with the great traditional approaches. This book seeks to build a bridge for such readers, by try ing to show tha t analytical philosophy contains a fu n d a­

Preface x

m ental question which can1 not only com pare with the traditional approaches but actually proves to be superio r to them .

This aim is a reflection o f my own developm ent, which started out from H eidegger and led to language-analytical philosophy. I becam e convinced tha t H eidegger’s question about the und erstan d in g of ‘B eing’ can only acquire a concrete and realizable m eaning w ithin the fram ew ork o f a language-analytical philosophy. A lthough the re is hardly any m ention o f H eidegger in these lectures I owe to him the specific m ode o f access with which I approach the problem s o f analyti­cal philosophy. For this reason the book is dedicated to him.

It has its origin in lectures I gave in H eidelberg in the Sum m er sem ester o f 1970. A lthough I have re-w ritten and expanded the text it seem ed to me sensible to reta in the lecture-form .

Starnberg, March 1976. E.T.

Translator’s Preface

My aim th ro u g h o u t this translation has been to com bine accuracy with readability, bu t at times the latter quality has had to take second place. W henever possible long sentences have been b roken down into several sh o rte r ones, bu t in som e cases to have d o n e this w ould have significantly a ltered the sense o f w hat is being said.

As for my translation o f individual w ords the following req u ire some com m ent. For Vorstellung I have used ‘rep re se n ta tio n ’ ra th e r than ‘idea’, for to have chosen the la tte r would have m ade it impossible to translate the verbal form s vorstellen, vorgestellt etc. I considered the m ore literal ‘p resen ta tion ’, bu t in the end settled for ‘rep resen ta tio n ’ because o f the cu rrency it has acquired th rough K em p Sm ith’s translation o f the Critique o f Pure Reason. F or both Bezugnahme, bezugnehmen and Ver­weisung, verweisen I have had to use ‘re fe ren c e’, ‘r e fe r ’. For the m ost p art it is clear from the context which sense is in tended , but w here there is the possibility o f confusion I have p u t the G erm an term in brackets. For gegenständlich I have used the artificial ‘objectual’ because ‘objective’ w ould have been positively m isleading. Gegenständlich m eans som ething like ‘having the character o f an object’. I t has n o th in g to do with ‘objec­tive5 in the sense in which, fo r exam ple, a ju d g m e n t may be objective (ra th e r than subjective).

In translating quotations from H usserl’s Logische Untersuchungen I have in the m ain followed J. N. F indlay’s translation. In the case o f W ittgenstein I have simply rep ro d u ce d the s tan d ard English transla­tions w ithout m aking any changes.

I would like to thank P rofessor T u g en d h a t fo r the tho roughness o f his com m ents at every o f the translation , my friends Eric M atthews and Guy Stock for some very helpful discussions o f points re la ting to the translation, and the G erm an A cadem ic Exchange Service (DAAD) fo r enabling me to have two periods o f study in G erm any with consequent

Translator's preface Xll

benefit to my know ledge o f G erm an philosophy and the G erm an la n ­guage. Finally, I wish to thank Professor H ans W erner A rnd t o f the University o f M annheim for having first d raw n my a tten tion to P ro fes­sor T u g e n d h a t’s book.

P. A. G O R N E R

University o f Aberdeen

Part One

Introduction: confrontation of analytical philosophy with traditional conceptions of philosophy

L E C T U R E 1

A question of method

‘In tro d u c tio n to language-analytical ph ilosophy’ - th a t is am biguous. From a lecture-course with this title one m ight expect a survey o f a philosophical m ovem ent, an historical o r system atic gu ide to the ph ilo ­sophical lite ra tu re com m only called language-analytical. T his is not w hat I shall be do ing , particu larly as such in troductions to language- analytical philosophy already ex ist.1 T h e title can also be in te rp re ted in an o th e r sense, by u n d e rs ta n d in g ‘ph ilosophy’ in the sense o f ph ilo soph­ical activity. T h e title would then deno te an in troduction to language- analytical philosophizing.

O ne in troduces som eone to a particu lar activity by dem o n stra tin g it to him by m eans o f an exam ple, so th a t he can im itate it. So I w ould have to d em o n stra te to you a characteristic language-analytical line o f th o u g h t in a way th a t would enable you to follow it and stim ulate you to carry ou t sim ilar patterns o f a rg u m e n t yourself. A nd indeed this is som eth ing I in ten d to do. B u t such a dem onstra tion by m eans o f an exam ple cannot, taken by itself, suffice fo r an in troduction if the activity in question is a way o f do ing philosophy.

A way o f do ing philosophy is not re la ted to o th e r ways o f do ing p h i­losophy in the way tha t one fo rm o f dance is re la ted to o th e r form s. Form s o f dance a re not m utually exclusive o r inclusive. O n the sam e evening one can, with equal en thusiasm , dance a tango , a boogie and a rock ’n ’ roll - an d sim ply no t bo th e r with the waltz. B ut one cannot philosophize in one way w ithout having rejected o r in co rpo ra ted the o thers. A dance can be ou t o f date; bu t it is no t on tha t account incor­rect. In philosophy, on the o th e r hand , as in every science, th e concern is with tru th . For this reason, although ways o f d o ing philosophy can be m o d e rn o r o ld-fash ioned , w orrying abou t this is th e business no t o f the ph ilosopher bu t o f the h istorian . I f I am asked why I do philosophy in this way ra th e r th an tha t I canno t answ er: ‘Because it is up to d a te ’, bu t

Introduction 4

only: ‘Because it is the correct way.’ B ut this implies an obligation to justify the claim to be correct. To in troduce som eone to a way o f doing philosophy, the re fo re , involves rela ting it to o th e r ways o f doing ph i­losophy and, by m eans o f such a confron tation , dem onstrating its cor­rectness.

B ut this m eans that one m ust debate the idea o f philosophy as such. I f on e’s aim is to in troduce som eone to a particu lar way o f doing p h i­losophy one canno t simply p resuppose the concept o f philosophy. T o in troduce som eone to a particular way o f philosophizing is, hence, always also to in troduce som eone to philosophizing as such.

If it is tru e tha t one can only in troduce som eone to language- analytical philosophy, o r any other sort of philosophy, by contrasting it with o ther ways of do ing philosophy, then this affects the question o f which line o f th o u g h t is to be chosen to illustrate it. We cannot be con­ten t with ju s t any exam ple. In confron ting language-analytical philos­ophy with o th e r ways o f philosophizing we are not ju st confron ting m ethods. T h e im portan t philosophical positions o f the past always took as the ir starting-poin t certain fundam ental substantive questions a ro u n d which the whole field o f possible philosophical questions was organized. In the case o f language-analytical philosophy it may be less clear what its central substantive question is, indeed w hether it has one. B ut then we may expect tha t it m ight be precisely in the confron tation with earlier philosophical positions tha t language-analytical philosophy will find its own central question. A nd this m eans tha t it is only in this confrontation th a t it will find itself.

I f this is correct we canno t assum e tha t language-analytical philoso­phy is already a fixed quantity which we can first in troduce and then contrast with earlier positions in an appendix . N ow here is it laid down what language-analytical philosophy is. I f we sought to arrive at a defi­nition o f ‘language-analytical philosophy’ by a process of induction and abstraction from the existing philosophical litera tu re which is described as language-analytical, then at best we would achieve an em pty charac­terization; it could not serve as the basis for a concrete way o f philoso­phizing.

So do I want to in troduce you to som ething which does not yet exist? In the case o f philosophy this is not as absurd as it sounds. A philosophy is only constitu ted in philosophizing. I t follows from this tha t philoso­phizing, and a way o f philosophizing, is an activity which only becomes what it is in the process o f being in troduced .

B u t in that case we m ust abandon yet ano ther prejudice: if what is being in troduced does no t exist p rio r to its in troduction then clearly

A question of method 5

the person who wishes to in troduce o thers to this activity cannot himself have it at his disposal. He can only in troduce o thers by at the same time in troducing himself.

Perhaps these reflections strike you as incredible and as a poor peda­gogical trick. Does not the pretension of seeking to in troduce someone to som ething which does not yet exist, and is first constituted in the in troduction , rem ind one o f M ünchhausen’s attem pt to pull him self up by his own bootstraps? Can I seriously wish to assert that I want to in troduce you to som ething with which I am myself not yet acquainted? Obviously one cannot look for som ething o f which one does no t already have a vague prelim inary conception (Vorbegriff). A nd obviously I do have a vague prelim inary conception o f linguistic analysis. B ut then no doub t so do you. O n the o th e r hand , it is unclear to us, and in general, in w hat linguistic analysis, as a philosophical position, really consists. We cannot expect to rem ove this unclarity by getting an answer from som ew here, bu t only by deepening the existing prelim inary conception. A nd it may no t be im plausible to expect tha t precisely from a confron­tation of linguistic analysis - initially on the basis o f the vague prelim i­nary conception we have o f it - with im portan t earlier philosophical positions there will em erge its own substantive fundam ental question. T o arrive at this fundam ental question is the aim o f the in troductory p art o f these lectures (Lectures 1-7). In the main p art which follows we shall, by analysing the predicative sta tem ent-form , take a first step in answ ering this question.

Let us begin, then , with th a t vague prelim inary understand ing (Vor­verständnis) which everyone can be assum ed to have, inasm uch as it is simply an explication of its designation. Clearly ‘language-analytical philosophy’ refers to a way of doing philosophy which involves the belief tha t the problem s of philosophy can be solved, o r m ust be solved, by m eans of an analysis of language.

Im m ediately the question arises: by m eans o f w hat sort o f an analysis o f language? T h e analysis o f language would seem to be the task of linguistics. Does this m ean, then, tha t philosophy, if it is understood as linguistic analysis, becomes linguistics o r a part o f linguistics? O r is the analysis of language carried ou t in philosophy d iffe ren t from tha t car­ried out in linguistics? A nd, if so, how is the d ifference to be character­ized? Notice how, from the very beginning, o u r en terp rise becomes m ore com plicated. Language-analytical philosophy finds itself con­fron ted , not only with a dem and to legitim ate itself vis-a-vis o th e r con­ceptions o f philosophy, bu t also with the dem and to define its relation­ship to a closely-connected em pirical science.

Introduction 6

We have here a specific instance o f a difficulty philosophy has always, faced when trying to define itself: how is it to define its relationship to the sciences? I t is characteristic of m odern philosophy that this question arises not ju s t in general in relation to all sciences, but in a special way in relation to one particu lar science. For classical m odern philosophy, particularly since Kant, this science was psychology. Now it is linguistics. Perhaps there is ano ther way of do ing philosophy for which sociology occupies a corresponding position. In m odern philosophy this peculiar collision with a specific em pirical science results from what is called its reflective character. It conceives o f its enquiries as consisting not in the d irect them atization o f such and such objects but in sim ultaneous reflection on how these objects can be given to us, how they become accessible to us.

In classical m odern philosophy the field o f givenness reflected upon was conceived as consciousness, a dim ension o f representations or ideas; whereas in the new conception of philosophy it is conceived as the sphere o f the understand ing o f ou r linguistic expressions. In every instance philosophy finds its sphere o f reflection already occupied by a particular em pirical science. And so each tim e the question arises: how is this sphere, if, from the point o f view of philosophy, it is not ju st one sphere am ong others, accessible to a specifically philosophical mode of study?

I know o f no satisfactory answer to the question of how language- analytical philosophy is to be distinguished from the em pirical science o f linguistics. Such an answer can certainly no t be given with the aid of traditional distinctions between philosophy and science, since this answer would have to depend essentially on the new conception of p h i­losophy. Anyway at the p resen t stage o f this in troduction we clearly lack all the presuppositions for m eaningfully tackling this question. All one can really say at presen t is: language-analytical philosophy differs from the empirical science o f linguistics in that it has to justify itself as ph i­losophy, and, hence, finds itself confron ted by o ther philosophical positions.

I re tu rn to the nom inal definition of ‘language-analytical philosophy’ as a philosophy which seeks to solve the problem s o f philosophy by m eans of an analysis o f language. How can we get fu rth e r if we start from this first prelim inary understanding? We can tu rn to the person who hears this definition for the first time and see what his initial reac­tion is.

I f he is a thinking person he will im m ediately raise the following objection (it is the standard objection that is always b rough t against the

A question of method 7

language-analytical conception of philosophy). ‘It is clear,’ he will say, ‘th a t verbal explanations belong to philosophy. T hey have always done so. B ut they rep resen t only a prelim inary stage and serve m erely to rem ove the unclarity and am biguity in the use o f philosophical terms. T his can only be a transitional stage on the way to the things with which we are concerned. A fter all, language is only a m edium , and if a philos­ophy regards the analysis o f linguistic usage as no t ju s t a prelim inary task, but as its real task, then it has clearly lost contact with the substan­tial questions, the things them selves.’

We begin then with the negative in which the idea of a language- analytical philosophy first appears to an outsider. However, the objec­tion ju s t raised only rem ains on the periphery . It speaks o f things in contrast to words w ithout saying w hat sort o f things it m eans, and w here they are to be found . Only when we get o u r th inking person to explain what he m eans will we have taken a first step into the real field o f dispute.

In which extra-linguistic sphere, we will ask him , are the things them ­selves to which he refers to be sought? If he is no t a philosopher, but simply a th ink ing person, then he will most likely reply: ‘T h e things themselves? Clearly they are given to us by experience. A nd the appeal no t to rem ain with m ere words had this m eaning: to reach know ledge one m ust have recourse to experience.’

W ith reference to em pirical know ledge the objection, thus in te r­p re ted , seems plausible, indeed conclusive. Precisely what it says is true o f an em pirical science: explanations o f words are necessary, but they constitute only a transitional stage in research. H ere the things them ­selves are the facts o f a sphere o f scientific experience. B ut if the objec­tion is pu t forw ard as an objection to a conception of philosophy, then this can only 'm ean either (a) that one denies tha t philosophy is a specific dim ension o f enquiry which is not reducible to the em pirical sciences (in which case it is not an objection specifically to language-analytical philosophy, bu t to philosophy as such) o r (b) th a t one supposes tha t philosophy has its own, and hence non-em pirical, m ode o f experience. I f the objection is not simply from a th inking person, bu t from a p h i­losopher, then the second of these alternatives m ust be the one he has in m ind.

T h e justification o f the above objection cannot, the re fo re , be ra tion ­ally discussed w ithout going into the question o f the specific subject- m atte r o f philosophy, and what it is about this subject-m atter which distinguishes philosophy from the em pirical sciences. A dom inant, though not und ispu ted , view of philosophy in the history o f philosophy

Introduction 8

is tha t it has to do, not with em pirical knowledge, but with a p rio n knowledge, th a t its propositions are valid a p r io n , i.e. they cannot be verified, or falsified, by (sensory) experience. O f course, this description applies equally to logic and m athem atics; so it cannot be used to define philosophy. M oreover, such an ex ternal description rem ains unsatisfac­tory so long as one does no t ask on w hat essential fea tu re o f philosophy it is g rounded .

T hose who have described the subject-m atter o f philosophy as a priori (Plato was the first) have done so because they believed that all u n d e r­standing contains presuppositions we norm ally do not attend to, but which when a ttended to ap p ear as som ething we know, for we cannot conceive that it could be d iffe ren t. B u t when we want to express this know ledge we becom e perp lexed . A classical exam ple o f this (used again by W ittgenstein2) is St A ugustine’s rem ark about time. ‘W hat then is time? I f no one asks me, I know w hat it is. I f I wish to explain it to him who asks me, I do no t know.’3 H ere then we seem to have a sphere o f know ledge w here ou r ignorance rests not on inadequate experience but on the fact tha t we are dealing with aspects o f ou r understand ing which are too close to us and too obvious. W hat we are here striving for is not the explanation o f som ething tha t is not yet understood , but the clarification o f what is already understood . A nd this clarification can only be achieved by reflection on o u r u n derstand ing itself, not by expe­rience.

T his explication o f the subject-m atter o f philosophy (though still, o f course, wholly abstract and thesis-like) also enables one to see how ph i­losophy differs from o th e r a p rio n form s of knowledge. Logic and m athem atics are also a prio ri, but they do not seek to articulate som e­th ing we already know; ra th e r they ask about w hat is im plied by things which we already know, or which we can hypothetically assume. St A ugustine’s rem ark about tim e is not applicable to the sentences o f logic and m athem atics.

Since Kant the analytic and the synthetic a priori have been distin­guished. Sentences are called analytic a priori if their tru th or falsity rests solely on the m eaning o f the linguistic expressions contained in them . T hus we arrive at analytic a priori sentences by linguistic analysis or, m ore precisely, by the analysis o f the m eaning of ou r linguistic expressions. By contrast, sentences would be synthetic a p rio n if, though not em pirical, the ir tru th did not rest simply on the m eaning of the expressions contained in them .

So it now becom es clear both which conception o f philosophy u n d e r­lies the language-analytical position and which alternative the objection

raised to it boils down to. Language-analytical philosophy corresponds to the traditional conception o f philosophy as an a priori form o f knowl­edge and in terp re ts the a priori as an analytic a priori. With reference to the explanation o f the subject-m atter o f philosophy ju s t given this m eans that the know ledge presupposed in all u n derstand ing is to be understood as knowledge o f the m eaning o f the linguistic expressions in which understand ing is articulated. A nd the objection to the lan­guage-analytical positon, as now becomes apparen t, comes down to the alternative: e ither deny tha t the re is an a priori subject-m atter o r claim that the re is a synthetic a priori.

In the objection as it was first abstractly form ulated were com bined (though at first this was not noticed) two diam etrically opposed posi­tions, one em pirical and the o ther metaphysical. T h e only p ro p er way o f dealing with the objection is to discuss each o f them separately. Against the em piricist the linguistic analyst can argue that in language we actually have a sphere o f the a priori as this was ju s t described: we know what our linguistic expressions m ean w ithout always being able to articulate what we thus know; w here we succeed the re result analytic statem ents.

B ut on what should a synthetic a priori rest? It seems tha t one m ust conceive, in the sphere o f the a priori, an analogue o f sense-experience. In this way there arises the idea of a non-em pirical experience, a sp ir­itual seeing, an intellectual intuition. Plato and A ristotle called this intellectual intuition nous; and in Latin this was translated as intuitus. M ore or less explicitly this idea of an intellectual in tuition plays an im portan t role in large parts o f the philosophical tradition. In o u r time it has been taken up and theoretically developed by phenom enology. T h e language-analytical thesis that there is only an analytic, only a lin­guistic a priori can therefo re be seen as a counter-thesis to the idea of an intellectual intuition.

T h e re is o f course ano ther conception of a synthetic a priori which does not involve an appeal to intellectual intuition. This is the K antian conception. K ant rejected the idea o f a non-em pirical experience, an intellectual intuition. H e also related all non-analytic know ledge to em pirical experience. How ever, he believed tha t one can know syn­thetic propositions a priori re lating to experience. T h e ir validity is not ap p reh en d ed in an intellectual intuition but rests on the fact th a t they form ulate the conditions of the possibility of experience. However, it is doubtfu l w hether K ant’s a ttem p t to find an alternative to the analytical and intuitive conceptions o f philosophy is successful. T h e propositions which K ant represen ts as conditions o f the possibility o f experience can

A question of method 9

Introduction 10

also be in te rp re ted as analytic. T o the ‘conditions o f the possibility’ o f experience belongs precisely what is analytically contained in the m ean­ing o f ‘experience’. T h u s one can say tha t w hat K ant has done is to analyse a certain concept o f experience.

Sum m arizing we can say: supposing (a) tha t the critique o f the K an t­ian conception o f a synthetic a priori (which I have here m erely h in ted at) had been carried th ro u g h an d (b) that the idea o f an intellectual intu ition h ad been re fu ted , then the language-analytical conception of do ing philosophy would have been shown to be the correct, because only possible, way o f do ing philosophy. A ssum ing, tha t is, tha t the a priori is characteristic o f philosophy.

In on e’s first confron tation with the specifically philosophical, a priori subject-m atter one is easily misled in to tran sfe rrin g to it the structures which are fam iliar from scientific o r even pre-scientific knowledge. T his is why one points away from w ords to th ings w ithout considering tha t philosophy does no t relate to things in the way the sciences do. A nd even w hen this is adm itted th e re is a tem ptation to distinguish the things o f philosophy and th e ir m ode o f availability from em pirical things bu t nonetheless to conceive o f them by analogy with em pirical things. T h e philosophical subject-m atter is not su rre n ­dered in th e language-analytical conception; it is m erely freed from a naive m isunderstand ing .

For philosophy the dem and th a t we shou ld tu rn o u r atten tion to the things can only m ean: tha t we should conceive o f the a priori sub­ject-m atte r in connection with experience. T h e d anger o f losing con­tact with the things (and that m eans: with experience) arises precisely when a philosophy constructs in the a priori sphere its own fictitious world o f things with its own non-em pirical m ode o f access. Precisely if experience is the only subject-m atter for philosophy then w hat is spe­cifically philosophical can only be linguistic analysis.

T h e last reflection serves to draw ou r atten tion to a questionable assum ption which the line o f th o u g h t p u rsu ed so far shares with the traditional conception o f philosophy. Even if the existence o f an a priori an d its distinction from th e em pirical seems undeniab le , it does not follow that it is m eaningfu l to set the sphere o f the a priori over against the sphere o f the em pirical as a self-contained sphere of knowledge. N or does it follow th a t it is m eaningfu l to distinguish from the em pirical sciences an exclusively a priori enquiry and subject- m atter called philosophy.

T his how ever is to touch on a question which points ahead and which one is unlikely to m ake progress with in the confron tation with

A question of method 1 1 ,

earlie r philosophical positions. A nd at the p resen t stage o f o u r a rg u ­m en t it canno t be tackled at all, fo r we do no t yet possess a un itary conception o f philosophy. Only on the basis o f such a conception could it be decided in w hat way a priori and em pirical en q u iry are to be com bined.

W e m ust no t d isregard this question , th o u g h it m ust be borne in m ind as a question tha t has yet to be decided. T h e prim ary aim of the line o f th o u g h t p u rsu ed so fa r has been to m ake us realize that we have not yet arrived at a definite conception o f philosophy (even if we do no t question the p resupposition o f a pu rely a priori conception o f philosophy). For even if we exclude, in the way previously indicated, th e sphere o f logic an d m athem atics, the rem ain ing sp h e re o f the a priori does n o t am oun t to a unified subject-m atter. We clearly do no t w ant to reg a rd all analytic statem ents which rest on som e defin ition or o th e r (e.g. ‘A bachelor is u n m a rrie d ’) as belonging to philosophy.

So ap rio rity is at best a generic fea tu re o f philosophy; it does no t suffice fo r its specific definition. N or did ea rlier conceptions o f philos­ophy consist simply o f the idea tha t philosophical know ledge is a prion. Similarly, the notions o f m eaning-analysis an d analyticity do not su f­fice to p rov ide the language-analytical conception with a un ita ry con­cept o f philosophy. I f one looks at th e language-analytical lite ra tu re one notices th a t it is n o t ju s t any w ords whose m eaning is investigated. B ut from w here does language-analytical philosophy get its criteria fo r decid ing which words, types o f w ord, and linguistic structu res are to be analysed? Obviously to a large ex ten t fro m its o rien ta tion tow ards trad itional philosophical disciplines and problem s. In so fa r as this is so the objection that th e language-analytical position is only a m ethod an d does n o t possess a unified cen tral question o f its own appears justified . H ow ever, the objection only applies to the existing language- analytical litera tu re . We have yet to see w h eth e r in the very idea o f a language-analytical philosophy th e re may no t be contained a un ita ry fu n d am en ta l question.

L E C T U R E 2

A philosopher in search of a conception of philosophy

T h e confron ta tion with earlier conceptions of philosophy with which I am beginning the in troduction to language-analytical philosophy not only has the aim of justify ing this way o f doing philosophy; it is also in ten d ed as a way o f finding its own central question. T h e first th rust rem ained on the periphery . We m erely in ferred what language-analyt­ical philosophy is from the definition o f the nam e. A nd the idea that philosophy is a pnon was simply taken over from the tradition . N one­theless we did succeed in taking a first step tow ards justification: the apparen tly superficial idea tha t the m ethod o f philosophy consists in an analysis o f ou r linguistic un d erstan d in g was shown to be the defensible core o f the trad itional conception of the a priori character o f philoso­phy. With this first step we have reached the cu rren t self-understanding of language-analytical philosophers. However, this self-understanding is no t adequate, for, as we have seen, it provides no criterion fo r distin­guishing the philosophically relevant words, o r w hat is philosophically relevant in language, from what is philosophically irrelevant. For this we clearly need a definition o f the subject-m atter o f philosophy, som e­th ing which is not given simply by saying tha t the subject-m atter of philosophy is a priori.

H ow should one proceed? O ne could try to m ake distinctions within the sphere o f the a priori, to distinguish d iffe ren t species o f the analytic. O ne could, for exam ple, exclude em pirical expressions which can be defined in term s o f a com bination of p roperties. For exam ple, the sen­tence ‘Bachelors are u n m a rrie d ’ is analytic because ‘bachelor’ is defined as ‘u n m arried m an ’. O ne could try to delim it a class o f expressions which one feels are not em pirical in this sense and which may be th o u g h t to be som ehow (I am deliberately expressing m yself in this vague way) philosophically relevant: words such as ‘good’, ‘tru e ’, ‘action’, ‘belief’, ‘experience’, ‘tim e’, ‘object’, ‘m eaning’.

In search of a conception of philosophy 13

I shall not take this path, though it seems to me prom ising and has yet to be developed. Even if by following this path one succeeded in arriving at useful distinctions one would still lack orientation with respect to the question of which linguistic sphere is philosophically rel­evant. For to be able to decide this one m ust start from a conception of the subject-m atter o f philosophy.

We m ust th e re fo re enquire directly as to the them e, or fundam ental question, of philosophy; and , specifically, about the them e, or fu n d a­m ental question, o f language-analytical philosophy. T o this end I shall again choose the m ethod o f confron tation with the philosophical tra ­dition. This time we cannot start out from a prelim inary conception of language-analytical philosophy, for as far as the substantive fu n d am en ­tal question is concerned we do not have such a conception at o u r disr posal. We must, the re fo re , take as our starting-point one or m ore fu n ­dam ental questions of the philosophical tradition and see w hether, when in te rp re ted language-analytically, they yield a fundam ental ques­tion o f language-analytical philosophy. I shall consider in tu rn th ree concepts that are central to traditional philosophy, and in this way we shall gradually in troduce ourselves into the fundam ental question of language-analysis. T hese are (1) the concept o f being which was ch ar­acteristic o f the ancient conception o f philosophy (2) the concepts o f consciousness and experience towards which m odern philosophy has been orien tated and (3) the concept o f reason which - in the Socratic question- stands at the beg inn ing of o u r philosophical tradition. O f course we cannot with these th ree points o f orientation exhaust what tradition can teach us about the fundam ental philosophical question. T h e orientation towards o th e r perspectives o f philosophical though t could also have o ther consequences for the sketch of a language-analytical position. So to this ex ten t the following a ttem p t rem ains consciously incom plete and one-sided.

You m ight also ask: can one in the end really be satisfied with a his­torical o rientation tow ards existing conceptions of philosophy? Even if it can be shown tha t these traditional conceptions are only genuinely realized in linguistic analysis this would only justify the language-ana­lytical m ode of enquiry relative to these positions. B ut how are these positions themselves to be justified? This leads to the question: how does one justify a certain conception of philosophy in and for itself ra the r than simply relative to o ther conceptions? T his question we should keep in m ind in the discussion o f the particular conceptions. A lready the discussion o f the first conception will lead to im portan t hints how to answer this question.

Introduction 14

It was Parm enides who first posed the question about being as the fundam ental question o f philosophy. But it is in Aristotle, at the beg in ­ning o f his Metaphysics, that we first find an a ttem pt at an in troduction to this question as the fundam ental question o f philosophy. T h e re is much tha t is im portan t for our u ndertak ing (independently of the o ri­entation towards being) to be learned from it; I shall th e re fo re p resen t it in ra th e r m ore detail.

A ristotle first develops a prelim inary idea o f philosophy, i.e. a form al indication o f what is to be understood by philosophy. Only a fu rth e r reflection on how this idea can be concretely realized will lead to a d e f­inite conception o f philosophy.

This prelim inary idea is that o f know ledge o f the highest and most general grounds. T his prelim inary conception o f philosophy as a m ost general an d ultim ately justifying form of know ledge rem ained , despite variations in in te rp re ta tion , largely dom inan t in the trad ition that fol­lowed. H usserl, for exam ple, still characterizes his prelim inary concep­tion o f philosophy as the ‘idea o f a science an d ultim ately a universal science with an absolute foundation .’1

A ristotle arrives a t this prelim inary conception of philosophy in two ways. O n e2 consists in an investigation o f w hat in general is understood by the w ord ‘philosophy’ - in A ristotle it is the word ‘wisdom ’, sophia. T h e o ther starts from the fact tha t we m ean by ‘philosophy’ a p re-em i­nent, highest m ode o f know ledge or enquiry. It would thus seem rea ­sonable to call tha t knowledge ‘philosophy’ which possesses those p ro p ­erties constitutive o f knowledge to the h ighest degree. Now it is constitutive of know ledge that it is general and that one can give re a ­sons fo r w hat is known. It follows from this th a t the highest know ledge is know ledge based on ultim ate and the m ost general grounds.

In this line of arg u m en t the assertion tha t know ledge is general may, in particu lar, seem implausible. Usually know ledge is d istinguished from belief o r opinion: we say o f som eone tha t he knows som ething, and does not merely believe it, if his belief is true and if he can justify it.3 T h e aspect o f justification figures in this definition b u t not th a t o f generality. A nd we would certainly say that th e re can also be know ledge of singular item s.4

To u n derstand A ristotle’s account one m ust bear in m ind tha t he is not speaking of know ledge in contrast to belief o r opinion. In term s of the distinction A ristotle here has in m ind, believing, know ing and ques­tioning all belong to the same level and are as a whole contrasted with a lower cognitive capacity he calls ‘experience’ (empeiria). In his in tro ­duction Aristotle no t only contrasts philosophy with ‘low er’ m odes of

knowledge; he regards it as the h ighest possibility of cognitive behav­iour.

In a quite m odern -sound ing discussion he distinguishes th ree levels of cognitive behav iour5 (or of the cognitive com ponen t o f behaviour). T h e lowest level is tha t o f perception . By m eans o f perception a hum an being or anim al responds to env ironm ental stim uli in accordance with a given behavioural schem a. A h ig h e r cognitive level is that which A ris­totle calls experience-capacity and which m odern psychology calls learning-capacity: th ro u g h association one learns from experience. T h e repeated percep tion o f a phen o m en o n A to g e th er with a phenom enon B has the resu lt that w hen A h appens we expect B; o r we learn tha t if we do B we achieve A an d in this way form a new behavioural schema. We have the th ird cognitive level, according to A ristotle, when th e re is fo rm ed betw een A and B not a m erely unarticu la ted association which m anifests itself m erely in behaviour, bu t when this connection is m ade explicit in the belief (or knowledge) tha t ‘w h e n e v e r^ , then B ’ or tha t ‘all A are B \ We do no t find this capacity in the o ther anim als, says A ristotle, only in m an; an d we can supp lem en t this by saying that only m an possesses a system o f signs in which it is possible to fo rm universal ‘if - th e n ’ sentences o r ‘all’ sentences. We can also form ulate this m ore precisely as follows. O nly m an possesses a language in which there are singular (‘this . . .’) and hence also particu lar (‘som e . . .’) and universal (‘all . . .’) sentences; and only when an organism possesses such a system o f signs can it d istinguish between singular, particu lar an d universal states o f affairs. T h e associative state o f affairs - the connection between A an d B - is already general; bu t this is a quasi-generality which does no t yet stand in any defin ite relation to singularity and in which, th e re fo re , th e re is as yet no distinction between ‘all . . .’ and ‘m any . . .’ C onnected with this is the fact th a t the general in ‘ex p eri­ence’ is no t yet an object. T h e organism m erely behaves in the a p p ro ­p ria te way an d w hat it has in m ind is the individual th ing given in p e r ­ception. For this reason it is only the cognitive level o f know ledge which A ristotle characterizes as general. O nly w hen the general becomes an object and stands in a d e term in a te rela tionsh ip to the individual is th e re a rela tionship to the general.

A fu rth e r characteristic o f know ledge in contrast to experience, according to A ristotle, is justification. T h e universality o f an ‘all’-sen- tence clearly belongs to a justification-context. W hoever asserts ‘all . . .’ can be asked fo r a justification or have his a tten tion draw n to coun ter­reasons. T h e relation to justification is in fact a fea tu re not ju s t o f ‘all’- sentences b u t o f all assertoric sentences, all assertions in so far as they

In search of a conception o f philosophy 15

Introduction 16

make a truth-claim . I shall exam ine this connection later. W hat A ris­totle calls ‘experience’, on the o th er hand , the cognitive faculty which is not articu la ted in sentences, does not involve reasons and counter-rea- sons.

So w hen A ristotle refers to justification as a characteristic of knowl­edge h ere too the contrast is no t with belief bu t with experience. Knowl­edge, it is true, is distinguished from belief by the fact that som eone who knows som eth ing can justify what he knows. B ut the person who only believes or doubts about w hat the o th er person knows also stands in a possible justification-context, som ething which is not tru e o f the person who disposes of a co rrespond ing state of affairs in the m an n er o f association on the basis o f ‘experience’.

T h e distinction o f the th ree levels o f cognitive capacity that Aristotle makes at the beg inning of his Metaphysics is still not obsolete. N everthe­less we may not find com pelling the construction of the idea of a h ighest science on the basis o f the two characteristics which result from this distinction. O ne can, however, reach the sam e result m ore directly, sim ­ply by conceding tha t by a h ighest science one can at any ra te not u n d ers ta n d a restric ted science. It follows from this th a t no science can be reg a rd ed as the h ighest science which is restricted in its scope and in its justification.

We shall now pause a m om ent and consider what we can learn from this in troduction o f a p relim inary conception regard ing the question of how it is possible to in troduce a conception of philosophy. A ristotle arrived at his prelim inary conception of philosophy by reflecting on (a) what one u n derstands by the w ord ‘philosophy’ (or ‘w isdom ’) and (b) a p articu lar aspect o f this understand ing , nam ely, that by ‘philosophy’, in contrast to the particu lar sciences, is m ean t a highest science, a p re ­em inent fo rm of knowledge. My appeal to tradition is simply a variant (b roadened by the historical dim ension) o f such an appeal to a p re lim ­inary understan d in g . B ut is such an appeal to the prelim inary u n d e r­standing o f the w ord com pelling? Can we no t free ourselves from it and sketch ano ther, p e rh ap s m ore adequate, conception o f philosophy? Certainly we can sketch o th er conceptions o f ‘philosophy’. B ut what does it m ean to say tha t they could be m ore adequate? M ore adequate to what? We see th a t it would only be a m atte r o f a d ispu te about words. T h e re is no such th ing as the correct m eaning o f a w ord. It is natu ra l when talking abou t ‘philosophy’ to m ean w hat corresponds to the o rd i­nary prelim inary u n d erstan d in g o f this word. But the re is no th ing sacred abou t this. A nyone is at liberty to in troduce an o th er m eaning, p rov ided he can distinguish it clearly from the usual m eaning.

In search of a conception o f philosophy 17

Now this may ap p ear unsatisfactory. In this question o f what is to be understood by philosophy, thus at the crucial starting-point o f our enquiry, are we to rem ain subject to arbitrariness and a boundless re l­ativism? Does not this appear wholly incredible?

In philosophy if one gets into a situation like the p resen t one, one should ask oneself w hether one does not in reality m ean som ething d iffe ren t from what one thinks one means. We think we know that the m eaning o f ‘philosophy’ cannot be anything arbitrary. But on the o th e r hand it is plainly absurd to argue about the correct m eaning of a w ord. I f the re fo re there is som ething correct about o u r feeling tha t the m ean­ing of ‘philosophy’ cannot be anyth ing arbitrary , then we m ust in reality m ean som ething else. It is possible tha t som eone will now say: ‘Precisely. T h e e rro r lies in your language-analytical p rocedure . T h e m eaning o f the word ‘philosophy’ may be arb itrary ; but not what philosophy is.3

W hoever talks like this does no t know what he is saying. For so long as the m eaning of the w ord is no t settled the question: w hat is philoso­phy? can only m ean: w hat does the word m ean?

Probably then we m ean som ething else w hen we assum e that the m eaning of ‘philosophy’ cannot be arbitrary. Let us consider w hat it would be like to ask a similar question of an o th er science, botany for exam ple. We would say: o f course it is a verbal question w hether a p a r ­ticular scientific subject-m atter is so called; b u t this subject-m atter - the study of plants - exists, irrespective o f which te rm we use to describe it. H ere we are dealing with som ething that is there , a specific sphere o f objects. B ut one could hardly say tha t philosophy deals with a specific sphere of objects. H ow ever, ju st as in the case o f the particu lar sciences we say ‘th e re exists a sphere of objects M ’ in the case o f philosophy we can say ‘th e re exists such-and-such a specific m ode o f know ledge’, e.g. ‘there exists - in idea at least - a h ighest form o f knowledge, regardless o f w hether we call it philosophy, o r som ething else’. A nd we could then perhaps go on to say: ‘the re exists — in idea at least - a most general an d ultimately justify ing form of knowledge, regardless of w hether we call it philosophy or som ething else’. However, this way of avoiding m ean- ing-relativism clearly leads to a fo rm of dogm atism and hence basically back to relativism. If we concede the word ‘philosophy’ to an opponen t who declares tha t he would like to m ean som ething d iffe ren t by ‘p h i­losophy’ and rem ain con ten t m erely to say ‘bu t there exists the m ode o f enquiry X ’, then others a re free to say: ‘there is also the m ode o f enquiry Y .’ B ut in th a t case which m ode o f enquiry one engages in would be a m atter o f arb itrary choice.

T h ere thus em erges a difficulty which has no parallel in the o th er

I ntroduction 18

sciences: for th e re we are not concerned to show th a t it is n o t arb itrary to engage in th e m ode of enquiry in question. W hat is given in the case of a science is sim ply the sphere o f objects. In the case o f philosophy, on the o ther h an d , we are not th ink ing o f a particu lar sphere o f objects bu t o f a m ode o f know ledge or enquiry , thus a particu lar activity. T h e non-arb itrariness (and tha t means: legitimacy) o f an activity, however, can only reside in the non-arb itrariness o f the motivation fo r engaging in this activity. I f we are reluctan t to accept th a t the w ord ‘philosophy’ stands for som eth ing arb itra ry then it would ap p e a r that w hat we m ean is th a t it canno t stand for som ething for which the re is an arb itrary m otivation. In this case w hat ‘exists’ is not a sp h e re o f objects. N or is it an activity. I t is a motivation. A conception o f philosophy is to be reg a rd ed as a suggestion to which o th er suggestions can be opposed. T h ese suggestions are to be understood , firstly (and incidentally), as suggestions as to how the w ord ‘ph ilosophy’ is to be understood , and, secondly (and essentially), as suggestions reg a rd in g the engaging in a particu lar fo rm of enquiry. T his provides us with an answ er to the question: w hat is it to legitim ate a conception o f philosophy, not ju s t historically, relative to a given prelim inary u n d erstan d in g , but abso­lutely? T o in troduce som eone to philosophy, and to a particu lar con­ception o f philosophy, is to show the m otivation to engage in this activ­ity to be p re-em inen t vis-a-vis o th e r motives.

A ristotle already saw it like this in his in troduction . F rom the outset his conceptual in troduction goes h an d in h an d with an in troduction o f m otivation. We delight, he says, in the cognitive as such, already in p e r­ception an d especially in seeing; an d the h ig h e r the cognitive level the m ore estim able it is.6 T o show th a t it is the cognitive as such to which we are m otivated A ristotle believes he has to isolate the cognitive from the context o f behav iour.7 T h e re fo re the h ighest m otivation within the cognitive sphere belongs to know ledge w ithout practical purpose, p u re theory .8 In this text A ristotle is con ten t to re fe r to existing opinions; besides he m erely shows that th e re is a m otivation to the cognitive an d tha t w ithin the cognitive sphere th e highest m otivation belongs to th e ­oretical science and, ultim ately, to philosophy. B ut A ristotle though t he could show tha t p u re theory is the activity m ost worth striving after, th e highest possibility o f happ iness.9 T his thesis is based on the following two prem ises: (1) tha t activity is th e one m ost w orth striving after which is self-sufficient and im m utable (2) theory is th e activity which possesses these properties. N either o f these prem ises can seem evident to us today. N or do I know o f any o th e r a rg u m en t tha t can justify the m oti­

In search o f a conception o f philosophy irr

vation to p u re theory as a way o f life w orth striving a f te r for its own sake, let a lone the one m ost w orth striving a fte r, as valid fo r everybody.

Let me sum m arise. (1) T h e u ltim ately decisive in troduction o f a con­ception o f philosophy which provides m ore th an a historical legitim a­tion and which is no t relative to a given prelim inary u n d ers ta n d in g is an in troduc tion in term s o f m otivation. (2) T h e A ristotelian in tro d u c ­tion in term s o f m otivation is no t convincing. W hether this invalidates the A ristotelian prelim inary conception, o r w hether it is still possible to p rovide a convincing in troduc tion o f this p relim inary conception in term s o f m otivation, a re questions to which I shall be re tu rn in g in con­nection with the th ird trad itional gu id ing notion , tha t o f reason (Lec­tu re 7).

We m ust now exam ine how A risto tle gets from the prelim inary con­ception o f philosophy as a science which (1) is universal an d (2) rad ica l­izes the aspect o f justification , to his particu lar conception o f philosophy as en qu iring into being or ‘being as being’.

In Metaphysics 1 , 2 , w here A ristotle in troduces the prelim inary con­ception, he indicates a specific concrete e labo ra tion10 o f it which is clearly P laton ic:11 every particu la r science is being conceived - on the m odel o f geom etry - as a deductive theory which justifies the p ro p o si­tions possible in this sphere by deducing them from the h ighest and most general prem ises in this sphere : the axiom s of this science. Both the justification and the generality possible w ithin the particu la r science are lim ited. T h e justification is lim ited because the axioms are assum ed as hypotheses; they are no t them selves justified . T h e generality is lim ited to the p articu la r sphere . T his perspective leads to the idea o f a h ighest science the task o f which would be to derive the prem ises o f the p artic ­u lar sciences from one o r m ore sup rem e principles, from which the prem ises o f all o th e r particu la r sciences a re to be derived. T his w ould provide a concrete conception o f philosophy and its tasks vis-a-vis the particu la r sciences th a t co rresponds exactly to the p relim inary concep­tion developed by A ristotle: the aspects o f h ighest generality and ulti­m ate justification coincide in th e idea o f a science which derives all know ledge from su p rem e princip les (= m ost general g rounds).

T h e P latonic idea o f philosophy as a deductive system based on a sup rem e princip le, o r sup rem e principles, reta ined a strong attraction in the subsequen t h istory of philosophy rig h t up to G erm an Idealism with its dialectical systems. H ow ever, this idea o f a substantive universal science was rejected as unrealizab le by A ristotle, on the g rounds th a t it p resupposes a m istaken theory of science. T h e u ltim ate substantive

Introduction 20

prem ises of the particu lar sciences, according to Aristotle, are ir re d u ­cible, not fu rth e r derivable.12 T h is criticism can be m ade even m ore stringent from a m odern perspective. Aristotle still accepted the Pla­tonic conception of the particular science as a deductive theory on the model o f geom etry .13 This contradicts the concept of an em pirical sci­ence which g rounds its propositions so to speak from below, by expe­rience, and not from above by reference to given prem ises. If already within the particu lar science the m ovem ent o f justification is from below upw ards ra th e r than from above downwards then the idea o f a radicalization of substantive justification by derivation from even h igher up is ru led out in advance.

For A ristode the Platonic conception of philosophy was the most na t­ural possibility of giving a concrete sense to his prelim inary conception. I f he nevertheless cam e to recognize that this conception is incapable o f realization and yet wished to hold fast to his prelim inary conception, then a new approach was requ ired which would determ ine, within the fram ew ork of this prelim inary conception, the relation o f philosophy to the sciences in a fundam entally d iffe ren t way. This new conception is that o f ontology.

L E C T U R E 3

Ontology and semantics

It is at the beginning of Book IV o f his Metaphysics that Aristotle first in troduces his new conception of philosophy. ‘T h e re is a science which studies being as being.’ Indeed the special character of this science vis- a-vis the o ther sciences is supposed to consist in the fact tha t whereas the latter investigate a particular sphere o f being philosophy investi­gates being as being.1 W hat distinguishes the concept o f being, for Aristotle, is tha t it is the most general concept.2 F or of everything and anything one can say tha t it is. Everything and anything, therefo re , can be called being.

Clearly Aristotle arrives at his new conception o f philosophy by d ro p ­ping the aspect o f justification from the prelim inary conception devel­oped at the beginning and settling exclusively for the aspect o f highest generality. T h e highest, p re-em inent science, called philosophy, is un i­versal, but does not have a justificatory role in relation to the particular sciences. This conception, since it is orientated towards the concept o f being {on), leads to the conception o f philosophy as ontology.

T o enable us to understand the specific character o f this conception o f philosophy as ontology (and this means: a conception tha t is based on the concept o f being) we can think of an analogous reflection using a concept o f m odern philosophy, that o f an object. Each science has to do with a specific sphere o f objects, objects o f a specific kind, and with a specific m ode o f accessibility. Can one say that it is also the task o f the particu lar science to them atize this object-sphere as such and the pecul­iar m ode o f givenness which distinguishes it from o ther object-spheres? O ne could argue about this. Since the concepts which characterize an object-sphere as such are not o f merely gradually h igher generality than the concepts within the object-sphere, one can say that the object- sphere as such, e.g. that o f physics, the arts, mathem atics, is the subject- m atter o f the philosophy o f physics, philosophy o f art, philosophy o f

Introduction 22

m athem atics. H usserl called such a them atization o f an object-sphere a ‘regional ontology’.3 W hat is discussed in such an ontology is what it is to be an object o f the relevant sphere. B ut one can go fu rth e r an d ask: w hat is m eant by an object as such, in abstraction from a particu lar sphere o f objects? In this way one arrives at the question about objects as objects, ju s t as previously one arrived at the question about beings as beings.

A re we here dealing with two analogous questions or do both ques­tions have the sam e m eaning? Obviously this depends on w hether or not the two expressions ‘being’ and ‘object’ have roughly the same m eaning. W hereas for us the expression ‘being’ (‘Seiendes1) is a philo­sophical term o f art, the G reek philosophers w ere able to take the expression ‘on from o rd inary language.4 As ‘being’ (on) is the participle of the verb ‘to b e ’ (einai), and the w ord ‘is’ is notoriously am biguous, difficulties arise here , about which I shall speak later. At presen t, how­ever, we can confine ourselves to the characterization ju s t given: every­th ing and anyth ing is a being, because o f everything it is significant to say th a t it is. O ne could doub t this, po inting ou t tha t th e re exist objects tha t are not, e.g., objects o f phantasy. B ut to this one could reply that by saying ‘they exist’ one says o f them tha t they are. T hus it seems that even those objects which in a certain sense are not in an o th er sense som ehow are.

Now what is m ean t by the w ord ‘object’? This w ord too, in the com­prehensive sense in which it is used in philosophy, is a term o f art. In o rd inary language we are inclined to call only m aterial objects - and then only those which are no t persons - objects, an d not e.g. events or num bers and o th e r abstract objects; although then again one does speak of the object o f a discussion. W hat is m eant by ‘objects’ in philos­ophy has its basis no t in w hat we call objects in o rd inary language but in what we m ean by the w ord ‘som eth ing’ in o rd inary language. One could say: by ‘object’ is m ean t everything tha t is som ething. However, this form ulation is linguistically faulty, because the word ‘som eth ing’ is not a predicate, bu t an indefinite p ronoun . T h e e rro r o f speaking in such a way would becom e even m ore strikingly evident were one to say: an object is a som ething. N onetheless philosophers have frequently talked like this in the tradition . T hus A ristotle, e.g., coined the expres­sion ‘a this’ (tode ti) fo r ‘object’. We m ust try to avoid such ung ram m ati­cal expressions, and to this end there is no alternative bu t to go back even fu rth e r to th e linguistic background. T h e re is a class of linguistic expressions which are used to stand for an object; and here we can also say: to stand fo r som ething. T hese are the expressions which can func­

Ontology and semantics 23

tion as the sentence-subject in so-called singu lar predicative statem ents and which in logic have also been called singular terms. T h e elucidation of the concept o f an object by recourse to singular term s is certainly also to be found p rio r to language-analytical philosophy. T hus H usserl determ ines the b read th in which he wishes the concept ‘object’ to be und ersto o d by defin ing it as follows: ‘any subject o f possible true pred ications’.5 A nd A ristotle too defined his concept of an object by m eans o f the concept of the hypokeimenon, the subject o f p red ications.6

B ut w hat is m ean t by this rem ains unclea r so long as it is not specified what is to be und ersto o d by a singular p redicative sta tem ent an d by its subject. Now it w ould seem n atu ra l to d istinguish singular predicative statem ents from o th e r predicative sta tem ents by saying that they are those which have a singular te rm as th e ir subject. W e are thus m oving in a circle and req u ire an in d e p e n d e n t criterion fo r recognizing singu­lar term s. T h e following criterion suggests itself: an expression X is a singular term if, w hen it is supp lem en ted by an o th er expression to form a whole assertoric sentence, one can deduce from this sentence an o th er sentence in which X is replaced by ‘som eth ing ’ (or ‘som ebody’).7 By this criterion ‘the n u m b er 3’, e.g., is a singular term , fo r from the sentence ‘T h e nu m b er 3 is sm aller than the n u m b e r 4 ’ th e re follows the sentence ‘Som ething is sm aller than the n u m b er 4 ’; and we can now add by way of elucidation: som ething, nam ely tha t fo r which the expression ‘the nu m b er 3’ stands. Obviously the m ode o f em ploym ent o f singular term s is connected with a system o f p ronom ina l expressions which can take the ir place (pro-nom ina!): ‘som eth ing’, ‘which?’, V h ic h ’, ‘the sam e’. A nd each of these p ronouns we can, trivially, supp lem en t by the w ord ‘object’, by saying, instead o f ‘w hich’, ‘which object’, instead o f ‘the sam e’, ‘the same object’, and , instead o f ‘som eth ing’, ‘some object’. If the w ord ‘object’ is used so broadly , thus if its m eaning is yielded by the use o f these p ronouns, or the singular term s which replace them , then it has the broad sense in tended in philosophy. Only la ter will I exam ine the in tim ate connection betw een the w ords ‘som eth ing’ and ‘the sam e’. But even now I can po in t o u t that, instead o f the criterion ju s t m en­tioned, one could also use th e identity-sign as a criterion of singular term s: an expression is a singu lar term if it can stand on e ither side of ‘is the sam e as’ (or ‘ = ’).8

T h e notion o f ‘objects’, as thus d e te rm in ed , is clearly directly con­nected with that o f ‘being’. T h e thesis th a t o f everything and any th ing we can say ‘it is’ m eans-som eth ing like: w hatever som ething may be at any rate it is. H ence, already accord ing to A ristotle the concept o f being was intim ately connected no t only with th a t o f unity bu t also with that

Introduction 24

of som ething (ti)\ and this connection - tha t every being is also som e­thing and one and vice versa - was also reta ined in Scholasticism u n d er the titles ens, unum, aliquid.

But how can one them atize such a th ing as being as being, objects as objects? W here and in what way do we find such a thing? Clearly not in experience, fo r although we find objects in experience we do not find the object as object, objecthood, being as being, Being (Sein). B ut nor do we reach it by abstraction. We can see this if we reflect on the pecu­liar certainty o f the statem ent ‘E verything is a being’ o r ‘Everything is an object.’ Clearly the certainty o f this statem ent is no m ere inductive o r hypothetical certainty; we have not arrived at it by com paring many objects and progressive abstraction. For if that were so, then it would be conceivable that th e re m ight still be som ething which we had not so far considered tha t could not be te rm ed an object. But this possibility is excluded a priori, because if it is som ething then it is eo ipso an object.

So we do not arrive at what we m ean by ‘an object’, o r by the exp res­sion ‘som eth ing’, by abstraction. H usserl drew attention , in this connec­tion, to the distinction between ‘generalization’ and ‘form alization’9 and called the them atization of the object as object, in contrast to the ‘regional ontologies’, ‘form al ontology’. T hus with the conception of philosophy as ontology the rela tion o f the highest science to the partic­u lar sciences is de term ined in a new way vis-a-vis the Platonic concep­tion: philosophy no longer em braces the particu lar sciences as regards the ir content, but form ally; as ontology it them atizes tha t which all sci­ences form ally p resuppose, ra th e r than principles from which the ir propositions could be derived. W ith this Aristotle explicitly m arked ou t for the first tim e a them atic field which was already in Plato, though only implicitly. Clearly this field o f the form al is a field o f a priori, an a ­lytic know ledge (we have ju st seen tha t we cannot arrive at the concepts in question inductively, em pirically). And equally it is clear that here the dem arcation of a narrower field within the analytic, which we missed when discussing the analytic character of philosophy, begins to take shape.

B ut now w hat is m ean t by ‘form alization’ in contrast to ‘generaliza­tion’? H usserl did not elaborate this and Aristotle did not even explicitly m ake the distinction. O ne could provisionally describe it by saying tha t it p resupposes a move o f reflection. W hereas the particu lar sciences are concerned with the objects o f a dom ain and their properties, the subject-m atter o f ontology is not to be sought in a transcenden t dom ain (for w here should this be?). B ut then the only alternative is that one arrives at this subject-m atter by reflecting on the m anner o f ou r re fe r ­

ence to objects. Such a m ode of enquiry was only developed in the m od­ern period and that explains why it was not possible for A ristotle to use a concept such as that o f form alization. T h e question now arises: in what way do we refer, or can we refer, to objects? We became acquainted with one possibility in the attem pt to fix the philosophical concept o f an object, viz. reference to objects by means o f linguistic expressions. T h ere is a specific class o f linguistic expressions - singular term s - whose m ode o f em ploym ent is such th a t by m eans of them we are able to re fe r to som ething, to an object. A nd if we can only specify what is m eant by the philosophical concept o f an object by recourse to singular term s, the w ord ‘som ething’ and o ther p ronouns, then to the question ‘how can one them atize som ething like the objecthood of objects (or being qua being)?’ it would seem plausible to answer: only by reflecting on the use o f the corresponding linguistic expressions.

From the linguistic perspective we can at any rate give a definite m eaning to the distinction between generalization and form alization. W hat I have called, in explaining generalization, progressive abstrac­tion clearly rests on the fact that we can subsum e concepts u n d e r o ther, m ore general, concepts. Concepts are principles o f classification and to them there correspond , in language, the so-called general terms o r pred­icates, which, accordingly, can be called classification-expressions.10 An exam ple o f a progressive abstraction would be the series o f predicates: ‘native o f the Palatinate’, ‘G erm an’, ‘hu m an ’, ‘living th ing ’, ‘spatio- tem poral object’. H ere the next p redicate is m ore general than the p re ­ceding one, because it is applicable to all objects to which the preceding one is applicable, whereas the converse is not the case. Now one can say o f each of these predicates, or the co rresponding concepts, tha t it is a classification-expression, or classification-principle. This is clearly to say som ething completely general, for it applies to all predicates, o r con­cepts. But it is clearly no t simply gradually m ore general than any o f the predicates o r concepts themselves, and, therefo re , does not belong to a series o f the kind I have ju st presen ted . F or it is the concepts o r predicates themselves and not the objects which fall u n d er the concepts, o r to which the predicates are applicable, which fall u n d er the descrip­tion ‘classification-principle’ or ‘classification-expression’.

In contrast to the previously nam ed predicates, which are applicable to objects, we arrive at the predicate ‘classification-expression’, o r the predicate ‘pred icate’, by reflecting on the m ode of em ploym ent o f those o r o f any predicates; and that m eans: by reflecting on th e m ode o f em ploym ent o f a species o f linguistic expression. But now this is a p ro ­cedure of basically the same kind as tha t by which we arrived at the

Ontology and semantics 25

Introduction 26

predicate ‘singular te rm ’ o r ‘expression which stands for an object’. Now the specifically language-analytical position would be th a t one can only explain w hat is m eant by ‘concepts’ by reference to the use o f p red ­icates, and that one can only explain w hat is m ean t by ‘objects’ by ref­erence to the use of singular terms.

B ut in what sense is this reflection on the m ode of em ploym ent of expressions to be understood as form alization? A b rief indication of linguistic distinctions is here required . T h e concept o f form is clearly closely connected with that o f structure. O ur linguistic expressions are structurally com posite at two levels: (a) that o f the com bination of phonem es to form the smallest m eaning-bearing units, i.e. words or m orphem es and (b) that o f the com bination o f m orphem es to form sentences. ‘S tructural com position’ m eans that the composition is rule- governed: the sm aller units cannot be com bined with others arbitrarily b u t only in so far as they are elem ents o f certain classes. Now at the level o f the com bination o f m eaning-bearing units to form sentences th e re are two possible m odes of analysis: on the one hand, the syntac­tical which investigates the external o r ‘surface’ com position o f sen­tences and has regard neither to the m eaning of the sentences nor to tha t o f the sentence-com ponents. T h e classes o f sentence-com ponent are no t defined semantically; ra the r they are determ ined exclusively by the so-called principle o f ‘d istribu tion’, i.e. the substitutability o f their elem ents for one another, the only condition being that the result m ust also be a sentence. By contrast one calls ‘sem antic’ any m ode o f analysis which concerns the meaning of linguistic expressions. H ere it can be a m atte r either o f the m eaning of the individual words or o f how the m eaning o f sentences depends on the m eanings o f their parts. In lin­guistics it was only a few years ago tha t attention began to be paid to this second m ode of enquiry, whereas within philosophical semantics it has been, since Frege, the dom inant one. T h e classes of sentence-parts to which one m ust re fer in this m ode o f enquiry are not the classes of syntactic sentence-parts, fo r the fo rm er cannot be defined indepen­dently o f m eaning; ra th e r they are sem antic classes which are defined in term s o f the sort o f contribution the m eaning o f th e ir elem ents makes to the m eaning o f a larger unit, ultim ately a sentence; o r in term s o f the elem ents o f o ther sem antic classes with which they can be com­bined. As exam ples of such sem antic classes one can m ention both sin­gu lar and general term s. T h e reason why the definition o f singular term s is so involved is because one is dealing no t with a syntactically definable class bu t with a semantic class which is defined by the m ode o f em ploym ent o f the expressions, o r by the so rt o f contribution they

Ontology and semantics 27

m ake to the m eaning o f a sentence. T hese two sem antic classes also provide a sim ple exam ple of the com bination o f the elem ents o f two sem antic classes; fo r the com bination o f an expression which stands for an object with a classification-expression yields a (singular) predicative assertoric sentence. How precisely this is to be u n derstood will occupy us at length later on.

Now one can call any structural them atization o f language - the pho ­netic, the syntactic and the sem antic - form al. It seems natu ra l, th e re ­fore, to distinguish between m aterial {inhaltlich) an d form al semantics. Form al semantics arrives at its subject-m atter th ro u g h a process o f fo r­malization which is linguistically symbolized by rep lacing the m aterial expressions by variables, with the u n d erstan d in g th a t these stand for an a rb itra ry expression of a sem antic class. In this way the sem antic form of a com posite expression can be exhibited, e.g. one can use the letters ‘a , ‘b’, V for singular term s, F \ £G’ for general term s and can then exhibit the form of a predicative sentence with one singular term (e.g. ‘P eter is crying’) by ‘Fa, and that o f a predicative sentence with two singular term s (e.g. ‘Peter hits Paul’) by ‘Fab’.

T h u s if it should be tru e tha t we can only arrive at, and can only them atize, the category o f object by reflecting on the co rrespond ing linguistic expressions, then it would be clear w hat is m eant by ‘form ali­zation’. For the them atization in question would no t be ju s t any the­m atization o f linguistic expressions; it would be a them atization o f their sem antic form . You could point out th a t I have now only described w hat is m eant by ‘form al’ and ‘form alization’ w ithin semantics in term s o f the contrast with m aterial sem antic questions, and that it is no t clear w hether this distinction corresponds to the distinction between form al­ization and generalization, thus betw een a form al and a m aterial approach to objects. T hese two things m ust indeed be d istinguished; they are, how ever, connected. O ne can show this by means o f the sym­bolization I have ju s t illustrated. T h e scientist w ho is dealing with objects o f a certain sphere, and the ir p roperties, uses, am ong o th er sen­tences, sentences o f the fo rm ‘Fa. W ith the singular term s which he uses in place of ‘a he refers to certain objects, e.g. with the expression ‘the m oon’ he refers to the m oon. T h e m aterial sem anticist can enquire abou t the meaning o f this an d o ther expressions. B ut when one fo rm al­izes w hat the scientist (or any o ther language-user) is do ing when he refers with this singular term to this object, thus if one asks what in general it m eans to re fe r to an object an d w hat in general it m eans to speak o f an object (‘w hat an object qua object “is” ’), then one does so precisely by form alizing the m aterial sub ject-m atter o f the sem anticist

Introduction 28

and asking about the form al m eaning o f singular term s. Objectual fo r­m alization finds its sense in sem antic form alization.

I f this should be the only possible way o f understand ing the form al- ization-step by which ontology is constituted, then one would have already shown tha t ontology is only realizable in a language-analytical philosophy u nderstood as form al sem antics.11 At this stage o f o u r reflections I cannot yet assert so m uch. In discussing the m odern , so- called transcenden ta l concept of philosophy we shall becom e acquain ted with an o th e r kind o f reflection in which reference to objects is not u n derstood linguistically; and only m uch later will I really attack this trad itional m o dern conception (Lectures 20 and 27). So far only this m uch has been said: (I) fo r Aristotle, and with him the whole p re ­m odern ontology, th e re was no possibility o f explaining what d istin ­guishes the form al concepts investigated in ontology from o ther con­cepts. (2) T h e recourse to form al semantics offers at least one possibility o f explain ing this distinction.

You will have felt it to be problem atic tha t for the A ristotelian con­cept o f being I substitu ted the concept o f an object, and th a t I o rien ta ted myself exclusively tow ards the latter. My reason for proceeding in this way is tha t the concept o f an object is less am biguous and that, conse­quently, certain aspects o f w hat is m eant by ‘being’ can be b rough t out by it m ore clearly. B ut when one proceeds in this way essential perspec­tives o f the traditional ontology rem ain unheeded . W hat makes the expression ‘being’ so difficult is its connection with the am biguous ver­bal expression ‘is’. F or the tim e being it will suffice to a ttend to two o f this w ord’s various m odes o f em ploym ent. We sometim es use the w ord ‘is’, though adm ittedly ra th e r seldom , with a singular te rm or a p ro ­noun an d w ithout a supp lem enting pred icate expression (e.g. in the sentence ‘G od is’). H ere it m eans ‘exists’. A second m ode of em ploy­m ent, and the one which is most frequen t in o u r language, is as the so- called copula in a predicative sentence (e.g. ‘T h e sky is b lue’). Now it would seem tha t w hen one speaks o f ‘a being’ only the use o f ‘is’ in the sense o f ‘exists’ is involved, for ‘a being’ m eans ‘som ething which is’, thus the w ord ‘is’ is h e re used w ithout a supp lem enting predicate- expression. So w hereas the expression ‘is’ is used am biguously, the sub­stantival expression ‘being’ seems to be univocal and have the sense ‘ex isten t’. W e m ust be all the m ore su rp rised then w hen we find th a t in his ontology A ristotle is prim arily o rien tated towards the copulative ‘is’. A nd it m ust seem even m ore su rp rising tha t he understands this ‘is’ as the ‘is’ o f a ‘being’12 and that he takes the ‘being’ to be tha t for which

the predicate stands, hence the being-thus-and-so {das So-seiend-Sein) o f the object.

O ne could try to in te rp re t this in a harm less way: why should not the predicate in a sentence like ‘T he sky is blue’ also stand fo r som ething, in this case the blueness o f the sky? This conception would be com par­atively harm less because som ething like blueness is indeed an object (something) and thus could also be designated as a being. O u r criterion o f objects fits: the expression ‘the blueness’ is a singular term . B ut in m oving from ‘T h e sky is b lue’ to ‘the blueness o f the sky’ we have to change the form of the expression; the predicate ‘is b lue’ has been changed by a so-called nom inalization into the singular te rm ‘the blue­ness’. A nd since singular term s and predicates are sem antic classes, we m ust also u n derstand this gram m atical change as one of sem antic form . L ater I will show that the nom inalized form is semantically secondary relative to the predicative form . I cannot assum e this here. B ut then I do not need to, for A ristotle himself, in his debate with Plato, does no t m erely regard objects like blueness, hence abstract objects, as sec­ondary; he rejects them altogether. W hatever on e’s attitude may be to the Platonic problem of the relation o f the blueness of an individual object to the blueness as such, A ristotle rejects not only the latter bu t also the fo rm e r.13 For him such abstract objects do not exist, only con­crete objects with the ir predicative determ inations. Aristotle was cer­tainly too casual in his approach to the com plicated problem o f abstract objects. H e was, however, undoubted ly right to reject the reduction o f predicative determ inations to abstract objects. Even if one accepts abstract objects, they in the ir tu rn have predicative determ inations.

B ut how should one positively understand predicative de term ina­tions if they are not objects? If one approaches the problem im partially then I th ink one has to say: if we divide a singular predicative sentence into its sem antic com ponents both o f them - the singular te rm and the predicate - have a m eaning, i.e. we u nderstand both of them ; but only one of them - the singular term - stands for an object. I f only singular term s stand fo r objects then it seems to follow tha t form alization of the linguistic expressions reaches fu rth e r than the form alization o f objects.

An explicitly sem antic enquiry was, however, unknow n to Aristotle. This is why he called predicative determ inations both onta (beings) and legomena (som ething said).14 In the M iddle Ages this undecidedness becam e the starting-poin t o f the nom inalism controversy. A ristotle refu sed to follow Plato in treating the m eaning o f predicates as an inde­p en d en t object. However, because he failed to perceive th e sem antic

Ontology and semantics 29

Introduction 30

dim ension he inevitably objectified the ir m eaning. T h e result is a pecu­liar extension o f the concept o f being (on). It is — together with the concepts of the one and o f som ething - m ore com prehensive than that o f an object (tode ti).

T h e title ‘ontology’ now begins to iridesce. It would have an unequivocal sense if one were to define it, as I initially did, and as is usual in analytical philosophy, in term s o f the concept o f an object, or, which am ounts to the same thing, in term s o f the concept of being in the sense of existence; ‘ontology’ would then m ean ‘theory o f objects’. In contrast to this the introduction o f ontology by Aristotle, which becam e standard in the tradition, contains a tension which was not resolved in the tradition. T his tension is a consequence o f A ristotle’s dual orientation: on the one hand, to the - objectual - fo rm ula ‘being as being’; on the o ther hand , to the verbal form ‘is’. H e lets him self be guided by this verbal form even where it does not connote being in the sense o f existence, i.e. w here the ‘is’ is not the ‘is’ o f a being; and since the form ula ‘being as being’ nonetheless rem ains the guiding principle, the form alizing approach, which in itself would have led away from the restriction to the problem o f objects, is again being cast into an objectual term inology. A ristotelian ontology transcends the form al theory of objects in the d irection o f a form al semantics, bu t in such a way that what em erges is m isin terpreted in term s o f an object-oriented perspec­tive, owing to the lack o f awareness o f the sem antic dim ension.

T hus if one views the traditional elaboration, essentially determ ined by Aristotle, o f the idea o f a philosophical fundam ental discipline as ontology from a language-analytical perspective (one of reflection on the m eaning o f words) it tu rn s out to be unsatisfactory in reg ard to both o f the aspects in A ristotle’s prelim inary conception o f philosophy. Firstly, in regard to its justification: the object-orientated Aristotelian fo r­m al discipline lacks a foundation in a m ethod of reflection; such a fo u n ­dation would be provided by a form al semantics (though w hether this is the only possible foundation we do no t yet know). Secondly, in regard to its scope: its claim to universality could only seem convincing so long as one rem ained orien tated to objects. But the orien tation to everything (and that m eans: to all objects) appears itself restricted as soon as one focusses on the realm o f the form al itself. T h e perspective on objects then corresponds to ju s t one sem antic form am ong others.

T h e re are two aspects o f this critique of ontology from a language- analytical po in t o f view which I m ust particularly em phasize. Firstly, the critique is not an external one. Both defects rep resen t im m anent diffi­culties. T h e language-analytical perspective was no t necessary to show

Ontology and semantics 31

them up , but only to rem ove them . Secondly, this critique has positive consequences fo r the question o f linguistic analysis as regards its own conception o f philosophy. We d id not u n d ertak e the confron tation with the trad itional basic conceptions m erely in o rd er to be able to contrast the language-analytical conception o f philosophy with them . R ather we did so in o rd er to arrive at such a conception in the first place, a fte r it em erged tha t the orien tation tow ards m eaning-analysis does no t suffice for this purpose. In this reg a rd the following possibility seems so far to em erge. We can follow A ristotle in his sketch of a prelim inary concep­tion o f philosophy (with the reservations about m otivation m entioned in the last lecture). We can also follow him in the developm ent o f his p relim inary conception into the conception of a philosophical fu n d a­m ental discipline which does no t deductively g round the know ledge of the o th e r sciences bu t ra th e r them atizes what they all formally p resu p ­pose. A nd now we only need to follow up the two weak points o f his conception o f ontology with an a ttitude tha t is p rep a re d to reflect on the m eaning o f linguistic expressions and we find th a t the traditional ontology itself points beyond itself to a new conception of the form al science, which, in the shape o f a form al sem antics, underlies all sciences. Form al semantics is, on the one hand , a language-analytical u n d e rta k ­ing: it is sem antics, it analyses the m eaning o f linguistic expressions. On the o th e r hand , it is form al, in the sam e sense tha t ontology was form al; and because it rem oves weaknesses o f ontology, which are incapable of im m anen t resolution , it can lay claim to being ontology’s legitim ate suc­cessor.

T h e re is also an o th er way in which we can come to see the superiority which form al sem antics possesses over form al ontology from the p e r­spective of A risto tle’s own idea o f philosophy. T his is by starting out not from the unclarities o f ontology bu t from the step which A ristotle took from the regional sciences to form al science. I f one construes ph i­losophy qua h ighest science as the transition from the particu lar sci­ences to the form al elem ent com m on to them all, then it is by no m eans obvious that the only such elem ent is objecthood. In science, as also elsewhere, we never simply re fe r to objects, but always in such a way that we m ake predicative statem ents abou t them . B ut most o f the state­m ents o f science, e.g. those in which laws are form ulated , do not con­tain singular term s and only re fer indirectly to objects. It is surely con­sistent then not to restrict the form alization to singular term s, but to ex tend it to whole sentences and all sentence-form s.

N ote m oreover tha t ontology is com pletely preserved w ithin form al semantics. T his is tru e no t only o f the parts o f trad itional ontology in

Introduction 32

which som ething was illegitimately objectified - the predicative d e te r­m inations - but also o f the theory of objects which now proves to be a p art of form al semantics.

W hat still rem ains unclear, however, is w hether this new conception o f a form al fundam en tal discipline has a unitary fundam ental question as ontology did in the question concerning ‘being as being’. So for the tim e being we have simply an ex tended subject-m atter; w hether this can be organized a ro u n d a cen tral question is still unclear. I shall take up this difficulty in the next lecture.

T oday I would simply like to lighten the shock you may have felt when I stated tha t ontology is p reserved within semantics. Even if you agree with me th a t the orientation tow ards the fo rm of linguistic expressions opens up a subject-m atter that is b roader than ontology, you will probably want to qualify this to the effect tha t the new subject- m atter is a linguistic one and thus no longer belongs to the dim ension of ontology, the d im ension o f ‘reality’.

T h e re is repeated here the sam e resistance to the language-analytical position that has already shown itself, in an o th er connection, in the first lecture. L anguage, one thinks, is som ething merely subjective. If one converts ontology, which has to do with reality, into linguistic analysis, then the m ost im p o rtan t thing is lost, even if som ething else is added.

Now it seems to m e tha t in objections o f this kind, even when raised by philosophers, a pre-philosophical motive is p resen t, a reservation sim ilar to that which is en terta ined against philosophy in general by those who are not fam iliar with philosophy. In what way, I would reto rt, does ontology have to do with reality? Certainly not in the way the sci­ences do. It does no t have to do with objects. O bjecthood is not itself som ething real like an object.

But is no t - you could now ask - objecthood, even if no t itself som e­th ing real, the reality o f the real? A nd is not this lost in the reflection on the m erely linguistic?

It is only in the fu rth e r course o f these lectures tha t we will gradually work o u t the categories which make it possible adequately to discuss such questions. So fa r I have barely touched on the essence of the lin­guistic. W e are, the re fo re , not yet in a position to see tha t it is a mistake to speak o f the ‘m erely linguistic’, of language as a m ere m edium between us and reality. At the p resen t stage of our enquiry it m ust suffice to say tha t the opinion ju s t expressed, that the reality of the real cannot be ap p re h en d ed in reflection on the use o f language, arises from a m ere feeling; fo r if it were to be m ore than this one would have to be able to say how som ething like ‘reality’ can be given to us, if no t in

Ontology and semantics 33

linguistic usage. We will soon have an opportunity critically to examine such a positive alternative (p. 61 ff).

But it would in any event be false to think that the distinction between object and meaning is levelled out by the semantic turn. The opposite is the case. It was the old ontology in which this distinction was levelled out; for want of other categories the meaning of a linguistic expression was interpreted as an object. By contrast, there is no reason, from the semantic perspective, for neglecting the object for which an expression stands in favour of its meaning. All linguistic expressions which we understand have, in so far as we understand them, a meaning. Some of these expressions, singular terms, stand for objects. An expression can presumably only have this function of standing for an object in virtue of how we understand it, thus in virtue of its meaning. If this is correct- and it is only later that we will precisely investigate these connections- then the objecthood of objects cannot be thematized independently of the meaning of singular terms. However, that does not mean that the object for which the expression stands and its meaning coincide. The dimension of objects does not lose anything as a result of the semantic approach; rather something is added to it and only thus does it become intelligible.

If we now proceed from singular terms to whole sentences (or to other component-expressions) then, although they have a meaning, it is not clear that we can say that they stand for objects. H ere then the dimension of objects does indeed fall away; and the impression there­fore arises that we are dealing with the ‘merely linguistic’. But the lin­guistic is not the mere sign; it is that which one understands and which many can understand in the same way. It is, therefore, nothing subjec­tive. At present I am only concerned that you should dwell on this understanding in its puzzling familiarity and resist the temptation to force it into traditional categories.

Another reason why I have examined objections I assume you to have towards a formal semantics as extended ontology is that they seem to me to be the same objections that have hindered traditional ontology itself from developing into formal semantics. All the decisive steps of Greek ontology resulted from semantic reflection. However, every step led to an objectifying reinterpretation which concealed the linguistic dimension of the reflection. Thus the problem of being (Sein) and not- being (Nichtsein) was twisted by Parmenides into a problem of what is and what is not (von Seiendem und Nichtseiendem), with the grotesque but con­sistent result that there is just a single unmoved being (Seiendes), because with what is not (dem Nichtseienden) not-being (das Nichtsein) was also

Introduction 34

excluded.15 Plato discovered for the first time, in reflection on questions of definition, the meanings of predicates - and immediately objectified them, in his doctrine of ideas, into supersensible entities. Finally, Aris­totle started out from the form of the singular predicative sentence and nonetheless developed on this basis an objectual ontology.

But what is it then that hindered traditional ontology from dealing with the semantic dimension as such, and which also makes it so diffi­cult for us? Why is it that we too unconsciously interpret the non- objectual understanding of linguistic expressions in an objectual way? Clearly it is due to the fact that when we speak of something this is, by definition, an object.

We can thus thematically direct ourselves only to objects; under­standing is essentially unthematic. So if we wish to investigate the mean­ing of our expressions we find ourselves faced with peculiar difficulties. The meaning is not what we are naturally directed towards; we must, therefore, carry out a reflection which inhibits our natural directedness. And then we must also be careful not to objectify that with which we are dealing in this reflection. My saying ‘with which . . . we are dealing’ already contains, in the pronominal ‘with which’, an objectification. And is there not an objectification already contained in speaking of ‘the meaning’ of an expression? Indeed, and we shall see later that, and to what extent, this is an im proper way of speaking.

L E C T U R E 4

Has formal semantics a fundamental question?

Because Aristotle orientated himself not only towards the expression ‘being’ (Seiendes) but also towards the expression ‘is/to be’, traditional ontology reaches beyond a m ere theory of objects or theory of sub­stance. This surplus really already belongs to the broader field of a formal semantics. If one asks why Aristotle included predicative deter­minations as well as objects in his formal thematic the answer must be: because predicative determinations, although not objects, are at least determinations of objects. So the orientation towards the category of an object, which is definitive of the whole of traditional philosophy, also determ ined which non-objectual semantic form Aristotle dealt with.

If we now envisage a formal semantics as the formal universal science in place of ontology we must consider whether the formal thematic, as thus extended, still has some sort of unified structure and a centre, so that here too a unitary fundam ental question can be formulated.

T he fundam ental question of ontology is: what is being as being? It is obvious that this formulation of the question is a makeshift solution, for it is framed as though one were asking about the what-being (Was- Sein) of an object. I have therefore reform ulated it so that what is being asked is what it means to speak of an object (or being). This is already to give the question a semantic formulation. But its real semantic coun­terpart is the question: ‘How can one refer to objects with linguistic expressions?’ and this question, it would seem (cf. p. 33), leads back to the question: ‘In what does the meaning of a singular term consist?’ And if we wish to avoid speaking of a meaning as an object, this ques­tion can be formulated as follows: ‘What is it to understand a singular term ?’ Analogously, if we extend the sphere of the formal thematic beyond that of singular terms, we can ask with respect to any semantic class: ‘What is it to understand an expression of this form?’ T he for­mulation ‘What is it to understand . . . ?’ is not completely clear. But at

Introduction 36

the present stage of our deliberations it is appropriate to content our­selves with a question-formula which contains only a vague indication of what we are looking for. As yet we do not know with what categorial means we can adequately thematize something like the understanding of linguistic expressions. To explain how such questions about the understanding of expressions of a certain form, or about the under­standing of this form, are adequately to be formulated is, therefore, a task which already belongs to the elaboration of formal semantics.

We can call the question: what is it to understand an expression of a certain semantic form, or the form of this expression? the formalized question of meaning. Its relation to questions about the meaning or understanding of a particular expression of this form is analogous to the relation of the study of being as being to the reference to an indi­vidual being. Thus it seems that we can call this question the fundam en­tal semantic question which corresponds to the fundam ental ontological question. Moreover, with this question concerning the structure of our linguistic understanding we have an answer to the question concerning a narrower and more fundam ental reflection on our understanding, something we still lacked when attem pting, in the first lecture, to char­acterize a language-analytical conception of philosophy. In contrast with the broad sphere of the a priori, of the clarification of meaning, we have here a narrow er sphere of reflection in which the understanding- structures already presupposed and understood in all understanding of particular linguistic expressions are to be analysed. If we define the subject-matter of philosophy in terms of this formal-semantic sphere of reflection, this implies the thesis that all specifically philosophical con­cepts are concepts connected with the analysis of the semantic struc­tures.

But if we construe the enquiry into semantic form as reflection on the presuppositions of all linguistic understanding, then we cannot be con­tent to see this question break down into as many questions as there are semantic forms. In this respect then the question about semantic form, as hitherto formulated, does not yet correspond to the fundamental question of ontology; for to speak of the question about semantic form obscures the fact that one is really dealing with several questions. How are they connected?

A first step suggests itself immediately: the form of the expressions of a semantic class is determ ined by which expressions of other classes they can be combined with and by the way in which the meaning of the expressions contributes to the meaning of the composite expressions which thereby arise (p. 26). The semantic form o f a class of compo­

nent-expressions is thus always an abstract moment of the structure of the corresponding complex expressions. Now it belongs to the essence of a structural composition that it must terminate in units which are no longer essentially parts of wholes, but essentially wholes of parts. In the linguistic sphere these independent units are sentences - or, to put it better, we can define sentences as follows (though this is not unproble­matic): they are those linguistic expressions which can still enter as parts into larger linguistic units, but no longer in such a way that they are essentially components of a more comprehensive syntactic or semantic structure (which does not rule out that some components of sentences, e.g. personal pronouns may involve a reference to components of other sentences).1 T he sentence thus seems to be the primary unit of mean­ing. True, one can also understand parts of sentences, but one then understands them as parts of sentences; and one can only say some­thing (etwas zu verstehen geben) with a sentence, not with a word or other sentence-part (save where these function as elliptical sentences).

It follows from this that the question concerning the semantic form of a class of non-independent expressions (such as, e.g., singular terms) is always to be understood as a component-question of the question concerning the semantic form of the corresponding sentence (in this case of the predicative sentence). The questions concerning the seman­tic forms of the various classes of component-expressions thus enter as parts into the questions concerning the semantic forms of sentences.

But now, because there are different sentences, the problem again arises, at the present level, of whether there are not just many similar questions in formal semantics, but a unitary question. Does our under­standing of linguistic expressions break down into the understanding of different sentence-forms or are these sentence-forms internally con­nected? Is there such a thing as a general form of all sentences, of which all particular sentence-forms can in some way be regarded as specifica­tions? It would seem clear that this question cannot be decided in advance, but only in the actual elaboration of the formal-semantic anal­yses. But it seems equally clear that the analyses must be directed towards this question if one does not wish to have different forms of understanding merely juxtaposed to one another. Accordingly, formal semantics has at least a hypothetical fundamental question, viz. the question concerning the form of all sentences, or the connection of all sentence-forms, a question we can also formulate thus: ‘What is it to understand a sentence?’ It is this question which corresponds to the question about being as being in ontology.2

All we can do at the present stage of our deliberations by way of

Has formal semantics a fundam ental question? 37

Introduction 38

preparation for answering this fundamental question is to ask whether traditional ontology does not already contain a perspective on some­thing which is common to all sentences. You may feel that it is a mistake even to surmise the presence of such a perspective, given that sentences did not even belong to the thematic of ontology. However, we have seen that ontology is an ambivalent discipline, one of whose tendencies points towards a formal semantics (though this tendency is blocked by an object-orientated counter-tendency). We may then surmise that ontology already contains a perspective on the form of sentences, par­ticularly when one considers that Aristotle, in his ontology, was orien­tated towards the concept of logos - more precisely the logos apophanti- kos, the assertoric sentence. This orientation towards the assertoric sentence was connected with the thematization of the copulative ‘is’. Aristotle’s interpretation of all (singular) predicative sentences with a verbal predicate (e.g. ‘Peter swims’) as sentences with a copula and a participle (e.g. ‘Peter is swimming’)3 implies that the copula had for him the significance of the indicative verbal form as such, or the form of the composition of predicative sentences. However, this perspective on the sentence-form came to nothing, because the interest was immediately directed towards ‘beings’, the predicative contents. But most important of all, because of the orientation towards ‘beings’, the singular predica­tive sentence-form remained the only one considered in ontology. And this sentence itself, partly on account of the orientation towards ‘is’ and partly on account of the interpretation of the predicates as standing for determinations of objects, was interpreted one-sidedly as a compound of singular term and predicate. The possibility, which became so crucial for the modern logic of relations, that sentences like ‘Peter hits Paul’ can just as well be interpreted as the combination of a so-called two- place predicate (= classification-expression) (‘. . . hits . . .’) with an ordered pair of singular terms (‘Peter’, ‘Paul’), was not seen. The remaining forms of assertoric sentence were partly not perceived by Aristotle at all (such as those of complex sentences) and partly (as in the case of general sentences) regarded as ontologically irrelevant because they do not express statements about an object.

Thus one cannot derive a perspective for the question concerning something common to all forms of sentence from these beginnings of a theory of the predicative sentence-form; for they only deal with a single sentence-form. In certain places, however, Aristotle encountered aspects of form which, though presented by him as formal aspects of the predicative sentence, can immediately be seen to apply also to other sentences. One such place is his treatm ent of the Principle of Contra­

Has formal semantics a fundam ental question? 39

diction and the Law of Excluded Middle. In the context of our discus­sion there are two respects in which Aristotle’s mode of treatm ent of these ‘most general principles’ is worthy of attention.

Firstly, Aristotle finds it difficult to bring the discussion of these prin­ciples into a systematic connection with the thematic of ontology - the doctrine of substance. T he discussion of these principles does indeed not belong to a theory of objects. Aristotle, however, justifies his inten­tion to deal with them in ontology on the grounds that they underlie all sciences.4 Here then Aristotle himself encountered formal foundations of the sciences which cannot be understood objectually. Nowhere does the tension between theory of objects and formal semantics, which per­vades the Aristotelian ontology, become more evident than here. On the one hand, Aristotle was undogmatic enough to place treatm ent of these principles, on account of their formal-universal character, in the context of the question concerning being as being. On the other hand, the discussion of them, in Book 4, does not stand in any systematic connection with the rest of the problematic of the Metaphysics.

Secondly, it is worth noting that Aristotle presents both principles only with reference to the form of singular predicative statements. Thus the Principle of Contradiction takes the form: ‘the same attribute cannot at the same time and in the same respect belong and not belong to the same object’.5 In other places, however, Aristotle also formulates it thus: ‘it is impossible that (something) at the same time is and is not’.6 If the ‘is’ of this formulation is understood as the copula, then this for­mulation is identical with the previous one. However, this ‘is’ can also be interpreted as having the sense of that ‘is’ which can be placed before any assertoric sentence. Instead of saying, e.g., ‘It is raining’ one can say ‘It is the case that it is raining’, and instead of saying ‘It is not rain­ing’ one can say ‘It is not the case that it is raining.’ If we orientate ourselves towards this use of ‘is’, the form ulation of the Principle of Contradiction just mentioned would have the meaning: ‘It is impossible for something at the same time to be the case and not be the case.’ We can perhaps say, provisionally, that this prefixed ‘is’ expresses the affirmative form of an assertoric sentence and the ‘is not’ expresses the negative form. The following re-form ulation o f the last-mentioned fo r­mulation of the Principle of Contradiction would correspond to this explanation: ‘It is impossible at the same time to affirm and deny some­thing.’

Why is it impossible? To this we also find an answer in Aristotle. It is clearly not impossible to utter such a sentence; it is only impossible for one to say something thereby, to signify something (etwas zu verstehen

Introduction 40

geben).1 I can elucidate this as follows: if one affirms and denies some­thing, this is like making a move in chess and then withdrawing it, or giving something away and then taking it back.8 In the relevant action- respect one has not done anything. Aristotle points out that one can only justify the Principle of Contradiction indirectly, namely by show­ing that whoever denies it makes speech senseless. Since this justifica­tion no longer rests on the special form of the predicative statement, but on the more universal aspect of affirmation and negation, inter­preting the Principle of Contradiction in this more general way is sug­gested by Aristotle himself. But how general is it?

If the more general interpretation holds for all sentences whose affirmative or negative form is expressed by that prefixed ‘is’ or ‘is not’, then clearly it holds for all assertoric sentences. We find a criterion for distinguishing so-called assertoric sentences or statements from other sentences in Aristotle (though not in th e Metaphysics)', and this criterion has been accepted ever since. It is: whether one can call what is said with the sentence true or false.9 By means of this criterion assertoric sentences are distinguished from sentences in other so-called modes, such as imperative sentences, optative sentences, interrogative sen­tences. In contrast to this semantic criterion it might seem plausible to characterize assertoric sentences by reference to the syntactic criterion of the indicative verbal form. But the two criteria do not correspond. T here are various types, or modes of employment, of indicative sen­tences in which one does not say something which can be called true or false and which, therefore, cannot be called assertoric sentences, e.g. many sentences in the 1st Person Future Indicative - such as ‘1 will come’ - express not a prediction, which can turn out to be true or false, but an intention.

If it makes sense to speak of the general form of assertoric sentences, then we must also be able to symbolize these sentences in a general way. And if the Principle of Contradiction holds for all assertoric sentences, then it must be capable of being form ulated by means of this formali­zation. From now on I shall use, as is usual in logic, the letters ‘p \ ‘q\ Y as symbols for arbitrary assertoric sentences. In this way we arrive at the standard m odern formulation of the Principle of Contradiction: a statement of the form ‘p and not^?’ is impossible (necessarily false).

If we characterize what is said with a sentence (e.g ‘It is raining’) as true or false, we express this by means of the formulation ‘that/?’, e.g. ‘That it is raining is true’. But this ‘that’ already occurred in the for­mulation with the prefixed ‘is’ and one can now easily see that, in the case of any sentence which one can equivalently transform into ‘It is

Has form al semantics a fundamental question? 41

the case that/?’, one can equally well say ‘It is true that/?’. Consequently, the use of that prefixed ‘is’ extends just as far as, and no fu rther than, the use of assertoric sentences. If it is correct that the prefixed ‘is’ and ‘is not’ express the affirmative or negative form of assertoric sentences, then we can now say that the Principle of Contradiction is grounded in this general form of assertoric sentences. The real result for our ques­tion concerning a unitary sentence-form would of course be that here a general form, if not of all sentences then at least of all assertoric sen­tences, seems to emerge. But does not the distinction between affirma­tion and negation extend beyond the use of that prefixed ‘is’ and ‘is not’? Does it not apply to imperative sentences and optative sentences as well as to assertoric sentences? And would we not then have to expect a more general form of the Principle of Contradiction, which extends beyond the sphere of assertoric sentences? Of course more im portant for us than this question would be the question whether perhaps there emerges here a general form of all sentences.

Before I investigate this question (and I will not do so until the next lecture) we must consolidate what has so far been achieved. I wanted to pursue the question concerning a general sentence-form today only to the extent that clues to it can be derived from traditional ontology. If the reflections so far can only be supported by reference to Aristotle’s treatm ent of the Principle of Contradiction, then they have a rather slender basis in his ontology. Since ontology is orientated towards the word ‘is’, a clearer basis would be given if that prefixed ‘is’, on which I based the reflections, occurs in Aristotle himself. Now this is in fact the case. In his discussion of the various meanings of the word ‘is’ Aristotle distinguished this prefixed ‘is.’10 He put forward the thesis, which a m oment ago I merely hinted at, that with this ‘is’ it is said that some­thing is true and with the corresponding ‘is not’ that something is false. This thesis he justifies by referring to the equivalence of ‘It is the case that/?’ to ‘It is true th a t/?’. Aristotle calls this meaning of being einai hos to alethes.11 We can accordingly speak of veritative being.

T he situation with regard to Aristotle’s treatm ent of veritative being resembles that with regard to his treatm ent of the Principle of Contra­diction. Here too Aristotle considers only predicative sentences, although again it is clear that what he says can be applied to all asser­toric sentences. And here too Aristotle is unsure about the systematic place of this subject in ontology. At first he states that it does not belong to metaphysics at all,12 but later on there is a short discussion of this meaning of being.13 This discussion is directed so strongly towards the predicative sentence form that it yields little for our problematic. Only

Introduction 42

one thing is of interest: Aristotle arrives here at a differentiation within veritative being which relates to the so-called modalities of being (pos­sibility, actuality, necessity). In his detailed discussion of possibility and actuality in the Metaphysics, on the other hand, Aristotle treats them as modalities of copulative being; and this means: in an objectual manner. The word ‘possible’, e.g., is not understood in such a way that, like the veritative ‘is’, it is prefixed to the assertoric sentence (and we must add: to an arbitrary assertoric sentence) (‘It is possible that/?’), but in such a way that one must say a is possibly F \ thus in such a way that it is the objects that are possibly or actually such-and-such. In the ontological tradition possibility, actuality and necessity have been called ‘modalities of being’; but because of the orientation towards objects one could not see that the being of which they are modalities is veritative being.

But what, you will ask, is this veritative being? From Aristotle we get no further information on this question. However, we must not over­look the fact that in this connection too Aristotle speaks not only of ‘to be’, but also o f ‘being’. Is the veritative ‘is’ the ‘is’ of an object? This con­ception may not seem mistaken, inasmuch as the grammatical trans­formation of (p ’ into ‘that /?’ must be conceived as nominalization. Although the expression ‘that it is raining’ still seems to have the same content as the assertoric sentence ‘It is raining’, ‘that it is raining’ is not a sentence; it is a singular term. We can tell this from the fact that ‘that/?’ requires supplementation by a (higher-level) predicate in order to become once more a whole sentence, e.g. ‘T hat it is raining/is true’, ‘T hat it is raining/is pleasant.’ We must also include here such rela­tional supplementations as ‘he /hopes/that p ’, ‘he/believes/that/?’.

Now if ‘that/?’ is a singular term, then we will have to say that every such expression stands for an object. Clearly, from any sentence ‘that p/Fy we can infer ‘something/F’. Now what sort of objects are they for which the nominalized form of an assertoric sentence stands? Instead of saying ‘It is the case that/?’ we can also say ‘It is a fact that/?.’ This circumstance might suggest the view that the objects in question are facts. However, this would be a mistake. For in the negative case we say ‘It is not a fact that/?.’ Here we still have to do with the object in ques­tion, but deny that it is a fact. T hat it snowed yesterday is a fact only if it is true that yesterday it snowed. We obviously use the predicate ‘is a fact’ as equivalent in meaning to the predicate ‘is true’. We must, there­fore, ask: what are these objects that are designated by an expression of the form ‘that/?’ (and which are only facts if they are true) in and for themselves? T hat which can be true or false and which, if it is true, is a fact, is what is asserted when we utter an assertoric sentence. So it seems

Has formal semantics a fundamerital question ? 43

that we can characterize the objects in question as what is said or asserted. This also seems to correspond to our ordinary use of lan­guage. We ask, e.g., ‘Is what he asserted (said) true? Is it a fact?’ In English philosophy the term of art ‘proposition’ has been adopted for what I here designate as what is said (das Gesagte). Frege called these objects ‘thoughts’. As this word is used by Frege ‘thought’ is not to be understood in the sense of thinking, but ra ther in the sense of what is thought. Perhaps this designation ‘what is thought’ or ‘the thought’ is more attractive to you than my linguistic talk of ‘what is said’. But per­haps you will also be unhappy with Frege’s designation and say: ‘All these designations of the object in question, whether as “what is said” or “what is thought”, characterize it merely subjectively as “what is meant by us”; but we would like to know what it is in itself.’ Although I doubt whether this dem and is justified, I shall for the present accede to it. As a m atter of fact, there is also a terminology for the objects in question which is free of all subjective connotations: they are referred to by Husserl, and in W ittgenstein’s Tractatus, as ‘states of affairs’ (Sach­verhalte). Thus, according to this conception, every nominalized asser­toric sentence ‘that /?’ stands for a state of affairs. Wittgenstein then proceeded to define a fact as the ‘obtaining’ (Bestehen) of a state of affairs.14

With this we would have reached a conception which suggests that the veritative ‘is’ is to be construed as the being (Sein) of an object, as a mode of existence (Existenz), namely as the obtaining (Bestehen) of a state of affairs. O f course this obtaining would be a strange kind of existence. For one will have to say even of states of affairs which we do not obtain, hence states of affairs that are not facts, that, as states of affairs, they somehow exist; for otherwise we could not talk about them. One could try (and it has been tried) to dispose of this difficulty by attributing to states of affairs another sort of being which they have in themselves whether or not they also obtain. But the real difficulty sets in earlier, namely, in the question: what then is a state of affairs? One can see nicely from this example how the allegedly thing-orientated thinking and language-analytical thinking compare. T here are people who breathe a sigh of relief when offered the term ‘state of affairs’ for the objects for which expressions of the form ‘that/?’ stand. But the alleg­edly language-independent object is then something merely suggested by the word. It is the linguistic analyst who first establishes the relation to the thing (Sachbezug) by not being content with the word and asking what is m eant by it. What then is m eant by it? Just try to answer this question without referring to sentences and their meanings! I cannot

Introduction 44

yet deal with this question adequately. As yet we lack the categorial means for tackling a question concerning a kind of objects. Also I shall only later (Lecture 10) show in what way the attempts by Husserl and Wittgenstein to construe the object-character of states of affairs without reference to language must be regarded as having failed. In these pre­liminary reflections we are not yet trying to solve the problems, but simply to arrive at a correct way of posing the problems. At certain points then I will have to be satisfied if an idea seems sufficiently plau­sible for you to be prepared to follow me further.

In the case of the present4 question it must suffice if I say: the concep­tion of veritative being as the obtaining of states of affairs at any rate does not correspond to our ordinary ways of talking, and seems merely to arise from a traditional tendency to assimilate this being to the exis­tence of perceptible objects. One can, if one insists, speak of the obtain­ing of a state of affairs. But if one asks for the criterion for deciding whether a state of affairs obtains or not, one will have to say: the state of affairs that (e.g.) it snowed yesterday obtains if and only if it is true that it snowed yesterday. And it is true that it snowed yesterday if and only if it snowed yesterday. Thus talk of the obtaining of a state of affairs points back to the understanding of the unmodified, not yet nominalized, sentence.

We had the following series of equivalences: (1) the state of affairs that/? obtains = (2) that/? is a fact = (3) it is the case that/? = (4) that/? is true = (5) /?. Now if the meaning of (2) is not explicated by (1), but the meaning of (2), and hence that of (1), by (4) and (5), then clearly we also cannot explicate the meaning of (3), hence the meaning of verita­tive being, in terms of its equivalence with (2) and (1), but only in terms of its equivalence with (4) and (5).

Let us begin with (4) and (5). However unclear the meaning of ‘that /?’ still is to us, this much seems clear: that the nominalized sentence ‘that/?’ contains not more than the original sentence *p\ but less. What was removed from the sentence ‘/?’ in the transformation into the sin­gular term ‘that/?’ is what we can call its assertion-moment. If instead of saying ‘Yesterday it snowed’ one simply says ‘that it snowed yester­day’, one refers to what was asserted in the preceding utterance; but in such a way that one no longer asserts it. If one only says ‘that it snowed yesterday’ then, in contrast to when one says ‘It snowed yesterday’ one does not yet signify (zu verstehen geben) anything; rather, by removing the assertion-moment, one creates, so to speak, an empty-place for other supplementations, in which other position-takings vis-a-vis the same state of affairs, which was previously asserted, can be expressed,

Has formal semantics a fundamental question ? 45

e.g. ‘that p - is doubtful’, ‘that/? - I regard as improbable’, etc. Now there is one such predicative supplementation the significance of which appears to be that it gives back to the expression precisely what was taken away by its nominalization. This is the supplementation by the predicate ‘is true’. For if one supplements ‘that/?’ with ‘- is true’, one says again precisely the same as one said with ‘/?’. From this it seems to follow that the predicate ‘is true’ expresses the assertion-moment. In the original expression ‘/?’ there is no sign for expressing the assertion- moment. But because the equivalence ‘/? = that/? is true’ holds, we can now say that the meaning of ‘/?’ is composed of a propositional content and the assertion-moment, and that the propositional content corre­sponds to what is expressed in the nominalized form by ‘that/?’.

You will ask: how is this assertion-moment to be understood? I can­not yet answer this question here; the answer again belongs to the elab­oration rather than to the exposition of the problematic (cf. Lectures 15 and 28). Provisionally, I think I can assume that you have a vague understanding of what is meant by the assertion of something, as opposed to other position-takings, such as supposition, doubt, etc. of the same state of affairs. But one thing it would seem can be inferred from the reflections just carried out, and this is that the assertion- moment, because it can be expressed in the predicate ‘is true’, contains a truth-claim. Whoever utters an assertoric sentence asserts something (e.g. that it is raining) and always also thereby asserts that what he asserts (e.g. that it is raining) is true.

Now if the veritative ‘is’ is used as equivalent to the predicate ‘is true’, then it follows that this ‘is’ also expresses the assertion-moment of the statement, and that this assertion-moment is to be understood as a truth-claim. This result can now be directly linked with the explication of this ‘is’ I gave in the discussion of the Principle of Contradiction, viz. that it expresses the affirmative form of the statement. T he difficulty which arose there: that this description is too general, because it also applies to non-assertoric sentences, would now be removed. For we can now say: the affirmative form of the sentence, in the special case of the assertoric sentence, has the significance of an assertion; and this is a definition of the class of assertoric sentences by a characterization of their form. This definition, of course, is very closely connected with the criterion that we have previously found, viz. that a sentence is an asser­toric sentence if one can call what is said with it true or false. The p re­sent definition, on the other hand, says that a sentence is an assertoric sentence if it is used in such a way that a truth-claim is thereby made. We can of course easily bring out the connection between the two cri­

Introduction 46

teria by saying: only in so far as the use of a sentence already contains a truth-claim can what is said with it be called true or false.

But now the following difficulty suggests itself. I had previously dis­tinguished the affirmative sentence-form from the negative. But the truth-claim, or the assertion-character, is supposed to belong to all assertoric sentences and hence also to negative assertoric sentences. Does this not show that the subordination of the assertion-form to the affirmative form is mistaken? W here does the mistake lie? Have I, per­haps, in the introduction of the truth-claim, based myself one-sidedly on the affirmative assertoric sentences and lost sight of veritative not- being? But what would this mean? Should we say that negative sen­tences, or sentences of the form ‘It is not the case that/?’, make a falsity- claim? ‘It is not the case that/?’ is clearly equivalent to ‘T h a tp is false’; so one can indeed say that with a sentence of this form it is asserted that what is said with *p* is false. But ‘It is not the case that/?’ is, of course, equivalent, not to ‘p \ but to ‘not-/?’. With ‘It is not the case that/?’, there­fore, the falsity of the opposite statement is asserted; and this is only possible because here too the statement asserts its own truth. Thus the use of any assertoric sentence, whether or not a ‘not’ occurs in it, involves a truth-claim. And that in the use of any assertoric sentence something is asserted is clear anyway. Although one can assert the opposite, one cannot do something that is the opposite of asserting. We thus find ourselves forced to look for the e rro r on the o ther side. The assumption that there is an affirmative and a negative sentence-form, or that affirmation and negation are on a level, cannot be correct.

Firstly, it is not possible to divide sentences into affirmative and neg­ative. True, for every assertoric sentence there is an opposite sentence. But there is no general criterion by which we could tell which of the two is the negative one; for the criterion that that sentence is negative in which a negation-sign occurs is of only limited application. Frege gives as an example the sentence ‘Christ is immortal.’15 It is, of course, the negation of the sentence ‘Christ is m ortal’; but it appears not to be meaningful to say that it is intrinsically negative, or, as I incautiously said, that it has a ‘negative form ’. The predicate ‘is immortal’ is as ‘pos­itive’ as the predicate ‘is.m ortal’. It may seem negative because it is equivalent to ‘is not mortal’; but the predicate ‘is mortal’ can equally well seem negative because it is equivalent to ‘does not live forever’.16

Thus we cannot regard negation as a property which belongs to a sentence; rather we must construe it as an operation which, when applied to a sentence, generates the opposite sentence. But what is it that is negated, or to what is the operation of negation applied? As both

Has formal semantics a fundam ental question? 47

assertoric sentences - e.g. ‘It is raining’ and ‘It is 'no t raining’ - are assertoric, one cannot construe the second sentence as though what is negated is the assertion of the first; rather what is negated is what the first asserts, its propositional content. If we symbolize the assertion- m om ent of ‘p ’ with Frege’s assertion-sign V , and the non-nominalized propositional content of with */?*, then our two sentences have the form V*/?*’ and V *not-/?*’ rather than the form V */?*’ and ‘not h *p*\

T hat the negation-sign pertains only to the propositional content can also be reinforced, following Frege, by considering the role of negation in component-sentences of complex sentences.17 When sentences enter as component-sentences into a complex sentence then, even if they are not nominalized, they lose their assertion-moment; only the whole sen­tence is asserted. Let us take e.g. the sentence-form ‘p or q\ It is obvious that here only the complex sentence as a whole is asserted, and that it would contradict the sense of such an ‘o r’-sentence if the two compo­nent-sentences were also asserted. Thus the two component-sentences lack the assertion-moment when they occur in this form of sentence, though there is no symbol for this lack in natural language. But this is only possible if (since the assertion-moment is absent) the ‘not’ belongs to the propositional content. This argum ent can only be evaded by the outlandish thesis that the word ‘no t’ has a different m eaning when it occurs in an independent sentence and when the same sentence func­tions as a component-sentence.

But is it not then misleading, you will ask, to continue to refer to assertion as affirmation? The notion of an affirmation seems essentially related to a denial. However, one can acknowledge such a correlation without contradicting the results just achieved. We simply have to remove an ambiguity in speaking of denial and negation. For every propositional content *p* there is always one opposed to it (*not-/?*), which we get by negating the first. To avoid ambiguities we can reserve the word ‘negation’ for this. If one asserts (assertorically affirms) this second propositional content (*not-/?*) one can call this the denial of the affirmation of the first propositional content (*/?*) and, again to avoid ambiguity, one can reserve the word ‘denial’ for this. Denial, as thus understood, is, therefore, the assertoric affirmation of a negated propositional content; and this means: of a propositional content which is negative relative to another. T hus the denial, which is itself an affir­mation, is opposed to another affirmation.

It is in this way that we must also understand the words ‘yes’ and ‘no’ (for the time being I shall disregard their non-assertoric use). We use these words in conversation when the propositional content which is

Introduction 48

affirmed or denied has already been expressed by a preceding assertion or question, so that the reply does not need to repeat the propositional content and can restrict itself to the mere affirmation or denial. Again it is easily overlooked that with the ‘no’ too something is affirmed, viz. the opposite propositional content. If we use the symbol ’ to indicate that the propositional content which is at issue is the one that had been mentioned in the conversation, then the assertoric ‘yes’ has the form

’ and the assertoric ‘no’ the form V notNow if every denial is an affirmation which is opposed to another

affirmation, then it also follows that every affirmation is opposed to another affirmation - namely, its denial, the affirmation of the opposite propositional content. Every ‘no’ expresses a position-taking against a ‘yes’. Does every ‘yes’ also express a position-taking against a ‘no’? If it is correct that one cannot divide sentences into intrinsically positive and intrinsically negative then this is clearly how we must view it. Thus, understanding the assertion-moment as affirmation turns out to be cor­rect. Every assertion is a position-taking against the one opposed to it; only this position-taking is not made explicit as it is in denial. Later I shall try to show that without consideration of this property - being a position-taking against - one cannot understand the mode of employ­ment of assertoric sentences and their various forms (Lectures 15, 17, 27).

Thus from the fact that what is negated is always only the proposi­tional content and not the assertion, it in no way follows that the asser- tion-moment is untouched by the negation. On the contrary. That the utterance of an assertoric sentence has the character of an assertion, and hence of an affirmative position-taking, would be meaningless if the propositional content were not opposed to one that negated it. Now where do we stand with respect to our question concerning a unitary form of all sentences? We have arrived at a form which is common to at least all assertoric sentences: (1) they have the structure ‘t- */?*’ and con­nected with this structure is the fact that (2) for every *p* there is a *not-/?*. Whereas the first part of this symbolization represents some­thing that remains the same for every assertoric sentence, behind the second part, the symbol */?*, is concealed the entire multiplicity of struc­tures of the propositional content. A unitary understanding of the form of assertoric sentences would only be achieved if a connection of the various structures of the propositional content were also to become vis­ible. But so far it is not even clear how one is to enquire into the p rop­ositional structures. I do not intend to pursue this aspect of the problem

Has form al semantics a fundam ental question? 49

any further in the introductory reflections; it will be our central concern in the main part of the lectures.

But I would also not wish to give the impression that with the refer­ence to the assertion-moment and the negateability of propositional content this structure that is common to all forms of assertoric sen­tences is already explained. The aim of this lecture was not to find an answer to the question concerning the form of assertoric sentences (and perhaps of all sentences), but simply to investigate whether there is a unitary form at all which would be amenable to such a question. At present we only know that the question: what is it to understand an assertoric sentence? aims at three structural moments and their inner connection: (1) what is it to understand an assertoric affirmation? (2) what is it to understand a propositional content? (3) what is it to under­stand the word ‘not’?

L E C T U R E 5

Consciousness and speech

If Aristotle or the tradition which followed him had taken veritative being as the guiding thread of their investigation, then there would have developed, within the framework of ontology, a semantics of the assertoric sentence-form. Instead of this, however, the problematic, which Aristotle had at least touched upon, became unrecognizable in the shape of the inadequate doctrine of theverum as another ‘transcen­dental’ determ ination of ens, together with unum and aliquid, a doctrine in which the veritative meaning of ‘is’ was assimilated to the others and thereby finally objectified. Moreover for mediaeval ontology the start­ing-point for the demonstration of the universality of being was no longer the usage of ‘is’ but the thesis that the determ ination ens is the first determ ination that is given to the m ind.1 How this proposition, which, to the impartial observer, must appear far from evident, indeed unintelligible, could be regarded as supremely evident by an entire tra­dition, can be explained only by reference to the concept of represen­tation (Vorstellen) which I shall examine at the end of this lecture.

W here the Aristotelian ontology, and indeed the entire traditional ontology from Parmenides to Hegel, came closest to veritative being was in the assumption that the question of being is always connected with the question of not-being. How far from obvious this is can be seen immediately one considers that a theory which started out from objects or from that mediaeval conception of ens would have no occasion to thematize the ‘not’. T he opposition of being and not-being is (as we could already see in connection with the Principle of Contradiction) an opposition that belongs to veritative being just as much as the so-called modalities of being; and other meanings of ‘is’ (such as that of the cop­ula or existence) only participate in this opposition because they are species of veritative being.2 On the other hand, our traditional orien­tation towards the opposition being - not-being contributes to our ten­

Consciousness and speech 51

dency to place denial on a level with affirmation and to overlook the peculiar status3 of the ‘not’ m the sentence-form.

What we were able to achieve in the previous lecture in connection with veritative being was a formal characterization of all assertoric sen­tences. This provided us with a fundam ental question for the semantics of assertoric sentence-forms. If in the question concerning a formal universal science we orientate ourselves towards the Aristotelian starting- point, namely towards the question concerning the form al presuppo­sitions o f all sciences, then with the formal semantics of assertoric sen­tences we have already achieved our aim, for in the sciences only assertoric sentences occur (at most-one would have to add in terroga­tive sentences).

On the other hand, once one has entered the sphere of formal semantics it would be artificial to restrict the thematic field to asserto­ric sentences, since such a restriction would be incompatible with this formal science’s claim to universality. It thus seems plausible no longer to sketch the formal thematic of the sought-for universal science by reference to the sciences but by taking as one’s point of departure, as I did in the previous lecture (p. 36), our understanding of linguistic expressions in general. But then we are faced with the question: how can we extend the unified perspective we have achieved for the inves­tigation of assertoric sentences in such a way that it can be understood as the unified perspective for the investigation of all forms of sen­tences?

Assertoric sentences are contrasted with imperatives, optative sen­tences and interrogative sentences. I leave undecided w hether this list of other sentence-forms is complete and systematically significant. And I shall not now give a semantic criterion for these classifications.4 It is usual to rely on grammatical criteria such as mode of verb, word-posi- tion and intonation. This is an indication of how little attention has hitherto been given in philosophical semantics to non-assertoric sen­tences. O ne can see in this an effect of the traditional ontology. Whereas Aristotle had with veritative being just entered the dimension of assertoric semantics, non-assertoric semantics seem to lie completely outside the scope of an ontology - unless one understands the orien­tation towards ‘is’ as an orientation not ju s t towards the indicative form of this verb, but towards all its modes. Such an orientation has never been developed in the traditional elaboration of the question of being; but it would be conceivable.5 Just as we can transform an asser­toric sentence into a prefixed ‘it is the case’ followed by the nominali- zation of the sentence, so we can clearly also transform an imperative

Introduction 52

sentence (e.g. ‘Let him come!’) into a prefixed ‘let it be the case’ fol­lowed by the nominalized form ‘that he comes!’, or, e.g., the interro­gative sentence ‘Is he coming?’ into ‘Is it the case that he is coming?’

This gives us a fundam ental insight into the structure of non-asser­toric sentences: just like assertoric sentences they can be divided into an affirm ation-m om ent and a propositional content; and in their case too the propositional content without the affirmation-moment can be expressed by means of the nominalized locution ‘that p ’. But not only do non-assertoric sentences also have a propositional content; one can also see immediately that one and the same propositional content can occur in an assertoric sentence and in the various non-assertoric modes o f sentence; e.g. ‘He is coming’, ‘Let him come!’ ‘If only he would come’, ‘Is he coming?’ all clearly have the same propositional content and only differ with respect to mode. There is thus confirmed what initially did not seem to follow necessarily, namely, that we must regard the semantic form of non-nominalized assertoric sentences as also composed of an affirmation-moment and a propositional content; although the propositional content is not grammatically isolable. Only this structure makes intelligible the connection between ‘He is coming’, ‘Let him come!’ etc. It thus emerges that the sentence-forms of the various sentence-modes only differ with respect to their mode, whereas the other element, the propositional content (thus that ele­ment which is the bearer of all further formal-semantic sub-struc­tures), is common to all sentence-forms (with of course certain quali­fications, such as, that imperatives can only refer to something in the fu ture).6

In this way then one arrives at a unitary structure of all sentences, which one can symbolize by ‘M */?*’, where ‘M ’ is a variable which is to be replaced by the symbols for the various modes e.g. by V for the assertion-moment, which gives us the already familiar Similarlyone can e.g. write ‘!*p*’ for the imperative and for the inter­rogative sentence.

Bringing in non-assertoric sentences also provides additional confir­mation for the view that the ‘no t’ belongs to the propositional content. The imperative ‘Do not come!’ demands the realization of the same state of affairs whose truth is asserted by the assertoric sentence ‘You will not come.’ T he imperative denial also has the form ‘!*not -/>*’ and is thus the imperative affirmation of a negated propositional content.

Only in passing shall I mention the problem of so-called ‘external’ negation, i.e. negation in such sentences as ‘I do not order that p .’1 Such so-called ‘perform ative’ sentence-forms, though they have the

Consciousness and speech 53

grammatical form of assertoric sentences, are not assertoric sentences. For in uttering them one is not only asserting something; one is at the same time doing what they assert (e.g. when I say ‘I order . . .’ I am not (only) asserting that I am ordering; I am ordering).8 Now if one could assimilate ‘I assert that/?’ and ‘I o rder that/?’ to ‘p y and ‘!*/?*’ then ‘I do not assert that /?’ and ‘I do not o rder that /?’ would clearly have the form ‘not h */?*’and ‘not ! */?*’. One may doubt whether this assimilation is correct and whether the semantics of performatives is not ultimately to be understood in terms of the semantics of assertoric sentences, so that one would again arrive at a unitary conception of negation. In any case external negation does not occur in the case of the non-nominalized sentences ‘/?’, *!*/?*’, etc. However, by reference to performatives one can see what meaning a sentence of the form ‘not-M*/?*’ would have and in this way convince oneself again that negative assertoric sentences, imperatives, etc., do not have this form.

Must we also regard the non-assertoric modes as a form of affir­mation? Do they too have the character of a position-taking against an opposing affirmation? For this view speaks the fact that the Principle of Contradiction clearly also holds for non-assertoric sentences and that one can justify it here in exactly the same way as in the case of assertoric sentences. If one says ‘Come here and do not come here’ one has said nothing. With the second step one has cancelled the first.

And of course there is also an imperative use of ‘yes’ and ‘no’. How­ever, even in the case of the explicitly negative sentence (‘Do not come!’) it seems less clear that it is to be construed as position-taking- against; for such an utterance seems directed not against another utterance but against an action. Only if it could be shown that the negative imperative is really directed against an opposite imperative by which the action is being directed would it be clear that it is to be construed as the denial of an affirmation. And then of course it would be plausible also to construe the imperative which does not contain a ‘not’, precisely as in the case of the assertoric sentence, as affirmation; and that means as position-taking against the imperatival affirmation of the opposite propositional content. The question of whether the various non-assertoric modes can be construed as modes of affirma­tion can, therefore, just as in the case of the assertoric mode, in the end only be decided by actually carrying out the semantics of these forms of sentence.

Only in the case of interrogative sentences can one see in advance that they do not conform to this schema. Although they too have a negatable propositional content, it seems not to be meaningful to

Introduction 54

describe the two interrogative sentences which have an opposed prop­ositional content as being themselves opposed. It would be more cor­rect to say that in both questions ‘Is he coming?’, ‘Is he not coming?’ one and the same thing is being asked. But it is also easy to see why interrogative sentences occupy this special position. A question is a dem and that a sentence be uttered, within a range specified by the interrogative sentence,9 normally an assertoric sentence or, if there are specifically practical questions (cf. Lecture 7), an imperative. Thus the only reason why interrogative sentences are not affirmative is that they contain the dem and for an affirmation, a position-taking. Con­sequently these non-affirmative sentences do not constitute a counter­example, to the result that is now emerging, viz. that yes/no position- taking is fundam ental to the use of all sentences with a propositional content.

Do all our sentences have this form? Clearly there are alsoindependent units of communication which have no propositional content, e.g. ‘ow’, ‘h u rrah ’, ‘thanks’, ‘hello’. But such expressions do not exhibit any semantic structures (or only very rudim entary ones). We shall see later that the analysis of the meaning of such situation- related expressions presents far fewer difficulties than the analysis of propositional sentences. One can therefore say that the question con­cerning the understanding of our linguistic expressions, though not completely reducible, is in essence reducible to the question of the understanding of sign-formations of the form 'M *p*\ And so the fun­damental question of formal semantics as a whole can be directly joined to the fundam ental question o f the semantics of assertoric sen­tences presented at the end of the previous lecture; only in place of the question about the understanding of V’ one has the question about the understanding of the various modes and their interconnection. The fundam ental question is: how is it that our entire linguistic understanding has the structure of yes/no-position-takings of various modes vis-a-vis propositional contents? Of course, whether one can speak of a unitary structure and not simply of common structural moments is still an open question. But at least we now know in what answering it positively would consist: firstly, in the dem onstration o f a systematic connection o f the structures o f the propositional content (p. 48) and, secondly, in the dem onstration o f a systematic connection between the different modes.

With this I can conclude this preliminary sketch of a formal seman­tics as the language-analytical successor-discipline to ontology in its capacity as formal universal science. We do not yet know anything

Consciousness and speech 55

about the conceptuality and methods with which such an enquiry can be tackled; in the main part of these lectures I shall likewise attem pt to pursue this enquiry by way of a step-by-step destruction of the con­ceptuality available from the tradition. At present I am only trying to mark out a thematic field that can be claimed as the field of a science which is somehow pre-em inent and for this reason merits the title ‘philosophy’.

In o rder to restore the connection with this our principal question and to prepare the next step I can now summarize the criticism that can be brought against the Aristotelian conception of a pre-em inent science as ontology in the following way. Firstly, one can call this con­ception in question absolutely, i.e. with respect to motivation (p. 18). It will be in the context of the guiding idea of ‘reason’ that I shall first attem pt to introduce a conception of philosophy absolutely, that is to say by reference to a pre-em inent motivation and not on the basis of a preliminary understanding of the word ‘philosophy’ (Lecture 7). Secondly, one can criticize the conception of philosophy as ontology relative to the prelim inary understanding which Aristotle takes as his starting-point. One could then question the idea of formalization. If this is to be m easured against Aristotle’s own preliminary conception, then the question arises: with what right is one of the two aspects of the preliminary conception, viz. the radicalization of the justificatory character of science, left out? I shall likewise be taking up this aspect under the heading ‘reason’. Thirdly, one can also criticize the ontolog­ical conception from the point of view of the other aspect of Aristotle’s preliminary conception, viz. that of universality.

It was this third aspect which led us to envisage the broader disci­pline of a formal semantics in place of ontology. So far then the only thing to recom m end the idea of a formal semantics as a pre-em inent science is essentially the fact that it is m ore comprehensive than ontol­ogy. If we continue to orientate ourselves primarily towards this aspect of universality one could ask: if it has now been shown that ontology is surpassed in its universality-claim by formal semantics, what guar­antee have we that formal semantics will not in tu rn be surpassed in its universality-claim by another discipline?

One could argue that the sense of formalization which em erged from the analysis of the concept of an object is only significant with reference to sentences and that, consequently, the idea of an extension of the formal discipline beyond the sphere of sentences does not make sense. I adm it that I am inclined to this view. However, we m ust be on our guard against dogmatism here. In the linguistic interpretation of

Introduction 56

Aristotle’s formal reflection-step I stressed that I must initially leave open the question of whether this is the only possible way of under­standing this step. Moreover, it seems undeniable that linguistic understanding is not reducible to the understanding of isolated sen­tences. Sentences are the smallest units of communication; but these belong - both in science and elsewhere - to larger contexts of com­munication and understanding. These contexts remain unexamined if we confine ourselves to the form of sentences. A thematic orientated towards sentences, just like one orientated towards objects, favours an atomistic approach. It is for this reason that there has recently arisen, under the title ‘pragmatics’, a discipline which seeks to transcend this limitation.10 Also, the question arises of whether we should leave out of account all modes of consciousness and experience which are not expressed in sentences. This question opens up a new perspective. It subsumes the understanding of linguistic expressions under the con­cept of consciousness and thus holds out the prospect of a broadening of the field of formalization, although off-hand it is not clear whether a formalization of non-linguistic experience makes sense, or what should take the place of formalization.

In consciousness we encounter the second traditional guiding-idea with which I wanted to confront the language-analytical conception of philosophy (p. 13). The orientation towards consciousness character­istic of classical m odern philosophy has - like the language-analytical conception - been understood as a critical extension of ontology. The relationship to ontology of philosophy of consciousness and linguistic analysis is in a certain way analogous. In both cases, and in contrast to the theory of objects, the new philosophical approach arises out of reflection. In the philosophy of consciousness this is represented as. reflection on the experience or consciousness of objects; in semantics as reflection on the sentences in which objects are spoken about. Phi­losophy of consciousness and language-analytical philosophy thus appear as competing undertakings, each of which can claim to out­reach the other. All consciousness of an object is always a component of the understanding of a sentence. On the other hand, the philoso­phy of consciousness can argue that all sentence-understanding is merely one mode of consciousness among others.

An introduction to language-analytical philosophy will therefore have to include a confrontation with the philosophy of consciousness. As in the debate with ontology the concern is just as much with addi­tional insights into the nature of language-analytical philosophy as with the justification of the language-analytical conception of philoso­

phy. T he link between the debate with ontology and the debate with the philosophy of consciousness is this: reflection on consciousness opens up a perspective which appears to outbid the universality-claim of the language-analytical conception.

In order to be able to carry through this debate with the philosophy of consciousness we must first get an idea - obviously a very simplified and schematic one - of the positions of the philosophy of conscious­ness relevant to our problematic. This we shall do by examining how ontology was made into a problem, and its point of departure extended, by reflection on consciousness. I shall present this develop­ment of the philosophy of consciousness, in contrast to ontology, in the form of three successive and increasingly radical steps. Having done this we shall then have to ask what the consequences are of each of these three steps for the language-analytical conception, or what the consequences are of a language-analytical conception for these three steps.

The first step is that which we can call the Cartesian step. Historically it was with this step that the turn from ontology to consciousness began. It resulted from a retu rn to that aspect of the Aristotelian preliminary conception of philosophy which Aristotle himself had neglected in his ontological interpretation of this preliminary conception, viz. the aspect of grounding and justification. T he question of justification in the sci­ences concerns their assertions in so far as they make a claim to knowl­edge. But knowledge is ultimately always knowledge of an individual. It always concerns something which someone believes. We say that he not only believes it, but knows it, if he can justify it. We also say then that he is certain of what he believes, that it is indubitable for him. Justification consists in an explicit elimination of possible doubts. In the question of doubt and certainty everyone finds himself thrown back upon him self: one can have one’s attention drawn to possible doubts by others, but the doubting itself and the corresponding certainty is a state of the individual. Now Descartes pointed out that this state of doubt or certainty itself cannot be doubted by the person whose state it is at the time he is in that state. It then turned out that there is a whole class of states - believing, wishing, intending, feeling, etc. - which are indubit­able for the person who is in them. If one asks by what this class is defined perhaps the only criterion is precisely this: that the person whose states they are has, at the time in which he is in them, an indu­bitable knowledge that he is in them. This criterion gives us a first broad concept of consciousness (we shall later become acquainted with a sec­ond, more narrow concept). It now seemed plausible to interpret this

Consciousness and speech 57

Introduction 58

sphere of consciousness as something inner which is somehow imme­diately given to the individual, thus to ‘me’ (to the ‘ego’, as was now said). And further it seemed plausible to assume that I know everything outer and dubitable somehow by means of what is inwardly given and indubitable. The radicalization of the question of justification thus leads to (1) the conception of the inner as something indubitable and (2) the question of how something outer is given to me, how it can be known by me. This question was called the epistemological question.

This first step in the philosophy of consciousness - the Cartesian, epistemological one - does not yet imply any extension of ontology. It signifies vis-a-vis ontology only a new centre of gravity of enquiry: before the question concerning being as such there is placed the ques­tion concerning its accessibility. The ontological structures can remain untouched.

However, the question of accessibility can affect the ontological ques­tions themselves. W here this happens there results the second step, the so-called transcendental-philosophical turn. This step consists in this: the question of the mode of givenness of objects is no longer regarded merely as a question about certainty but as constitutive for the object­hood of objects.

We can clarify this thesis of transcendental philosophy most easily by reference to Husserl’s conception of regional ontologies, a conception I have already touched on in the introduction of the ontological posi­tion. At the time I pointed out that the fundam ental concepts which characterize a sphere of objects as such - concepts such as ‘material thing’, ‘state of consciousness’, ‘num ber’ - are not merely of gradually higher generality than the concepts which belong to the object-sphere; they are fundamentally different from them. If there are fundam en­tally distinct object-spheres which are not simply sub-divisions of one universal realm of being, this must be because there are not only objects of different kinds in the distinct object-spheres, but objects whose mode of objecthood is different. Now for a philosophy which does not yet reflect on language there is only one possible interpretation of this state of affairs. The distinctions in question are not distinctions of the objects and their contents; they can therefore only concern the mode of giv­enness of objects. For Husserl the form of objecthood is constituted in the form of mode of givenness of these objects (Ideen hi §7). Corre­spondingly one can also, according to Husserl, only clarify the sense of objecthood as such in a transcendental study of givenness.

We find a basically similar position in Kant, only with the difference that Kant does not thematize objecthood as such. Nor does he distin­

Consciousness and speech 59

guish different spheres of object; he is concerned only with the object­hood of the objects of experience in space and time. As with Husserl the possibility of experience is, for Kant, constitutive of the objecthood of objects. T hat is the sense of the celebrated statement: ‘The conditions of the possibility of experience as such are at the same time conditions of the possibility of the objects of experience.’11

With this second step, then, the reflection on consciousness has an effect on ontology itself. Ontological analysis is now conceived as the analysis of the possibility of experience or of what makes it possible that objects as such and the objects of the various regions can be given. But as I have presented the transcendental-philosophical position so far it still remains dependent on ontology in that it allows the latter to furnish it with its fundam ental concept, that of an object. The transcendental turn merely brings it about that what was already the theme, object­hood, is now analysed in a new way; but the thematic field is not thereby extended.

This extension of the thematic field by reflection on consciousness is the third step. The latter consists in this, that by reflecting on conscious­ness the orientation towards objects is transcended, for it emerges that there are modes of consciousness which cannot be understood as con­sciousness of an object. Historically this step is only hinted at in the modern philosophies of consciousness, in particular in Kant.

Reflection on the experience of objects made Kant aware that objects are given to us in space and time. Space and time, however, are not themselves objects. But how we should positively in terpret the con­sciousness of space and time, which is not itself an objectual conscious­ness, remains unclear. On the basis of formal analogies (Critique of Pure Reason, B39, B47) Kant calls it an ‘intuition’ (Anschauung) but that is clearly a makeshift solution. Intimately connected with the conscious­ness of space and time (cf. B 39 f.) is the consciousness of the world, the (definite or indefinite) totality of experienceable things (B 446 f.). This totality of objects is clearly not itself an object.

Unlike Husserl, Kant arrived at the necessity of a transcendental enquiry precisely through the recognition that all experience of objects is essentially always spatial and temporal and contains a world-relation (Weltbezug). The subjective turn was made necessary by something which does not become accessible in an objectual approach. On the other hand, Kant only thematized that non-objectual consciousness which belongs to the context of objectual experience.

Kant recognized that concepts like totality and infinity can only be understood on the basis of the concept of a repeated (‘successive’) act. 12

Introduction 60

Moreover, the concept of a synthetic act - the synthesis of the manifold according to rules - was, for Kant, fundam ental to the understanding of consciousness as such, or at any rate of that consciousness which Kant called experience: the knowledge of objects (B103, 195 f.). Now the consciousness that someone has o f his action (and that means: of the rule which he follows in his action) is again not the consciousness of an object. So not only did Kant also take account of modes of consciousness that are non-objectual; he thought that a certain non-objectual con­sciousness - an act-consciousness - was essential to the consciousness of objects.13 This, o f course, shows that Kant nonetheless remained orien­tated towards the consciousness of objects: for this reason he consid­ered only that act-consciousness which, in his view, is essential to the consciousness of objects and their connection in space and time. He did not develop a general theory of the consciousness of acts; hence the question of what it is to be conscious of one’s action and what it is to be conscious of an action-rule rem ained unclarified.

It is in H eidegger’s Sein und Zeit that we first find an attem pt explicitly to carry through the third step and free the understanding of con­sciousness from the orientation towards objects. In doing so Heidegger abandoned the term ‘consciousness’. This was because this term had, on account of its descent from ontology, been so tied to the concept of an object that it seemed that consciousness means eo ipso consciousness of objects.

For the term ‘consciousness’ H eidegger substituted a term of art ‘dis- closedness’ (Erschlossenheit),14 In particular Heidegger tried to show that the disclosedness which the person has of himself, of his own being (thus speaking traditionally: his self-consciousness) is not to be under­stood objectually. This problematic was, for him, connected with the non-objectual ‘disclosedness’ o f ‘world’, where ‘world’ stands not for the totality of objects but for the totality of significance in which a person understands himself (Sein und Zeit §§ 18, 32).

After this crude survey of the relationship of the philosophy of con­sciousness to ontology we must ask: what implications does this have for the language-analytical conception?

For the present I shall pass over the problem which we encountered in the first, Cartesian, step, for this problem had to do not with an extension of ontology, but with a completely different conception of philosophy which results from a radicalization of the aspect of justifi­cation. We shall return to this problem later under the heading ‘reason’.

T he second, transcendental step is an expression of the insight that one can only thematize that which constitutes an object as object and

Consciousness and speech 61

the objecthood of the various spheres of object by at the same time reflecting on our relationship to objects. In the discussion of the Aris­totelian formalization-step I pointed out that one can only thematize the objecthood of objects by reflecting on our relationship to them (p. 24). We then saw how one can achieve and thematize the concept of an object if one understands our relationship to objects in such a way that it rests on the use of certain linguistic expressions. At the time I left open the question of whether one can also conceive of this relation­ship, and the reflection on it, in non-linguistic terms. It is this concep­tion which we now find ourselves confronted with in the transcenden­tal-philosophical position.

In so far as our present guiding question is the universality of the mode of enquiry this much at least is clear: so long as it restricts itself to objects the transcendental mode of enquiry is not more comprehensive than the language-analytical, for the relationship to objects which tran­scendental philosophy of course regards as primarily non-linguistic, is, even from its perspective, not linguistically inaccessible. On the other hand, however, the language-analytical reflection on our relationship to objects places this in the more comprehensive context of predication and veritative being. What is to be understood by the ‘givenness’ of objects in this context is something we shall have to consider later. But it will perhaps already make sense if I say, by way of anticipation, that from the point of view of linguistic analysis the problem of the ‘acces­sibility’ of objects becomes a part of the problem of the verifiability of the predicative statements which can be made about objects. T he prob­lem of regional ontologies thereby acquires a new and more compre­hensive sense than it has in Husserl. If the difference between different spheres of object concerns the objecthood of objects, this means that it concerns the form of the corresponding predicative statements. That two objects differ not only with respect to certain predicates but with respect to their objecthood means, as Husserl says, that they are acces­sible in essentially different ways. But that would now also have to mean: that the applying of predicates to them is verified in fundamentally different ways. And that means: the difference in mode of givenness is grounded in the difference of veritative being.

But the transcendental-philosophical reflection on our relationship to objects is not just more narrowly conceived than the language-analyti­cal. The question arises: is not the idea of a pre-linguistic relationship to objects which one can reflect on an illusion? I can today merely indi­cate this language-analytical criticism of the transcendental-philosoph- ical approach in order to provide at least a preliminary orientation;

Introduction 62

only later (Lectures 21 and 27) will we be in a position to elaborate this criticism. T he idea of a pre-linguistic relationship to objects implies that one thinks of this relationship as a having before oneself (ein Vorsichha- ben). T he fundam ental modern concept for this having before oneself is the concept of ‘representation’ (Vorstellung). Consciousness has a rela­tionship to objects by ‘representing’ them.

This term is the principal point of attack for a language-analytical critique of the transcendental-philosophical position. It is not the fact that transcendental philosophy orientates itself towards objects that lin­guistic philosophy objects to. Rather it is that transcendental philosophy conceives of the consciousness of objects in too simple a fashion; it fails to take account of the fact that we refer linguistically to objects by means o f expressions which - as singular terms - belong to a certain logical (formal-semantic) sentence-structure. How then is one to conceive of a pre-linguistic relationship to objects? As representation. What is meant by this?

Here I must become somewhat more precise. In ordinary linguistic usage the expression ‘sich etwas vorstellen’ (lit. to represent something to oneself) is used in two different senses. In one sense ‘er stellt sich vor . . .’ is completed by a noun clause, e.g. ‘Er stellt sich vor, dass es jetzt in Berlin regnen könnte’ (‘He imagines that it could now be raining in Berlin’). This is representing something to oneself in the sense of imag­ining that something is so and so. In the second sense ‘er stellt sich vor . . .’ is completed by an expression which stands for an object, e.g. ‘Er stellt sich den Kölner Dom vor’ (‘He pictures Cologne Cathedral to him ­self’), ‘Ich kann mir meinen Grossvater nicht m ehr vorstellen’ (‘I can no longer picture my grandfather’). Used in this way ‘sich etwas vor­stellen’ means something like ‘make something present to oneself’ (‘sich etwas vergegenwärtigen’) in the sense o f : form an inner picture, a phantasy-image of, bring it before one sensuously (anschaulich). It is this second mode of employment from which the philosophical terminology derives. It came to be used in early m odern philosophy in the context of an epistemological ‘theory of representation’ (Repräsentation): inner representations (Vorstellungen) as representatives (Vertreter) [repraesen- tationes] of outer objects. However, transcendental philosophy still retained the term ‘representation’ (Vorstellung) even when it rejected this theory of representation (Repräsentation) and made it clear that con­sciousness relates to objects directly and not via inner representatives. This is true, as indeed Husserl himself showed,15 even of phantasy- representations (when I imagine Cologne Cathedral I mean it itself directly; I do not see a picture which ‘stands fo r’ it). But it is also true of

Consciousness and speech 63

all other forms of being conscious of an object. Hence the representa­tion (Repräsentation) aspect in the concept of representing (Vorstellen) drops out. And as this concept is also not restricted to making present sensuously (anschauliches Vergegenwärtigen) it becomes so refined that it stands for any, even non-sensuous, conscious having before oneself. And thus ‘representation’ became the general concept for our conscious relationship to objects. But in reality this concept is vacuous. With it som ething which belongs to a sensuous relationship (Anschauungsbezie­hung) is transferred to a relationship that is logical. T he consciousness of an object is like the sensuous having before oneself of a picture, only this having before oneself is not sensuous. From the beginnings of Greek philosophy up to Husserl philosophers, through the neglect o f language-analytical reflection, have operated with a sensuous and even optical model. T he philosopher sits at his desk and thinks about the world and nothing is more natural than to look at the objects he has before him: things on the desk and, outside the window, trees and houses. O f all these things he has a sensuous image. And it is just so, he thinks, only not sensuously, when one refers to objects. But what does ‘ju st so - only not sensuously’ mean?

We can now also understand the mediaeval idea that being is the prim ary object of the intellect (ens primum objectum intellectus nostri).16 Although the term ‘representation’ is not yet used here the conception of an intellect which has something before it (objectum) is basically the same. It was then thought: if from this content which the intellect has before it, as imagination its image, one removes all determ inateness, what results is the concept of being.17 And it was this conception of ‘being’, which no longer has anything to do with the actual use of ‘is’, which Hegel took as his point of departu re in his Logic. 18

We shall later have to see w hether this conclusion, that transcenden­tal philosophy in its attem pt to construe the relationship to objects1 non- linguistically falls back on an empty concept, can also withstand a more precise examination. You could think: even when we refer to an object linguistically we must still also represent it to ourselves. But if someone uses a singular term, e.g. ‘Peter’, we do not ask him: ‘Who do you rep ­resent to yourself with “Peter”?’ but ‘Who do you mean with “Peter”?’ We do not represent objects to ourselves, we mean objects.

But what, you will ask, is it to mean an object? Now that is precisely the question that will have to be investigated. It is a question which can be investigated - the language-analytically purified question of tran ­scendental philosophy - whereas the question of what it is to represent an object is a pseudo-question. Even the ordinary language ‘Vorstellen’

Introduction 64

which of course, unlike the philosophical term, is meaningful (making something present in the imagination) is only possible in the context of meaning something: if someone, e.g., tries to sensuously represent to himself an object, this means that he tries sensuously to represent to himself the object which he means.

But has the whole of transcendental philosophy been so exclusively orientated towards representation? You will point out that I myself have drawn attention to the fact that Kant conceives of the conscious­ness of objects as a synthetic act. However, if we ask what it is that this act synthesizes the answer is: representations. It is of secondary im por­tance that Kant actually uses this term .19 What is decisive is that the manifold which consciousness synthesizes consists of simple data (‘sense-data’). And just as little as the consciousness of objects can be thought of as representation can it be thought of as a combination of representations. In general one can say: so long as one failed to con­sider logical structure it was not possible to think of the relationship of consciousness to objects in any other way than by analogy with a sen­suous having before oneself, arid that means - whether this expression is used or not - as ‘representation’. Though other perspectives were always being added to it, it was the orientation towards ‘representation’ that rem ained decisive. Particularly questionable is the development of this conception in Germ an Idealism (though this would require a criti­cal interpretation on its own). H ere representation was further formal­ized into a ‘subject-object relation’ and this one sought to grasp with general logical-ontological concepts such as those of identity and oppo­sition. These concepts, for their part, were taken over naively without considering the way they function in sentences. And then, since con­sciousness allegedly cannot be grasped with the understanding, by means of logic, these concepts were combined in paradoxical fashion in a so-called dialectical logic. Once one has started with a problem in a wrong way, and there are prejudices which prevent things from being put right, then all that seems to remain is the escape into the apparent profundity of the paradoxical.

L E C T U R E 6

The argument with the philosophy of consciousness continued

The debate with the second step of the philosophy of consciousness - the transcendental approach - went in favour of the language-analyti­cal position. How is it with the third step, in which the transcendental question concerning the conditions of the possibility of the experience of objects led to modes of consciousness which are no longer objectual?

In this extension of the enquiry beyond objects transcendental phi­losophy failed to take account of sentences. It thereby passed over a whole dimension of non-objectual consciousness without which there is also no objectual consciousness. Thus in extending the thematic it started out from an unclear basis.1 On the other hand, with the world- problem there is opened up, in both the Kantian and the Heideggerian version (connection of objects in space and time; connection of signifi­cance), a dimension of consciousness which goes beyond the under­standing of sentences just as much as beyond the relationship to objects. Also the other non-objectual modes of consciousness, such as the con­sciousness of action-rules (the same is true of the experience of a sen­sible manifold such as the view of a landscape or the hearing of a mel­ody) are not ‘logical’ modes of consciousness, are not articulated in sentences. Here then we encounter a limitation of the language-analyt­ical approach, if this is understood as formal semantics.

But if we recognize such a limitation from whence do we get our criterion of universality? Clearly we orientate ourselves towards a broad concept of consciousness in the sense of what Heidegger meant by ‘dis­closedness’. But what is to be understood by consciousness in general? Plainly we have no clear concept of this. Nor do we have a clear concept of the various non-logical modes of consciousness; what is actually meant by a consciousness of spatial and temporal connections, by an action-consciousness and so on? Until we can see more clearly here we

Introduction 66

can have no idea how we should concretely conceive an extension of the universal approach beyond the sphere of sentences. In particular it must remain unclear whether, and if so how, a formal analysis of a non- linguistically articulated consciousness is possible, what formalization would mean here, or what would have to take the place of formaliza­tion. One can clearly only make progress here by clarifying our under­standing of the various modes of consciousness and of the notion of consciousness in general. How can modes of consciousness be analysed? One might think: by introspection, by inner perception. But is con­sciousness something we can inwardly look at? Is there such a thing as inner perception, inner observation? I ask you to seriously try to per­ceive what is inside you. Is it not immediately evident that this is a non­sensical idea? We can observe with our senses, and if we speak of observing something inwardly this can mean: attending to one’s bodily sensations. But this is not what one means when one speaks of what is inner. What is inner - that is consciousness, and here there can be no question of anything like observation or perception. Perhaps you will say: ‘But I know indubitably that I am conscious of this and that; this consciousness must therefore be somehow inwardly given to me.’ Must it? It would be over hasty to argue: what we do not know by means of outer observation we know by means of inner observation. Perhaps the so-called inner differs much more fundamentally from the so-called outer.2

Instead of philosophizing in this way by postulation we would do bet­ter to look at how things actually stand. But where should we look if not by inner perception? What is to be done if we do not even know where and how we have to look, if we do not know how something about which we speak is to be made evident? If all that is given to us of something is our speaking about it then we can only elucidate it by examining how we can speak about it. It seems then we can only clarify even that the­matic which extends beyond the understanding of sentence-forms by means of linguistic analysis. To be sure, I am speaking now o f ‘linguistic analysis’ in the broad sense of an analysis of meaning, not in the narrow sense of an analysis of sentence-forms. Linguistic analysis in this broad sense takes the place of descriptive phenomenology, if one rejects as fictions the peculiar forms of intuition - inner intuition and intuition of essences - presupposed in phenomenology.

I would like to try to demonstrate to you the methodological superi­ority of linguistic analysis to phenomenology by reference to that way of speaking of consciousness which has become decisive for phenom e­nology itself (in Husserl).

The argument continued 67

Husserl distinguishes two concepts of consciousness.3 The one derives from our speaking of someone having a consciousness of some­thing or a conscious relation to something. Husserl calls this conscious­ness of something ‘intentionality’ or ‘intentional experience’ (intention­ales Erlebnis). It is for Husserl the decisive concept of consciousness. However, ‘intentional experiences’, according to Husserl, fall under the more comprehensive genus of experiences in general. In each person experiences belong to the unity of a ‘stream of experiences’. We also call this ‘stream of experiences’ ‘consciousness’. In this sense we say, e.g., this and that is contained in my consciousness - that is, it is a part of the whole of my consciousness in the sense of my stream of experi­ences. Consciousness in this sense - Husserl’s second concept of con­sciousness - is clearly founded upon the concept of ‘experience’ (Erleb­nis). The two terms which are fundam ental to Husserl’s concept of consciousness, and which stand in need of clarification, are, therefore, ‘experience’ and ‘intentionality’. Husserl seemingly carries out the explication and clarification of both concepts phenomenologically, using the method of inner intuition. I would like to show that in neither case is there really any question of an inner intuition and that the deci­sive criteria are exclusively linguistic.

By ‘experiences’ Husserl understands anything which can be inwardly perceived by the person whose experiences they are. Husserl claims that the possibility of such an inner perception is evident. How can Husserl claim something to be evident which we have just seen can­not even be found? From Husserl’s reference to the Cartesian sphere of inner certainty we can immediately see what he has in mind. I have already drawn attention, in the previous lecture, to the fact that one can define a first concept of consciousness in such a way that it embraces all states in which the person whose states they are has, at the time at which he has them, an indubitable knowledge that he has them. These states are what Descartes called cogitationes and what Husserl calls experi­ences. W hen I say: I am in such-and-such.a mood, have such and such phantasies, feelings, intend, believe, wish this or that, it is clearly inap­propriate to ask: ‘How do you know that, are you sure of that?’ (one can only ask: ‘Are you telling the truth?’) - as opposed to when, e.g., I say: ‘I weigh 70 kg’ or ‘My upper right wisdom-tooth is hurting.’ This is the fact to which Descartes drew attention and which Husserl too starts out from. Husserl, however, immediately placed the following in terpre­tation on it: if I cannot doubt a state of myself then it is directly given to me, I directly perceive it. Husserl believed that he had an evident perception of this perception. However, m ust one not say: this allegedly

Introduction 68

evident perception is as fictitious as what it is supposed to be a percep­tion of?

Wittgenstein, in his Philosophical Investigations,4 has pointed out that an approach such as Husserl’s illegitimately assimilates statements about what is inner to statements about what is outer. If we make a statement in which an ‘experience’ is expressed it is clearly not based on an external observation. From this philosophers like Husserl conclude that it is based on internal observation. But must a statement always be based on something? I have just pointed out that in the case of a state­ment about an experience it does not make sense to ask: ‘How do you know that?’ Husserl presupposes that such a question always makes sense and that one can reply: ‘I know it by inner perception.’ That a statement is not based on anything appears strange to us. But this is only because we automatically assimilate all statements to one another and construe them on the model of observation-statements. If we examine how things actually stand we see that in this case the statement itself is something ultimate.

T he characteristic feature of experience-words, for Wittgenstein, is ‘that the third person present is to be verified by observation, the first person not’ (Zettel §472). T he statem ent ‘I am worried’ is not based on the worry in the sense that it rests on the observation that I am worried but in the sense that in it the being worried is expressed, as in a cry (Phil. Inv. §244). On the other hand - and this distinguishes it from the cry- the statement is uttered by me as one which can be repeated as this same statement by others and can then be verified by observation. Both these things - that the word in the 1st Person Present is not to be veri­fied and that in the 3rd Person it is to be verified by observation - belong essentially to the mode of employment of such a word. ‘I do not say it on the basis of the observation of my behaviour. But it only makes sense because I so behave’ (§357). (And that means: because another can say it on the basis of the observation of my behaviour and my utter­ances.) If it were not so then we could not learn and understand such a word.

By dem onstrating this two-sidedness in the use of experience-words Wittgenstein made possible an understanding of the so-called inner which is distinct from both behaviourism and introspectionism. In the context of our discussion we need not pursue this problem any further. Our sole concern here is to show that, contrary to Husserl’s view, the certainty-criterion which Descartes used to define cogitationes does not rest on an inner perception. T he certainty is not something positive, but simply the negative fact that here ‘the expression of uncertainty is

The argument continued 69

senseless’ (Phil. Inv. §247) because the request for a justification is sim­ply not applicable and a ‘doubt is logically excluded’ (Phil. Inv. p. 221). The criterion of an experience-sentence in the 1st Person Present such as ‘I am in pain’ is, therefore, that it can be automatically converted into ‘I know that I am in pain.’

Because this conversion, though always possible, contains no addi­tional information it hardly occurs in ordinary language. Where, how­ever, philosophy orientated itself towards this sentence ‘I know that I am in pain’ without noticing that its meaning is precisely that of the sentence ‘I am in pain’, the illusion arose that the person who utters such a sentence observes himself and perceives the experience-state with absolute certainty. The orientation towards this form of sentence in which the word ‘I’ occurs twice resulted in self-consciousness being characterized as a relation o f ‘the self’ to itself (so-called reflection). The real peculiarity of sentences in which someone says something about his experience-states is not the double occurrence of the word T , but rather that they - and indeed precisely in the simple form - are not statements about the state, but its expression, and hence exclude doubt. And it is precisely this last-mentioned fact which is expressed by the sentence with the double ‘I ’ (‘I know that I . . .’).

With this the language-analytical basis for Heidegger’s analysis o f the disclosedness of one’s own being is reached. Heidegger rejected the theory of reflection just as much as Wittgenstein;5 his mistrust of sen­tences, however, hindered a clear destruction of the theory of reflec­tion. But Heidegger started precisely where Wittgenstein finished. The obsession with reflection had prevented philosophers from seeing what one is actually conscious of in self-consciousness. This can only be apprehended by attending to the simple sentences (without a double ‘I ’) in which someone says something about himself by saying e.g.: I am in such-and-such a mood, I intend to do this and this. I am not directed towards myself in the utterance of such sentences, but express my being-thus-and-so (being-in-a-mood, being-bent-on). It is the analysis of this being which, according to Heidegger, is the task of a properly understood theory of self-consciousness.6

I come now to the other, and for Husserl decisive, concept of con­sciousness: that of ‘intentional experience’. This is intended to desig­nate that class of experiences the peculiar character o f which is to be directed towards an object. Consciousness in this sense, then, is essen­tially consciousness of something.

Now we must ask ourselves: how does Husserl establish the existence of this object-relation, and in what does it consist? What is meant ‘strikes

Introduction 70

us unmistakably in any illustration we choose. In perception something is perceived, in imagination something imagined, in a statement some­thing stated, in love something loved, in hate something hated, in desire something desired, etc. It is the common feature that can be appre­hended in such examples that Brentano has in mind when he says: “Every mental phenom enon is characterized by what the scholastics of the middle ages called the intentional (or mental) inexistence of an object and what we . . . would call the direction to an object.” ’7

How does Husserl establish the directedness to an object by means of these examples? Here too Husserl appealed to the evidence of the intu­ition of essences (Wesensanschauung). However, the examples show that again the criterion is simply a linguistic one. He notices that verbs such as ‘perceive’ ‘state’ ‘hate’, etc., are transitive, that they have a gram m at­ical object. Intentionality is thus a relation. Since intentionality is sup­posed to constitute what is to be understood by ‘consciousness in the strict sense’,8 everything depends on making clear what it is that distin­guishes these transitive verbs from non-intentional transitive verbs.

Husserl gives no further explanation. I will now try to work out what is specific to the intentional relation by asking: of what kind are the objects referred to by the grammatical object of these verbs? We find that the majority of these verbs are not completed by singular terms which stand for concrete (perceptual, spatio-temporal) objects but by nominalized sentences, thus by linguistic expressions of the form ‘that p ’ which stand for ‘states of affairs’. When Husserl, in his series of examples, says that in a statement something is stated, the word ‘some­thing’ clearly stands not for a concrete object, but for a state of affairs. The same is true of most intentional verbs or modes of consciousness e.g. knowing, believing, doubting, wishing, questioning. The sentence ‘I know . . .’ can only be completed by an expression of the form ‘that p’. Something corresponding is true of asking a question and intending, except that the completing sentence has a different grammatical struc­ture. We can therefore say that the characteristic feature of these inten­tional verbs is that they stand for a relation which holds not between two concrete objects but between a concrete object (a person) and a state of affairs. In English philosophy, in which the term ‘proposition’ has been adopted for ‘Sachverhalt’, these modes of consciousness are called propositional attitudes. We can speak of propositional modes of conscious­ness.

Now clearly not all the examples Husserl gives of intentional experi­ences belong to the class of propositional modes of consciousness. In the case of some of them, e.g. loving, pitying, admiring, only a singular

The argument continued 73

term which stands for a concrete object can appear as transitive object. Moreover, there is a group of intentional verbs which can be used in both ways. To this mixed group belong perceiving, seeing, rem em ber­ing, desiring. One can say ‘I see that the sun is rising’, but also ‘I see the sun’, ‘I desire to eat a piece of b read’ but also ‘I desire a piece of bread.’ We can distribute this mixed group according to their two modes of employment among the two other groups, and we would then have two fundam ental classes of intentional modes of consciousness: proposi- tional and non-propositional, those which relate to states of affairs and those which relate to concrete objects.

Now the question is: what do these two classes have in common that distinguishes them from all other relations? We cannot appeal to the fact that both are modes of consciousness of something; for what this means is supposed to be explained by the characterization of intention­ality. Husserl says that common to all modes o f consciousness is a direct- edness to something. But what does this mean? ‘Directedness’ is clearly a m etaphor. Sign posts and guns are directed towards something; how­ever they are not intentional. Again the appeal to inner evidence may appear tempting. But if someone were to say to me: ‘You see when you are conscious of an object, perceive it, rem em ber it, fear it, that you are directed towards it; and you see how this relation differs from other relations’, then I would say: I see nothing at all. I sense a difference of course, but here it is a question of making clear what is unclearly sensed; and for this I have no intuition at my disposal - only linguistic usage.

If there was only the one class of intentional modes of consciousness, viz. the propositional, then we would have a clear distinguishing crite­rion. For we do not find relations between a concrete object and p rop­ositions outside intentionality. Opposed to this criterion, however, stands the o ther class of experiences designated as intentional. What is to be done in such a case? T hree possibilities are conceivable, (a) the modes of consciousness of the two classes have nothing in common, in which case the concept of intentionality turns out to be a pseudo­concept or (b) we succeed in finding a completely different common characteristic; the supposition that the orientation towards the propo­sitional could be a way of achieving a general criterion of intentionality would then be shown to be mistaken, or (c) it would have to be shown that the non-propositional modes of consciousness are only apparently non-propositional, that in reality they imply propositional conscious­ness. It is this last possibility which, in my opinion, can be realized.

I assert then: all non-propositional intentional modes of conscious­

Introduction 72

ness imply propositional modes of consciousness. We can take as our point of departure a thesis of Brentano according to which what distin­guishes intentional relations from other relations is the fact that the second term of such a relation need not exist.9 We shall see in a moment that this characterization does not apply to all non-propositional inten­tional relations. It is, however, clearly correct for most cases. X can fear, love, desire, etc.,7V, even though TV does not exist. Non-intentional rela­tions, by contrast, are not possible unless both terms of the relation exist. If N does not exist, then I cannot hit N, eat him, or sit on him. But how is this peculiarity of intentional relations to be explained? Shall we say that in the case of an intentional relation the object is, so to speak, in the mind of the person concerned and, hence, that the rela­tion is also possible when the object does not exist in reality? But this way of speaking is again clearly metaphorical. How can we give it a clear meaning? Probably by saying that X must at least believe that N exists. I can fear the devil without him existing but not without believing that he exists. To believe that N exists - or, to put it more precisely, that there is exactly one object to which the properties implied in ‘AT belong - is a propositional mode of consciousness. What Brentano drew attention to, viz. that the object of an intentional mode of consciousness need not exist, is therefore simply a consequence of the fact that one can only relate intentionally (consciously) to an object in such a way that one believes it to exist.

T hat it is really this that is characteristic of the non-propositional intentional relation is shown by those cases to which Brentano’s thesis does not apply. If I say ‘X sees, hears, recognizes N ’, then that N does not exist is ruled out. T h a t would only be possible if I had said: ‘X believes that he sees, hears N .’ However, even here the consciousness that there exists an object that = N is implied. When we say of X that he sees N this means: he knows on the basis of his optical perception (1) that there is something which = N (2) that here (in his optical surround­ings) there is something, (3) that this =N. T hat the apparently simple sentence ‘He sees N ’ contains such a complex assertion can be seen from the fact that if someone says ‘I see N ’ one can dispute this assertion in three ways: (1) there is nothing which = N (N does not exist) (2) there is nothing here or (3) this (what you see) is not =N. Thus seeing, etc., also implies a propositional consciousness that N exists. At the same time the account just given makes it clear why Brentano’s thesis does not apply to these cases. If one says of someone that he sees, hears, recog­nizes N, this means that he does not just believe the implied state of affairs; he knows it.10 Now if one says of someone ‘He knows that/?’

The argument continued 73

(and not: ‘He believes he knows th a tp ’), then one asserts (among other things) that it is true that p. Thus, whoever says ‘X sees N ’ (and not just ‘X believes he sees AT) at the same time endorses the truth of the state­ments implied by X’s statement ‘I s e e N ’, hence the truth of the implied existential statements; and fo r this reason it is implied in these cases that N exists.

To reinforce my thesis that all non-propositional intentional con­sciousness implies a propositional consciousness — or, more precisely, a belief that that to which it refers exists - I would like to deal with another apparent counter-example. How is it when we represent some­thing to ourselves in the imagination? The specific character of this mode of consciousness seems to be precisely that that to which it refers is meant as non-existent. But how is this non-existence to be under­stood? For example, one is telling a story, a joke or the like and one begins ‘A man . . .’ Implied is: ‘Imagine that there existed a man . . T he man is thus meant as non-existent. But that can only be done by thinking of him as existing. It would be incorrect to think that when we mean something in the imagination the existence is taken away and the mere object is left over. T he phantasy-modification has the character ‘it is not so, but I imagine that it were so’. T he phantasy-modification is a modification of veritative being. It therefore concerns not just objects but, e.g., a whole story. Everything which is told there or thought up is not meant as really being so, but only thought of as being so. It is, however, thought of as being so. And for the objects this means that they are not meant as really existing, but are merely thought of as exist­ing. They are, however, thought of as existing. Thus phantasy-con- sciousness too is implicitly propositional.

In the previous lecture I drew attention to the fact that to have a consciousness of an object is not to have a representation of it but to mean it, and to mean an object by means of a singular term is a non- independent component in the understanding of predicative sentences. T he reflections which have now been carried through take us a step further: to mean an object is not only a non-independent component of a propositional consciousness; it in turn rests on a propositional con­sciousness, which consists in holding to be true an existential sentence.11 And thus the general proposition would be proved that all so-called intentional consciousness is explicitly or implicitly propositional con­sciousness. The essence of the intentional relation consists in this: it is a relation between a concrete object and a state of affairs. And that means: that it is grounded in the understanding of a sentence. Thus the attem pt to recover in a language-analytical way Husserl’s ‘strict’ con­

Introduction 74

cept of consciousness has led us to an unexpected discovery: we find that a consciousness of something that is not founded in a holding-to- be-true of an existential sentence does not exist. The peculiar ‘quality’ of consciousness which Husserl called ‘intentionality’ and which in sup­posedly intuitive description he characterized as a directedness to an object turns out to be sentence-understanding.

Let me rem ark, incidentally, that the description of consciousness as a subject-object relation, of which the transcendental philosopher is so fond, is also thereby rendered untenable. T here is no such relation. W here a subject has a conscious relationship to an object this is never a simple relation but is always founded in the understanding of a sen­tence. A particularly unfortunate consequence of the idea of the ‘sub­ject-object relation’ was that the attem pt was also made, in German Idealism, to interpret self-consciousness in accordance with this schema. In the case of self-consciousness the subject is conscious of him­self. This consciousness was, therefore, interpreted as a subject-object relation in which subject and object are identical; thus as a relation of something to itself.

This idea, which was simply a result of the underinterpretation of the actual facts as manifested in the sentences in which someone speaks about himself led the already inadequate conception of self-conscious­ness as reflection into absurdity. Instead of simply orientating oneself towards sentences with the double T (‘I know that I . . .’) one took these two occurrences of T out of their sentence-context and constructed an abstract self-relation of ‘the’ self to itself. Insoluble problems were obviously bound to result from such an approach.12

The demonstration that all intentional consciousness is propositional gives an additional historical significance to the language-analytical pro­gramme of a theory of sentences: the question about consciousness, like the ontological question about being as being, turns out to be a question about the understanding of sentences. O f course this only applies to consciousness in the sense of intentionality, not to the non-objectual modes of consciousness.

A language-analytical clarification of the non-objectual modes of con­sciousness, unlike that of experiences and intentionality, would not be able to base itself on work that has already been done. More extensive preparation would therefore be necessary here. I thus leave open the question of whether, and, if so, how, the question concerning the understanding of sentences can be surpassed by invoking either a more comprehensive concept of understanding such as that aimed at by ‘pragmatics’, or a more comprehensive concept of consciousness of the

The argument continued l b

kind to which the transcendental-philosophical attempts point. And so the question also remains open as to how the'universality-claim of the language-analytical conception of philosophy I have developed is to be judged and whether, and if so how, this conception could be extended.

But does this mean then that the decisive question remains unan­swered? The decisive questions never find an answer in philosophy. This is not to say that they are unanswerable, but only that, when the questions left open by a philosophy (e.g. ontology) are answered by a new philosophical approach, this new approach, so long as it is philo­sophically alive, itself reaches into the dark, inasmuch as it encounters questions which, for the time being at least, it cannot answer. In the case of such questions, which concern the limits of a philosophical posi­tion, it is already a gain if one at least perceives these limits. You could ask: if I see the limits of a sentence-orientated theory of understanding why do I not go beyond them? My answer is: because I do not know how the concept of formalization can be extended beyond these limits or what would there take the place of formalization. Perhaps you will draw my attention to the fact that I myself expressed doubts, in Lecture 1, about an a priori conception of philosophy. Should one not be even more dubious about a purely formalistic conception? Certainly. From such doubts, however, there would not immediately result a new posi­tive conception of philosophical method, but only a syncretism. You will see that it will not be at all easy for us even to find an adequate concep­tuality for the new semantic thematic. For in the main the categorial means available to us still stem from an object-orientated tradition; and it seems to me doubtful w hether it is possible to develop a new concep­tuality other than by debating with the inadequacies of previous ones (cf. Lecture 8). Only somebody who does not see these specifically philosophical difficulties of conceptual clarification and the formation of categorial means adequate to a m ode of enquiry can wish to take two or more steps at a time.

L E C T U R E 7

A practical conception of philosophy

Today I come to a new and final attempt at an introduction of the language-analytical conception of philosophy. It is orientated towards the idea of reason. T he word ‘reason’ (Vernunft) is not used univocally. In the German Enlightenment vernünfteln was used to translate the Latin ratiocinari, just as today raisonner in French and to reason in English are still in use. The faculty of reason would accordingly consist in the capacity to argue. It is this that Kant has in mind when he defines rea­son in its logical sense as the faculty of making inferences (Critique of Pure Reason, B355). But from this there follows a second ‘transcenden­tal’ definition, according to which reason is designated as ‘the faculty of principles’ (B356). ‘Principles’ are the first and, hence, unconditional propositions of a deductive system. From this there resulted for Kant a concept of reason according to which reason stands for the conscious­ness of unconditioned totality (B378 ff.). On this basis German Idealism then came to oppose reason, wholeness, dialectic, to the ‘mere’ under­standing, and there began that disdain for the logical that is character­istic of the German development of the last century and a half. Despite this peculiar development of German philosophy the words Vernunft and vernünftig retained, in ordinary language, their original Enlight­enm ent sense, whereas vernünfteln and räsonnieren are now only used in a pejorative sense. In ordinary language vernünftig means something like ‘well grounded’. And the dem and to use one’s reason means: one should not take over opinions unexamined, but enquire as to their grounds and counter-grounds. The capacity to argue is not only a capacity to make deductive inferences but, more generally, the capacity to justify statements. Ratio, raison, reason can mean both ‘ground’ (Grund) and ‘reason’ (Vernunft). T he faculty of reason is the capacity of being able to answer for one’s beliefs and actions (Latin: rationem red- dere. Greek: logon didonai).

A practical conception of philosophy 77

A conception of philosophy which is orientated towards the idea of reason thus takes up that aspect of the Aristotelian preliminary concep­tion of philosophy which had been neglected in its ontological realiza­tion. There have been many attempts in the history of philosophy to develop a conception of philosophy as the science which radicalizes the aspect of justification which is present in all sciences; for example, in Plato, in Descartes, in German Idealism, in Husserl. However, I shall not undertake the introduction of a language-analytical conception of philosophy which is orientated towards the idea of reason in connection with these historical positions. For now, at the end of the introductory reflections, I would like to make an attempt to justify the language- analytical conception of philosophy - and with it an idea of philosophy as such - absolutely, rather than relatively to given historical concep­tions or a given understanding of ‘philosophy’.

With this I link up with the reflections of Lecture 2: to justify a par­ticular conception of philosophy can mean: to show that it corresponds to our preliminary understanding of philosophy or that in it the inten­tions of earlier conceptions of philosophy can be realised or better real­ised. The previous part of my introduction moved within this frame­work. One can, secondly, try to justify a particular conception of philosophy directly. But, as I tried to show in the second lecture, this can only mean: to justify the motivation for this activity. By this I mean: to demonstrate that it is advisable (ratsam) to engage in this activity. A justification of a conception of philosophy which is not merely relative to a presupposed preliminary understanding of the word ‘philosophy’ can, therefore, only have the sense of a practical justification.

If such a practical introduction is not itself to presuppose a particular understanding of a word then we cannot start from a particular concep­tion of philosophy and only subsequently attempt to justify it practi­cally. Rather we must ask: is there a theoretical activity - for we can assume that this is what is at issue - which it can be shown to be advis­able to engage in? And whatever the answer will be, it will be this theo­retical activity which - being practically pre-eminent - we shall call phi­losophy. Or we can drop the assumption just made and simply ask: is there any activity which it is advisable to engage in? Or better: what is it advisable to do? This is clearly the most comprehensive practical question that it is possible to ask, and we can now say: if it motivates or requires a particular theoretical activity then we will call this theoretical activity ‘philosophy’.

Let us first try to get clear in a preliminary way about the meaning of such practical questions in general and about the meaning of the most

Introduction 78

comprehensive practical question. What are practical questions and to what context do they belong?

There is a class of actions which are characterized by being inten­tional.1 Some authors even define ‘action’ in terms of intentionality, for of an unintentional movement and its consequences one does not say ‘He did it* but ‘It happened to him.’ What criterion do we have for recognizing an intentional action? The distinguishing criterion o f intentional action seems to be that the action can be characterized by reference to its intention. Someone is making movements at a window. What is he doing? He is letting fresh air in. That is his intention. But the opening of the window - as means to an end - is also intended. So to the question ‘What is he doing?’ we can also reply: ‘He is opening the window.’ How do we tell that the action is really intentional and has precisely this intention? Ultimately only from the fact that the person concerned is himself prepared to express his intention. If one accepts this criterion, then one can only speak of intentional acting, and hence of doing or acting in a narrow sense, with respect to beings who can speak. In our context we can ignore the problem of unconscious inten­tions,2 for what concerns us is that our entire conscious life, that which can express itself linguistically, is always characterized by intentions and intentional activity. We can always ask someone who can speak and who is not asleep or unconscious: ‘What are you doing?’

We achieve an additional understanding of the relation o f intention and action if we consider that we can also intend future actions. If someone has the intention o f doing something in the future he can also simply say: ‘I will do this*, e.g. ‘I will come.* It is to be noted that such sentences in the 1st Person Future which look like assertoric sentences are not assertoric sentences (cf. above p. 49). The sentence ‘I will come’ is not a prediction but a sentence in which an intention is expressed. I shall call such sentences intention-sentences. If the person said ‘I will come* and he does not come we do not say ‘He made a mistake’, but ‘He did not stick to his declaration.’ When I say to someone ‘I will certainly come* this ‘certainly’ expresses, not the theoretical certainty of a predic­tion, but the resolution to stick to my word.

Just as the denial o f the negation, considered as a possibility, under­lies theoretical certainty (I am sure that p when I am convinced that it is excluded that not-p) so too the practical certainty of resolution is grounded in a denial o f the negation, considered as a possibility, o f the intention (the intention-sentence): ‘I will certainly come’, that is: for me there is no question of my not coming. Now we encounter here a pecu-

liar feature of intentional activity that is conscious (capable of linguistic articulation): because the act is determined by an intention and this is articulable in a sentence, but the sentence is negateable, with the con­sciousness of the act there is always given the consciousness of the pos­sibility of refraining from the act. Conscious intentional action is there­fore always situated in a range (Spielraum) of possibilities against which one can of course close oneself but for which one can also open oneself in deliberation.

With this we have reached the phenomenon of so-called human free­dom.3 It is better if we speak of a free action than of a free will. All conscious intentional action is free. However, the mere consciousness of alternative possibilities does not amount to freedom; freedom requires that the consciousness of alternative possibilities can be action- determining. Where this is not the case we speak of compulsive acts; and we do not say of compulsive acts that they are intentional. The compulsive act happens to the person concerned. How can one estab­lish whether the consciousness of alternative possibilities is action-deter­mining? By connecting with the act which someone claims not to be able to avoid a disproportionate punishment. If he still cannot refrain from the action then he is acting unfreely.

You will now perhaps ask: how can one know that action, in which the consciousness of alternative possibilities can be action-determining, is really free, that it is not itself determined? This question is, however, no longer permissible if a free act is defined as one in which the con­sciousness of alternative possibilities is action-determining. The ques­tion just posed presupposes a concept of ‘real* freedom which would be opposed to determinism. I hold this metaphysical concept of freedom to be fictitious; I would therefore say that in this sense of freedom intentional action is not free. By contrast, freedom as I have just defined it is a phenomenon, for which there are empirical criteria. And it seems to me that it fulfils what we mean when we, e.g., say ‘I am free to do it’ or ‘It depends on me whether I will do it/

Now just as there is deliberation with reference to the range of free­dom, so too there is questioning, for all deliberation is guided by a ques­tion. This question must be understood as the question-counterpart of the intention-sentence, just as the theoretical question is the question- counterpart of the statement. When we deliberate and ask what is to be done we take counsel with ourselves. We can therefore say that the question-counterpart of the intention-sentence is the question: what is it advisable to do? - thus what I have called a practical question.4

A practical conception of philosophy 79

I ntroduction 80

However, it is stili unclear w hat is really being asked about w hen one asks: what is advisable? Is there a typical question-form ula for this ques­tion which brings ou t the logical character o f this question?

If the practical question is the question-counterpart of the inten- tional-sentence, then its general form would have to be that which results from “ the question-m odification o f such a sentence, thus, e.g., ‘Will I com e?’ and , as the com prehensive question, ‘W hat will I do?’ Now if in an intention-sentence an in tention is expressed and ‘I will com e’ m eans som ething like ‘I in tend to com e’, then the question m ust also be able to take the form ‘Do I in tend to come?’ and, as com prehen ­sive question: ‘W hat do I in tend to do?’ This question, however, unless fu rth e r qualified is at least am biguous, in tha t it gives the im pression of being a theoretical question (‘which intentions, motives are p resen t in me?’). Only at a pinch can it be understood as a practical question, i.e. as a question which initiates a deliberation and is directed to a decision, not a statem ent. T h e previously m entioned form ‘W hat will I do?’, on the o ther hand, is am biguous in ano ther way: it can have the sense of a request for a theoretical prognosis.

N onetheless a deliberation can be expressed in this question and also in the question ‘W hat do 1 in tend to do?’ You see, it is not the g ram ­matical form which is decisive, bu t the m ode of em ploym ent.

T h e re is, however, still an o th er gram m atical form for the practical question in which the possibility o f m isin terpreting it as a theoretical question seems to be excluded. W hen we deliberate on what is to be done we are m ost likely to say ‘W hat ought I to do?’ A nd yet this form too is am biguous, for such a question can equally well be used as a request for an o rder: it is at th e same tim e the question-counterpart of the im perative. This is connected with the fact that semantically inten- tion-sentences and im peratives are closely related; intention-sentences are the equivalent in the 1st Person of im peratives in the 2nd (or 3rd) Person. In the case of a statem ent (‘H e will come') the relationship between sentence and event is such tha t the sentence is supposed to correspond to the event, to be correct (or incorrect) relative to it. In the case o f the im perative and the intention-sentence, on the o th er hand, the relationship is such tha t the event - namely the action - is supposed to correspond to the sentence, to be correct (or incorrect) relative to it. (Com pare: ‘H e presents the m atter as it is’; ‘H e acts as he was o rdered to do/as he had resolved to do ’.) C onnected with this affinity between im perative and in tention-sentence is the fact that ‘O ught I to come?’ is used both as a question-counterpart o f ‘Com e!’ and as a question-coun­te rp a rt o f ‘I will com e.’ T h e question ‘W hat ought I to do?’ is, therefore,

A practical conception of philosophy 81

also am biguous; it can be used as a request fo r an o rd e r as well as fo r advice.

T hus to reach an u n d e rstan d in g o f practical questions we cannot confine o u r atten tion to any one o f the gram m atical form s ‘W hat will I do?, ‘W hat do I in tend to do?’, ‘W hat oug h t I to do?’ In the case o f each o f the form s we m ust add: i f it is u n d ersto o d as a request for advice. A nd so we find ourselves again having to fall back on the form ulation we started ou t from : ‘W hat is it advisable to do?’ We m ust the re fo re ask ourselves: what are we asking for w hen we ask for advice, what do we expect from advice?

Suppose we ask som eone: ‘W hat oug h t I to do (in this situation)?’ and suppose he understands that we are not asking fo r an o rd er bu t for advice, and suppose he replies: ‘Do w hat is best (in this situation).’ Surely we would say: ‘W hat is best, clearly, bu t what is the best (in this situation)?’ W hat seems ridiculous in the contex t o f a concrete question, because it is trivial, can be philosophically significant. T h e trivial answer to a practical question is ‘T h e best’. T his gives us a sim ple linguistic criterion fo r telling w hether the question ‘O u g h t I . . .?’ is m eant as a practical question, as a request fo r advice, and not fo r an order: nam ely, w henever it occurs in explicit o r implicit connection with the expres­sions ‘the best’ or ‘g ood’.

W hat do we m ean by the w ord ‘good’? W hen do we call som ething ‘be tte r’ than som ething else?5 T o this the following general answer can, I believe, be given: w hen the one is to he preferred to the o ther fo r objective reasons. T h e definition I am suggesting thus has two com ponents: (1) the w ord ‘b e tte r’ is used to express a p reference, and ‘to p re fe r’ m eans: with reference to a plurality o f possibilities, to decide fo r one, to choose it. T h e w ord ‘good’ thus belongs, as a p reference-w ord , to the context of choice and freedom . Only a being which is free in the sense previ­ously described can u n d erstan d the w ord ‘good’. (2) In contrast to o ther preference-w ords like ‘m ore p leasan t’ o r ‘like m ore’, which express that som ething is m erely subjectively p re fe rre d by som eone, the w ord ‘be t­te r’ expresses tha t the p reference is an objectively justified one.

T h e preference — both the subjective one and the objective one — can relate to things, states o f affairs o r actions. F or exam ple, we say o f a thing o f a certain k ind, a car fo r instance, th a t it is good o r is better than another. T his m eans: if one needs a th ing o f this kind, this is to be p referred to the o th e r fo r objective reasons (which does not ru le out that som eone m ight consciously choose the worse because he nev erth e­less likes it m ore. A lthough he ought to p re fe r the one, he chooses the other).

Introduction 82

T alk o f good things is d ep e n d en t on talk o f good actions. A thing is good if one can p erfo rm well with it the action for which it is servicea­ble. H ow ever, this is no t to say tha t ‘good’ m ust ultim ately refer to actions. For, th a t an action is to be p re fe rred to ano ther for objective reasons may in tu rn be g ro u n d ed in this, that the condition - the state of affairs - tha t is b ro u g h t about by it is to be p referred , for objective reasons, to th a t which w ould be b ro u g h t about by the o ther action. T he answ er ‘W hat is best’ is thus the trivial answer to the question ‘W hat oug h t I to do?’; but no t only to this question. T h e question ‘W hat ought I to do?’ is itself a special fo rm of the completely general question ‘W hat o u g h t I to choose?’ ‘W hat is to be p re fe rred for objective reasons?’ T he scope o f the trivial answ er ‘the best’ extends as far as this question. It is not only in questions concern ing o u r own actions tha t we find ourselves faced with a choice. ‘G ood’ m eans com pletely generally som ething like ‘w orthy of being desired ’. W hat we regard as w orthy of being desired, i.e. w hat we believe is to be desired (preferred) fo r objective reasons, determ ines w hat we believe we o u g h t to do. We can leave open the question o f w hether w hat is ultim ately (i.e. not relative to som ething else) w orthy o f being desired can itself only be though t o f as activity.

Practical questions then are questions concerning the good, the bet­ter, the best. In the practical question the questioner is not asking about w hat is (veritative being) b u t about w hat it is good (better, best) for him to do. Now w hat does it m ean to say tha t som ething is to be p re fe rred for objective reasons?

T o g ro u n d som ething objectively means to justify it. T his co rre­sponds to the concept o f reason which I in troduced at the beginning of today’s lecture. So instead o f saying o f that which is good th a t it is that which is to be p re fe rre d for objective reasons we can equally well say th a t it is tha t which is rationally to be p referred . A nd now we can also say: when we ask som eone for advice, o r take counsel with ourselves reg a rd in g w hat we oug h t to do, we are asking w hat it is rational to do, i.e. which course o f action we can g round , justify.

B ut w hat does tha t m ean? Justification (Ausweisen) is a species o f leg­itim ation (Rechtfertigung). W hen we legitim ate an action we are saying tha t it is correct; and by this we m ean that it conform s to a rule. All conscious action is ru le-gu ided and the re fo re the w ord-pair ‘co rrec t- inco rrec t’ belongs essentially to all conscious actions. An action can, e.g., be co rrec t relative to a ru le o f w riting, a rule o f a gam e, a social conven­tion, a technical ru le, a legal norm . O ne can speak o f a relative legiti­m ation here. O ne legitim ates on e’s action, renders an account o f it, rel­ative to a certain norm ; an d this then also m eans: vis-ä-vis the particu lar

partners who adhere to this rule. This relative legitim ation can already be called a grounding: one grounds one’s action, renders an account of it by reference to the ru le which it follows.

Justification differs from this relative legitim ation in being an absolute legitim ation, absolute in the sense tha t (a) it is not accom plished relative to a given ru le and hence also (b) not vis-a-vis particular partners but vis-a-vis arb itrary partners (and is in this sense ‘objective’). It is precisely the capacity for this absolute legitim ation tha t we call ‘reason’. W hat one is dealing with here then is a possibility o f answering for, rendering an account of, an action in a way tha t is not relative to a given rule.

Such an absolute legitim ation exists, in a p rim ary sense, only in re la­tion to linguistic acts, statem ents. T h e correctness o f statem ents does not hold relative to a rule, bu t absolutely; this correctness one calls tru th . I f we can also ask (in a secondary sense) of o ther actions w hether they can be legitim ated (in the absolute sense), this is because in ten ­tional acts imply intention-sentences which can be g rounded by state­m ents o f the form ‘It is good (better) that . . (including the form ‘It is good to do x.’) Such statem ents, which are possible replies to practical questions, we can call practical statem ents. T o every intention-sentence ‘I will do x*, but also to every action-sentence ‘I am doing x \ there can be attached a why-question about the reason for the action, o r for the intention. H ere reason m eans som ething like motive. If I am asked why I am doing som ething I can reply by giving either a subjective o r an objective reason. In the fo rm er case I say ‘Because I like doing it’, in the latter ‘Because it is good (better) to do this.’ Now when I d o x because I believe that the best th ing (in this situation) is to dox , then the justifi­cation o f this practical statem ent carries over to the action which implies this statem ent. O ne can therefo re also legitim ate non-linguistic acts in an absolute sense, show them to be rational, by justifying the statem ents about what is good tha t are implicit in them (and tha t o f course m eans: justifying them as true). T h e good is thus a species of the true , which is simply to say that practical statem ents are a k ind of statem ent. And the true in tu rn is a species o f the correct, that species, namely, in regard to which we can speak o f absolute legitim ation, justification. O ne can also oppose the tru e to the good; it then stands for those statem ents in which the word ‘good’, o r a corresponding word like ‘advisable’ or ‘o u g h t’, does not occur. It then stands only for those statem ents which, in con­trast to practical statem ents, one can call theoretical statem ents.

B ut now how are statem ents o f the form ‘It is good tha t/?’, ‘T h a tp is better than tha t q to be justified? Only when that is explained would the com plicated path we have followed in today’s lecture - a path which

A practical conception o f philosophy 83

Introduction 84

started with practical questions and led, via the phenom ena o f freedom , choice, deliberation , advice, the good as w hat is objectively to be p re ­fe rred , to reason as the faculty o f justification - lead beyond an appeal to ever new w ords to a genu ine result.

Let us first m ake clear to ourselves th a t sentences of this fo rm are used in such a way tha t they at any ra te make a claim to objectivity and justifiability. T h e characteristic fea tu re o f the justification-claim of statem ents is the possibility o f bracketing it by the use of expressions like ‘it seems (to me, to us) th a t’, ‘I believe t h a t . . T h e m eaning o f the justification-w ord ‘is tru e ’ (just like th a t o f any sim ple use o f an asser- toric sentence) lies in this contrast with ‘it seems th a t . . Now this con­trast is also p resen t in sentences in which the w ord ‘good’ is used. We can say: ‘I t seems to them to be good, they believe tha t it is good. B ut is it good?’ T his does not apply to subjective preference-w ords: we cannot (except u n d e r special circum stances) say ‘I believe th a t I like it’ 'That p seems to me to be m ore p lesan t than that q.’ T h e whole context o f ques­tioning, doubting , disputing , justify ing is absent here; bu t no t in the case o f those sentences in which it is asserted tha t som ething is good. T hese sentences the re fo re also have no subjectivity-indicator; their validity claim is objective, fo r all rational beings (e.g. fo r all who can ask abou t the justification o f sentences). T h is seems to be contradicted by the fact tha t we can say ‘T h a t is good fo r him , for m e.’ B ut th e re is an am biguity in the w ord ‘fo r’. ‘It is good fo r h im ’ m eans: it is conducive to his well-being; if his well-being is to be p rom oted th e re are objective reasons for p re fe rrin g this. B ut these reasons for this being b e tte r for him are valid no t ju s t fo r him bu t for anyone and , hence, som eone else can advise him with reference to his well-being.

T h u s it is clear tha t practical questions are actually rational questions, justification-questions. B ut now how is a justification o f practical state­m ents conceivable? T his question can easily be answ ered fo r all cases in which we ask w hat we o u g h t to do (what it is good to do) in order to achieve a certain end. T h e question o f w hat are the best m eans to p re ­supposed ends can be answ ered by recourse to theoretical reasoning, particularly abou t the relevant causal connections. But w hat about w hen we do no t ask ‘W hat oug h t I to do in o rd e r to achieve A V bu t simply ‘W hat o u g h t I to do (in this situation), w hat is the best to wish and to do (in it)?’, e.g. ‘Is a capitalistic o r a socialistic econom ic system better?’ Also in the answ ering o f such a question theoretical considera­tions take up the greatest space; we canno t significantly answ er such a question w ithout knowing the situation and theoretically pene tra ting

A practical conception o f philosophy 85

the alternatives envisaged in the question . B u t m ere know ledge is no t adequate fo r answ ering the question: w hat is it best to do? T h e legiti­m ation o f o u r u ltim ate aims is no t a m a tte r o f theoretical reason. K now ledge o f w hat o u g h t to be canno t be reduced to know ledge o f w hat is. So th e re arises the question concern ing the possibility o f the justification o f the irreducibly practical com ponen ts o f practical ju s tifi­cation. I t is the question concern ing the possibility o f practical reason.

So w hen we ask practical questions which a re no t m erely relative to p resupposed ends, w hen we ultim ately call in question o u r actions and desires, we are re fe rre d back to th e p rio r question o f th e possibility o f practical reason. B u t m ust we ultim ately call in question o u r action? W hat induces us to do so?

We have previously seen tha t practical questions arise fo r us to the ex ten t th a t we are conscious o f ourselves as faced with a ran g e o f pos­sibilities. Mostly we are no t conscious of a ra n g e o f possibilities fo r o u r action; an d w hen we are , this usually only concerns the question o f the correct choice o f m eans. Seldom do we also call in question o u r aims and ou r way o f life as a whole. B u t on w hat th e n does th e ex ten t o f o u r consciousness o f o u r ran g e o f possibilities d ep en d ?

T h e question: w hat induces us to ask practical questions? thus refers us back to th e question: w hat induces us to becom e conscious o f a ran g e o f possibilities? We can now say th a t it is the in te rest o f reason in ju s t i­fication. F or the read iness to ju stify the theore tical an d practical beliefs implicit in o n e ’s own action presupposes th a t one considers alternative possibilities. Reason presupposes freedom a n d ex tends only as fa r as freedom . If, th e re fo re , one wants to act rationally , on e has an in te rest in becom ing conscious o f on e’s ran g e o f possibilities an d in ex tend ing it. A im ing a t reason in one’s actions can n o t sim ple m ean: asking w hether w hat one is actually do ing is good. R a th e r it m eans asking w hat it would be best to do. It is th e re fo re not as th o u g h o u r action m oved in a fixed ran g e o f possibilities which is th e re in d ep en d en tly o f o u r con­sciousness an d to which we only have to a tten d . How fa r the ran g e o f possibilities ex tends dep en d s on o u r in te rest in acting reflectively, and tha t m eans: rationally. This rational in te rest can be restric ted to the m eans to given ends, b u t it can also rela te to th e ends them selves and , ultim ately, to o u r life as a whole. T h e question: w hat is w orthy o f being desired in general (and no t ju s t fo r me)? an d the question connected with it: w hat ough t I to do? (understood as a question re la ting to my whole life), I w ould like to call the fu n d am en ta l practical question . W hat induces us to ask this question is thus th e in te rest in reason, an d

Introduction 86

tha t means: the in terest in existing responsibly in an absolute sense, i.e. in such a way tha t we can be answerable for o u r actions not only with reference to given norm s and no t only with reference to given aims.

O ne could go a step fu rth e r and ask: but why should one want to be rational?

T his why-question is am biguous. I t can either be understood as being itself a question of reason , in which case it has the sense: is it rational to w ant to be rational? O r it can be understood as a question o f motivation: w hat as a m atte r of fact induces us to be rational?

Is it rational to want to be rational? We cannot, so it seems, legitim ate reason th ro u g h itself; this would be to presuppose the very thing we wish to justify. M ust we not then say that the rational in terest is itself irrational, canno t itself be rationally justified?

In o rd er to be able correctly to assess the logical situation in which we h ere find ourselves it will be best if we think o f it in the form of a dialogue. T h e re are then two possibilities; e ither the person who rep ­resents the in te rest o f reason (A) is challenged by ano ther person (B) to justify it; o r A tries to con v in ced o f his standpoint. In the first case A can reply tha t in asking him for a justification B is already presupposing the in terest o f reason; and in this sense - as the presupposition o f all rational questions - the in terest o f reason is itself rational. This legiti­m ation of the in terest o f reason corresponds to the way in which, since A ristotle, one justifies the Principle o f Contradiction. T h e Principle o f C ontradiction can also no t be justified directly, because it m ust itself be p resupposed in any justification. B ut for this very reason it cannot be called in question; if one speaks at all - in sentences - then one has presupposed the Principle o f C ontradiction. A ristotle therefo re says that, to be consistent, the oppo n en t (of the principle) would have to give up speaking .6 In the case of the in terest o f reason we are not in so favourable a position: h e re we can only say that to be consistent the o p p o n en t may not ask for a legitim ation. (One could still allow him to ask relative rational questions concerning m eans to ends; bu t the ques­tion asked by B, nam ely, how the in terest o f reason is to be legitim ated, was not in tended in this relative sense.) But, conversely, if he does not ask a rational question B also canno t be convinced by A o f the stand­poin t o f reason, fo r som eone who does not en ter into rational argum ent can clearly also not be rationally convinced.

T h u s we arrive at the following result. T h e in terest of reason is itself rational in the special sense that it is the presupposition o f all rational questions. B u t there is no absolute standpoin t outside reason from which one could persuade an in terlocu to r - perhaps the one in one’s

A practical conception of philosophy 87

own soul - who lives according to the principle of immediacy (non­deliberation) to accept the in terest of reason. O ne can only awaken and streng then the in terest o f reason th rough p roper education.

This also answers the second question, which concerned not the leg­itim ation of the in terest o f reason but its factual genesis. But does this m ean that everything is relativized again by the question: in what does the p ro p e r education consist? No, for the correct education is defined by this line of though t as tha t which cultivates the in terest o f reason. And although such an education is only one of many conceivable ed u ­cations, it follows from what has previously been said that it is the only one tha t is correct in the absolute sense, the only one which can be legitim ated as rational.

I can now re tu rn to my real purpose, a practical in troduction of a conception of philosophy. T his was supposed to arise from the context of practical questions and for this reason we had first to get clear about the n a tu re of practical questions. My suggestion at the beginning of the lecture was: if the practical question ‘W hat is it advisable to do?’ dem ands a certain theoretical activity then we can call this activity - as a practically pre-em inent activity - ‘philosophy’. Now the question ju st re ferred to is the question which in the m eantim e I have called the fundam ental practical question and which has em erged as the most com prehensive rational question. And the fundam ental practical ques­tion, we saw, points in tu rn to the prio r question concerning the possi­bility o f practical reason.

Now one could designate as philosophy both the concrete process of answering the fundam ental practical question with all its theore tical- veritative and practical-veritative implications and also the prior ques­tion concerning the possibility o f practical reason. T h e two belong together; and from what has gone before it is clear tha t the re is fo r both of them not just some m otivation but the highest rational motivation.

If we com pare this in troduction of philosophy with tha t of A ristotle then it already differs from it in the prelim inary stage which in Aristotle was characterized by the concept of science. To this prelim inary stage there corresponds in the presen t introduction the concept of reason so far as this is not yet specifically related to the fundam ental practical question. It com prehends both practical and theoretical reason and is thus m ore com prehensive than the concept of (theoretical) science. In the concept of reason there is taken up the aspect o f justification which was em phasized by Aristotle h im self but neglected in the second stage of his introduction.

But, secondly, the presen t in troduction is prim arily distinguished

Introduction 88

from the A ristotelian by that aspect by which philosophy is d istin ­guished as a p re-em inen t science, or rational question, from the o thers. For in the p resen t in troduction this p re-em inence is a practical p re ­em inence. T h ere is a practically p re-em inen t rational question. I f we call it philosophy then com ponents in the usual prelim inary u n d e r ­standing o f the w ord sophia becom e im portan t which A ristotle neglected: ‘philosophy’ and ‘w isdom ’ as the question abou t the h ighest good, as the question about the m eaning o f life and as a general p rac­tical-theoretical w orld-orientation. We call him a wise m an who can advise us well with respect to ultim ate aims and life as a whole.

In the fundam en ta l practical question we also find again the aspect o f universality which Aristotle em phasized; only now the universality is not u nderstood in term s of spheres o f objects but in term s of the p rac ­tical question which ideally depends on a theoretical penetra tion o f the total concrete w orld-situation. Em pirical sciences would have to en ter into ‘philosophy’ as thus understood; the o rd e r in which they would do so would be d e term ined by the practical aim. I am beginning to speak in the subjunctive because as yet a m ethodologically clarified conception o f such a ‘philosophy’ does not exist, no r am I in a position to presen t such a conception. My practical in troduction the re fo re breaks o ff at the crucial po in t w here th e prelim inary conception o f philosophy obtained from the practical fundam ental question should acquire definite m eth ­odological contours.

Only at the h ighest point o f the fundam ental practical question, in the prelim inary question concern ing the possibility o f practical reason, do we reach m ethodologically fam iliar g round . With the question about the possibility of-practical and theoretical reason, i.e. with the question of how statem ents can be justified , the theory o f veritative being, o r the assertoric sentence-form , is taken up again with particu lar reference to the aspect o f justification. T h e aspect o f justification which was neglected by A ristotle in the form alization-step is in tu rn accessible to a form al them atization and the resulting them atic coincides with the th e ­matic o f ontology ex tended to veritative being in the widest sense (which also em braces statem ents in which w ords like ‘good’ occur). We shall see later tha t one understands an assertoric sentence if and only if one knows its tru th-conditions; and that m eans: knows how it is to be justified. A nticipating this la ter result I can say tha t the explanation o f the m eaning of a sta tem ent-form is identical with the explanation o f how statem ents o f this form are to be justified . T hus the question con­cerning the possibility o f reason finds its answ er in a semantics o f asser­toric sentences. T o be sure, the question concerning the possibility of

A practical conception of philosophy 89

reason does no t directly lead beyond veritative being to a general fo r­mal sem antics. N on-assertoric sentences have no relation to reason. H ow ever, the justification an d necessity o f such an extension does not simply follow from the fact tha t one can only elucidate som eth ing by them atizing the whole genus to which it belongs and by con trasting it with o th e r species o f the sam e genus. R a ther it has already becom e clear th a t we cannot hope to analyse practical sta tem ents w ithout analysing in ten tion-sen tences, im peratives an d optative sentences. T h e w ord ‘go o d ’ can only be explained in this way; thus the u n d e rs ta n d in g o f those assertoric sentences in which th e word ‘go o d ’ occurs is g ro u n d ed on th e u n d ers ta n d in g o f these non-assertoric sen tence-form s. T h u s my practical in troduc tion o f philosophy leads, a t least at its h ighest point, back to the language-analytical conception o f philosophy as this em erged in connection with the ontological realization o f the A risto te­lian prelim inary conception of philosophy. By analogy with the A risto­telian fo rm ulation ‘th e re is a science which studies being as being’ (above p. 21) we can now say: th e re is a form al question w hich we have a p re-em inen t rational m otivation fo r asking: the question concern ing the possibility o f practical reason. T h is question belongs to the m ore general question concern ing the possibility o f reason as such, a ques­tion which is identical with the question concern ing the u n d ers tan d in g o f assertoric sentences. T h is la tter question is posed w ithin the fram e­w ork o f a general form al sem antics, whose fu n d am en ta l question is: w hat is it to u n d e rs ta n d a sentence? A nd this question (cf. above p. 36) coincides with the elucidation o f th e question o f w hat it is to u n d e rs ta n d the m eaning o f a linguistic expression.

T h e p re-em in en t m otivation we have fo r the question concern ing the possibility o f practical reason thus leads us in to the sam e sem antic th e ­matic tha t resu lted w hen we took ontology as o u r starting-poin t. H ow ­ever, it leads in to it via a particu la r sem antic s tru c tu re - th a t o f practical statem ents - which fo r a general analysis o f the sem antic sp h e re canno t stand at the beg inning. T h e practical h ie rarchy based on the p re ­em inen t m otivation does no t co rresp o n d to the theoretical sequence which results from the in n e r connections o f th e them atic field. This is also the reason why it is so difficult to work o u t in m ethodologically clarified fo rm th a t conception of philosophy which rep resen ts the con­crete carry ing th ro u g h o f the fu n d am en ta l practical question. A nd yet from the p o in t o f view o f this practical in troduc tion th a t w ould be the most im p o rtan t philosophical task.

Part TwoA first step: analysis of the predicative sentence

L E C T U R E 8

Preliminary reflections on m ethod and preview of the course of the investigation

My aim in the in tro d u c to ry p a rt o f these lectures was to w ork ou t a question which can be re g a rd e d as a fu n d am en ta l question o f language- analytical philosophy. At th e sam e tim e the language-analytical concep­tion o f philosophy was to be co n fron ted with o th e r philosophical posi­tions and we w ere to exam ine w hether, an d if so how, it can be justified vis-ä-vis these o th e r positions. A nd this involved also discussing the idea of philosophy in general.

T h e fact tha t this has no t yielded a un ita ry conception o f philosophy is no disadvantage. T h e object o f such reflections is to get clear about the d iffe ren t possibilities o f u n d ers ta n d in g som eth ing (in o u r case the idea o f a ‘p re-em in en t science’) and ab o u t how these d iffe ren t possibil­ities are re la ted to one an o th er. W hich o f these one then calls ‘philoso­phy’ is a secondary m atter. Essentially we have becom e acquain ted with th ree ways in which ‘ph ilosophy’ could be u n d ersto o d . Firstly, on the basis o f the discussions o f the first lecture, one could designate ‘p h i­losophy’ all elucidation o f p rio r u n d ers tan d in g , all clarification o f con­cepts o r m eanings. Such enquiry w ould be a ‘p re -em in en t’ enquiry inasm uch as it concerns th e understan d in g -p resu p p o sitio n s o f d irect, non-reflective know ledge an d enquiry .

Secondly, from o u r exam ination o f th e A risto telian in troduc tion th e re em erged a conception o f philosophy as a universal fo rm al science, which is to be u n d ers to o d as form al sem antics.

T h e first o f these two conceptions o f philosophy rep resen ts a vague, bu t indispensable, m ethodological conception o f language-analytical philosophy. By contrast the second conception has clear them atic con­tours and a defin ite fu n d am en ta l question. It rep resen ts, if one holds on to the first, b road conception o f language-analytical philosophy, the g round-d iscip line of language-analytical philosophy. It contains a ques­tion which, vis-ä-vis questions in accordance with the first conception, is

Analysis o f the predicative sentence 94

‘p re -em in en t’; for it concerns the universal presuppositions o f all u n d erstand ing .

T h e conceptions o f philosophy as ontology and as transcendental philosophy have shown them selves to be inadequate approxim ations to the second conception o f philosophy. In so far as transcendental p h i­losophy contains elem ents which po in t beyond this conception these elem ents can them selves only be clarified by linguistic analysis (Lecture 6), thus by m eans of a p ro ced u re in accordance with the first conception o f philosophy.

Finally, a third conception o f philosophy represen ts the question I have called the fundam en ta l practical question. Historically, since Soc­rates and Plato, this question concern ing the good - W hat is to be done? How should we live? - has influenced the conception o f philosophy. It is possible to see the w hole history o f E uropean philosophy as a debate between this practical conception o f philosophy and the A ristotelian th eo re tica l-fo rm a l conception. T h is practical fundam en tal question is also a ‘p re -em in en t’ question. B ut it is p re-em inen t in ano ther sense, nam ely, in the sense th a t it is the only question fo r which there is an im m ediate an d absolute rational m otivation. Now which sense o f ‘p re ­em in en t’ shou ld we settle for: tha t which leads to the second conception o f philosophy or tha t which leads to the th ird conception? If by this question one m eans: which sort o f enquiry should be called ‘philoso­p h y ’? then it is undecidable and also un im portan t. It is a m atter o f ind ifference to which so rt o f enqu iry one attaches the label ‘philoso­p h y ’. B ut if the question m eans: fo r what sort o f enquiry should one rationally decide, which question o u g h t one to ask? then it follows an a­lytically tha t it can only be the th ird conception o f philosophy.

I was no t in a position to give clear contours to this kind o f philoso­phy. O f course, it does no t requ ire m uch reflection to see tha t the ques­tion: what o u g h t to be done? is no t a sem antic question. T h e highest conception o f philosophy, highest in the sense tha t it is the only one th a t is p re-em in en t in term s o f its m otivation, is thus no t a language- analytical conception o f philosophy.

Now we have seen th a t this th ird conception o f philosophy also refers back to ou r second conception, in as m uch as one can only clarify the prelim inary question concern ing the possibility o f practical reason - the possibility o f the justification o f practical statem ents - by m eans o f an analysis o f the fo rm o f these sentences (p. 8 8 -9 ).1 However, it would be sophistical - though very attractive as a way o f easing the theoretical ph ilo so p h er’s conscience - to in fer from this connection tha t the uncon­ditional rational m otivation the re is fo r the th ird conception of philos­

Preliminary reflections on method 95

ophy carries over to the second conception. It is tru e that the question o f the possibility o f asking the practical question is m ethodologically p rio r to the practical question; but it does not follow from this that it should precede it practically. For the fundam ental practical question does not adm it o f postponem ent. O ne would be simply evading the practical question were one to try to persuade oneself that one cannot ask it w ithout having first clarified the m ethodologically necessary p re­lim inary questions. M oreover, what guaran tee is th e re that one will not rem ain stuck in the form al prelim inary questions? In these lectures at any ra te I will no t even get as far as a semantics o f practical statem ents;I will thus rem ain stuck in a question which is prelim inary to the prelim ­inary question.

A fter this w arning I can begin the m ain part o f the lectures. In the first lecture I announced th a t I would not be providing any results or surveys. Instead 1 would like to pursue with you a fundam en tal ques­tion o f linguistic analysis. W e now have a rough idea o f the outlines of the fundam ental discipline o f linguistic analysis. How can we begin to w ork this them atic field?

T h e following p rocedu re m ight suggest itself. Since it has becom e clear tha t what is unsatisfactory about ontology is the restriction to one or at best two sem antic form s - singular term s an d predicates - a gen­uine reflection on the sentence-form s would have to m ake su re tha t the totality o f the relevant form al distinctions were included. It would then be possible to carry out the enquiry into the u nderstand ing o f sentences at the requ ired level o f systematic generality.

Such a systematic p rocedu re is how ever not possible if one does not yet possess the categorial m eans o f analysis. W e do not yet know how one can enqu ire into the m eaning o f a linguistic expression, let alone into w hat it is to und erstan d a sem antic class o f expressions, or the com bination o f expressions o f d iffe ren t classes. As yet we do no t even have a clear conception o f sem antic fo rm , hence we do not yet have a p ro ced u re for semantically classifying sentences, an d thus for arriv ing at the totality o f form al distinctions ju s t dem anded . T h e few distinc­tions I have m ade in the in troducto ry reflections w ere purely provi­sional ones which could simply claim a certain plausibility; they m erely served to give us a view o f the field of investigation. I only distinguished the sem antic class o f singular term s, which was m arked out by the tra ­ditional concept o f an object, and the com plem entary class o f predicates (general terms). A nd even in the presen tation o f these classes I had to anticipate unclarified notions such as tha t o f a m ode of em ploym ent. Equally tentative w ere the distinctions between propositional con ten t

Analysis o f the predicative sentence 96

and m ode and , in particu lar, the distinction o f d iffe ren t m odes. T h e dem and to ex ten d the analysis from objects to the u n d erstan d in g o f sentences can th e re fo re no t be taken to m ean th a t in sentences we have a field o f investigation which we simply have to en ter. I f we wish to reg a rd form al sem antics as the successor discipline to ontology then we m ust first w ork ou t a conceptuality, as fundam en ta l as the ontological, with which this field o f investigation can be m ade accessible.

O f course a conceptuality which is adequate to a subject-m atter can only be achieved in the analysis o f this sub ject-m atter itself. How ever, th e re results a com pletely d iffe ren t kind o f p ro ced u re , d ep en d in g on w hether (as in a particu la r science, in ou r case linguistic science) one em barks directly on the theoretical w orking o f a field, using an existing conceptuality and leaving the developm ent o f a m ore adequate concep­tuality to scientific p rogress, o r (as in philosophical research into fo u n ­dations) one approaches the field o f investigation with the prim ary in ten tion o f w ork ing ou t an adequate conceptuality. In the first case the investigations o f the concrete m aterial a re carried ou t in such a way tha t the available trad itional conceptuality (which is know n to be inade­quate), and hence any conceptuality, is as little in evidence as possible. In the second case we m ust carry ou t such analyses as a re aim ed at p u ttin g the adequacy o f the trad itional conceptuality to the test and, if it proves to be inadequate, developing a new conceptuality.

In concrete term s tha t m eans: the fundam en tal concept o f previous philosophy is th a t o f beings o r objects. T h e fundam en ta l concept o f m o d ern philosophy - th a t o f consciousness - is also und ersto o d in the sense o f a consciousness o f objects. T his leads to every new philosophi­cal o r scientific them atic being construed in term s o f this concept. T h e consequence fo r sem antics is th a t the answer given to the question: w hat is it to u n d ers tan d a linguistic expression? is th a t w hat one u n d e r­stands when one u n d erstan d s an expression is the meaning o f the expression, this being conceived as follows: the expression stands fo r the m ean ing which the person who un d erstan d s the expression represents to him self, and w hat one rep resen ts to oneself is an object. T hus although stand ing for objects is pecu liar to one specific class o f expressions (sin­gu la r term s), and although even in the ir case m eaning and object should be d istinguished, the fact th a t the re is no o ther conceptuality available in th e trad ition leads one to construe even the m eaning o f these expressions and th a t o f every expression as an object. W here the in app rop ria teness of this m ode o f conception is felt in linguistic research , bu t a fundam entally new one is not available, such an objec­tionable concept as tha t o f an ‘object’ is avoided by using a m ore neu tra l

Preliminary reflections on method 97

term inology: one speaks, fo r exam ple, o f ‘co n ten ts’, o r one sim ply con­fines oneself to saying th a t every expression stands fo r som ething, leav­ing it com pletely o p en how this ‘so m eth in g ’ is to be u n d ersto o d .

We shall have to ad o p t precisely the opposite p ro ced u re . W e m ust avoid concepts w hose only advan tage is inde term ina teness. I f instead o f saying th a t the expression stands fo r an object one says th a t it stands fo r a con ten t, o r sim ply fo r som eth ing , then one has m erely lost clarity and gained noth ing . Crucial fo r us m ust be th e question o f w h e th e r one can say th a t the expression stands jor som eth ing a t all. I f this tu rn s out to be false then we m ust try to w ork o u t a new conception which is as fu n d am en ta l as th e objectual conception .

T h e first, critical—destructive step - th e testing o f th e adequacy o f the trad itional object-orien ted ap p ro ach in its app lication to the u n d e r ­s tand ing o f linguistic expressions - seem s relatively easy to u n d ers tan d . I t is perfectly intelligible th a t a conceptuality th a t was developed fo r a specific and m ore narrow them atic shou ld prove inadequa te w hen applied to an o th e r an d b ro ad e r them atic. B u t how should one conceive the second , constructive step? I f we p u t the trad itional perspective ou t o f action and e n te r o u r them atic field as it were w ithou t any perspec­tive, th e n far from a p p e a rin g in a new ligh t it will a p p e a r in no ligh t at all. A new conceptuality can never be a tta ined d irectly from an u n co n ­ceptualized them atic field, b u t only in reflection on the w eaknesses or limits o f a previous conceptuality . T h u s the critique o f the object- o rien ta ted position in its application to th e question o f m ean ing is by no m eans in ten d ed to have the m erely negative significance o f clearing away an inadequa te conceptuality; ra th e r it constitu tes the ind ispens­able first step in the positive task o f w ork ing o u t a m ore ad eq u ate con­ceptuality. O f course no t every kind o f critique o f a given conceptuality is constructive; how ever, th e re is no p rogress in th e field o f fo u n d a ­tional research which does no t arise fro m a critique o f the previous conceptuality. T h is connection betw een destructive and constructive conceptual w ork is exem plified by a s tru c tu re which in the course o f these lectures will repeated ly prove to be fu n d am en ta l to a new step: if an ex isting conceptuality tu rn s o u t to be in ad eq u a te when app lied to a w ider them atic field, one is obliged to reflect on p resuppositions o f this conceptuality which did n o t have to be reflected u p o n w ithin th e p re ­vious conception. O n e thus arrives at a m ore general conceptuality , one which underlies the prev ious conceptuality b u t w hich also perm its other conceptual in te rp re ta tio n s. Such new conceptual in te rp re ta tio n s of course never arise in a m erely abstract reflection on foundations, bu t only u n d e r the p ressu re o f new data o f the them atic field which canno t

Analysis o f the predicative sentence 98

be explained in term s of the previous conceptuality. We shall see that this critica l-constructive s truc tu re is no t only im portan t fo r the w orking out o f a specifically language-analytical conceptuality, as opposed to the trad itional ob ject-orien tated conceptuality, bu t that it also repeats itself within the language-analytical position.

T h e debate with trad itional philosophical conceptions of trad itional philosophy was thus by no m eans merely a m atter o f in troducing you to the them atic o f language-analytical philosophy. In so far as language- analytical philosophy, like all previous form s of fundam ental philoso­phy, is conceptual research in to foundations and is not m erely con­cerned with p articu la r problem s (many analytical philosophers o f course a re only concerned with particu lar problem s; for us how ever this is irre levant), it is carried o u t by m eans o f a critique of trad itional philosophy. In the in troducto ry p a rt o f these lectures the debate with trad itional conceptions concerned only the dem arcation o f the them atic field. Now, how ever, the debate will concern the conceptuality.

T h e aims o f this m ain p art o f the lectures are, (1) to d em onstra te that the m eaning o f sentences cannot be construed objectually (Lectures 9 - 10), (2) to w ork ou t a new conceptuality , in term s o f which we can ex­plain, if not the u n d e rs ta n d in g o f all sentences, then at least the u n d e r­standing o f all assertoric sentences (Lectures 11-19, 27) and (3) to show tha t the trad itional fu n d am en ta l concept o f an object can itself only be understood on the basis o f this new conceptuality (Lectures 20-27).

I f it is correct tha t we can only achieve a new conceptuality with which the u n d ers ta n d in g o f linguistic expressions can be explained by explic­itly taking leave o f the o b jec t-o rien ta ted conception, then as regards the m ethod o f o u r enqu iry it w ould be advisable, firstly, initially to restric t the investigation to those sem antic structures which fall w ithin the purview o f trad itional ph ilosophy and, secondly, to start with the criticism o f a theory o f m ean ing which though m odern is still trad itio n ­ally o rien ta ted an d which a ttem pts to m ake as m uch as possible o f the object-orien tated conceptuality in semantics.

R egard ing the first point, it would seem reasonable to ignore com ­pletely fo r the tim e being non-assertoric sentences; for these sentences do no t fall w ithin the scope o f the traditional conceptuality at all. M ore­over it w ould ap p e ar sensible to concen tra te above all on the predicative sen tence-form . A lthough the trad ition was also fam iliar with o th e r form s o f assertoric sentence, the trad itional sem antic conception m akes sense, if a t all, only in connection with predicative sentences (for o th e r form s cf. L ectures 17 and 18); an d the struc tu re of the predicative sen ­

tence was widely reg a rd ed as the universal structu re of the assertoric sentence, o r ju d g m en t, as such.2 It is no accident that in the debate with ontology the two sem antic classes o f singular and general term s, thus the two constituents o f the predicative sentence, em erged relatively easily. I f o n e’s o rien tation is tow ards objects then the predicative sen­tence seems the one m ost easy to un d erstan d , for besides an object- designation it contains only one o th er com ponent. M oreover it also seems reasonable to assum e that the predicative statem ent in which an individual is classified is the most elem entary form of assertoric speech and the one which underlies all o th e r sem antic structures.

Husserl is a good represen tative o f a trad itional position with which to begin. O n the one han d , H usserl was already aw are o f the problem o f the m eaning o f linguistic expressions and , u n d e r the influence of Frege, tried to avoid constru ing the m eaning o f an expression as its object. O n the o ther h an d , he, unlike Frege, app roached th e problem from a philosophical position which did not exclude the subjective (psychological-epistem ological) aspect; and it was a decidedly object- orien tated position. F u rth e rm o re , in his theory o f categorial synthesis H usserl m ade an effort, un ique in the pre-analytical trad ition , to solve by m eans of the trad itional conceptuality the problem of how the m ean­ing of a com plex expression (in particu lar tha t o f a sentence) arises ou t o f the m eaning o f its com ponents. H ence, in exam ining the possibility o f explaining the u n d erstan d in g o f linguistic expressions and , in partic­u lar, their com bination, in term s o f an object-orientated conceptuality, the difficulties and blind alleys o f H usserl’s sem antics can be regarded as exem plary.

T h e collapse o f the traditional conception, according to which lin­guistic expressions are always used to stand fo r som ething, will lead us to them atize the phenom enon , p resupposed bu t no t reflected upon in the traditional conception, o f mode o f employment, and to look ou t for a new conception. In analytical philosophy th e re are two approaches to the explanation of the u n d ers tan d in g of linguistic expressions and, in particular, th a t o f assertoric sentences. T h e one approach , which stems from the later W ittgenstein, says: to und erstan d a sentence is to know how it is to be used. T h e o th e r approach , which extends from Frege via the early W ittgenstein, C arnap and Tarski to Davidson and others, says: to un d erstan d an assertoric sentence is to know u n d e r w hat conditions it is tru e o r false. We shall see tha t a satisfactory analysis o f the m eaning o f assertoric and , in particu lar, o f predicative sentences an d the ir com ­ponents, can only be achieved by com bining these approaches in a ce r­

Preliminary reflections on method 99

Analysis o f the predicative sentence 100

tain way. Even now it may ap p ear plausible tha t the concept o f tru th occupies a place in the conceptuality o f a theory o f the un d erstan d in g o f assertoric sentences which is com parably fundam en tal to th a t occu­pied by the concept o f an object in the conceptuality o f ontology and transcenden ta l philosophy. O f course the concept o f tru th also occurs in trad itional philosophy. M oreover the concept o f an object is also reta ined in analytical philosophy. H ow ever, w hereas in the trad ition the tru th -re la tion was understood in term s o f reference to objects, I shall try to show tha t reference to objects is essentially an elem ent in the tru th -re la tion , ju s t as the function of singular term s is only to be u n d e r ­stood in term s o f the ir role in a sentence. B ut we canno t be con ten t to trea t the concept o f tru th as som ething simply given. O f course the w ord ‘tru e ’ is no t definable. It will be one of ou r first tasks a fte r the dem onstration o f the collapse o f the ob ject-orientated approach to explain how the m eaning o f expressions and sem antic classes o f ex p res­sions can be investigated philosophically. It will th en em erge th a t the explanation o f the w ord ‘tru e ’ coincides with the explanation o f the assertoric, and, ultim ately, the predicative sen tence-form (Lectures 18, 27).

In these lectures I shall no t get beyond this first step in the w orking out o f a basic analytical conceptuality. W ith the dem onstra tion tha t the reference to objects is an elem ent in the u n d ers tan d in g of a species o f sentences the step in the conceptual research into foundations to which the debate with the pre-analytical, object-orientated trad ition is essen­tial is concluded. T h e nex t fundam en ta l step, which concerns the extension of the problem atic from assertoric to non-assertoric sen­tences, I shall m erely indicate in the last lecture. T his step is reg ard ed by m any analytical philosophers today as the nex t to be taken, their a tten tion having h ith e rto been largely confined to assertoric sentences. It seems clear th a t one cannot transfe r the notion o f tru th -cond itions - o r at least not autom atically - to sentences which do not characteristi­cally contain a truth-claim . So again we are faced with an inadequate concept, this tim e tha t of the tru th -re la tion . A nd one can ask w hether, by reflecting on its foundations, one can so ex tend this concept as to yield a concept which achieves for all sentences w hat the concept o f tru th achieved fo r assertoric sentences. So again the next step in the question o f foundations w ould not consist simply in a ttend ing to h ith ­erto neglected sem antic form s. It w ould also involve a sim ultaneous debate with the previously available fundam en tal conceptuality; except that this would no longer be a trad itional conceptuality, bu t already an analytical one.

Preliminary reflections on method 101

O ne m ust o f course leave open the question of w hether the concept o f tru th (always assum ing tha t it does p rove to be fu n d am en ta l to the analysis o f assertoric sentences) is really su ited to serve as the basis for the sough t-afte r b ro ad e r conceptuality. B u t if this app roach fails then the question arises: from w hat o th e r conceptuality can one start? T h e idea tha t th e re is no need to start from a previous, m ore restric ted con­ceptuality , and that a new conceptuality will autom atically arise from the co n fron ta tion with the new data , is naive. T his is the reason why, if we are seeking no t som e kind o f inven to ry b u t ra th e r conceptual clari­fication, we canno t begin im m ediately a com plete form al sem antics, bu t m ust p en e tra te the them atic field step by step. T h e sequence o f steps here is p rescribed n o t by factors o f the them atic field bu t by the avail­able conceptuality .

T h e s tru c tu re o f all sentences which I b ro u g h t ou t in the in troduc tion -M * p * - could o f course give rise to doub ts ab o u t this p ro ced u re . I had po in ted ou t tha t all fu rth e r fo rm al su b stru c tu rin g belongs to the p ro p ­ositional conten t. W e m ust th e re fo re expect th a t th e re is also a p re d i­cative fo rm , etc., in th e case o f non-asserto ric sentences. W ould it no t then be m uch m ore satisfactory from a system atic p o in t o f view if instead o f first restric ting ourselves to assertoric sentences we w ere to proceed like Jo h n Searle in his book Speech Acts, i.e. m a first section investigate the sem antics o f the d iffe re n t m odes and , in a second sec­tion, th e sem antics o f propositional s tru c tu res in abstraction from the m odes? T h is w ould indeed be m o re satisfactory from a system atic po in t o f view, bu t I do no t see the conceptual m eans with which we can carry ou t such an u n d ertak in g . In p a rticu la r I m ust po in t to two difficulties: firstly, bo th m ode an d propositional co n ten t are abstract elem ents of sentences the significance o f which we can n o t h o p e to clearly grasp righ t at the beg inn ing o f o u r u n d e rta k in g ; and connected with this is the fact th a t in developing the conceptuality necessary fo r the analysis o f these com ponen ts th e re is, in n e ith e r case, a trad itional conceptuality which we can s ta rt ou t from . Secondly, as one is dealing with two m om ents, no t with two constituen ts, one can n o t so conceive the s tru c­tu re M*^>* th a t the conceptuality req u ired fo r the analysis o f the one elem ent could be in d e p e n d e n t o f the conceptuality req u ired fo r the analysis o f the o th e r elem ent. B ut in th a t case it w ould seem m ost unlikely th a t an analysis first o f th e m odes an d then o f th e propositional con ten t could lead to a satisfactory resu lt. Since one canno t do every­th ing at once th e re is no a lternative b u t to p re p a re th e general sem an­tics o f th e form s o f p ropositiona l co n ten t by first analysing som e o f these form s in connection with one m ode. A nd the m ode m ost suitable

Analysis o f the predicative sentence 102

for this purpose is clearly the traditionally most fam iliar one, viz. the assertoric m ode. O nly tha t subsequent abstraction-step to which I have already re fe rre d which would have to lead beyond the conceptuality tha t p roved successful fo r the analysis o f assertoric sentences would make possible both an u n d ers tan d in g o f the o ther m odes and a concep­tuality in which we can u n d ers ta n d the form s of propositional conten t independen tly o f the individual m odes. We thus find ourselves once m ore re fe rre d to a step-by-step p rocedure . H ow ever convincing Searle’s p rog ram m e may be it is a p rog ram m e for the fu tu re - as can be seen from the conceptual un fru itfu lness o f Searle’s own carrying out o f this p rog ram m e (which I will only be exam ining later. See p. 398). O ne m ust recognize that, as in any research , so too in philosophical research into foundations, the sequence o f steps in the analysis does not co rrespond to the sequence in the subsequent systematic exposition.

B efore we begin o u r u n d ertak in g with the exam ination o f H usserl’s analysis o f the predicative sen tence-form I would like to specify the questions which m ust gu ide us. We will be dealing for the first tim e with questions o f m ean ing and we should fram e such questions in a way which involves the fewest possible assum ptions. I f one asks: ‘W hat is the m eaning o f the expression “re d ”?’ one clearly m eans the same as when one asks: ‘How is the expression “re d ” to be u n derstood?’ We cannot speak o f the m ean ing o f an expression w ithout re fe rrin g to an u n d e r­standing; bu t we can perfectly well speak o f the u n d erstan d in g o f an expression w ithout re fe rrin g to a m eaning. It would th e re fo re seem to involve few er p resuppositions if one w ere to fo rm ulate a m eaning- question as follows: ‘How is the expression “A ” to be u nderstood?’ ra th e r than as follows: ‘W hat is the m eaning o f the expression “A ”?’ An object-orientated theory o f m ean ing will o f course p re fe r the second version because it is analogously fo rm ulated to the question: ‘W hat is the object fo r which the nam e ‘W ” stands?’ B ut the object-orientated philosopher can also have no objection to the first version. In fo rm u la t­ing the question we should no t already p red e te rm in e a particu lar answer. So although we should no t puristically avoid speaking o f the m eaning o f an expression (which is a perfectly natu ra l way o f speaking), we should no t be m isled into asking questions which are only suggested by this m ode o f speech.

H ow ever we are no t dealing with so sim ple a sem antic question as tha t concern ing the m ean ing o f the word ‘re d ’. T h e question concern ­ing the u n d ers ta n d in g o f the predicative sentence-form differs from such a question in two respects. Firstly, it concerns the m eaning o f an expression (such as ‘T h e ball is r e d ’) which is com posed o f two elem ents

each of which already has a m eaning bu t which when com bined yield a un itary m eaning. As we are in terested in the question o f how the un d erstan d in g of the sentence-w hole depends on the u n d e rstan d in g o f the two sentence-parts we m ust now, secondly, abstract from the partic­u lar m eaning of this sentence o r its com ponents and ask about the un d ers tan d in g o f th e sentence-form ‘Fa o r ‘Fab’ (cf. p. 26).

I f one asks how a whole is m ade up of parts, th e re is clearly a relation of in te rdependence betw een a correct understan d in g o f the parts (as parts o f this whole) o f the whole and of the com bination. We m ust the re fo re expect an analysis o f the predicative sentence-form to p ro ­vide an answ er to the following fo u r questions:(1) how is a singular te rm understood?(2) how is a p red icate understood?(3) how is the com bination of a singular term with a p red icate u n d e r­stood?(4) how is a (predicative) assertoric sentence understood?

It is no t so easy to form ulate these fo rm al-sem antic questions in term s o f the w ord ‘m ean ing’. It may seem plausible to use the ‘as’ fo r­m ula of ontology and ask, e.g., ‘In w hat does the m eaning o f a predicate as such consist?’ But the question ‘In what does the m eaning consist?’ has no clear m eaning; it could tu rn ou t to be a pseudo-question. In question (3), finally, the reference to a m eaning no longer seems ap p ro ­priate , unless tha t is one assumes tha t to the com bination too th e re somehow corresponds ‘a m eaning’.

Because o f these difficulties I have im m ediately fo rm ulated the four questions w ithout using a word like ‘m eaning’ o r ‘sense’ (like H usserl I reg a rd these two expressions as synonyms). B u t even if we confine o u r­selves to the w ord ‘u n d ers tan d in g ’ it is far from clear how such fo rm a l- sem antic questions are to be form ulated . O n an earlier occasion I used the form ulation ‘W hat is it to u n d erstan d a pred icate?’ B ut already at the time I rem arked tha t this form ulation can only be reg a rd ed as p ro ­visional (p. 35 f). We could only arrive at a clear form ulation o f o u r questions if we were clear about the type of answ er we expect. We thus find ourselves in the awkward, bu t in philosophy not unusual, position o f not know ing clearly w hat it is we are really asking about. It is im p o r­tan t in such cases tha t one at least knows that one does no t know.

We will thus have to bear in m ind tha t the questions themselves also becom e clearer in the course of being answ ered. In particu lar we m ust see to it tha t the form al sem antic questions (e.g. concerning the u n d e r­standing o f predicates) reta in the necessary connections with the con­crete sem antic questions (e.g. about the m eaning o f ‘re d ’) from which

Preliminary reflections on method 103

Analysis o f the predicative sentence 104

they resu lt th rough form alization. We m ust th e re fo re find a fo rm for concrete sem antic questions which can be tran sfe rred to the form alized questions. O ne such possibility (which I m ention by way o f anticipation) is the following. O ne can also de-substantify the question concern ing ‘the u n d e rs ta n d in g ’ of, e.g., the expression ‘re d ’ and ask: ‘U n d er w hat conditions do we say o f som eone that he und erstan d s the expression “re d ”?’ B ut to speak o f conditions (or the ‘conditions o f the possibility’) also involves substantivization. Clearly we canno t express what is m eant by m eans of a substantive, bu t only by m eans o f a sentence, nam ely, by saying: the answ er we are aim ing at to the question concern ing the u n d ers tan d in g o f the expression ‘re d ’ m ust have the form : ‘som eone u n d erstan d s the expression “re d ” if and only if . . .’ T his fo rm ulation can now easily be tran sfe rred to the form alized sem antic question. W hen we ask abou t the u n d erstan d in g o f predicates we are looking for an answ er o f the form : ‘Som eone un d erstan d s a pred icate “F” if and only if . . .’ Now it seems plausible to conceive the com pletion o f this if- sentence in such a way tha t both in the concrete question and in the form alized sem antic question the m ode o f em ploym ent o f the exp res­sion is being re fe rre d to. As this seems to involve a distinct p rejud ice against the object-orientated position I w anted at p resen t sim ply to m ention this perspective, no t to adop t it. I shall com e back to this ques­tion o f the form in which fo rm al-sem antic questions are to be posed at the beg inn ing o f the positive analytical reflections, in connection with a princip le o f W ittgenstein (Lectures 11 and 12).

Let us now re tu rn to the fou r questions re fe rred to above. Clearly question (3) (how is the com bination of the singular te rm with the p re d ­icate understood?) is the key-question. F rom its clarification the re would im m ediately have to follow the answ er to the fou rth question: how is the predicative sentence understood? We can also expect tha t an adequate answ er to the first two questions is only possible if the th ird question is already included; for only if it can be shown tha t in the u n d e rstan d in g o f each o f the two sen tence-com ponents the u n d e r ­standing o f its com bination with the o th e r com ponen t is already included can one expect a un ita ry m eaning of the whole sentence and not a m ere agglom erate o f two m eanings to result.

It w ould seem th e re fo re th a t we should begin with question (3), o r at least include it from the very beginning. H ow ever, if we want to take as ou r sta rting-po in t the object-orientated , and specifically the H usserlian , theory then we m ust first observe the sequence which the basic approach of this theory dictates. It is characteristic o f the object-orien­

Preliminary reflections on method 105

ta ted app roach tha t it does not take as its sta rting -po in t the com position o f the sentence, bu t is prim arily o rien ta ted tow ards th a t sentence- co m p o n en t which stands for an object, viz. th e singular term . It then construes the whole sen tence as also stand ing fo r an object. In the in te r ­p re ta tio n o f H usserl we will th e re fo re have to s ta rt with question (1), and from th e re m ove tow ards a p re lim inary decision ab o u t question (4). Only a fte r this will we be able to deal with questions (2) and (3) toge ther. H usse rl’s ap p ro ach , an d the ob ject-o rien tated approach in general, fo u n d ers on th e question o f how pred icates a re u nderstood . In pred icates th e re fo re we will find a sta rting -po in t fo r the develop­m en t o f a new, specifically language-analytical, conception. T h u s in w orking o u t the language-analytical conception I will proceed in reverse o rd er. I shall begin with question (2) - pred icates - (Lectures11 and 12) and then proceed to a p re lim inary investigation o f question(4) - the u n d e rs ta n d in g o f assertoric sentences as such - (Lectures 13- 16). Only at the end (Lectures 20-27) will I develop the language- analytical conception o f singular term s. It will em erge tha t singular term s in fact p resen t sem antic analysis with m uch g rea te r difficulties than do predicates, an d tha t the ob ject-orien tated analysis explains the sem antic category o f singular term s, fo r whose exp lanation it was seem ­ingly p red estin ed , in a wholly in ad eq u a te way. In th e language- analytical analysis s ingu lar te rm and p red icate are essentially u n d e r ­stood as sen tence-parts; the answ er to questions (3) an d (4) will th e re ­fo re follow autom atically from th e clarification o f the first two ques­tions.

In conclusion I w ant to deal with an objection tha t has probably occurred to som e o f you d u rin g these last reflections. I am beginn ing with the question o f th e m eaning o f a com posite expression. B ut would it not be m ore ap p ro p ria te from a system atic p o in t o f view to first explain how a non-com posite expression is to be u n d ersto o d ? For only then would we have a basis fo r the question o f how a com posite ex p re s­sion arises ou t o f the m eanings o f its co m p o n en t expressions.

H ow ever, if it is co rrec t that the prim ary sem antic units are sentences (and hence com posite expressions) then th e question concern ing the sem antics o f elem en tary com posite expressions is the fundam en ta l sem antic question. B ut in tha t case one can n o t hope to be able to clarify the essence o f the m ean ing o f expressions in general in d ep en d en tly of answ ering this question. We m ust th e re fo re also expect tha t all n o n ­com posite expressions a re essentially com ponen t-expressions and tha t th e ir m ean ing can only be u n d ersto o d in te rm s o f the sentence-w hole.

Analysis o f the predicative sentence 106

If this is correct then any sem antic theory is m istaken which believes it can first say som eth ing about the m eaning o f sim ple expressions and only then move on to the question of how the m eaning of a com posite expression arises ou t o f the m ean ing of its com ponen t expressions.

L E C T U R E 9

Husserl’s theory of m eaning1

H usserl develops the foundations o f his theory o f m eaning in his Logical Investigations, particularly in Investigation I, which is en titled ‘Expres­sion and M eaning’. T h e in troductory paragraphs are devoted to distin­guishing ‘m ean ingfu l’ signs - linguistic expressions - from indicative signs. T h e concepts essential to his theory o f the m eaning o f linguistic expressions are in troduced by H usserl in §§9-14. T h e first, fu n d am en ­tal step is taken in §9: if an expression is not ju s t ‘a m ere w ord-sound’ bu t a sign and , m oreover, a sign o f a specific kind, then this is due to the fact tha t it can be ‘in te rp re ted ’ as som ething which has a m eaning. T h e m ere p a tte rn o f sounds or m arks on paper does not have a m ean­ing in itse lf; ra th e r the m eaning is ‘co n fe rred ’ up o n it by its being in te r­p re ted in a particu lar way.

T h is first step in H usserl’s investigation seems to me to be unobjec­tionable, though not obvious. By taking it H usserl placed his analyses on a deeper though m ore hazardous basis than Frege: a satisfactory theory o f m eaning canno t confine itself to talking abstractly about m eanings; it m ust also take into account the psychological o r a n th ro ­pological factor o f the sign-user. M eanings do no t exist in a Platonic heaven; they are m eanings of signs. A nd they are m eanings o f signs only in virtue o f the fact tha t certain sensible form s are used (‘in te r­p re te d ’) as signs.

I f this is so, then it is fundam en ta l to a satisfactory theory o f m eaning tha t one correctly characterize the m ode o f behaviour, o r conscious­ness, in which an expression is in te rp re ted as m eaningful. In the p re ­vious lecture I drew atten tion to the fact that one only speaks o f m ean­ings o f expressions in connection with an u n d erstan d in g o f these expressions. O ne would the re fo re have expected H usserl to re fe r to tha t which ‘confers’ m eaning on the expression as understanding, so th a t

Analysis o f the predicative sentence 108

the further question would then have to be: What is it to understand an expression?

From the outset, how ever, H usserl speaks, as though this were obvious, o f m ean ing -con fe rring acts. A nd in H usserl ‘act’ is a technical te rm fo r ‘in ten tional experience’. As indicated in L ecture 6, by an in ten ­tional experience H usserl m eans a m ode o f consciousness o f an object. T h e concept o f intentionality , which is no t explicitly developed until Investigation V - in an analysis which, significantly, is no longer o rien ta ted tow ards the m eanings o f linguistic expressions - is already p resu p p o sed as th e only possible basic concept o f the theory o f con­sciousness in the sem antic Investigation I. ‘In virtue o f these acts [the m ean ing -con ferring acts] the expression is m ore than m erely a word- sound. I t m eans som ething, an d by m eaning it it refers to an object’(§9).

You will notice how, in an alm ost parad igm atic m an n er, th e problem o f m ean ing is h e re overlain with a trad itional, object-orientated concep­tuality. W hat it m eans to u n d ers tan d a linguistic expression is a question which is no t asked. I t is taken fo r g ran ted that it is a m atte r o f in te n ­tional consciousness, consciousness d irec ted to an object. F rom the o u t­set H usserl approaches the problem o f m eaning with this concept o f consciousness. O ne m ight have expected , as the logical consequence of this approach , th a t the m eaning o f an expression w ould be sim ply iden ­tified with the object to which the m ean ing-conferring act is directed . H ow ever, H usserl does no t m ake such an identification. H usserl is so rew ard ing as a critical po in t o f d e p a r tu re because, on the one h an d , his app roach is, explicitly and as a result o f philosophical reflection, an object-orien tated one, and because, on th e o ther han d , he nonetheless recognized tha t the m eanings o f expressions canno t simply be con­s trued as objects. So by exam ining H usserl’s theory o f m eaning one can investigate the question o f how far a consistently object-orientated app roach can as it were be stretched beyond itse lf; thus how far, if one starts o u t from the assum ption tha t linguistic expressions are used to rep rese n t objects, one can nonetheless take account o f the fact th a t the u n d ers ta n d in g o f expressions is no t identical with th e rep resen ta tio n of objects.

H usserl took account o f this fact in two respects, and in so doing ind icated d iffe ren t possible ways in which the m eanings o f linguistic expressions can be inco rpo ra ted into a consciousness o f objects w ithout them selves having to be reg a rd e d as objects.

Firstly, he acknow ledged tha t the re a re expressions which, though they have a m eaning , do no t in any way stand fo r an object: the so-

Husserl’s theory of meaning 109

called syncategorem atic expressions. T h is concept stem s fro m the sem antics o f trad itional logic.2 C ategorem atic expressions were those which can occur as terms, i.e. in th e position o f subject o r p red icate in the p ropositions o f a syllogism. T h e no tion o f singu lar and general term s is o f th e sam e orig in . Expressions which canno t function as term s w ere re g a rd e d as m ere auxiliary words. T h e ob ject-o rien ta ted app roach also lies b eh ind this division o f expressions in to categorem atic and syncategorem atic. In the A risto telian ontology, every expression which falls u n d e r one o f the ‘categories’ (every kategorema), i.e. every expression which can function as a s ingu lar o r general te rm , stands fo r som ething. T h ese expressions which stand fo r som eth ing have an in d e ­p en d e n t m eaning . In c lu d ed am ong them w ere also w hole assertoric sentences. A ccording to this conception , o th e r expressions have a m ean ing only in connection with categorem atic expressions and w ere, fo r this reason , called ‘syncategorem atic’.

H usserl takes over the division in to ca tegorem atic an d syncategore­m atic expressions (Investigation IV §4 ff.). W e will see la te r how he fits syncategorem atic expressions into his ob ject-o rien ta ted app roach . In i­tially, in Investigation I, the d istinction is igno red . H ere H usserl seem s, w hen speak ing of expressions in a com pletely general way, to be th in k ­ing only o f ca tegorem atic expressions. B ut even in th e case o f catego­rem atic expressions, in o th e r w ords, expressions which stand for som e­th ing, H usserl does n o t th ink th a t the object fo r which an expression stands is th e m ean ing o f tha t expression : ‘Every [!] expression . . . has no t only its m eaning , b u t also refers to certa in objects . . . B ut the object never coincides with the m ean in g ’ (Investigation I §12).

In the no tion that, in the case o f all (categorem atic) expressions, one m ust d istinguish betw een the object an d the m ean ing o f the expression, H usserl is invoking a distinction m ade som e years ea rlie r by F rege, in his essay ‘O n Sense a n d R efe rence’.3 In this essay F rege s ta rted o u t from those expressions he called p ro p e r nam es an d w hich m ore o r less c o rresp o n d to the singu lar term s o f the o ld e r trad ition . A nd H usserl too states: ‘N am es o ffe r the clearest exam ples of th e separation o f m ean ing an d referen ce to an ob ject’ (Investigation I 12). A lthough H usserl’s term inology is no t wholly unam biguous, h e m eans by nam es principally expressions which ‘can p e rfo rm the sim ple sub ject-function in a s ta te m en t’ (Investigation V §34). W hat we have h e re then is H u s­serl’s sem antics o f singu lar term s. W e are th e re fo re faced with the first o f the fo u r questions which I singled o u t a t the en d o f the prev ious lecture as being im p o rta n t fo r th e clarification o f th e p red icative sen ­tence.

Analysis o f the predicative sentence 110

How does H usserl answ er the question about the m eaning o f singular term s? Every such expression, he says, designates an object. A nd if one defines ‘object’ as I suggested in L ecture 3 and as it is also defined by H usse rl,4 then one will take no exception to this sta tem ent (though one may p erh ap s fo rm u la te it m ore carefully as: ‘Every singular te rm claims to designate an object’5 in o rd e r to allow for the fact tha t what is desig­nated m ay not exist). But, he continues, every such expression also has a m ean ing , and this is to be d istinguished from the object. In support o f this claim, H usserl, like Frege, points to the fact tha t two singular term s can designate the same object and yet not have the same m ean­ing, e.g. ‘the victor o f J e n a ’ and ‘the vanquished o f W aterloo’, or, to quote F re g e’s celebrated exam ple, ‘the Evening S tar’ and ‘the M orning S tar’. O n e can u n d e rs ta n d both expressions, and thus know what m ean ­ing they have, and yet no t know tha t they stand fo r the same object.

F rege and H usserl are h e re re fe rrin g in particu lar to a specific class o f singu lar term s know n as ‘defin ite descrip tions’. So as to be able som e­how to classify these expressions we should first get clear abou t the various sem antic types o f those singular term s with which concrete (perceptible) objects can be designated. I will re tu rn to these distinc­tions w hen, following the discussion o f predicates, I give my own anal­ysis o f singular term s.

A first possible way in which expressions can designate concrete objects is the deictic, by m eans o f dem onstrative p ronouns such as ‘this’ or ‘th a t’ an d personal p ro n o u n s such as ‘I ’, ‘you’, ‘it’. I t is characteristic o f this m ode o f designation th a t it depends on the context o f speech which object the expression stands for. In the case o f such a w ord, one canno t ask for which object does it stand, only for which object does it stand in this o r th a t context. T h e object-reference depends on the p a r­ticular use. I f one takes this p ro p e rty as the criterion o f an expression’s m em bersh ip o f this class, then we will also have to include in the class o f deictic subject-expressions those expressions which consist in the com bination o f a dem onstra tive p ro n o u n or a possessive adjective with a substantive, e.g. ‘this ho rse’, ‘o u r h o rse’, likewise the com bination with the defin ite article (‘the h o rse ’) w here the expression is not used to des­ignate a species (‘the horse is a dom estic an im al’) bu t in such a way tha t an ind iv idual object o f this k ind is m eant, w here it is again given by the context which object it is.

A second class is constitu ted by definite descriptions: expressions such as ‘the victor o f J e n a ’, ‘the Evening S tar’. Expressions of this kind designate an object by specifying a certain characteristic (e.g. th a t of being the victor o f Jen a) which, it is supposed, belongs only to a single

Husserl’s theory of meaning 111

object, this being expressed by the definite article. Russell therefo re aptly called these expressions ‘defin ite descrip tions’.6

A th ird class consists o f p ro p er nam es, such as ‘N apo leon’, ‘V enus’, ‘B onn’. I t is characteristic o f these expressions that, although they des­ignate an object, they do not do so by m eans o f a con tex t-dependen t or con tex t-independen t description. It th e re fo re seems correct to say of these words tha t they have no m eaning. For it makes no sense to ask for the m eaning o f such a w ord or to ask how it is to be understood : one can only ask which object does it designate. In the tradition determ ined by ontology the fact tha t these words designate objects but that one cannot in addition ask for their m eaning m ade them seem the linguistic expressions par excellence. From a naive po in t o f view p ro p e r nam es seem also to be the singular term s which it is m ost easy to understand . In contrast to a definite description a p ro p er nam e seems to designate the object as such, im m ediately and directly .7 We will see later that this is a mistake and that the m ode o f em ploym ent o f p ro p e r nam es belongs to a h igher level than that o f the o ther two classes o f referring- expression and indeed presupposes them .

At p resen t all that needs to be m ade clear is that the thesis o f Frege and H usserl, tha t every ‘nam e’ both designates an object and has a m eaning, is valid for only one class o f concrete singular term s: definite descriptions. Deictic singular term s do not in themselves stand for an object; and p ro p er nam es have no m eaning. For us the im p o rtan t ques­tion is w hether, and if so how, one can m ake sense of the notion o f a m eaning o f an expression which is distinct from the object from an object-orientated point o f view. W hat is to be understood by the m ean­ing o f a ‘nam e’?

Frege had already given a plausible answer to this question. O f course one m ust take note o f his d iffe ren t term inology. Frege, unlike H usserl, does not use ‘sense’ (Sinn) and ‘m ean ing ’ (Bedeutung) as synonyms; ra th e r for tha t which H usserl p rim arily calls ‘m eaning’ F rege exclu­sively uses the word ‘sense’ and, dep a rtin g from ord inary linguistic usage, calls the object the m eaning (Bedeutung) o f the expression. This term inological d ifference does no t involve any additional substantive problem s; it is ju s t that w hen speaking about F rege’s views one m ust be clear w hether one is using the w ord ‘m ean ing’ in his or the usual sense. Now according to Frege the sense o f the expression contains the ‘m ode o f p resen ta tion ’ o f the object.8 It is clearly constitutive of m aterial objects tha t they can ap p ear in an un lim ited n u m b er o f perspectives or m odes o f presen tation . F rege’s thesis is tha t every definite description designates an object as the object which is given in such-and-such a way.

Analysis o f the predicative sentence 112

T h e expression ‘the Evening S tar’, for exam ple, stands for the object which appears at a particu lar tim e in a particu lar p a rt of the sky. T he sam e object appears at an o th er time in an o th er p a rt o f the sky and if one designates it as the object which appears thus one calls it ‘the M orn­ing S tar’.

Now tha t is an account which H usserl was well able to adopt. In Investigation I though he gives yet an o th er account. H e starts from the assum ption that expressions which designate the sam e object bu t have d iffe ren t m eanings a re distinguished from one an o th e r by ‘the specific m anner in which the object is m ean t’ (Investigation I §13). It th e re fo re seem ed plausible to construe the m eaning as a characteristic o f the act. H owever, an expression has one identical m eaning, w hereas any n u m ­ber of acts can in te rp re t it in the same way. H usserl th e re fo re cam e to the view tha t the m eaning consists in the essence (the ‘ideal species’) of the relevant act (‘o f m eaning the particu lar object’). ‘T h e m eaning (.Bedeutung), th e re fo re stands to the various acts o f m eaning (Meinen) . . . as redness in specie stands to the strips o f p aper lying here which all “have” this sam e redness’ (Investigation I §31).

T his at any ra te is a possible way in w hich H usserl can accom m odate m eanings within the fram ew ork o f his ob ject-orientated approach . But it is by no m eans obvious tha t w hen we m ean an object with a definite descrip tion what we u n d erstan d is the essence of this (act of) m eaning. T his account is also unable to answ er the question of how it is th a t th e re are d iffe ren t m odes in which one and the sam e object can be m eant. Does no t a particu lar m ode o f p resen ta tion of the object co rrespond to each such act-essence? L ater, in his Ideas. (§94), H usserl m odified his conception in this direction, which had already been anticipated by F rege’s account: the ‘sense’ is the ‘object in its specific m ode o f p resen ­ta tion’ (§131).

T his is a second possible way o f in teg rating m eanings which are dis­tinct from the object into the object-orientated approach : the m eaning is no t the object bu t the m ode of p resen tation o f the object. A nd it is, m oreover, quite a plausible view.

Reflection on the m ode o f p resen ta tion of objects is characteristic of the so-called transcenden ta l tu rn in ontology. In its transcendental form the theory o f objects thus had a perspective in which it could m ake intelligible the m eanings o f at least those expressions which also desig­nate objects. In the theory th a t the m eaning of an expression is the m ode o f p resen tation of the object, the concept of m eaning rem ains d ep en d en t on the concept o f an object. Indeed from the outset we m ust expect tha t it is simply not possible for the object-orientated approach

Husserl’s theory of meaning 113

to develop a concept o f m eaning which would be in d e p en d e n t o f the concept o f an object. H u sse rl’s first exp lanation of the m eaning of a ‘n am e’ - the m ean ing is the essence o f the act o f m eaning - is an exp la­nation in which the m ean ing is u n d ers to o d in term s o f the object-ref- erence; for it rests on th e assum ption tha t th e re is an act, an d an act is the consciousness o f an object.

T h e concept o f m ean ing which is d e p e n d e n t on re feren ce to objects is o f course unp rob lem atic so long as we are dealing with singular term s, in o th e r words, with expressions which designate objects. B ut now w hat is th e position as regards th e m eaning o f the rem ain ing ca te­gorem atic expressions, viz. (a) general term s an d (b) w hole assertoric sentences? T h ese questions co rresp o n d to the second an d the fo u rth o f the four questions to which I re fe rre d at the en d o f the previous lec­tu re.

Let us first consider question 4, th a t o f th e m ean ing o f the whole assertoric sentence. H usserl gives an answ er to this question in Investi­gation I, befo re answ ering the th ird question - the question about p re ­dicative s tru c tu re - in Investigaton VI. T ackling things in this o rd e r also makes sense in term s o f the subject-m atter. For one can certainly say som eth ing abou t how in general one should in te rp re t the m ean ing o f whole asserto ric sen tences - w hether they be predicative o r no t - even befo re one has investigated th e ir structu re . We shall see tha t the account which H usserl gives o f th e m eaning o f the w hole assertoric sentence p red e te rm in es a particu la r answ er to the th ird and crucial question ab o u t pred icative structu re .

H usserl does no t wish to restric t his thesis th a t every expression both refers to an object an d has a m ean ing to singu lar term s. It holds fo r all (categorem atic) expressions and, in particu la r, fo r whole assertoric sen ­tences (§12). B ut now w hat are we to u n d ers tan d by the m eaning and the object o f a whole assertoric sentence? H usserl provides no unequ iv ­ocal answ er to this question . O n th e one h an d , he says, one can reg a rd the object o f the subject-term o f th e sentence, in o th e r w ords, th a t ‘ “abou t” which the s ta tem en t is m a d e’, as the object o f the sentence. O n this view, th e object o f the s ta tem en t ‘a is la rg e r than b’ would be a o r perhaps a an d b. ‘B ut th e re is also possible,’ he then says, ‘an o th er co n ­ception which views the whole state o f affairs co rresp o n d in g to the sta tem ent as th e ana logue of the object n am ed by a nam e and which d istinguishes it from the m eaning o f the asserto ric sentence. If one does this then one will give as exam ples sen tence-pairs such as “a is la rg e r than b” an d is sm aller than a”. T h e two sentences clearly assert som e­th ing d if fe re n t . . . B ut they express the sam e state o f affairs . . . Now

Analysis o f the predicative sentence 114

w hether we define the object o f the sta tem ent in this o r the o th e r sense . . . statem ents a re always possible which d iffe r in m eaning bu t which rela te to the sam e “object” ’ (§12).

T h is gives the im pression tha t it is m ore im p o rtan t to distinguish object and m ean ing at any price than to specify w hat is to be understood by the m eaning and the object o f a statem ent. H usserl’s wavering betw een two possible ways o f constru ing the m ean ing-ob ject distinction as app lied to w hole assertoric sentences reveals a fundam en tal u ncer­tainty as to how talk o f objects in reg a rd to linguistic expressions is to be defined . O n the one hand , H usserl too defined an ‘object’ as that which is the subject o f possible p red ications.9 O n the o th er hand , his ob ject-orien tated approach - the assum ption tha t all ‘m eaning-confer­ring consciousness’ is consciousness o f an object - obliged him to hold tha t every linguistic expression, o r at least every categorem atic expres­sion, stands for an object. It follows from the first conception th a t the only th ing that can be reg a rd e d as the object o f an assertoric sentence is the object (or objects) fo r which the subject-term (or subject-term s) of the sen tence stands (and one can only speak of an object o f the sentence in th e case o f singu lar predicative sentences). In this way m ean ing and object o f the sen tence are indeed sharply d istinguished, for th e re is now no object co rrespond ing to the m eaning o f the whole sentence. How­ever, fo r the ob ject-orien tated app roach this is an unacceptable result, fo r from such a po in t o f view a m ean ing not su p p o rted by conscious­ness o f an object hangs as it w ere in a void and is simply inconceivable.

So H usserl found him self com pelled by his object-orientated ap p roach to draw the distinction betw een object an d m eaning in regard to whole sentences in a d iffe ren t way. T h e ob ject-orientated approach d em an d e d a conception which assigns to the sentence an object co rre­sp ond ing to the whole m eaning. A nd as we have already seen we can indeed speak o f an object fo r which a whole sentence ‘p ’ stands, viz. the state o f affairs that p. M oreover such a conception is com patible with the defin ition o f an object as the subject o f possible predications, for, even if the assertoric sentence itself is no t a subject-expression (a singular term ), its nom inalized fo rm ‘tha t p ’ certainly is.

B u t now if the object o f an assertoric sentence is u nderstood in this way can we in add ition distinguish a m eaning of the sentence? This is w hat H usserl tries to do in the discussion to which I have ju s t re ferred . T h e two sentences ‘a is bigger than b’ and lb is sm aller than a ’ rep resen t the sam e state o f affairs bu t have d iffe ren t m eanings. H usserl clearly wants to assim ilate the m ean in g -o b jec t distinction as applied to state­m ents as closely as possible to the distinction as applied to singular

Husserl’s theory of meaning 115

term s: the m eanings o f the two sentences are so to speak two m odes of p resen tation o f one and the sam e state o f affairs.

B ut here talk o f d iffe ren t m odes o f p resen tation o f one and the sam e th ing has only a m etaphorical sense. W hat lies behind it? W hat are the criteria fo r deciding (a) when two sentences have the sam e or d iffe ren t m eanings and (b) w hen they stand for the sam e o r d iffe ren t states of affairs? As regards (a) no th ing specific can be in ferred from H usserl’s exam ple. I t may be doub ted w hether he had definite criteria, and hence a definite concept of m eaning, in m ind. As fo r (b) H usserl probably used the s tandard criterion , viz. two sentences stand for the same state o f affairs if they have the same tru th-conditions, i.e. if we can assert a priori (analytically), sim ply on the basis o f o u r u n d erstan d in g o f the sen­tences, th a t if the one (e.g. ‘a is b igger than b’) is true then the o th e r (e.g. (b is sm aller than a’) is also true , and if the one is false then the o th e r is also false.

T h e co rrespond ing criterion in the case o f descriptions would be: two descriptions stand for the same object if we can assert a priori (analyti­cally), simply on the basis of ou r u n d erstan d in g of the expressions, tha t they designate the sam e object. H ow ever this is not the criterion which holds for all descriptions which designate the same object, fo r norm ally we cannot establish w hether they designate the same object simply on the basis o f o u r u n d erstan d in g o f the expressions. We canno t in fer th a t ‘the victor o f J e n a ’ stands for the sam e object as ‘the vanquished o f W aterloo’ m erely from o u r u n d erstan d in g o f these expressions; we can only establish this by experience. An exam ple of descriptions analogous to ‘a is bigger than b’ an d ‘b is sm aller than a in this respect would be, perhaps, ‘the victor o f J e n a ’ and ‘the com m ander o f the victorious arm y at Je n a ’. T h u s we see tha t the criterion by reference to which H usserl determ ines which statem ents stand fo r the sam e state of affairs is in fact d iffe ren t from , and narrow er than , the criterion for descrip tions’ hav ­ing the sam e object. In the case o f statem ents a classification-criterion as wide as tha t which connects all descriptions which designate the sam e object would have to be: statem ents have the same object if they have no t only the sam e tru th-conditions, b u t the sam e tru th -va lue .10 But one w ould then have to say tha t all tru e statem ents designate one and the sam e object; and likewise all false statem ents.

This intuitively u n n a tu ra l but form ally consistent thesis was p ro ­p ou n d ed by F rege.11 H e too th o u g h t tha t one must also distinguish betw een ‘sense’ (Sinn) and ‘object’ (Bedeutung) in the case o f statem ents. H ow ever he in te rp re ted the object o f the assertoric sentence no t as th e state o f affairs fo r which it stands bu t as its truth-value, i.e. as ‘the cir­

Analysis o f the predicative sentence 116

cum stance tha t it is tru e o r th a t it is false’. This conception, in contrast to H usse rl’s, at least allowed a clear distinction to be draw n between m eaning and object. T h e two sentences ‘B erne is the capital o f Switz­e rla n d ’ and ‘B onn is situated on the R h ine’ have d iffe ren t m eanings, bu t they stand fo r the sam e ‘object’, nam ely, the sam e tru th-value, for they are both true . O ne can reach a b e tte r intuitive u n d ers tan d in g of this idea by th ink ing o f the ‘object’ fo r which all tru e statem ents stand as ‘reality’ o r th e ‘w orld’; the m eanings (senses) o f tru e statem ents would then be th e d iffe ren t m odes o f p resen ta tion in which reality m anifests itself (false statem ents would th en not have the ir own object; ra th e r th e ir m eanings would rep resen t the aspects in which reality does no t m anifest itse lf) .12 B ut this idea o f reality as an object as suggested by the substantival expression m ust itself ap p e a r suspect to us. T h e real substance of the analogy which Frege exhibited betw een the m eaning and object o f descrip tions and the m ean ing and tru th -value o f state­m ents lies in a n o th e r d irection and already points beyond the object- o rien ta ted approach . I canno t go in to this h e re 13 bu t would m erely po in t o u t tha t F rege was clearly using the w ord ‘object’ in a non-stan ­d ard sense.14 I f it is taken in its usual sense - as ‘subject o f possible p red ications’ - th en F rege’s theory , w hatever its positive con ten t may be, provides no possible answ er to the question concerning the object o f assertoric sentences. H usserl on the o th e r h an d probably d id not u n d ers ta n d the form al connections Frege had in m ind , and in any case he was bound to be p u t o ff by the intuitive unnatu ra lness o f F rege’s result; he thus m ade his own suggestion in which the analogy with the co rrespond ing distinction in regard to descriptions is no longer p re ­sent. A nd he d id no t develop it fu rth e r.

H ow ever we should no t over-estim ate this negative result. I t simply consists in this: th a t the distinction betw een m eaning and object which is m ade in reg a rd to descrip tions can scarcely be tran sfe rred to asser­toric sentences in the m an n er suggested by H usserl. B ut H usserl was able to d ispense with the very vague idea o f the m ean ing o f assertoric sentences which he in troduced on this occasion; he was nevertheless able to hold on to the view th a t every assertoric sentence ‘p ’ stands for an object, viz. the state o f affairs that p. Only the concept o f m eaning had to be newly defined and its rela tionship to the object newly d e te r­m ined. It now occu rred to H usserl to com bine th e two possible ways, considered as alternatives in §12, o f d istinguish ing betw een th e object and the m eaning of a statem ent. You will recall that, according to the first o f these, the object o f a predicative sentence is the object o f its subject-term ; the whole sentence as such only has a m eaning and not,

Hass er Vs theory of meaning 117

in addition , an object. T h e re was no real a rg u m e n t which told against this conception, only th e ob ject-o rien ta ted app roach , accord ing to which th e re also had to be an object c o rre sp o n d in g to the whole ex p res­sion. B ut now this req u irem en t is fulfilled by sim ply in co rp o ra tin g in to the first possibility th e one u np rob lem atic idea o f the second possibility, viz. the idea th a t every nom inalized asserto ric sentence ‘th a t p' stands fo r a state o f affairs.

T h u s in §34 o f Investigation I th e re em erges the follow ing concep ­tion (the one which is finally accep ted by H usserl). T h e object o f the sen tence ‘Ju liu s is w eep ing’ is th a t o f which it asserts som ething, viz. Ju lius. B ut one can also speak ab o u t the m ean in g o f the sentence. In this case the m ean ing o f the sen tence becom es itself the object-about- which o f a fu r th e r sta tem ent. T h is new object which is designated by the nom inalized expression ‘tha t Ju liu s is w eep ing ’ is the state o f affairs. C o rresp o n d in g to the gram m atical m odification o f nom inalization is the sem antic m odification o f the objectification o f m eaning.

T h e question concern ing the re la tionsh ip betw een th e m ean ing an d object o f th e assertoric sen tence is thus m ade m ore com plicated by the fact that now two objects are involved. T h e object-about-w hich o f the sta tem ent is the object o f the sub ject-term o f the sentence; it is d istin ­gu ished from the m ean ing in the way described by H usserl in the first a lternative o f § 12. T h e state o f affairs, on the o th e r h an d , th e object fo r which the expression ‘th a t/? ’ stands, is the — objectified - m eaning.

T h e question I left o p en w hen I in tro d u c ed objects o f the type that p in L ecture 3, viz. how these objects - states o f affairs o r propositions - should be conceived, w ould thus be answ ered in a specific way: the object that p is the m ean ing o f th e sen tence ‘p ’. T h is idea could a p p e a r plausible, fo r it is na tu ra l to say th a t two states o f affairs that p and that q a re identical if the two sentences and ‘q have the sam e m ean ing (it is o f course assum ed th a t the sen tences an d ‘q con tain no deictic expressions).

B ut even if we d isreg a rd deictic expressions the identification o f the state o f affairs that p with the m ean in g o f ‘p ’ is no t tenable. We can already see this from linguistic usage: we canno t transla te sta tem ents abou t states o f affairs with sta tem ents ab o u t m eanings. For exam ple, one canno t say instead o f ‘the state o f aff airs th a t it was snow ing yester­day is p leasing’ ‘the m ean ing o f th e sen tence “it was snow ing yesterday” is p leasing’. T his fact which is o ften rem a rk e d u p o n 15 does no t itself p rove anyth ing . W hat on e m ust go on to ask is w hat the g ro u n d o f this discrepancy in linguistic usage is.

I t is useful to look again a t th e c o rre sp o n d in g theo ry o f Frege. T o

Analysis o f the predicative sentence 118

the superficial rea d e r F rege seems to hold the sam e view as Husserl. I w ould rem ind you that F rege’s te rm for what H usserl calls a ‘state o f affa irs’ is ‘th o u g h t’. Now Frege also says that the sense o f an assertoric sentence is a th o u g h t.16 Also according to Frege the object o f the no m ­inalized expression ‘tha t p ’ is the sense o f ‘p ’. 17 However, it is easy to overlook the fact that w hat Frege calls ‘sense’ is a technical term and does not co rrespond at all to what one ordinarily understands and what H usserl u n d ers tan d s by ‘m ean ing’ (or ‘sense’). In particular the co rre­lation betw een sense and u n d erstan d in g from which I started in the previous lectu re does no t hold for Frege. T h e re seems to be no com ­prehensive te rm in F rege for what we understand when we understand a linguistic expression. I f we continue to call th a t which we u n derstand ‘m ean ing’ (Bedeutung) - and thus use the w ord quite differently from Frege - we will have to say that fo r Frege the sense o f an assertoric sentence constitu tes only a p a rt of its m eaning. Because, for Frege, the sense is the th o u g h t and th e though t is that which can be tru e o r false,18 only what is re levan t to the question o f tru th o r falsity belongs to the sense .19 H ere , as D um m ett has show n, Frege anticipated the m odern view tha t the m ean ing o f a sentence consists in its tru th-conditions. It is only later tha t I shall re tu rn to this really crucial aspect o f F rege’s th e ­ory. I shall also at p rese n t d isregard those o th er constituents o f m ean­ing20 in which according to Frege a feeling is expressed o r a ‘h in t’ is given to the h e a re r .21 W hat is im p o rtan t from the point o f view of ou r discussion is tha t, for F rege, an assertoric sentence and the co rrespond­ing in te rrogative sentence have the same sense; this m eans tha t the assertion-m ode o r question-m ode (Frege speaks of ‘assertoric force’) does no t belong to the sense bu t is an additional elem ent o f m eaning .22 Now this view corresponds in the m ost precise fashion to actual linguis­tic usage, w hen we consider tha t the sense is tha t for which the expres­sion ‘th a t / / stands. ‘T h a t/? ’ d iffers from (p ’ in tha t it lacks the assertion- m om ent (cf. above p. 64).

W e can now re tu rn to H usse rl’s thesis tha t the state o f affairs that p is the (‘objectified’) m ean ing o f ‘p \ It is now clear why this thesis is false. T h e m eaning o f ep ’ always contains m ore than that for which the expression ‘th a t/? ’ stands. H ow ever justified an d however natu ra l it is to say tha t an expression ‘tha t p* stands for som ething - w hether one calls it a state o f affairs, a p roposition o r a th o u g h t - it is false to say this o f the unm odified expression ‘p ’. Som eone who says ‘p ’ is no t simply designating a state o f affairs, bu t at the same time asserting that it is tru e o r ‘ob ta ins’; and this add itiona l factor which is included in the m ean ing o f ‘p ’ can no lo nger be construed objectually.

Husserl’s theory of meaning 119

H usserl did not overlook this factor o f the sentence-m ode; on the contrary he held th a t a ‘positional quality’ belongs to the essence o f intentional acts as such, including all ‘nom inal’ acts.23 H e thus tried to neutralize the th rea t to his fundam en tal position, which is orien tated towards nam es and represen tation , posed by the natu re o f sentences by simply incorporating this aspect o f sentences into the object-reference. Now this would m ean: every singular term not only stands for an object; with such a term som ething is implicitly asserted. B ut w hat is it tha t is implicitly asserted with a singular term? A ccording to H usserl it is the existence o f the object.24 A pplied to singular term s tha t stand for (m ate­rial) objects this thesis is not im plausible and I shall be re tu rn in g to it later (Lect. 26). B ut even now we can anticipate tha t if this thesis turns ou t to be correct the conclusion it suggests is the opposite o f tha t draw n by H usserl: it would follow that these nam es re fe r back to statem ents (those which they implicitly assert) (cf. above p. 73). H owever, applied to the nom inal expressions with which we are now concerned, expres­sions o f the form ‘tha t/?’, the thesis would seem to be false. T h e idea tha t w hen we say ‘tha t p ’ the obtaining o f the state o f affairs (or the tru th o f the thought) is implicitly asserted contradicts the fact tha t we can ju s t as well com plete ‘that/? . . with ‘is false’, ‘is doub tfu l’, etc., as with ‘is tru e ’. Som eone who is ju s t beginning to say ‘tha t p . . .’ has not yet implicitly p re judged how he will com plete it.

So we have no alternative bu t to reject H usserl’s identification o f the state o f affairs thatp with the m eaning of ‘p ’. How ever, H usserl’s object- orien tated approach gives rise to ano ther an d even m ore serious p ro b ­lem. W hether one identifies the state o f affairs that p with the m eaning o f *p’ or in te rp re ts th e ir relationship in a d iffe ren t way there still arises the fu rth e r and crucial question o f w hether ou r u n derstand ing o f the m eaning o f ‘/?’ is g rounded in ou r know ing for which object ‘tha t p ’ stands o r w hether it is the o th e r way round .

H usserl himself, in Investigation I, characterizes the connection between m eaning and state o f affairs thus: the state o f affairs is the objectified m eaning. This would seem to imply that the identification o f the state o f affairs that p already presupposes the understan d in g of the m eaning o f ‘p \ B ut then this would m ean tha t the m eaning would fo r its part have to be in te rp re ted non-objectually. But, as we have already seen, fo r H usserl’s object-orientated approach a m eaning which is not p ro p p ed up by an object-consciousness is impossible.

So H usserl draws the conclusion which, given his starting-point, is the only possible one: as the state o f affairs thatp is in any case supposed to be identical with the m eaning o f *p’ it is easy to project the conscious­

ness o f the state o f affairs back in to the u n d e rs ta n d in g o f m eaning. Consistently with this H usserl no longer speaks, in Investigations V and V I, o f the m eaning, b u t only o f the state o f affairs. T h e u n d ers ta n d in g o f the no t yet nom inalized sta tem en t is also in te rp re ted as ‘conscious­ness of the state o f a ffa irs’; only in this case the state o f affairs is no t yet ‘objectual in the strict sense’ (Investigation V §§36, 38).

We m ust th e re fo re reg a rd H u sse rl’s a ttem p t to apply the m e an in g - object d istinction to statem ents as well as to nam es as a failure. In itself this would be no d isaster, no r is this resu lt th e essential one. W hat is essential is th a t H usserl construes the m ean ing o f the sen tence as an object. How can the soundness o f this conception be tested? By asking: if th e state o f affairs is no t founded in the m ean ing then how, according to this conception, is it positively u n derstood? I f one canno t have recourse to th e m ean ing of the sta tem en t th e only way o f explain ing the ontological status o f the state o f affairs is to regard it as a composite object. T h e idea tha t the m eaning o f an assertoric sentence is an object inevitably results in the view tha t the m anner in which the m eaning o f the whole s ta tem en t arises ou t o f the m eanings o f its parts can only be th o u g h t o f as com position. O ne can leave open the question o f w hether one speaks o f m eanings o r objects in reg ard to th e sentence-parts. W hat is crucial is th a t it is a m atter o f com position. Com position implies objects both as its elem ents and as its results. Even if one avoids speak­ing o f objects and thinks o f the meaning o f the com posite expression as com posed o f the meanings o f the com ponen t expressions, in the concept o f com position one is m aking use o f an objectual category and hence constru ing m eanings as objects.

We are thus faced with the crucial th ird question o f the fou r ques­tions em phasized by m e at the en d o f the last lecture viz. how is th e com bination o f the two sen tence-parts, the singu lar te rm an d the p re d ­icate, to be understood? H usserl’s answ er to th e fou rth question (how is the whole assertoric sen tence understood?), viz. its m ean ing is an object (a state o f affairs), largely p rede term ines the answ er to th e th ird ques­tion: to the com position o f the expression th e re m ust co rrespond a com position in the object, o r m eaning.

This may at first a p p e a r harm less, and p e rh ap s you will even ask in astonishm ent: how else is the m eaning of the whole expression to be u n d ersto o d , if no t as com posed o f the m eanings o f its com ponent- expressions?

In the nex t lecture we shall have to exam ine how far the objectual conception o f the m ean ing o f a predicative sentence can be p reserved by m eans o f an acceptable concept o f com position.

Analysis o f the predicative sentence 120

L E C T U R E 1 0

Collapse of the traditional theory of meaning

T h e g u id ing question o f o u r investigation is: W hat is it to u n d e rs ta n d a sen tence? T h is question I reg a rd as th e fu n d am e n ta l question o f p h i­losophy, which shou ld take the place o f th e trad itio n a l fu n d am e n ta l question - w hat is being as being? o r w hat is it to re p re se n t an object? We a re thus n o t asking this question fo r th e sake o f asking it b u t ra th e r with th e aim o f reach ing a new philosophical ap p ro a ch . T h e conviction exp ressed in th e first p a r t o f these lectu res th a t th e question - W hat is it to u n d e rs ta n d a sentence? - has th e form al universality o f th e ques­tion concern ing objects as objects an d in fact is m o re com prehensive than th e la tte r, does no t a m o u n t to a new ap p ro ach vis-ä-vis th e object- o r ien ta te d ap p ro ach ; it m erely points in the d irec tion o f such a new ap p ro ach . O n e can only develop it by w ork ing o u t a new basic concep­tuality a p p ro p ria te to the new subject-m atter. T o achieve this we shall first look at w hat hap p en s w hen the trad itiona l basic conceptuality is ap p lied to the u n d e rs ta n d in g o f a sen tence, in the expecta tion th a t the resu ltin g tension will yield a basis fo r th e d ev e lo p m en t o f a new horizon o f exp lana tion .

In the last lec tu re we w ere able to see how th e ob ject-o rien tated ap p ro ach d id n o t p rev e n t H usserl fro m giving a p lausib le account o f th e d istinction betw een th e m ean ing an d the object o f nam es b u t th a t in th e case o f th e m ean ing o f whole sentences it lands h im in grave difficulties. T h e state o f affairs that p, which H usse rl h im self initially in te rp re ts as a subsequen t m odification, m ust, given the ob ject-o rien ­ta ted ap p ro ach , be p ro jec ted back in to the o rig inal consciousness o f the m ean in g o f A ccordingly, if the consciousness o f a sta te o f affairs ca n n o t be e lucida ted by refe ren ce to th e u n d e rs ta n d in g o f a sentence, b u t the u n d e rs ta n d in g o f the expression com posed o f s in g u lar te rm an d p red ica te is an originally objectual consciousness, th en the m a n n e r in which the m ean ing o f th e pred icative sen tence d ep en d s on the m ean ­

Analysis o f the predicative sentence 122

ing o f its term s can only be explained by m eans o f the only object-orien­ta ted conceptuality available for this purpose, i.e. it m ust be construed as composition, as synthesis.

So we are faced with the fu n d am en ta l question of sem antics, here in connection with the special case o f the predicative sentence: how does th e m eaning o f a com posite expression result from the m eanings of its co m ponen t expressions? Proving itself adequate in this fundam en ta l sem antic question is the decisive criterion for th e applicability o f a ph il­osophical conceptuality to the questions of semantics.

I t was clear to H usserl th a t the com position constitutive o f a state o f affairs canno t be w hat one ord inarily understands when one speaks o f com posite objects. O rd inarily w hen one com bines objects to form a com posite object - e.g. w hen one com bines pearls to form a pearl-neck- lace o r build ing-stones to form a build ing - the com posite object is ju s t as m uch a concrete spatio-tem poral object as its parts. A state o f affairs o r fact, on the o ther h an d , is no t a concrete spatio-tem poral object. W hen we speak o f the fact that C aesar was m u rd e red in Rome in 44 B.C. C aesar is a concrete spatio-tem poral object. Equally the event o f his m u rd e r is spatio-tem porally locatable. It h ap p en ed in th a t place and at th a t time. T h e fact th a t C aesar was m u rd ered in tha t place at that tim e, on the o th e r hand , is no t locatable and datable. T h u s the object fo r which a nom inalized sentence stands, like tha t for which a nom i­nalized p red ica te stands, is not a concrete spatio-tem poral object: states o f affairs are , like a ttribu tes , so-called ‘abstract’ objects. H usserl calls concrete objects ‘rea l’ objects and abstract objects ‘ideal’ objects. For him the criterion o f a ‘rea l’ object is th a t it can be sensibly perceived (Inves­tigation VI §46).

T h u s a lthough his ob ject-orien tated approach obliged H usserl to construe states o f affairs as com posite objects they are nonetheless objects o f a n o th e r o rd e r than the objects of which they a re com posed. B ut now this m eans tha t the com position in question m ust be o f a spe­cial kind. H usserl a ttem p ted to overcom e this difficulty by m eans o f his theo ry o f categorial synthesis. T his rep resen ts the most far-reaching a ttem p t so fa r m ade to explain states o f affairs and the m eaning of sentences fro m an ob ject-orien tated position.

B efo re ex p o u n d in g the basic fea tu res o f this theory I would like briefly to d raw your a tten tion to an o th e r object-orientated position in which the no tion o f com position is app lied naively to states o f affairs, viz. the position rep resen ted by W ittgenstein’s Tractatus. Inasm uch as W ittgenstein in the Tractatus is, un like H usserl, o rien tated prim arily tow ards sentences ra th e r th an nam es his position is already a decidedly

Collapse of the traditional theory of meaning 123

language-analytical one: ‘Only propositions have sense; only in the nexus o f a proposition does a nam e have m eaning’ (3.3). H ow ever this idea is still in te rp re ted ontologically. T h e Tractatus g rounds the sem an­tic prim acy o f sentence over nam e in an ontological prim acy o f facts over things: ‘T h e w orld is the totality o f facts, not o f th ings’ (1.1). Now the question arises: w hat are we to und erstan d by a ‘fact’? W ittgenstein answers: ‘W hat is the case - a fact - is the obtaining o f states o f affairs’ (2.). A nd what is a state o f affairs? T o this W ittgenstein answers: ‘A state o f affairs is a com bination o f objects’ (2.01).

T his view is open to the criticism at which I have already h in ted , viz. that it depicts the state of affairs as a com posite concrete object. T h e Tractatus invites this criticism by explicitly stating: ‘In a state o f affairs objects fit into one an o th er like the links o f a chain’ (2.03).

W ittgenstein rejected this view h im self when he abandoned the object-orientated position o f the Tractatus. From this period stem some notes which, u n d e r the title ‘Com plex and Fact’ have been published as an append ix to Philosophical Remarks. 1 H ere W ittgenstein writes: ‘Com ­plex is no t like fact. For I can e.g. say o f a com plex tha t it moves from one place to ano ther, bu t no t o f a fact . . . A nd a com plex is a spatial object, com posed o f spatial objects . . . B u t that this com plex is now situated here is a fact . . . T o say tha t a red circle is composed o f redness and circularity, o r is a com plex with these com ponent parts is a misuse o f these words and is m isleading (Frege was aware o f this and told me). It is ju s t as m isleading to say the fact tha t this circle is red (that I am tired) is a com plex whose com ponent parts are a circle and redness (myself and tiredness) . . . O f course we also say: “to po in t ou t a fact”; but th a t always m eans; “to po in t out the fact tha t T o po in t ou ta fact m eans to assert som ething, to state som ething. “T o point ou t a flower” doesn’t m ean this . . . T h e roo t o f this m uddle is the confusing use o f the word “object”.’

W hat W ittgenstein here calls a com plex is a com posite concrete object. In now going so far as to say th a t the fact does not consist o f som ething he is rejecting the object-orientated approach altogether. H usserl, however, has shown tha t on the basis o f an object-orientated approach one can still perfectly well distinguish between a com plex and a fact.

T his brings me to his theory o f categorial synthesis. T h e task H usserl set h im self was to distinguish from the real com position o f an object a special non-real com position which is constitutive o f a fact. Let m e try to m ake the distinction clear by m eans o f an exam ple. A ham m er is a real object which is com posed of two parts: a shaft and a head. I f we

Analysis of the predicative sentence 124

ascertain this an d say ‘this ham m er is com posed o f a shaft an d a h e a d ’ th e re co rresponds to this sta tem ent the state o f affairs th a t this h am m er is com posed o f a shaft an d a head. T h e state o f affairs, it is assum ed, is itself an (ideal) com posite object. Now what a re its constituents? Should we say: the (real) com position is (ideally) com bined, on the one hand , with the h am m er, on the o th e r h an d , with the shaft an d the head? T h e state o f affairs in question would then be com posed o f two constituents: (1) the real com position an d (2) the o rd e red object-trio [ham m er, shaft, head]. H usserl p re fe rre d an o th er way o f looking at it (Investigation VI §48) accord ing to which only the real objects, thus ham m er, shaft an d head , function as constituen ts of the state o f affairs and the real com ­position (the p a rt-w h o le relation) rep resen ts th e way in which these objects a re (ideally) com bined in th e state o f affairs. Now w hichever way the m o d e o f com position of the state o f affairs is to be u n d ersto o d (I shall com e back to this) it is clearly fundam entally d iffe ren t from the m ode o f com position o f the ham m er. T h e h am m er itself en ters into the state o f affairs as a p a rt, and , although a real com posite object can always be itself a real p a r t o f a la rg e r whole, it can never be so in such a way th a t its own parts can be the com plem entary constituen ts o f the new whole. T h e state o f affairs th a t the h am m er consists o f head an d shaft, is, in con trast to the ham m er, no t a percep tib le object. A nd equally we can n o t perceive its com position as we can perceive the com ­position o f th e h am m er o u t o f head and shaft. I t is th e re fo re tem pting to say: ideal com position is no t ascertainable by percep tion bu t only in thought.

H usserl can here appeal to a long trad ition accord ing to which th in k ­ing, the ‘u n d e rs ta n d in g ’, is a faculty o f synthesis and a synthesis which is no t a species o f real com position .2 Let m e illustrate this by m eans o f an o th e r exam ple. W hen we ap p re h e n d the fact tha t A is separate from B then A a n d B a re certainly no t really com bined, they are separate . A nd yet in th e state o f affairs th a t they are separate , they a re connected. T hey are b ro u g h t in to this connection, which is not a real connection, by though t. T his does n o t m ean th a t the connection does not actually (wirklich) exist (A an d B a re actually separate). T h a t facts are no t real objects (concrete objects in space and tim e), th a t they are only consti­tu ted in tho u g h t, does no t m ean th a t they a re no t actual.

T h in k in g too, like all consciousness, is u n d ersto o d by H usserl as objectual consciousness and hence as an ‘act’.3 Acts o f th ink ing he calls ‘categorial’ acts in con trast to ‘sensory’ acts in which concrete objects a re rep resen ted . It is characteristic o f a categorial act tha t it rep resen ts an object which is com posed in such an d such a way as com posed in such

Collapse o f the traditional theory o f meaning 125

and such a way - which it can only do by sim ultaneously re p re se n tin g its co m p o n en t objects. Now th e re p re se n tin g o f each com ponen t-ob ject is (by definition) itself an act. A categoria l act is th e re fo re a synthetic act which is fo u n d ed on o th e r acts, ultim ately on sensory acts w hich re p ­resen t th e real objects which e n te r in to th e synthetic object. T h e syn the­sis o f th e objects o f th e fo u n d in g acts is accom plished by th e fo u n d ed categorial act; a n d in this synthesis th e new synthetic object is consti­tu ted . T h u s the la tte r canno t, even in princip le, be rep rese n ted in a sim ple (sensory) act.

So H usserl tries to explain the d istinction betw een ideal an d real objects and the pecu lia r com position o f ideal objects by d istinguish ing betw een the c o rre sp o n d in g acts, thus by d istinguish ing the ways in which the re levan t objects a re given (thus by giving a ‘tran scen d en ta l’ exp lanation). T h e o rd e r o f types o f object is g ro u n d e d in the o rd e r o f acts. T h e exp lana tion he gives is supposed to ho ld fo r all ideal objects, also fo r species (which are constitu ted in acts o f ‘ideative abstraction ’), a ttrib u tes and equally fo r sets. H ow ever, I shall confine m yself to states o f affairs. T h e com position o f a state o f affairs, which is fundam entally d if fe re n t from all real com position , is exp la ined by saying th a t this syn­thesis is the synthesis accom plished by a categorial act. It thus becom es clear ‘th a t categorial functions, in “fo rm in g ” sensible objects, leave th e ir real essence u n to u c h ed . . . C ategoria l fo rm s d o no t glue, tie o r p u t parts to g e th er so th a t a real, sensibly percep tib le w hole results. T h ey do no t fo rm in the sense in which the p o tte r form s. F or then th a t which was originally given to sense-percep tion w ould be m odified in its own objecthood: re la tin g an d connecting th o u g h t an d know ledge w ould no t be th o u g h t an d know ledge o f w hat is, b u t ra th e r falsifying tra n sfo r­m ation into som eth ing else’ (Investigation VI §61).

P erhaps you will ask: to w hat ex ten t then can one say th a t particu la r states o f affairs actually {wirklich) ob ta in (and th a t the co rresp o n d in g sta tem ents a re true), if these objects a re no t ‘re a l’ (real) an d are only constitu ted in th e synthetic acts o f th ink ing? T o this H usserl can reply: a state o f affa irs is one th a t actually obtains (and the co rresp o n d in g s ta tem en t is tru e ) if the re levan t categoria l synthesis on the foundation o f th e real objects th a t e n te r into it can be p e rfo rm e d (is possible) (e.g. the state o f affa irs th a t sha ft and h ea d are com bined , actually obtains if the co rre sp o n d in g synthesis o f these real parts can be p e rfo rm ed ).4

Now befo re I com e to th e application o f this theo ry o f categorial syn­thesis to o u r concrete question co ncern ing th e sem antic s tru c tu re o f p red icative s ta tem en ts I w ould like to draw y o u r a tten tio n to a p a rticu ­lar sem antic p ro b lem which H usserl th o u g h t this theory could solve:

Analysis o f the predicative sentence 126

the problem o f the sem antics o f syncategorem atic expressions. As I rem arked in the last lecture (p. 109) such expressions constitute, even for H usserl, a class o f expressions which have m eaning b u t do not stand fo r an object. This conception can be in teg rated with the object-orien­ta ted conception, with which p rim a facie it seems incom patible (Inves­tigation V §§4 ff.), by m eans o f the theory o f categorial synthesis. A ccording to H usserl, syncategorem atic expressions a re com bination- words; they have no ‘in d e p e n d e n t’ m eaning. Only expressions which stand for an object (categorem atic expressions) have an ‘in d e p en d e n t’ m eaning. Now categorem atic expressions can only be com bined with o th er categorem atic expressions to form a com plex expression with a new un ita ry m eaning if this com bination is m ediated by one or m ore syncategorem atic expressions. T h is sem antic-syntactic conception co r­responds directly to the on to log ica l-transcenden ta l conception o f cate­gorial synthesis. T h e synthesis o f a categorial act is expressed in the n o n -in d ep e n d en t m eanings of the syncategorem atic term s (e.g. ‘a n d ’, the predicative ‘is’, ‘ = ’). T hey a re objectually in te rp re ted by these acts no t in the sense th a t they them selves stand fo r objects, bu t ra th e r in the sense tha t they rep rese n t the fo rm of unity in which the synthetic object is constitu ted on the basis o f the found ing objects. Since it is again an act which confers m ean ing on th e syncategorem atic expression, and since the to ta l-m eaning o f this synthetic act is again an object, the ob ject-orien tated app roach can be upheld in im pressive fashion even fo r the u n d e rs ta n d in g o f these expressions.

B ut in th e end we have to ask w hether the theory o f categorial acts is really capable o f m aking the consciousness o f states o f affairs, o r the u n d e rs ta n d in g o f the m eaning o f com posite expressions, intelligible. In particu lar, how do th ings stand as regards th e m eaning o f the p red ica­tive sentence? I have deliberately p resen ted H usserl’s theory o f non- real com position as abstractly as H usserl h im self presen ts it, because the subsum ption u n d e r this theory o f the predicative sentence gives rise to an add itiona l difficulty. I f we take a sim ple predicative sentence, e.g. ‘H eide lberg Castle is r e d ’, then if the theory o f categorial synthesis is to be app lied we m ust assum e th a t no t only the singular te rm ‘H eide l­berg Castle’ bu t also the predicate-expression ‘re d ’ stands for an object; fo r if we do no t have at least two objects th en we canno t speak o f a com position, a synthesis. We thus com e up against the second o f my fo u r questions, viz. th a t concern ing the m ean ing of the predicate.

A n objectifying conception o f predicates o f the kind one could have in fe rred from the general s tru c tu re o f the theo ry o f categorial acts is in fact to be fo u n d in H usserl. T h e analysis o f the predicative sentence-

form in Investigation VI §48 is carried out toge ther with an analysis o f sentences in which it is said o f som ething that it contains som eth ing else as a part. As a unitary schem a fo r both predicative sentences and w h o le-p art sentences H usserl suggests: A is (has) a A is a ’ is the fo r­malization o f a predicative sentence with a copula such as: ‘T h e castle is re d .’ H usserl attaches g rea t im portance to separating the copula, as a syncategorem atic com bination-w ord which is supposed to rep resen t (repräsentieren) the synthesis, from the predicate. A has a , on the o ther hand , is a (not very happy) attem pt to form alize a w h o le-p a rt sentence, e.g. ‘T h e castle has the banquet-hall.’ T h e converse form which H usserl gives both for A is a’ and fo r A has a , viz. ‘a is in A ’ (e.g. ‘T h e banquet- hall is in the castle’) is clearer. If we also apply this converse fo rm to the predicative sentence we get: ‘(the) redness is in the castle’. Obviously we can now convert this fo rm itself into the converse form ‘T h e castle has redness’, which, in contrast to ‘T h e castle is red ’, H usserl regards as the fo rm of expression in which the synthetic s truc tu re becomes explicit.

We also find this assim ilation o f the sub ject-p red icate sentence to the w h o le -p a rt sentence elsew here in H usserl. In Investigation I II , which is headed ‘O n the theory o f wholes and parts’, he says tha t predicates stand for ‘n o n -in d ep en d en t parts’. ‘We in te rp re t the concept part in the widest sense which allows one to call any thing a “p a r t” that can be distin­guished “in ” an object or, speaking objectively, tha t is “p rese n t” in it . . . T hus every non-relative “real” p red icate points to a p a r t o f the object o f the subject-term . T h u s e.g. “re d ” and “ro u n d ” . . .’ (§2)

O ne m ight question w hether the w h o le-p a rt relation is the m ost suit­able relation to which to assimilate the sub ject-p red icate structure . Instead o f ‘the redness is in the castle’ one could suggest ‘the redness is on the castle’, an d instead o f ‘the castle has redness’, ‘the castle is com ­bined with redness’. T h e real question, however, is no t which relation is to be p re fe rre d bu t whether a predicative sentence can be in te rp re ted as a relational sta tem ent at all. T h a t it m ust be thus in te rp re ted how ever follows necessarily from th e idea th a t the pred icate stands fo r som e­th ing; and this idea is itself inevitable if one starts from the assum ption tha t the state o f affairs is constitu ted in a categorial synthesis. In d ee d the idea that the predicate stands for som ething does not even d ep en d on the peculiarities o f H usse rl’s theory o f categorial synthesis; ra th e r it rests on the fundam en tal presupposition , which we also fo und in the Tractatus, that the state o f affairs is som ething com posite; fo r this p re ­supposes tha t it is com posed o f at least two constituents.

So you see, the specific way in which H usserl has answ ered the fou rth question (the m eaning of the whole sentence is the state o f affairs) p re ­

Collapse of the traditional theory o f meaning 127

Analysis o f the predicative sentence 128

determ ines a specific answ er to the th ird question, how the m eaning of the whole sentence arises ou t o f the m eaning of the com ponent-expres- sions (namely, as a result o f com position, o r m ore precisely: as a result o f categorial synthesis). A nd this answer to the th ird question in tu rn presupposes a specific answ er to the second question, tha t concerning the m eaning o f the predicate , nam ely, th a t the m eaning o f the p red i­cate (e.g. ‘re d ’) is the object for which its nom inalized m odification (‘redness’) stands. I t is to be noted tha t every step in this thought- sequence (if one d isregards the peculiarities o f the theory o f categorial synthesis) is a necessary consequence o f the object-orientated approach as such and no t som ething peculiar to H usserl’s philosophy.

In H usserl the in te rp re ta tio n o f predicates as objects does no t only arise in this way as a necessary consequence from the systematic con­text, bu t also sim ply from the conception o f a p red icate as a categore­m atic expression, or, m ore fundam entally , because an alternative con­ception o f m eaning which d id not requ ire to be p ro p p ed up by objects was sim ply no t available w ithin the fram ew ork o f the object-orientated approach . It is tru e that in Investigation I. §12, with which I began, H usserl also claim ed tha t the distinction between object and m eaning which had been shown to obtain in reg a rd to nam es m ust also be m ade with regard to predicates; indeed he even assum ed there tha t a p red i­cate does not designate an object at all and that, the re fo re , one cannot speak o f the object o f a p red icate bu t simply of a ‘relation to objects’, these being the objects to which a pred icate can be applied. T hus he can explain the distinction between object and m eaning as applied to predicates in a way fam iliar to us from m odern semantics: two p red i­cates - e.g. ‘an equilateral triangle’ and ‘an equ iangu lar triang le’- can have ‘the same relation to objects, the same range o f possible applica­tion ’ and yet no t have the sam e m eaning. But if we ask how this m ean­ing which is d istinct from the relation to objects is itself to be u n d e r­stood the answ er is exactly analogous to tha t in the case o f the m eaning o f the whole sentence: (a) in speaking o f ‘redness’ one is speaking objec- tually of the m eaning o f the p red icate ‘re d ’ (b) in the absence o f any o th e r conception o f m eaning the objectual consciousness (of redness) is p rojected back in to the orig inal m eaning-consciousness o f the predicate ‘re d ’. A lthough in und erstan d in g the pred icate o f a sentence the act of consciousness is no t objectually d irected to the m eaning o f the p red i­cate, bu t only to the object o f the subject-term o f the sentence, the m eaning o f the pred icate is nonetheless an object, nam ely, the co rre­spond ing attribu te.

T h u s the objectual conception o f predicates canno t be shaken by sim ­

Collapse o f the traditional theory of meaning 129

ply po in ting to the fact tha t H usserl h im self defined an ‘object’ as the subject o f possible pred ications (cf. p. 418 n. 4). T his defin ition does no t con trad ic t the idea th a t p red icates too stand fo r objects. F or every p re d ­icate can be nom inalized and one can then say: ju s t as, though the state o f affairs that p is no t the object-about-w hich o f the sentence ‘p ’ the sentence nevertheless stands fo r this state o f affairs, so, in the same way, a lthough in using the pred icate ‘re d ’ one is n o t d irec ted to the a ttribu te o f redness as an object, nevertheless the p red ica te stands fo r this object, and this object is its m eaning.

Are we better eq u ip p ed in the case o f p red icates th a n we w ere in the case o f assertoric sentences to show th a t this p ro ce d u re o f H usserl’s involves a hysteron-proteron? I believe we are . For in th e m eantim e we have inco rpo rated in to o u r enqu iry the question o f the construction of the whole expression from the m eanings o f the com ponen t-expres- sions; and it is in rela tion to this p rob lem , the p rob lem upon which everything else h inges, tha t we can show that the ob ject-orientated approach founders.

T h e object-orien tated app roach req u ired tha t the way in which the m eaning o f the whole expression arises o u t o f the m eanings o f its com ­po n en t expressions be in te rp re ted as composition. T h a t this conception is un tenab le if com position is th o u g h t o f in th e usual sense as real com ­position we saw by referen ce to W ittgenste in ’s Tractatus. T h e p u rp o se o f the theory o f categorial synthesis was to overcom e this difficulty. Was it successful? Yes, to the ex ten t tha t th e com position o f the state o f affairs canno t be in te rp re ted as-real com position. B ut this only tells us how the com position may not be in te rp re ted . W hat is still lacking is a positive characterization o f this com position. In the case o f real com ­position we have defin ite criteria fo r dec id ing w hether o r no t an object A is com bined with an object B (e.g. the shaft and head o f the ham m er); and this is equally so in the case o f a real p a rt-w h o le rela tionship . Now if the talk o f com position is not to be com pletely em pty we m ust also have a criterion fo r decid ing w hether o r not an ideal com position obtains. We canno t establish, e.g., w h eth e r redness is in the castle or com bined with the castle in the way th a t we can establish th a t the d raw er is in the table o r is com bined with it. Redness fo r its p a r t is not a real object, but an a ttribu te and this canno t be a ttached in a real way to the castle o r occur in it as a real, separa te part. Indeed this was stressed by H usserl him self. B ut then w hat sort o f positive criteria do we have?

It seems to me we only have one: th a t redness is in (or on) the castle is the case if and only if the castle is red . In o th e r w ords, if we a re asked

Analysis o f the predicative sentence 130

which relation we m ean w hen we speak o f the relation between the a ttribu te and the object we can only reply: that relation which obtains w hen the co rresp o n d in g p red icate applies to the object. I f this is correct - and so long as no a lternative way o f u n d ers tan d in g this rela tion is offered we m ust accept it - th en th e hysteron-proteron o f the ob ject-orien­tated conception o f p red icates is established. W hat a sentence such as ‘Redness is in the castle’ o r ‘R edness is com bined with the castle’ m eans can only be exp lained by recou rse to the sentence ‘T h e castle is re d ’ and no t the o th e r way ro u n d . It is im m aterial which preposition we use in the objectual re n d e rin g - w h eth er we say the redness is in o r on or upon the castle o r com bined with it; fo r what we m ean by all such awk­ward fo rm ulations (aw kw ard because they are parasitic on d iffe ren t real relations) can be m ade precise (and can only be m ade precise) by re ference to the s tra igh tfo rw ard predicative sentence in which no re la ­tion is expressed .

We a re thus at th e tu rn in g p o in t o f the whole discussion. For if it is tru e tha t we can only define the rela tion between a ttr ib u te and object by m eans o f the orig ina l p redicative sentence then we cannot seek to explain the u n d e rs ta n d in g o f the predicative sentence itself by m eans o f tha t rela tion . B u t then this m eans th a t we requ ire a com pletely new exp lanation o f th e u n d e rs ta n d in g o f a predicate, an explanation which does no t have recou rse to the nom inalized form o f the p red icate and which does no t take the fo rm o f saying tha t the p red icate stands for som ething. For any such exp lanation would again have to speak o f a com bination o f the object o f th e subject-term with the object (or m ean ­ing) o f th e p red icate , and w hen asked fo r the criterion o f the existence o f this com bination w ould again have to have recourse to an u n d e r ­standing o f the pred icative sen tence which one already possesses. We m ust th e re fo re com pletely ab an d o n the ob ject-orientated explanatory m odel o f a com position o r synthesis.

T his m odel which consisted in the assimilation o f a logical stru c tu re to a real relation (and com position unless specially defined is a real re la­tion) offers only two alternatives: either one does no t distinguish the com position o f a state o f affairs from tha t o f a real th ing a t all (Tracta­tus), or one does d istinguish them bu t is then unable positively to ch a r­acterize them (H usserl). I f we now look back at the theory o f categorial synthesis it becom es clear th a t w hat gave it plausibility was m erely the negative advan tage o f avoiding th e absurdities o f a real com position. T h e vagueness o f the concep t o f ideal com position by which this advantage is pu rchased is no t rem oved by p ro p p in g it up with categorial acts, for such acts can them selves no t be directly exhibited. O u r only evidence

Collapse of the traditional theory of meaning 131

tha t a categorial act o f a particu lar type is involved is the fact tha t an expression o f a particu lar sem antic fo rm is being used.

So far o f course I have only dem onstra ted the failure o f the theory o f categorial synthesis in reg ard to one-place predicative sentences. T h e inadequacy o f the theory is particularly obvious in their case because such a sentence has only one object-about-which and it was necessary th e re fo re first to transfo rm them befo re one could speak o f a synthesis o f two objects at all. O ne m ight th ink that the theory can at least be sustained in the case o f relational statem ents, thus in the case o f many- place predicative sentences. W hen I first in troduced the theory in a general way 1 also gave relational sentences as exam ples. H ow ever let us now look at the m atter m ore closely.

Let us take the exam ple 1 have already used, the sentence ‘T his ham ­m er is com posed o f shaft and h ead .’ In my discussion of this exam ple I po inted ou t tha t the com position o f this state of affairs can be though t o f in two d iffe ren t ways. T h e one th a t seems to me to be logically co r­rect is as follows: in the state o f affairs the relation of real com position is ideally com bined with, on the one hand , the ham m er and , on the o th e r hand, with the object-pair shaft and head . In this conception a relational sta tem en t is trea ted as a m any-place predicative statem ent. T h e relation (in this case the real com position) is the object fo r which the nom inalization o f the m any-place pred icate (‘com posed o f ’) stands and thus corresponds to the a ttribu te in the case o f a one-place p red i­cative sentence. T h e ideal com bination of the relation with the real objects (the ham m er on the one h an d , the object-pair shaft and head on the o ther hand) thus corresponds exactly to the com bination o f the a ttribu te with the one real object in the case o f a one-place predicative sentence. B ut then this conception is open to precisely the sam e objec­tion as the previous one: asked to give a criterion for the presence o f this ideal com position one can only reply that it obtains between the real com position and the objects if the original sentence is true , in o u r exam ple: if the ham m er is com posed o f shaft and head.

As I have already po in ted out, H usserl p re fe rre d an o th e r concep­tion, according to which in a relational sta tem ent it is only the real objects which are synthesized in the categorial act; the real relation on the o ther h and is so to speak incorporated into the categorial synthesis. T h u s the categorial act will be a d iffe ren t one d epend ing on the type o f rela tion concerned . T his view seems to me to be un tenable . T h e re is no reason to reg a rd any relation betw een two real objects as not a real relation. H usserl how ever th ough t th a t to the d iffe ren t real relations th e re co rrespond d iffe ren t ideal relations. Consistently app lied this

Analysis of the predicative sentence 132

idea would resu lt in a duplication o f all types o f relation. ‘In the fo r­m ation o f ex ternal relations a sensuous form may serve as the fo u n d a­tion for the constitution o f a co rrespond ing [!] categorial form ; as when we ap p reh en d , an d p erhaps express, th e sensible adjoin ing o f contents A and B given in the percep tion o f a com prehensive whole G in the synthetic form s “A adjoins B ” o r “B adjoins v4” W ith the constitution o f the la tter form s, however, th e re arise new objects belonging to the class “states o f affairs” ’ (Investigation VI § 48). F rom th e fact tha t the state o f affairs tha t A adjoins B is an ideal object H usserl mistakenly infers tha t the relation o f ad jo in ing expressed in the two sentences is itself an ideal relation. T h e only ideal relation involved is the relation between the real relation o f ad join ing and the object pair {A, B}; we would thus be back with my original conception which leads to the sam e difficulty as arose in the case o f one-place predicative sentences.

L ater on I will exam ine an o th er aspect o f H usse rl’s theory o f cate­gorial synthesis which concerns the m ean ing o f the words ‘a n d ’ and ‘o r’. B ut o u r first task m ust be to work ou t a new conception o f predicates which does no t trea t them as standing fo r objects.

L E C T U R E 1 1

Predicates: the first step in the development of an analytical conception of the m eaning of sentences. The dispute between nominalists and conceptualists

T h e object-o rien ta ted conception o f the m ean ing o f predicative sen ­tences fo u n d e re d on the question o f how th e m ean ing o f the w hole sen tence results from the m eanings o f the sen tence-com ponen ts. T h e only answ er the ob ject-orien tated position could give was: the m ean ing o f the w hole sen tence is com posed o f tha t fo r which the singular te rm stands an d th a t for w hich the pred icate stands. T h is answ er led to the dilem m a: e ith e r the com position m ust be co n stru ed as th e real com po­sition o f a com plex object o r one canno t say w hat is to be u n d ersto o d by com position h e re w ithout p resu p p o sin g precisely th a t u n d e rs ta n d ­ing o f the sentence which was to be exp lained .

T h is resu lt is no t pu rely negative, inasm uch as it prescribes a specific d irection o f enqu iry fo r a new, no longer ob ject-o rien ta ted a ttem p t at an exp lanation . Firstly, u n d e rs ta n d in g o f the p red icate has em erged as the - from an ob ject-orien tated p o in t o f view - critical e lem ent in the u n d e rs ta n d in g o f a p redicative sentence. W e will thus first have to try to achieve a new an d no longer ob ject-orien tated conception o f the u n d e rs ta n d in g o f a predicate . Secondly, it has a t the sam e tim e becom e clear th a t the p rob lem o f the u n d e rs ta n d in g o f a p red ica te - the second o f my fo u r questions (see p. 103) - is directly connected w ith o u r th ird question , viz. how we u n d e rs ta n d th e com bination o f th e singular te rm with the pred icate . It would th e re fo re seem plausible th a t these two questions should now be com bined. T his gives a concrete clue to the enquiry in to the u n d e rs ta n d in g o f a p red icate . W ere we sim ply to fo r­m ulate the question concern ing the u n d e rs ta n d in g o f th e p red icate thus: w hat is it to u n d e rs ta n d a p red icate if this u n d e rs ta n d in g can n o t consist in th e consciousness o f an object? we w ould have no positive clue as to how we shou ld proceed . If, on the o th e r h an d , we com bine the second question with the th ird an d hold fast to the idea th a t a t any ra te the singu lar te rm stands for an object, we can ask: if th e su p p le m e n ta ­

Analysis o f the predicative sentence 134

tion o f the s ingu lar te rm by a p red icate does not have the function of com bin ing the object o f the singular te rm with an o th e r object (that of the predicate) how then is it to be understood? Clearly we canno t now look fo r som eth ing o th e r than an object with which the object o f the singu lar te rm w ould be com bined, fo r w hatever is com bined with som e­th in g is som eth ing and , hence, an object. W hat we have to abandon is precisely this idea o f a com bination o r synthesis. So we have to ask: if the su p p lem en ta tio n o f a singular te rm by a p red icate does no t have the function o f com bining the object fo r which the singular te rm stands with som eth ing else then w hat is its function?

W e can h e re link o u r discussion to a concept which was already p rese n t in the trad itional theory o f th e pred icate . T h e only aspect o f the trad itional theo ry I have so far considered is the idea tha t the p re d ­icate stands fo r an object. I have not gone into the question o f what that object was conceived to be. I t is only in passing th a t I have called these objects designated by nom inalized pred icates ‘a ttr ib u tes’: we find in the trad itio n , in ad d itio n to ‘a t tr ib u te ’, a series o f o th e r designations such as ‘species’, ‘un iversa ls’, ‘concep ts’, ‘p ro p e rtie s’. T h e designation ‘univer- sals’ expresses the idea o f ‘general objects’ which can belong to any n u m b e r o f ind iv idual objects. It is this ‘belonging’ which is expressed in the designation ‘a ttr ib u te ’. ‘Species’, th e designation p re fe rre d by H us­serl, is the Latin transla tion o f the G reek eidos (‘look’, ‘ap p earan ce’); this w ord con tribu tes little to th e characterization o f th e objects in question except p e rh ap s this: tha t they are objects o f an intellectual in tuition. T h e w ord ‘co n cep t’ occupies a special position, for one hesitates to re g a rd concepts as objects. So already in trad itional philosophy the re seem s to exist, in the idea tha t predicates stand fo r concepts, an ap p ro ach which leads away from the object-orien tated conception; the te rm ‘concep t’ should th e re fo re be specially investigated .1 Finally, the designation o f universals as ‘p ro p e rtie s’ can be understo o d as a fu rth e r specification o f th e ir designation as attribu tes: it is characteristic o f u n i­versals th a t w hen they a re ascribed (‘a ttr ib u ted ’) to a real object it ap p ears as characterized by these attribu tes; the a ttr ib u te is its ‘quality’ o r ‘p ro p e r ty ’. H ow ever, this designation is insufficiently co m p re h en ­sive. A n a ttrib u te such as redness is a p ro p erty o f the castle bu t we would not call the a ttr ib u te o f being a castle a quality o f this building. ‘T his is a castle’ answ ers th e question ‘W hat is this?’ w hereas one only calls qual­ities those characteristics o f objects which rep re se n t answers to the question: ‘H ow is this qualified?’ W hat is m ean t then by calling un iver­sals ‘p ro p e rtie s ’ is simply th a t the object is som ehow characterized by

Predicates 135

them , in w hatever respect this may be. For this reason it would seem be tte r to call an a ttribu te a ‘characteristic’ ra th e r than a ‘p ro p erty ’.

I f the object for which a p red icate stands is conceived as a character­istic then the object-orientated conception com es closest to the p resen ­tation o f the problem I have ju s t elaborated. For as a characteristic tha t fo r which the pred icate stands is viewed in its relation to the object o f the singular term . I had asked: w hat function has the supp lem entation o f the singu lar term by a predicate? Using the concept o f a character­istic it is possible to give an answ er to this question that still conform s to the object-orientated approach . Such an answ er would be: the supp le­m entation o f the singular term by the pred icate has the function of characterizing the object o f the singu lar te rm and it does this by com ­bining this object with a characteristic. Now we were supposed to d ro p the idea o f a com bination with som ething. I t was fo r this reason th a t I fo rm u lated the question thus: if the supp lem entation o f the singular te rm by a p red icate does not have the function of com bining the object o f the singu lar term with som ething then w hat function does it have? We can derive an answ er to this question directly from the object-orien­tated account ju s t given by d ro p p in g the specifically ob ject-orien tated addition . Instead o f saying: the supp lem entation o f the singular te rm by a p red icate has the function o f com bining the object with som ething an d thereby characterizing it we can say: the supplem entation by the pred icate does not have the function o f com bining the object o f the singular te rm with som ething, but instead th a t o f characterizing it.

With this we en co u n te r a new and no longer object-orientated thesis abou t w hat it is to u n d e rs ta n d a predicate. A ccording to this new con ­ception th e function o f the p red icate is not to stand fo r som ething, bu t ra th e r to characterize som eth ing (the object o f the singu lar term ) an d hence to u n d ers ta n d the p red icate is to u n d erstan d its characterization- function. A sentence such as ‘T h e castle is re d ’ is no longer to be explained by saying th a t the p red icate stands for a characteristic (re d ­ness) which is synthesized with the object, b u t ra th e r by saying tha t the object - th e castle - is characterized in a specific m an n er by the p re d i­cate ‘is r e d ’.

I can see th a t you will raise a whole series o f fundam en tal objections and queries. Firstly, you m ight ask, with w hat righ t do I suddenly s ta rt speaking o f a function o f a linguistic expression. Secondly, you will w ant to know w hat precisely is m ean t by a ‘characterization -function’. T h ird ly , you would be justified in asking how far my suggestion is an alternative to the ob ject-orien tated account at all. T h a t th e object o f the

Analysis o f the predicative sentence 136

singular te rm is characterized by the p red icate is, you will say, obvious and alm ost trivial. B ut doesn’t this simply say less th an the object-orien- ta ted conception? T h e la tter also allows one to speak o f a characteriza­tion o f th e object o f the singular term by the p red icate , bu t instead o f simply leaving it a t th a t it also provides an explanation o f how this ch a r­acterization is achieved, viz. by the com bination with a characteristic for which th e p red icate stands. Only if it becam e clear how the new exp la­nation com pensates fo r this a p p a ren t lack could it really claim to be reg a rd ed as an alternative to the object-orientated explanation . In this lecture I want to deal in tu rn with these th re e question-com plexes.

So, firstly, with w hat righ t do I ask fo r the function o f a linguistic expression? In so do ing I seem to be sm uggling in a new idea which has no t arisen from my critical in te rp re ta tio n o f the ob ject-orien tated approach . H ow ever, I would ask you to consider th a t ou r task m ust now be th a t o f find ing an alternative, in the specific case o f predicates, to the ob ject-orien tated in te rp re ta tio n o f linguistic expressions. We thus find ourselves throw n back to th e sta rting -po in t o f H usserl’s sem antics and m ust so to speak take an o th e r step beh ind this starting- point so th a t the a lternative possibilities can reveal them selves. H usse rl’s sta rting -po in t was th e ‘con fe rring o f m ean ing’ by the use o f signs (above p. 107). W e saw how H usserl im m ediately in te rp re ts this ‘co n fe rrin g o f m ean ing ’ as an ‘act’ and thus as consciousness o f an object. I f we go beh ind this first m ove o f H usserl we arrive at the m ore general p e r­spective o f the use o f signs. Because from the ou tset H usserl so con­ceived the use o f a sign th a t a sign can only be used to stand fo r an object, th e underly ing concept o f the use o f a sign could not as such becom e salient fo r him . Now this concept o f sign-em ploym ent has, as a basic concept, the advantage th a t it still encom passes the object-orien­tated approach bu t a t th e sam e tim e opens up alternatives to it. It p e r­mits the view tha t all o r som e signs are used to stand fo r an object. On the o th e r hand , how ever, we can now ask: fo r w hat o th e r purposes are signs used? A nd one possible answ er would be: to characterize. Now th a t fo r which som eth ing is used is w hat we call its function .

Notice tha t in this new way o f looking at things signs assum e an im portance which they did no t have in the ob ject-orientated concep­tion. For the la tte r th e sign is a m ere in te rm ed iary betw een conscious­ness an d object. It has the function of m aking p resen t to consciousness the object fo r which it stands. B u t consciousness can also be conscious o f the sam e object th a t is m ade p resen t to it by the sign w ithout the sign. T h u s the whole theory of categorial acts can be carried th rough purely as a theory o f th ink ing w ithout having to re fe r to signs. T h e situation is

Predicates 137

com pletely d if fe re n t as soon as one considers o th e r functions o f signs. For exam ple, if th e function o f a sign consists in characterizing , th en this function o f characteriz ing is d e p e n d e n t on a sign. H ere the sign does no t serve as a m ere in s tru m e n t fo r som eth ing which could also be achieved w ithou t it.

Now if the function o f a th in g is w hat th a t th in g is used fo r then the no tion o f th e function o f a sign is closely connected with th a t o f the employment o r use o f a sign. W hen we ask w hat th e function o f som eth ing is we p resu p p o se tha t it belongs to a con tex t o f pu rposive action. I f we en q u ire as to the function o f som eth ing , e.g. a h am m er, we a re asking: fo r w hat p u rp o se is it norm ally used? A nd this question refers to a p articu la r h u m a n activity (e.g. the knocking in o f nails). So the ques­tion: w hat is th e function o f a sign? is directly connected with the ques­tion: w hat is the no rm al use o f this sign? and th e la tte r question refers to the question: w hat is th e action o f which this use o f a sign is the (or a possible) condition? F or exam ple, if a p articu la r class o f signs has the function o f characteriz ing som eth ing this m eans th a t these signs a re norm ally used to characterize som eth ing an d th a t the action which som eone p erfo rm s w hen he uses a p red icate is th a t o f characteriz ing som eth ing as thus o r thus. All o f this is clearly also tru e in th e p articu la r case which th e ob ject-o rien ta ted conception reg a rd s as the only case. A sign with the function o f s tand ing fo r som eth ing is used to indicate which object one m eans. A nd the answ er to th e question: w hat is a p e r ­son do ing who uses a sign in this way? is th a t h e is ind icating which object he m eans.

W hen I discussed H u sse rl’s tak ing th e ‘co n fe rrin g o f m ean in g ’ as his s ta rting -po in t I po in ted o u t tha t the in te rp re ta tio n (Auffassung) o f a sign which confers m ean ing on it should really be called ‘u n d e rs ta n d in g ’ fo r we had ea rlie r seen th a t to ask for th e m ean ing o f a sign is to ask how th a t sign is to be u n d ers to o d . W e saw th a t H usserl im m ediately by­passed u n d e rs ta n d in g in favour o f th e in ten tional act. C onsideration o f an u n d e rs ta n d in g o f th e sign was fo r him as superfluous as considera­tion o f th e use o f the sign, although the ob ject-o rien ta ted conception can equally well be based on u n d e rs ta n d in g as on use: to u n d e rs ta n d a sign is to know fo r which object it stands.

Now how is th e u n d e rs ta n d in g o f a sign connected with its use? Clearly in this way: to u n d e rs ta n d a sign is to know w hat function it has o r how it is used. Now if we know o f som eth ing how it is used then this m eans tha t we know its em ploym ent-ru les. I f this is so then u n d e rs ta n d ­ing the function o f a sign m ust consist in know ing th e rules o f its em ploym ent.

Analysis of the predicative sentence 138

It thus em erges tha t by going back to the function o f linguistic expressions we arrive at a new com prehensive explanatory perspective o f which the following concepts are characteristic: function, use of som ething, ru le o f this use, ac tion-rule and an u n d erstan d in g tha t is an u n d ers ta n d in g o f a rule. A fter the discussion o f predicates we will have to a ttem p t to apply the sam e set o f concepts to singular term s. O n the basis o f the thus re-fo rm u la ted questions about singular term s and predicates we will be able to re tu rn , with m ore prospect o f success (though only a t the end o f the w hole lecture-series [Lecture 27]), to ou r th ird question - viz. w hat is it to u n d ers tan d the com bination o f a sin­g u la r term with a predicate? For we no longer feel com pelled to enquire in to a com bination o f w hat the singular term stands for with what the pred icate stands for. R a ther the question now acquires the sense: how is the em ploym ent-ru le o f the one sort o f expression connected with the em ploym ent-ru le o f the o ther so rt o f expression? T h e aim of the whole investigation is an answ er to th e question: what is it to u n d e r­stand a whole predicative sentence? It should now be possible to pu t this question thus: w hat is the function of such a sentence? or what are we doing w hen we em ploy a predicative sentence? H usserl’s answer was: w hat we are do ing w hen we use a predicative sentence is re p re ­senting , by m eans o f it, a state of affairs. O f course on the p resen t basis it is to be expected tha t we will arrive at quite a d iffe ren t sort o f answer.

B u t now let us re tu rn to predicates. T h e second question 1 expected from you was: w hat does it m ean to speak o f ‘characterization’? As I th ink this is a basic da tum o f our u n d erstan d in g in so far as we u n d e r­stand how to use predicates, I cannot define this w ord, m erely elucidate it. A pred icate fulfils its characterization-function by acting as a crite­rion. A criterion (from th e G reek krinein, to separate) is som ething which serves to d istinguish. In applying a p red icate to some objects but not to o thers we classify all those objects to which we apply it and at the sam e tim e distinguish them from those to which we do no t apply it. W hen we apply a p red icate to an object we declare it to be an object which is like the o ther objects to which we apply the pred icate and un like those to which we do no t apply it; and that m eans: we character­ize it as such an object. T h e characterization-function consists in this classifying-and-distinguishing.

T h e th ird question I expected from you can be directly linked to this elucidation. For you could now a rgue against me as follows. I f charac­terizing is a form o f classifying then fo r precisely this reason the objects characterized by a p red icate m ust have som ething in com m on. A nd do we no t then have to say th a t it is really this com m on som ething tha t

Predicates 139

characterizes the object and are we not thus led back to the objectual characteristic? T h e predicate, you will say, cannot itself function as a criterion, fo r the application o f the pred icate to ju s t these objects and not o thers requires a foundation in the objects, o r ‘m ark o f d istinction’; and indeed the characteristic was also so called in traditional philoso- phy.

Should this arg u m en t prove com pelling, then the object-orientated conception would be restored . I t would then have been shown tha t the functional conception does not o ffe r a genu ine alternative, fo r it is itself obliged to have recourse to the object-orientated conception: the p re d ­icate only characterizes the object by standing for a characteristic which itself characterizes the object in a prim ary sense. It is only now th a t the real weight o f the object-orientated conception shows itself. It does not simply arise from the philosophical prejud ice tha t every linguistic expression stands for som ething; ra the r it seems to be the only in telli­gible epistem ological explanation o f the use o f predicates. Vis-ä-vis my earlier reference to the priority o f the predicative form (‘re d ’) over its nom inalized m odification (‘redness’), the object-orientated philosopher could now argue tha t this priority tu rns ou t to be a m erely gram m atical one; epistemologically the un d erstan d in g o f the pred icate is fo u n d ed in the rep resen ta tion o f the co rrespond ing characteristic. My accusa­tion of a hysteron-proteron would thus rebound against my own exp lana­tion.

So if this functional explanation is to rep resen t a genu ine alternative to the object-orientated one then its real con ten t m ust be deeper than anything suggested by my account so far; and it m ust be in term s o f this d eep er con ten t tha t the functional explanation is able to answer the object-orientated counter-criticism ju s t p resen ted . F rom the po in t o f view of m ethod it would seem reasonable to start the p resen tation o f the new approach by argu ing with this criticism, for such a co n fro n ta­tion should disclose the real substance o f the functional explanation . I will conduct this confron tation in the form o f a dialogue between the two positions. Such a dialogue should bring us step by step to the rea l kernel o f the problem .

T h e functional explanation fits into the so-called nom inalist trad ition according to which th e re are no general essences for which predicates stand; and the only objects given to us in u n d erstan d in g these signs a re the signs them selves, the nomina. T h e contrary position has som etim es been called realism, som etim es conceptualism ; the fo rm er when the theory has been understood in a m ore ontological sense (predicates stand for real objects), the latter w hen it has been understood in a m ore

psychological o r epistem ological sense. T h e traditional ontological cri­tique o f realism by nom inalism is no t the one I p resen ted in my last lecture, which concerned the p rob lem o f how the object o f the p re d i­cate is connected with th a t of the subject; ra th e r it re la ted to the o th e r p roblem o f how one an d the sam e general object can sim ultaneously be in m any concrete objects. T h e debate into which we m ust en ter (in view o f the counter-criticism ju s t raised), on the o th e r hand , is no t prim arily an ontological debate b u t a psychological-epistem ological one.

A ccording to the conceptualist thesis one can only u n d erstan d the characterization-function o f a p red icate if the em ploym ent o f the p re d ­icate is connected with the rep resen ta tion o f som ething fo r which the predicate stands. For if this w ere no t so the pred icate would have no objective foundation ; it would be arbitrary .

Let us first allow th e nom inalist to m oun t his counter-attack . H e will deny tha t we in fact always rep re se n t som ething w hen we significantly em ploy a p redicate. T ake , e.g., the sentence ‘H eide lberg Castle is re d .’ I f we m ake this s ta tem en t in the percep tua l situation, thus in a situation in which we perceive the castle an d perceive th a t it is red , then clearly we do so on the basis o f a particu lar co lour-rep resen tation . B ut suppose we u tte r such a sentence h ere in th e lecture-room , for exam ple, w here we cannot see the castle; we can have a co rrespond ing co lou r-rep resen ­tation in o u r im agination , bu t clearly we can also u n d e rs ta n d the sen­tence w ithout having any such rep resen ta tio n co rrespond ing to the w ord ‘re d ’.

T his first attack by the nom inalist can easily be bea ten o ff by his opponen t. I t tu rn s o u t to be a m isunderstand ing , for it th o u g h t o f the rep resen ta tion o f som eth ing for which the pred icate is supposed to stand as a sensuous rep resen ta tion . T h a t we can use a pred icate with u n d ers tan d in g w ithout having sensuous im ages is som eth ing the con ­ceptualist will im m ediately concede. H usserl h im self showed this in im pressive fashion, in C hap ter 2 o f Investigation I. B u t it is not a ques­tion o f sensuous rep resen ta tions, the conceptualist will say; the charac­teristic o f redness is som ething which is com m on to m any things, it is redness in general.

So this was a m isunderstand ing . H ow ever, it was no t an unproductive m isunderstanding . F o r it is now clear tha t th e only kind o f rep rese n ta ­tions in question a re non-sensuous rep resen ta tions. Previously the nom inalist had asserted that u n d e rs ta n d in g a p red icate is not always accom panied by a sensuous rep resen ta tion ; h e will now claim tha t th e re simply are no rep resen ta tions o f the kind claim ed by the conceptualist. T h e suspicion arises tha t, a lthough the com m on characteristic is n o t an

Analysis o f the predicative sentence 140

Predicates 141

object o f sense-percep tion , this represen ta tion is still being th o u g h t o f on the m odel o f sense-percep tion . T h e common characteristic, because it is som eth ing general, canno t be sensuously rep resen ted ; bu t it can nonetheless be rep resen ted . B ut how? T h a t sense-percep tion h ere serves as the m odel seems particu larly clear in H usse rl’s case, fo r he says: the general object is given to us in an in tu itio n o f essence (Wesen­sanschauung). A co rresp o n d in g trad ition , accord ing to which we re p re ­sen t these general essences in an intellectual in tu ition (nous), has existed since Plato. T h e nom inalist can leave und ec id ed how far his o p p o n en t th inks o f the general characteristic o n the m odel o f sense-percep tion (though one may well th ink th a t u ltim ately he has to th ink o f it in this way). His a rg u m e n t will now be tha t in any event we do no t en co u n te r such a rep resen ta tio n o f som eth ing general th a t we recognize in an object. Let us assum e, w hen we say o f th e castle th a t it is red , th e most favourab le case, viz. tha t o f percep tion ; we have a particu la r sensuous co lo u r-rep resen ta tio n , b u t do we have, in add ition to this re p re se n ta ­tion (or fo u n d ed in it), a fu r th e r non-sensuous rep re se n ta tio n th ro u g h which we a p p re h e n d th a t com m on characteristic - the redness - in the object?

T h is question is m ean t to hit the o p p o n e n t a t precisely tha t po in t which he h im self em phasized as the decisive one. We canno t, he had claim ed, recognize an object as one characterized by a p red ica te unless we can d iscern in it a characteristic fo r which th e p red icate stands. A nd now it is objected to him th a t we sim ply canno t find such a re p re se n ta ­tion o f a characteristic. T h is b rings o u t a pecu liarity o f his a rg u m en t which should be noted: his insistence on the ap p re h en sio n o f a charac­teristic which m ust u n d erlie the charac teriza tion -function o f the p red i­cate was no t a discovery b u t a postu late . It was n o t show n th a t it is so; ra th e r it was a rg u ed th a t it m ust be so. T h is arouses the suspicion th a t again it is only the ob ject-orien tated p rejud ice tha t is at w ork here , not, adm itted ly , in the sim ple fo rm o f sta ting in advance th a t every sign m ust stand fo r som eth ing , b u t ra th e r in the fo rm o f an inability to th ink o f th e characteriza tion-function o f p red icates in any o th e r way. O ne can n o t th ink o f it d iffe ren tly ; th e re fo re it m ust be so. B ut if one canno t ascertain th a t it is so th en th e re w ould seem to be som eth ing w rong wich on e’s p resuppositions.

B u t at the stage the d eb a te has now reached , th e streng th o f the con­cep tualist vis-ä-vis the nom inalist consists in th e fact th a t a lthough the conceptualist canno t exh ib it the non-sensuous rep resen ta tio n he pos­tu lates, the nom inalist fo r his p a rt has no t yet given a positive account o f how the characteriza tion -function o f p red icates is to be u n d ersto o d .

Analysis o f the predicative sentence 142

So long as no alternative account is available the conceptualist account, how ever hypothetical it may be, reta ins the advantage.

T h e nom inalist m ust th e re fo re now be pressed into giving his own positive account. T h e weakness o f trad itional nom inalism was that its s treng th lay only in its critique o f th e opposing position. Only since W ittgenstein’s Philosophical Investigations has th e re been a positive expla­nation o f how one can u n d e rs ta n d the characterization-function o f p redicates th a t does not involve p ro p p in g up this function with objects.2

1 shall take as my p o in t o f d e p a r tu re a general rem ark about the m ean ing o f linguistic expressions which is found in §560 o f the Investi­gations: ‘-“T h e m eaning o f a w ord is w hat is exp lained by the exp lana­tion o f the m ean in g ” i.e. if you want to u n derstand the use o f the w ord “m ean in g ”, look fo r w hat are called “explanations o f m ean ing” ’.

A t the beg inn ing o f th e Blue Book, w here a sim ilar rem ark occurs, W ittgenstein explains its significance by saying th a t it is in tended to b ring the question ‘W hat is m eaning?’ ‘down to e a r th ’. It is in ten d ed to free us from th e com pulsive idea th a t m eaning m ust be an object. T h e sense o f this rem a rk is th e re fo re on a level with my earlier suggestion th a t instead o f asking fo r the m ean ing o f an expression we should ask how we u n d e rs ta n d it.

H ow is th e understanding o f m ean ing of which I had spoken con­nected with the explanation of m ean ing re fe rre d to in the quotation from W ittgenstein? U n d erstan d in g an d explain ing are frequently used as opposing concepts, as fo r exam ple in the fam iliar con trast between in te rp re ta tiv e (verstehend) an d exp lanato ry psychology. H ow ever, the w ord ‘exp lain ’ is used in two d iffe ren t senses. O ne can say ‘Explain to me why that is thus and so’: exp lanation in the sense of the giving o f reasons. I t is only this k ind o f explanation th a t can be opposed to u n d e rs ta n d in g o r describing. Explaining-w hat o r -how differs from this explaining-w hy: ‘Explain to m e how th a t w orks’, ‘explain to me what tha t m eans’, ‘explain to m e the m ean ing of the expression’. It is expla­nation in this second sense tha t W ittgenstein is re fe rrin g to. T h e re la­tionship o f this kind o f exp lanation to u n d erstan d in g is as follows: som eone who explains som eth ing shows what he un d erstan d s or how he u n d e rs ta n d s som eth ing . I f I ask som eone to explain to m e how a m achine works I assum e th a t he u n d erstan d s how it works and how one opera tes it; an d the exp lanation is successful if it results in my u n d e r­stand ing how it is o p era ted . C orrespondingly , if I ask som eone to explain the m ean ing o f a linguistic expression to m e I assum e tha t he un d erstan d s the expression; and the explanation is successful if it

Predicates 143

results in my und erstan d in g the expression. T hus we can say: to explain (in this sense) m eans to show w hat one understands, to com m unicate an understan d in g , to m ake som ething understood .

If now we apply W ittgenstein’s rem ark to predicates we m ust say: the m eaning of a p red icate is w hat we explain w hen we explain its m eaning. A nd how do we explain the m eaning o f a predicate? If to u n d erstan d a p red icate is to know how it is used to characterize, i.e. to classify and distinguish, then the explanation o f the m eaning o f a predicate m ust consist in explaining how it is used to classify and distinguish. How can this be done, e.g. in the case o f the pred icate ‘re d ’? Clearly not by po in t­ing to the general characteristic o f redness, for, as the conceptualist rightly em phasized, this is supposed to be a general essence and hence no thing which can be simply po in ted to. T his general essence simply plays no p art in the explanation o f the m ean ing of a predicate. In real­ity we explain the m eaning o f a pred icate - if we cannot explain it by m eans of o th e r w ords, by a definition - by m eans o f exam ples. We p resen t the person to whom we wish to explain the m eaning o f the word ‘re d ’ with objects which we characterize as red (‘tha t is r e d ’) and others o f which we deny the p red icate (‘tha t is no t re d ’). T h e positive exam ples show how th e pred icate classifies and the negative ones show from w hat it d istinguishes th a t which it classifies. W hat we show in this way is how the pred icate is used. We can then ascertain w hether the person to whom we have explained the use o f the pred icate has u n d e r­stood the explanation by getting him to use the p red icate himself. A nd if he then uses it d ifferen tly from w hat we in tended we correct him by m eans o f such expressions as ‘co rrec t’ and ‘no t co rrec t’ until he has understood us. W hat we explain to him by m eans o f the exam ples is thus the employment-rule o f the predicate. F or an activity which in all its stages is regu la ted by ‘correct’ an d ‘incorrec t’ is an activity which follows a rule, even if the ru le cannot be fo rm ulated in words. T h e ru le shows itself only in its co rrec t use, and in o u r case this m eans: in the correct application o f the p red icate to exam ples.

We thus arrive at the following result: if the m eaning o f a pred icate is simply w hat we explain when we explain the m eaning o f the p re d i­cate, then the m ean ing o f the pred icate canno t be identified with the com m on feature; ra th e r the la tte r does no t figure in the explanation of m eaning at all. T h e factor which conceptualism held to be essential proves to be superfluous in the explanation o f m eaning. T h e rejection of the rep resen ta tion o f a general essence is now no longer m erely neg­ative. R a ther a new positive conception has taken its place. I f he is

Analysis o f the predicative sentence 144

asked: how can a m ode o f em ploym ent o f a p red icate avoid being arb i­tra ry if it is w ithout a foundation in objects? the nom inalist can now po in t to the ru le obtained by applying the p red icate to exam ples.

T h e conceptualist could try to relativize this resu lt by calling in ques­tion the presupposition , taken over from W ittgenstein, th a t the m ean ­ing o f a linguistic expression is simply what we explain w hen we explain its m eaning. E xplanation, he could say, is sim ply an intersubjective com m unication of understand ing . I t does no t follow from this tha t w hat we u n d ers ta n d can be identified with w hat we can explain.

A nd indeed this does not necessarily follow. How ever, the nom inalist will now give the conceptualist the following points to consider: (1) L in­guistic signs do as a m atte r o f fact belong to intersubjective com m uni­cation. As a m atte r o f fact the individual only acquires a language in this way. Is it no t then superfluous to posit, in addition to this u n d e r­s tanding tha t one can explain, make understood, an o th er special inner- subjective understand ing? (2) T h e po in t previously reached in the debate was th a t the conceptualist was not able to dem onstra te the exis­tence o f his postu lated rep resen ta tion of a general essence and dem anded from the nom inalist an alternative positive explanation. T h is has now been provided , at least fo r that m eaning which is in tersub- jectively com m unicated . T his explanation which is actually valid for intersubjectively com m unicable m eaning is, then , a possible exp lanation o f the u n d erstan d in g o f m eaning in general. A nd as the conceptualist had no dem onstrab le alternative to o ffer fo r inner-subjective u n d e r­standing he o u g h t now to accept this explanation-possibility exhibited in intersubjective explanation as holding also for inner-subjective un d erstand ing . A nd this, the nom inalist would say, seems all the m ore plausible if we consider th a t each individual, if he wishes to becom e clear about his u n d erstan d in g of a predicate, m ust em ploy the sam e m ethod he uses when explain ing the em ploym ent-ru le o f the predicate to som eone else. I f one wishes to get clear abou t the m eaning with which one is using the w ord ‘red ’, even if th e re were such a thing as th a t general rep resen ta tion of redness, it w ould be useless to appeal to it: even to oneself one can only explain (m ake clear) one’s own u n d e r­standing o f a pred icate by elucidating to oneself by m eans o f exam ples how one uses it. We canno t say tha t we only u n d erstan d w hat we are able to explain; we can, however, say that we only u n d erstan d clearly w hat we are able to explain, and tha t we can only becom e clear about w hat we u n d erstan d by explaining it.

We will no t be able to leave the debate betw een nom inalism and con­ceptualism in this state. T h e conceptualist can still get ro u n d this last

Predicates 145

arg u m e n t o f the nom inalist. B u t before passing final ju d g m e n t in the controversy we will first have to get c lea re r abou t the m ethodological significance o f the new perspectives in tro d u ced by th e nom inalist.

Appendix on speaking of concepts

W hen listing the various trad itiona l designations o f w hat p red icates s tand for p. 134) I also m entioned a term inology which one m ight th in k could p o in t beyond the ob jec t-o rien ta ted fram ew ork : th e term ino logy acco rd in g to which pred icates stand fo r concepts. T h is way o f speak ing seem s to avoid two difficulties to which H usserl’s concep tion gave rise. Firstly, it is n o t c lear th a t a co ncep t is an object. Secondly, if one speaks o f concepts this no lo n g er suggests th e idea o f a com po­sition. W e do no t say th a t th e object fo r w hich th e sin g u lar te rm stands is com­bined w ith a concept; ra th e r we say th a t it is subsumed u n d e r it, th a t the objectfalb under the concept.

So it m ig h t seem th a t H u sserl’s concep tion o f p red icates is n o t a t all re p re se n ­tative o f th e trad itio n , in d eed th a t in choosing H usserl I h a d picked a p a rticu ­larly weak rep resen ta tiv e o f the trad ition . A n d o f course if this w ere so then m erely d e m o n s tra tin g th a t H u sse rl’s concep tion can n o t w ithstand critical analy­sis w ould be no cause fo r ab an d o n in g the e n tire trad itio n a l position an d ad o p t­ing a specifically language-analy tical conception .

W hat th en is th e significance o f speak ing o f concepts? K an t’s concep tion can p e rh ap s be re g a rd e d as exem plify ing the trad itio n a l u n d e rs ta n d in g o f concepts. In his Lectures on Logic § 1 K ant defines ‘co n cep t’ as ‘g en era l re p re se n ta tio n ’ and in b rackets adds: repraesentatio per notas communes. C o rrespond ing ly , in th e Cri­tique of Pure Reason, he says a concep t is a ‘re p re se n ta tio n ’ which in co n trast to in tu ition rela tes to an object no t im m ediately b u t ‘m ed ia te ly ’ ‘by m eans o f a fea­tu re w hich several th ings may have in co m m o n ’ (B377 cf. also B93 f.). T h u s a concep t is u n d e rs to o d by K an t a n d in early m o d e m ph ilosophy generally as a species o f re p re sen ta tio n (G erm an: Vorstellung. Latin: repraesentatio. English: ‘idea’). In early m o d e m philosophy th ere is a d d e d to th e fu n d am en ta l d ifficul­ties involved in speak ing o f ‘re p re se n ta tio n s’ the am biguity o f the term as betw een re p re se n tin g a n d re p re se n ted , i.e. betw een the sta te o f consciousness and the object a p p re h e n d e d by it, its objective co rre la te . T h is am biguity also m akes it u n c lear w hat is m ean t by speak ing o f a ‘co n cep t’ (conceptus). T h e defi­nitions o f K ant which have ju s t been given, how ever, a re relatively u n am b ig u ­ous. T h ey decide th e am biguity in favour o f the subjective m ean in g (concipere, conceive) fo r, in th e defin ition th e objective m ean in g is again expressly em p h a­sized as som eth ing to which th e concept relates: nota communis. B u t now this objective co rre la te o f th e subjectively in te rp re te d co n cep t - th e com m on m ark - is sim ply H u sse rl’s ‘species’, th e ‘a ttr ib u te ’ o f the o b jec t-o rien ta ted conception . Now th e subjective co rre la te o f th e a ttrib u te - th e re p re se n tin g or m ean in g (Mei­nen) o f th e a ttr ib u te — is clearly n o t a possible theoretical substitu te fo r the a ttr i­bute. I t w ould a p p e a r to m ake no p a rticu la r sense to say th a t the p red ica te stands

Analysis o f the predicative sentence 146

fo r the re p re sen tin g o f th e a ttribu te ; an d even if it d id m ake sense one w ould still be left with the unc larified no tion o f an a ttribu te .

In te rp re te d in this way th en the concept-term inology does no t rep resen t an a lternative to the ob ject-orien ta ted conception; the im pression o f a non-objec- tual concep tion results sim ply from th e am biguity o f this term inology.

In F rege, how ever, we find, in co n trast to this earlie r m o d ern trad ition , a theory o f p red icates accord ing to which (1) p red icates stand fo r concepts (2) concepts a re explicitly u n d e rsto o d objectively (objektiv) and (3) concepts a re nonetheless sharply d istingu ished fro m objects (Gegenstände). A b rie f descrip tion o f the essential featu res o f his theory will have to suffice h e re .3

For F rege th e re are - a p a r t from syncategorem atic expressions - two kinds o f linguistic expression: ‘com ple te’ and ‘incom plete’. C om plete expressions a re (1) nam es (singular term s) an d (2) whole (assertoric) sentences. O f bo th we can say that they stan d fo r an object. I can h e re ignore the peculiarity in F rege’s concep­tion th at a whole sentence .also stands fo r an object (a tru th -va lue).4 Incom plete expressions a re expressions with one o r m ore gaps; they are thus ‘in n eed o f su p p lem en ta tio n ’. Exam ples are: ‘T h e b ro th e r o f . . . is a ho rse’. I f anincom plete expression is su p p lem en ted by a com plete expression (by a nam e) th e re results a n o th e r com plete expression , w hether it be a nam e o r a sentence, e.g. ‘the b ro th e r o f C h arles’, ‘C atalina is a horse’. All expressions which need su p p lem en tin g in this way Frege calls ‘functional expressions’. Predicates - e.g. ‘. . . is a h o rse ’ - thus constitu te a species o f functional expression; they are those expressions whose su p p lem en tation results no t in a nam e bu t in a sentence. A ccording to F rege functional expressions also designate som ething; how ever this is no t an object, fo r ‘object’ is d e fin ed as w hat is designated by a com plete expression . F rege calls w hat is designated by a functional expression a ‘fu n c tio n ’; and if the functional expression is a p red icate he calls the function a ‘concept’.

Functions in the w ider sense I can h e re d isregard . T h e im p o rtan t aspect o f F reg e ’s theo ry in this con tex t is the idea th a t a p red icate too stands fo r som e­th ing, b u t th a t this is n o t an object b u t a concept. F rege em phasizes that this holds fo r the p red icate precisely in its character as predicate. T h e concept is ‘essentially p red icative’.5 If, on the o th e r hand , one wants to say som ething about a concept th en one m ust designate the concept by m eans o f a ‘nam e’ which results fro m the nom inalization o f the predicate. B ut now a nam e can only des­ignate an object. If, th e re fo re , one speaks of a concept, if one says som ething about it, it m u st ‘first be tran sfo rm ed in to an object’.6

T h is object into which th a t which th e pred icate stands fo r is tran sfo rm ed is roughly th e sam e as H u sse rl’s ‘species’, the trad itiona l a ttribute . Frege thereby gets h im self in to the follow ing aw kw ard position: when speaking o f such an object individually he calls it a ‘concep t’ even though being an object it cannot be a concept. T h is leads to th e p a rad o x ‘th a t in my term inology expressions such as “the concep t F” do not designate concepts bu t objects’; ‘the concept horse is not a concep t’.7

In assessing F reg e’s th eo ry o f p red icates we can distinguish two com ponents,

Predicates 147

a syntactical one an d a sem antic one. T h e syntactical fo u ndation o f F reg e’s th e ­ory is his idea that a pred icate is an essentially incom plete expression, th a t it is to be u n d e rsto o d as essentially a sen tence-part, a sta tem ent-fragm ent. W ith this p ioneering step F rege p re p are d the way fo r the language-analytical conception. H e b reaks with th e trad itional idea, which H usserl took over, th a t a singular predicative sentence is com posed o f subject, copula and predicate, th a t both pred icate and subject are in d ep en d en t units, th a t each stands fo r an object and th at betw een the two objects fo r which they stand there m ust be a com bination- elem ent, the sem antic co u n te rp a rt o f the copula, a synthesis betw een the two objects. For Frege the copula no longer exists; w hat was called the copula is a p a rt o f the predicate.

T h is conception o f the syntax o f predicative sentences also enables F rege to break with the trad itiona l idea th a t the p red icate stands fo r an object. T h e pecu­liarity o f F rege’s conception finds expression in the fact that he applies to what the p red icate stands fo r and what he calls a concept, the sam e term s as he applies to the predicate: ‘n eed ing sup p lem en ta tio n ’, ‘u n sa tu ra ted ’. T h a t the concept is ‘essentially pred icative’ m eans th a t it is som eth ing incom plete, u n sa tu ra ted and fo r precisely this reason canno t be an object. Frege thus rem ains tied to the trad ition in so fa r as he holds fast to the idea that the p red icate too stands for som ething; what it stands for, how ever, is no t an object. T h e question is w hether this rep resen ts a g en u in e th ird possibility betw een the ob ject-orienta ted concep­tion and the language-analytical conception. I f the p red icate does no t stan d fo r an object then one w ould be rid o f the difficulties o f the ob ject-orien ta ted con­ception an d yet w ould no t simply be left with the view th a t it is all a m a tte r o f the sign and its em ploym ent-ru le .

Does such a th ird possibility really exist? T h e re is, to begin with, th e basic difficulty th at a sign is supposed to stand fo r som ething th a t is nevertheless not an object. T h e con trad ic tion which seems to be involved h e re is show n by the fact th a t F rege finds h im self having to m ake such statem ents as ‘T h e concept horse is no t a concep t.’ H ow ever I do not wish to pu rsu e this problem fu rth e r; I re fe r you instead to the instructive in te rp re ta tio n and criticism in Searle .8

F rege him self says: ‘L anguage h e re is in an aw kw ard situation which justifies d ep artin g from w hat is usual.’9 A ssum ing one is p re p are d to accept this, it still rem ains to ask (1) w hat positive in te rp re ta tio n are we to give to this som ething which is no t an object? and (2) how are we to u n d e rs tan d the relation in a p re ­dicative sentence betw een this som ething an d w hat the subject o f the sentence stands for? T hese two questions a re directly connected.

Frege calls that which a p red icate stands fo r, like that which a singular term stands for, its ‘re fe ren ce ’ (Bedeutung).10 I t is no t necessary fo r me h e re to go fu r th e r in to the problem s connected with this term inology .11 Suffice it to say that with reg ard to bo th predicates and nam es Frege distinguishes th e ir re fe r­ence from th eir sense, an d th a t in each case th e re ference is w hat the expression designates, stands fo r .12 Frege is th ere fo re able to call the concept the re fe ren ce o f a predicate . T h e question: w hat is one to u n d e rs tan d by a concept? can th e re ­

Analysis o f the predicative sentence 148

fore also be construed thus: what is it for a predicate to have a reference? Frege’s answer to this question can be inferred from his account of what it is for a pred­icate (concept-word) to have no reference: ‘Concept-words . . . which have no reference . . . are not such as combine contradictory elements - for a concept can very well be empty - rather they are such as have vague boundaries. In regard to every object it must be determinate whether it falls under the concept or not; a concept-word that does not satisfy this requirement in its reference has no reference.’13 This definition shows that for Frege a concept is a criterion by which objects are distinguished into those which fall under it and those which do not. That the question of what a concept is is so closely connected with the ques­tion of how we are to understand its relation to what a name stands for is the consequence of Frege’s idea that the concept is something essentially in need of supplementation. The relation o f ‘an object’s falling under a concept’ Frege calls ‘the fundamental logical relation’.14

Of course these answers to the two questions just referred to which can be inferred from Frege immediately give rise to counter-questions: (1) to speak of an object falling under a concept is to speak metaphorically. How are we to conceive this falling of something under a concept? (2) Even though we may understand in general what it means to speak of a criterion there still remains the question of how in a given case one can recognize a concept or how one can decide whether an object falls under a particular concept.

One can find no answer to these questions in Frege. For him they are episte- mological questions and as such no concern of logic. In the context of a funda­mental philosophical enquiry, however, we cannot rest content with this. State­ments about what is to be understood by the sense or the reference of an expression which do not tell us how we can know what sense or reference an individual expression has remain empty. In contrast to Frege, both Husserl and linguistic analysis give an answer to this question. According to Husserl one rec­ognizes the attribute in an act of intuition of essence; and against this the lan- guage-analytical position argued that one explains the application-rule of a predicate by means of examples. Frege’s theory of concepts does not offer a third possibility; it provides no answer at all.

One can, however, try to get clear about what Frege calls the ‘fundamental logical relation’. What does it mean to say that an object ‘falls under’ a concept? What does the corresponding notion of ‘subsumption’ mean? The metaphor of ‘falling under’ goes back to the terminology Aristotle used for the subject-pred- icate relation. The term for ‘to predicate’ - kategorein - means something like ‘say down upon’; correspondingly the correlative term ‘subject’ (,hypokeimenon) means ‘that which lies under’. There is a peculiar ambiguity in the way Aristotle uses this terminology: it is not clear whether he is referring to the linguistic expression or to something for which the expression stands. It was this ambigu­ity which concealed from Aristotle the later alternatives of nominalism and con­ceptualism. However, the expression being said down upon’ stems from the orientation towards the use of the expression. Just as in Greek one could use the

Predicates 149

expression ‘saying dow n u p o n ’ we speak o f th e application - o f the p red icate ! - to (.Anwenden auf). T h a t th e n o tio n o f ‘su b su m p tio n ’ a n d the falling o f objects ‘u n d e r ’ concepts seem s m ore plausib le th an the ob jec t-orien ta ted n o tio n o f a combination o f the object with an a ttr ib u te is d u e simply to th e fact th at th e fo rm er re fe rs back directly to the language-analytical exp lanation .

W e can now try to fo rm a m o re precise idea o f how th e no tion o f an object falling u n d e r a co n cep t re fers back to th a t o f the applicability o f a p red ica te to the object. In this I follow P. L o re n z e n .15 O n e can first m ake clear to o n ese lf that instead o f a p a rticu la r p red icate -ex p ressio n which on e uses in accordance with a ru le (which one has ex p la ined by m eans o f a defin ition o r by sam ples) one can always use o th e r expressions in accordance with the sam e rule. Sem antically what m atters is n o t the expression b u t th e em plo y m en t-ru le , the ch aracte riza ­tion-function . Now one can abstract from all pecu liarities o f the sign a n d re fe r to two d iffe re n t predicate-expressions - e.g. ‘r e d ’ a n d ‘ro u g e ’ - as the sam e predicate if they a re app lied in accordance with the sam e ru le. H ow ever, to avoid am bi­guities on e can as b e fo re reserve the w ord ‘p red ica te ’ fo r th e p red icate-expres- sion a n d use the w ord ‘co n cep t’ fo r the abstrac tion ju s t in tro d u ced . T h is w ord w ould th en be so d e fin ed th a t all p red icates th a t a re used in accordance with the sam e ru le re p re se n t the sam e concept. A n d F reg e ’s fu n d am en ta l logical relation of ‘fa lling u n d e r’ w ould th en be defined thus: an object falls u n d e r a concep t if a p red ica te which rep re sen ts this concep t applies to it (auf ihn zutrifft).

T h u s the questions to which no answ er cou ld be fo u n d in F rege w ould now be answ ered by g ro u n d in g the talk o f concep ts on that o f predicates.

T h e re is o f course a d iffe ren ce betw een this defin ition o f ‘concep t’ a n d F reg e’s concep tion . T h e defin ition ju s t given is a so-called ‘in tensiona l’ defin ition: acco rd in g to it two pred icates re p re se n t th e same co ncep t if an d only if they have the sam e m ean ing , a re used in acco rdance with the sam e application-ru le . F rege on the o th e r h a n d has a so-called ‘ex ten sio n al’ co ncep tion o f ‘co n cep t’. For him two concept-w ords stand fo r the sam e concept if a n d only if ‘th e c o rre ­sp o n d in g concept-ex tensions coincide’.16 T h u s fo r in stance the two pred icates ‘anim al with a h e a r t ’ a n d ‘anim al with k idneys’ rep re sen t, accord ing to the defi­n ition we have ju s t given, two d iffe re n t concepts, w hereas, on the basis o f F rege’s defin ition , they stand fo r one a n d the sam e concept. T h is d iffe ren ce how ever is n o t a fu n d am en ta l one; one can o p e ra te with an ex tensional defin ition o f ‘con­cep t’ in m uch the sam e way as one does with the in tensional defin ition . T h e iden tity -crite ria o f th e concep t w ould still re la te to the app lication o f predicates; only o n e would now have to say: two p red icates re p re se n t the sam e concep t if they app ly to the sam e object.

L E C T U R E 1 2

The basic principle of analytical philosophy. The dispute continued. Predicates and quasi-predicates

B efore I b rin g to an en d the in te rru p te d debate betw een the nom inalist and the conceptualist we should try to get a c learer grasp o f the m e th ­odological significance an d scope of the two decisive perspectives on which the nom inalist’s a rg u m en t rested . T hese perspectives constitute o u r first steps in the d irection o f a new, no longer ontologically-orien- ta ted sem antic conceptuality . T h e first o f these perspectives was con­ta ined in th e question concern ing the function o f a lingustic expression th a t was in tro d u ced at the beg inn ing o f the previous lecture, the second in W ittgenstein ’s d ic tum , which was only b ro u g h t in in the course of the d ispu te bu t is in fact fu n d am en ta l: T h e m eaning o f a w ord is what the exp lanation o f its m ean ing explains.’ In both cases it is a m a tte r o f p e r ­spectives which I had app lied specifically to o u r question concerning th e m eaning o f p red icates but which are in fact o f universal scope; fo r they concern the question o f the m ean ing o f all linguistic expressions an d thus reach beyond th e special controversy between nom inalism and conceptualism .

How far am I en titled to claim th a t these perspectives have som ething intrinsically com pelling about them and are n o t a rb itra ry alternatives to the ob ject-orien tated conception? O ne cannot dem and o f a new way of looking at th ings th a t it be intrinsically com pelling, m erely that it be m ore fu n d am en ta l than th e previous one and hence can at least not be called in question by th e latter.

I have a lready show n in the last lecture th a t this is the case with the functional conception vis-ä-vis th e object-orien tated conception. T h a t signs are used and used to p e rfo rm a particu lar function is no t den ied by the ob ject-orien tated conception, bu t p resupposed as obvious; and the only reason why this fea tu re is no t m ade them atic is because the ob ject-o rien tated p h ilo sopher sim ply takes it fo r g ran ted th a t the fu n c­tion of the sign is to stand for an object. B ut as soon as one explicitly

retraces tha t step the particu la r function o f standing for an object turns ou t to be m erely one possibility am ong others. O f course the object- o rien ta ted ph ilosopher may say tha t we cannot conceive o f any o ther function for a sign than th a t o f standing fo r an object. B ut we have already seen that, in the special case o f predicates at least, it is not d if­ficult to get him to adm it th a t a p red icate has the function o f character­izing the object fo r which a singular te rm stands. O f course he had to qualify this adm ission by saying tha t this is m erely a function that the pred icate also has and th a t it can only fulfil it by p e rfo rm in g its alleged basic function o f standing fo r an object: it characterizes the object o f the subject-term of the sentence by stand ing fo r an object - the attribu te- which itself characterizes the object in a p rim ary sense. W ith what w onderfu l capacities the trad itional ph ilosopher finds it necessary to endow his objects! A re we no t indulg ing in m ythology if we say of objects th a t they characterize o th e r objects? A nd this they are supposed to accom plish by ad h erin g to the o th e r objects, by being com bined with them . It already em erged in the H usserl-critique tha t the conception o f the m eaning o f a predicative sentence as com posite or - this always sounds better - as a ‘synthesis’ o f two objects - cannot be carried th rough . It now em erges th a t even if it could, it could no t m ake the function o f characterizing intelligible. In reality the object-orientated conception has here pro jected a function which is only intelligible as a function of signs back in to tha t s tru c tu re which was the only one avail­able to it.

T h e m ethodological significance o f the second perspective which is contained in W ittgenstein’s rem ark - ‘T h e m eaning o f a w ord is w hat the explanation o f its m ean ing explains’ - rem ained less clear in the last lecture. W hat is m eant by this rem ark is simply this: when philosophi­cally we ask abou t the m ean ing o f linguistic expressions we are asking w hat tha t is in general (‘as such’) abou t which we ask w hen pre-philo- sophically we ask about the m eaning o f an individual expression. T h u s W ittgenstein’s rem ark has fo r analytical philosophy a significance co r­respond ing to th a t which the question about ‘being as being’ has for ontology. Ju s t as ontology did not m ean by ‘being’ a m etaphysical con­struction bu t was asking w hat the beings with which we have to do pre- philosophically a re as beings, so too fo r analytical philosophy what is m ean t by ‘m ean ing ’ canno t be a m etaphysical o r scientific construct: w hen philosophically we ask how we use linguistic expressions we are asking about the sam e th in g we ask ab o u t when pre-philosophically we ask how an individual expression is used; only now it is a question o f m ode o f em ploym ent as such, in form al generality. O ne could th e re ­

The basic principle of analytical philosophy 151

Analysis o f the predicative sentence 152

fo re call W ittgenstein’s d ic tum the fundam en tal princip le o f analytical philosophy, i.e. the fundam en tal princip le o f th a t philosophy which conceives itself as a question about the u n d e rstan d in g o f o u r linguistic expressions and which seeks to win the conceptuality in which it poses this question from this question itself.

Now how is this second perspective connected with the first, viz. that o f the function o f linguistic expressions? T h e first perspective yielded the maxim : if you want to clarify the m eaning o f a form o f linguistic expressions then ask: w hat a re expressions o f this form used fo r ? T h e second perspective yields the maxim: if you want to clarify the m eaning o f a fo rm of linguistic expressions then ask: how a re expressions o f this fo rm used? T h u s both perspectives em phasize the m ode of em ploym ent o f an expression, bu t the second proves to be m ore fundam en tal and general; as opposed to the first perspective it does no t p resuppose that the u n d e rstan d in g o f a linguistic expression belongs to a teleological context, the contex t of an in tentional action.

I do not know w hat arg u m en t one could use to cast doub t on W itt­genstein ’s p rincip le so long as it is understo o d in this fundam en tal gen ­erality; fo r one would then have to detach the m ean ing which the words ‘m ean ing’, ‘u n d ers ta n d in g ’, etc., have in linguistic theory from the m eaning they have in the ir p re-theoretical em ploym ent. T h e re are con­ceptions of linguistic theory for which this is a possibility. A philosoph­ical semantics how ever would lose its p u rpose if this happened , fo r ph i­losophy only seeks to m ake explicit w hat we already u n d erstan d pre- philosophically. W ittgenstein’s p rincip le th e re fo re also lays dow n the limits o f a possible philosophical sem antics; and this also contains a first clue to the answ er to the question I left open, viz. how philosophical sem antics could be d istinguished from linguistic semantics.

So the object-orientated ph ilosopher could not evade W ittgenstein’s princip le either. H ow ever, he w ould im m ediately - and rightly - explain tha t w hen he says th a t every expression (or at least every ‘cate- gorem atic’ expression) stands for an object he also m eans tha t in each individual case w hen one is asked abou t the m eaning or the use or the explanation o f an expression one is to indicate th e object fo r which the expression stands. So from the po in t o f view o f m ethod we have the sam e situation as we had in the case o f the first perspective: a question is posed which had not been m ade explicit in the object-orientated tra ­dition but which the object-orientated ph ilosopher im m ediately accepts and m oreover im m ediately answers in an ob ject-orientated way, as though this w ere the only possible answer. W e how ever can ask: are th e re not o th e r possible ways o f explain ing an expression’s m ode of

The basic principle o f analytical philosophy 153

em ploym ent? In the particu la r case o f p red icates it seems obvious tha t we canno t explain the use o f an expression in the way the object- theorist claims we can (cf. p. 143).

Now one could object that in today ’s elucidations I have unjustifiably sim plified W ittgenstein’s princip le. Did we no t already see in the p re ­vious lectu re tha t this princip le by no m eans m erely ties the ph ilo soph­ical analysis o f m ean ing to the answ ering o f individual questions of m eaning but, in its stress on the explanation o f m eaning , reduces m ean ­ing to the intersubjectively available m ode o f em ploym ent o f the expression? A nd if the princip le is u n d ers to o d in this way it clearly canno t be acknow ledged by the ob ject-o rien ta ted ph ilosopher as being m erely trivial. W ittgenstein did in fact u n d e rs ta n d his princip le in this n a rro w er sense an d d ispu ted the possibility o f an in trospective fixing of m eanings. H ow ever W ittgenstein’s den ial o f the possibility o f a so-called private la n g u ag e1 rests on a rg u m en ts th a t are d isp u ted in analytical ph ilosophy .2 I f we are to reg a rd W ittgenste in ’s p rincip le as the fu n d a ­m ental princip le o f analytical ph ilosophy then a t the p resen t stage o f ou r reflections w here we are only getting a foo ting in analytical philos­ophy we should no t b u rd en it with this narrow in te rp re ta tio n . We m ust take it in a fo rm in which it is adm itted by everyone. We obtain such an undogm atic in te rp re ta tio n o f the princip le if we allow not only th e pos­sibility o f som eone explain ing an exp ressio n ’s m ode o f em ploym ent to him self in the sam e way th a t he would explain it to som eone else but also, as a lim iting case, the possibility o f th e re being m eanings o r m ean- ing-com ponents o f expressions th a t som eone can only explain to h im ­self. T h e la tter w ould clearly be the case if som eth ing tha t is only inw ardly accessible, a sensation o r rep resen ta tio n , constitutes o r partly constitutes the m ean ing o f an expression.

W e can now take up again th e controversy betw een the nom inalist and the conceptualist. In response to th e challenge to p roduce his own positive conception o f the m ean ing o f a p red icate the nom inalist, invok­ing W ittgenstein’s p rincip le, had re fe r re d us to the em ploym ent-ru le which is explained an d can only be exp lained by m eans of positive and negative exam ples.

T ak in g account o f the distinctions I have ju s t m ade the conceptualist could now reply th a t he has no objection to W ittgenstein’s princip le, bu t th a t one m ust d istinguish betw een the genu ine explanation and grasp ing o f m ean ing and the in tersubjective explanation . ‘It is perfectly tru e ’, h e would say, ‘tha t a p red ica te can only be explained in tersub jec­tively in the way described by th e nom inalist. H ow ever all this proves is tha t the intersubjective exp lanation is never ad eq u ate to the g rasp ing

Analysis o f the predicative sentence 154

o f a m eaning; it can only m ediate it. For such an explanation o f the em ploym ent-ru le by exam ples still leaves open what it really is that determ ines the ru le. T h e individual exam ples do not suffice for this for we only say o f som eone tha t he has u nderstood the explanation if he is able to apply the p red icate correctly to new exam ples. But this can only m ean th a t the p erson to w hom we explain the pred icate by m eans of exam ples form s a m ental co rrela tion betw een the pred icate and som e­th ing th a t is com m on to the positive exam ples but absent from the neg­ative ones; and it is only because he can also perceive this com m on fea tu re in new objects that he can apply the predicate correctly beyond the exam ples given to new cases. T hus the positive explanation given by the nom inalist canno t p resen t itself as a genuine alternative to the ob ject-orientated explanation bu t m ust itself ultim ately fall back on the la tte r.’

T h e re a re analytical philosophers who are simply closed to this a rg u ­m ent. T h u s it frequen tly looks as though in the end th e re is again ju s t one view con fron ting ano ther. We cannot leave it like this. W e m ust settle the controversy in a convincing m anner. T h ere are two dogm atic a rgum en ts with which the conceptualistic line o f a rg u m en t ju s t p re ­sented is som etim es parried .

Firstly, one occasionally hears it said tha t such an arg u m en t refers to psychological conditions with which sem antics has no th ing to do. As we shall see th e re is an elem ent o f tru th in this statem ent. In this vague form , how ever, it is unacceptable; fo r it is not clear in advance w hat the boundaries between sem antics an d psychology are. O ne m ust first take seriously the conceptualist’s a rg u m e n t in o rd e r to establish precisely which aspect no longer belongs to sem antics and why it does not do so.

Secondly, it is som etim es stated tha t it is only th rough language - and m ore precisely pred icates - tha t o u r experience is s tru c tu red into types. T h e conceptualist’s rep ly thus violates the princip le o f the ‘impossibility o f going beh ind lan g u ag e’ (‘Unhint ergehbarkeit. der Sprache’)* However, this is an im plausible thesis which can be shown to be absurd by the fact tha t we perceive linguistic signs them selves (w hether acoustically or optically) as typical. I f for exam ple we hear the sound ‘re d ’ we respond to it in so fa r as it exhibits a p articu la r s tructu re ; we react to the individ­ual sound in so far as it is a rep resen ta tive o f a type of sound. B ut now the sam e is true o f all percep tion , not ju s t hum an perception . Every s tim u lus-response schem a, w h eth er conditioned or uncond itioned is such that the sam e so rt o f response follows the sam e sort o f stim ulus, o r a determ ina te ran g e o f sim ilar stimuli. A nd to the ex ten t that p e r ­

The basic principle of analytical philosophy 155

ceptual phenom ena are introspectively accessible we can likewise ascer­tain that o u r sensation- and represen tation-conten ts are, w ithout excep­tion, typical. A colour-perception , fo r exam ple, is as regards its contents never unique; ra th e r a specific shade o f red is perceived as one that can be perceived as the sam e on any num ber o f occasions.

O ne m ust here note a rem arkable confusion which th ro u g h o u t the centuries has b u rd en ed the en tire discussion o f nom inalism . T ra d i­tional philosophy, since A ristotle,4 carried with it a dogm a according to which perception - ‘sensibility’ - relates to individuals. T his prem ise was presupposed by both sides in the nom inalism controversy. T h a t the re are represen tations o f individuals appeared unproblem atic to both sides in th e dispute; thus if there a re also rep resen tations o f universals o r types this m ust be a non-sensuous m ode o f cognition. In reality how ­ever percep tion relates neither to individuals n o r to universals; ra th e r it is typical. T h u s the contrast is not between the individual and the universal and typical bu t between the typical on the one h and and the individual and the universal on the o ther. W hat distinguishes the u n i­versal from the typical is tha t we only call universal what is com m on to m any individuals (an attribute) o r can be applied to m any individuals (a predicate). T h e concept o f a universal is correlative to that o f an individual, ju s t as predicates are essentially expressions which supp le­m ent singular term s. F rom the po in t o f view of linguistic analysis the consciousness o f what is individual is no m ore a sensuous phenom enon than consciousness o f a universal; it is a logical phenom enon . A nd as we shall see the logical - o r linguistic - constitution o f reference to individuals poses m uch g rea ter problem s of analysis than does the con­sciousness o f universals which in consciousness o f types has a sensuous precursory form . A nd on the o th e r hand we can now u n d erstan d how trad itional philosophy could fail to notice the evident fact, never doub ted in psychology, tha t percep tion relates to the typical.5 It had to accom m odate the category o f the individual som ew here and as it d id no t reflect on the form o f sentences all tha t rem ained was to assign the individual to sensibility.

W hat follows from all this reg ard in g the evaluation o f the a rg u m en t b ro u g h t by the conceptualist against the nom inalist? T h e correct app li­cation of the pred icate to exam ples according to the conceptualist’s th e ­sis is only possible on the basis o f a m ental-correlation o f the predicate with som ething that is com m on to all exam ples. Does this no t m ean: with som ething typical given in consciousness? A nd would not the con­ceptualist’s view be vindicated by the typicality o f perceptions ju s t described? B u t in tha t case the pu rpose for which the conceptualist

Analysis of the predicative sentence 156

found it necessary to postulate an abstract object w ould already be achieved by sensuous represen tations themselves.

In o rd e r to attain clarity h ere we m ust d istinguish betw een the behav- iouristic and introspective conceptions o f the psychological p h en o m e­non o f the typicality o f perceptions. T h e typical sensation o r the typical m ental im age tha t we encoun ter in in trospection (a specific shade of red , a specific triangle-shape) does not co rrespond to th e em ploym ent- ran g e o f o u r o rd inary predicates (‘re d ’, ‘triang le’). For precisely this reason the conceptualist found it necessary to postulate a non-sensuous rep resen ta tion o f som ething tha t is com m on to all shades o f red , every­th ing th a t is called ‘re d ’, o r all form s o f triangle. T h e situation is d iffe r­en t for th e behaviouristic conception. T h a t an organism perceives typ­ically is de term ined by its respond ing to stim uli o f a certain kind in the sam e way. T h e ran g e of similarity o f the stimuli to which the organism responds in the sam e way can be w idened o r narrow ed. For exam ple the organism can learn to respond in the sam e way to all colour-shades one calls ‘re d ’. W hereas in the introspective conception the re is no sen­suous rep resen ta tio n co rrespond ing to all colour-shades tha t are not merely the sam e b u t similar, in the behaviouristic conception this dis­tinction no longer applies. W hether we call the various stimuli, from an introspective standpo in t, the sam e o r m erely sim ilar - if the .organism is appropria te ly conditioned then it responds to them in the sam e way. In this sam eness o f reaction to similar stim uli behaviouristic psychology of percep tion has a presen tab le datum with which the in trospectionist conception has no th ing com parable.

T h e problem o f sim ilarity played a large role in the trad itional, introspectively conducted discussion of nom inalism . For nom inalism , H um e a rgued th a t one does no t need the rep resen ta tion of som ething identical tha t is com m on to all sensuous rep resen ta tions which co rre­spond to the em ploym ent o f a predicate; it suffices tha t the rep rese n ­tations g roup them selves in to circles o f sim ilarity.6 B ut in tha t case, it was a rgued against H u m e,7 th e re m ust be a rep resen ta tion o f the simi­larity of the sensuous represen tations, for w hat m atters from an in tro ­spectionist standpo in t is tha t consciousness recognizes tha t the sim ilar rep resen tations are similar; the nom inalist thereby finds h im self com ­pelled to again adm it at least the rep resen ta tion o f one abstract object. B ut if one adm its the rep resen ta tion o f similarity then why no t re p re ­sentations o f any relations and attributes? This difficulty does no t arise in the behaviouristic conception which is no t concerned with rep re se n ­tations and is able to po in t to the physical phenom enon o f the same

The basic principle of analytical philosophy 157

response w hereas the in trospective conception is unab le to po in t to any c o rresp o n d in g psychic pheno m en o n .

B ut w hat follows from this? At p rese n t only this: one can only appeal to the typicality o f percep tion as the psychological fo u n d a tio n o f the un ifo rm ity in the em ploym ent o f a p red icate if one takes as o n e’s basis the behaviouristic concep t o f percep tion . T h e conceptualist may ag ree with this; fo r a fte r all he is claim ing tha t the psychological foundation o f the un ifo rm ity in the em ploym ent o f a p red ica te is no t a percep tion , is no t a m erely sensuous rep resen ta tion . B ut if the behaviouristically- in te rp re ted percep tua l m echanism can even be considered as the psy­chological fou n d atio n o f the u n ifo rm em ploym ent o f p red icates, then m ust no t th e recourse to abstract objects, which only becom es necessary from the in trospective perspective, ap p e a r suspect (especially w hen we recall th a t these rep resen ta tions a re not fo u n d in in trospection bu t m erely postulated)? T h e conceptualist m ay reply th a t n o th in g is exp lained by the re feren ce to th e s tim u lu s-re sp o n se schem a an d its conditionability; that, on the con trary , the learn ing-capacity o f intelli­g en t organism s is itself ju s t as puzzling and in need o f exp lanation as the special capacity fo r learn ing predicates. T h is is correct. But, firstly, this is to sh ift the p rob lem to an o th e r level, th a t o f causal explanation ; and , secondly, one m ust po in t ou t to the conceptualist th a t if the le a rn ­ing o f p red icates is only one case o f such a universal behavioural p h e ­nom enon as the learning-capacity o f in te lligen t o rganism s then the req u ired exp lanation can only be conceived as a physiological one; unless one also wants to a ttribu te the capacity fo r rep re se n tin g abstract objects to mice and fish (always assum ing th a t the rep rese n tin g of abstract objects could in any way explain the capacity fo r a un ifo rm response to sim ilar stim uli).8

W e arrive at a definitive clarification o f the situation if we now re tu rn to W ittgenstein’s p rincip le, which I assum ed would be accepted, in its m ost general in te rp re ta tio n , by th e ob ject-o rien tated ph ilosopher. I had conceded to the la tte r the possibility tha t a linguistic expression is used (and hence also explained) in such a way th a t it is assigned to a r e p re ­sentation o r stands fo r som eth ing tha t is only inw ardly accessible. We can now use W ittgenstein’s p rincip le as a criterion fo r dec id ing w hether the conceptualist was re fe rrin g to the m ean ing o f an expression a t all w hen he asserted th a t exp lanation by exam ples is n o t enough , th a t the expression m ust stand fo r som eth ing identical tha t we inw ardly re p re ­sent. C learly he was no t if the m ean ing is w hat we explain w hen we explain the m ean ing o f the expression. F or even if we d isregard in te r-

Analysis o f the predicative sentence 158

subjective exp lanation and assum e tha t a person explains the w ord ‘red ’ to h im self in trospectively he can clearly only do so by placing befo re his inner-eye, no t th a t a ttr ib u te o f which th e re is no sensuous rep re se n ta ­tion, b u t exam ples. So it now becom es clear tha t w hat the conceptualist m issed in exp lanation by exam ples was no t som ething that belongs to the question o f m ean ing bu t som eth ing tha t concerns the causal expla­nation o f u n d ers tan d in g . I f one accepts W ittgenstein’s principle (in its w idest in te rp re ta tio n ) then hypotheses concern ing the causal exp lana­tion o f u n d e rs ta n d in g no m ore belong to the philosophical exp lanation of u n d e rs ta n d in g th an they belong in the individual case to explain ing how a particu la r expression is used. By o rien ta ting ourselves tow ards W ittgenstein’s p rincip le we can also see th a t one does not over-step the limits o f a specifically sem antic enquiry simply by bring ing in psycho­logical p h enom ena ; one only does so if the enquiry assum es the char­acter o f an exp lana tion why.

O f course th e re is no law th a t one m ust accept W ittgenstein’s princi­ple. B u t if one does no t accept it one m ust be clear what it is one is really asking about. A nd we have ju s t seen tha t if w hat interests us is causal exp lanation then physiological hypotheses a re p referab le to the in trospection ist hypothesis. I f the conceptualist nonetheless holds on to the in trospection ist s tan d p o in t this is because he was originally m oving w ithin the limits o f W ittgenstein ’s d ictum and was no t th ink ing o f a causal exp lanation at all. H e had s ta rted ou t from the assum ption that one explains an ind ividual p red icate-expression to oneself by assigning it to the co rresp o n d in g a ttribu te . B ut this thesis had to be abandoned righ t at the beg inn ing o f the debate with the nom inalist, for in tro spec­tion fails to reveal such an object. Only w hen the conceptualist asserted, in his critique o f the nom inalist, tha t th e re must be such an object did he slip unno ticed in to trea tin g the question as one o f hypothetical causal exp lanation .

So we can now finally reg a rd the ob ject-orientated conception of p red icates as hav ing been disposed of. B u t have we already achieved a new conception o f the m ean ing o f predicates tha t is fundam en tal enough an d precise enough to replace the object-orientated concep­tion? W hat we have so far reach ed are two definitions: (1) to explain (or u n d erstan d ) a p red ica te is to explain (or understand ) w hat characteri- za tion-function it has, (2) we explain a pred icate or the m eaning o f a p red icate w hen we show (or know) the use o f the pred icate by m eans of positive an d negative exam ples. I t is easy to see how these two defin i­tions are connected : we explain which characterization- (distinction-, classification-) function a p red ica te has by dem onstra ting , by m eans o f

The basic principle o f analytical philosophy 159

positive an d negative exam ples, its m ode o f em ploym ent, o r we u n d e r­stand w hat characterization-function the pred icate has i f we can use it correctly, positively an d negatively. For this conception o f the m eaning o f predicates the use o f the linguistic sign fo r characterization is essen­tial and does not m erely m ediate this characterization; one can th e re ­fore call it a specifically language-analytical conception. T h a t the p re d ­icate stands for an a ttribu te is n o t d ispu ted by the new conception; it m erely asserts tha t the existence, o r know ledge, o f the attribu te canno t g round the u n d erstan d in g o f the predicate, bu t is itself g ro unded on this understand ing .

As regards fundam entality this explanation o f the m ean ing of p re d ­icates can com pare with the object-orientated explanation , bu t no t yet as regards clarity an d distinctness. For it is no t yet clear w hat precisely is m eant by ‘m ode o f em ploym ent’, o r ‘correct application to exam ples’, notions tha t a re fundam en ta l to the new explanation . This can be shown by the following consideration . T h e explanation o f a p red icate is supposed to consist in the dem onstration by m eans o f exam ples o f its correct an d incorrect em ploym ent. T h a t the person to whom a p re d i­cate has been explained has und ersto o d th e explanation is shown by the fact th a t he uses it and only uses it as it was explained to him. W ere he to use it d ifferen tly and were we still to say tha t he is using it correctly this would m ean th a t the explanation was incom plete; we could no longer claim tha t w hat was being explained was the m eaning o f the expression. But then it would seem tha t a m ode o f em ploym ent o f predicates would resu lt tha t does no t at all co rrespond to the ir actual m ode o f em ploym ent. T h e p red icate ‘red ’, fo r exam ple, is explained by applying it to ap p ro p ria te objects given in perception . W ould it no t then follow tha t he an d only he has correctly understo o d the w ord ‘re d ’ who uses it if and only if som ething red is p resen t in th e percep tual situation; and tha t he and only he correctly understands the w ord ‘ra in ’ who uses it if and only if it is rain ing?

We can of course imagine a prim itive language in which expressions a re used in this way. I shall call expressions used in this way quasi-pred­icates. O ne can probably say tha t the characterization-expressions chil­d ren learn in the first stage o f language-acquisition are used in this way. A child learns to say ‘bow-wow’ w hen it sees a dog, ‘m am a’ w hen the percep tua l p a tte rn o f the m o ther shows itself. We are no t now con­cerned with w hether it is a co rrec t hypothesis o f developm ental psy­chology to say tha t the characterization-expressions which ch ildren first learn are quasi-predicates. I am using the quasi-predicate m erely as a though t-m odel o f an expression com parable to a pred icate bu t sem an­

Analysis o f the predicative sentence 160

tically very m uch sim pler and re fe r to child-language sim ply by way o f illustration.

T h e characteristic fea tu re of quasi-predicates is tha t in the ir case em ploym ent-situation and explanation-situation are o f the same kind. T h e explanation shows in what situation the expression is to be used. In the case o f predicates, by contrast, it is clearly characteristic that th e ir norm al em ploym ent-situation is no t o f the sam e kind as the ir exp lana­tion-situation. I f then we wish to hold on to the view that w hat is explained w hen an expression is explained is its m ode o f em ploym ent (and we m ust hold on to it fo r this connection is analytically contained in the m eaning o f the w ord ‘exp lanation’) then it follows tha t the expla­nation o f a p red icate , an d hence too its m ode o f em ploym ent, is m ore com plicated than could have been gathered from the account given so far. As I have so fa r described the m eaning o f predicates it rem ains inde term ina te w hether we are speaking o f predicates o r quasi-p red i­cates, o r m ore correctly: so far w hat has been explained has no t been the m eaning o f predicates, but tha t o f quasi-predicates.

T h e difficulty we h e re com e u p against is no m ere question o f detail. It cannot be dealt with by m erely supp lem enting what has been said so far, for it concerns the foundation of the sem antic theory that is to be w orked out. It concerns the question o f how the m ode o f em ploym ent o f predicates, and hence tha t o f the o ther expressions o f o u r language, is to be understood . T h e notion o f the use (or em ploym ent) and o f the rules o f use is, as we have seen, still extrem ely vague. I t still also fits the objectual conception. T h e question which now arises is: if the use o f linguistic expressions is not regu la ted by reference to objects then by reference to what is it regulated? T h e answ er tha t m ost readily suggests itself is: by reference to the circum stances o f use. It looks as though if one does no t assign linguistic expressions to rep resen ta tions o f objects- from the inner-perspective so to speak - then the only alternative is to assign them to the circum stances o f use - in the ou ter-perspective so to speak. T h u s we read in W ittgenstein th a t when one asks fo r the m eaning o f a linguistic expression one should ask oneself in what sort o f circum stances it is used .9 Now it is precisely this conception which breaks dow n when applied to predicates, fo r it results in their being explained as quasi-predicates. O f quasi-predicates we can indeed say that to explain them is to explain in w hat circum stances - in w hat situ­ation - they are to be used; and that a speaker o f the quasi-predicate language understands such an expression is shown by his using it in the correct circum stances. I f predicates cannot be und ersto o d in this way and if it should em erge tha t the o th er expressions canno t be u n d e r ­

The basic principle of analytical philosophy 161

stood in this way e ith e r th en the question arises: w hat o th e r possible way is th e re o f u n d e rs ta n d in g th e ir m ode o f em ploym ent if it is to be n e ith e r by re fe ren c e to rep resen ta tio n s o f objects no r by re fe ren ce to circum stances o f use?

O u r fu r th e r analyses o f the sem antics o f p redicative sentences m ust be o rien ta ted tow ards this question. W hat is at issue is the fo rm ation o f a conceptuality adequate to the u n d e rs ta n d in g o f o u r linguistic ex p res­sions, initially in the partia l area o f the pred icative assertoric sentence. A new conceptuality does not sim ply fall from heaven. W e can only w ork it ou t step by step, taking as o u r sta rting -po in t previous concep­tualities and th e ir defects revealed in the a ttem p t to apply them .

By way o f p rep a ra tio n fo r the n ex t steps I w ould today ju s t like to d raw your a tten tio n to a fu rth e r d istinction betw een pred icates and quasi-predicates. T h e quasi-p red ica te ‘re d ’ also d iffers fro m the p re d i­cate ‘re d ’ in th a t it is used as an in d e p e n d e n t linguistic expression; it is a one-w ord sen ten ce ,10 w hereas the p red icate requires su p p le m e n ta ­tion , in the sim plest case supp lem en tation by a singular te rm . This d if ­fe rence betw een p red ica te and quasi-p red ica te is so characteristic th a t we can expect tha t w herever a w ord functions analogously to a p re d i­cate (nam ely as a characteriza tion-expression) bu t is used in d e p e n ­dently it is a quasi-pred icate, i.e. a characterization-expression w hose em ploym ent is situation-rela ted . W hat follows from this fo r the m e an ­ing o f predicates? T h a t it will only be possible to u n d e rs ta n d them in th e context o f the supp lem entab ility by singu lar term s, certainly. But h e re two opposing hypotheses a re conceivable. Firstly o n e could su p ­pose tha t it is only because pred icates canno t be em ployed in d e p e n ­dently tha t they canno t be u n d ers to o d by refe ren ce to the situation o f th e ir em ploym ent; to this supposition would co rresp o n d th e hypothesis th a t the m ean ing o f whole sentences, pred icative sentences, and th en o thers, can be u n d e rs to o d com pletely by referen ce to the circum stances o f the ir em ploym ent. T h e second possibility w ould be th a t it is precisely the supp lem en ta tion o f pred icates by singular term s which makes the em ploym ent o f p red icates, and hence that o f whole predicative sen ­tences, situation-independent. If th a t is so then only if we app roach the question concern ing the u n d e rs ta n d in g o f singu lar term s in the sam e specifically language-analytical m a n n e r in which, in the last two lec­tu res, we have ap p ro ach ed the question concern ing the u n d e rs ta n d in g o f pred icates can we a rriv e at a new u n d e rs ta n d in g o f the m eaning of p red icates which relates n e ith e r in trospectively to the rep resen ta tio n o f objects no r behaviouristically to the circum stances o f em ploym ent.

You will already have g a th ered from the way in which I have c h a r­

Analysis o f the predicative sentence 162

acterized these two hypotheses that I ho ld the second to be the correct one. I in tend , how ever, to exam ine the first hypothesis first and will not exam ine the second till m uch later. F o r it will em erge tha t fo r a p ro m ­ising clarification of the em ploym ent-ru les o f the two com ponents o f a pred icative sen tence we req u ire a tenable prelim inary conception o f the em p loym ent-ru le o f the w hole sentence. Such a conception canno t be achieved at a stroke.

L E C T U R E 1 3

The meaning of an expression and the circumstances of its use. Dispute with a behaviouristic conception

I f the m ean ing o f a linguistic sign cannot be understood in term s o f the sign’s stand ing fo r an object then the view which m ost readily suggests itself is tha t to u n d e rs ta n d a sign is to know in which circum stances it is to be used. A t the en d o f the last lecture I tried to show tha t as regards the use o f p redicates at any ra te this conception will no t do; how ever I also raised the question o f w hether this conception m ight nevertheless be co rrec t in the case o f all in d e p en d e n t u tterances an d hence in the case o f whole assertoric sentences. B efore we reject the thesis tha t the m eaning o f an expression consists in the circum stances of its use we m ust subject it to a m ore fundam ental exam ination.

In §117 o f his Philosophical Investigations W ittgenstein writes: ‘If, fo r exam ple, som eone says tha t the sentence . . . has m ean ing fo r him , then he should ask h im self in w hat particu lar circum stances this sentence is actually used .’ A nd as W ittgenstein says in ano ther place the use is thereby connected with our o th e r activities. By way o f explanation he presents, a t the beg inn ing o f the Investigations, some exam ples o f ‘lan­guages m ore prim itive than ou r ow n’ (§2). A n exam ple, which is fu r th e r elaborated in the following parag raphs, is described in §2: ‘Let us im ag­ine a language . . . which is m ean t to serve fo r com m unication betw een a b u i l d e r a n d an assistan t# . A is building with building-stones: th e re are blocks, pillars, slabs and beam s. B has to pass him the stones, an d tha t in the o rd e r in which A needs them . F or this p u rp o se they use a language consisting o f the w ords “block”, “pillar”, “slab”, “b eam ”. A calls them out; - B b rings the stone which he has learn t to bring at such- and-such a call.’ W ittgenstein adds: ‘conceive this as a com plete p rim i­tive language’. Som ew hat later we read: ‘W e can also th ink o f the whole process o f using w ords in §2 as one of those gam es by m eans o f which children learn the ir native language. I will call these gam es “language-

Analysis o f the predicative sentence 164

gam es” . . . I shall also call the whole, consisting o f language and the actions into which it is woven, the “language-gam e” ’ (§7).

T h e language o f §2 is a m ore realistic m odel o f a situation-rela ted language th an the quasi-predicate language I p resen ted in the last lec­tu re , m ore realistic because this language-gam e fulfils an intelligible com m unicative purpose. I f the em ploym ent-ru le o f expressions simply consisted in all m em bers o f the linguistic com m unity u tte rin g such-and- such an expression in such-and-such circum stances this would have no obvious com m unicative significance and one could no t explain how such a language developed in a biological species, unless it represen ts (as we can assum e in the case o f child-language) a ru d im en tary stage o f a higher-level language. O f course this in no way affects the usefulness o f the quasi-predicate language as a thought-m odel. H ow ever, even fo r the child-language it is m ore realistic to conceive o f its sem antics as being en riched in the following way: the child does no t only u tte r a p articu lar sound w hen a particu lar percep tual situation is given but also w hen it wishes it to be given; it learns to say ‘m am a’ no t only w hen its m o th er is th e re bu t also w hen it w ould like h er to come. T h u s the quasi­p red icate is used both quasi-indicatively and quasi-im peratively-opta- tively; and th e im perative-op tative m ode o f em ploym ent is clearly p u r ­poseful. T h e p a tte rn o f explanation o f such expressions rem ains in p rincip le the same and we can th e re fo re also speak o f quasi-predicates th a t are used both indicatively an d im peratively. In both cases the explanation has the form : ‘if such-and-such circum stances obtain, such- and-such an expression is used’; the only d ifference would be th a t in o ne case the circum stances consist o f the percep tual situation and in the o th e r case o f the need-situation; in the one case they are ex ternal stim ­uli in the o th e r in te rnal stimuli which, how ever, are d irected to the p roduction o f external stimuli.

H ow ever, as W ittgenstein him self em phasized one can hard ly speak o f ‘exp lanation’ in the case o f child-language in the first stage o f its developm ent; it is m ore ap p ro p ria te to speak o f ‘tra in in g ’ (§5). T o speak o f ‘explain ing’ p resupposes th a t the one to whom an expression is explained already u n derstands the words ‘correct’ an d ‘incorrec t’. T h u s he learns a ru le to which he conforms; h e learns the m ode o f em ploym ent tha t is co rrec t relative to an action-norm . (For the exis­tence o f an action-norm it is no t necessary tha t it be capable of being form ulated . Its existence am ounts simply to this: tha t certain actions can be called ‘correct’ an d o thers ‘incorrect’.) I f the m ode o f em ploy­m en t is no t being explained then we are dealing simply with a causal connection the m echanism o f which can be u n d ersto o d purely in term s

The meaning o f an expression 165

o f th e behaviouristic theo ry o f learn ing . T h e child learns, th ro u g h so- called ‘instrum en ta l cond ition ing ’, th a t in such-and-such ex ternal o r in te rn a l circum stances (in the en v iro n m en t o r the organism ) it can th ro u g h such-and-such an activity - th e u tte r in g o f a particu la r sound- p ro d u ce a positive stim ulus o r p rev en t a negative one. T h e positive stim ulus functions as a so-called ‘rew ard ’ o r ‘re in fo rcem en t’ which causally m otivates the cond itioned response. In th e case o f the im pera­tively em ployed quasi-pred icate the rew ard consists in the p roduction o f the stim ulus associated with the expression ; in the case o f its indica­tive em ploym ent it consists in the p leased reac tion o f the adults. For o u r purposes, how ever, we can d isreg a rd the question o f w hat m oti­vates the association betw een circum stances (in te rna l o r ex ternal) and the use o f th e sign; indeed we can igno re a lto g e th er th e question o f w hether this association is explained o r causally p ro d u ced by condition­ing. We can th e re fo re also ignore the question o f w hether one can speak o f the meaning o f a sign and understanding at all w here the use o f the sign is learn ed by condition ing (clearly n o t if one can only speak o f ‘m ean ing ’ and ‘u n d e rs ta n d in g ’ w here one can also speak o f ‘explana­tio n ’; on the o th e r h an d one shou ld never get bogged dow n in such verbal questions: ra th e r one should leave o p en the possibility of the re being, e.g., a w ider and a narro w er concept o f m eaning). W e can dis­reg a rd this distinction - which in o th e r contexts is fu ndam en ta l - because the im p o rta n t p o in t in o u r con tex t is u n affec ted by it. W hether the ru le in question is a normative ru le , i.e. a ru le which th e sign-user follows, o r a causal ru le , i.e. a m ere objective regu larity which can be n o ted by an observer - thus w hether th e ru le is o f the form ‘if such and such conditions obtain this expression is to be u se d ’ o r ‘if such and such conditions ob ta in this expression is u se d ’ - m bo th cases it is a m atter o f the association o f the use o f a sign with a p articu la r percep tua l situation, with particu la r perceived circum stances or conditions. In so far as such a ru le , w hether it be construed as a causal ru le o r a norm ative ru le, assigns the use o f the sign to certa in circum stances (conditions) I shall call it a ‘conditional ru le ’. T h e question o f w h e th e r the m ean ing o f ou r linguistic expressions consists in th e circum stances o f th e ir use can th e re fo re also be fo rm u la ted thus: a re the em ploym ent-ru les o f o u r linguistic expressions to be co nstrued as in this sense conditional rules?

T h e language-gam e which W ittgenstein p resen ts in §2 is also d istin ­gu ished from the quasi-p red ica te language en riched by the quasi­im perative em ploym ent-fo rm in th a t it takes account o f th e communica­tive aspect o f all na tu ra l - w h eth er h u m a n o r anim al - languages. A sign is no t only used by som eone; it is also d irected to som eone (or

Analysis o f the predicative sentence 166

several persons). We m u st th e re fo re distinguish betw een speaker and h e a re r or, m o re generally , betw een th e em itter and the receiver(s) o f the sign. In W ittgenste in ’s exam ple it is only in respect to the speaker th a t one can say tha t the sign is associated with the circum stances o f its use. In reg a rd to the h e a re r - in th e exam ple the assistant - on the o th e r hand , w hether o r no t he u n d erstan d s the sign is shown by w hether w hen he perceives the sign he perfo rm s a particu lar act. H ow ­ever, the h e a re r-ru le is also a conditional rule; the only d ifference is th a t now the condition is the sign-event itse lf: if th e sign is h ea rd such and such an activity is ca rried out. T h is ru le can also be construed both norm atively an d causally.

I t is only in this m odel th a t we have a realistic m odel o f a prim itive language, fo r it is only in this m odel th a t an in tersubjective p u rpose o f sign-em ploym ent becom es evident. In fact one can say th a t all actual p rim itive languages w hich in con trast to child-language are no t p re lim ­inary stages b u t already function purposively in th e ir own righ t, hence all anim al-languages, con fo rm to this schema. H ence the en tire behav­iouristic theo ry o f lan g u ag e takes its d e p a r tu re from this schema. A ccording to this concep tion the function o f the sign is to m ediate betw een stim ulus and resp o n se and in this way to m ake it possible th a t the one co m m u n ica tio n -p a rtn e r has o r receives th e stim ulus and the other exhibits th e a p p ro p r ia te re sp o n se .1 T hus in W ittgenstein’s exam ­ple A needs a particu la r bu ild ing -stone and instead o f p e rfo rm in g the a p p ro p ria te action h im self h e p erfo rm s a substitute-action which for its p a rt exerts a stim ulus on B which brings it abou t th a t# perfo rm s the action. T h u s th e sign m akes it possible fo r the one to act fo r the o ther, b u t also for th e one to perceive fo r th e o ther. T h e latter is the case, fo r exam ple , with w arn ing o r feeding-signals: one p a rtn e r perceives the situation an d does no t h im self resp o n d (or is no t the only one to respond) b u t ra th e r p e rfo rm s an action which serves as a substitu te stim ulus fo r the o th e r p a r tn e rs such th a t they can respond a p p ro p ri­ately to the situation w ithou t perceiv ing it them selves.

W e can call signs with this sort o f com m unicative function ‘signals’. T h is term inology is no t u n d isp u ted . T h e re are au th o rs2 who only call signs o f this k ind ‘signals’ if th e ir em ploym ent both by the em itte r and the receiver is n o t lea rn ed b u t inna te , as in the language o f bees. H ow ­ever, in the con tex t o f o u r investigation the distinction between the lea rn ed an d in n a te em p loym en t o f such signs is no t im portan t, partic­ularly as I shall also be ig n o rin g the (m uch m ore fundam ental) distinc­tion betw een those signs o f this k ind whose ru les a re causal and those w hose ru les a re norm ative and which we m ust u n d e rs ta n d as conven­

The meaning o f an expression 167

tional signals. A lthough W ittgenstein says tha t the language-gam e that he describes is learn t th ro u g h ‘tra in in g ’ it em erges clearly from later parts o f the Investigations th a t he u n d ers tan d s the rules as norm ative and the signs as conventional. A m ore serious uncertain ty concerning the in te rp re ta tio n o f W ittgenstein’s language-gam es has to do with the question o f w hether he in tends to con trast them as ‘prim itive languages’ with the language we actually speak o r w hether he in tends them to be und ersto o d as a sim ple m odel o f how even o u r language functions. T h e thesis tha t ou r language is a signal-language and that it is to be u n d e r­stood on the m odel o f conditioned responses (only in a m ore com pli­cated way) is p ro p o u n d ed by the behaviourist theory o f language; w hether W ittgenstein held a co rrespond ing view, though with a n o r­mative ra th e r than a causal in te rp re ta tio n o f rules, is a question we can leave open. Anyway it is this thesis - tha t ou r linguistic expressions function as signals o r tha t the ir rules a re conditional rules - th a t we have to exam ine. It is the m ore general and m ore fu ndam en ta l fo rm o f the thesis tha t the m eaning of an expression consists in the circum ­stances o f its em ploym ent.

I would first like to p o in t ou t two fu n d am en ta l difficulties which im m ediately becom e ap p a re n t if one a ttem pts to construe the sentences o f o u r language as signal-signs. T h e first is tha t the m ean ing o f a signal- sign is not the sam e fo r speaker and hea re r. T hus the behaviourist lan- guage-theorist L eonard Bloom field w rites tha t the ‘m ean ing’ o f an expression is ‘the situation in which the speaker u tte rs it and the reac­tion which it calls fo rth in th e h e a re r’.3 A ttem pts have been m ade to gloss over this fact, pointing out that every m em ber o f a linguistic com ­m unity can, dep en d in g on the situation, assum e both the ro le o f the speaker and th a t o f the h e a re r and on e can th e re fo re say tha t som eone only u n d erstan d s the expression if he knows both conditional rules. B ut even so this conception does not fit the sentences o f o u r language. O r p u ttin g it m ore cautiously: it at least does not co rrespond with o u r o rd in ary u n d ers tan d in g to say tha t a sentence does no t have one and the sam e m eaning for speaker and h ea re r.

T h e re is h ere revealed a peculiar con trast betw een the two theories o f m ean ing tha t I have so fa r discussed. T h e ob ject-orientated theory o f m ean ing had simply ignored the com m unicative function o f lan­guage; as it d id not expressly reflect on the fact tha t an expression is used it was also able to overlook the sp e a k e r-h e a re r relation. F rom this position the fact that an expression has only one m eaning fo r speaker an d h ea re r appears entirely unproblem atic . T h e behaviourist concep­tion on the o th e r h and rightly took the com m unicative situation as its

Analysis o f the predicative sentence 168

starting-point. H ow ever, its o rien tation tow ards signal-languages led to speaker-m ean ing and h ea rer-m ean ing being separated . A satisfactory analysis o f the m ean ing o f o u r linguistic expressions can ignore neither the com m unicative aspect o f language n o r the iden tity o f m ean ing for speaker and h ea re r; and we will th e re fo re have to ask how this identity o f m eaning is constitu ted from w ithin the sp e a k e r-h e a re r situation. T his is a way o f fo rm ulating the question which is sim ilar to the way in which it had already been fo rm ulated in the first th ird o f this cen tury by the A m erican social-psychologist G eorge H. M ead .4 Instead o f sim­ply taking the identity o f m eaning fo r g ran ted as in the ob ject-orien­ta ted conception one should , according to M ead, take as o n e ’s starting p o in t the biologically m ore prim itive and theoretically unprob lem atic signal-languages and ask w hat in the way o f behav iour m ust be added fo r speaker an d h ea re r to be able to re fe r to som ething identical. A ccording to M ead this is only possible if the speaker im plicitly co- perfo rm s the response o f the p a rtn e r and likewise the h ea re r the action o f the speaker. I f tha t is correct, th en an identical m eaning which is in d e p en d e n t o f th e specific com m unication-roles o f speaker an d h ea rer w ould only be constitu ted by the division o f the com m unication-event in to speaker- an d hearer-ro les becom ing explicit for speaker and h e a re r them selves, in th a t each one in the p e rfo rm ance o f his role co- perfo rm s tha t o f the o ther. T h is theory , tha t each o f the p a rtn e rs in te r­nalizes the ro le o f the o th e r, also explains, according to M ead, how it is th a t som eone can talk to him self. H ow ever, M ead’s theory rem ained program m atic . H e did no t show how the im plicit co-perform ance o f the p a r tn e r ’s response in o n e ’s own action is concretely to be conceived. B u t we can re ta in his conception as a hypothetical perspective for o u r fu r th e r analysis.

I f despite this obvious d iffe rence betw een sentences and signals viz. th a t sentences have only one m eaning , we exam ine the thesis o f th e signal-character o f o u r language in relation to those sentences which are o u r im m ediate concern - predicative assertoric sentences - then we find ourselves im m ediately facing a second difficulty. F or a signal- language clearly offers no room for assertoric sentences. I f one seeks to apply the distinction betw een indicative and im perative sentences, which is essential to ou r language, to a signal-sign one can in te rp re t the signal both as a sta tem ent th a t som eth ing is the case and as an im pera­tive tha t som ething is to be done; b u t precisely because bo th in te rp re ­tations are possible both a re ou t o f place. Is a bee-dance to be construed as in fo rm ing tha t th e re is honey in a particu la r place o r as a com m and to fly there? Is a w arning-cry to be construed as in fo rm ation tha t th e re

The meaning o f an expression 169

is d a n g e r o r as a com m and to ru n away? Such a d istinction would clearly only be m ean ing fu l if it belonged to the ru les o f the language- gam e th a t the rec ip ien t does no t m erely resp o n d with the response a p p ro p ria te to the em itte r’s stim ulus b u t also has the possibility o f resp o n d in g to th e em itte r’s giving o f a sign as such. A lthough we are n o t yet able to say w hat the em ploym ent-ru les a re which d e te rm in e w h e th e r a sen tence is assertoric o r im perative this m uch at least is clear: a sign can only be used assertorically o r im peratively if the rec ip ien t has the possibility o f resp o n d in g to the giving o f the sign as an asserto ric o r im perative giving o f a sign. T h e possibility o f resp o n se to th e giving o f the sign itself is in o u r sen tence-language con tained in the p h en o m en o n o f position-tak ing (Stellungnahme). T ak e a situation in which a signal could also be given (e.g. th e alarm -cry ‘fire ’). T h e rec ip ien t can respond to th e sen tence in the sam e way tha t he can resp o n d to th e signal. B ut it is also possible fo r him , instead o f resp o n d in g , to rep ly to th e sentence, to take u p a position tow ards it. T h e sim plest fo rm o f position-tak ing is den ial an d d e p e n d in g on how the w ord ‘n o ’ is used an d w hat it is tow ards which a position is taken the sen tence is revealed as an im p er­ative o r an indicative.

It m ay be su rm ised tha t the two fea tu res which already a t first glance seem to d istinguish the sentences o f o u r language from signals belong toge ther. It w ould now seem plausible to look u p o n the position-taking o f th e rec ip ien t as tha t response o f the p a r tn e r which, accord ing to M ead, is im plicitly co -perfo rm ed by th e em itter. For the negative o r affirm ative position-tak ing o f the rec ip ien t is a response o f the kind being looked for; one which is a lready an tic ipated in the em itte r’s use o f the sign. W hen we u tte r a sentence w hether assertorically o r im p er­atively, we m ean it as one th a t can be den ied . T h e a p p ro p r ia te response o f th e p a r tn e r to a sentence, in con trast to th a t to a signal, is no t the p e rfo rm an ce o f an action b u t the affirm ation o r den ial o f th e sentence. In th e case o f the im perative the p e rfo rm an ce o f the action is also a possible response. B ut this p e rfo rm an ce o f th e action is to be u n d e r ­stood as a ffirm ation , som eth ing essentially d if fe re n t from th e m ere re sp o n se ;5 fo r instead o f p e rfo rm in g th e action on e can rep ly with a ‘n o ’. In yes/no position-tak ing , as in the o th e r position-tak ings such as question ing , d o u b tin g and so on which a re g ro u n d e d in the possibility o f yes/no position-taking, the rec ip ien t is clearly re fe rrin g to the same thing as the em itte r, albeit in a d iffe ren t way. A nd if the ru les for the em p loym en t o f th e sen tence by the em itte r can be shown to be such as re la te its em ploym ent to the possible position-tak ings o f the recip ient, th en the em itte r is already re fe rrin g to the sam e th ing as the possible

Analysis o f the predicative sentence 170

recipients. In this way M ead’s p rog ram m atic hypothesis w ould be given defin ite conten t. O f course these reflections are them selves still p ro ­gram m atic, fo r one w ould first have to show th a t and how yes/no posi­tion-tak ing belongs to the em ploym ent-ru le o f sentences.

Initially we are only concerned with assertoric sentences and in p a r ­ticu lar with predicative assertoric sentences. A fter these very general an d an ticipato ry reflections I com e now to the prom ised exam ination o f the thesis th a t the em ploym ent-ru les o f predicative sentences can be un d ers to o d as conditional rules and , specifically, in accordance with the schem a o f a signal-language. T h e general difficulties ju s t re fe rre d to do no t re n d e r a detailed exam ination superfluous. For it is only by m eans o f the precise critical exam ination o f existing theories tha t we can hope to m ake new positive approaches.

I w ant to conduct th e discussion with reference to an exam ple. I shall choose an exam ple which because it is ap p ro p ria te to a context o f action is as accom m odating as possible to the theory to be criticized: the sen­tence ‘T h e tow n-hall is on fire.’ O ne can easily im agine situations in which it w ould be ju s t as ap p ro p ria te to use this sentence as the signal- expression ‘F ire!’ Now accord ing to the theory to be exam ined the m ean ing o f th e sen tence consists o f two conditional rules: the correct u n d e rs ta n d in g o f the h e a re r is show n by the fact that he responds to th e sen tence in a specific way; an d the co rrec t use by the speaker is show n by th e fact tha t he uses the sentence in specific circum stances. So if the theory is correct it m ust be possible to specify the particu lar reac­tion o f the h e a re r and the p articu la r situation o f the speaker which to g e th e r a re supposed to constitu te the m eaning o f this sentence.

I shall begin with the hearer. Suppose som eone comes here into the lectu re-room an d u tte rs the sen tence ‘T h e tow n-hall is on fire.’ How w ould we react? Probably m any o f us would no t react at all, o thers w ould ask th e m an to leave us in peace, o thers would ask him how h e knows this, o th e rs w ould p erhaps laugh and o thers m ight ru n out with various aims; in sho rt th e re is no question o f a ru le-governed connexion betw een the h ea rin g o f the sentence and particu lar actions. T o avoid this m anifest nonsense to which th e theory seem s to lead behaviourist theorists in tro d u ced the concept o f an action-disposition .6 A sentence can only have a practical m eaning , one m ight say, in certa in situations. F or exam ple, the sen tence ‘T h e tow n-hall is on fire’ does not have a p ractical m ean ing if som eone u tte rs it h e re in the lecture-room b u t it does if som eone is r ing ing the fire-brigade. T o escape the objection th a t we nevertheless u n d e rs ta n d the sen tence w hen it is u tte red here in the lectu re-room an d we do no t belong to the fire-brigade it is explained

The meaning o f an expression 171

tha t the m eaning consists no t in an action bu t in an action-disposition; tha t we here and now u n d erstan d the sentence ‘T h e town-hall is on fire’ m eans that it awakens a disposition in us such tha t if we w ere the fire- b rigade we w ould rush ou t to extinguish the fire. B ut one could ju s t as well say the m ean ing consists in the disposition to ru n ou t in to the street if one were an occupant o f the town-hall. T h e re are any num ber o f d iffe ren t situations in which this sentence can be practically relevant. A m ong arsonists, fo r exam ple, it can be used as an announcem ent o f success. T h e a ttem p t to construe an assertoric sentence as a signal thus leads to the resu lt that no t ju s t two m eanings m ust be a ttr ib u ted to it b u t innum erab le m eanings.

T his disqualifies the theory at least as regards the hearer-side; for if th e re are indefinitely m any m eanings o f a linguistic expression then the m eaning o f th e expression cannot be som ething tha t one can learn according to a ru le (or several rules). I would also like to po in t ou t once again tha t 1 have chosen an exam ple which is specially favourable to the theory. In the case o f m ost assertoric sentences it is difficult to think o f any kind o f ap p ro p ria te reactions. W hat, fo r exam ple, would be the ap p ro p ria te reactions to the sentence ‘T h e tow n-hall is r e d ’ o r ‘T h e town-hall dates from the eigh teen th cen tu ry ? We will th e re fo re have to suppose tha t th a t with which the h e a re r connects the sentence accord­ing to a ru le is som ething o th er than an action, action-disposition o r bund le o f action-dispositions, and re tu rn to the m uch m ore natu ra l conception th a t the action o r action-disposition which the hearing o f an expression effects is a resu lt requ iring the com bined opera tion o f the u n d ers tan d in g o f the expression and o f the action-m otivations o f the person concerned.

B ut in w hat th en does the m eaning for the h ea re r consist? O ne could perh ap s th ink th a t the ru le which in signal-languages is the h ea re r-ru le fits im peratives ra th e r than assertoric sentences, and that the ru le we have to follow in using assertoric sentences is the ru le which in the signal-language is the speaker-ru le . T h is is certainly an attractive sug­gestion, particularly when one considers tha t in this w eaker fo rm the behaviouristic theory would also be able to cope with both o f the diffi­culties to which I re fe rre d a t the beginning. In contrast to signals, sen­tences would be e ither indicative o r im perative. T h e m ean ing o f the indicative sen tence would be d e term ined even fo r the h e a re r by the speaker-ru le an d the m eaning o f th e im perative sentence would be determ ined even for the speaker by the hearer-ru le . In both cases then the m eaning fo r speaker and h ea re r w ould be the same.

So in the case o f the sentences with which we are at p resen t con­

Analysis of the predicative sentence 172

cerned (i.e. assertoric sentences) every th ing d ep en d s on w hether at least the speaker-side o f th e signal-schem a fits. Can one say th a t the m ean ing o f an assertoric sen tence consists in the circum stances in which it is used?

T his thesis fo r its p a rt allows o f two in te rp re ta tions. Firstly one could try to construe the circum stances practically. T h e m ean ing o f th e sen­tence ‘T h e tow n-hall is on fire’ would then be defined by the circum ­stances which give point to the u tte rance o f the sentence, e.g. th a t help is n eed ed to p u t o u t the fire. T his explanation how ever am ounts to a repe tition o f the explanation th a t was previously given for the h ea re r and it would be subject to the sam e weaknesses. Besides we w anted to assim ilate the h ea re r-ru le to the speaker-ru le and no t vice versa. This in te rp re ta tio n is th e re fo re elim inated.

T h e second a n d m ore plausible in te rp re ta tio n o f the re feren ce to circum stances o r em ploym ent-situation is th a t which co rresponds to the referen ce to the stim ulus in the behaviourist m odel: the circum stances to which the m ean ing-ru le ties the use o f the sen tence are defined by the percep tua l situation o f the speaker. T h is view can be held in a sim­ple an d a revised version. A pplied to o u r exam ple the sim ple version w ould be: always (or in m ost cases) w hen it is perceived tha t th e town- hall is on fire one says ‘T h e town-hall is on fire’, o r in its norm ative form ulation : w henever it is perceived th a t the tow n-hall is on fire one is to say ‘T h e tow n-hall is on fire.’ O ne could also reg a rd this ru le for the use o f the predicative sentence as a specification o f a ru le fo r the use o f th e predicate, which w ould then be: w henever it is perceived tha t som eth ing is on fire one is to say th a t it is on fire. This explanation w ould be an explanation o f the p red icate as a quasi-predicate. It is obvious tha t this explanation is inadequate, fo r norm ally we use neither p red icates no r predicative sentences in the co rrespond ing percep tua l situation.

O f course, signals too are som etim es stored an d transm itted no t in the percep tua l situation, b u t only when they can be relevant to a p a r t­ner. Bees, fo r exam ple, p e rfo rm th e ir language-dance no t in the presence bu t in the im m ediate afte rm ath o f certain perceptually received factors; it is the in fo rm ation -requ iring p artn e rs tha t first trig ­ger o ff the language-dance. This com plication can be easily incorpo­ra te d in to the behaviourist theory by m eans o f th e following additions:(1) the notion o f percep tua l circum stances can be ex ten d ed to include percep tions which continue to exercise an influence th ro u g h m em ory.(2) H ere too o n e can in tro d u ce the concept o f a disposition and say: from the m om en t o f the relevant percep tion the speaker is in a co rre­

spond ing d isposition which is only actualized if he encoun te rs a p a r tn e r n eed ing in fo rm ation ; the p a r tn e r is th en h im self a fac to r tha t belongs to the circum stances. T h e conditional ru le is now an ex tended situation- ru le an d can be fo rm u la ted som eth ing like this: if xy was perceived and a p a r tn e r P is perceived , S is (or in the n o rm ative fo rm ulation : ‘is to be’) u tte red .

T his revised version o f the behaviouristic exp lanato ry schem a can easily lead one to sup p o se th a t th e lack o f situation -re la tedness which is characteristic o f the use o f a p red icative sen tence is in princip le n o th in g b u t the ex ten d e d situation -re la tion th a t a lready holds fo r bee-language, only m uch m ore com plicated . T h e sp eak e r need n o t have perceived tha t the tow n-hall is on fire; he may have in fe rred it o r learn t it from som eone else. All these possibilities w ould have to be in co rp o ra ted in the ex ten d ed situation -ru le . As in th e case o f the h ea re r-ru le the em p loym ent-ru le th rea ten s to becom e so com plicated tha t it ceases to be d e te rm in a te - an d this in th e case o f so sim ple a sentence as ‘T h e tow n-hall is on fire’! But, you m ay ask, why sh o u ld n ’t a sentence which is easy fo r us to u n d e rs ta n d rest u p o n a com plicated m echanism ? H ow ­ever, we are n o t en q u irin g ab o u t a m echan ism u n d erly in g the use o f the sen tence b u t ab o u t the ru les o f its use which one m ust be able to explain to som eone; o therw ise we are n o t en q u irin g abou t m eaning .

B ut we do no t n eed to dwell on this inde term inacy arg u m en t. T h e thesis th a t the m ean in g o f a p red icative sen tence can be defined in term s o f the situation of its use, how ever ex ten d ed , fo u n d ers fo r re a ­sons o f princip le. In o rd e r to show this I w ould like to draw a tten tion to two aspects o f th e use o f p red icative sen tences which contrad ict the account which rela tes use to circum stances. I can b rin g these two aspects in to re lie f by se tting th em against the two additions in the revised version o f th e behaviouristic theo ry . T h ese w ere (a) to th e p re ­sent circum stances w ere ad d e d th e past circum stances which con tinue to exert an in fluence th ro u g h th e m ed iu m o f m em ory (b) the p resen t circum stance w hich calls fo rth th e actual use o f the sen tence is th e p res­ence o f a p a r tn e r n ee d in g in fo rm atio n . _ j

1. I shall begin w ith the second o f these two points. I t is essential to the view tha t use is d e te rm in e d by circum stances fo r if th e re is no p re ­sent fac to r which calls fo rth th e actual use, th e actual use at a p articu la r tim e an d a p a rticu la r place w ould no t be g o verned by the circum stances at all. B u t now an assertoric sen tence is n o t governed by a ru le which ties its use to a specified p artn e r-s itu a tio n . Even if one ignores the fact tha t one can also significantly u tte r a sen tence w ithou t an addressee being p re se n t it is nonetheless tru e th a t on e can u tte r it to any p a r tn e r

The meaning of an expression 173

Analysis o f the predicative sentence 174

at any tim e in any place w ithou t its m ean ing changing. (This o f course does no t apply to sentences with deictic expressions; it is only la ter tha t 1 shall speak abou t these expressions, which in the contex t o f o u r dis­cussion possess a special significance.) B u t this m eans then tha t the use o f the sen tence is no t ju s t less s ituation -re la ted than a signal b u t posi­tively situ a tio n -in d ep en d en t. I t can be re p o r te d on the radio, fo r exam ­ple, th a t H eide lberg tow n-hall is on fire, and som ebody som ew here in Alaska o r in A fghan istan can tell som e people, who may or may no t be in te rested to h ea r it, ‘H eide lberg tow n-hall is on fire .’ O f course if the m an in A laska does use this sen tence th e re will be certain circum stances in his situation which occasion his rep ea tin g of-this sentence; b u t these circum stances a re certa in m otives and in terests and have no th ing to do with the m ean ing o f th e sentence. So the em ploym ent-ru le which d e te r­mines th e m ean ing o f a p red icative sen tence cannot be a conditional ru le o f w hatever con ten t; it can n o t be a ru le o f the form ‘if . . . use sen tence “p ” H ow ever to recognize the s itua tion-independence o f a pred icative sen tence is not yet to u n d e rs ta n d it. T h e elucidation o f this peculiarity o f p red icative sen tences is in my opin ion o f decisive im p o r­tance fo r the u n d e rs ta n d in g o f the use o f predicative sentences and of assertoric language generally; I will la te r go into this problem th o r­oughly. F or the tim e being how ever I can at m ost claim plausibility for the thesis th a t it is th e s ingu lar term s in predicative sentences which m ake possible this situ a tio n -in d ep en d en ce . W hat d istinguishes th e sen­tence ‘H eide lberg tow n-hall is on fire’ fro m the quasi-pred icate ‘fire’ is the fact th a t the s ingu lar te rm holds on to the percep tua l situation , or som eth ing in it, as identical an d in this way m akes it possible to re fe r to the p ercep tu a l situation fro m any o th e r situation. T his reference to the situation from a n o th e r situation is som eth ing fundam entally d iffe ren t from an ex ten d ed situation -re la tion in th e sense o f situation-depen- dence.

2. In th e case o f th e em p loym en t o f a p redicative sentence th e re are not only no p rese n t circum stances which belong to its m eaning, the re are no past circum stances e ither. I t m igh t at first seem tha t although the significant use o f the sen tence ‘H eide lberg tow n-hall is on fire’ by the m an in Alaska d id no t req u ire a ca lling-forth fac to r tha t belongs to the m ean ing it d id req u ire th e circum stance tha t he h ea rd this sen tence on the radio . Som e experien tia l fo u n d atio n , even if no t a percep tua l one, seem s to be req u ired fo r th e significant use o f the sentence. H ow ­ever it is easy to see th a t this assum ption is m istaken. T h e m an may from w hatever m otive have go t it in to his head to s ta rt a false ru m o u r. This exam ple may at first a p p e a r som ew hat artificial. You m igh t even

be inclined to say th a t if som eone uses the sentence in this way then he is using it in a way th a t is con trary to the ru le and tha t uses which are contrary to the ru le do no t need to be, indeed may not be, included in the em ploym ent-ru le . B ut th ink once m ore o f the m an who com es in here and cries: ‘T h e town-hall is on fire.’ We u n d e rs ta n d this sta tem ent im m ediately although we do no t know w hat experien tial basis the m an has for it; indeed we do no t know w hether he has any experien tial basis at all. N aturally we will only take his sta tem en t seriously to the ex ten t tha t we suppose it to be w ell-founded. B ut o u r u n d ers tan d in g o f the sta tem ent is in d e p en d e n t o f w hether we take it seriously o r not. I f we ask ourselves o r ask him w hat a re th e circum stances which have occa­sioned his u tte rance o f this sentence o r w hat g rounds he has for it we are p resupposing tha t fo r him an d fo r us independen tly o f these c ir­cum stances o r g rounds the m eaning of the sentence is already fixed. T h e conclusion is thus unavoidable tha t no t only the p resen t circum ­stances b u t also the past circum stances o f a sen tence’s em ploym ent a re com pletely irre levan t to the m ean ing o f the sentence. In reg a rd to the first point, th e s ituation -independence o f th e em ploym ent o f a p re ­dicative sentence, one could still suppose th a t a conditional rule w hich fixes its m eaning is m erely not sufficient for the actual use o f a p red ica­tive sentence. I t was still conceivable tha t th e re should be a conditional ru le which fixes the m eaning which concerns th e speaker’s past circum ­stances an d rep resen ts at least a necessary condition o f the use o f the predicative sentence. I t has now been shown tha t the m eaning canno t ' be contained in a conditional ru le at all.

This negative resu lt raises once m ore the question: w hat should be the d irection o f o u r enquiry into th e m eaning o f a predicative sentence? I f we canno t construe the m eaning as an object and if it does not consist in the circum stances o f use w hat th en rem ains? In p articu la r you may ask: if the circum stances o f use a re irre levant to the m ean ing then m ust we not also abandon th e p resupposition tha t to u n d erstan d a linguistic expression is to know its em ploym ent-ru le? B u t to argue thus would be to assum e th a t it is only to the circum stances o f its em ploym ent th a t som eth ing can be re la ted by its em ploym ent-ru le ; and this would be a mistake. W hen I first in troduced th e question abou t use it em erged tha t w hen we speak o f the use of som ething it is norm ally som eth ing th a t has a function (p. 136f). A nd w hen we ask for the em ploym ent-ru le o f som eth ing th a t has a function we are no t asking u n d e r w hat circum ­stances it is to be used bu t how it is to be used if one wishes to achieve the p u rp o se fo r which it exists. P redicates, it becam e clear, have a fu n c­tion , the function o f characterizing; so it is plausible to suppose th a t

The meaning o f an expression 175

one can also speak o f a function in the case o f the whole sentence. B u t in what should the function o f a sentence consist? Does the use o f a sentence have a purpose? Do we use a sen tence with the in ten tion o f bring ing som eth ing about? T his would be th e sim plest way o f in te r ­p re ting the idea tha t the use o f a sentence has a function . T h e em ploy- m ent-ru les o f the sen tence w ould on this view be rules concern ing how the sentence is to be used to achieve the in ten d ed effect; one can call rules o f this kind instrumental rules. We shall th e re fo re have to exam ine the hypothesis th a t the em ploym ent-ru les o f assertoric sentences are not to be construed as conditional rules bu t ra th e r as instrum ental rules.

Analysis o f the predicative sentence 176

L E C T U R E 1 4

The employment-rule of an assertoric sentence. Argument with Grice and Searle

‘T h e m ean ing o f a w ord is w hat the ex p lan a tio n o f th e m ean ing explains.’ I have called this d ic tum o f W ittgenste in ’s th e fu n d am en ta l p rincip le o f analytical philosophy; and we have so fa r seen no reason fo r no t con tinu ing to a d h e re to it. (In the last lec tu re it reced ed into the back g ro u n d inasm uch as it could n o t fo rm th e basis o f th e b eh av io u r­istic version o f the thesis th a t the m ean ing o f an expression is d e te r ­m ined by a conditional rule. F or precisely this reaso n how ever I stressed th a t one can u n d e rs ta n d th e cond itiona l ru le n o t only as a causal ru le b u t also as a ru le which the perso n w ho uses th e sign follows an d which on e can th e re fo re explain . My critique o f th e conditional- ru le theory was no t specially d irec ted against its causal version and can th e re fo re also be u n d e rs to o d as a critique o f a theo ry w hich takes W itt­g enste in ’s p rincip le as its basis.)

F u rth e rm o re it still seem s trivial to say th a t to explain th e m ean ing of an expression can only be to exp lain the ru le o f its em ploym ent.

I began th e analysis ö f the m eaning o f p red icative sen tences with th e enqu iry in to the m ean ing o f p red icates. T h e sequence was as follows: in analysing the pred icative sen tence it seem ed n a tu ra l to sta rt with pred icates because it was this p a r t o f th e p red icative sen tence on which th e ob ject-o rien ta ted conception o f the m ean in g o f p red icative sen ­tences fo u n d e re d . In th e m ean tim e a second reason fo r this o rd e r o f p roceed ing , an d one m ore closely re la ted to th e m a tte r in hand , has becom e a p p a ren t. I f un like the ob jec t-o rien ta ted p h ilo so p h er one starts n o t with rep resen ta tio n s bu t with m odes o f b ehav iou r an d if fu r th e r ­m o re one takes into accoun t languages m o re prim itive th an sentence- language, th e n it becom es clear th a t characteriza tion -expressions a re th e linguistic expressions which a re th e easiest to u n d e rs ta n d . P re d i­cates have a p re -fo rm which I have called quasi-p red ica tes and which a lready function as in d e p e n d e n t expressions, hence do no t req u ire to

Analysis o f the predicative sentence 178

be su p p lem en ted by singu lar term s. T h eir explanation is u n p ro b ­lem atic for it is achieved by m eans o f exam ples by sim ple assignm ent to particu la r circum stances (perceived by the sign-user or to be b ro u g h t ab o u t by the recipient). In the last lecture I d id not directly re fer to signals as quasi-pred icates, fo r I w ished to avoid confusion in the exam ­ina tion o f th e question o f w hether predicative sentences can be con­s tru ed as signals. B ut subsequently th e re is no reason for no t also calling signals quasi-predicates.

O ne can thus th ink o f p red icative sentences as arising o u t o f quasi­p red icates th ro u g h the add ition o f singular term s. T h e analysis o f sin­g u la r term s which to th e ob ject-orientated ph ilosopher ap p eared so unp ro b lem atic constitu tes fo r the behav iour-o rien ta ted approach the ; rea l crux in th e analysis o f the sem antics o f predicative sentences. It is the singular te rm s by m eans o f which - as I have so far only been able to suggest - th e use o f predicative sentences becom es situation-inde­p en d en t.

Now it is because pred icates, though close to quasi-predicates, are not quasi-pred icates (for they are essentially incom plete expressions and this is som eth ing th a t w ould also have to be em bodied in th e ir em ploy­m ent-ru le) tha t my first account« o f the way in which predicates are exp lained fo u n d e red (p. 160). T h is account p roceeded in accordance with the m odel ap p ro p r ia te to quasi-predicates: to explain an expres­sion is to show in which circum stances it is to be used. T h is m odel, how ever, had to be fundam en tally criticized. I t had to be shown tha t it is useless no t only for the exp lanation o f predicates bu t also fo r that o f w hole sentences. Since this has been done in the last lecture we could r e tu rn to the discussion o f p redicates. A nd accord ing to w hat has ju s t b een said a satisfactory accoun t o f the explanation o f p redicates would now at the sam e tim e have to include the explanation of singu lar term s. Now the com bined analysis o f the em ploym ent-ru le o f predicates and singu lar term s is sim ply the analysis o f the predicative sentence in its p redicative s tru c tu re . I f p redicates an d singular term s are expressions which essentially com plete each o th e r then the ir m eaning can only con­sist in the con tribu tion which each o f these types o f expression makes to th e m ean ing o f the w hole sentence. B u t if this is so then it w ould be advisable befo re sta rting th e discussion o f the two sen tence-com ponents to achieve a t least a p re lim in a ry ‘conception (Vorbegriff) o f the em ploy­m en t-ru le o f the whole sentence. Since the objectual approach (which arose from a one-sided concern with singular term s) and now the thesis th a t m ean ing consists in th e circum stances of em ploym ent (which arose fro m a one-sided concern with signal-signs function ing like predicates

The employment-rule o f an assertoric sentence 179

m ade in dependen t) are to be reg ard ed as having fo u n d ered we at p re ­sent lack a positive perspective that could guide an analysis o f p red i­cates and singular term s. A nd since o u r aim in the counter-m ove to those one-sided approaches is to u n d ers ta n d both predicates and sin­gu lar term s in th e ir m utually supp lem enting function we can only expect to find such a guid ing-perspective in the em ploym ent-ru le of the whole sentence. O f course if we should be able to discover som e­th ing about its use as a sentence before the elucidation o f the sentence- com ponents and hence o f the struc tu re o f the predicative sentence, then obviously this can only concern the sentence as an assertoric sen­tence in general, no t as a predicative sentence. A nd if the predicative sen tence-form is the most elem entary sen tence-form , to which all o ther form s, as h igher-level form s, re fer back, then w hat we can find out about the use o f the assertoric sentence in general before the analysis o f the predicative s tru c tu re can only be regarded as a provisional account which can itself only becom e fully intelligible later th ro u g h the analysis o f the predicative structu re . In an investigation such as the p re ­sent one an area has first to be opened up ; one is no t simply describing one th a t is already know n. C onsequently such going to and fro is hard ly to be avoided.

T h e question o f how the em ploym ent-ru le o f an assertoric sentence is positively to be understo o d was raised at the end o f the last lecture, afte r we had arrived at the negative resu lt th a t it canno t be u n derstood as a conditional rule. O u r only reliable gu ide in this quest is the princi­ple tha t we can only coun t as the m eaning o f a sentence that o f which one can significantly say tha t we explain it w hen we explain to som eone how the sentence is to be used. T h e explanation , we saw, is not achieved by giving exam ples o f the em ploym ent o f the sentence in particu lar circum stances. Now if the p resen t circum stances have no th ing to do with the m eaning o f the assertoric sentence then the addressee o f the in fo rm ation is the only elem ent p resen t in the situation tha t could have som ething to do with the m ean ing o f the sentence. Could it be th a t the em ploym ent-ru les we are seeking are ru les which som ehow rela te the em ploym ent o f the sentence to the addressee? I f so th en this clearly cannot be in such a way tha t the addressee is u nderstood simply as a triggering-factor, for, as we have seen, if this were hisTuncti'on, th en he would m erely be an o th e r item in the circum stances.

B ut how else is the rela tion o f the use o f the sentence to the addressee to be understood? I f the em ploym ent-ru le in which the m eaning is sup ­posed to reside does not rela te the em ploym ent o f an assertoric sen­tence to som ething given (the p resen t and past circum stances) an d if

Analysis o f the predicative sentence 180

m oreover we wish to avoid relating it to som ething inw ardly re p re ­sented , then the only th ing left is to relate it to certain consequences o f the sen tence’s em ploym ent. A nd the sim plest way to construe this would be as follows: the em ploym ent-rule o f the sentence relates to a purpose, to an intended effect. T h e relation betw een the em ploym ent-ru le an d the p a r tn e r would then be this: w hen we say som ething to som eone we thereby in tend to b rin g som ething about in him. I am thus taking up the suggestion m ade at the end o f the last lecture tha t one should take as on e’s sta rting -po in t the fact tha t linguistic expressions have a function and enquire w hether the em ploym ent-ru le which we explain w hen we explain the m eaning could be construed as an instrumental rule.

B ut first we m ust dispose o f a m isunderstanding . T h e question o f w hether the em ploym ent-ru le o f an assertoric sentence relates to a p u r ­pose should no t be confused with the trivial question o f w hether we use sentences to achieve purposes. T h a t by u tte rin g a sentence we norm ally in tend to b rin g som ething about is trivially tru e - but we can in tend to b ring about all sorts o f things so it cannot be this which determ ines the m eaning o f the sentence. As well as using a ham m er for the p u rpose for which it exists we can also use it for various o ther purposes; and even when we use it f o r t h e p u rp o se for which it exists,,in its own p ro p e r function , we do so in o rd e r to achieve various o th e r purposes. A nd we can also use a stone for all sorts o f purposes, a lthough in itself it has no function. So w hen we ask w hether the em ploym ent-ru le o f sentences relates to a pu rpose we are not asking w hether we use them fo r purposes: R ather we are asking w hether a sentence has a s tandard purpose, in th e way th a t a ham m er has and a stone has no t and w hether this stan d ard p u rpose is w hat we explain w hen we explain its m eaning.

O ne possible way o f rela ting the m eaning o f a sentence to an effect in the addressee which the sentence-user in tends w ould be this: the sentence is used to b ring about a particu lar action or action-disposition o f the p a rtn e r. This conception w ould be very close to the conception, refu ted in the last lecture, that the h earer u n d erstan d s th e sentence if he responds to it by acting in a particu lar way. T h e p resen t suggestion is - at least as regards assertoric sentences - like the previous one and , fo r the sam e reasons, un tenable. T h e actions we; can in ten d to p roduce in a p a rtn e r with an assertoric sentence are so m ultifarious tha t they cannot determ ine its m eaning.

So this possibility is excluded. I f there is any question at all o f an in tended effect on th e p artn e r determ in ing the m ean ing then it can only be a disposition which corresponds to the indicative sense of the

The employment-rule o f an assertoric sentence 181

assertoric sentence. O ne could say: th e use o f an assertoric sen tence - th a t is to say th e s ta tem en t - has the function o f in fo rm ing tha t som e­th in g is the case. A nd one could re n d e r this m ore precisely by saying: if som eone, vis-ä-vis a p a r tn e r, uses an assertoric sentence *p’, he in tends to b ring it ab o u t tha t the p a r tn e r holds ‘p ’ to be true , or, in an o th er fo rm u lation , th a t he believes tha t p. B uild ing on this thesis, an d a co r­resp o n d in g one for im peratives, H. P. Grice, in an influential p a p e r ,1 developed a theo ry ab o u t w hat it is to mean som eth ing with a sign. We m ust sharply d istinguish th e two m eanings o f th e G erm an w ord 'meinen’ which play a ro le in G rice’s theory and fo r which in English th e re are two d iffe ren t w ords: (1) to m ean som eth ing with a sign: English: ‘m e an ’, F rench : ‘vouloir dire’ (in G erm an too instead o f £was meinst du ' damit?’ we can equally well ask ‘was willst du damit sagen?’) (2) to believe th a t som eth ing is the case, hold som eth ing to be true . T h e fu n d am en ta l im portance fo r sem antics o f ‘m ean in g ’ in th e sense o f vouloir dire becom es clear w hen one realizes th a t the action which som eone p e r ­fo rm s w hen he uses a sen tence consists in m ean in g som ething-w ith the sentence; an d one can now also u n d e rs ta n d th e ‘u n d e rs ta n d in g ’ o f the h e a re r as a concept correlative to this m ean ing (Meinen): the h ea re r u n d ers tan d s w hat the speaker m eans. A ccordingly w hat is u n d ersto o d is n o t the sign b u t the action o f the speaker or, be tte r, w hat is done with the sign. G rice’s p ro g ram m e consists in a ttem p tin g to construc t the con­cep t o f the m ean ing (sense) or u n d e rs ta n d in g o f linguistic expressions on the basis o f the concep t o f m ean ing (vouloir dire).

Previously I said th a t th e function o f the assertoric sen tence can be u n d e rs to o d as in fo rm in g (Mitteilen). O ne could now build the concept o f in fo rm in g in to the G ricean conceptuality by saying: in fo rm ing is a special case o f m ean ing (Meinen) nam ely th a t which consists in the em ploym ent o f an asserto ric sentence. Now G rice shows th a t w hen one does som eth ing with the in ten tion o f b ring ing it about th a t o n e’s p a r t­n e r believes som eth ing this canno t always be u n d e rs to o d as in fo rm in g o r as m ean ing (in the sense o f vouloir dire). F or exam ple, in o rd e r to b rin g it ab o u t th a t a p e rso n believes som eth ing one can get the perso n in to the a p p ro p ria te percep tua l situation o r on e can see to it tha t he perceives a sym ptom (Anzeichen) o f its being th e case. A sym ptom is a so-called n a tu ra l sign, a state o f affairs whose existence enables one to in fe r the existence o f a n o th e r state o f affairs. W hen I po in t ou t to som e­one the p aw -p rin t o f a b ea r this can have the significance th a t I in ten d to b ring it ab o u t tha t he believes th a t a bear is in the vicinity; b u t such an act is n o t in fo rm ing . G rice shows tha t we can only speak o f in fo rm ­ing or, m ore generally , o f m ean ing (vouloir dire) if one in tends, in a

Analysis o f the predicative sentence 182

particular way, to b ring it abou t th a t o n e ’s p a rtn e r believes that/?, nam ely in such a way tha t the p a r tn e r (1) recognizes the in ten tion and (2) the recognizing o f the in ten tion is fo r him the g ro u n d for the form ation o f the belief. F or ou r purposes how ever we can ignore these refinem ents, fo r what has first got to be decided is w hether it is at all correct to say th a t som eone who, vis-a-vis a p a rtn e r, uses an assertoric sentence ‘p ’ always thereby in tends - in w hatever m ore precisely definable way - to b ring it abou t that the p a rtn e r believes that/?.

A lthough this is certainly most o ften the case, it is not always the case. A pupil, fo r exam ple, who answ ers the teach er’s question does not in ten d to in fo rm the teacher. O r if A u tters, vis-ä-vis B, an assertoric sentence ‘p \ know ing th a t# is convinced o f the opposite, he clearly does no t do so in o rd e r to b rin g it about t h a t # believes that/?. Grice him self, in a later p a p e r ,2 ab an d o n ed his theory for ju s t such reasons; how ever he con tinued to hold th a t the em ploym ent-ru le o f an assertoric sen ­tence consists in its being used to b ring it abou t tha t a p a r tn e r believes som ething. A ccording to the new theory: w hen A uses, vis-ä-vis B, an assertoric sen tence *p\ then he in tends (in the m ore precisely defined way) to b ring it about t h a t # believes tha t A believes th a t/? .3

T his sta tem en t seems to me to be correct. W hen I u tte r, vis-ä-vis som eone, an assertoric sentence ‘p ’ I do n o t necessarily in tend to bring it abou t th a t he believes th a t p; b u t I do necessarily in tend to bring it abou t tha t he believes th a t I believe tha t p (even if I’m lying I in tend to b ring it abou t th a t he believes tha t I believe that/?).

H ow ever, it does no t follow from the correctness o f this statem ent th a t the in ten tion to which it refers is the prim ary in ten tion connected with the use o f an assertoric sentence, o r tha t this is the function of the sentence. It is conceivable tha t this in ten tion is m erely a consequence o f th e prim ary in ten tion with which an assertoric sentence is used and by referen ce to which its function is to be understood . But above all it does n o t follow fro m the correctness o f th a t sta tem ent that the m eaning of the sentence is con tained in the function o f an assertoric sentence as this is h e re defined. I shall deal only with this last point which for us is the m ain one.

I f we apply W ittgenstein’s p rincip le to G rice’s suggestion then we w ould explain to som eone the m ean ing o f an assertoric sentence ‘/?’ by telling him th a t it is used to bring it abou t tha t a h ea re r believes that the speaker believes tha t p. O ne w ould thus explain the m eaning o f the sen tence ‘p ’ by m eans o f a longer sen tence ‘q’ which contains the expres­sion ‘tha t/?’ as a part. T o this it can be objected, firstly, tha t *q’ is clearly no t synonym ous with ’/?’; and , secondly, that one cannot u n d erstan d

such an explanation unless one already understands the m eaning o f ‘tha t p ’ and hence the m eaning o f

B ut even if we would o r could explain, in a particu lar case, the m ean­ing o f a sentence in this way this would not be a possible basis for a theory of m eaning, at least not if the la tter regards itself as fu n d am en ­tal, thus in explaining the m eaning o f sentences is not con ten t to p re ­suppose the u n d e rstan d in g o f co rrespond ing sentences in ano ther lan­guage, a so-called m eta-language. T h e re are various types o f such circular m eta-linguistic theories o f m eaning. W hat distinguishes the G ricean theory from the o thers is th a t it not only presupposes that the m eaning-theorist understands m eta-linguistic sentences bu t also that he already knows w hat it is to believe that p. B ut this is to presuppose every­th ing that is to be explained. Grice’s theory does no t o ffer an alternative to the approach which has recourse to ideas and the behaviouristic approach; it simply leaves the problem open.

I do not wish to suggest tha t G rice’s proposal is worthless. However, the im portance o f his contribu tion lies elsewhere. Grice attem pts p re ­cisely to define a com prehensive concept o f m eaning (in the sense o f vouloir dire) which goes beyond m ean ing in sentences and also em braces signals, those nam ely which are no t to be understood causally, bu t o f which one can say tha t som ething is signified by them . I believe that such a use o f signals only occurs w here the basic language o f those who use the signals is already a sentence-language. In the case o f signals o f this kind it is indeed tru e tha t they a re explained in the way envisaged by Grice. T o appeal to a sentence ‘q’ which presupposes the u n d e r­stand ing o f a sentence in explain ing a signal x is no t circular. B ut precisely because Grice’s concept o f m eaning {vouloir dire) is so com pre­hensive, he cannot, as we shall see, cap tu re w hat is specific to assertoric speech.

Finally, 1 w ould like to m ention one m ore difficulty. Even if we reg a rd the intersubjective em ploym ent o f sentences as basic, no theory o f m eaning can be satisfactory which does not allow for the fact that we can also use assertoric sentences, and with no change o f m eaning, w hen talking to oneself. So if my suggestion th a t the em ploym ent-ru le is to be understo o d as som ehow related to th e addressee is to w ork then this can only be if th e role o f the addressee can also be in ternalized. B ut in G rice’s theory this is not so. I t would be absurd to suppose th a t when I say som ething to myself I in tend to b ring it abou t that I believe tha t I believe that som ething is the case. W hen one speaks to oneself one is clearly not try ing to b ring som ething about. N onetheless it does seem tha t one uses the linguistic sign to som e end, th a t it has a function.

The employment-rule of an assertoric sentence 183

Analysis o f the predicative sentence 184

T his would suggest tha t we should con tinue to be gu ided by the idea tha t the sen tence has a s tan d ard pu rp o se , i.e. a function , bu t tha t the function does n o t relate to the p ro d u ctio n o f an effect. Is this conceiv­able? In any event every th ing seems to favour ab andon ing th e sugges­tion th a t we shou ld rela te th e m ean ing o f assertoric sentences to an in te n d ed effect an d in te rp re t the ir em ploym ent-ru les as in strum en ta l rules. P erhaps th e re is a n o th e r way o f u n d e rs ta n d in g the function o f a sentence; and p erh ap s th e re is a way o f rela ting the em ploym ent-ru les to th e addressee which does not involve reg a rd in g the la tte r as the object o f an in te n d ed effect.

Above all we m ust s ta rt a t a m uch low er level an d not o p era te from the ou tset with such high-level w ords as ‘believe’ an d suchlike. T h a t the p rim ary effect o f an assertoric sentence on an add ressee is no t an action b u t a belief is som eth ing which, if correct, itself requ ires to be explained. I f in seeking to u n d ers tan d the word ‘believe’ we do no t wish to reso rt once m ore to in te rn a l rep resen ta tions th en we m ust be clear th a t outside th e use o f sentences th e re are no actions o r action-disposi- tions th a t one can simply describe as th e expression o f a belief. I t is tru e th a t we speak o f th e actions o f in telligent anim als as being d e term ined by beliefs and in ten tions. F or exam ple, the cat ru n s tow ards the spot because it believes th a t th e re is som eth ing th e re th a t has the p roperties x y which it can perceive and because it desires th ings which have these p roperties. T h e in ten tional action is the expression of a belief and a desire; bu t th e re are no actions ou tside the use o f assertoric sentences in which a belief-disposition could m anifest itself independen tly and no t as a m ere com ponen t. T h a t the cognitive and the voluntative factors are contrasted with one a n o th e r at all in behaviour seems thus to be a consequence o f th e use of, on the one hand , assertoric, and , on the o th e r hand , im perative an d in ten tional sentences; bu t, if this is so, then we canno t m ake use o f th e concept o f belief in explain ing the m ode o f em ploym ent o f an assertoric sentence.

In stead o f p resu p p o sin g th a t only a belief can co rrespond - in w hat­ever m ore precisely definable way - to an assertoric sentence, we should sta rt with the actual m odes o f behav iour with which an addressee responds to th e u tte ran ce o f an assertoric sentence. W hat d istinguishes the p a r tn e r ’s response to an assertoric sentence from the response to a signal is not th a t the p a r tn e r does no t resp o n d - b u t only believes som e­th ing - bu t ra th e r tha t the only actions o f the h e a re r which are rela ted in a ru le-governed way to the s ta tem ent o f the speaker a re them selves speech-acts an d consist in th e use o f a linguistic expression. Tw o lin­guistic responses to a sta tem en t tha t a re always possible a re the u tte r ­

The employment-rule of an assertoric sentence 185

ances ‘yes’ and ‘n o ’. E quivalen t to these a re the rep e titio n o r denial o f the s ta tem ent, likewise the u tte ran ces ‘th a tp is t r u e ’ an d ‘th a tp is false’. C learly ‘n o ’ and ‘yes’ a re no t ju s t two possible responses, ra th e r they belong toge ther: the h e a re r can answ er with ‘yes’ or ‘no ’. In this ‘can’ is g ro u n d e d every o th e r linguistic resp o n se to th e u tte ran ce o f the speaker (a) in th e sense th a t th e re a re o th e r position-takings and equally the possibility o f abstain ing from tak ing a position and these possible responses are all g ro u n d e d in the u n d e rs ta n d in g o f the yes/no a lte rn a­tive (b) in the sense th a t every o th e r linguistic resp o n se which can be re g a rd e d as a rep ly to the sp eak e r’s u tte ran ce p resupposes explicitly or im plicitly one o f the position-takings. B u t if you th in k tha t in saying this I am already asserting too m uch it d o e sn ’t m atte r; it is en o u g h if you ad m it th a t the h e a re r can always rep ly with ‘n o ’ o r ‘yes’.

H ow can a th eo ry such as th a t o f G rice accom m odate such facts? T h e ‘yes’ reaction could be accom m odated relatively easily: the hea re r the reby m akes it know n, on e m igh t say, th a t he accepts the belief o f the speaker. B ut w hat abou t ‘n o ’? Clearly o n e canno t say tha t the h e a re r is the reby m aking it know n th a t he does n o t accep t the belief o f the speaker. I f he says ‘n o ’ this does no t m ean tha t he does not believe tha t som eth ing is the case b u t th a t he positively believes th a t som eth ing is no t th e case. Shall we th en say th a t som eone w ho replies with ‘n o ’ the reby m akes it know n th a t he believes the opposite o f w hat the sp eak er believes? B u t th en h e a re r a n d speaker w ould no t be co n tra ­d ic ting one an o th e r . T h e two sen tences ‘A believes tha t p ’ and ‘B believes tha t n o t-p’ do no t co n trad ic t each o ther; an d o f course this is equally tru e o f th e sentences A in tends to b rin g it ab o u t th a t# believes th a t A believes that/>’ and *B in ten d s to b rin g it ab o u t tha t A believes tha t B believes tha t not-^>.’

W e have now a first im p o rta n t clue to the em ploym en t-ru le o f an asserto ric sen tence. W e m ust first clarify w hat it is th a t som eone is do ing who uses an asserto ric sen tence i f th e p erson to w hom he speaks can resp o n d to it w ith ‘n o ’ an d this u tte ran ce is to be u n d ers to o d in such a way th a t the h e a re r is co n trad ic ting th e speaker. I f w hat th e speaker is do ing is to be in te rp re te d as try ing to b rin g som eth ing abou t then it rem ains un in te llig ib le w hat it is th a t th e h e a re r is con trad ic ting o r what it is th a t is d en ied o r affirm ed by the h e a re r . I f we ask ourselves w ithout p reconcep tions w hat is it th a t is d en ied by th e h e a re r the answ er is tha t clearly it is th a t which the speaker asserted .

W e w ould now have a new hypothesis as to w hat the action o f the speaker and th e function o f th e sen tence consist in; w hat th e speaker is do ing w hen he uses an asserto ric sen tence is asserting som eth ing and

Analysis o f the predicative sentence 186

the function of the sentence consists in its being used to assert som e­th ing o r to m ake an assertion.

B ut o f course this im m ediately raises the question: w hat does it m ean to assert som ething? We have already seen in the provisional charac ter­ization o f assertoric sentences in the in troduction tha t the assertion of a sentence ‘p ’ contains a tru th-claim ; tha t it is asserted tha t it is tru e tha t p. B ut we canno t be con ten t with such explanations. T h e question is: if the function of the sen tence is so defined then how are the em ploy­m ent-ru les o f the sen tence to be understood? Negatively we can say: when^f asserts tha t p this is an act, som ething he does, bu t contrary to G rice’s way o f defin ing the speech-act - the em ploym ent o f the sign - this is no t an act th a t can be defined as bringing, o r try ing to bring, som ething about. A ccordingly we can also say o f the sign, the assertoric sentence, tha t the p u rp o se fo r which it is used is simply the act itself, the assertion.

In the case o f an action tha t is d irected towards an effect one first determ ines the in te n d ed effect; and then by m eans o f this one defines the action. O ne can th en ask fo r the ru le which the action - o r the em ploym ent o f m eans - m ust follow in o rd e r to achieve the purpose. In the case o f an action which contains its purpose in itself this division in to two stages does no t apply. As the assertion-act is not defined by reference to an in ten d ed effect it can only be defined by the action-rule itself (which o f course is always to be understood as the em ploym ent- ru le of the sign) and again this m eans tha t the em ploym ent-rules are not to be construed as in stru m en ta l rules. As Searle has show n,4 the rules m ust be rules th a t are constitutive of an action. B ut now w hat sort o f rules a re they?

T h e thesis tha t the genu ine speech-act, the act which Grice called m eaning (in the sense o f vouloir dire), is no t to be defined in term s o f the in ten tion o f p ro d u cin g an effect orig inates from J. L. Austin. A ustin distinguishes th ree acts which are p erfo rm ed in using a sentence.5 T o speak of d iffe ren t acts which are fo u n d ed in one an o th e r is standard in action-theory and co rresponds to o u r o rd inary way o f speaking. It derives from the fact th a t an act is defined by the in ten tion which gov­erns it, by its pu rpose, and tha t bodily acts are p erfo rm ed in o rd e r to achieve som ething, in o rd e r to achieve som eth ing fu rth e r , etc. T h u s ends a re p u rsu ed which in tu rn are in ten d ed as m eans to fu rth e r ends. F or exam ple I p e rfo rm a certain hand-m ovem ent on the window catch in o rd e r to open the window, in o rd e r to let fresh air in. A nd this m eans: I let fresh air in (act C) by open ing the window (a c t# ) and I do this by m oving my h an d in a p articu la r way (act A). O f course w hether

The employment-rule o f an assertoric sentence 187

it is correct to speak here of several acts ra th e r than several ways of describing one and the sam e act is d ispu ted in action-theory:6 bu t we do not need to bo ther about this here.

Back now to speech-acts. A ccording to A ustin we are to distinguish(1) th elocutionary act, th e u tte rance o f certain structu red sounds (2) the illocutionary act, the m eaning (in the sense o f vouloir dire), thus in ou r case the asserting; the relation betw een these two acts is such that one can say: he u tte rs the sentence ‘p y in o rd er to assert that p or a lte rna­tively: he asserts that/? by u tte ring o r em ploying the sentence ‘p ’ and (3) the perlocutionary act, th a t act o r those acts which in tend effects and which one seeks to achieve by m eans o f the illocutionary act e.g. ‘he asserts tha t p in o rd e r to convince the p a r tn e r ’ (in o rd e r to b ring it about that the p a rtn e r believes tha t p)\ or pu tting it the o ther way round : ‘he seeks to convince the p a rtn e r by asserting tha t p ’.7

A ustin’s contribu tion consists in having highlighted th e illocutionary act as a distinct act o r act-description, and the semantically relevant one, as against the perlocutionary act tow ards which G rice’s theory is o rien ­tated. A ustin ’s line o f though t, the central concern o f which lay else­w here (though I need no t go into this here), has subsequently been developed as a theory o f m eaning by Alston and Searle;8 b u t in ne ither o f these au thors do we find a satisfactory characterization o f the em ploym ent-rules o f this act.9 Searle, it is true , has constructed a whole system of rules in g rand style and in the process has p roduced a n u m ­ber o f ideas which can serve as guidelines; these how ever have rem ained undeveloped . T hus he gives the following rules for the illo­cutionary act o f assertion, and hence for the em ploym ent o f an asser­toric sentence: (1) th a t an assertoric sentence fp* is only to be used w hen th e speaker believes th a tp and (2) th a t the em ploym ent o f this sentence ‘counts as an un d ertak in g to the effect that/? represen ts an actual state o f affairs’.10 H ow ever, it is not clear, no r does Searle explain, what he m eans by ‘und ertak in g to the effect tha t’. F u rth e rm o re it is now here shown what the connection is betw een these two rules - which Searle calls the ‘sincerity ru le ’ and the ‘essential ru le ’.11 So it rem ains unclear w hether these are two in d e p en d e n t conditions which m ust both be fu l­filled or w hether the one is the consequence o f the other. I shall re tu rn to this difficulty later (p. 214ff) and for the p resen t confine my a tten ­tion to the ru le which Searle obviously regards as the fundam ental one: the ‘essential ru le ’. I shall deal with the first p a r t o f Searle’s fo rm ulation o f this rule - ‘an u n d ertak in g to the effect th a t . . .’ - la ter (p. 199); it m ust somehow concern th e act-character o f assertion.

It is the second p a rt o f the fo rm ulation (‘th a t p represen ts an actual

Analysis o f the predicative sentence 188

state o f affairs’) th a t is crucial. This explanation , you will perceive, rem ains com pletely em pty, fo r in it every th ing tha t shou ld be explained is simply p resupposed ; m oreover with its re feren ce to actual and n o n ­actual states o f affairs it w ould lead stra igh t back to th e object-orien­tated approach . I t is obvious th a t one canno t explain the use o f a sen­tence to som eone by saying th a t som eone who uses p ’ wishes thereby to say tha t ‘p ’ rep resen ts an actual state o f affairs. T h e inevitable co u n te r­question is: w hat a re states o f affairs an d how does one tell w hether they are ‘actual’? Now because Searle uses *p' to re fe r no t to the asser­toric sen tence b u t to the propositional co n ten t - thus as I have used ‘*p*’ _ We m ust re fo rm u la te S earle’s explanation in th e following way: som eone who uses ‘p ’ wishes thereby to say th a t the state o f affairs tha t p actually obtains. B u t the state o f affairs that/? actually obtains w hen it is tru e that/? (above p. 44). A nd as we also*have to om it the expression ‘wishes to say’ (‘m eans’) in explain ing the use we can say even m ore simply: w hen a speaker uses an assertoric sentence ‘/?’ he asserts that/? and w hen he asserts that/? he asserts th a t it is tru e that/?.

You will perhaps find this a ttem p t to sim plify Searle’s ‘essential ru le ’ ridiculous and say th a t all it achieves is th a t this ru le loses all exp lana­tory force. W hat is the use, you may ask, o f know ing th a t som eone who asserts tha t/? asserts tha t it is tru e that/?? In the first place, th e w ord ‘asserts’ occurs again in the explanation . In th e second place, I can now be accused o f w hat I always accused the o thers o f doing: nam ely of sm uggling into the explanation a w ord which itself has first to be explained: the w ord ‘tru e ’.

T o this I should reply: (1) Searle’s sta tem ent does indeed lose its a p p a ren t explanatory value as a result o f this sim plification; how ever we m ust dem olish pseudo-explanations to clear the way for real exp la­nations; (2) tha t th e w ord ‘assert’ occurs again in the explanation shows tha t this sta tem ent can at m ost rep resen t a first step tow ards an exp la­nation; (3) - and this is the crucial po in t - I have no t sm uggled in the word ‘tru e ’; ra th e r this word belongs to assertoric speech itself. N either S earle’s com plicated sta tem ent nor its sim plified fo rm have any value in them selves; its sim plified form how ever is in te resting for us because it belongs directly to the contex t o f m odes o f behaviour reflection on which has led us to conclude th a t the use o f an assertoric sentence is to be construed as assertion. It is these m odes o f behaviour them selves to which we m ust tu rn if we w ant to know w hat an assertion is, fo r the question o f w hat an assertion is can be n o th in g bu t the question: in accordance with w hat rules is this act perfo rm ed?

Let m e recall: th a t the use o f an assertoric sentence is to be construed

as assertion resu lted from the fact th a t th e h e a re r can rep ly to th e sp eak e r’s u tte ran ce with ‘yes’ (or ‘th a t is tru e ’) o r ‘n o ’ (or ‘tha t is false’) an d th a t th e u tte ran ce ‘n o ’ is to be u n d e rs to o d as m ean ing th a t the h e a re r is con trad ic ting th e speaker. In the m ost im p o rta n t trad ition o f m o d ern philosophical sem antics (a trad itio n which ex tends from F rege th ro u g h W ittgenste in ’s Tractatus to T arsk i, C a rn ap and D avidson an d to which I shall be re fe rr in g again) the w ord ‘tru e ’ occupies a cen tra l place, b u t only as a w ord which the sem antic th eo ris t uses; an d in sem antic theo ry the w ord ‘tru e ’ can also be u se fu l w here a language is them atized in which th e w ord ‘tru e ’ does no t occur (e.g. a signal- language).12 O n the o th e r hand th e fact th a t asserto ric speech is a fo rm o f speech in which th e use o f the w ord ‘tru e ’ itself has a constitutive ro le has so fa r been m ore o r less ig n o red with th e rem ark ab le exception o f M ichael D u m m ett’s recen t book on Frege.

How th e fact th a t the use o f th e w ord ‘t r u e ’ belongs to assertoric speech is to be in te rp re te d is som eth ing th a t will con tinue to occupy us fo r a long tim e; indeed we shall see th a t th e analysis o f the use o f the w ord ‘tru e ’ coincides with the analysis o f th e use o f assertoric sentences. W hat can be said h e re an d now is:

(1) the rep lies by m eans o f (a) ‘n o ’, (b) th e den ial o f w hat is asserted , (c) ‘th a t is false’, a re clearly equivalen t in m ean ing . Likewise the replies by m eans o f (a) ‘yes’ (b) the repe tition o f th e assertion (c) ‘th a t is t r u e ’. O ne can th e re fo re say th a t a lthough the yes/no reply has a w ider field o f application than asserto ric sen tences th e analysis o f ‘yes’ and ‘n o ’ as used in th e con tex t o f assertoric speech is iden tical with the analysis o f th e w ords ‘t ru e ’ and ‘false’.

(2) It is n o t only th e h e a re r w ho can say ‘th a t is tru e ’; th e speaker too instead o f sim ply asserting that/? can assert th a t it is tru e that/? (it was precisely this th a t my sim plified version o f S earle’s s ta tem en t b ro u g h t out). B u t in so do ing h e h im self is deny ing th e possible negative rep ly o f the h ea re r . A nd as ‘it is tru e th a t/? ’ is equ ivalen t in m ean ing to the orig ina l s ta tem en t ‘p ’ th e sp eak e r’s s ta tem en t already im plicitly contains a denial o f th e possible den ial o f the h e a re r (cf. p. 47f).

Now this does n o t yet enable us to give an analysis o f the em ploy­m ent-ru les o f an assertion; it does, how ever, enable us to take a first step tow ards such an analysis. F or now we can at least say: (1) N ot only a re the speech-acts with which th e h e a re r responds to the speaker - and , above all o th e r speech-acts, the yes- o r n o -response - re la ted in a ru le-governed way to the sp e ak e r’s u tte ran ce ; the o rig inal use o f the assertoric sen tence by th e speaker is also re la ted in a ru le-governed way to the yes/no-reaction o f the h ea re r. B u t this m eans th a t since the

The employment-rule o f an assertoric sentence 189

Analysis o f the predicative sentence 190

em ploym ent-ru les o f the assertoric sentence re la te neither to the cir­cum stances o f use n o r to an in ten d ed effect it can now be expected tha t the em p loym en t rules which we are seeking are som ehow m ediated th ro u g h the possible con trad ic tion o f the h ea re r. (2) T h e h e a re r’s response , at any ra te in its basic form s o f ‘yes’ and ‘n o ’, is itself an asser­tion. B u t th e re is an essential d iffe rence between this kind o f h e a re r’s response and th e o th e r th ings one w ould describe as responses. I t is not ju s t th a t the h e a re r , as reg a rd s the ru le-governed relation to the u tte r ­ance o f the speaker, can equally well respond with ‘yes’ o r ‘n o ’ o r an absten tion an d th a t hence th e re is a specific a re a o f freedom here; r a th e r the characteristic fea tu re at any rate o f den ial is th a t it refers back to the u tte ran ce o f th e speaker. A nd indeed this coun ter-u tterance o f th e h ea re r is re la ted to th e sp eak er’s u tte rance in precisely th e same way th a t the sp e ak e r’s u tte ran ce is re la ted to the h e a re r ’s u tte rance ; this is because, as we saw earlier, (p. 46f) the re is no absolute distinction betw een affirm ative and negative statem ents. W e can only say th a t the la tte r is the den ia l o f th e fo rm er. B u t then the fo rm e r is equally the den ial o f the la tte r. T his results in a far-reach ing relativization o f the d istinction betw een speaker and h ea re r. If the h e a re r responds with ‘n o ’ th e d istinction reduces to this: th a t the original speaker m akes so to speak the first ‘m ove’. T h u s in so far as the rela tion between speaker and addressee is n o t a one-w ay stree t it co rresponds neither to the stim- u lu s-re sp o n se schem a n o r to the G ricean conception o f a purpose- re la ted act. I t is no t ju s t th a t the act o f the h e a re r reacts u p o n the speaker o r his act; ra th e r bo th acts clearly relate - though o f course in a way th a t has yet to be exp lained - to the sam e th ing : the one denies what th e o th e r affirm s. M oreover, the affirm ing, an d likewise th e ques­tion ing , d oub ting , etc., responses o f the h e a re r re fe r back to the sp eak e r’s u tte ran ce in fundam en tally th e same way as denial, nam ely as d iffe ren t position-takings to the sam e th in g whose negation is asserted in the denial. F or all these responses take place against the background of the possibility o f denial, hence p resu p p o se denial as a possibility. A nd because all o th e r possible responses by m eans o f speech-acts also p re ­suppose one o f these position-takings they too a re not m ere responses to the sp eak er’s u tte rance . O ne can call all these speech-responses which p resu p p o se th e possibility o f denial, includ ing denial itself, answers instead o f responses.

W hat have we thereby achieved? Still very little. I f we confine o u r­selves to the possibility o f answ ering ‘yes’ or ‘n o ’ as the basic possibility then th e only characteristic o f the use o f an assertoric sentence so fa r to em erge is tha t it - as an assertion - anticipates a denial o r is itself the

denial o f an o th er assertion. I have already po in ted tha t ou t in the in tro ­duction (p. 47); an d this o f course does n o t am ount to an em ploym ent- rule. I t is conceivable how ever tha t this confron ta tion o f two opposed assertions provides an initial basis on which to conduc t our search for the ru les which d e term in e the em ploym ent o f an assertion. As the two assertions are re la ted to one an o th er in such a way tha t the one calls ‘false’ w hat the o th e r designates ‘tru e ’ the confron ta tion clearly con­cerns the tru th o f the statem ent. But w hat is ‘tru th ’ and how can the relation to it be reflected in em ploym ent-rules?

The employment-rule o f an assertoric sentence 191

L E C T U R E 1 5

Positive account of the employment-rule of assertoric sentences in terms of the truth-relation

In the last lecture I s ta rted ou t from the assum ption tha t before we can determ ine the em ploym ent-rules o f predicates, and o f the singular term s which supp lem en t p red icates so as to form elem entary assertoric sen ­tences, we m ust first possess a prelim inary conception o f the em ploy­m ent-ru les, an d tha t m eans o f th e m eaning, o f whole assertoric sen­tences. W e need a reasonably solid conceptual basis for the enquiry in to the u n d ers tan d in g o f th e com ponents o f an elem entary assertoric sen ­tence, even if this basis can itself only be subsequently consolidated by the analysis o f the u n d ers ta n d in g o f the com ponents o f the predicative sentence.

Even this question o f a m erely provisional u n d erstan d in g o f the em ploym ent-ru les o f assertoric sentences tu rn s ou t to be exceedingly difficult. I will sum m arize the results achieved so far:

(1) A lready in the lecture befo re last we w ere able to exclude the behaviouristic o r quasi-behaviouristic conception according to which the em ploym ent-ru le rela tes to circum stances, the em ploym ent-situation.

(2) Likewise excluded is the object-orientated conception which relates the em ploym ent o f the sentence to a rep resen ta tion or idea (however this is to be in te rp re ted ) o f a state of affairs o r actual state o f affairs; such an exp lanation w ould be a hysteron-proteron because the state of affairs can itself only be identified by m eans o f sentences.

(3) I t n ex t seem ed plausible to in te rp re t the em ploym ent-ru les as functional ru les and to rela te them , on the one hand , to the addressee and , on the o th e r han d , to the consequences o f the speech-act. B ut the question was: How? T h e m ost na tu ra l th ing seem ed to be to view the act as being defined by an in tended effect and the rules as in strum enta l rules; this conception led to the speech-act being defined as an act o f in form ing . H ow ever, the correct sta tem en t th a t som eone who, vis-ä-vis a p artn e r, u tte rs an assertoric sen tence *py thereby inform s him tha t he

believes that/?, an d th a t this m eans th a t in a specific way he in tends to b ring it about tha t the p a r tn e r believes tha t he believes that/?, also proved to be an unsu itab le basis fo r a rriv ing at the em p loym en t-ru le o f the sentence.

(4) I then suggested , appealing to the illocutionary-act theory , tha t we should u n d e rs ta n d th e sem antically re levan t act o f em ploying an asserto ric sen tence no t as an act o f com m unication but as an act o f assertion. H ow ever, we could find no satisfactory answ er in Searle to the question concern ing the em ploym ent-ru les o f an assertion.

(5) I f the em ploym ent-ru les can only be re la ted to the p a r tn e r and to the consequences o f the act then the only adequate p rocedu re seem ed to be to ask: w hat a re the possible responses o f th e add ressee that a re re la ted by a ru le to th e speaker’s u tte rance? It em erged tha t these responses a re them selves speech-acts and tha t u n d erly in g them all a re the answers by m eans o f ‘n o ’ or ‘yes’ o r an in te rm ed ia te position-taking, o r an absten tion from tak ing a position , and th a t th e possibility o f ‘n o ’ possesses a fu n d am e n ta l significance fo r all o th e r position-takings, a ‘n o ’ which is clearly used as equivalen t in m ean ing to the expression ‘th a t is false’. I t is only this circum stance which perm its one to call the use o f an asserto ric sen tence ‘assertion ’. A nd it also becam e clear th a t th e ‘no ’ itself expresses an assertion. I t belongs to the sense o f an assertion th a t it contains a re fe ren c e to an assertion con trad ic ting it.

T h is is how fa r we h ad got. T h e question we m ust now ask is w hether this is a possible basis fo r finding th e em ploym ent-ru les o f assertoric sentences the exp lanation o f which could claim to be the exp lanation o f th e ir m eaning . I t is only th e sense o f these em ploym ent-ru les tha t could d e te rm in e ju s t w hat is m ean t by calling a speech-act ‘assertion ’ - a des­ignation w hich so far I have m erely taken as a fact from o rd inary usage.

B u t first I w ould like to deal with two doub ts tha t m igh t be raised abou t my p ro c e d u re at th e end o f th e last lec tu re , a p ro ce d u re which co n cen tra ted entirely on one reply, viz. th a t with the w ord ‘n o ’.

(1) Since I have m yself em phasized the im p o rtan ce o f the fact th a t in assertoric speech the w ord ‘tru e ’ occurs, it w ould seem necessary th a t th e fu r th e r analysis shou ld no t only be o rien ta te d tow ards the ‘no’ bu t shou ld also take account o f the fact th a t the ‘n o ’ o f assertoric speech has th e specific sense o f ‘th a t is false’. A nd do we no t th e re fo re have to prov ide a t least a provisional exp lana tion o f th e w ords ‘tru e ’ and ‘false’?

(2) How fa r is one ju stified in accord ing such a special position to the negative reply? Even if it is clear th a t all o th e r position-takings take place against th e back g ro u n d o f th e possibility o f denial it could n o n e ­

Positive account o f the employment-rule 193

Analysis of the predicative sentence 194

theless be argued that, for the addressee, prior to any position-taking, prior also to the withholding from any position-taking, is the under­standing o f the speech-act. We must o f course accept that this under­standing o f the assertion already belongs to the context o f possible position-takings towards the assertion; however it is not identical with any position-taking, though it does seem to be the primary hearer-cor- relate o f the speech-act. If we free the notion, introduced by Grice, o f meaning in the sense o f vouloir dire from Grice’s own interpretation o f it as communication and apply it to the present view that it is an assertion, then we can now also say: the hearer understands what the speaker means, he understands what he wishes to say, he understands the asser­tion. And this understanding is not just a theoretical assumption that we can make in order to explain the transition from the hearing o f an assertion to the hearer’s taking up his own position; rather there are also responses belonging to the hearer’s behaviour in which the pure understanding o f the assertion is expressed. T he hearer says, for exam­ple: ‘(I understand what he is saying:) He is asserting that/?.’ The analysis of the employment-rules o f an assertoric sentence will have to pay par­ticular attention to this understanding on the part o f the hearer, for when we ask about the meaning o f a linguistic expression we are asking what it is to understand it; and indeed we can now say: what the hearer understands when he understands the speaker’s assertion is, precisely, the employment-rules o f the assertoric sentence. True, one must distin­guish between the understanding o f the linguistic expression and the understanding o f the speech-act, but if what is understood when the linguistic expression is understood is its function, then the two things belong together: to understand an assertoric sentence is to understand what assertion it can be used to make; and one can then say o f this assertion itself that it is understood. However, the reference to the pos­sible denial o f the hearer is not relativized by this inclusion o f the hearer’s understanding; it is merely supplemented. Someone who understands an assertion understands it precisely as one to which an assertion denying it can be opposed. And as we have seen this means: the assertion is so understood that another assertion can say that what it asserts is false - or that it is true. But this means: whoever understands the assertion understands it as one which can be true or false.

But what do we mean by these words ‘true’ and ‘false’? I thus come to the other point. Before attempting, on the basis o f what has now been achieved, to look for the employment-rule o f an assertoric sentence must we not first explain the use o f the words ‘true’ and ‘false’? Now one could say that this has already been done, for I pointed out that

Positive account of the employment-rule 195

‘that p is false* is used equivalently with the denial o f the sentence and ‘that/? is true’ equivalently with the denial o f this denial, hence with the original sentence. T he meaning o f the word ‘true’ would then be given by the equivalence: that p is true = p.

It is customary to refer to the theory that the meaning o f the word ‘true’ is defined by this equivalence as the Redundancy Theory,* for it appears to make the word ‘true’ superfluous: instead o f saying ‘that/? is true’ we can always simply use the original statement 7?’ itself.

Almost everyone who is confronted with this theory for the first time has the feeling that it suppresses something essential, something one could perhaps call the statement’s relation to reality. We must try to form a clearer idea o f what underlies this feeling. To this end the best thing to do would be to start with the way in which this relation to reality was dealt with in the truth-definition o f the philosophical tradition. T he traditional definition o f truth goes back to a definition o f Aristotle: ‘For to say that what is the case is not the case or that what is not the case is the case is false; but to say that what is the case is the case and that what is not the case is not the case is true.’2 Since Aristotle thought that there are negative and positive statements, in an absolute sense, he defined their truth and falsity separately. I f we ignore this peculiarity o f his definition it turns out to be identical with the Redundancy Theory: a statement that something is the case is true if it is the case.

However, Aristotle also explained his definition by saying that truth consists in a correspondence between statement and thing3 and this in turn led to the traditional formula o f the adequatio intellectus et rei,4 the agreement (Übereinstimmung) o f the thought with the thing. T he inde­terminateness o f the expressions employed in this formula led in the philosophical tradition (which for the most part simply took this formula, rather than the actual use o f the word ‘true’, as its point o f departure), to the most phantastic theories such as, e.g., that truth is the coincidence (Zusammentreffen) o f thinking and reality, the unity o f subject and object;5 and that things become true by being thought; and it would then also seem plausible to suppose that a statement only becomes true by being verified, for only then would thinking come into contact with the thing itself. Unbridled speculations o f this kind, which merely spin out an uncomprehended traditional formula and have lost contact with the matter itself (die Sache selbst) - the actual understanding o f words - are clearly not worth debating with.

If the traditional formula permits a meaningful interpretation at all then it can only be this: since that with which the thought is supposed to agree is construed as an object (as ‘thing’) ‘the thought’ too is to be

Analysis o f the predicative sentence 196

construed as an object and no t in the sense o f ‘th inking’. T he only version o f the fo rm ula w orth discussing is th a t which construes the ag reem ent as one betw een what is believed (or w hat is asserted) and w hat is real, such that what is believed is the objective correlate o f the actual statem ent an d w hat is real is the objective co rre la te o f the tru e statem ent. Since the objective correla te o f a sta tem ent is a state o f affairs th e re would resu lt the following definition:(1) the asserted state of affairs, that p, is true if and only i f it agrees with the corresponding real state of affairs, the corresponding fact.

T his conception o f two states o f affairs, a believed or asserted state o f affairs on the one hand , a real state o f affairs on the o ther hand , which in any event som ehow ‘co rresp o n d ’ (entsprechen) and in the case o f tru th also ‘ag ree’ (übereinstimmen) founders on the impossibility of cashing the im ages o f co rresp o n d en ce and agreem ent. In particu lar it cannot be specified w hat the real state o f affairs ‘co rrespond ing’ to the asserted state o f affairs is supposed to be w hen the assertion in question is false, n o r w herein the relation o f ag reem ent is supposed to consist. W e can how ever re -fo rm u la te (1) in such a way that the reference to a co rre­spondence o r an ag reem en t is d ro p p ed but w hat was in tended by the fo rm ulation is preserved:(2) the asserted state of affairs, that p, is true if and only i f it is a real state of affairs (a fact).

T h e re is now no longer any talk o f two states o f affairs; it is the same state o f affairs th a t is being asserted th a t in the case o f tru th is real and which we then call a fact. O ne can still call (2) a form ulation o f the ag reem en t-theo ry o f tru th although the word ‘ag reem en t’ no longer occurs in the definition. H ow ever, the question now arises w hether we should so to speak read this equivalence from left to righ t o r from righ t to left. W hat is defined by what? T h e claim o f the object-orientated ag reem en t-theo ry is tha t the word ‘tru e ’ is explained by reference to the reality o f the state o f affairs. But this presupposes (1) that we already u n d e rs ta n d w hat is m eant by a state o f affairs that/? before we u n d e r­stand the sen tence and (2) th a t the re is a p roperty o f states o f affairs W th a t e ith er is reality or is th e criterion o f reality, and that we have to exam ine states o f affairs with respect to this p roperty in o rd e r to decide w hether the assertion that/? is true. H owever, things are obviously the o th e r way ro u n d : if we a re to explain to som eone w hat the p roperty in question is all we can say is: the state o f affairs asserted by m eans of a sen tence ‘/?’ is real (a fact) if and only if it is tru e that p.

I f we now d ro p the object-orientated com ponents o f (2) we can attem pt to construe the rela tion to reality to which the w ord ‘tru e ’ is supposed

Positive account of the employment-rule 197

to re fe r not as the p roperty o f an object bu t adverbially and in this way arrive at the form ulation:(3) that p is true = really p, fo r exam ple: ‘(the assertion) tha t it is raining is true if and only if it really is ra in ing’. B ut now it is obvious tha t we can say ‘it really is rain ing’ if and only if we can also simply say ‘it is ra in ing’. This how ever reduces the last-m entioned form ulation of the agreem ent-theory to the form ula of the R edundancy Theory . T he word ‘really’ m erely underlines a contrast which clearly already belongs to the use o f the assertoric sentence itself. W hat is this contrast?

T h e o ther fea tu re which I b ro u g h t in today by way o f supp lem enta­tion, viz. the understand ing o f the hearer which precedes any position- taking tow ards the assertion o f the speaker, is o f help here too. W hen S says ‘the town-hall is on fire’ the response o f the h ea re r in which his u nderstand ing which precedes his own position-taking is expressed is the sentence ‘it is asserted tha t the town-hall is on fire’. A nd he can add: ‘is it really on fire?’ o r ‘is it tru e tha t it is on fire?’ T h u s in an assertoric com m unication-situation we always have these two things: the speaker’s sentence ‘p ’ and the h ea rer’s sentence (which does not have to be u ttered bu t always could be u ttered): ‘it is asserted tha t /?’. Inasm uch as what the speaker is doing, viz. asserting som ething by m eans o f the assertoric expression p \ is understood and possibly stated by the hearer (‘it is asserted . . .’) the now m odified expression ‘that /?’ loses its assertion- m om ent and can thus serve th e h ea re r as the basis for a position-taking of his own; for ‘that/? . . .’ can now be supplem ented by ‘is tru e ’, ‘is false’, ‘is doubted by m e’, etc. T h e w ord ‘tru e ’, like the word ‘real’, is a contrast- word. L earn ing to und erstan d it is learn ing to understand the contrast between '/?’ and ‘it is asserted tha t /?’ o r m ore generally: the contrast between ‘p ’ and ‘that/?’. C om pared with the incom plete expression ‘tha t p ’ the expression ‘p ’ contains a plus and it is this plus tha t is expressed in the supplem entation by ‘is tru e ’: the expression ‘is tru e ’ is tha t expression by m eans o f which we are able to so supplem ent the reduced expression ‘that/?’ as to obtain an expression which is again equivalent to the original assertoric expression ‘/?\ It is precisely this tha t the form ula o f the redundancy-theory states. It only appears trivial if one overlooks how essential the d ifference betw een '/?’ and ‘that/*’ is. T o explain the m eaning of the word ‘tru e ’ is to explain the difference in the m eaning o f ‘/?’ and ‘that /?’ which is essential to the use of assertoric sentences. W hoever moves from ‘(it is asserted) that /?’ to (p ’ moves from m erely u n d erstan d in g ‘p ’ to asserting that/?. H ence the explanation o f the w ord ‘tru e ’ is identical with the explanation o f the act o f asserting. For this reason the understan d in g of the word ‘tru e ’ as this is specified in the

Analysis o f the predicative sentence 198

form ula o f th e redundancy theory is only trivial if one assum es tha t one already u n d erstan d s the em ploym ent o f assertoric sentences.

W hat have we achieved with all this? We h ad already seen earlier th a t the ‘tru e ’/‘false’-response corresponds to the ‘yes7‘no’-response, o r a ffir­m ation an d denial. In addition it has now em erged th a t to the sense o f ‘trueT false’ th e re belongs not only the opposition betw een these w ords them selves which can be rep ro d u ced in the affirm ation and denial o f the relevant sentence bu t also w hat distinguishes the sen tence ‘p ’ from the reduced expression ‘that p’. This contrast-m om ent which is expressly m anifested in the w ord-pair ‘tru e ’/Talse’ th e re fo re also already belongs to the assertorically used Yes/No an d m ust be taken into account in the analysis o f the act o f assertion.

Let us now re tu rn to the po in t reached at the end of th e last lecture. I f assertion essentially anticipates the possibility o f a denial - a co u n te r­assertion - then it can be u n derstood as a challenge, in the sense in which one challenges som eone to take u p the counter-position in a gam e, e.g. in a bet. T h a t the em ploym ent-ru les of assertoric sentences can be u nderstood as rules o f a game is an idea which stems from W ittgenstein, which Searle too has taken up b u t no t really exploited, and which has been developed above all by D um m ett.6 F rom D um m ett com es the fu r ­th e r suggestion tha t assertoric speech can be com pared with the type o f gam e in which two p artn e rs play against each o th er and the rules are such tha t following them leads to a final-position which consists in the one having won and the o ther having lost.

Let us first be clear tha t the moves in a gam e are acts o f the kind we a re looking for, nam ely, acts the rules o f which relate no t to the circum ­stances in which they are p erfo rm ed but to th e ir consequences. But now these a re consequences no t in the sense o f in tended effects, bu t in the sense of consequences reg ard in g the outcom e o f the gam e which resu lt from following the ru les o f the gam e. T h e ru les also re la te the acts to a p artn e r; b u t they do so in such a way tha t the p a rtn e r is no t the object o f an in ten d ed effect, bu t the o p p o n en t who is essential to the gam e. T h e gam e is defined by the way in which the moves o f th e two p artn e rs a re rela ted by the ru les o f the gam e to the outcom e o f the game.

Now w hat is the gam e o f assertoric speech like? T h e assertion-act th a t consists in th e em ploym ent o f an assertoric sentence ‘p ’ is the open ing move. T h e rules o f th e gam e are such tha t th e p a r tn e r’s counter-m ove is already fixed by the opening-m ove; it consists in the u tte ran ce o f ‘no t p .’ By way o f analogy th ink o f bets (though in fact it is m isleading to speak o f ‘analogy’ h e re fo r a bet can itself only be und ersto o d as a m o d ­ification o f the assertoric gam e; how ever, the re ference to a bet may

serve to illustrate what I have in mind). Now how is the opening move to be understood?

I can now come back to that part o f Searle’s rule whose discussion I had postponed. As the essential rule for the employment of an assertoric sentence he offered: its em ploym ent ‘counts as an undertaking to the effect that/? represents an actual state o f affairs’ (above p. 187). I have already shown that the second part of this form ulation presupposes what has first to be explained and is therefore useless. But how is the first part to be in terpreted? What does ‘counts as an undertaking to the effect that . . mean? This rem ained unclear. However, it becomes sig­nificant if we in terp ret it as: ‘stands for a guarantee th a t . . .’ T hat some­one who uses an assertoric sentence ‘p ’asserts som ething means, we might say, that he offers a guarantee that it is true that p. His opponent on the other hand guarantees that it is true that not -p. W hat does this mean? Well, whoever guarantees som ething guarantees that certain conditions specified by him are fulfilled. W hat would these conditions be in the case o f the em ploym ent o f an assertoric sentence? Can we say: whoever uses an assertoric sentence guarantees that the truth-conditions of his assertion are fulfilled?

But what, you will ask, is m eant by this talk of truth-conditions? A condition is expressed in an if-sentence. So if that which someone asserts is to have a truth-condition we would have to envisage a form ulation of the following kind: ‘that/? is true, if . . .’ And the thesis just m entioned would therefore mean: if someone uses an assertoric sentence ‘p ’ (if he guarantees that it is true that p) then he guarantees that the condition referred to in the protasis of the above form ula is fulfilled. To be able properly to understand the significance of this suggestion one should com pare it with the idea - which was rejected - that the employment- rule relates the employment to the circumstances of employment. There we were dealing with a conditional rule. It had the form: ‘if . . . the em ploym ent o f the expression* is correct’. Now the ‘if . . .’ appears on the o ther side: ‘the em ploym ent of the expression* is correct (true), if . . T here it was said u n d er what conditions it is correct to use the expression. Now one is saying: if (for whatever reasons) the expression is used what then are the conditions under which it is correct. This reversed relation to conditions presupposes ( 1 ) that those conditions in which the expression is used (the em ployment-situation) are irrelevant to the correctness of its use - this is the situation-independence of em ploym ent that I m entioned earlier (but which will only be explained later) - and (2 ) that those conditions on which the correctness of the use o f the expression depends are those whose fulfilment is guaranteed by

Positive account of the employment-rule 199

Analysis of the predicative sentence 200

the use of the expression itself. W hat the expression guarantees is that the conditions of its correctness (truth) are fulfilled. It is this, then, that is meant by speaking o f truth-conditions. T he speech-act of assertion consists in guaranteeing its own truth-conditions.

We can now also begin to understand how it is that someone who uses an assertoric sentence ‘/?’ can equally well say ‘that/? is true’. The equivalence ‘p = that/? is tru e ’ is g rounded in the fact that someone who asserts something is always asserting the correctness (truth) of his asser­tion; that this is so is grounded in the essence of assertion as an act of guaranteeing.

In the preceding prelim inary discussion of the word ‘tru e’ I pointed out that this word expresses the contrast between '/?’ and ‘it is asserted that/?’ and that it is in the hearer’s understanding that this contrast first becomes prom inent. We can now see why this is so. Someone who gives a guarantee m ust always do two things: ( 1 ) he specifies the conditions whose fulfilment he is guaranteeing; (2 ) he guarantees their fulfilment. W hoever gives a guarantee does both these things; but there would be no act o f guaranteeing if he did not do both things at once. Now the person to whom the guarantee is given can only be said to understand the guarantee if he also understands both these things. But he would not understand the guarantee as a guarantee if these two things - what is guaranteed and that it is guaranteed - were not kept apart in his understanding.

Applied to the understanding of an assertion this means: someone understands the assertion made by means of an assertoric sentence if, firstly, he knows the truth-conditions of the assertion and, secondly, if he knows that the speaker is guaranteeing that these conditions are fulfilled. W hat he does not know, what is open for him, is whether the conditions are actually fulfilled, in o ther words, whether the assertion is true. Its being open for the person who understands the assertion w hether it is true is as essential to his understanding as his knowing that the person who makes the assertion asserts that it is true.

T he insight that one understands an assertoric sentence if and only if one knows its truth-conditions was first form ulated in W ittgenstein’s Tractatus: ‘T o understand a proposition (Satz) means to know what is the case if it is tru e ’ (4.024). This definition is, however, incomplete, for the understanding of an assertion is also characterized by the second of the above-mentioned features: it is also understood that the person employing the sentence is guaranteeing that it is tru e . 7 But now this second feature remains the same for every assertoric sentence. So if it is simply a m atter of explaining to someone the meaning of a particular

Positive account of the employment-rule 201

sentence and it can be assumed that he knows that it is an assertoric sentence, one can also simply say: the meaning of the sentence is explained by giving its truth-conditions.

The result so far achieved seems unsatisfactory, for two reasons: firstly, nothing has so far been said about how one can explain what the truth- conditions of an assertion or sentence are. One possibility would be (as the reference to an if-sentence suggests) that the truth-condition of a sentence is itself given by means of a sentence. The tradition of semantic theories that was inaugurated by Tarski8 is grounded in this possibility. It presupposes, of course, that in explaining a sentence one always has at one’s disposal another sentence which is already understood or, if it is a question of the meaning of a whole system of sentences, another language, a so-called meta-language. I pointed out in the debate with Grice that such a meta-linguistic theory is not sufficient for our funda­mental question of how linguistic expressions are understood. I would also rem ind you that so far we have seen no reason to abandon the idea that to explain a linguistic expression is to explain its employment-rule, and that means: its mode of employment. If we specify the truth-con­dition of a sentence by means of another sentence, this can only mean that the first sentence is used in the same way as the second; it does not mean that the mode of employment itself is shown. Besides, by the thesis that the meaning is given in a conditional rule we did not mean that the conditional rule is formulated in words, but, rather, that it is shown under what conditions the sentence is used. And clearly we must now hold on to the same theoretical claim.

Secondly, the purpose of speaking of a ‘guarantee’ and a ‘bet’ can only be to point in the right direction; such terms must now be put aside, for the following two reasons. Firstly, every act of guaranteeing something itself presupposes the use of an assertoric sentence. The explanation by means of the term ‘guarantee’ would thus be a pseudo­explanation. We would again be committing a hysteron-proteron. Secondly, the notion of a guarantee involves something more than the anticipation contained in an assertion, for it is essential to the concept of guaranteeing that in the event of the anticipated condition not being fulfilled the guarantor must reckon with certain negative consequences from the side of the partners. The same is true of the notion of a bet. Even some­one who bets ‘merely for the honour’ loses the honour that would be due to him were he to win the bet.

These two defects belong together. I used the word ‘guarantee’ in order to describe the act of assertion as the opening-move in a game. But if we want not merely to name the opening-move of a game but to

Analysis of the predicative sentence 202

define it, this can clearly only be done by specifying the consequences it has in the game, and that means: how it is related by the rule of the game to the outcome of the game. And only by defining an assertion in terms of the rules of the game-moves that follow it can we expect to be able to explain the truth-conditions guaranteed by the assertion by ref­erence to employment-rules.

It is only now that the crucial defect of Searle’s system of rules becomes clear. This system of rules ended precisely where it should have begun, namely with the mere nam ing of the opening-move of the game. This is no merely external criticism, for in his general preliminary reflections Searle himself compared speech-acts with game-acts and pointed out that a game-act is regulated by its consequences in the gam e . 9

By what rules then is the game of assertoric speech defined? AlthoughI want to drop the notion of a guarantee we can still be guided by it. It is essential to a guarantee that there are decision-criteria for its fulfilment. Someone who understands a guarantee knows the criteria by reference to which it is decided w hether it is fulfilled or not. Likewise someone who understands an assertion, though he does not know w hether it is true does know how it can be established w hether it, or the opposite assertion, is true; in o ther words he knows how it would be decided w hether the asserted truth-conditions are fulfilled or not. T he estab­lishing of w hether an assertion is true is called its justification or verifi­cation. Thus from a completely different starting-point we arrive at a thesis made famous by Logical Positivism, viz. that one understands an assertion if and only if one knows how it is to be verified, and that means: if one knows its verification-rule. Now if the person who under­stands the assertion knows how one establishes whether it is true, then the assertion m ust consist in the guarantee that, if it is tested as to its tru th , it will be established that it is true, and that means: that following its verification-rule will lead to success. This state of affairs could be expressed in the complicated formulation: the person who asserts that p in guaranteeing that its truth-conditions are fulfilled guarantees that the fulfilment of the truth-conditions, and hence the tru th of the asser­tion, is verifiable. But the interm ediate clause now becomes superfluous and there is no reason for not regarding the verifiability itself as the tru th condition of the assertion. This step is crucial for it removes the unclarity that has so far surrounded the question of what one is to understand by the truth-conditions of an assertion and how one can explain them to someone. If the truth-condition consists in the fact that following the verification-rule will lead to success, then giving the truth-

Positive account of the employment-rule 203

condition of an assertion will consist in dem onstrating its verification- rule or, to put it m ore simply, in showing how it is verified.

So instead of speaking of a ‘guaran tee’ we can now specify the rules of the gam e whose opening-move is the use of an assertoric sentence. T he gam e is like this: a speaker utters an assertoric sentence ‘p \ T he hearer is free to regard him self as a m ere spectator o r as a p a rtn e r in the game. In the latter case he assumes the role of opponent by uttering the negation of ‘p*. However, it is also enough if speaker and hearer know that the hearer (or som eone or other) could assume the role of opponent. The rule of the game consists in the verification-rule. T he ver- ification-rule is such that following it leads to a positive result either for the speaker or for his opponent. T he gam e-outcom e is defined by the consequence that an agreem ent is reached between speaker and oppo­nent such that either the speaker assents to the opponen t’s original assertion or vice versa.

T h e act of assertion can now be defined. Rem em ber I only appealed to the notion of assertion for the purpose o f defining anew the use of an assertoric sentence, after it had become clear that the use of such a sentence can be defined neither by the circumstances of its use nor by an in tended effect. From the outset our purpose was to find the em ploym ent-rules which we explain (or understand) when we explain (or understand) the meaning of an assertoric sentence. And now we can say: one understands an assertoric sentence if one knows what function it has, viz. the function of being used to perform a particular assertion-act. This act or the use o f the sentence is defined as the open­ing-move of the gam e just described. A nd that means: following a certain rule - the verification-rule - leads to a result the consequence of which is that the opponent agrees with the speaker or vice versa. T he specifi­cation of the complex of rules and actions (the ‘gam e’) to which the use of an assertoric sentence belongs spells out what was merely h inted at by describing som eone who uses an assertoric sentence as guaranteeing that its truth-conditions are fulfilled and, subsequently, as guaranteeing that his assertion is verifiable, or that the application of the verification- rule will have a positive outcom e for him. W hat it means to guarantee a positive outcome shows itself in the way in which the consequence of the gam e-outcom e is connected with the opening-m ove by the verifica- tion-rule. T he consequence of the gam e-outcom e does not have the character of an effect. R ather it is a consequence which is draw n by the players in accordance with the rule of the game, such that if someone refused to draw the consequence which results from following the ver­

Analysis of the predicative sentence 204

ification-rule we would say that he did not understand what an assertion is, or that it showed that although he u ttered a sentence he did not understand its meaning.

It could be objected to my analysis that I have defined an assertion as the opening-m ove of a game at the end of which there are again asser­tions, namely the agreeing statements of the two opponents. Isn’t this circular? We must first try to get a clearer picture than so far achieved of the connection between the result of following the verification-rule and the game-outcome. Following the verification-rule obviously leads to a situation in which no one who understands the assertion is any longer free to affirm it or to deny it. I say obviously it leads to such a situation, for there is no other way of interpreting the fact that even the person who denied it must now affirm it. T hat there actually are such rules the following o f which has a result that one can characterize by saying that the assertion turns out to be true (or false) could of course only be shown by the actual explanation of these rules and so far I have not done this. So far all we can say is: that following the verification- rule leads to such a situation is shown by the fact that playing through the verification-rule has the consequence that one of the two opponents sees himself compelled to accept the assertion of the other. T he assertion at the end of the gam e thus has a pre-em inent character: one cannot contradict it in this situation - at the end of the game - without laying oneself open to the charge of not understanding it. But is it possible to treat the use of an assertoric sentence which occurs at the end of the game as itself an assertion if I define assertion as an act o f guaranteeing, or as the opening-m ove in the game? I think it is. We must allow the limiting-case of a trivial act of guaranteeing, or the trivial case in which opening-move and concluding-move coincide.

T o fu rth e r characterize the game-outcome by saying that the speaker has won if the opponent has to agree with him and otherwise has lost seems at present to be superfluous; it is only in a later connection that we will see how far this description is necessary. For the present we can say that the rules of this game are not such as lead to an outcome which consists in a player having won o r lost but ra ther in an assertion proving to be true or false. This is exactly how the outcome of the game is described by the opponents in the game itself. T he assent of the oppo­nent forced upon him by the rules of the game is expressed in the sentence ‘Your assertion has tu rned out to be true, mine false.’

Verification-rules are distinguished from other game-rules - and this constitutes their unique character - by being rules of justification, i.e. rules the following of which decides whether the assertion of the speaker

Positive account of the employment-rule 205

or that of his opponent is correct - where the word ‘correct’ has the sense of ‘true’. In contrast to all the rules so far referred to - conditional rules, instrum ental rules and other kinds of game-rules - the character­istic feature of justification-rules is that the correctness of an act does not simply consist in its following the rule; ra ther it is only the correct following of the rules that decides whether the original act is correct in the absolute, no longer rule-relative, sense of ‘tru e’.

W hen I provisionally characterized the employment of an assertoric sentence as guaranteeing that its own truth-conditions are fulfilled (a characterization which of course continues to be valid; it was merely insufficient) I pointed out that when it is simply a m atter of explaining the meaning of a particular sentence to someone and it can be assumed that he knows that it is an assertoric sentence one can also simply say: the meaning of the sentence is explained by giving its truth-conditions. Similarly I can now say: as all other features of the verification-game are the same for all assertoric sentences then, when it is only a m atter of explaining the meaning o f a particular sentence and one can assume that it is known that it is an assertoric sentence, one can simply say: the m eaning of the sentence is explained by showing how it is verified; to understand an assertoric sentence is to know its verification-rule.

But precisely this aspect of the present theory, which is clearly central and of decisive importance for particular explanations, has so far rem ained undeveloped. I have not yet shown how a verification-rule is explained. But nothing can be said about this in general. How an asser­toric sentence is verified is something that must be shown separately for each form of assertoric sentence. One cannot get further than the result so far achieved so long as one speaks about assertoric sentences in gen­eral. My immediate aim was simply to arrive at a - necessarily hypo­thetical - preliminary conception of the employment-rules of assertoric sentences as a foundation for the enquiry into the employment-rules of the components of a predicative sentence, the most elementary form of assertoric sentence (p. 179).

So the next step would seem to be to set about the task of clarifying the employment-rules of predicates and singular terms. If the conception now arrived at is correct then the employment-rule of the singular term and the employment-rule of the predicate together constitute the veri­fication-rule of the predicative sentence. From this fact alone, viz. that the verification-rule of the predicative sentence is founded in two other rules, it is clear that the elucidation of the verification-rule of the pre­dicative sentence-form will present peculiar difficulties. Consider also the following point. In the refutation of the thesis that the employment-

Analysis of the predicative sentence 206

rule of assertoric sentences relates to the circumstances of em ploym ent we encountered the peculiar situation-independence of the employment- rule of assertoric sentences. It is this situation-independence of employ­m ent which makes it possible for the em ploym ent to be determ ined by rules of another kind (with which we have now become acquainted). In particular this situation-independence is clearly essential to the fact that one can use not only the word ‘correct’ but the word ‘true’: an assertion is once and for all true or false. So far I have presupposed this aspect of the employm ent of the word ‘true’, not explained it. How this situa­tion-independence is constituted is som ething that the analysis o f the most elem entary sentence-form would have to show. I have already hinted that it is the function of singular terms to make possible this situation-independence (p. 161). The analysis o f the verification-rule of predicative sentences is thus made even more difficult. I will therefore postpone once again the treatm ent of predicates and singular terms in o rder first to show by reference to simple cases how the m eaning of a sentence can be explained by giving its truth-conditions and how these can be explained by dem onstrating the sentence’s mode of verification. These simple cases are those in which the truth or falsity of an assertion depends simply on the tru th or falsity o f other assertions. Sentences employed in this way are the complex sentences form ed by means of ‘and ’ and ‘o r’, and so-called general sentences. T he discussion of these sentence-forms will also provide an opportunity to dem onstrate, from another angle, the inadequacies of object-orientated semantics.

L E C T U R E 16

Supplements

T he multiplicity of semantic theories I have touched on in the last two lectures in the process of trying to achieve what seems to me to be a tenable prelim inary concept o f the m eaning of an assertoric sentence may have left behind a certain confusion. So before taking up the prob­lem to which the line o f thought of the last lecture led it seems to me to be necessary to insert a lecture devoted to surveying what has been achieved. This will enable me to say som ething about the connections between the various positions and to add a supplem ent that will be im portant for what will follow.

/.

T he various theses about the m eaning of assertoric sentences (a.s.) are as follows.(1) One understands an a.s. if one knows for which state of affairs it stands.(2) O ne understands an a.s. if one knows in which circumstances it is to be used.(3) One understands an a.s. if one knows what its truth-conditions are.(4) O ne understands an a.s. if one knows what its verification-rules are.(5) O ne understands an a.s. if one knows what belief the person who uses it communicates to a hearer.(6 ) O ne understands an a.s. if one knows which assertion-act a speaker can perform with it (illocutionary act theory).(7) One understands an a.s. if one knows the verification-game whose opening-m ove is perform ed with it.

T he most striking thing both about the line of thought as it has turned out and about virtually all im portant m odern theories is the central position which the concept of tru th suddenly acquires. You could say:

Analysis of the predicative sentence 208

that statements can be true or false is something one has always known. Before Frege, however, no one had hit upon the idea of defining the meaning - the sense - of a sentence by means of its truth-conditions. T he ‘reality-relation’ of statements was also understood as tru th in the philosophical tradition, but this relation always rem ained pre-defined as a relation to things (res), to beings. It is only when one is primarily orientated towards sentences ra ther than towards names that it seems natural to start with the possible tru th of a sentence and by reference to this understand even its meaning. But how is this tru th that is no longer grounded in an orientation towards objects to be understood? T here is a great tem ptation simply to presuppose the concept of tru th as an unanalysed basic concept, as was done with the concept of an object in traditional philosophy. And in fact this is precisely what most analytical philosophers do, both those who have regarded the formula of the Redundancy Theory as an answer to the question of the meaning of ‘true’ ra ther than merely a starting-point, and also those who, in the question o f the m eaning of an assertoric sentence, have been content with thesis (3). These two views, which are often held simultaneously, contradict one another. For the idea that the word ‘true’ is eliminable, as the Redundancy Theory claims, is incompatible with the idea that it is indispensable to the determ ination of the meaning of a sentence.

If one is w ondering w hether there is an alternative to the object- orientated approach - thesis ( 1 ) - and the orientation towards the con­cept of tru th , then it is clear that this is not provided by the action- theoretical account given in (5) or (6 ), for this must either itself have recourse to the concept of tru th or revert to the object-orientated approach. T he only alternative seems to be the conception represented by (2 ) which relates the sentence to the circumstances of its employment. T he three great conceptual alternatives for understanding the reality- relation of a sign thus seem to be ( 1 ) the relation to an object (2 ) the relation to circumstances of use (3) the relation to truth. It is of course a crude simplification to speak of alternatives here. We shall see that just as the object-orientated approach took some account of the truth- relation, so too the tru th-orientated approach includes the relation to objects as a necessary component.

Thesis (2), we have seen, is hopelessly inadequate. Now that we have arrived at the tru th-orientated conception we can also see why it is. Despite a ra ther dangerous ambiguity which the word ‘circumstances’ thereby acquires we can contrast theses (2) and (3) by saying that according to (2 ) to understand a sentence is to know in which circumstances it is to be used, whereas, according to (3), to understand a sentence is to

Supplements 209

know in which circumstances it is true. T he proponent of thesis (2) found it necessary to re-interpret what we can now recognize as the grounds of the truth of a statement as the conditions under which the statement is used. Why this was hopeless can now easily be seen. Employment- grounds and truth-grounds are two entirely different things. T he con­dition of our significantly using a statem ent is not that we have grounds for its truth but only that we know what they are; and this is precisely what is meant by saying that we know what its truth-conditions are.

From a methodological point of view, however, thesis (2), which is false, has an advantage over (3), which is correct. It meets the require­ment, which thesis (3) does not meet, that the meaning of an expression must be explained by explaining how it is used. Admittedly Wittgenstein’s principle, which I called the fundam ental principle of analytical philos­ophy, does not go quite as far as this. It says only that the meaning is what the explanation of the meaning explains. And one can say that we explain the meaning of a sentence by giving its truth-conditions. But such an explanation is bound to be circular so long as one is unable to explain the truth-conditions themselves. And this can only be done by showing what one has to do to justify the statement or, putting it another way, by showing how one verifies the statement. Therein lies the supe­riority of thesis (4) over thesis (3), a superiority one can also describe by saying that to speak of tru th , both in general and in reference to a particular statement, remains empty so long as one does not explain the word ‘true’ itself; and this one can only do by showing how one recognizes that a statem ent is true.

Thesis (4) is superior to (3) inasmuch as it explains truth-conditions by reference to a rule of action, but it does not show in what sense this rule of action is the employment-rule of the sentence itself. Besides, the explanation it gives would apply just as much to the reduced expression ‘th a t/ / as to the assertoric sentence itself, *p\ T he same truth-conditions, or the same verification-rule, hold for both expressions. But whereas if one only says ‘that p ’ one leaves it open whether the truth-conditions are fulfilled, whether following the verification-rule will lead to success, if one uses the sentence ‘p ’ one asserts that the truth-conditions are fulfilled or that following the verification-rule will lead to success. It is only thesis (7) which does justice to this aspect of the meaning of a sentence. In this thesis I have given a precise meaning to thesis (6 ) - which as it stands is vague - that an assertoric sentence is used to make an assertion, by combining it with (4) and hence also with (3).

It is only on the basis of (7) that it becomes intelligible how by explain­ing the verification-rule one explains the employment-rule of the sen­

Analysis of the predicative sentence 210

tence. T he em ploym ent-rule is not identical with the verification-rule - the em ploym ent-rule is the rule which relates the em ploym ent via the verification-rule to the game-outcome - the verification-rule is, however, that com ponent of the em ploym ent-rule of an assertoric sentence that distinguishes it from the employment-rule of another assertoric sentence. Perhaps it still strikes you as strange that the rule which concerns the verification of a sentence should be its employment-rule. It is not the verification-rule, you might say, which determines the use of a sentence. Certainly not. But the employment-rule of something which has a func­tion is not what ‘determ ines’ its use, if this means: what motivates or somehow causes its use, or what determ ines in which circumstances it is used. What we are looking for under the heading ‘em ploym ent-rule’ is what we explain to someone when we explain to him the use of a lin­guistic expression. And when we explain to someone the use of an assertoric sentence we are not explaining what the occasions, circum­stances or motives of its use are. Rather we are showing him what some­one who uses it is guaranteeing and how he does this.

Summarizing we can say: thesis (1) commits a hysteron-proteron. (2) is false but contains an im portant methodological principle. (3) and (4) are correct but insufficient. (6 ) is correct but indeterm inate and all three are incorporated in my thesis (7). T h ere remains (5), the thesis which agrees with (7) inasmuch as it too starts out from the assumption that the em ploym ent-rule of an assertoric sentence relates to its function but which interprets this function as communication.

II

We have already seen that thesis (5) is incorrect because it founders on W ittgenstein’s fundam ental principle (p. 182). O n the o ther hand, the statem ent th a t / 2 when, vis-ä-visB, he utters *p\ intends to bring it about that B believes that A believes that/?, is correct. Accordingly if (7) is a correct analysis o f the em ploym ent-rule of ‘p ’ then this statem ent must follow from (7). I must therefore supplem ent the critique of (5) from the standpoint of my own conception by placing the two conceptions into a positive relationship to one another.

But first a rem ark about terminology. Concerning Grice’s ‘m eaning’ (vouloir dire) there are two possibilities: Either the notion of m eaning is understood as being correlative to that o f understanding (such that one can say: the hearer understands, or does not understand, what the speaker ‘m eans’) - and this appears to be the m eaning which ‘m ean’ or vouloir dire (and the Germ an meinen) actually has in ordinary linguistic

S upplements 211

usage. But in that case it is false tha t what a speaker means by ‘p ’ is that he wants to bring it about, etc.; ra th e r this is what he would mean if, he said ‘I want to bring it about, etc.’ W hat he means by u ttering ‘p ’ is to assert that p. It is this that the o th er person understands. Or I g ran t Grice his terminology. I then have to say that the function of an assertoric sentence, or the intention with which such a sentence is used, is not that o f m eaning som ething with it bu t ra th e r tha t of asserting something with it.

And now to business. Both in Grice and in my in terpretation the use o f a sentence is understood intersubjectively. A ccording to Grice, how­ever, the addressee is the object o f an in tended effect; whereas in my account he is a partner in a game. This m eans (1) that in Grice’s account the communication-act is one-sided, whereas in my account it is in trin ­sically reciprocal: the speaker addresses the hearer as som eone who can take up a position towards what he says. This anticipation of position- taking belongs to the m eaning o f an assertoric sentence (cf. p. 189f). From this, however, it follows that (2) in my in terp re ta tion the speaker can assume the role o f the addressee in the capacity o f ‘N o’-utterer. T he phenom enon o f speaking to oneself presents no difficulty for this con­ception; even if we are speaking to ourselves the use o f an assertoric sentence consists in an assertion in the sense described. It is only this conception which makes it possible, on the one hand, to take account o f the intersubjective character o f speech and, on the o ther hand, to avoid the absurd consequence that a sentence does no t have the same m ean­ing in soliloquy as it has in conversation with an o th e r . 1

On the o ther hand I m ust now adm it that in my account an essential function o f intersubjective assertoric speech, viz. its communication- function, has so far been omitted. W hen I u tte r a sentence ‘p ’ I do not normally do so with the intention of challenging the h earer to a verifi- cation-game, but ra th e r with the intention o f informing him (bringing it about in a specific way that he believes), by means o f my assertion, that p or at least that I believe that p. Even if my thesis that the communi- cation-function does not belong to the m eaning of the sentence is correct I must still be able to explain this function. As the notion of believing th a tp is clearly essential in the form ulation of the communication-func- tion we m ust first ask: what is m eant by belief, and in what relation does it stand to assertion?

I am not in a position to give a definition of ‘belief’; no r do I know w hether there is a satisfactory explanation of this word. One of the chief difficulties is this: we also speak of belief in regard to beings which do not speak and in regard to ourselves in contexts in which we do not

Analysis of the predicative sentence 212

speak and hence do not use an expression such as ‘p ’ or ‘that p \ ‘Belief’- as Peirce was the first to em phasize2 - refers to an act-disposition. We say of a being that it believes that p if in its actions it takes account of the fact that p. Alternatively one can say that it relies on or banks on that p, that in its actions it presupposes that p. Belief is the cognitive disposition which, together with the voluntative or instinctual disposi­tions, determ ines intentional action; and, as I have already pointed out (p. 184), p rior to the use of sentences there is no action in which this disposition is manifested by itself (so it can only be an external description when we speak of a belief o f an animal by means of the expression ‘that p ’ which refers to a sentence).

In contrast to this broad concept o f belief (Glauben) we can define a narrow er concept, using for purposes of terminological contrast the term opine (Meinen):3 A opines that p = def. if A is presented with the question ‘p or no t-p?’, and if he has no intentions which go beyond the game-outcome, he will assert that p. This concept thus refers to the tendency or readiness of a person to guarantee the truth of the assertion that/? in so fa r as he has no intentions which go beyond the outcome of the game. This qualifying clause can also be expressed thus: ‘in so far as the person’s speech-act is determ ined only by the intention that the game should have a positive outcom e for him ’. So here we do find it necessary to speak of losing and winning the game, for the reason that here it is a question of the motivation for taking one side or the other or for abstaining. In describing the verification-game in the last lecture I ignored this question. I was able to do so because one can explain the game to som eone without reference to the question of the motivation for playing on one side o r the other. I said in the last lecture that the outcom e of the game does not consist in a player winning or losing but rather in an assertion proving to be true or false (p. 204). But this means that if we now consider the game from the point o f view of the intentions of the players, or their readiness to take one side or the other, the inten­tion of winning the game can equally well be described as a truth-inten- tion: the speaker intends to assent to an assertion that is tru e . 4 One can call this intention in the em ploym ent of an assertoric sentence, which is concerned only with the positive outcome of the verification-game, a purely theoretical intention. It is this purely theoretical intention which enters into the definition o f ‘opining’ through the qualifying clause ‘if he has no intentions going beyond the game-outcome’. Opining, like belief, is defined as a specific act-disposition. But here the act consists only in the em ploym ent of an assertoric sentence with the purely theo­retical intention ju st described. O ne can therefore call this act a purely

Supplements 213

theoretical act. O f course it is not these labels which m atter; what m at­ters is that in opining, as thus defined, we have a belief-disposition which represents an isolable cognitive disposition. Whereas the broad concept of belief refers to a cognitive disposition which can only be extrapolated as an item co-existing with voluntative dispositions to explain in ten­tional acts, in the case of the disposition o f opining one can speak of a purely cognitive disposition, inasmuch as the acts in which it is mani­fested are determ ined solely by the intention o f truth.

However, opining as thus defined - in other words assertoric belief - can now be subsumed under the broad concept of belief: someone who without regard to fu rther intentions has a tendency to assert that/? will also take account of thatp in his actions. On the other hand one cannot say that, whenever A believes that/?, he also opines that/?. The contra­position o f the previous sentence, however, clearly does hold: if he does not believe that p then neither does he opine that p; hence one can convict someone on the basis of his actions of the untruthfulness of his assertorically expressed opinions.

It would, however, be a mistake not to differentiate actions and lin­guistic utterances, but simply regard them as manifestations of one belief- disposition. The purely cognitive disposition of opining outlined by the definition just attem pted is an independent disposition whose definition contains the concept of assertion; through the connection thereby given with the verification-game it acquires certain differentiating and con­trastive features which do not belong to non-assertoric belief. (1) Whereas in the case of any belief one can speak of its causes, opinions also have grounds. One might even be tem pted to substitute the following defi­nition for the one I have given: A opines that/? = def. A expects that the assertion that/? can be shown to be true. T he definition I have given would then be the consequence of the definiens that has now been given. However, as a definition this suggestion would be circular, for ‘expects’ is simply another word for ‘opines’. (2) T he readiness to assert that/? is grounded in an explicit or implicit decision between the assertion that p and the assertion that not-p; thus to opinion, unlike non-assertoric belief, there belongs the contrast with the possibility of undecidedness, or of doubt w hether the assertion that p is true or false. Doubt too concerns the justifiability of the assertion. (3) At the other end o f the scale, if the opinion is combined with consciousness of the indubitability (and that means: the complete verification) of what is believed then the person concerned says not only that he opines that/?, but that he knows that p. We also make a distinction between belief and knowledge in the case of non-assertoric belief (e.g. of animals). But that is another concept

Analysis o f the predicative sentence 214

of knowledge according to which knowledge is correct belief; the classical definition of knowledge as a belief that is not just true but also adequately g rounded 5 clearly fits only assertoric belief. I f we now return to the question of the relation between assertion, belief, communication and the meaning of an assertoric sentence, then clearly as regards belief we can restrict ourselves to the narrow er concept of assertoric belief, i.e. to opining.

I shall begin with the connection between opining and assertion. T he definition of opining I have given defines this connection in a specific way. We can now test the adequacy of this definition by reference to the way in which assertoric speech itself expresses itself. Clearly, the hearer can receive the speaker’s assertion in such a way that he says: ‘He asserts th a tp; but he does not opine that/?.’ T he speaker himself, on the o ther hand, cannot say ‘p; but I do not opine that /?’. Searle has attem pted to in terpret this feature - that it is not possible to assert that p and at the same time openly adm it that one does not opine that/? - as one of the rules of assertion, or of the employment of an assertoric sentence. How­ever, the connection of this rule — which he calls the ‘sincerity ru le’ - with the in any case inadequately developed main rule (‘essential ru le’), in place of which I have put thesis (7), remains unclear. And, of course, this connection is bound to remain unclear so long as one has not decided whether opining is to be defined by reference to assertion or vice versa. The syncretism of Searle’s theory here reaches its high-point. On the one hand, the speech-act which consists in the employment of an asser­toric sentence is characterized as assertion and determ ined by the ‘essential ru le’. On the other hand, the em ployment of a sentence ‘p ’ is supposed to consist in the speaker’s taking the ‘responsibility’ for opining that p; but we are not told anything about the meaning of ‘opine’.

In fact there is a clear alternative here. Either one defines assertion - the em ployment of an assertoric sentence - by reference to opinion, thus for instance: ‘a sentence “p” is used to express the opinion that /?’- a definition that would be unobjectionable if only it were possible to give an explanation of the word ‘opine’ and of the meaning of the expression ‘that/?’ which did not have to appeal to the employm ent-rule of the sentence ‘/?’ defined by the verification-game. Or one assumes that this is not possible and defines opinion as I have done, by reference to assertion. T hen the fact that when one uses the sentence ‘/?’ one expresses that one opines that p - or in o ther words: cannot openly adm it that one does not believe that/? - cannot be a component of the employment- rule of *p\ Rather it would have to follow from the meaning of ‘opine’

Supplements 215

as I have just defined this word. T h a t it does indeed follow from this definition is easy to see. For if someone asserts that p, and hence guar­antees that it is true that p, then he cannot at the same time openly adm it that it is not his intention to assert something that is true. Someone who asserts tha t p w ithout opining that p perform s the opening-m ove of the verification-game without the intention of w inning the game; his opening-m ove is the opening-move of one who would have chosen this opening-m ove with the intention of w inning the game. On the basis o f the definition o f opinion we can also form ulate this as follows: he expresses by his assertion an opinion which is not his but that of a person who asserts the same with the intention of asserting som ething true, and this means: he deceives his p a rtn e r about his gam e-intention, and that means: about his opinion.

B ut now what does this mean for the act of assertion? Let us look at the m atter from the perspective of the hearer. If the latter notices that the speaker does not opine what he asserts then for the hearer this means that the player has m ade his move without the intention of win­ning. Nonetheless he has m ade a move which has its significance with reference to the game-outcome independently of the player’s intention of winning. Otherwise expressed, it means for the h earer that the speaker does no t stand behind his assertion, nonetheless he has made the assertion. T h e hearer will not now take the speaker seriously; he can, however, take the assertion seriously. W hether or not the speaker stands behind his assertion is im portan t to the hearer as regards the question of what weight he gives to the speaker’s utterance. But this circum stance does not belong to the sentence’s em ploym ent-rule, which is defined by reference to the game-outcome.

T h e re is m anifested here a peculiar independence of the game from the players, which in the previous lecture I described by saying that the outcom e of the game does not consist in a player’s having won or lost, but ra ther in an assertion’s proving to be true or false. T he same is true of the opening-move. In this game one and the same move can be per­form ed by several persons, one and the same guarantee can be given by several persons. If one of you says to me ‘T h e re ’s a m an waiting outside for you’, and immediately afterw ards ano ther person uses the same sentence, I will say ‘I ’ve already heard. Why do you say it again?’ On the other hand , they would clearly not both be com m unicating the same thing to me if one were to say ‘I believe that a m an is waiting outside’ and the other were to add ‘I believe so too.’ T he one assertion would thereby acquire grea ter weight for me. W hat is cancelled out if

Analysis o f the predicative sentence 216

someone openly admits that he does not opine what he asserts is not the assertion, the guarantee, but the dispositional participation of this individual in the guarantee.

Now that the connection between assertion and opinion has been explained and it has been made clear that the recourse to opinion is not necessary to explain the meaning of an assertoric sentence, but that ra ther the dependence is the other way round, we must ask: given that this is so, how are we to understand the communication-function emphasized by Grice? What I have so far said about the connection between assertion and opinion also applies to speaking to oneself, though of course in this case - trivially - the possibility of asserting something one does not opine no longer applies. In every monological employment of an assertoric sentence there is expressed an opinion of the person; and in the fact that all opinion is always also a belief, in the broad sense of this word, is grounded the practical significance of assertoric speaking to oneself.

It may seem strange that although it started out from the intersubjec­tive employm ent-situation, my description of the employment-rules of sentences does not so far contain the aspect of communication. It is im portant to be clear what the reason for this is. When one explains to someone how an assertoric sentence is used it is presupposed that the partner is an arbitrary person, so long as he knows the same employment- rules. In the same way, when the rules of chess are explained to someone it is assumed that one’s partner is an arbitrary person who knows the same game-rules. This arbitrariness of partner is a reason (though not the only one) why one can also play the verification-game with oneself (one cannot play chess with oneself). To understand how a sentence is used to communicate involves more than knowledge of the employment- rules, inasmuch as the communication is always a communication to specific persons, not arbitrary persons. If I wish to be understood by a particular person, then, in addition to intending to assert that/?, I must also intend to bring it about that this person notices that I am asserting that/?. In precisely the same way, if I am playing chess with someone and make a certain move, then, in addition to having the intention which governs the move and is related to the outcome of the game, I must also intend to bring it about that my partner notices that I have made this particular move. This ancillary intention is realized - in the assertoric gam e as in chess - by the player (a) seeing to it that his partner perceives which sign he is using and (b) - on the assumption that there are several signs or sign-systems for the same game-act - choosing a sign or sign- system (a language) which (i.e. the rules of which) is familiar to his part-

Supplements 217

ner. However, these presuppositions of communication (viz. that the partners reciprocally know of one another that they are using the same signs in accordance with the same rules, and that they use the signs in such a way that each notices which sign the other is using), no more belong to the employment-rules of linguistic signs than they do to the employment-rules of other game-signs.

If, in someone’s perceptual range, I use an assertoric sentence *p' the rules of which he knows, then he knows that I am asserting that/?. Now, if I have occasion to intend that the o ther knows that I am asserting that p, then I also have occasion to intend that he opines that I am seriously asserting that/?, and this means that I opine that/?. O f course I cannot bring it about that he knows that I opine that/?. But, since, for the most part, someone who asserts that/? also opines that/?, my partner, if he knows that I am asserting that p, also has a reason (even if not a sufficient reason) to opine that I opine that/?. And since, for the most part, when someone opines that /?, his opinion is more or less well- founded, my partner, if he knows that I am asserting that/?, also has reason himself to opine that/?. Thus by seeing to it that someone notices that one is asserting that/? one can inform (mitteilen) him that/?. For what is here called communication (Mitteilung), however, it is not only essential that one intends to bring it about, by means of conventional rules, that the hearer believes something (which is what Grice’s theory essentially comes down to), but that one intends to bring it about, by means of an assertion, that he opines something. This means that the communication is essentially exposed to the possibility of denial, and hence doubt, and belongs, therefore, to a potential dialogue; the effect is only achieved to the extent that the partner regards the o ther’s assertion as justified.

I ll

With this I conclude the debate with thesis (5) and would now like to indicate a difficulty which has probably been troubling you for some time and which will give me the opportunity to make a few supplem en­tary remarks. W hen I drew attention to the Yes/No response of the hearer there seemed no way of avoiding having to say: there is something that is affirmed or denied. And this was even clearer when the same response was expressed in the utterance ‘that is true/false’. It is to this same thing that the other position-takings relate and it was this too of whose truth-conditions I afterwards spoke. Is there not revealed here, you will have asked yourselves, an objectual component in my own conception? No doubt this is so. What we have to ask ourselves is (1)

Analysis of the predicative sentence 218

how should we interpret this objectual element? (2 ) how far can I none­theless claim that the recourse to it does not represent a reversion to the object-orientated approach?

Following ordinary linguistic usage I have called that which is true or false ‘the assertion’. One might therefore be inclined to think that the objectual elem ent that we have here encountered - that which is true or false - is the speech-act, the use of an assertoric sentence. For it was this that I have called assertion. However, it is easy to see that the expression ‘the assertion’ is ambiguous and that I have also in fact used it ambiguously : 6 ( 1 ) we can call an individual assertion-act one assertion (2 ) in ordinary speech we say that someone repeated one assertion or that several people made the same assertion. In both (1) and (2) it is th e a ^ r - tion-act that is referred to, but obviously we must distinguish between the act-event and the act-type. To this distinction there corresponds an analogous distinction on the part of the sign. When an assertoric sign ‘p ’ - or any other sign - occurs more than once, or is used more than once, we can say that there are several signs (sentences), that is, several physical occurrences of the same structure. But we can also say (and this is the more usual way of speaking) that it is one sign that occurs or is used several times. Following Peirce, the sign-event is referred to as the sign-token, and the one sign that occurs several times as the sign-type. It is clear that when we enquire about the rule of the employment of a sign we mean the sign in the sense of the sign-type. T he ambiguity in speaking of the assertion-act corresponds m ore or less exactly to this ambiguity in speaking of a sign.

But we m ust distinguish a fu rther ambiguity in the expression ‘the assertion’. We have seen that one can automatically supplem ent the sen­tence ‘This assertion is true’ as follows: ‘This assertion that/? is true .’ Can we still say that what ‘the assertion that/?’ refers to is the assertion- act (in the sense of the act-type)? Clearly we have to distinguish the assertion in the sense of the asserting (the assertion-act whether as act- type or act-event) and the assertion in the sense of what is asserted. It is only the assertion in this third sense - in the sense of what is asserted - which is that which we call true or false. The distinction ju st indicated corresponds not only to linguistic usage (we do not say of the speech- act that it is true or false) but also to a distinction we encountered in describing assertion as an act of guaranteeing: someone who uses a sen­tence ‘p ’ (1) indicates what he is guaranteeing and (2) guarantees it. This distinction is a necessary one, for it is possible to take up a position towards that which the sign-user asserts - thus towards that which he guarantees - which is o ther than that of guaranteeing. O ne can assert

that it - the same thing that had been asserted - is false or doubtful, etc. T he assertion-act which someone perform s when he employs the sentence thus consists in this: it asserts something; and that means: takes up a position towards the same thing towards which a different position can also be taken up, but to which, equally, the same position can be taken.

This implies that, because the Yes/No relates to tru th , there belongs to assertoric speech a basic relation to som ething identifiable, and this means: a relation to an object. If one considers assertoric speech, in contrast to a more primitive, circumstances-related language, its char­acteristic feature is that it relates to tru th and by virtue of this to objects. One m ust therefore say that the prim ary objects o f assertoric speech - those to which it relates qua assertoric speech - are those of which it can be predicated that they are true or false.

T he objects which we here encounter under the title ‘that which is asserted’ are of course the same as those we met earlier un d er the description ‘states o f affairs’ or ‘thoughts’ or ‘propositions’ (p. 43). The problem of these objects which can be true or false is usually treated by contrasting them with spatio-tem poral objects, which are taken for granted and regarded as unproblem atic. Within a limited enquiry such a procedure is justified, inasmuch as spatio-tem poral objects are iden­tifiable in space and time, whereas a state of affairs that/? can only be identified by saying: it is that which is asserted when one uses the sen­tence . So although talk of objects that can be true or false is as fun­dam ental as the use of assertoric sentences (as became clear in the analysis of the employment-rule of assertoric sentences in the last lecture) the identification of such an object that p presupposes the understanding of the em ploym ent-rule of the corresponding sentence ‘p’. H erein lies the difference from the object-orientated position (the explanation of the employment-rule of the sentence does not have to refer to the object that p. cf. above p. 2 0 2 ).

T he explanation of the ontological status of a state of affairs we have just given - that an object that p is identified as that which is asserted when the sentence *p* is used - is, of course, not satisfactory as it stands; it can merely serve to indicate the direction in which we have to enquire. For the explanation as it stands is peculiarly elusive and invites the fu r­ther question: what is that which is asserted when the sentence *p* is used? With this I take up a question which I left open in the introductory reflections on form al semantics (p. 44).

T he simplest answer to this question would be: it is the sentence itself, clearly in the sense of sentence-type. In that case, that which is called

Supplements 219

Analysis of the predicative sentence 220

true or false, and to which position-takings relate, would be the sentence. This view would not only be simple, but seems also to be presupposed by a widely-held thesis to which I too have appealed. I am referring to thesis (3) (above p. 207). I f we can only explain the meaning of a sentence by giving its truth-conditions, then this means that it is sentences that have truth-conditions, hence that it is sentences which can be true or false. This idea, that the so-called tru th-bearer is the sentence, would, of course, be subject to at least the following qualification: it is the sen­tence-type in so fa r as it is employed in a particular way. If the same sentence were employed differently (e.g. as a sentence of another lan­guage) then it would have different truth-conditions. T he state of affairs would thus be the sentence-in-a-particular-mode-of-employment. Hence, what I just asserted in general would also hold for this conception: the identification of the state of affairs presupposes the understanding of the em ploym ent-rule and is thus grounded in the understanding of meaning.

However, even with this qualification the view that the state of affairs is the sentence is not tenable. On the surface even linguistic usage speaks against this identification of sentence and state of affairs. It is true that in G erm an, in contrast to o ther languages, the word ‘Satz is used in both m eanings - both for the sentence-type and for what is asserted by means of a sentence-type. T ha t there are two meanings here is shown by the fact that the term is com pleted differently. On the one hand, we speak of the Satz ‘p \ on the o ther hand of the Satz that p. W hat lies behind this surface difference?

In the first place it is easy to see that (as I have previously pointed out) not only can the same sentence stand for different states of affairs when, namely, it is used according to different rules, but also that dif­ferent sentences stand for the same state of affairs if they are used according to the same rule, thus have the same meaning. If we say ‘Copernicus asserted that the earth revolves around the sun’, we do not mean that he used the English sentence ‘T he earth revolves around the sun’, but ra ther some sentence o r o ther that has the same meaning as this sentence. T hus the identity-criteria for the state of affairs that p and for the sentence *p* overlap.

However, if that were all, one could still hold that the state of affairs that/? - like the sentence-type ‘/ / - is som ething by reference to which we identify many sentence-events as one Satz. In the case of the sentence- type the unifying-principle is the form of the sound- or script-structure; in the case of the proposition that p it is the employment-rule. So this conception am ounts to equating the identity-criteria for states of affairs

Supplements 221

with those for employment-rules, and this means, for the meanings of sentences. And this means that the state of affairs that/? would be the m eaning of ‘p \ Now we have already seen (p. 118f) that this conception cannot be correct, for the state of affairs that/? lacks the assertion-mode that belongs to the m eaning of'/?’ and which, in the case o f the expres­sion ‘th a tp ’, is only added by the supplem entation ‘is true’.

Besides, the view that the state of affairs that/? is the meaning of the sentence ‘/?’ collapses when applied to those sentences that contain deictic expressions. H ere we can no longer say that all sentence-occurrences with the same meaning stand for a proposition that p. For sentence- occurrences with one and the same meaning can have different truth- conditions, depending on the situation in which they are used (by which speaker, at what time); and sentence-occurrences with d ifferent m ean­ings can have the same truth-conditions, depending on the employment- situation. T he two sentence-occurrences ‘I am hungry now’ and ‘You were hungry yesterday’ do not have the same meaning; but they do have the same truth-conditions, or stand for the same state of affairs that I am hungry now, if the first sentence is used by me now and the second by you tomorrow.

T he consideration of deictic expressions calls in question not only the attem pt to construe the tru th-bearer as in some way the sentence or as a classification-principle of sentence-occurrences. I f one cannot speak of the truth-conditions of a sentence at all, then, clearly, the definition of the meaning of a sentence as given by me in the last lecture by spec­ification of truth-conditions or the dem onstration of its verification-rule is also untenable.

However, both difficulties can be removed by an appropriate supple­mentation. It is not the sentence-occurrence, but the speech-event, that we m ust regard as the elem entary unit which provides the basis for the truth-bearer. In this way we return to the beginning of the present reflection where I started from the idea that the object that p is the assertion - in a particular sense of this word - and represents an abstract elem ent of the assertion in the sense of the speech-act. And the speech- act, we can now say, is not identified simply by the meaning of the sen­tence which it uses, but by the meaning of the sentence together with the situation in which it uses it. If a sentence contains deictic expressions, then it is only through the combination of a sentence used according to a certain rule with a certain employment-situation that there arises som ething that can be true or false. T he consequence of this for the meaning of the sentence is that to understand a sentence ‘/?’ that contains deictic expressions cannot be to know its truth-conditions, but rather

Analysis of the predicative sentence 222

what the truth-conditions of the speech-act are which uses it in situation x, where ‘x’ is a variable. This can also be expressed in somewhat technical language as follows: the meaning of the sentence is a function whose argum ents are the employment-situations of the sentence and whose values are assertions, i.e. objects which have determ inate truth-values (or have determ inate truth-conditions for the person who has identified them on the basis of the meaning of the sentence employed and the employment-situation).

However, we must go a step further. Not only does the employment of a sentence with one and the same meaning have different truth- conditions, depending on the employment-situation. It is also true, con­versely (as was shown by the examples ‘I am hungry’ and ‘You were hungry’), that sentences with different meanings, employed in different situations, stand for one and the same assertion with the same truth- conditions. Consequently what I just said, viz. that it is only through the combination of a sentence employed according to a particular rule with a particular employment-situation that there results something that can be true or false, must be correspondingly supplemented. T he combi­nation just described would characterize only an individual speech-event, and such a fleeting event is not the identifiable situation-independent something that can be true or false. The assertion — both in the sense of the speech-act-type and in the sense of what is asserted (the state of affairs that/?) - is rather the identical something whereby all speech- events which, through the use of different sentences in different situa­tions, have the same truth-conditions are united into one class. T he one assertion is thus the unifying principle relative to the many speech-events determ ined by the meanings of the employed sentences and the situa­tions, in the same way that the sentence-type and the proposition that/? were, in their different ways, the unifying principles relative to the many sentence-occurrences. W hat is meant by ‘unifying principle’ is that we count the many occurrences or events as one sentence-type, or as one Satz that/?, if they fulfil a certain condition. In the case of the sentence- type this condition was the form of the sound- o r script-structure; in the case of the Satz that/? it was the employment-rule. In the case of the assertion, on the other hand, it is not immediately clear what it is in virtue of which the many speech-events, which have neither the meaning of the employed sentence nor the situation in common, are counted as one assertion. To say that they have the same truth-conditions would be correct. But this would be to invoke as the ground of explanation the very thing that has to be explained. For the question is precisely: how

Supplements 223

is it that we can apprehend differenr. speech-events as having the same truth-conditions?

T he following conjecture, which we must later examine, suggests itself: if the meanings of sentences are functions whose argum ents are the em ployment-situations, and whose values are truth-conditions, this can only mean that the employment-rules o f the various sentences which, in their respective situations, represen t the same truth-conditions recip­rocally refer to one another; this systematic relation between the expressions must belong to their em ploym ent-rule. This systematic reciprocal relation, whereby the identical elem ent required for assertoric speech is constituted, m ust be the specific achievem ent of the employ­m ent-rule of those sentence-com ponents which relate the use of sen­tences to the situation: the deictic expressions. You may perhaps find strange this conjecture that it is o f all things the function of situation- related expressions to make the use of linguistic expressions situation- independent. But it can really only appear strange if one takes for granted the situation-independence of assertoric speech, as something obvious. I f we view assertoric speech against the background of m ore primitive, situation-dependent languages then we m ust ask: what are the linguistic means whereby the em ploym ent o f expressions is made independen t of situation in the m anner presupposed in speaking of ‘tru e’ and ‘false’. And what is then more natural than to suppose that this situation-independence is m ade possible by those expressions which expressly relate to the situation?

T hat this positive function of deictic expressions in the constitution of identifiability has hitherto not been seen is connected with the fact that, in the context o f the usual meta-linguistic semantic theories, one did not have to worry about how the expressions which refer to situa­tions, and which, as everyone admits, m ust figure in the specification of the truth-conditions of sentences with deictic expressions, are themselves explained. In the meta-language the em ploym ent-situation can be referred to by means of expressions which locate the situation in an already presupposed objective system of objects in space and time. T he nature of ou r enquiry clearly rules out presupposing the understanding o f any meta-linguistic expressions. We shall have to ask how the use of expressions which locate the situation in an objective system of objects is itself to be explained. It will em erge - this too I can now merely anticipate as a thesis - that far from it being the case that the use of situation-related deictic expressions can be explained by means of such objective expressions, the use of these objective expressions, and with it

Analysis of the predicative sentence 224

the use of all singular terms which refer to spatio-temporal objects, can only be explained by recourse to the deictic expressions.

This suggests a whole new perspective for the problem of propositional objects and spatio-temporal objects. H itherto in analytical philosophy this problem has been regarded simply as one of identifiability and the two kinds of object have been discussed only in contrast to one another. In the course of a fundam ental discussion of the meaning of assertoric sentences, however, the basic question is: to what extent is something like a relation to objects essential to assertoric speech (in contrast to more primitive, situation-dependent languages)? And the primary object-relation, without which there could not be assertoric speech, con­cerns those objects which can be true or false. Now if the identifiability of these objects is grounded in the use of deictic expressions, and if the deictic expressions make possible the situation-independence essential to this identifiability by producing through their reciprocal relation a primary level o f identifiability in space and time, then one must conclude that a reference to objects in space and time is the condition of the possibility of the use of the expressions ‘true’ and ‘false’.

So you see, my opposition to the object-orientated tradition should not be interpreted as implying that a relation to objects is no longer essential. On the contrary, it emerges that a relation to objects is as basic as the understanding of assertoric sentences. The difference between my position and what I call the object-orientated position consists simply in this: the latter presupposes the object-relation as something self-evi­dent, or it views objects as correlates of the pseudo-concept of represen­tation (Vorstellen) and then puts them in place of the meanings of lin­guistic expressions; whereas what we really have to do is to show how something like an object-relation is only constituted in the rule-governed use of linguistic expressions, and this means: in the understanding of their meaning.

If the hypothesis regarding the positive function of deictic expressions for the identifiability of states of affairs, and hence for the possibility of using the words ‘true’ and ‘false’, is correct, then clearly we can no longer regard sentences with deictic expressions, as is usual in metalinguistic semantics, as a special case or as an unavoidable complication of so- called natural languages. In the case of sentences which contain no deictic expressions one can speak of the truth-conditions of the sentence and say that to understand the meaning of the sentence is to know its truth- conditions or its verification-rule. And correspondingly, in the case of these sentences, we can also regard the state of affairs th a tp as a classi- fication-principle of sentence-occurrences (as the ‘Satz that/?’).

But, firstly, it now seems desirable to regard this case as the special case and view the state of affairs that/? in all cases as a classification- principle of speech-events. In the special case in which there are no situation-components of the truth-condition the speech events can be identified by the rule of the sentence that is used, and the truth-condi­tions of the speech-events are therefore, trivially, reducible to the truth- conditions of the sentence-occurrences.

Secondly, if the thesis referred to is correct, we can clearly no longer assume that the two cases, deictic sentences and other sentences, are of equivalent status. If it is only through deictic expressions that the identity required for the use of the word ‘true’ is constituted, then the use of an assertoric sentence - a sentence with a truth-claim - that does not refer back in one way or the other to deictic sentences is simply inconceivable. If we disregard more complicated sentences - those for instance which refer to abstract objects - then we must envisage two forms of such a reference-back:

If it is correct that the use of all singular terms refers back to deictic expressions, then this means that in the case of all elementary sentences- all predicative sentences in which concrete (perceptible) objects are referred to - the truth-condition, and hence the verification-rule of the sentence, can itself not be explained without the use of deictic expres­sions. In other words, the use of an elementary sentence is a speech-act which, even if it employs a sentence without deictic expressions, only stands for a particular assertion that p if it belongs to a class of speech- acts which all have the same truth-conditions, and of which some use sentences with deictic expressions. Thus in the case of such sentences it is a mere illusion that one, can specify their truth-conditions without deictic expressions, an illusion that can persist only as long as one spec­ifies the truth-conditions by means of a meta-language, and which, as we shall see, disappears as soon as one attempts to explain the employ­ment-rule.

It is much simpler in the case of a second form of reference-back. T here are sentences which are such that the tru th or falsity of the asser­tion depends simply on the truth or falsity of other assertions. Thus no situation-components enter into the truth-conditions of these so-called truth-functional sentences or assertions. But this is only because they already have a situation-independent stratum of assertions that can be true or false as their foundation. So because it is essentially simpler to explain the verification-rules of these higher-level, truth-functional sentences I will begin the concrete presentation of the employment- rules of assertoric sentences with this most simple case and only then

5 upplemen ts 225

Analysis of the predicative sentence 226

come back to predicative sentences. At the same time it has now become clearer why, prior to the discussion of predicative sentences, we could only achieve a prelim inary concept of the general essence of assertoric sentences (cf. p. 179). T he situation-independence and identity funda­mental to the use of all assertoric sentences, and thus an essential aspect o f the m eaning of the words ‘true’ and ‘false’, is constituted at the level of elem entary sentences.

L E C T U R E 1 7

And’ and ‘or’

In the lecture before last we arrived at a first, prelim inary result in the enquiry into the m eaning of assertoric sentences. In the last lecture I related this result to o ther possible conceptions and then supplem ented it in an essential respect. I had started from the assum ption that to understand a linguistic expression is to understand its employment-rule. It became clear that the understanding of its em ploym ent-rule consists not in knowing in what circumstances it is used, but ra th er in knowing what its function is; and that this function consists in asserting something. So far we have only been able to determ ine in a general way what this means. To assert som ething is to perform the opening-m ove in a veri­fication-game. Such a gam e has the following defining features. T here are two mutually negating opening-positions. T h e rules of the game are verification-rules. T he outcom e of the game is characterized thus: on the basis o f following the verification-rule the one assertion turns out to be true the o ther false or one assertion must be withdrawn in favour of the other. It followed from this general characterization of the em ploym ent-rules of assertoric sentences that the explanation of the use o f such a sentence m ust consist in the explanation of its verification- rule. It was this question that rem ained open and to which we must now turn . It had to rem ain open, because in general nothing can be said about the verification-rules of assertoric sentences. They m ust be shown separately for the different sentence-forms or, better: different sentence- forms are distinguished precisely by having different sorts o f verification- rule. (Should that prove to be correct then we would have a basis for rendering precise the concept of semantic form that has h itherto rem ained vague.)

T h e account given so far is thus not only abstract; it is also incomplete so long as it has not been shown how the verification-rules of specific sentence-form s can be explained. Only by doing this will we be able to

Analysis o f the predicative sentence 228

see how far this conception really provides a basis for the explanation of the meaning of individual sentences and sentence-parts.

Now we saw at the end of the last lecture that the view that the em ploym ent-rule (meaning) of a sentence is its verification-rule must be corrected. T hat which can be said to have truth-conditions or a ver­ification-rule is in general not the sentence but what is asserted. The sentence as such has in general no specific truth-condition or verification- rule. To know its em ploym ent-rule is not to know its verification-rule, but ra th er the verification-rule of the assertion which uses the sentence in situation*. At this point, however, we can disregard this complication, for I wanted to begin with simpler, though higher-level, forms of sen­tences which do not necessarily contain deictic expressions, thus sen­tences whose em ploym ent-rule does not contain a situation-reference. In the case of these sentences, therefore, one can say that the truth- condition, or verification-rule, of the assertion is also the truth-condition or verification-rule of the sentence. Hence, the explanation of the meaning of the sentence is reducible to the explanation of this verifi­cation-rule.

T he simplest type of such sentence-forms are the so-called truth- functional sentences, i.e. sentences whose truth or falsity depends on the tru th or falsity of other sentences. Sentences of this kind are (a) certain forms of complex sentences, in particular sentences of the form ‘p or q and (b) the sentences, traditionally known as general sentences, in which words like ‘all’ and ‘some’ occur. In both cases one has to do with expressions or sentence-forms which had also been taken account of in the tradition, though of course they were interpreted there in an object-oriented way. These forms, together with that o f the simple pre­dicative statem ent, played an essential role not only in m odern logic but also in traditional logic (the entire Aristotelian syllogistic is exclusively concerned with general sentences), and, as a result, the traditional idea of thinking included these forms as well as the predicative form. It would therefore seem to me worthwhile not to begin immediately with the truth-oriented analysis of these sentence-forms, but again take as our starting-point the traditional object-orientated conception. In this way we can give the confrontation of the language-analytical approach with the traditional position a broader basis.

How could the meaning of the word ‘and’ be explained from an object- orientated position? It would clearly be absurd to claim that this word stands for an object. But it is not necessary for traditional semantics to do so. For it had at its disposal the concept of a syncategorematic expression (p. 109). Rather it is characteristic of the traditional view

‘And’ and lor 229

to regard the ‘and’ as standing for an ‘aggregate’ (Zusammen) of objects. However, this can hardly be a m atter of real composition (Zusammenset­zung). When we say ‘Peter and Paul and Simon’ we do not mean that the three are joined together or in some way really (real) connected. But what then do we mean? If it is a m atter of composition at all it can only be a composition in ‘thought’ of the kind contained in Husserl’s concept of categorial synthesis. So if there is any possibility at all of interpreting the meaning of the word ‘and’ in an object-orientated way, then Husserl’s theory of categorial synthesis seems tailor-made for this purpose; and in fact in constructing his theory of categorial synthesis Husserl from the outset had also this case in mind.

T he thematic discussion o f ‘and’ and ‘or’ takes place in §51 of Inves­tigation VI of Logical Investigations under the heading ‘Collectiva and Disjunctiva’. However, the analysis is only carried out for ‘and’. This is probably no accident, for it is difficult to imagine what an object-orien­tated interpretation of ‘o r’ could look like. Now in the case of ‘and’ Husserl speaks of the ‘conjunctive combination of names or statem ents’. In other words, he treats the case in which the word ‘and’ occurs between names and that in which it occurs between statements as analogous. This is a logical consequence of his view that assertoric sentences also stand for objects, namely, states o f affairs (above p. 119). From this it follows that the ‘and’ that occurs between assertoric sentences represents a combination of states of affairs in the same way that the ‘and’ between names represents a combination of objects. This led to Husserl only explaining ‘conjunctive combination’ with reference to names, for on his view the transference of this explanation to the other case follows automatically.

He gives the following explanation: ‘T hat which corresponds to the words “and” and “o r” . . . cannot be grasped with one’s hands or app re­hended with some sense; just as it cannot be represented pictorially, e.g. in a painting . . . T here is here only the one possibility which is always open to us: that we perform a new act of conjunction (collection) on the basis of the two individual acts of perception and thereby mean the aggregate o f the objects^ and B .’ In the first sentence Husserl rejects a conception o f ‘and’ as the representative of a real combination. In the second he gives his positive explanation which grounds the synthesis in a ‘new’ (categorial) act of a special kind, an act of conjunction. To this act there corresponds ‘a unitary object which can only be constituted in this act-combination’. Husserl calls this object a ‘collectivum’ or a lter­natively an ‘aggregate’ (Inbegriff); by this he would appear to mean the same abstract object which in logic is usually called a ‘set’.

Analysis of the predicative sentence 230

Is it an intelligible explanation of the word ‘and’, at least as far as it occurs between names, to say that the objects that are designated by the names connected by the word ‘and’ together constitute a set? We should not let ourselves be impressed by the scientific respectability of the word ‘set’. The explanation that through the word ‘and’ several objects are combined to form a set explains nothing so long as we do not know what a set is. Now Husserl by no means merely presupposes this. On the contrary, he tries to explain what a set - au aggregate - is by re fe r­ring to the synthetic act in which, as he thinks, the word ‘and’ is expressed. Now how is this act described? As one which ‘means’ ‘the aggregate of the objects’. This description is Husserl’s explanation of the meaning of the word ‘and’.

Now the question is: is this description intelligible? Do we understand what it really means to mean the ‘aggregate’ of several objects? What is meant by this ‘aggregate’? The first thing we think of when we hear this word is some kind of spatial proximity. But Husserl has rightly ruled this out in rejecting the idea of a ‘real’ combination. When we say ‘Peter and Paul’ no spatial proximity is implied. T he one may be in Australia, the other in Costa Rica or no longer be alive. But what then is positively meant by the ‘aggregate’? Husserl would no doubt reply: we are here dealing with something entirely sui generis which can only be understood by perform ing the relevant act. ‘If we want to make clear to ourselves the meaning of the word “and” then we must actually perform an act of collection and in the thus genuinely presented aggregate bring to fulfilment a meaning of the form “a and b” ’ (Investigation IV §9 (b)). But now the question once m ore arises: how is an act of collection char­acterized? How do I tell that I am perform ing such an act and not some other synthetic act? Husserl is clearly moving round in circles: an act of collection is one which means the aggregate of objects and this aggregate is precisely what is constituted in such an act. Perhaps you will say: what enables one to tell that one is dealing with an act of collection is the fact that it is expressed in the word ‘and’. But that would mean that what we are to understand by ‘aggregate’ is explained by means of ‘an d ’; whereas Husserl wanted to explain the meaning of ‘and ’ by reference to the ‘aggregate’. But if we can only explain the ‘aggregate’ by means of ‘and’ we must be able somehow to understand the latter.

But how do we understand it? Since the object-orientated in terp re­tation has turned out to be a game with empty words, we shall have to try to find the step back which we took in the case of predicates. We must interpret the question about the m eaning of the word ‘and’ as a question about the explanation of its use. But then the prior question

‘A n d ’ and ‘or’ 231

immediately arises of whether the use of the word ‘and’ can be explained at all in the way proposed by the object-orientated conception, namely by having it simply stand between names - ‘Peter and Paul’ — without this word-sequence being supplem ented by a predicate. I believe it can be shown that such an explanation of the word ‘and’ is simply incon­ceivable. For how could we explain the use of such a part-expression ‘a and b and c’ by means of examples? Suppose we try to do so by somehow placing the objects together, or by drawing a circle and placing them in it. W hat we would then have explained would already include a certain predicate, e.g. ‘a and b and c are in this circle’. And we would thereby have included a real relation, som ething which Husserl has already rightly ruled out. Perhaps you think one could keep rearranging the three objects until it is understood that one is to abstract from the specific m anner in which they are placed together. But it is an illusion to think that we could in this way bring someone to understand that we m ean ‘and’; we will rather be understood to mean that the objects are somehow together. But there is no reference to such a being together or spatial proximity in ‘a and b\

Thus the expectation that one could explain the use o f the word ‘an d ’ purely in connection with names turns out to be an illusion. And it would be a corresponding illusion to suppose that one could explain what is to be understood by a set w ithout m aking use of predicates. Sets can only be determ ined with the aid of predicates; we say of all objects to which a predicate is applicable that they are elements of a set. I f we introduce the concept of a set in this way then it should also be noted that, contrary to H usserl’s opinion, an expression of the form ‘a and V does not represent a set at all. So long as we only use the expression ‘a and b\ then although we can say that a and b are elements of a set, we cannot say that together they constitute a set; for the expression ‘a and b’ leaves open the possibility that the set also contains o ther elements. You might ask: why shouldn’t one also be able to form a set consisting only of the elements a and b? Clearly one can do this - but not by means o f the m ere expression ‘a and b\ I f we wish to speak of the set consisting only of a and b then our form ulation must express the exclusion from this set of all other objects. So we would have to say: the elements of this set are a and b and no object that is not identical with a or b; and this can be expressed by means of a more complex predicate by saying: all objects to which the predicate ‘identical with a or W is applicable constitute the intended set. Notice that to introduce a set consisting only of a and b one has to use, not the word ‘and\ but the word ‘or\

You could make the following objection: from the fact that ‘a and b’

Analysis of the predicative sentence 232

does not stand for a set in the sense of set-theory it does not follow that H usserl’s idea o f an ‘aggregate’, for which the expression ‘a and b* is supposed to stand, is not meaningful. T hat is correct. However, perhaps it is precisely on the basis of this excursus into set-theory that we can now have a better idea of what possible meaning can be connected with this ‘aggregate’. We have seen that one cannot understand the expression (a and b’ to mean that a and b constitute a set, but merely that they are elements of a set. But elements of which set? For this set is not already defined by these two elements. Obviously, precisely that set that is defined by the predicate that follows the sentence-part a and b\ I believe this to be the only rem aining sense for H usserl’s idea of an ‘aggregate’. What ‘connects’ a and b is that they are both elements of a set which has been left open and has yet to be specified. But if this is correct then we can spare ourselves the whole detour into set-theory and say much more simply: the sense o f that ‘aggregate’ consists in this, that a predicate that has yet to be specified applies to both a and b. T he expression ‘Peter and Paul’ should be understood as merely a sentence-fragment. It points forward to a predicate F and the thus completed sentence - ‘Peter and Paul are F’ - clearly has the same meaning as the complex sentence ‘Peter isF and Paul is F.’ With this we would have reached an im portant intermediate result: not only do expressions of the form ‘a and b’ involve a reference Verweisen) to sentences, in the sense that they are to be understood as sentence-fragm ents; their completion to form a whole sentence shows that ‘and’ as it occurs between names refers back to ‘and’ as it occurs between whole sentences. This sheds new light on the pos­sibility of explaining the use of the expression ‘and ’ by means of exam­ples.

However, this result requires qualification. Not every sentence of the form ‘a and b are F’ has the same m eaning as ‘a is F and b is F\ We cannot convert a sentence such as ‘Peter and Paul are standing next to each other’ into ‘Peter is standing next to each other and Paul is standing next to each o ther.’ We must regard this sentence-form as a special case. In such a case one is dealing with symmetrical relations. A relation is term ed ‘symmetrical’ if whenever it holds between a and b it also holds between b and a. If the sentence ‘Peter is next to Paul’ is true, then the sentence ‘Paul is next to Peter’ is also true. Only in the case of non- symmetrical relations (e.g. ‘Peter hits Paul’) is it necessary for there to be a syntactical indication which shows the order of the two terms of the relation. N atural languages do this in two ways: by inflection and by word-order. In the case of symmetrical relations this requirem ent no longer applies. Hence in English we have the sentence-form in which

'And’ and ‘or 233

the order of the names of the two terms of the relation before the pred­icate is optional. T ha t the word ‘and’ also occurs between the two names has no additional semantic significance. ‘Peter Paul stand next to each o ther’ would have the same meaning. T he mode of employment of this ‘and’ therefore requires no special explanation.

We can thus ignore this special case and turn our attention to the employment of the word ‘and’ as it occurs between sentences. We can now also include the word ‘o r’ in the discussion. It is in any case obvious that when ‘or’ occurs between names (e.g. ‘Peter or Paul’) the expression requires to be supplem ented by a predicate and that the sense of the ‘or that occurs between names is reducible to the ‘or’ that occurs between sentences. I do not know how Husserl may have conceived, in the case o f ‘or’, the relation between objects analogous to the ‘aggregate’ o f ‘and’. A disjunction between objects? What could that be?

Suppose Husserl had followed me up to this point and would admit that the words ‘and’ and ‘or’ must be understood primarily in their role of combining sentences to form complex sentences. From his object- orientated position he would have to say that the ‘and’ and the ‘o r’ stand for certain syntheses between states of affairs. But this would be simply to transfer the interpretation-schem a which Husserl applied to the ‘and’ of names to the new level; and this would result in the same difficulties as before. If it were really a m atter of a synthesis between states of affairs then we should expect, not the expression ‘p and q\ but rather ‘thatp and that^’. Such expressions do actually occur and they are indeed analogous to ‘a and b\ for they too require to be supplem ented by a predicate, e.g. ‘T hat it is raining and that it is warm is pleasant.’ But with regard to this expression we would now have to say that it is equiv­alent to the higher-level complex sentence ‘T hat it is raining is pleasant and that it is warm is pleasant.’ We would thus find ourselves in an infinite regress.

So Husserl could not even have accepted the interm ediate result reached so far. To accept it is to have already given up the object-orien­tated interpretation of ‘and’ and ‘o r’.

What o ther interpretation is conceivable? Someone coming from the object-orientated tradition will now be inclined to pose the question like this: if the word ‘and’ combines neither objects nor states of affairs what then does it combine? However, as we saw in our treatment of predicates, what is mistaken in the object-orientated conception is not its view of what is connected by the predicate with the object of the subject-term of the sentence. T he error was more deep-seated. It consisted in the assumption that something is combined with something at all. The same

Analysis of the predicative sentence 234

holds for ‘an d ’ and ‘o r’. We must free ourselves completely from the presupposition that the meaning of these words consists in their com­bining in some way something with something. T he e rro r in the object- orientated conception of ‘and’ is not that it speaks of states of affairs (clearly there corresponds to the complex sentence ‘p and q a state of affairs that p and q, likewise to the constituent sentences *p' and ‘q the states of affairs that p and that q)\ ra ther its e rro r consists in thinking that the state of affairs that p and, q is somehow composed of the other two states of affairs similar to the way in which the sentence ‘p and q is composed of the sentence ‘p \ the word ‘and’ and the sentence ‘q.

As in the case of predicates we again require a more general concept which, though embracing the possibility of composition, also leaves open other possibilities. W hen dealing with predicates I introduced the con­cept of function and then that of mode of employment: if the predicate does not have the function of combining, if that is not its mode of employment, then what function does it have, how then is it used? Sim­ilarly we could now ask: what o ther function or mode of employment does the word ‘and’ have? However, we will make more progress from a theoretical point of view if we ask this question not of the sentence- part but of the sentence-whole or the corresponding state of affairs. In that case the question would have to be: if the state of affairs that p and q does not depend on the states of affairs that p and that q in the sense that it is composed of them , in what sense does it depend on them? T he move from the concept of composition to the m ore general one of depen­dence, like that to the more general concepts of function and mode of employment, is still entirely uncontroversial and the object-orientated philosopher would have no reason to resist it. It is uncontroversial that what someone asserts when he employs the sentence ‘p and q must somehow depend on what he asserts when he uses the sentence ‘p ’ and on what he asserts when he uses the sentence ‘<7 ’. Now if this dependence does not consist in the one state of affairs being composed of the others then the most natural thing would be to assume that the state of affairs that p and q has some property which depends on some properties of the states of.affairs that p and that q. Now let us recall that someone who uses an assertoric sentence Y not only somehow refers to a state of affairs but also asserts that it has the property of truth. Someone who uses the sentence ‘p and q’ says the same as someone who uses the sen­tence ‘It is true that p and q.’ It is therefore at least very plausible to suppose that the property of the state of affairs that p and q which depends on certain properties of the states of affairs that p and that q is its truth. T he question: on what does the tru th of an assertion depend?

‘And’ and 4o f 235

is clearly just ano ther form ulation of the question: what are its truth- conditions?

T h a t the properties of the states of affairs that p and that q on which the tru th of that p and q depends are likewise their tru th or falsity is not self-evident; but it would at any rate be the simplest hypothesis.

Before exam ining this hypothesis I would like, on the assumption that it is correct, to take a step fu rther. T he words ‘and’ and ‘o r’ are obviously not deictic expressions. Now in the last lecture we saw that if a sentence contains no deictic expressions we can ju st as well speak of the tru th and truth-conditions of the sentence as of the tru th and truth- conditions of the assertion chat can be m ade with it; and that in this case to understand the m eaning of the sentence is to know its truth-condi- tions. Let us for the present assume that the constituent sentences ‘p ’ and ‘q’ also contain no deictic expressions. In that case the truth-value of the assertions that can be made by means of ‘p ’ and ‘q\ hence the truth-value of that p and of that q is also the truth-value of ‘p* and ‘q ; and since therefore the truth-value of the state of affairs that p and q is also the truth-value of the sentence ‘p and q’, there follows from the hypothetically assumed dependence o f the truth-value of the state of affairs or assertion that p and q on the truth-values of that p and that q a corresponding dependence of the truth-value of the sentence ‘p and q on the truth-values of the sentences ‘p* and ‘q\ And we can now say: we understand the m eaning of the sentence ‘p and q if we understand the m eaning of ‘p ’ and understand the m eaning of ‘q’ and if we know that the truth-value o f ‘p and q depends in a specific way on the truth-values of ‘p ’ and ‘q\ O r if we understand ‘p ’ and ‘q as m ere variables, thus ifwe in terp re t ‘p and q as ‘ . . . a n d ---- ’ (where \ . .’ and ‘----- ’ are emptyplaces for arbitrary assertoric sentences) then we can now simply say: to understand the m eaning of ‘p and q - and hence the m eaning of the word ‘and’ - is to know how the truth-value o f ‘p and q depends on the truth-values of ‘p* and ‘q\ I said before that this all holds provided ‘p*and ‘q’ contain no deictic expressions. B ut the m eaning o f ‘ . . . a n d ---- ’is clearly the same w hether the sentences we substitute for ‘ ’ and‘ ---- ’ contain deictic expressions o r not. We can therefore say: tounderstand the m eaning of ‘ . . . a n d ---- ’ is to know in a completelygeneral way how the truth-value of an assertion that p and q depends on the truth-values of the assertions that p and that q: and in the specific case in which *p’ and ‘q’ do not contain any deictic expressions this dependence on the truth-values of assertions is equally a dependence on the truth-values of sentences.

In logic one abstracts from sentences with deictic expressions. For

Analysis of the predicative sentence 236

this reason the words ‘an d ’ and ‘o r’ are defined in m odern logic in terms of a dependence of the truth-value of the complex sentence on the truth-values of the constituent sentences. Meta-linguistic semantics has taken over this definition as a general definition of the words ‘and’ and ‘o r’. This substitution of the special case for the general case is harmless, precisely because the m eaning of ‘an d ’ and ‘o r’ is itself situation-inde­pendent, and the form of tru th-dependence remains the same whether one regards the assertion or the sentence as the truth-bearer.

By reference to the concrete definitions of the words ‘and ’ and ‘o r’ that are given in logic or meta-linguistic semantics we can now examine the hypothesis that the dependence of the state of affairs thatp and q on the states of affairs thatp and that q is one of truth-value on truth-values. T h e definition is: the sentence ‘p and q is true if and only if ‘p ’ is true and ‘q is true and it is false in the o ther three cases, thus if ‘p* or ‘q or both are false. In regard to ‘o f one distinguishes whether the word is understood in the exclusive sense of the Latin aut or in the sense of ‘and /o r’ (Latin vel). In the form er case, ‘p or q is true if only ‘p ’ is true or only ‘q is true; it is false if both are false or both are true. In the latter case, ‘p o r q’ is true if one of them is true or both are true and false only if both are false. So the thesis now is that the m eaning o f ‘an d ’ and ‘o r’ is explained by giving, in this way, the truth-conditions of com­plex sentences form ed by means o f ‘and’ and ‘o r’.

Doesn’t this explanation in fact correspond to our understanding of these words? If som eone says ‘It is raining and it is warm ’ the person who understands this assertion does not represent to him self some ‘aggregate’ o f the two states of affairs; rather he knows that the assertion is true if it is both true that it is raining and tha t it is warm. If you now ask how I can prove that it is so, hence that the object-orientated con­ception is false and the tru th-orientated conception correct, then I must re fe r back to the foundation of the entire discussion, to W ittgenstein’s dictum: ‘T h e m eaning of a word is what is explained by the explanation o f its m eaning.’ We have seen that we cannot explain that supposed aggregate; on the o ther hand it is clear that we can explain the meaning of these words in the way just described.

You will object to this (1) that it has not yet been shown that this explanation explains the actual m ode of employment of these words, and (2) that the explanation I have given is circular. Both objections are justified; and they belong together.

I shall start with the second objection. The circularity-objection points out that in specifying u n d er what conditions ‘p and q is true one again makes use of the word ‘and’, and in specifying under what conditions

‘And’ and ‘o f 237

‘p and q is false the word ‘o r’ is used: ‘p and q is true if and only if (p ’ is true and ‘q is true; *p and q is false if and only if ‘p ’ is false or (and/or)‘q is false. And in specifying the truth-conditions of ‘o r’ the situation is analogous. This circularity is not so evident when the truth-conditions of complex statements are given in so-called ‘truth-tables’, as is the prac­tice since W ittgenstein’s Tractatus.1 But this is clearly only an illusion. Though the words ‘and’ and ‘o r’ do not appear the tables are so con­structed that they must be supplem ented by these words when we read them.

However, the fact that the word to be explained occurs in the expla­nation does not render the explanation worthless. T he explanation shows that and how the truth of a complex sentence, or what is asserted with it, depends on the truth and falsity o f its constituent sentences, or of what is asserted with them. And the thesis that to understand this dependence of the truth-value on the truth-values of the constituent sentences is to understand the meaning of the word ‘and’ does not lose its content because of the reappearance of the word ‘and’ in the speci­fication of the truth-conditions. For if you now ask: ‘What does this “and” m ean?’ the answer is: obviously the same. It also holds for the sentence ‘T hat p is true and that q is true’ that it is true if and only if it is true that it is true that p and true that it is true that q. Certainly there is a regress here. But this regress does nothing to alter the content of the original statem ent that we understand the word ‘and’ if and only if we know in what way the truth-value of the complex sentence form ed with it (or of what is asserted by means of this sentence) depends on the truth-values of its constituent sentences (or of what is asserted by means of them).

You could point out that the explanation that has been given can also be achieved by m ere substitution, by recourse to the equivalence ‘T hat p is true = p’ (W), and thus proves itself to be tautological. T he expla­nation says: that p and q is true = thatp is true and that q is true. Now on the basis of W we can substitute ‘p and q for the left side of this equiv­alence and, within this expression, again on the basis of W, we can sub­stitute ‘T h a tp is true’ for ‘p ’ and ‘T hat q is true’ for ‘q’, with the result that the left side of the equivalence is now exactly the same as the right side. But that the explanation reveals itself to be tautological is precisely what we would expect. T he verbal explanation of the m eaning of a word must be analytic, for otherwise it would be false. And when, as in the case of the word ‘and’, it is a basic word, i.e. a word that is not definable by means of other words, a verbal explanation of this word can only be effected by means of this word; and this means: tautologi­

Analysis of the predicative sentence 238

cally. (There is it is true a possibility of defining the word ‘and’ by means o f ‘or’ and ‘not’; but then we are still left with one of the two expressions.) T hat the above definition of ‘p and q is tautological does not mean that nothing is gained by it. One must distinguish between illuminating and empty tautologies. T he tautology ‘ “p and q” is true if and only if “p and q” is tru e’ is empty. The tautology ‘ **p and q ’ is true if and only if “p” istrue and “q” is tru e’ explains the meaning of ‘. . . a n d ---- because itgives the truth-condition of sentences of this form and thus says what it is to understand this word.

I would thus like to emphasize that although a semantics that confines itself to giving truth-conditions and does not go beyond this and ask about the employment-rules of sentences does not achieve much, it does achieve something. A theory according to which it is the task of semantics (or at any rate that of the semantics of assertoric sentences) to give for all sentence-forms ‘Tarskian truth-definitions’ of the kind whose simplest case we have now become familiar with - truth-definitions which can be form ulated not in another language but in the same language whose semantics is being explained - has recently been developed by Donald Davidson . 2 Davidson starts from the now generally acknowledged requirem ent that a satisfactory semantic theory must be capable of showing how the meaning of sentences depends on the meaning of the sentence-parts. It is this requirem ent on which, as we saw, the object- orientated conception foundered. T hat the dependence of the meaning of a sentence on that of its parts is to be construed as a dependence of its truth-value, and can thus be presented for the relevant sentence- form in the shape of a truth-definition, is a plausible hypothesis, once it is recognized that to understand a sentence is (if for the sake of sim­plicity we think only of sentences without deictic expressions) to know its truth-conditions. This at any rate is Davidson’s thesis. He regards it as the special virtue of this truth-definition m ethod that it enables one to analyse the semantic structure of sentences without using words not themselves contained in these sentences . 3 When Davidson says that it cannot be the task of a semantic theory to explain our linguistic expres­sions with means which are not themselves contained in the latter this corresponds to W ittgenstein’s statem ent that the meaning is what we explain when we explain the meaning of the sentence. His conception is directed against attempts to explain the meaning of words and the way in which they are combined to form more complex expressions - thus the forms - by other verbal means, attempts which lead one to say, e.g., that the word ‘and’ stands for an aggregate, or an act of collection, or a conjunction, or that predicates stand for attributes. What is special

‘And’ and ‘or’ 239

about the truth-definition is that although it explains such a form-word as ‘and’ in the only possible verbal way, namely, by means of this word itself, the definition nonetheless has a cognitive value in that it clarifies the semantic form designated by this word.

Nevertheless, the charge of circularity is clearly justified; for only someone who already understands the word ‘and’ can understand the analysis of its meaning given in terms of the truth-definition of the word ‘and’. A nd when we ask how we can explain the m eaning of a word, clearly we want to know how we can explain it to someone who does not yet know this word (or a corresponding one in another language). Only it is now clear that such an explanation can no longer be accomplished by words. It can only be an explanation in which we dem onstrate the word’s m ode of em ploym ent by means of examples, ju s t as the m ode of employment of predicates, we saw earlier, is only explained by examples (though it subsequently tu rned out that what was explained in this way was not predicates at all, but quasi-predicates). In the meantime we have seen that the explanation of the em ploym ent of an assertoric sentence can only consist in the explanation of its being used to make an assertion, and that means of its being used to make the opening move in a verifi­cation-game; hence that the em ploym ent-rule that is explained is to be construed as a verification-rule.

Notice that we are now applying to the example of ‘an d ’ - and ‘o r’ - sentences the various theories which in previous lectures I discussed in abstracto. T he truth-definition of ‘and’ and ‘o r’ has enabled us to see what a semantics which, in explaining the m eaning of (non-deictic) sen­tences, confines itself to giving their truth-conditions, can achieve; but also what it does not achieve. W hat it is no t able to do is to explain an expression without using the expression itself (or a corresponding one in a meta-language). And what it equally cannot do is to give an expla­nation of the expression’s mode of employment. T h e two objections raised against explaining the word ‘and’ by means of truth-conditions thus belong together; for if an expression cannot be explained by means of another expression it can only be explained by dem onstrating its mode of employment.

But I do not want merely to apply those abstract theses about the m eaning of assertoric sentences to the case of ‘and ’ and ‘o r’; on the contrary, this case should serve to confirm them. T herefo re I will not proceed as though it had already been settled that the employment- rules of these sentences can only be verification-rules. Could they not also be conditional-rules which relate their em ploym ent to circum ­stances? In the case of ‘and’ at least, such a suggestion seems extremely

plausible. Could one not say that one explains the use o f the word ‘and’ to som eone by teaching him always to use the sentence ‘p and q if p and if q and always to deny it if not-p or not-q? T he objection of circularity does not apply here, because although I can only describe the expla­nation of the word’s em ploym ent by again using the words ‘and’ and ‘o r’, in the explanation itself they would not be used. Think, e.g., of a person A who wants to explain the word ‘and’ to a person B who does not yet know it. Let us suppose that it is warm and it is raining. It is I who am now giving this description of the situation, not A. A only says ‘It is warm and it is raining’ in the appropriate situation. In so doing he can assume that B already understands the constituent-sentences ‘It is warm’ and ‘It is rain ing.’ In another situation in which it is warm, but not raining, A denies the sentence ‘It is warm and it is raining.’ In this way he goes through the four possibilities by means of this and other examples. Eventually B will be able to imitate him in regard to o ther examples; and this means: he has understood the employment-rule that was explained to him.

But does the em ploym ent-rule which was explained to him in this way correspond to the mode of employment of our word ‘an d ’? I described what is explained by saying that/4 teachesB always to use the sentence ‘p and q if p and if q. But if the employment is to relate to the circumstances then this ‘i fp and if q can only mean: ‘if B perceives that p and he perceives that q. this was also presupposed in the example I gave. But in that case it is clear that the mode of employment of the word ‘and ’ explained there does not correspond to the actual mode of em ploym ent of this word. We neither only use the sentence ‘p and q if we perceive th a tp and that q nor always use it if we perceive th a tp and that q.

It would seem plausible to widen the suggested explanation in the following way. A teaches B always to use the sentence ‘p and q not only when he perceives, but whenever he believes that p and believes that <7. This explanation already seems m ore closely connected to the expla­nation by means of truth-conditions; for ‘believe’ (meinen) means the same as ‘hold to be tru e ’. But, firstly, this would not correspond to the actual use of the word ‘and ’ either, for we do not always say ‘p and q when we believe th a tp and believe that <7; and besides it is not clear how the explanation by means of examples that corresponds to this concep­tion should look.

We at least get closer to the actual use if instead of saying ‘always if we believe’ we say ‘only if we believe’. This step corresponds to the decisive step we took in the general discussion of the employment-rules

Analysis of the predicative sentence 240

‘And’ and ‘o f 241

of assertoric sentences, when it became clear that the employment-rules are not conditional rules. The rules are not such as determ ine under what conditions - situational or cognitive - the employment of the expression occurs; ra ther they are such as determ ine what the use of an expression implies i f the expression (for whatever reasons or from whatever causes) is used.

But now it is also plainly false that someone who asserts that/? and q implies that he believes that q and believes that/?. He implies only that he is prepared both to assert that /? and to assert that q. However, it is not immediately clear what the corresponding explanation would be for the assertion that/? or q. Someone who asserts this at any rate does not imply that he is prepared to assert th a tp or to assert that q. He is precisely not prepared to assert one or the other by itself, but only that one or the other is true.

So far in the discussion I have neglected the ‘o r’. With this word it could never have occurred to us that its employment-rule could relate to circumstances. In the case of ‘and’ one could at least give this sort of explanation of the mode of employment of a word that by analogy with quasi-predicates we could call a quasi-‘and’. But how should a corre­sponding mode of employment of the word ‘or’ have looked? Something like this perhaps: one can always say ‘p or q if one perceives that/? or perceives that q? But there is always less said with the sentence ‘p or q’ than with one of the two sentences ‘/?’ or ‘q’; so on this account there would never be any occasion to use the word ‘o r’. T he employment of ‘o r’ is clearly prospective, not retrospective. But prospective towards what? Is there any alternative but to admit that it is verification?

In that case we would have the following explanation: whoever says ‘p and q’ asserts that both assertions - that p and that q - are true, and that means: verifiable; whoever says ‘/? or q’ asserts that one of the two assertions is true, and that means: verifiable. W hat this means can be explained by means of examples by showing, not in which circumstances the sentences are used, bu t in which circumstances the assertions made with them can be upheld against the assertions negating them , or must be withdrawn.

T he use of the sentence-form and that of its denial are explained at the same time. T he explanation takes the form of showing what attitude each of the two speakers - the one who affirms and the one who denies- takes to his assertion and that of the other when, on the basis of the verification of the two constituent sentences, they have reached agree­m ent on their tru th and falsity. Let us take as an example the sentence ‘A hippopotam us is sitting at the fron.t-door or a lion is lying in the

Analysis of the predicative sentence 242

yard.’ Mr and MrsX want, by means of this example, to explain the use of the word ‘or’ to their little boy who, let us assume, understands the two constituent sentences but does not yet understand ‘o r’. Mr X affirms the sentence, Mrs X denies it. Together they go outside with the boy and look around. Now as far as the verification and falsification of the constituent sentences is concerned there are four possibilities - clearly the same as those already distinguished in giving the truth-conditions. It will not be difficult for father, m other and son to reach agreem ent about the truth or falsity of the two constituent sentences. In doing this neither the word ‘o r’ nor the word ‘and’ is used. But the boy will perceive that, depending on the result of the verification of the constituent sen­tences, either the father or the m other will withdraw the assertion they previously made in favour of the other’s assertion. Once he has u n d er­stood on what the withdrawal of the assertion depends, thus once he has understood on which verification-rules the outcome of the gamedepends, he has understood the employment-rule of ‘---- or . . .’ (or ‘- -- and . . .’). However, this understanding is not expressed in some meta­language - for what should this be? - but simply by the boy himself now using ‘and ’- and ‘o r’-sentences in the same way.

Thus the conception that was developed abstractly in the lecture before last has been confirmed in a simple concrete case. We were able to give the truth-conditions for a specific form o f sentence, and these truth- conditions were then explained in terms of the dependence of the with­drawal (or upholding) of the affirmative (or negative) assertion on the tru th (or falsity) of the constituent assertions. To explain the truth-con- ditions in this way is to explain how the truth-value of sentences or assertions of this form is established, and this means: how they are ver­ified. And this explanation also explains the mode of employment of sentences of this form without making renewed use o f the words that determ ine this form.

If we compare the conception of the words and ’ and ‘or’ that has em erged with the object-orientated conception, then, as in the case of predicates (p. 158), we can say that it is a specifically ‘language-analytical’ conception in the sense that the sign no longer appears as a mere in ter­mediary by means of which something in ‘thought’ is represented which could also be achieved without language. T o understand the words ‘and’ and ‘o r’ is to be master of a particular kind of sign-employment; the idea that these words stand for something else is revealed as absurd as soon as one has become clear about how the meaning of these words is explained.

L E C T U R E 18

General sentences. Resumption of the problem of predicates

T he two form s of complex sentence I dealt with in the previous lecture belong to a particularly simple and semantically transparen t species of complex sentence, those namely whose truth-value is determ ined by the truth-values of their constituent sentences, thus sentences - or the assertions m ade with them - which are truth-functions of their constit­uen t sentences (or of the assertions m ade with them). T h ere are o ther forms of complex sentence whose truth-value (or that of the corre­sponding assertions) does not depend, or does not only depend, on the truth-value of their constituent sentences (or constituent assertions). T hus, for example, the forms ‘q because/?’ and ‘i f p then q. e.g. ‘T he bus crashed because the driver was d ru n k .’ Clearly such a sentence can only be true if both its constituent sentences are true. But this is not sufficient. It can be true that the bus crashed and that the driver was drunk , but false that the bus crashed because the driver was drunk. Thus the tru th of such an assertion does not only depend on the truth-value of its constituent assertions. In this case there is clearly in addition a certain grounding-relationship between the two constituent assertions. T he situation is similar in the case of the form s ‘If the driver had been drunk , the bus would have crashed’ and ‘If the driver is drunk , the bus will crash.’ A connection of ground and consequence is also asserted in statements of the ‘if - th e n ’ form, only now it is presupposed - in the case of the counter-factual conditional - that the constituent sentences are false or - in the o ther case - that the truth-value is left open.

T he distinction between truth-functional (so-called extensional) and non-truth-functional (so-called intensional) complex statements stems from Frege. T he discussion of the different forms of intensional complex statements in Frege’s paper ‘On Sense and R eference’ is probably still the most comprehensive. T he semantics of intensional complex state- m ent-form s is still today a m atter of controversy. I shall not go into it

Analysis of the predicative sentence 244

here. At the m om ent I am only concerned that you do not draw too far-reaching conclusions from the fact that intensional complex state­ments are not truth-functional. It does not follow from the fact that the truth-value of these forms of sentence does not depend, or does not only depend, on the truth-value of the constituent sentences that it is not true of these forms of sentence that to understand them is to know what the truth-conditions are of the assertions made with them. We have no reason to think that this extremely general characterization of the meaning of assertoric sentences should not hold for intensional complex sentences. O f course, not much is thereby achieved, for the question immediately arises: on what then does the truth-value of these complex sentences depend, if it is not the truth-value of their constituent sentences? Is it the m eaning of the constituent sentences? Is it their m eaning together with their truth-value? Is it the truth-value of other assertions which are not expressed in the constituent sentences of the complex sentence but are implied by it? So the hypothesis that it is also true of intensional complex sentences that, to understand them is to know what their truth-conditions are, does not am ount to an answer. However, it does indicate the only direction in which the enquiry can proceed. H ere too the alternative would be an object-orientated con­ception, according to which the m eaning o f ‘because’, ‘if . . . then’, etc., would have to be conceived as a certain combination of the states of affairs for which the constituent sentences stand. W hether there exists a concrete attem pt at carrying out an object-orientated explanation of intensional complex statements, I do not know; however, it should be possible to recognize in advance the senselessness of such an attempt now that we have shown by reference to the example of extensional complex sentences the absurdity of the idea of the composedness (Zusammengesetztsein) of a state of affairs.

I also wanted to examine the statements which in the traditional logic were called general statements, in contrast to singular predicative state­ments. Two kinds of general statement were distinguished, the universal (e.g. ‘All ants are poisonous’) and the particular (e.g. ‘Some ants are violet’). T he distinction is problematic, because the denial of a universal statem ent is a particular statem ent (‘it is not the case that all’ = ‘some are not’) and vice versa (‘it is not the case that some’ = ‘none are ’ = ‘all are not'), and because the so-called quantifiers (‘all’, ‘some’) can occur m ore than once in a statem ent (e.g. ‘All ants have some ancestors’). Nonetheless one can say that the positive universal statement and the positive particular statem ent are the two simplest forms of general statement.

General sentences 245

How should we conceive the semantics of such statements? What is it to understand a sentence of this form? T he traditional view was that they are a species of predicative sentence; that is: the expressions ‘some ants’ and ‘all ants’ were construed as subject-expressions. So long as one has a purely grammatical, syntactical concept o f ‘sentence-subject’ there is nothing wrong with this. In traditional logic however the notion of the sentence-subject was an undifferentiated syntactical-semantic one. Thus not only in grammar but also in logic expressions of the form ‘som e/7’, ‘a lii7’, were construed as subjects, and that means: as compo­nents of predicative judgm ents. In the traditional logic books1 it was customary first to introduce ‘the’ predicative judgm ent in general as consisting of subject, copula and predicate. T he singular, the universal and the particular judgm ent were then distinguished as three species. Occasionally the singular judgm ent was subsumed under the universal because in both cases, and in contrast to the particular judgm ent, the predicate is valid for the ‘whole extension’ of the subject-term . 2 Thus, the general judgm ent was conceived on the model of the singular ju d g ­ment; the relation between predicate or concept and that for which a singular term stands became the model for the relation between predi­cate and that for which the expression ‘all F’ or ‘some i7’ stands.

The traditional conception thus presupposes that the expressions ‘all F’, ‘some F’ stand for something, for an object. 3 But what sort of an object can this be? In the case of ‘all ants’ one might say it is the class of ants. But when we say ‘All ants are poisonous’ we do not mean that the class of ants is poisonous - a class cannot be poisonous - but that ants - all ants - are poisonous. It becomes even more difficult in the case of ‘some ants’. It used to be said that this expression stands for a sub-set of the class of ants. But does this mean a determ inate sub-set? If, for example, the statement ‘Some ants are violet’ is false, does this mean that there is a determ inate sub-set of ants which are not violet? Clearly not. T he sentence is false, if, am ong all sub-sets of ants, there are none whose elements are violet. ‘T here is a certain sub-set . . however, is simply an elaborate formulation of ‘there are some ants . . .’ T he expression ‘an (indeterminate) sub-set of F’ does not stand for a (deter­minate) sub-set of F, and hence does not stand for something at all.

What then do we mean by the expression ‘some ants’? If we free ourselves from the idea that such an expression must be understood as somehow analogous to a singular term (‘this ant’), then there is no longer any necessity to articulate the sentence semantically on the model of a singular predicative sentence: ‘Some ants/(are) violet.’ ‘Some ants are violet’ clearly means the same as ‘T here are violet ants.’ This equivalence

Analysis of the predicative sentence 246

suggests that we should articulate the sentence as follows: ‘Some//ants/violet.’ Correspondingly, we would have to articulate the universal sentence thus: ‘All//ants/poisonous.’ Does this get us any fu r­ther? If we isolate the words ‘some’ and ‘all’, then their meaning seems to become even more unintelligible. This is so as long as we ask what the expressions stand for. But because the question, as thus posed - given the suggested articulation - makes no sense, we are forced to pose the question differently.

Instead of ‘all’ we can also say ‘every’. If we write ‘every//ant/poisonous’ we are rem inded that there is also a sentence with the corresponding grammatical structure: ‘Everything that is an ant is poisonous.’ This clearly means something like: ‘Everything : if it is an ant then it is poi­sonous.’ The word ‘everything’ does not stand for something, rather it refers to other sentences, namely, all singular predicative sentences ‘This is an ant.’ This opens up the possibility of conceiving the understanding o f the word ‘every’, and hence that o f a universal sentence, as an u nder­standing o f the truth-conditions of this sentence, such that the sentence ‘Every ant is poisonous’ is true if and only if in every case in which one can say o f something that it is an ant one can also say that it is poisonous. T he word ‘every’ does not stand for something. Rather it contains a directive:-‘Take everything in tu rn ’; and to this directive is connected the assertion: ‘If it is F, then it is G. ’

We can in terpret the particular statement in exactly the same way: ‘(take in turn) everything: (then you will find) one or some of them that are F and G\ H ere too the assertion refers one to singular predicative assertions, but in this case in such a way that the assertion is true if any singular sentence ‘This is F and it is G’ is true. It may surprise you that the reference to all (‘take everything in tu rn ’) is now also taken up with the paraphrase of the particular statement; however one can easily see that a reference to all is indeed contained in our understanding o f ‘some’. ‘Some’ means ‘some of all’. To make clear the contrast, and the parallel, between universal and particular statements we could paraphrase thus: ‘of all every: . . .’, ‘of all some (a): . . .’

This interpretation of general statements stems from Frege . 4 With it Frege succeeded in breaking out of the narrow confines of traditional predicate-logic - the Aristotelian syllogistic - and developing the general predicate-calculus of m odern logic. The syntax of this calculus only seems artificial because looked at from the standpoint of the syntax of natural language it is unusual. But this is only because its structure makes explicit precisely the semantic structure of the general statements of natural lan-

General sentences 247

guage which does not become visible in their own syntactic structure.It is only in one detail that the semantic conception of m odern logic

seems not to correspond to the semantics of natural language . 5 The universal statem ent is written thus in m odern logic: \x){Fx-*Gx)' or in words: ‘for all x: if x is F then it is G\ This is ju s t another form ulation of the sentence I ju st proposed: ‘everything: if it isF then it is G\ Now it seems much m ore natural to in terp ret a statem ent like ‘All ants are poisonous’ not, as I have ju s t done, as equivalent to ‘Everything that is an ant is poisonous’, but thus: ‘every ant: it is poisonous5 and, corre­spondingly, the particular statem ent thus: ‘some one of all F: it is G\ T he difference is that the singular term s o f the predicative statements to which the general statement, in Frege’s interpretation, refers stand for all objects o f a presupposed dom ain of objects, whereas, according to the o ther conception, they stand only for the objects that are F. According to the first conception the verification o f the statem ent ‘All ants are poisonous’ would consist in the exam ination of all real objects with respect to whether, i f they are ants, they are also poisonous; according to the second conception the directive is restricted to the totality o f ants. It is a consequence o f the first conception that, if there are no ants, the universal statem ent is still true; whereas it would cor­respond to the semantics o f natural languages to say that if there are no ants the question of w hether or not all ants are poisonous cannot be m eaningfully raised and that a corresponding statem ent is neither true nor false.

T hus Frege has at this point been prepared to depart from the semantics of natural language in the interests of logical system. However, I have only m entioned this difference in order to make clear that what is decisive in the context of our enquiry is common to both conceptions, namely, that the understanding of these sentence-form s consists in knowing how their truth-value depends on the truth-value of other sen­tences. Thus in regard to general sentences we reach a result analogous to that reached in the case of truth-functional complex sentences. T he only difference is that the assertions on whose truth-value the truth- value of a general sentence depends are not expressed in the components of this sentence. And a fu rth er difference is connected with this: the assertions on whose truth-value the truth-value of a general sentence depends can be of an unlim ited num ber, and in that case we cannot finally verify the all-sentence. In those cases where we are dealing with a finite num ber of objects we can convert the universal statem ent into an and-statem ent and the particular statem ent into an or-statement.

Analysis o f the predicative sentence 248

T he sentence ‘All ants in this box are poisonous’ is, if the num ber of ants in the box = n, equivalent to the sentence ‘Ant 1 is poisonous and ant 2 is poisonous . . . and ant n is poisonous.’ I have already spoken many times of the ‘equivalence’ of statements without having defined this expression. We can now say: two sentences are to be term ed equiv­alent when they have the same meaning, and that means: the same truth- conditions. In our case one can see immediately that the truth-value of the two sentences depends in the same way on the truth-value of the same sentences ‘Ant 1 is poisonous’, etc.

Clearly the explanation by means of truth-conditions that I have given of the forms of general sentences is circular in the same way as the corresponding explanation of ‘and’ - and ‘or’ - sentences. The universal statem ent is true if every insertion of a singular term yields a true p re­dicative statement; the particular statement is true if one such insertion yields a true predicative statement. As the quantifiers refer not to objects but to the insertions or the sentences which result from such insertions there is again no m ere circle. T he truth-definition shows how the truth- value of the sentence depends on the truth-value of other sentences.

And here too we can take the fu rther step and avoid the circle by explaining the em ploym ent of such sentence-forms by means of exam ­ples. In his Remarks on the Foundation of Mathematics Wittgenstein writes: ‘One learns the m eaning of “all” by learning that from “(x)Fx” ‘F a” follows. (And this means that if any statement “This is F” is false then the universal statem ent is also false.) The exercises which train one in the use of this word, teach one its meaning, always aim at showing that an exception may not be m ade’ (§10). How can we explain this - ‘that an exception may not be m ade’ - to someone who does not yet u n d er­stand it? For example, we have beans in a sack and assert that all are black and depending on what emerges we will uphold the statement or withdraw it or say it has been shown to be true or false.

With this I conclude the discussion of the semantics of truth-functional sentence-form s and can now move on to the twice-postponed treatm ent of the form of elem entary sentences - the predicative sentence-form. I postponed the discussion of the predicative sentence-form for the first time when it had become clear that neither the meaning o f predicates nor that o f whole predicative sentences can be understood by relating them through their em ploym ent-rule to the circumstances of their use. A fter it became clear that the isolated discussion of predicates leads ju st as much into a blind-alley as the orientation towards names it seemed natural to look for the meaning of the two components of the predicative sentence-form in their contribution to the m eaning of the whole pre-

General sentences 249

dicative sentence. But this required that before resum ing the discussion of predicates and the treatm ent of singular terms we should arrive at a preliminary conception of the meaning of assertoric sentences. Such a conception was reached in the explanation that to use an assertoric sen­tence is to assert that its truth-conditions are fulfilled, or that following its verification-rule will lead to confirmation of the assertion.

With this a basis was achieved for the analysis of the semantics of the predicative sentence-form. To understand a sentence of this form must be to know its truth-conditions, or its verification-rule. At this point I again postponed the discussion of the predicative sentence-form in order to illustrate the abstract notion of truth-conditions or verification-rules by reference to the simpler case of truth-functional sentence-forms. What makes this case so simple is not only the fact that in dealing with it we can disregard deictic expressions, and hence the distinction between sentence and assertion and hence the whole problem of the identifiability of what is asserted which is basic to our speaking of ‘true’ and ‘false’. T he peculiar simplicity of the explanation of truth-functional sentence- forms is grounded in the fact that it makes the tru th of these assertions dependent simply on the tru th and falsity of other assertions. In the truth-definition this was expressed by the reappearance of the word ‘true’ in the definiens - in the specification of the truth-conditions - and, in the explanation of the mode of employment, by the fact that the withdrawal, or upholding, of the assertion depended on the with­drawal, or upholding, of other assertions. The tru th and verifiability of truth-functional sentences refers (verweist) to the tru th , or verifiability, of other sentences and hence ultimately to the truth, or verifiability, of the assertions made by means of elementary predicative sentences. The explanation of the employment of truth-functional sentences presup­poses that the word ‘true’ or the employment of o ther sentences is already understood. But then this means that my exemplification of the abstract explanation of the employment-rule of assertoric sentences and of the notion of truth-conditions and verification-rules by means of the truth-functional sentence-forms does not stand on its own feet. In the end this abstract explanation can only be made concrete by being applied to the employment of elementary predicative sentences. So one can now understand my earlier claim (p. 179) that what can be said about the meaning of assertoric sentences in general prior to the analysis of the semantics of the predicative form of sentence can only be regarded as provisional.

This can be made particularly clear by reference to the word ‘true’. I have already made ample use of this word. However, it has not yet been

Analysis of the predicative sentence 250

explained. If someone wishes to object that it is explained by the redun- dancy-formula I would rem ind you that the redundancy-form ula itself presupposes that one already understands the difference between ‘(it is asserted) that p ’ and ‘p and this difference is only understood if the verification-game whose rules are verification-rules is understood. Thus one can only regard the word ‘true’ as having been explained by the redundancy-form ula if one presupposes that one already understands how assertions can be verified, and that means: how their tru th can be recognized; the word ‘true’ does not have to be used here but it is natural to use it (together with the expression ‘that p'). Thus one could only dispense with an explanation of the word ‘true’ that goes beyond the redundancy formula if one were to omit the first step in the explanation, viz. to understand an assertion is to know its truth-conditions, and pro­ceed immediately to the next, viz. to understand an assertion is to know how it is verified.

If we again begin at the level of the first explanation-step - as in the case of truth-functional statements - we see immediately how far we still are from an understanding of the word ‘true’. For whereas in the explanation of what it is for a truth-functional assertion to be true the word ‘true’ appeared again it cannot figure in the explanation of what it is for a predicative assertion to be true; so it is here that we must expect the explanation of the word ‘tru e’ that has so far been lacking. However, this cannot be taken to mean that a general m eaning of the word ‘true’ would be arrived at which could then be transferred to the tru th of truth-functional assertions. W hat it means for truth-functional assertions to be true is already fixed by their truth-definitions. These truth-definitions can only be completed by explaining the word ‘true’ that occurs in their definiens by means of a truth-definition of predicative assertions. What results in this way is a so-called recursive truth-definition o f the kind first developed by Tarski. T he predicate ‘true’ is defined in such a way that it is first defined for the class o f elem entary assertions and is then defined for the others - at any rate those we have so far become acquainted with - as a function o f its application to the elemen­tary assertions.

How should we approach the question of the truth-condition of a predicative assertion? Are we not once again at one of those dangerous places on our path at which no definite direction is prescribed for our enquiry, and at which I can only make some suggestion or o ther which would be bound to appear more or less arbitrary? I think not. If the entire previous procedure has not been a blind-alley then the next step must follow automatically once we bring together the various lines of

General sentences 251

thought - in part carried through, in part broken-off - that are relevant to the present enquiry.

From the outset the question concerning the semantics of predicative sentences could only be understood as the question concerning the p re­dicative form , and that means: as the question of how the m eaning of the whole sentence results from the m eaning of its two components. O f course the object-orientated position also understood the question in this way. However, its attem pt to construe the m anner in which the m eaning of the sentence-whole arises out o f the sentence-parts as com­position foundered. As this conception was conditioned by the objectual in terpretation of predicates I tried to work out a new conception of predicates. But because predicates were treated in isolation the attem pt failed to overcome the ambiguity between predicates and quasi-predi- cates. It thereby became clear that the m eaning of the predicate can only be understood by reference to the m eaning of the sentence-whole. And since in the m eantim e it has em erged that to understand an asser­toric sentence (if we disregard deictic expressions) is to know its truth- conditions we are now in a position to understand the alternative to the object-orientated conception of the predicative sentence in terms of the sentence-whole. If we attem pt a formulation of this alternative analogous to the one for truth-functional sentences (and for the sake of simplicity let us continue for the time being to disregard deictic expressions) we can again say: the e rro r of the object-orientated conception is not that it speaks of objects that can be true or false - states of affairs, or assertions- but that it thinks of a predicative state of affairs, that a is F, as composed of a and that for which the expression ‘F ’ stands; and fu rth er we can now again say: the dependence of the assertion is to be construed not as composition but as a dependence of truth-value.

But when we attem pt to say what it is on which the truth-value is dependen t we get into difficulties. If we were to say that the truth-value of the predicative assertion is dependen t on the object for which the singular term ‘a stands and on the attribute or the class for which the predicate ‘F ’ stands, this, unlike the untenable synthesis-theory, would be correct but (a) following our discussion of the nom inalism-problem we know that such an explanation can at most have a secondary justifi­cation and that the prim ary explanation by recourse to the attribute or class would represen t a hysteron-proteron and (b) it is clearly not suffi­cient to say that the truth-value of the predicative assertion depends on this and that; ra th e r we require a complete form ulation of the form: ‘the assertion that a is F is true if and only if . . .’ T he truth-value of truth-functional assertions did not depend simply on assertions but on

Analysis of the predicative sentence 252

a property of these assertions. This property was that of truth. T here is no question of this in the present case, but we must still expect the tru th of the predicative assertion to depend on certain properties of the objects concerned or on a relation between them; and it will probably have to be a relation if the if-sentence which has just been left open - thus the truth-condition - is to yield a coherent formulation. If, however, the above-mentioned hysteron-proteron is to be avoided then at any rate in the case of the predicate the object concerned can only be the sign itself.

So it would now seem plausible to ask: is there a relation between the predicate ‘F* and the object for which ‘a stands which is such that one can say: the assertion that a is F is true if and only if this relation obtains? We can here refer back to the result of our earlier investigation into the function and mode of em ploym ent of predicates (p. 135 f), and since we are now re turn ing to it from the perspective of the question of the truth-condition of the assertion the result achieved at that time loses the ambiguity which led us into the blind-alley of quasi-predicates. T he purpose for which a predicate is used was shown to be characterization (p. 135). In introducing this characterization-function it seemed wholly natural to say: that which is characterized by means of the predicate - or, putting it more abstractly, to which the predicate is applied - in the employment of a predicative sentence is the object for which the singular term which supplem ents the predicate to form a sentence stands. It was then the question of the explanation of the use of the predicate which as it were led to the relation of the predicate to the employment-situation taking the place of the relation to the object and so gave rise to the ambiguity with quasi-predicates. Initially I shall leave open the question of how the problem of the explanation of employment is to be conceived from the present perspective, for I want to keep to the same sequence which em erged in the abstract description of the meaning of assertoric sentences and which I also followed in the explanation of the meaning o f truth-functional sentences. Only after this shall I move on to the crucial question of the explanation of employment-rules. So ignoring for the time being this wider question we can tackle the question of how the characterization-relation is to be understood relative to the object if it is to be understood as a truth-condition. It is not enough merely to speak in an unqualified way of a characterization of the object by the predicate or an application of the predicate to the object. An assertion that a is F is true not when the predicate *F' is merely applied or is applicable bu t when the predicate can legitimately be applied to it. This relation of the legitimate applicability of a predicate to an object is what

General sentences 253

is denoted by the word ‘applies’ (‘Zutreffen ). Thus we arrive at the fol­lowing truth-definition for predicative assertions: the assertion that a is F is true if and only if the predicate F ’ applies (zutnfft) to the object for which the singular term ‘a’ stands.

Perhaps you will find this result ridiculous, for it appears so trivial. However, it is the sort of triviality we must expect from any truth-defi­nition of a sentence-form. What we now have to do is to see exactly what is achieved with this definition, but also what it is about it that leaves us unsatisfied.

This is a significant result in the following two respects:(1) The truth-definition of the predicative sentence enables one to

give a first, no longer object-orientated answer to the question of what it is to understand a predicative sentence. Continuing for the time being to ignore deictic expressions, we can say: to understand a predicative sentence is not, as was thought in the object-orientated tradition, to represent a synthesis between two objects which correspond to the two sentence-components and for which these stand; rather it is to know that the sentence (or the assertion m ade with it) is true if and only if the predicate applies to the object for which the subject-term of the sentence stands. This is a far from negligible result at that level of semantic theory which corresponds to Davidson’s conception.

(2) It is a significant result to have succeeded in giving a definition of the word ‘tru e’ for the form of elementary sentences in whose definiens the word ‘tru e ’ no longer occurs. You will point out to me that I have only been able to do this by bringing in the unexplained word ‘applies’ (‘zutrijft’). O f course I admit this; even worse I do not believe it is possible to give a verbal definition o f the word ‘applies’ other than by means o f the word ‘tru e ’, namely by reading the truth-definition just given in the opposite direction, from right to left. What is meant by the word ‘applies’ can only be defined by saying: a predicate F ’ applies to an object a if and only if the assertion that a is F is true. At the level of predicative assertions then we move in a circle between the two words ‘true’ and ‘applies’. Nonetheless this represents a first step towards an analysis of the condition of the possibility of the truth of elementary statements. The truth-definition suggests that what is asserted by an elementary sentence can only be true or false because the elementary sentence is not only structured {gegliedert) but so structured that the two components have different and mutually supplem enting functions of a kind which make possible such a thing as a truth-condition: an assertion is true if the predicate applies to an objectfor which the subject-expression5 to<i$.

Historically it was Plato who first showed that the possibility of sen­

Analysis of the predicative sentence 254

tences being true or false is grounded in their predicative structure . 6

For Plato, however, the problem was simply that of explaining the pos­sibility of falsity (how it is possible to assert and believe something that is not) and he could solve this problem by exhibiting the complex struc­ture of the (predicative) statement. In regard to this question the fact that the two components have different functions could be ignored. And in the tradition this fact was not seen because it was taken for granted that both sentence-components have the function of standing for objects. In this levelled-down form - the condition of the possibility of being true or false is the synthetic form of the statement - Plato’s insight became, via Aristotle , 7 a perm anent item in the philosophical tradition. (Statements other than predicative were not considered.) The foil against which Plato and Aristotle posed the problem of falsity was the idea of a representation (Vorstellen) of objects that could only be true . 8 Both the truth-relation and the object-relation were thus taken for granted and treated as unproblematic. The more radical question of the conditions of the possibility of a relationship to truth - or in a language of the use of the word ‘true’ - could not be posed at all. Of course until now this question has not been recognized in analytical philosophy either, for the simple reason that, through lack of reflection on the relation of their own position to the philosophical tradition, analytical philosophers confined themselves to the traditional problems.

Like Plato’s question this question too is posed against the foil of a more primitive language which lacks the specifically predicative struc­ture. But whereas for Plato this was the fiction of a language consisting only of names, in the present enquiry it is a language consisting of more primitive characterization-expressions (quasi-predicates) of the kind we actually encounter in signal-languages and in the first language-acqui- sition of children. For the characterization-expression to be able to apply to something such expressions must first be supplem ented by singular terms which have the function of standingfor objects to form predicatively structured expressions.

Thus what makes possible a relation to truth in a language is the addition of expressions which make possible a relation to objects. This again shows why the question of the condition of the possibility of a relation to truth lies outside the horizon of traditional philosophy: the traditional phi­losopher took the relation to objects for granted and, given his orien­tation, was unable to see it as a problem. It has also become clear that the one-place semantic predicate ‘true’ in the truth-definition of predic­ative sentences involves a reference (verweist), not only to the two-place semantic predicate ‘applies’, but also to the two-place semantic predicate

General sentences 255

‘stands for’. It is true that one can omit the latter expression from the truth-definition by simply saying . . if the predicate “F ” applies to a*. But that a predicate applies to an object means that the assertion made by means of a sentence form ed by the supplem entation of the predicate by an expression that stands for the object is true.

We can now also say what remains unsatisfactory about the explanation of the meaning of predicative sentences by means c f the truth-definition. Let me rem ind you again that I have so far disregarded deictic expres­sions. We were thus able to speak o f the truth-condition of the sentence and to say that to understand a predicative sentence is to know the truth-condition which is given in the truth-definition. If we now recall W ittgenstein’s principle: ‘T he m eaning is what the explanation of the meaning explains’, then it is clear that if we wanted to explain to someone the meaning of a predicative sentence by means of this truth-definition we would have to assume that he already understands the expressions ‘applies’ and ‘stand fo r’. But since a verbal definition o f the word ‘applies’ is only possible by means of the truth-definition read in the reverse direction, this means that the explanation either rem ains circular or is dependent on a non-verbal explanation of what is meant by a predicate’s ‘applying’. And since, as we have just seen, one cannot understand what is meant by a predicate’s ‘applying’ without at the same time u n d er­standing what is m eant by saying that an expression - which for this reason is called a singular term - ‘stands fo r’ an object, the truth-defi- nition is, if it is to be an explanation of m eaning, equally dependent on a non-verbal explanation of what is m eant by an expression’s standing for an object.

It would be a mistake to suppose that I would now suddenly transfer the dem and for the explanation of m eaning from the prim ary level of expressions of natural language to a meta-level of expressions of semantic theory. In the present line of thought such a separation which is made in the usual meta-linguistic semantic theories is not possible. Rem em ber that I started out from the question of how one explains a predicate. In this (philosophical) question one is only asking about what in general happens when the m eaning of a particular predicate is (pre- philosophically) explained (p. 150). From the question o f how a predicate is explained, however, we were referred to the question of how a p re­dicative sentence is explained. A nd if the answer to this question is: ‘by giving its truth-condition’ and if doing that involves using the words ‘applies’ and ‘stands fo r’, then, if the explanation is not to remain in the air, we have no alternative but to explain these words themselves.

H ere, just as in the case of truth-functional sentences, we find that if

Analysis of the predicative sentence 256

the explanation is not to rem ain circular we must move from the level of the specification of truth-conditions to that of the explanation of mode of employment. Now we have already seen in the case of the semantic predicate ‘true’, in that part of its recursive definition that has so far been given, that one cannot explain such a word in abstracto; what is explained is how one establishes w hether an assertion is true; and this can only be done by explaining the mode of employment of the corre­sponding form of sentence. Likewise we must expect that one can only explain the word ‘applies’ by explaining how one establishes that a predicate applies; and this can only be done by explaining the employ- m ent-rule of predicates. A nd in precisely the same way we must expect that the expression ‘stands fo r’ can only be explained by explaining how one determ ines for which object an expression stands, and this can only be done by explaining the em ploym ent-rule of singular-terms. But now if it is correct that singular terms and predicates are complementary expressions such that a predicate’s applying to the object for which a singular term stands constitutes the truth-condition of an elementary statement, then we must also expect ( 1 ) that the explanation of the employm ent-rule of predicates presupposes an understanding of the employm ent-rule of singular terms and vice versa, and (2 ) that by explaining the em ploym ent-rule of predicates, together with that of singular terms, one explains the employment of predicative sentences. Or to put it another way: by explaining what it is for predicates to apply and what it is for singular term s to stand fo r objects one explains what it is for an elem entary statem ent to be true.

L E C T U R E 1 9

The mode of employment of predicates. Transition to singular terms

‘The assertion that a is F is true if and only if the predicate “F” applies to the object for which the singular term “a” stands.’ With this truth- definition a start is made in the analysis of the meaning of predicative sentences. But it is only a start. The transition from this first level of semantic analysis to the crucial second level, which consists in the expla­nation of use, will prove much more difficult in the case of predicative sentences than in the case of truth-functional sentences. You will there­fore not be prepared to undertake the lengthy analyses that are now required unless I am able to convince you that this first level of expla­nation of meaning does not suffice for the fundam ental analysis of the meaning o f predicative sentences we are here aiming at.

In the last lecture I explained the inadequacy of this explanation by saying that it presupposes that we already understand what it is for a predicate to apply to an object and what it is for an expression to stand for an object. But we can also see the inadequacy of this explanation from another angle. T he truth-definition with variables F ' and ‘a ’ is of course m eant to provide a framework for the specification of the truth- condition of any actual assertion by the substitution of a particular predicate and a particular singular term. But this means that it would also have to provide the framework both for the explanation of the meaning of predicates and for the explanation of the meaning of sin­gular terms; and it would have to do this both in general and for the explanation of particular predicates and particular singular terms. But that the truth-definition, as it now stands, allows for an explanation of the sentence-components could at best be claimed for the singular term. O f the latter one could say that one understands it if one knows for which object it stands. But what would it be, in terms of this tru th- definition, to understand the predicate? T he best I can think of is: to understand a predicate ‘F ’ is to know what it is for it to apply to an

Analysis of the predicative sentence 258

object. But this of course is to say nothing so long as one does not explain what it is. T he inadequacy of such an explanation becomes immediately clear when one considers how one would explain the meaning of particular predicates within the framework o f the general explanation.

Now there is a possibility of modifying the truth-definition in a way that removes this difficulty. In the usual meta-linguistic semantic theo­ries the truth-definition for a predicative assertion goes roughly as fol­lows: ‘the assertion that a is F is true if and only if the singular term “a” stands for an object which is an elem ent of the class for which the p red ­icate “F ” stands’. This definition can then be supplem ented by a speci­fication in a meta-language of the object to which the expression ‘a* is assigned and a specification of the class to which the predicate F” is assigned. This truth-definition enables one, if one has a list on which all singular terms and predicates of a language are assigned to specific objects and classes, to determ ine the truth-conditions of all the sen­tences which can be form ed by all combinations of the singular terms and predicates on the list. But this also makes it clear what the limited theoretical interests are within which such a truth-definition has a value. It is im portant that we realize the extent to which this truth-definition is even less satisfactory than the previous one.

Firstly, the explanation of predicates which was lacking in the p re­vious truth-definition is achieved in the present one at the cost of a relapse into the object-orientated position; in place of a predicate’s applying to something one has the inclusion of an object in a class. A variant of this would have been to say that the assertion is true if the attribute for which the predicate stands belongs to the object. Since, however, two predicates which stand for d ifferent attributes, or have different meanings, determ ine the same class if they apply to the same objects, the truth-value of the assertion, if it depends on the predicate’s applying to the object, depends only on which class the predicate de ter­mines. Hence if one is introducing something objectual for the pred i­cate at all then the situation is m ade clearer if one speaks of the class than if one speaks of the attribute.

Secondly, just as the previous definition presupposed that one already understands the term ‘applies’, the present definition presup­poses that one already understands what it is for an object to be an element of a class. On this score then neither of the definitions has any advantage over the o ther except that, as we have seen on an earlier occasion (p. 231), the inclusion of an object in a class can be defined in terms of the predicate’s applying to the object (an advantage which of

The mode of employment of predicates 259

course will only become effective if we can find an explanation of a predicate’s applying which is independent of the truth-definition).

Thirdly, what makes the new truth-definition totally unacceptable within our line o f thought is that it assigns the singular term to an object and the predicate to a class by means of a meta-linguistic expression. T he assignm ent is really only an assignm ent of the object-language expression to the meta-linguistic expression. H ence it is presupposed that the one to whom a singular term or predicate is explained in this way already understands the corresponding expression in the o ther lan­guage.

Fourthly, I would also point out that predicates and singular terms, as they are explained in this second truth-definition, are not explained as com plem entary expressions, and are not explained as essentially com ponents of the predicative sentence. This point is, of course, a mere consequence o f the first, the relapse into the object-orientated position.

Summarizing, we can say that the first truth-definition was only unsatisfactory in the sense that it did not go far enough; it left open the explanation both of the semantic expressions in the definiens and, what is directly connected with this, of the sentence-com ponents themselves. T he second truth-definition, on the o ther hand, is unsatisfactory in the wholly d ifferen t sense that, what the first truth-definition left open, it answers with a pseudo-explanation given in a m eta-language and in this way merely covers up the problem facing us. T he advantage of the first truth-definition is that although it does not give an answer it does put the question concerning the mode of em ploym ent of predicates and that concerning the mode of employm ent of singular terms on the right track. For from the outset it reciprocally relates predicates and singular term s to one another and both of them to the tru th of the assertions m ade with them. So we can expect that the explanation of the mode of em ploym ent of predicates, and that of singular terms, will at the same time be an explanation of the mode of em ploym ent of whole predica­tive sentences, and, hence, of the m eaning of the word ‘tru e ’ in its ele­m entary m ode of employment.

We already know in a general way that the transition from the first level of semantic explanation of the form of an assertion by means of a truth-definition to the second level, in which the mode o f em ploym ent o f such sentences is explained, m ust take the form of presenting the verification-game between an assertion of this form and its denial; and in essence this means: showing how the assertion is verified. In the case of predicative assertions this can only mean that the verification-rule of the sentence is founded in the em ploym ent-rules of the two sentence-

Analysis of the predicative sentence 260

components. For the employment-rules of the two sentence-compo- nents this means that they must be so constituted that, if one knows the em ploym ent-rule of a particular singular term and that of a particular predicate, then one knows the verfication-rule of the sentence com­posed o f this singular term and this predicate. And this means that the em ploym ent-rules of the two components must consist in their contri­bution to the sentence’s verification-rule. Thus our understanding of each of the two sentence-components is not independent of its being the com ponent of a predicative sentence, though our understanding is independent of its being combined in a sentence with precisely this par­ticular expression of the complementary semantic class. Now since knowing a sentence’s verification-rule consists in knowing how its truth is established - this is simply a verbal definition - it follows from the truth-definition that the verification-rule of the assertion made by means of a predicative sentence ‘Fa is grounded (a) in the knowledge of how it is established for which object of arbitrary predications the singular term \ . . a stands and (b) in the knowledge of how it is estab­lished that the predicate ‘F — ’ applies to an arbitrary object.

With this we have now achieved at the level of the explanation of mode of em ploym ent what was lacking in the truth-definition (at any rate in its first form which for us is the only relevant one), viz. a question about the explanation of the two sentence-components that is inte­grated into the question about the explanation of the predicative sen­tence. You could object that I could have already reached this result in connection with the truth-definition. I said there that in the framework of the truth-definition one cannot understand in what the explanation or understanding of the predicate consists; at best one can say that to understand a predicate is to know what it is for it to apply to an object. W ould it not have been easy to supplem ent this as follows: to under­stand a predicate is to know how it is established that it applies to an object? However the question is not whether such a supplementation would have been easy but ra ther what its methodological significance is. Talk of establishing that a predicate applies already belongs to the con­text of establishing the truth, and that means: to the context of verifica­tion.

We shall see that in the case of the singular term too by asking how one establishes for which object the expression stands a perspective is opened up which was not yet contained in the truth-definition in its first form , and which in the second form is covered up by speaking of an assignment. However, in terms of m ethod it would seem correct to begin the analysis with the explanation of the employment-rule of

The mode of employment of predicates 261

predicates. This is not only because it presents fewer difficulties, but, above all, because in such an analysis we can appeal to the description of the explanation of the mode of employment of predicates that I have already presented but have broken off on account of the resulting ambiguity with quasi-predicates. O ur task will now be to see whether this ambiguity is avoided by taking into account the fact that predicates are classification-expressions which essentially require to be supple­mented by singular terms; and that means: classification-expressions which characteristically apply or do not apply to objects. We would also expect from an explanation of how one establishes whether a predicate applies that it should explain the word ‘applies’ which verbally is only definable by means of the word ‘true’, and that in this way a first step would be taken towards the analysis of the meaning of the word ‘true’. And, finally, since, in describing the employment-rule of predicates, we cannot avoid taking account of the fact that a predicate is an expression that requires to be supplem ented by a singular term, the explanation of the mode of employment of predicates should also give us a perspective for tackling the question of the mode of employment of singular terms.

T he procedure of establishing w hether an assertion is true is called the verification of this assertion. There is no corresponding term for the procedure of establishing whether a predicate applies to an object or not; but there would seem to be no harm in also using the term ‘verification’ for this purpose. T ransferring the term in this way seems natural because the procedure of establishing whether a predicative assertion that a is F is true is, according to the truth-definition, identical with that of establishing whether the predicate *F' applies to the object a. Nothing is prejudiced by this transference beyond what is asserted in the truth-definition. What is achieved with the transferred terminology is merely a handy term for speaking about a rule for establishing whether a predicate applies. This rule we can now call the ‘verification- rule’ of the predicate. Accordingly, to understand a predicate is to know its verification-rule, i.e. to know how it is established whether it applies to an (arbitrary) object or not. And, correspondingly, to explain a pred­icate would be: to explain its verification-rule.

How can this be done? I can here refer back to the description which I gave of the explanation of the mode of employment of a classification- expression. (The concept o f a ‘classification-expression’ provides a generic concept for predicates and quasi-predicates.) A classification- expression, we saw, if it is not explained by means of other words, is explained by explaining its mode of classification by means of positive and negative examples in perception, or, putting it ^another way: by

Analysis of the predicative sentence 262

employing it positively and negatively in appropriate perceptual situa­tions. We saw at the time (p. 159) that if what is explained in this way is the expression’s mode of employment, in other words, if the classifica­tion-expression is employed in the same way in which it is explained, then it is a quasi-predicate and not a predicate (or this is precisely what defines a quasi-predicate). On the other hand, what is explained by the exemplary positive and negative employment in the perceptual situa­tion when what is being explained is a predicate, we can now see, is not the expression’s employment-rule but its verification-rule. I shall eluci­date this by means of an example. If the explanation of the word ‘red ’ by its exemplary employment in appropriate perceptual-situations is understood in such a way that the person to whom it was explained employs it in the same perceptual situations and only in them (thus if and only if he perceives something red) then he has in terpreted the word as a quasi-predicate. By contrast he has in terpreted the same explanation of the word as an explanation of the predicate ‘red ’ if he also employs it outside the perceptual situation in a way that indicates that he has understood that what was explained to him is not the employm ent-rule of the predicate but its verification-rule.

O f course, this description of the difference between predicate and quasi-predicate is still unsatisfactory. Two questions arise: (1) How is a predicate employed outside the perceptual situation if its verification- rule has been understood, and in what relation does this employment stand to the employment in the perceptual situation? (2) Can one really say that the same expression - e.g. the word ‘red ’ - is understood now as a quasi-predicate and now as a predicate depending on how the expla­nation is understood? Is it not rather characteristic of the explanation of the predicate that it can only be accomplished by means of a sen­tence, so that, for example, we can only explain the word ‘red ’ as a predicate by means of sentences of the form ‘This is red/not red ’?

I shall start with question (2). It correctly points out that a predicate can only be explained as an expression that supplements a singular term . Clearly the singular term that is of particular relevance here is ‘this’, a term which is characteristically used to stand for an object which is present in the perceptual situation. However, it would be a mistake to regard the supplem entation of the classification-expression by ‘this is . . in the explanation-situation, as a sure sign that the expression is understood as a predicate. On the contrary, a m ode of employment of the expression ‘this is red ’ can be conceived in which ‘red ’ still functions as a quasi-predicate. So what is essential cannot be the m ere fact of supplem entation by ‘this is . . .’; ra ther it must be the special mode of

The mode of employment of predicates 263

em ploym ent of ‘this’. By investigating this question we shall achieve a prelim inary insight into the essence o f the expressions which supple­m ent predicates to form sentences, viz. singular terms.

To clarify the distinction with which we are here concerned I shall start from the sim pler expression ‘red h e re ’. And, as in the introduction of the notion o f quasi-predicates (p. 159f), I can again appeal to how children speak in an early stage of acquiring a language (though again it is only by way o f illustration; it is for ou r purposes irrelevant w hether it be empirically correct or not). If a child says, not simply ‘bow-wow’, but ‘bow-wow h e re ’, simultaneously pointing with his finger to a partic­ular place or in a particular direction, then, by the criterion given, the expression ‘bow-wow’ is functioning as a quasi-predicate, not as a p red­icate. For the em ploym ent o f such an expression is explained like this: if som ething bow-wow-like appears at a particular place in the percep­tual situation, one points at this place and says ‘bow-wow h ere ’. The child has correctly understood the explanation if it uses the expression in the same way in which it was used in the explanation-situation. This was the criterion for the classification-expression being a quasi-predi­cate. T h e em ploym ent-rule of the whole expression ‘F here’ (combined with the corresponding gesture) is a conditional rule. To understand it is to know in which circumstances it is to be used.

T h e expression ‘this is F \ like F h e re ’, is also used by simultaneously pointing to a particular place in the perceptual situation. A nd we can easily imagine a language-gam e in which ‘this isF ’ is used in accordance with the same em ploym ent-rule as ‘F h e re ’. It would thus be conceivable that we explain the word ‘re d ’ to a child by exem plary use of the expres­sion ‘this is re d ’. I f our criterion for the child’s having understood the explanation is that it uses ‘this is red ’ in corresponding situations, then the word ‘red ’ is functioning as a quasi-predicate.

As we use the expression ‘this is F' in our language, however, ‘F ’ is not a quasi-predicate, for although we also use the classification-expres­sion ‘F ’ in combination with the word ‘this’ we do not only do so. What decides that it is a predicate, and not a quasi-predicate, is that we also use it in combination with other supplem entary expressions, in combi­nation with which it was not explained to us. This means that when it is combined with these o ther supplem entary expressions it is being used in a way which does not correspond to the explanation-situation. But how, you may ask, can the possibility o f also using ‘F ’ with o ther supple- m entary-expressions have the consequence that already in ‘this is F ’ it is not functioning as a quasi-predicate? T he answer is that we can not only combine the classification-expression ‘F ’ with other supplem entary

Analysis o f the predicative sentence 264

expressions to say something different from what we say with ‘this is F\ but also to say the same thing from another situation. What we express in the perceptual situation with ‘this (to which I am now pointing) is F’ we can take up again as the same, outside the perceptual situation, with ‘that (to which I then pointed) is F\

With this we touch upon the decisive point that I anticipated in an earlier lecture (p. 223f ): it is the substitutability o f (situation-related) sin­gular terms for each other such that we can say that in the different situations we mean som ething identical, the same concrete object, that makes it possible to say something identical by means of the sentences uttered in the different situations; and it is only in this way that the use of the sentence becomes an assertion. T he criterion for the word ‘this’ in ‘this is F’ being a singular term is that it is combinable with another expression by means of the two-place predicate ‘is the same as’ (* = ’). (cf. p. 23) It is only in this second mode o f employment of the word ‘this’ that, in contrast to the first (where it is analogous to ‘here’), the situation-item pointed at is m eant as something identifiable; and this means: as something, as an object. And if predicates are expressions which are supplem ented by singular terms to form elementary sen­tences then ‘F’ is not a predicate merely by virtue of being supple­mented in the explanation-situation by ‘this is . . .’; rather it must be supplem ented by ‘this is . . .’ in such a way that other expressions can be substituted for ‘this’ by use of the identity-sign.

This provides us with a basis for the inquiry into the mode of employ­m ent of singular terms on which we can later build. For the time being our investigation of singular terms is limited to what is necessary for understanding the m ode of employm ent of predicates. T hat a classifi­cation-expression which we explain in the perceptual situation by means of examples only functions as a predicate if it is supplem ented by an expression which can be replaced by other expressions by means of the use of the identity-sign also follows from the truth-definition. For only if we employ the word ‘this’ in such a way that it stands for an object, for something identifiable, can one say that the predicate applies, or does not apply, to an object - the object for which the word ‘this’ stands.

We are now also equipped to answer the first of the questions just raised, viz. in what relation does the ordinary use of a predicate stand to its use in the explanation-situation in which one shows how it is ver­ified? How the predicate is verified is explained by using the sentence ‘this is F (or not F)’ in different appropriate situations, thus where the word ‘this’ stands for a d ifferent object on each occasion, but always for

The mode of employment of predicates 265

one that is given in perception. In this way one shows how it can be established in regard to an arbitrary object whether the predicate applies; and this is what it means to explain its verification-rule. The ordinary use of the predicate, on the other hand, is that in which it is not combined with the word ‘this’, but with a singular term V where ‘a stands for an object that need not be given in the perceptual situation. But what then is the employment-rule of this general employment of ‘F ’ in ‘Fa, if the employment that was explained was the particular employ­m ent of ‘F’ in sentences of the form ‘this is F I This question would be unanswerable if the employment-rule were determ ined by the employ- ment-situation. The answer follows automatically, however, if to under­stand the predicate is to know its verification-rule. For we can now apply the general characterizations arrived at in the abstract analysis of the use of assertoric sentences: to employ a predicate F ’ in combination with a singular term ‘a’ is to assert that it can be established, in accor­dance with the verification-rule that was explained by means of sen­tences of the form ‘this is F\ that the predicate applies to the object referred to by V; but this means: it is to assert that a certain sentence ‘this is F\ namely that which we can use in the situation in which we perceive a, thus when we can at the same time say ‘this = a , can be correctly used in accordance with the verification-rule of ‘F \

With this we have achieved two things: (1) it is now clear that the mode of employment of ‘F' that is explained by the particular employ­m ent of F ’ in sentences of the form ‘this isF ’ is already the general mode of employment of ‘F ’ in sentences of the form F a. For what is explained when the verification-rule is explained by means of sentences of the form ‘this isF ’ is what is asserted when the predicate is employed in any predicative sentence. (2) with the description I have given of how the employment of predicates is explained I have also explained the word ‘applies’; I was able to dispense with the word in the explanation I finally gave. This means that we now have a specification of the truth- condition of a predicative assertion that a is F in which the word ‘applies’ no longer occurs: the assertion that a is F is true if, in the situation in which one can substitute the word ‘this’ for la (can say ‘a is this’), one can correctly use the sentence ‘this is F\ in accordance with the presupposed explanation of the verification-rule of ‘F \

T here is no circle here between the word ‘true’ in the definiendum and the word ‘correct’ in the definiens. For ‘correct’ here does not yet have the meaning of ‘true’; it means no more than ‘corresponding to the rule’. T he correct employment of the predicate F ’ in the verifica- tion-situation is explained in exactly the same way as the correct

Analysis of the predicative sentence 266

employment of the quasi-predicate ‘F \ T ha t this correct employment has now the sense of establishing that the predicate applies to an object, and hence of establishing that an assertion is true, is due to the fact that the supplem entary expression ‘this’ is understood in such a way that it is replaceable by other expressions by means of the identity-sign. T hat the explanation of the classification-expression is understood as the explanation of a predicate (and this means: as the explanation of how one establishes that this expression applies to something) rests on the fact that in the explanation it is presupposed that the word ‘this’ is understood as a singular term . And this much has already become clear: the fact that ‘this’ is understood as a singular term cannot be disposed of simply by saying that this expression ‘stands for’ an object. Rather we shall now have to ask what it means to say that an expression ‘stands for’ an object. And if the criterion for an expression’s being a singular term is that other expressions can be substituted for it by means of the identity-sign (which expressions are then likewise singular terms) then we shall have to ask what the rules are which govern such substitutions; and to ask this question is to ask about the employment- rule of singular-terms, or to ask how they are explained. Only when this is clarified will we be able to understand what it is to establish that a predicate applies and hence what it is for a predicative assertion to be true.

We have now progressed far enough to begin the enquiry into the semantics of singular terms. With this we bring the language-analytical critique of the object-orientated position to bear on the latter’s own point of departure. So far the critique of the object-orientated position concerned only the tendency of the latter to transfer the only formal category at its disposal, that of an object, to the other linguistic expres­sions. As regards this category itself, however, and as regards singular terms, there seems, at first sight, nothing to find fault with in the tra­ditional conception. So in what way, one might ask, is the traditional idea that singular terms - and in particular proper-nam es - stand for objects incorrect? Isn’t a singular term by definition an expression that stands for an object? Indeed it is. So one certainly cannot say that the traditional view, as far as it goes, is not correct. However, the suspicion arises that it does not go far enough. By this I mean that the object- orientated position is probably not in a position - this is something we will have to examine - to explain what it is for an expression to stand for an object. In that case it may also be surmised that the object-orien­tated position cannot explain its own basic concept: that of an object. It might seem paradoxical that a philosophical position should be inca­

The mode of employment of predicates 267

pable of explaining its own basic concept; but not if you consider that if it starts out from a particular concept a philosophical position has no dimension into which it can step back and from which it could explain that concept.

By contrast, the language-analytical approach has at its disposal an explanatory dimension which com prehends the relation to objects. In very general terms this counter-position is expressed in the idea that singular terms are non-independent expressions and that the prim ary semantic unit is not the name but the sentence. I f singular terms are essentially expressions needing supplem entation this means that an object is essentially som ething classifiable and that the relation to objects must be understood, in a way still to be clarified, in term s of the relation of assertions to tru th .

In the truth-definition of predicative assertions: ‘T h e assertion that a is F is true if and only if the predicate “F ” applies to the object for which the singular term “a ” stands’ (which I also took as my starting-point in the explanation of the semantics of predicates) we already have a for­mula for expressing the idea that the function of singular terms - that o f standing for objects - is em bedded in the function of predicates - that of classifying objects - and hence in the context o f the tru th and falsity of predicative assertions. However this form ula cannot do more than suggest an essential connection between the ‘standing for’ o f sin­gular term s and the applying (Zutreffen) of predicates and the tru th of assertions. It does not by itself show that one cannot explain what it is for a singular term to stand for an object independently of this form ula.

Indeed you could even argue that this form ula appears to suggest the opposite: whereas the questions of w hether the assertion is true and w hether the predicate applies are mutually dependent (cf. p. 253) nothing corresponding holds for the question of which object the sin­gular term stands for. In the case of the predicate we could reverse the truth-definition and say: the predicate tF> applies to the object for which the singular term ‘a stands if and only if the assertion ‘Fa is true. Clearly we cannot say analogously: the singular term ‘a stands for that object which, if the predicate ‘F' applies to it, makes the assertion ‘Fa* true. For then the object a would not be distinct from all o ther objects to which the predicate ‘F’ applies.

Thus we here encounter an asymmetry regarding the roles o f the predicate and the singular term in the predicative sentence.1 O ne can, it is true, say that the tru th of the assertion depends both on which object the singular term stands for and on w hether the predicate applies to this object. But then one m ust be clear that, whereas the first

Analysis of the predicative sentence 268

of these conditions can be form ulated independently of the second, the second cannot be form ulated independently of the first: the question of whether the predicate applies to the object presupposes that we know to which object, thus which object it is for which the singular term stands. T hough one does not have to, one can describe this state of affairs as follows (this is Strawson’s view):2 the tru th of the assertion does not depend on which object the singular term stands for and on whether the predicate applies to it; ra ther the question of the tru th or falsity of the assertion presupposes that we know for which object the singular term stands. T he tru th of the assertion depends (then only) on w hether the predicate applies to this object. W hether the state of affairs should be thus described is disputed. I f one describes it thus this has the consequence that if there is no object, or more than one, for which the singular term stands, the assertion does not count as false. We must then say either that it is neither true nor false or that the person who uttered such a sentence did not assert anything. A lthough Strawson and others have made much of this question of whether one should in ter­pret it in this way or that it is of only limited importance. W hether the assertion is called false or neither true nor false in such cases is a m atter of convention and may vary from language to language. I shall be re turn ing to this question (Lecture 22). T he asymmetry between pred­icate and singular term , however, from which this question starts out, represents a non-conventional, substantive difference. W hereas the question o f w hether the predicate ‘F' applies to the object a depends on our knowing for which object the singular term ‘a’ stands, the reverse is not the case: the question of which object the singular term ‘a’ stands for m ust be settled independently of, and is the condition of, being able to establish w hether the predicate ‘F’ applies to it.

However, we should not read too much into this result. All that fol­lows from it is that in the question of the tru th of a particular assertion Fa the question o f which object the singular term ‘a’ stands for must already be decided independently of the question of the tru th of this assertion (or of w hether this predicate applies to this object). On the other hand it does not follow from this that what in general it means for a singular term to ‘stand fo r’ an object can be explained indepen­dently of the fact that it is something to which predicates can apply, or not apply. Thus from the truth-definition of predicative assertions we can arrive at neither a positive nor a negative decision regarding the question of w hether what it is for a singular term to stand for an object can be understood independently of the context of predication. Q ur line of thought so far suggests that the object-relation cannot be under­

stood independently of the truth-relation. On the other hand the asym­metry that has just been described should make us cautious. We cannot expect that what has been shown for predicates can simply be trans­ferred to singular terms. The question can only be decided by applying to singular terms the same basic question which guided us in the case of predicates, viz. the question regarding the mode of employment of these expressions, and this means: the question of how we can explain the employment of expressions of this type.

The mode of employment of predicates 269

L E C T U R E 2 0

W h at is it for a sign to stand for an object? T he traditional account

In the enquiry into the semantics of singular terms there is no reason why we should not follow the traditional philosopher in describing the relation between the singular term and the object by saying that the singular term stands for the object. But at present we should regard this expression ‘stands for’ as a cipher, as an expression whose sense has yet to be specified.

I earlier (p. 15If) called W ittgenstein’s dictum ‘The m eaning o f a word is what the explanation of its meaning explains’ the fundam ental principle o f analytical philosophy, because it instructs one to frame the philosophical question about the m eaning of linguistic expressions in a way that exactly corresponds to the pre-philosophical question and in fact merely formalizes the latter. According to this principle the philo­sophical question concerning the understanding o f a form o f expres­sion should be construed as the question of how pre-philosophically we explain expressions of this form. In the case of singular terms we have to do with expressions which not only have a meaning but also stand for an object. In whatever way these two aspects may be connected, if the expression stands for an object at all then the question about the expression’s mode of employment must embrace, together with the question of how its meaning is explained, that of how one explains for which object it stands. And we can only answer the philosophical ques­tion of what it is for an expression to stand for an object by explaining how pre-philosophically one explains in a particular case which object an expression stands for. To explain what it is for a predicate to apply to an object we asked how we would explain how we establish that a predicate applies to an arbitrary object. Likewise we can only explain what it is for a singular term to stand for an object by asking how we establish for which object a singular term stands.

The question ‘How do we establish it?’ is quite indispensable. W ithout

What is it for a sign to stand fo r an object? 271

it any explanation of an expression would rem ain in the air. Within semantics this recourse to establishing corresponds to the move we made in the enquiry into the semantics of whole assertoric sentences from truth-conditions to verification-rules. In precisely the same way we were able to see in the case of predicates that the m eaning of ‘applies’ can only be explained by asking how it is established that a predicate applies to an object. W ithin ontology this move corresponds to the epistemolog- ical (‘transcendental’) tu rn which ontology took in m odern times; so if the question concerning the m eaning of ‘stands fo r’ assumes this form there is nothing to prevent even the traditional m odern philosopher from going along with us. W hoever refuses to pose the question in this form thereby refuses to enquire into the m eaning of the relation between name and object at all, however this relation is designated.

Such an attitude is not characteristic of the traditional position but of certain m odern semantic theories, those, namely, which in the specifi­cation of the m eaning of assertoric sentences rem ain at the level of truth-conditions and specify the latter in a meta-language. In these the­ories the nam e-object relation is so construed that each singular term is ‘assigned’ to an object. This expression ‘assignm ent’ is ju st another word for ‘standing for’; and like the latter it rem ains vacuous so long as one does not show how it can be established to which object a singular term is assigned. W hat actually happens in theories of this kind is that the singular term is simply assigned to another name, a nam e in the m eta-language. These theories are merely logical or linguistic tech­niques and do not am ount to a philosophical position at all. By merely presupposing, ra ther than explaining, how we are able to assign a sin­gular term to an object these techniques merely presuppose, ra ther than explain, how we are able to refe r to objects and what it means to speak of an ‘object’. By contrast it is characteristic of both the episte- mologically based m odern theory of objects - transcendental philoso­phy - and the language-analytical theory of objects that they seek to clarify the object-relation and what it means to speak of ‘objects by recourse to the question of how objects can be given to us. T he d iffer­ence is that whereas the traditional theory posed this question indepen­dently of the question o f the function of the corresponding signs, the language-analytical theory understands the object-relation as a re fe r­ence (Bezugnahme) which can essentially only be achieved by means of a sign.

B ut if we do not wish to commit a petitio principii vis-a-vis the tradi­tional philosopher we cannot just assume that we can only refer to objects by means of linguistic expressions. Once again what matters is

Analysis of the predicative sentence 272

that we should so frame the question that the traditional philosopher has no difficulty in accepting it. And he can and must agree to clarify the question of what it is for a sign to ‘stand for’ an object by asking how we would explain, or establish, in an individual case, for which object a singular term stands. He must agree to it for one cannot see in what other way the m eaning of ‘standing for’ could be explained. He can agree to it because that there are signs which stand for objects is not a m atter of controversy. W hat the traditional philosopher wishes to stress is simply that referring by means of linguistic expressions is not our only and not our prim ary way of referring to objects. T h e tradi­tional philosopher would thus not approach the question of the object- relation as such from the sign-relation. Nevertheless he too must be capable of explaining this sign-relation. And we can expect that it will be precisely in the question of how we can establish for which object a sign stands that he will find it natural to have recourse to his pre-lin- guistic way of referring. And I imagine that you too, when confronted with this question for the first time, will find it natural to think that we must be able to refer to an object in a non-linguistic way. For how, one m ight ask, can one explain for which object the sign stands unless one is somehow able to refer directly to the object?

But how is this pre-linguistic object-reference to be understood? We shall obtain the answer to this question from traditional philosophy itself by taking as our point of departu re a traditional theory of singular terms which can be regarded as representative. We find such a theory in John Stuart Mill. By examining his theory we will be able to get a clear picture of what a pre-analytical object-theory must presuppose in asking how we can explain, or establish, for which object a singular term stands. As before if we do not wish to proceed dogmatically we must allow the way forw ard to be determ ined only by the difficulties in which the traditional way of posing the problem itself becomes entangled. A nother reason for choosing Mill’s theory to debate with is that after having long been regarded by analytical philosophers as disposed of it has recently been finding supporters.1

Before I begin the systematic enquiry into the employment of singu­lar term s we should recall the hints for the understanding of singular terms yielded by the analysis of the m ode of employment of predicates. These hints cannot have a justificatory function in the investigation of singular terms that is now to be undertaken. However, though com­pletely hypothetical and vague, they can indicate a direction in which we m ust expect the explanation of singular terms to proceed if our explanation of the mode of em ploym ent of predicates is to be retained.

What is it for a sign to stand fo r an object? 273

T he distinction between predicates and what I have called quasi-pred­icates involved no empirico-genetic hypothesis; and it served only inci­dentally to contrast predicate-language with more primitive languages. The indispensable significance of this distinction arose internally from the requirem ents of an explanation of the mode of employment of predicates. As was shown in the last lecture this is relatively complex. T he explanation of quasi-predicates enabled us to isolate a first level at which we explain a classification-expression without using words like ‘true’ and ‘applies’, but simply the word ‘correct’ in the sense of ‘con­form ing to a ru le’. If we could ignore the fact that a predicate is employed outside its verification-situation it would be a quasi-predicate. Because an expression which would otherwise be a quasi-predicate becomes a predicate by virtue of its need to be supplem ented by a sin­gular term it seemed plausible to suppose that the function of the sin­gular term must be (a) to make the classification-expressions - now sup­plem ented to form sentences - situation-independent and (b) to refer (verweisen) from the employment-situation to the verification-situation. T hat which is classified no longer results of itself from the employment- situation as in the case of the quasi-predicate but is taken up into the linguistic expression by being represented by the singular term ; T here is thus constituted a speech-act which belongs to a class of speech-acts in o ther situations which all ‘say the same’. The correctness of these speech-acts is independent of the employment-situation and points for­ward to a special situation in which this correctness is established, a correctness which we call ‘tru th ’. I then also indicated that it seems plau­sible to suppose that in the constitution of situation-independence a special role is played by the situation-relative singular terms, the deictic expressions. And in addition we saw in the last lecture that deictic expressions also have a special role inasmuch as the reference (Verwei­sung) of the singular term to the verification-situation is to be under­stood as the reference to a deictic expression to be used in this situation.

These connections suggested by the explanation of the mode of employment of predicates do not correspond to the usual, not even the usual language-analytical, conception of singular terms which starts out from the relation of singular terms to objects as something obvious and requiring no fu rther explanation. But since, on the other hand, it can­not be doubted that singular terms stand for objects, then, if these per­spectives for the understanding of singular terms should turn out to be correct, one would have to be able to show, firstly, that such a thing as an object-relation is only made possible by the situation-independence of speech accomplished by means of deictic expressions, a situation-

Analysis of the predicative sentence 274

independence which at the same time makes possible a truth-relation. Secondly, we would have to expect that the most basic (ursprünglich) objects by reference to which all other objects in their objecthood are to be understood are the speech-situations themselves.

T he perspectives for the semantics of singular terms arising from the discussion of predicates to which I have so far referred are relatively vague. However, the connection between the semantics of singular terms and that of predicates acquired a more determ inate form through the truth-definition. I said that the aim of the investigation is a non-verbal explanation of the employment o f the word ‘tru e’ in pre­dicative assertions (and that could only mean: an explanation of how one establishes that assertions of this kind are true) by means of a non­verbal explanation of (1) how one establishes that a predicate applies to an object (i.e. the employment-rule of predicates) and (2) how one establishes for which object a singular term stands (i.e. the employment- rule of singular terms). Thus what we are looking for is an explanation of the employment of (elementary) predicative sentences in which not only the word ‘true’ but also the words ‘applies’ and ‘stands for’ no longer occur (and in this way all three words - or the employment of the corresponding semantic classes of expressions - would be explained). Now in the last lecture I gave an explanation of what it is to establish that a predicate applies which no longer depends on an understanding of the word ‘tru e ’ but (a) rests only on an understanding of the word ‘correct’ (in the sense of ‘conforming to a ru le’) (b) presupposes an understanding of the employment o f singular terms. In this explana­tion I used neither the expression ‘stands fo r’ nor the expression ‘object’, both of which themselves require explanation; I merely made use of the understanding of the two-place predicate ‘is identical with’ (' = ’).

T he following additional perspectives for our enquiry into the employment-rule of singular terms result from this. (1) In the last lec­ture the close connection between objecthood and identity - that objects are essentially something identifiable - was simply presupposed as plau­sible. This was an anticipation which can only be justified in the context of the systematic discussion of singular terms. (2) What is meant by predicates applying to something cannot be understood without at the same time understanding what is m eant by objects, or what it is for a singular term to stand fo r an object. Now that it has proved possible to explain this object-relation that is implicit in the notion of applying by, instead of speaking of a singular te rm ’s standing for an object, merely appealing to an understanding of the identity-sign in connection with

What is it fo r a sign to stand fo r an object? 275

singular terms and in particular deictic singular term s, it would seem natural to expect that the standing-for of singular terms, even when it is explained by itself, can be reduced to a corresponding understanding of the identity-sign. Identity would then not merely be essential to objecthood (as is generally acknowledged); the m eaning of object­hood could be understood without rem ainder in terms of identifiability. The consequences of this for the concept of tru th would be as follows. Since the understanding of the word ‘tru e ’ in its application to predic­ative assertions rests on the understanding of ‘applies’ and ‘stands for’ the elem entary use of the word (which recursively underlies the other uses) could be explained purely on the basis of an understanding of ‘correct’ and the identity-sign.

Having looked at these hypothetical perspectives for a possible lan- guage-analytical explanation of ‘standing for’ suggested by the discus­sion o f predicates I shall now begin the systematic discussion of the question of how one can refer to objects by means of singular terms with a critical assessment o f the traditional conception. I shall for the most part ignore singular term s which stand for abstract objects - such as properties, or states of affairs - and concentrate on those which stand for concrete (and that means perceptible) objects. In connection with the interpretation o f Husserl’s theory o f meaning (p. 150) I pointed out that one can take as one’s starting-point a crude distinction of concrete singular terms into deictic expressions, p roper names and descriptions. In the traditional theory, apart from proper-nam es only descriptions have played a role. It was prim arily orientated towards p roper names because these seemed to exhibit the standing of a linguistic expression for an object in the most simple and direct way. Already influenced by Frege, Husserl here departed from the usual conception. Like Frege he orientated him self prim arily towards descriptions, for in their case one m ust distinguish the object and the m eaning o f the expression. As the classical traditional conception was incapable of dealing with a m eaning that could not be construed as an object it took the proper name to be the prototype of all linguistic expressions. Even for Russell the m eaning of an expression was the same as its object.2

A definite description - ‘the so-and-so’ - seems somehow to refer indirectly to an object. It does not stand simply and directly for the object, but - w hether or not one wants to speak o f a m eaning here - by means of an attribute, which, as the use of the definite article ‘the’ implies, belongs to only one object. As soon as one considers descrip­tions and p roper names the question of their m utual relationship arises. In Husserl there is no explicit theory o f p roper names. Frege, on the

Analysis of the predicative sentence 276

other hand, took the revolutionary and, for the tradition which fol­lowed him, decisive step of treating proper names, not as the more elementary, but as the higher-level expressions: to establish for which object a p roper nam e stands we have to have recourse to descriptions.3 For example, if we use the p roper name ‘Aristotle’ and are asked for which object, for which person, this nam e stands, we can only give descriptions: Aristotle, that is the person who was born in such-and- such a place at such-and-such a time, the pupil of Plato, the teacher of A lexander the Great, the philosopher who wrote most of the texts that have come down to us under his name, and who died in Chalkis in 322 B.C. All these descriptions stand for the same object, and their d ifferent meanings correspond to its different ‘modes of presentation’. Thus underlying the primacy of descriptions over proper names urged by Frege is the ontological-epistemological idea that there is no such thing as an ‘object as such’ behind its modes of presentation, which could be somehow directly referred to by means of a proper name. Though im portant, this insight, which was also taken over by Husserl, does not, as such, explain what it is for a singular term - and for Frege this means: a description - to ‘stand for’ an object.

Let us now leave this conception and return to a conception which also distinguishes between proper names and descriptions but which regards the object-relation of proper names as the more elementary. We find such a view in Mill.4 Mill calls all categorematic expressions, thus all so-called ‘term s’ of traditional logic, ‘names’ and distinguishes general names and individual names. T he for us im portant distinction between connotative and non-connotative names he also applies to gen­eral terms; for us however it is only im portant in regard to individual names, singular terms. All individual names according to Mill are ‘de­notative’, i.e. they stand for an object. But only descriptions are in addition ‘connotative’, i.e. they stand for their object by means of an attribute which they ‘connote’. Proper names, by contrast, are non-con- notative. They do not refer to their objects as to the sole bearers of an attribute but are ‘attached to the objects themselves’.

O f course, one immediately asks: what is one to understand by such a direct assignment? Unlike the meta-linguistic semantic theorists Mill does not evade this question. T he robber in the tale of Alibaba marks a house with a chalk-sign so as to be able to recognize it again as the same, and distinguish it from others; this is how one should think of the func­tion of p roper names. But the use of a name, one will object, does not consist in its being affixed to the object for which it stands. (For certain purposes, of course, this can happen, e.g. in the case of localities, ships,

What is it fo r a sign to stand for an object? 277

participants at a congress. But one would hardly seek to understand the general name-relation by reference to these special cases. On the con­trary, one will only be able to explain those special cases on the basis of an understanding of the general name-relation.) Mill himself raises this objection and answers it thus: what we ‘m ark’ with a name is not the object itself but the idea of the object.

I do not think this view can be dismissed as a peculiar lapse on Mill’s part. I have already pointed out in the introductory discussions (p. 62 f) that it is characteristic of the object-orientated tradition that it conceives of the object-relation as representation (Vorstellung). At the same time I distinguished between a narrower version of the representation-theory which is characteristic of the early modern period and according to which that to which we directly relate is not the object but the represen­tation as its representative (Repräsentant), and a broader version to which even the critics of this theory continued to subscribe and which - even if not under this name - is also characteristic of the ancient and mediaeval conception. According to this broader version, the relation - now understood as direct - to the object is understood as a relation of representation: the object is represented; an object is essentially some­thing representable.

If one asks what is m eant by ‘representation’ here, one can find no answer to this question in traditional philosophy, because the concept of ‘representation’ is a basic concept for m odern traditional philoso­phy.5 And of ancient philosophy one can say, retrospectively, that the concept of representation implicitly underlay its distinction between ais- thesis and naus, for which there was no comprehensive generic concept.6 One cannot get a clear grasp of this concept from the outside either for it rests on an uncashable metaphor. T he only thing we can do is to make it clear to ourselves what this m etaphor is. T he point of departure of the m etaphor is a representation in the sense of a perceptual (anschau­lich) or imaginative (optical) image, or image consciousness. Taking this as the basic model for the consciousness-relation in general, the relation to objects was conceived as a having-before-oneself (Vorsichhaben) of an optical image; only this having-before-oneself is no longer to be un d er­stood as sensuous perception (Anschauung). How it is positively to be understood was bound to remain undecided, for the m etaphor becomes vacuous. A non-sensuous representation, a quasi-representa­tion is a pure fiction. For the concept of the object itself the conse­quence was that it had to be essentially thought of as the correlate of such a non-sensuous quasi-representation. Every representative of tra ­ditional philosophy will resist this apparently malicious interpretation.

Analysis o f the predicative sentence 278

But then ask how he positively understands the relation between con­sciousness and object. You can read the history of pre-analytical philosophy from Parmenides on from beginning to end and back again and you will find no answer, unless it be one in which one adds to the confusion by saying, as in G erm an Idealism, that the relation is to be understood as a relation between subject and object. W hen we ask what this is we are told: the identity of the different (or: the identity of iden­tity and non-identity) and suchlike. It is, of course, simply nonsensical to say that a= b and ai=b. And even if this were not nonsensical it is impossible to see how this formal ontological relation, simply by being twisted in upon itself can be made to yield something like a consciousness-relation, regardless of how often this twisting is ‘reflected’ and repeated (as in Hegel’s Logic). O f course the advocate of such theories will claim that he is here being criticized merely from the standpoint of the ‘understanding’. But in that case one should ask him how he can explain his special (‘reason’) concept o f identity, non­identity, etc., without presupposing it. A nd for our part we can see his belief that he has to appeal to such a special insight or structure inacces­sible to ordinary hum an understanding as a desperate attem pt to escape from the dilemma of object-orientated philosophy without su rrender­ing its basic presupposition which consists in not recognizing that we relate to objects by means o f linguistic signs which for this purpose must be employed in a specific way. It was not an avoidable accident but in­evitable that traditional philosophy should take the metaphorical con­cept of representation as its starting point; for if one ignores the logical, the linguistic, nothing is given to us in consciousness but the sensible. T herefore the only alternative to a language-oriented explanation of the relation to objects is an explanation in terms of a m etaphor based on perceptual images.

Mill’s conception of the object-relation of names, according to which what is directly designated is not the object but the idea, belongs to the narrow er version of the representation-theory. We can ignore this peculiarity; only when we view Mill’s conception as representative of the broader version of the representation-theory will its fundam ental significance for the whole tradition become evident. His theory can easily be detached from the narrow version of the representation-theory if instead of saying that the name does not stand directly for the object, but for its representation (idea), we say that it stands for the represented object, thus for an object as object of an act of representing. Mill’s specific conception is merely a variant of this conception, which we can regard as the traditional conception pure and simple.

What is it fo r a sign to stand for an object? 279

W hat follows from this general orientation towards the concept of representation for the nam e-object relation? Representations in the non- m etaphorical sense, that is perceptual images, are simple data which can be taken in at a glance. It is this aspect that is carried over in the m etaphorical extension of the concept of representation to the object- relation, or at any rate can be carried over. Thus because objects are understood as what can be represented it becomes possible to in terp ret them as simple data like perceptual images and it thereby becomes pos­sible to conceive of the object-relation of a nam e as m ere association or assignment. A nd it is only if the object is conceived as a simple datum that it is possible to assume a direct relation o f a ‘non-connotative’ name, a proper nam e, to ‘th e’ object, a relation which does not involve - as on Frege’s view it m ust - a reference to descriptions. F rege’s conception, by contrast, presupposes that an object is not som ething representable, a simple datum , but som ething which essentially manifests itself in manifold modes of presentation.

T he idea that the object-relation of p ro p er names is somehow more basic than, or at least independent of, that o f descriptions implies that there can be a relation between sign and object which can be understood as simple assignment. And this idea can only be either the product of a m ere failure to think - and we m ust suppose this to be the case with m odern meta-linguistic assignm ent-theories - or if it is to be philosoph­ically justified it implies that objects can be simply given, as simply as the signs to which they are assigned; and this means tha t it is based on a representation-theory of objects. At any rate I do not know of any other possible way of conceiving of objects as som ething simply given as this is required by the assignm ent-theory.

1 should here m ention a possible objection. You could ask: by p re­senting. things in this way, do I not presuppose that at least the signs themselves are representable, and are not the signs also objects? T he question is to be answered in the affirmative. T he sign-types are indeed representable and m oreover — this is crucial - in a non-m etaphorical sense. And clearly we must also regard them as objects. However, as sign-types they are abstract objects, whereas the objects ‘fo r’ which those signs that are (concrete) singular terms stand are concrete objects. My opposition to the traditional idea that reference to objects is essentially to be understood as representation, w here ‘represen tation’ has an uncashable m etaphorical meaning, does not rule out that am ong other objects there are also objects which are actually - and in a harmless, non-m etaphorical sense - representable.

Let me again point out that Mill’s theory, w hether in his own narrow

Analysis of the predicative sentence 280

version or in my generalized version, is a theory of the object-relation of names which implies that our relation to objects is as such not a lin­guistic one. H ere too it seems to me that the representation-theory is the only possible alternative to understanding the object-relation and what is meant by objects in terms of the em ployment of a species of sign (at any rate the only one with which we are familiar). Although in another place (in the same §5 of Chapter 2 of his Logic) Mill says it is the function of names to distinguish objects, in the place where he speaks of the relation of names to representations (‘ideas’) he restricts the function of names, consistently, to that of arousing in the mind those representations of objects with which they are associated. And this clearly, is what it means, for the traditional position, for a name to ‘stand for’ an object. It means that it ‘stands fo r’ a representation of an object in the sense of being associated with it. It is now also clear why for the traditional con­ception the semantics of singular terms is independent, does not involve reference to a sentence-context. For a representation in the non-m eta­phorical sense does not belong to such a context; thus representation in the metaphorical sense and accordingly what is to be understood by an object does not belong to such a context either.

You could object that so far all I have really shown is, at most, that the assignm ent-theory of nam e and object and the idea of the primacy of p roper names are grounded in the representation-theory. On the other hand it has not been made clear on precisely what my hostility to this theory itself rests. W hat are the criteria for deciding that it is incor­rect or even impossible? We must here distinguish between the doubtful character of the extended representation-concept as such and that of the idea that what we mean by objects is to be construed as representa­tions, or som ething representable.

As far as the extended, metaphorical representation-concept as such is concerned I cannot add anything essential to what I have already said. My thesis is that the m etaphor cannot be cashed; that it does not make sense to speak of a non-sensuous or somehow intellectual repre­sentation. This first step in my critique is not conclusive, for an assertion that a word is meaningless, that a m etaphor cannot be cashed, cannot be explained, may simply reflect the narrowness of the person who makes it. In principle then I m ust leave open the possibility that someone will succeed in m aking good sense of this extended use o f the term ‘repre­sentation’.

Secondly, ju st supposing that it does make good sense to speak of representations, can we understand what is meant by objects, and ref­erence to objects, in terms o f this notion? Let me begin by showing that

What is it for a sign to stand for an object? 281

this thesis of the representation-relation to objects is by no means merely something we find in the philosophical tradition; it also seems natural to us. I shall leave it open whether this is because we belong to this tradition. It will be best if I remain subjective and merely relate to you how it was in my own case when I became aware of this problem; I assume it could be similar with many of you. Ordinarily one does not reflect on what one means when one speaks of objects. When I reflected on this I noticed how strong my tendency was to think of an object as something over against me (ein Gegenüber). Indeed this seems to be already implied by the words ‘Gegenstand’, ‘ob-ject’. I reflected further that what can most indubitably be term ed an over-against is the optical, pictorial view in which things manifest themselves to us (particularly when we are sitting still). And then I noticed that I was thinking of the relation to objects on this model. Of course I then immediately said to m yself: it is clearly not this optical view of the object, e.g. of this desk, that I mean when I mean the object. I can walk round the desk, I can also perceive it with other senses. I can also talk about it - or, if you do not want to hear anything about talking, think about it - when I am not in the room. And without doubt when I do so I mean the same object, namely this desk. So I clearly cannot stick to the naive idea of an optical over-against. However the idea of an over-against seems even then so natural that I say: the object is an over-against, only not a pictorial one. These reflections show that the thesis that we relate to objects by means of representations in the metaphorically extended sense can arise quite naturally.

T he doubt whether this metaphorical sense still has any content should not now disturb us. We leave this question open, act as if there were such a meaning and ask: could this really be what we mean when we speak of an object? How can we approach such a question? Clearly only by already having some preliminary conception of what we mean by ‘object’. And how do we arrive at such? Surely not by mere analysis of the use of the word ‘object’. I shall come back to this question (cf. also p. 22). We do not need to answer it now for it suffices for our present purpose to fall back on one thing which is not a m atter of controversy between the traditional philosopher and his analytical critic, viz. the fact that there are certain linguistic expressions - singular terms - which somehow ‘stand fo r’ objects. O f course we cannot now understand what in general it is to relate to objects in terms of the use of these expressions; that would be a petitio principii in favour of the language-analytical posi­tion. But we can assume this much: that one must be able to explain how these expressions stand for objects, even if this should not be

Analysis of the predicative sentence 282

essential to the sense of the object-relation. And we can now apply to our problem that basic principle of analytical philosophy, which, as we have seen, must also be accepted by the traditional philosopher: that we must be able to explain how we use a linguistic expression.

It is this principle on which the representation-theory appears to founder, even if there were such things as representations in the extended metaphorical sense. For it would follow from the represen­tation-theory that when we are asked to explain for which object a name stands the question concerns what we represent when using this name. And that the question cannot be thus understood seems clear, at least as regards intersubjective speech. In the case of representations in the non-m etaphorical sense - images - one can understand what it might mean intersubjectively to exhibit them. Representations in the meta­phorical sense, on the other hand, have not been understood intersub­jectively by traditional philosophy and it is not clear how they could be understood intersubjectively.

Again this result is not conclusive. If we ask how a singular term is used, thus if we ask for its explanation, we are not asking what (which object) the person who uses it represents to himself, but rather what (which object) he means, that we too can mean (cf. above p. 63f). T he linguistic object-relation is to be understood as meaning (Meinen) where this word is understood not in the sense in which it is supplem ented by a nominal sentence (‘I believe (meinen) that/?’) but in the sense in which it is sup­plemented by a singular term (‘I mean N ’). How this meaning is to be understood has of course still to be explained. This word does not am ount to an answer to our question; all we know is what we must ask if we are enquiring about the explanation of singular terms. Now the traditional philosopher could also try to bend this question to fit his conception. He would say: one can only explain which object one means with a singular term by recourse to the corresponding representation.

Thus it would seem that we cannot conclusively refute the traditional answer until we have given a new positive explanation of what it is to explain which object one means with a singular term. But this question must be orientated towards how one actually answers, in a particular case, the question ‘Who or what do you mean by that?’ And in fact nobody answers this question by pointing to representations. W eshall see that the question thus posed leads us to a completely new conception of how we refer to objects and of what is meant by ‘objects’.

Of course the following m anoeuvre is still open to the traditional ph i­losopher. Regarding the explanation that is now to be given of how it is established what or who we mean with an expression he could say that

this is precisely what he m eant by his metaphorically extended talk of the representation of an object. However, this m anoeuvre would rep ­resent an abandoning of the traditional position. For it would imply the admission that we cannot explain what it is to mean an object by recourse to representation but that, on the contrary, we can only explain what is m eant by ‘representation’ by explaining what it is to mean an object. The word ‘representation’, if it is newly defined, is, of course, harmless- like any word.

What is it for a sign to stand fo r an object? 283

L E C T U R E 2 1

The function of singular terms

At the end of the last lecture it again became clear how limited the possibilities of internally refuting a philosophical basic conception are. One cannot show - assuming one is prepared to accept the uncashed m etaphor of the representation-concept - that a representation-theory is impossible. Sponge-like thoughts have the advantage that they cannot be smashed. But to dem and an internal refutation would also involve an unrealistic view of the extent to which obsolete philosophical ideas can be put out of action. It is enough to show that the object-relation which we encounter when we examine how we actually establish for which object a sign stands is not a representation-relation and that the meaning of objecthood which emerges from this examination cannot be understood in terms of representation. The genuine refutation of the traditional conception cannot be accomplished internally but only by means of the positive construction of a new conception which right from the start can claim the advantage that it takes its departure from an actual form of reference to objects, namely that by means of linguis­tic expressions (for the present I leave it open whether it is the only one). The traditional conception is thus not shown to be impossible but - disregarding the unclarity of its basic concept - only false.

Before I begin the positive construction of the new conception I wish to add som ething to what I said about the representation-theory. I have shown that the doctrine of the primacy of proper names over descrip­tions, which implies that ‘standing fo r’ is to be understood as simple assignment or association, is grounded, if it has any philosophical foun­dation, on the representation-theory. However, it would be false to sup­pose that the converse holds, viz. that the representation-theory neces­sarily leads to this conception of the name-relation and such a simplistic object-concept. T hat it need not do so is particularly well illustrated by the case of Husserl. H usserl’s object-theory is, on the one hand, a rep ­

The function of singular terms 285

resentation-theory: what he calls intentionality is the not linguistically but representationally construed object-relation of consciousness. On the other hand, we have already seen that Husserl follows Frege in thinking that descriptions are semantically more fundam ental than proper-names. In Husserl this is connected with the ontological-epis- temological idea that objects - spatial ones at any rate - are only acces­sible to us in a manifold of modes of presentation, and that it is only in such modes of presentation that they are constituted as objects. If a presentation of the object in a multiplicity of something like modes of presentation turns out to be also characteristic of the language-analyti­cal theory of objects, then the language-analytical conception must be distinguished from the traditional conception (in so far as the latter is already able to see the object as a unity of modes of presentation) by what it understands by ‘modes of presentation’. We shall see that the modes of presentation to which the enquiry into the employment of singular terms leads us are to be understood as rules for identifying an object. In Husserl, on the other hand, the modes of presentation of the object are understood as ‘adum brations’ (Abschattungen), as the mani­fold perceptual perspectives in which an object presents itself to an observer, depending on his standpoint and the surrounding circum­stances. These adum brations are thus themselves representations, and, moreover, representations in the non-metaphorical sense. For Husserl all spatial objects present themselves in a rule-governed synthesis of simple data - perceptual representations - and this synthetic object- consciousness, as intentional consciousness, is itself understood as rep ­resentation (though now in the metaphorically extended sense).1

Husserl is here clearly following in the tradition of Kant, who was the first to develop a theory of this kind. According to Kant the represen­tation of an object depends on a ‘rule’ which makes a ‘synthesis’ of man­ifold ‘representations’ necessary.2 However Kant’s talk of the ‘object’ is misleading. He would have characterized his problem more clearly if, instead of speaking of ‘objects’, he had spoken of ‘objective connec­tions’. His question is not: how is it possible for us to refer to things, to something (an object)? Rather it is: how far are the connections of our representations not merely subjective but objective? This can be seen clearly from the fact that the text from which I have just quoted is concerned with the justification of the law of causality, the question of why it is that we must regard the sequence of events as not merely subjective but objective. H ere then the problem is in fact a completely different one. To imagine that the problem of what makes possible a relation to objects is approximately the same as the problem of what

Analysis of the predicative sentence 286

justifies what we refer to to have a claim to objectivity would be to let ourselves be deceived by an accident of language. W hereas the contrast with ‘subjective’ is essential to ‘objective’, no such contrast belongs to the meaning of ‘object’. The two problems are, it is true, not completely independent of each other. But it is safer if we first separate them over- sharply and say that they overlap: thus it makes perfect sense to speak of merely subjective objects - e.g. objects of which we dream. On the other hand it is not merely the existence of objects that is objective (or subjective) but also properties of objects (predications) and connections of objects (e.g. sequences of events). T he lack of linguistic-logical reflection in the tradition also explains the failure - despite certain beginnings in K ant3 - to see that ‘objective’ is an adverb which qualifies veritative being and whose opposite is expressed by, for instance, such a term as ‘seeming’. Even when we speak of objects as ‘objective’ we mean their existence; and this means that in this case too the term ‘objective’ refers, not to the object, but to the proposition, the state of affairs.

Thus when Kant says that a relation to objects becomes possible through the synthesis of representations that is subject to a necessary rule, what he really means is not objects but something objective. His problem is not the one which concerns us here, viz. how we can refer to something (where this means: to something for which a singular term can stand). In Husserl’s case, on the o ther hand, the situation is more complicated. In so far as he defines ‘object’ as the ‘subject of possible true predications’,4 what he means by ‘object’ corresponds exactly to what we are here concerned with, i.e. something for which certain lin­guistic expressions (which can be supplem ented by predicates) can stand. But Husserl does not base this definition, which appeals to the function of singular terms, on a language-analytical conception of the object-relation. For he holds the view that the relation to objects, for which his definition serves merely to provide a mark of identification, is grounded in intentionality; and this I have in terpreted as represen­tation in the metaphorically extended sense. If I can call Husserl a rep­resentative of the traditional conception of what is meant by ‘objects’, then I must be able to assume that when he speaks of ‘objects’ he actually means objects and not, like Kant, something else; and the above definition is a clear indication of this. Now when Husserl speaks of a synthesis of modes of presentation he does not mean objects in general but specifically spatial objects. However, if one looks more closely, one sees that the problem which he deals with under the heading ‘synthesis of modes of presentation’ is one which concerns not the spatial object

The function of singular terms 287

as such (as the subject of possible true predications) but the various predicative determ inations of this object: every property of such an object - its form , its colour, its o ther sensible qualities - is constituted in a rule-governed synthesis of adum brations. If, for example, we describe this desk as brown, this description does not stand for a simple sensation-quality, but for a manifold of such qualities, which vary according to a rule, depending on the illum ination and the position of the percipient. Now this problem is on a level with the Kantian problem which concerns not the object qua object but the objectivity of properties of objects: the specifically objective colour-quality, for example, is con­trasted with its m anifold subjective appearance-m odes. However, Hus­serl starts the problem at a m ore basic level than Kant: the properties of im agined, or dream t of, spatial objects, for example, are also objec­tive unities which are distinct from subjective modes of presentation and are constituted in the rule-governed sequence of the latter. For Husserl the objective in K ant’s narrow er sense o f the word is simply a rule of the same kind though on a higher-level.5 T hat the modes of presentation dealt with by Husserl are not modes o f presentation of the object as such but only of its predicative determ inations is obscured by the fact that he speaks indiscriminately of things and properties of things. T hat he does so we can again explain by reference to his fun­dam ental, traditionalist position: by construing the object-relation as representation he reduces the object as it is given in perception to that in the object which is given representationally in the non-m etaphorical sense. And although Husserl occasionally distinguishes between the object as such and its predicative determ inations, the object as such he merely calls ‘the pure X in abstraction from all predicates’.6 In other words, all that Husserl can say about what the singular term as such stands for is that it is a pure X which underlies the predicative deter­minations. T he manifold modes of presentation which Husserl has in mind are thus not such as concern the relation of the singular term to the object. O r putting it ano ther way: they are not modes of presenta­tion to which a plurality o f singular term s (descriptions) which some­how all designate the same object correspond. It is not surprising that in the enquiry into the constitution of the object-relation Husserl was not orientated towards these modes o f presentation. For although he defined objects as the subjects of possible predications, thus as what a singular term stands for, he did not look to an analysis of the employ­m ent of singular term s for the explanation of the object-relation. Only in Investigation I of his Logical Investigations where, following Frege, Husserl explicitly goes into the distinction between the m eaning and

Analysis of the predicative sentence 288

the object of ‘names’ (by which like Frege he understands descriptions as well as proper names), does he encounter those different modes of presentation of an object to which different singular term s correspond. By contrast where he investigates the object-relation as such, and the constitution of spatial objects, he is not orientated towards the employ­ment of linguistic expressions at all. Frege, on the other hand, left open the question of how the different modes of presentation of the object which correspond to its descriptions are to be understood. Even less did he attem pt to investigate the question of what it is for a linguistic expression to ‘stand fo r’ an object, how we refer to objects by means of such expressions and what follows from this for the notion of objects as such.7

T hrough his enquiry into the meaning of singular terms Frege p re­pared a new way of posing the problem. But he was not interested in the philosophical question of how such a thing as an object-relation is constituted in the em ployment of such expressions; he dismissed it as psychological and epistemological. Mill and Husserl, by contrast, had a philosophical concept for the object-relation, namely, the concept of representation, a concept which Frege rejected without being able to put another in its place. O f course Mill, like Frege, and also the majority of analytical philosophers, simply took for granted the notion of objects. Husserl, on the other hand, made an attem pt analytically to elucidate the object-relation and with it the meaning of the term ‘object’. However, he did so on a pre-linguistic basis, in terms of the concept of representation. T hus'although Husserl does not have a sim­ple object-concept like Mill, and although he did not simply presuppose the existence of an object-relation and a universe of objects, his approach is just as traditional as that of Mill.

As we now turn to the question of the actual mode o f employment of singular terms I would like once m ore (cf. p. 271 f) to rem ind you that this question does not, as such, imply any prejudice in favour of a lan- guage-analytical conception of the object-relation. T hat these expres­sions stand for objects is not a m atter of controversy. An explanation of how one establishes for which object such an expression stands is nec­essary in any case, regardless of what position one holds. It is under­standable that in explaining the object-relation the traditional philoso­pher did not take the function o f the sign as his starting-point. But he can have no objection to such a precedure for however the object-rela­tion is finally in terpreted it must be possible to explain what it is for signs to stand for objects in terms of this interpretation. It may still turn out that the explanation of singular terms requires a non-linguistic

The function of singular terms 289

object-relation. Only if it should tu rn out that the object-relation which the explanation of singular terms presupposes cannot be understood independently of a specific use of signs would we have achieved a spe­cifically language-analytical object-theory.

If we understand the question of which object an expression stands for in terms of the use of that expression (and this means: its use by someone) then it becomes the question of which object someone means with the expression (above p. 282). It is of course possible for someone in a given case not to mean with a singular term the object for which it stands, e.g. if he mistakenly believes that the expression stands for an object other than that for which it in fact stands. But since there can be no sign-object relation independently of the use of signs by persons this discrepancy can only mean that in a particular case someone means with the expression a different object from that which is (or was) nor­mally (or originally) meant with it. This is all that can be m eant by saying that an expression really stands for a particular object. So the question of which object a singular term stands for can only be understood as the question of which object is meant with it, if not by an individual then by a linguistic community.

Now this question - ‘who or what is meant with “a”?’ — clearly belongs to the context of whole sentences. We can see this if, in addition to the question about mode of employment, I also apply to singular terms the other question I applied to predicates, viz. the question concerning their function. The question of function arises as soon as the sign-rela­tion is brought into connection with an employment of the sign by per­sons; and that means: with an action of persons. Every action is deter­mined by an intention and if the use of a thing - in this case the use of a sign - belongs to an action then the act-intention determ ines what we call the function of the thing (what it is used for). When I distinguished the function of predicates from that of singular terms I characterized the function of singular terms in the same way as it is characterized by the traditional theory: it is the function of a singular term to stand for an object. But it is precisely when we reflect on the aspect of function that we see how much in need of explanation this characterization itself is. For it remains unclear what is achieved by using something to stand for an object, what the standard intention is that determines an action which consists in the employment of something with this function.

T he traditional theory gives no answer at all to this question. But we should also be clear that the answer given by a pragmatist—behaviourist tradition, viz. that the function of the sign consists in its representing (vertreten) the object for which it stands in the sense that it evokes in

Analysis of the predicative sentence 290

the hearer the same (or a similar) response as that which the object itself would evoke8 is incorrect. A sign-theory of this kind is orientated towards so-called natural signs, i.e. symptoms (cf. p. 181) which are defined in the following way: a state of affairs functions as a symptom of another state of affairs if the belief that the first obtains (e.g. that it is now lightning) leads one to believe that the second obtains or is about to obtain (e.g. that it will thunder). One responds then to the perception of the first as one would respond to the second. One can describe this symptom-relation by saying that one state of affairs represents the other in a specific fashion; and one can in terpret this relation pragm at­ically in the sense described. We saw earlier (Lecture 13) that we can perhaps understand quasi-predicates in this way (which would then be functioning as signals) but not whole sentences. That one could u n d er­stand singular terms in this way is ruled out in advance by the fact that the objects to which symptoms refer are not concrete objects but states of affairs. Also from the point of view of the pragm atist-behaviourist interpretation it is clear that the only objects in question are states of affairs. One cannot respond to a concrete object. What one responds to is the presence of this object; and the presence of an object (that it is now here) is a state of aff airs. Those theorists who interpreted a sign’s stand­ing for an object as representation (Vertretung), and sought to interpret the latter pragmatically, overlooked the sharp distinction between con­crete objects and states of affairs.

Only when we simultaneously keep in mind these two things, viz. that the traditional description of the sign-object relation as ‘standing for’ tells us nothing, and that the behaviourist-pragm atist representation- theory (Stellvertretertheorie) is inapplicable to it, will we become aware of just how little we understand this seemingly so obvious sign-relation. Just as the traditional psychological representation-theory, according to which the sign represents (vertreten) a representation (Vorstellung), so the behaviourist-pragm atist representation-theory presupposes that what the sign merely represents could also be given to us independently of the signs. This applies both to representations (Vorstellungen) and to states of affairs. But what about objects? How can we refer to them at all?

It thus begins to emerge that we will only understand what is meant by ‘objects’, and what it is to refer to objects, by analysing the corre­sponding sign-relation. But how, you will ask, is the relation of the sign to the object to be understood if it can in no way be construed as rep­resentation (,Stellvertretung)? Doesn’t something like a representation- relation belong to the essence of any sign-relation? Certainly not. For

The function of singular terms 291

we have already seen in the case of predicates that their function - that of characterizing - has nothing to do with a representation-relation.

T he singular term is supposed to have the function of standing for a certain object. But if we view the singular term in isolation the signifi­cance of this function remains unintelligible. T he singular term ’s func­tion of standing for something thus seems to be an essentially non-inde­pendent function. This would mean: one needs signs with this function so that o ther signs can fulfil their function. T he act which consists in the employment of such an expression seems to have no independent intention and hence not to be an independent act. This is shown by the fact that someone who utters a singular term - a name for instance - by itself, has not done anything at the level of significant speech. He has of course done something at the level of utterance: he has uttered this sign. But with what intention? If someone begins ‘T he so-and-so’ and does not continue we will ask: ‘Well, what about it? W hat do you want to say about it?’ Someone who has merely u ttered a nam e, then, has not yet said anything. He has only said something if he supplem ents the nam e so as to form a complete sentence. A nd at the level of assertoric sentences (to which for the time being the whole discussion is confined) this means: if he supplements the name with a predicate. (The use of names in the vocative constitutes an apparen t counter-example. How­ever this use is always more than a m ere naming. When we use a name in the vocative by itself it does not just stand for the object; it does not function as a singular term but contains a dem and to the person named to answer or to come.)

T hat the function of the singular term essentially needs supplem en­tation becomes clear if we characterize the function of standing for an object as that of indicating which object is meant. If an action consists in indicating which object is m eant then it essentially refers {verweisen) to a complementary action which says something about a particular object and which therefore requires an indication of which object this is. At the most elementary level such an act which says som ething about a particular object is the characteriz.ation-act perform ed by means of a predicate. A predicate is as we have seen a classification-expression the use of which is such that one cannot tell simply from the situation in which it is used what the expression relates to. It therefore needs a supplem entary expression by means of which one indicates what it is (which object) - and that means: which of all - that is being classified by the predicate.

In these reflections as well as using the form ulation ‘which object is m eant’ I have also said simply ‘what (which) is m eant’; and finally have

Analysis of the predicative sentence 292

also used the locution ‘which of all’. W hat do these differences imply? As regards the transition from ‘which object' to ‘which’ this step is clearly unproblem atic. T he interrogative or relative pronouns ‘what’ or ‘which’ are an even better guide for our enquiry than the word ‘object’, for only they ensure that the enquiry is carried out at the requisite level of form al generality. T he expression ‘what’ or ‘which’ can always be replaced by ‘which object’. B ut by using only the expression ‘what’ or ‘which’ we make sure that the purely form al notion of something is not given an extra m aterial connotation by the addition of the substantive ‘object’. We thus link up again with my original introduction of the term ‘object’ in Lecture 3. T here I started out from the statement: everything and anything is an object. This implies tha t we cannot achieve the full extension which the term ‘object’ is in tended to have by saying: an object is everything which . . . and then supplem enting this by a predi­cate. For by so doing we would exclude o ther things which . . . , thus other things that are also something, from the extension of the term ‘object’. T he expression ‘what’, understood as a relative pronoun, already implies that one is speaking of objects. Since, therefore, the term ‘object’ is not delimitable by a predicate, we must conclude that the word ‘object’ itself, in so far as it functions grammatically as a pred­icate is a pseudo-predicate, precisely because it does not serve to dem ar­cate a class of objects from others (and it was this function by reference to which predicates were semantically defined). But this means that for that which is here in question the expression ‘object’ is as such inappro­priate. It is merely that predicate that comes closest to what is m eant by words which are never predicates because they are essentially words that supplem ent predicates, viz. singular terms and the corresponding pronouns: ‘which’ (‘what’), ‘som ething’, ‘this’. This is the reason why I said in the last lecture that we cannot clarify our prelim inary conception of what we mean by ‘objects’ by investigating the use of the word ‘object’. T he words of natural language towards which we have to ori­entate ourselves are ra ther the pronouns ju st referred to and singular terms. You may think that this is to smuggle in a language-analytical assumption. But this is not so. T he philosophical tradition was orien­tated towards the equation ens —unum —res =aliquid, an equation which is not unproblem atic bu t which we may take seriously at least in connection with the word aliquid (‘som ething’). Equally we could start, as we did at the end of the last lecture, from the traditional admission that every singular term stands for an object, i.e. for something.

T hese considerations make it appear probable that the explanation of what it is for a singular term to stand for an object - for something

The function of singular terms 293

- will essentially involve an explanation of how the use of singular terms is connected with the use of the corresponding (definite and indefinite) pronouns. The fu rther characterization of the function of singular terms that I have given, viz. that by means of a singular term one indi­cates which of all is meant, leads directly to this connection.

But was I justified in supplem enting the characterization of the func­tion of the singular term in this way, i.e. by saying not just ‘which’ but ‘which of all’? It seems to me that this supplementation is already implicitly contained in the question ‘which is meant?’, and that it is only when we make this supplem entation explicit that the meaning of this question, and hence that of the function of the singular term, becomes clear. For the need to indicate which object is meant, thus which object it is that is classified by a classification-expression, only arises when a plurality of objects is presupposed. T he function of the singular term is to indicate which of all objects that could come into question is meant. T hus it is the function of the singular term to pick out one thing from a plurality as what is meant - and this means: as that to which the pred­icate is supposed to apply. I shall call this function of picking out one thing from the presupposed plurality specification.9 We can now describe what happens in a predicative assertion as follows: by means of the predicate that which is specified by the singular term is characterized.

O f course, this is not yet an answer. W hat this definition provides is merely a description of the function of singular terms, which, in con­trast to the not incorrect, but empty, notion of ‘standing for’, says what these expressions really achieve. And now the question to answer will be: how do they achieve it? Establishing the function of these signs has given a definite direction to the enquiry into their mode of employ­ment: the explanation of their mode of employment must show what the employment-rules of these signs are which make it possible for them to specify an object (pick out which of a presupposed plurality is meant). T he broadening of horizon which the question has thereby acquired vis-ä-vis merely speaking o f ‘standing for’ may not be imme­diately clear. It consists in this: whereas it seemed plausible to construe ‘standing for’ as a relation which holds merely between the sign and its object, once the function of the singular term is understood as specifi­cation a relation of the singular term to all objects (of a domain) is p re­supposed. This is implied in the form ulation ‘indicate (pick out) which of all’. T he relation of the singular term to the object for which it stands is thus mediated by a relation to all objects of the presupposed plurality; and this shows that in fact it is not a relation at all. A sign has this ‘relation’ to an object in so far as someone who uses it can refer by

means of it to this object. And this reference - as specification - presup­poses that the person who uses such an expression, like the person who understands it, has the presupposed plurality somehow present in his consciousness. If this were not so then there would be no need to indi­cate which of all is meant. T he dem and for such an indication presup­poses that it is initially open which of all is meant; and that means: that it could be any one of all. Thus if to mean an object with a sign is to m ean an individual object, and if this means: to indicate which of all it is, then an object-relation implies a simultaneous relation to a multiplic­ity of objects. Thus we will not be able to explain the function of sin­gular terms without at the same time explaining this multiplicity-con- sciousness; and the question will arise whether this too is not constituted in a certain employment o f linguistic expressions. If then the function of a singular term consists in indicating which of all is meant, then the understanding of the word ‘all’ is as basic as the employm ent of singular terms.

But at the present level of reflection (where we merely have an abstract description of the function of singular terms and are not yet clear how they are able to perform this function) there is still more that m ust be adm itted to be essentially connected with the use of singular terms. We have a plurality; and from this one thing is to be singled out as what is meant. However this is concretely achieved, the idea of such a specification presupposes that we are able to distinguish one thing from others, and not just in so far as other predicates belong to it - if that were all we could never be sure that there is not also another thing to which exactly the same predicates belong and exactly the same p red­icates do not belong - but in so far as it is ano ther object. It is presup­posed that every thing can be distinguished from every o ther thing as an individual. To distinguish one thing from others, however, means to establish that the one is different from the others; and this means - I am thereby merely saying the same thing in d ifferen t words - that it is not the same as the others. Thus the use of singular terms also presup­poses an understanding o f the expression ‘is the same as’ or ‘is identical with’ (‘ = ’). I f it is essential to what we mean by an object that it is an individual (or putting it in terms of the function of singular terms: if the specification of which is m eant implies that it is specified which individual is meant), then essential to the understanding of what is m eant by ‘objects’, or to the understanding of singular terms, is an understanding of identity and non-identity and the possession of crite­ria of identity.

T h e possibility of being able to distinguish objects from one another

Analysis of the predicative sentence 294

The function of singular terms 295

is the condition of the possibility of our being able to count them. Objects are essentially countable. T he understanding o f ‘all’ and ‘some’, the understand ing of ‘is identical with’ and ‘is not identical with’ and the ability to count are on all fours with each o ther and with the possi­bility of using singular terms.

These connections are completely general and hold for all singular term s w hether they refer to concrete or to abstract objects. For all that has so far been explained followed completely generally from the func­tion which singular term s have as expressions tha t supplem ent p red i­cates; it was not yet assum ed that the objects that are specified by means of singular term s are perceptible objects. Thus even speaking of individ­uals does not as.such imply that one is dealing with concrete perceptible objects any m ore than does speaking of countability. The term ‘individ­ual’ is to be understood relative to the term ‘universal’ju st as singular term s are com plem entary to predicates. T here are classification-expres­sions whose use does not involve the com plem entary use of singular term s. These are quasi-predicates and as their use is situation-relative those singular term s which supplem ent classification-expressions which are analogous as regards content to quasi-predicates (and this means: which are explained analogously to quasi-predicates, as we saw in the lecture before last) designate only objects that are specified by reference to situations and are thus spatio-tem poral objects. Thus quasi-predi- cates only have analogues am ong predicates which classify concrete objects; they have no analogues am ong higher-order predicates, i.e. those which classify abstract objects. B ut clearly there are such higher- o rd er predicates; and this means that there are also h igher-order sin­gular terms. No doubt one will be able to say that reference to abstract objects presupposes reference to concrete objects; but at p resen t we are only concerned with the latter. On the o ther hand there have already been indications that reference to at least one species of abstract object- states o f affairs - is as basic as reference to concrete objects (p. 224). However this may be, we can and m ust from the outset distinguish those aspects o f the use of singular term s which distinguish them as singular term s as such - as expressions which have the function of spec­ifying wThich o f all is m eant - from those aspects which specially distin­guish those singular term s which specify concrete objects. Now if (a) the distinction between concrete and abstract objects is one which concerns them as objects, and if (b) specificatory reference to objects by means of singular terms is essential to speaking of objects, then we must also expect that what distinguishes that species of singular term that speci­fies concrete objects within the genus of singular terms is precisely the

Analysis of the predicative sentence 296

particular form which the general function of specifying takes in their case; and in that case concrete objects themselves would be distin­guished from abstract objects by the m anner in which they are speci­fied. Having established the general function of those expressions which by supplem enting predicates distinguish them from quasi-pred­icates we can now pursue our enquiry into their mode of employment by asking how these expressions specify the objects for which they stand.

L E C T U R E 22

Russell and Strawson

To tackle the question of what it means to speak of ‘objects’ and how a reference to objects is possible by asking how we can refer to objects by means of signs is already to adopt a language-analytical approach. However we have seen (p. 288) that such an approach still leaves it open whether the answer to this question itself results in a specifically lan­guage-analytical position. This would consist in holding that the re fe r­ence to objects to which the explanation of such signs points cannot be understood independently of the use of just such signs.

As with all signs the question of the mode of employment of these signs can only be tackled by asking how they can be explained. And in this case that means: how it can be explained, or established, for which object the expression stands (p. 270). To prepare for this question I enquired in the last lecture into the function of these expressions. What em erged there regarding the purpose for which singular terms are used and what it means to speak of ‘objects’ holds generally for all sin­gular terms and all objects. We now know what in general is being asked when it is asked how it can be established for which object a singular term stands: one is asking which object is specified by the singular term , where ‘specify’ means: to pick out what is meant from a presupposed plurality.

If we now ask, not about the function, but about the explanation, or m ode of employment of these expressions then precisely because of what we have seen regarding the function of these expressions the question can no longer be posed in a formal, general way. Rather a specific type of object, a specific plurality (that of perceptible objects or states of affairs or attributes, etc.) is presupposed and one asks how it is possible to pick out one object as the one meant from among all objects o f this type. Thus the question left open in the last lecture, viz. how a singular term can specify an object, must be put separately for each type

of object (and this is because the different types of object - though I have not shown this - are distinguished precisely by the m anner in which they are specified). And this question is identical with the ques­tion about the m ode of employment, or the explanation, of the type of singular term concerned.

T hus by appealing to the function of expressions a new and specifi­cally language-analytical approach is achieved, both for the explanation of what it means to speak of ‘objects’ and for the distinction of objects into what Husserl called ‘regions of objects’. In essence this approach derives from P. F. Strawson who first introduced it in his paper ‘On R eferring’ (1950) in a confrontation with Russell’s Theory of Descrip­tions and subsequently developed it in his book Individuals (1958). Sin­gular terms, according to Strawson, have the following function: some­one who uses them can by means of them refer to an object. A nd this referring consists in the singular term ‘identifying’ the object with which the rem aining part of the sentence has to do .1 This notion of ‘identifying’ roughly corresponds to what I have called ‘specifying’; though, as we shall see, certain additional distinctions will need to be drawn. Though his views have not rem ained undisputed Strawson is the standard au thor for this problem. It will therefore be appropriate in enquiring into the mode of employment of concrete singular terms to take his view of how objects can be identified by means of such expressions as our starting-point. And because Strawson arrived at his view via a critique of Russell’s Theory of Descriptions and because the latter theory is presupposed positively or negatively by the entire liter­ature on our them e I shall begin with a brief exposition of Russell’s theory.2

It is called the ‘theory o f definite descriptions’, and by ‘definite descriptions’ Russell understands expressions which have the form ‘the so-and-so’. Russell started out from the traditional view, that every sin­gular term stands for an object, as a premise that is not fu rther ques­tioned; he did not ask how this relation is to be understood. However, if it is a relation then it seemed impossible for the object for which an expression stands not to exist. But now singular terms, and in particular definite descriptions, are used which have no object. For example some­one using the sentence ‘T he King of France is bald’ in 1905, when Rus­sell wrote his paper, makes the claim that the singular term ‘the King of France’ stands for an object. In fact, however, this object does not exist. And if someone corrects him by saying ‘T h e King of France doesn’t exist’ then his use of the singular term does not even involve the claim that it stands for an object. Meinong, from whom Russell had

Analysis of the predicative sentence 298

Russell and, Strawson 299

taken his departure , had tried to overcome this difficulty by saying that the expression stands for an object but that an object, need not exist. Russell gave no conclusive argum ent to show that M einong’s view is impossible (this is som ething we will have to clarify). In any event he found M einong’s view unsatisfactory, and thus the question arose of how singular terms which it is possible do not stand for an object are to be understood.

Russell held fast to the traditional view that a p ro p er name stands for an object in a m ore genuine sense than a description in that it desig­nates the object ‘directly’3 and not, as a description does, as that to which a certain attribute uniquely belongs, e.g. the attribute of being the present King o f France. Russell concluded from this that in the case of expressions which can correctly be called p roper names it is not pos­sible for them not to have an object. W hat follows from this in regard to sentences such as ‘H om er did not exist’ we shall see in a moment. Anyway this view led Russell to think that the problem of how expres­sions which possibly do not stand for an object are to be understood reduces to the problem of how definite descriptions are to be u n d er­stood.

Russell’s answer to this question is: definite descriptions are not really singular terms at all. A sentence of the type ‘T he so-and-so is F’ only seems to be like a singular predicative sentence because of its gram m at­ical form . I f we logically analyse it it becomes clear that it is a general sentence and has the logical form ‘there is one, and only one, object which is so-and-so and this is F ’. W hat Russell m eans by ‘logical form ’ corresponds to what I call ‘semantic fo rm ’; and his phrase ‘if we logi­cally analyse the sentence’ we can replace, in the context of our enquiry, by ‘if we ask how the sentence is used’.

According to this view, then, the semantic structure of a sentence like ‘T he King of France is bald’ only becomes visible in the sentence ‘T here is one, and only one, object which is King of France and this is bald.5 In this form ulation there no longer appears a definite description. It no longer contains the definite article ‘the’; there only remains the predic­ative com ponent of the definite description (‘King o f France’). T h e case in which the object does not exist no longer presents any difficulties. We no longer have an expression that stands for an object, and the case where the object does not exist now reduces to the case where there is not one, o r only one, object o f that kind. Since in the sentence itself, as now form ulated, it is asserted that there is one, and only one, such object, if there is not one or only one such object then the sentence is simply false. Thus the sentence is not only false when the one so-and-

Analysis of the predicative sentence 300

so is n o t /7, but equally when there is no object, or more than one, that is so-and-so. And one can now easily understand the sentence ‘T he King of France does not exist.’ It is simply the negation of the sentence ‘T here is one, and only one, object that is King of France.’

Thus Russell’s solution to the problem consists in in terpreting the singular predicative statem ent as an existential statement, and the latter as a general statem ent. And this general statement he interprets in terms of the semantic structure which was brought to light by Frege and which we have already encountered (p. 247). A presupposition of Russell’s Theory of Descriptions is thus the m odern theory of general existential statements (i.e. such sentences as ‘Unicorns exist’), which was first developed by Frege, but had already been anticipated by K ant.4 ‘Existence’ is not only not a real predicate in such a sentence, as Kant taught; the word ‘exist’ does not have the semantic function of a p red ­icate (classification-expression) here at all. Rather the sentence is to be understood in precisely the same way as the sentence ‘T here are uni­corns.’ A nd this, as we have seen (p. 246), has the sense of a particular sentence. It says: ‘Some of all objects are unicorns.’ This view seems compelling, for it enables one to relate the understanding of the sen­tence to a truth-condition which corresponds to the way in which we verify such sentences. A ccording to the traditional view, in a sentence such as ‘Unicorns exist’ one would be saying of unicorns, thought of as possible, that they exist. But to establish whether unicorns exist we do not exam ine the possible unicorns with regard to whether the predicate of ‘existence’ applies to them; ra th er we examine the objects of the spa­tio-tem poral world with regard to whether the predicate ‘unicorn’ applies to some of them .

This result can be transferred directly from general existential sen­tences to so-called singular existential sentences, to sentences of the form ‘T h e so-and-so exists’, but with the following difference: such a sentence not only says that at least one object (‘some’) is so-and-so but also that at most one object is so-and-so, in other words that one and only one (a single) object is so-and-so. This conception also seems com­pelling in the case of the singular existential sentence, for only this con­ception explains the sentence’s semantic structure in a way that corre­sponds to how we verify such a sentence. Now this of course has the consequence that, contrary to appearances, so-called singular existen­tial statem ents are also not statem ents about individuals but are always general statements. This also provides us with the, in my view, decisive argum ent against M einong’s view. This view that we can refer to objects which are merely objects and may or may not exist and of which we can

Russell and Strawson 301

predicate existence or non-existence is refuted by the fact that it contra­dicts the way in which we establish the existence or non-existence of something.

Thus one does not have to accept Russell’s basic premise, viz. that it is impossible for the object for which a singular term stands not to exist, in order to find his Theory of Descriptions plausible. R ather its plausi­bility rests on the following considerations: (a) with sentences of the form ‘the so-and-so is F* it is always possible that there does not exist a single such object, and that, therefore, (b) the existence of this object must be implied by such a sentence, and (c) the existential statem ent in which this implication would be articulated must be in terpreted as a general statement. Indeed one could even say that the result of Russell’s theory contradicts the premise from which he started out. For it has now become clear that speaking of existence always presupposes that one is speaking of all objects and that, consequently, one cannot say of an individual object that it exists.5

But how then according to Russell are proper names, which it is claimed cannot possibly not stand for an object, to be understood? Rus­sell’s answer is that what we call p roper names in natural language are not p roper names at all. For it is perfectly possible in the case of such expressions - e.g. ‘H om er’ - to discover that there is no object for which it stands. T o explain this Russell applies to the proper names of natural language Frege’s theory of proper names. T he proper names of natural language do not have the direct relation they appear to have to the object. Rather they depend on descriptions; they are ‘really abbrevia­tions for descriptions’.6 For example, Romulus is the person ‘who did such-and-such things, who killed Remus, and founded Rome, and so on’.7 This Fregean view of p roper nam es,8 albeit in various form s,9 went unchallenged in analytical philosophy for a long time. It has only recently been called in question, by Kripke and D onnellan.10

If the proper names of natural language do not refer to objects in the direct way that Russell’s view and the traditional view would have led one to expect, m ust one not conclude that the idea of such a direct relation is simply a fiction? But for Russell taking such a step was out of the question, for his Theory of Descriptions led to the conclusion that what looks like a singular predicative statem ent is, if its subject-expres- sion is a description, really a general statement. And if the proper names of natural language are reducible to descriptions, then the state­ments whose subject-expression is such a proper name are also really general statements. So the result would be that there are no singular statements. But if we can always only say of all objects that one of them

Analysis of the predicative sentence 302

is such-and-such then it would seem that we cannot directly refer to an individual object at all. Quine, who has radicalized Russell’s theory, has accepted this conclusion: the basic statements are general statements and there are no singular statem ents.11 Such a conclusion appeared unacceptable to Russell, however, for the same reason that Strawson later rejected Q uine:12 general statements themselves - as we have already seen (p. 246) - refer (verweisen) via the specification of their truth-conditions to singular statements. T hough Russell did not pu t it like this, one cannot explain to someone the mode of employment of a general sentence without presupposing that he already knows the mode of employment of singular sentences.

So on Russell’s view there must be some genuine singular sentences. And for him this could only mean: sentences whose subject-expression stands for an object in the way that the philosophical tradition assumed that p roper names stand for objects. Now the required direct sign-rela­tion appears to presuppose a direct epistemological relation. For us to be able directly to assign a sign to an individual object, this object m ust be directly given to us - in perception. W here this is the case we can obviously use the deictic expression ‘this’. Now it is characteristic of ordinary objects that can be given to us in perception that they continue to exist outside the perceptual situation; if we still refer to them when they are outside the perceptual situation, we can no longer do so with the word ‘this’. It would seem plausible to say that if the object is given to us in perception, then not only can we refer to it with ‘this’, we can also assign to it an ordinary proper name; and that when it is no longer perceived, though we can no longer refer to it by means of the expres­sion ‘this’, we can still do so by means of the proper name. No doubt this is in a sense what actually happens. But is is not at all clear how it is possible. For if we have only introduced the proper nam e in the p er­ceptual situation, and hence as an equivalent of the expression ‘this’, then how do we know that we can still refer to it, the same object, by means of the proper name when the object is no longer present to us? As which object do we mean the object with the proper name if it is no longer this now present object? T o answer this question must we not have recourse to some descriptions or other? But in that case the proper name would again presuppose descriptions and is thus ruled out for Russell. All that remains then is a proper nam e whose referential func­tion ceases as soon as the object is no longer present, thus a proper name whose referential function reaches no further than that o f the deictic expression. In this way Russell arrives at the view that ‘this’ is the

Russell and Strawson 303

only logically proper nam e,13 i.e. the only one which in contrast to the merely ap p aren t p roper names of ordinary language functions sem an­tically in the way in which on Russell’s view p roper names must func­tion; it is an ambiguous proper name. O f course understood as a deictic expression the word ‘this’ would not be ambiguous. But if it is u n d e r­stood as a proper name then it m ust be term ed ambiguous (though this ambiguity could be rem oved by subscripts). Russell combined this view with an epistemological thesis according to which the only objects which can be directly given to us are sense-data. As these are ephem eral objects this means that the logically proper nam e not only designates an object in so far as it is p resent bu t designates this object as such.

It is im portan t that with Russell’s theory of logically p roper names in addition to descriptions and p roper names the third class of singular term s, viz. deictic pronouns, is included in the question of how an expression can stand for an object and indeed takes precedence over the other two classes. Russell, it is true, does not construe the word ‘this’ as a deictic expression but as an am biguous p ro p er name; the function of ‘this’ as a deictic expression rem ains unclarified. However, an im por­tant feature of deictic expressions, namely, that they refer to an object that is present in the perceptual situation, is retained.

As we shall see, Strawson in his positive conception latches onto this deictic com ponent of Russell’s conception, though of course with the difference that he recognizes the deictic expressions as such and does not in terp re t them as proper names. However, Strawson has not emphasized this positive link with Russell. As he does not discuss Rus­sell’s theory of logically proper names, but only his Theory of Descrip­tions, we should ask ourselves, before I move on to Strawson’s critique, how we should judge Russell’s theory of logically proper names. This question provides us with an opportunity to test the fruitfulness of what was achieved in the previous lecture regarding the function of singular term s (though at this stage of course only from the point of view of criticism). Can Russell’s logically p roper names do what they are sup­posed to do, namely, function as singular term s, if a singular term essentially specifies an object? We can immediately answer this question in the negative if we recall that ‘specify’ means ‘indicate which of all’. In the object-relation of a logically p roper nam e there is no relation, not even an implicit one, to a plurality of o ther objects. Consciousness is only related to the presen t sense-datum and there is no possibility, if these objects are not present, of holding on to them by using a sign, not even in the sense of being able to re fe r back to them as past sense-data.

Analysis of the predicative sentence 304

And this has the consequence that the expression ‘this’ because it does not distinguish something from others fulfils no function at all and could just as well be omitted.

It is instructive to com pare this language of logically proper names with that variant of the quasi-predicate language in which the word ‘this’ was used but in such a way that it was not replaceable by other expressions (above p. 263). In the case o f that language too I said that the word ‘this’ is not yet functioning as a singular term . Now, after the clarification of the function of singular terms, we can see m ore clearly why this must be so. T he predicative sentences which can be formed with Russell’s logically p roper names represent a language that is an exact counterpart, based on an introspectionist conception, to the extended quasi-predicate language based on a behaviouristic concep­tion. In both cases the speaker is tied to his perceptual situation, and the perspective of his speech does not extend beyond this situation. The word ‘this’ is idling, because it is not understood as a word that contrasts with other words which could take its place. (It has perhaps - as we saw in the case of quasi-predicate language - the minimal function of point­ing to a particular place in the situation; and something similar could be m aintained for Russell’s theory with respect to a place in perceptual space.) Russell believed that in the direct relation of the logically proper nam e to a perceptual datum he had found that relation to an individual that is characteristic of singular terms. But singular terms do not simply have a relation to an individual - every object has this to every other object to which it stands in some relation. Rather it is characteristic of singular terms that by means of them we refer to an individual as an individual by indicating which it is; and this they can only do by at the same time relating to all. T h ere is thus confirmed from another per­spective what I had already anticipated in connection with the distinc­tion between predicates and quasi-predicates: that it must be the func­tion of singular terms to make the employment of predicates independent of the perceptual situation. Even the deictic expression ‘this’ only functions as a singular term if it refers to what is present in such a way that the latter appears as what is meant in contrast to all others. How it can do this is a question we have still to investigate.

We can now see the peculiar dilemma to which Russell’s theory of singular terms leads. Because Russell started out from the traditional view that a singular term relates directly to an object he denied those expressions that are regarded as singular terms - definite descriptions and the proper names of ordinary language - the status of singular terms; for he correctly perceived that they imply a relation to every­

Russell and Strawson 305

thing. As such they cannot, in his opinion, relate to an individual. But those expressions from which Russell removed all relation to a plurality, precisely so that they can relate to individuals, can for this very reason not refer to individuals. It is true that the use of general sentences can only be explained if familiarity with the mode of employment of sin­gular sentences can be presupposed. On the other hand, however, a peculiar generality seems already to characterize the singular term and, hence, the singular sentence. This is, as we shall see, a dilemma which Strawson too failed to see through and which has rem ained unresolved to this day.

W hat form does Strawson’s critique of Russell’s Theory of Descrip­tions take? His article ‘On R eferring’ is divided into five parts. T he first three are devoted to criticism of Russell. In the last two parts Strawson indicates his own positive conception.

Part I of the article is supposed to describe the premises and the question from which Russell started. Strawson points out that Russell did not distinguish between the meaning of an expression and the object for which it stands. It would follow from this, Strawson claims, that a singular term which has no object also has no meaning; Russell developed his Theory of Descriptions in order to avoid this absurd con­sequence.

Now it is true that Russell did not distinguish between meaning and object.14 But it is unfruitful to reduce the basis of Russell’s problem to this easily refutable premise. Quine for example, although he accepted Russell’s theory and fu rther developed it, emphatically rejected this prem ise.15 Russell’s real problem was that of non-existent objects; and this problem remains even if one does not confuse the meaning of a singular term with its object. By presenting a weakness of Russell’s view that is not essential to his theory as though it were its foundation Straw­son has made criticism of it far too easy.

To refute Russell’s theory Strawson produces two argum ents in the second and third parts of his article which are presented as if they con­cerned the same problem but are in fact independent.

T he first argum ent is as follows. One might distinguish with regard to an expression such as ‘the King of France’ between the expression and the use of the expression. T he expression as such has a meaning, but does not stand for a specific object. T he expression only stands for a specific object if it is used in a specific situation - in this case, if it is used at a specific time. And it stands for a different object (whoever is King of France at the time) or for none, depending on the time at which it is used. A corresponding distinction is to be made in regard to the

Analysis of the predicative sentence 306

whole sentence. The sentence ‘T he King of France is bald’ has a m ean­ing, but in itself no truth-value. Only when it is used at a particular time can it have a truth-value. Once one distinguishes between the expres­sion and its use the confusion between m eaning and object which underlies Russell’s theory can, Strawson thinks, no longer arise.

Against this argum ent Russell has himself pointed out (in his reply ‘Mr. Strawson on Referring’) that the distinction that Strawson makes, which specifically concerns deictic expressions, and which he, Russell, in no way denies, has nothing to do with the real problem. One can replace the situation-reference implicit in the expression ‘the (present) King of France’ by an objective specification by saying for instance ‘the King of France in 1905’ and see immediately that the real problem, which concerns not a confusion of meaning and object but the talk of non-existent objects, is independent of the special problem of deictic expressions. It is true that with deictic expressions Russell’s problem only arises at the level of the use of the sentence but it arises at this level in just the same way as before.

O r perhaps not? This Strawson tries to show in the third section of his article. T he thesis of this section is that when someone uses a sen­tence like ‘T he so-and-so is F' he presupposes that one, and only one, object exists that is so-and-so but does not assert this. What is the d iffer­ence? Strawson replies that if someone thinks that the existential p re ­supposition a person makes in employing such a sentence is mistaken he will not say that the latter’s statem ent is false but ra ther that ‘the question of whether his statement is true or false’ simply does not arise.16 T he criticism of Russell then simply boils down to this: that when such an object does not exist the statem ent is not false but neither true nor false.

Russell reacted sharply to this criticism too - and rightly so. Firstly, what is at issue, he claims, cannot be whether one says one thing or the o ther in ordinary language; besides one can give examples which show that even in ordinary language one says in such a case that the state­m ent is false, ra ther than that the question of its truth or falsity does not arise.

This criticism must be made even sharper. Firstly, Strawson writes as though this second point in his criticism were connected with the first, whereas in fact it is independent of it. Secondly, the second point is as irrelevant to the real problem as the first point. Strawson seems not to notice that the alternative which he presents to Russell’s view already involves acceptance of the essentials o f Russell’s theory. T h e existential statement, w hether merely presupposed or implicitly asserted, is con­

Russell and Strawson 307

ceived by Strawson in exactly the same way as by Russell:17 consequently if the expression has no object the non-existence of the object cannot be understood by Strawson differently than by Russell. W hether the consequence of this for the original statem ent is that the existence is asserted or merely presupposed, is a nuance the relevance of which is not obvious.

Strawson’s appeal to actual usage in ordinary language, and Russell’s reply that this cannot be what is at issue, has led people to think that the controversy between Russell and Strawson can be characterized as follows: the form er is concerned with the construction of an ideal logi­cal language, the latter with the peculiarities of ou r actual ordinary lan­guage. This contrast, which in expositions of analytical philosophy is used virtually as a criterion for distinguishing the ‘ordinary language philosophy’ of W ittgenstein and the O xford School from those s email- deists, such as Russell, Q uine and C arnap, who are orientated more towards logic and belong to the tradition of Frege, is absurd when understood in this sense. T he appeal of W ittgenstein and others, including Strawson, to ordinary language was not aimed at bringing out the wealth of semantic nuance of ordinary language - an undertaking which, it could properly be objected, could be better carried out by the empirical science of linguistics.18 Philosophy, even philosophy of ordi­nary language, cannot be concerned with purely factual nuances of ordinary language; the merely factual was never the object of philoso­phy, only the possible. T he appeal to natural language in W ittgenstein and those who have understood him has had a m ore fundam ental sig­nificance. T he appeal to natural language does not involve opposition to the idea of an ideal language as such, but only to the idea of an ideal language built in a vacuum. W hat m ade necessary the recourse to nat­ural language was the realization that the ideal language rem ains itself unexplained, o r is only explained by means of an ordinary language m eta-language. It is motivated, not by an interest in factual nuances, but by methodological ideas, about the m eaning of an ultimate semantic explanation (ideas of the kind we have become acquainted with in con­nection with the fundam ental principle of analytical philosophy).

From our point of view it is a m atter of indifference whether the view that the existential statem ent is only presupposed in the use of the singular term is better attested for the natural languages with which we are familiar than the view that the existential statem ent is asserted. If both are possible, but the existential statem ent m ust at least be presup­posed, then Russell’s theory is vindicated.

O ne must therefore conclude that Strawson’s Russell-critique in ‘On

Analysis of the predicative sentence 308

Referring’ fails in every respect. Even Russell’s starting point is p re­sented in a way which fails to make clear what is really at stake in decid­ing the correctness or uncorrectness of Russell’s theory. And the two objections which Strawson then brings tu rn out to be both irrelevant. It is only in the concluding section of Strawson’s article, in which he sketches his own view, that it becomes clear what the real issue is. H ere he speaks of the function (task) of singular terms. If one is speaking about individuals one needs expressions by means of which one can uniquely refer to individuals. Expressions can only fulfil this function if they function in such a way that through them a speaker enables a hearer to ‘identify what is being talked about’. But this can only be achieved by a specific reference to the speech-situation (the context of utterance); to identify the object referred to, the object must ‘be in a certain relation to the speaker and to the context of utterance’.19

We shall have to see in the next lecture how Strawson elaborates this view in his book. But what we m ust now ask is whether by reference to Strawson’s positive view we can explain what the real difference is between him and Russell.

Strawson rightly regards Russell’s logically proper names as fictions. However, he only criticizes this aspect of Russell’s view in passing. He thinks he can dispose of it by pointing out that the conception of ‘this’ as an ambiguous p ro p er name again rests on the confusion of meaning and object. Strawson is, of course, right in thinking that the word ‘this’, as it is actually used in natural language, is not an ambiguous proper name but, as a deictic expression, has a unitary meaning and designates d ifferent objects depending on the situation in which it is used. But Russell did not claim to be using this word as it is used in natural lan­guage; nor can one oblige him to do so. T he test of his theory is not whether it corresponds to the ordinary-language use of words, but whether it can explain how we can refer to individuals. T he reason why it founders on this is not because Russell fails to see that despite the fact that a deictic expression stands for d ifferent objects (depending on the situation in which it is used) it has a unitary meaning. Rather it is because he fails to take into account another peculiarity of deictic expressions, viz. that the same object for which a deictic expression is used in the perceptual situation can be referred to outside the percep­tual situation by means of another deictic expression and then also by means of o ther singular terms. It is this substitutability of deictic expressions which makes it possible to refer to the same object when it is no longer perceived. Strawson also failed to go into this aspect of the use of deictic expressions which is crucial to the specification of objects, although he deserves credit for having raised the question of the func­

Russell and Strawson 309

tion of singular terms and, in particular, deictic expressions. It is, how­ever, philosophically irrelevant to reproach Russell for having failed to take account o f certain aspects of the use of deictic expressions. This cannot touch him, for he does not regard ‘this’ as a deictic expression at all. It can only touch him, and only become philosophically relevant, if the aspect o f the use o f deictic expressions in question can be shown to be a necessary condition o f the possibility of referring to individual objects.

Strawson wanted to conduct his critique of logically proper names in a more indirect fashion. He wanted to dem onstrate that it is not neces­sary to postulate such expressions by rehabilitating the singular terms of natural language which, in Russell’s Theory of Descriptions, are in terpreted away. This was to be achieved firstly (and negatively) by the critique of Russell’s theory and secondly (and positively) by the appeal to the function of singular terms. O ne must, he says, recognize the use of these terms, whose purpose it is to refer to individuals, as the ‘harm ­less necessary thing’20 that it is. But is it really so harmless? And has Strawson really succeeded in rehabilitating singular terms as such, if, as he believed, this requires a refutation of Russell’s theory? Does the neg­ative critical part of the article provide that critique which the positive part presupposes? Strawson holds fast to the same presupposition which Russell took as his starting-point, viz. that if one is to be able to refer to an individual with an expression that expression cannot imply a general statement. A refutation of Russell’s theory would have required Strawson to destroy its foundations rather than simply offer an alternative version in the shape of the thesis that the general state­m ent is not asserted but merely presupposed. If one only criticizes Rus­sell in the way Strawson does then precisely that aspect o f Russell’s the­ory to which Strawson took exception is preserved. Strawson himself admits even in regard to the deictic expression ‘this’ that its use presup­poses a general statement: if someone points somewhere and says ‘This so-and so . . .’ then it is either asserted or presupposed that in that place there is one, and only one, so-and-so.21 The function of singular terms particularly of deictic expressions emphasized by Strawson thus in no way contradicts Russell’s theory; ra ther it would even seem to presup­pose it. But in that case this function is harmless neither in the sense intended by Strawson - that it makes Russell’s theory dispensable - nor in the sense that it requires no explanation. We shall have to ask: how can singular statements be understood if they already presuppose a re f­erence to a totality and an existential statem ent that is a general state­ment?

L E C T U R E 23

W h at is ‘identification’?

At the end of the lecture before last we arrived at the result that the function of singular terms is to be understood as that of specifying. By ‘specifying’ was meant: indicating which of all objects it is that is classified by the o ther sentence-component. It has not yet been decided whether the meaning of our reference to objects is exhausted by the thus char­acterized object-relation that is made possible by the employment of a species of linguistic sign, nor whether we are to understand what is meant by ‘objects’ in terms of it. And in particular it rem ained still unexplained how these linguistic expressions - singular terms - must be employed if they are to be able to fulfil the function of specifying as thus described.

In order to prepare ourselves for this enquiry into the concrete form which the specificatory function of singular terms takes we shall begin by orientating ourselves towards the conception of Strawson (who was the first to tackle this problem). We saw in the last lecture that one cannot be satisfied with the way in which Strawson, in his early paper ‘On R eferring’, tries critically to distinguish himself from Russell. How­ever, his own positive conception (viz. that singular terms have a so- called identificatory function) is merely sketched in that paper. It is only worked out in his book Individuals (1958) and in his paper ‘Singular Term s and Predication’ (1961) in which he argues with Quine. So it is these writings towards which we must orientate ourselves in the hope of reaching, via Strawson’s talk of identification, a concrete understanding of what I have term ed the function of specification.

In the 1961 paper Strawson states that it belongs to the essence of a singular term that it is ‘used for the purpose of identifying the object’. And this identificatory function is said to consist in ‘bringing it about that the hearer knows which object it is, of all the objects within the hearer’s scope of knowledge or presum ption, that the other term ’ - the

What is ‘identification’? 311

predicate - ‘is being applied to’.1 This definition of what is m eant by ‘identification’ comes very close to the definition which I have given of ‘specification’; one m ight therefore think that they stand for the same concept. However, as we shall see, the m atter is more complicated. But it is worth noting that Strawson also takes account here o f the fact (which 1 have stressed but which he has otherwise neglected), that reference to an object implies a reference to all objects o f a domain.

T he only respect in which Strawson’s account differs from the one I gave is in the explicit mention of a hearer. In Individuals the problem of identification is, from the start, treated as one that concerns com­m unication between a speaker and a hearer. Thus it is tha t for Strawson the person who prim arily identifies an object is the hearer; he does so when he knows which object the speaker means. In the case of the speaker Strawson speaks only in a derivative sense of identification: that the speaker identifies an object means that the hearer identifies the object which he, the speaker, m eans.2

From this we can see why Strawson chose the term ‘identify’ for the function o f the singular terms in question: the hearer knows which object is m eant by the speaker if he knows that the object which the speaker means is identical with an object so-and-so that is accessible to him , the h earer.3 It might at first sight seem a particular virtue of Strawson’s conception that from the outset he tries to understand reference to objects in term s of the com m unication-situation. On the o ther hand we saw in the semantics o f whole assertoric sentences (in the debate with Grice) that, although linguistic expressions m ust be explained as they are used in in ter subjective communication, no explanation is acceptable that essentially presupposes that there are two separate speech-partners. For then the use of the expression in soliloquy could not follow the same rules. In my description of the function of singular terms, viz. that they serve to pick out, or indicate, which object is m eant, the use of these expressions is understood in terms o f the function they have in discourse; bu t this can ju st as well be soliloquy. T he exaggerated ten­dency that is to be observed today in some quarters towards a com m u­nication-orientated semantics at any price can only lead to a semantics that construes language as simply an instrum ent of communication. Such a semantics would, like traditional philosophy, again presuppose as pre- linguistic the very epistemological structures which require explanation; such structures are then not understood in terms of the speech-situation. T he apparently more radical communication-orientated approach (more radical because for such an approach the separation of speaker and hearer is essential) is in reality less radical, for it inevitably rem ains on

Analysis of the predicative sentence 312

the surface. T hus we saw that a purely com m unication-orientated con­ception of the use of assertoric sentences, such as that of Grice, has to presuppose judg m en t or belief, and hence all the logical structures of semantics, and that it therefore cannot explain them in term s of dis­course. In the same way a theory of singular term s according to which their function consists in this: that by their means one speech-partner communicates to another which object he is referring to, must presup­pose reference to objects as such as something that does not first need to be explained in terms of language and discourse. Such a theory of singular terms could therefore contribute nothing to the fundam ental enquiry into the object-relation and the concept of an object.

You might say: ‘So what. One does not refute a theory by pointing out that a particular end cannot be achieved by it; for it could be that this end is overdraw n.’ T h a t is correct. It can, however, be shown that Strawson’s explanation presupposes the very thing it is supposed to explain. T he hearer identifies the object referred to by the speaker if he knows that the object referred to by the speaker is identical with the so-and-so. The second component of this identity-statement is a singular term which, it is assumed in this explanation, the hearer understands. So the concept of identification, as thus understood, can contribute nothing to the explanation of singular terms. This weakness in Strawson’s conception has been drawn attention to particularly by B. A. O. Williams in a review of Strawson’s book.4 Consequently, the authors who have continued to work in the field of enquiry opened up by Strawson have no longer used the term ‘identification’ in the communication-orientated sense in which Strawson first introduced it.5 A nd Strawson too finds him self compelled to speak of ‘identifying’ in a sense other than that initially introduced. T he hearer, he says, must be able to identify the in tended object ‘for h im self’.6 And, one might well add, the speaker m ust also be able to identify the in tended object ‘for him self’. In a later place in his book Strawson writes: ‘For each of us can think identifyingly about particulars without talking about them .’7 It is clear that the term ‘identify’ is here being used in a completely different sense from that in which it was introduced at the beginning of the book. For whether the ‘thinking identifyingly’ of which Strawson here speaks is understood as soliloquy or as something pre-linguistic, it can on no account be conceived as an internal counterpart of the identification introduced by Strawson at the beginning of the book. For that identification, when articulated, had the form of an identity-statem ent the first com ponent of which was: ‘the object which the speaker means’. Trivially, such a reference has no place in soliloquy. So if one also speaks of identifying in soliloquy then

What is ‘identification ? 313

it would seem no longer to be a reference that is articulated in an identity- statement. In an identity-statement two singular terms are used. If the reference to an object which both speaker and hearer perform by means of one singular term is called ‘identification’, then ‘identification’ cannot mean ‘holding something to be identical with som ething’.

But what then does it mean? We find no real answer to the question in Strawson. In this lecture and the next I shall have to put your patience to a severe test as, building on the meagre beginnings we find in Strawson and others, I attem pt to reach a clear concept of identification which will tu rn out to be a pre-em inent special case of the specification which, in the lecture before last, I showed to be the general function of singular terms. I f you want to know, before letting yourself in for this undertak­ing, why you should be so patient, my reply would be that I believe that it is only by means of this narrow er concept of identification (which Strawson has not explained but hinted at) that the general function of specifying, and hence the meaning of reference to objects, can be understood. Strawson does not explain what he means by ‘identify’ in the non-com m unication-orientated sense. He simply paraphrases the word, mainly by the expression ‘to pick ou t’; and Searle and others have followed him in this.8

W hat is meant by the English ‘to pick ou t’? It is not only used for the physical act of picking out an individual object from a collection, e.g. a berry from a basket o f berries, but also for the merely designatory pick­ing out of an individual as the one intended, e.g. an officer standing before his company picks out an individual as the one who is to perform a certain task, by pointing to him. And it is this picking out in the sense of singling out that Strawson has in mind when he uses ‘to pick out’ as a paraphrase for ‘to identify’. This is also shown by the fact that on one occasion he paraphrases ‘to pick out’ itself with the expression ‘to single out’.9 It means roughly: ‘to bring something into prom inence as the individual that is m eant’.

So it now seems clear (1) that Strawson uses the word ‘identify’ in two completely d ifferent senses (2) that in its first sense - ‘speaker- hearer identification’10 - it presupposes in the case of both speaker and hearer identification in the second sense, and that, therefore, this is the basic sense, (3) that what is m eant by ‘identify’ in this second sense, paraphrased by ‘pick out’ and ‘single out’, appears roughly to correspond to what I have called ‘specification’: the act of indicating which of all it is that is classified by the supplem enting predicative expression.

Strawson himself considered using the expression ‘specify’ (which has also been used by Quine), but rejected it as too vague.11 It seems that

Analysis of the predicative sentence 314

he wished to exclude the following ambiguity. One could specify12 what is m eant simply by saying: it is that object to which such and such prop­erties belong. Clearly there are singular terms which specify the object for which they stand in this way, viz. definite descriptions. At any rate one can say this of all those definite descriptions which do not contain a deictic expression or a spatio-temporal specification. A deictic expres­sion serves subjectively to locate something, a spatio-temporal specifi­cation objectively to locate something. We shall have to deal with this in detail later on. Descriptions which do not locate in either of these ways we can call ‘purely descriptive definite descriptions’ (rein deskriptive Kennzeichnungen). One may doubt whether one can distinguish an individual object from all others by the m ere accumulation of merely descriptive expressions, i.e. predicates. However this possibility is rela­tively clear in the case of ‘ordinal’ properties, i.e. properties with respect to which objects can be ordered in a series.13 Examples of this sort o f singular term would be ‘the highest m ountain’, ‘the second highest m ountain’. An object can fail to be distinguished from all others, and hence specified, by such descriptions, because there are two or more objects to which the property in question belongs to the same degree, e.g. if there are two m ountains of exactly the same height. But the fact that the specification fails does not mean that an expression is not used in this way (with this intention). Even the use of a demonstrative expres­sion - ‘this so-and-so’ - which is regarded by many authors, and also by Strawson, as the most unam biguous form of identification can fail if it turns out that there is no so-and-so, or more than one so-and-so, at the place which is pointed to.

So it cannot be because it can sometimes fail that Strawson does not recognize this form of specification as identification; he must have o ther reasons. If therefore Strawson paraphrases his second use of the word ‘identification’ so that it appears to correspond to what I have term ed ‘specification’ and if nevertheless he does not count as identification cases which clearly fulfil the function of specification then we must assume that this talk of ‘identification’ contains yet another hidden ambiguity. W hat Strawson intends by ‘identify’ in the sense of ‘pick ou t’ m ust be a special, a pre-em inent case of the general function of indicat- ing-which-of-all. T he problem is clearly connected with Strawson’s opposition to Russell’s Theory of Descriptions which I discussed in the last lecture. For it is obviously the possibility ju s t m entioned, of purely descriptive specification, which Russell’s Theory of Descriptions fits. Whereas in ‘On R eferring’ Strawson completely rejected this theory he now seems to accept it at least for one type of description; so now the

What is ‘identification’? 315

objection would only be that Russell has im properly generalized what is true of this type and in so doing has overlooked that use of singular terms in particular by means of which an object is ‘identified’.

We should now be in a position to, on the one hand, understand better what it was in Russell that Strawson took exception to and, on the o ther hand, obtain an indication from this o f what he means by ‘identification’. A lthough Russell of course does not himself speak of ‘specification’ and although Strawson too does not call it ‘specification’, one can nevertheless say: there is a kind of specification by means of a singular term in which it is said: ‘there is an object that is so-and-so’ or (if we render implicit the explicit assertion of existence which Strawson m ade much of, but which is in itself indifferent) ‘the one which is so- and-so’.

We can now easily see what it is that Strawson still misses in such an expression and why he will still not grant it the function of identification. I f we supplem ent such an expression with a predicate to form a whole sentence and say: ‘the one which is so-and-so is F’, then we have not indicated, one m ight say, o f which object it is asserted that it is F; we have only said that the only thing that is so-and-so, whichever this may be, is F. O f course, this explanation is still unclear, for to som eone who were to argue in this way one could reply: ‘But we have specified which it is, precisely by saying “the only one that is so-and so”.’ However, the o ther person could reply that in this way the object is only indirectly designated as the sole bearer of a property; an object is only identified if it is itself directly designated. This contrasting of an indirect and a direct reference to an object is o f course still unclear. One can, however, clarify this contrast by means of an example. I f som eone says som ething about the highest m ountain on earth it is not yet clear of which object, of which m ountain, he is speaking. O ne can ask ‘and which m ountain is the highest then?’ Only when this question had been answered would we have ‘identified’ a specific mountain. It is still unclear how this further specification of which it is is to be achieved; but we can now say that however it is achieved it is only this specification that should be called ‘identification’. Such a definition of the word ‘identify’ occurs in Searle. He says: ‘By “identify” here I mean that there should no longer be any doubt or ambiguity about what exactly is being talked about. At the lowest level, questions like “who?”, “what?”, or “which one?” are answ ered.’14

You will still find this unclear and ask: ‘what is this distinction of levels in regard to the question “which of all is m eant?” and what is it that distinguishes the lowest level?’ So far as I can see the literature has up till now provided no means for systematically dealing with this question.

Analysis of the predicative sentence 316

Now in the fundam ental principle of analytical philosophy we have a natural clue to guide us in the enquiry into the semantics of singular terms. But though I referred to this at the beginning we have not so far put it to use. The question o f what it is for a singular term to stand for an object, we saw then (p. 270), can only be answered by asking how it is established for which object a singular term stands. In the meantime we have seen that the function of the singular term is to specify an object, hence that ‘standing for’ is to be understood as specifying. Thus the question we have to ask is obviously: how is it established which object a singular term specifies? And if there are different kinds of singular term and different levels of specification then we can expect that these can be distinguished by reference to the different ways in which it is established which object the expression specifies. Moreover, this should also explain how that narrow er concept of specification, which Strawson calls ‘identification’, and whose meaning he did not explain but merely hinted at by means o f paraphrases, is to be understood.

However, in preparation for such a systematic analysis we should first make clear how far one can get with Strawson. So far I have only dis­cussed the question of what in general Strawson means by ‘identification’. We must now try to form a clearer picture of the possibilities of identi­fication of perceptible objects by singular terms which he envisages. In this way we can prepare the ground for the systematic clarification of what distinguishes the narrower concept of identification from the more comprehensive concept o f specification.

Strawson distinguishes two sorts of identification: direct (dem onstra­tive) and indirect (non-demonstrative). As one would expect he sees identification by means of the deictic expression ‘this’ as the simplest case of identification. T he question ‘which is the highest m ountain?’ can, so it would seem, be definitively answered by taking the questioner to a particular m ountain and saying: ‘it is this one’. This ‘demonstrative identification’ consists in this, that ‘the hearer can sensibly discriminate the individual object referred to’ and thus ‘pick it ou t’. He is thereby able ‘directly to locate’ the object referred to.15 But an object can also be identified indirectly, in a non-dem onstrative way. This can only be done by relating it to the situation of demonstrative identification, thus to the perceptual situation of speaker and hearer. Such an identification is achieved therefore by means o f ‘a description which relates the p a r­ticular in question uniquely to ano ther particular which can be dem on­stratively identified’.16 Now there is a comprehensive system of unique relations to which all spatio-tem poral objects belong, namely the system of spatio-tem poral relations. Every perceptible object has a place within

What is 'identification ? 317

this system and thus stands in a unique spatio-temporal relation to every o ther perceptible object. As the system of spatio-temporal relations is unique, and comprehensive, every perceptible object can be identified, if not by another unique relation at any rate by its spatio-temporal rela­tion to a directly identifiable object.17

If we measure these two possible ways of identifying concrete objects which Strawson envisages against the idea - an idea which of course is not form ulated by Strawson himself - that a specification by means of a singular term V shall only count as identification if it no longer gives rise to the fu rther question ‘and which object isx?’, then one must raise the following objections:

1. If we take someone who wants to know what the highest mountain is blindfold to Mt Everest, remove the blindfold and say to him ‘It is this’, it is quite likely that he will reply: ‘But which m ountain is it that I see here?’ Normally one would reply to such a question by giving the name of the mountain. But that the answer ‘the highest m ountain (or: the mountain which you see here) is Mt Everest’ is usually accepted as definitive is due simply to the fact that it is presupposed that one is clear which mountain it is that is so nam ed. By the question ‘which is it?’ we mean here ‘where is it? in what spatial relations does it stand to other objects, to o ther geographical data?’ It is only after a reply to a question o f this kind that the fu rther question ‘and which is the m ountain that is at such-and-such a place?’ is meaningless. Thus one can only agree with Strawson that demonstrative specification is a specification at the lowest level, an ‘identification’, to the extent that normally when we use the demonstrative pronoun ‘this’ or the corresponding spatio-temporal expressions ‘here’ and ‘now’ we presuppose that we can replace these expressions by other expressions which designate the objective spatio- tem poral position in which the object is situated, thus the position which belongs to the object not just subjectively, from the standpoint of the speaker, but the position which belongs to it relative to all o ther possible standpoints. But then it is not possible to conceive of the identification of perceptible objects in that way in two steps as Strawson does. Demon­strative identification if it is to be genuine identification itself presupposes spatio-temporal, non-demonstrative identification. What Strawson calls ‘direct location’ (above p. 316) presupposes the possibility o f an objective location. Strawson’s failure to see this results from his taking over the traditional assumption, which is also implicit in Russell’s theory of logi­cally proper names, that an expression’s standing for an object is to be understood as a direct relation. Like Russell, Strawson conceived this relation as a demonstrative relation in which something present in p er­

Analysis o f the predicative sentence 318

ception is simply pointed at. He failed to notice that to specify an indi­vidual means: to indicate which of all. This does not m ean that one must be able to give the relations of the object referred to to all other objects (this is not the case with, e.g., most types of abstract objects). It does however mean that through the ‘indication which’ the object referred to must be distinguishable from everything that is not identical with it. And in the specific case of the specification of perceptible objects this has the consequence that the indication of which object is m eant by means of a m ere dem onstrative does not suffice for specification. One must also be able to distinguish the object meant from all others. Why this can only be done by indicating in which spatio-temporal relation it stands to all others is something we have still to investigate.

2. We have seen that Strawson distinguishes two sorts of non­demonstrative identification. In both cases the object m eant is identified by means of a unique relation to something demonstratively identified, in the first sort by means of its spatio-temporal relation, in the second by means of another relation. We have now seen that demonstrative identification itself presupposes the first kind of non-dem onstrative identification. T he presupposition is of course a reciprocal one. All locating of something by indicating its spatio-temporal relations to other objects is only an identification if the spatio-tem poral relations of these other objects to the perceptual situation are known. I shall come back to this point. But how is it with the o ther kind of non-dem onstrative identification - that by means of other than spatio-temporal relations? One cannot regard this kind of identification as restricted to those cases in which something stands directly in a unique relation to something perceived. All specification by means of a unique relation to something already identified clearly belongs to the same class; hence the relation to the perceptual situation can be as indirect as you please. Examples are: ‘the m urderer of this m an’, ‘the wife of Mr M aier’, ‘the President of France’, ‘the first man to climb Everest’. Now it is easy to see that the specification of an object by means of such a unique relation to something already identified in no way differs from specification by means of a merely descriptive expression, to which Strawson has denied the status of ‘identification’. A specification of the form ‘the only one that stands in the relation R to this here’ does not differ qua specification from a specification of the form ‘the only one that isF .’ In both cases something is only singled out as an individual by saying that it is the only thing which has a certain property or is related in a certain way to something identified. And in both cases the question remains open what it is that has this property or relation, e.g. ‘the m urderer of Mr X—’. A nd then

What is ‘identification ? 319

the question arises: ‘who is that, who m urdered Mr X?’ This question, like the question ‘which is the highest m ountain?’ can initially be answered by giving a p roper name. But again this is a reply that can only count as identification if it is assumed that one knows which person it is that is so nam ed. And here too this can only mean: if one can spatio- tem porally identify the person.

It is difficult to see why Strawson believed that specification by means of a unique relation differs essentially from specification by means of a unique property. Probably he thought that only a unique relation ensures that what is specified is really only one. B ut if this were so one could not adm it all relation-descriptions which in virtue o f their m eaning p re­suppose that only one object stands in this relation (e.g. ‘the m urderer of . . .’) but only those where not m ore than one object can have the relation concerned to a particular object (e.g. ‘the m other of . . .’). Spec­ifications by means of such relations, unlike those by means of a unique property, cannot fail in the sense that it could tu rn out that the descrip­tion applies to m ore than one object. But without fu rth e r explanation it is not clear why a specification should not be an identification merely because it can fail. And, besides, all the kinds of identification distin­guished by Strawson can fail in the o ther sense, i.e. in the sense o f it tu rn ing out that there is no such object. T hus in the case of dem onstra­tive identification it can turn out that when one says ‘this beetle’ there is no beetle in the place at which one points; and the same clearly holds for identification by means of the objective identification by means of the spatio-tem poral position at which the object is situated.

Strawson has not sharply defined his concept of identification, for he has not distinguished, as I have done, a narrow er concept of identifi­cation from a wider one of specification. It is therefore also not clear what his criterion is for distinguishing a specification in the narrow er sense from one in the wider sense. Consequently, my assertion that specification by means of a unique relation does not differ, in the sense relevant to identification, from specification by means of a unique p rop­erty, does not tell decisively against him. O ne cannot argue definitively against an imprecise position. This assertion of mine is however conclu­sive if, as I assumed in connection with the quotation from Searle, the relevant criterion for distinguishing identification from m ere specifica­tion is that in the form er case the question ‘which o f all is m eant?’ is answered at the lowest level, in o ther words, that this question can no longer be repeated.

If we now look at the objections I have b rough t against the two forms of identification distinguished by Strawson together, they seem to point

Analysis of the predicative sentence 320

from two d ifferen t sides to one and the same thing: namely that Straw­son, who deserves the credit for drawing attention to the special signif­icance of the system of spatio-tem poral relations for the identification of perceptible objects, has yet underestim ated this significance. For Strawson this significance consists simply in this, that the spatio-temporal relations between objects constitute a comprehensive and unitary system of relations and thereby make it possible to identify an object, in so far as it is not demonstratively identifiable, if not by means of o ther unique relations, then at least by means of its spatio-temporal relations. But it has now been shown, (a) tha t dem onstrative identification is only an identification if it is presupposed that it can be replaced by spatio-tem­poral relations by which the objective spatio-temporal position which the object occupies is designated and (b) that specification by means of o ther unique relations is not identification at all. It emerges then, not just that the system of spatio-tem poral relations is specially significant for the identification of perceptible objects, but that there is only one sort o f identification of perceptible objects: if an individual perceptible object is to be spoken of in such a way that the question ‘And which is that?’ is no longer possible, then it m ust be located in space and time.

W here have we got to? T h e result at which we have now arrived has its foundation in an internal critical in terpretation of Strawson and, at the same time, in the attem pt to pursue the question ‘which is meant?’ through its various levels to the final one at which it can no longer be repeated. In this way I have situated my own enquiry and its fu rther systematic elaboration in the context o f the state o f the problem in con­tem porary analytical philosophy, which, as far as I can see, has not essentially advanced beyond Strawson. T he correction of Strawson’s view that it is most im portan t to hold on to is the refutation of his conception of an isolated dem onstrative identification. This conception is a residue of the traditional theory o f the object-relation as a simple relation to som ething immediately given and contradicts the insight that reference to an object is to be understood as specification, as the indication of which of all is meant.

W hat remains as the positive result o f the internal interpretation of Strawson however will scarcely strike you as transparent. T h a t all spec­ification of perceptible objects at the lowest level is location by means of their spatio-tem poral relations is a result that must first be made intel­ligible, in several respects.

1. So far it has not become clear why it is that there are several and essentially two levels of specification o f perceptible objects and what it is, in principle, that distinguishes the final level. We could get no answer

What is ‘identification f 321

to this question from Strawson, because his use of the term ‘identification’ is too unclear. So I confined my attention to the m ere fact that such levels are shown in the m anner in which we specify an object. But if reference to an object is essentially to be understood as specification, and specification exhibits such differences, then the analysis o f these differences is clearly essential for understanding our reference to objects.

2. Although the specification of perceptible objects at the lowest level has tu rned out to be spatio-temporal location it has so far remained unexplained how this specification actually functions in the interplay of demonstrative and objectively locating expressions.

3. Should we accept it simply as a fact that the specification of per­ceptible objects at this lowest level turns out to be spatio-temporal loca­tion? For Strawson this question did not need to arise because spatio- temporal location appeared to him to be merely one kind of identification among others. But if this fact is somehow to become intelligible then it must be possible to explain it by recourse to the concept of identification- specification at the lowest level - and we do not yet have such a concept.

In o rder to answer this question and in this way advance our enquiry into the mode of employment of singular terms and reference to objects we will now have to bring to bear the conceptual means which we derive from our systematic approach:

1. I have already pointed out that from the standpoint of W ittgen­stein’s principle it would have been obvious from the start to pose the question of the m ode of employment of the various kinds of singular term (and that means: the question of the particular form which the specificatory function assumes) by asking how it is established for which object the expression stands (p. 316). Moreover we will now have also to bring in the question of meaning; and we already know that the philosophical question concerning the m eaning of a form of expression is the question concerning how individual expressions of this form are explained - and this means: how their meaning is established.

2. In the introduction to the problem of singular terms I left open the question of whether the ‘standing’ of a singular term ‘fo r’ an object can only be understood in relation to the context of predication, for this question could not be decided on the basis of the truth-definition of the predicative sentence-form (p. 268f). But as it then em erged that ‘standing for’ must be understood as specification it also became clear that singular term s are, as regards their object-relation, essentially incomplete expressions - they require to be supplemented by something that is said about the object meant, i.e. by a predicate (p. 291). T he explanation of the mode of em ployment of singular terms can thus be

Analysis of the predicative sentence 322

reinforced by approaching the m atter from the o ther side and asking how perceptible objects m ust be specified if they are to be capable of being characterized by corresponding predicates, i.e. by predicates which are explained in perceptual situations.

3. There is a third methodological perspective directly connected with this second one. If the specificatory function of singular terms is to be understood in connection with the need to be supplem ented by pred i­cates this means that the establishing of which object the singular term stands for must be understood in connection with the establishing o f w hether the predicative statements into which the singular term enters as a dependent com ponent are true or false.

L E C T U R E 24

Specification and identification. Specification and tru th

T he attem pt to clarify Strawson’s vague talk o f ‘identification’, which at first seemed to correspond to what I had called the specificatory function of singular term s, led to the following result. W hat is thereby intended is not the general concept of specification, but a p re-em inent special case of this function by means of which is indicated at a lowest level which object is m eant, that is, in such a way that the question ‘And which object is the one thus specified?’ can no longer be repeated. And it em erged that in the case of perceptible objects this special case of specification is only given when the object is spatio-tem porally located by the singular term . However, we have-so far not succeeded in describ­ing this special case of specification in a way that would justify speaking of a clear concept of identification. It therefo re also rem ained unclear what relevance this special case has for specification in general and hence for the possibility o f referring to objects. A nd equally it rem ained unclear how this identification - specification by spatio-tem poral location - actually functions and in what way spatio-tem poral location has such a pre-em inence for the ultim ate specification of a perceptible object.

At the end o f the last lecture I re fe rred to th ree methodological per­spectives which can guide us in answering these questions which still rem ain open:

1. the orientation towards the actual mode of employment of singular term s and, in particular, towards the question o f how it is to be estab­lished for which object the expression stands.

2. consideration of the fact that singular term s are expressions which supplem ent predicates.

3. consideration of the function which establishing which object the expression stands for has in establishing the tru th or falsity o f the sen­tences that can be form ed with it.

O f these th ree perspectives the first is the most concrete; and it is also

Analysis of the predicative sentence 324

decisive, for we clearly cannot attribute to singular terms through the expressions which they supplem ent or through those which arise from such supplem entation, an object-relation which they do not themselves yield. On the o ther hand, if the specificatory function of singular terms is essentially a partial function (Teilfunktion) which supplements the function of characterization-expressions in such a way that expressions with an assertoric function result, then from the other perspectives one should at least be able to shed light on those questions which concern not so much the descriptive phenomena themselves as their intelligibility. The third perspective is of particular importance to us, for after all we are only undertaking the investigation of singular terms so as ultimately to arrive at an explanation of the use of the predicative sentence-form, or of the word ‘tru e ’ at this lowest level of its recursive definition. The best thing therefore is for us to start from a global consideration of the th ird perspective as the most comprehensive. The resulting orientation for the understanding of singular terms will automatically lead us to the detailed investigation of singular terms from the point of view of how one establishes which object is specified by a singular term. Since we shall only be able to answer a part of the open questions in this way, I shall revert, in the next lecture, to the second methodological per­spective, to consideration of the fact that singular terms are expressions which supplem ent predicates - predicates which if they were not thus supplem ented would be quasi-predicates - and this perspective will lead to a deepening of our understanding of the use of singular terms which, finally, should lead to a full understanding of singular terms from the third perspective, with the result that it is no longer necessary to use the locution ‘standing for’ in the truth-definition of predicative state­ments and that the predicative sentence-form becomes intelligible.

It would also seem appropriate to begin with the third perspective because, of the three, it is the only one to have so far been considered in analytical philosophy by some authors. T he reason for the unsatis­factory state of the semantics of singular terms, despite the vast literature, is that the discussion of such terms was only placed in the more general context of the semantics of assertoric sentences where the latter was pursued at the level o f meta-theory, thus without enquiring about the mode of em ploym ent of the expressions. W here the mode of employ­m ent of singular terms was enquired about this happened without an orientation towards more general semantic principles and, most im por­tantly, without consideration o f the function of singular terms in the sentence-whole. This holds for Strawson and it also holds for other im portant m odern investigations of particular types of singular terms

Specification and identification 325

which represent views opposed to Strawson, in particular those of Kripke and Donnellan. Only in Dummett do we find a clear orientation towards the third perspective, inasmuch as, in his book on Frege, he maintains that the meaning of a singular term, which is understood by him as a directive for establishing which object the expression stands for, is to be understood as the contribution it makes to the meaning of the assertoric sentences into which it enters as a constituent. And to understand the meaning of an assertoric sentence is also, for Dummett, to know how its truth or falsity is to be established. However, Dummett has not devel­oped the semantics of singular terms in detail. A similar conception is also to be found in Wiggins. However this conception has no conse­quences for the details of his analysis in which Wiggins sticks to the framework marked out by Strawson.1

The formulation in Wiggins is so concise that we can take it as our starting point, especially as it presents itself as an attempt (albeit a ‘rough and unsatisfactory’ one) to define the term ‘identify’. ‘I identify a if, for any <p such that I understand fully what it is for something to be tp, I (i) know what it is for “<pa” to be true and (ii) know without preliminaries what it is to investigate whether “<pa” is true.’2

What Wiggins means by the qualification that the two conditions only hold for those predicates V of which one knows what it is for something to be ip can be explained as follows: they hold only for predicates with regard to which one knows how one establishes that they only apply to an object, in other words, only for predicates whose verification-rule one knows. Of the two conditions the first concerns the knowledge of the truth-condition of l<pa’, the second the knowledge of the verification- rule; and since, as we have seen, the verification-rule is a more thorough­going formulation of the truth-condition (p. 202 f ) the first condition can be omitted. So we can summarize Wiggins’ definition as follows: a sin­gular term ‘a identifies the object a if - assuming that one knows the verification-rule of the predicate V’ (that one knows without prelimi­naries how one establishes whether ‘<px’ is true) - one knows without preliminaries how one establishes whether ‘<pa is true.

Let us first be clear what is plausible about this definition. If (a) we have a sentence that consists of two components V’ and ‘a ’ and (b) it is assumed that to understand a sentence is to know how one establishes whether it is true and (c) we understand V’, where this means: we know the contribution this sentence-component makes to establishing the truth of the sentence, then (d) to understand a’ must be to know the contri­bution of other sentence-component to establishing the truth of the sentence. I think this is clearly correct as far as it goes. However, the

Analysis of the predicative sentence 326

explanation remains abstract inasmuch as it fails to reveal the difference of the roles which the singular term and the predicate play in establishing the truth of the predicative sentence. Besides, Wiggins does not intend his definition to explain what it is to understand a singular term, but what it is for a singular term to identify an object. Now one can easily build a bridge here and say: to understand a singular term is to know how one establishes which object it identifies. This would make the con­nection between the meaning of the singular term and its identificatory or specificatory function. And you might think that this provided an answer to the question just raised concerning the special role of the singular term in the establishing o f the tru th of the sentence. But this is not so, for, in the first place, although speaking o f specification or identification brings out a peculiarity o f the singular term , it remains unclear how far this describes its role in the establishing o f the tru th of the predicative sentence form ed with it. And, secondly, the word ‘iden­tify’ is the word that Wiggins defines in his definition; what the role of the singular term is in the establishing of the tru th of the predicative sentences that are form ed with it can therefore not be presupposed by reference to a previous understanding of the word ‘identify’, but would have to result from the definition (the definiens).

However, the most im portant problem is as follows. It emerges clearly from the context that Wiggins does not intend his definition to be understood as a general definition of the function of singular terms. Rather the criterion of identification, as he defines it, is supposed to dem arcate a special sub-class of singular terms. What Wiggins has in mind is the narrow concept of identification, which Strawson aimed at but left unclear, and not the broad concept of specification. At the end of the previous lecture we found a possible external criterion for this narrow concept of identification in the non-repeatability of the question ‘and which is that?’ But one of the questions left open was how this distinction of levels is to be understood, and why it is that it comes to an end in the locating descriptions. Thus we are so far lacking a concept of identification. Could it be contained in Wiggins’ definition?

Certainly Wiggins intended this with his definition. But that this def­inition actually achieves this, that it really only applies to the specificatory function of a special class of singular terms, is bound to seem doubtful as soon as we consider how then the meaning and specificatory function of the o ther singular terms is to be understood. It ju st seemed to us perfectly natural to say that one understands a singular term - any singular term - if one knows how the tru th of the predicative sentences into which it enters as a constituent is to be established (assuming that

Specification and identification 327

one also understands the predicates of these sentences). I f this is not so, then how could the m eaning of those predicative sentences that contain singular term s which are not identificatory in the narrow sense be understood? So it would seem that Wiggins’ definition fits the general concept of specification rather than the narrow concept of identification. T hus, although Wiggins’ approach is more fundam ental than that of Strawson, his attem pt to dem arcate the in tended narrow concept of identification would also seem to fail. For like Strawson he too has failed to consider the broad concept of specification and hence has made no provision for the necessary dem arcation against it.

Nevertheless, Wiggins’ definition contains ano ther restrictive qualifi­cation which could perhaps lead us out of the present impasse. He says that a is identified if one knows ‘without prelim inaries’ how one estab­lishes that ‘(pa* is true. Is this any help? We could now say: one understands a singular term ‘a\ or an object is specified by ‘a , if one knows how one establishes w hether ‘(pa’ is true; but an object is identified by ‘a if we know without preliminaries how one establishes w hether ‘<pa’ is true (in both cases it is assumed that we know how the predicate is to be verified).

But of course you will ask: what does it mean to know such a thing with or w ithout ‘prelim inaries’? Wiggins says nothing about this and I do not believe we will get any fu rth e r by merely trying to speculate what could be m eant by ‘preliminaries’ in this context. T he qualification which rem ains unclear in Wiggins could, however, point a way out of the impasse if we simply regard it as an indication that there could be dif­feren t ways of knowing how one establishes (given that one already knows the verification-rule of V ’) w hether ‘<pa’ is true; and this would give us a perspective for explaining not only the concept of identification but also its relation to the general concept of specification, and the reason for that sequence of levels in the question ‘Which is it?’; for it could tu rn out that, depending on the kind of singular term ‘a , there are different ways of establishing that the predicative sentences ‘<pa form ed with it are true.

But what basis do we have, after so many blind alleys, for adopting this perspective, which has resulted from an obscure qualifying clause in Wiggins’ definition, and for seeing in it the possibility of an answer to our question? Well, the various considerations I have raised in con­nection with Wiggins’ definition were not simply random , but were guided by the aim of connecting Wiggins’ specific intention - a definition of identification - with the basic thesis which em erged from our third methodological perspective, viz. that to understand a singular term must be to know its contribution to the establishm ent of the tru th of predic­

Analysis of the predicative sentence 328

ative sentences. The perspective for the clarification of the differentiation of level in the specification-question which has now em erged is simply that which we get by focussing on a differentiation yielded by the third methodological perspective.

Assuming you accept this, you will now ask: but how should we now proceed? How can we tackle the question of the different ways in which one can establish - assuming one already knows the predicate’s verifi­cation-rule - whether a predicative statement is true? As we are dealing with that aspect of the establishment of the tru th of the statem ent which does not concern the predicate’s verification-rule but that with respect to which the predicate’s verification-rule is to be applied, the question at issue is how one establishes what it is with respect to which the p red ­icate’s verification-rule is to be applied. But this means: the question at issue is how one establishes for which object the singular term stands - and this is nothing o ther than the first of our three methodological perspectives.

Thus as soon as we try to carry through the third perspective con­cretely we are forced to make the transition to the first perspective. But the question presented by the first perspective has now received from the third perspective a direction that was not obvious from the start. T he question ‘How does one establish which?’ when it is incorporated into the question of how one establishes whether the corresponding predicative statements are true has not only the sense of ‘How does one establish which object it is that is classified?’ but also ‘How does one establish with respect to which object the verification-rule of the classi- fication-expression is to be applied?’ In other words, the question ‘How does one establish for which object the singular term stands = Which object is specified by it?’ not only belongs essentially to the context of something being classifiable by predicates, but it belongs to this context in such a way that it concerns the verifiability of this classifiability. Is this a hypothesis, you will ask, or is it compelling? It is compelling if two things are adm itted: (1) that we understand an assertoric sentence if and only if we know how its tru th is to be established and (2) that the meaning of the component-expressions of predicative sentences consists in their contribution to the m eaning of the sentences. But as we did not want to impose on the m eaning of singular terms anything that does not follow from their own explanation it is advisable in applying the first methodological perspective initially to disregard the third perspec­tive and only bring it in later.

I f we confine ourselves to those singular terms which also have a m eaning - and this means: disregard proper names - then in the p re ­

Specification and identificationn 329

vious lecture we had to do with four ways in which a perceptible object can be specified by a singular term: (1) by a demonstrative expression (‘this m ountain’, ‘this beetle’), (2) by a description by means of spatio- tem poral relations (‘the mountain that is situated at the intersection of such and such a parallel of latitude and such and such a circle of lon­gitude’), (3) by other unique relations to something already identified (‘the m urderer of Mr Maier), (4) by a unique property (‘the highest m ountain’). We have seen that Strawson sees no structural difference between (2) and (3), but that he sharply distinguishes between (3) and (4) and regards (1) as independent of (2). However, reflection on the possibility of repeating the question ‘And which is that?’ showed that there is no difference between (3) and (4), that (1) and (2) are closely connected and that between (3) and (4) on the one hand and (1) and (2) on the other there is a sharp distinction.

Applying the question ‘How does one establish which is meant?’ should reveal the correctness of this initially tentative reclassification; and if at the same time we bring the third methodological perspective to bear, the ground of the distinction of levels in the question ‘Which is it?’ should become intelligible.

What distinguishes the first case, the use of a demonstrative singular term , is clearly that establishing which is meant can be achieved directly through perception. It is decided by perception whether the specification succeeds or fails; one establishes whether there really is one and only one object of the relevant kind at the place to which the demonstrative expression points.

In the second case, that of a locating description, it is clearly the per­ceptual situation in which the existence of the object referred to would be established that is designated; and here too perception would decide whether one (and only one) object was specified.

If the reflections of the previous lecture on the dependence of demonstrative specification on locating specification were correct then clearly one will have to supplem ent what has now been said about the first case by reference to the second case. In the first case it is not enough to say that which object is m eant is established by perception. It is true that in a perceptual situation the presence of one and only one object of a certain kind is established; but which it is (and that means: which of all) is established by specifying what the perceptual situation is relative to all other perceptual situations.

We can now begin to understand why locating descriptions have the role of a final answer in the question ‘Which is it?’ (in the previous lecture we could merely point to this as a fact). What is special about

Analysis of the predicative sentence 330

spatio-temporal relations for the problem of specification resides not only in their universality, as Strawson thought, but in this: they are not just relations between objects; every spatio-temporal position is a per­ceptual situation. Strawson, like Russell, had assumed that all reference to perceptible objects must be dem onstratively-perceptually grounded. W hereas for Russell the object-relation did not reach beyond the par­ticular demonstrative act, Strawson conceived of a. system of identification; however, he presented it as though it consisted merely of relations between objects and was demonstratively anchored at only one place. In reality, however, a system of specification could never arise in this way. W hat Strawson failed to see is that the system of spatio-temporal relations is not just anchored in dem onstration and perception; ra ther it is a system of possible perceptual positions - and that means: a system of demonstrative specifications.

W hat is special about spatio-temporal descriptions, therefore, is that they specify perceptible objects as perceptible objects, that they specify them as objects of possible perceptions. But this does not yet capture what is essential about this state of affairs. I am speaking as though it were self-evident that there are perceptible objects, whereas it is the present analysis of the use of singular terms that is supposed to explain what it is for us to be able to refer to objects in perception. It is at this point that it becomes necessary to explain the significance of what was worked out by adopting the first methodological perspective by reference to the third perspective. For what is a perceptible object? W hatever else it may be it is an object to which indeed other predicates apply, but to which first and foremost perceptual predicates apply (i.e. predicates whose use can be explained in perception by means of examples); it is thereby constituted as a perceptible object. In a predicative sentence whose predicate is a perceptual predicate the distinguishing character of dem onstrative and locating singular terms consists in this, that they specify the perceptual object in such a way that one can establish of it as thus specified whether the predicate applies to it. One can, it is true, say: singular term s of this kind are such as specify the perceptual situ­ation in which it can be established by perception which the object is that they specify. But what this means is that they specify the verification- situation in which one can establish whether the perceptual predicates asserted of the object apply to it. This then would be, at least as regards singular terms o f the first two types, the special role they have in estab­lishing the tru th of predicative sentences: they indicate the verification- situation to which the predicate’s verification-rule is to be applied.

It is thus true to say that by means o f the spatio-temporal descriptions

Specification and identification 331

a perceptible object is specified qua perceptible object. T he significance of this statem ent, however, can easily be underestim ated, for it can be mis­in terpreted as m eaning that by means of this specification an object that is already the object it is independently of this specification is specified in a certain respect - as perceptible - ju s t as it can also be specified in other respects, e.g. as regards its causes or effects o r ordinal properties. But an object that is essentially a perceptible object cannot already be the object it is independently of its specification as perceptible. Thus the state of affairs is more appropriately expressed if in the first instance we form ulate it without using the word ‘object’: with the specification of an object by spatio-tem poral descriptions a verification-situation for the application of perceptual predicates is specified. T o refer to a per­ceptible object is to specify a verification-situation for perceptual p red ­icates; and since one can only specify such a perceptual Situationen such by spatio-tem porally locating it, the perceptible object as such is only specified if it is spatio-temporally located. With this definition of the specification of the object as such we have arrived at the narrow concept of identification for which we have been looking.

With this we now have a basis for rendering W iggins’ definition p re­cise, so that it can really be understood as a definition of the intended narrow concept o f identification. However, to obtain the p ro p er back­ground for this I would first like to deal with the still outstanding ques­tion o f how one establishes in the case of the o ther two types of singular terms which object is specified. Singular term s of the third and fourth kinds like those of the first two kinds contain a directive as to how it is to be established which object is m eant. In this case, however, the direc­tive does not take the form of specifying the perceptual situation in which the object would be perceived; ra th e r it takes the form of speci­fying a criterion — in the th ird case a relative one, in the fourth an absolute one - by which the object m eant can be recognized. T hus the singular term contains the directive to investigate all objects of a certain kind to establish which uniquely possesses the p roperty in question.

T h e ways in which all objects are investigated to establish which has the characteristic can be very d ifferen t depending on the property con­cerned. T he procedures required for establishing who the m u rd erer is are d ifferen t from those for establishing which is the highest m ountain. In no case is the establishing of the simple type we have become acquainted with in the case of perceptual predicates, where it could be established by simple perception w hether a given object has the char­acteristic. O rdinal properties are also implicitly relational. H ere estab­lishing which object it is is preceded by an investigation which compares

Analysis of the predicative sentence 332

the objects of the kind specified (e.g. mountains) with respect to the relevant ordinal property (e.g. height). With the relational properties it is mostly a m atter of causal properties (e.g., ‘the m other of . . ‘the m urderer of . . .’) or institutional properties (‘the wife of . . ‘the pres­ident of . . .’) and, depending on which it is, there are different, some­times complicated, decision-procedures for establishing whether a given object has this relation to another. For our purposes, however, all that matters is that in all cases of both these types of specification all objects of a certain kind must be individually investigated - directly or indirectly- to establish which one it is that stands in this relation.

But this means that to establish which object it is that is specified by means of a singular term of type (3) or (4) is to establish which of the perceptible objects, which are specifiable by means of locating descrip­tions, it is that uniquely fulfils the relevant criterion (being the so-and- so). In other words, specification of types (3) and (4) presupposes locating specification of types (1) and (2), not ju st as a m atter of fact, but in virtue of its meaning. T he m eaning of an expression of the form ‘the so-and-so’ is understood if (a) its form is understood (and that means: if one knows how in general statements of the form ‘T here is am ong all things one and only one that is so-and-so’ are verified) and (b) the p re­dicative, or relational, expression ‘so-and-so’ is understood (and that means: if one knows how it is established whether this predicate applies to an object).

We can now understand why Strawson (1) could not see where the relevant dividing line is between the types of singular term, (2) why he failed to give appropriate expression to the uneasiness he rightly felt about Russell’s Theory of Descriptions. Both these things result from his not asking how the meaning of singular terms is explained, or how one establishes which object a singular term specifies. Only when this question is asked does it become clear that although locating descriptions and descriptions by means of other unique relations have the same grammatical form - ‘the so-and-so’ - they have completely different semantic functions. T he ordinary relational description specifies a rela­tional characteristic and contains the directive to investigate all objects to establish which uniquely has the property. Locating descriptions, on the o ther hand, specify the situation in which one and only one object of a certain kind can be perceived. Although the locating description also has the form ‘am ong all objects there is one that is at such and such a time in such and such a place’ this does not contain a directive to run through (durchlaufen) all objects individually and examine them to establish which has the characteristic of being in such and such a place

Specification and identificationn 333

at such and such a time. This is because the running through of indi­vidual objects is itself perform ed by running through the different spa­tio-temporal positions (perceptual situations), thus because the individual perceptible objects are constituted as individuals precisely by the spatio- temporal positions they occupy. T he last explanation does not clearly follow from what has gone before. I shall be returning to this aspect. On the o ther hand, it is already clear that, although both kinds of description imply a universal statement, only in the one case can that which is m eant only be established by a running-through of all; in the o ther case no such detour is required. This is because, (a) that which is referred to is a perceptual situation, but also because (b) the specification of an individual spatio-temporal position implies a reference to all spatio- temporal positions in a way which differs from that in which the speci­fication of an individual object implies a reference to all objects. But I will also be coming back to this (p. 369).

We are now in a position to give a clear sense to Wiggins’ definition. Indeed it can now claim to be both a definition of specification in general, of what it is for a singular term, of whatever kind, to specify an object, and, when suitably qualified, of the narrow concept of identification (which is what Wiggins intended).

We can easily see that the definition also completely fits singular terms of types (3) and (4), at least if we ignore the qualification ‘without p re­liminaries’ which, due to its unclarity, is ultimately empty. If one speaks of ‘the m urderer of H ans’ one is clearly not directly in a position to verify whether a predicate, in particular a perceptual predicate, e.g. ‘having blue eyes’, applies to him. But one is also not directly in a position to verify such a thing about a person who is specified by a locating description. Thus if one were to in terpret Wiggins’ ‘without prelim inar­ies’ qualification very narrowly then his definition would only apply to demonstrative specification, which, as we have seen, is, if taken in iso­lation, not specification at all. On the other hand, of course, the person who understands the expression ‘the m urderer of H ans’ understands perfectly well what he has to do to establish whether the person so described has blue eyes; only what he has to do to establish this is some­thing different from what he has to do to establish whether the person who was born at such and such a time in such and such a place has blue eyes. It is not a question of which is easier or more complicated; that depends among other things on who it is who wants to establish it. Someone who was present when Hans was m urdered can probably establish m ore easily and with greater certainty whether the m urderer has blue eyes than if the person is specified for him by his date of birth

Analysis of the predicative sentence 334

and he does not belong to his immediate family. Rather it is a question of a fundamental difference in the method of verification of a perceptual predicate. A locating description specifies an object directly, that is by specifying the situation in which it can be established w hether a percep­tual predicate asserted of this object applies to it. In the case of objects that endure and change their place and to which different perceptual predicates apply at different times the locating, if it does not specify the complete life-path of the object but only, for example, its birth, does not designate the perceptual situation in which it can be decided whether a predicate that belongs to it at some time applies; however the life- path of the object can be followed from perceptual situation to perceptual situation to that time at which the predicate is supposed to apply to it. This problem of the duration of objects, to which we shall also have to return , introduces a complication into the connection between locating specification and the verification of perceptual predicates, but no dif­ference of principle. By contrast we have a fundamentally different situation if the object is not specified as a perceptual object by being spatio-temporally located but by means o f a characteristic; for in this case before one can know what the perceptual situation would be in which one can establish w hether a perceptual predicate applies to this object, it must first be established which o f all perceptual objects uniquely has the property concerned. It is this interm ediate procedure which is absent in the case o f locating descriptions; so in this sense one can say with Wiggins that in their case one knows ‘without prelim inaries’ how one establishes w hether the relevant sentence is true.

W hat is missing from Wiggins’ definition, in so far as it is supposed to define that narrow concept o f ‘identification’, is thus (a) the elucidation (which I have just given) of his qualification ‘without preliminaries’ and(b) the restriction to perceptual predicates. T he reason why Wiggins’ definition, if intended as a definition o f ‘identification’, must be restricted to perceptual predicates is this: if the special pre-eminence of locating expressions for establishing the tru th of predicative statements rests on their designating perceptual situations then this special role only obtains in the case of predicates which are such that it can be verified by per­ception whether they apply.

Let me clarify this by means of an exam ple.3 To establish whether the statem ent that the author of the Divine Comedy was a great poet is true it would clearly represent a detour were we first to establish which of all spatio-temporally locatable persons it was that wrote the Divine Comedy. For in o rder to establish whether it is true that the author was a great poet all that is relevant is that he was the author of this and

Specification and identificationn 335

possibly other poems, not anything to be established by perception about a locatable person. And even if we want to know w hether D ante - the person who was born in such-and-such a place and had such-and-such a life-path - was a great poet, the perception of this person is only relevant to the decision w hether this predicate applies to him in so far as we can only decide by perception, or appropria te inferences to p e r­ceptible actions, w hether this person is the au th o r o f the Divine Comedy and o ther poems. Thus if we want to know in a final sense who this g reat poet was we must clearly have recourse to locating singular terms and perceptual predicates; ‘if we want to know who in a final sense’: we now know what this means. It means: if we want to specify the object <25

a perceptible object - and this means: identify it. But as we have seen we can also specify an object in another way. And we can clearly also verify with respect to the object thus specified that it has such-and-such p ro p ­erties (if they are not perceptual properties); only we then mean: the so-and-so, whichever that may be, is so-and-so.4 We now know what is m eant by the clause ‘whichever that may be’. It means ‘which of all objects that are locatable and distinguishable as individual objects of perception by means of their spatio-tem poral relations’. We have thus answered one of the questions which rem ained open at the end of the previous lecture, namely, why it is that there are two levels o f specification of perceptible objects and what it is that distinguishes the narrow concept of specification - which we can call ‘identification’ - from the general concept of specification. T h a t a distinguishing o f individual perceptible objects is possible at all depends on there being a multiplicity of employ­m ent - s i t 11 ations of elem entary (perceptual) predicates. T h a t there are perceptible individual objects - i.e. things that are perceptible and clas­sifiable, such that from all of them one can be singled out — is connected, in a way that has still to be explained, with the fact that there is a m ul­tiplicity of perceptual situations, such that from any perceptual situation one can refe r to any o ther and thus can indicate of each one which of all it is. In this way it becomes possible to specify individual perceptible objects as perceptible; and that means: to identify them. But it is also possible to make general uniqueness-statem ents, that is, specify some­thing as som ething that uniquely possesses a relative or absolute p ro p ­erty. And if som ething is specified in this way then one can always also ask (though one need not), how it is to be identified.

You may ask: is not this conception of two levels of reference to indi­vidual perceptible objects very similar to Russell’s conception? Certainly. But it is distinguished from Russell’s conception in so far as the two levels now belong closer together; and this implies a new concept of the

Analysis of the predicative sentence 336

reference to objects. Even at the fundam ental level at which one refers directly to individual perceptible objects one refers to the object - con­trary to Russell’s opinion and the traditional opinion in general - in such a way that one singles it out as one of all; and this seems to be essential to understanding what it is to mean an object. And because one also singles out one from all at the other level it appears necessary to establish a unitary concept of the specificatory function of which the identificatory function can be shown to be a particular form (though of course a fundam ental one).

T he second question which rem ained open at the end of the last lec­tu re was: how should one interpret the fact that the specification of perceptible objects at the lowest level is by means of spatio-temporal location? This question which can also be put thus: how should one in terpret the fact that the objects of our perception are essentially spatio- tem poral objects? seems also to have now been answered, though only in a prelim inary way. T he employment-situations of elementary classi- fication-expressions, i.e. perceptual predicates, are distinguished from one another by their spatio-temporal relations. From this it would seem to follow that if we can hold apart, distinguish and identify the multi­plicity which these classificatory expressions classify, thus if we can refer to what is classified by them as objects, then the elementary objects resulting in this way must be perceptual situations; and that means: spatio-temporal positions. O f course, this answer cannot be regarded as satisfactory for it goes fu rther than the question and at the same time contradicts the facts inasmuch as instead of explaining why our elemen­tary objects are identified by spatio-temporal relations it would lead to the view that our elementary objects simply are spatio-temporal positions. So we now find ourselves confronted with the question: how are the identification of spatio-temporal objects and the identification of spatio- tem poral positions connected? And this question is in turn obviously closely connected with the third of the questions that rem ained open at the end of the last lecture, viz. how does spatio-temporal identification operate?

L E C T U R E 25

Spatio-temporal identification and the constitution of the object-relation

What we achieved in the last lecture was a preliminary clarification of the concept of identification which sufficed to explain the division into levels in the specification-question and to confirm the sharp distinction between locating and non-locating descriptions. We can now disregard non-locating descriptions and confine ourselves to the problem of iden­tification, since it has become apparent that the question of what is meant by ‘perceptual objects’ and how we are able to refer to such objects centres on this problem.

To this end we must again take up the question of how singular terms are used and how one establishes for which object they stand. Whereas in the previous lecture we were concerned to distinguish the use of demonstrative and locating singular terms from non-locating singular terms, our concern will now be to explain, from the same methodological perspective, the interdependence of demonstrative singular terms and objectively locating singular terms as regards their function in the iden­tification of perceptible objects. We shall have to so approach this prob­lem that we simultaneously aim at a clarification of the difficulty we encountered at the end of the last lecture when the hypothesis arose that perhaps spatial and temporal positions rather than objects that are in space and at times are to be regarded as the most elementary objects. We must therefore put the question about spatio-temporal identification in such a way that initially we leave open the question of w hether the identification of spatial (extended) objects and temporal objects (events) depends on the identification of places and times or whether the reverse is true or whether there is an interdependence between them.

Of course, one can only speak of places and times as objects if there are corresponding singular terms. But clearly there are such terms. In particular the demonstratives ‘here’ and ‘now’ are such. Despite their syntactic difference from the other singular terms they fulfil all the

Analysis of the predicative sentence 338

semantic conditions for singular terms with which we have become acquainted. By means of the words ‘here’ and ‘now’, if they are used in a specific situation, something is specified and even identified. An indi­vidual spatio-temporal position is singled out from all spatio-temporal positions as that which can be classified by a predicate (for example if we say ‘Here and now it is hot’). And this specification is an identification, since what is referred to is specified in such a way that, if the predicate is a perceptual predicate, then with the specification of place and time the perceptual situation for the verification of this predicate is specified. But what applies to ‘here’ and ‘now’ applies equally to all expressions which specify a place or a time by its relations to other places and times, or to spatial and tem poral objects. And just as in the case of other types of object, we can also apply the so-called quantifiers to places and times; we can say something about all or about some places and times. Again it was Frege who first recognized that ‘places, instants, stretches of time are, logically considered, objects’ and hence that ‘the linguistic desig­nation of a definite place, a definite instant, or a stretch of time’ is to be regarded as a singular term (‘proper nam e’).1

Thus w hether places and times are to be regarded as objects in the formal sense, which is the only relevant one for us, is not in doubt. W hat is in question is whether these objects, because they cannot as such be perceived, are not such that they can only be identified by means of objects which mark places and times by their spatio-temporal relations. What the relations of dependence are here can only be decided after we have clarified the actual mechanism of spatio-temporal identification. T he singular terms towards which we have to orientate ourselves in this question are thus no longer only those demonstrative and locating expressions which locate extended objects and events, but also those which locate places and times.

I introduced the considerations of the previous lecture by combining the question of how one establishes for which object a singular term stands with a further question, that concerning the contribution of the singular term to the establishing of the truth of the predicative sentences into which it enters as a component. Now before we em barked on these analyses I m entioned a third methodological perspective towards which we can orientate ourselves: the question of how singular terms must function if they are to be able to supplem ent predicates, especially p er­ceptual predicates. These two approaches - the discussion of singular terms from the perspective of the predicative sentences into which they en ter as parts and the discussion of them from the perspective of the predicates which they supplem ent - are clearly not independent of one

Spatio-temporal identification 339

another; and they are all the less so if, as we have seen in distinguishing different types of singular term s on the basis of their function in the verification of predicative statements, one has to consider what sort of predicates are involved. So this o ther m ethodological perspective was al­ready present in the reflections o f the previous lecture. However, it re ­m ained on the periphery and I postponed discussion o f the fu rther ques­tions to which it gave rise. Now, when the analysis is to concentrate on those singular term s whose distinctive character as identifying expres­sions is grounded in the fact that they refe r to the verification-situations o f perceptual predicates, it would seem natural expressly to bring in this methodological perspective, all the more so as it is this perspective which has given rise to the additional problem of w hether places and times should not perhaps themselves be regarded as the prim ary objects.

In order properly to understand how the various threads of the problem hang together you m ust also see that in adopting this perspec­tive — asking how singular terms must function if they are to supplem ent perceptual predicates - all we are doing is connecting up the discussion of singular term s with our earlier discussion of predicates. In terms of that earlier perspective the problem is: how must expressions function if they are to supplem ent classification-expressions which, if they were used independently, would function as quasi-predicates, in such a way that the classification-expressions acquire the character of predicates, in other words, in such a way as to yield expressions which can be true or false (or can be used to make true or false assertions)? So you see, we can approach the veritative function of singular term s - or the truth - context to which our speaking of ‘objects’ belongs - from two sides:

On the one hand, we can start, as in the previous lecture, from the fact that the statem ents to which the recursive definition of the word ‘true’ leads are statements in which som ething is asserted of an individual object. And by proceeding in this way we were able to see how the rela­tion to objects is functionally included in the relation to tru th ; for it became apparent that we know for which object the singular term stands if and only if we know - assuming we already know the verification-rule of the predicate - how the tru th of the statem ent is to be established.

On the o ther hand, instead of starting from this fact we can start from a hypothetical problem by asking how that part of the predicative sentence which already has a possible isolated m ode of em ployment would have to be supplem ented if som ething capable of tru th is to result. T he advantage of this second approach is that it does not presuppose the object-relation as a fact and, hence, can contribute som ething to understanding it. O n the o ther hand, this approach has, for just this

Analysis of the predicative sentence 340

reason, a constructive character and can therefore only provide a hypothesis that can only be decided by reference to the actual use of linguistic expressions.

Now the ‘on the one hand - on the other hand’ does not just concern the m ethod of proceeding. O ne must see that the two modes of analysis approach the unitary phenom enon from the two opposite ends and would therefore have to meet. T he level at which they meet is that of the verification-situation. If one starts so to speak ‘from above’, from predicative sentences, or singular terms, then, as we have seen, in the question of verification the other singular terms refer (verweisen) to those which specify objects as perceptible, thus to those which specify them by reference to the perceptual situations in which one can verify (a) w hether they exist and (b) w hether the perceptual predicates asserted of them apply to them. If, on the other hand, one starts so to speak ‘from below’, from the use of a predicate in a perceptual situation, the question arises: what else is needed if a perceptual situation is to be not merely a perceptual situation but a verification-situation? And this is identical with the question: what must be added to the use of a classifi- cation-expression in a perceptual situation for the classification-expres- sion to be related to an object? Thus the analysis ‘from below’ enables one to explain, by means of the contrast with a more primitive language that does not yet contain reference to objects and, hence, no truth-rela- tion, how a reference to objects is constituted linguistically. (I still say: how they are linguistically constituted. At the end of the discussion we shall have to ask w hether a non-linguistic reference to objects is still conceivable.)

T he level at which the two modes of analysis, which, somewhat clum­sily, I have called the analysis ‘from above’ and the analysis ‘from below’, meet, is m arked by the dem onstrative singular terms ‘this’, ‘here’, ‘now’. On their correct understanding there ultimately depends the u n d er­standing of what it is to refer to a perceptible object. Viewed from ‘above’, dem onstrative expressions are those to whose use the use of all other singular terms refers (verweist). T o understand the o ther singular terms is to understand this reference. On the other hand, viewed from ‘below’, if the dem onstrative expressions are used to designate the verification- situation of a predicate, then this use in turn refers back to other singular terms which refer to them. This reciprocal reference is not an in terp re­tation but can be dem onstrated in the use of these expressions: they are used in such a way that they can be connected with other expressions by means of the identity-sign, a procedure which gives expression to the fact that they can be substituted for one another in such a way that,

Spatio-temporal identification 341

depending on the situation in which they are used, the one speech- event is described as correct if and only if the other is described as correct; when the word ‘correct’ is used in this way it has the meaning of ‘true’. This substitutability which is expressed in the use of the identity- sign and in which reference to objects is constituted is thus the condition of the possibility of expressions being employed in such a way that they- in contrast to the expressions of a quasi-predicate language - can be described in the language itself as true and false. Now this dependence of a reference to objects on singular terms and their substitutability can be more easily explained in terms of the approach from ‘below’, because the approach from ‘above’ already presupposes that one is dealing with assertoric sentences, with expressions capable of being true and referring to objects, whereas from the perspective of the approach from ‘below’ we can explain the constitution of the object-relation by means of the contrast with a language that is still pre-objectual, viz. the quasi-predicate language.

I shall begin therefore by answering the question mentioned at the beginning of the lecture about how the mechanism of the identification of perceptible objects functions - and this means: how those singular terms are used by means of which we not only specify but identify p er­ceptible objects. Notice that this question is a purely descriptive question about the mode of employment o f certain types o f singular term. The constructive aspect to which I just referred will enter into the analysis only in so far as, in answering this question, I start with the singular terms closest to the quasi-predicate language, the demonstrative terms, and only after this pass on to the objectively locating expressions. Only in the next lecture will I tackle the constructive enquiry ‘from below’.

Now that we have seen that the words ‘here’ and ‘now’ are also singular terms and that, together with the word ‘this’, they constitute, as far as the demonstrative character is concerned, a unitary class, ou r concern is with this entire class of singular terms; and one can also include here the word T . I pointed out in the previous lecture that one can say of singular terms in general, in so far as they have a meaning at all, that to understand them is to know how one establishes which object they identify. We call ‘deictic expressions’ (and all simple deictic expressions are singular terms) those which are such that establishing which object they identify is dependent on the situation in which they are used. To understand the meaning of such an expression is to know how an object is identified by the use of the expression in relation to the employment- situation. Borrowing from the language of mathematics one can for­mulate this as follows: the meaning of such an expression is a function

Analysis of the predicative sentence 342

whose arguments are the speech-situations and whose values are objects. By means of the expression’s employment-rule a particular object is assigned to the speech-situation or, as one also says, the meaning ‘maps’ the speech-situations onto the objects.

This holds for all deictic singular terms. It is characteristic o f dem on­strative deictic singular terms that they identify an object not just in relation to a speech-situation but as one which is directly connected with the speech-situation. One understands the word ‘here’ if one knows that it is used in accordance with a rule according to which it identifies the place at which the person using it is situated, or to which he is pointing. Something similar holds for the other demonstrative expressions: with T the person speaking identifies himself, with ‘now’ the point of time at which he is speaking, and with ‘this’ an object to which he can point at the place at which he is speaking, or which (in the case of events) can be perceived at the point of time at which he is speaking.

T he meaning of deictic expressions in general and of demonstrative expressions in particular is usually presented in roughly this way. How­ever, this explanation is still a characteristically meta-linguistic one. It presupposes that the explanation already has at its disposal an objectual vocabulary in which one can speak about persons, places, times and objects generally. But one cannot yet understand the expression ‘a place that . . .’ at all if one does not yet understand the word ‘here’; and the same holds for the other expressions.

You will perhaps suggest that we could attem pt to describe how one can demonstrate (vorführen) the use of demonstrative expressions in a way which corresponds to the meta-linguistic explanation. And certainly it is such an explanation by dem onstration, similar to that which I gave for predicates and truth-functional expressions, that we need. But if we try to dem onstrate the use of words like ‘here’ and ‘this’ simply in accor­dance with the rule which was contained in the meta-linguistic expla­nation the result would be that variant of the quasi-predicate language which I dem onstrated with expressions o f the form ‘this isi7’, when they are used in such a way that ‘this’ is not yet replaceable by other expres­sions (p. 263). I f we wanted to explain to a child the m ode of employment of, for example, the word ‘now’ by simply saying ‘now hot’ whenever it is hot, and likewise with other predicates, such an explanation would correspond to the meta-linguistic explanation just given. But the child would not have learnt that meaning which the word ‘now’ has in our language. T he words ‘here’, ‘this’, if used simply in this way, would not yet be singular terms; they would not be being used to ‘specify’ a point of time, a material object, etc., but would be functionless appendages in

Spatio-temporal identification 343

a situation-relative use of predicates. If they are to be m ore than that then, although relating to the situation, they must already refer beyond (hinausweisen) the situation. Even within the situation one can only specify this situation - and that means (you recall): single it out vis-ä-vis o ther situations as the one referred to - if one already refers (bezugnimmt) to other situations and designates the present situation as one which can also be designated as the same from the perspective o f other situations. But to achieve this, to designate the present situation in such a way that it can also be designated from the perspective of o ther situations, we have at ou r disposal no names, no free-floating signs which could be assigned to the situation as such. R ather it is the dem onstrative expres­sion itself which refers beyond the situation in the requisite m anner by being used in such a way that one knows that it can be replaced by other deictic expressions if the same thing is referred to from another situation. It is in ju st this way that the present situation is first constituted as some­thing identifiable; and this means: only in this way is it referred to as an object.

It is im portant to see that this rule - that in the event o f a correspond­ing change of situation the demonstrative expressions are to be replaced by o ther deictic expressions - belongs to the meaning of dem onstrative expressions. If someone were to use dem onstrative expressions as they would be used in the extended quasi-predicate language to which I have just referred , thus, if, in the event of a change in situation, he w^ere simply to drop them , instead of replacing them by another deictic expression, we would say that he has not understood their m eaning in ou r language. The meta-linguistic explanatory schema that has ju st been presented is therefore not only circular, inasm uch as it already presup­poses a reference to objects; as an explanation o f the m eaning of these expressions it is also incomplete.

So one can only explain the m eaning of dem onstrative expressions, only dem onstrate their mode of em ploym ent, by also taking into account the group of o ther deictic expressions that belong to them. T o every dem onstrative expression that designates an object by its immediate rela­tion to the situation o f speech there belong other deictic expressions that likewise designate an object in relation to the situation of speech, but in such a way that - with ‘th ere ’, ‘then’, ‘later’ - another perceptual situation is designated, or - with ‘you’, ‘h e ’/‘she’ - a person o ther than the speaker.

T he m eaning - and that means: the identificatory function of deictic expressions - is thus characterized by two peculiarities: (1) it depends on the em ployment-situation which object is identified by the use of

Analysis of the predicative sentence 344

such an expression (2) the same object that in the perceptual situation is identified by a dem onstrative expression can, in another situation, no longer be identified by this expression; it can, however, be identified by another expression of the same deictic group. T he first of these pecu­liarities is the one which Strawson appealed to in his criticism of Russell; the second he failed to notice and it is precisely this one that should have been invoked against Russell (p. 308f).

It is this second peculiarity of deictic expressions which is fundamental for understanding how a reference to objects is possible, because only it explains how it is possible to refer to a perceptual situation in which one no longer finds oneself. Only because the use of the demonstrative expression in the perceptual situation already anticipates the specification of the same situation, or the same object that is perceived in the situation, by means of a non-dem onstrative deictic expression from outside the situation can one say that the use of the demonstrative expression itself specifies something, refers to an object.

W hat is crucial about this peculiarity of deictic expressions is that here at a lowest level the use of the identity-sign comes into play; and it is in precisely this way that the object-relation is constituted. We had previously seen that a reference to objects - a specification of which of all is meant - implies that the object meant can be distinguished from all others. But it is only from the circumstance that we are beings whose perceptual situations change that it follows that, if we are to be able to refer to objects of perception, we must be able, amid the changes of perceptual situation, to refer to the perceptual situations as identical; and because this reference to a situation takes place from the perspective of a situation that is always different, any situation can only be held on to as identical by another deictic sign being used in the new situation. Far from it being the case that an object can be distinguished from the others by a certain sign being simply assigned to it as a name, it can only be held on to as identical by different signs being substituted for one another according to a rule.

But what is this rule like? Obviously the mere substitution of a non­demonstrative deictic expression for a demonstrative one is not enough, for, in contrast to dem onstrative expressions, non-demonstrative deictic expressions are indeterminate. With ‘here’ a specific place is meant. With ‘th ere ’, on the o ther hand, a place is not yet specified; it is only said that a place other than the one here is meant. Merely to use the word ‘there’ invites the question ‘there where?’ In the case o f ‘that’ one asks ‘which?’, in the case of ‘th en ’ ‘when?’ And even in the case of ‘you’, if several are present, one can ask ‘Which of us do you mean?’ Making clear which

person is meant by ‘you’, which object by ‘that’, can clearly only be achieved by spatio-temporal specification. The problem therefore centres on the questions ‘where?’ and ‘when?’

T he most natural answer to these questions takes the form ‘then - when such-and-such happened’, ‘there - where such-and-such is situ­ated’. But these replies remain in the air if the tem poral and spatial objects referred to in them cannot themselves be identified by their spatio-temporal relations. This, however, can only be done by specifying the distance and (in the case of spatial objects) the direction from a certain reference-point; the latter serves as the coordinate zero-point. Depending on w hether this is objective or subjective one can speak of objective or subjective localization.

It is subjective, localization which connects up directly with the last considerations. For through subjective localization non-demonstrative deictic expressions acquire the determinateness they initially lack. In subjective localization the here-and-now is the coordinate zero-point and all other perceptual situations are specified by giving their distance and direction from the present perceptual situation. Only in this way can we identify another perceptual situation from the perspective of the present perceptual situation, by replacing the indeterm inate expression ‘then’ by, for example, ‘an hour ago’.

However, one should bear in mind, firstly, that the units of m easure­ment in such subjective localizations - hours, metres, points of the com­pass - already belong to a system of objective localization and, secondly, that localization by reference to the subjective coordinate zero-point yields no identification if it is not underpinned by an objective coordinate zero-point. This is clearly so in the case of communication between sev­eral persons. If each person could only locate something by reference to his own here and now there would be no possibility of the different communication-partners identifying something for one another. But even an individual speaker could bring no spatio-temporal relations capable of being held on to as identical into the manifold of his percep­tion if his only reference-point were his changing situation.

A stable spatio-temporal system of reference for the identification of perceptual situations can only arise relative to a spatio-temporal coor­dinate zero-point whose position in the time series and in space is objec­tively determ ined. Now there is no absolute time and no absolute space. What makes possible a stable coordinate zero-point in space, in contrast to the changing subjective coordinate zero-point, is the contingent fact that a sufficient num ber of the spatial objects around us remain invariant in their spatial relations to one another and hence constitute a fixed

Spatio-temporal identification 345

Analysis of the predicative sentence 346

spatial frame of reference from which we can then single out an arbitrary object - e.g. the place Greenwich - and agree upon it as coordinate zero-point.2 Likewise a stable zero-point for tem poral location is made possible by the contingent fact that there is a regularity of events in nature which provide a fixed unit of tem poral distance, such that a unique series of events form a stable frame of reference that makes it possible to pick out one such event (e.g. the birth of Christ) as zero- point. Both in the case of space and in that of time the zero-point is conventional, moreover its objective position is uncertain; but both these things are unproblem atic, because what is really fixed is not the point but the set o f spatial or tem poral objects whose relations to one another are invariant.

You could now ask: if one can only identify something relative to an objective spatio-temporal system of reference, could I not have spared myself the trouble of entering into the problem of specification by means of deictic expressions which is relative to the subjective coordinate zero- point? However, the subjective coordinate zero-point is as indispensable to identification as the objective one. I f a speaker does not know his own position in the objective spatio-temporal system of relations, then he does not know how he can, or could, establish what is meant. T he identification of a perceptual object by specification of its spatio-temporal relations prescribes a path which leads to the spatio-temporal position or positions at which the object could be perceived; and to be cognizable as a path it must be capable of being connected with one’s present posi­tion. In o ther words, to be able to orientate oneself one must know where one is. To know where the subjective coordinate zero-point is relative to the objective zero-point is, of course, the same as to know where the objective zero-point is relative to the subjective. W hen we locate an event or a material object we normally only give objective specifications, but these specifications only have a meaning for us because we assume that we know where we ourselves are in the objective system and hence can in terp ret the objective specifications subjectively. Thus although objective localization contains no deictic expressions it can only become effective for us as identification if it can be deictically interpreted. Conversely one can say that (a) the objective spatial and tem poral units of measurement and (b) the fixed objective spatial and temporal relations are the extra factors that must be added to the transition from the demonstrative to the non-demonstrative reference to situations if past perceptual situations are to be capable of being held on to as identical (or identifiable) and if, furtherm ore, all o ther possible perceptual situ­ations are to become identifiable.

Spatio-temporal identification 347

With this the question raised at the beginning of this lecture concern­ing the mechanism of the identification of perceptible objects is answered in outline. It has become clear how this identification is achieved by an interplay of dem onstrative singular term s and objectively, spatio-tem ­porally locating singular term s, an interplay that depends on the use of the identity-sign. T he use of the identity-sign has shown itself to be the decisive factor in the transition from the quasi-predicate language to the predicate language; and that means: the decisive factor in the use of expressions as singular terms. Until now this factor has rem ained on the periphery of our discussion, and its precise significance has still to be explained. With the clarification o f this aspect we shall be able to bring the discussion of singular term s to its conclusion, for it carries one over into that context fo r the sake o f whose clarification we have undertaken the whole discussion of singular terms: the explanation of the truth-definition of predicative sentences. Even then (Lecture 19) the hypothesis was suggested that the notions of a singular term ‘standing for’ an object, and of a predicate ‘applying’ to an object, which figure in this definition, become dispensable if we only presuppose an un d er­standing of the identity-sign.

However, it m ust first be decided w hether the questions which at the beginning of today’s lecture I connected with the question about the mechanism of identification can be answered on the basis of our analysis of this mechanism. These were (1) the question of w hether the identi­fication of spatio-tem poral positions is prior to the identification of objects in space and time, (2) the question of w hether we can support the analysis of the actual use of singular terms, and make its results m ore intelligible, by asking, hypothetically, how expressions would have to be used which supplem ent classification-expressions, which if they were used independently would be quasi-predicates, in such a way as to yield expressions capable of being true. It will em erge that these two questions are closely connected.

L E C T U R E 26

Supplements

I would like to divide this lecture into two parts and in each part discuss one of the two questions m entioned at the end of the previous lecture.

I The connection between object-relation, situation-independence and the truth- capacity of assertoric speech

The enquiry into the mode of employment of singular terms has shown that, if one disregards names, there are two ways in which singular terms can specify a perceptual object: (a) by means of an - absolute or relative— property and (b) by spatio-temporally locating the object. We were also able to show that and how the first kind of specification presupposes the second. Although I only considered the use of names in passing, in their case too it became apparent that the way in which they specify an object rests on specification by spatio-temporal localization. The analysis of the mode of em ploym ent of singular terms thus led to the conclusion that the only way to directly specify a perceptible object - and following Strawson I have called this ‘identification’ - is by specifying it by means of spatio-temporal localization.

T hat the fundam ental singular terms are the spatio-temporally locat­ing ones - by which we now mean not just the objectively locating ones but the whole system, presented in the last lecture, of demonstrative, deictically locating and objectively locating expressions - is a descriptive fact. It is, therefore, astonishing that hitherto it has not been clearly recognized. For the older tradition names were the fundamental singular terms - a view which could only be maintained so long as the specificatory function of singular term s had not been recognized and the mode of employment of the expressions had not been enquired about. And if recently Kripke thought he could reverse the conception, which has been characteristic of the m odern tradition since Frege, of the primacy

Supplements 349

of descriptions over names, this could only achieve a semblance of plau­sibility because of his failure to take account of the distinction between locating and other descriptions.1 Strawson and those who more or less followed him have, it is true, attached great weight to locating expres­sions; but they have not recognized their fundam ental significance, for, although they treated demonstrative singular terms as basic, they regarded them as independent in their identificatory role. T he basic error common to all previous conceptions was the assumption that there is, or even could be, a distinctive class of singular terms which can stand for objects in isolation without presupposing in their mode of employ­ment their replaceability by singular terms of other types.

O ne may simply accept the descriptive fact that the spatio-temporally locating singular terms are the fundam ental ones. But one can also seek to make this fact intelligible by starting out from the expressions which supplem ent singular terms to form sentences (i.e. classification-expres­sions) and asking: what function and mode of employment must we expect those expressions to have which are able to supplem ent classifi­cation-expressions to yield expressions capable of being true? This question ‘from below’ was only incidentally considered in the two pre­vious lectures, and it is one of the two questions which I mentioned as still being open at the end of the last lecture.

It is im portant to see that one could not ask the converse question. T hat is to say, one cannot start from singular terms and ask how those expressions which supplement them to yield expressions capable of being true must function. For an isolated use of a singular term would have no meaning and hence there is no language or part-language that con­sists only of singular terms or a pre-form (Vorform) of singular terms corresponding to the quasi-predicates as a pre-form of predicates. Such a language cannot even be conceived. By contrast, a language consisting only of classification-expressions (‘quasi-predicates’) is not only conceiv­able; such languages actually exist in the form of signal-languages and not only in the case of animals, where their rules are to be construed as causal rules, but also in the case of hum an beings, where their rules are conventional and their use is explained by means of the word ‘correct’.

But now how are we to pose the question if we start in this way ‘from below’? In any event we must expect a formulation of the following kind: how must the classification-expressions be supplem ented if such- and-such is to be achieved? But how is this condition to be formulated? A weak formulation that I have already employed (p. 32If) would be: how must they be supplem ented if they are to be capable of applying to objects? O r putting it the other way round: how must singular terms

Analysis of the predicative sentence 350

be used if they are to be expressions that supplement those classification- expressions that are explained in perceptual situations? Inasmuch as it already presupposes as self-evident that the supplem enting expressions specify objects such a form ulation does not involve much risk. And something can be achieved with it, though not much: it can be used to explain the nature of our elementary objects or their specification - that it m ust be spatio-temporally locating specification - but not how refer­ence to objects is possible.

So how can one form ulate the question without already mentioning objects in the condition that is to be fulfilled? Since the distinctive char­acter of quasi-predicates, in contrast to predicates or the expressions form ed by the supplem entation of predicates, is that their mode of employment is essentially situation-relative, the question can be for­m ulated thus: how must the classification-expressions that are used in a situation-relative way be supplem ented for that which they achieve - a classification - to become situation-independent? Reference to percep­tible objects would then be made intelligible by showing that only if the classification-expressions are supplemented by expressions which specify the perceptual situations by spatio-temporally locating them and thus objectifying them, and by means of these situations also specify other spatio-temporally locateable objects, does the classification-act become situation-independent.

O f course I believe this to be the case. But even so it seems to me that the question has not yet been given a broad enough formulation, for one can ask: what is achieved by our speech becoming situation-inde­pendent? To this it can be replied: only in this way is assertoric speech constituted; only in this way does it become possible to assert something which involves a claim to truth and to which o ther partners in other situations can take a position of affirmation or denial. For this reason I form ulated the question at the beginning of today’s lecture and also in the previous lecture, thus: how must classification-expressions be sup­plemented if they are to yield expressions with which something can be said that can be true or false? Only if we form ulate our question ‘from below’ in this way does it also correspond to the methodological per­spective from which I carried out the enquiry ‘from above’ in the lecture before last, the enquiry into the actual mode of employment of singular terms. Just as there the question of the actual mode of employment of singular terms was placed in the context of establishing the tru th of the sentences into which they enter as components, so now the question: how does the use of singular terms arise at all? would be placed in the

Supplements 351

context of the question of what makes it possible to speak of ‘true’ and ‘false’.

Now so far I have not adequately discussed this connection between the situation-independence of speech and its truth-relation. You could express doubt as to w hether it has really been established that the word ‘true’ cannot occur in a quasi-predicate language. It seems to me to be necessary to take issue with this doubt.

I earlier drew attention to the fact that if som ething is true or false it is so once and for all, and that it is essential to what can be called ‘tru e ’ or ‘false’ - to som ething asserted - that a position, o f affirm ation or denial, can be taken towards it by anybody and hence from any situation. It therefore seemed clear that there could be no question of ‘tru e’ and ‘false’ in a situation-relative language. But now one could adm it that this holds for ‘true’ and ‘false’ as applied to assertions, but doubt whether it only makes sense to speak of ‘tru e ’ and ‘false’ in this context. I f one asks w hether one cannot also speak o f ‘tru e’ and ‘false’ in a quasi-pred­icate language, my reply that assertions have a situation-independent sense could look like a petitio principii. T h e opponen t could say that ju st as I speak of quasi-predicates, so he m ust be allowed to speak of quasi-truth. What would he have in mind? I believe the following: in the quasi­predicate language the classification-expressions are also used with ref­erence to som ething - the actual situation. A lthough it is characteristic of the quasi-predicate language that that with reference to which the classification-expression is used is not referred to in the expression itself, one seems nonetheless able to say: depending on w hether the situation is actually correspondingly qualified or not the expression is used truly or falsely.

Clearly we need an informal understanding of what is meant by ‘true’, or an analogous expression extended beyond statem ent-truth, if we are to be able to decide whether such a word can already have a m eaning in a quasi-predicate language. Underlying the argum ent of the opponent of my thesis is the, in my opinion, correct idea that the use of a classifi­cation-expression with reference to som ething - and this may simply be the situation - is true o f that with reference to which the classification- expression is used is actually thus qualified.

Now two questions will have to be distinguished here: (a) the question of w hether it is possible for an observer who possesses a language of his own in which the word ‘true’ and singular term s occur to in terp re t the sign-acts o f beings who use a quasi-predicate language as true and false and (b) the question of w hether these beings themselves, who have no

singular terms, could already use the words ‘tru e’ and ‘false’, thus w hether they themselves could understand their sign-acts as true and false.

I f the user of a quasi-predicate language uses a classification-expres­sion in conform ity with the situation - as I shall say for the sake of brevity - then doubtless this use can be described as true by an observer. B ut in this sense even the reading of an instrum ent of m easurem ent, or a natural sign, can be called true or false. O ur problem obviously concerns not this question but the second question, viz. whether the user of a quasi-predicate language can himself use a word like ‘true’.

In the reflections which we must undertake to clarify this question I assume that we have to do with a quasi-predicate language whose rules are already conventional and whose classification-expressions can be ‘explained’ in the sense described earlier. And this means that I assume that the word ‘correct’ already exists in this language in the sense in which it is used to affirm the rule-conformity of an act. So we can refor­m ulate our question m ore pointedly by asking: is it possible, within the fram ework of a quasi-predicate language, for the word ‘correct’ to be used both to affirm the rule-conform ity of a speech-act and to affirm its situation-conformity? In this second mode of employm ent the word ‘correct’ would have the m eaning of ‘tru e’; and to avoid ambiguity, for this second m eaning of ‘correct’ I shall only use the word ‘tru e ’.

Both in the case of ‘correct’ and in that of ‘tru e ’ it will be better if we orientate ourselves towards the negations. We have occasion to use the word ‘correct’ because we have occasion to use the word ‘incorrect’ to register an act’s non-conform ity to a rule, to correct an act relative to a rule. And we have occasion to use the word ‘true’ only if we have occasion to use the word ‘un true’ or ‘false’, i.e. to correct a speech-act with respect to its not conform ing to a situation or its not conform ing (in a language with singular terms) to an object. O ur question can therefore also be form ulated thus: can there be occasion in a quasi-predicate language to correct the use o f a classification-expression with respect to its situation- conformity? And it seems that this question comes down to this: is there the possibility within a quasi-predicate language of correcting an error; and that means: is there the possibility in such a language of experiencing som ething as error?

Let us consider how things stand with regard to erro r in the case of predicative assertions. In the case of predicative statements relating to perceptible objects there seem to be essentially two possibilities of error: (a) the statem ent concerns an object which is not present. One errs because the object is not perceptible; and the e rro r is corrected by

Analysis of the predicative sentence 352

recourse to the verification-situation. T here can clearly be nothing cor­responding to this in the case of quasi-predicates, for these by definition relate to what is present, (b) the verification-rule of the predicate is so complicated that, even if the object to which the predicate is supposed to apply is present, one cannot establish by a simple act of perception whether the predicate applies to the object, but only by a certain oper­ation. It is difficult to think of an analogue of this in the case of quasi­predicates, because one can only speak of an error, or the correction of an error, if it is assumed that that to which the operations are applied remains the same during their application; and such a consciousness of identity is again excluded by definition in the case of quasi-predicates. The same holds for that possibility of erro r which also obtains in the case of simple perceptual predicates, in so far as their verification depends on certain perceptual conditions, such as proximity, illumina­tion, etc. When a user of quasi-predicate language says in the dark first ‘man’ and then ‘bush’, this cannot be construed as the correction o f an error; rather he has applied a word correctly to two perceptual situations. This is a conceptual consideration that cannot be empirically falsified, but could itself be used for correctly interpreting empirical findings. Thus, if one were to discover that children who one had supposed pos­sessed only a quasi-predicate language understand the correction of illusion and error, one would have to conclude that they already possess the rudim ents of singular terms.

Thus, any use of the words ‘tru e ’ and ‘un tru e’ which is grounded on the verifiability, or falsifiability, of classificatory utterances or - which is the same thing - on the experience of e rro r and the correction of error in regard to such utterances, presupposes that that which is classified can be held on to as the same. And if there is no possibility of holding on to th'e employment-situation as the same without the use of corre­sponding signs then the use of classification-expressions can only become erroneous, and in this sense untrue, if it takes up the situation-relation, which it has in the quasi-predicate language, into the utterance, by means of a sign representative - or, as we saw in the last lecture, by means of a series of such representatives which, depending on the speaker’s change of situation, must be substituted for one another. It emerges then that my thesis, that a speech act, if it is to be capable of being designated true or false in the same language, must be situation-inde­pendent, was not simply a petitio principii (in so far as it is precisely asser­tions that can be called true and false). Rather what is characteristic o f assertion, that a position of affirmation or denial can be taken vis-a-vis one and the same thing, is the condition of the possibility of correcting

Supplements 353

Analysis o f the predicative sentence 354

an error, hence of experiencing error, and hence of being able to use the word ‘u n tru e ’ in this sense.

Now you could object that the word ‘u n tru e ’ is not restricted in the predicate language to the context of error. We also use it to designate a speech-act as a lie; and although any speech-act that can be erroneous can also be a lie, there are also speech-acts in which a lie, but not an error, is possible. This is in particular familiar from the discussion that originated with W ittgenstein concerning the expression of inner states, e.g. the statem ent ‘I have toothache’ or the cry ‘ow!’ can be un true in the sense of ‘deceitful’, ‘insincere’, but there is no room here for error. Could not something corresponding to this hold also for the quasi­predicate language? Can we not, for example, suppose that a child that speaks a quasi-predicate language, though it cannot err, can lie?

What is wrong with this supposition can be seen immediately if we consider that a lie is an act defined by the intention of deceiving others; and that means: of producing in them mistaken beliefs. T o be able to have such an intention the person who lies m ust already have at his disposal a language in which an e rro r is possible. Although the person who says ‘I have toothache’ cannot himself be mistaken, this can only be a lie if the utterance belongs to an intersubjective context in which others can be mistaken about this state of affairs. The same is even true of the act o f pretending in which one aims to deceive others, not by the use of symbols, but by the production of natural signs. Even pretending which is not at all a linguistic act is only possible in beings who have at their disposal a language in which one can be mistaken. Thus, if a speech- act which can be called untrue because it is un tru th fu l is only possible in a language in which an utterance can be called untrue in so far as it is mistaken, then a lie is no more possible in a quasi-predicate language than an error.

T here is, finally, a more general argum ent to show that the word ‘untrue’ has no possible employment in a quasi-predicate language. T hat a classification-expression is used untruly (not in conformity with the situation, or object) - whether out of erro r or as a lie — presupposes that it is used correctly (in conformity with the rule). But what is it for a classification-expression to be used correctly? One can clearly only explain how such an expression, e.g. ‘red ’, is used correctly, in the sense of ‘in conformity with the rule’, by using it in conformity with the situ­ation in many situations. What we mean by ‘tru e’ in the sense of ‘in conformity with the situation’ is contained in the way in which the word ‘correct’ is used in the explanation o f a quasi-predicate, but without it being isolable. For one to be able to use a word like ‘true’ the two aspects must be separable, for the utterance is only incorrect in the sense of

untrue on the assum ption that it is correctly used in the sense of in conformity with the rule. How can this separation be achieved? It is clearly not enough that the speaker uses the expression correctly in o ther cases; for it does not follow from this that when he uses it incor­rectly in the presen t case he uses it untruly, for it could be that he has not yet learned the em ploym ent-rule well enough, or that at the present m om ent he has forgotten it. Rather one can only speak of an un true use if the p artner can say that the speaker is using the expression in the present situation in exactly the same way as in the o ther situations, and if the speaker can agree with this. Such a thing cannot be form ulated if one does not explicitly refer to the present object of application, or the present situation, and, equally explicitly, to ‘all o thers’. And this in turn is only possible if singular term s and quantifiers are used. This seems to me to prove that, even if we disregard the experience of e rro r and the correction o f error, which is the condition o f someone being able to describe his own utterances as un true, and even if we confine ou r­selves to the case o f someone describing the utterances o f another as un true in his own language, this language cannot be a quasi-predicate language.

T he doubt concerning the claim that the word ‘tru e ’ cannot be used in a quasi-predicate language which rested on the fact that quasi-pred­icates can also be used in conformity and not in conformity with the situation would therefore be disposed o f by the following considera­tions: (1) situation-nonconform ity does not suffice for speaking of un tru th ; ra ther we must have the possibility of distinguishing situation- nonconform ity and rule-noncom form ity; (2) the m ere fact o f situation- nonconform ity does not suffice. For it to be experienceable by those who use the language they must understand what it means to correct an erro r. And for this it is not enough that som ething is classified; it is necessary to hold on to it as identical amid changes in the speech-situation of the speakers.

My thesis that the situation-independence of speech is the condition of its capacity for truth thus seems vindicated. We are therefore justified in form ulating the question about how classification-expressions must be supplem ented if one is to be able to use them independently of sit­uation as follows: how m ust they be supplem ented if one is to be able to say som ething with them that can be true or false? T h e reflections which have just been carried out have also confirmed what we saw earlier in the transition from truth-conditions to verification-rules, namely, that the use of the words ‘tru e ’ and ‘false’ can only be explained by showing how one establishes that som ething is true or false. T he word ‘u n tru e ’ is a correction-word and can therefore only be explained by dem on­

Supplements 355

Analysis o f the predicative sentence 356

strating how an error is corrected. The question: how must classification- expressions be supplem ented if something is to be said with them that can be true or false? thus has the more precise sense: how must they be supplem ented if that which they achieve - a classification - is to be verifiable or falsifiable?

And this question is not simply a reform ulation of the question: how must classification-expressions be supplem ented if that which they achieve is to be situation-independent? For situation-independence is only one, negative side of the total phenom enon. The other positive side is the reference of a speaker, whose situation changes, to an earlier or some other (also the present) perceptual situation, to the same one to which any other speech-partners can also refer from other situations. Thus it is not an independence of situation that would consist in no longer having anything to do with situations; rather it is an independence of situation of speech-partners, who occupy perceptual situations and whose situations are constantly changing, in their reference to perceptual situations which they either occupy or do not occupy.2 T here are these two sides - situation-independence in situation-reference - that are presupposed in speaking of the verifiability of classifications that relate to perceptible objects.

The reflections that have just been carried out, which were intended to justify the question: by what expressions must classification-expres­sions be supplem ented if that which is achieved by their use is to be capable of being true or false? already contain the elements required for answering it. T he answer can be divided into two steps:

Firstly: the supplem enting expressions, it has emerged, must be sin­gular terms, expressions that have a specificatory function. This confirms the hypothesis expressed earlier (p. 224) that the use of singular terms, or reference to objects, is the condition of the possibility that signs can be used to say something that can be true or false. Of course there are also assertoric sentences that contain no singular terms. But we have seen that the truth-definitions of the other forms of statement refer back recursively to the predicative form of sentence. The latter emerged from the point of view of the truth-definitions as the most elementary form and the one which underlies the others. On the basis of the reflec­tions ‘from below’ that have now been carried out we no longer have to accept this as a mere fact. It has become intelligible why, if one starts out from the only form of pre-veritative language with which we are familiar - that of quasi-predicates - the level of expressions capable of truth is reached precisely with the predicative sentence-form, thus where classification-expressions are supplem ented by expressions whose

Supplements 357

employment-mechanism is that of singular terms. We now no longer need regard the circumstance that our most elementary assertoric sen­tences have the form of predicative sentences, are composed of a spec­ifying expression and a classifying expression, as a brute fact.

But secondly something else has become intelligible: not only why the level of ‘true’ and ‘false’ is reached with the constitution, in the use of signs, of a reference to objects, but also why the basic singular terms (those which ‘identify’ an object) to which the o ther singular terms refer (verweisen) are spatio-temporally locating expressions, thus why objects of perception are identified by being spatio-temporally located; or, put­ting it another way: why objects of perception are not only factually spatio-temporal but such that what decides their identity and non-iden­tity (thus what constitutes them as individual objects) is their spatio- tem poral location. For, as we have seen, the expressions which supple­ment classification-expressions to yield expressions with which something can be said that is true or false are singular terms whose objects are perceptual situations; and as perceptual situations are distinguished from one another as individuals only by their spatio-temporal relations, the objects for which these expressions stand can be nothing other than spatial and tem poral positions.

II Reciprocal dependence of the identification of spatio-temporal objects and the identification of spatio-temporal positions

We are thus faced once more with the question (which we have already encountered several times pp. 336, 337) o f whether we should regard spatial and tem poral positions or the objects that exist in space and happen in time as the primary objects. This is the second of the questions 1 announced for today at the end of the previous lecture. By trying to reach a decision about it we can deepen our understanding of the mechanism of the identification of spatio-temporal objects and answer some of the questions that have so far remained open.

It is the approach ‘from below’, in which one does not analyse singular terms in their actual mode of employment, but rather asks, in construc­tive fashion, how singular terms must function if they supplem ent clas­sification-expressions which are explained in perceptual situations to produce expressions capable of truth, which suggests the view that our prim ary objects must be spatio-temporal positions. If one stuck exclu­sively to this constructive approach it might even seem that spatio-tem- poral positions (perceptual situations) are the only objects to which per­ceptual predicates can apply. However, such a view would already

Analysis of the predicative sentence 358

founder on the simple fact that there are singular terms which do not stand for spatio-temporal positions but for perceptible objects that are merely identified by spatio-temporal positions. But even the weaker thesis implied by this description, that spatio-temporal positions, though not the only objects to which perceptual predicates can apply, are the prim ary objects, founders, as we saw in the previous lecture (p. 345f), on the fundam ental consideration that there is no absolute space and no absolute time and that spatio-temporal positions cannot be perceived by themselves. To be able to distinguish and identify spatial and temporal positions we need perceptible objects which m ark the positions. But we cannot be content with such a vague statement. O ur whole train of thought is conducted in such a way that at no point may we presuppose the notion of objects, and of a reference to objects, as something given. If one has succeeded in explaining a reference to spatio-temporal posi­tions in terms of the objectification of perceptual situations that is required to make possible a truth-relation, then the question arises of whether spatio-temporal objects, without which such an objectification seems impossible, can be explained from the same perspective, or w hether this requires another perspective.

So we must first ask: how are perceptible spatio-temporal objects con­stituted? It is clearly not enough that a spatial and tem poral extension (Erstreckung) is characterized by one or more perceptible qualities. If we are in a desert or at sea we have something extended that is qualitatively determ ined, but there are no objects, for there is nothing delimited that could be distinguished as an individual.

Frege drew attention to the fact that there is a class of predicates which are distinguished from others by virtue of the fact that they ‘def­initely delimit’ that to which they apply ‘and perm it no arbitrary divi­sion’.3 ‘Cat’, for example, is such a predicate: one cat is definitely delim­ited from another, and a part of a cat cannot itself be called a cat. ‘Red’ or ‘water’, on the other hand, are not predicates of this kind. If two red objects are definitely delimited from each other this is not in virtue of being red. And the predicate does not resist an arbitrary division; every part of a red surface is itself red.

In m odern analytical philosophy predicates of the first kind are referred to as ‘sortals’.4 What is characteristic about sortals is that such predicates (as emerges from Frege’s characterization) contain a criterion of identification and distinction; and this means: the predicate deter­mines which of the positions to which one can point belong to the one object - e.g. a cat - and which do not. By such predicates one object is definitely delimited from another and it is this which first makes it pos­

S upplements 359

sible to speak of several individual objects and hence to speak of objects at all (p. 294f). Grammatically this manifests itself in the fact that we use sortals with the definite or indefinite article (‘the cat’, ‘a cat’), but not other predicates. Connected with this is the fact that sortals contain a principle of countability. This is the context in which Frege introduced this concept: because such a predicate enables one to hold on to an object of a kind (a ‘sort’) as the same, and distinguish it from others of the same kind, it is possible to count how many objects o f this kind there are (for instance in a certain spatio-tem poral area). Finally, we can now also see that even the dem onstrative pronoun ‘this’ can only funcdon as a singular term if it is used together with a sortal. I anticipated this in earlier discussions by com pleting the expression thus: ‘this m ountain’, ‘this beetle’, ‘this F’ (where ‘F' m ust be a sortal).

If we consider what it is that enables such predicates as ‘cat’, ‘m ountain’, ‘chair’, ‘coin’ to definitely delimit that to which they apply, it is clearly the fact that they are shape-predicates. If such a predicate is to apply to something, that thing m ust have a definite spatial configuration. This is why an object of this kind is definitely delim ited (even if not sharply, as, for example, in the case of a mountain) from another object of the same kind, and why a part of such an object is not itself such an object.5 A predicate of this kind also makes possible som ething which again would not be possible with the word ‘this’ alone, namely, to follow an object through a stretch of time during which it rem ains one and the sam e.6 If spatially configurated objects did not also have duration, but were only m om entary phenom ena, we could not count them . It is essential for the identification of such objects that we be able to say, for example, that the cat which earlier came through the door is the same as the one which is now lying on the sofa, and is not the same as one that is now standing on the table. A m aterial object F that is observed at time h at a place si is identical with the object F that is observed at time tn at place sm if and only if it could be observed that during this time-interval the object has traversed a continuous path between si and sm- I f we call the object F at time t\ a and the object F at time t2 b, then the identity of a with b depends on the continuity of a trajectory in space; but this is the spatial continuity o f the trajectory of an F , and it would not be the same object if there were not a sortal lF ’ that applies to the object at every m om ent o f its life-path through space.7

However, the significance of the sortal fo r the identity of the object during the several tem poral phases of its life-duration is essentially dif­ferent from its significance for its spatial configuration. T he spatial con­figuration of such an object is essentially connected with the sortal p red ­

Analysis of the predicative sentence 360

icate. The duration of the object, on the other hand (and clearly also the path in space it traverses during its life-duration), is accidental to the sortal predicate. A five mark piece, for example, would not be a five m ark piece if it did not have a specific form, and that means: specific parts. But this predicate prescribes to the object to which it applies nei­ther a specific life-duration nor a specific life-path. Directly connected with this is the fact that the life-phases of a material object cannot be regarded as its parts, by analogy with its material parts. A material object is composed as a so-and-so out of its spatial material parts. But it is a so- and-so irrespective of how long it exists, and it is not composed as a so- and-so out of its life-phases. And it is in this that the following differ­ence is grounded. Pointing to the object in its various phases we can say ‘T h a t is the same cat’, o r if, not knowing this, we had previously called the cat at time t x a and the cat at time t2 b, we can make such statements as that a = b. W hereas we cannot say, pointing to different parts of the cat, ‘T hat is the same cat.’ For none o f the parts is itself a cat, thus it cannot be the same cat as another part. Here we can only say: these are all parts of the same cat a, whereas this is a part o f the cat b.

So although we can only follow a material object as one and the same on its life-path by subsuming it under a sortal, the sortal only contains a principle of spatial delimitation, not one of tem poral delimitation. Although it is essential to material objects to endure in time we can describe them as essentially spatial objects, inasmuch as the sortal that constitutes them prescribes no tem poral delimitation; they only have spa­tial parts, and hence only spatial positions can be constituted by them, not tem poral positions.

Now if tem poral positions can no more be perceived by themselves than spatial positions we m ust ask: by which perceptible objects can tem poral positions be constituted? T he answer of course is obvious: they are constituted by events. In comparison with the question of how we can refer to material objects this question has been strangely neglected in analytical philosophy. In the literature philosophers frequently speak as though every state of a material object, or of a place during a certain period of time, could be called an event, e.g. the sounding of a tone, a quarrel between two persons.8 To be consistent one would also have to describe the being red of a leaf as an event. Such a view gives rise to a series of difficulties. It seems that if it were correct we would be able to convert any predicative sentence whose singular terms stand for material objects, by substantivization of its predicate, into a sentence about an event. Thus, for example, the sentence ‘Peter and M arianne are quar­relling’ would be converted into the expression ‘the quarrel between

Peter and M arianne’, the sentence ‘The leaf is red’ into ‘the being red of the leaf’.9 And this leads to the question: how are events, which after all are supposed to be spatio-temporal (and so-called ‘extensional’) objects, distinguished from the corresponding abstract (and moreover ‘intensional’) objects, e.g. the attribute of redness that belongs to the leaf, or the state of affairs that the leaf is red. I do not want to go into these difficulties and will confine myself to that aspect which in our context is significant. States are temporally divisible, as material objects are spatially divisible. The sound or the quarrelling can be temporally divided. But their parts are again sound, or quarrel, and the temporal duration of the state is accidental. Such a temporal object is not ‘definitely delimited’. Thus, whether or not one can call such states ‘events’ temporal positions cannot be marked by such events. Tem poral objects of this kind seem rather to be analogous to those spatial objects one can call ‘masses’, such as sand or water. In both cases a spatial, or temporal, extension is in a qualitative state F. The one is spatially divisible, the other temporally divisible. And in both cases the predicate ‘F ’ does not itself contain a principle of divisibility and delimitation.

It seems, then, that we need sortal predicates which apply to what is temporally divisible and temporally delimit this in a way analogous to that in which the sortals of material objects delimit material objects spa­tially.10 And we would expect that the temporal objects that are consti­tuted by these sortals are of such a character that they too only are what they are in virtue of a certain configuration of heterogeneous (in this case temporal) parts, so that the sortal predicate that applies to the whole cannot apply to its individual parts. Now there actually are such temporal objects and the corresponding sortals. They are changes (Veränderungen), e.g. a sunrise, a revolution of the earth around the sun, a person’s birth, Peter’s flight from Berlin to Moscow, or, to connect up with the previous examples, the cessation of a sound, the outbreak, or the ending, of a quarrel, a leaf’s changing colour from green to red. Changes are defined by the transition from one state to another state. Thus a change consists of temporal parts: at least the two states from which and into which the change takes place, and, normally (always if there are such), the in ter­mediate phases. To none of these parts can the predicate used to describe the change as a whole be applied. Changes are thus temporal wholes analogous to the spatial wholes which are the material objects subsumable under sortals. Tem poral sortals contain a principle o f distinction and identification for temporal parts, just as spatial sortals contain a principle of distinction and identification for spatial parts. By virtue of the tem ­poral sortals what lies in the temporal dimension becomes ‘definitely

Supplements 361

Analysis of the predicative sentence 362

delim ited’ and countable; changes are tem poral objects. And just as through material objects spatial positions become markable, so through changes tem poral positions become markable.

It might seem that by having recourse to the sortal predicates I have surreptitiously fallen back on elements of a traditional object-concept. My talk of configurated formations, which, in the spatial case at least, would also be representable (in a non-m etaphorical sense), seems to fit this concept. I do not have any need to deny the representation-aspect- as an aspect. The im portant thing is to see that this aspect does not define the objecthood of these objects and, above all, that sortals make possible an aspect that is essential to spatio-temporal identification and which, in my account so far, has been simply passed over.

Let us cast a glance back to make this clear to ourselves. T he function of a singular term, we saw, consists in indicating which of all (of a p re ­supposed plurality) it is to which the expression supplem enting the sin­gular term applies (see end of Lecture 21). This holds for all singular terms, not just for those which refer to perceptible objects. To explain how singular terms can fulfil this specificatory function we had to explain the employment-rule of singular terms. And that means: we had to ask how - in terms of this employment-rule - does one establish which thing it is to which the classification-expression applies? Since this question already depends on the type of classification-expressions that supplement the singular term, it was necessary to restrict the investigation to the basic case of perceptual predicates (see beginning of Lecture 22). I say: to perceptual predicates; and not: to perceptual objects; for we cannot assume that we know what a perceptual object is. This must be shown by the way in which singular terms that supplem ent perceptual p red i­cates are used. This was why the question ‘How is it established which one is m eant?’ had to be built into the question of how the truth of the corresponding predicative statements is established (Lecture 24). It was this methodological step which made it possible for us to explain the distinction of a narrow er concept of specification (a concept for which I took over the term ‘identification’ which Strawson had introduced). This distinction is obvious in the mode of em ploym ent of singular terms but had hitherto not been understood, not even by analytical philoso­phers. Also, in making this concept precise it was particularly im portant not to presuppose some preliminary understanding of perceptible objects, but only what had shown itself to be the function of singular terms in connection with the verifiability of predicates. T he question ‘How does one establish to what a supplem enting predicate-expression applies?’ has, therefore, in the last resort - i.e. where it is a question of

identifying the object - the sense: ‘How is the situation to be established in which it is to be established w hether the predicate applies?’ So in the lecture before last we reached a result that has now been confirmed in the approach ‘from below’, namely, that we have to regard perceptual situations - and that means: the veriftcation-situations of perceptual predicates - as the prim ary objects. A nd one can see that what distin­guishes these situations - thus the correlates of classifications that in the quasi-predicate language are unarticulated, not yet objectified - from one another (not as regards their classifiability but as perceptual situa­tions) are their spatio-tem poral relations to one another. This led to the question which I tried to answer in the last lecture: how beings whose perception- or speech-situations change (temporally in any case, but also spatially) can, by the use of certain signs, so designate any past perceptual situation, but also the presen t one, tha t one knows from the perspective of any situation which of all it is in which one could establish that the supplem enting predicative expression applies. This is simply an application to the case of perceptual situations of what belongs to the essence of specification and, in particular, to the essence of identi­fication: (a) the consciousness of a multiplicity, o f which one is the object m eant (b) the possibility of designating any object as the object meant; and that means: the possibility of distinguishing it from others (it became clear that this designating is only possible th rough the rule-governed substitution of various expressions, so that, in addition to the negative em ploym ent of the identity-sign in distinguishing, a positive, but none­theless non-tautological, employment of this sign is essential) and, finally,(c) (the specific condition o f identifying specification): the possibility of not ju st distinguishing it in general from others (by means of a charac­teristic) but as such, and this means: as individual perceptual situation. This is precisely what is achieved by distinguishing it as a spatio-temporal position from all other spatio-tem poral positions.

Now in this analysis one aspect was taken for granted as obvious and thus passed over: if spatial and tem poral positions are to be distinguish­able from one another by their d ifferen t relations to all o ther positions, then they m ust first be capable of being m arked o ff (delimitable); but this they are not, taken by themselves, bu t only th rough the sortal p red ­icates by which objects are constituted, which, though essentially spatio- tem poral, are not mere spatio-tem poral positions. T h e modification of the provisional thesis that the prim ary objects o f our perception are spatio-tem poral positions that this makes necessary arises, therefore, not from a new perspective, let alone from a reversion to the traditional object-concept, but from an aspect, hitherto simply passed over, which

Supplements 363

Analysis o f the predicative sentence 364

belongs to the essence of specification, because it belongs to the essence of distinguishing an individual.

But now what is the modification that is required by this aspect? When we pursue this question it will em erge that the conclusions resulting from the introduction of sortals do not destroy the previous conclusions, but simply refine them . One could easily get the opposite impression. If the distinguishability of spatial and tem poral positions depends on the perception of spatial and tem poral objects that fall under sortals, then the w'hole approach, according to which the prim ary objects to which perceptual predicates can apply are perceptual situations, is called in question. If the objects determ ined by sortals are the primary objects, then do we not need a completely different approach if we are to make a prim ary reference to such objects intelligible? It is at this point that the danger arises of slipping back into a representation-theory. At least in the case of the objects determ ined by spatial sortals one could say: they are objects because they are spatially configurated and, as such, representable forms (Gebilde). But what is representable here would be merely a type, an image, and would not yet constitute what is m eant as an individual and as som ething distinguishable from others of the same kind. So this non-m etaphorical representation-aspect is not very signif­icant and has not even been dom inant in traditional philosophy. I only m ention it because it has, nonetheless, always played a part and could lead even us astray at the present juncture.

In reality, the impression that the priority of objects determ ined by sortals would point to another kind of reference to objects would be false. Objects determ ined by sortals are themselves essentially spatial and tem poral objects, and indeed not just in the sense that they are spatially or tem porally extended (divisible). W hat is special about sortal predicates is not that, like other predicates, they are only applicable to what is spatially and temporally extended, but that their application presupposes a specific configuration of what is spatially, or temporally, extended. We can only apply a predicate like ‘cat’ or ‘5-mark piece’ if we can say, pointing to many spatial positions: ‘Those are parts of the same cat (or the same 5-m ark piece).’ T hat we can point to these different spatial positions as individual, distinct positions is, of course, only possible on the basis of the em ploym ent of sortal predicates. But quite apart from the fact that we do not need to refer to the same object whose parts they are (we can, e.g., indicate the spatial positions which belong to the one cat or the one 5-mark piece with our finger or the point of a pencil) the fact that we can only indicate spatial positions by means of objects determ ined by sortals does nothing to alter the fact that, con-

Supplements 365

versely, we can only explain the sortal predicate by means of the indi­cation of spatial positions, together with the identity-sign. Indeed it is precisely this character of sortal predicates which makes it possible for spatial positions to be marked by such objects.

T he dependence of the distinguishability of spatial and temporal positions on that of spatially or temporally configurated objects is thus a reciprocal one. Although it is true that we can only mark spatial posi­tions because of the contingent fact that there are spatial objects deter­mined by sortals, we can only recognize these objects if we can delineate the combination of spatial positions prescribed by the sortal. Although we cannot m ark a perceptual situation as such without a structured object, and hence cannot speak of perceptual situations without the perception of such objects, these objects are, on the other hand, nothing but a combination, defined in such-and-such a m anner by the sortal, of qualitatively determined perceptual situations. T he perceptual situations in which quasi-predicates are used are essentially indeterminate, diffuse. T here are also singular terms, place- and time-specifications, whose objective reference is relatively indeterm inate. If someone says ‘It is snowing here’, then, although the place is determ ined by reference to a material object, namely, the speaker, the extent of the place thereby referred to normally remains open. On the other hand, it is now possible to delimit as precisely as one wishes the perceptual situation (which it would perhaps be better to call a perceptual position), a delimitation which would not be possible without objects determ ined by sortals, but is no longer restricted to them.

T he reciprocal dependence of reference to material objects and ref­erence to spatial positions, however, concerns not only the aspect of the markability of spatial positions or the divisibility of material objects; this reciprocal dependence manifests itself particularly in the aspect of identification which is decisive for reference to objects. Already in the last lecture we saw (p. 345 f ) that we can only identify spatial positions by reason of the contingent fact that the majority of the extended objects that surround us remain constant with respect to their spatial relations. We can thereby identify spatial positions relative to an objectively fixed coordinate zero-point; and something similar is true of events and tem ­poral positions. On the other hand, however, material objects and events can only be identified by the spatial and tem poral positions at which they occur. T h a t all identification of perceptible objects is identification by spatio-temporal localization is not itself a contingent fact, because perceptible objects are essentially objects to which perceptual predicates can apply, and, hence, can be nothing other than verification-situations,

or agglomerations of verification-situations, of perceptual predicates, and because verification-situations are distinguished as individuals (this, however, is an empirical fact) by their spatio-temporal relations. The system of spatio-temporal relations is, therefore, not merely the only comprehensive system of identification (as Strawson believed); it is the system of identification of perceptible objects.

You might now ask: why must this function of spatio-temporal relations with respect to the identification of perceptible objects imply that one must also speak of spatial and temporal positions? Does it not suffice to speak of the spatial relations between extended objects and the temporal relations between events? It does not suffice, because once the system of spatio-temporal relations is constituted an infinite ordered multiplicity of spatial and temporal positions are thereby distinguishable and iden­tifiable, of which only some coincide with the extensions of objects determ ined by sortals. Arbitrary spatial positions can be identified between and beyond material objects, likewise arbitrary positions within each material object. W ithout this system of spatial positions a universal identification of spatial objects is simply not possible. Firstly, we can only mark, and then identify, masses - e.g. a snowfield, an ore-mine, an area of rain - which are not already spatially determ ined by a sortal by reference to the spatial positions which they occupy. Secondly, we can also only identify material objects by means of spatial positions, which, of course, are in turn identified relative to a material coordinate zero-point (or the material fram e of reference that makes this possible).

So we have no alternative but to recognize the identificatory system of perceptible objects - the system of spatio-temporal relations - as the unitary and comprehensive system of indefinitely many spatial and temporal positions, which is, however, dependent on a finite multiplicity of material objects and events as perceptible points of reference (.Bezug­spunkte). There exists between the two multiplicities - that of material objects and events and that of spatial and temporal positions - a rela­tionship of reciprocal dependence. T he question is not, which of the two multiplicities has a pre-eminence over the other, but in what the pre-eminence of each consists.

For the multiplicity of material objects and events this has already been explained. With respect to the multiplicity of spatial and temporal positions, on the other hand, all that has so far become clear is that it cannot be reduced to the other multiplicity. What its pre-eminence con­sists in has so far only been hinted at. It rests on the fact that this mul­tiplicity of spatial and temporal positions stands in a universal (unitary and comprehensive) and at the same time systematic (ordered) connec­

Analysis of the predicative sentence 366

Supplements 367

tion; whereas the multiplicity of material objects and events is diffuse (zerstreut) and only acquires the ordered connection necessary for their identification by being contained in the multiplicity of spatial and tem ­poral positions (which is in turn only made possible by the demarcations of those objects). But now this is the basis of further essential distinctions of spatial and tem poral positions in the mechanism of identification, without which the latter would rem ain ultimately unintelligible.

T he singular terms which identify spatial and tem poral positions are also distinguished from the others by the fact that their objective ref­erence cannot fail.11 If I say ‘this F* it can tu rn out that there is not an F here and now, thus that the object that is identified by this expression does not exist, thus that the objective reference fails; whereas the objec­tive reference of the expressions ‘here’ and ‘now’ cannot fail. This dif­ference is evident not only in the case of dem onstrative identification, but also in the case of every other kind of localization. I f we say ‘the F that is in such-and-such a place’, this identification of the material object can fail, but not the identification of the spatial position that it implies.

Notice that the reason why the identification of the material object or event can fail is because these objects are identified by being placed in a relation to a spatio-temporal position by means of an existential asser­tion. W hen we say ‘this F\ the tru th of the existential statem ent ‘T here is one and only one F that is here and now’ is implied or presupposed. And it is clear that the identification of the spatio-temporal position that is implied by such a statem ent does not itself presuppose the tru th of an existential statement. So although the identificatory system of spatial and tem poral positions as a whole depends on the identification of some material objects and events, the identification of each individual material object and event refers (verweist) to the identification of corresponding spatial and tem poral positions. For the latter represent the verification- situations in which one establishes w hether the identifying reference to the material object or event fails or not, and that means: w hether the object exists (or occurs) in this position or not. Because the reference to the individual spatio-temporal positions cannot fail, it is also senseless to enquire about their possible non-existence. It is therefore senseless to speak of the existence of such objects at all.

T he fact that in the employment of those singular terms that identify spatial and tem poral positions the objective reference cannot fail is thus identical with the fact that one cannot speak of the existence or non­existence of these objects. We here encounter a question which I have hitherto neglected, the question, namely, of w hether every use of a sin­gular term , and, hence, every predicative assertion, implies the existence

Analysis of the predicative sentence 368

of the object specified by the singular term. T here are authors who regard this as self-evident. Searle has even called it an ‘axiom’12 - a rem arkable view, when one considers that it seems to contradict the m odern understanding of the concept of existence, according to which existential statements are general statements. It would thus seem plau­sible to accept Searle’s ‘axiom’ for those predicative statements which in Russell’s sense are not genuine predicative statements at all but general statements, but reject it for those predicative statements whose singular term does not specify an individual indirectly, by means of a character­istic and a uniqueness-clause, but directly as an individual. Russell himself at any rate drew this conclusion. To him it seemed that it is only where the specification of an object can fail that the existence of this (to be more exact one would have to say: of such an) object is implicitly asserted by means of a general statem ent, and that, hence, where an individual is directly designated, the reference cannot fail, and, hence, that it does not even make sense to speak of the existence of the individual object (above pp. 300f).

But to what purpose, you wiil ask, do I appeal to Russell’s theory of logically proper names, the abstruseness of which is undeniable and the weakness of which I have myself shown up? We have seen, however, that Russell’s theory contained some real insights. So it is better to link my discussion to it than to so contradictory a conception as that of Strawson and Searle who (a) share the modern interpretation of the concept of existence but (b) reject Russell’s analysis of ordinary predica­tive statements as general statements, and yet (c) admit that every predi­cative statem ent presupposes an existential statement.

What seemed correct in Russell’s theory of logically proper names was the idea that the prim ary reference to perceptible objects must be, not via proper names, but via demonstrative expressions (p. 303). Its weakness was the idea that the objective reference did not reach beyond the m om entary act. T he word ‘this’, we then saw, can only identify an object if it is used in such a way that it cän be replaced by other deictic expressions and then by objectively spatio-temporally locating expres­sions. We were able to retain Russell’s fundam ental distinction between two kinds o f statem ent about individuals (p. 335): on the one hand, statements which specify something by means of a characteristic; on the other hand, statements which designate an individual directly as an individual. Except that, Russell’s use of isolated demonstrative expres­sions was now replaced by the system of demonstrative identification as a system o f localization (p. 330). This modification seemed necessary for the particular reason that we were able to make clear to ourselves

Supplements 369

that an objective reference (meaning something) is only conceivable as specifying, and that means: indicating which of all (p. 303). On the other hand, it could seem that Russell’s distinction thereby loses its point, for now even statements which refer to an individual directly as an individual are existential statements and, hence, general statements. And the dif­ficulty, upon which I have already rem arked (p. 304f), that we could not explain the form of general existential statements if there are not sin­gular statements which cannot in turn be interpreted as existential statements, has so far rem ained unresolved. Implicitly the solution had, of course, already been almost achieved as it became clear that the two kinds of existential statem ent are verified in fundamentally different ways: the statement ‘T here is one (and only one) F at such-and-such a spatio-temporal position’ is not verified by examining all Fs for whether exactly one is in that position, but by establishing whether in the specified verification-situation there is exactly one F. In the case of a locating existential statement one does not establish which the material object referred to is by running through (durchlaufen) all material objects or all Fs, for what one would be examining them for would already be presupposed in this running through; rather one establishes which is the material object referred to by recurring to the spatio-temporal position specified in the localization as the verification-situation for this existential statement.

We can now see that the identificatory system which takes the place of the use of isolated demonstrative expressions in Russell’s theory o f logically proper names is distinguished from this theory by two steps: firstly, the isolated demonstrative act is replaced by a system of dem on­strative identifications, but, secondly, the elements of this identificatory system are not objects to be designated by ‘this’, for these do not consti­tute a system; rather they are spatial and temporal positions. It is this second step which makes it intelligible how Russell’s sharp distinction between general statements and singular predications which can no longer be construed as existential statements is preserved, despite the fact that the identification of material objects and events is accomplished by means of existential statements. The identification of material objects and events is indeed accomplished by means o f existential statements, but (a) these are existential statements of a special kind which relate the objects concerned to the spatio-temporal positions ‘at’ which they ‘occur’ or ‘exist’ and (b) this presupposes an identification of spatio-temporal positions which is itself not accomplished by means of existential state­ments. With respect to such positions, therefore, one can no longer speak of existence or non-existence (as was the case in Russell with

Analysis of the predicative sentence 370

respect to the objects of logically proper names); and, hence, the objective reference of the corresponding singular terms cannot fail.

But is there not then repeated at the level of those singular terms which stand for spatial and temporal positions the same difficulty I found with Russell’s logically proper names, namely, that an expression that does not contain an implicit reference to all cannot function as a singular term, because it cannot specify which of all is meant? However, because spatial and temporal positions are members of an ordered series, a space- and time-specification contains the required implicit reference to all other places, without implying a general existential statement. Which of all is meant is not specified by a characteristic for which one item after the other would have to be examined, but by indicating its place in the ordered series. In such a procedure is constituted the reference to an individual as an individual It is such a procedure which m ust be presupposed in the explanation of how a general statem ent is verified, and it does not in tu rn presuppose this explanation.

With regard to the existential statements by means of which material objects and events are located relative to the spatial and tem poral posi­tions, it has already become clear that in regard to their verifiability, and hence their meaning, they are not comparable with ordinary exis­tential statements. This can be made clear in the following way. Just as in a singular statement the singular term specifies the verification-situation in which it is to be established whether the predicate applies, so in a general statem ent the expression ‘of a ll/7: - ’13 specifies the verification- field, which is to be run through in order to establish w hether the p red ­icate applies. But now because the system of spatio-temporal positions, as the totality of possible verification-situations, is our universal verification-

field, in the case of a locating existential statem ent (‘O f all F: there is one and only one that is in such-and-such a position’) the specified part of the universal verification-field takes the place of the verification-field of all F. W hat is asked is not whether exactly one F has a property, but w hether there is exactly one F in this part of the universal verification- field.

But now what does the ‘is’ which expresses the presence of an object in a spatio-temporal area mean? We speak of the occurrence of an event at a particular time, likewise of the occurrence of a material object at a place, o f a material object’s being situated at a place, and of the existence of a material object during a period of time. I do not want to go into this problem, but would merely point out that here we have a genuine case of the existence of individuals. It is genuine because it does not fall foul of the criticism of the traditional concept of individual existence.

S upplements 371

If this tem poral existence can be understood as a predicate at all, then it is not, like the traditional concept of existence, a one-place predicate, but a two-place predicate (x exists at tn).14 W hether x still exists at time tn is not tested by exam ining x - if x no longer exists, there is nothing there to be exam ined - but by exam ining the point of time tn. It is precisely this which reveals the pre-em inence of spatio-tem poral posi­tions, as the ultimate verification-field, over the objects that occur in this field.

T hus there are various aspects which give spatio-tem poral positions a pre-em inence in the mechanism of identification over material objects and events - on the existence of which the possibility of identification nonetheless depends. These aspects, however, are only d ifferen t sides of one and the same state of affairs: (1) the ordered multiplicity of spatial and tem poral positions, as the totality o f perceptual situations, is the universal and ultim ate verification-field for the applying (Zutreffen) of perceptual predicates; (2) for this reason all perceptible objects, which are not themselves parts o f this field - spatial and tem poral positions - can only be identified th rough their presence in parts of this field by locating existential statements; and, hence, (3) - this is merely the o ther side of the same phenom enon - the identification of spatial and temporal positions cannot itself fail.

L E C T U R E 27

Results

Despite some obvious gaps and defects in execution I now want to regard the analysis of the mode of employment of elementary singular terms- those with which one can refer to perceptible objects - as concluded. In this lecture I shall merely summarize what follows from this analysis for the understanding of reference to objects, the explanation of the predicative form of sentence, and the elementary use of the word ‘true’. At the outset I predicted that, contrary to the traditional prejudice, the analysis of the mode of employment of expressions which ‘stand for’ objects would be essentially m ore complicated than the analysis of the mode of employment of predicates. The extent to which this is grounded in the facts of the case has meanwhile become apparent. The analysis of the mode of employment o f these expressions is more complicated because the mode of employment is itself more complicated. It is so because the use of individual singular terms, in contrast to the use of elementary perceptual predicates, is not isolable. T herefore the indi­vidual expression cannot be explained by itself. Rather the use of any type of singular term refers (verweist) systematically to the use of singular terms o f other types; and in part, as we have seen, this reference is even a reciprocal one.

These references (Verweisungen) are expressed in the replaceability of one expression by other expressions in the predicative sentences in which it is used. This replaceability can itself be asserted in an identity-state- ment. T he replaceability of V by ‘b’ expressed in the identity-statement ‘a =b’, however, does not fully bring out the phenom enon which I here call ‘reference’, because in an identity-statement neither of the two terms has a priority over the other, whereas in what I call reference it is always the one term that refers to the other; and this is not convertible except where the reference is a reciprocal one. We first encountered this phe­nom enon in connection with the question: ‘and which object is a?’ (Lee-

Results 373

ture 23). And we saw then that the reason for the precedence of certain types of singular term over others which manifests itself in the answering of this question resides in the m anner in which they fulfil the function of specifying common to all singular terms; and this means: in the dif­ferent role which their use has in the verification of those predicative sentences whose predicates are perceptual predicates (Lecture 24). I can therefore define what I mean in this context by ‘reference’ (Verwei­sung) thus: a singular term of typeX - - refers to a singular term of type Y - to ‘n j1 - if and only if (1) mx —n y and (2) any sentence Fmx , where *Fy is a perceptual predicate, can only be verified by verifying ‘Fny\ The peculiar complexity that characterizes the mode of employ­ment of singular terms is grounded in the fact that there are no singular terms of any type which do not refer to singular terms of other types. One can, it is true, understand a singular term and also in a certain sense know for which object it stands (which it specifies) without having at one’s disposal a specific term 'n-J to which it refers and by which would be identified which object mx is. But it belongs to the understanding of any singular term (mx that one knows to which- other type or types it refers, thus that even if one does not have a specific ‘n y* at one’s disposal to which the *mx refers one nonetheless knows that it refers to some ‘n j . If that were not the case then ‘mx’ would simply not be understood as an expression which stands for a perceptible object.

To render concrete what has just been said and to recapitulate let us glance back at the reference-system that the last few lectures have yielded. The comparatively simple cases are those in which the reference is one-sided rather than reciprocal. In the one-sided sense both non­locating descriptions and proper names refer to locating expressions. I have shown this for non-locating descriptions (Lecture 24), whereas in the case of p roper names I have confined myself to hints and left open the question of the precise understanding of their mode of employment. This is one of the gaps in my presentation; but it is a gap which seems to me tolerable. For the fact that is crucial in the context of our enquiry, viz. that any sentence Frn in which F ’ is a perceptual predicate and ‘m’ a proper name can only be verified by verifying ‘Fn, where (n is a locat­ing expression and m —n, cannot be doubted.

Things became more complicated in the case of the locating expres­sions themselves. The expectation that the singular terms o f this type to which those of the other types refer do not themselves refer to others was disappointed when it turned out that locating expressions are themselves divided into two types - subjectively locating (or deictic) and objectively locating expressions - and that every use of a singular term

Analysis of the predicative sentence 374

of one of these types refers to singular terms of the other type (Lecture 25). It was a fu rther defect in my presentation that I did not show in detail how objectively locating expressions can be explained nor what the identification-criteria are for the identification of something objec­tively located (i.e. by reference to an objective coordinate zero-point) with something subjectively located (i.e. by reference to the speaker). Nonetheless the fact of reciprocal reference showed itself to be indu­bitable. I can now form ulate it in a m ore informative way in terms of the definition just given: if ‘mo is an objectively locating singular term and ‘F ’ a perceptual predicate then ‘Fmo can only be verified if there is a subjectively locating expression ‘ns’ such that the speaker knows that mo = ns; ‘Fns’ is verifiable for the speaker in the sense that he now knows in what relation the verification-situation for determ ining w hether ‘F’ applies stands to his own position, ‘ns’ contains a directive as to what the speaker has to do (or would have to do if ns is a past, or in some other way unattainable, perceptual situation) in order to verify w hether ‘F’ applies: he must get himself into the situation in which ‘mo can be replaced by ‘Id , where ‘Id stands for a dem onstrative expression. But, as we have seen, the converse reference also holds: an object is only identified by a dem onstrative ‘lD’ or another deictic expression ‘ns' if there is a ‘ko such that the speaker knows that lD — k0. This too is to be understood as reference in the sense defined because, as we have seen, the correct use of ‘Flo in the perceptual situation only has the significance of a verification of ‘F ’ ’s applying to something, or of a verification of the tru th of an assertion, if the demonstrative ‘lD’ functions not as a mere ornam ent to a quasi-predicate (as in the case of F here’) but is replaceable by o ther deictic expressions and, ultimately, by an objectively locating singular term by means of which can be specified from the point o f view of an arbitrary speaker-position what it is that ‘F’ is being verified of. Every objectively locating expression (and m ediated through this, all non-locating singular terms: names and descriptions) refers to the cor­responding demonstrative expression (‘here’, ‘now’, ‘this’), because the assertion that a is F is only verifiable in the situation in which ‘a is replaceable by a dem onstrative expression. And on the other hand the use of the dem onstrative expression ‘Id’ refers to an objectively locating expression, because only then is something asserted by using the dem on­strative expression F/d, i.e. only then is something expressed that is true or false and hence verifiable.

It then emerged in the last lecture that the reciprocal dependence in the use of singular terms of various types at this most basic level of locating expressions is complicated by the fact that in addition to the

reciprocal dependence o f objectively locating and deictically locating expressions we have a reciprocal dependence o f two other types of sin­gular terms, namely those which identify spatial and tem poral positions and those which identify m aterial objects and events. This second recip­rocal dependence does not of course have the character o f a ‘reference’ in the sense just defined. Nonetheless here too it holds that we can only use the expressions o f the one type if we can also use expressions of the o ther type. T h e special complication in the ‘explanation’ of singular term s at this lowest level o f identifying expressions consists therefore in this, that we can only explain the m ode of em ploym ent (and this means: the identificatory function) of any type of locating expression if we include the employment of both complementary types, thus if we already presuppose, or simultaneously explain, their m ode of employment.

As it has not become clear how this circle is to be broken into1 I must emphasize that my analysis of identifying expressions has not achieved any real clarity and can only be regarded as a provisional sketch. I believe that it can nonetheless suffice for our purposes, i.e. for answering the question we started out from and for the sake of which I have carried out this whole discussion of singular terms. This was, firstly, the question concerning a positive alternative to the traditional understanding, which rests on the m etaphorical concept of representation, of what it is for an expression to ‘stand fo r’ an object and how the - linguistic or non-lin- guistic - reference to an object is to be understood. A nd secondly (this was really the question from which we started out) how in giving the truth-definition of the predicative form of sentence we can replace the singular te rm ’s ‘standing for’ an object by its em ploym ent-rule (analo­gously to the replacem ent, carried out earlier, of the predicate’s ‘apply­ing’), so as to arrive at an explanation o f the use of the predicative sentence as a whole or o f the word ‘true’ at the lowest level of its recursive definition.

I The analytical concept of an object

It is clear that the two questions are very closely connected, for they both depend on a positive understanding o f what it is for an expression to ‘stand fo r’ an object.

Let us start with the first question. We have to consider w hether the analyses of the actual mode o f em ploym ent of singular terms we have in the m eantim e carried out enable us to resum e the dialogue with the traditional philosopher in such a way that we do not ju st subject him to an im m anent critique but offer him a positive alternative. Let us first

Results 375

recall how far we had come before we embarked on the concrete anal­yses.

What I call the traditional position vis-ä-vis this question is the view that a sign’s standing for an object is to be understood as the assignment of this sign to that object. This view presupposes that the person who uses the sign can also refer to the object without signs, as something that he has ‘before’ him (or his consciousness); for otherwise he could not assign the sign to the object. This pre-linguistic reference to an object is in terpreted in the tradition with the help of the metaphorically extended notion of ‘representations’ (Vorstellungen). T he characteristic traditional theory of signs was a theory of representation (Stellvertreter­theorie)'. the sign represents (vertritt) something that could also be given independently of the use of this or some other sign, namely in repre­sentation (Vorstellung). I can imagine that by now you are inclined to smile at the notion of representation but that you still regard the view that the object for which a sign stands must also be capable of being given independently of signs as self-evident. But if you do then you m ust give an account, other than the representation-orientated one, of how you conceive this sign-free givenness of an object. I know of no other.

Probably to speak o f *a’ standing fo r a is already to invoke the theory of representation (Stellvertretertheorie) - ‘a’ represents (vertritt) a - and if this were so one would really have to say that the traditional view of singular terms consists in the view that their function is that of standing for objects. I have p referred to use the term ‘stand for’ in an abstract sense which does not prejudice the traditional conception. This verbal indeterm inateness seems to me to be harmless, for if one regards the statem ent that singular term s stand for objects as an answer to the ques­tion concerning their function or mode of employment then it is clear that one holds the traditional view. If on the other hand one understands the term ‘standing for’ as simply a cipher whose meaning has yet to be specified (p. 270) then it is equally clear that one regards the traditional view as simply one possible answer to the question concerning the ‘rela­tion’ of the singular term to an object.

Prior to raising the question of the actual mode of employment of singular terms I could only criticize the traditional conception by seeking to show that the metaphorical character of the extended concept of representation is uncashable. T he traditional conception could not be refuted thereby; but the readiness to consider a wholly different con­ception could be awakened.

I then chose for the developm ent of a positive alternative to what

Analysis of the predicative sentence 376

Results 377

seems self-evident to the traditional philosopher a starting point that seemed sufficiently neutral for me to be able to assume that the tradi­tional philosopher could agree to it (p. 288). The neutrality was achieved by initially leaving open the question of whether it is possible to refer to objects independently of signs. That objects are referred to by means of singular terms is uncontroversial. The traditional philosopher (and by him I always mean the traditional voice in ourselves) was simply invited to follow an unprejudiced analysis of the actual mode of employment of these signs. The traditional philosopher - at any rate one belonging to the reflective tradition of transcendental philosophy - could also accept without difficulty the principle which I made the basis of my analysis and which derived from W ittgenstein’s basic principle: that what it is in general for an expression to stand for an object is answered by explaining how it is established in the concrete case for which object an expression stands. So none of this yet contained any specifically language-analytical prejudice.

T he further step of seeing the function of singular terms in their contribution to the function of the sentences whose parts they can be (p. 289) implicitly contained the decisive turn. For if the function of singular terms is built into the sentence-whole, their function can hardly consist in the mere representation (Vertretung) of a function which can also be accomplished outside this linguistic context. Nonetheless it was still possible to think that the question of which object a singular term stands for can ultimately only be answered by means of an ostensive explanation, that is by pointing to an object. And what else can this mean but that something is pointed to that is given to us in representation? As far as I can see all schools of analytical philosophy have rem ained in this question explicitly or implicitly within the traditional point of view.

The first real blow to the traditional conception came when it emerged that the function of singular terms is that o f specification. If the ‘relation’ of the singular term to the object for which it stands consists in this, that by means of it is specified which of all is m eant (p. 293), then this ref­erence (.Beziehung) to all entails that it cannot be a mere relation between the sign and its object; and hence that this reference cannot be construed as assignment. However, this reference to all remained initially unclear. This much of course seemed clear: we only know which of all objects of a domain is meant by a singular term ‘a if we have a criterion for decid­ing w hether a =b or a ^ b , where ‘b’ is any other singular term with which an object of this domain is specified. But it was still not clear how deeply the identity-relation reaches into the specificatory function itself. One could admit (this opinion is probably widespread in analytical phi­

Analysis of the predicative sentence 378

losophy) that it belongs to the sense of the use of expressions for the specification of objects that one possesses a criterion for replaceability by other expressions of the same domain, but still be of the opinion that this is merely an additional aspect which is added to the ‘standing’ o f the expression ‘for’ an object but which does not determ ine the sense of this ‘standing for’.

This view is only tenable so long as one believes oneself able to assume that for every object-domain there is a pre-em inent type of singular term which is such that, whereas in the case of all other singular terms the question ‘Which object is m eant by “a,”?' is answered by an identity- statem ent the second term of which is ultimately a singular term of this pre-em inent type, in the case of this pre-em inent type itself the question ‘Which object is m eant by “a”? ’ is no longer to be answered by an identity- statem ent but by assigning “a” to the object, which must therefore be accessible independently of signs. T he role of functioning as this p re ­em inent type of singular term, which is supposed to refer directly to the object, has been ascribed both to proper names and to dem onstra­tives. But, as we have seen, both views are untenable. Not only do they contradict the actual m ode of employment o f these types of expression; one cannot even invent a type of singular term - as e.g. Russell did with his logically proper names - capable of fulfilling the postulate of the theory of assignment.

We are now in a position to see wherein the decisive result of the analysis of the actual m ode of employment of singular terms lies: in the fact of reciprocal reference that obtains between those types o f singular term to which the other types of singular term refer. We have seen that this reciprocal reference is not a merely contingent characteristic of our language but is grounded in the essence of the supplem entation of p e r­ceptual predicates to form expressions with which something can be said that can be designated true or false. O ur finding that there is no singular term that does not refer to others means that every question of the form ‘which object is meant by ‘V 7 ’ is answered by another sign that can be substituted for V. This does not lead to a vicious circle in which one sign is arbitrarily substituted for another. Rather the rule- governed substitution of singular term s for one another is the condition of the possibility of a being whose perceptual position constantly changes being able to refer from any position to any other perceptual situation. T he latter can only be held on to as identical if, in accordance with changes in one’s own position, o ther terms are being substituted. This concerns at first the substitution of subjectively locating expressions for one another. But in addition to this, as we have seen, the subjectively

Results 379

locating expression m ust be replaceable by an objectively locating one. It could appear as though the objectively locating expressions themselves require no substitution; so that one m ight think that here we have the pre-em inent type o f singular term postulated by the traditional theory which does not refer to any others but is directly assigned to an object. But, firstly, an objectively locating expression identifies an object not by being assigned to it but by specifying its spatio-tem poral relations to other objects. Such an expression does not function as representative (.Stellvertreter) for a reference to the object that is also conceivable without the sign. Secondly, though it is true that an objectively locating expression stands once and for all for the same object, if the speaker did not know which subjectively locating expression he can substitute for it (and this will always be different depending on his own situation) then this would not be an identification o f the object for him; and it is clearly senseless to speak o f an identification o f an object which is not an identification

f o r someone. Replaceability by objectively locating expressions ensures that the identification is an identification/or everybody; but it is so only if everyone can replace the objectively locating expression by a subjectively locating one. T he reference o f every ‘m0’ to an ‘n 8*, and vice versa, does not give rise to a vicious circle for these two types o f reference are the mutually supplem enting sides o f the identification in which every speaker can specify for every other (but also for himself) which verification- situation o f a perceptual predicate he means.

If we now return to the principle conceded to us by the traditional philosopher, that the question o f what it is for an expression to stand for an object is answered by explaining how it is established in the con­crete case for which object an expression stands, then we reach the result that this establishing is never accom plished by assigning the expression to an object (which in that case would have to be som ehow given inde­pendently o f signs) but always in accordance with the reference-rules which hold between the various types o f singular term. We explain to som eone what it is for an expression to stand for an object (of an object- domain) by demonstrating to him the reference-rules (Verweisungsregeln) which govern the expressions that supplem ent the predicates (of this object-domain).

With this we have now achieved for singular terms too what I have several times called a ‘specifically language-analytical position’ (cf. p. 158, p. 242), viz. what is achieved by a certain use o f signs does not m erely take the place o f som ething that could also be achieved without the use o f these signs; rather it requires this use o f signs. In the present case this means that the reference to an object - perhaps I should say

Analysis o f the predicative sentence 380

m ore concisely: to an individual - which is made possible by the use o f singular terms is simply not conceivable without the use o f such expres­sions and, in particular, o f the identity-sign.

This then is the outcom e o f the dialogue with the traditional philos­opher which started from the dem and that he should first agree to an analysis o f only that reference to objects which is achieved by the use of signs which stand for the objects. Assuming that the traditional philos­opher accepts this result as far as it goes he can still say that this analysis leaves untouched that reference to objects which is achieved indepen­dently o f signs. Formally this is incontestable, if one can also say what one is to understand by this ‘reference to objects’ if it is not that kind of reference one can explain in terms o f the use o f singular terms. Should one seek to explain what is meant by saying that it is the reference to objects that is accom plished in representation (Vorstellen) then clearly one unintelligible thing is being explained by another. Such a view cannot be refuted. But it does not need to be; for it can do no damage. So I think we can summarize the result by saying that there is no such thing as a sign-free reference to an object (an individual).

It would be a com plete m isunderstanding were you to suppose that this m eans that the rule-governed use o f signs would somehow take the place o f objects, as happened with predicates. This misunderstanding would correspond to the prejudice that linguistic analysis is concerned with linguistic usage rather than with things (die Sachen). Faced with this objection one m ust always ask what is m eant by ‘things’. Here it can only mean the objects in space and time. But not only do these objects remain intact; it is only the m ode o f em ploym ent o f singular terms, as this has been presented, that makes it intelligible both that and how one can mean an individual spatio-temporal object. T he rule-governed use of these signs does not take the place o f these objects; rather it takes the place o f a fictitious sign-free reference to these objects. In fact it was this reference, interpreted as representation, which replaced spatio-temporal objects by som ething else, namely by representations.

It is im portant to see how far the recourse to the use o f signs has a systematically different significance in the language-analytical destruction o f the traditional conception o f singular terms than it has in the lan­guage-analytical destruction o f the traditional conception o f other lin­guistic expressions, e.g. predicates. It was characteristic o f the traditional conception o f all linguistic expressions that it did not raise the question of the m ode o f em ploym ent o f a sign, but instead just took it for granted that every sign stands for an object. Now in the case o f singular terms the language-analytical recourse to the m ode of em ploym ent has the

Results 381

significance o f explaining the sign-object schema, whereas in the case o f the other expressions it has the significance of preven ting the sign - meaning schema from being replaced by the sign-object schema. In this second case, that o f predicates, for exam ple, the possible objectual ref­erence to som ething for which the predicate stands, the reference to an attribute, is indeed based on the m ode o f em ployment o f the sign. We can now formulate this more clearly: the identity-criterion for a = b , where ‘a and ‘b ’ stand for attributes, consists in the predicates whose nominalization yields the expressions ‘a and ‘b’ having the same mode o f em ployment (synonymy). On the other hand, i f ‘a’ and *6’ stand for spatio-temporal objects, then the identity-criterion for a = b is the identity o f the spatio-temporal relations o f a and b. Thus in the case of spatio- temporal objects the identity-criterion does not concern the mode o f em ploym ent o f certain signs. To specify which is the object meant one does not, as with attributes or states o f affairs, point to the m ode o f em ploym ent o f a linguistic expression, rather one locates the object in space and time. This locating, however, can only be achieved by means o f the rule-governed use o f a species of linguistic expression.

With this I can now conclude the elucidations regarding the outcome o f our analyses for the first o f the two questions from which we started, viz. for the question o f what it is for a singular term to ‘stand for’ an object. In the case o f singular terms that stand for perceptible objects one knows for which object the expression stands if one knows which object it identifies; and one knows that if one knows to which other singular terms it refers.2 Since in the definition o f ‘ “a” refers to “b” ’ singular terms already appear as com ponents o f predicative sentences and reference is made to the verifiability o f these sentences (above p. p. 373) the implications for our second question - the consequences for the truth-definition o f the predicative form o f sentence - can be directly attached to the foregoing results.

II The mode o f employment o f predicative sentences and the explanation o f the w ord ‘true’

T h e truth-definition for the predicative form o f sentence was (p. 253):‘The sentence “F a” - or the assertion that a is F - is true if and only

i f the predicate “F ” applies to the object for which the singular term “a ” stands.’

Our problem was: how can we, in place o f this meta-theoretical spec­ification of the truth-condition, dem onstrate a verification-rule in such a way that we thereby explain how sentences o f the form ‘F a ’ are used?

Analysis o f the predicative sentence 382

We had seen in a general way for all assertoric sentences that they are used to assert something: whoever uses such a sentence guarantees that certain truth-conditions are fulfilled (Lecture 15). From this it followed that one explains the use o f a form o f sentence by showing by means of exam ples, not under what conditions a sentence o f this form is used, but under what conditions it is confirmed or refuted; and this shows itself in the conditions under which it, or its negation, is to be withdrawn. T he peculiar difficulty o f explaining the predicative form o f sentence consisted in this, that (1) whereas the explanation o f the truth-condition o f the higher-level forms o f sentence could fall back on the word ‘true’ as applied to more elementary sentences, this is not possible in the case of elementary sentences whose truth does not depend on the truth of other sentences; (2) that we must assume that in contrast to the unitary verification-rule o f the higher-level forms o f sentence the truth o f pre­dicative sentences depends on two rules, an em ploym ent-rule o f pred­icates and an em ploym ent-rule o f singular terms. If we form ulate this at the level not o f the verification-rule but of the truth-condition then this state o f affairs can be expressed thus: whether the sentence ‘Fa is true depends on (a) for which object ‘a ’ stands and (b) whether the predicate F ’ applies to this object. We can now expect that in relation to employment-rules this state o f affairs is to be construed thus: whoever uses a sentence o f the form ‘F a guarantees that if the em ploym ent-rule of the singular term is followed the predicate can be correctly applied in accordance with its em ployment-rule to the result o f following this rule o f the singular term.

This provides us with a framework into which we have to insert the two rules. T he transition to em ployment-rules is accomplished in every case by the question: how is it to be established that -? The question: how is it to be established that a sentence o f the form ‘Fa is true? is answered by showing how it is to be established that the predicate *Fy applies to an object and how it is to be established for which object of perceptual predicates the singular term V stands. Now in the case of predicates we have seen that the criterion of our knowing how it is to be established that a perceptual predicate F ’ applies to an object is that we know how the truth o f the dem onstrative expression F m D’ is to be established in arbitrary situations. And correspondingly we have seen in connection with W iggins’ definition that the criterion o f our knowing how it is established which object a singular term ‘a’ identifies is that we know how it is established that ‘cf>a’ is true where for ‘cf)’ an arbitrary perceptual predicate is to be inserted, a perceptual predicate that we already understand in the sense o f the explanation given in the previous

Results 383

sentence; and this means: whose verification-rule we know. With this explanation o f the singular term we can now generalize the definition o f the criterion o f the understanding o f predicates by dropping the restriction to demonstratives: the criterion o f our understanding a (per­ceptual) predicate ‘F ’ is that we know how one establishes that F m is true where for ‘m an arbitrary singular term can be substituted which we understand in the sense o f the explanation given in the previous sentence; and this means: whose identification-rule we know.

However, these explanations do not yet give us what we are looking for, for they explain the use o f the com ponent-expressions in such a way that they assume that we already know how one establishes that a predicative sentence is true; whereas it is precisely this that is to be explained. But having thus brought to light the rules o f the com ponent- expressions involved in establishing the truth o f a predicative assertion we will simply have to use these rules whose explanation we already know - the verification-rule o f the predicate, the identification-rule o f the singular term - for the explanation of the truth o f the predicative assertion. T h e abstract explanation which has just been given o f what it is that is guaranteed in using a predicative sentence - an explanation which did not yet take into account the peculiarity o f the two rules - can now be rendered concrete in the follow ing way: it is guaranteed that if one has iden tified the object - and that means ultimately: the verification-situation - in accordance with the em ploym ent-rule o f the singular term, then the predicate can be correctly applied in this situation in accordance with its em ploym ent-rule. But this can equally well be inserted into the truth-definition as a truth-condition: a sentence ‘F a ’ is true if and only if the predicate ‘F ’ can be correctly applied in the situ­ation identified by ‘a .

I think it will be helpfu l before taking the final step to bring out certain basic features o f the conception which now em erges. Let us first get clear about how it differs from the traditional theory o f judgm ent and truth. T h e latter, which was exclusively orientated towards the elem entary case o f the predicative sentence, had, it is true, already rec­ognized that the condition o f the possibility o f a linguistic expression (or its psychological or ontological counterpart) being called true or false is its structure (above p. 253). This corresponds to the fact just em phasized that the truth o f a predicative sentence essentially depends on two rules. In the tradition, however, due to its object-orientation, the connection o f these two factors was interpreted as synthesis, the connection o f two objects. So one was unable to see that the two com ­ponents o f the predicative sentence have d ifferent and, in their d iffer­

Analysis o f the predicative sentence 384

ence, com plem entary functions. In the meantime we have seen that the fact that one com ponent has an identificatory function must be regarded as just as much a condition of the possibility o f being able to speak of truth and falsity as the fact that there are two components.

We had already seen (p. 253) that this weakness in the traditional conception is already overcom e in the first, still meta-theoretical, step in the specification o f the truth-condition - thus in the truth-definition. According to the truth-definition truth depends on two factors: firstly- as presupposition - on which object the singular term stands for and, secondly - as condition - on whether the predicate applies to this object.

Thus it could already becom e clear at this level that there are two different factors involved. But how these two factors, or the two semantic expressions ‘stands for’ and ‘applies’, are connected cannot be made clear at this level, but only by asking how it is established for which object the singular term stands and how it is established that the predicate applies to it. Even then o f course the clarification o f this connection does not follow automatically but only by making clear to oneself that the identification o f the object consists in the identification of the situ­ation in which it can be established whether the predicate applies to it; and to get this clear our analysis o f singular terms was necessary.

This essential complementarity o f identification-rule and verification- rule also explains how the function o f the singular term in the sentence is dependent on the com plem entary function o f the predicate and yet in a certain sense is independent o f it. This is the problem o f the asym­metry between singular term and predicate to which I had referred at the beginning o f the discussion o f singular terms (p. 267f). In asking in regard to a predicative sentence ‘F a ’ whether the predicate applies to the object we must already have identified the object before we can exam ine whether the predicate applies to it. And although we do not need to have identified it we must have specified it in order even to be able to form ulate the question o f whether the predicate applies to it. Whereas the identification or specification o f the object must be achieved independently o f whether the predicate applies to it. For this reason it seems more correct if instead o f saying that the truth o f the predicative assertion depends on two factors - (1) for which object the singular term stands and (2) whether the predicate applies to it - we say, with Strawson, that it only depends on the second factor, whereas the first represents the presupposition o f being able to form ulate the truth-con- dition at all. This also explains why the verification-rule of the predicative sentence is not based on two partial verification-rules but only on one. T he identification-rule is not a verification-rule; rather following it con­

Results 385

stitutes the presupposition for the application of the verification-rule of the predicate (in that a verification-situation is thereby picked out). The fact however that the following o f the specification- and ultimately the identification-rule always precedes the application o f the verification- rule does not mean that the specification o f which object is meant has an independent significance apart from the function it has in the veri­fication o f predicates.

We can now move on to the final step in our analysis. The reason why the formulations o f the truth-condition o f predicative sentences cannot be regarded as an explanation o f the m ode of em ployment o f these sentences is clearly that semantic expressions like ‘identification-rule’ and ‘verification-rule’ still occur in them. What has so far been achieved is that in the truth-definition the expressions ‘stand for’ and ‘applies’ no longer occur; instead reference is made to the employment-rules o f the two com ponent-expressions which are to be regarded as the rules con­cerning how one establishes for which object the singular term stands and how one establishes that the predicate applies to an object. What is still lacking is an explanation o f what is meant by ‘verification-rule’ and ‘identification-rule’, or the replacement o f such locutions by the description of the actual m ode o f em ployment o f the two kinds of expression. T o replace the reference to rules in this way is the same as to explain them; for what is meant in general by such a rule is explained by demonstrating by means o f examples how the use o f expressions is explained in the particular case in accordance with such a rule.

But we already possess a description o f how the use o f a particular predicate (and that means: its verification-rule) is explained, and likewise of how the use o f a particular singular term (and that means: its iden­tification-rule) is explained. For predicates I already gave such an account before we started the discussion o f singular terms (p. 264f). And we have seen in the last few lectures (with o f course the defects which I pointed out today) how the identification-rules o f singular terms o f the different types are explained. In both cases it became apparent that the explanation o f an expression o f one of the two kinds (singular term or predicate) essentially presupposes that the expression is understood as an expression that supplements expressions of the other kind and that the mode o f em ployment of expressions o f the other kind is already known. Thus we understand a singular term of a certain type only if we know to which other types it refers (verweist)', and this implies a ref­erence (Bezugnahme) to the verification o f perceptual predicates. On the other hand, in describing the explanation o f predicates one could not avoid anticipating what is essential to the explanation o f singular terms

Analysis of the predicative sentence 386

- their character o f referring to other singular terms. This reciprocal presupposition does not represent a vicious circle but is precisely what we should expect if the two kinds o f expression are essentially com ple­mentary expressions.

Besides, one can think o f an explanation o f the expressions o f the two kinds in stages that makes clear that there is no circle in the expla­nation: Stage 1: classification-expressions are explained as quasi­predicates. Stage 2: the classification-expressions are supplem ented by quasi-demonstrative expressions which do not yet refer to other expres­sions. The m ode o f em ploym ent o f the resulting expressions F m o (‘this is F \ ‘here F \ ‘now F*) does not yet d iffer essentially from that o f ex ­pressions o f the first stage. *F* is not a predicate; ‘m D’ is not a singular term (cf. p. 263). Stage 3: the use o f is explained by dem onstrating its systematic replaceability by other deictic expressions in connection with the change in the speaking-situation. If the expression ‘m ^ is used in accordance with this explanation then it has an identificatory func­tion: it is a singular term. This explanation then autom atically has the consequence for the supplem entary expression *F* that it is no longer understood as a quasi-predicate but as a predicate. Although ‘F ’ is still explained in exactly the sam e way as at the first and second stages what is now explained is no longer the situation-related em ployment-rule o f ‘F ’ but its verification-rule and hence the situation-free em ploym ent- rule o f ‘F ’ as a predicate, i.e. its m ode o f em ployment in combination with other than demonstrative expressions (cf. p. 265):

This explanation of the replaceability o f the demonstrative expression by other deictic expressions implies (1) that if the utterance ‘F m D’ in the one situation can be described as correct one can also say o f the utterance ‘Fris in the other situation that it is correct and vice versa.3 T o understand this is to understand the m eaning o f substitutability - or the expression ‘ = \ But to understand the systematic connection o f demonstrative and deictic expressions and hence the use o f the corresponding predicative sentences one must also understand (2) that ‘n s’ ‘refers’ to ‘m D\ This not only means that if the expression F m o is correctly used in its situation then the expression F n s’ is correctly used in its situation and vice versa, but also that the correct situation-related use o f ‘F m D’ is the criterion o f the correct situation-free use o f *Fn8y. This reference of the correct use o f one expression to that o f the other can be explained (as is the case with all forms o f assertoric speech) by dem onstrating the corresponding verification-game by means o f exam ples, e.g. (1) A says F n s\ B denies it (or calls A ’s utterance ‘not correct’). (2) Both get into the perceptual situation in which they can say ‘mD is the same as that which we previously

Results 387

designated with n s\ (3) T ogether they determ ine whether F m D\ or ‘F \ can be used in this situation in conform ity with the rule. (4) D epending on the r e su lts or B says that the utterance o f the other at stage 1 was correct and that his own was not correct.

With this further explanation, which presupposes the preceding one (the explanation o f * = ’), the word ‘true’ has been explained at a most elem entary level o f elementary (predicative) assertions. For the word ‘correct’, as it was used in these two explanations, no longer has the meaning o f ‘conform ing to the rule’ (except in step (3) o f the verification- game). T hat the word ‘correct’ in steps (1) and (4) o f the verification- game has a m eaning other than ‘conform ing to the rule’, for which therefore a new word - ‘true’ for exam ple — is needed , can not merely be asserted from without; it shows itself in the way in which this cor­rectness is established (checked). Determ ination o f correctness in the sense o f rule-conform ity and determ ination o f correctness in the sense o f truth now become separate (cf. 3 5 4 f). O ne shows how correctness in the sense o f truth is established by means o f the verification-gam e that has just been dem onstrated. W hether som ething true is said with ‘F n sy in a situation other than the perceptual situation is established by recourse to the perceptual situation in which the same thing is said with ‘Fm o- On the other hand, whether som eone is using F m n or the com ponent expression ‘F ’ in ‘F m ^ (but also in ‘Fra/) correctly, in the sense o f ‘in conform ity with the rule’, is determ ined by exam ining how he uses the same dem onstrative expression ‘F m D’ in other situations. Only in the spe­cial case o f the use o f (F m D’ (thus in step 3 o f the verification-game) is the characterization o f the use as correct am biguous. But this ambiguity in the use o f an expression in that situation in which truth is tested is (a) necessary and (b) harmless. It is necessary because the testing of truth presupposes that the expressions are used in conform ity with the rule. It is harmless because the characterization o f the use o f the expression as correct is only ambiguous so long as it is viewed in isolation. T h e characterization o f a use o f F m o as correct/incorrect has the mean­ing o f true/false in so far as this characterization entails the correspond­ing characterization of the other expressions F k s\ ‘F ls’, etc., with which the same thing is said (and this also m eans that what is called ‘correct’ [‘true’] is not the use o f the expression but what is said by m eans of the use o f the expression and, what is the sam e thing, what is said by means o f the other expressions in so far as they are used in the corresponding situations). On the other hand, the characterization o f the use of F m ß as correct/incorrect has the m eaning o f ‘in conform ity with the rule’/‘not in conformity with the rule’ in so far as this characterization is confirmed

Analysis o f the predicative sentence 388

or disconfirm ed by the m ode o f em ploym ent o f the same expression in other situations.

O f course as long as the verification-game is only demonstrated in terms o f exam ples o f the kind just described it can only explain the word ‘true’, even as restricted to predicative sentences about perceptible objects, at a m ost elementary level. But to this first level the others can be easily attached. I can restrict m yself to indications:

(1) That to identify the intended object demonstratively is already to identify the verification-situation of the predicate holds universally only for spatio-tem poral positions. For material objects it only holds in the case o f predicates o f which it is presupposed that they apply to the object during the whole duration o f its existence and everywhere uni­formly. We cannot decide whether the assertion that the Neckar flows into the Rhine is true simply by perceiving it at som e point o f its course and referring to it with a dem onstrative expression. However, we know on the basis o f the identity-criterion o f spatio-temporal continuity which holds for material objects how we must proceed in order to identify that section o f this object which represents the verification-situation o f the predicate. What we have here then is simply a more complicated form o f step 2 in the verification-model I have just presented.

(2) A more complicated form o f step 3 in the verification-model results when we consider that even perceptual predicates never have so simple a verification-rule that the assertion that such a predicate applies to an object can be decided by a single act o f perception. That the table at the door o f the lecture-room is red cannot be conclusively verified by looking at it. It is possible that further acts o f perception will reveal that it only seem ed red because o f the red illumination or because one only per­ceived a part o f it on which there was a red spot, etc. T o the contrast between false and true there also belongs this special contrast between appearance and reality. However, it was correct to ignore this com pli­cation at first for one can only explain perceptual predicates in this sequence. Only the simple predicate F ’ can be explained on the basis o f the quasi-predicate F ’; and only if one already understands the simple predicate F ’ can one learn the contrast between ‘it is really F* and ‘it seems to be F ’.4 This explanation is achieved by showing by what further perceptions an assertion already supported by perception can be dis­confirm ed or further supported. Thus in place o f the first, simple deci- sion-procedure one learns a graduated decision-procedure for deter­m ining the truth o f an assertion. This complication essentially concerns the predicate; however as regards the singular term it must be ensured that although the perceptions are always new the same object is still

Results 389

meant; and this presupposes that the previous point has already been taken care of.

(3) T he verification-model is also incomplete in so far as I have restricted it to deictic singular terms. However, the extension to other singular terms follows automatically from what I have said about the reference-connections between the different types of singular term.

However, there are some complications here which I must at least indicate. If the singular term V in the utterance at stage 1 of the veri­fication-game is not a deictic expression then its replacement by a demonstrative expression ‘mo at stage 2 rests on the no longer analytic but empirical truth o f ‘n = m D’. But in that case one can no longer say that what is said with TV is the same as what is said with F m D\ for it is possible for som eone to hold what is said with *Fmß to be true and what is said with TV to be false or vice versa (that is if he does not believe that n — m D). We here come up against the question o f the identity- criteria o f states o f affairs or assertions. T he view that two sentences F n and ‘G m ’ stand for the same state of affairs if and only if their com ponents are extensionally equivalent, thus if F ’ applies to the same objects as ‘G’ and n = m, has also been maintained;5 but it hardly cor­responds to our ordinary understanding. T he usual view is that the two sentence-components must be intensionally equivalent,6 thus that, in' particular, ‘m = n ’ is analytically true. And it seems appropriate to accept the even narrower criterion o f Frege that we believe that what is said with an utterance is the same as what is said with another utterance if and only if it is not conceivable that somebody should hold the one to be true and the other false,7 thus in particular if it is not conceivable that the speaker should hold ‘m ~ n to be false. Whichever o f these two con­ceptions we adopt, we shall have to draw the boundaries o f the identity o f what is being asserted more narrowly even where both singular terms are deictic expressions. I f today I say ‘today’ and mean by this the same day which some time ago I meant when I said ‘tomorrow’, I can nonetheless be mistaken in the belief that it is the same day (for example if I have slept through a day). Even here then the presupposed identity-sentence is empirically falsifiable; thus what is asserted in the two cases is at least not necessarily identical. It is different if one says: ‘The day which I mean when today I say “today” is the same as the day which I shall mean when tomorrow I will say “yesterday”.’ This is analy­tic. So it also follows that what I assert today with the sentence ‘Today it is raining’ is the same as what I would assert tomorrow with the sentence ‘Yesterday it was raining.’8 Here an error is impossible because the statement that I have just made no longer concerns the question

Analysis o f the predicative sentence 390

whether what is being said in different situations is in fact the same; rather this statement only claims that to express what is being asserted from the perspective o f one situation from the perspective o f other situations certain other expressions have to be used. Now the em ployment-rules o f these different expressions, and with them the analytical knowledge that in every other situation one can assert the same thing only with the corresponding other expression, are presup­posed by the possibility that what is being asserted in the second situa­tion can turn out as an empirical fact not to be the same as what had been asserted in the first situation, although it was meant to be the same. It is these identity-sentences both o f whose terms are deictic expres­sions, and which are analytic, through which at the level o f assertions there is constituted over against the fleeting utterance (the act o f saying) som ething identifiable, som ething said, to which one can com e back as the same from any other situation and which is a possible bearer o f the predicates ‘true’ and ‘false’. In the other cases, however, in which in the verification-game we have to assume the truth o f empirical identity- sentences in order to ensure that both outside and inside the verification-situation we mean the same spatio-temporal object, this (extensional) identity o f the spatio-temporal object does not entail the (intensional) identity o f what is asserted. Thus in verifying an assertion whose singular term is not a deictic expression we have recourse to another assertion. Since however in the verification-game the truth o f the identity-sentence (‘n = % ’) is assumed by both partners, what is as­serted with ‘F n ’ turns out to be true, or false, if and only if what is asserted with ‘F m ^ is true, or false (cf. above p. 386f), although the two assertions are not identical. More than this is not required by the verification-game.

L E C T U R E 28

The next steps

T he analysis o f the predicative form o f sentence was only intended as a first step in an enquiry into foundations that concerned the m eaning o f all linguistic expressions; and that means: the semantic forms o f all sen­tences. Measured against this aim we have not achieved much. Firstly we have not even arrived at a general theory o f singular terms nor, consequently, at a general theory o f predicative sentences. T he next step would be an analysis o f those predicative sentences whose singular terms stand for abstract objects. Secondly, the conception thus far reached o f the m ode o f em ploym ent o f assertoric sentences would have to be widened into a general theory o f all forms o f assertoric sentences. A nd, finally, we would have to rem ove the restriction to assertoric sen­tences so as to arrive at a general theory of all sentences with proposi- tional content.

Each o f these three steps relates to a new and extensive field o f inves­tigation and in none o f them, least o f all in the third, can we assume that the conceptuality so far worked out can be simply transferred to the extended field o f investigation. Thus in these steps the field o f investigation would not simply be extended; rather each o f the three steps would represent a further step in the enquiry into the foundations. In each o f them one would be concerned with a reexamination o f the previous conceptuality and possibly with the working out o f a new, more fundam ental conceptuality.

T he three steps would not necessarily follow one another. The analysis o f those predicative sentences in which abstract objects are referred to is not a presupposition for the analysis o f the other forms o f assertoric sentence; and equally the investigation o f non-assertoric kinds o f sen­tence does not have to wait until we are in possession o f a com plete semantics o f assertoric sentences. T h e three steps thus do not constitute a series; rather they represent so to speak three directions in which one

could continue from the stage we have now reached. This shows that this first step, even if it does not go far enough, is nevertheless funda­mental for the further enquiry into the foundations.

What is characteristic o f this first step is the dissolution o f the tradi­tional conceptuality (which it would now be better to call representa­tion-orientated than object-orientated.) We were able to establish a new, specifically language-analytical formal conceptuality with which we were able to transcend, but also re-interpret, the basic formal concept o f tra­ditional philosophy: the concept o f an object. T he traditional idea o f a language-free subject—object relation was thereby shown to be devoid o f meaning. There is no such thing as a reference to an object that is detached from a context o f sentences. The use o f signs thus acquires a fundam ental significance for the theory o f consciousness: linguistic signs are not representatives o f other functions which would also be possible without them. Hand in hand with this ‘up-grading’ o f signs goes the new conception o f their m ode o f em ployment. Whereas according to traditional semantics one understands a sign - any sign - if one knows what it stands for, according to the language-analytical conception one understands a sign - including a sign that stands for an object - if one could explain its m ode o f em ploym ent to som eone who does not yet understand it. And in the particular case of the com ponents o f predic­ative sentences this means: explain their identification-rule or verifica­tion-rule; and that means: explain the contribution they make in estab­lishing the truth o f sentences into which they can enter as components.

This characterization o f the stage now reached makes it clear that from it one could begin with each o f the three next steps to which I have referred; for it contains perspectives for each o f these steps. With respect to the first step this is obvious. For the second step it contains the hypothesis that o f every meaning-bearing part o f an assertoric sen­tence we can say that we understand it if we know what contribution it makes with regard to the determ ination of the truth of sentences into which it can enter as a part. As regards the third step, the analysis o f the meaning o f non-assertoric sentences, we cannot, to be sure, expect that their m ode o f em ploym ent can be explained by the demonstration o f verification-rules. H owever, we have no reason to doubt that here too we can proceed on the basis o f W ittgenstein’s principle, and that if we are enquiring about their m eaning we must ask for the explanation o f their mode o f employment.

T h e task o f this final lecture is to keep alive for you, and for me, the dynam ic o f the enquiry beyond the path traversed so far. It is charac­teristic o f a philosophical enquiry into foundations that all answers are

Analysis o f the predicative sentence 392

The next steps 393

merely steps to new questions. A philosophical line o f thought can thus only be concluded as a philosophical line o f thought by making oneself aware o f the new difficulties into which it leads. It is in this sense that the following indications about the next three steps are to be understood.

For the first o f these tasks, a general theory o f the predicative sentence, we are not without a hint. When before the clarification o f the mode of employment o f concrete singular terms I enquired about their fun ction the argument was still completely general (pp. 289-95). It em erged that the function o f a singular term, whether concrete or abstract, is to specify which o f all is meant; and this means: which object it is that is classified by the supplem enting predicate. I called this function ‘specification’. We then saw that in the case of concrete singular terms we must distin­guish between specification in a wide sense and specification in a narrow sense (‘identification’). We can accept as a hypothesis that an analogous differentiation also holds for abstract singular terms. The notion of abstract singular terms (or abstract objects) is defined negatively. They are singular terms with respect to which the question ‘which is meant?’ can not be answered by a spatio-temporal identification. Since every sin­gular term has the function o f specifying which o f all is meant the use of an abstract singular term also presupposes an underlying object- domain, a totality o f objects o f a kind from which one is picked out by the singular term. ‘O f a kind’; that means here: a totality o f objects that can be distinguished from one another and for which, therefore, there is a criterion o f identity. T o give the criterion o f identity would be to give a positive definition o f the object-domain in question. Thus it becomes apparent that the negatively defined concept o f abstract objects is a collective concept which covers different object-domains with distinct criteria o f identity. For exam ple, if ‘a ’ stands for a proposition, ‘b ’ for a number, and V for a spatio-temporal object, then it is as senseless to ask whether a = b as to ask whether a = c . ‘x = y ’ is only m eaningful if one has a criterion for deciding whether x —y.

T he first step - one widely neglected in the literature - towards an analysis o f abstract singular terms1 would consist in a comprehensive compilation o f the different domains o f abstract objects. By way of illus­tration I shall simply mention some important examples: (1) Attributes (for which nominalized predicates stand). (2) States o f affairs (for which nominalized assertoric sentences stand). (3) Types of signs in contrast to the corresponding sign-events (cf. above p. 218). Abstract objects o f a similar kind are, e.g., geometrical figures, tones, melodies, pieces of music. (4) Institutions and their parts, e.g. games and moves in a game. (5) Classes. (6) Numbers. A peculiar intermediate position between con­

Analysis of the predicative sentence 394

crete and abstract objects is occupied by concrete institutions such as firms or states which in a certain way are spatio-temporally locatable. T he listed cases are in part themselves collective concepts, e.g. what is indicated under (3) and (4) clearly embraces several object-domains.

Since singular terms are expressions that are essentially com plem en­tary to predicates the domains of objects must be distinguished not only by their identity-criteria but also by the predicates that can be applied to them. The clarification o f a domain o f objects, or o f the use o f the corresponding singular terms, would therefore really consist in a clari­fication o f the relevant whole predicative sentences; and this means: in the clarification o f how these sentences are used; and this means: how they are verified. By means o f this enquiry into the verification o f sen­tences one also reaches a decision concerning the question o f whether one is dealing with genuine objects, with a genuine distinctive domain of objects. If in asking about verification one finds oneself referred to other, but equivalent, sentences about concrete objects, in which the abstract singular terms only appear in modified form as predicates, then one will have to answer this question in the negative. This question o f the relation o f sentences about abstract objects to sentences about con­crete objects and about the possible reducibility o f the former to the latter arises separately for each o f the cases just listed.

An extrem e case is that in which abstract singular terms can be used, so to speak, in isolation, thus where one cannot specify a corresponding domain o f objects.2 For exam ple, if one says ‘T he lack o f vitamin C causes scurvy’ the question ‘Which o f all is meant by “the lack o f vitamin C”?’ is clearly idle. We would not know with which predicate designating an object-domain we would have to supplem ent the word ‘all’. But the question does not even arise, for it has no bearing on the question of the verification o f the sentence. One clearly does not verify the sentence by exam ining an object that is identified by ‘the lack o f vitamin C’ to see whether the predicate of this sentence applies to it, but rather by inves­tigating whether the sentence ‘Most people who lack vitamin C get scurvy’ is true. So of those substantival expressions which are immediately elim­inated when the question o f verification is raised one will have to say that they are not really singular terms, because they are not em ployed as such.

If all abstract singular terms were eliminated in this way in the verification-question then we would not need to speak o f abstract objects at all. In the case o f those object-domains which turn out not to be eliminable, however, the question arises: why are they not eliminable? How is it that we can, and possibly must, speak o f abstract objects o f

The next steps 395

such and such domains? How are such ways o f speaking constituted and what would be lost without them? This is the question about abstract objects which would correspond to that question about concrete singular terms which I called the question ‘from below ’. Just as then the question was set against the foil o f a poorer language which did not yet contain any singular terms, so it would now be posed against the foil o f a poorer language, namely, that which only contains concrete singular terms.

On the basis o f these clarifications one could now resum e in the required generality the enquiry into what it is to refer to an object. How far the results o f our analyses o f the m ode o f em ploym ent of concrete singular terms could be generalized is com pletely open. For precisely that result that was decisive for the confrontation with the traditional conception, namely that every singular term ‘refers’ (verw eist) to others (p. 373ff), seems not to apply to the abstract domains. T h e reciprocal reference characteristic o f the identification o f concrete objects was, after all, merely the consequence o f the special circumstance that the objects concerned were perceptual situations to which it had to be possible to refer from every other perceptual situation. In the case o f abstract objects this complication does not apply. Does this m ean that for abstract objects the traditional representation-theory would be vindicated? Or must we say that in their case too there can be no question o f a simple assignment (and this would mean: to a representation) though for differen t reasons? Whatever the answer may be it is at any rate clear that we must anticipate a thorough-going revision o f the conception o f reference to objects developed so far.

Now to the second step, the project o f a general theory o f the semantic forms o f assertoric sentences. Do we have a concept o f ‘semantic form ’ that is determ inate enough to enable us to form a definite idea o f such a project? I introduced the term ‘semantic form ’ by saying that two sentences have the same semantic form if their m eaning depends on the m eanings o f their m eaning-bearing com ponents in the same way; and that two m eaning-bearing sentence-com ponents have the same semantic form, or belong to the same sem antic class, if they contribute in the same way to the m eaning o f a sentence (p. 26). In the meantim e it has em erged that one understands an assertoric sentence if and only if one knows its truth-conditions, or knows how one establishes whether it is true or false. It follows from this that we understand a sentence- com ponent if we know what contribution it makes in establishing the truth o f sentences into which it can enter as a com ponent; and that we understand the semantic form o f this expression if we understand in what way it makes this contribution. O ne can therefore, with Davidson,3

Analysis o f the predicative sentence 396

call the semantic form o f a sentence - its ‘meaning-relevant’ composition- its ‘truth-relevant’ composition; or, if one takes the further step to the establishm ent o f truth which has each time proved to be necessary, its ‘verification-relevant’ com position. If we disregard this further step we can say that the semantic form o f a sentence is expressed in its Tarskian truth-definition (cf. above p. 238f); for, firstly, such a definition speci­fies on what the truth or falsity o f a sentence depends and, secondly, we dem and o f such a definition that it does not give a m ere translation o f each individual sentence but that it be valid for all sentences o f a certain form in such a way that we can obtain the truth-condition for all individual sentences o f this form by merely substituting particular com ponent-expressions for the semantic variables (cf. above p. 258). So we can now, following Davidson, give a determinate sense to the pro­gramme o f a general theory o f the semantic forms o f assertoric sen­tences: it would consist in the construction o f as many truth-definitions as would take account o f every aspect o f the semantic structure o f asser­toric sentences that we intuitively hold to be such.

That the construction o f such truth-definitions is by no means a trivial affair becom es im mediately apparent if one attempts to go beyond the simple cases considered so far. I have already referred to the problem o f com plex sentences which cannot be interpreted truth-functionally (p. 243). A further task is the analysis o f the other semantic structures o f simple sentences, in particular the structures within singular terms and above all predicates, that I have not considered. Take for exam ple the sentence ‘That is a h uge fly.’4 One can, o f course, treat the whole predicate as an unstructured unit and thus restrict the form o f the sen­tence to its predicative structure. But if one did so one would clearly not be doing justice to the understanding of such a sentence, for this would mean that the expressions ‘huge fly’, ‘huge animal’, ‘huge water­fall’, etc., belong to our primitive vocabulary. Each expression would have to be explained separately, and the explanation o f ‘huge fly’ would be unrelated to that o f the words ‘h uge’ and ‘fly’; whereas in fact if we understand the words ‘h u ge’ and ‘fly’ we understand the expression ‘huge fly’, even if we have not previously heard this combination o f words. A truth-definition o f the sentence referred to which considered only its predicative structure would ignore this inner-predicative struc­ture just as much as a truth-definition that simply repeated the sentence itself would ignore its entire structure. Now what would a truth-definition be like that did justice to this inner-predicative structure? It would be tem pting to answer: ‘ “T hat is a huge fly” is true if and only if the object pointed at is huge and is a fly.’ But this suggestion clearly misses the

mark. Such a truth-definition applies to other, apparently similar, sen­tences such as ‘That is a red fly’, for som ething is indeed a red fly if and only if it is both red and a fly. ‘H uge’, on the other hand, is an attributive adjective; and if som ething is a huge fly this means that it is huge as a fly, not huge simply and not, for exam ple (as would follow from the suggested definition), a huge animal.

You could ask how one can be so sure that the formal-semantic con­nection between attributive adjective and substantive must be understood by means of a truth-definition. But do you have an alternative? Besides, this conception corresponds to what we would expect if we view the problem from the perspective o f the explanation o f mode o f employ­ment. If one has to explain to som eone how the word ‘huge’ is used as an attributive adjective, one explains to him how one establishes that som ething is a hugeF , where F ’ is a variable; and correspondingly for all attributive adjectives.

Another problem which likewise concerns inner-predicative structure is that o f the semantics o f modal adverbs, e.g. ‘Peter runs fast.’ Here too, for reasons similar to those just given, one clearly cannot regard the whole predicate as an unstructured expression. How does one explain to som eone a word like ‘fast’? Obviously by explaining to him how it is established whether the classification-expression ‘fast’ applies to an event, a happening or a state. Adverbs are thus a kind o f predicate, as one can easily see from the transformation: ‘Peter’s running is fast.’ T he truth-definition would have to show how the truth o f a sentence like ‘Peter runs fast’ depends on (a) the object for which the singular term stands being in a certain state and (b) this state being classifiable in a certain way.5

We can see how far we still are from a general theory o f assertoric forms o f sentence by reference to any sentence chosen at random from the newspaper. We have here a wide field which philosophers have barely begun to work at. Despite all the difficulties o f execution we have for the time being no reason to think that the semantic concepts which we have reached so far and the hypothesis that the meaning-relevant com­position of a sentence is its truth- or verification-relevant composition will founder on this problem area.

This would seem much more likely in the case o f the third step, which concerns the semantics o f non-assertoric sentences. Because such sen­tences - e.g. imperatives - cannot be true or false it seems clear from the start that the thesis that one understands a sentence if one knows its truth-conditions, or that one explains it by showing how it is verified, cannot apply to these sentences. But equally there can be no question

The next steps 397

Analysis o f the predicative sentence 398

o f a semantics o f non-assertoric sentences that is unconnected with the semantics of assertoric sentences, for the sentences o f the various modes differ only as regards m ode and not (with certain qualifications) as regards propositional content (above p. 52). And it would be absurd to suppose that we must explain (and learn) the various sentence- com ponents such as singular terms and predicates, and also words like ‘and’ and ‘or’, separately for the different modes. There would thus seem to be only two possibilities: (1) Because it cannot be extended to non-assertoric sentences the conception developed so far is also shown to be false for assertoric sentences or (2) it must be possible to discover a concept with the function for the semantics o f non-assertoric sentences analogous to that which the concept o f truth has for the semantics of assertoric sentences: or, putting it another way: it must be possible to find a concept that is wider than the concept o f truth and o f which the latter is a special case.

We are thus here faced with the task o f a revision o f basic conceptuality similar to that involved in the transition from the traditional to the analytical conceptuality. Just as there it was a question o f seeing that the ‘standing’ of an expression ‘for’ an object is only one function and mode o f em ploym ent among others, so here one would have to show that truth is only a special case o f som ething m ore general. I referred to this methodological structure - the step-by-step deepening or extending o f the conceptuality - at the beginning o f our undertaking and also used it to justify initially restricting myself to assertoric sentences (p. 13 Iff); at any rate I do not know how we would have been able to reach that wider concept that is now envisaged before we had conceptually exhausted the traditionally more familiar area. In those preliminary m ethodological deliberations I drew attention to the different system­atic procedure o f Searle, and I said that Searle pays for this systematic advantage with a comparative conceptual unfruitfulness. I must now give my reason for this claim, and by doing so we shall be led on to a clarification o f the problem itself .

The characteristic feature o f Searle’s procedure is that he first specifies the semantic rules for the modes or ‘illocutionary forces’ o f all kinds of sentence and only subsequently proceeds to the semantic rules o f the propositional content (common to all modes). Now I have shown, in the specific case o f the assertion-mode, that the rules which Searle specifies are, in various respects, wholly inadequate (pp. 187f, 202, 213ff). I have no time now to go into m ore detail but if you look at the rules (the ‘essential rules’) that Searle gives for the various kinds o f sentence or ‘illocutionary acts’6 you will see that they all am ount simply to para­

The next steps 399

phrases and that no systematic connection can be discerned between these paraphrases. It is therefore all the m ore surprising that Searle later, in that part o f the book where he deals with the propositional content, does produce a systematic connection and that the governing concept is the concept o f truth, a concept which did not figure at all in the thematic treatm ent o f the m odes.7 T o understand the connection you must realize that if one wants to give a general theory o f predicates that is not restricted to assertoric sentences one can clearly no longer carry along the assertoric aspect which they have in their usual expla­nation as assertoric sentence-components. T h e predicates must therefore be m ade either m ode-neutral or mode-variable. Searle settled for this second alternative (rightly so as we shall see). So far so good. But of course the question now arises: what sort o f a preliminary conception o f the understanding o f a predicate do we have? At this point Searle’s procedure is that which I earlier (in Lecture 8) claimed to be m ethod­ologically unavoidable (though he does not reflect on the methodological significance o f this procedure): he starts from the concept o f an assertoric predicate that initially is the only one at our disposal and gives the fol­lowing definition: we understand a predicate if and only if we know under what conditions it is true or false o f an object (p. 91). And then he undertakes what I have here dem anded: he tries so to extend this notion o f ‘being true o f ’ (or ‘applying’) as to make it also applicable to predication in the other modes. In a predication, whatever its m ode, the ‘question o f truth’ is always, he says, in one way or another ‘raised’.8 What Searle means by this becom es clearer if we take the modification which he considers as an aspect o f the predicate out o f the propositional content and thus regard it as a modification o f what I earlier called veritative being. O ne can then say that in all m odes the question o f truth is som ehow ‘raised’: with an assertoric sentence one says that som ething (a state o f affairs) is true; with an interrogative sentence one asks whether som ething is true; with an im perative som eone is com m anded to make som ething true; with an optative sentence one wishes that som ething were true.9

O nce again Searle has stopped just where the real problem begins (cf. above p. 202); but in this case he has at least reached the real problem. B efore we enquire further I m ust summ arize my criticism o f Searle. In his discussion o f the m odes o f predicates Searle has, without realizing it, m ade it clear that he still lacked the conceptual m eans for his system­atic procedure. Because the m odes and the propositional content are two non-independent m om ents that reciprocally refer to one another the conceptuality in terms o f which the m odes are interpreted must

Analysis o f the predicative sentence 400

harmonize with the conceptuality that is relevant to the propositional content. We have seen that we cannot explain a single assertoric sentence- com ponent without referring to the concept o f truth; and something analogous is to be expected in the case of non-assertoric sentences. Searle was not clear about this connection, with the result that the paraphrases he constructs in his thematic treatment of illocutionary forces in Chapter3 prove useless for the analysis, in Chapter 5, o f the predicate-modes. So he found him self obliged to develop a unitary concept o f the mode of a predicate on the basis of the concept that initially is the only one at our disposal — that o f ‘being true o f ’. And in this way he arrived at the concept o f ‘the question o f truth being raised’. What he failed to notice however was that this result conflicts with his paraphrases in Chapter 3 and that it is in any case superior to them because it suggests a unitary concept of modes and hence provides a possibility of constructing a general semantics o f sentence-components.

But are we justified in assuming, as Searle’s concept o f ‘the question of truth being raised’ seems to suggest, that this unitary concept must still be the concept of truth? T he possibility o f so reformulating imper­atives, etc., that the word ‘true’ occurs in them proves little. In the sen­tence ‘Make it true that p ! ’ we still have in the ‘make’ the unreduced imperative element. And, conversely, assertoric sentences could be reform ulated in an analogous way: instead o f saying ‘It is raining’ one could say ‘T h e command that it should rain is complied with’ or ‘T he wish that it might rain is fulfilled.’10 However, it could be precisely these correspondences which point the way ahead. Could it not be that com ­pliance and non-compliance, fulfilment and disappointment have a sig­nificance for the imperative and the optative sentence analogous to that which truth and falsity have for the assertoric sentence? Would this not provide us with a basis for what it is to explain an imperative, or optative? Just as we explain an assertoric sentence by showing how we recognize whether it is true so it seems plausible to say that we explain an imperative by showing how we recognize that it is com plied with. Analogously to speaking o f truth-conditions in the case o f assertoric sentences we could now say: we understand an imperative if we know under what conditions it counts as complied with; we understand an optative sentence if we know its fulfilm ent-conditions.11

We can easily see by reference to the simplest case - ‘and’ and ‘or’ sentences - that in the analysis o f the structure o f non-assertoric sen­tences these concepts really ‘bite’. Analogously to truth-tables one can construct compliance - or fulfilment - tables.12 For example, an imper­ative o f the form 7 *p or q * ’ (e.g. ‘Shut up or get out!’) is complied with

if one complies with ‘!*p*’ or or both; otherwise it is not compliedwith. And as with assertoric sentences (cf. above p. 242) we could now explain the use o f the imperative ‘or’ without repeating it (as in the spec­ification of obedience-conditions) by demonstrating how compliance and non-compliance with the whole sentence depends on compliance and non-compliance with the constituent sentences.

Thus a semantic programme would take shape in which for all forms of imperative sentence we construct compliance-definitions and for optative forms of sentence fulfilment-definitions. These would corre­spond exactly to the truth-definitions o f assertoric forms o f sentence, except that the word ‘complied with’ or ‘fulfilled’ would occur where the word ‘true’ occurs in the truth-definition. From this we can see that if the propositional structures have already been explained in the case of assertoric sentences then, provided one knows what it is to comply with a sentence, one does not need to explain the propositional structures of imperatives (let alone the words which occur in them) separately. But, by the same token, the converse also holds: assertoric sentences have no primacy. This is the basis o f what is justified about Searle’s view that one must be able to give a unitary explanation o f the propositional structure of all modes.

However, it is important to understand this correctly. T h e reason why the different forms o f sentence have a comparable propositional content is not because this content is mode-neutral, but because - as Searle correctly recognized in his treatment of predicates - it is mode- variable, that is to say is repeated in each o f the modes; as the truth - or compliance - definitions show the m ode is sense-determining for the whole propositional content and its parts. You could reply that we have the mode-neutral form ‘that p \ However we have already seen that understanding ‘that p ’ depends on understanding ‘p ’ rather than vice versa. On the other hand understanding 7*/?*’ appears to be on the same level as understanding ‘p ’^ ^ p * ’). But if that is so we must now ask what it is that truth, compliance, fulfilment, etc., have in common that enables them to have this sense-determining function. O f course it is not at all obvious that to every one of the ‘illocutionary forces’ listed by Searle or Austin there corresponds a pair like true-false or fulfilled- disappointed. Probably there are only a few basic modes that are sense- determining for the whole sentence and the other illocutionary forces are simply additions to a content that is neutral with respect to them.

If one asks what it is that a true assertoric sentence, a complied with imperative and a fulfilled optative sentence have in common it is clearly this, that in each case the sentence - or what is said with it - and reality

The next steps 401

agree. If an assertoric sentence is true, an imperative is com plied with, an optative sentence is fulfilled - in each case this means that things stand as the sentence says. T here seem to be two, and only two, possible forms o f this agreem ent. Either - in the case o f the assertoric sentence- reality is the measure; if there is non-agreem ent then the sentence does not correspond with reality. Or - in the case o f the optative sentence or imperative - the sentence is the measure; if there is non-agreem ent then reality does not correspond to the sentence.13 W hoever plays the assertoric game sincerely (truthfully, i.e. without intentions that do not belong to the rules o f the game cf. above p. 215) wants to say what is the case; whoever plays the other game sincerely wants what he says to be the case.

I am o f course here using unexplained concepts, but for the purpose o f merely indicating how the enquiry might continue they might suffice. With the wide concept o f agreem ent that has now been indicated we would have found the concept more com prehensive than that o f truth required for a general semantics. This concept would do for all propo­sitional sentences what the concept o f truth does for the special case of assertoric sentences. At the same time the vague and negative talk o f ‘all other sentences’ or ‘the non-assertoric sentence m odes’ would have acquired a delimitation and positive meaning. Because there are two and only two possible forms o f agreem ent one would have to acknowl­edge that there are only two semantically fundamental sentence-m odes:14 agreem ent-conditions are either truth-conditions or fulfilment-condi- tions.

Thus whereas in Searle the ‘illocutionary forces’ are multiplied far beyond the num ber of grammatical m odes, even different grammatical m odes would now belong to one basic m ode. As I have already pointed out, interrogative sentences can be regarded as a special kind o f im per­ative.15 That optative sentences and imperatives belong closely together can be seen from the fact that in both cases one can speak o f fulfilment- conditions and that in both cases the speaker expresses his desire that som ething be the case. If a grammatical optative is addressed to the person who is referred to in it as the subject o f the desired action then semantically it has in fact the meaning o f an imperative. Another class of sentences that are closely related to imperatives are the volitional sentences and those sentences in the 1st person future indicative which I earlier called ‘intention-sentences’ (p. -78). I already pointed out at that time that these sentences, which grammatically are in the indicative m ood, are semantically not to be construed as assertoric sentences but as analogous to imperatives. T he connection between imperative and

Analysis o f the predicative sentence 402

The next steps 403

intention-sentence is revealed in the fact that if an imperative is affirmed or denied by the person to whom it is addressed this ‘yes/no’ is an inten­tion-sentence. ‘Yes I will do it; no, I will not do it.’ T h e relationship between sentence and executing act is the same whether the sentence stems from another person (hence is an imperative) or from the person who is to carry the action out (hence is an intention-sentence).16 T h e use o f an im perative also presupposes that the addressee can intention­ally perform the desired action. T h e capacity for intentional, deliberate action thus underlies all these form s o f sentence. Som eone who did not possess this capacity could not understand any o f these forms o f sentence which, in their different ways, are so used that they serve as the measure for reality, thus that they say what ‘ought’ to be. T he sim plest case is where such a sentence guides one’s own action (intention-sentence). T he second possibility is that it is directed to others who can fulfil it (im per­ative). Optative sentences, which have no addressee, represent a limiting case. T he wish is fulfilled or not fulfilled without one being able to hold som eone responsible for this. It would seem plausible to say that som eone can only understand what it is to wish som ething if he already knows what it is to will som ething and to act deliberately.17 T h e analysis o f optative sentences would be especially important; for, if my explanation that we call that ‘good’ which is ‘rationally to be preferred’, hence worthy to be desired, was correct, then the understanding o f the word ‘good’ too is based on the understanding o f optative sentences (cf. above p. 8 1 f).18

O ne could call the semantically unitary class o f sentences to which optative sentences, imperatives and intention-sentences belong ‘practical sentences’ (Kenny calls them ‘fiats’) .19 In the T ractatus (4.022) W ittgen­stein characterized assertoric sentences as follows: ‘T he proposition (Satz) shows how things stand i f it is true. And it says that they do so stand.’ Correspondingly one could say o f practical sentences that the sentence shows how things stand if it is fulfilled; and that it (or the person who uses it) says that things ought so to stand.20 Accordingly we would have, as a consequence o f the two possible specifications o f the concept o f agreement, two basic kinds o f sentence: theoretical (assertoric) sentences and practical sentences. This is not to say that it is not m eaningful to distinguish a plurality o f illocutionary acts (or forces). Only one must realize that they are specifications o f the two basic m odes, and that, unlike the two basic m odes, these specifications are not sense-determ in- ing with respect to the com ponents o f the propositional content. Rather they contribute additional expressive or com m unicative aspects which no longer affect the propositional content, and whose explanation does

Analysis of the predicative sentence 404

not involve the fundam ental difficulties we face in explaining the m ode o f em ploym ent o f the two basic forms. The considerable lack o f orien­tation one finds in Austin, Searle and the work which starts out from them regarding the question o f an adequate classification o f ‘illocutionary forces’ is due to a neglect o f form al-sem antic questions. Searle started from the analysis o f sentences in which som ething is promised. H e treated this class o f sentences as though it were on as fundam ental a level as assertoric sentences. It is, however, obvious that promising-sen- tences are a species o f intention-sentences, which in turn are only a species o f practical sentences.

Imperatives and intention-sentences cannot, o f course, be formed with just any propositional content. A dem and or an intention can only con­cern som ething in the future. And they can only have an action, not a state, as their content. M oreover they can only be directed to free beings who can take up a position in regard to them by means o f ‘yes’ and ‘n o.’ But these restrictions do not hold for practical sentences as such. It is likely that one will be able to say that to every assertoric sentence there is a corresponding optative sentence, and that there is a one-to-one cor­respondence between the class o f practical sentences and the class o f assertoric sentences.

O ne must not, o f course, confuse what I am here calling practical sentences with the practical statements I spoke o f in the discussion of the question o f practical reason (Lecture 7). T he analysis o f practical state­ments, sentences about the good, would be the next step but one, the step that would have to follow the analysis o f simple practical sentences. Practical statements, the only sentences for the clarification o f which we have an absolute rational motivation (Lecture 7) are, o f course, seman­tically the m ost perplexing; they are theoretical-practical hybrids. As statements they contain a truth- or justification-claim, but what is asserted with them to be objectively justified, rational, is that som ething should be wished or done. T here is, therefore, a great temptation to overlook either their justification-claim or their practical character. Perhaps one will have to suppose with H are21 that, although sentences about the good are capable o f being justified, they would lose their practical char­acter, and we our freedom as responsible beings, if this justification were a com plete justification and could be regarded as analogous to the justification o f theoretical statements. In so far as they have both an assertoric side and a practical side practical statements contradict my thesis that there are two basic kinds of sentence: assertoric and practical.

I shall leave this contradiction as it stands. The purpose o f the present survey is not to answer these questions but merely to give some indication

The next steps 405

of the direction in which one would have to proceed. T he basic unity of practical sentences and the possibility o f explaining these sentences by means o f a concept analogous to that o f truth are merely hypotheses.I must also emphasize that I have only indicated the first step in a semantics o f practical sentences, which would correspond to that first step in the semantics o f assertoric sentences that consisted in the con­struction o f truth-definitions. T he further question o f how one would explain the employment-rules o f practical sentences - or the corre­sponding ‘gam e’ - is completely open; in particular, the question o f how one would explain the words ‘complied with’ or ‘fulfilled’, which for the understanding o f practical sentences would have a significance corre­sponding to that which the word ‘true’ has for the understanding of assertoric sentences.

In explaining the mode o f em ploym ent o f practical sentences one would also have to take into account the connection between the use of these sentences and the expression o f corresponding psychological states. Just as the person who uses an assertoric sentence lp ’ expresses that he believes that/?, som eone who utters an imperative ‘1*/?*’ expresses that he would like that p to be realized. With respect to assertoric sentencesI have tried to show that one can explain the concept o f belief in terms o f the concept o f assertion, which in turn is defined by the rules o f the verification-game (p. 212). T he corresponding hypothesis with respect to imperatives would be that these are used to demand o f som eone that he do som ething and that what this means could be explained, without recourse to volition, by the rules o f the compliance-game. Thus the concept o f volition would be explained by means o f the speech-act of dem anding, just as the concept o f belief was explained by means o f the speech-act o f asserting. A corresponding procedure in regard to inten­tion-sentences could also appear promising. Such a hypothesis, however, seems implausible in the case o f optative sentences. For, whereas in the case o f imperatives and intention-sentences we can demonstrate by means o f the correction o f actions what it is to correspond to the sen­tence, we cannot explain what it is for a fact to be ‘incorrect’ relative to an optative sentence without already presupposing that the sentence expresses a wish. A plausible way o f dealing with this difficulty would be to introduce the concept o f wishing via imperatives, so that this con­cept could then be made use o f in the explanation o f optatives. However,I must also point out that even when explaining the psychological state o f believing by means o f the speech-act o f assertion I had to make use o f another psychological state, namely that o f intending (p. 212). The problem o f ‘speech acts and psychological states’ would thus have to be

Analysis o f the predicative sentence 406

taken up anew. And I cannot claim to have shown that there is a practical alternative to the sort o f semantic program m e sketched by Grice and D. Lewis, in which the concepts o f intending and believing are treated as basic and presupposed as given independently o f language.22

What we have now achieved is at least a view o f formal semantics as a whole. So we should also have a perspective for the clarification o f the question which, in the introductory part o f these lectures, I called the fundamental question o f formal semantics and which has taken the place of the ontological fundam ental question about being and not-being, the question, namely, o f what it is to understand a sentence (p. 37f), thus the question regarding the essence o f all sentences, the general structure o f sentences. In the introductory reflections I got as far as form ulating this fundam ental question thus: ‘how is one to understand the fact that our entire linguistic understanding has the structure o f yes/no position- takings o f various m odes vis-ä-vis propositional contents?’ (p. 54). If the three aspects - propositional content, negatability, m ode - are the essential characteristics o f ‘our’ language (which we can therefore call ‘propositional language’) then the peculiar character o f these aspects should be revealed by contrasting our language with other, non-prop- ositional languages; and at the same time the inner connection o f these three aspects should become intelligible. It is, o f course, signal-languages that serve for this contrast. I have called the characteristic rules o f signal- languages ‘conditional rules’ (p. 165). They have the form: if such and such is perceived the signal 5 is used or if the signal S is perceived such- and-such is to be done. O ne might try to bring out what is characteristic of the employment-rules o f the independent expressions o f propositional language - sentences - in contrast to these conditional rules by calling them projective rules. I have borrowed the expression ‘projective’ from W ittgenstein’s Tractatus. W hen W ittgenstein described the sense o f sen­tences as pictures he did not have in mind a copying-function; rather the picturing-function was understood in a projective, or anticipatory sense: ‘In the proposition (Satz) a state o f affairs is as it were put together in an experim ental way’ (4.031). T h e sentence shows how things could stand. I have described the use o f an assertoric sentence as the opening m ove in a verification-game, and we have seen that the employment- rules are not conditional, but must be understood by reference to the outcom e o f the verification-game; we can now call this the ‘projective’ character o f these employment-rules. In the case o f imperatives it would only seem plausible to regard their rules as conditional rules if one were able to interpret them one-sidedly, by reference to the hearer, hence not as em ployment-rules. T he rule o f an imperative, as the speaker’s

The next steps 407

em ploym ent-rule, can also only be understood projectively. And as the hearer’s understanding is not shown by whether he com plies with the imperative, but by whether he, just like the speaker, knows what it is to comply with it and can by saying ‘yes’ or ‘n o’ com ply with it or not comply with it, the hearer understands the sentence in exactly the same way as the speaker; and that means: projectively. A nd with regard to the other practical sentences it seem s clear from the outset that they can only be understood projectively.

What is it that makes this projective m ode o f em ploym ent possible? In the case o f assertoric sentences we have seen that this is ultimately the achievem ent o f singular terms. It is due to the addition o f a singular term that the classification-expression loses the capacity for being used independently which it had in the signal-language, and instead o f relat­ing in a conditional manner to the given situation anticipates a verifi- cation-situation. T h e bearer o f this projection which specifies to what the predicate relates is the singular term. A lthough such a quasi-genetic form ulation is questionable, on account o f the empirical hypotheses it suggests, the ‘transition’ from a conditionally regulated language to a projectively regulated language is m ade possible by the debasem ent (.D epotenzierung) o f the independent classification-expression that is characteristic o f primitive signal-languages to a predicate and its exten­sion to a propositional expression through its supplem entation by a sin­gular term. W ithout using genetic metaphors the hypothesis can be for­m ulated as follows. Only propositionally structured expressions can be governed by projective rules; and all propositionally structured expres­sions must be projectively regulated. T hat this hypothesis must also be confirm ed for practical sentences, thus that in their case too projective regulation is ultimately m ade possible by singular terms, is clearly sug­gested by the fact that practical and assertoric sentences correspond as regards their propositional content.

T h e next point, to express it once m ore in the genetic m etaphor, is that the mere addition o f a singular term to a classification-expression not only debases the classification-expression - in that it is now in need o f supplem entation - but also debases the whole expression that results from this combination; for the m ere showing o f how things could stand- the ‘experim ental putting together o f a state o f affairs’ - would be, in contrast to the use o f a signal, a purposeless and m eaningless activity. T h e gain that is achieved by the propositional-projective language vis- a-vis conditionally regulated language can be visualized by saying that language now no longer concerns only the perceptually given, but reaches as far as the spatio-tem poral im agination. Now this extension

Analysis o f the predicative sentence 408

into the realm of possibilities would remain without purpose if language did not acquire a new relation to reality. An expression like ‘the cas­tle/burning’ shows how things could stand with the castle. The proposi­tional expression only acquires a relation to reality if in addition it is said that this possibility is actual or that it is to be actualized; thus if either it is asserted that it is or dem anded that it be as the propositional expression represents it as being possible. The speech-act’s initial loss of purpose resulting from its ‘transformation’ into a propositional expression can thus only be com pensated for by combining a reality- claim with the use o f the expression. Once more the thesis can be for­mulated without using genetic metaphors: a propositional language without modes is inconceivable and, conversely, a modal language that is not propositional appears to be inconceivable.

Now why is it that propositional language also seems to be distin­guished from signal-languages by the presence o f negation? Since the ‘not’, as we saw earlier, belongs not to the mode but to the propositional content (p. 46f), it would seem natural to look for the reason in the propositional content. And within the propositional content it would again seem natural to think first and foremost of predicates. As classi­fication-expressions are used to classify, and thereby distinguish, the use o f the word ‘not’ seems to be necessarily connected with the use of classification-expressions. Against this supposition, however, speaks the fact that the expressions o f signal-languages are also classification- expressions, and yet these languages manage without the word ‘not’. On the other hand, one can now easily see that it is the agreement asserted or dem anded in the basic modes which makes the use o f a negation-sign necessary; for it is essential to the understanding o f the relationship between sentence and reality that there is agreem ent or non-agreem ent. T o understand a sentence is to know under what con­ditions it is true and under what conditions false (or fulfilled and dis­appointed). We can make this clear to ourselves by reference to a sub­class o f the sentences at our disposal, in which the word ‘not’ happens not to occur. If one is to establish o f a sentence of this class whether it (if it is an assertoric sentence) is true or false, then if the sentence turns out to be false the word ‘not’ must be used, and there would thus be generated, if it is a predicative sentence, a sentence with a negated predicate. This is the reason why to every propositional content there belongs one that is opposed to it. Thus, although negation belongs to the propositional content, it does so by virtue of the fact that a propo­sitional expression can only be used according to projective rules. There would thus result the further thesis: negation has its origin in the pro­

The next steps 409

jective essence o f our language. In this perspective on the connection o f yes/no with the projective essence of our language there would be taken up again both (1) H eidegger’s basic thesis that the question of the meaning o f Being and Not-being must be posed from within the ‘horizon’ o f time23 and (2) an essential aspect o f the theory of propositions o f the early Wittgenstein, the only analytical philosopher who was also con­vinced that the key to understanding the essence of the proposition lies in the ‘mystery of negation’.24

It would thus now become clear why - as we were already able to see in the introductory reflections (p. 53f) - the modes (we can now say more precisely: the basic modes) are essentially Yes/No-position-takings. Since a propositional expression is, as Wittgenstein put it, ‘bipolar’25 - true/false, fulfillable/disappointable - and merely as it were strikes a theme and leaves open projectively the decision between Yes and No, the use o f such an expression is only meaningful if the speaker decides between Yes and No, and this means: projectively anticipates the agree­ment of reality with the expression in one way or the other, i.e. either asserts or demands agreement. But with this a position is always taken vis-a-vis the opposite assertion or demand (i.e. that which asserts or demands the opposite propositional content). The basic modes, there­fore, have the character of affirmations-and-denials, o f Yes/No-position- takings. That the speaker must decide between Yes and No implies that the hearer can also decide between Yes and No. T he order can be rejected, the assertion doubted: the origin o f freedom and reason.

Such high-sounding theses, o f the kind one likes to express at the end of a course of lectures, could, o f course, easily disguise how fragile the whole line of thought that I have presented to you is. The thesis about the origin o f negation to which it has finally led us directly invites a counter-thesis which is calculated to throw my whole procedure into confusion. It could be objected that we must already use a negation- expression as an action-correcting word together with that expression ‘correct’ which we already need if we are to be able to explain the m ode of employment o f a linguistic expression and the understanding of whichI have presupposed in explaining the word ‘true’. This objection does not only touch the thesis about negation. It must awaken doubts as to just how fundamental one can regard an enquiry into the understanding o f our linguistic expressions which leaves unclarified the use of action- correcting words. Now this is not an accidental omission, but a necessary consequence of the methodological orientation towards Wittgenstein’s principle that the meaning of an expression is what we explain when we explain its meaning to someone. We cannot explain the words ‘cor­

Analysis o f the predicative sentence 410

rect’ and ‘incorrect’, for in any explanation o f a word these words must already be used. Should we not therefore re-examine the methodological orientation towards this principle? It suggested itself as the adequate principle for a specifically philosophical analysis o f language, if one understands by a philosophical analysis an im manent analysis, a mere making explicit o f what is implicitly already known in understanding (p. 152). But is this assumption from which I started in Lecture 1 sac­rosanct? And would not the assumption o f a ‘specifically philosophical’ analysis o f language that is distinct from language-scientific, empirical methods have to be re-examined? But what other m ethods, whether philosophical or scientific, would enable us to proceed further? Clearly we have reached a point at which a new and this time more fundamental revision o f the basic conceptuality would be necessary. T he question o f what it is to understand a linguistic expression seems, if we do not deceive ourselves, as unclear as ever.

Notes

Lecture 1

1 cf. among others Passmore, A Hundred Years of Philosophy, Chaps. 9 and 15-18 (this can be especially recommended); Urmson, Philosophical Analysis (instructive for the beginnings but does not take in more recent develop­ments); Rorty, The Linguistic Turn (a collection of programmatic papers with a detailed introduction); von Savigny, Die Philosophie der normalen Sprache; von Kutschera, Sprachphilosophie; Stegmüller, Hauptströmungen der Gegen­wartsphilosophie Vol. i, Chaps. 9-11, Vol. ii, Chaps. 1 and 2.

2 Philosophical Investigations, §89.3 Confessions xi, 14.

Lecture 2

1 Cartesian Meditations §5.2 It determines the thought-sequence at the beginning of Chap. 2 of Book i

of the Metaphysics; the other is in Chap. 1.3 cf. Plato, Theaetetus 201c and Russell, Problems of Philosophy, Chap. 13. More

recent papers on the concept of knowledge can be found in the collection edited by Griffiths Knowledge and Belief.

4 This, however, was not Aristotle’s view (cf. Post Anal. A33). However we can here disregard this particular aspect of the Platonic-Aristotelian concept of science, which does not correspond to the ordinary understanding of the word.

5 The Greek expression is gnorizein, gnosis. In the chapter in Metaphysics it only occurs incidentally (180a26); cf. however in the parallel text in Post. Anal. B19, 99b38 and 100b4 and above all De Anima 427a21.

6 This is not explicitly justified here by Aristotle where he is merely repeating the existing opinions; it is however justified in an earlier version of the same thought in the lost writing ‘Protreptikos’, Fragment 7.

7 cf. 980a21-25.8 981bl3ff., 982b ll-983al 1.9 Nichomachean Ethics Book 10, Chaps. 7-8.

10 cf. 982a24ff.11 cf. Plato, Republic 510.

Notes to pp. 2 0 —3 4 412

12 Post. Anal. A9.13 Post. Anal. A2ff.

Lecture 3

1 1003a22-25, I025b7-10.2 998b20f.3 Ideen zu einer reinen Phänomenologie, I, §9.4 Homer, Iliad 1, 70.5 Ideen §3.6 Metaph. VII, 3.7 I am here following Dummett, Frege, p. 59. As Dummett shows (59 f.) this

criterion of course provides only a necessary condition which must be fur­ther qualified if it is to be claimed as a sufficient condition. In particular, the expression ‘something’ itself, or an expression which contains ‘some­thing’ or another indefinite pronoun as a part, would also fall under this criterion. These expressions are thus explicitly to be excluded. However I do not share Dummett’s view - which is also held by Strawson (‘Singular Terms and Predication’, Section n) and Geach (‘On What There Is’) contra Quine (From a Logical Point of View, p. 13, Word and Object, p. 240) - that predicates too are replaceable by ‘something’. Rather one will have to say: if that which was expressed by a predicate is referred to by ‘something’ then we are already in what is called in logic a higher-level predicate-language and are speaking of a higher kind of objects, namely, attributes. The proof of this seems to me to be that wherever a predicate is replaced by ‘some­thing’ it can also be replaced by ‘the same’. Dummett and Geach rightly hold that ‘ = ’ can only be used with reference to objects. But if * = ’ can be used wherever ‘something’ is used then it follows that if ‘something’ takes the place of a predicate it is really taking the place of the corresponding designation of the attribute.

8 As with the previous criterion we must here exclude expressions which already include indefinite pronouns.

9 Ideen, §13.10 ‘Predicate’ is really a grammatical-syntactical concept and does not corre­

spond exactly to the semantic concept of a general term. In the sentence ‘The horse is tired’ only one predicate (‘is tired’) occurs, but two general terms (‘horse’ and ‘tired’). Many analytical philosophers (e.g. Strawson) nonetheless speak of predicates when what they mean is general terms. I shall also adopt this improper way of speaking because the correct way (which is used e.g. by Quine) would prove too involved in the course of these lectures, for it does not permit an adjectival modification (‘predica­tive’) and suchlike.

11 A programmatic sketch of formal semantics is to be found in the essays of Davidson, ‘Truth and Meaning’, and ‘Semantics for Natural Languages’.

12 cf. esp. the beginning of Book 7 of the Metaphysics.13 cf. my Ti kata tinos, §§5—6.14 cf. e.g. 1045b30 f.15 I have interpreted the specific situation with regard to Parmenides in my

essay ‘Das Sein und das Nichts’.

Notes to pp. 3 7 -5 0 413

Lecture 4

1 cf. Lyons, 5.2.2 It could appear strange that the semantic counterpart to the fundamental

question of ontology can dispense with the ‘as’, thus that we need not use the formulation:‘what it is to understand a sentence as a sentence’. Aristotle however only had to introduce the ‘as’-formulation because he formulated the question objectually. That made it necessary to exclude the misunder­standing that ontology studies beings in the sense of individual beings, indi­vidual objects. As soon as one formulates the ontological question semanti­cally (what does it mean to speak of ‘being’?) a corresponding misunderstanding can no longer arise and the ‘as’-formulation drops out.

3 10l7a27-30.4 996b26-997al5, 1005al9-b5.5 1005b 19 f.6 1006a3.7 1006a21 f. The Greek word is semainem.8 cf. Strawson, Introduction to Logical Theory, p. 2 f.9 De Interpretatione, Chap. 4.

10 10l7a31-35.11 1051b33.12 Metaph., VI, 4.13 Metaph., IX, 10.14 Wittgenstein, Tractatus 2.15 ‘Die Verneinung’, p. 149. All smaller writings by Frege are quoted by me in

accordance with the original pagination for this is also given in the margin of the two volumes edited by G. Patzig.

16 A criterion for the distinction between positive and negative predicates and hence for positive and negative sentences can be obtained from the distinc­tion between merely exclusive predicates (like ‘red’ and ‘blue’) and contra­dictory predicates (like ‘red’ and ‘not-red’); its applicability is, however, restricted, cf. Ayer and Gale.

17 ‘Die Verneinung’, pp. 153 f. cf. also Geach, ‘Assertion’ and Dummett, pp. 316 f.

Lecture 5

1 cf. e.g. St Thomas Aquinas, Veritate, I, 1: ‘illud autem quod primum intellectus concipit quasi notissimum et in quo omnes conceptiones resolvit est ens’. Duns Scotus, Ord. I, dist. 3 pars. l,q.3.no. 137: ‘primum objectum intellectus nostri est ens’.

2 1 shall not go into the problem of the ambiguity of the Indo-European and in particular the Greek word ‘to be’ and the questions of whether the differ­ent meanings are nonetheless connected and whether this connection is suf­ficiently universal for philosophy to be able to orientate itself to this word. Instead I refer you to the work of C. Kahn, in which for the first time this problem is discussed comprehensively and at the level of contemporary lin­guistic and philosophical knowledge. For someone today simply to start out from talk of ‘Being’ (‘das Sein) as Heidegger has done is unsurpassably naive.

Notes to pp. 5 1 - 6 0 414

3 I myself have fallen into this error for this reason in my essay ‘Die spra- chanalytische Kritik der Ontologie’, p. 492.

4 I shall come back to these questions in Lecture 28.5 In a certain way it is to be found in Heidegger, cf. note 14 to this lecture.6 A formal semantics which is built on this insight and is accordingly divided

into two parts - a semantics of the modes and a semantics of the proposi­tional content common to all sentences is developed in Searle’s book Speech Acts.

7 cf. Hare, ‘Meaning and Speech Acts’ (1970) in: Hare, Practical Inferences, pp. 82 ff. The term ‘external negation’ is however also used in another sense; cf. the article ‘Negation’ by A. N. Prior in the Encyclopedia of Philoso- Phy.

8 It was Austin, in his How to Do Things with Words, who first called attention to the so-called ‘performative’ use of sentences.

9 cf. Frege, ‘Verneinung’, first sentence; Lewis, Convention pp. 186 f.; Hare, loc. cit., pp. 80-2. In the case of so-called sentence-questions (such as ‘Is he coming?’) the range of possible replies is determined by the propositional content which can be either affirmed or denied. In the case of so-called word-questions like ‘Who is coming?’ an element of the propositional con­tent is left open and it is left to the respondent to replace it with an appro­priate content.

10 The subject-matter (also its demarcation from semantics) and in particular the conceptuality of this discipline are still very uncertain. J. Habermas seeks to develop a philosophical conception of pragmatics in his paper: ‘Was heisst Universalpragmatik?’

11 Critique of Pure Reason, B 197.12 cf. the interpretations in F. Kambartel, Erfahrung und Struktur, pp. 113 ff.13 Connected with this is the fact that what Kant means by ‘object’ (‘Gegen­

stand’) is really what one calls ‘objectivity’ (‘Objektivität’); the latter, however, is a mode of veritative being. But Kant could not explicitly conduct his enquiry in this way, for despite the fact that he started out from the forms of judgment he was not orientated towards sentences.

14 That which is ‘disclosed’ Heidegger called ‘Being’ (<Sein,). He was thus able to interpret the analysis of disclosedness at the same time as the resumption of the question about Being which for its part is to be detached from the traditional fixation with objects. This unusual use of the word ‘being’ can be made intelligible in the following way: from among the various meanings which the word ‘be’ has in language Heidegger orientated himself primarily to veritative being. This is shown in particular by the fact that for him the question about Being is at the same time a question about Not-being. All disclosedness which is expressed in sentences is to this extent a disciosed- ness of (veritative) Being. Now as we have seen we can also transfer the talk of ‘being’ to the more general concept of affirmation, so that one can now say that all disclosedness which is expressed in sentences is a disclosedness of Being. Once one has got this far one can understand that Heidegger was able to extend the talk of ‘Being’ yet again in such a way that one can now say that all disclosedness, even that which is not articulated in sentences, is a disclosedness of Being. Heidegger himself, however, has never given a clear

Notes to pp. 6 2 -7 415

account of his use of the word ‘Being’. On the one hand he has naively assumed that the meanings of the word ‘being’ constitute a unitary connec­tion; on the other hand it seemed to him natural to say with the object- orientated tradition that all being is a being of beings, although this does not fit veritative being, let alone the extended concept. 1 have attempted to give a precise meaning to Heidegger’s equation of ‘world’ and ‘being’ in sec­tion 4 of my essay ‘Das Sein und das Nichts’.

15 cf. Logische Untersuchungen, i i , 1, pp. 421 ff. (‘Beilage zur Kritik der “Bilder­theorie” und der Lehre von den “immanenten” Gegenständen der Akte’).

16 cf. above n. 1.17 cf. the quotation from Aquinas in n. 1.18 cf. my interpretation in section 3 of my essay ‘Das Sein und das Nichts’.19 cf. B129 ff.,B234-6,242; and for the use of the word B376 f and Erste Ein­

leitung zur Kritik der Urteilskraft, m (Werke, xx, p. 205 f.

Lecture 6

1 Both Kant and Heidegger overlooked veritative being, to which they were nonetheless orientated. For Kant cf. above p. 414 n. 13. The situation is especially confused in Heidegger. He was not only concerned with an extension of the disclosedness-thematic beyond objects; he also wanted to show that the ‘more basic’ disclosedness is one that does not relate to objects at all. By ‘objecthood’ - in Sein und Zeit ‘presence-at-hand’ (Vorhandenheit’) - he meant not only that for which singular terms stand but the whole onto­logical perspective which results from the orientation to the statement (§§13,33). In contrast to the disclosedness which is expressed in sentences he sought to exhibit a pre-logical, pre-linguistic disclosedness as more basic (but for the analysis of which he nonetheless took the statement-structure - the ‘as’, §32 - as his clue). This exclusion of sentences from the core-area of the analysis which results from the rejection of the logical contradicts the central importance which Heidegger attributed to language (‘Language is the house of being’). In his statements about language Heidegger therefore reverted to the level of the most primitive theories of language, in that he emphasized the significance of the word for the disclosedness of beings. Because Heidegger restricted the notions of objecthood and objectification to the level of statements, objects could once more gain access through the backdoor of another terminology (that of ‘beings’ and ‘things’) and take up a dominant and analytically uncontrollable position. Heidegger’s concep­tion o f ‘world’ is correspondingly ambiguous: on the one hand it appears as a whole of significance, on the other hand - and increasingly so in the later works - as the sphere of disclosedness of things (cf. Tugendhat, Der Wahr­heitsbegriff bei Husserl und Heidegger, pp. 399-402).

2 cf. Wittgenstein, Philosophical Investigations, §§305,308.3 Logische Untersuchungen, v, Chap. 1. Husserl also mentions a third concept

of consciousness but which can here be passed over: consciousness in the sense of inner-perception. This concept falls under that of intentionality and in turn forms the basis for the delimitation of consciousness in the

Notes to pp. 6 8 —73 416

sense of the unity of experiences {Erlebniseinheit) (§6). What I have called Husserl’s second concept of consciousness is the one he mentions first.

4 Philosophical Investigations, esp. §§244 ff. and pp. 221 ff., Zettel, §§472 ff.5 cf. Sein und Zeit, §25.6 cf. Sein und Zeit, §§28 ff.7 Logische Untersuchungen, v, §10.8 Ideen, §89.9 Brentano, Psychologie vom empirischen Standpunkt, Appendix, Section 1; cf.

also R. Chisholm, Perceiving, §11.10 The same is true where ‘perceive’, ‘see’, ‘hear’ are used propositionally. *x

sees that the sun has risen’ means: x knows on the basis of his optical per­ception that the sun has risen.

11 cf. also Searle, Speech Acts, p. 93. The reflections so far do not amount to a proof of this statement; this requires further reflections on the form of existential statements (reflections which I shall not carry out till later). So long as one conceives the form of an existential sentence predicatively (TV exists’) the above thesis appears to contradict itself because the understand­ing of an existential sentence would in turn presuppose the consciousness of an object (of N ). We shall see however (p. 300) that existential sentences cannot be understood predicatively. The sentence ‘The devil exists’ really has the form: ‘There is one and only one object that is devilish’. It might also be thought that even if it cannot be shaken by phantasy-consciousness the above thesis is refuted by the holding-to-be-true of negative existential statements. For in such a case we seem to have the consciousness of an object which we neither mean as existent nor think of as such. However, this argument also fails if we do not understand existential statements pre­dicatively. For when I say ‘There is no object that is devilish’ I do not have a consciousness of a non-existent object.

However, it will later emerge (p. 367) that the above thesis cannot be sus­tained in this generality. It holds indeed for all objects that exist (occur) in space and time, but not for the spatial and temporal positions themselves. One can believe that an object exists at a particular time in a particular spa­tial position and this belief can turn out to be false. But this cannot be repeated for the spatial and temporal positions themselves. One cannot say of spadal and temporal positions that they do not exist and consequently one also cannot say that they exist. Meaning a spatial or temporal position is thus not founded in the holding-to-be-true of an existential statement. It is propositional only in the weak sense that it is accomplished by means of a singular term which is a non-independent element of a predicative sen­tence. Connected with this difference is the fact that there cannot be a phantasy-modification for spadal and temporal positions in the same sense that there is for material objects and events. One can say: ‘Imagine that the Neckar does not exist, that 200 metres northwards from here the Neckar is not flowing, but there is an autobahn on which a princess is riding a bike.’ But one cannot say: ‘Imagine that the spatial position that is 200 metres northwards from here does not exist.’ A phantasy-modification is only pos­sible with regard to spatio-temporal positions in the sense that one can say ‘Imagine a place . . ‘There was once a time . . .’ and not locate this spatio-

Notes to pp. 7 4 -1 0 9 4.17

temporal position in our spatio-temporal system (i.e. specify no distance from the spatio-temporal position at which one is speaking).

12 cf. Henrich, ‘Fichtes ursprüngliche Einsicht’ and ‘Selbstbewusstsein’.

Lecture 7

1 For the next two pages cf. Hampshire, Thought and Action, Chap. 2 (‘Inten­tion and Action’).

2 cf. the essentially more differentiated characterizations in Kenny, Will, Free­dom and Power, Chaps. 2 and 4.

3 The explanation of freedom of action in terms of the action-determining understanding of a sentence to which there always simultaneously belongs a consciousness of the negation of the sentence stems from Aristotle, cf. his Metaphysics, ix, 2 and 5.

4 For the concept of a practical question cf. Hare, Freedom and Reason, 4.3 and 4.5.

5 The question concerning the meaning of the word ‘good’ is the fundamen­tal question of language-analytical ethics; cf. Moore, Pnncipia Ethica; Ste­venson, The Emotive Meaning of Ethical Terms and esp. Hare, The Language of Morals. Standing outside this tradition but particularly instructive is von Wright, The Varieties of Goodness, cf. also P. Ziff, Semantic Analysis, Chap. 6.

6 Metaph., 1006a 11-25.

Lecture 81 The connection between the understanding of sentences and the concept of

the good was first brought out by Aristotle in the famous passage in which he characterizes man as a political animal and grounds this precisely through his capacity to understand sentences {Politics 1253a 1 Off.): ‘Of all animals only man speaks in sentences. The mere making of sounds serves to indicate pleasure and pain, and is thus a faculty that belongs to other animals. Only in sentences can one signify what is advantageous and what is harmful; and thus one can only signify what is just and what is unjust in sentences. For it is peculiar to man in comparison with the other animals that he alone has a perception of good and evil and of the just and the unjust. It is participation in such which constitutes a household and ulti­mately a polis.’

2 cf. Aristotle, De Interpretation, Chap. 4; Kant, Critique of Pure Reason, B93; Heidegger, Sein und Zeit, §33. The first working out of the predicative structure is to be found in Plato’s Sophist.

Lecture 9

1 The critique of Husserl’s theory of meaning carried out in this and the next lecture can already be found, in its basic features, in my essay ‘Phänomenol­ogie und Sprachanalyse’. The impatient reader can skip these lectures and go straight on to Lecture 11.

2 cf. J. S. Mill, System of Logic, Bk. i, Chap. 2, §2.

Notes to pp. 1 0 9 -^ 2 418

3 Frege, ‘Über Sinn und Bedeutung’ (SB), pp. 26 f.4 Ideas §3: ‘Every possible object, logically speaking: “every subject of possible

true predications” ’ cf. also Logical Investigations, ii, §8.5 Thus Quine: ‘every singular term names or purports to name just one

object’ (Word and Object, pp. 90, 95 f.).6 cf. e.g. Russell, Introduction to Mathematical Philosophy, Chap. 16. In Russell’s

classical discussion of these expressions in the essay ‘On Denoting’ this ter­minology is not yet used.

7 cf. Mill, loc. cit., Bk. i, Chap. 2, §5.8 SB.9 cf. above n. 4.

10 One recognizes this, according to Frege (SB pp. 32-5) by the fact that if two descriptions ‘a’ and V designate the same object, thus if a — b, their mutual substitution in an arbitrary (non-intensional) sentence leaves the truth-value of the sentence unchanged. The identity ‘a = b ’ is even defined by the sameness of truth-value in the sentence: a = b = def. (F) (Fa =Fb).

11 SB, p. 34.12 cf. C. I. Lewis, An Analysis of Knowledge and Valuation, p. 52.13 cf. my paper ‘The Meaning of “Bedeutung” in Frege’.14 Frege’s definition of ‘object’ is to be found in his paper ‘Funktion und

Begriff’, p. 18: ‘An object is anything that is not a function, so that an expression for it does not contain any empty place.’

15 cf. e.g. Cartwright, ‘Propositions’, p. 101: Pitcher, Truth, p. 8.16 SB, p. 32, ‘Der Gedanke’ (G), p. 61.17 SB, p. 37.18 G pp. 60 f.19 cf. G 64, SB, p. 32.20 Since he lacks this broad concept of meaning Frege speaks, strangely, of

‘constituents of the sentence’ (G 63).21 G 63 f. cf. Dummett, Chap. 1.22 G 62 f. cf. Dummett, p. 295 f.23 Investigation v, §§34 f.24 loc. cit.

Lecture 10

1 pp. 301-3. The same notes are also printed in the appendix o t Philoso­phische Grammatik, pp. 199-201.

2 cf. e.g. Aristotle, Metaphysics, vi, 4; De Anima, hi, 6; Kant, Critique of Pure Reason, §15.

3 For what follows cf. Logical Investigations, vi, §46 and v, §§17-18.4 cf. Investigation vi §62 (p. 190) and my Der Wahrheitsbegriff bei Husserl und

Heidegger, p. 131.

Lecture 11

1 cf. the appendix to this lecture.2 cf. Philosophical Investigations, §§65—73, Philosophische Grammatik, §§47 ff. and

Lorenzen, Methodisches Denken, p. 30.

Notes to pp. 1 4 6 -8 2 419

3 For what follows cf. the essays ‘Funktion und Begriff’ and ‘Begriff und Gegenstand’ and also the treatise published in Frege's Nachgelassene Schriften under the title ‘Ausführungen über Sinn und Bedeutung’.

4 cf. above p. 115 f.5 ‘Begriff und Gegenstand’, p. 201. cf. also p. 193.6 loc. cit., p. 197.7 loc. cit., p. 198 and p. 197.8 Speech Acts, Chap. 5, §1.9 loc. cit., p. 197.

10 loc. cit., p. 198.11 cf. my paper ‘The Meaning of “Bedeutung” in Frege’.12 cf. ‘Sinn und Bedeutung’, p. 31.13 Nachgelassene Schriften, p. 133.14 loc. cit., p. 128.15 Lorenzen, Methodisches Denken, pp. 35 f.16 Nachgelassene Schriften, p. 133. On ‘extension’ and ‘intension’ cf. Carnap,

Meaning and Necessity.

Lecture 12

1 Philosophical Investigations, §§243 ff.2 cf. Jones (ed.), The Private Language Argument.3 cf. Mittelstrass, ‘Die Prädikation und die Wiederkehr des Gleichen’.4 cf. De Anima 417b22.5 This was even seen by Aristotle himself cf. Post. Anal. 87b28: ‘Even if per­

ception as a faculty is of “the such” and not of a “this”, yet that which is being perceived is a “this” and at a definite present place and time.’

6 Hume, Treatise of Human Nature, Vol. i, Book i, Pt. 1, Sect. v i i .

7 cf. Russell, The Problems of Philosophy, Chap. 9.8 Wittgenstein’s arguments, in particular, are directed against this assump­

tion, that the representation of something can explain the unified employ­ment of a sign. cf. Phil. Inv., §§73, 86, Phil. Gram., Part 1, Sect. iv.

9 Phil Inv., §117.10 cf. Quine’s notion of ‘occasion-sentences’ (Word and Object, §9).

Lecture IS

1 cf. Bloomfield, Language, p. 24.2 cf. Brown and Dulaney, ‘A Stimulus—Response Analysis of Language and

Meaning’, pp. 85 f.3 Bloomfield, p. 139 (my emphasis).4 G. H. Mead, Mind, Self and Society, especially §§7-10.5 cf. Hare, The Language of Morals, i, 1.7.6 Stevenson, Ethics and Language, Chap. 3; Morris, Signs, Language and Behav­

iour, p. 12.

Lecture 14

1 H. P. Grice, ‘Meaning’.2 ‘Utterer’s Meaning, Sentence-Meaning and Word-Meaning’.

Notes to pp. 1 8 2 -2 1 4 420

3 loc. cit., p. 59.4 Speech Acts, Chap. 2, §5.5 How to do Things with Words, Lecture 8.6 cf. A. Goldman, A Theory of Human Action, pp. 1 ff.7 The distinction between locutionary and illocutionary act, as I have described

it, does not correspond exactly to Austin’s own description, but to the revised version of Searle. cf. Searle, loc. cit., Chap. 2, §1. On the contrast with Austin see especially ‘Austin on Locutionary and Illocutionary Acts’.

8 W. Alston, Philosophy of Language, Chap. 2, and Searle, loc. cit.9 Perhaps this is the reason why Grice developed his theory, which is orien­

tated towards a conception which one might think had been overcome by Austin, at a time when Austin’s theory was already known.

10 pp. 64, 66.11 cf. the relevant places on p. 60 (transition from 6 to 7) and p. 63. One could

call my criticism of Searle unfair, because he only applies his theory inci­dentally to assertions and primarily presents it with reference to the speech act of promising. However, in what follows it will emerge that my objections by no means concern only accidents of formulation, but are objections of principle.

12 cf. Lewis, Convention, pp. 147 ff.

Lecture 15

1 The Redundancy Theory is usually traced back to Ramsey; cf. Ramsey, ‘Facts and Propositions’ (1927) in Ramsey, The Foundations of Mathematics, pp. 142 f. The passage is also re-printed in G. Pitcher (ed.), Truth, a volume in which several papers of the discussion connected with Ramsey are col­lected.

2 Metaphysics iv, 7, 1011b26f.3 cf. Metaphysics ix, 10, 105 lb2-5.4 cf. St Thomas Aquinas, Quaestiones Disputatae de Veritate, quest. 1. art. 1.5 cf. e.g. Schelling, System des transzendentalen Idealismus, §1.6 Searle, Speech Acts, Chap. 2, §5; Dummett, Frege, Chap. 10. cf. also his paper

‘Truth’.7 cf. also Tractatus, 4.022.8 cf. Tarski, Der Wahrheitsbegriff in den formalisierten Sprachen.9 cf. Chap. 2, §5.

Lecture 161 Grice has seen this difficulty but not solved it (cf. ‘Utterer’s Meaning and

Intentions’, pp. 174-6).2 cf. the essay ‘How to make our Ideas clear’, Collected Papers, Vol. v, pp. 248-

71.3 I introduce this distinction between ‘believing’ (Glauben) and ‘opining’ (Mei­

nen) merely to simplify my exposition.4 cf. Dummett, pp. 298 ff.5 cf. above p. 14.

Notes to pp. 2 1 8 -8 8 421

6 For the distinctions in the next paragraphs cf. R. Cartwright ‘Propositions’, pp. 84-6.

Lecture 17

1 Tractatus, 4.31 ff. A simple exposition can be found at the beginning of most modern text-books on logic.

2 cf. his papers ‘Truth and Meaning’ and ‘Semantics for natural languages’.3 cf. the second paper, pp. 178 f.

Lecture 18

1 cf. e.g. Pfänder, Logik, i, 9.2 Kant, Critique of Pure Reason, B 96.3 For criticism of the traditional conception cf. Geach, Reference and General­

ity, esp. Chap. 1.4 cf. his Begriffschrift, §§11 f.; ‘Funktion und Begriff’, p. 23 ff.5 cf. Strawson, Introduction to Logical Theory, Chap. 6, §7.6 Plato, Sophist 262-263.7 cf. De Anima in, 6.8 cf. Oehler, Die Lehre vom noetischen und dianoetischen Denken bei Platon und

Aristoteles, esp. Pt. 1, Sect. 2.

Lecture 19

1 cf. Strawson, Logico-Linguistic Papers, pp. 56 f.2 loc. cit., pp. 11 ff.

Lecture 20

1 cf. S. Kripke, ‘Naming and Necessity’, p. 322.2 e.g. Logic and Knowledge, pp. 186 f., 201.3 cf. the papers ‘Über Sinn und Bedeutung’ and ‘Der Gedanke’.4 A System of Logic, Bk. i, Chap. ii, §5.5 cf. Kant, Critique of Pure Reason, B 376.6 cf. my paper ‘Das Sein und das Nichts’, pp. 140-2.

Lecture 21

1 A detailed exposition of Husserl’s conception is to be found in my Wahr­heitsbegriff, §4. The most important texts for this problem are also cited there.

2 Critique of Pure Reason, B 236. The other most important texts are A 104 and B 137.

3 cf. esp. §19 of the second edition of the Critique of Pure Reason.4 Ideen, i, § 3 cf. also Logische Untersuchungen, n, 1, 125.5 cf. Ideen, n, §§15, 18.6 Ideen, i, §131.7 This statement must be considerably qualified, inasmuch as Frege, in his

Notes to pp. 2 8 8 -3 0 5 422

discussion of abstract objects, discovered the perspective of criteria of iden­tity, which is decisive for an analytical theory of objects. ‘If we are to use the sign a to designate an object, we must have a criterion for deciding in all cases whether b is the same as a, even if it is not always in our power to apply this criterion,’ (Foundations of Arithmetic, §62). In the same connection the concept of representation is also sharply rejected as a possible orienta­tion for the notion of objects (§60).

8 cf. Charles Morris, Foundations of the Theory of Signs, pp. 3 f.; see also his Signs, Language and Behaviour, p. 12; and Brown and Dulaney, ‘A Stimulus- Response Analysis of Language and Meaning’, pp. 75 ff.

9 In the English literature too the expression ‘to specify’ is occasionally used for the function that is here described, e.g. by Quine, Word and Object, p. 177. cf. Strawson, Papers, p. 60. The best expression with which I am famil­iar for what I mean here is the English expression ‘to single out’ (cf. Straw­son, ‘Singular Terms, Ontology and Identity’, p. 438); there is, however, no exact German equivalent for it.

Lecture 22

1 ‘On Referring’, pp. 17-19.2 The original and definitive exposition is to be found in the article ‘On

Denoting’. There is a more simple exposition in the lectures ‘The Philoso­phy of Logical Atomism’, in Logic and Knowledge, pp. 242 ff.

3 cf. the essay ‘Knowledge by Acquaintance and Knowledge by Description’, in Mysticism and Logic, p. 203.

4 Russell himself deals with general existential statements in The Philosophy of Logical Atomism, Lecture v. For Frege, c f.Foundations of Arithmetic, §53. Kant’s classical discussion of the problem in the Critique of Pure Reason, B 620 ff. does not go as far as his treatment of the concept of existence in the earlier writing ‘Der einzig mögliche Beweisgrund zu einer Demonstration des Daseins Gottes’ (1763) (Werke, ii, pp. 72 f). In the Critique Kant only says that existence is not a ‘real’ predicate, whereas in the early work he maintains that it is not a predicate at all and already interprets existential statements in the way Frege does.

5 Russell himself says this, Logic and Knowledge, p. 252. However he can uphold his original premise by saying: if something is ‘really a name, the question of existence could not arise, because a name has got to name something or it is not a name’ (p. 243).

6 Mysticism and Logic, p. 203, Logic and Knowledge, pp. 200 f.7 Logic and Knowledge, p. 243.8 ‘Über Sinn und Bedeutung’, note 2, ‘Der Gedanke’, pp. 38 ff.9 Wittgenstein, Phil. Inv., §79; Searle, ‘Proper Names’.

10 Kripke, ‘Naming and Necessity’; Donnellan, ‘Proper Names and Identifying Descriptions’.

11 From a Logical Point of View, pp. 5-8.12 Individuals, pp. 196 f.13 Logic and Knowledge, p. 201, Mysticism and Logic, p. 211.14 cf. above, p. 275 n. 2.15 From a Logical Point of View, p. 9.

Notes to pp. 3 0 6 -3 5 423

16 Logico-Linguistic Papers, p. 12.17 loc. cit., p. 11.18 cf. e.g. the introduction of Fodor and Katz to their collection The Structure

of Language.19 Papers, pp. 17-19.20 loc. cit., p. 16.21 loc. cit., p. 15.

Lecture 23

1 Logico-Linguistic Papers, p. 59.2 Individuals, p. 16.3 Individuals, pp. 31 f., Papers, p. 63.4'B . A. O. Williams, ‘Mr. Strawson on Individuals’, pp. 312 ff.5 Explicitly Wiggins, ‘The Individuation of Things and Places’, p. 183;

implicitly Searle, Speech Acts, p. 85.6 Papers, p. 63.7 Individuals, p. 61.8 Strawson, Papers, p. 63, Searle, p. 81, Wiggins, loc. cit., p. 183, Donnellan,

‘Reference and Definite Descriptions’, p. 285.9 ‘Singular Terms, Ontology and Identity’, p. 438.

10 Individuals, p. 61.11 Papers, pp. 60 f.12 cf. Papers, pp. 62, 63 f., Individuals, pp. 26 f.13 cf. Quinton, The Nature of Things, p. 15; Strawson, Individuals, pp. 26 f.14 Speech Acts, p. 85 (my emphasis).15 Individuals, pp. 18 f.16 loc. cit., p. 21.17 loc. cit., pp. 22-5.

Lecture 24

1 His distinction between direct and indirect identification (‘The Individua­tion of Things and Places’, p. 183) corresponds exactly to Strawson’s dis­tinction between demonstrative and non-demonstrative identification. The pre-eminence of locating identification is acknowledged (cf. esp. ‘Identity- Statements’, p. 44) but not argued for.

2 ‘The Individuation of Things and Places’, p. 184.3 I am here following Dummett, pp. 232-9.4 The distinction of K. Donnellan in his important paper ‘Reference and Def­

inite Descriptions’ between an ‘attributive’ and a ‘referential’ use of definite descriptions (p. 285) does not coincide with, but is connected with, my dis­tinction between an identifying and a non-identifying specification. One frequently uses non-locating definite descriptions in context in such a way that the speaker expects that it is understood which object he is identifying with the expression. One says, e.g., ‘the murderer of Hans’ and means this (locatable) man here. If the expression ‘the murderer of Hans’ is used in this way then according to Donnellan it is being used referentially. This gives rise to ambiguities to which Donnellan rightly draws attention. If I use the

Notes to pp. 3 3 8 -4 9 424

expression ‘the murderer of Hans’ on the tacit assumption that with it the (locatable) Mr XY is meant and I say, e.g., ‘The murderer of Hans is mad’ then if Hans was not murdered at all the statement can still be true pro­vided only that Mr XY is really mad. If on the other hand I am using the expression ‘the murderer of Hans’ attributively then with the sentence ‘The murderer of Hans is mad’ I mean: the one of all who murdered Hans, whoever (whichever locatable person) he may be, is mad; and in this case of course if Hans was not murdered the statement cannot be true. However correct these distinctions are and however devastating they appear to be against Strawson’s views with their unclear concept of identification, Don­nellan has neglected to place them in the context, intended by Strawson, of the question: How can one refer to perceptible objects? The peculiar role of locating definite descriptions thus remains unconsidered by him and conse­quently it does not become clear (1) that the non-locating definite descrip­tions only acquire an identificatory function by being combined with locat­ing definite descriptions (2) that his distinction between an attributive and a referential use can no longer be applied in the case of locating definite descriptions and (3) that even the so-called ‘attributive’ use of a definite description is in a broad sense of the word ‘referential’, in that though it does not identify an object it does specify it. Donnellan comes close to this last point when he says that even in the case of the ‘attributive use’ one can speak of ‘reference in a very weak sense’ (p. 303).

Lecture 25

1 ‘Sinn und Bedeutung’, p. 42.2 cf. also Wiggins, ‘The Individuation of Things and Places’, p. 179 and his

quotation from Leibniz on pp. 181 f.

Lecture 26

1 I do not think that pointing this out suffices to dispose of Kripke’s view; but it is here that the criticism would have to begin. Kripke himself points in this direction by making it clear that what he is concerned with is not the concept of a proper name but what he calls a ‘rigid designator’. A ‘rigid designator’ is a singular term ‘which in every possible world designates the same object’ (p. 269). Now Kripke expressly concedes that demonstratives can also function as ‘rigid designators’ (n. 16). And as he himself appears to admit that proper names can only be introduced by means of a demonstra­tive (p. 302) he has thereby implicitly retracted his thesis of the primary and free-floating proper name—object relation which is independent of any descriptions. There are, however, several other important aspects of his theory which are not touched by this, in particular his view (which is also held by Donnellan in ‘Proper Names and Identifying Descriptions’) that with the name of an object which we only know by hearsay we do not, as Wittgenstein (Phil. Inv., §79) and Searle (‘Proper Names’) believed, mean the object to which the majority of the descriptions which are transmitted to us with the name apply but the object which those who have transmitted the

Notes to pp. 3 5 6 -7 1 425

name so designated. For criticism of Kripke, cf. Dummett, Frege, pp. 110— 51.

2 The essence of this, and of the thoughts connected therewith, can already be found in Hampshire, p. 16.

3 Grundl. d. Arithmetik, §54.4 For the discussion of sortals cf. esp. Strawson, Individuals, pp. 168 ff.;

Quine, Word and Object, §19; Wiggins, Identity and Spatio-Temporal Continuity, §2; Carl, Existenz und Prädikation, §6. In a certain way the modern concept of the sortal predicate represents a re-discovery of the Aristotelian concept of the substance-predicate (cf. Wiggins, loc. cit., p. 28); however the two concepts do not correspond completely. Aristotle distinguishes a hierarchy of substance-predicates; the lower ones he also calls matter-predicates (e.g. ‘water’, ‘sand’), while what distinguishes the genuine substance-predicates is that they are shape-predicates and contain a principle of countability (hen arithmo). In the early modern period the Aristotelian insight was no longer understood; substance appeared as a substrate that, itself not perceptible, underlies a bundle of perceptible qualities (Locke) and could therefore be rejected (Hume). The conception that results from this, that objects are spatio-temporally instantiated bundles of properties which could be referred to with the word ‘this’, persisted in British Empiricism until Russell and was first overcome through the rediscovery of the sortal predicate (cf. Geach, Reference and Generality, pp. 43 f.)

5 This is not to say that the application-rule of such a predicate is simply a matter of the occurrence of a particular shape; Aristotle in particular pointed out that we do not call something a chair or a cat because it has a certain shape, but because it fulfils a certain function, and that the particu­lar configuration is only a consequence of the fact that something can only fulfil its function under this condition. But this does nothing to alter the fact that it is the configuration diat contains the criterion of identification and distinction.

6 on this aspect cf. esp. Wiggins, loc. cit., pp. 29-36.7 cf. Wiggins, loc. cit.8 cf. Strawson, Individuals, pp. 46 ff., Goldman, p. 10. In the important

papers by Davidson about events (esp. ‘The Individuation of Events’) this error is not committed; but it is also not said what distinguishes events from states. By contrast cf. von Wright, Norm and Action, pp. 27 f.

9 However, differentiations are possible and necessary here: cf. Vendler, ‘Facts and Events’, where of course events are likewise not explicitly distin­guished from states.

10 The following reflections on the nature of events I owe to a suggestion by Miss R. Kronseder. cf. also von Wright, loc. cit.

11 This conception has been developed independently by Quinton (pp. 38 f.) and myself (‘Existence in Space and Time’).

12 Speech Acts, p. 77.13 That this formulation is also implicitly contained in existential (particular)

statements has been shown above, p. 246.14 I have dealt with this ‘temporal concept of existence’ in my paper ‘Existence

in Space and Time’ and have put forward the thesis there that temporal

existence is not a two-place predicate but is reducible to the existential quantifier; I do not know whether this is correct.

Notes to pp. 3 7 5 -8 9 426

Lecture 27

1 On this problem of circularity with respect to the second of the reciprocal dependencies cited above cf. Wiggins, ‘The Individuation of Things and Places’.

2 Here and in what follows I am using the term ‘identification’ in a more comprehensive sense than hitherto. I had hitherto called only the specifica­tion of locating expressions identification. The phenomenon of reference (Verweisung), however, allows one also to speak of an identification and of a rule of identification in the case of non-locating singular terms. When we understand a non-locating definite description we know which object it specifies though we do not yet know which it identifies, but we know how one establishes which it identifies and that means: we know its identification-rule (cf. p. 331).

3 That is Leibniz’s law according to which a —b if and only if (0) 0a = 0b, hence if every statement in which ‘a’ is supplemented by an arbitrary predi­cate is true (or false) if and only if the statement which results when ‘6’ is supplemented by the same predicate is true (or false). This law holds abso­lutely if one restricts predicates to perceptual predicates, hence if one does not allow intensional predicates for which the law does not hold. Dummett elucidates the law thus: ‘if a predicate applies to an object, it applies to it by whatever means we refer to that object’ (Frege, p. 265). One can take over this elucidation and yet reject Dummett’s view that reference to an object and the standing of a name for an object has a meaning before the use of the identity-sign, a meaning which supposedly is given through the osten- sive reference to an object (p. 406).

Can one say that through the equivalence of the Leibnizian law the mean­ing of * = ’ is defined? Only then can what is said above in the text be under­stood as a description of an explanation of ‘ = \ However, one can only say: that someone has understood the meaning of ‘ = ’ is shown by the fact that under certain circumstances for any *F' he infers from the truth of Fa that of F b’ and vice versa. But this does not yet tell us what these ‘certain cir­cumstances’ are which justify his doing this. In the case of perceptible objects these circumstances consist in this, that a and b have the same spatio- temporal relations to other objects. This, which one can call the identity- criterion (Dummett, pp. 544 f., above p. 381), differs from object-sphere to object-sphere. And indeed it is precisely the identity-criterion through which an object-sphere is constituted (or through which objects of this sphere are constituted as objects (of this sphere)). Only if someone learns the use of the singular terms of an object-sphere in such a way that on the basis of the identity-criterion which obtains for this object-sphere he can substitute them for one another in the manner described by the Leibnizian law has he understood the sign ‘ = ’ for this object-sphere, and hence what it is for a singular term of this kind to ‘stand for’ an object (of this sphere).

4 cf. Wittgenstein, Zettel, §§411-425.5 cf. Lemmon, ‘Sentences, Statements and Propositions’.

Notes to pp. 3 8 9 -4 0 6 427

6 cf. e.g. Patzig, ‘Satz und Tatsache’.7 cf. ‘Sinn und Bedeutung’, p. 32, and the still narrower identity-criterion in

Chisholm, ‘Problems of Identity’, pp. 24 f.8 cf. Frege, ‘Der Gedanke’, p. 64 and the interpretation in Dummett, Frege, p.

384. Neither philosopher, however, heeds the distinction to which I seek to refer in the sentence which follows in the text.

Lecture 281 The treatment in Dummett is instructive, Frege, pp. 70—80, 370-81 and

Chap. 14; also in Specht, parts in 8c iv. Cf. also Carl, pp. 153-68.2 cf. Dummett, Frege, pp. 70 f.3 ‘Semantics for Natural Languages’, p. 178.4 For what follows cf. Davidson ‘Truth and Meaning’, p. 317.5 On the semantics of adverbs cf. Davidson, ‘The Logical Form of Action

Sentences’ as well as Bartsch, ‘Die logische Analyse von Modaladverbien’ and Bartsch, Adverbialsemantik. Cf. also for the solution of a related semanti­cal problem Bartsch and Vennemann, Semantic Structures, Chap. 3.

6 cf. the table on pp. 66 f.7 Speech Acts, 5.6.8 cf. p. 122 8c p. 124.9 cf. Dummett, Frege, p. 303.

10 cf. Hare, The Language, of Morals, 2.3.11 cf. Dummett, Frege, p. 303.12 cf. Dummett, Frege, p. 303.13 cf. Anscombe, Intention, p. 56 and esp. Kenny, Action, Emotion and Will, pp.

22 f., and Kenny, Will, Freedom and Power, p. 38.14 cf. Kenny, Will, Freedom and Power, p. 38, also Stenius, ‘Mood and Language

Game’, above all p. 274.15 cf. above p. 54 n 9 and the literature cited there.16 Anscombe gives on p. 56 the example of a man who makes purchases in a

grocery shop by means of a list: if the list comes from his wife then it is an imperative; if it is his own list then it is an intention-sentence.

17 The most extensive analysis so far of these connections is probably achieved in the two books by Kenny. See esp. Chap. 4 of the new book.

18 At a lowest level the word ‘good’ is already used as a simple practical affir­mation: instead of replying to a demand or a suggestion with ‘yes’ one can also say ‘good’; thus we also call something good when we agree with it or when it fulfils our wishes. In most contexts, however, the word is used in such a way that it implies not just practical agreement but an objectively jus­tifiable agreement.

19 Will, Freedom and Power, pp. 39 f.20 cf. Kenny, Wittgenstein, p. 121.21 cf. his book Freedom and Reason.22 The connection between the theories o f Grice and Lewis has been made

clear by Bennett in ‘The Meaning-Nominalist Strategy’. Bennett who has sketched a conception based on both authors has in contrast to Grice and Lewis explicitly drawn attention to the problem of a language-independent

Notes to pp. 409 428

explanation of meaning and intending which then becomes necessary (p. 145).

23 cf. the title of the first part of Sein und Zeit.24 c£. Notebooks, p. 30 130e.25 cf. Notebooks, pp. 93 f.

Bibliography

Alston, W. P., Philosophy of Language, Englewood Cliffs 1964.Anscombe, G. E. M., Intention, Oxford 1957.Aristotle, Opera, ed. of Prussian Academy, Berlin 1831 ff.Augustine, Confessions.Austin, J. L., How to do Things with Words, Oxford 1962.Ayer, A. J. ‘Negation’, Journal of Philosophy, 49 (1952); reprinted in: A. J. Ayer,

Philosophical Essays, London 1963, pp. 36-65.Bartsch, R., ‘Die Logische Analyse von M odaladverhien’, Linguistische Bericht, 10

(1970) pp. 27-34.- Adverbialsemantik, Frankfurt 1972.Bartsch, R. and Vennemann, T., Semantic Structures, Frankfurt 1972.Bennett, J., ‘The Meaning-Nominalist Strategy’, Foundations of Language, 10

(1973), pp. 141-68.Bloomfield, L., Language, London 1935.Brentano, F., Psychologie vom empirischen Standpunkt, 2 vols. (1874) ed. by

O. Kraus, Leipzig 1924.Brown, R. W. and Dulaney, D. E., ‘A Stimulus-Response Analysis of Language

and Meaning’, in P. Henle (ed.), Language, Thought and Culture, Ann Arbor 1958.

Carl, W., Existenz und Prädikation, Munich 1974.Camap, 'K., Meaning and Necessity, Chicago 1947.Cartwright, R., ‘Propositions’, in R. J. Butler {ed.), Analytical Philosophy, Oxford

1962, pp. 1-103.Chisholm, R., Perceiving, Ithaca (N.Y.) 1957.- ‘Problems of Identity’, in M. K. Munitz (ed.), Identity and Individuation, New

York 1971, pp. 3-30.Davidson, D., ‘The Logical Form of Action Sentences’, in N. Rescher (ed.), The

Logic of Decision and Actwn, Pittsburgh 1966, pp. 81-95.- ‘Truth and Meaning’, Synthese, 17 (1967) pp. 304-23.- ‘The Individuation of Events’, in N. Rescher (ed.), Essays in Honor of Carl G.

Hempel, Dordrecht 1969, pp. 216-34.- ‘Semantics of Natural Languages’ in Linguaggio nella societa e nella technica,

Milan 1970, pp. 177-88.Donnellan, K. S., ‘Reference and Definite Descriptions’, Philosophical Review, 75

(1966), pp. 281-304.Dummett, M., ‘Truth’, Proceedings of the Aristotelian Society, suppl. vol. 59 (1958-

Bibliography 430

9), reprinted in G. Pitcher (ed.), Truth, Englewood Cliffs 1964, pp. 93-111.- Frege; Philosophy of Language, London 1973.Duns Scotus, J., ‘Ordinatio’, in: Opera Omnia (ed. Balic), Vatican 1954, vol. 3. Fodor, J. A. and Katz, J. J. (eds.), The Structure of Language, Englewood Cliffs

1964.¥ rege, G., Begriffsschrift, Halle 1879.- Die Grundlagen der Arithmetik, Breslau 1884.- ‘Funktion und Begriff’, Jena 1891. Reprinted in G. Frege, Funktion, Begriff,

Bedeutung, ed. by G. Patzig, Gottingen 1962.- ‘Sinn und Bedeutung’, Zeitschrift fü r Philosophie und philosophische Kritik, NF

100 (1892), pp. 25-50. Reprinted in G. Frege, Funktion, Begriff, Bedeutung, ed. by G. Patzig, Göttingen 1966.

- ‘Begriff und Gegenstand’, Vierteljahrschnft fü r wissenschaftliche Philosophie, 16(1892), pp. 192-205. Reprinted in G. Frege, Funktion Begriff, Bedeutung, ed. by G. Patzig, Göttingen 1962. (

- ‘Der Gedanke’, Beitrage zur Philosophie des deutschen Idealismus, I (1918-19) pp.58-77 reprinted in G. Frege, Logische Untersuchungen ed. by G. Patzig, Göt­ti ngen 1966.

- ‘Die Verneinung’, in Beitrage zur Philosophie des deutschen Idealismus, I (1918—19), pp. 143-75. Reprinted in: G. Frege, Logische Untersuchungen, ed. by G. Patzig, Göttingen 1966.

-Nachgelassene Schriften, edited by H. Hermes, F. Kambartel and F. Kaulbach, Hamburg 1969.

Gale, R. M., ‘Negative Statements’, American Philosophical Quarterly, 7 (1970) pp. 206-17.

Geach, P. T., ‘On What There Is\P roc. Arist. Soc., suppl. vol. 25 (1951), pp. 125-36.

-Reference and Generality, Ithaca (N.Y.) 1962.- ‘Assertion’, Philos. Rev., 74 (1965), reprinted in P. T. Geach, Logic Matters,

Oxford 1972, pp. 254-69.Goldmann, A ., A Theory of Human Action, Englewood Cliffs 1970.Grice, H. P., ‘Meaning’, Philos. Rev., 66 (1957), pp. 377—88, reprinted in: P. F.

Strawson (ed.), Philosophical Logic, Oxford 1967.- ‘Utterer’s Meaning, Sentence-Meaning and Word-Meaning’, Found of Lang., 4

(1968), reprinted in J. R. Searle (ed.) The Philosophy of Language, Oxford 1977, pp. 54-70 (references to latter).

- ‘Utterer’s Meaning and Intentions’, Philos. Rev., 78 (1969), pp. 147-77. Griffiths, A. P., Knowledge and Belief, Oxford 1967.Habermas, J., ‘Was heisst Universalpragmatik?’ in K. O. Apel (ed.), Sprachprag-

matik und Philosophie, Frankfurt 1976.Hampshire, S., Thought and Action, London 1959.Hare, R. M., The Language of Morals, Oxford 1952.- Freedom and Reason, Oxford 1963.- ‘Meaning and Speech Acts’ (1970), reprinted in R. M. Hare, Practical hifer-

ences, London 1971.Heidegger, M., Sein und Zeit, Halle 1927.Henrich, D., Fichtes ursprüngliche Einsicht, Frankfurt 1967.- ‘Selbstbewusstsein’ in Bubner, Cramer, Wiehl (eds.), Hermeneutik und Dialektik,

Tübingen 1970. vol. I, pp. 257-84.

Bibliography 431

Hume, D., A Treatise of Human Nature, London 1738.Husserl, E., Logische Untersuchungen, 2 vols., 2nd edition, Halle 1922.- Ideen zu einer reinen Phänomenologie und phänomenologischen Philosophie, 3 vols.

in Husserliana, vols. m -v, The Hague 1950-52.- Car tesianische Meditationen, in Husser liana, vol. i, The Hague 1950.Jones, O. R. (ed.), The Private Language Argument, London 1971.Kahn, C., The Verb ‘Be’ in Ancient Greek, Dordrecht 1973.Kambartel, F., Erfahrung und Struktur, Frankfurt 1968.Kant, L, Critique of Pure Reason (references to 2nd edition (‘B’)).- Gesammelte Werke, ed. by Prussian Academy, Berlin 1902 ff.Kenny, A ., Action, Emotion and Will, London 1963.- Wittgenstein, London 1973.- Will, Freedom and Power, Oxford 1975.Kripke, S., ‘Naming and Necessity’ in D. Davidson and G. Harman (eds.),

Semantics of Natural Languages, Dordrecht 1972, pp. 253-355.Kutschera, F. von., Sprachphilosophie, Munich 1971.Lemmon, E. J., ‘Sentences, Statements and Propositions’ in B. Williams and A.

Montefiore (eds.), British Analytical Philosophy, London 1966, pp. 87-108. Lewis, C. I., An Analysis of Knowledge and Valuation, La Salle 1946.Lewis, D., Convention, Cambridge (Mass.) 1969.Lorenzen, P., Methodisches Denken, Frankfurt 1968.Lyons, J., Introduction to Theoretical Linguistics, Cambridge 1968.Mead, G. H., Mind, Self and Society, Chicago 1934.Mill, J. S., A System of Logic, London 1843.Mittelstrass, J.., ‘Die Pradikation und die Wiederkehr des Gleichen’ in H. G.

Gadamer (ed.), Das Problem der Sprache, Munich 1967, pp. 87-96.Morris, C., Foundations of the Theory of Signs, Chicago 1938.- Signs, Language a7id Behaviour, New York 1946.Oehler, K., Die Lehre vom noetischen und dianoetischen Denken bei Platon und Aris­

toteles, Munich 1962.Passmore, J., A Hundred Years of Philosophy, London 1957.Patzig, G., ‘Satz und Tatsache’, in G. Patzig, Sprache uiui Logik, Göttingen 1970,

pp. 39-76.Peirce, C. S. Collected Papers, 6 vols., Cambridge (Mass.) 1931.Pfänder, A., Logik, Tübingen 1963.Pitcher, G. (ed.), Truth, Englewood Cliffs 1964.Plato: Opera Omnia (ed. Burnet).Quine, W. V. O., From a Logical Point of View, Cambridge (Mass.) 1953.- Word and Object, Cambridge (Mass.) 1960.Quinton, A., The Nature of Things, London 1973.Ramsey, F. P., The Foundations of Mathematics, London 1931.Rorty, R., The Linguistic Turn, Chicago 1967.Russell, B., ‘On Denoting’, M ind 14 (1905), pp. 479-93, reprinted in B. Russell,

Logic and Knowledge.- The Problems of Philosophy, London 1912.- Mysticism and Logic, London 1918.-Introduction to Mathematical Philosophy, London 1919.- Mr Strawson on Referring’, Mind 66 (1957), pp. 385-89, reprinted in B. Rus­

sell, My Philosophical Development, London 1959, pp. 175-180.

Bibliography 432

- Logic and Knowledge, London 1956.Savigny, E. von., Die Philosophie der normalen Sprache, Frankfurt 1969.Schelling, F. W., System des transzendentalen Idealismus (1800), in Werke (ed.

Schröter) vol. 2, pp. 327 ff.Searle, J., ‘Proper Names’, M ind, 67 (1958), pp. 166-73.- ‘Austin on Locutionary and Illocutionary Acts’, Phil. Rev., 72 (1968).- Speech Acts, Cambridge 1969.Specht, E. K., Sprache und Sein, Berlin 1967.Stegmüller, W., Hauptströmungen der Gegenwartsphilosophie, 2 vols., Stuttgart

1975.Stenius, E., ‘Mood and Language-Game’, Synthese, 17 (1967), pp. 254-74. Stevenson, C., ‘The Emotive Meaning of Ethical Terms’, MzW, 46 (1937), pp.

14-31.-E th ics and Language, New Haven/London 1944.Strawson, P. F., ‘On Referring’, Mind, 59 (1950), reprinted in P. F. Strawson,

Logico-Linguistic Papers.- Introduction to Logical Theory, London 1952.- ‘Singular Terms, Ontology and Identity’, Mind, 65 (1956), pp. 433-54. -Individuals, London 1959.- ‘Singular Terms and Predication’, Journal of Philosophy, 58 (1961), reprinted

in Strawson, Logico-Linguistic Papers (references to latter).- Logico-Linguistic Papers, London 1971.Tarski, A., ‘Der Wahrheitsbegriff in den formalisierten Sprachen’, Studia Philo-

sophica, I (Lwow 1936), pp. 261-405.St Thomas Aquinas, ‘Quaestiones Disputatae de Veritate’, Quaestiones Disputa-

tae, I, Rome 1949.Tugendhat, E., Ti kata tinos, Eine Untersuchung zu Struktur und Ursprung aristotel­

ischer Grundbegriffe, Freiburg 1958.- Der Wahrheitsbegriff bei Husserl und Heidegger, Berlin 1967.- ‘Die sprachanalytische Kritik der Ontologie’ in H. G. Gadamer (ed.), Das

Problem der Sprache, Munich 1967, pp. 483-93.- ‘Phänomenologie und Sprachanalyse’, in Bubner, Cramer, Wiehl (eds.), Her­

meneutik und Dialektik, Tübingen 1970, vol. 2, pp. 3-24 (English: ‘Pheno­menology and Linguistic Analysis’ in F. A. Elliston and P. McCormick (eds.), Husserl: Expositions and Appraisals, Univ. of Notre Dame Press 1977, pp. 325-37).

- ‘Das Sein und das Nichts’, in V. Klostermann (ed.), Durchblicke; Festschrift fü rM. Heidegger, Frankfurt 1970, pp. 132-61.

- ‘The Meaning of “Bedeutung” in Frege’, Analysis, 30 (1970), pp. 177-89.- ‘Existence in Space and Time’, Neue Hefte fü r Philosophie, 8 (1975), pp. 14-33. Urmson, J. O., Philosophical Analysis, Its Development between the Two World Wars,

Oxford 1956.Vendler, Z., ‘Facts and Events’ in Z. Vendler, Linguistics in Philosophy, Ithaca

(N.Y.) 1967, pp. 122-46.Wiggins, D., ‘The Individuation of Things and Places’, Proc. Anst. Soc., suppl.

vol. 37 (1963), pp. 177-202.- ‘Identity-Statements’, in R. J. Butler {ed.), Analytical Philosophy, 2nd series,

Oxford 1965, pp. 40-71.

Bibliography 433

- Identity and Spatio-Temporal Continuity, Oxford 1967.Williams, B. A. O., ‘Mr Strawson on Individuals’, Philosophy, 36 (1961), pp.

309-32.Wittgenstein, L.,Tractatus Logico-Philosophicus, London 1961.-B lu e and Brown Books, Oxford 1958.- Philosophical Investigations, Oxford 1973.- Remarks on the Foundations of Mathematics, Oxford 1974.-Z ettel, Oxford 1967.- Philosophische Bemerkungen.- Philosophische Grammatik.Von Wright, G. H., The Varieties of Goodness, London 1963.- Norm and Action, London 1963.Ziff, P., Semantic Analysis, Ithaca (N.Y.) 1960.

Index of names

Alston, W. P., 187 Anscombe, G. E. M., 427 Aquinas, 413, 415, 420 Aristotle, 9, 14ff, 18-21, 22f, 28-31,

34f, 38-42, 50, 86-9, 148, 155, 195,254, 413, 417, 418, 419, 425

Augustine, 8Austin, J. L., 186f, 401, 404, 414, 420 Ayer, A. J., 413

Bartsch, R., 427 Bennett, J., 427 Bloomfield, L., 167, 419 Brentano, F., 70, 72 Brown, R. W., 419, 422

Carl, W., 425, 427 Carnap, R., 99, 189, 307, 419 Cartwright, R., 418, 421 Chisholm, R., 416, 427

Davidson, D., 99, 189, 238f, 253, 395f,412, 425, 427

Descartes, 57f, 67ff, 77 Donnellan, K. S., 301, 325, 423, 424 Dulaney, D. E., 419, 422 Dummett, M., 118, 189, 198,325,412,

413, 418, 420, 421, 424, 426, 427 Duns Scotus, 413

Fodor, J. A., 423Frege, G., 26, 43, 46f, 99, 107, 109f,

l l l f , 115f, 117f, 146-9, 189, 208, 243, 246f, 275f, 279, 287f, 300, 301, 338, 348, 358f, 389, 414, 421, 422, 425, 426, 427

Gale, R. M., 413 Geach, P. T., 412, 413, 421, 4-25 Goldman, A., 420, 425 Grice, H. P., 181ff, 185ff, 190, 194,

201, 21 Of, 216f, 31 lf; 406, 419

Habermas, J., 414 Hampshire, S., 417, 425 Hare, R. M., 404, 414, 417, 419, 427 Hegel, 50, 63, 278 Heidegger, M., 60, 65, 69, 409, 413,

414, 415, 416, 417 Henrich, D., 417 Hume, 156, 425Husserl, E., 14, 22f, 24, 43f, 58f, 61,

62f, 66-71, 73f, 77, 99, 102f, 104f, 107, 132, 134, 136f, 140f, 145-8, 229-33, 275f, 284-8

Kahn, C., 413 Kambartel, F., 414 Kant, 6, 9-11, 58ff, 64f, 76, 145f, 285f,

300, 417, 418, 421 Katz, J., 423Kenny, A., 403, 417, 427 Kripke, S., 301, 325, 348, 421, 424 Kronseder, R., 425 Kutschern, F. von, 411

Leibniz, 424, 426 Lemmon, E. J., 425 Lewis, C. I., 418 Lewis, D., 406, 414, 420 Locke, 425 Lorenzen, P., 149 Lyons, J., 413

Index o f Names 435

Mead, G. H., 168ffMeinong, A., 298ffMill, J. S., 272, 276-80, 288, 417, 418Mittelstrass, J., 419Moore, G. E., 417Morris, C., 419, 422

Oehler, K., 421

Parmenides, 14, 33, 50, 378 Passmore, J., 411 Patzig, G., 427 Peirce, C. S., 212, 218- Pfänder, A., 421 Pitcher, G., 418Plato, 8f, 19f, 29f, 34, 77, 94, 411 Prior, A., 414

Quine, W. V. O., 302, 305, 305n.l5, 307, 313, 412, 418, 419, 422, 425

Quinton, A., 423, 425

Ramsey, F. P., 420 Rorty, R., 411Russell, B., I l l , 275, 298-310, 314f,

3l7f, 330, 332, 335f, 344, 368-70, 378, 411, 425

Savigny, E. von, 411 Schelling, F. W., 420

Searle, J., lOlf, 147, 186ff, 193, 198f, 202f, 313, 319, 368, 398-402, 404,414, 416, 420, 422, 423

Socrates, 13, 94 Specht, E. K., 427 Stegmüller, W., 411 Stenius, E., 427 Stevenson, C., 417, 419 Strawson, P. F., 268, 298, 302, 326,

329f, 332, 344, 349, 362, 366, 384, 412, 421, 422, 423, 425

Urmson, J. O., 411

Vendler, Z., 425 Vennemann, T., 427

Wiggins, D., 325-8, 331, 333f, 382, 423, 424, 425

Williams, B. A. O., 312 Wittgenstein, L., 8, 43f, 68f, 99, 104,

122f, 142f, 144 ,150-3 ,153n.l, 157f, 157n.8, 160, 160n.9, 163-7, 177, 182, 189, 198, 200, 209f, 237f, 248,255, 270, 307, 321, 354, 377, 403, 406, 409, 422, 424, 426

Wright, G. H. von, 417, 425

Ziff, P., 417

Index of subjects

affirmation, 46ff, 52f ‘all’, 244-7analytic statements, 8ff ‘and’, 228-42 answering, 190 applying, 253-6, 265f a priori, the, 8ff, 24 assertion, 185, 188f, 202f, 218f, 222,

350assertion-moment, 44f assertion-sign (in Frege), 47 attributes, 128f, 134, 158f

behaviouristic theory of language, 166f, 170-3

belief, 184, 211-14

categorematic terms, 109 categorial synthesis, 99, 122-6, 130ff changes, 36If characterization, 135f, 138f classification, 138classification-expressions, 25, 26If,

407fcommunication, 2.16f composition (of meanings or states of

affairs), 120f, 127f, 130, 234, 244 concepts, 145-9 conceptualism, 139-44, 153-8 conditional rules, 165f, 170, 175, 239f consciousness, 56-60, 62, 65-74 coordinate zero-point (for spatio-

temporal identification), 345f correctness, 82f, 143, 352, 354f, 386ff,

409correspondence theory of truth,

195ffcountability, 258f, 295, 425

deictic expressions, 110, 221-4, 308f, 341, 343, 389

demonstrative identification, 316ff,329

descriptions, 11 Of, 275fdescriptive descriptions, 314 locating (spatio-temporal) descrip­

tions, 325n.l, 329-32, 335n.4 relational descriptions, 3 l7ff Russell’s theory of descriptions,

298-301 determinism, 79disclosedness (in Heidegger), 60, 414,

415

error, 352ff events, 360ffexistence, 28, 72, 299ff, 367-71, 416,

425explanation (of an expression), 142f,

164, 238f, 255f, 392

formalization, 24-7, 55f, 66, 75 freedom, 79, 81, 85, 409 function, 136-7

in Frege’s terminology, 146 of linguistic expressions, 135f, 289 of predicates, 134ff of singular terms, 289ff, 293 of assertoric sentences, 183f, 185f

general statements, 244 good, 81-5, 88f, 403f, 427 guarantee, 199f, 201 f

identification, 298,310-21, 325ff, 331, 334f, 345f, 378f, 383, 426

kinds of identification, 316-19

Index o f Subjects 437

identity, 23, 294f, 340f, 344, 347, 358, 364, 379, 386f, 426

illocutionary acts, 187, 398f, 403 imperatives, 53, 80, 400f, 402f, 406f individuals, 155, 294f, 304f, 370 informing, 18If, 211 inner perception, 66ff, 71 intellectual intuition, 9, 141 intentional actions, 78f intentional consciousness, 69-74 intentionality, 67, 69-74, 108, 284f intention-sentences, 40, 72f, 402f intuition of essences, 70, 141

justification, 14ff, 82ff, 87

linguistics, 5f, 96, 152, 410 localization, 314, 316f, 329, 331, 345f,

373f

meaning, 102f, 104f, 142f, 152, 221f (in sense o f ‘meaning an object’), 63,

282, 289, 293, 336 (in sense of vouloir dire), 182f, 186f,

194, 210f meta-linguistic theories of meaning,

183, 201, 223, 258f, 271, 307, 342 modes (of sentences), 40, 51f, lOlf,

401-4, 408 modes of presentation (or givenness),

58, 61, 1 Ilf, 285ff

negation, 46ff, 52f, 408f ‘no’, 47fnominalization, 29, 42 nominalism, 139-44, 153-8 ‘not’, 408f

objectivity, 285ff objects, 21-25, 280-1, 292, 294f

abstract objects, 122, 295f, 393ff concrete objects, 275, 295f, 33Of,

356ffcontrast of analytical with tradi­

tional conception of objects, 375-81

opining, 212ff optative sentences, 400-5 ‘or’, 233, 236f, 241

particular statements, 244 perceptual predicates, 330f perceptual situation, 329, 340, 344,

356

performatives, 52f phenomenology, 66f position-taking, 169, 185, 190 pragmatics, 56, 74predicates, 25, 133-6, 138f, 143, 158f,

261-6, 385-6, 399f Principle of contradiction, 38-41 proper names, 111, 275f, 301 f, 317,

373, 420 in Frege’s terminology, 109 logically proper names, 303f, 368f

quantifiers, 244quasi-predicates, 159ff, 164, 178,

261-4, 273, 351-5

recursive truth-definition, 250 redundancy theory of truth, 195-8 reference (Frege’s Bedeutung), 11 If reference, Bezugnahme (to objects),

27If, 279f, 294, 335f, 343f, 379ff, 392, 394f

reference, Verweisung (of singular terms to other singular terms), 340f, 372-5, 378f, 395

representation (Vorte Hung), 62ff, 140f, 145,276-83, 375f

representative theory of signs, 289ff, 376

self-consciousness, 69, 74 sets, 230ffsignals, signal-language, 166f, 168f,

172f, 183, 406ff situation-conformity, 35If situation-dependence (-indepen­

dence), 161, 163f, 174, 206, 223f, 350,355f

‘some’, 244ff ‘something’, 22-40 sortals, 358-65spatio-temporal positions, 330, 336ff,

357f, 363, 364-7, 369f, 416 spatio-temporal relations, 316f, 319f,

330specification, 2 9 3 f, 313ff, 326 f, 384f,

426kinds and levels of specification,

319f, 329, 332, 335 standing for, 253-6, 270, 279f, 293f,

376ff, 380f states of affairs, 43f, ll7f, 122,219-24

criterion of identity for states of affairs, 389

Index of Subjects 438

subject-object relation, 64, 74, 278, 392

substitutability (replaceability of sing­ular terms by other singular terms), 308, 340f, 342f, 378f, 386

symptoms, 181, 290 synthesis, 122, 124f, 130, 134, 146,

151, 233f, 254, 285ff

‘this’, 262-5, 302ff, 342, 359 thought (in Frege), 43, 117 transcendental philosophy, 58-65 truth, 83f, 135f, 189, 194-8, 207ff,

224f, 249f, 254, 341, 350-6, 387f, 398

truth-conditions, 99f, 199-206, 22Iff of predicative sentences, 381-4

truth-definition (Tarskian), 327ff, 381, 396ff, 401

truth-functional senences, 228, 243 truth-tables, 237

universal statements, 244

verification-field, 370f verfication-rule, 202-5, 227

of predicates, 261, 264f, 283f verification-situation, 340f, 363, 370

what is asserted (Behauptetes), 42f, 185, 218-19, 222, 390