1 Absolute Truth in a Changing World Peter Simons University of Leeds The fundamental things apply As time goes by Herman Hupfeld 1 Philosophical Method and the History of Philosophy, Especially in Poland With Jan Woleński I share a great admiration for the achievements of Polish philosophers and logicians from Twardowski onwards, and we likewise share a fascination for their personalities, foibles and vicissitudes. More importantly, we both agree that Polish logic and analytical philosophy got the balance about right between philosophy and its history. Other things being equal, it is better – for philosophy – to be a good philosopher who is ignorant of the subject’s history than a good historian with no good sense of what is philosophically important. But other things are not equal, and it is possible to both have a good nose for philosophical importance and plausibility, as well as being sensibly informed of relevant portions of the subject’s history. The reason is not simply that by knowing the history one is able to avoid tumbling into the pitfalls of the past or wasting time reinventing theories that have already been invented. It is also that the historical dimension lends depth to one’s appreciation of the problems themselves, and gives one a sense of the historical element in any current discussion. No one philosophizes in a vacuum and it is folly to suppose otherwise. When the history is relatively recent, as with the history of Polish philosophy from 1895 to 1939, some of the issues are likely still to be with us. One such issue is the philosophy of truth, about which Jan and I collaborated some years ago in a long historical essay. i An aspect of it is the subject of this essay. Knowledge of the past should not imply slavish adherence to past views. Which ones would they be? Philosophers were as divided then on doctrine as they are now. Even
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Absolute Truth in a Changing World
Peter Simons
University of Leeds
The fundamental things apply
As time goes by
Herman Hupfeld
1 Philosophical Method and the History of Philosophy, Especially in Poland
With Jan Woleński I share a great admiration for the achievements of Polish philosophers
and logicians from Twardowski onwards, and we likewise share a fascination for their
personalities, foibles and vicissitudes. More importantly, we both agree that Polish logic and
analytical philosophy got the balance about right between philosophy and its history. Other
things being equal, it is better – for philosophy – to be a good philosopher who is ignorant of
the subject’s history than a good historian with no good sense of what is philosophically
important. But other things are not equal, and it is possible to both have a good nose for
philosophical importance and plausibility, as well as being sensibly informed of relevant
portions of the subject’s history. The reason is not simply that by knowing the history one is
able to avoid tumbling into the pitfalls of the past or wasting time reinventing theories that
have already been invented. It is also that the historical dimension lends depth to one’s
appreciation of the problems themselves, and gives one a sense of the historical element in
any current discussion. No one philosophizes in a vacuum and it is folly to suppose
otherwise. When the history is relatively recent, as with the history of Polish philosophy from
1895 to 1939, some of the issues are likely still to be with us. One such issue is the
philosophy of truth, about which Jan and I collaborated some years ago in a long historical
essay.i An aspect of it is the subject of this essay.
Knowledge of the past should not imply slavish adherence to past views. Which ones
would they be? Philosophers were as divided then on doctrine as they are now. Even
2
philosophy makes progress, albeit somewhat crabwise, so mere repetition is a waste of time.ii
That is why it is better to have a good new idea in ignorance of its novelty than to simply
bang on about what X said then, as if that solved a problem. There are outstanding examples
of historically-informed philosophers and logicians among the Lvov–Warsaw School:
Łukasiewicz among the logicians, Ajdukiewicz and Kotarbiński among the philosophers. Yet
all three were considerable and innovative thinkers. Indeed the history of logic as a modern
subject started with Łukasiewicz. And here is the historical part of the explanation. The
founder of the School, Kazimierz Twardowski, was himself historically aware. His teacher
Brentano was both a philosophical innovator and a knowledgeable historian who derived
inspiration from the past. Twardowski also – rare for his time, rare even today– appreciated
Bolzano, whose own great Wissenschaftslehre of 1837 was subtitled “Attempt at a
comprehensive and mainly new exposition of logic with constant attention to its previous
authors”.iii
Perhaps Twardowski’s most important and influential paper was one which he
published in 1900, called “O tzw. prawdach względnych”, “On so-called relative truths”.iv In
it, Twardowski takes issue with those who claim that truth should be considered a property
relative to time, or place, or speaker, or anything else. His argument is that since ordinary
language is mainly a practical tool, it is replete with ellipsis. If someone says “It’s raining”,
the words of which alone fail to determine a unique place and time, and thus fail to determine
a unique truth-value, this may mean in a particular context what is more adequately expressed
by the words “On 1 March 1900 by the Gregorian Calendar at 12 noon Central European
Time it is raining in Lvov on Castle Hill and its surroundings.” Twardowski practises what
has been called the “decontextualization” of thought.v The problem with decontextualization,
which has been a common move in logical semantics from Bolzano and Frege through
Leśniewski to Tarski, Carnap and beyond, is that it most naturally suggests and goes together
with semantic platonism, the view that the proper or primary bearers of truth are timeless
propositions and that their constituents are themselves timeless abstract ideas.vi The only
person among those cited who disagreed with this view is Leśniewski , and his way of doing
things has been widely sidelined, most influentially by Tarski himself. In any case
3
Leśniewski’s very narrow interest in an artificial logical language meant that he did not
address at length the contextuality problem of Twardowski.vii
In this paper I shall first show how Tarski, for mathematical reasons, deviated in the
direction of platonism in his theory of truth, against the views of his teacher Leśniewski. I
shall suggest that platonism about truth-bearers and their parts makes no essential
contribution to our understanding of the real phenomena of language and truth. I shall then
outline the magnitude of variety of potential truth-bearers needing to be accounted for in a
realistic but non-platonist account of truth. Finally I shall show how in principle such a
plethora of different truth-bearers, all participants in the hurly-burly of the real world, may be
true or false, and yet truth and falsity remain absolute in the spirit of Twardowski.
2 Tarski’s Heresy and Modern Semantic Platonism
Standard formal theories of truth of the sort pioneered by Alfred Tarski are designed for the
languages of deductive sciences. Tarski explicitly rejected the possibility of producing an
adequate and consistent theory of truth for vernacular languages because their semantic
closure means that it is possible to formulate semantic antinomies. More recent theories of
truth have extended Tarski’s methods to larger and more ambitious languages more closely
akin to the vernacular. I suggest that one may and should modify the theory of truth in a
different direction, one which is more closely related to Tarski’s own background and the
logical heritage in which he grew up.
Tarski’s doctoral supervisor was Stanisław Leśniewski. Jan Woleński and I share the
opinion that Leśniewski was one of the finest as well as one of the most remarkable in a
century not short of great logicians, and in our private conversations we habitually refer to
him as ‘Big Stan’. The reason I mention Leśniewski is that his way of conceiving of
languages for everyday as well as for logic was quite different from the way adopted by
Tarski in his 1933 work O pojęcie prawdy w językach nauk dedukcyjnych (hereafter PP),viii
and which came to be the standard for later theories of truth and for model theory. In standard
semantics it is usual to assume that the language is a fixed structure of abstract or eternal
4
entities, in an abundant, countably infinite, supply. There is never any shortage of expressions
in such a language. In such ‘platonic languages’ the linguistic expressions of which they are
composed, whether modelled on natural or on formal languages, are construed as platonic
entities existing independently of space and time and irrespective of whether there are
actually any physical embodiments or tokenings of any particular expression. Anyone who
accepts at face value the statements of such a semantics is a platonist about expressions, if
about nothing else. To utilize Peirce’s terminology, in such a semantics linguistic expressions
are conceived one and all as types, and the question whether this or that expression is as a
matter of fact tokened in the history of the physical universe is an incidental and empirical
one of little or no concern to semantics as such.
Leśniewski was a nominalist, or as near a nominalist as makes no difference,ix and
for this reason he was unable to accept philosophically the way in which Tarski treated
expressions as abstract entities. In his own metalogical work, Leśniewski was scrupulous
about treating linguistic expressions as individual tokens rather than platonic types.x Tarski
was of course acutely aware of Leśniewski’s views. In a footnote of PPxi he excuses himself
for formulating his theory so as to “give the appearance of a widespread error which consists
in identifying expressions of like shape” and using his metalinguistic terms to denote not
individual expression-tokens but whole classes of such tokens, by saying that this “is
convenient” and is “to avoid…the introduction of superfluous complications into the
discussion, which would be connected among other things with the necessity of using the
concept of likeness of shape”.xii In this footnote Tarski appears to endorse Leśniewski’s
view and to apologise for not following it, but for pragmatic reasons only. One gets the
impression that if he had chosen to do so, Tarski could have recast his theory of truth – with
considerable complications no doubt – in a nominalistically acceptable form which would
have placated his teacher, who did not welcome the methodological innovations of the truth
paper.
However Tarski could not in fact have recast his theory in such a way without
significantly weakening its results. Tarski’s metalogic includes the assumption that the set of
consequences of a set of sentences may be denumerably infinite, an assumption used in the
5
truth paperxiii but explicitly stated earlier in the 1930 paper ‘Fundamental Consequences of
the Methodology of the Deductive Sciences’,xiv where Tarski states that he regards the
assumption that there are denumerably infinitely many sentences as “quite sensible…and …
even … useful from a metamathematical standpoint”.xv Tarski’s misgivings about the
mathematically cramping restriction imposed by an assumption that sentences are tokens
emerges in the somewhat tortured discussion between Definitions 17 and 18 of PP.xvi There
he admits that without the assumption that sentences exist such as those needed in the middle
of proofs of often simpler sentences than themselves, using the definition of ‘provable
sentence’ that he gives (Definition 17), statements to the effect that all sentences of a given
kind are provable become impossible to interpret without introducing existential assumptions
which are intuitively weaker than those already eliminated from the axioms. Further, such
concepts as consistency and completeness require similarly strong existential assumptions.
The furthest that Tarski is prepared to go in weakening his existential assumptions is to
consider interpreting his metatheory within the natural numbers, but even here he would need
to rely on a strong existential assumption, namely the Whitehead–Russell axiom of
infinity.xvii
Having somewhat salved his Leśniewskian conscience with this writhing, Tarski
thereupon elects to largely ignore such worries and proceed with his platonistic theory. There
can be no doubt that he was caught on the horns of a dilemma: either to make his metalogic
ontologically unproblematic by assuming only the existence of expression tokens, but then be
unable to derive the mathematical results he wanted, or to embrace a limited platonism for
the sake of the mathematical results and risk Leśniewski’s ire. As it was, the lure of the
mathematics was greater than his desire to placate his teacher, and despite his own no doubt
deeply harboured misgivings about the methodological platonism,xviii he continued in that
vein. In due course the expected rejection came from Leśniewski , but Tarski was by this
time able to withstand the severe disapproval.
The rest is history – the history of modern logical semantics. In the early part of the
twentieth century the mathematical approach to logic was in the ascendant, and there was a
kind of scramble for the major metalogical and metamathematical results. This scramble in
6
metalogic was largely completed, barring details, by Church’s and Turing’s proofs of the
undecidability of first-order predicate logic in the 1930s. Nowadays, mathematical logic is a
branch of mathematics, and its relevance to the traditional philosophical concerns which gave
it birth, except for a few issues such as the semantics of possible worlds or the significance of
relevance logic, has receded. Meanwhile, Leśniewski’s nominalistic alternative and his
honest if inconvenient scruples have been largely sacrificed on the altar of mathematical
advance. No one – myself not least – will deny the significance and beauty of the
metamathematical results achieved within the platonist framework. But – I shall suggest in
the next section – it is largely irrelevant to understanding how language works and how we
have truth in the real world. So the question then is how the shortfall in a real theory of truth
should be remedied.
3 The Irrelevance of Platonism
Even if it were true that there are timeless, spaceless, abstract entities which are the senses of
expressions of actual languages, and of which a subset (the propositions) are the primary
bearers of truth and falsity, of what would this avail us? We should be guaranteed that truth,
falsity, logical validity, consistency, compatibility, and other semantic concepts, defined
platonistically, are well-defined and furthermore defined in the simplest and most language-
invariant possible way. This will guarantee the objectivity of logic, as indeed was the
principal aim of those who invoked such platonic meanings - Bolzano, Frege, Husserl.
Now let us turn our attention groundwards to ourselves and our endeavours. Is the
assurance of a platonic heaven of meanings any help to understanding what we do when we
talk, refer, judge, and reason? I suggest it is an extravagant irrelevance. Every human being
learns to speak in space and time from other human beings in space and time via causal
signals in space and time and with reference to their enveloping spatio-temporal environment.
That the spatio-temporally produced words they utter and hear and understand have to
express things, say things, be right or wrong, needs to be appreciated and understood in terms
of what resources are at their evident disposal. These are all spatio-temporal. Miracles – or
7
surd and incomprehensible grasping or intuiting – aside, we have no way of communing with
the extra-causal realm of meanings. We need a theory of how we can speak, understand,
infer, be right or wrong, which makes use just of those things which a naive observer can
observe. This is bound to be messy, untidy, mathematically inconvenient and unlovely, with
holes and gaps, anything but fit to ground the elegant results of mathematical logic. So what?
Are we interested in mathematical beauty or in explaining the real facts? If someone wishes
to idealize and simplify, or to come along afterwards and give a nice elegant mathematically
invariant theory of what is going on, they are heartily welcome. Let them not imagine they
are explaining. They are merely summarizing that for which another explanation is required.
This is no knock-down argument against platonism. There is none. Like all good
metaphysical theories (and I mean ‘good’ literally, not ironically), platonism is not to be
refuted by gut feelings or gauche incomprehension. The best way to look at it is associated
with the name of Brother William. If we can give a reasonable account of what is going on
when people refer, predicate, tell the truth or not, infer validly or otherwise, which both
covers the plethora of things to be explained and does not drag in any supernatural entities,
then even if platonic entities exist that explanation is to be preferred as the more parsimonious
over the platonic. Platonism may be true. In the absence of a convincing account of the
relationship between us and the platonic objects, we cannot afford to rest content with a
platonistic explanation of logical notions such as truth. There is real work to be done.
4 The Variety of Truth-Bearers
In the history of semantics from Plato to the present, several kinds of entities have been put
forward as candidates for having the properties of being true and being false, and standing in
logical relations such as consequence or incompatibility. Most frequently – in terms both of
the length of time over which such views have been held and the variety of champions – have
been mental occurrences and dispositions such as judgements, assumptions, and beliefs.
From the twentieth century – and here the Lvov–Warsaw School played a leading part in the
shift; one observes it when progressing from Twardowski to Łukasiewicz – it has been
8
linguistic entities such as sentences or statements or assertions which have played the part.
From Bolzano onwards, notably with Frege, Husserl, and Church, the truth-bearers have been
abstract propositions, construed as the meanings of sentences and the contents or objects of
judgements and beliefs. With the disagreements about what should be regarded as truth-
bearers have gone disagreements as to whether truth is absolute or relative, or indeed whether
it can properly be defined for anything so unruly as a vernacular language.
The problem of the plethora of truth-bearers is much worse than is dreamt of in
Plato’s worst nightmares. Practically anything can be a truth-bearer, and very many diverse
kinds of things in fact are.
Mental events may be true or false: occurrent judgements, assumptions, and the
propositional acts which are part of others. One need not believe or judge something to
entertain that thing mentally. Someone who judges a disjunction entertains both disjuncts
even if she judges neither. Even someone who expressly denies or doubts something that is
true entertains the true proposition non-believingly.
Mental states such as beliefs, whether they are dispositions or not, likewise admit of
evaluation as true or false. We are all familiar with people who live all their lives subjectively
convinced of falsehoods, as well as truths. But a belief may also be short-lived. I may judge
on the basis of a reported aeroplane accident that a loved one has met a tragic death, only to
be reassured seconds after acquiring the belief by a telephone call that she was fortuitously
delayed and missed the fatal flight. The belief may last a shorter or a longer time.
The vast majority of our mental convictions, the thousands per minute arising
continuously during our waking life by perception, having to do with the humdrum properties
and configurations of things in our immediate environment, rarely rise to the dignity of being
expressed in words or of forming themselves into salient thoughts. For all that they may be
true or, less frequently, false, and prove their existence on those occasions when need for
witness drags them from previously mute memory.xix
Some of our thoughts and beliefs are unconscious. Eliciting them or finding this out is
clearly a more indirect process than simply asking the subject, but they do exist just the same.
9
Let us move on to linguistic items. Primarily of course there is spoken language,
which consists in events of production of intelligible sounds with a meaning. There are in
general three phases to such events: production, propagation of the sound from the speaker,
and reception by the audience, who hear and in normal circumstances understand what is
said. Then there is writing, in which language is fixed graphically in some way as relatively
stable signs, whether written by hand or printed. Finally there are now manifold other ways
of recording and presenting linguistic items: sound recording, video recording and film, and
other relatively stable media such as computer disks. Also many people now read much of
their language on a computer screen, which is a more evanescent medium than the printed
page but less fleeting than spoken language. In the case of all the media which store spoken
or written language, such as tape, disk, CD and so on, the actual form of the stored
information is not immediately intelligible to people, being in such forms as magnetized
domains or the pits and bumps of a CD surface, and which require equipment and often
software to translate into a form intelligible to human beings. Similar remarks apply to the
modes of propagation of language from place to place, whether via the electric pulses of
landlines, radio waves carrying (by amplitude or frequency modulation) mobile and satellite
telephone conversations, radio and television channels, or the encoded bytes of internet
communication. There is no obvious barrier short of physical impossibility to the ingenuity of
scientists and engineers in discovering new ways to encode and transmit linguistic
information from one place to another. In the past other ingenious propagation systems such
as smoke signals, drums, semaphore, maritime flag signals, heliographs and Aldis lamps
using morse code were used to overcome the problems of sending linguistic messages long
distances.
We are indeed very familiar with the many methods by which linguistic
communication takes place: I am simply recalling their unbridled variety. Without for one
moment denying the ontological primacy of spoken language in all this, and the secondary
but still important position of written language, all the carriers of linguistic messages in
whatever form may in a more or less derivative sense be called true or false whenever what
they carry is interpretable at either end propositionally. A string of bits on a computer disk
10
may be true or false as can a series of frequency modulations carrying part of a mobile
telephone conversation. They may not be immediately intelligible to us, but they carry or
bear a truth-value.
In all this variety a general ontological duality is present, that between persisting
things and events, or continuants and occurrents. This duality entered perceptible language
with writing and pervades all aspects of the communicative process. Writing a sentence in a
letter is an event, but the resulting written sentence is a thing which may persist long after the
event of writing it has ceased. That of course is the main point of writing: to record
permanently or at least for a longer time what otherwise dies away. But the duality is actually
present from the beginning in a less obvious way, since the memory traces of what someone
has said persist within us. People can still remember years afterwards what they heard on
some special occasion such when they proposed marriage, or a dramatic public event such as
President Kennedy’s utterance of “Ich bin ein Berliner”. Less dramatically, we may
remember more everyday spoken sentences for some seconds, minutes or hours after they are
spoken, and may remember their “gist” rather than the actual words for much longer.
This brings me to what may be the most numerous class of truth-bearers directly
concerning human beings, which are events of hearing or reading (and understanding). A
single production event may propagate and cause reception events in many thousands or even
millions of people, as when one person addresses a crowd or makes a broadcast. An author
may write a sentence only once but once it is printed and published, many millions of people
may read and understand it. Each of these events is truth-evaluable. When a TV newsreader
says “Concorde has crashed” and millions of viewers hear it then each of these events of
understanding has the same truth-value as the original statement, and the number of tokens of
this truth increases by millions in a second.
5 Primary Truth-Bearers
In all this multiplicity, it would be surprising if there were not some candidate truth-bearers
which play the role more centrally and more fundamentally than others. Proposition theorists
11
have always assigned this role to propositions: thoughts or utterances are true or false in so
far as the propositions they express or have as their content are true or false. The rationale for
this is twofold. Firstly, propositions are objective and mind-independent and guarantee the
same truth or falsity for all. Secondly, their truth or falsity is not subject to the vicissitudes of
circumstance: since they are eternal objects, truths are true and falsehoods false without
qualification or variation. Truth is absolute. Without propositions, these desirable qualities
have to be underwritten in another way.
There are some general principles which can guide us to a reasoned decision as to
what are the primary truth-bearers. If one item is meant as a record of another item, whether
written or audio or video recording or some other way, then clearly the item which is
recorded has priority over the recording. Thus diaries are posterior to the thoughts the diarist
has which are set down, and recordings of conversations posterior to the conversations
recorded. A second principle is that an item whose truth-value can be uniquely recovered
from the situation and the facts is preferable to an item whose truth-value cannot be so
recovered. A scribbled note on a locked office door saying “Not in this afternoon” does not
allow the casual reader to assess whether it is true or false because the note leaves no trace of
when it was posted. There are several ways to “disambiguate” such a note: one is to replace
the relative time expression by an absolute one: “Not in on the afternoon of 21 July”, or
indeed to leave the relative time but note the date of posting: “21 July – Not in this
afternoon”. But with the message as it stands we would be able to determine its truth-value if
we knew on what day it was posted. So the physical act of posting the written message, rather
than the written words themselves, should be considered the primary truth-bearing item. But
it is not the only prior truth-bearer connected with this notice. Anyone coming by and reading
the message will probably think to themselves something like “The occupant of this office is
not in this afternoon”. They are thereby assuming it was posted on the day they see it, which
may be false: perhaps the note has been there for one or more days. So those reading the
notice on subsequent “wrong” days think a thought not intended. the friend of propositions
would say that the note’s text does not determine a unique unambiguous proposition. Without
propositions, I say: the note itself conveys one thing when posted and that act as well as any
12
act of understanding on the same day share their truth-value, whereas any act on subsequent
days might have a different truth-value. Here the items whose truth and falsity are up for
consideration are acts of “uttering” (in this case, the physical act of posting the notice on the
door), and any number of acts of understanding the notice.
When a speaker addresses a large audience, whether in one place or via the mass
media, and says for example “I shall bring forward legislation in this parliamentary session to
improve our schools”, whereas the friend of propositions would say the speaker and the
audience all understand the same proposition, I say the speaker commits one truth-bearing
act, the act of uttering those words assertively with understanding, while each (linguistically
suitably competent) hearer experiences another truth-bearing act, that of understanding them.
There are as at least as many acts as there are participants, active or passive, in the
communication. Even if the speaker is lying, or if some in the audience do not believe what it
said, these acts have the same truth-value (true or false as the case may be). If the speaker
indeed is lying, he or she is “saying in his or her heart” the opposite, and those who
disbelieve the speaker are thinking to themselves “I don’t believe this”. These are acts whose
truth-value may vary (and in the speaker’s case must) differ from that of the acts of asserting
and understanding.
So it will be seen from these examples that brief, occurrent acts of thinking, uttering
or asserting, and understanding are what I am putting forward as primary truth-bearers. Many
of these are mental acts not marked by immediate outward activity, and in that respect I
return to the views of Twardowski, who held, following Brentano, that it is individual
judgements which are truth-bearers, rather than sentences, as in later Polish philosophy.
However sentences, in the sense of occurrent, spoken sentence-tokens, may be considered to
have their truth-values coevally with the acts of asserting or propounding them: there is little
point in forcing a priority on the act of uttering a token over the token itself, or vice versa,
since the two are so intimately interdependent. These public items (aural linguistic events)
may thus be considered primary truth-bearers. Sentence-token-understandings by contrast are
not public linguistic events but private, mental linguistic events. Other linguistic items may
have truth-values, but they have them derivatively. A written or recorded sentence token, one
13
which perdures as a continuant after the event of producing it has ceased, may by virtue of its
persistence come to be systematically misleading, as in our note on the door example. We
have a choice of ways of regarding its derived truth-value. We could say it inherits it from the
act of production, and so remains true or false despite the fact that due to its linguistic
meaning it becomes misleading as time goes by. Or we could say its meaning forces a
different truth-value on it as circumstances change around it. Both of these ways of assigning
it a truth-value are consistent, and by carefully labelling them differently they are consistent
with one another. Each form of derivation has its drawbacks: the first leaves truth-value
invariant but decouples it from the token’s meaning, while the second keeps the link between
meaning and truth-value at the cost of relativizing truth. But in either case the truth-value is
derivative so the absoluteness and transparency of primary truth is unaffected.
A continuant linguistic token may vary with other factors than time. Imagine a
travelling circus which takes around with it portable notices saying “Our circus here tonight”.
At each town they visit they put up the notices, but the place referred to on each occasion is
different, as well as the time. One could imagine the circus going bankrupt and the signs
being sold to a different circus, so the signs refer to a different circus from then on.
Obviously when we move from sentence-tokens to sentence-types the scope for
variation is much greater and more apparent. One can well understand the attractiveness of
“eternal” or “standing” sentences, whose truth-value, given their linguistic meaning together
with the way the world is, is invariant across the type. Such sentences share with propositions
their absoluteness without suffering the same ontological obscurity. Even if one admits
abstract types in addition to concrete tokens, their relationship to the tokens is more
transparent than that of propositions to – what? Sentences? Acts of uttering and
understanding? Sadly though, eternal sentences are a tiny minority of those actually
produced. Nearly all sentences have a meaning which makes their truth-value dependent on
the circumstances of their utterance. Far from being central, they are a sideline, a welcome
exception to the hard work which needs to be put in to guarantee the absoluteness of truth.
Beliefs, considered as standing or continuant dispositions to assent, share similar
vagaries of truth-value variation to sentence-tokens, though being private states they are not
14
obvious to others for inspection except as the person whose beliefs they are manifests them
through utterance or other linguistic act.
These central examples may stand as representatives for the whole variety of truth-
bearers. Primacy is accorded to occurrent events of production or understanding or concrete
linguistic tokens, and other things that are ascribed a truth-value derive theirs from these
primary truth-bearers. The question whether truth is absolute or not should therefore
concentrate on the primary truth-bearers.
6 Terminological Repossession
It is convenient to have a word for primary truth-bearers. I shall for the remainder of this
essay re-appropriate the word ‘proposition’ for this purpose. The reasons are threefold.
Firstly, it is short and suggestive. Secondly, the sense of a proposition as an abstract Satz an
sich or Gedanke — a use going back no further than to Moore and Russell in the early 20th
century — is not being employed constructively for the duration of this essay. In future I
shall use the expression ‘abstract proposition’ for such things. Thirdly, and most importantly,
this is by far the oldest of uses of the term ‘proposition’ in English, dating from the time (and
pen) of Wyclif in the 14th century and used in the first logic book in English, Thomas
Wilson’s Rule of Reason of 1551: “A Proposition is, a perfecte sentence spoken by the
Indicatiue mode, signifying either a true thyng, or a false.”xx Sense 4. a. (a), dating originally
from around 1340, of the noun ‘proposition’ in the Oxford English Dictionary is given as
“The making of a statement about something; a sentence or form of words in which this is
done; a statement, an assertion”. This definition corresponds closely with our analysis to the
effect that primary truth-bearers include both acts of proponing (I shall use this word, dating
back to c. 1375, for the act) and the (token) utterance produced. The only slight discordance
with our findings of the previous section is that acts of understanding would not typically
have been called ‘propositions’.
15
7 Context-Dependency of Truth-Value, What
The obvious threat to the absoluteness of truth comes from the apparent context-dependency
of truth-value for propositions. But we need to be clear what this context-dependency
amounts to. Propositions, as we are now using the term, are concretely situated events. They
are indeed typically associated with forms of words or sentence-types (which means they are
associated with particular languages), and these types tell us which linguistic competence a
speaker or hearer has to bring to the understanding of a particular utterance. However the
assessment of the truth-value of what someone says depends in most cases on more than the
meaning associated with the expression token uttered. It is the dependence of truth-value of
linguistic expressions of this type on their context which is what context-dependency is, not
dependence of the particular proposition on its context. The context is in general richer in its
descriptive content than is exhausted by the relevant linguistic type. When my friend John
says at five on a Friday afternoon “I need a drink” then my knowledge of English does not
suffice for me to know who needs a drink when, but knowing who says it when is sufficient,
and my knowledge of the meaning of the English first-person singular nominative pronoun
and the present tense do enable me, in conjunction with the knowledge of who speaks when,
to compute the likely proposition (and its truth-value). Were my friend Anne to say the same
thing the next day at lunchtime then similar linguistic competence would come into play in
conjunction with my particular knowledge of the situation of her utterance to enable me to
pull off a similar feat. But the propositions themselves, their proponings and my
understandings, are not context-dependent, since they are already embedded as the acts they
are in their contexts. Propositions as such are not context-dependent: what is context-
dependent are the relevant sentence or expression types. Context-dependence affects
expression-types in respect to reference, truth-value, and other dimensions of disambiguation
such as the use of multivocal words or proper nouns naming more than one object –
Manchester (England) vs. Manchester (New Hampshire), or my friend Ewa Kowalska vs.
Maria’s cousin Ewa Kowalska. When a proposition (of whatever ontological type, whether
mental or linguistic) suitably matches the type of its associated sentence-token (produced or
16
understood), i.e. is in context about what the sentence would be properly taken to be about,
and means what the sentence would properly be taken to mean, then I shall call the
proposition appropriate to the sentence.
8 How To be Absolutely True — In Context
Let us consider how a proposition (in our sense) can be absolutely true, and yet how the
meaning of any associated type-expression may contribute towards determining the truth-
value. What we are looking for is a generalisation of Tarski’s T-schema which takes account
of context. Here is a typical case. I say to Jan ‘Maria’s cousin Ewa is getting married in St.
Mark’s tomorrow’. Many tokens of such a sentence type could and some most probably have
been uttered in proponing different propositions. Consider then such a proponing act. Assume
for a start that it is a genuine assertion. Then provided the speaker and the addressee both
know who is being spoken about and which church is in question, they understand one
another and their proponings match in truth-value. The link to the linguistic meaning is
regular and can be captured in a quantified conditioned biconditional as follows:
(1) For all P, D, E and C:
if P is a proposition appropriate to the utterance on day D of a token of the English
sentence ‘Maria’s cousin Ewa is getting married in St.Mark’s tomorrow’ such that
‘Maria’s cousin Ewa’ refers to the person E and ‘St.Mark’s’ refers to the location C
then:
P is true if and only if E gets married in C on the day after D.
There are one or two perhaps surprising aspects of this analytical suggestion but let us first
consider a few more examples.
17
(2) For all P, S, H and T:
if P is a proposition appropriate to the utterance by S to H at time T of a token of the
Polish sentence ‘Kocham ciebie’ then:
P is true if and only if S loves H at T.
(3) For all P, L and T:
if P is a proposition appropriate to the utterance at T by a speaker indicating location
L of a token of the German sentence ‘Ist dieser Platz frei?’ then:
P is true if and only if the location L is unoccupied (free) at T
(4) For all P, H and T:
if P is a proposition appropriate to the utterance to H at time T of a token of the
French sentence ‘Asseyez-vous!’ then:
P is true if and only if the H sits down immediately after T.
Here is a case of a standing sentence:
(5) For all P:
if P is a proposition appropriately connected to a token of the English sentence type
‘17 is a prime number’ then:
P is true if and only if 17 is a prime number.
Finally, here is a surprising example:
(6) For all P, L and T:
if P is a proposition appropriate to the utterance at time T in external surroundings L
of the Polish sentence ‘Śnieg pada’ then:
P is true if and only if it is snowing at T in L.
18
These and similar analyses are meant to stand in for an open array of examples, simpler and
more complex. Let me now draw attention to the most salient features of the examples.
(A) Within the context of the universally quantified implicational sentence, the embedded
biconditional is a T-sentence schema.
(B) The predicate ‘is true’ in this schema does not contain any bound variable within it and
is accordingly absolute, and in particular untensed.
(C) The dependence of truth-conditions for the proposition on factors of context is carried
by the universally quantified variables affecting the T-schema and by the conditions in
the antecedent.
(D) These variable factors and conditions affect different aspects of context of utterance
and the form of dependence, common but by no means invariable factors including
speaker, addressee, time and place of utterance.
(E) There is no uniform recipe for all cases: how the dependence works depends on
features specific to the language and expression type in question.
(F) The common denominators to all cases are the biconditional form, the variable for the
proposition P, the absoluteness of the truth predicate, and therefore of the right-hand
side of the biconditional once the variables are replaced in any actual case by
constants, and the need to specify the language to which the sentence token belongs.
(G) It is not always necessary to know who the speaker is or the time of utterance. In some
cases, as in (5), knowledge of the utterance act itself is unnecessary beyond minimally
knowing to what language the token is intended to belong.
(H) The proposition need not be directly connected to an assertion. It may be involved in
uttering or understanding a question (Example (3)) or a command (Example (4)). A
competent German speaker will propone to herself a proposition to the effect that the
indicated place is free when hearing the question – whether she accepts that the place
is free or not. If she knows it is not, she will almost certainly immediately afterwards
propone to herself a proposition negating this – ‘(No,) This place isn’t free’. A
competent French speaker will propone to herself in case (4) a proposition to the effect
19
that the addressee will sit down – whether or not she expects it. Such a proponing is
not necessarily assertive, it comes with simply understanding what is said and so
makes clear the general dependence of propositions like those considered on specific
linguistic competences (which may involve much less than a good grasp of the
language in question).
(I) It is not assumed that the speaker must be speaking in good faith or in full knowledge
or that speaker and addressee or speaker and hearer (if hearer is a third party) share all
the relevant knowledge. For instance in (2), the speaker may be lying, or self-deluded,
or making a linguistic mistake (e.g. thinking that ‘kochać’ means ‘to like’), or have
mistaken the addressee in the indifferent lighting of a discotheque. These are all forms
of what Austin would call “infelicity” which do not affect the embedded truth-
conditions. In some cases it will not be the speaker who is proponing P. If I, under the
mistaken impression that ‘kochać’ means ‘like’, say to Ewa ‘Kocham ciebie’ then my
proponing is to the effect that I like her whereas hers and that of competent Polish
speakers who might chance to overhear is to the effect that I love her. We make
symmetrical mistakes: due to my linguistic incompetence they do not understand my
proposition and I invoke inappropriate propositions in them (especially in Ewa!) But
their propositions are appropriately connected to my misleading utterance, mine is not.
They misunderstand me because they correctly understand my utterance.
(J) It is in general possible for certain kinds of mistake to be unimportant for the truth-
conditions of a proposition: for example both speaker and hearers in Example (1) may
wrongly think that the church C is called ‘St. Mark’s’ when it is in fact called ‘St.
Matthew’s’. But in the context, our parallel mistakes do not matter: ‘St. Mark’s’ does
refer in such a case (inappropriately) to St. Matthew’s. Someone who knew that C is
called ‘St. Matthew’s’ would therefore misunderstand the speaker’s utterance or could
be even more knowledgeable and know the speaker was wrongly referring to St.
Matthew’s church as ‘St. Mark’s’. There is no obvious upper bound to the kinds of
infelicity that may occur. For example perhaps speaker and listeners know that St.
Matthew’s is meant but have made a prior arrangement to use the wrong name because
20
they fear the bride’s revengeful ex-boyfriend may turn up and cause trouble, and
suspect he or a friend of his may be lurking around to find out when and where the
wedding is to take place. All it takes is a lively imagination or a surfeit of reading spy
novels to come up with still more devious cases.
(K) Standing sentences (Example (5)) are particularly straightforward and yield – modulo
the fact that ‘P’ varies over tokens rather than types – Tarski’s T-schema.
(L) Speaking of Tarski, Example (6) is his and is meant by him to illustrate the basic
(absolute and context-free) T-schema. It cannot be both absolute and context-free: I
say to preserve absoluteness of truth it is doubly context-dependent: on time and
location.
(M) Because the analysis always brings in tokens of sentences of a particular language, it
will not work for cases of mental propositions not connected to utterances or
occurrences of sentence-tokens. Nor should there be any expectation that it should.
How such mental propositions work and get their truth-conditions and therefore get to
be true or false is a difficult and delicate matter and I shall simply duck the issue here
for reasons of lack of space, but it will clearly entail delving into the mental
equivalents of reference, predication and other overtly linguistic manifestations of
thought.
(N) The analysis provides only partial relief from Liar-type paradoxes. Suppose we lay
down the obvious valuation principle that a sentence-token be deemed true iff all
propositions appropriate to the token are true and false if all propositions appropriate
to the token are false. Then if T is the token sentence ‘This sentence is not true’ and P
is any proposition appropriate to it, then P is true iff T is not true, and the assumption
that T is true leads to its opposite and vice versa. However we could simply deny that
just because appropriate propositions are true, the sentence T is true. That would allow
us to consistently allow that T is not true and so any appropriate proposition P is true,
without inferring on the rebound that therefore T is true. In this case, as in the
analogous truth-teller case of a sentence-token saying of itself that it is true, there is no
way out of the circle of assumptions if we accept the obvious valuation principle, but
21
we do not need to accept it. More tricky would be a sentence-token U of ‘no
proposition appropriate to this sentence-token is true’. If there were such an
appropriate proposition then if it were true it would not be true and since this would
apply to all such propositions by the obvious valuation principle U would be not true.
Here though the buck stops as something’s not being true does not entail its being
false: cats, cups and computers are neither true nor false. Hence there can be no
proposition appropriate for U. However Liar-type self-reference can be more devious:
there appears to be nothing to stop a proposition being about itself, expressible in
words tantamount to ‘This very proposition is not true’, so by the expected T-sentence
it would be true if and only if it were not true. Hence although certain kinds of
linguistically-mediated paradoxes may be disarmed, at rather little cost to our
intuitions, the possibility of paradoxical propositions cannot be ruled out, provided
only propositions can refer to or quantify over propositions. Thus the switch from
abstract to concrete propositions as primary truth-bearers only avoids some paradoxes
and not others. The switch to concrete propositions may safeguard the absoluteness of
truth where truth is well-defined, but it is not a panacea against paradox.
Woleński, Jan and Simons, Peter: 1989, De Veritate: Austro–Polish Contributions to the
Theory of Truth from Brentano to Tarski. In: K. Szaniawski, ed., The Vienna Circle
and the Lvov–Warsaw School. Dordrecht: Kluwer, 391–442.
Yourgrau, Palle: 1991, The Disappearance of Time. Cambridge. Cambridge University Press.
23
i Woleński and Simons 1989. ii I stress, for philosophy. History of philosophy has to be brought to each new generation anew. iii The fact that the historical passages refer mainly to figures now wholly forgotten serves only to highlight the magnitude of Bolzano’s leap forward. iv Twardowski 1900. v Cf. Yourgrau 1991, pp. 104 ff. vi The first and still perhaps the greatest proponent of semantic platonism is Bolzano, ably seconded by Frege. vii An exception is tense, which Leśniewski treats – like Bolzano and Carnap – by invoking time-slices of objects. See Leśniewski 1992, 379–82. viii Cited according to the English translation of 1956 and the Polish reprint of 1995. ix Leśniewski did not describe himself as a nominalist because he believed in phenomena such as after-images, but after-images and any other mental qualia are not necessarily universal entities though they might not be physical. There is no reason why a physical/phenomenal dualist may not believe that all entities are particular, and hence be a nominalist. x Cf. Leśniewski 1992, 471: “Two expressions equiform to each other written in two different places are never the same expression.” xi Tarski 1995, 19n.5; 1956, 156n. xii Ibid. xiii Cf. Definition 18, Tarski 1995, 57;1956, 185. xiv Cf. Tarski 1956, 63. xv Tarski 1956, 64. xvi Tarski 1995, 55–7; 1956, 183–5. xvii Tarski 1995, 57; 1956, 185. xviii Tarski’s anti-platonism even to the end of his days is tellingly described in Suppes 1988. xix See Simons 1997. xx Wilson 1551, cited after the Oxford English Dictionary from the edition of 1580, p. 18.