Top Banner
8/3/2019 Todorov on Linguistic Symbolism 1/25 On Linguistic Symbolism Author(s): Tzvetan Todorov and Richard Klein Source: New Literary History, Vol. 6, No. 1, On Metaphor (Autumn, 1974), pp. 111-134 Published by: The Johns Hopkins University Press Stable URL: Accessed: 04/08/2009 11:33 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected]. The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to  New Literary History.

Todorov on Linguistic Symbolism

Apr 06, 2018



Welcome message from author
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.
Page 1: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 1/25

On Linguistic SymbolismAuthor(s): Tzvetan Todorov and Richard KleinSource: New Literary History, Vol. 6, No. 1, On Metaphor (Autumn, 1974), pp. 111-134Published by: The Johns Hopkins University PressStable URL:

Accessed: 04/08/2009 11:33

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to

 New Literary History.

Page 2: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 2/25

On Linguistic Symbolism

Tzvetan Todorov

I. Lexical Symbolismand PropositionalSymbolism

HE FOLLOWING EXAMPLE is found in the rabbinical commentaries

on the Pentateuch: it is said in the Bible that God will reward

even animals; and the commentary adds: "Can it not bereasoned a fortiori if it is thus for a beast, for men it is all the more

certain that God will not withhold his reward?"1 A single propositionis present in the text to be interpreted: "Animals will be rewarded";but it allows us (a) to understand its meaning, which is the proposi-tion "all animals will be rewarded" and (b) to give it a secondary, in-

direct interpretation which is that men will be rewarded. Leavingaside for the moment the question of the a fortiori, or qal wahomer,an essential feature of the rabbinical gloss, we may arrive at the general

conclusion: the signifier of a single proposition allows us to under-stand two signified meanings, one direct, the other indirect.

Let us now imagine that in the proposition "all animals will be

rewarded," the term "animals" is being used metaphorically to desig-nate, for example, the humble of spirit. The word animal on the

one hand will evoke, directly, the meaning animal; indirectly, on

the other hand, the meaning humble of spirit. Once again, a single

signifier leads us to understand two signified meanings.

Linguistic symbolism is defined by that excess of meaning with

which the signified overflows the signifier;2 thus we have before ustwo examples of the symbolic function of language. Their kinship is

evident, but what is their difference? It is in the nature of the segmentthat gears the symbolic process: a proposition in the first case, a word

in the second. More exactly, a language segment like "animals" has

a meaning but involves no predication; while the segment "animals

will be rewarded" has a meaning (can refer to a certain situation)but also involves predication: it is being said of "animals" that theywill be "rewarded." This initial difference entails others: in the first

case, the initial proposition animals will be rewarded can be submitted

i J. Bonsirven, Exegese rabbinique et exegese paulinienne (Paris: Beauchesne,1939), p. 87.2 Cf. "Recherches sur le symbolisme linguistique, I," Poetique, i8 (I974), 220.

Page 3: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 3/25


to the test of truth; in the second case, on the contrary, there is no pointin asking whether animals in a literal sense will actually be rewarded;that is not the question. Only the second proposition concerning men

can be said to be true or false. Again, if one wants to explicitate every-

thing those two segments tell us, one would get, in the first case:

(I) animals will be rewarded; (2) men are like animals (only more-

so); (3) men also will be rewarded-three propositions. And, in the

second case: (I) certain men are like animals; (2) those men will be

rewarded-two propositions.I will call the first kind of linguistic symbolism, in which one proceeds

from aproposition

with itsmeaning intact, propositional symbolism;and the second kind, in which one proceeds from a word, having

cancelled the meaning of the initial proposition, lexical symbolism. The

opposition between lexical (or predicational) and propositional sym-bolism is continuous with the one that allows logic to be divided into

two great divisions, the calculus of predicates (or functions) and the

calculus of propositions.To my knowledge, no one has sought to compare and distinguish

these two language phenomena in precisely this way-which does not

at all mean that the distinction itself has gone unnoticed; but it hasreceived other descriptions and, I will now try to show, less satisfyingones.

The best-known description originates in the writings of St. Augustine.We know that the originality of St. Augustine lies above all in the

synthesis he forged of divergent traditions. In On Christian Doctrine,he integrates the rhetorical heritage, among other things, into a gen-eral semiotic theory: tropes acquire the status of "transposed signs"(signa translata). But the definition St. Augustine gives of a trope

is no longer the same as that found in rhetorical manuals, that is,a word used in a sense which it does not usually have. He writes:

"Signs are transposed when that thing which we designate by a literal

sign is used to signify something else; thus we say 'ox' and by that

syllable understand the animal which is ordinarily designated by that

word, but again by that animal we understand an evangelist, as is

signified in the Scripture, according to the interpretation of the Apostle,when it says: 'Thou shalt not muzzle the ox that treadeth out the

corn.' "3 The trope is defined here as a symbolism of the things trans-

mitted by language. The sentence, attributed in Deuteronomy toGod, speaking of the ox, is interpreted by St. Paul in the First Letterto the Corinthians as referring to those who announce the Gospel. But

3 On Christian Doctrine, tr. D. W. Robertson, Jr. (Indianapolis, 1958), II. x. 15(translation modified).

I I2

Page 4: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 4/25


the words themselves do not change their meaning; it is the object ox

which, in a second instance, evokes theEvangelist.A page later, however, St. Augustine cites another example of an

indirect sign. This is the way he glosses the sentence of the prophetIsaiah (Is. 58:7) : "Despise not the family of thy seed." . . . 'Familyof the seed' may be taken in a transposed sense so that it is under-

stood to mean 'Christians' born spiritually from the seed of the word

which produced us" (On Christian Doctrine, II. xii. 17). Here there

is no symbolism of things; the words are to be understood in another

sense, as in the case of rhetorical tropes.

These two contrasting examples ought not to be taken as evidenceof confusion on St. Augustine's part but of his desire to enlarge the

category of the "transposed"or "figurative." In discussing the problemsraised by transposed signs that are unknown, he recognizes two typesof solution (and thus, implicitly, two types of transposed signs): those

which depend on an understanding of language and those which

depend on our understanding of things (II. xvi. 23). The former

type of solution designates transposed meaning inherent in the word;the latter, the thing the word designates: linguistic knowledge and

encyclopedic knowledge are here distinctly separated (II. xvi. 23-24ff.).The opposition is formulated even more clearly in On the Trinity, in

which St. Augustine glosses St. Paul's allegorical interpretation of the

two wives and the two sons of Abraham as figures of the celestial and

the earthly Jerusalem (Gal. 4:22): "Nevertheless, when the Apostle

speaks of allegory, he is speaking not of words but of a thing in fact:

in the passage where he shows that the two sons of Abraham, the son

of the servant and the son of the free woman (those are not words but

facts) signify the two Testaments."4 That formulation is at the originof one of the most important distinctions in Christian hermeneutics,the one between allegoria in factis and allegoria in verbis.5 "Allegory"here designates the whole of symbolism; "factual" (or "real") alle-

gories as distinct from "verbal" ones comprise only one of its species.It can be seen from the examples above that we are dealing with

facts identical in respect to the previously evoked instances of lexical

and propositional symbolisms. We could have said that "animals"

understood in the sense of humble men was a change in the sense of the

4 De la Trinite, Oeuvres de st. Augustin, XVI (Paris: Desclee de Brouwer,1955), XV. ix. 15.

5 Concerning this opposition, cf. J. Chydenius, The Theory of Medieval Symbolism(Helsinki, 1960) ; H. de Lubac, Exegese medievale (Paris: Aubier, I959), II,

493-98; IV, 131-37; and the unpub. study of Armand Strubel, "Allegoria in factiset allegoria in verbis."

I I3

Page 5: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 5/25


words; and that in the other case, conversely, the very thing evoked

(thereward of

animals)allowed us to derive a new

meaning (con-cerning the reward of men). Which of the two descriptions is to be

preferred?The shortcoming of the opposition between verbal and factual alle-

gory (or symbolism) consists not merely in its substantialism and in its

failure to reveal the mechanism itself that produces two distinct facts of

language. It also errs in another direction: tropes (i.e., verbal allegory)are all just as "factual" as factual allegories themselves. If I use the

word ox to designate metaphorically an obtuse man (which is not

what St. Paul suggests), I must indeed refer to the animal itself in orderto determine some comparison with a certain kind of man. In that

respect the case is in no way different from the one in which the words

designate an ox but the ox in turn evokes the Evangelist. The opposi-tion here between words and things is a somewhat awkward way of

indicating that the meaning of the initial statement is maintained in

one case and cancelled in the other. In "verbal allegory" the statement

concerning the animal disappears; in "factual allegory" the statement

remains. This fact itself results from the linguistic difference between

the segments that serve as the starting point of the interpretation:either a word or a proposition.

The example of the two wives of Abraham is a case of the same

sort. If by "wife" were understood weakness, for example, that would

be a case not of forgetting the thing but of cancelling the first statement:

nothing would have been said about the (literal) wives of Abraham.

St. Paul interprets the sentence in another way: Abraham indeed has

two wives (the sense of the initial statement is kept) but they announce

the two Jerusalems. Here, as before, one passes through the world of

"things"; the only variable is the size of the segment one reinterprets:either a word or a proposition.

The same can be said for another, slightly different formulation of

the same opposition such as one finds in St. Thomas Aquinas. There

the opposition is more accentuated, for while St. Augustine allows for

every kind of symbolism in the Bible, St. Thomas leaves lexical sym-bolism to poets and claims as the prerogative of divine expressiononly (one of the forms of) propositional symbolism. On the basis ofthe same opposition factis/verbis, he insists on the fact that one

of the interpretations is successive, the other simultaneous.

These various readings do not set up ambiguity or any other kind ofiiiixture of meanings, because, as we have explained, they are many,not because one term may signify many things, but because the thingssignified by the terms can themselves be the signs of other things ....


Page 6: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 6/25


The parabolicalsense [hissynonymfor verbal allegory,like that which men

also employ] is contained in the literal sense, for words can signify some-

thing properlyand somethingfigurativelyat the same time; in the last casethe literal sense is not the figure of speech itself, but the object it figures.When Scripture speaks of the arm of God, the literal sense is not thathe has a physical limb, but that he has what it signifies, namely the

powerof doing and making.6

Our concern is not with the distribution proper/meaning-trans-

posed meaning/literal meaning-spiritual meaning which differs in

St. Augustine and St. Thomas. The fact remains that in the case of

"factual allegory," according to St. Thomas, one must first interpretwords, then the things they designate; while in the case of "verbal

allegory" (or parable) the two meanings are given simultaneously.But here again only one of the descriptions is exact. To return to our

initial example, it is true, one must first understand the sentence

"animals will be rewarded" in order then to conclude that men will

be too. But the same is true for the other case: one first understands

the meaning animal, and only then the meaning humble of spirit;it is through the initial meaning animal that we arrive at the second

meaning men. And despite what St. Thomas says, hearing aboutsomeone's arm, we think first about an arm, and then only secondarily,

having decided that the first meaning is inadmissable, do we go from

arm to active power. At the same time, one sees what St. Thomas has

in mind: in one case one understands the first proposition, and then

a second one is added to it; in the other case, one envisages a first

interpretation, and then another one is substituted for it. But this

difference can be derived from the inequality of the interpretable

segments; indeed, in the case of tropes, meaning is also added, but it is

the meaning of a word, not of a proposition. The process is the samein the two cases; if the result is different, it is because the process appliesto different entities, words and propositions.

It has been necessary to insist on the reinterpretation of the opposi-tion factual/verbal in propositional/lexical symbolism, because even

modern historians of exegesis (Chydenius, de Lubac, or Pepin, for

example) continue to use the old terms. Some other attempts to

formulate the same dichotomy may be quickly considered.

Perhaps one should rather speak of the absence of any such attemptsif one thinks of classical or modern rhetoric. The fact is all the more

striking insofar as the opposition word/proposition is present in both

6 Summa Theologiae, tr. Thomas Gilby (London and New York, 1964), I, 38-39:Question I, Article io, Solutions I, 3 (my italics).


Page 7: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 7/25


and even at times plays an important role. I have already shown that



in Fontanier for

example,conceals the



absence of a figure indicating the trope; but the trope itself does not

change.7 The same is the case for the opposition metasememic/meta-

logical, in the Rhetorique generale of J. Dubois et al.8 From the

beginning (pp. 32-33) one is told that the two are opposed to one

another as words are to sentences; however, later in the book that

criterion is given up ("the number of units of meaning has no

discriminative value here," p. 132) in order to be replaced by the

presence or absence of an "egocentric circumstantial." Thus the opposi-tion

word/proposition disappearswitiout

having played anyrole

what-soever; one would further like to know the meaning of the sentence

which introduces the previously quoted remark: "In this respect the

symmetry of the general framework [precisely the one which opposesmetasememes and metalogisms] has only a didactic value" (p. 132):

by what sort of metonomy has the term "didactic" become synonymouswith "false"?Recent rhetoric i; thus unaware of propositionalsymbolism

and retains only the symbolism of tropes.The problem is touched upon in the rhetoric of Quintilian, without

being explicitly posed. Quintilian opposes tropes and figures, withoutdetermining the categories underlying the opposition: words (tropes) /

propositions (figures) or form (figures)/meaning (tropes); which re-

sults in the uneasy tripartite division of tropes, figures of words, and

figures of thought. Thus irony appears both as a subdivision of allegory

(therefore as a trope) and as a figure of thought-affirmations that

Quintilian tries to reconcile in the following manner: "Thus, as a con-

tinued metaphor constitutes an allegory, so a continuation of ironical

tropes forms the figure irony."9 But if allegory is opposed to metaphor

as the figure is to the trope, is it any longer a trope? Another signpoints in the same direction: the example is presented as a subdivision

of allegory; whereas, in the light of already-cited illustrations, examplebelongs to propositional symbolism, not lexical. Instances of allegory,

however, point in an opposite direction: allegory is only the accumu-

lation of several metaphors drawn from the same domain: " 'O ship,shall new waves bear thee back into the sea? O what are thou doing?Make resolutely for the harbor,' and all that ode in Horace, in which

he puts the ship for the commonwealth, the tempests of the wazes for

civil wars, and the harbor is peace and harmony" (Institutes, VIII. vi.44). As the commentary of Quintilian reveals, we are dealing here with

7 "Recherches," 240-41.

8 Rhetorique generale (Paris: Larousse, 1970).9 Institutes of Oratory, tr. John Watson (London, 1876), IX. ii. 46.

Page 8: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 8/25


several words (metaphors) but not a single proposition: there is no

statement about real ships which would allow us, secondarily, to com-

pare it to the State. The extended metaphor is still a metaphor that has

nothing to do with propositional symbolism, and only such are explicitlyconsidered in The Institutes of Oratory.

There is yet another attempt to circumscribe the same categories,but I can only touch on it briefly. It is the distinction made bySanscrit theorists between two language functions, figuration (laksand)and resonance (dhvani) or manifestation (vyanjaii). "One dis-

tinguishes between direct denotation (abhidhd) and indirect or in-

ferred denotation (laksand), which is the basis of metaphor; somepostulate tatparya or 'intention,' that is the value that words derive

from being connected to one another in a sentence, independently of

their inherent value. Beyond lies 'suggested meaning,' vyangyartha

('the meaning to be manifested') or dhvani (strictly speaking,'resonance')."10 The opposition between the two direct meanings is

constructed on the word/proposition axis; it appears that the same can

be said for the two kinds of indirect meaning, although dhvani

occasionally has a more general sense.

In a preceding section of this investigation, I noted that, since thelinguistic sign includes three elements-signifier, signified, and refer-

ent-one could conceive three kinds of symbolism, each one of the

elements capable of evoking another of the same sort.11 One could

thus distinguish a symbolism of the signifier (an infralinguistic one),of the signified (a metalinguistic, or, simply, a linguistic one) and the

symbolism of the referent (that is, extralinguistic). For this first cate-

gorization, which is coherent but much too substantialized, I would

like to substitute another, once again founded on language, or more

exactly, on the nature of the segment which becomes the starting pointof the symbolic process. It will include propositional symbolism, based

on the proposition; lexical symbolism based on a purely referential

segment (in practice, a word); and finally, a symbolism of the ultimate

elements which result from the segmentation of discourse, sounds or

letters: that is, phonetic or graphic symbolism.

For the last time, let us quickly illustrate these three kinds of

symbolism. If one considers the story of Adam as describing a real

event but at the same time as announcing the coming of Christ, weare dealing in propositional symbolism. If we interpret Adam, by

io L. Renou et al., L'Inde classique, II (Paris and Hanoi, 1953), ? 1576, p. 117.I am relying here on the unpub. work of Mme. M-CI. Porcher.

II "Introduction a la symbolique," Poetique, I I (1972), 285.

I I7

Page 9: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 9/25


antonomasis, as signifying man (in general), as in the sentence "This

Adam has beentempted

by a new Eve," it is a case of lexicalsymbolism.Finally, we would have graphic symbolism if we proceed like St.

Augustine who writes:

Who does not know that from him all nations are sprung; and that in

the four lettersof his name the four quartersof the globe, by their Greek

appellations, are indicated? For if the east, xwest,north, and south are

expressed in Greek even as Holy Scripture mentions them in various

places, the initial letters of the words, thou wilt find, make the wordAdam: for in Greek the four quartersof the world are called Anatole,

Dysis, Arktos, Mesembria. If thou write these four words, one underthe other, like four verses, the capital letters form the word Adam.12

Comparing this distribution with the preceding one, the field appearsto have been narrowed to the firsttwo types, leaving aside extralinguisticsymbolism. In fact, there is no such restriction, since any nonlinguisticsymbolism can be translated into terms of language and analyzed in

that form, as St. Augustine himself has pointed out: "And I could

express the meaning of all signs of the type here touched upon in words,

but I would not be able at all to make the meanings of words clear bythese signs" (On Christian Doctrine, II. iii. 4). It is no accident that

what Peirce cites as an example of nonlinguistic signs-smoke for fire-

served, in the distant time of the Stoics, to illustrate propositionalsymbolism: if there is smoke, there is fire. Extralinguistic symbolism,

according to whether it arises from "substance" or from an "event,"is distributed between lexical symbolism and propositional symbolism.It is not at all a question of seeking a linguistic basis for the notions of

"substance" or "event," nor of affirming that the structure of language

is to be likened to the structure of the world; but simply of demon-strating that such facts can be analyzed within the framework of the

study of linguistic symbolism.

II. The Symbolic and the Discursive

It may be recalled that Aristotle, in the Poetics, identifies four

varieties of metaphor (the name with which he designates tropes):

"either transferred from the genus and applied to the species or fromthe species and applied to the genus, or from one species to another

2 Lectures or Tractates on the Gospel According to St. John, tr. John Gibb and

James Innes, A Select Library of the Nicene and Post-Nicene Fathers of the

Christian Church, ed. Philip Schaff (New York, x888), VII, 67.

I 8

Page 10: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 10/25


or else by analogy."13

Leaving analogy aside for the moment, it can

be seen that we are dealing here with a two-dimensional combinatory

system, species/genus and starting point/end point (or exposed mean-

ing and imposed meaning), three of whose results are enumerated while

the fourth is missing: from genus to genus. We might designate them

with the terms in current use: from genus to species equals what is

called a particularizing synecdoche; from species to genus equals gen-

eralizing synecdoche; and from species to species equals metaphor.The missing variation, from genus to genus, corresponds to metonymy:for while metaphor implies two terms (species) having a property



example"love" and


"burning,"metonymy requires that one term (species) be liable to be qualified

by two independent properties, for example (to return to the metonymyevoked in the preceding section) "didactic" and "false," both con-

sidered as properties, we might say, of "textbooks."14

An extremely similar enumeration is found in certain other writingsof Aristotle. Here is an extract from the first book of the Rhetoric (a

paraphrase of it appears in the Prior Analytics, 6ga): "We have said

that example is a kind of induction and with what kind of material it

deals by way of induction. It is neither the relation of part to whole,nor of whole to part, nor of one whole to another whole, but of part to

part, of like to like, when both come under the same genus, but one of

them is better known than the other." 15 The resemblance between the

two sentences is not merely structural: one of the dimensions of the

new combination is frankly identical to that of the preceding system

(starting point / end point, from . . . to . . .); the other identity is

barely concealed behind a transparent synonymy (genus equals the

whole, species equals the part). What then is example which soclosely resembles a trope?

In his statements of principle, Aristotle defines example and enthy-

meme as the imperfect, nonrigorous, rhetorical version of two logical

operations, induction and syllogism. But in the more descriptive pas-

sages, like the one just quoted, example is described as a third varietyof reasoning, alongside induction and syllogism, and not as the degradedform of one of them-a fact which has led one translator of the Prior

Analytics, J. Tricot, to specify in a note: "An example is not an induc-

13 Aristotle, Poetics, tr. W. Hamilton Fyfe (1925; rpt. London and Cambridge,Mass., 1968), I457b.

14 The analysis of tropes in terms of the logic of classes was performed in

Rhetorique generale, pp. 9off. I have modified its presentation.

I5 Aristotle, Rhetoric, tr. John H. Reese (London and New York, 1926), I357b.


Page 11: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 11/25


tion, which goes from the particular to the general, nor a syllogismwhich


from the

generalto the

particular;it is a form of reason-

ing that proceeds from the particular to the particular when the two

particularia are contained within the same genus." 16 It may be noted

that a third pair of synonymous terms has made its appearance here,and we can determine, more correctly than before, two new termi-

nological equations: genus equals general, species equals particular.

Example, therefore, is a form of reasoning alongside syllogism and

induction; it appears that we are still missing the fourth term, one

nevertheless theoretically envisaged by Aristotle in the passage from

the general to the general, from the whole to the whole. Induc-tion, like the syllogism, has already been identified in exactly those terms.

Thus: "induction [proceeds] from particulars to the universal";17and "For let E be the conclusion reached by means of the premisesA, B, C, and D. Then some one of these must have been assumed to

be related to some other as whole to part; for it has already been shown

that where there is a syllogism certain of the terms must be so related."18

Indeed, there is no need to consider argumentation as a whole in order

to observe the passage from the general to the particular, and so forth;

those movements can be found inside the propositions which composesyllogism or induction. We know that Aristotle's logic is a logic of

classes, not of propositions; its argumentation describes the relation of

predicates that make up the proposition and not the relations between

propositions. With the condition that one relativizes the concepts"general" and "particular" (that one takes them on a semantic level

as that which is "more general" or "more particular," and not as a

predicate preceded by "all" or "some"), every proposition of a syllogism

(or an induction) is constituted by the passage from the general to the

particular, or conversely (for example, "all men [the particular] are

mortal [the general]"). One should thus be able to rediscover, amongthe variety of syllogisms, which depend (among other things) on the

permutation of terms, the four-termed combinatory system we have

already described.

Such, indeed, is the case. It may be recalled that in systematizinghis teaching the disciples of Aristotle came up with four types of

s!llogism. If M stands for the major term, O for the middle term,

16 Aristotle, Organon, tr. Tricot (Paris: Vrin, 1965-71); Prior Analytics, pp.315-I6, n. 5.

17 Topics, tr. E. S. Forster (Cambridge, Mass. and London, 1938), 156a; cf.


18 Prior Analytics, tr. Hugh Tredennick (Cambridge, Mass. and London, 1938),42a.


Page 12: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 12/25


and m for the minor, and if one abstracts the variety of quantifiers and

relations, thearrangement

of thepremises

of those fourtypes

can be

represented schematically as follows:

Figure I Figure 2 Figure 3 Figure 4

0-M 0-M M O M-0

m-O O-m m-O 0-m

Figures I and 4 are simpler: the same movement of generalization

(I) or of particularization (4) is repeated twice; a single proposition

from each figure can thus be used to represent the logical operationwhich is repeated. The same is not the case for figures 2 and 3, where

the two propositions are heterogeneous. To bring these figures to the

same level of abstraction, we will use only one of their features: in

figure 2, the same subject receives two different predicates (as "text-

book" was both "didactic" and "false"); in figure 3, two subjects are

characterized by the same predicate, like "love" and "flame," both

"burning." To speak in simpler terms, figure 2 describes the coexistence

of twoproperties; figure 3

theequivalence (a partial one,

to besure)of two terms.

Neither the terminological variations, nor the gaps in the combinatory

system, can conceal the identity of the two matrices. Aristotle uses the

same means to describe the discursive (the domain of reasoning) and

the symbolic (tropes). The same word figure which appears in both

places (in the rhetorical tradition if not in Aristotle) symbolizes this

unity. And the lesson is not entirely forgotten in the later tradition:

Father Lamy includes a chapter on the figures of the syllogism that

follows his study of other figures; Du Marsais cites as an example of

synecdoche "mortal" for "men." The proportional metaphor which

we have left aside obviously can be seen to have its discursive double as

well: it is the famous analogy catalogued as follows in the Topics:"Likeness must be examined in things belonging to different genera--asA is to B, so is C to D (for example, As knowledge is related to the

object of knowledge, so is sensation related to the object of sensation"

and so forth (I o8a; cf. also I24a). The difference between the two

is that on the one hand, subjects and predicates are being described,on the other, exposed and imposed meanings (or vehicle and tenor,

etc.). And thus the difference is that either two entities are present

(that is the discursive) or that one of them is omitted but evoked by the

meaning of the other (which produces the symbolical). Tropes are


Page 13: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 13/25


submerged propositions. The symbol is condensed discourse.19

We have passed a little too rapidly over example, in particular wehave not observed any examples of it. Aristotle distinguished two

kinds: historical and nonhistorical (atemporal) examples, which in

turn can be parables or fables. Here is an historical example: the

war that the Thebans waged against the Phocidians was an evil; it

follows that the Athenians should not make war against the Thebans

if they want to avoid evil; these two particular cases are linked by a

general property: Thebans and Phocidians, like Thebans and Athen-

ians, are neighbors (cf. Prior Analytics, 6ga [translation modified]).

And here is a nonhistorical example: ". . . magistrates should notbe chosen by lot, for this would be the same as choosing as repre-sentative athletes not those competent to contend, but those on

whom the lot falls" (Rhetoric, II. i393b). The logical form of these

arguments is not identical to that of syllogism or induction; it is no

longer a relation of classes which is being described but one of proposi-tions. If the first war was harmful, the second will be too; if athletes

are not chosen by lot, magistrates should not be either. The validityof this kind of reasoning no longer depends on the relation between

the functions (predicates) which constitute each proposition-as inthe case of the syllogism-but on the global truth of each proposition.Without noticing, we have gone from the logic of predicates to the logicof propositions (that is what classical logic calls a hypothetical syl-

logism: one whose major premiseis constituted by an implication). An

example, in the Aristotelian sense, is a real or possible event, and not

a substance; we should not be surprised to find it situated alongside

propositional calculus.

Imagine that the Athenians, speaking among themselves and con-

sidering a possible war against the Thebans, were content to evoke the

example from the past without connecting it to the present situation.

By the bias of condensation, we would have passed from the discursive

realm to the symbolical. But what kind of symbolism? The assertion

concerning the first war would have kept its initial meaning (its truth

value); one would have simply added to it a second assertion con-

cerning the present war. In other words, we would have a case of

propositional symbolism. Moreover, in our first example, "animals

19 The use of the word trope in Stoic logic is revealing in this respect. TheStoics recognize two forms for the transcription of reasoning: complete and

abridged. Complete, as in the example: "If it is day, it is light"; abridged: "If the

first, the second" (the complete is a concrete application of the abridged). Theycall complete reasoning logoi, and the others tropoi. Cf. R. Blanche, La logique etson histoire (Paris: Armand Colin, 1970), p. I 3.


Page 14: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 14/25


will be rewarded," we might equally have passed from the symbolicto the discursive realm-which is what the exegete does when he un-

covers a hidden proposition. The rabbinical gloss catalogues not onlyindirect expression under the rubric qal wahomer but also direct expres-sion as a fortiori; thus in the Bible, Moses says to his people: "For

I know how defiant and stubborn you are; even during my lifetime

you have defied the Lord; how much more, then, will you do so when

I am dead?" (Deut. 31:27). Qal wahomer is closely related to the

nonhistorical example, just as the story of Abraham's two wives is re-

lated to historical example.

Let us take another example. Commenting on the quality of theHoly Scriptures, St. Augustine expresses himself this way: "To the

extent that the wealth of gold and silver and clothing that people took

with them from Egypt was less than that they afterwards acquired at

Jerusalem, especially during the reign of Solomon, the knowledgecollected from the books of the pagans, although some of it is useful,is also little as compared with that derived from the Holy Scriptures"

(On Christian Doctrine, II. xlii. 63). In its explicit form, this

reasoning is related to the nonhistorical example, but supposing the

sentence cut in two (before the second "as little") with the secondmeaning suggested by the context, then suddenly one would find one-

self plunged into a form of propositional symbolism.We will modify the Aristotelian terminology in order to correspond

with current usage: example will be called that species of propositional

symbolism in which one passes from one particular to another, and the

corresponding reasoning will be designated by a term borrowed from

child psychologists (Stern), namely, transduction. The latter term-in

a somewhat arbitrary but useful fashion-will be distinguished from

induction (the passage from the particular to the general) and fromdeduction (from the general to the particular) as well as from implica-tion (from the general to the general), all of them being species of

reasoning or inference.

One suspects that each of these discursive forms can, but for the

suppression and indirect evocation of its second part, produce a sym-bolic form. Let us review them rapidly. In one case one states a par-ticular fact in view of illustrating a moral, or a general truth-one

which itself remainsunexpressed;

that will be a case of illustration.In another instance, conversely, the particular case to which the words

apply will be understood implicitly and it will suffice to enunciate the

general truth; we could designate that kind of discourse as a sentence,which is what seems to be intended by the fourth rule of Tichonius, theauthor of the first hermeneutic treatise, Liber de septem regulis, from


Page 15: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 15/25


which St. Augustine borrows in his On Christian Doctrine: in the

speciesis understood the

genus,in the


species.The last variety of propositional symbolism, the evocation of a general

proposition by another, is worth pausing over by virtue of the attention

it has received in classical logic where both Aristotle and the Stoics

give it the name sign. Here is the Aristotelian definition: "That

which coexists with something else, or before or after whose happen-

ing something else has happened, is a sign of that something's having

happened or being" (Prior Analytics, 7oa). Thus, to take the canon-

ical example of the logicians, " a woman has milk" is the sign

that "a woman has given birth." The definition of the Stoicsis almost identical: "Thus, for instance, the Stoics, who seem to have

defined it exactly, in attempting to establish the conception of the sign,state that 'A sign is an antecedent judgement in a valid hypothetical

syllogism, which serves to reveal the consequent."20 The difference

between the two is that Aristotle still thinks he is staying within the

domain of the classical syllogism (consequently within the logic of

classes), while the Stoics correctly identify the propositional nature of

the relation. But, in both cases, one affirms the coexistence of two

predicates which are different in the same subject; one thereforepasses from one (general) genus to another. We will abandon the

term sign which, like that of symbol or allegory, is too general and

too polysemous to usefully serve our purposes; we will replace it with

the term allusion, which, without being precise, is not in contradiction

with current usage.21

We can now summarize the results of this survey in a table-which

needs to be taken only as provisional, not at all as a rigid schema.

The firstpart

of thisstudy

was devoted to the discussion of thecoupleword / proposition; we will not return to it. The last opposition poses

no problem at the level of its components (the dichotomies between

topic and comment, exposed meaning and imposed meaning are suf-

ficiently established), but it may become problematical, at the level

of their assimilation. It is not at all evident, in other words, that the

symbolic copula is "is"; it could also be "and." For the moment

we will leave this question aside, in order to examine at some lengththe two other oppositions.

The choice of the opposition general / particular is the most arbitrary

20 Sextus Empiricus, Outlines of Pyrrhonism, tr. R. G. Bury (London and New

York, 1933).

21 Freud at times uses the term in his sense. Cf. Le Mot d'esprit (Paris: Galli-

mard, 1971), p. 35, n.


Page 16: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 16/25


lexical relations propositional relations

discursive symbolic discursive symbolic

general-I particular particular- particular- deduction sentence

ization izing synec-doche

particular-* general generali- generalizing induction illustra-

zation synecdoche tion

particular-b particular equivalence metaphor trans- exampleduction

general-o general coexistence metonomy implication allusion


The sixteen variables in the table are produced by the play of four

oppositions: word/proposition, discursive/symbolic, general/particular,and topic/comment (this last distinction to be assimilated with the one

between exposed meaning and imposed meaning).

of all; we have simply taken it from the tradition represented here by

Aristotle. We might add that it still has its logical relevance, since itis founded on the relation of inclusion, and it has demonstrated its

efficacy in rhetorical descriptions. But in saying that we are merely

repeating the claims of the tradition; nothing proves that this notion

is the one most appropriate to categorize statements and symbolic evoca-

tions (nothing, however, proves the contrary). One could defend this

choice on the basis of its not preventing us from adding other categorieswhich would additionally complicate the table; nor does this initial

choice exclude the possibility of other choices or in any way prejudice

their hierarchy. Here are some supplementary categories which illus-trate this possibility. First, speaking of inclusion, we have insisted on

the relation general / particular (genus / species), leaving aside the

relation between the whole and the part (the decomposition of a totalitycan be material or conceptual). All discursive and symbolic forms


Page 17: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 17/25


will therefore have two variants, depending on which analysis one per-forms. This distinction which was not unknown in

antiquity (St.Augustine spoke of "that subtlety taught by the dialecticians who dis-

pute most acutely the difference between a part and a species": On

Christian Doctrine, III, xxxiv. 47; but the distinction can alreadybe found in Quintilian, Institutes, VIII. vi. I9). This distinction will

permit us, among other things, to envisage without difficulty sentences

in which the copula is not "to be" but "to have": "war is harmful"

but "the woman has milk."

Another subdivision, which at least is useful in the study of symbolic

forms, is analogous to the one which permits the opposition in logicbetween inclusion and appertaining: it is the distinction between propernoun and common noun. If in the place of the "particular" one sub-

stituted "singular" (a proper noun) one would obtain a figure that

rhetoric calls personification. Strictly speaking, this term designates

only the medieval practice by which notions like temperance become

characters, through the simple play of capitalization (Temperance).One should rather speak of depersonification (which is only a variant

of antonomasis) in the contrary case, when a proper noun is read asa common noun, in the manner of Cratylus or the Stoics ("Saturnis time, as confirmed by the assimilation of their Greek names Kronos-

khronos; if in Latin he is called Saturnus, it is because he is 'saturated

with years,'satureturannis").

Finally, the description can be refined through specification of the

nature of the elements placed in a symbolic relation. The result, accord-

ingly, will be different depending on whether we consider a discourse

as that which isenunciated,

an utterance(which

has been the case

until now), or as an enunciation, a speech act. "Lend me some bread"

symbolizes (illustrates) poverty in terms of what is stated (the semantic

content); the same sentence, considered as a part of the statement,

symbolizes a marginal universe in which slang is used. In this case, in

place of the opposition particular/ general, it is more suitable to

envisage a continuum within the realm of what is symbolized, which

proceeds from the individual (relations of "intertextuality") to literarystyle or to the social milieu which employs a certain jargon (such are

the "effects produced by the evocation of a milieu," in Bally's terms).Other categories may naturally be added to these.

Let us return now to the opposition discursive / symbolic, whose

single relevant feature we have taken to be condensation (omission).Once again we could refer to the tradition and notably to the descrip-


Page 18: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 18/25


tion poets give of their art (whose relationship with the symbolic has


acknowledged):Mallarme said that "to name an object

is to suppress three-quarters of the pleasure of the poem which derives

from the satisfaction of guessing little by little; to suggest the objectis the dream"; and Pound epitomized his poetic doctrine with the

formula: Dichten = condensare. But that does not suffice to resolve

the question which must be viewed from several angles.

One must first inquire whether the simple fact of not saying some-

thing even though it may be deduced from our words deserves to be

called symbolic (expression). Or from another point of view: is all

linguistic implicitness symbolic? I think it is preferable to apply thatterm to only one of the kinds of implicitness. One finds somethinglike that distinction in Aristotle: on the one hand there is what he

calls sign, by which one designates an implied meaning; on the

other hand, an inert implication that could be explicitated but is not

intended by the speaker. That implicit meaning belongs either to

collective memory (encyclopedic knowledge), as in this example: "If

any one of these premises is well known, there is no need to mention it,for the hearer can add it himself. For instance, to prove that Dorieus

was the victor in a contest at which the prize was a crown, it is enoughto say that he won a victory at the Olympic games; there is no need

to add that the prize at the Olympic games is a crown, for everybodyknows that" (Rhetoric, I. I357a: the implicit-the major premise-is clearly distinguished here from the sign-the conclusion); or the im-

plicit meaning resides in the very meaning of words [analytic implica-

tionl, as follows: "Furthermore, a man who has made an assertion

of any kind whatsoever, has in a way made a number of assertions,

because each assertion necessarily involves a number of consequences.For example, he who has said that 'X is a man,' has also said that

X is an animal and a biped and is animate and is receptive of reason

and knowledge" (Topics, I 12a).

We might agree that the symbolic is merely a focalized form of the

implicit, but what does "focalized" mean? Elsewhere I have formulated

some suggestions on that subject; 22for the moment let it suffice to say

that it is a process concerning the perception of language and not

language itself; that it is thus a psycholinguistic category and not alinguistic one. More generally, it is clear that the omission bv which

the svmbolic process is set in motion gives rise in the speaker to a

22 "Recherches," 244. Cf. also Dan Sperber, Le symbolisme en general (Paris:Hermann, 1974), Ch. v.


Page 19: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 19/25


type of activity different from the one he performs in understandinga nonsymbolic statement. We will content ourselves here with merely

indicating the problem without studying it: signs are understood, sym-bols are interpreted.

But, if interpretation is a process different from understanding,

something within each particular speech act must indicate to the

speaker the appropriate attitude. Here again the question is too vastto be treated within the present framework. We will only add thaton that question classical hermeneutics when it examined the "indices

of allegory" as well as modern logic (Grice, Ducrot) have already

staked out the initial elements of such a study.Finally, the criterion of condensation may also be questioned from

another point of view: can any discursive form by simple suppressionproduce a symbolic form? I have given some examples of such trans-

formations; that does not prove that the movement is always possible.Since the Middle Ages rhetoric has sought to be the taxonomy of

already existing forms; it has forgotten the generative perspective ithad at its origins and which could be concretized through questionssuch as: do impossible metaphors exist? One might say: the fork is

like the knife (insofar as both tear food: a discursive equivalence);but one can never indicate a fork through the metaphor of a knife or

conversely. (The same goes for door and window, fox and wolf,sweater and shirt: could it be that the "common species" which is the

ground of metaphor should not work as a lexical class?)

Alongside these constraints inherent in language we dimly perceivethat there are others linked to perceptual norms to which classicalrhetoricians were undoubtedly referring when they used the term

"usage": "It should not be

thought,"wrote Du

Marsais,"that one

name can be taken indifferently for another either by metonymy or by

synecdoche: once again, it is necessary that figural expressions beauthorized by usage.... If one were to say that a fleet was composedof a hundred masts or a hundred oars instead of saying a hundredsails to mean a hundred ships, one would appear ridiculous: every partcan not be taken for the whole, and every generic noun can not betaken for a particular species nor every species noun for the genus; itis only usage which as it chooses gives that privilege to one word ratherthan another."23 But what establishes the basis of

"usage"? Onesuspects that there lies a vast field of study, again of a psycholinguisticrather than a linguistic order, towards which new rhetoricians will haveto turn (the same questions arise in the study of propositional svm-

23 Des Tropes (18i8 ed.), p. 127.

I 28

Page 20: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 20/25


bolism). Such tasks belong strictly to the realm of the symbolic, while

the logical structure of symbolic forms, since they are identical to dis-

cursive forms, have to be studied within a framework anterior to the

very division between the symbolic and the discursive.

It should be added that the identity of logical structure appears onlyif we pass within the discursive realm to the hierarchically superior

segment. The symbolism of letters within discourse produces words;word symbolism produces propositions; propositional symbolism pro-duces arguments. Recall again our example of Adam: each letter of

his name signifies a word; the antonomasis is translated by a proposition

("Adam is man par excellence"); the "implied meaning" resultsfrom reasoning that consists of several propositions.24 That, after all,is merely the natural result of the work of condensation.

That both discourse and symbol participate in the same logicalstructures should not surprise us: unlike the sign, a ternary structure,both the one and the other are founded on binary relations,

Aristotle, in his way, said the same thing in affirming that the investiga-tion of resemblances is as fundamental for the creation of tropes ("For

right use of metaphor means an eye for resemblance," Poetics, I459a)

as for the constitution of discursive reasoning: "It is not necessary,in order to make demonstration possible, that there should be Forms

or some One apart from the Many; but it is necessary that it should be

true to state a single predicate of a plurality of subjects. Otherwise

there will be no universal term; and if there is no universal there will

be no middle term, and hence no demonstration. Therefore there must

be something which is one and the same above the several particulars,and does not share a common name with them."25

III. The Theory and History of Linguistic Symbolism

One might think that a theoretical (and typological) approach to

linguistic symbolism turns its back on the concrete history of symbol

production and on the description it has received in the past. It was

to avoid that misunderstanding that in the preceding discussions I chose

examples from actual exegeses. More generally, I think that only a

24 We should pay homage here to William Empson (The Structure of ComplexWords [London, I950]), who was the first to have understood the necessity of

confronting tropes not with words but with propositions; his title "Statements inWords" is a whole program in itself.

25 Posterior Analytics, tr. Hugh Tredennick (Cambridge, Mass. and London,I938), 77a.


Page 21: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 21/25


rigorous theory-which the preceding sketch is not-will allow prob-lems of history to be clarified, whether it be the


of symbol

production or of exegetic schools. Nevertheless the preceding sketch

might become such a theory by virtue of being confronted by the real

complexity of facts. As an example, I would like to examine here an

historical problem in order to illustrate the mutual advantages to be

derived for theory and history from their being taken together.

This problem is still being debated today: it is the question of the

originality of Christian allegory in relation to contemporaneous or early

pagan allegory, like that which was practiced, notably, in ancient

Greece. One can make out the two opposing theses: according tocertain authors, the difference is purely substantial; an already existingform (pagan allegory) was applied to a new subject matter (Christian

ideology); according to other authors, including several churchmen,Christian allegory is entirely different from pagan allegory, even in its


In order to pass any judgment on the issue one must first examine

a little more closely what is meant by Christian allegory. We know

that since the patristic era the Scriptures have been thought to containmultiple meanings. The most common variant of this theory consists

in saying that its meaning is quadruple, articulated first on an opposi-tion between a literal (or historical) meaning and a spiritual (or

allegorical) meaning, the latter being then subdivided into three: an

allegorical (or typological) meaning, a moral (or tropological) mean-

ing, and an anagogic meaning. St. Thomas' formulation for a longtime had been common opinion:

That first meaning whereby the words signify things belongs to the sensefirst mentioned, namely the historical or literal. This meaning, however,whereby the things signifiedby the words in their turn also signify otherthings is called the spiritual sense; it is based on and presupposes theliteralsense.

Now this spiritualsense is divided into three. For, as St. Paul says, TheOld Law,is the figure of the NewV, nd the New Law itself, as Dionysiussays, is the figure of the glory to come. Then again, under the New Lawthe deeds wroughtby our Head are signsof what w,eourselvesought to do.

Well then, the allegoricalsense is brought into play when the things ofthe Old Law signify the things of the New Law; the moral sense whenthe things done in Christ and those who prefiguredhim are signs of whatwe should carry out; and the analogical when the things that lie aheadin eternalgloryare signified.26


Page 22: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 22/25


First, let us clarify some matters of terminology. Moral meaningis also called tropological-a term it is better to avoid in order not to

confuse it with "trope." "Allegory" sometimes designates the three

last meanings taken together, sometimes only one of them; again, to

avoid confusion, we will speak of spiritual meaning in the first case,

typological meaning or more simply typology in the second case-

although this last term is not in ancient usage. Here is an example of

an interpretation according to the four meanings used by Dante in the

famous, although perhaps spurious letter to Cangrande:


ourprocedure may appear more clearly we will apply it tothe following lines: "When Israel went out of Egypt the house of

Jacob from out of a barbarouspeople, Judea became its sanctuary and

Israel its domain" (Psalm CXIV). For if we keep to the letter of the

text, the meaning is the exodus of the children of Israel out of Egyptin the time of Moses. If we considerthe allegory, the sense is our redemp-tion effected by Christ. In the moral sense, it is the conversion of thesoul from the pain and povertyof sin to the state of grace. In the anagogi-cal sense it is the passage of the holy soul from the slavery of present

corruptionto the freedomof eternal


Without going into any further detail it can be observed that the

three spiritual meanings are uncovered on the basis of whole proposi-

tions; in other words, we are dealing with propositional symbolism.This observation is usually formulated as the necessity of maintain-

ing the literal meaning. And very often, it is precisely in the mainte-

nance of the literal sense that the specificity of Christian allegoryhas been found: indeed, pagan allegory calls for its abolition. Thus

Auerbach says, for example:

In the case of allegory or symbolism,at least one of the two elementsthat combine is a pure sign; w-hile in typological relations the facts

signifying and signified are both real, concrete historical events. In alove allegory or in religioussymbolism,at least one of the terms does not

belong to human history; it is an abstraction or a sign. Conversely, inthe sacrificeof Isaad considered as the figure of Christ'ssacrifice,neitherthe prefiguringnor the prefiguredevent, despite their meaning and their

26 Summa Theologiae, I, 40-41; Question I, Article Io, Conclusion. The classical

exposition of the question of the four meanings is found in E. von Dobschutz, "Vomvierfachen Schriftsinn. Die Geschichte einer Theorie," Harnack-Ehrung. Beitrdgezur Kirchengeschichte (Leipzig, I921). The four-volume Exegese medievale, byH. de Lubac (Paris: Aubier, 1959-64) examines the question from all anglesbut is not easily usable.


Page 23: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 23/25


figurative relation, lose their literal, historical reality. This point is

essential and has been emphatically underlined, as least in the Western


Or de Lubac writes:

Two meanings [as in Christian allegory] which are added to one another,or two meanings in which the first, in itself very real although exterior,

eclipses itself behind another or transforms itself into the other, beginningwith a creative or transfiguring event, are not two meanings which [as

in Greek allegory] exclude one another the way appearance and reality

or truth and "falsehood" exclude one another. Not any more than ap-pearance or "falsehood" in Greek mythology correspond to the "letter"

or the "history" of the Christian exegete, the "truth" of the first does

not correspond even from a wholly formal point of view, to the truth of

the second.... Thus far from constituting the even approximate analogueof the Greek terms to which we might be tempted to assimilate them,the Christian terms rather constitute their antithesis.28

Pagan allegory, in effect, belongs to lexical symbolism. But that in

no way proves the originality of Christian allegory; it is not alone inbelonging to propositional symbolism, and not only in practice, which

goes without saying, but also in theory (as in the case of the sign in

Aristotle or the Stoics, or certain figures of thought, example for

instance, among the rhetoricians). The difference, if it exists, must be

sought at a more specific level.

Let us return rapidly to the three kinds of spiritual meanings. Moral

meaning is the one that poses the least problems (as to its identifica-


resembles,in every

way,the form of

thoughtthat Aristotle

described under the name of example-even including the examplesthemselves: a certain past action (of Sacred History) must be placedin parallel with present actions-and serve as their guide. There is no

formal difference here; from the point of view of allegorical theorythe Thebans' war is equivalent with the war of the children of Israel.

But the same is not the case for typology; and, in fact, one is usually

thinking about that when one speaks of Christian allegory. Adam is

the type of Christ, who is his antitype. It is not a question here of

example. But of what then?

27 "Typological Symbolism in Medieval Literature," Gesammelte Auifsiitze zurromanischen Philologie (Bern: Francke, i967), p. Ill. Cf. also his study

"Figura," ibid., pp. 55-92.

28 Exexgesemedievale, II, 517.


Page 24: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 24/25


Historians of theology have answered that question. Thus we could

enumerate the features of typology proceeding from the most generalto the most specific:

(i) It belongs to propositional symbolism.

(2) It participates in the intersection of properties, and not in ex-

clusion or inclusion; in this sense it belongs to the realm of example.

(3) The two facts which constitute it belong to the past; they are

two historical facts. That is not enough however to characterize

typology; indeed in the history of exegesis a sentence of Plutarch (De

Fortuna Alexandri, X) is quoted, according to which in the verse ofHomer "both a good king and an excellent warrior" is taken simply in

praise of Agamemnon, but also predicts the greatness of Alexander;29

this is an historical example, like those in Aristotle, but not a typology,for the events repeat themselves without the one being considered to

be the fulfillment of the other.

(4) Only a particular relation between two facts can be spoken of

as typology, within historical examples; this relation in a strict sense

nolonger belongs

to our formalframework;

it is a relation offulfillment.There must be a gradation of the two meanings in favor of the second

one; the first announces the second, the second fulfills the first. To placethem on the same level within a Christian perspective would be heresy.

(5) The following restriction is purely one of content; we may

agree to call Christian typology one which is realized within the frame-

work of a certain ideology. This restriction is necessary because as

Goppelt has clearly shown, there exists a non-Christian typology.

(6) Finallywithin Christian

typology,we can isolate

testamentarytypology, according to which the events of the Old Testament announce

those reported in the New Testament. That is what the "second" mean-

ing refers to (in the theory of fourfold meaning), which we have

just designated as "typology." This new restriction is necessary because

the fourth meaning, the anagogical, shares certain properties with

typology without however being a testamentary typology. Anagogical

meaning concerns eschatology. Starting from a series in which the Old

and New Testaments are taken together we can infer another, yet to

come (the end of the world). The difference then is double: we are

dealing with a prophecy and not an interpretation of the past; there is

29 Quoted by L. Goppelt, Typos, Die typologische Deutung des Alten Testamentsim Neuen (Gutersloh. C. Bertalsmann, 1939), p. 20 (my italics). This book

contains an illuminating discussion of the problems of typology.


Page 25: Todorov on Linguistic Symbolism

8/3/2019 Todorov on Linguistic Symbolism 25/25


no text here that plays a role analogous to that played in testamentary

typologv bythe New Testament in relation to the Old.

It is necessarytogive a definition of "typology" which does not link

it to Christian doctrine-in order to observe whether the same "historical

example of fulfillment" is found elsewhere. Without wanting to pursuethat line I would suggest that here are many "typologies" within the

great interpretative strategy of our times, psychoanalysis. The two

events are no longer situated within the history of humanity but within

that of the individual; it remains the case that a recent event (neurotic

symptoms, for example) is perceived as the fulfillment of a past event

(the infantile trauma), which in turn announces the other.This example shows how history and theory may help clarify one

another. It is by virtue of the clarification of descriptive concepts in use

that we have been able to observe the error of earlier observations:

the difference between Christian and pagan allegory is not where we

expected to find it, in our terms, between lexical and propositionalsymbolism. The detailed consideration of an historical situation has

led us in turn to refine our conceptual apparatus: in order to circum-

scribe the specificity of typology, we are obliged to consider closely the

relation between the events designated by the two propositions; the

opposition between repetition and gradation proves here to be funda-mental.

This being the case, it should be recalled that all forms of linguisticsymbolism are found at work in language and, as a result, of every

exegetical school; the differences between them, therefore, will neverbe as clear as those that are determined abstractly. One will ratherhave to wait for an exegetical school to render obligatory what has

remained until nowdiscretionary,

or for aliterary

school that uses a

procedure abundantly while another uses it only exceptionally.Lovers of purity in the meantime will have to be content with an

abstract theory.



(Translated by Richard Klein)