Paul de Man

The Resistance to Theory

   This essay was not originally intended to address the question of teaching directly, although it was supposed to have a didactic and an educational function - which it failed to achieve. It was written at the request of the Committee on Research Activities of the Modern Language Association as a contribution to a collective volume entitled Introduction to Scholarship in Modern Languages and Litera-tures. I was asked to write the section on literary theory. Such essays are expected to follow a clearly determined program: they are supposed to provide the reader with a select but comprehensive list of the main trends and publications in the field, to. synthesize and classify the main problematic areas and to lay out a critical and programmatic projection of the solutions which can be expected in the foreseeable future. All this with a keen awareness that, ten years later, someone will be asked to repeat the same exercise.

   I found it difficult to live up, in minimal good faith, to the requirements of this program and could only try to explain, as concisely as possible, why the main theoretical interest of literary theory consists in the impossibility of its definition. The Committee rightly judged that this was an inauspicious way to achieve the pedagogical objectives of the volume and commissioned another article. I thought their decision altogether justified, as well as interesting in its implications for the teaching of literature.

   I tell this for two reasons. First, to explain the traces in the article of the original assignment which account for the awkwardness of trying to be more retrospective and more general than one can legitimately hope to be. But, second, because the predicament also reveals a question of general interest: that of the relationship between the scholarship (the key word in the title of the MLA volume), the theory, and the teaching of literature.

   Overfacile opinion notwithstanding, teaching is not primarily an intersubjec-tive relationship between people but a cognitive process in which self and other are only tangentially and contiguously involved. The only teaching worthy of the name is scholarly, not personal; analogies between teaching and various aspects of show business or guidance counseling are more often than not excuses for having abdicated the task. Scholarship has, in principle, to be eminently teachable. In the case of literature, such scholarship involves at least two com-plementary areas: historical and philological facts as the preparatory condition for understanding, and methods of reading or interpretation. The latter is admit-tedly an open discipline, which can, however, hope to evolve by rational means, despite internal crises, controversies and polemics. As a controlled reflection on the formation of method, theory rightly proves to be entirely compatible with teaching, and one can think of numerous important theoreticians who are or were also prominent scholars. A question arises only if a tension develops between methods of understanding and the knowledge which those methods allow one to reach. If there is indeed something about literature, as such, which allows for a discrepancy between truth and method, between Wahrheit and Methode, then scholarship and theory are no longer necessarily compatible; as a first casualty of this complication, the notion of "literature as such" as well as the clear distinction between history and interpretation can no longer be taken for granted. For a method that cannot be made to suit the "truth" of its object can only teach delusion. Various developments, not only in the contemporary scene but in the long and complicated history of literary and linguistic instruction, reveal symptoms that suggest that such a difficulty is an inherent focus of the discourse about literature. These uncertainties are manifest in the hostility directed at theory in the name of ethical and aesthetic values, as well as in the recuperative attempts of theoreticians to reassert their own subservience to these values. The most effective of these attacks will denounce theory as an obstacle to scholarship and, consequently, to teaching. It is worth examining whether, and why, this is the case. For if this is indeed so, then it is better to fail in teaching what should not be taught than to succeed in teaching what is not true.

   A general statement about literary theory should not, in theory, start from pragmatic considerations. It should address such questions as the definition of literature (what is literature?) and discuss the distinction between literary and non-literary uses of language, as well as between literary and non-verbal forms of art. It should then proceed to the descriptive taxonomy of the various aspects and species of the literary genus and to the normative rules that are bound to follow from such a classification. Or, if one rejects a scholastic for a phenomenological model, one should attempt a phenomenology of the literary activity as writing, reading or both, or of the literary work as the product, the correlate of such an activity. Whatever the approach taken (and several other theoretically justifiable starting-points can be imagined) it is certain that consid-erable difficulties will arise at once, difficulties that cut so deep that even the most elementary task of scholarship, the delimitation of the corpus and the etat present of the question, is bound to end in confusion, not necessarily because the bibliography is so large but because it is impossible to fix its borderlines. Such predictable difficulties have not prevented many writers on literature from proceeding along theoretical rather than pragmatic lines, often with considerable success. It can be shown however that, in all cases, this success depends on the power of a system (philosophical, religious or ideological) that may well remain implicit but that determines an a priori conception of what is "literary" by starting out from the premises of the system rather than from the literary thing itself - if such a "thing" indeed exists. This last qualification is of course a real question which in fact accounts for the predictability of the difficulties just alluded to: if the condition of existence of an entity is itself particularly critical, then the theory of this entity is bound to fall back into the pragmatic. The difficult and inconclusive history of literary theory indicates that this is indeed the case for literature in an even more manifest manner than for other verbalized occur-rences such as jokes, for example, or even dreams. The attempt to treat literature theoretically may as well resign itself to the fact that it has to start out from empirical considerations.

   Pragmatically speaking, then, we know that there has been, over the last fifteen to twenty years, a strong interest in something called literary theory and that, in the United States, this interest has at times coincided with the importation and reception of foreign, mostly but not always continental, influences. We also know that this wave of interest now seems to be receding as some satiation or disappointment sets in after the initial enthusiasm. Such an ebb and flow is natural enough, but it remains interesting, in this case, because it makes the depth of the resistance to literary theory so manifest. It is a recurrent strategy of any anxiety to defuse what it considers threatening by magnification or minimi-zation, by attributing to it claims to power of which it is bound to fall short. If a cat is called a tiger it can easily be dismissed as a paper tiger; the question remains however why one was so scared of the cat in the first place. The same tactic works in reverse: calling the cat a mouse and then deriding it for its pretense to be mighty. Rather than being drawn into this polemical whirlpool, it might be better to try to call the cat a cat and to document, however briefly, the contemporary version of the resistance to theory in this country.

   The predominant trends in North American literary criticism, before the nine-teen sixties, were certainly not averse to theory, if by theory one understands the rooting of literary exegesis and of critical evaluation in a system of some conceptual generality. Even the most intuitive, empirical and theoretically low- key writers on literature made use of a minimal set of concepts (tone, organic form, allusion, tradition, historical situation, etc.) of at least some general import. In several other cases, the interest in theory was publicly asserted and practiced. A broadly shared methodology, more or less overtly proclaimed, links together such influential text books of the era as Understanding Poetry (Brooks and War-ren), Theory of Literature (Wellek and Warren) and The Fields of Light (Reuben Brower) or such theoretically oriented works as The Mirror and the Lamp, Language as Gesture and The Verbal Icon.

   Yet, with the possible exception of Kenneth Burke and, in some respects, Northrop Frye, none of these authors would have considered themselves theoreti-cians in the post-1960 sense of the term, nor did their work provoke as strong reactions, positive or negative, as that of later theoreticians. There were polemics, no doubt, and differences in approach that cover a wide spectrum of divergencies, yet the fundamental curriculum of literary studies as well as the talent and train-ing expected for them were not being seriously challenged. New Critical approaches experienced no difficulty fitting into the academic establishments without their practitioners having to betray their literary sensibilities in any way; several of its represenatives pursued successful parallel careers as poets or novelists next to their academic functions. Nor did they experience difficulties with regard to a national tradition which, though certainly less tyrannical than its European counterparts, is nevertheless far from powerless. The perfect embod-iment of the New Criticism remains, in many respects, the personality and the ideology of T. S, Eliot, a combination of original talent, traditional learning, verbal wit and moral earnestness, an Anglo-American blend of intellectual gen-tility not so repressed as not to afford tantalizing glimpses of darker psychic and political depths, but without breaking the surface of an ambivalent decorum that has its own complacencies and seductions. The normative principles of such a literary ambiance are cultural and ideological rather than theoretical, oriented towards the integrity of a social and historical self rather than towards the impersonal consistency that theory requires. Culture allows for, indeed advocates, a degree of cosmopolitanism, and the literary spirit of the American Academy of the fifties was anything but provincial. It had no difficulty appreciating and assimilating outstanding products of a kindred spirit that originated in Europe: Curtius; Auerbach, Croce, Spitzer, Alonso, Valery and also, with the exception of some of his works, J. P, Sartre. The inclusion of Sartre in this list is important, for it indicates that the dominant cultural code we are trying to evoke cannot simply be assimilated to a political polarity of the left and the right, of the academic and non-academic, of Greenwich Village and Gambier, Ohio. Politi-cally oriented and predominantly non-academic journals, of which the Partisan Review of the fifties remains the best example, did not (after due allowance is made for all proper reservations and distinctions) stand in any genuine opposition to the New Critical approaches. The broad, though negative, consensus that brings these extremely diverse trends and individuals together is their shared resistance to theory. This diagnosis is borne out by the arguments and complicities that have since come to light in a more articulate opposition to the common opponent.

   The interest of these considerations would be at most anecdotal (the historical impact of twentieth-century literary discussion being so slight) if it were not for the theoretical implications of the resistance to theory. The local manifestations of this resistance are themselves systematic enough to warrant one's interest.

   What is it that is being threatened by the approaches to literature that developed during the sixties and that now, under a variety of designations, make up the ill-defined and somewhat chaotic field of literary theory? These approaches cannot be simply equated with any particular method or country. Structuralism was not the only trend to dominate the stage, not even in France, and structuralism as well as semiology are inseparable from prior tendencies in the Slavic domain. In Germany, the main impulses have come from other directions, from the Frankfurt school and more orthodox Marxists, from post-Husserlian phenomenol-ogy and post-Heideggerian hermeneutics, with only minor inroads made by structural analysis. All these trends have had their share of influence in the United States, in more or less productive combinations with nationally rooted concerns. Only a nationally or personally competitive view of history would wish to hierarchize such hard-to-label movements. The possibility of doing liter-ary theory, which is by no means to be taken for granted, has itself become a consciously reflected-upon question and those who have progressed furthest in this question are the most controversial but also the best sources of information. This certainly includes several of the names loosely connected with structuralism, broadly enough defined to include Saussure, Jakobson and Barthes as well as Greimas and Althusser, that is to say, so broadly defined as to be no longer of use as a meaningful historical term.

   Literary theory can be said to come into being when the approach to literary texts is no longer based on non-linguistic, that is to say historical and aesthetic, considerations or, to put it somewhat less crudely, when the object of discussion is no longer the meaning or the value but the modalities of production and of reception of meaning and of value prior to their establishment- the implic4tion being that this establishment is problematic enough to require an autonomous discipline of critical investigation to consider its possibility and its status. Literary history, even when considered at the furthest remove from the platitudes of positivistic historicism, is still the history of an understanding of which the possibility is taken for granted. The question of the relationship between aesthetics and meaning is more complex, since aesthetics apparently has to do with the effect of meaning rather than with its content per se. But aesthetics is in fact, ever since its development just before and with Kant, a phenomenalism of a process of meaning and understanding, and it may be naive in that it postulates (as its name indicates) a phenomenology of art and of literature which may well be what is at issue. Aesthetics is part of a universal system of philosophy rather than a specific theory. In the nineteenth-century philosophical tradition, Nietzsche's challenge of the system erected by Kant, Hegel and their successors is a version of the general question of philosophy. Nietzsche's critique of metaphysics includes, or starts out from, the aesthetic, and the same could be argued for Heidegger. The invocation of prestigious philosophical names does not intimate that the present-day development of literary theory is a by-product of larger philosophical speculations. In some rare cases, a direct link may exist between philosophy and literary theory. More frequently, however, contemporary literary theory is a relatively autonomous version of questions that also surface, in a different context, in philosophy, though not necessarily in a clearer and more rigorous form. Philosophy, in England as well as on the Continent, is less freed from traditional patterns than it sometimes pretends to believe and the prominent, though never dominant, place of aesthetics among the main compo-nents of the system is a constitutive part of this system. It is therefore not surprising that contemporary literary theory came into being from outside philosophy and sometimes in conscious rebellion against the weight of its trad-ition. Literary theory may now well have become a legitimate concern of philosophy but it cannot be assimilated to it, either factually or theoretically. It contains a necessarily pragmatic moment that certainly weakens it as theory but that adds a subversive element of unpredictability and makes it something of a wild card in the serious game of the theoretical disciplines.

   The advent of theory, the break that is now so often being deplored and that sets it aside from literary history and from literary criticism, occurs with the introduction of linguistic terminology in the metalanguage about literature. By linguistic terminology is meant a terminology that designates reference prior to designating the referent and takes into account, in the consideration of the world, the referential function of language or, to be somewhat more specific, that considers reference as a function of language and not necessarily as an intuition. Intuition implies perception, consciousness, experience, and leads at once into the world of logic and of understanding with all its correlatives, among which aesthetics occupies a prominent place. The assumption that there can be a science of language which is not necessarily a logic leads to the development of a terminology which is not necessarily aesthetic. Contemporary literary theory comes into its own in such events as the application of Saussurian linguistics to literary texts.

   The affinity between structural linguistics and literary texts is not as obvious as, with the hindsight of history, it now may seem. Peirce, Saussure, Sapir and Bloomfield were not originally concerned with literature at all but with the scientific foundations of linguistics. But the interest of philologists such as Roman Jakobson or literary critics such as Roland Barthes in semiology reveals the natural attraction of literature to a theory of linguistic signs. By considering language as a system of signs and of signification rather than as an established pattern of meanings, one displaces or even suspends the traditional barriers between literary and presumably non-literary uses of language and liberates the corpus from the secular weight of textual canonization. The results of the en-counter between semiology and literature went considerably further than those of many other theoretical models-philological, psychological or classically epistemological - which writers on literature in quest of such models had tried out before. The responsiveness of literary texts to semiotic analysis is visible in that, whereas other approaches were unable to reach beyond observations that could be paraphrased or translated in terms of common knowledge, these analyses revealed patterns that could only be described in terms of their own, specifically linguistic, aspects. The linguistics of semiology and of literature apparently have something in common that only their shared perspective can detect and that pertains distinctively to them. The definition of this something, often referred to as literariness, has become the object of literary theory. Literariness, however, is often misunderstood in a way that has provoked much of the confusion which dominates today's polemics. It is frequently as-sumed, for instance, that literariness is another word for, or another mode of, aesthetic response. The use, in conjunction with literariness, of such terms as style and stylistics, form or even "poetry" (as in "the poetry of grammar"), all of which carry strong aesthetic connotations, helps to foster this confusion, even among those who first put the term in circulation. Roland Barthes, for example, in an essay properly and revealingly dedicated to Roman Jakobson, speaks eloquently of the writer's quest for a perfect coincidence of the phonic properties of a word with its signifying function. "We would also wish to insist on the Cratylism of the name (and of the sign) in Proust. ... Proust sees the relationship between signifier and signified as motivated, the one copying the other and representing in its material form the signified essence of the thing (and not the thing itself). . . . This realism (in the scholastic sense of the word), which conceives of names as the `copy' of the ideas, has taken, in Proust, a radical form. But one may well ask whether it is not more or less consciously present in all writing and whether it is possible to be a writer without some sort of belief in the natural relationship between names and essences. The poetic function, in the widest sense of the word, would thus be defined by a Cratylian awareness of the sign, and the writer would be the conveyor of this secular myth which wants language to imitate the idea and which, contrary to the teachings of linguistic science, thinks of signs as motivated signs."' To the extent that Cratylism assumes a convergence of the phenomenal aspects of language, as sound, with its signifying function as referent, it is an aesthetically oriented con-ception; one could, in fact, without distortion, consider aesthetic theory, including its most systematic formulation in Hegel, as the complete unfolding of the model of which the Cratylian conception of language is a version. Hegel's somewhat cryptic reference to Plato, in the Aesthetics, may well be interpreted in this sense. Barthes and Jakobson often seem to invite a purely aesthetic reading, yet there is a part of their statement that moves in the opposite direction. For the con-vergence of sound and meaning celebrated by Barthes in Proust and, as Gerard Genette has decisively shown, 2 later dismantled by Proust himself as a seductive temptation to mystified minds, is also considered here to be a mere effect which language can perfectly well achieve, but which bears no substantial relationship, by analogy or by ontologically grounded imitation, to anything beyond that particular effect. It is a rhetorical rather than an aesthetic function of language, an identifiable trope (paronomasis) that operates on the level of the signifier and contains no responsible pronouncement on the nature of the world - despite its powerful potential to create the opposite illusion. The phenomenality of the signifier, as sound, is unquestionably involved in the correspondence between the name and the thing named, but the link, the relationship between word and thing, is not phenomenal but conventional.

   This gives the language considerable freedom from referential restraint, but it makes it epistemologically highly suspect and volatile, since its use can no longer be said to be determined by considerations of truth and falsehood, good and evil, beauty and ugliness, or pleasure and pain. Whenever this autonomous potential of language can be revealed by analysis, we are dealing with literariness and, in fact, with literature as the place where this negative knowledge about the reliability of linguistic utterance is made available. The ensuing foregrounding of material, phenomenal aspects of the signifier creates a strong illusion of aesthetic seduction at the very moment when the actual aesthetic function has been, at the very least, suspended. It is inevitable that semiology or similarly oriented methods be considered formalistic, in the sense of being aesthetically rather than semantically valorized, but the inevitability of such an interpretation does not make it less aberrant. Literature involves the voiding, rather than the affirmation, of aesthetic categories. One of the consequences of this is that, whereas we have traditionally been accustomed to reading literature by analogy with the plastic arts and with music, we now have to recognize the necessity of a non-perceptual, linguistic moment in painting and music, and learn to read pictures rather than to imagine meaning.

   If literariness is not an aesthetic quality, it is also not primarily mimetic. Mimesis becomes one trope among others, language choosing to imitate a non-verbal entity just as paronomasis "imitates" a sound without any claim to identity (or reflection on difference) between the verbal and non-verbal elements. The most misleading representation of literariness, and also the most recurrent objec-tion to contemporary literary theory, considers it as pure verbalism, as a denial of the reality principle in the name of absolute fictions, and for reasons that are said to be ethically and politically shameful. The attack reflects the anxiety of the aggressors rather than the guilt of the accused. By allowing for the necessity of a non-phenomenal linguistics, one frees the discourse on literature from naive oppositions between fiction and reality, which are themselves an offspring of an uncritically mimetic conception of art. In a genuine semiology as well as in other linguistically oriented theories, the referential function of language is not being denied - far from it; what is in question is its authority as a model for natural or phenomenal cognition. Literature is fiction not because it somehow refuses to acknowledge "reality," but because it is not a priori certain that language functions according to principles which are those, or which are like those, of the phenomenal world. It is therefore not a priori certain that literature is a reliable source of information about anything but its own language.

   It would be unfortunate, for example, to confuse the materiality of the signifier with the materiality of what it signifies. This may seem obvious enough on the level of light and sound, but it is less so with regard to the more general phenomenality of space, time or especially of the self; no one in his right mind will try to grow grapes by the luminosity of the word "day," but it is very difficult not to conceive the pattern of one's past and future existence as in accordance with temporal and spatial schemes that belong to fictional narratives and not to the world. This does not mean that fictional narratives are not part of the world and of reality; their impact upon the world may well be all too strong for comfort. What we call ideology is precisely the confusion of linguistic with natural reality, of reference with phenomenalism. It follows that, more than any other mode of inquiry, including economics, the linguistics of literariness is a powerful and indispensable tool in the unmasking of ideological aberrations, as well as a determining factor in accounting for their occurrence. Those who reproach literary theory for being oblivious to social and historical (that is to say ideological) reality are merely stating their fear at having their own ideological mystifications exposed by the tool they are trying to discredit. They are, in short, very poor readers of Marx's German Ideology.

   In these all too summary evocations of arguments that have been much more extensively and convincingly made by others, we begin to perceive some of the answers to the initial question: what is it about literary theory that is so threatening that it provokes such strong resistances and attacks? It upsets rooted ideologies by revealing the mechanics of their workings; it goes against a powerful philosophical tradition of which aesthetics is a prominent part; it upsets the established canon of literary works and blurs the borderlines between literary and non-literary discourse. By implication, it may also reveal the links between ideologies and philosophy. All this is ample enough reason for suspicion, but not a satisfying answer to the question. For it makes the tension between contem-porary literary theory and the tradition of literary studies appear as a mere historical conflict between two modes of thought that happen to hold the stage at the same time. If the conflict is merely historical, in the literal sense, it is of limited theoretical interest, a passing squall in the intellectual weather of the world. As a matter of fact, the arguments in favor of the legitimacy of literary theory are so compelling that it seems useless to concern oneself with the conflict at all. Certainly, none of the objections to theory, presented again and again, always misinformed or based on crude misunderstandings of such terms as mimesis, fiction, reality, ideology, reference and, for that matter, relevance, can be said to be of genuine rhetorical interest.

   It may well be, however, that the development of literary theory is itself overdetermined by complications inherent in its very project and unsettling with regard to its status as a scientific discipline. Resistance may be a built-in con-stituent of its discourse, in a manner that would be inconceivable in the natural sciences and unmentionable in the social sciences. It may well be, in other words, that the polemical opposition, the systematic non-understanding and mis-representation, the unsubstantial but eternally recurrent objections, are the dis-placed symptoms of a resistance inherent in the theoretical enterprise itself. To claim that this would be sufficient reason not to envisage doing literary theory would be like rejecting anatomy because it has failed to cure mortality. The real debate of literary theory is not with its polemical opponents but rather with its own methodological assumptions and possibilities. Rather than asking why liter-ary theory is threatening, we should perhaps ask why it has such difficulty going about its business and why it lapses so readily either into the language of self-jus-tification and self-defense or else into the overcompensation of a programmati-cally euphoric utopianism. Such insecurity about its own project calls for self-analysis, if one is to understand the frustrations that attend upon its practitioners, even when they seem to dwell in serene methodological self-assurance. And if these difficulties are indeed an integral part of the problem, then they will have to be, to some extent, a-historical in the temporal sense of the term. The way in which they are encountered on the present local literary scene as a resistance to the introduction of linguistic terminology in aesthetic and historical discourse about literature is only one particular version of a question that cannot be reduced to a specific historical situation and called modern, post-modem, post-classical or romantic (not even in Hegel's sense of the term), although its compulsive way of forcing itself upon us in the guise of a system of historical periodization is certainly part of its problematic nature. Such difficulties can be read in the text of literary theory at all times, at whatever historical moment one wishes to select. One of the main achievements of the present theoretical trends is to have restored some awareness of this fact. Classical, medieval and Renaissance literary theory is now often being read in a way that knows enough about what it is doing not to wish to call itself "modern."

   We return, then, to the original question in an attempt to broaden the discussion enough to inscribe the polemics inside the question rather than having them determine it. The resistance to theory is a resistance to the use of language about language. It is therefore a resistance to language itself or to the possibility that language contains factors or functions that cannot be reduced to intuition. But we seem to assume all too readily that, when we refer to something called "language," we know what it is we are talking about, although there is probably no word to be found in the language that is as overdetermined, self-evasive, disfigured and disfiguring as "language." Even if we choose to consider it at a safe remove from any theoretical model, in the pragmatic history of "language," not as a concept, but as a didactic assignment that no human being can bypass, we soon find ourselves confronted by theoretical enigmas. The most familiar and general of all linguistic models, the classical trivium, which considers the sciences of language as consisting of grammar, rhetoric, and logic (or dialectics), is in fact a set of unresolved tensions powerful enough to have generated an infinitely prolonged discourse of endless frustration of which contemporary liter-ary theory, even at its most self-assured, is one more chapter. The difficulties extend to the internal articulations between the constituent parts as well as the articulation of the field of language with the knowledge of the world in general, the link between the trivium and the quadrivium, which covers the non-verbal sciences of number (arithmetic), of space (geometry), of motion (astronomy), and of time (music). In the history of philosophy, this link is traditionally, as well as substantially, accomplished by way of logic, the area where the rigor of the linguistic discourse about itself matches up with the rigor of the mathematical discourse about the world. Seventeenth-century epistemology, for instance, at the moment when the relationship betwen philosophy and mathematics is particu-larly close, holds up the language of what it calls geometry (mos geometricus), and which in fact includes the homogeneous concatenation between space, time and number, as the sole model of coherence and economy. Reasoning more geometrico is said to be "almost the only mode of reasoning that is infallible, because it is the only one to adhere to the true method, whereas all other ones are by natural necessity in a degree of confusion of which only geometrical minds can be aware." 3 This is a clear instance of the interconnection between a science of the phenomenal world and a science of language conceived as definitional logic, the pre-condition for a correct axiomatic-deductive, synthetic reasoning. The possibility of thus circulating freely between logic and mathema-tics has its own complex and problematic history as well as its contemporary equivalences with a different logic and a different mathematics. What matters for our present argument is that this articulation of the sciences of language with the mathematical sciences represents a particularly compelling version of a con-tinuity between a theory of language, as logic, and the knowledge of the pheno-menal world to which mathematics gives access. In such a system, the place of aesthetics is preordained and by no means alien, provided the priority of logic, in the model of the trivium, is not being questioned. For even if one assumes, for the sake of argument and against a great deal of historical evidence, that the link between logic and the natural sciences is secure, this leaves open the question, within the confines of the trivium itself, of the relationship between grammar, rhetoric and logic. And this is the point at which literariness, the use of language that foregrounds the rhetorical over the grammatical and the logical function, intervenes as a decisive but unsettling element which, in a variety of modes and aspects, disrupts the inner balance of the model and, consequently, its outward extension to the nonverbal world as well.

   Logic and grammar seem to have a natural enough affinity for each other and, in the tradition of Cartesian linguistics, the grammarians of Port-Royal experienced little difficulty at being logicians as well. The same claim persists today in very different methods and terminologies that nevertheless maintain the same orientation toward the universality that logic shares with science. Replying to those who oppose the singularity of specific texts to the scientific generality of the semiotic project, A. J. Greimas disputes the right to use the dignity of "grammar" to describe a reading that would not be committed to universality. Those who have doubts about the semiotic method, he writes, "postulate the necessity of constructing a grammar for each particular text. But the essence (le propre) of a grammar is its ability to account for a large number of texts, and the metaphorical use of the term . . . fails to hide the fact that one has, in fact, given up on the semiotic project. "4 There is no doubt that what is here prudently called "a large number" implies the hope at least of a future model that would in fact be applicable to the generation of all texts. Again, it is not our present purpose to discuss the validity of this methodological optimism, but merely to offer it as an instance of the persistent symbiosis between grammar and logic. It is clear that, for Greimas as for the entire tradition to which he belongs, the grammatical and the logical functions of language are co-extensive. Grammar is an isotope of logic.

   It follows that, as long as it remains grounded in grammar, any theory of language, including a literary one, does not threaten what we hold to be the underlying principle of all cognitive and aesthetic linguistic systems. Grammar stands in the service of logic which, in turn, allows for the passage to the knowledge of the world. The study of grammar, the first of the artes liberales, is the necessary pre-condition for scientific and humanistic knowledge. As long as it leaves this principle intact, there is nothing threatening about literary theory. The continuity between theory and phenomenalism is asserted and preserved by the system itself. Difficulties occur only when it is no longer possible to ignore the epistemological thrust of the rhetorical dimension of discourse, that is, when it is no longer possible to keep it in its place as a mere adjunct, a mere ornament within the semantic function.

   The uncertain relationship between grammar and rhetoric (as opposed to that between grammar and logic) is apparent, in the history of the trivium, in the uncertain status of figures of speech or tropes, a component of language that straddles the disputed borderlines between the two areas. Tropes used to be part of the study of grammar but were also considered to be the semantic agent of the specific function (or effect) that rhetoric performs as persuasion as well as meaning. Tropes, unlike grammar, pertain primordially to language. They are text-producing functions that are not necessarily patterned on a non-verbal entity, whereas grammar is by definition capable of extra-linguistic generalization. The latent tension between rhetoric and grammar precipitates out in the problem of reading, the process that necessarily partakes of both. It turns out that the resis-tance to theory is in fact a resistance to reading, a resistance that is perhaps at its most effective, in contemporary studies, in the methodologies that call them-selves theories of reading but nevertheless avoid the function they claim as their object.

   What is meant when we assert that the study of literary texts is necessarily dependent on an act of reading, or when we claim that this act is being systemat-ically avoided? Certainly more than the tautology that one has to have read at least some parts, however small, of a text (or read some part, however small, of a text about this text) in order to be able to make a statement about it. Common as it may be, criticism by hearsay is only rarely held up as exemplary. To stress the by no means self-evident necessity of reading implies at least two things. First of all, it implies that literature is not a transparent message in which it can be taken for granted that the distinction between the message and the means of communication is clearly established. Second, and more problematically, it im-plies that the grammatical decoding of a text leaves a residue of indetermination that has to be, but cannot be, resolved by grammatical means, however extensively conceived. The extension of grammar to include para-figural dimensions is in fact the most remarkable and debatable strategy of contemporary semiology, especially in the study of syntagmatic and narrative structures. The codification of contextual elements well beyond the syntactical limits of the sentence leads to the systematic study of metaphrastic dimensions and has considerably refined and expanded the knowledge of textual codes. It is equally clear, however, that this extension is always strategically directed towards the replacement of rhetor-ical figures by grammatical codes. This tendency to replace a rhetorical by a grammatical terminology (to speak of hypotaxis, for instance, to designate anamorphic or metonymic tropes) is part of an explicit program, a program that is entirely admirable in its intent since it tends towards the mastering and the clarification of meaning. The replacement of a hermeneutic by a semiotic model, of interpretation by decoding, would represent, in view of the baffling historical instability of textual meanings (including, of course, those of canonical texts), a considerable progress. Much of the hesitation associated with "reading" could thus be dispelled.

   The argument can be made, however, that no grammatical decoding, however refined, could claim to reach the determining figural dimensions of a text. There are elements in all texts that are by no means ungrammatical, but whose semantic function is not grammatically definable, neither in themselves nor in context. Do we have to interpret the genitive in the title of Keats' unfinished epic The Fall of Hyperion as meaning "Hyperion's Fall," the case story of the defeat of an older by a newer power, the very recognizable story from which Keats indeed started out but from which he increasingly strayed away, or as "Hyperion Fall-ing," the much less specific but more disquieting evocation of an actual process of falling, regardless of its beginning, its end or the identity of the entity to whom it befalls to be falling? This story is indeed told in the later fragment entitled The Fall of Hyperion, but it is told about a character who resembles Apollo rather than Hyperion, the same Apollo who, in the first version (called Hyperion), should definitely be triumphantly standing rather than falling if Keats had not been compelled to interrupt, for no apparent reason, the story of Apollo's triumph. Does the title tell us that Hyperion is fallen and that Apollo stands, or does it tell us that Hyperion and Apollo (and Keats, whom it is hard to distinguish, at times, from Apollo) are interchangeable in that all of them are necessarily and constantly falling? Both readings are grammatically correct, but it is impos-sible to decide from the context (the ensuing narrative) which version is the right one. The narrative context suits neither and both at the same time, and one is tempted to suggest that the fact that Keats was unable to complete either version manifests the impossibility, for him as for us, of reading his own title. One could then read the word "Hyperion" in the title The Fall of Hyperion figurally, or, if one wishes, intertextually, as referring not to the historical or mythological character but as referring to the title of Keats' own earlier text (Hyperion). But are we then telling the story of the failure of the first text as the success of the second, the Fall of Hyperion as the Triumph of The Fall of Hyperion? Manifestly, yes, but not quite, since the second text also fails to be concluded. Or are we telling the story of why all texts, as texts, can always be said to be falling Manifestly yes, but not quite, either, since the story of the fall of the first version, as told in the second, applies to the first version only and could not legitimately be read as meaning also the fall of The Fall of Hyperion. The undecidability involves the figural or literal status of the proper name Hyperion as well as of the verb falling, and is thus a matter of figuration and not of grammar. In "Hyperion's Fall," the word "fall" is plainly figural, the representation of a figural fall, and we, as readers, read this fall standing up. But in "Hyperion Falling," this is not so clearly the case, for if Hyperion can be Apollo and Apollo can be Keats, then he can also be us and his figural (or symbolic) fall becomes his and our literal falling as well. The difference between the two readings is itself structured as a trope. And it matters a great deal how we read the title, as an exercise not only in semantics, but in what the text actually does to us. Faced with the ineluctable necessity to come to a decision, no grammatical or logical analysis can help us out. Just as Keats had to break off his narrative, the reader has to break off his understanding at the very moment when he is most directly engaged and summoned by the text. One could hardly expect to find solace in this "fearful symmetry" between the author's and reader's plight since, at this point, the symmetry is no longer a formal but an actual trap, and the question no longer "merely" theoretical.

   This undoing of theory, this disturbance of the stable cognitive field that extends from grammar to logic to a general science of man and of the phenomenal world, can in its turn be made into a theoretical project of rhetorical analysis that will reveal the inadequacy of grammatical models of non-reading. Rhetoric, by its actively negative relationship to grammar and to logic, certainly undoes the claims of the trivium (and by extension, of language) to be an epistemolog-ically stable construct. The resistance to theory is a resistance to the rhetorical or tropological dimension of language, a dimension which is perhaps more explicitly in the foreground in literature (broadly conceived) than in other verbal manifestations or - to be somewhat less vague - which can be revealed in any verbal event when it is read textually. Since grammar as well as figuration is an integral part of reading, it follows that reading will be a negative process in which the grammatical cognition is undone, at all times, by its rhetorical displace-ment. The model of the trivium contains within itself the pseudo-dialectic of its own undoing and its history tells the story of this dialectic.

   This conclusion allows for a somewhat more systematic description of the contemporary theoretical scene. This scene is dominated by an increased stress on reading as a theoretical problem or, as it is sometimes erroneously phrased, by an increased stress on the reception rather than on the production of texts. It is in this area that the most fruitful exchanges have come about between writers and journals of various countries and that the most interesting dialogue has developed between literary theory and other disciplines, in the arts as well as in linguistics, philosophy and the social sciences. A straightforward report on the present state of literary theory in the United States would have to stress the emphasis on reading, a direction which is already present, moreover, in the New Critical tradition of the forties and the fifties. The methods are now more technical, but the contemporary interest in a poetics of literature is clearly linked, tradition-ally enough, to the problems of reading. And since the models that are being used certainly are no longer simply intentional and centered on an identifiable self, nor simply hermeneutic in the postulation of a single originary, pre-figural and absolute text, it would appear that this concentration on reading would lead to the rediscovery of the theoretical difficulties associated with rhetoric. This is indeed the case, to some extent; but not quite. Perhaps the most instructive aspect of contemporary theory is the refinement of the techniques by which the threat inherent in rhetorical analysis is being avoided at the very moment when the efficacy of these techniques has progressed so far that the rhetorical obstacles to understanding can no longer be mistranslated in thematic and phenomenal commonplaces. The resistance to theory which, as we saw, is a resistance to reading, appears in its most rigorous and theoretically elaborated form among the theoreticians of reading who dominate the contemporary theoretical scene.

   It would be a relatively easy, though lengthy, process to show that this is so for theoreticians of reading who, like Greimas or, on a more refined level, Riffaterre or, in a very different mode, H. R. Jauss or Wolfgang Iser-all of whom have a definite, though sometimes occult, influence on literary theory in this country - are committed to the use of grammatical models or, in the case of Rezeptionsdsthetik, to traditional hermeneutic models that do not allow for the problematization of the phenomenalism of reading and therefore remain uncritically confined within a theory of literature rooted in aesthetics. Such an argument would be easy to make because, once a reader has become aware of the rhetorical dimensions of a text, he will not be amiss in finding textual instances that are irreducible to grammar or to historically determined meaning, provided only he is willing to acknowledge what he is bound to notice. The problem quickly becomes the more baffling one of having to account for the shared reluctance to acknowledge the obvious. But the argument would be lengthy because it has to involve a textual analysis that cannot avoid being somewhat elaborate; one can succinctly suggest the grammatical indetermination of a title such as The Fall of Hyperion, but to confront such an undecidable enigma with the critical reception and reading of Keats' text requires some space.

   The demonstration is less easy (though perhaps less ponderous) in the case of the theoreticians of reading whose avoidance of rhetoric takes another turn. We have witnessed, in recent years, a strong interest in certain elements in language whose function is not only not dependent on any form of phenomenalism but on any form of cognition as well, and which thus excludes, or postpones, the consideration of tropes, ideologies, etc., from a reading that would be primar-ily performative. In some cases, a link is reintroduced between performance, grammar, logic, and stable referential meaning, and the resulting theories (as in the case of Ohmann) are not in essence distinct from those of avowed grammarians or semioticians. But the most astute practitioners of a speech act theory of reading avoid this relapse and rightly insist on the necessity to keep the actual performance of speech acts, which is conventional rather than cognitive, separate from its causes and effects - to keep, in their terminology, the illocutionary force separate from its perlocutionary function. Rhetoric, understood as persuasion, is forcefully banished (like Coriolanus) from the performative moment and exiled in the affective area of perlocution. Stanley Fish, in a masterful essay, convincingly makes this points What awakens one's suspicion about this conclusion is that it relegates persuasion, which is indeed inseparable from rhetoric, to a purely affective and intentional realm and makes no allowance for modes of persuasion which are no less rhetorical and no less at work in literary texts, but which are of the order of persuasion by proof rather than persuasion by seduction. Thus to empty rhetoric of its epistemological impact is possible only because its tropological, figural functions are being bypassed. It is as if, to return for a moment to the model of the trivium, rhetoric could be isolated from the generality that grammar and logic have in common and considered as a mere correlative of an illocutionary power. The equation of rhetoric with psychology rather than with epistemology opens up dreary prospects of pragmatic banality, all the drearier if compared to the brilliance of the performative analysis. Speech act theories of reading in fact repeat, in a much more effective way, the grammati-zation of the trivium at the expense of rhetoric. For the characterization of the performative as sheer convention reduces it in effect to a grammatical code among others. The relationship between trope and performance is actually closer but more disruptive than what is here being proposed. Nor is this relationship properly captured by reference to a supposedly "creative" aspect of performance, a notion with which Fish rightly takes issue. The performative power of language can be called positional, which differs considerably from conventional as well as from "creatively" (or, in the technical sense, intentionally) constitutive. Speech act oriented theories of reading read only to the extent that they prepare the way for the rhetorical reading they avoid.

   But the same is still true even if a "truly" rhetorical reading that would stay clear of any undue phenomenalization or of any undue grammatical or performa-tive codification of the text could be conceived - something which is not neces-sarily impossible and for which the aims and methods of literary theory should certainly strive. Such a reading would indeed appear as the methodical undoing of the grammatical construct and, in its systematic disarticulation of the trivium. will be theoretically sound as well as effective. Technically correct rhetorical readings may be boring, monotonous, predictable and unpleasant, but they are irrefutable. They are also totalizing (and potentially totalitarian) for since the structures and functions they expose do not lead to the knowledge of an entity (such as language) but are an unreliable process of knowledge production that prevents all entities, including linguistic entities, from coming into discourse as such, they are indeed universals, consistently.defective models of language's impossibility to be a model language. They are, always in theory, the most elastic theoretical and dialectical model to end all models and they can rightly claim to contain within their own defective selves all the other defective models of reading-avoidance, referential, semiological, grammatical, performative, log-ical, or whatever. They are theory and not theory at the same time, the universal theory of the impossibility of theory. To the extent however that they are theory, that is to say teachable, generalizable and highly responsive to systematization, rhetorical readings, like the other kinds, still avoid and resist the reading they advocate. Nothing can overcome the resistance to theory since theory is itself this resistance. The loftier the aims and the better the methods of literary theory, the less possible it becomes. Yet literary theory is not in danger of going under; it cannot help but flourish, and the more it is resisted, the more it flourishes, since the language it speaks is the language of self-resistance. What remains impossible to decide is whether this flourishing is a triumph or a fall.

   Notes   I . Roland Barthes, "Proust et les noms," in To Honor Roman Jakobson (The Hague: Mouton, 1967), part I, pp. 157-58.   2. "Proust et le language indirect," in Figures II (Paris: Seuil, 1969).   3. Blaise Pascal, "De 1'esprit gdom6trique et de fart de persuader," in Oeuvres completes, L. Lafuma, ed. (Paris: Seuil, 1963), pp. 349ff.   4. A. J. Greimas, Du Sens (Paris: Seuil, 1970), p. 13.   5. Stanley Fish, "How to Do Things with Austin and Searle: Speech Act Theory and Literary Criticism," in MLN 91 (1976), pp. 983-1025. See especially p. 1008.