[A Bulgarian-born French critic, Todorov is a significant scholar in structuralist and post-structuralist theory. His writings include Littérature et signification (1967); Introduction à la littérature fantastique (1970; translated and published as The Fantastic: A Structural Approach to a Literary Genre, 1973); and Théories du symbole (1977). In the essay below, Todorov explores Bakhtin's theory of the utterance as rooted in social context.]

Bakhtin formulates his theory of the utterance on two occasions: once during the late twenties, in the texts signed by Medvedev and especially by Voloshinov; and in several works published at the end of the fifties, some thirty years later. I will present these two syntheses separately, although there is no great difference between them (in fact, the only changes involve accentuations of various aspects of the utterance).

The first general formulations concerning the utterance are already to be found in Freudism (1927); one page of The Formal Method in Literary Studies (1928) evokes this problem from a similar viewpoint, with an insistence on the social rather than the individual nature of the utterance; but Bakhtin introduces here a new notion, which is not reiterated in subsequent writings: that of a discursive strategy.

Discursive strategy plays a particularly significant role in daily verbal communication by determining its form as well as its organization. It gives form to everyday utterances by establishing both the style and the genre of the verbal expression. Strategy is to be understood here in a broad sense: politeness represents but one of its moments. This strategy can pursue different directions, moving, as it were, between two poles—the compliment and the insult. The strategy is determined by the set of all social interrelations between the speakers, by their ideological horizons, and finally by the concrete situation of the discussion. Whatever may be its particular nature, such a strategy determines our every utterance. There is no discourse without strategical consciousness.

In Marxism and the Philosophy of Language (1929), Bakhtin accomplishes a major step by forsaking his general theories to propose instead a detailed description of the utterance: this will constitute Chapter 3 of the second part, entitled "Verbal Interaction."

One may recall the criticism which Bakhtin voiced about the "individualist subjectivism" school (Vossler and his disciples): although superior to that of Saussure insofar as it does not ignore the utterance, yet it mistakenly believes that this utterance is individual.

Any moment of the expression-utterance one may observe will invariably be determined by the real conditions of the speech-act, primarily by the nearest social situation. Verbal communication will never be understood or explained without a reference to its link with the concrete situation.

In other words, the difference between an utterance and a proposition (or a sentence), a unity of language, is that the former is necessarily produced within a particular context which is always social. This sociality has a dual origin: first of all, the utterance is addressed to someone (this implies the existence of a micro-society comprising two people, the speaker and the addressee); secondly, the speaker himself is always a social being to begin with. These are two primary elements of the speech-act context which we need to consider in our interpretations of an utterance.

Let us first observe the role of the addressee. The utterance is established between two socially organized people: should there be no real interlocutor, then he is presupposed, in a certain sense, as a normal representative of the social group to which the speaker belongs. The discourse is oriented towards the interlocutor, towards what the interlocutor is.

Instead of the individual interlocutor we can thus imagine a certain type of addressee or, in other words, a certain horizon of reception; a notion we shall again encounter in an article published the following year (1930):

From the daily primitive utterance to the achieved poetic utterance, each one invariably comprises, as a necessary ingredient, an "implied" extra-verbal horizon. We can analyze this living and concrete horizon in terms of three components: spatial, semantic, and of values. The value horizon assumes the most important role in the organization of a literary work, especially in its formal aspects. [V. Voloshinov, "O granizakh poetiki i lingvistiki," in V bor'be za marksizm v literaturnoi nauke, 1930]

As we shall see, Bakhtin later returns to this question of values (although the suggestions formulated above will not be pursued).

The sociality of the speaker is equally important, albeit less evident. After taking certain precautions (acts of acoustical phonation and perception are indeed individual but they do not concern the essential aspect of language: its significance; a biological and individual "I-experience" does indeed exist, however, unlike the "we-experience," it remains inaccessible), Bakhtin states that the expression of an individual is not individual in the least.

There can exist no experience beyond its incarnation in signs. This immediately precludes the possible principle of any qualitative difference between interior and exterior.(…) Expression is not organized by experience, but on the contrary, experience is organized by expression which, for the first time, imbues this experience with form and direction. Aside from material expression, there is no such thing as experience. Moreover, expression precedes experience; it is the cradle of experience.

A footnote to the last sentence declares that this "assertion was in fact originally drawn from certain statements of Engels" which are to be found in Ludwig Feuerbach; beyond this, we can perhaps perceive a more distant and common source in the work of Humboldt (the inspiration for "individualist subjectivism"): an experience is preformed by the possibilities of its expression. Once we have located the formative traces of an expression at the very core of the expressible, then whatever its sources may be, there can no longer exist any sphere which is entirely devoid of sociality (since words and other linguistic forms do not belong to the individual).

Only the inarticulate animal cry is truly organized within an individual physiological system.(…) But even the most primitive human utterance, produced by the individual organism, is already organized in external terms, through the inorganic conditions of a social milieu which shapes its content, significance, and meaning. The very howls of an infant are "oriented" towards its mother.

We might formulate this observation by saying that every utterance can be perceived as part of a dialogue, in the general sense of the word; only in his subsequent writings will Bakhtin define this more specifically (as a dialogue between discourses).

Verbal interaction is the fundamental reality of language; and dialogue, in its narrow sense, is a single form, though clearly the most important one, of verbal interaction. But dialogue can be interpreted in a much broader manner, as referring not only to the direct verbal communication which is voiced between interlocutors, but also encompassing every form of verbal communication.

As a first important consequence of this new framework, we must radically distinguish between meaning in language from meaning in discourse or, to use the terminology adopted by Bakhtin at the time, to distinguish meaning from theme. In and of itself, this distinction is nothing very new; however, it will quickly become so, due to the increasing importance Bakhtin attaches to the theme. Indeed, the standard oppositions of that period between current and occasional meaning, between fundamental and marginal meaning, or between denotation and connotation, are equally fallacious in that they favor the first term, while in fact discursive meaning, or theme, is never marginal.

Thus we will strictly reserve the term "meaning" for language ("langue"); meaning is recorded by dictionaries, and any one meaning is always identical to itself (since it is merely potential): in other words, like all other linguistic elements, it can be repeated.

Meaning in opposition to theme, will represent those moments of an utterance which can be repeated and yet remain identical to themselves. Meaning actually signifies nothing except for the potentiality, the possibility of meaning within a concrete theme.

In contrast, the theme—like the utterance as a whole—is unique and cannot be repeated, since it arises from the interaction of meaning with the equally unique context of the speech-act.

Let us call the significance of an entire utterance its theme.(…) In fact, like the utterance itself, the theme is individual and cannot be repeated. It is an expression of the concrete historical situation which engendered the utterance.(…) It must then follow that the theme of an utterance is not only determined by the linguistic forms which compose it—words, morphological and syntactical forms, sounds, and intonation—but also by the extra-verbal aspects of the situation. And if we should lose these aspects, we will not be able to understand the utterance, as if we had lost the most important words themselves.

One essential feature of a theme, and therefore of an utterance, is that it is endowed with values (in the broad sense of the term). Vice versa, meaning, and therefore language, do not share this relation with the world of values:

Only an utterance can be beautiful, just as only an utterance can be sincere, delusive, courageous, or timid, etc. These value determinations are linked to the organization of utterances and literary works insofar as they involve the functions assumed by the latter within the unity of social existence and, above all, within the concrete unity of an ideological horizon.

The idea of an evaluative dimension in the utterance is further pursued by the article already referred to, "On the Boundaries Between Poetics and Linguistics." Bakhtin investigates the possible formal embodiments of this value judgment; and first considers the use of non-linguistic means.

Let us say that any evaluation which is incarnated through the (verbal) material is an expression of values. The human body itself will provide the original raw material for such an expression of values: gesture (the signifying movement of the body) and voice (outside of articulated language).

Within language itself, phonetic means are naturally to be distinguished from semantic means; and somewhat more remarkably, these are classified according to a dichotomy between selection and combination; this division is familiar today, but was unpublished at the time (although one may seek its origin in the work of Kruszewski).

We must distinguish two forms of value expression [in poetic creation]: 1) phonic and 2) structural [tektonicheskuju], whose functions can be separated into two groups: first, elective (selective), and secondly, compositional (organizational). The elective functions of the social evaluation emerge through the choice of lexical material (lexicology), the choice of epithets, metaphors, and other tropes (the entire range of poetical semantics), and finally, through the choice of a "content." In this way, most stylistics and certain elements of thematics belong to the elective group.

The compositional functions of the evaluation determine the level and hierarchical positioning of each verbal element in the work as a whole; they also determine its general structure. This involves all problems of poetic syntax, of composition in its literal sense, and finally of genre.

In the first book signed by Bakhtin himself, which is devoted to the work of Dostoevsky, the utterance will assume a new dimension, whose importance will steadily increase: every utterance can be linked to preceding utterances, thereby giving rise to intertextual relations. In this first edition, Bakhtin does not concern himself with general theories but rather with a typology of the utterance, thus he merely states:

No member of the verbal community will ever discover any words in language which are totally neutral, devoid of another's aspiration and evaluations, or free of another's voice. No, a word is apprehended through the voice of another which will remain forever imbedded within it. A word reaches one context in terms of another context, penetrated by the intentions of another; its own intentionality encounters a word which is already inhabited. (In the second edition of the work, 1963, the instances of "intention" will disappear to be replaced by osmyslenie, interpretation, and mysl', thought.)

In a previously cited article, signed by Voloshinov, these contentions, as well as several others, are paraphrased with one curious variation: "intonation" here replaces "intention":

For the poet, language is permeated with living intonations; it is entirely contaminated by social considerations and by the embryonic phases of social orientations. The creative process must continually struggle with such elements; it is from among their midst that one must choose one linguistic form or another, one expression or another, etc.… An artist never receives any word in a linguistically virginal form; it has already been 'impregnated' by the practical circumstances and poetic contexts in which it is encountered.(…)

This is why the work of a poet, like that of any artist, can only accomplish certain transvaluations, or certain displacements of intonation; these will be perceived by the artist as well as his public through the perspective of previous evaluations and intonations.

Let us now turn to the second synthesis which appears in the notes written during the fifties, and published after Bakhtin's death, under the title "The Problem of the Text"; the "Methodological Remarks" of the second edition Dostoevsky presents a summary of these issues. The frame of reference is no longer sociology, as it was thirty years earlier, but now involves translinguistics, the new discipline Bakhtin intends to found, whose primary object will be the utterance. Three factors are immediately set forth to distinguish an utterance from a sentence: an utterance has a speaker and an object, moreover it partakes in a dialogue with previous utterances.

The utterance is determined not only by its relation to the object and the speaking subject—the author (and by its relations to language as a system of potential possibilities, or givens) but, most importantly from our perspective, it is directly determined by other utterances within the framework of a certain field of communication. In simpler terms: purely linguistic relations (that is to say the object of linguistics), comprise the relations between one sign and another, or several others (in other words all systematic or linear relations between signs). The relations an utterance may have with reality, the real speaking subject, and other real utterances, that is to say, those relations which render the utterance true, false, or beautiful, etc., can never become an object of linguistics.

We must make a slight digression at this point concerning the speaking subject, the speaker. He is viewed as a constituent element of a speech-act and thus of an utterance; at the same time, one refers to the image of the author which is deduced from the utterance; and one naturally tends to project the second onto the first. However, a clear distinction between the two must be maintained. An author produces an entire utterance which does comprise the "image of the author" but he himself is a producer and never a product, natura naturans instead of natura naturata.

Even if an author-creator could create the most truthful autobiography of confession, he would still remain excluded from the universe he has portrayed simply insofar as he has produced it. If I should recount (or write) an event I have just experienced, then the mere act of narrating (or writing) this event will place me outside the time-space in which it has occurred. It is impossible to be absolutely identified with one-self, to reconcile one's veritable "I" with the "I" of his narration, just as it is inconceivable to lift oneself up by his own hair. However realistic and authentic a represented universe may be, yet it can never be chronotopically identical to the real representing universe in which the author-creator of the representation is located. For this reason, it seems to me that the term "author's image" is quite unfortunate: what has become an image of the work and thereby entered its chronotope, is a product, not a producer. "The author's image," when perceived as the image of the author-creator, is a contradictio in adjecto; each image represents something which has been produced and cannot be a producer.

Let us return to the general scheme of the utterance. We have seen that language ("langue"), the speaker, the object, and other utterances are all to be taken into consideration; we must not forget the addressee.

Discourse (like any sign in general) is interindividual. All that is said or expressed exists outside the "soul" of the speaker; it does not belong to him. Discourse cannot be attributed to the speaker alone. He clearly holds inalienable rights over the discourse, but the auditor has certain rights as well, as do those, whose voices reverberate in the words chosen by the author (since there are no words which do not belong to somebody). Discourse is a drama with a cast of three characters (not a duet, but a trio). It is performed outside the author, and one may not introject it (introjection) back into him.

Meaning, a property of language, will be opposed here to significance; this more familiar term replaces theme and links the utterance to the world of values which language does not know.

Isolated signs, and linguistic or textual systems (insofar as they represent a unity of signs) can never be true, false, or beautiful, etc. Only an utterance can be exact (or inexact), beautiful, just, etc.

We can summarize the preceding observations by reconstituting a communication model according to Bakhtin, and by comparing it with the currently more familiar model which Roman Jakobson has presented in his article "Linguistics and Poetics."

      Bakhtin
      object
  speaker utterance auditor
      intertext
      language
 
      Jakobson
      context
addresser message addressee
      contact
      code

Two kinds of differences are immediately apparent. Jakobson isolates "contact" as an independent factor. This is absent from the Bakhtinian model, yet the relation to other utterances (which I have designated here as the "intertext") is absent from Jakobson's schema. There are then a series of differences which would seem to involve minor questions or terminology. Jakobson uses rather general terms (semiotic as well as linguistic) and they reveal the influence of his frequent associations with communication engineers. "Context" and "object" both correspond to that which other language theorists would call the "referent."

But after a more careful scrutiny, it is clear that the differences are much more important, and that the terminological discrepancies betray a deeply-rooted opposition. Jakobson sets forth these notions as a description of "the constitutive factors in any speech event, in any act of verbal communication." While for Bakhtin, there exist two radically distinct events, so distinct that they necessitate the use of two independent disciplines, linguistics and translinguistics. In linguistics, words and grammar rules provide the initial basis for the formation of sentences; in translinguistics, one starts off with sentences and the speech-act context eventually to obtain utterances. From Bakhtin's point of view, any attempt to formulate a proposal concerning "any speech event," that is to say, of language as well as discourse, would be futile. In the very schema I have drawn above, the "language" factor is not to be considered on a par with the others.

Moreover, it is no accident that Bakhtin says "utterance" instead of "message," "language" rather than "code," etc.: he quite deliberately rejects the use of engineering language to speak of verbal communication. This language could all too easily lead us to perceive a linguistic exchange in terms of telegraphic work: in order to transmit a certain content, one telegrapher first encodes it with a key and then broadcasts it; once contact has been made, the other uses the same key to decode the message and recover the initial content. This image does not correspond to discursive reality: in fact, prior to the speech-act, the speaker and the addressee literally do not exist as such; it is only the discursive process which thus defines them in relation to each other. For this reason, language is not to be considered as a code; for this reason as well, Bakhtin cannot possibly isolate one "contact" factor amidst the others: the entire utterance is contact, but in a stronger sense of the word than the "contact" of radiotelegraphy or electric work.

It is quite curious to find a page in the book signed by Medvedev which criticized the Jakobsonian model of language, thirty years before it was actually formulated; however one must note that the critique was written as a reply to certain theories of the Formalist group—to which Jakobson belonged.

That which is transmitted cannot be separated from the forms, the means, and the concrete conditions of the transmission; whereas the Formalist interpretations tacitly presuppose an entirely predetermined and immutable communication, as well as an equally immutable transmission. This might be explained schematically in the following manner: let us take two members of Society, A (the author), and B (the reader); for the time being, the social relations between them are unchangeable and immutable; we also have a prepared message X, which A must simply deliver to B. In this prepared message X, the "what" ("content") is distinct from the "how" ("form"), since literary discourse is characterized by the "set toward the expression" ("how") [this is a quotation from the first published text of Jakobson].

(…)

The schema set forth above is completely wrong. In actual fact, the relations between A and B are in a state of continual formation and transformation; they are further modified during the very process of communication itself. There is no prepared message X; it is established by the communicative process between A and B. Moreover, it is not transmitted from one to the other but is built between them like an ideological bridge through the process of their interaction.

Thus, in 1928, we can discern a rather precise prefiguration of certain recent French language theories which are sometimes based on the work of Benveniste (for example those of Oswald Ducrot or François Flahault).

As we now turn from the model of the particular utterance to the set of utterances constituting the verbal life of a community, we should note the fact which would appear to be most striking in the eyes of Bakhtin: there exists a large, but nonetheless limited, number of utterance or discourse types. One must indeed beware of two possible extremes: first, to recognize the diversity of languages and ignore that of utterances; secondly, to consider this variety as being individual and therefore limitless. Besides which, Bakhtin accentuates difference rather than plurality (one need not attempt to conceive of any common denominator which would reconcile various discourses; the argument here runs counter to the idea of unification). To designate this irreducible diversity of discursive types, Bakhtin introduces a neologism, raznorechie, which I translate (literally but in Greek) as heterology; this term is flanked by two parallel neologisms, raznojazychie, heteroglossy, or diversity of languages, and raznogolosie, heterophony, or diversity of voices (individual).

We will recall that every utterance is oriented towards a social horizon which comprises semantic and value elements. The number of these verbal and ideological horizons is quite high but not unlimited; and every utterance necessarily falls within one or several of the discursive types determined by a horizon.

There are no longer any words or forms in language which are neutral and belong to no one: it appears that language has been pillaged, pierced through and through by intentions, and accentuated. For a consciousness which exists within language, it is not an abstract system of normative forms but a concrete heterological opinion of the world. Each word evokes a profession, a genre, a trend, a party, a particular work, a particular man, a generation, an age, a day and an hour. Each word evokes a context and the contexts within which it has experienced an intense social life; every word and every form is inhabited by intentions. Contextual harmonies found in a word (of the genre, of the trend, of the particular individual) are inevitable.

Through the preceding enumerations we can already see that the stratification of language in discourse is not restricted to one dimension. In the course of the most detailed study which he devoted to heterology (in "Discourse in the Novel," text of 1934–35), Bakhtin discerns up to five types of stratification: genres, profession, social levels, ages and regions (dialects strictu sensu). Let us merely note that social class does not play a different role from that of profession or age group: it is simply one diversifying factor among several others.

In a certain sense, heterology is inherent to society; it is engendered spontaneously by social diversification. But just as the unique state attempts to contain this social diversity by means of its laws, so do the authorities fight the diversity of discourse by aspiring to a common language (or rather idiom).

The category of common language is a theoretical expression of the historical processes of unification and centralization—an expression of the centripetal forces in language. A common language is not a given; in actual fact it is always ordered, and opposes genuine heterology at every instant throughout the life of a language. Yet at the same time, this common language is perfectly real when seen as a force which overcomes this heterology, constrains it within certain limitations, assures a maximum mutual comprehension, and is crystallized in the real, albeit relative, unity of literary and spoken (everyday) language, which is the "proper language."

Bakhtin will refer, as one can see, to this tendency towards unification as a "centripetal force" and by the same token, to heterology as a "centrifugal force." Different types of discourse themselves favor one force over the other for varying reasons. For example, the novel (or what Bakhtin defines as such) reinforces heterology, while poetry does not; for heterology is linked to the representation of language, which is a characteristic feature of the novel.

While the principal sorts of poetic genres develop within the flow of the centripetal unifying and centralizing forces which inform verbal and ideological existence, the novel, as well as other related genres of literary prose, emerged historically within the flow of decentralizing, centrifugal forces.

Therefore, the high periods of the novel correspond to those which witnessed a weakening of centralized power.

The embryonic forms of novelistic prose appear in the heteroglossic and heterological world of the Hellenistic epoch, in imperial Rome, also in the decomposition and decadence of the verbal and ideological centralism of the medieval church. Similarly, the period of fruition of the modern novel is always tied to a general decomposition of verbal and ideological systems, to a process of reinforcement and intensification which opposes linguistic heterology in the literary dialect but also outside it!

On the other hand, as Bakhtin remarks, the different theories or philosophies of language are always born in the wake of unifying movements; this moreover explains their helplessness when confronted by heterology. Thus, for example, the sad fate of stylistics when it tackles the novel: a "Ptolemaic" discipline cannot account for a "Galilean" genre.

Traditional stylistics ignores the kind of combination whereby languages and styles merge in a superior unity; it has no means of approaching the particular social dialogue of languages within a novel. This is why stylistic analysis is not oriented towards the novel seen as a whole but only towards one or the other of its subordinate stylistic aspects. The specialist bypasses the distinctive characteristic of the novelistic genre; he transforms the object of his study, and instead of the novelistic style he in fact analyses something completely different. He transposes an orchestrated symphonic theme in the place of a piano.

Bakhtin enumerates several other examples of such helplessness in the face of heterology:

The poetics of Aristotle, the poetics of Augustine, Medieval religious poetics of the common language of truth, the Cartesian poetics of Neo-Classicism, the abstract grammatical universalism of Leibniz (the idea of universal grammar), the concrete ideologism of Humboldt—whatever may be the distinguishing nuances—these all express the same centripetal forces of sociolinguistic and ideological existence; they all serve the same objective: the centralization and unification of European languages.

The rather surprising name in this roster is Humboldt, a distant source of inspiration for Bakhtin, as we know, and an advocate of linguistic diversity, that of languages as well as that of individuals (language expressing a national spirit, the utterance—an individual one). However, Humboldt forgets a crucial gap between these two: social diversity. Beyond the unicity of Classicism and the Romantic infinite variety, Bakhtin seeks a third path: that of typology.

Tzvetan Todorov, "Bakhtin's Theory of the Utterance," translated by Claudine Frank, in Semiotic Themes, edited by Richard T. De George, University of Kansas Publications, 1981, pp. 165-78.