The Metaphor-Metonymy Binarism

Much theoretical speculation in the recent history of formalism, structuralism, and deconstruction depends on a relatively simple dichotomy between the linear and referential aspects of language. As soon as one understands this basic dichotomy, however, everything else falls into place with relative ease. Grammar and the syntagmatic advance of words are linear, while the meaning of words and their figurative implications inclusive of metaphor and punning is referential. Every word and group of words necessarily partakes of both dimensions.

The dichotomy was first proposed by Ferdinand de Saussure in Course in General Linguistics (1915; McGraw Hill, 1959--pb., 1966), when he differentiated (p. 66) between langue (language as a collection of rules and habits) and parole (the use of phrases and sentences that draw upon these rules and habits). In the words of Saussure: \"Language [langue] is speech less speaking [parole]. It is the whole set of linguistic habits which allow an individual to understand and to be understood\" (p. 77). Again: \"Language [langue] and speaking [parole] are then interdependent; the former is both the instrument and the product of the latter. But their interdependence does not prevent their being two absolutely distinct things\" (p. 19). And yet again: \"Language [langue] is necessary if speaking [parole] is to be intelligible and produce all its effects; but speaking [parole] is necessary for the establishment of language [langue], and historically its actuality always comes first\" (p. 18).

Saussure also proposed that linguistics may be plotted on two axes, the axis of simultaneities and the axis of successions: the axis of simultaneities \"stands for the relations of coexisting things . . . from which the intervention of time is excluded,\" while the axis of successions advances through time such that \"only one thing can be considered at a time . . .\" (p. 80). Fixed grammatical rules occur on the axis of simultaneities, while their gradual evolution through linguistic history occurs on the axis of successions. According to Saussure, these are two entirely different relationships. Just as celery may be cut across the grain to display its cross section or may be cut with the grain to display its striations, all linguistics can be divided into the synchronic and diachronic varieties. Synchrony emphasizes the axis of simultaneities that embraces grammar's relative permanence as a collection of syntactic rules and habits shared by individuals in the same language group, while diachrony emphasizes the \"evolutionary phase\" of language on the axis of selections such as occurs from one century to the next (p. 81). Saussure primarily used these coordinate axes to differentiate structural linguistics from historic linguistics, but their use can obviously be extended to express the distinction also suggested by Saussure between langue (the habitual grammar and word definitions shared by readers) and parole (their spoken use to communicate ideas). Langue would be understood to occupy the axis of simultaneities, while parole comprises the linear sequence of words equivalent to the linear history of language on the axis of successions.

Roman Jakobson found a comparable dichotomy between metaphor and metonymy in his seminal paper, \"Two Aspects of Language and Two Types of Aphasic Disturbances,\" published in his monograph, Fundamentals of Language (Mouton & Co--Gravenhage, 1956). Here Jakobson discussed two types of aphasia based on complementary disorders

in comprehending language: (a) a similarity disorder whereby one primarily depends on syntactic context to draw words into use (pp. 63-); and (b) a contiguity disorder whereby one's style becomes a telegraphic \"word heap\" without much, if any, evidence of syntax (pp. 71-72). According to Jakobson, two faculties are thus involved in the use of language: (a) selection in the choice of words to express an idea (metaphoric); and (b) the combination of words, again to express an idea (metonymic). Elaborate sentences without a particularly impressive vocabulary (for example in the prose of Henry James) illustrates the similarity disorder, while big vocabulary in loosely constructed sentences (for example in the prose of James Joyce) illustrates the contiguity disorder. Joyce heaped together his words with apparent abandonment, while James strenuously belaboured his syntax to produce exactly the right effect--an effect he found difficult to articulate with words alone as opposed to their combination in intricate sentences. An inferior choice of words, Jakobson claimed, is at the sacrifice of metaphor, whereas an inferior combination of words is at the sacrifice of metonymy (p. 76).

Jakobson's specific use of metaphor also emphasized Saussure's signifier/signified binarism that unavoidably occurs in the choice of words. The particular sound of a word (e.g., horse) is its signifier, while the object this sound refers to (the animal one ordinarily identifies as a horse) is its signified. I think it important to recognize that this signified/signifier binarism may also be extended to include I.A. Richards's vehicle/tenor interaction between the image of a metaphor (its so-called vehicle) and its final and otherwise elusive meaning that is evoked by this image (its so-called tenor). Just as ordinary signification draws upon the conventional meaning of words on the axis of simultaneities, the use of metaphor to intensify the effect of words (by using an image to signify something entirely new) likewise draws upon the meaning of words on the axis of simultaneities. Metaphor simply doubles signification by signifying an image that in turn signifies a new and more elusive meaning (for example the horse face or horse feathers or to horse around). On the other hand, Jakobson's use of metonymy applies to contiguous relationships among words in their forward motion on the axis of successions (pp. 78, 82), giving the interreferentiality among nearby words greater importance relative to their signifier/signified relationships. The conventional definition of metonymy based on traditional rhetoric has emphasized the use of a closely associated image (e.g., the use of crown to signify royal authority, as in \"the crown of England). Jakobson's specific definition of metonymy involved the use of later words to clarify earlier words and earlier words to clarify later words. These words are associated because of their sequential position, and their meanings necessarily intersect on a metaphoric basis within this linear context. Hence they are metaphoric not through referential simultaneity but through sequential interplay: let them be called metonymic.

According to Jakobson, the stark opposition between these two aphasiac disorders discloses the inevitable polarization between the selection of words (metaphoric) and their combination (metonymic) (pp. 58, 60). In his later paper, \"Closing Statement: Linguistics and Poetics,\" in Style and Language, edited by Thomas Sebeok (The MIT Press, 1960), pp. 350-377, Jakobson went on to schematize this opposition by proposing an axis of selection equivalent to Saussure's axis of simultaneities, and an axis of combination equivalent to Saussure's axis of successions. The advantage of Jakobson's shift in

nomenclature was that selection emphasizes the actual choice of words and locutions drawn from the realm of simultaneities identified as langue, whereas combination emphasizes the actual bringing together of these words in verbal arrangements typical of parole as a flow of spoken and written language. Jakobson was thereupon able to offer a simple but ingenious formalist definition of poetry based on this binary distinction: \"the poetic function projects the principle of equivalence from the axis of selection into the axis of combination\" (p. 358). In other words, metaphoric referentiality on the axis of selection can be displaced to the axis of combination through metonymic interplay. As illustrated by Jakobson's explications of Baudelaire's \"Les Chats\" and Shakespeare's Sonnet 129, this is achieved through the use of dualisms or parallel elements each of which \"signifies\" the other by drawing attention to their parallel relationship. This occurs, for example, with rhyme, rhythm, alliteration, etc. Here Jakobson emphasized a formalist pursuit of of resemblances, but the no less formalist pursuit of deviations (foregroundings, surprises, or \"stylistic devices\") by deviationist critics such as Shklovsky, Mukarovsky, Riffaterre, and others can just as easily be explained to feature this displacement to the axis of combination, with the similarity among elements subsumed to their unexpected differences (the surprising use of similarities included).

The primary value of Jakobson's dichotomy was in shifting Saussure's paradigm from its initial distinction between historic and structural linguistics to the more useful hermeneutic distinction between the referential and linear aspects of language:

For those who remain mystified by this paradigm, it is possible to understand each axis by its linear direction alone. The horizontal axis of combination depicts the right-to-left advancement of words on the typewritten page, while the vertical axis of selection combines the signifier/signified and tenor/vehicle paradigms of Saussure and I.A. Richards.

Louis Hjelmslev among others somewhat complicated this model by identifying the vertical axis as being paradigmatic to emphasize grammatical categories and the horizontal axis as being syntagmatic to emphasize the linear and irreversible flow of words one at a time. The important French structuralist and deconstructionist critic Roland Barthes devoted chap. 5 of his useful text Elements of Semiology (Hill and Wang, 1967) to an explanation of Hjelmslev's theory to help clarify Saussure and Jakobson's respective paradigms. For example, Barthes fused his paradigmatic/syntagmatic dualism with Saussure's langue/parole dualism in his argument: \". . . because the syntagms of speech [parole] are constructed according to regular forms which thereby belong to the language [langue]\" (p. 62--italics added). Barthes likewise mixed Hjelmslev's dualism with Jakobson's metaphor/metonymy and selection/combination dualisms in his ingenious but confusing argument:

. . . that any metaphoric series is a syntagmatized paradigm, and any metonymy a syntagm which is frozen and absorbed in a system; in metaphor, selection becomes contiguity, and in metonymy contiguity becomes a field to select from. It therefore seems that it is always on the frontiers of the two planes that creation has a chance to occur. (p. 88)

Quite aside from the paradox Barthes was trying to articulate here, elaborating Jakobson's earlier paradox that poetry projects the axis of selection onto the axis of combination, one cannot ignore the extent to which Barthes mingled Saussure, Jakobson, and Hjelmslev's binarism in his critical usage.

In his essay, \"The Structural Study of Myth,\" Jacques Levi-Strauss likewise drew upon the assumptions implicit in the use of coordinate axes when he explained how the elements of myth are combined as gross constituent units. He declared that these elements are brought together on \"a two-dimensional time referent which is simultaneously diachronic and synchronic, and which accordingly integrates the characteristics of langue on the one hand, and those of parole on the other.\" As a friend and protege of Jakobson, Levi-Strauss subsumed Saussure's categories--langue and parole, and diachronic and synchronic--to Jakobson's paradigm to explain the intra-referential complexity of the Oedipus myth based on story segments (or \"functions,\" in Vladimir Propp's nomenclature) rather than words and phrases. Langue, for example, entails the lexicon of relevant story (or action) segments the teller of myths can draw upon (killing one's father, etc.) from the axis of selection, while parole entails the actual combination of these segments in the context of the myth being told--for example repeatedly advancing in the Oedipus myth among two alternatives on the axis of combination: (a) of overrating and underrating blood relations and (b) of monsters slain and human physical impediments acknowledged. As with words, the collection of story segments that comprise langue is different from the axis of selection that includes it, but to draw upon this axis necessarily brings it into play; likewise, the combination of these story segments through parole is different from the axis of combination in and of itself, but the flow of episodes in telling stories necessarily brings this axis into play. Moreover, the synchronic existence of segments comprising langue is not exactly the same as the axis of selection, but it is the synchronic availability of these functions that makes selection possible; and likewise the diachronic process of combining functions may be differentiated from the axis of combination in and of itself, but the actual combination of functions as a diachronic process occurs as measured by the axis of combination.

What Levi-Strauss did once he established the framework of his approach with the Saussurean/Jakobsonian coordinate axes was to map out the Oedipus myth on its two-dimensional plane, with the repetition of functions symbolized vertically and the

syntagmatic advancement among these functions symbolized horizontally. Vertical relationships signified equivalent retrievability from the synchronic langue of mythical functions, while horizontal relationships represented the diachronic forward inertia as a story is being told that combines these functions. In Levi-Strauss's article, \"The Story of Asdiwal\" in The Structural Study of Myth and Totemism, ed. by Edmund Leach (Tavistock, 1967), this sequence actually reaches closure suggestive of Aristotle's unity of action. Lacan and Derrida's deconstructionist emphasis upon deferred signification (or differance) likewise drew upon the Saussurean/Jakobson's paradigm, since the metonymic imposition of metaphor upon the axis of combination inevitably transgresses the limits of any particular text. Other, presumably external associations necessarily intrude, so the principle of textual unity is inevitably violated. Metonymy ranges at will within a text, but also beyond a text, and as a result the limits of any particular text expand to become the entirety of discourse. Roland Barthes explained this principle of intertextuality in his article \"From Work to Text\": \"The Text . . . practices the infinite deferment of the signified, is dilatory; its field is that of the signifier, and the signifier must not be conceived of as the 'first stage of meaning,' . . . but in complete opposition to this, as its deferred action.\" What might seem mystifying here is clear enough when Barthes's argument is understood in light of Jakobson's paradigm, once again with the dynamics of signification on the axis of selection superimposed as metonymy through \"deferred action\"--in this case metonymy without limits on the axis of combination. And finally, Paul de Man's dichotomy between grammar and rhetoric in \"Semiology and Rhetoric\" (Diacritics, 1975) might seem free of Jakobson's influence, but not so. De Man was obviously using his dichotomy as a semiotic binarism in the tradition of Saussure and Jakobson. The connection seems plain, for example, where de Man drew upon Hjelmslev's dichotomy between the syntagmatic and paradigmatic aspects of language, and where he quoted Genette to the effect that there is a \"combined presence, in a wide and astute selection of passages, of paradigmatic, metaphorical figures with syntagmatic, metonymic structures.\" Here he was very clearly linking Hjelmslev's idea of paradigm with Jakobson's idea of metaphor on the axis of selection, and Hjelmslev's idea of syntagm with Jakobson's idea of metonymy on the axis of combinations.

But how did de Man intend to locate grammar and rhetoric on these two axes? Obviously, he identified rhetoric with metaphor: \". . . the study of tropes and of figures (which is how the term rhetoric is used throughout this essay . . .).\" However, it is not clear whether his particular tropes and figures referred to metaphor itself on Jakobson's axis of selection or to metonymy as a subcategory of metaphor on his axis of combination. Based on Saussurean/Jakobsonian convention in the use of coordinate axes, grammar would normally be identified with langue on the axis of selection, thereby relegating rhetoric as trope (or metaphor) to the use of metonomy on the axis of combination. Then again, de Man might have wanted to treat grammar as a linear use of structures to organize the syntagmatic flow of words, or to include both tropes and grammar on the axis of selection, as for example when he argued, \"the study of tropes and of figures . . . becomes a mere extension of grammatical models.\" But then again, when de Man favorably mentioned a new perspective whereby metaphor \"would not be defined as a substitution but as a particular type of combination,\" it would seem he wanted to bring metaphor itself onto the

axis of combinations, presumably based on its linear contextual dynamics, as explained, for example, by Max Black's theory of the metaphoric \"frame.\" But it remains unclear whether de Man meant to pursue this line of inquiry himself or simply wanted to mention it before launching into his own argument.

When de Man spoke of the \"illocutionary realm of grammar\" as opposed to the \"perlocutionary realm of rhetoric,\" he apparently drew upon J.L. Austin's two categories in How to do Things with Words (Oxford, 1962, 1975) to link rhetoric with the axis of combination, in this instance by isolating the referential aspect of language (illocutionary) from its rhetorical production of effects (perlocutionary) through both figurative tropes and the persuasive skills usually associated with rhetoric. De Man also seems to have had the same purpose when, perhaps influenced by Michel Foucault's theory of insanity in Madness and Civilization (1961; Random, 1965), he later treated verbal behavior on the axis of combination as the primary source of dysfunctional experience on the axis of selection: \"Rhetoric radically suspends logic and opens up vertiginous possibilities of referential aberration.\"

Later in his article, however, de Man accepted at its face value Jakobson's metaphor/metonymy distinction: \". . . here it is the substitutive totalization by metaphor which is said to be more effective than the mere contiguity of metonymic association,\" thus affording figurative primacy to the axis of selection. And still later he brought grammar into the metonymic advancement of sentences as the actual combination of words instead of their repository in langue to be drawn upon in making such combinations: \"The passage from a paradigmatic structure based on contingent association, such as metonymy, shows the mechanical, repetitive aspect of grammatical forms to be operative in a passage that seems at first sight to celebrate the self-willed and autonomous inventiveness of a subject.\" In other words, grammar as a \"paradigmatic structure\" on the axis of selection directly exerts itself on the axis of combination to dominate metonymy's \"contingent association\" that only seems to declare a subject's \"self-willed and autonomous inventiveness.\"

De Man himself finally explained the source of these difficulties to have been that the two categories of grammar and rhetoric merge through a \"rhetorization of grammar\" and a \"grammatization of rhetoric.\" Here it seems that de Man, like Barthes, was once again drawing on Jakobson's formalist principle that literature imposes the axis of selection on the axis of combination, but with the important caveat also implied by Barthes that reciprocity occurs through each axis being imposed on the other. Interestingly, De Man identified the rhetorization of grammar with indeterminacy, as opposed to the grammatization of rhetoric, which he identified as a reaching for truths, and which I argue in Negative Poetics necessarily entails closure through the achievement of false truths. De Man thus linked the grammatization of rhetoric on the axis of combination with the narrative structures that feature the pursuit of truth, but he gave priority to the rhetorization of grammar on the axis of selection, thus giving indeterminacy a more basic role in the creation of a text, thus justifying the theory of deconstruction: \"The deconstruction is not something we have added to the text; it constitutes the text in the first place. A literary text simultaneously asserts and denies the authority of its own rhetorical mode.\" All fair enough, except that de Man still later argued, \"the deconstructive reading revealed a first

paradox: the passage valorizes metaphor as being the 'right' literary figure, but then proceeds to constitute itself by means of the epistemologically incompatible figure of metonymy.\" Here once again we are confronted with Jakobson's original location of metaphor (hence rhetoric) primarily on the axis of selection. One can only conclude that the principle of indeterminacy was built into de Man's essay itself, and that perhaps its primary charm lies in his constant vacillation among potentially contradictory uses of the Saussurean/Jakobsonian paradigm. So there is no simple answer to de Man's use of this paradigm, and if he had any clear idea of what he was doing he primarily linked rhetoric with metonymy, not metaphor. For De Man was no less larcenous in his theoretical speculation (i.e., confused about his mine/thine boundaries) than in the life he led, though without diminishing the value of his speculations, since he wanted to defend the principle of indeterminacy as the most essential ingredient of the deconstructive reading. He actually \"went indeterminate\" to justify its literary value.

To summarize, Jakobson's coordinate axes of selection and combination define and help to differentiate the following binarisms that have been important in the history of the formalist/structuralist/deconstructionist critical tradition: Saussure Jakobson Hjelmslev Derrida Austin de Man metaphor paradigm(atic) paradigm(atic) illocutionary grammar? metonymy syntagm(atic) syntagm(atic) syntagm(atic) rhetoric? axis of selection langue synchrony axis of simultaneities axis of combination parole diachrony axis of successions Of course this list can be extended even further. Other possible binarisms that can be mapped on coordinate axes are listed in my article, \"The Dialectics of Paranoid Form,\" Genre (spring, 1978), pp. 131-57--revised and reprinted as Chapter Five in Negative Poetics (Iowa, 1992). By respectively linking the Freudian displacements of identification and denial typical of paranoid delusions respectively with the axes of selection and combination--projection through metaphoric signification and denial through the narrative achievement of closure--I open Pandora's box for a variety of non-formalist alterities, including anima and animus, Nietzsche's Dionysian and Apollonian, Adler's inferior feelings and masculine protest, Caudwell's natural and social roles, etc.--in all instances with the first term plotted by the axis of selection and with the second plotted by the axis of combination (see esp. pp. 152-53, in my article, p. 150 in my book). In my Centennial paper, \"Metaphoric Hypersignification, Metonymic Designification,\" I also treat at greater length metaphor's dynamics of signification, its absorption by metonymy resulting from the

interaction between the short and long term memories, and the paradoxical interdependence between these experiential alternatives within the context of literary form.

So I think the metaphor/metonymy binarism mapped on coordinate axes by Saussure and Jakobson can be appropriated for critical purposes beyond its essentially conservative intra-referential use at the expense of history, psychology, etc. by the formalists/structuralists/deconstructionists. Right now, of course, criticism is elsewhere engaged, but such might be our task in the years to come.