PART I—
LANGUAGE

Chapter One—
Flight from Eden:
Myths about Myths about Language in Modern Times

It's an age-old problem with language: since it has never done what it was meant to do, since no one has ever established a lasting notion of what it was meant to do, and since it has never submitted to any plausible account of why it functions in the inadequate way it does, the only way out has been to resort to myths. Most of these myths are myths of origin: language today is imperfect in performing its assigned task, but in the beginning it was different in a way that can be ideally hinted at but not fully described. This is the way Rousseau sees it. "The first language of man," he says, "the most universal, the most energetic language, and the only one he needed before it became necessary to persuade men in an assembly, is the cry of Nature."^[1] In those early days "people spoke only in poetry."^[2] Man had not yet learned to reason, but that was fine since his life was an unspoiled vision of simplicity and his needs were few.

Some myths are not quite like this one. The myth about language that has served as a point of reference for almost all philosophy of language in the West is a myth of the present, and it belongs to Cratylus in the dialogue of that name by Plato. Cratylus is a naive etymologist who thinks that language is meant to name things and that names consequently bear a necessary connection with the things they designate. But even Cratylus's argument requires an appeal to an unknowable and ideal past, a past when the words of language sprang into existence by a kind of natural necessity.

The distance of the myth, whether it's a distance in time or just a

distance in thought, a sense of distinctness from the way things are today, is essential. It is a correlate to the present inadequacy of language: if language is imperfect today, it must be that it has fallen from grace, from some distant Eden where it did all the things it no longer does. But language is inadequate only in relation to a standard that has been assumed, often without any explicit argument, from the outset. Thus the pure, natural language that Rousseau's primitive man spoke—or sang—in that idyllic landscape of uncorrupted, precivilized virtue was ideal because Rousseau expects from language a perfect medium for the expression of man's passions. Love, hate, pity, and anger are the things that first raised men's voices, and this is why the first languages were "singing and passionate" rather than "simple and methodical."^[3] The birth of language in the passions has to be accepted before the myth makes sense. And the pure language that Cratylus sees in operation in the present is pure because he expects from language a perfect medium for the transmission of things in nature. The primacy of this function of language has to be accepted before his myth makes sense.

We can see what makes myths about language so valuable. It is not at all important to ask whether the authors of these myths (if there is such a thing as the author of a myth) really believe them to be plausible, historical accounts of the emergence or origin of language. It is important only to ask what the myths tell us about their authors' notions of language in the present—for instance, that Rousseau saw the unity of speaker and speech as a fundamental obligation of human language, whereas Cratylus saw the unity of name and thing (signifier and signified) as an equally fundamental obligation.

Why does this need to be said? Because the history of the theory of literary language in modern times is a history of the attempted flight from Eden and the subsequent, irresistible pull of Eden. This history parallels, in fact is intertwined with, the history of modern linguistics, a distinctly modern product that is modern precisely because it has tried to be "scientific" and dispense with disobedient, metaphysical ideas like myths of the origin of language. But modern linguistics, for all its efforts to demythify, hasn't resisted the temptation of the traditional myths any more than Rousseau or Cratylus did; or any more than the visionary poets of the late nineteenth and early twentieth centuries who dared, in what appeared to be a last, desperate appeal for the metaphysical integrity of the human spirit in a scientific and skeptical age, to propose their own myths about the nature and origin of language. It wasn't really a last, desperate appeal, because the myths were carried on—by even the

scientifically inclined literary theorists. And it wasn't an appeal for the metaphysical integrity of the human spirit either, because the myths served the same function for the visionary poets as for the theorists. The myths were there to be rejected in favor of a far lonelier view of language. At least that's what they all thought. The only problem was that the myths never really went away.

The Myth of the Fractured Myth

The curious thing is that there is more than one myth—or more than one level of myth. There are the various myths about language, its primeval unity, its innate expressiveness, its state of perfection in some other time and place than now and here. But there is also the myth of how these myths came to be abandoned. For modern criticism not only has perpetuated many of the myths of its predecessors but also has told its own story with a pretty fair mythical imagination of its own.

The story of one of the most fundamental and "radical" tenets of modern criticism and linguistics is a perfect illustration of this point. The lost unity of language in the early modern era was most forcibly asserted by the Swiss linguist Ferdinand de Saussure, who used the expression l'arbitraire du signe (the arbitrariness of the sign). This cardinal concept, the concept of Saussure's Cours de linguistique générale (Course in general linguistics, published in 1916), the frontal assault on the myth that the sign, or word, bears a motivated connection with the thing it designates, has been seen as a necessary condition for modern linguistics and criticism. It is the cornerstone of the "lonelier" modern view of language.

The myth of modern criticism is that this concept was new. It was not. Eugenio Coseriu, a Romanian linguist teaching at Tübingen, wrote an article in the 1960s in which he demonstrates first, how pervasive the belief is that the theory of l'arbitraire du signe originated in the twentieth century and, second, how far this belief is from the truth. In fact, Coseriu shows, the notion has to be traced back to Aristotle. For one thing, Aristotle sees language from a purely functional rather than a Cratylic point of view, rejecting the Cratylic idea of any sort of motivated or necessary connection between words and the things they stand for. For another, the very concept of arbitrariness and the expressions in Latin, French, German, and English to convey it can themselves be traced back to Aristotle and a troublesome little two-word phrase he

used in his treatise On Interpretation . The phrase in Greek is kata syntheken, and in Coseriu's understanding it means something like "through what is already established." In a famous passage from On Interpretation Aristotle defines language as "sound that is meaningful kata syntheken, " which is to say that the words of language have meaning not through any kind of natural necessity, as Cratylus believed, but through historical necessity, through tradition, through what is already established.

Next, Coseriu says, the concept of kata syntheken is adopted by a long line of Western philosophers and translated into various modern languages by a number of different phrases. What is striking is that if one is willing to accept Coseriu's idea that the various words and phrases (ad placitum, institutio, and conventio in Latin, arbitraire in French, Willküur in German, arbitrary/arbitrariness in English) are equivalent, a great many thinkers steadily from the seventeenth century up until Saussure had already proposed the notion of the arbitrariness of the sign. In fact, many of them had done so without any particular claim to originality, in some cases treating the idea as a self-evident truth. Coseriu lists about twenty philosophers, among them Hobbes, Locke, Leibniz, Christian Wolff, Berkeley, Condillac, Lessing, Fichte, and Hegel.^[4]

Coseriu, whose business it is to know the facts about the history of language philosophy, expresses shock and indignation at this common oversight. And who can blame him? After all, Hobbes, Locke, and Leibniz are not insignificant thinkers. The writers whose ignorance is in question are not insignificant in their fields either; they are prominent linguists and historians of linguistics. Is the problem that Coseriu's list of names is largely restricted to philosophers who were known less for language philosophy than for other things? This may well be part of it. The twentieth-century philosopher and historian of philosophy Ernst Cassirer devotes a brief section of his monumental work The Philosophy of Symbolic Forms to the history of language philosophy, mentioning a good many of the same names as Coseriu does. But then Cassirer is hardly a standard to measure others by, since he was enormously erudite and saw everything in such broad terms as only his uncannily vast learning permitted. The fact remains that Coseriu's list does not look like the list of figures one would include in a history of language philosophy, linguistics (to the extent that these two are different), or modern literary theory.

But something else is going on here, something that Coseriu doesn't mention and that he really has no reason to mention. Look at the writ-

ings of the literary theorists who wrote in the years right after Saussure and to whom the notion of the arbitrariness of the sign (to choose just this one concept of "modern" language theory) may be attributed, of any theorist who is committed to undermining the notion that language functions because of a mystical unity between sign and world (or sign and speaker, or sign and national spirit). What about the Russian Formalist critics, for instance, who in the 1910s and the 1920s were so intent on rejecting the old myths and retaining only the most functional, "scientific" view of language? (Never mind, for the moment, that they didn't really abandon the myths; we'll get to that shortly.) How many of them cite Hobbes, Locke, Leibniz, Wolff, Berkeley, Condillac, Lessing, Fichte, or Hegel as their models and predecessors? These names almost never appear in the writings of the Formalists or in writings about them.

The names that do appear are either from the tradition that still entertained the old myths, and are thus cited antagonistically, or from a new, revolutionary tradition, and are thus cited favorably. Viktor Shklovsky, one of the most famous members of this group, in his early writings uses a critique of the nineteenth-century philologist Aleksandr Potebnia to launch his own theory of poetic language, citing the contemporary Russian Futurists as illustrations of the "new" ideas he is advocating.^[5] When the renowned linguist Roman Jakobson, whose early career is an integral part of the history of Russian Futurism and Formalism, comes to talk about the formative influences in his early linguistic theories—the principle of relativity and the belief that everything in language "is based on relationship"—he mentions not the linguists or philosophers of language but "the great men of art born in the 1880's": Picasso, Joyce, Braque, Stravinsky, Khlebnikov, and Le Corbusier. It was in studying Khlebnikov's poetry, Jakobson says, that he first tested one of his most basic concepts, that of the phoneme.^[6]

Is this preference for recent thinkers just a manifestation of the cult of the new that is a trademark of modernism, a bold refusal to acknowledge any predecessors except either the latest ones or those who are to be discarded for having old, worn-out ideas? Maybe. Is it simple ignorance? Possibly in some cases, though certainly not in the case of Jakobson. But two things are clear. One is that the ideas of the Formalists (to take just this one example) were not as new as the Formalists' mythical account of their own ideas made them out to be. The other is that the Formalists, while perpetuating the myth of the novelty of their own ideas (the myth of the rejection of myth in favor of functionalism and

science), were still holding on to precisely those myths they claimed to be rejecting.

L'arbitraire du signe is only one idea that the twentieth century has canonized, and canonized as specifically new. The Russian Formalists are only one group of critics and did not, besides, explicitly propose a doctrine of the arbitrariness of the sign. It would be wrong to suggest that their failure to recognize predecessors in this doctrine is by itself evidence of much of anything. Still, there is a pattern here. Modern literary theory and criticism has always promoted its own novelty in cases where it is not completely novel. It has defined itself by a relation of antagonism with its past while continuing a great many of the myths it rejects. The whole process appears to be one vast evolutionary rejection of the myths. When Shklovsky criticizes Potebnia for overlooking the formal elements of language in his distinction between poetic and prosaic language, for ignoring the "facts" of sound and rhythm and thus being unscientific, there are two things he doesn't realize (apart from the circumstance that what he says about Potebnia isn't really true): first, that Potebnia, like Shklovsky himself, is engaged in a valiant attempt to reject the old, unscientific myths about poetic language and, second, that he, Shklovsky, was just as unsuccessful as Potebnia. Both merely perpetuated the myths they sought so earnestly to overturn.^[7]

And that is pretty much the story of this chapter. Over and over again modern criticism, like a neurotic lover foolishly, ridiculously vowing every day to escape the clutches of the abusive and overbearing tyrant who was once a source of tenderness and security, seeks a radical break with the myths of the past, only to find itself helplessly drawn back to them. The poets I've chosen as characters in this story define the terms of the whole ritual. They supply the idea of the break, and they sustain the myths to run back to. One of the most prominent myths is precisely that there is no myth, that we have finally made that break. We haven't.

The Fracture

Before we describe this history, we need to have a sense of where it goes. How will this struggle for freedom from the old myths play itself out in twentieth-century criticism? I've used the word break to describe the desired liberation from mythical forms of thought. As it happens, the notion of a break or fracture comes to be an integral part of the very theories that seek that break with the past. Now, however, the break has

to do with language itself, not just the relation of theorists and poets to their past.

The late French philosopher Michel Foucault has described the modern view of language in Les mots et les choses (Words and things, translated into English for some reason as The Order of Things ) in a way that is useful, although his context is different from ours. For Foucault, a view of language is part of an episteme, by which he means an entire way of understanding and knowing in a given epoch. An episteme is different from a worldview or Weltanschauung in that it has as its specific object the human, social world. The three broad classifications into which the objects of a single episteme can be divided are life, work, and language.

Foucault is interested irf three epistemai, the most recent of which extends from shortly before the beginning of the nineteenth century to the present. The big change that begins to occur in the thinking about language early in this period is the change from a view of language as a representational medium to a view of language as functional. In the period of Foucault's first episteme, roughly corresponding to the sixteenth century, the prevailing notion of language is of a medium that operates through a principle of resemblance between words and designated objects. In the period of the second episteme, comprising approximately the seventeenth and eighteenth centuries, the notion of resemblance is rejected in favor of a binary notion of language. That is, language is seen as a neutral and conventional means for representing—rather than imitating—the world, and the emphasis is on the split between representing word and represented world. In the modern period representation gives way to a notion of pure function, as philologists discover laws of similarity among the Indo-European languages, particularly those laws that show resemblances among such purely grammatical attributes of languages as inflectional endings. "Language is no longer composed only of representations and sounds that, in turn, represent these representations and order themselves as strings of thought require. It is instead composed of formal elements that are grouped in a system and that impose on the sounds, the syllables, and the roots an order that is not that of representation." This is how there comes to appear in the modern age a "dimension of the pure grammatical."^[8]

As a result, language in the modern age takes on a kind of opacity and becomes an object in its own right, instead of being merely a transparent, representational medium. It "folds back on itself, acquires its own density, displays a history, laws, and an objectivity that belong to

it alone." Language exists now only in a "dispersed mode." "For the philologists, words are like so many objects composed and passed on by history; for those who wish to formalize, language must rid itself of its concrete content, leaving only the universally valid forms of speech to appear. If one wishes to interpret, then words have become a text to be fractured in order that one can then see the emergence of this other meaning they hide."^[9]

Foucault's presentation of the modern episteme goes a long way toward describing the status of language in the writings of twentieth-century literary critics, theorists, and aestheticians. But critics, theorists, and aestheticians are not Foucault's subject, at least not exclusively or even prominently. Foucault speaks largely of the professional philosophers—in addition to speaking of authors and poets from time to time—and thus provides a perspective similar to the one we find in Coseriu's article on the arbitrariness of the sign. When Foucault speaks of Mallarmé, saying that with him (and Nietzsche) "thought was led, and violently so, back towards language itself, back to its unique, difficult being," thus making Mallarmé a crucial figure in the twin processes of the "fracturing of language" and its transition to a state of "philological objectivity,"^[10] he speaks almost as though the whole question had been settled with Mallarmé. It had not. But the value of Foucault's discussion of language for us lies in its capacity to describe the limits of the trend he is talking about, to describe a view of language that has come to dominate much twentieth-century thinking.

The term fracture is apt, and it certainly gives a good idea of the situation in which language finds itself in the twentieth century. The fracture operates on levels that Foucault does not enumerate (this is not his purpose). But one may classify a great many twentieth-century theories of language according to the terms in which they formulate a fracture. Its most basic sense may be found in the Saussurian doctrine of I'arbitraire du signe, in which the fracture severs the sign from its referent. Any aesthetics that is formalist in the general sense—any aesthetics emphasizing functions and relations over elements (in this case words)—will accept this doctrine as a self-evident truth.

Other schools of criticism, aesthetics, and language theory have seen the fracture as operating at a higher level than sign and referent, namely the level of the extended utterance. Theorists loosely grouped under the rubric of phenomenology tend to see the problems of language arising not out of the relation between words and things, or signs and referents, but instead out of the relation between larger signifying units and a

newly problematized entity called meaning. Edmund Husserl, the founding father of this movement in philosophy, had initiated this trend in his Logische Untersuchungen (Logical investigations, 1900–1901) by referring to meanings as an independent class of ideal objects that don't need to have anything to do with the signs associated with them.^[11] In fact, the nature of this association is far less simple than has commonly been supposed, Husserl thinks. Signs may well be signs for something, he says, but they do not in all cases possess a meaning or sense that is "expressed" by them.^[12]

When Roman Ingarden, a Polish pupil of Husserl, comes to talk specifically about literary texts in the late 1920s, he organizes language into what he calls meaning units, or unified utterances. Like Husserl before him, Ingarden distinguishes sharply between meanings, which are created by subjective acts of consciousness, and the units of language associated with those meanings. The meaning function of language—the function, that is, that associates a meaning with a particular meaning unit—is conferred on that meaning unit from without by a subjective act of consciousness and can therefore in no way be considered as automatic or absolute.^[13] The very premise of Ingarden's theory is thus that there is no necessary, simple, or natural relation between language and its meanings. Three decades later the French phenomenologist Maurice Merleau-Ponty also was to insist on the lack of a necessary and direct relation between utterances (by which Merleau-Ponty understands units larger than individual words) and their associated meanings. For him, meanings are ideas that cannot be directly expressed but can only be approached through a series of "adumbrations" (Abschattungen, a Husserlian term). In his essay "On the Phenomenology of Language" (1951) he describes how the adumbrations in a given utterance suddenly "contract into a single meaning" by a process that can't even be described. For Merleau-Ponty, there can never be any question of a direct signifiedsignifying relation because the signified always "surpasses" the signifying.^[14]

A much more radical fracture, however, is the one we find in the work of Jacques Derrida and his many American epigones. In fact, the whole trend known as postmodernism, which helped set the tone for literary studies in the 1970s, is dedicated to an idolatrous fascination with the lonely, nihilistic worldview that is implicit in Derrida's fractured theory of language. For Derrida, language can be characterized only as a play of differences among an intricate network, or "tissue," of signs. Language is marked by the "impossibility for a sign, for the unity of a

signifier and a signified, to occur in the plenitude of a present and an absolute presence."^[15] There is no present or absolute presence in any sign other than the divine logos, and the logos is a purely metaphysical, "infinitist" notion.^[16] In the language of man, meaning is always in flight, always beyond the horizon; there is no presence, no unity, only the trace of what was just there. Thus language is not only the differences at play in a tissue of signs; it is also différance (with a instead of e ), a word Derrida invented to combine the two principal senses of the French verb différer, "differ" and "defer." In the intricate complex of differences that is language, meaning is always deferred, put off, scattered. Derrida introduces the concept of the trace to describe the way in which language is never fully present (as we would like to think it is) but exists instead in a state that looks back to something that was already there. The trace, always flitting ahead of the presence that it leaves behind, thus describes a double process of "retention" (since it retains what has been present) and "protention" (since it tends forward, away from that present). Signification is born at this juncture of retention and protention, and Derrida finds in the French language a word that happily combines the notion of juncture and rupture he sees united in the trace—brisure, which means "hinge," as in the hinged articulation between two parts of a shutter (this definition, of course, allows Derrida to exploit the double sense of "articulation"), and "fracture."^[17]

There is another kind of fracture that at one time had become canonical in literary aesthetics, and that is the fracture that separates speaker from language. This fracture is the premise of any objectivist literary aesthetic, any aesthetic that regards the word as independent of its speaker and the literary work of art as independent of its creator. Such a view had become virtually an article of faith in the teaching of literature in many of the most respected American universities in the 1960s and 1970s, when it was fashionable for professors to shock undergraduate students by telling them that inquiries into an author's life or into the author's avowed understanding of his or her own work were irrelevant to a true understanding of that work—hence the license to throw lecture notes away and spend hours of class time in an exercise known as close reading, wherein students, with their instructor's minimal participation, take a piece of writing, like a bit of organic tissue, pick it into tiny bits, and comment on how those bits interact either with each other (the bits) or with them (the students).

Objectivist criticism became entrenched in the 1950s, when many of the next decade's academics were still in graduate school. Cleanth

Brooks published The Well Wrought Urn in 1947, proposing a view of literary artworks as organic structures whose constituent parts were to be understood and judged only in terms internal to the artwork as a totality.^[18] In "The Intentional Fallacy," first published in 1946 and then again as a chapter of The Verbal Icon in 1954, W. K. Wimsatt and Monroe Beardsley had argued that consulting an author to discover the meaning of that author's words was illegitimate. "Critical inquiries are not settled by consulting the oracle," the essay ends.^[19] And in Aesthetics: Problems in the Philosophy of Criticism (1958), a work designed as an introductory college textbook in aesthetics, Beardsley forcefully espoused objectivism as a principle of all art criticism.^[20]

Wimsatt and Brooks, in their Literary Criticism: A Short History, published in 1957, trace the origins of the objectivist aesthetic, or "impersonality," as they put it (since the author's "person" does not appear in his works). Credit goes to T. S. Eliot, they say, not because he was the first to come up with the idea but because it was he who "brought this matter of impersonality squarely to the attention of his generation." Eliot's position in separating the personal experience of the poet from the poem itself is stated in 1919 "with almost shocking emphasis," as Wimsatt and Brooks put it.^[21]

The idea of severing the speaker from his words is by no means restricted to such New Critics as Wimsatt and Brooks. Nor am I persuaded that Eliot is the one who brings it into the open. It is implicit in any formalist or structuralist view of language (in the general sense of those terms), where the emphasis is on relations between words rather than on their expressive qualities. And it emerges in the phenomenological views of language I mentioned a moment ago, since for thinkers like Ingarden and Merleau-Ponty the meanings of language escape and surpass the speaker as much as they escape and surpass language. But no matter what the specifics of the view are, language is a long way from the organic unity it enjoys in the myths—unity with man in Rousseau and many of the romantics, unity with God in the Gospels.

Many of the relativizing tendencies in the twentieth century are evidence of a fracture in the primitive unity of language. This notion is true of any theory asserting that the meanings of language are determined by a context of some sort. It is essential to the thought of Mikhail Bakhtin, the Russian philosopher and literary theorist whose works became immensely fashionable in the 1980s. One of Bakhtin's central notions is raznorechie (translated as "heteroglossia"), which describes the necessary interaction in verbal utterances between the fixed system of a

given language and the context of a particular utterance.^[22] It is certainly there in E. D. Hirsch, who, some twenty years before he gained notoriety for his Cultural Literacy (a kind of companion best-seller to Allan Bloom's Closing of the American Mind in 1987), wrote a book in which he gave a theory of literary interpretation. Hirsch distinguishes between the "meaning" of a literary work as a fixed thing established in the context of its writing, and the "significance" of a literary work as something that changes over history as a function of the contexts of future generations of readers.^[23] It is there in so-called reception theories, which see the meaning of language as largely determined by characteristics of the listener (or reader). And it is there in Marxist theory, which views consciousness and, consequently, language as determined by social existence.

The twentieth century is not responsible for inventing the skepticism necessary to produce language theories as lonely and unsettling as some of these. The fracture is there, at least in germinal form, in the early modern period I am discussing. But it is never asserted wholeheartedly. Instead, the old myths dwell side by side with the fractured notion of language. And the forum that allows for this cohabitation is the one where the distinction between poetic language and ordinary, or prosaic, language is investigated. The typical symptom of the surviving neurotic attachment to the old myths is the tendency to distinguish between two types of language, the language of poetry and the language of prose (or whatever you want to call language that is not poetic), and to assign a privileged, Edenic status to poetic language.

The Myth of the Poetry-Prose Distinction . . . and the Myth That There Is No Myth

Maybe it would be most accurate to say simply that the whole poetryprose distinction is a myth, at least as it is framed in this period. Whether or not this is true, the distinction presents a problem since so much of modern literary criticism and language theory has been intent on fostering an image of itself as the product of rational reflection, of a carefully, thoughtfully elaborated "methodology" (a buzzword of literary study today). It has in many cases sought to earn itself a place next to the other "human sciences," which are also a product of modern times and whose central obsession has been with their place next to the meth-

odologically more secure natural sciences. The scientifically inclined schools of literary criticism, the formalisms and the structuralisms, have ridden the coattails of twentieth-century linguistics, something that is evident not only because modern criticism has often borrowed the methods and terminology of linguistics but above all because it has focused on the same object of study, namely language. Even where the notion of science and scientificity has not had an overriding appeal, the concern has been with detachment, with sober and impassive judgment—hence the preference for a spirit of objectivism in so much twentieth-century criticism.

But literary criticism is of necessity either implicitly or explicitly an aesthetic. Its object of study is not language in general, signifying objects in general, or meaning in general; it is specifically literary language, literary signifying objects, literary meaning. The problem has been, then, to account for the specific "literariness" of literary language (what Jakobson called literaturnost' ), but to do so without abandoning the highly esteemed posture of scienrificity. It is questionable whether modern criticism or modern linguistics has successfully done this, though both have certainly tried. The task presents some significant difficulties. For to set aside a certain body of written texts and call them literary is to assume the existence of an aesthetic value, some quality that mysteriously places these texts in an entirely different category from the ones in which we place "ordinary" written texts. And the minute we start talking about values, we are talking about murky, essentialist, metaphysical qualities whose existence is suspect and whose functioning is hopelessly obscure, at least from a "scientific" viewpoint.

Nonetheless, modern thinking has had trouble abandoning the romantic notion that literary texts are somehow the object of reverence and awe. It has also not quite been able to abandon the notion that poetic literary texts are more literary than prose texts and that they are, consequently, more the object of reverence and awe. And there is an assumption, usually unwritten, that literary works of art owe the specificity they possess as works of art to qualities in the language that composes them. It follows that there must be a difference between literary language and "ordinary" language and that this difference accounts either entirely or in part for the difference between, say, a poem by Blake and a paragraph from a car repair manual. Even deconstructionist thinkers like Derrida and Paul de Man, for all their skepticism about language and meaning, retain a faith in the specificity of poetic language. Derrida

drew a clear distinction between literary and philosophical texts, and de Man, as I mentioned in my introduction, used the term rhetoricity for his version of the specificity of poetic language.^[24]

And so we see in modern criticism the complicated situation I described earlier. The myth subsists that there is an intrinsic distinction between poetic and prosaic language. I use the word myth not because I wish to assert that the poetry-prose distinction is false in all of its versions (although I do happen to believe it is false when it is expressed as a distinction between two types of language ); I use it because belief in the poetry-prose distinction is framed as a myth, usually a Rousseauistic myth of original purity. But as literary aesthetics and its corresponding philosophy of language grow increasingly "scientific," there is an increasing tendency toward formalism in philosophy of language, and with formalism comes a tendency to demythify. Hence the coexistence, in the early modern period, of the myth itself (that is, the myth of the poetry-prose distinction) and the myth that there is no myth.

The idea of poetry (a term that can mean anything from imaginative writing to specifically lyrical verse) as a distinctive form of verbal expression that stands in natural opposition either to other forms of expression or to whole ways of apprehending the world is an idea of long standing. Aristotle, in the Poetics, had opposed poetry (understood in its broadest sense as what we might call today simply literature) to history, saying that poetry is more philosophical, indeed is "greater" than history, because poetry expresses universals whereas history expresses particulars.^[25]

The more familiar distinction since Aristotle has been between poetry and prose, poetry being understood in a more specific sense than the one it has in Aristotle. This distinction is standard in the poetics of romanticism. Rousseau sets the tone for the entire era in his "Essay on the Origin of Languages," where poetry is identified as the earliest and, of course, most natural of languages. The English and German romantics return again and again to the same theme: poetry, specifically as distinguished from prose, is the purest, the most original (in the sense of being close to the origin) form of verbal expression. The German romantic critic Johann Gottfried von Herder, in his essay on the origin of language, asserts, like Rousseau before him, that poetry is older than prose and that in its capacity to express passions in all their immediacy lies the primitive source of all verbal communication.^[26]

By the late romantic era in England something new has happened: the poetry-prose distinction has given way in a great many instances to

a distinction between poetry and science. M. H. Abrams, in his classic work on English romanticism The Mirror and the Lamp, documents this transition, citing Thomas Babington Macaulay's 1825 essay on Milton as evidence that Aristotle's poetry-history distinction had been converted into the distinction between poetry and science.^[27] But the poetry-science opposition does not arrive directly from a rejection of Aristotle; it comes along instead to replace the distinction that was dominant at the time, namely the distinction between poetry and prose. This, at least, is how Wordsworth sees it. As early as 1800, in the preface to the Lyrical Ballads, he goes so far as to say that "there neither is, nor can be, any essential difference between the language of prose and metrical composition." Then, in a footnote, he says that "much confusion has been introduced into criticism by this contradistinction of Poetry and Prose, instead of the more philosophical one of Poetry and Matter of Fact, or Science."^[28] The poetry-science opposition in most cases does not represent an abandonment of the poetry-prose distinction; it represents merely a clarification, in epistemological terms, of the perceived province of prose, since science in the new opposition corresponds to prose as the vehicle exclusively suited to its methods and worldview. Thus poetry is to prose as feeling is to intellect, or as expression is to description, or as emotion is to scientific cognition.

This view is taken over in the work of Wilhelm von Humboldt, author of the highly influential über die Verschiedenheit des menschlichen Sprachbaues und ihren EinfluB auf die geistige Entwicklung des Menschengeschlechts (On the diversity of human speech structure and its influence on the spiritual development of the human race, published 1827–29) and in many ways a pivotal figure both for the history of language philosophy and for our story here. Humboldt divides the territories of poetry and prose in a manner similar to that of the romantics. Poetry apprehends reality in its sensible (sinnlich ) manifestations; it is "inseparable from" music and is thus more appropriately the vehicle of aesthetic beauty than is prose. It is given to the rendition of "individual moments" in human experience rather than to universals. Prose, by contrast, is concerned with facts and concepts rather than with ideas or sensation. Since it is suited to the rendition of ordinary life, it is the province of the objective, of science and scientific terminology; and where poetry "belongs to the individual moments of life," prose accompanies man "in all the expressions of his spiritual activity."^[29]

But there is a significant difference between the opposition that Humboldt makes and the one the romantics made, and that is that for Hum-

boldt the opposition does not of necessity entail the privileging of poetry over prose. Humboldt's distinction is simply descriptive, and this is because his approach to language is in certain fundamental respects different from that of the romantics. Humboldt sees language above all as a function of our inner being. It is an organic medium, representing the perpetual spiritual activity of an entire nation (not the individual). As Humboldt put it in his famous formulation, language is not a mere dead created thing (ein totes Erzeugtes ) but an ongoing creation (eine Erzeugung ); it is not (finished) work (he uses the Greek word ergon ) but ceaseless activity (energeia ).^[30] Any definition of language must, then, be a genetic one, that is, one that sees language not as the product of an act of genesis in some distant, mythic past but as a continuing genesis in the present. Thus Humboldt has taken the romantic myth of origin, by which language is a creature fallen from the grace it had "in the beginning," and replaced it with a myth of origin where that origin is always now .

The secret to the creative, perpetually generative force of language in the present is what Humboldt calls the inner form of speech, a notion that will be taken over by the Russian philologist Potebnia, with lasting consequences for the Russian tradition. Inner form is the thing that makes language what it is. It is the key to the connection between language production and the sound of language. It is what allows language to express those thoughts that the greatest minds entrust to it.^[31] Language deals not with objects but with concepts.^[32] Its principal task is to mediate between objective reality and subjective inwardness, and it successfully accomplishes this task when sound and inner form are harmoniously joined. At that moment the process of language production and the external sound qualities of language work together to form a language that is living and rich, particularly in metaphorical content. This happens especially in the works of poets and philosophers, who contribute to the development of a nation's language by infusing its words with spiritual content. The success of a language for Humboldt is always measured by the progress of that language through history, and this notion leads him to mention a phenomenon that Potebnia and the entire Russian tradition later canonize. Harmony of inner and outer form is lost when a language stagnates. At that moment metaphors that had previously had a "youthful sense" become, through daily use, "worn out" so that they are "barely perceived anymore."^[33]

Why is Humboldt a pivotal figure? To begin with, he has sought to abandon all forms of naive Cratylism and romantic primitivism in his

approach to language. If Humboldt believes in the specificity of poetic language, it is not because he sees poetry as based in some mystical unity of either word and object or word and speaker. For him, poetic language is distinguished from prosaic language more by its function, by the way it is used, than by any intrinsic properties. The nonexistence of a necessary or motivated connection between signs and the things they designate in language in general, whether that language is poetic or prosaic, is already implied in the subject of Humboldt's work. Humboldt's point of departure, as the title of his work suggests, is precisely the fact of the diversity (Verschiedenheit ) of human speech, the obvious fact that, owing to the existence of a multitude of human languages, there is no single, universal, and necessary way of saying any one thing.

The flight from Cratylism or natural-sign language is further signaled by Humboldt's insistence that language has to do not with things but with concepts. This is because language does not spring into existence ready-made to serve poets as a vehicle for the expression of their feelings; on the contrary, we ceaselessly create language for the purpose of meaningfully organizing our own experience. And since Humboldt was a good Kantian, believing that conscious beings can never "know" things in and for themselves, to him language, like any activity that generates meaning, cannot deal directly with objects.

Humboldt's Kantianism is another reason for his status as a pivotal figure. Ernst Cassirer writes about it in the brief history of language philosophy that he includes in the first volume of the Philosophy of Symbolic Forms . As Cassirer sees it, Humboldt betrays his underlying Kantianism in at least two ways. The first is in the notion of language as a subjective medium whose function is to objectify sensory impressions, thus serving as a kind of bridge between subjectivity and the objectivity that, according to Kant, we can never know in and for itself. The second is in the distinction between the matter (namely sound) of language and its form. In Cassirer's reading of Kant form is "the truly objectifying principle of knowledge" because it is the source of unity in our apprehension of objects. The function of form in Humboldt is entirely analogous since inner form for him is exactly what serves as a mediating function between subjectivity and the objective world of phenomena. Humboldt's entire conception of language thus appears in Cassirer's eyes to be founded in the "basic principle of the transcendental method" (the method in Kant by which the philosopher logically deduces the categories that the mind uses to order experience). Cassirer calls this principle "the universal application of philosophy to science." Hum-

boldt's innovation consists in making language the object of that "universal application" rather than mathematics and mathematical physics, as Kant had.^[34]

The Russian tradition that is central to the development of language theory in modern literary aesthetics begins precisely in the context of the transition that Cassirer identifies in the work of Humboldt. It is important to keep in mind that the period in question is transitional and that the notion of the perpetuation of myths remains fully in force. Humboldt certainly has his own myths, in spite of any Kantian formalism that Cassirer sees in his thinking. The notion of inner form is scarcely an empirically scientific one; even less scientific is Humboldt's concept of nationhood and the way in which some vaguely defined and almost mystical national character finds itself expressed in language. The Russian tradition begins in the shadow of Humboldt, with a bow to Humboldt, but also, strangely enough, with a reassertion of some of the older myths Humboldt had apparently rejected—the myth of origin, for instance, which now shows itself in a slightly new form.

The Russian Tradition from Potebnia to Shklovsky, with Some Poets in Between

Aleksandr Potebnia:
From Myth to Science—
And Back Again

The history of the modern tradition in Russia can safely be said to begin with the Ukrainian-born philologist Aleksandr Potebnia (1835—1891). Until the 1980s few people in the West paid much attention to Potebnia's role in the history of modern language theory and criticism. He customarily received a few pages in historical accounts of Russian Formalism and even Russian symbolism. Victor Erlich mentioned him decades ago in the book that introduced the English-speaking world to Russian Formalism.^[1] A voluminous, more recent study of Russian Formalism, in German, contains one fairly long discussion of Potebnia and numerous brief references to him.^[2] There have been sporadic articles on him both in the Soviet Union and in the West,^[3] though generally not in widely read journals, and there have even been a few books on him published in the Soviet Union.^[4] In the 1980s, however, Potebnia began to attract some attention in the West, owing in large part to the surge of interest in early twentieth-century Russian language theory and psycholinguistics, and in 1986 a full-length book in English on Potebnia was published.^[5]

The work that established the outline of Potebnia's thought for his entire career was a collection of articles called Mysl' i jazyk (Thought and Language). It was published in 1862, the year Potebnia turned twenty-seven. His purpose in this book was to take Humboldt's work

as a point of departure, correct certain of Humboldt's misconceptions, and explore an area left unclear in Humboldt, namely, the relation between thought and language. As Potebnia sees it, a satisfactory account of this relation is precisely where his contemporaries and immediate predecessors are most deficient. Potebnia devotes the second and third articles of his collection to the work of Humboldt and two of his successors, Karl Friedrich Becker (1775–1849) and August Schleicher (1821–1868), showing how the system of each of these linguists collapses on this point. Both Becker and Schleicher focus on the problem of whether thought precedes language or language precedes thought, but, as Potebnia sees it, both assert one position only to contradict it unwittingly later on.^[6] Humboldt is not much better on this count, in Potebnia's estimation, because although he does not actually contradict himself, he certainly leaves the matter confused. Thought and language simply do not coincide in any simple formula, Potebnia points out, and Humboldt does not seem to have been aware of that.^[7]

The thought-language problem itself is not exactly the point here since Potebnia's impact on succeeding generations had little to do with this issue. It has more to do with the progress of the myths I have been talking about. But there is a matter of chronology in Potebnia's discussions of his immediate predecessors that is fascinating because it partially explains why his impact on Bely and subsequent thinkers was what it was. Potebnia reduces the flaws in Becker's thinking to a contemptuous phrase, referring to the German linguist's "fruitless formalism."^[8] The whole problem in Becker and Schleicher appears to be a lack of logical and scientific rigor, as the contradictions in their arguments demonstrate.

If Potebnia had waited another ten years or so, however, he would have been much less likely to fault Schleicher for lack of methodological rigor. Schleicher's most important contribution to the history of linguistics was his Darwinian Theory and Linguistic Science, published in 1873. Ernst Cassirer assigns this book a cardinal position in the same transitional phase of the history of linguistics that Humboldt belongs to. Schleicher's book, in Cassirer's view, represents the last step needed after Humboldt "to dissolve linguistics completely into natural science and linguistic laws into pure natural laws."^[9] Schleicher's book was a vigorous assertion of positivism in the study of languages and resolved the earlier thought-language issue by proposing a doctrinaire monism that denied the independence of thought or spirit from matter.

Potebnia formed his theory in the late 1850s largely in response to

recent developments in language philosophy and did not change it substantially for the rest of his career. The date is crucial because ten years later the picture would have been very different, and chances are that his thought would not have taken the direction it did had he started writing then. If Potebnia's view of language is infused with a particular type of mythic thinking, it is no doubt partly because of this circumstance. Later thinkers like Bely owed their conception of language in no small measure to this accident in the history of linguistics.

The most important impulse Potebnia gave to modern literary aesthetics was his redefinition of inner form, which profoundly affected his own version of the poetry-prose distinction. The notion of inner form arises in the context of a psycholinguistic distinction that is central to Potebnia's doctrine. In a section on "the language of feeling and the language of thought" Potebnia distinguishes between the two cognitive processes of understanding and association. Understanding of language takes place when sound precedes meaning; association takes place when meaning precedes sound. Understanding is thus a sort of immediate, necessary apprehension of the meaning of language, whereas association promotes only an indirect apprehension through the assignment of purely conventional meanings to signs. Inner form is relevant only to understanding, not to association, because it is directly tied to human thought. "The inner form of a word," says Potebnia, "is the relation of the content of thought to consciousness; it shows how man's own thought represents itself to him."^[10]

What does this statement mean? Potebnia tells us in the most concrete terms, by making yet another distinction. A word whose inner form is in evidence is one whose meaning is objective, as opposed to one whose meaning is subjective . Words accessible to the understanding have objective meanings since the meaning is simply there, whereas words apprehensible only through association are subjective since listeners must rely on their own subjective faculties to supply a meaning. What, then, is an objective meaning? Easy: it is the "closest etymological" one. Thus the Russian word for table, stol, has an objective meaning because it contains the Slavic root /STL/, "to spread." The presence of this root confers on the word stol a kind of immediate meaningfulness and thus makes it objectively understandable to a speaker of Russian.^[11] In a word like this the inner form is successfully joined to the outer form, or the sound of the word.

Of course, this theory sounds glib and naive today. But Potebnia has done something extraordinary: he has attempted to take the Humbold-

tian notion of inner form and bolster it firmly on a "scientific" foundation of philology, thus making concrete what had been vague and intangible in his German predecessor. But at the same time, apparently without knowing it, he has replaced Humboldt's truly original contribution to the philosophy of language, namely the energeia concept, with the old myth of origins. Potebnia thus claims the existence of a state of primeval purity and thinks that because it has been identified with scientifically established etymological roots, it won't be seen as mystical or mythical.

But mystical and mythical is just what it is, and the person who noticed this was Andrei Bely. In a review article of Potebnia's Thought and Language published in 1910, Bely pointed to the most prominent paradox of Potebnia's system. He shows how Potebnia's notion of the relation between inner and outer form leads to a view of the ideal word that privileges the artistic over the ordinary, since the ideal word for Potebnia, containing as it does an inherent, objective meaning, must necessarily be an autonomous entity, a symbol. "This conclusion," Bely says, "again and again brings [Potebnia] back to the poet-romantic's dream of an unexpressed, fluid, momentary, and rationally incommunicable meaning that shines through as if from the depths of every word. Behind the cover of its ordinary sense the word conceals inside itself a primevalelemental, magical-active force. . . . From linguistics, grammar, and the psychology of verbal symbols Potebnia arrives at the assertion of a mysticism of the word itself."^[12]

Bely's comment is valuable for two reasons. The first is simply the shrewdness he shows in calling attention to a pitfall in Potebnia's thinking. The second is the light he sheds, perhaps unwittingly, on Potebnia's historical role. Bely finds it shockingly contradictory that Potebnia should take as a point of departure the supposedly scientific disciplines of linguistics, grammar, and psychology and arrive from them at a position as mystical as his doctrine of inner form.

But from our perspective this contradiction is exactly what is so important in Potebnia. Potebnia was apparently convinced that his emphasis on the thought-language problem and his psycholinguistic foundation were a timely corrective to the least scientific aspects of Humboldt's theory. He was not willing to consider himself a part of the romantic legacy in language philosophy that unabashedly espoused mystico-mythical ideas. And if, as Bely accurately noticed, Potebnia ended up regressively asserting a mythical tradition even older than Humboldt's, this is only an indication to us that he is a transitional figure.

He is a perfect example of the process I have described because he subscribes at once wholeheartedly and unwittingly to one of the oldest myths in existence but views himself as rejecting all myth. To a significant extent for Potebnia, the myth—beyond the actual myth of origins—is that there is no myth.

The Poetry-Prose Distinction

Since there is no myth and, consequently, no murky, unscientific conceptual framework, Potebnia blithely makes his way from his conception of inner form to a poetry-prose distinction that will be adopted in many significant respects by even the most formalistic of subsequent Russian linguists, critics, and aestheticians. In the last article in Thought and Language Potebnia systematizes his conception of inner form and gives it more clarity. He now speaks of words as containing three elements: outer form, which is defined as the "articulate sound"; content; and inner form, which he unabashedly defines as the "closest etymological meaning" (Mysl' ijazyk, p. 134). What does this system have to do with poetry? The successful unity of sound and meaning is the "form" of poetic production, Potebnia says. The force responsible for creating this unity, the "third link," is none other than inner form (p. 138). Now inner form is also the symbolic meaning of a word, by which Potebnia means the meaning that is directly accessible. Symbolic meaning, or "symbolism," is the province of poetic language, as Potebnia says at the beginning of the article. "Poeticity," he had said, is the symbolism of language, whereas "prosaicity" is the "oblivion of inner form" (p. 134).

There are profound analogies, according to Potebnia, between words and art. Art, like language, is a means for the creation of thought, and its goal, like that of the word, is to produce a certain subjective mood (nastroenie) in both creator and perceiver. Art, because it is an active, creative force, is a form of energeia, just like language (p. 143). But not all language, of course: only poetic language enjoys the privileged status that makes it analogous to art. There is an excellent reason for this, too. The privileging of poetry does not arise from an arbitrary decision that judges it superior to prose. Poetry is privileged because it is earlier . We have here the myth of origins all over again, and Potebnia easily slips into a Rousseauistic mode of evocation to make his point. Everything is "first," "in the beginning," back in a mythic past of perfect linguistic transparency. "The first word is poetry," says Potebnia; this is why poetry precedes all the other arts. "The most perfect words of folk poetry

date back to a time when people would not have been in a condition either to conceive of or to produce anything worthy of the name of paintings or statues." "In the beginning the word and poetry concentrated within themselves the entire aesthetic life of the people" (p. 150).

Poetry precedes not only all other forms of art but all other forms of speech as well. So primal is poetry, in fact, that the "late" division between poetry and prose can be said to arise from poetry itself (p. 152). The distinction comes about as a result of a gradual process of loss over time. Poetic language is language still imbued with inner form, whereas prosaic language develops in proportion to the loss of inner form. Inner form is lost as the objectivity of language (that is, objectivity understood in Potebnia's sense) gives way increasingly to abstraction. With the rise of abstraction in a word comes its increasing distance from sensual perception. This distance, incidentally, is why prosaic language tends to be the domain of science: prose is given to analyzing reality, whereas poetry seizes reality directly in its sensible manifestations, says Potebnia, loosely translating from Humboldt (p. 152).

Potebnia's opposition thus not only sets up poetry in a favored position of mythic purity but also specifically ties aesthetic value to poeticity and, by extension, to the myth of origin. A poetic word is valued in direct proportion to its symbolism, which means its inner form, which means its proximity to a primal state, which means its proximity to its own etymological root. And the corollary is that the more abstract a word, the more detached from its origin, the more intellectual effort required to discern its meaning, the less poetic and, consequently, the less aesthetically valued it is.

The Word Is the Work

Since primal language is poetry, and since poetry is at the root of art, it follows that there is a fundamental analogy between art and language, artworks and words.^[13] Put one way, in fact, there really is no difference between a word and an artwork, since the word (poetic, of course) is art. But the analogy works on the level of specifics, too: the artwork in general contains the same three moments as the word, namely outer form, inner form, and content, and these three moments interact in the name of the same function for the artwork as for the word, namely the creation of thought.^[14]

Thus the word, in its strict sense as a discrete lexical unit of meaning, is a kind of microcosm of the larger artwork of which it might form a

part. As long as the word is a poetic one, the same propositions that are true of the word are true of the artwork. The artwork, by implication, functions like a symbol in the way the word does. And it stands in the same relation to the world as the word does: like the word, it creates something that was not there before, it is energeia rather than ergon, and, as a consequence, it contains more than what the speaker or artist put into it.

In one way this kind of thinking sounds perfectly trite and entirely unworthy of anyone's attention. After all, isn't the lack of distinction between individual lexical units and the broader units of meaning in language more a regressive idea than a progressive one? Ancient cultures did not distinguish sharply between the individual word, of the sort that we expect to find listed in a dictionary, and speech or utterance. In those days (and in oral cultures today, I'm told), the only parts of language that really are perceived as intrinsically discrete are proper names.

But Potebnia is pointing to something that really is new and something that will be prominently featured in the following generation or two. Apart from the mythic aspects of Potebnia's theories, apart from the silly idea that the inner essence of a word is nothing more than the word's Slavic or Indo-European root, Potebnia's conception carries a level of philosophical complexity that merits attention. He sees the act of speaking as a deeply relational act, one that touches each of a complex of coordinates without actually rendering any one of them or being identical to any one of them. These coordinates are things like the word (as lexical item, "articulate sound"), the image or concept, the speaker, and the world, with inner form as a kind of pure (but only vaguely defined) mediating function. That is to say, the word is a structure of relations, and this is the source of its analogy to the larger units of language and, ultimately, to works of art composed of language. Words, phrases, language, poems—all these things are there in our perception as relational complexes. In a literal sense they are nowhere, and yet they truly have the power to create a "somewhere."

The notion of symbol will serve to designate this class of things for Andrei Bely. When he comes along to define symbols, he does so in a way that makes it impossible to determine whether he is talking about words, poems, specific figures of poetic speech, or, for that matter, signifying objects in the very broadest sense of the term. This is not only because Bely has decided to return to an archaic notion of language as logos, where logos can mean, as it did in Greek, a range of things from discrete utterances to language broadly conceived. It is also because Be-

ly's interest transcends the limited philological subject of words as lexical items. He is concerned with a much more modern subject, and that is meaning, specifically its structure and function. But it is Potebnia, standing on the border between the realm of linguistic myth and the realm of modern science and functionalism, who had set the stage. And to judge from what he had to say about his predecessor, Bely knew it.

Bely:
The Value of Formalism and the Formalism of Values

Everything came together for Bely in 1909, if you can call what happened that year "coming together." It came together in the sense that Bely wrote most of his greatest theoretical works in a stupendously concentrated period of activity. He claims to have written his crowning article on the theory of symbolism, "The Emblematics of Meaning," in a week (the article is close to a hundred printed pages long). After writing his trend-setting article on metrics, "Lyric Poetry and Experiment," Bely wrote three accompanying pieces on metrics and his famous "Magic of Words" in a single month, then in two additional months dashed off the two-hundred-odd pages of commentary to his collection Symbolism and reread Potebnia.^[15]

Bely's concern for the previous twelve years or so had been a theory of symbolism, something to define the cultural and philosophical era of which he was a part. Symbolism is the monument to this whole period since in it are collected Bely's most important theoretical pronouncements. But the book has a strangely schizoid character that would appear to make nonsense of any superficial attempt to find a unified vision in the different things Bely says. Bely himself knew this and wrote about it toward the end of his life in his memoirs. This "heap of screamingly contradictory articles," he says of Symbolism, was a reflection of his "stormy and agonizing personal life" at the time.^[16]

It's true that the articles are "screamingly contradictory." There are articles proposing purely formalistic and scientific systems of aesthetics; there are the articles on metrics, which rely on the extensive use of statistics for verse analysis; and there are the difficult idealist articles on theory of symbolism, like "The Emblematics of Meaning" and "The Magic of Words." It's not true, however, that all these contradictions are simply the reflection of Bely's personal life. They are the reflection

of a deep division in Bely's thinking, one that arguably lasted his entire adult life.

Less self-conscious remarks that Bely made about his theoretical writings will show what I mean. For both in the period when the essays were written and later on, in his memoirs, Bely seems confused about the true orientation of his studies. One moment he appears to think that the secret to a theory of symbolism lies in the direction of Kantian epistemology, or what he calls criticism (since Kant arrives at his epistemology through a critique of pure reason), and the next moment he appears to think that it is some sort of religion. In the introductory footnote to "The Emblematics of Meaning," for example, Bely states that the theoretical foundation for a theory of symbolism (a foundation, incidentally, that has not yet been established, he says) is to be found "in the context of a critical reappraisal of the basic epistemological premises concerning reality."^[17] In his memoirs he speaks of the "epistemology of symbolism" and how he had wanted his theory to be critical (in the Kantian sense) rather than mystical.^[18] But in the preface to Symbolism Bely unabashedly writes that symbolism is "a certain religious creed, having its own dogma," that this is the positive side of a doctrine of symbolism whose negative side alone he has given in "The Emblematics of Meaning." He then goes on in a bizarre and rambling paragraph to explain that theory and dogma are irreconcilable because "dogma is the Word become Flesh." About this, he continues, it is best to read in the Gospel according to John (S, p. ii).

Bely has good reason to be confused. His readers have been confused for more than half a century. It is at least to his credit that he later recognized just how chaotic the contents of this book were in subject, method, and, most of all, ideological orientation. My own view, however, is that there is a great deal less confusion in it than even Bely thought. To be sure, it is difficult to reconcile the contradictory things Bely says about his aims in approaching a theory of symbolism. But that difficulty assumes one takes him seriously when he characterizes his own work. As it happens, in spite of the breadth of knowledge that came from his desultory nature and amazing eclecticism, Bely showed an extraordinarily narrow understanding of what he himself did. His theory of symbolism can't be completely defined by being identified as a form of religion any more than it can be completely defined by being described as critical. It has the ability to encompass both forms of knowledge, because for Bely a theory of symbolism is a universal theory of

meaning. It accounts for all types of signifying objects and thus can be used to analyze, say, religious icons as well as words in a poem.^[19]

This universality has to be recognized before any sense can be made of Bely's theory of symbolism. But once it is, then that theory takes on a tremendous number of implications for our study of the theory of language, for a theory of language necessarily becomes a subset of a theory of symbolism. Bely's confusion then falls into place as another symptom of a transitional age. Like Potebnia, Bely was in the process of abandoning the traditional myths but was unwilling to let go entirely.

The essay where Bely presents what might plausibly be characterized as a theory of language is "The Magic of Words." Bely's rhetoric sounds as if he had lifted it directly from the pages of Potebnia, and this is not especially surprising since Bely had reread Potebnia (whom he mentions repeatedly) and written his review of Thought and Language in the same period when he wrote "The Magic of Words." He evokes the myth of origin, for instance, in terms largely reminiscent of Potebnia. "In the beginning (pervonacal'no^[*] ), poetry, the process of cognizing, music, and speech were all one and the same," he says in a passage strikingly similar to the one in Thought and Language where Potebnia asserts that poetry is the first expression of the aesthetic life of a people (S, p. 431).^[20] He refers at one point in the essay to inner form, without mentioning either Potebnia or Humboldt, and speaks freely of representations, images, contents, and all the other terms that Potebnia had established as integral parts of his theory. The entire second part of "The Magic of Words," in fact, is borrowed (in this case with appropriate attribution) from a posthumously published work by Potebnia called Notes on the Theory of Literature .^[21]

The central concern of the essay is Bely's poetry-prose distinction. Poetic speech for Bely is "living" speech, and the thing that distinguishes it from prosaic, or "dead," speech is that it is generative, creative. "The word creates a new, third world: a world of sound symbols by means of which both the secrets of a world located outside me and those imprisoned in a world inside me come to light," Bely says. "The aim of communication is to kindle, through contact made between two inner worlds, a third world that is indivisible for those communicating and that unexpectedly deepens the individual images of the soul. . . . The original victory of consciousness lies in the creation of sound symbols. For in sound there is recreated a new world within whose boundaries I feel myself to be the creator of reality" (S, p. 430; SE, p. 94).

But only the poetic word is creative in this sense. In opposition to it

stands the dead and deadly "word-term." Such is "the common prosaic word," which "has lost all its sound and pictorial imagery." It is a "fetid, decomposing corpse," says Bely. Word-terms come about as the result of a process of abstraction like the one Potebnia describes. As in Potebnia, the process begins in the mythical era with the poetic word and ends with the abstract term. In fact, Bely outdoes Potebnia by presenting the process as a kind of genesis of human knowledge: "The word begot myth; myth begot religion; religion begot philosophy; and philosophy begot the term" (S, p. 440; SE, p. 103).

Although Bely's rhetoric sounds a great deal like Potebnia's, and many of the terms of his discussion are borrowed from Potebnia, there is a significant difference of emphasis. Bely has restored to Potebnia's philosophy of language the geneticist orientation of Humboldt, but with a modification that firmly establishes Bely's modernity. Language is certainly an organic and creative force for Bely, just as it had been for Humboldt. But Bely is concerned more immediately with the individual act of speech; Humboldt had seen the creative force of language at the level of a whole people and a whole language. And because Bely focuses on the individual act of speech, his interest is attracted to the ontological dimension of language and speech. He even uses this term at one point in a discussion otherwise lifted from Potebnia. Speaking of the creation of a metaphor, Bely says that in a metaphor, which joins two images into one figure of speech, a new, third image is generated from the combination of the two original images. This third image, he says, is independent of the two images that "engendered" it because "creation endows it with ontological being independently of our consciousness" (S, p. 446; SE, p. 109). Where Potebnia had sought to account for the representational powers of language, Bely seeks to account for language's ability to generate being.

Even Bely's reformulation of Potebnia's derivation of the poetry-prose distinction shows this difference in orientation. For Bely, it is not enough to point to a historical (or mythico-historical) process that begins with poetry and ends with prosaic abstraction. He insists that the steps of this progress are linked by the generative force of the word: each step begets the next.

Another aspect of Bely's essay shows a reorientation that is distinctly Belyan if not specifically modern. Bely repeatedly uses the word cognition without saying clearly what he means by it. In "The Emblematics of Meaning" he develops at great length a polarity between cognition, by which he essentially means ordinary scientific understanding, and

creation, by which he means the sort of power he attributes to speakers of poetic language in "The Magic of Words." Both are forms of knowledge; hence language for Bely is a form of knowledge. This is why he says at the beginning of "The Magic of Words" that "cognition is impossible without words" (S, p. 429; SE, p. 93). Words are the vehicles of knowledge—word-terms for scientific knowledge (cognition) and poetic words for creation.

Recognizing that Bely's orientation in "The Magic of Words" is toward issues like ontology and theory of knowledge has the effect of knocking the wind out of his visionary rhetoric. In fact, one can easily come around to the view that the magic Bely talks about and the myths he evokes are meant in a figurative sense. Words have the power to generate being by a process analogous to that by which a magician would generate being, if there were such things as magicians. The analogy, of course, is limited to the idea of generating being and does not suggest that the being that is generated is material, as it is in magic.

Why, then, does Bely write in the style he does, and why does he insist on invoking the same old myths of origin as Potebnia? Something strange is certainly happening, and it all looks even stranger in the light of what Bely says about Potebnia in his critique of Thought and Language . There Bely accuses Potebnia of practicing a form of mysticism. He refers to the distinction between geneticist theories of language like Humboldt's, according to which language is always being created in the present, and nativist theories, according to which language was already handed down in perfect form at some point in a mythic past. Because of his regressive reinterpretation of inner form, Potebnia was not the Humboldtian geneticist he considered himself to be, Bely says, but had returned to a mystical, nativist conception of language.

But Bely offers an explanation for this shortcoming on Potebnia's part, and it confirms the view that Bely is using Potebnia's rhetoric in "The Magic of Words" for some other purpose than to propose a mysticism of his own. What Potebnia is really presenting in disguise, according to Bely, is a theory of the value of the word in which the word is considered "from the point of view of its irrational content." The problem for Potebnia was that he allowed himself to become entangled in all kinds of contemporary psychological theories, thus losing sight of what he was really doing. He lacked an understanding of values as an epistemological problem; but then that was not his fault, Bely says, because the problem had not been articulated at the time Potebnia wrote.^[22]

Bely doesn't articulate it here either, but he had made it the basis for

his most extensive statements on the theory of symbolism in "The Emblematics of Meaning." That essay too, like "The Magic of Words," is filled with mystico-mythical rhetoric, but in the end it proposes a formalistic theory of meaning. What are values, and why is their presence in a theory of meaning an indication that that theory is formalistic rather than mystical?

In "The Magic of Words" the emphasis is on the creation of meaning. In "The Emblematics of Meaning" the emphasis is on epistemology, which is to say the reception of meaning. I will speak at length of "The Emblematics of Meaning" in chapter 5. For now, all that needs to be said is that Bely uses the analogy between objects of meaning (a generic category that includes words) and religious icons. Religious icons in the Russian theological tradition are seen as embodying an essence that we might refer to as divine grace. What Bely does is to replace the theological conception of the icon with a secularized conception by substituting "value" for the essence that is embodied in an icon. Value is a term that had gained currency in the recent writings of neo-Kantian philosophers like Heinrich Rickert (1863–1936). In a book called Der Gegenstand der Erkenntnis (The object of cognition) Rickert had sought to redefine Kantian epistemology by asserting that every act of ordinary cognition, instead of merely proceeding according to rules prescribed by the structure of the mind (Kant's Categories, or Pure Concepts of Understanding), contains an element of will, a kind of positive, ethical affirmation of truth.^[23] The object of this affirmation is a value, and a value is simply something that the human subject, again by will, esteems. Rickert's system thus provides a means for eliminating the strict separation that Kant had made between practical reason, which has to do with morals, and ordinary scientific understanding, which has to do with the way we organize experience in the physical world.

Rickert's philosophy served Bely particularly well because even though it blurs the boundaries between scientific understanding and something as nasty and metaphysical as morals, Rickert proposed his own system as entirely free of any metaphysical implications. The valuebased theory of knowledge is a purely formal system, both because values are not metaphysical objects (according to Rickert) and because the focus is on the process of cognizing. Thus when Bely adopts the Rickertian notion of value and integrates it into an otherwise theological structure where it takes the place of divine grace, he too is adopting a formal system.^[24]

Bely's assertion that Potebnia was proposing a theory of the value of

the word and at the same time was considering the word "from the point of view of its irrational content" should now be clear. The problem in Potebnia is the "irrational content." Had Potebnia had the benefit of neo-Kantian value-based epistemology, he could have elaborated a theory that was free from the irrational elements that in fact contaminate what he wrote. And if we now consider Bely's theories in this light, then his own quasi-mystical and religious rhetoric appears all the more either to be the result of some sort of schizoid confusion or, more likely, to serve as an analogy in a structural description of something that resembles, but isn't quite, mysticism and religion.

Bely described the key notion of his theory of symbolism years later in a passage striking for its concreteness. The passage comes from a work with the ungainly title Why I Became a Symbolist and Why I Never Ceased Being One in All the Phases of My Intellectual and Artistic Development, and if ever there was evidence to support the assertion that Bely's theory of symbolism was attempting to be formalistic, it is here. The account Bely gives is credible both because it has the advantage of distance from the symbolist period (Why I Became a Symbolist was written in 1928) and because it was written at a moment when Bely had a spiritual commitment different from the one he professed to have during his earlier period. The purpose of this memoir is to show that the central idea of symbolism, what he calls here "symbolization," was a constant in his career that saw him through such ideologically diverse periods as the symbolist period and the current period, when he claimed allegiance to a system of thought known as anthroposophy.

Symbolization is the process by which a symbol comes into being. To demonstrate the universality of the process and its lifelong presence in his own life, Bely evokes a childhood experience. The whole thing begins with a state of consciousness, in this case fear. Here is how a symbol comes to be:

Wishing to reflect the essence of a state of consciousness (fear), I would take a crimson-colored cardboard box top, hide it in the shadows, so that I would see not the object but the color, and then walk by the crimson spot and exclaim to myself, "SOMETHING PURPLE." This "SOMETHING" was the experience. The purple spot was the form of expression. The two, taken together, constituted the symbol (in the process of symbolization). The "SOMETHING" was unidentified. The cardboard box top was an external object bearing no relation to the "SOMETHING." But this object, having been transformed by the shadows (the purple spot), was the end result of the merging of that (imageless) and this (objective) into something that is neither

THIS nor THAT, but a THIRD. The symbol is this THIRD. In constructing it I surmount two worlds (the chaotic state of fear and the object given from the external world). Neither of these worlds is real. But the THIRD world exists.^[25]

Bely does not use the vocabulary of value philosophy here, but the process is unquestionably the same as the one he outlines in "The Emblematics of Meaning." A vague state is embodied in a physical object, with the result that the physical object becomes a symbol, thus generating a "third" state of being that is independent from both the vague state and the physical object. This is symbolization, and Bely would have us believe that it fits all the different systems of belief (and there were many) that he adopted in his career.

Future generations of critics and poets would ridicule Bely for espousing an aesthetics, and a worldview underlying that aesthetics, that always pointed towards the otherworldly. The poets known as Acmeists, whose most famous representatives were Anna Akhmatova and Osip Mandelshtam, prided themselves on their rejection of the symbolist commitment to mystical ideals. As early as 1913 the Acmeist Sergei Gorodetsky, in his manifesto titled "Some Currents in Contemporary Russian Poetry," describes the "battle" between Acmeism and symbolism as a "battle for this world."^[26] The characteristic attitude of antiaestheticism that so many members of the Russian avant-garde adopted in Bely's day allowed little sympathy for the supposedly mystical orientation of symbolist poets. And, of course, it was part of the polemical posture of Formalist critics like Viktor Shklovsky to reject symbolist notions of mystical and allegorical content in favor of a "scientific" pursuit of functions and techniques in language.

But to call Bely a mystic is entirely unfair because it ignores the most basic characteristic of his thought and his personality. Bely was a man of conflicting temptations, and his favorite thing seems to have been to place himself in various systems of thought, explore all their twistings and turnings, but remain at a distance from the center, where the unwary wanderer might find himself trapped into actually believing something. This is why Bely fits so neatly into our story. His spiritual diffidence, his tendency toward formalism, and his ironist's pose in his "adoption" of different religious and mystical systems of thought all indicate the struggle with myth characteristic of this age. Yet the perpetuation of these systems, even if they present themselves only as temptations, is evidence that Bely's escape is not complete. As far as theory of language

goes, we can see Bely caught in the same conflict as everyone else I've mentioned. In "The Emblematics of Meaning" his theory of meaning is formalistic (this is his "fracture"), but the entire formal edifice of his thought is borrowed from Russian Orthodox theology. And when he comes to talk about poetry in "The Magic of Words," we find him succumbing to the alluring power of the poetry-prose myth just as Potebnia and everyone else before him had.

The Zaumniks

The word zaum' was coined in 1913 by Aleksei Kruchenykh (1886–1969). It is a difficult word to translate—for two reasons. The less important is that there is no English word that renders clearly what it means. "Transrational language" is often used, since za - corresponds to trans - and -um ' has to do with mind or reason. But "transrational" has a technical, philosophical ring to it that is entirely missing in zaum', and since no one has come up with a more plausible English equivalent, many writers simply use the Russian word. The more important reason is that zaum' was used to mean more than one thing.

The principal exponents of zaum' language theory are Kruchenykh and Velimir Khlebnikov, both commonly classified as members of the avant-garde Cubo-Futurist group. Kruchenykh and Khlebnikov show the transitional stage of thinking that is typical of the era. They are at home in an Edenic world of mysticism—much more than Bely is—yet we see in them a pull toward the modernist "fracture."

Their pronouncements on language can be divided into two broad categories. In both versions of zaum' theory the constant factor is that zaum' language is distinct from any existing language and is ideally referential, which is to say that it shows an ideal correlation with the things it designates. The difference between the two versions is one of orientation.

The first version of zaum' theory is an old-fashioned nativist theory of language. Poetic language is seen as being motivated directly by the objects it designates. There is thus an intrinsic connection between any object and a set of sounds that will express that object. In his first zaum' manifesto, "Declaration of the Word as Such" (1913), where the word zaum' was introduced, Kruchenykh invented a word for "lily" composed entirely of vowels since he felt that in this sequence of sounds the object's

"original purity" was restored.^[27] In an essay called "The Simple Names of Language" (1916) Khlebnikov listed words beginning with the consonants m, v, k, and s, hoping to demonstrate that the words in each group have an idea in common and that this idea is expressed by the intrinsic quality of the initial sound. Thus words beginning with m often signify "the smallest members of certain varieties," an initial v often denotes the "action of subtraction," words starting with k often have to do with death, and so forth.^[28] If one envisions the sign-signified problem as a polarity between word and world, then in this version the point of inception is the world since objects in the world motivate the formation of words appropriate to them.

In the other version, things are reversed, and the point of inception is the word. In a manifesto called "New Ways of the Word" Kruchenykh points out that in the archaic conception of language, thought takes precedence over words. He feels this view is wrong and has undertaken to correct the error. "Until now," he says, "it has always been asserted that 'thought dictates laws to the word, and not vice versa.' We have pointed out this error and come up with a free language that is both transrational (zaumnyj ) and universal. Previous artists have proceeded from thought to the word, but we proceed by means of the word to direct comprehension."^[29]

If words precede thought and comprehension, it is because they create meaning, indeed even surpass it. As the theory comes to be elaborated in the writings of Khlebnikov, language creates its own worlds through the same necessary sign-signified correspondence that exists in a nativist conception of language, but with the obvious difference that the signified comes into being only at the moment it is named. This is where Khlebnikov's version of the poetry-prose distinction comes in. In more than one place Khlebnikov distinguishes between the function of a word when it serves reason and its function when it serves poetry. "The word lives a double life," he says in an essay titled "About Contemporary Poetry." "Either it simply grows like a plant, putting forth a cluster of neighboring sound stones, and then the principle of sound lives a self-spun life, while the portion of reason named by the word remains in shadow; or else the word goes over into the service of reason, and sound ceases to be almighty and autocratic; the sound then becomes a 'name' and obediently fulfills the commands of reason."^[30] In another essay, titled simply "About Verses," Khlebnikov likens poetic language to magical invocations, saying that in the case of invocations "the demand

may not be made on the word, 'Be easy to understand like a billboard.' "^[31] The poetic word thus shows a certain density, whereas the ordinary word is transparent.

The creative power of poetic language is not exactly the same thing for Khlebnikov as it is for Bely in "The Magic of Words." Language certainly created its own worlds for Bely, but those it created were recognizable worlds marked by the subjectivity of the speaker. Bely's chief concern in asserting the creativity of language was to emphasize the existential autonomy of meaning. Khlebnikov has in mind the creation of a realm of existence that has nothing to do with the one we are familiar with. In his notion language generates objects and worlds that not only never existed before but were never even conceived of. Bely's magic is thus a form of ontogenesis, whereas Khlebnikov's zaum' is a form of mythopoiesis.^[32]

Needless to say, when Khlebnikov and Kruchenykh put their ideas into practice and actually create transrational words, the results are predictably unrecognizable and look like little more than gibberish. But this is not really a problem for us, even if they took all their zaum' poetry seriously. The theory simply has to be seen as idealist, like Bely's, and not designed to produce practical results.

Why view zaum' theory as anything but madness or, at best, regressiveness? In both versions it is clearly mystical, as Kruchenykh himself was not ashamed to point out. In "New Ways of the Word" he characterizes the poetic word in general as transrational (zaumnoe ) and then, by way of explaining the term, adds "mystical" in parentheses, along with other adjectives.^[33] But there is something modern and even philosophically interesting about zaum' . The mystical leap from signifying word to signified object (an object that is actual in the first version and mythical in the second) leads to the assertion that poetic language is objective and autonomous. The phrase that the Futurists used to speak of the autonomy of language was similar to the one Bely used in his critique of Potebnia. The Futurists talked of "the word as such" (slovo kak takovoe ); Bely had used the phrase "mysticism of the word itself (mistika samogo slova ).^[34] The word "as such" has an existence all its own in both versions of zaum' theory. In the first, nativist version, word and object exist in a relation of solidarity, owing to the perfect, intrinsically motivated referentiality of the word. The word thus shares in the existential autonomy of the object. In the second version it is autonomous because it is the existential point of departure for the mythic "concrete" reality it creates, this reality itself being seen as separate and

independent. This is where the modernist fracture modestly appears in zaum' theory.

The Early Shklovsky, or How It All Becomes Official in the Work of an Actual Critic

What does language theory of this sort look like when it turns up in the work of a critic of the same era? Viktor Shklovsky (1893–1984) is a perfect example because he shows just how intimate the relations were between poets and critics in this age. In fact, when you place Shklovsky next to writers like Bely and Khlebnikov, you begin to wonder whether the distinction between poet and critic (or, for that matter, between symbolist, Futurist, and Formalist) is all that meaningful. Shklovsky was not a poet, but he was not exclusively a critic either (he wrote novels).

Like so many others, Shklovsky saw himself as a demystifier and a demythifier. Among his earliest writings are an article on Potebnia and one on transrational language, both published for the first time in 1916. They are probably more interesting for what they say about Shklovsky than for what they say about their subject because they both show Shklovsky implicitly defining himself in relation to the recent past and the present. The past and the present for him are exactly the intellectual context I have just been talking about.

The article "On Poetry and Transrational Language" is usually seen as Shklovsky's defense of the theories of the zaumniki . If defending the zaumniki is what Shklovsky thought he was doing, then he certainly went about his task in a strange way. More than half the article is devoted to listing examples of transrational language from traditional literature, where, Shklovsky shows, the authors in question had no intention of writing in anything like transrational language. The point is that transrational language exists (here he apparently is lending support to the Futurists' theories), but not only in its pure form. For the most part, in fact, it exists "in a hidden state, as rhyme existed in ancient verse—alive, but something one was not aware of."^[35] Where it does appear in pure form is in the language of mystical sects and in glossolalia. Perhaps this assertion can be construed as validating the Futurists' idea of zaum' by demonstrating that it existed before the term had been invented. At the same time, however, it has the effect of devaluing their discovery. After all, the zaumniki were not claiming to have found something that was already in plain sight; they were claiming to have either

invented or discovered something that no one else had ever known about. When Shklovsky says it was there all the time, but not necessarily in its pure form, he shows that the zaumniki were doing nothing more than supplying a name for an existing phenomenon.

Besides, the whole thrust of the article seems to be toward a theoretical point independent of what the zaumniki were doing. Shklovsky's notion is that language and poetry are mutually determining things, that one can no more say poetry is a phenomenon of language than one can say language is a phenomenon of poetry. Transrational language in its pure state demonstrates this because of the way in which its unrecognizable speech "wants to be language."^[36] Shklovsky's thinking goes well beyond anything the zaumniki had done. In its emphasis on the perceptual aspect of language—that is, the tendency of the listener to take sound combinations that do not exist in his or her own lexicon and assimilate them into a structure that is somehow on the verge of meaningfulness—Shklovsky's article looks ahead to the psychoperceptual orientation his work will start to assume in the article "Art as Device."

The article on Potebnia is more aggressive. The curious thing about it is that Shklovsky's critique of Potebnia is based on a misunderstanding and is used to assert a position that is not much different from Potebnia's. Shklovsky outlines his predecessor's theory of the poetic word, referring to the tripartite structure of outer form, image (or inner form), and meaning. He then points out, rightly, that for Potebnia the imaginality (obraznost' ) of a word means its symbolism, which is directly proportional to its poeticality. Shklovsky disagrees. In his opinion what distinguishes poetic language from prosaic language is not images but something he calls the palpability (oscutimost')[o&!;&!;utimost'] of its construction." The poetic image is only one of several means for rendering the construction of language palpable: language can be made palpable as well in its acoustical, pronunciative, or semasiological aspects.^[37]

In fact, as Shklovsky saw it, Potebnia was so concerned with images that he committed a cardinal error, which was to argue away the importance of external form. "From the position that 'the clarity of a representation or its absence (that is, the imaginality of a word) cannot be made out in its sounds' and that 'imaginality equals poeticality,'" says Shklovsky, "Potebnia draws the conclusion that the poeticality of a word cannot be made out in the sounds of that word, that external form (sound, rhythm) may be left entirely out of account in a definition of the essence of poetry."^[38] Victor Erlich thinks that Shklovsky may have learned about Potebnia from secondhand sources.^[39] The suggestion ap-

pears to make sense, for it is difficult to recognize Potebnia in Shklovsky's critique: no one who has read Thought and Language can possibly think that Potebnia ignores outer form since he devotes many pages to the subject of the relation between the outer form and the inner form of a word. Still, Erlich's statement is puzzling since in the critique of Potebnia Shklovsky cites page numbers from Notes on the Theory of Literature . In addition, there is evidence to show that Shklovsky in one of his earliest articles appropriated words and phrases from Potebnia (without citing his source). There is no question, then, that Shklovsky knew Potebnia firsthand; the only question is whether he knew more than a few pages of his work.

Let's return for a moment to the article on Potebnia. Shklovsky's first reproach, that Potebnia was concerned only with images and not with the palpability of language, is just as unfounded as the second, although it is much easier to see why Shklovsky thought this. In fact, it is a flaw in Potebnia's system that makes Shklovsky's statement inaccurate. For what Shklovsky really wishes to see is a doctrine of the self-valued word or word "as such," a doctrine by which poetic language distinguishes itself from prosaic because it calls attention to itself as language . But isn't that exactly the doctrine Potebnia has unwittingly subscribed to, as Bely pointed out in his shrewd critique? Isn't Potebnia's "mysticism of the word itself," as Bely calls it, the very thing Shklovsky would like to see, though he certainly would not like to see it called mysticism?

He also would not have liked to see it be mysticism. But here is where Shklovsky's critique of Potebnia is most interesting. One commentator has suggested that the most important concept of Shklovsky's early career, the concept of ostranenie, arose as a direct result of his misunderstanding of Potebnia.^[40] The term is introduced in the 1917 article "Art as Device," where Shklovsky again brings up Potebnia, referring to his theory slightingly as one according to which art is "thinking in images," and then goes on to give a more complete account than previously of the distinction between poetic and prosaic language. Since it is the business of art to give us a perception of things that allows us to see them, not just recognize them, an must use the technique of "making strange" (ostranenie ), showing things in a new and different way, so that the perceiver "sees" them as if for the first time. Poetic language, which is the medium of verbal art, has the responsibility of promoting this type of vision, and it does so above all if words themselves are used in surprising and different ways, so that they call attention to themselves. Thus poetic language differs from prosaic in that prosaic language has become

automatized; its constituent words are not noticed in and for themselves but simply exist to point to their objects and then disappear.^[41]

In "Art as Device" Shklovsky has dressed up his ideas in the garb of scientific respectability by discussing language in the context of something that sounds like perceptual psychology. But what he is presenting is not fundamentally different from what he had said three years earlier, even before the Potebnia study, in an article titled "The Resurrection of the Word." There Shklovsky had made the same distinction between recognizing and seeing as modes of perception associated, respectively, with prose and poetry. The same appeal is made to restore the word to a position of value by using words that call attention to themselves as words. Hence the "resurrection" promised in the title. What is striking about the article, however, is how much Shklovsky's style resembles that of Potebnia and Potebnia's successors. Yet this is not surprising, because Shklovsky, without telling us, has borrowed a great deal of material for his article from Potebnia and other sources close to Potebnia.

The basic ideas are the same as in Potebnia. Potebnia had seen the process by which poetic language becomes prosaic as one of increasing abstraction. So does Shklovsky. Potebnia had talked about the objectivity of poetic language, meaning its nearness to sensual perception. Shklovsky speaks of the disappearance from consciousness of everything that is habitual, including prosaic language. And, most important of all, Potebnia had based the whole thing on a myth. Again, so does Shklovsky. Here is how Shklovsky begins his article:

The creation of words is man's most ancient [drevnejsim^[*] form of poetic creation. Today, words are dead, and language resembles a cemetery, whereas the word that had just come into being was imaginal and full of life [zivo^[*] - obrazno ]. Every word in its basis [osnova ] is a trope. For example, the word "moon" [mesjac ]: the original [pervonacal'noe][pervona&!;al'noe] meaning of this word was "measurer." . . . And often, when one succeeds in reaching the image that had formed the basis of a word but that has been lost, obliterated, one is amazed at its beauty, a beauty that once was but is no more.^[42]

Sound familiar? It should, because it is the same myth of original purity that Potebnia had found himself irresistibly drawn to. The same words and themes are there. There is talk of origins and tremendous separation in time. Shklovsky even uses the favored adjective to describe the linguistic Eden, pervonacal'nyj^[*] (from pervyj, "first," and nacalo^[*] , "beginning"), Potebnia had used this word repeatedly in his own evocations of the myth of origins (as had Bely).

A great deal of the material Shklovsky uses in "The Resurrection of the Word" is borrowed, and an examination of his sources shows how dependent he was on the Potebnian myth. In one instance he borrows from exactly the same page of Potebnia's Notes on the Theory of Literature as Bely had done in the second part of "The Magic of Words": the introductory section to Potebnia's lengthy discussion of literary tropes. Shklovsky casually mentions the commonly used metaphor "foot of a mountain" (podosva^[*] gory ),^[43] but he does not tell us that this is the very same example that Potebnia had used to illustrate the notion of metaphor.^[44] Bely's adaptation of that page from Potebnia had included an analysis of an expression containing the word mesjac (moon) (S, pp. 443–46; SE, pp. 106–8); mesjac is the first example Shklovsky gives of a word whose original (Potebnia would say etymological) meaning is more poetic than its current one. Perhaps Shklovsky was borrowing from Bely, too.

Most of Shklovsky's borrowings, however, are from another nineteenth-century linguist, a contemporary of Potebnia named Aleksandr Nikolaevich Veselovsky (1838–1906). In 1895, four years after the death of Potebnia, Veselovsky had published an article titled "From the History of the Epithet."^[45] This is where Shklovsky gets a great many of the examples he uses to illustrate his point about poetic language.^[46] For instance, in a discussion of poetic figures that through continued use have lost their vividness, Shklovsky mentions the epithets in the common Russian expressions solnce jasnoe (bright sun), belyj svet (white world), and grjazi topucie^[*] (mucky mud), all of which appear in Veselovsky's article as examples of what Veselovsky calls tautological epithets.^[47] When Shklovsky then talks about epithets that have become so habitual that they are used even in contexts where they don't make sense, he uses the following examples: the "white hands" (belye ruki ) belonging to a Moor in a Serbian epic; "my true love," used in English ballads of both faithful and unfaithful lovers; the "starry sky" to which Nestor, in the Iliad, extends his hands in broad daylight; and a fragment of folk poetry, which Shklovsky misquotes.^[48] All examples come directly from Veselovsky.^[49] This second group of examples in Veselovsky occurs in a discussion of what Veselovsky terms the oblivion of the real sense (zabvenie real'nogo smysla) of the epithet. Oblivion results in petrification (okamenenie), the phenomenon where an expression has become so hardened through habitual use that it survives in the most contradictory of contexts for the simple reason that it is no longer noticed by speaker or

listener. These are really the same concepts Shklovsky is using in "The Resurrection of the Word," except that Shklovsky prefers terms like renewal and rebirth .

But there is more to this story, and it shows that something strange is going on. Most accounts of nineteenth-century Russian linguistics and philology make a great deal of the difference between Potebnia and Veselovsky. Veselovsky is usually cited as the less psychologically inclined of the two, the more inductive, the more concerned with depersonalized facts of literary production. Since he sought to deny the importance of individual creativity in literary production, viewing it as a vast, historical process definable in terms of "formulas," he is seen as the precursor to the "antipsychological" tendencies of the early Formalists.^[50] And since Shklovsky has borrowed most heavily from Veselovsky, we might be tempted to think that his theoretical inclination would be toward the more sober and inductive methodology of Veselovsky rather than toward Potebnia's image-based, psychological system.

The odd thing, however, is that Veselovsky has apparently done some borrowing of his own, for in the pages Shklovsky has drawn from, virtually every example Veselovsky cites in support of his discussion appears also in precisely the same few pages of Potebnia we have been talking about. Some examples thus appear in all three writers: the passage from the Iliad describing Nestor, the "white hands" from the Serbian epic, and "mucky mud," to cite only three.^[51]

If Veselovsky's examples are the same as Potebnia's, it seems to be because in this essay his concepts are the same, too. Potebnia, on one of the pages from which Veselovsky has borrowed examples, speaks of certain epithets that are tautological, then mentions the oblivion (zabvenie ) of the representation (predstavlenie ) and its renewal.^[52] The next page contains Potebnia's list of permanent (postojannye) epithets. Veselovsky, too, refers to the permanence (postojanstvo) of certain epithets. The point is that the subject is the same in all three writers. All three are speaking of a central, vivid element that a word loses as it is subjected to continued use. In the pages in question Potebnia calls the element the representation (predstavlenie ) and links it up with the word's etymological root; Veselovsky calls it the word's real sense; and Shklovsky, unabashedly using the terminology of Potebnia, calls it the word's inner form.

Thus three years after "The Resurrection of the Word," when Shklovsky comes to talk of ostranenie as a technique (priem ) for restoring the

vitality of language, his point of departure is the same as in the earlier essay, namely, the feeling that ordinary language has suffered the loss of a vital element. The solution is really the same too, namely, the proposal of a means for restoring the capacity of the poetic word to make things concrete and visible instead of just recognizable.

Does all this mean that Shklovsky has nothing new to say and is merely a helpless victim of an age-old mythical tradition? By no means. As for Bely and the avant-garde theorists, what is new is largely a question of emphasis. Shklovsky's insistence on the distinction between seeing and recognizing suggests a move away from grounding the old poetry-prose distinction in language itself. The Formalist critic in 1917 is distinctly uncomfortable with a theory of language that unhesitatingly ascribes the difference between an everyday form of language and an aesthetically valued one to something contained in language. Much more solid is the notion that the difference is located in the mechanisms by which the subject perceives language.

After Shklovsky the field will open up to all kinds of new methodologies purporting to help define poetic language in the most objective and nonmystical of terms. Other Formalist critics will wrestle with the problem of defining verse language. Above all, there will be Jakobson's definition of poetic language in terms of the distinction between figures of speech: metaphor, where concepts are substituted for each other along the "vertical" axis of selection, and metonymy, where the substitution takes place along the "horizontal" axis of combination.^[53] The later theories are surely not free from the charge of mysticism that can be leveled at the Russian tradition inaugurated by Potebnia. But there is no doubt that for all his tendencies to move the emphasis from language itself to perceptual phenomena in the listener, Shklovsky is still very much in the grips of the old tradition. The tendencies are just that—tendencies. And Shklovsky is perfectly at home in the milieu I've been describing, just as Jakobson will be a few years later, having partially broken with the past but still perpetuating the myth about himself that there is no myth.

Chapter Three—
Mallarmé and the Elocutionary Disappearance of the Poet

To judge from what he wrote, Stéphane Mallarmé never made up his mind whether he was a mystic. There is no denying the existence of a cult of poetic art in him, nor is there any mistaking the tone of almost religious awe when he comes to speak of poetry. But there is also the persistent hint in Mallarmé's writings that he is not entirely serious, that he views his own idolatrous attitude with considerable irony.

One measure of his idolatry is his tendency to liken art to religion. The artist becomes a privileged figure enjoying powers similar to those of a priest or magus. The early "L'art pour tous" (Art for everybody) shows Mallarmé fully in the grip of this idea. "Every sacred thing that wishes to remain sacred," Mallarmé begins, "envelops itself in mystery. Religions take refuge in arcana that are revealed only to the predestinate: art has its arcana, too."^[1] Of all the arts, poetry has suffered the most from a democratization that has left it unhappily accessible to the vulgar masses. Only poetry is "without mystery against hypocritical curiosities, without terror against impieties," and Mallarmé nostalgically recalls the "gold fasteners of ancient missals," the "inviolate hieroglyphs of papyrus scrolls" (OC, p. 257). "L'art pour tous" appeared in 1862, five years after the publication of Baudelaire's Les fleurs du mal, and it is not surprising to see the young Mallarmé mention that work and adopt the aggressive, antidemocratic tone of its author. The poet soars over the vulgar masses, like Baudelaire's ungainly albatross. "Man may be democratic; the artist splits in two and must remain an aristocrat" (OC, 259).

Later on, in his last years, Mallarmé takes a more cynical view of art and the artist. Perhaps it is out of despair at seeing that his ideal of a sacred art of poetry has not been realized in this life. Perhaps he has simply arrived at a more hardened, demystified view of the whole subject; it's difficult to say. But the tone is unmistakably different from what it had been thirty-odd years earlier. In one of the four pieces collected under the general title "Quant au Livre" (Concerning the Book), after a convoluted passage in which he considers and then rejects the notion of suicide or abstention from writing, Mallarmé makes these comments:

Apart from headline news entrusted with spreading a faith in the everyday nothingness, unskilled if the scourge measures out its own period as a fragment, significant or not, of a century.

So look out for yourself and be there.

Poetry, consecration; that attempts, in chaste crises in isolation, during the other gestation in progress.

Publish.
(OC , p. 372)

This is Mallarmé at his tortuous best. What exactly does it mean? All that really concerns us now is what it means in its tone, and the tone is unquestionably less earnest than it had been in "L'art pour tous." It had become Mallarmé's custom, when he sought to portray the least exalted side of a profession in letters, to resort to wry descriptions of it as a commerce, both in the literal sense as the marketing of books and in the figurative sense as the marketing of meanings. Mallarmé enjoys depicting the man of letters as a shameless pander to the whims and financial needs of the moment. He characterizes his craft by using vocabulary drawn from the world of newspapers and money, treating the sacred side of literature with a contempt that is almost masochistic. Thus the glib mention of poetry as consecration is followed with the blunt, sarcastic command to publish, that is, to produce literature for consumption.

There is a significant analogy between Mallarmé's attitude toward art and his attitude toward language. His attitude toward art expresses itself as a tension between religious idolatry and skepticism. When Mallarm>CH:233> turns to the topic of language, he shows the same tendency, but without regularly having recourse to the vocabulary and imagery of Catholicism. Instead, the tension turns up as the same one we have seen in the Russian theorists. It is the tension created by the twin temptations

of linguistic mysticism—a kind of naive Cratylism, on the one hand, and modern scientific conventionalism, on the other. We can see these temptations primarily in two texts: the relatively early Cratylic fantasy called Les mots anglais (English words), written in 1875 and published in 1877, and the later "Crise de vers" (Crisis in verse), a patchwork of material written between 1886 and 1895.

English Words and the Game of Cratylism

One sign that something is amiss in English Words is that the text is far longer than any other published work by Mallarmé. The French have a fondness for starting at the absolute beginning of a subject and proceeding by a kind of irresistible logic to the heart of the matter. Even French cookbooks open with an essay devoted to the question, Why do we nourish ourselves? But not Mallarmé. Ellipsis, truncated syntax, and a kind of twisted economy of expression were his trademarks. Few things he wrote in prose were more than several pages long. So it is with considerable suspicion that the reader must regard a treatise that opens in the following manner and then continues for almost two hundred pages in the standard edition: "What is English? A serious and lofty question; treat it in the sense in which it is posed here, that is, absolutely—one simply cannot until the last of these pages, once all has been analyzed. For the moment, it behooves us to answer, keeping in mind the external and commonly noted characters of English, that this idiom is one of those in the world that a contemporary must know" (OC, p. 899).

Maybe Mallarmé is simply making fun of the very tendency that leads his compatriots to write introductions to cookbooks. After all, this first paragraph doesn't give the kind of answer a reader would expect in a learned philological discussion, and in fact it doesn't really make much sense. Besides, it isn't the true logical point of origin. Mallarmé poses a more basic question two pages later, and his answer to this one is hardly more satisfying than his answer to the first: "What is Language, among scientific materials to be studied? From each of them, Language, entrusted with expressing all the phenomena of Life, borrows something; it lives: and, since (to help childhood grasp) it is unavoidable that the outside world should lend its images, any figure of discourse, relative to this or that manifestation of life, is fine to use for speaking about language" (OC, p. 901).

Much of what Mallarmé writes in English Words is borrowed from the works of various nineteenth-century linguists and philologists. The work that served as the source of the bulk of his historical observations on English is called The Philology of the English Tongue, written in 1873 by a man named John Earle (1824–1903). Mallarmé's "borrowings" are so extensive, in fact, that scholars today don't even speak of his text's originality; rather, they find themselves proposing explanations for the cases where Mallarmé merely altered Earle's text slightly instead of shamelessly translating it word for word. Not that he is trying to hoodwink us: any reader can easily discover Mallarmé's heavy reliance on his sources. Using an old-fashioned scholarly study for so much material undoubtedly serves the same purpose as the formulaic Cartesian opening. Both create the illusion that the author is presenting a traditional treatise on language of the sort that might contain a thesis like the one Mallarmé advances.

And what of that thesis? Here is how Mallarmé states it: "What it behooves us to realize now seems to me to be the relation that exists between the meanings of words that I will assume to be unknown to you, and their external configuration" (OC, p. 918). Where does this relation emerge? As he explains a little later, it emerges in the beginning of the word, in what he calls the "attack" (OC, p. 926). An intrinsically meaningful opening sound is common to languages of the north, Mallarmé explains. For we mustn't forget that the close sound-sense relation is something that English has in far greater measure than French: hence the need to write a book about English words and not about French, or Hottentot. Mallarmé's notion is extremely simple. It is that in a great many cases the opening sounds of an English word give an indication of its meaning—though not precisely that opening sounds contain in themselves a clue to a particular meaning. In other words, there is nothing magical about, say, an initial b sound that causes it to signify a certain thing or concept. It's just that in English certain initial sounds commonly denote the same thing or concept. Mallarmé wants us to believe that the force of his theory is in frequency of correlation, not in a mystical explanation. Thus his method: "to group and eliminate" (OC, p. 918).

Once the logical foundation for the essay has been established, Mallarmé can proceed to the principal task of his enterprise: to make lists. His lists are arranged by initial letter and consist first of the group of words that belong to the same "family" and then of the group of "refractory" words that do not—"isolated words," as he calls them. Are the isolated words a problem for Mallarmé's theory? Not at all. "Noth-

ing could be more practical," he says, than to isolate the refractory words. "Nor could it be more in agreement with the theory of a Language, or with intelligent mnemonics. Separated after having so often come together since their common origin, these words now succeed in coming together once again, owing to your reflection, in a state of the Language treated with order" (OC, p. 922).

What this means exactly I hesitate to say, but perhaps it doesn't matter much. The essential thing in each list is the group of words belonging to the same family, and Mallarmé now offers his lists up in the section of his text called "Table." Here is where it is most difficult to take Mallarmé seriously or to believe that he took himself seriously. Under the letter b, for instance, we read that words in this group have meanings that are "diverse, yet secretly connected." The meanings have to do with "production and giving birth, fecundity, amplitude, swelling and curvature, boastfulness; also mass or boiling and sometimes goodness and blessing" (OC, p. 929); hence the common thread in baby, back, bat, bear, beech, beck, bell, bend, bind, better, bet, bid, big, bite, black, blend, blink, bless, block, blot, blow, and so on (OC, pp. 926–28). Further on we read that m (may, make, mash, maze, meet, melt, merry, mid, milk, mildew, mingle, moon, moor, morn, mow) "translates the power to do, thus joy, male and maternal; also, through a meaning that has come down to us from far in the past, measure and duty, number, meeting, melting and the middle term: and, finally, by a turnabout less abrupt than it might appear, inferiority, weakness or anger" (OC, p. 960). And k (keg, kedge, kin, kind, king, kill, quell, knit, knot, knop, knob, knuckle) "generally carries the sense of knottiness, knuckle, etc., but only by allying itself with n and becoming silent for the benefit of this nasal. Note also the group kin, kind, king, from which a notion of familial goodness emerges" (OC, p. 941).

This last group, because it is so small and because so many of the words are etymologically related to each other (as Mallarmé knows), shows better than many others the absurdity of Mallarmé's claims. He might just as well have said that k in English mysteriously unites the meanings of kegs, kedges, kin, kind, kings, killing, knitting, knots, knops, knobs, and knuckles. But of course Mallarmé is writing in French, which allows him to disguise what he is doing; for when he uses French words to give the common ideas from a list of English words, the French words do not resemble each other. Keg, kedge, kin, and kind, for instance, in Mallarmé's translation are caque, flotteur d'ancre, parenté, and familier et bon . Translate these French words into English,

however, and you come up with a list of words for common ideas that looks suspiciously like the original list of words from which the common ideas were extracted. The list of common ideas, to be sure, is shorter than the original list of words, but usually because the original list contains words that are etymologically related. Hence there is nothing amazing in their beginning with the same letter or, for that matter, in their having a common meaning.

What do we make of Mallarmé's linguistic speculations? Certainly not that he meant to be taken at his word. Jacques Michon, who wrote a full-length book on English Words, says that in the debate that has been carried on since Cratylus, Mallarmé places himself squarely on the side of the Cratylists. But he adds that Mallarmé is "a disenchanted occultist, for whom absolute language cannot exist."^[2] Edouard Gaède is a little more skeptical, but in being so he shows more faith in Mallarmé. English Words, he says, though it pretends to be a scientific work, is in reality "a vast poem on the nature of language." There can be no sincere talk of naivete in Mallarmé's absurd etymologies, Gaède feels, because Mallarmé's whole enterprise is a fictional one. Mallarmé "sets out to complete, fictionally, the jumble of scattered data that these [etymological] dictionaries offer, by placing the data in a succession that would introduce a law of continuity and thus a principle of intelligibility." In fact, Gaède says, Mallarmé's real point is to propose a theory of language that is thoroughly at odds with traditional Cratylism. The implicit theory of language that emerges from Mallarmé's classifications is one according to which language renders not objects but "a certain plan as a function of which the real takes shape." Gaède thinks that the very fact Mallarmé's work is a fiction signifies that for him language is not a representative, but rather a productive, medium.^[3]

This last statement, in my opinion, is excessive. But there can be no doubt that Gaède is right about the fictional quality of the project. Whether Mallarmé is trying to be funny, or whether he is simply giving expression to a kind of Cratylist nostalgia, English Words is a document that exists in the moment of hesitation between an older, mythical view of language and a newer, more skeptical one.

"Crisis in Verse"

The facts that shine through the irony or fiction of English Words —namely, that there is no necessary correspondence between the sound

of a word and its meaning in French, English, or any other language, that the sound-sense relation is arbitrary and capricious—serve as the logical point of departure for the more somber and skeptical view of language that Mallarmé proposes almost two decades later in "Crisis in Verse." But if these facts show that the system in English Words is a lie, and if they thus precipitate the crisis that Mallarmé refers to in the title of his later essay, they also serve as the basis for a hopeful resolution to that crisis.

The section of Mallarmé's patchwork essay in which he presents his view comes from a piece he published in Variations sur un sujet (Variations on a subject) in 1895. The point of departure is two marvelously simple perceptions:

Languages being imperfect in that they are several, the supreme one is lacking: since thinking is writing without accessories, or whispering but the immortal word still being tacit, the diversity, on earth, of idioms prevents anyone from uttering words that otherwise would turn out to be, by a unique stamp, materially truth itself. . . . Next to ombre, opaque, ténèbres darkens little; what disenchantment in the perversity that bestows upon jour as upon nuit, contradictorily, a timbre that is dark in the former case, light in the latter.
(OC, p. 363–64)

Mallarmé is making the same point that Humboldt made in his principal work on language. Both refer to the fact of the diversity (Verschiedenheit, diversité ) of human languages. Any theory of language that pretends to see a necessary correspondence between sound and sense is contradicted by the simple fact that there are so many languages and that each has its own word for any one thing or concept. Otherwise we would be able to utter words that "would turn out to be, by a unique stamp, materially truth itself." That is the first point. The second applies to an individual language considered apart from "the diversity, on earth, of idioms." Take a language like French, for instance, and put it to the test to see if there is a necessary connection between the sounds of its words and their meanings. You will soon come upon examples like the one Mallarmé mentions: jour, which has a dark sound but a light referent (day), and nuit, which has a bright sound but a dark referent (night). As Mallarmé well knows, this single example is sufficient to demonstrate the fallacy of supposing with Cratylus that even in a single language there can ever be an intrinsic correspondence between sound and sense. And as Mallarmé undoubtedly also knows, having demonstrated the fallacy of the sound-sense correspondence, he has demonstrated the fallacy of English Words as well. For even if English did show a statistically high

correlation between words with certain initial letters and a small group of related concepts (and Mallarmé has not shown that it does), this would still not be enough to establish the inevitability of the correlation. Either language is magic or it's not; the existence of even a few exceptions shows that it's not. English Words, filled as it is with faulty methodology (intentionally, no doubt), really demonstrates the same thing as "Crisis in Verse," only ironically.

But that's only the crisis in language; the essay is about a crisis in verse . Here is how Mallarmé makes the transition. Immediately after the last phrase I quoted he says, "The wish for a term of brilliant splendor, or that it should be extinguished, the opposite; as for simple, luminous alternatives—Only know that verse would not exist : it, philosophically compensates for the shortcoming of languages, superior complement" (OC, p. 364). This appears to be Mallarmé's crabbed way of saying that if our wish for a universal match between bright words and bright ideas, dark words and dark ideas, were to come true, then we would no longer have verse. Verse owes its existence to this patent shortcoming of language. This is certainly a novel idea, particularly since other, contemporary poets and thinkers asserted the distinctness of verse by claiming for it the mythical, Cratylic condition.

But Mallarmé has a somewhat more imaginative notion of (verse) language than many of his contemporaries. In fact, what he has done is to invert the poetry-prose distinction of his contemporaries and successors by claiming a Cratylic aim for prose and a different notion of language for verse. Here is how he introduces the distinction, in a characteristically complicated and humorous passage:

Abolished the pretension, aesthetically an error, even though it governs masterpieces, to include on the delicate paper of a volume anything other than, for example, the horror of the forest, or the silent, scattered thunder in the foliage; not the intrinsic and dense wood of the trees. A few spurts of deeply felt pride veraciously trumpeted abroad arouse the architecture of the palace, the only one fit to live in; apart from any stone, on which the pages would close with difficulty.
(OC, p. 365–66)

It's not just that Cratylism is false in the real world; it's that, even if it were true, it would be an aesthetically wrong concept. In the privileged language of poetry, language mustn't strive to render the forest—that is, the trees themselves, or the palace—in its physical reality. After all, Mallarmé quips, we wouldn't be able to close the book on all those big stones. No, poetic language instead must strive to communicate intan-

gible things, like the horror the forest inspires or the mute thunder—not the actual crashing thunder, but the residue of feeling it leaves behind in the trees.

Ordinary language doesn't function in this way. It functions, so Mallarmé's favorite joke goes, like money in a commercial exchange: "Narrate, teach, even describe, that's fine and even if it were enough for each of us perhaps, in order to exchange human thought, to take from or place in the hand of someone else a coin in silence, the elementary use of discourse serves the needs of universal reportage, of which, with the exception of literature, all the genres of contemporary writings contain elements" (OC, p. 368).

How is poetic language different? The key is what Mallarmé calls transposition. "What is the use of the marvel of transposing an act of nature into its vibratory almost-disappearance through the play of speech, however; if it is not so that from it should emanate, without the encumbrance of a close or concrete reference, the pure notion" (OC, p. 368). The most famous sentence in this essay once again stresses the nonrepresentational functioning of poetic language. "I say: a flower! and, apart from the oblivion to which my voice relegates any contour, understood as something other than the known calyxes, musically there arises, idea itself and pleasant, the flower absent from all bouquets" (OC, p. 368). The result of the use of language is not the transfer of a concrete thing; the thing is "transposed" so that only the pure notion of it remains. As Mallarmé says in another famous statement, "Divine transposition  . . . goes from fact to the ideal " (OC, p. 522; Mallarmé's emphasis). The "flower absent from all bouquets" is the ideal flower evoked in the sound "flower," and it is absent for the good and sufficient reason that it is ideal. Poetic language, rather than having "the function of facile, representative cash," rediscovers a prized quality: virtuality (OC, p. 368).

Concrete things are not all that disappears in the act of poetic speaking. "The pure work implies the elocutionary disappearance of the poet, who instead yields the initiative to words, mobilized by the clash of their inequality; they light up from their reciprocal reflections, like a trail of fire on gems, taking the place of that palpable breath in the lyric inspiration of yore or the enthusiastic personal direction of speech" (OC, p. 366). Meaning no longer comes from the poet; nor does it come as the result of aggregating the individual meanings of words, the way one can aggregate the individual values of coins and bills in a fistful of cash.

Instead, it emerges from the "reciprocal reflections" of the words, from the relational complex formed by the poem, which is seen as a structure, not a linear accumulation of word-references.

Mallarmé has come a considerable distance in this essay from the apparent impasse he faced in the observation that words don't mean by a "unique stamp." In fact, he has found a way out of the archaic notion of language that makes such an impasse possible. In the old notion, both for those who are confirmed Cratylists and for those who nostalgically yearn for Cratylism in an ideal past, language is a vertical medium made up of distinct units of meaning, each of which has the responsibility of transmitting a particular thing or concept. The naive Cratylists believe that this is how language really works. The skeptics realize that language doesn't work this way in the real world, but they feel that it should, that real languages represent a kind of fall from grace. Mallarmé, however, has abandoned the view of language that both groups presupposed. If there is no "unique stamp," then that is all to the good since language, at least good language, has a higher purpose. In this view it's not even relevant to be "materially truth itself." Truth doesn't have the rudimentary sense that the old view assumes.

Where Mallarmé's reflections lead is to a relational notion of language. But they also lead to another cardinal moment in the history of language theory. In chapter 1 I mentioned Mallarmé's role in the fracturing of language. I quoted Foucault's comment that Mallarmé played a major role in that process and in the movement of language to a state of "philological objectivity." We can now see how true Foucault's comment is, and we can see the logic by which Mallarmé arrived at those ideas. Once we have abandoned the Cratylic comfort of the sound-sense correlation, placing ourselves in a world where meaning assembles itself from the spaces between words, we have also abandoned the bond that unites poet and language. Transposition takes place in the language, not in the poet. It renders the pure notion of things. This pure notion comes to us from the clash of inequality of words; hence the "elocutionary disappearance of the poet," that is, the disappearance of the poet as a speaking presence. Because language is a system of clashing inequalities, the poet has yielded the initiative to words. When it comes right down to it, words are all there is; they are the only real thing that we encounter in language. That's why Mallarmé chooses to compare them with gems—hard, cold, objective gems. And that's why Foucault can speak of the "unique, difficult being" of language that Mallarmé leads our

thought back to. Mallarmé has fractured language by separating it from the poet, and he has rendered it hard, opaque, objective, precisely because it is separate.

Mallarmé is not the only French poet to address problems of language in a context of literary aesthetics. Paul Valéry comes to mind, too. His "Poetry and Abstract Thought" ("Poésie et pensée abstraite") revives the old poetry-prose distinction, suggesting that the peculiarity of poetic language is that it calls attention to itself as language. Valéry also modifies the Cratylic conception of language by proposing that poetic language merely gives the illusion of a necessary harmony between sound and sense.^[4] But "Poetry and Abstract Thought" was written in 1939 and doesn't form part of our history. Besides, the most interesting things Valéry had to say about language have more to do with relationalism than with Cratylism or the poetry-prose distinction, so we'll leave them for later.

The value of Mallarmé's theory of language is that it so compactly presents the whole complex of issues in modern literary aesthetics. The argument in "Crisis in Verse" leads naturally, logically, to a theory of relationalism. Mallarmé's hesitation between mysticism and skepticism shows him still in the grips of an older tradition based on faith ("You can't do without Eden") but pulled also toward the realm of the purely speculative and the thorny, characteristically twentieth-century question of aesthetic ontology. All these areas of thought are related. Poised on the edge of the twentieth century, Mallarmé conveniently bequeaths them to modernity as a system.