1. Meaning and Semantic Change

Symbols—despite the indirectness of their signification—are the species of (legi)sign capable of constituting linguistic discourse. ‘The most characteristic aspect of a symbol is its aspect as related to its interpretant; because a symbol is distinguished as a sign which becomes such by virtue of determining its interpretant’ (NE 4:260). Here we note the full power of Peirce’s definition as it applies to the notion of language as a symbolic system. And if we accept Jakobson’s statement that ‘everything language can and does communicate stands first and foremost in a necessary, intimate connection with meaning and always carries semantic information’ (1972:76), we are forced to the conclusion that all linguistic phenomena must ultimately be comprehended as instantiations of symbolicity. The semeiotic perspective articulated by Peirce encompasses symbols as the epitomical signs, and ‘language is an example of a purely semiotic system’ (Jakobson 1971:703). In order to make progress in the exploration of meaning, it thus appears we must probe the ontology of the symbol.

The successful investigation of semeiosis, including symbols, cannot proceed, however, without taking due account of Peirce’s three categories. It is Thirdness that figures most prominently in the structure of a semeiotic system such as language. Whereas Firstness is fundamentally a quality of feeling or a mere possibility and Secondness ‘the experience of effort, prescinded from the idea of a purpose’ (H 25), Thirdness alone informs the notion of mediation or representation; that is to say, precisely the triadic relation that obtains between sign, object, and interpretant (cf. H 31). Thirdness is generality governing lawlike change and transformation, which distinguishes it from Firstness, the possibility of a single unitary and immutable quality. Thirdness is accorded functional prominence in the Peircean concept of the interpretant. Since it is the interpretant that is indispensable to the integrity of the sign relation, semeiosis—and, consequently, meaning—are ineluctably contingent on Thirdness via the role of the interpretant: ‘every genuine triadic relation involves meaning, as meaning is obviously a triadic relation’ (1.345).

The categorial framework of Firsts, Seconds, and Thirds provides (in addition to other advantages) a way of understanding the different, seemingly incompatible aspects of meaning to which linguistis have been attuned traditionally. To begin with the perceived ‘ungraspability’ (Lyons 1977) of meaning which has led some to relegate the concept to the status of a ‘pre-theoretical term’ (Lyons 1977:1), we can now correctly judge this property to be inherent and highly necessary as the systematically built-in quotient or margin of inchoateness which allows meaning to subsist and to change. Firstness of meaning is tantamount to infinite regress, which is actually the inherent potential of an infinity of translatable signata indispensable to the fulfillment of new communicative needs as they arise in the course of cultural growth. The relation between sign and object cannot, of course, remain ultimately inchoate, and the mediational or interpretative system which comprises semantic structure comports an element of Secondness, of what Peirce terms ‘brute reaction’ between actual meanings as they are differentiated from each other, in the imposition of pragmatically definite choice that language makes on the potentially infinite continuum of reference. Simple reaction as represented by selection (the segmentation of the semantic continuum) does not complete the picture, however. This role is reserved for Thirdness, manifested in the rule or law, the patterning and coherence, which facilitate understanding between speakers through the system of interpretants that places practical communicative limits on infinite regress, thereby creating the condition of structure in meaning. Peirce himself puts it tellingly: ‘Reality is compulsive. But the compulsiveness is absolutely hie et nunc. It is for an instant and it is gone .... The reality only exists as an element of the regularity. And the regularity is the symbol. Reality, therefore, can only be regarded as the limit of the endless series of symbols’ (NE 4:261).

Symbols by their very nature, while producing an endless series of interpretants that are themselves symbols, tend towards definiteness, towards interpretants that are continually more determinate than their antecedents. Growth—a tendency to become determined via interpretation—inheres in the structure of the symbol as its fundamental definiens. That is why Peirce goes so far as to assert that ‘a symbol is an embryonic reality endowed with power of growth into the very truth, the very entelechy of reality’ (NE 4:262). (So much for truth-conditional semantics!)

The structure and development of tropes is particularly suited to an illustration of the teleological nature of symbols. The dynamic or life cycle of tropes (perhaps more aptly to be called their ‘life spiral’) typically involves (as we shall see in the next section) the lexicalization of an initially living metonymy or metaphor (by what Stern 1931:390 calls adequation). This aspect is especially common to the diachronic accretion of terminology in a specialized (e.g. scientific) sector of the vocabulary. As Quine has it, ‘the neatly worked inner stretches of science are an open space in the tropical jungle, created by clearing tropes away’ (1978:162). To take an example from the idiom of contemporary sports, specifically basketball in America, within the last decade an expression—back door—has entered common parlance to describe a maneuver whereby one of the offensive team’s players (without the ball) manages to penetrate unnoticed behind the opposing team’s defenses close enough to the basket to attempt an unobstructed short shot (usually a so-called ‘layup’). When this play first came into existence, its linguistic designation was patently metaphorical (cf. getting in through the back door). Both the play and the expression, indeed, arose spontaneously, by chance, as instantiations of the ‘brute force’ of Secondness. But as the play spread and grew in frequency, it ceased to be the spontaneous result of chance configurations in game situations. It became what is known as a ‘set play,’ that is a maneuver planned and perfected in practice sessions, designed to take its place among the stock of plays in a team’s repertoire or ‘play book.’ At this crucial point, the metaphor had largely lost its figural status, having become a term. Now, back door became a deriving base for terminological exfoliations such as to go back door, to back door, to pull a back door (play), etc. Indeed, here we can legitimately speak of a lexicalization of the original trope. In a Peircean perspective, this is the predicted teleological result of the very ontology of a symbol. In its status as the epitomically indeterminate sign, the symbol is defined by its preeminent power of growth and development, of determining itself increasingly, of making its meaning more concrete (= clear, not non-abstract) by engendering successive interpretants with which it maintains a semantic affinity along a gradient simultaneously synchronic and diachronic.

What is particularly significant about the playing out of this preordained development in terminology is the semeiotic connection between the gradual attenuation of indeterminacy and the concept embedded in the word term itself. The end result of the growth of symbols is precisely the etymologically veridical term(inus)!

Linking this process to Peirce’s pragmatistic conception of habit as the ultimate logical interpretant, we can see with clarity that the intellectual, rational core of habit presupposes a concomitant loss of the spontaneous, the emotive quotient of sign function. In the Peircean idiom, we have here a progression away from the emotional interpretant toward the final interpretant, viz. a mental habit subtending a habit of action. An immediate morphological parallel that suggests itself is lexicalization in affective (expressive) formations. In a language as replete with diminutives and hypocoristics as Russian, it is common to observe the transition of an original expressive derivation into the category of terms broadly conceived (i.e., neutral substantives with no affective meaning). A word such as nosok ‘sock, spout, toe (of footwear or hose)’ shows an effacement (not completely) of the historically primary meaning ‘little nose’ via an intermediate stage involving figuration, whereby the protuberance characteristic of a nose is transferred onto objects occupying the same position vis-à-vis an anchoring mass. Hardly an isolated example in Russian, it has, moreover, analogous counterparts in many other languages.

But the power of a symbol does not stop with the demise of a trope or the eclipse of affective meaning resulting from fading and lexicalization. The entombed history of the linguistic sign can always be disinterred—by poets, or by those who recur to the poetic function for aesthetic effect in an otherwise discursive context (such as that of advertising slogans). Thirdness, with its definitional esse in futuro, always teleologically transcends the limitations of Secondness, of the here and now.

The telos of the symbol, its teleological thrust, is not confined to this kind of sign; it is only most prominent therein. For signs are entelechies (NE 4:299), and entelechy is just ‘the third element... which brings things together . . . the element which is prominent in such ideas as Plan, Cause, and Law’ (ibid., 195-6).

We thus come to the notion that linguistic semeiosis necessarily presupposes the involvement of the most important member of the semeiotic triad, the interpretant, whose role is defined by the only fallible mode of inference—abduction. This definition accords perfectly with the essential nature of the symbol and with symbolicity. Collocated within the matrix of Peirce’s categories and his pragmaticism, semeiotic fallibilism renders change in meaning understandable as an inalienable part of the sign situation and, consequently, as an aspect of semantic continuity assuring interpretability across discontinuous generations. What is even more important, change itself becomes an ontological component of linguistic meaning and the nature of meaning. A semeiotic perspective thus leads without fail to the position that synchronic grammar(including lexis) cannot be validated without recourse to diachrony: every structure incorporates a dynamic.

Or as Jakobson put it (1971:562), ‘when the time factor enters into such a system of symbolic values as language, it becomes a symbol itself.’

2. Tropes and the Structure of Meaning

From antiquity to the present day the study of tropes, or figures of speech, has always been at or near the center of scholarly interest in language. The preponderant mass of research on tropes (both traditional and more recent¹) has had as its primary goal the explication and typologization of the semantic relations contracted by figures of speech, rather than the analysis of what can be called the ontological structure of tropes. As conceptualized with reference to Schema A below representing the simultaneous aspects inherent in all tropes, the emphasis heretofore has been on setting forth the precise details of the ‘preexisting conditions’ entered into by the two signata, figural and literal, i.e., a kind of relational algebra; and on the combinatorial logic or ‘operations’ to which the relations are subjected when the signata are juxtaposed. Furthermore, although many studies purport to deal with the ontological structure of tropes (their essential definientia), none appears in my opinion to have had noteworthy success.

Nor has there been much effort by modern investigators, with the prominent exception of Stern (1931; rpt. 1965), to collocate tropes in a general theory of semantic change. This is, nonetheless, precisely where tropes ought to be placed in order to be understood in their diachronic subaspect as verbal entities with a structurally marked dynamic. What has been insufficiently apperceived, indeed, is the fact that all figures of speech are a particular kind of abductive innovation. As such they necessarily project a historical dimension, which involves change in meaning; and even a kind of ‘life cycle.’

To elaborate on the ontological structure of tropes, one might begin by noting that for Aristotle the genus/species relationship in its four exhaustive substitutions of terms—genus for species, species for genus, genus for genus, and species for species—defines the domain of metaphor. Indeed, metaphor is the all-encompassing trope for Aristotle, as for all succeeding analysts of figural language. The opulence of studies focusing on metaphor and the marked paucity of those focusing on metonymy issue directly from this (tacit) understanding (I return to this matter below to suggest a natural and precise cause).

In a pioneering article (1965; rpt. 1971:345-59) on the semeiotic essence of language, Jakobson includes in passing the following perspicuous characterization (355) of metaphor and metonymy:

A partial similarity of two signata may be represented by a. . . total identity of signantia, as in the case of lexical tropes. Star means either a celestial body or a person—both of preeminent brightness. A hierarchy of two meanings—one primary, central, proper, context-free; and the other secondary, marginal, figurative, transferred, contextual—is a characteristic of such asymmetrical couples. The metaphor (or metonymy) is an assignment of a signans to a secondary signatum associated by similarity (or contiguity) with the primary signatum.

In this passage the notions of similarity and contiguity are somewhat subordinate to the notion of hierarchy, which tends to belie their prominence in Jakobson’s conception of the two tropes promulgated elsewhere. An earlier and better-known exposition is to be found in his Fundamentals of Language, in connection with the linguistic analysis of aphasia. There the notion of hierarchy is only implicit at best, and the entire thrust of the argument hinges on defining metaphor as a relation of similarity and metonymy as a relation of contiguity between signata. Clearly, both the ranking of signata vis-à-vis each other and their specific relational connection are equally pertinent to an understanding of the structure of tropes. But the relative importance of these two aspects of the ‘figural situation’ (as I shall call it) has not heretofore been fully assessed.

The figural situation is comprised by three fundamental components or conjugate aspects which are copresent and rank ordered in every trope. There is first the preexisting condition or quality in which the terms of tropes, the signata, must inhere; which is to say the requirement of relatability—the perceptual and cognitive potential underlying the actual relationship of two signata—that must be satisfied by the terms in order for tropes to have an esse in futuro. The state of being relatable is defined logically by three partially unranked specificity conditions on trope signata: (1) membership in a certain species, with a concomitant relatability by similarity; (2) membership in a certain context, with a concomitant relatability by contiguity (logical as well as spatial); (3) adumbration, vis-à-vis each other, of a potential rank order tantamount to value (in the strict axiological sense) or dominance. Second, there is the actualization of that substance represented in posse by the preexisting condition of trope formation; which is to say the juxtaposition or relation of signata. This juxtaposition is, more accurately, an opposition of terms, since the configuration of meanings in tropes is necessarily dyadic. In parallel to the first aspect, the second is defined logically as a partially unranked triad of operations comprehended by the signata: (1) selection, implying simultaneous reference to the paradigms of which the signata are representative; (2) combination of the signata into simultaneous syntagms; (3) imposition of ranking on the signata so selected and combined, or their hierarchization. Notice should be taken that the second operation is, strictly speaking, not prescindable from the third. Their relationship is, however, not biunique: ranking implies combination, but combination does not imply ranking.

In reviewing the parallel tripartite subdivisions of the first two conjugate aspects of the figural situation, then, we should say that species denotes class membership; the signata of tropes are either of the same or of different species. Context denotes the ground in which the signata are embedded; the ground can be either shared or discontinuous. And rank is the evaluative component which is superimposed on relatable, ultimately correlate, signata. These three terms are coordinate with three distinct types of relation which they implement: similarity, contiguity, and value (dominance). In turn, the three types of relation are coordinate, respectively, with three types of juxtapositional operations: selection, combination, and ranking. Again, ranking or hierarchization represents an evaluative superstructure imparting form to selection, or paradigmatization, and combination, or syntagmatization. Note that an implicit claim is here being made to the effect that similarity and contiguity relations are copresent in and underlie all tropes, not just metaphor and metonymy. The traditional association of metaphor with similarity and metonymy with contiguity is not thereby rendered incorrect. It must, nevertheless, be revised to render explicit the fact that these two kinds of relation are present in both major tropes but are so in opposed rank order; in metaphor, similarity predominates over contiguity; in metonymy, the order is inverted.

The preexisting condition, which acts as a springboard for the onset of the figural situation, simultaneously imposes a constraint on the combinability of the signata: they must differ in respect of at least one of the signata. This is the constraint, for instance, which prevents knife from being a trope for fork, or vice versa (ad Todorov: 1974:128), since both words do not differ in species, context, or rank. Note, characteristically, that in aphasic (i.e., pathological) speech knife and fork can be commuted for each other (Jakobson 1971:250).

Finally and most significantly, the figural situation manifests a third aspect—the resultant state or the representation articulated by the signata. This is the ontological definiens of the trope. It hinges on the concept of rank or hierarchy; here, more precisely, on the rank order of the signata in the set presented by the figural syntagm. Specifically: Metonymy is defined as that trope in which a hierarchy of signata is either established or instantiated. Complementarily: Metaphor is defined as that trope in which the (simultaneously) established hierarchy of signata is either reversed or neutralized. Metonymy places two previously unconnected signata into a dominance relation vis-à-vis each other; or it gives overt semantic expression to the specific hierarchical relationship which the signata contract independent of the trope. Metonymy goes no further than the establishment or the instantiation of a hierarchy; metaphor, on the other hand, reverses or neutralizes a hierarchy, even if the hierarchy is one established during the metaphoric process only to be reversed.

The entire purport of the foregoing fairly abstract argument can perhaps be made more palpable by resort to the following tabular framework (Schema A).

COPRESENT RANK ORDERED ASPECTS OF THE FIGURAL SITUATION

1. preexisting condition or quality

a. species (similarity)

b. context (contiguity)

c. rank (value)

2. juxtaposition or relation

a. selection (paradigmatization)

b. combination (syntagmatization)

c. ranking (hierarchization)

3. resultant state or representation

a. metonymy: establishment/instantiation of hierarchy

b. metaphor: reversal/neutralization of hierarchy

c. relation between metonymy and metaphor

SCHEMA A

It is important to clarify at this juncture just what is meant by the phrase ‘establish a hierarchy.’ We may start by noting that Jakobson (1971:232) considers contiguity to be an ‘external’ relation and similarity an ‘internal’ relation. This nomenclature can be understood in the terms we adopted by observing that semantic hierarchies (simultaneous syntagms) necessarily project two dimensions, referential and significational, corresponding to the processes, respectively, of denotation and connotation (Frege’s Bedeutung and Sinn). Within the dimension of reference the syntagm is characterized by the several denotata (referents) which correspond to the several signata. Here, then, is an articulation of the patterned set of correspondences between physical or material objects or events, on the one hand; and their mental or conceptual correlates as embodied linguistically in words and sentences, on the other. A focus on the referential dimension is tantamount to an emphasis on the external aspect of meaning. Simultaneously, each semantic syntagm has an internal organization—a focus on the relations contracted by the signata amongst each other—which corresponds to and is interdependent with the denotata that constitute the universe of referents we call reality.

In diagrammatic form these relations can be represented as follows (where S = signatum and D = denotatum) (Schema B).

The distinction between external and internal can be aligned, following Jakobson, with the distinction between metonymy and metaphor. In metonymy the referential dimension predominates over the significational, whereas in metaphor this dominance relation is inverted. An external or referential focus is thus tantamount to the priority of message over code, while an internal or significational focus is equivalent to the reverse priority of code over message.

SCHEMA B

To take a pair of concrete examples (borrowed and adapted from Le Guern 1973:14), if it is suggested to someone that he ‘reread his Shakespeare,’ this does not entail an internal modification in the sense (Sinn) of the word ‘Shakespeare.’ The metonymy constituted by the implementation of an author’s name to designate a work by that author hinges on a shift of reference; the internal, or semic, organization has remained unaltered even though the referent has been displaced from author to book. Similarly, when Zola writes about the ‘loud voices [that] wrangled in the corridors’ (‘de grosses voix se querellaient dans les couloirs’), the hierarchy of semantic features or range of connotations is unchanged. The only thing that changes is the referent with which the semantic syntagm is associated. To put it another way, the message demands a partly figural interpretation while leaving the code unaltered.

It is precisely this referential shift that is captured by the phrase ‘establish a hierarchy.’ The two denotata that are correlated with the two feature hierarchies are not paratactic vis-à-vis each other because one is more general than the other: ‘voice’ stands for itself in the Zola example as well as for ‘person’ but not vice versa. The fact that ‘voice’ retains its meaning while simultaneously referring to something other than itself is precisely one of the differentiae specificae of metonymy as a trope over and against metaphor. It is the contextual locus or, more fully, the necessary presence of the contiguity relations facilitated by the context that allows the figural meaning of a metonymy to be perceived and understood as such. In saying ‘figural meaning’ I am designating in more general terms the establishment of an asymmetric relation between denotata as embodied in their correlative signata.

The dominance of external reference characteristic of metonymy is absent from metaphor. The relation between the metaphoric term and the object to which it normally makes reference is rendered discontinuous or indirect. In Pascal’s ‘Man is a thinking reed,’ the primacy of a similarity relation means not that two referents are arranged in rank order but rather that the meaning ‘reed’ is incorporated for the moment or ad hoc as a nonce signatum of the simultaneous syntagm comprising ‘man.’ Furthermore—and this is the crucial point—there is a concomitant reversal of the rank order of the two signata ‘man’ and ‘reed’ without which the figurality of the sentence would not ensue. Another way of characterizing this reversal is to say (Le Guern 1973:15) that ‘the mechanism involved in metaphor is to be analyzed ... as a suppression, or more precisely, a bracketing (“mise entre parenthèses”)’ of one member of the set of meanings which constitute the word’s semantic structure.

All metonymies, which are based on an inclusion relation, either establish or instantiate hierarchies of signata. The simplest non-literary example of this mode of signification is a taxonomy (cf. Bunge 1969:19), since any classification is founded on (or is transformable into) the principle of binary opposition, and the relation of binary opposition is metonymic. Specifically, the negative entails (includes) the positive term of an opposition but not vice versa. Negation is consequently a metonymy. Indeed, the negative term of an opposition necessarily and overtly makes reference to the positive term, whereas the positive term only implies or makes covert reference to its negative opposite. An opposition being a minimal (binary) paradigm, one term is in absentia where the other is in praesentia and vice versa. However, the positive is explicitly included in the negative since the latter is a specified absence of the former. The positive, on the other hand, includes its (negative) counterpart only implicitly, in virtue of the very nature of (logical and privative) opposition. This necessary, overt inclusion of the positive in the expression of the negative is what renders the negative a metonymy of the positive and demonstrates the inclusional nature of metonymy.

It is important to understand the rhetorical implications of binary opposition precisely because tropes presuppose the copresence of two signata, roughly labeled the literal and the figural. The opposition between these two signata is functionally construed as privative: figural meaning is juxtaposed to its opposite, non-figural meaning. Moreover, figural meaning is in a relation of dominance over the non-figural. If this were not so, the trope would cease to be understood as such (as happens not infrequently; see below).

In metonymy the relation of dominance (i.e., hierarchy) is something that is established with the onset of the figural situation and goes no further. This is especially clear in pars pro toto. The word sail, used to denote the totality ship of which it is but one (albeit an important) part, involves the differentiation of that totality. Differentiation—another mode of characterizing the process inherent in what was earlier termed taxonomy—is precisely a singling out or individuation of constituents. However, in metonymy it is not differentiation pure and simple. A necessary concomitant is the relative ranking or hierarchization of the constituents, here the two signata. In the case of sail for ship, no wonder the part chosen to represent the whole is the most prominent one; prominence is, after all, a relation of rank. One might even assert that the choice of the item singled out in metonymy is perceptually and/or cognitively well-motivated (natural). That is to say the hierarchy established by metonymy rests on yet another hierarchy. In the sail for ship case the relation between a sail and the remaining parts of the vessel is also one of rank order.

Similarly, abbreviations (a form of metonymy)—visual as well as verbal—reflect the hierarchies which inform their structure. The hieroglyph, or the pictorial part of an emblem, consists of a visual representation. While different from a verbal representation it is nonetheless fundamentally metonymic because it represents one or more parts of an object or of an utterance taken as an object. The aspect of condensation was perceived by the quatrocento philosopher Marsilio Ficino, who praised hieroglyphs for their power of contracting many thoughts into one single form. It happens more often, however, the pictorial translation of language being at issue, that the hieroglyphic designs of emblems consist of several symbols rather than one alone. An apt example might be the amorous motto dulce et amarum ‘sweet and bitter’ illustrated by the pseudo-Theocritan idyll of Cupid stung by bees while tasting honey. The metaphoric, dominant aspect of the emblem, which submerges the aspect of narrow iconicity, is simultaneously its generalizing impetus and power. It demands that contexts be sought outside the realm of direct entailment by the pictured material itself. A massive diachronic parallel to this is instanced by Chinese writing, which began as pictographic but soon became ideographic. Over time the shape of the characters underwent an internal development entailed by functional demands which resulted in a continuous attenuation of the relation between the meaning of the character and the object which it originally depicted or imitated.

Indeed, investigation of the representational aspect of forms in the visual arts tends to corroborate the finding that the tertium comparationis of a symbol and its object is not external form but function; more specifically, that formal feature which fulfills the minimum requirement for the performance of the function. Accordingly, a hobby horse provides a child with a functional substitute for a real horse as long as it can be ‘ridden’ (Gombrich 1963), even as the metonymic aspect of the figural situation includes the metaphoric component owing to similarity. This example shows appositely the primal character of metonymy as well as the greater immediacy of its link to perception as compared to metaphor.

On the more strictly linguistic side of the ledger, an abbreviation, e.g. USA, reflects at least two hierarchies which condition its structure. First, only the categorematic words United States America are selected from among those which make up the full unabbreviated version, the syncate-gorematic words the, of being deleted. Second, not just any letters of the unabbreviated counterpart are implemented in the abbreviation but the functionally prominent initials. A further—grammatical—hierarchy underlies the variant US: the adnominal genitive of America is subordinate to the adjectival phrase United States; but contrast America (for The United States of America) where the lexical hierarchy reverses that of the grammatical one. In contemporary Soviet Russian, which abounds in abbreviations and acronyms, their structure typically reflects two further hierarchies within the unabbreviated version, viz. the syntagmatic precedence of (1) morphemes; (2) letters in morphemes. Thus a word like ispolkom ‘executive committee’ which abbreviates the adjectival phrase ispolnitel’nyj komitet presupposes a selection of the leftmost segment of each constituent word up to and including the first three letters of the first lexical morpheme (ispol- + kom-).

In these strictly linguistic, non-literary examples the metonymic relations are defined by the instantiation of a preexistent or implied hierarchy rather than by the ab ovo creative establishment of a hierarchy. Indeed, the difference between instantiation and establishment serves as a typological demarcation, respectively, between linguistic and poetic metonymy. If Jakobson (1971:295) is correct in labeling the relations obtaining between members of a word family as metonymic (via their ‘semantic contiguity’), then the juxtaposition of items like editor-edition-editorial-editorship represents the instantiation of a hierarchy defined by the lexical differentiation which the grammatical signata (here: suffixes) effect in the meaning of the deriving base.

The dichotomy between linguistic and poetic is, of course, not peculiar to metonymy and is inherent in metaphor as well. Here the difference hinges, respectively, on the distinction between the neutralization and the reversal of the signata in a hierarchy. To continue with Jakobson’s example, the ascription of metaphoric status (via ‘semantic similarity’) to the relations obtaining between words with the same suffix such as editor-auditor-solicitor is to be understood as the equivalence resulting from the neutralization of the base/suffix hierarchy which accompanies the juxtaposition of words with identical suffixes but different bases. This equivalence is the very same relation underlying synonymy, since two or more words are construed to possess essentially the same meaning when the difference between their respective syntagms of semantic features (signata) are suspended or neutralized. In poetic language, however, it is the reversal of a semantic hierarchy rather than its neutralization that defines metaphor. Reverting to Jakobson’s statement (cited above), it is just when the ranking of the two signata of star, viz. ‘heavenly body’ and ‘luminary,’ is inverted that the figural situation can be said to have ensued. The meaning considered primary in a literal context (hence superordinate to other meanings in other contexts) loses its primacy in a figural context, which is to say that it becomes subordinated to the figural meaning.

When metonymy and metaphor are thus viewed in terms of hierarchies of signata, it becomes evident that these two cardinal figures cannot be hermetically sealed off from each other. More importantly, indeed, it becomes clear that their continued separation tends to distort their immanent interdependence and to obstruct a unitary understanding of the dynamic that binds them. While taking account of the high incidence of semantic fading among tropes in non-poetic discourse, we must also acknowledge the fact that a metonymy, even at its original creation in literary texts, is already on the way to becoming a metaphor. This is, as I shall explain more fully below, a function of the inherent dynamic of tropes. Homer’s ‘thirty sails’ is a metonymy, to be sure, since the trope establishes a hierarchy between signata. At the same time, however, the very establishment of a hierarchy sows the seeds of metaphor, since the figural meaning must predominate over the literal in order for the figural situation to obtain. Every metonymy thus contains the potential for sliding into metaphor. Moreover, of course, both tropes can coexist in the same linguistic vehicle, facilitating all the more the slide of an original metonymy into metaphor. The word head used to denote the person in charge is metonymic in that one part of the human body is taken to stand for the entire body (person); it is simultaneously metaphoric in that an analogy (semantic similarity) is drawn between the dominance of the head physically and physiologically over the rest of the human body, on the one hand, and the dominance of a leader over the group, on the other. These two coexistent figural components of head are themselves, naturally, not unranked: the metaphoric component markedly dominates the metonymic. This rank order, notably, is consistent with the principle that metonymy tends strongly to be superseded by metaphor. This means that metonymy is the more basic, less complex of the two tropes (cf. Lévi-Strauss 1962:101, cit. Donato 1975:6).

All metonymies which ‘pass into the public domain’ become metaphors. This accounts for the overwhelming emphasis on metaphor (noted inter alia by Jakobson 1971:258) in the history of rhetoric and the impoverished state of studies in the theory of metonymy.² Metonymies are resistant to metaphorization only when localized in texts; similarly, metaphors are genuinely resistant to lexicalization (fading) only when firmly anchored to the literary text. Once metaphors have also passed into the public domain, they can be said to have commenced on a trajectory which will lead ultimately to their effacement: ‘a language is nothing but a necropolis of dead metaphors’ (Sparshott 1974:84).

The context-sensitive nature of metonymy as a modus significandi comports well with its high incidence in works of art, including literature, which fall under the rubric of realism (Jakobson 1971:255 ff.). Metonymy subsists on overt terms of reference for its very figurality, while metaphor (as e.g., the pun) can get by on the eduction of an implied context.

Certain types of metonymy depend on context in resisting metaphorization. In personification, for example, the new signatum does not enter into the syntagm of signata because the focus is invariably on the establishment of a hierarchy that did not exist before rather than on the reversal of a preexistent or newly generated hierarchy. Harald Weinrich (1970:31) has pointed out that in the Middle Ages personifications approach in function certain mythological personages who play an extremely active role within the boundaries of the courtly universe. These symbolic personifications arise upon the attachment of degenerate mythological structures that have lost their narrative element to the allegorism of medieval thought.

For both metonymy and metaphor, their figural power is thus either crucially dependent on and/or significantly enhanced by a textual locus. In most non-poetic discourse communication may be impaired directly as the quotient of figural expressions rises; that is to say, as the number of coded (fully lexicalized) units decreases. Beyond the obvious difficulties attendant on speaking in verse (for the encoder), people communicate in ‘prose’ precisely because the prefabricated coded units, including idioms (among which are no small number of dead figures), need no spontaneous glossing since their meanings are relatively fixed and well-known. Literary discourse, on the other hand, provides the author with a great latitude of creativity in fulfilling the poetic function but entails a possibly burdensome interpretative load for the reader (decoder). Aristotle is clearly very close to the mark when he avers that the poet is to be measured in part by his skill at metaphor, meaning not simply the high incidence of tropes but their freshness. This notion is convergent, for instance, with the Russian Formalist tenet of deautomatization or defamiliarization. The spontaneous creative aspect of novel (not necessarily ‘unusual’) locutions is conveyed by Monroe Beardsley’s ‘metaphorical twist’ or by Paul Ricoeur’s notion of ‘event.’ Ricoeur correctly stresses the role of context in regulating and limiting the figural power of tropes (1974:99-100):

This contextual action [shift in meaning] creates a word meaning which is an event, since it exists only in this context; but it can be identified as the same when it is repeated; in that way the innovation of an ‘emergent meaning’ (Beardsley) may be recognized as a linguistic creation; if it is adopted by an influential part of the speech community it may even become a standard meaning and be added to the polysemy of the literal entities, contributing in that way to the history of the language as langue, code, or system. But, at that ultimate stage, metaphor is a dead metaphor. Only genuine metaphors are at the same time ‘events’ and ‘meaning.’

As mentioned earlier, tropes which are or become part of the language at large (the public domain) have a decided tendency to lose their figural status (cf. Anttila 1972:145). Examples abound in all languages. To take a familiar Romance one, the literal meaning of Latin pastor ‘shepherd’ was augmented by the Christian figural meaning ‘head of the congregation.’ Over time, however, the figural meaning faded even though, notably, the original metaphor of shepherd and sheep survived. As pointed out by Anttila (1972:145-6), the fading of tropes (i.e., a kind of lexicalization) is particularly frequent in the formation of euphemisms. Since euphemisms tend to run in diachronic chains whose individual links are subject to progressive loss of marked status, it is not uncommon to find a synchronic distribution of such items which makes reference to parameters of style or speakers’ age (so with words, for instance, denoting the locus for discharging bodily functions).

Lexicalization actually involves two distinct stages of what might be termed the dehierarchization of trope structure. Where the opposition between figural and literal has faded without being totally effaced, the item (word or phrase) has entered the category of idiom. Thus at some early point in its historical trajectory, pastor faded but continued to maintain a connection with ‘shepherd.’ Ultimately, of course, this bond was severed for all practical purposes; only a specialized acquaintance with the word’s etymology permits the connection to be revivified. While the connection still lived a common life, the new meaning was an idiom: there was an idiomatic (here: religious) transference of meaning. The lexicalization was complete only when the transferredness ceased to be a part of the perceived meaning. The intermediate stage between the life and death of tropes constituted by idioms is most aptly exemplified by proverbs, for it is in these longest of idioms that the semantic parts fail to add up to the meaning of the whole despite the presence of a delineated constituent structure. The absence or demise of a perceived constituent structure in idioms is, indeed, the proximate cause of their petrification, i.e., of the passage of idioms into the category of genuine lexicalizations.

Idioms and other species of lexicalization are, however, not immutably petrified. The poetic function, whether in verse proper or in other varieties of discourse parasitic on poetry, often involves the revivification of dead or faded tropes. This process results in paronomasia or punning, of a kind especially favored by the slogans of modern advertising.

The specious equivalence relation established by the pun reverses the hierarchy between signans and signatum. This reversal and transposition mocks the referential function of language by substituting a virtual (covert) meaning for the contextually anticipated one. The resultant metaphor identifies two objects, each of which retains its own form. A literary variation of the pun is the use of a word in play, alternately in its normative dimension, on the one hand, and its connotative (figural) dimension, on the other.

The general framework of our structural analysis suggests that paronomasia (including punning) constitutes a rehierarchization of signata into a new figural syntagm, and thereby a remetonymization and/or remetaphorization of lexicalized material. This return to tropehood (as it were) means that the process has come full circle. Indeed, we have every reason to construe the voyage from metonymy to paronomasia via metaphor, idiom, and lexicalization as the ‘life cycle’ of tropes. While not every trope need complete the entire cycle, many obviously do. It should be emphasized that the full dynamic of trope development is inherent in all figures regardless of the arrested growth manifested by this or that particular example. To summarize in schematic form see Schema C.

SCHEMA C

In an analysis of the structure of tropes the figure known as oxymoron is a particularly useful diagnostic, for as Beardsley points out (1962:298), ‘in oxymoron we have the archetype, the most apparent and intense form, of verbal opposition.’ Now, as was noted earlier, an opposition is a minimal paradigm, one term being in praesentia and the other in absentia and vice versa. In oxymoron, however, both terms are in praesentia, since qualitative opposites are understood to be or are reinterpreted as privative opposites. In an example like ‘the living dead,’ if we represent the two terms as A (living) + B (dead), we then note that A = non-B and B = non-A, which is to say that both terms are negations of each other. This logical condition on the structure of the oxymoron as a figure means that oxymoron is fundamentally metonymic (via the mutual negation). At the same time, the oxymoron has a metaphoric component which predominates—in accord with our expectations. Oxymoron is metaphoric, more precisely, in that the paradigm/syntagm hierarchy is reversed: mutual exclusion (the condition of paradigmaticity) is subordinated to mutual inclusion (the condition of syntagmaticity). Put another way, disjunction is superseded by conjunction. But this resultant state or definiens of the trope is only the form of the meaning of an oxymoron. So long as the substance of the linguistic items in oxymoron is genuinely polar the meaning of their juxtaposition will remain epitomically abstract, which explains, incidentally, just why one is so hard pressed to distill a meaning from pure oxymoron. The superlative degree of vividness and intensity noted by Beardsley derives, then, from oxymoron’s canonic status as pure form.

Since it is the very terms of the conception of figurality themselves (as well as their rank order)—i.e., paradigm and syntagm—that are subject to manipulation in oxymoron, one is tempted to characterize oxymoron as a metafigure, corresponding to Beardsley’s archetype.

In underscoring the dynamism implicit in the structure of tropes and articulating their life cycle, we have actually approached a trichotomy of ontological categories corresponding to the triad of Peirce’s phenomenological categories viz. Firstness (quality), Secondness (reaction), and Thirdness (mediation).

Echoing the discussion of chapter 1, the categories are defined by Peirce (8.328) as follows:

Firstness is the mode of being of that which is such as it is, positively and without reference to anything else.

Secondness is the mode of being of that which is such as it is, with respect to a second but regardless of any third.

Thirdness is the mode of being of that which is such as it is, in bringing a second and third into relation to each other.

It is the category of Thirdness that is of special pertinence to the study of tropes since all linguistic units, including figures of speech, are thirds. Indeed, tropes are epitomical instances of Thirdness, since Peirce himself conceives of it as involving the idea of ‘composition’ (1.297) or combination, defined as ‘something which is what it is owing to the parts which it brings into mutual relationship’ (1.363). Peirce also recognizes the inherent complexity which is embodied by all thirds and hence by thirdness itself, which is a ‘conception of complexity’ (1.526). This requires, of course, that thirds have a part-whole structure. Peirce’s description of the conceptual correlates of this hierarchy are quite precise and strikingly relevant to tropes (which are genuine thirds):

In genuine Thirdness, the first, the second, and the third are all three of the nature of thirds, or thought, while in respect to one another they are first, second, and third. The first is thought in its capacity as mere possibility; that is, mere mind capable of thinking, or a mere vague idea. The second is thought playing the role of Secondness, or event. That is, it is of the general nature of experience or information. The third is thought in its role as governing Secondness. It brings the information into the mind, or determines the idea and gives it body. It is informing thought, or cognition. But take away the psychological or accidental human element, and in this genuine Thirdness we see the operation of a sign. (1.538)

This passage is helpful in understanding that Peirce’s semeiotic is a theory of cognition. Not only is the sign in the category of Thirdness but so is the interpretant, defined by Peirce as ‘that determination of which the immediate cause, or determinant, is the Sign, of which the mediate cause is the Object’ (6.347). Moreover:

A sign therefore is an object which is in relation to its object on the one hand and to an interpretant on the other in such a way as to bring the interpretant into a relation to the object, corresponding to its own relation to the object. (8.334)

As I have argued in chapter 2, Peirce’s interpretant in relation to linguistic signs is to be understood as the evaluative component of sign structure, i.e., the expression of hierarchy among coordinate sets of signantia and signata in syntagms (both simultaneous and linear). Hierarchy is, as we saw, the syntagmatic counterpart of the paradigmatic asymmetry known as markedness.

Peirce’s notion of Thirdness, his tripartite division of it, and its manifestation in the semeiotic are in perfect accord with the ontological analysis of trope structure summarized in Schema A and the diachronic progression of the inherent dynamic of tropes articulated in Schema C. The qualitative aspect of the figural situation corresponds to ‘thought in its capacity as mere possibility’; it is the relatability of some two signata as conceived by mind. Their actual relation in tropes is the potential of quality realized by specific juxtapositions of signata. Finally, the representation achieved thereby, which gives shape to the informational bond forged by the correlate signata, is a cognitive structure defined by reference to an interpretant, viz. hierarchy or the embodiment of Thirdness.

3. Style as an Aspect of Meaning

Style as a term is so much a part of common parlance and has such a wide range of application (language, literature, art, music, dance, clothing, coiffure, behavior generally), yet paradoxically it has never been studied as a global phenomenon, across disciplinary lines. There is no dearth of studies of style in individual fields, particularly literature, art, and music. Nor have scholars refrained from attempting general definitions of style (e.g. Enkvist 1969, Cassirer 1975, Thoma 1976), although such essays have inevitably failed to go beyond ever more inclusive cataloguings of previous attempts. This lack of success has even led one investigator to deny any legitimacy to style as a conceptually discrete object of study (Gray 1969, 1973).

This last position is, of course, a reductio ad absurdum. The complexity of style simply requires a far richer conceptual framework and a greater concern for its ontological specificity than have either been available or heretofore been accorded the problem. Peirce’s semeiotic provides just such a framework and all the tools of analysis as well. In particular, it is Peirce’s notion of interpretant that will be seen to suggest an articulated solution.

Style in language has been characterized as ‘a marked—emotive or poetic—annex to the neutral, purely cognitive information’ (Jakobson 1972:77). In literature, painting, architecture, dress, etc., style is usually associated with a choice of features expressive of the affective intention of the subject (utterer, writer, maker, etc.). Style involves the notion of appropriateness, a goodness of fit between the context (whole) and the features chosen (parts); and/or between proper function and actual use. Where a choice is possible, the particular feature or set of features selected from a pre-established stock (paradigm, code) are in virtual opposition to those which are absent. Something is given prominence, something suppressed by being excluded. Indeed, as one goes from the maximally coded signs in human culture—the distinctive features, or diacritic categories, of phonology—to those aspects of behavior which transcend language, there is a corresponding increase in freedom of choice. While features or units do not cease to be coded as one ascends the hierarchy of parts and wholes, there is a graded gamut of codedness, including important distinctions of degree of freedom or variability at certain crucial nodes.

When Jakobson speaks of a ‘marked annex’ to the purely cognitive information, we can interpret this to mean the investment of a certain linguistic form with a certain value. The notion of markedness defines the asymmetry attending the terms of a paradigmatic opposition: one term is marked, the other unmarked. These terms are signs. Terms of oppositions combine into sign complexes called syntagms. To a varying degree (depending on the kind of complexus), the terms in syntagms are ranked, which is to say that the paradigmatic asymmetry know as markedness is mirrored in the syntagmatic asymmetry known as hierarchy or ranking. Hierarchy corresponds to value, and since value can be defined as preferential behavior, there emerges an obvious connection between style and markedness/hierarchy. The interpretant of a true sign mirrors the latter’s fundamentum relationis and is itself a true sign. The triadic relation defining the non-degenerate sign hinges on the interpretant in that it is the interpretant (and only the interpretant!) that gives the semeiotic triad its ‘assurance of Form’ (Sanders 1970:10) which subsumes assurances of Experience and Instinct. In short, the form of relation is hierarchy.

This definition of hierarchy, which involves the notion of interpretant and thereby places the subject in a semeiotic frame, appears to facilitate the search for a global conception of style in a remarkably productive fashion. The sought-after definiens must apply uniformly to undifferentiated feelings or mere intimations of style (as signified in the phrase ‘to have style’) and to its full-blown manifestations as recurrent patterns in synchrony and in diachrony.

In global terms there is no principled difference between the instantiations of style as an aspect of semeiosis in objects of variable aesthetic significance. Hence, the analysis of styles of popular clothing is no less revealing theoretically than that of art objects. When a person ‘has style’ by virtue of the clothing he or she is wearing (or wears habitually), the aggregate of clothes and accessories is marked vis-à-vis possible other ensembles which manifest no prominence of features primarily by imputation, i.e., in which the semeiotic dimension of hierarchy (value) is relatively absent. The unmarked aggregate thus can either be a specified absence (the more usual function of unmarked generally) or it can be neutral, i.e., semeiotically vague or noncommittal. The latter value of the unmarked term encompasses not only the stylistically neutral but what is called customary or normal as well. A functional match between items of clothing and the context in which such items are customary is tantamount to the absence of a hierarchization which is anything other than purely ‘factual.’ The choice of denim as a cloth for use in work clothes does not rise to the status of a semeiotic datum as long as there is no interpretant and, therefore, no semeiosis. But a student who wears jeans to a college classroom for the first time in an institution’s history elevates denim to the status of a sign. The act is perceived as an innovation, i.e., as something with incipient intellectual purport. This is the beginning of a style, semeiotically completely parallel to all such stylistic beginnings whether they occur in the history of art (cf. Schapiro 1953:304-6) or, more proximately, in the rarefied world of haute couture. At this stage wearing jeans is marked; its semeiotic value is in its perceivability/interpretability as preferential (albeit singular) behavior. This fledgling sign is what Peirce calls ‘rheme’ and defines as a ‘sign which is represented in its signified interpretant as if it were a character or mark’ (H 34). If it continues to be manifested, this innovation may become a change, having a social rather than purely individual dimension. Peirce categorizes a relation of this sort as a gradience in ‘the Mode of Being of the Dynamical Object’ (8.344) or in the ‘Nature of their [i.e., the signs’] dynamic objects’ (8.366): a ‘concretive’ dynamic object is becoming a ‘collective’ one (8.365-367). Simultaneously, and in conformity with the nature of style as a semeiotic phenomenon, the signatum corresponding to the signans becomes more intelligible in that the interpretant becomes what Peirce calls a dicent and defines as ‘a sign represented in its signified interpretant as if it were in a Real Relation [i.e., a relation of factual contiguity] to its Object’ (H 34). Since ideas (interpretants) have, according to Peirce, a tendency to grow by begetting related ideas, what started as an innovation can become a change—a social datum in the full sense—when ‘the Nature of the Influence of the Sign’ (8.373) reaches its full fruition as an ‘argument,’ defined by Peirce as ‘a sign which is represented in its signified interpretant as if it were a Sign of the interpretants’ (H 34).

What is particularly important in this application of Peircean semeiotic for the global theory of style is the structured dynamic inherent in even as trivial an example as the spread of jeans as an item of clothing among college students. We know that jeans have lost their marked status; indeed, they have become customary (the ‘norm’), a habit. But while they were still perceived as a phenomenon of style proper, the signatum or object was identified with the wearer. In Peircean terms, the sign was predominantly a dicent (more specifically, a Dicent Indexical Legisign). But the teleology of all semeiotic situations predetermines a movement towards Thirdness, towards full ‘Lawlike-ness.’ This telos is, however, at odds with the ontology of style. Style loses its particularity and becomes custom when the signantia of style lose their epitomically dicential common signatum—Man as Object. Buffon’s famous dictum, Te style est [de] l’homme même’ (cit. Cooper 1907:179), is semeiotically perfect. Peirce himself expressed just this view when he wrote: ‘The word or sign which man uses is the man himself (5.314). This implied object of the dicent explains (inter alia) the preoccupation in art-historical circles with connoisseurship and attribution as an unceasing search for the missing signatum.

Style as code is, accordingly, not style in its proper semeiotic sense. However, style extended to include fully coded units and matrices is, like all such semantic transfers, an inevitable result of the teleology of signs and meanings. As Anttila has noted (1977:38), ‘affective words are largely signs for signs.’ Moreover, there appears to be a change in ranking (‘shift in semantic mode’) as between affective and cognitive: ‘The object of the affective sign seems to be the cognitive signs’ (loc. cit.). This kind of semeiotic relationship appears to be almost completely analogous to the structure of tropes. The ontological definiens of metonymy and metaphor (along the lines of Peirce’s pragmatic maxim) resides, as we have seen, in their significate effects. In both species of cardinal trope these effects have to do with hierarchy, with the ranking of figural and literal signata in a semantic syntagm. The ontological structure of tropes, moreover, determines—as we saw was true of style—a characteristic dynamic or ‘life cycle.’ Metonymy, the primary trope, tends to be submerged in metaphor, and metaphor in turn has a decided tendency toward lexicalization (fading, petrification, idiomatization). The fully coded metaphor is no figure at all. Its demise is defined by a dehierarchization.

The semeiotic parallel between troping and style suggests a common modus significandi, determined by what appears to be a common definiens. In terms of Peirce’s interpretant, style can be seen to establish a hierarchy of interpretants (a hierarchy of hierarchizations) in which the emotional interpretant is dominant vis-à-vis the energetic and the logical interpretants; it is thus a metonymization. As style loses its quotient of freshness (the ‘gratific’ portion of Peirce’s logical interpretant, 8.372) and gains in codedness, the hierarchy of interpretants is inverted or metaphorized. Drawing the parallel to its natural conclusion, this means that style is to be defined as a trope of meaning: the annexed, or stylistic, signatum is ranked higher than the cognitive, neutral, or nonstylistic signatum.

Just as the teleology inherent in the relation of phenomena to ends leads ineluctably (as Peirce has it) to the triumph of Law or Thirdness, so the lexicalization of tropes and the normalization of style furnish us with the ground, via patterning and regularity, for the recovery of meanings embedded in texts and artifacts from past generations (cf. Schapiro 1953). Change as an aspect of continuity in human culture thus arises as a concomitant of the teleology of function in all semeiosis.

Peirce brings this matter down—in his typically idiosyncratic way—to a question of ‘ethics.’ For him, ‘the only moral evil is not to have an ultimate aim’ (5.133).