The work of Roland Barthes is apparently highly varied, both in its object (literature, clothes, cinema, painting, advertising, music, news items, etc.) and in its method and ideology. Le Degré zéro de l'Écriture (1953) seemed to extend into the domain of "form" the reflection begun by Sartre some years earlier on the social situation of literature and the responsibility of the writer before history—a reflection on the frontiers of existentialism and Marxism. His Michelet (1954), though offered as a simple, "precritical" reading, borrowed from Gaston Bachelard the idea of a substantial psychoanalysis and showed what a thematic study of the material imagination could bring to the understanding of a work regarded hitherto as essentially ideological. His work for the review Théâtre populaire and in the struggle waged around that review to introduce the work and theories of Bertolt Brecht into France brought him a reputation, in the next few years, of being an intransigent Marxist, although official Marxists never shared his interpretation of Brecht's theory; but, at the same time, and contradictorily, two articles on Les Gommes and Le Voyeur made him the official interpreter of Robbe-Grillet and the theoretician of the nouveau roman, which was widely regarded as a Formalist offensive and as an attempt to "disengage" literature. In 1956, Mythologies revealed a sarcastic observer of the petty-bourgeois ideology concealed in the most seemingly innocuous manifestations of contemporary social life; a new "critique of everyday life," clearly Marxist in inspiration, which marked an unequivocal political attitude. In 1960, there was a new metamorphosis, a commentary on Racine for the Club français du Livre (revised in 1963 as Sur Racine), which seemed to effect a return to psychoanalysis, but this time closer to Freud than to Bachelard, though to the Freud of Totem and Taboo, an anthropologist in his own way: Racine's tragedies are interpreted in terms of the prohibition of incest and Oedipal conflict, "at the level of this ancient fable (that of the 'primal horde'), situated far beyond history or the human psyche." Lastly, the latest texts collected in Essais critiques (1964) seem to express a decisive conversion to structuralism, understood in its strictest form, and the abandonment of any responsibility towards meaning; literature and social life are now merely languages, which should be studied as pure formal systems, not for their content, but for their structure.

This many-sided image is obviously a superficial and even, as we will see, a highly unfaithful one. Not that the scope of Barthes' reflection is actually circumscribed, open as it is in principle to the most varied tendencies of modern thought. He himself admits that he has often dreamed "of a peaceful coexistence of critical languages or, perhaps, of a 'parametric' criticism which would modify its language to suit the work proposed to it," and, speaking of the fundamental "ideological principles" of contemporary criticism (existentialism, Marxism, psychoanalysis, structuralism), he declares: "For my part, in a certain sense I subscribe to each of them at the same time." But this apparent eclecticism conceals a constant in his thought that was already at work in Le Degré zéro, and which has become ever more marked, more conscious, and more systematic. If criticism can claim allegiance to several ideologies at once, it is, Barthes hastens to add, because "ideological choice does not constitute the being of criticism and because 'truth' is not its sanction": its task is not to uncover the secret truth of the works of which it speaks, but to cover their language "as completely as possible," with its own language, to adjust as closely as possible the language of our period to that of the works of the past, "that is to say, to the formal system of logical constraints worked out by the author in accordance with his own period." This friction between literary language and critical language has the effect not of bringing out the "meaning" of a work, but of "reconstituting the rules and constraints governing the elaboration of this meaning," in other words, its technique of signification. If the work is a language and criticism a metalanguage, their relation is essentially formal, and criticism no longer has to concern itself with a message, but with a code, that is to say, a system the structure of which it is its task to uncover, "just as the linguist is not responsible for deciphering the sentence's meaning but for establishing the formal structure which permits this meaning to be transmitted." In consideration of which, out of the varied languages that criticism can try on the literary works of the past (or of the present) "would appear a general form, which would be the very intelligibility our age gives to things and which critical activity helps, dialectically, both to decipher and to constitute." The exemplary value of critical activity, then, derives clearly from this double semiological character: as a metalanguage (a discourse on literary language), it studies a system from the viewpoint of that metacriticism, or "criticism of criticisms," which is simply semiology in its most general form. Thus criticism helps both "to decipher and to constitute" the intelligible, since it is at the same time semantics and semanteme, subject and object, of the semiological activity.

These remarks lead us then to the central point of Barthes' thought: the problem of signification. Homo significans: man the sign-maker, "man's freedom to make things signify," "the strictly human process by which men give meaning to things," such is the essential object of his research. It is a traditional, even fundamental, orientation, since already Le Degré zéro studied the various ways in which the writer, beyond all the explicit contents of his discourse must in addition—perhaps essentially—signify Literature, and this book was offered as acontribution to "a history of literary expression which is neither that of a particular language, nor that of the various styles, but simply that of the Signs of Literature," that is to say, of the signs by which literature draws attention to itself as literature, and points out its mask. It is an old question, then, but one that has continued to reflect upon itself and to define its terms.

As we know, it was the linguist Ferdinand de Saussure who first conceived of the idea of a general science of significations, of which linguistics would be no more than a particular case, "a science that studies the life of signs within society," which would show "what constitutes signs, what laws govern them," and which he proposed to call semiology. The natural languages (langues) being by far the most elaborate and best-known systems of signs, linguistics necessarily remains the irreplaceable model for all semiological research, but the domain of signs goes beyond that of articulated language. Indeed, there exist on the one hand signs outside language, which function so to speak beside it, such as those emblems or signals of all kinds that men have always used, from "primitive totemism" to the various sign-posts and symbols that modern civilization constantly proliferates before our eyes. Some of these signs have already constituted highly complex systems: one has only to think of the degree of elaboration once attained by the art of the coat-of-arms and its corresponding science of heraldry; the ability to constitute a system is precisely the characteristic of any set of signs, and it is this constitution that marks the passage from pure symbolism to the strictly semiological state, since a symbol becomes a sign only at the moment when it ceases to suggest of itself, and by virtue of an analogical or historical relationship (the Crescent, the emblem of Islam, the Cross, the symbol of Christianity) which it maintains with its "referent," in order to signify in an indirect way, mediatized by the relation of kinship and opposition that it maintains with other concurrent symbols; the Cross and the Crescent, taken in isolation, are two autonomous symbols, but the use of an Arab Red Crescent with a European Red Cross sets up a paradigmatic system in which red holds the place of a common root, and the opposition Cross/Crescent that of a distinctive inflection.

What we have, then, or at least it would seem so, is a series of extralinguistic semiological systems; but their social importance, and still more their autonomy in relation to articulated language appear to be highly questionable: "Until now semiology has had to concern itself only with codes of little interest, such as the highway code; as soon as one passes to systems possessing real social depth, one meets language once again." This is because nonlinguistic objects actually become signifiers only insofar as they are duplicated or relayed by language, as is clear enough in advertising or newspaper photography, which invariably accompany the visual image with a verbal commentary intended to confirm or to localize its virtual or floating significations, or again, in fashion writing, which gives objects (clothes, food, furniture, cars, etc.) their symbolic value by "speaking" of them, that is to say, by analyzing the signifying parts and naming the signifieds: the image might represent a man wearing a tweed jacket, standing in front of a country house, but the commentary will state more precisely "tweed jacket for the weekend," designating by name tweed as a sign and weekend as the meaning. "There is only meaning when it is named, and the world of signifieds is simply that of language." The extralinguistic domain rapidly gives way therefore (or is absorbed into) that other domain of semiology, which is the translinguistic, or metalinguistic order, and which embraces techniques of signification situated not beside, but above, or within, language. Semiology is thus brought back into the linguistic fold, which leads Barthes to reverse the Saussurian formula: semiology is no longer seen as an extension, but on the contrary as a specification of linguistics. However, it is not a question of assimilating the semiological fact to the linguistic fact, for language used in this way concerns semiology only as a secondary language, either because the verbal text is supposed to impose a signification on a nonverbal object, as in the case of the blurbs attached to press photographs or advertising images, or because it duplicates itself as it were in order to add to its own explicit, literal signification, or denotation, an additional power of connotation, which enriches it with one or several secondary meanings. Many pages of literature, as Valéry more or less remarks, mean nothing more than "I am a page of literature," a sentence which, however, is nowhere to be found in them; and Sartre rightly stresses that the meaning or intrinsic quality of a text is never after all directly designated by the words of this text, and that "the literary object, though realized through language, is never given in language."

This oblique language that suggests some unstated meaning is the language of connotation, of which literature is the domain par excellence, the study of which may avail itself of an illustrious, if sometimes decried precedent, that of Rhetoric. When a rhetorician of the classical period taught, for example, that the use of the word "sail" to designate a ship is a figure called synecdoche, and that this figure achieves its finest effect in an epic poem, he simply brought out, in his own way, the epic connotation implied in the use of this figure, and a treatise of rhetoric was a code of literary connotation, a collection of the means by which a poet could signify, over and above the explicit "content" of his poem, its quality of being epic, lyrical, bucolic, etc. Such is the case of those obscenities with which the prose of Père Duchêne is dotted, not to signify anything in the discourse, but to signal, obliquely, a whole historical situation: the precious figures of revolutionary rhetoric.

In fact it is the phenomena and techniques of connotation that, since Le Degré zéro de l'écriture, have particularly commanded Barthes' attention. Writing, we should remember, is that responsibility for Form which, between the Nature represented by the horizon of the language (imposed by time and place) and that other Nature determined by the vertical thrust of style (dictated by the depths of the body and psyche), manifests the writer's choice of a particular literary attitude, and therefore indicates a particular modality of literature; the writer chooses neither his language nor his style, but he is responsible for the methods of writing that indicate whether he is a novelist or a poet, a classicist or a naturalist, bourgeois or populist, etc. All these facts of writing are means of connotation, since over and above their literal meaning, which is sometimes weak or negligible, they manifest an attitude, a choice, an intention.

This effect of super-signification may be represented by a simple schema, for which we will borrow from rhetoric once again its classic example: in the synecdoche sail ship, there is a signifying word, "sail," and an object (or concept) signified, the ship: that is the denotation; but since the word "sail" has been substituted for the literal word "ship," the relation (signification) that links the signifier to the signified constitutes a figure; this figure in turn clearly designates, in the rhetorical code, a poetic state of discourse: it functions then as the signifier of a new signified, poetry, on a second semantic plane, which is that of rhetorical connotation; the essence of connotation is in effect to establish itself above (or below) the primary signification, but in a dislocated way, using the primary meaning as a form to designate a secondary concept; hence the schema (which might be expressed more or less in some such formula as: the semiological system in which the word "sail" may be used to designate a ship is a figure; the secondary semiological figure in which a figure, such as the use of the word "sail" to designate a ship, may be used to signify poetry, is rhetoric). [In an endnote, Genette renders this Relationship as: "[(sail = ship) = poetry] = Rhetoric."]

Readers of Mythologies will recognize a similarity between this schema and the one used by Barthes to represent the dislocation of myths in relation to the semiological system onto which it is grafted. This is because we are dealing with an effect of the same order, and Barthes says quite rightly that Le Degré zéro "was, all told, nothing but a mythology of literary language" in which he defined writing "as the signifier of the literary myth, that is, as a form which is already filled with meaning and which receives from the concept of Literature a new signification." From the point of view that concerns us here, what distinguishes Mythologies from Le Degré zéro is, on the one hand, an explicit recourse to the notion of semiological system and a clear view of the superimposition and dislocation of the two systems and, on the other hand, the application of this analysis to non-literary objects and even, in some cases, non-linguistic objects, such as the photograph of a black soldier saluting a French flag, which adds to this uncoded and purely denoted visual message a second connoted, ideological message, which is the justification of the French empire.

Thus a whole world is opened up to semiological analysis, a much vaster world than that of literature and one that still awaits its rhetoric: the world of communication, of which the press, the cinema, and advertising are the most obvious and best-known forms. But the field of signification does not stop there, for the language of connotation shows that man can endow with an additional meaning any object that has previously been provided either with a primary meaning (verbal statement, graphic or photographic image, film shot or sequence, etc.) or with a non-signifying primary function, which may, for example, be some kind of use: "Food is to be eaten; but it also serves to signify (conditions, circumstances, tastes); food is therefore a signifying system and must one day be described as such." Similarly clothes are intended to be worn, a house to provide shelter, a car to move around, but clothes, houses, and cars are also signs, indications of a condition or personality, instruments or a "showing." Semiology thus becomes coextensive with a whole civilization and the world of objects becomes a universe of signs: "In a single day, how many really non-signifying fields do we cross? Very few, sometimes none." What we call history, or culture, is also that "shudder of an enormous machine which is humanity tirelessly undertaking to create meaning, without which it would no longer be human."

But it has to be realized that this signifying activity is always carried out, for Barthes, as an addition of use imposed on things, and therefore on occasion as a distortion or an abuse. For Barthes, signs are almost never, like ships' flags, roadsigns, or any other of the clarion calls with which semiology has traditionally concerned itself, signifiers deliberately invented for explicit, limited signifieds, in short the elements of a recognized and overt code. The systems that interest him are always, as he says of literary criticism, "semiologies that dare not speak their name," ashamed or unconscious codes, always marked by a certain bad faith. To decide that a red or a green lamp will signify "stop" or "go" is not in the least equivocal: I have created a sign that could not be clearer, I have abused nothing and nobody. To decide that a leather jacket "suggests sportiness" and therefore to turn leather into a sign of "sportiness," is something quite different: for leather exists outside this imposition of meaning, as a substance that one might like for profound reasons having to do with its feel, its consistency, its color, its texture; by turning it into a signifier, I obliterate these substantial qualities and substitute for them a social concept of doubtful authenticity; but, on the other hand, I confiscate to the benefit of this signifying link the perceptible properties of leather, which are always available as a reserve of natural justification: leather is sporty because it is supple, convenient, and so forth; I wear leather because I am sporty: what could be more natural? The semiological link is concealed beneath an apparently causal relation, and the naturalness of the sign exculpates the signified.

It is clear that semiological reflection has shifted here from the level of facts to that of values. There is for Barthes an axiology of the sign, and it is doubtless not excessive to see in this system of preferences and rejections the deeper motive for his activity as a semiologist. Barthesian semiology is, both in its origin and in its active principle, that of a man fascinated by the sign, a fascination that no doubt involves, as it does for Flaubert or Baudelaire, an element of repulsion, and which has the essentially ambiguous character of a passion. Man makes rather too many signs, and these signs are not always very healthy. One of the texts collected in Essais critiques is entitled "The Diseases of Costume." It begins with this characteristic sentence: "I should like to sketch here not a history of an esthetic, but rather a pathology or, if you prefer, an ethic of costume. I shall propose a few very simple rules which may permit us to judge whether costume is good or bad, healthy or sick." The diseases of theatrical costume, which is obviously a sign, are three in number and all three turn out to be hypertrophies: hypertrophy of the historical function, archeological accuracy; hypertrophy of formal beauty, estheticism; hypertrophy of sumptuosity, money. In another text on the theater, Barthes reproaches traditional Racinian diction for its "hypertrophy of detailed significance" (signification parcellaire), a plethora of details that spread over the text like a film of greasy dirt and impair the clarity of the whole; the same criticism is leveled, with more violence, at the performance of a modern actress in the Oresteia: "a dramatic art of the intention, of the gesture and the glance heavy with meaning, of the signified secret, an art suitable for any scene of conjugal discord and bourgeois adultery, but which introduces into tragedy a cunning and, in a word, a vulgarity utterly anachronistic to it." It is an indiscretion comparable to that of the rubato dear to the romantic pianists, and which Barthes finds again in a particular interpretation of a Fauré song: "this pleonasm of intentions muffles both words and music, and chiefly their junction, which is the very object of the vocal art." All these redundant, overfed significations, like Michelet's "lacteous and sanguine" Englishwomen or the apoplectic burgomasters of Dutch painting, arouse a disapproval that is indissolubly of a logical, moral, and esthetic, but perhaps above all physical, order: it is nausea, that "immediate judgment of the body" which Barthes finds so easily in his Michelet, who judges history "at the tribunal of the flesh." The bad sign is bloated because it is redundant, and it is redundant because it wants to be true, that is, both a sign and a thing, like the costume for Chanticleer of 1910, made up of several pounds of real feathers "sewn one over the other." The good sign is arbitrary: it is the common word, the name "tree" or the verb "to run," which has value only through an express convention, and does not try to deceive by adding to this conventional value the oblique power of natural evocation. It is the flag in the Chinese theater, which on its own signifies a whole regiment, the masks and costumes of the Commedia dell'arte, or better still, the red gown of the Caliph in the Thousand and One Nights, which signifies "I am angry." The bad sign par excellence is the meaning-form which serves as signifier to the mythical concept, because it uses the natural character of the meaning surreptitiously in order to justify the secondary signification. The naturalization of culture, and therefore of history, is in Barthes' eyes, as we know, the major sin of petty-bourgeois ideology, and its denunciation the central theme of Mythologies.

Now the semiological instrument of this naturalization is the fraudulent motivation of the sign. When a Racinian actress utters the words "Je brûle" (I burn) in an obviously burning tone, when a singer interprets "tristesse affreuse" (terrible sadness) by terribly saddening the sounds of these two words, they commit a pleonasm and an imposture: they have to choose between the sentence and the cry, "between the intellectual sign and the visceral sign," which latter is really no longer a sign, but a direct manifestation of the signified, an expression, in the full sense of the term; but such effects are practically outside the reach of art, which must be accepted fully as a language. Now "if there is a 'health' of language, it is the arbitrariness of the sign which is its grounding. What is sickening in myth is its resort to a false nature, its superabundance of significant forms, as in those objects which decorate their usefulness with a natural appearance. The will to weigh the signification with the full guarantee of nature causes a kind of nausea: myth is too rich, and what is in excess is precisely its motivation." The health of an art, its virtue, its elegance, lies in its strict fidelity to the system of conventions on which it rests: "The exercise of a signifying system … has only one requirement, which will therefore be the esthetic requirement itself: rigor": this is the case of Brechtian dramaturgy, cleansed by the effect of distancing, which knows that "the responsibility of a dramatic art is not so much to express reality as to signify it"; it is the case with the sober acting of Helene Weigel, the literal performances of a Panzera or a Lipatti, the photographs of Agnès Varda, shot with "exemplary humility," the cathartic writing of Robbe-Grillet, determined to kill the adjective and to restore to the object its "essential thinness."

Barthes does not see the semiological activity, then, as exclusively, or even essentially, belonging to the order of knowledge. For him, signs are never the neutral objects of disinterested knowledge, as Saussure conceived them when he contemplated the founding of a semiological science. The normative choice is never far behind analytical discourse, and this ethical origin that he recognizes in the work of the mythologist is to be found throughout his work. "Brechtian criticism will therefore be written by the spectator, the reader, the consumer, and not the exegete: it is a criticism of a concerned man." This attitude marks all Barthes' critical activity, which is constantly underpinned by the question: in what sense does this work concern us? This criticism is and is always intended to be profoundly and aggressively subjective, because every reading, "however impersonal it forces itself to be, is a projective test" into which the critic "puts all his 'profundity,' i. e., his choices, his pleasures, his resistances, his obsessions." It has nothing to do, we realize, with the intersubjective participation which animates criticism like that of Georges Poulet, and which always operates to the benefit of the "thought criticized," before which critical thought stands back and falls silent, its sole raison d'être being to recreate a space and a language for it. Barthesian criticism is not the resumption of one subject by another, of one speech by another: it is a dialogue, and a dialogue that is "egoistically shifted toward the present." Thus, paradoxically, this notorious representation of the "newest" new criticism is alone in honoring in his work the ancient meaning of the word "criticism," which designates a militant act of assessment and challenge. His literary criticism is certainly a semiology of literature; but his semiology, in turn, is not only a study of significations, but also, in the most vivid sense of the term, a critique of signs.

Noting in the final section of Mythologies the imposture involved in the ambiguity of the mythical sign, "this turnstile of form and meaning," Barthes adds that one can escape this imposture, stop this turnstile, only if one focuses on form and meaning separately, that is, by applying to the mythical object a semiological analysis. Semiology, then, is not only a tool of knowledge and criticism: it is also, for the man besieged by signs, the only possible recourse, the only defense. To analyse the sign, to distinguish between its constitutive elements, to place on one side the signifier, on the other the signified: this activity, which, for Saussure, was a simple technique, a methodological routine, becomes for Barthes something like the instrument of an ascesis and the beginnings of a salvation. The semiological discipline stops the vertigo of meaning and authorizes a liberating choice: for it is the privilege of the semiologist to turn away from the signified in order to devote himself to the study of the signifier, and therefore to an exclusive commerce with it. He has given himself as his "moral goal," as Barthes says of the critic, "not the decipherment of the work's meaning but the reconstruction of the rules and constraints of that meaning's elaboration": thus he avoids "good conscience" and "bad faith." His gaze stops at the frontier of meaning and does not cross it: like the linguist, he is concerned only with forms. But this prejudice in favor of forms is no mere methodological rule, it is an existential choice.

We have to remember that the forms in question are not sentences, words, phonemes—they are objects; and when the semiologist has operated the semiological reduction, the epoché of meaning on the object-form, he is presented with a matte object, cleansed of the varnish of dubious, abusive significations, with which social speech had covered it, restored to its essential freshness and solitude. Thus the formalist enterprise opens up, in an unexpected way, an adhesion to, a very profound conformity with, the reality of things. The paradox and difficulty of such a deviation have not escaped the author of Mythologies, who devotes the last page of that book to them: the mythologist wishes "to protect reality" against the "evaporation" with which it is threatened by the alienating speech of myth, but he fears he has himself contributed to its disappearance. The "goodness of wine" is a French myth, but at the same time wine is good and the mythologist is condemned to speak only of its mythical goodness. This abstention is regrettable, and Barthes recognizes that he has been unable to avoid it altogether: "Finding it painful constantly to work on the evaporation of reality, I have started to make it excessively dense, and to discover in it a surprising compactness which I savored with delight, and I have given a few examples of 'substantial psychoanalysis' about some mythical objects." All critical irony laid aside, he gives himself up for example to praising old wooden toys, the nostalgic associations of which are characteristic: "A sign which fills one with consternation is the gradual disappearance of wood, in spite of its being an ideal material because of its firmness and its softness, and the natural warmth of its touch…. It is a familiar and poetic substance, which does not sever the child from close contact with the tree, the table, the floor…. Wood makes essential objects, objects for all time." Material intimacy, access to the "essence of things" is here, as in Proust, a lost paradise, which he must try to recover, but by some indirect way. For Barthes, semiology plays the role of a catharsis, but this ascesis, which rejects the meaning added by history, is in its own way a return, or an attempted return, to reality. His method is almost the opposite of that of (modern) poetry, that language without writing by which man "confronts the world of objects without going through any of the forms of history or of social life": the semiological procedure seems to consist on the contrary in accepting the deviation as inevitable, in the belief that ideology and its rhetoric overlie the entire surface of reality, that the only way of obviating this is to confront them in order to traverse them, and therefore that the poetic project of an immediate speech is a sort of utopia. But the opposition of means must not conceal the kinship of ends: the semiologist as Barthes understands him is also in search of "the inalienable meaning of things," which he uncovers beneath their alienated meaning. The movement from the (ideological) signified to the (real) signifier is only apparently therefore an abandonment of meaning. It would be better to say that it leads from the ideological meaning, which is an (abusive) speech, to the poetic meaning, which is a silent presence. "Things must taste of what they are," Curnonsky demands. The rediscovery of this profound taste is perhaps the unacknowledged aim of the semiologist.

This explains the privilege accorded, and preserved throughout his work, to Literature. For Barthes, literature uses signs, following Kafka's lesson, not to name a meaning but to "deceive," that is to say, both to offer it and to suspend it. In the literary work, the transitive movement of the verbal message stops and is absorbed into a "pure spectacle." To the proliferation of meaning, literature opposes a resistance that is all the more effective in that its instruments are exclusively of a semantic order, and that all its works are composed of language. Far from turning away from that rather sickening technique which Barthes calls the "cooking up of meaning," literature is wholly and entirely committed to it, but in act, in order to free itself from it, preserving the significations, but diverting them from their signifying function. The literary work tends to turn itself into a monument of reticence and ambiguity, but it constructs this silent object, so to speak, with words, and this work of abolishing meaning is a typically semiological process, liable as such to an analysis of the same order: literature is a rhetoric of silence. [In an endnote Genette observes: "We know that on certain jukeboxes one can get, for the same price as the latest tune, a period of silence equal to that of a record: it may in fact be a blank record specially made for this purpose. But whatever the means, the lesson of this invention is clear, namely, that in a civilization of noise silence must also be a product, that it is the fruit of a technology and a commercial object. There is no question of stopping the racket, which ought to be muffled as quickly as possible, but on occasion one is able, by paying the price, to get it to run silently. Likewise, in a civilization of meaning, though there is no longer any place for 'truly insignificant objects,' it is still possible to produce objects loaded with significations, but conceived in such a way that these significations cancel each other out, disperse, or are reabsorbed, like mechanical functions in Ashby's homeostat. No one is really able (or permitted) to be entirely silent, but the writer has the special, indispensable, indeed sacred function, of speaking 'in order to say nothing,' or to say 'What?'."] Its art consists entirely in making language, a vehicle of knowledge and rather hasty opinion, a locus of uncertainty and interrogation. It suggests that the world signifies, but "without saying what": it describes objects and people, relates events, and instead of imposing on them definite, fixed significations, as does social speech (and also, of course, "bad" literature), it leaves them, or rather restores to them, by a very subtle technique (which is still to be studied) of semantic evasion, that "shaky," ambiguous, uncertain meaning, which is their truth. Thus it breathes new life into the world, freeing it from the pressure of social meaning, which is a named meaning, and therefore a dead meaning, maintaining as long as possible that opening, that uncertainty of signs, which allows one to breathe. Thus literature is for the semiologist (the critic) a permanent temptation, an endless vocation postponed until later, experienced only in this dilatory mode: like the Proustian Narrator, the semiologist is a "writer postponed"; he constantly intends to write, that is to say, to turn over the meaning of signs and to send language back to the silence that forms part of it; but the postponement is only apparent, for this intention to write, this "Moseslike gaze" on the work to come is already Literature.

Gérard Genette, "The Obverse of Signs," in his Figures of Literary Discourse, translated by Alan Sheridan, Columbia University Press, 1982, pp. 27-44.