Plato’s Cratylus, the matchless dialogue on the relationship between words and things, is the first work in which one finds a detailed discussion of the question which preoccupied philosophers of language in Ancient Greece like no other, namely that of ‘the correctness of names’ (orthotes onomaton). The eponymous hero of Plato’s work, Cratylus, takes the position, espoused before him by Heraclitus, that language attaches form to content ‘by nature’ (physei), whereas his opponent, Hermogenes, follows Democritus in maintaining that things get their (Greek) names ‘by convention’ (thesei). Socrates, who inclines towards the position of Cratylus, is called upon by Hermogenes to demonstrate in his accustomed manner just how words are suited naturally to represent the things they name. For his part as moderator, Socrates, after adducing a series of examples calculated to vindicate Cratylus, comes in the latter section of the dialogue to conclude that the apparent superiority of representation by likeness over the use of arbitrary signs must be attenuated by the complementary presence of ‘custom’ (ethos) or conventionality. Cratylus accepts the view of Socrates, and the question which so engaged the protagonists remains unresolved.
Later Greek philosophy continues to be preoccupied with this controversy, the Epicureans and Stoics aligning themselves with the physei side and the Sceptics with the thesei side. In the Hellenistic period the topic reappears in a somewhat altered guise as a dispute over whether language is governed by ‘analogy’ (the Alexandrian grammarians) or by ‘anomaly’ (the Stoics of Pergamum). (Roughly, as far as linguistics is concerned, these terms were used to mean something like ‘regularity’ and ‘irregularity.’)
Although the controversy ceased to have the theoretical acuity it enjoyed among Greek philosophers and grammarians in the subsequent history of linguistics, one or another form of it is implicit in thinking about the foundations of language throughout the medieval and modern period. A kind of benchmark as far as the nineteenth century is concerned is an article by the pioneering American linguist William Dwight Whitney in the Transactions of the American Philological Association (1879) entitled ‘Physei or Thesei—Natural or Conventional,’ in which the ancient argument is raised anew. Whitney comes down on the side of those, like Plato’s Hermogenes, who view language as a system of arbitrary signs based on custom, habit, and convention. In fact, Whitney propagated this view in a number of books of the 1860s and 1870s which had a profound influence on the course of linguistic theorizing in Europe. Most prominent among those who accepted the doctrine of the arbitrariness of the linguistic sign was the Swiss, Ferdinand de Saussure.

Saussure, who along with the Poles Jan Baudouin de Courtenay and Mikołaj Kruszewski (teacher and student) is considered to be a founder of structural linguistics, made the notion of arbitrariness into a dogma of his conception of linguistic structure, which was set forth in the posthumous Cours de linguistique générale (1st edition, 1916) compiled by Saussure’s students from lecture notes. Citing Whitney at a number of points, Saussure declared the bond linking signifier (signifiant) and signified (signifié) in the linguistic sign to be arbitrary. After the publication of the Cours, this Saussurian principle became a staple of thinking about the nature of language and was endorsed by such important linguists as Charles Bally, Antoine Meillet, Joseph Vendryes, and Leonard Bloomfield.
However, even at its publication Saussure’s principle did not meet with unanimous acceptance. The prominent Danish linguist Otto Jespersen in his review of the Cours was quick to express the opinion that the role of arbitrariness in language had been grossly exaggerated. The most-influential and oft-cited rejoinder of the inter-war period was Émile Benveniste’s article (1939) ‘La nature du signe linguistique, ’which showed that what was arbitrary from the viewpoint of the outsider was necessary from that of the native speaker. Relations between components of the linguistic sign which appear to be mere accidents to the person with no knowledge of the language involved are seen as quite natural by the person to whom no other means of expression are available. Roman Jakobson, who himself contributed significantly to amending the Saussurian doctrine, repeats the anecdote of the Swiss-German peasant woman who was supposed to have queried her French-speaking countrymen as to why cheese was called fromage, remarking: ‘Käse ist doch viel natürlicher!’ (‘Käse is so much more natural!’). He reminds us also (echoing Franz Boas) that languages differ not in what they can express but in what they must express.
Despite the vigor and insistence which accompanied Saussure’s espousal of the doctrine of arbitrariness, there are passages in the Cours that represent a qualified retreat from the monolithic position usually ascribed to Saussure’s teachings. A distinction between absolute and relative arbitrariness is introduced (part 2, chapter 6, section 3), which attenuates the fundamental principle by allowing degrees of arbitrariness and a concomitant gradedness in the unmotivated nature of the linguistic sign. Not all signs are completely unmotivated; indeed, where words have constituent structure along the syntagmatic axis and an attendant identification of such constituents as members of paradigms, Saussure speaks of the ‘limiting of arbitrariness.’ In what seems like a striking about-face, the best possible way of approaching the study of language as a system is identified with this revisionist methodological tenet. Relative motivation is a necessary consequence of the human mind’s natural propensity to introduce order into the mass of irrational facts, argues Saussure, and language structure must therefore oscillate in actuality between two impossible extremes, complete arbitrariness and total motivatedness. On this view, a typology could be articulated whereby languages would be classified according to where they were judged to lie along the continuum between ‘a minimum of organization and a minimum of arbitrariness.’ Using strictly morphological criteria, Saussure ranges Sanskrit, for instance, as an ultra-grammatical type near one end and Chinese as an ultra-lexical type at the other end. They conform in structure to the two drifts he identifies in language, the tendency to prefer the grammatical instrument (contructional rule), on the one hand, and the opposing preference for the lexical tool, or unmotivated sign, on the other.
Even with the introduction of relative arbitrariness (‘degrees of arbitrariness’) into the scheme, Saussure remains unequivocally biased towards an absolutism which is basically incompatible with naturalist tendencies, and steadfastly regards absolute arbitrariness to be indeed ‘the proper condition of the linguistic sign.’
In the 1930s efforts were made to overcome this bias, notable among which (besides the Benveniste article mentioned earlier) was J. R. Firth’s book Speech (1930), where the term phonaestheme is coined and applied for the first time to describe the ‘partial’ or ‘submorphemic’ element si- with affective meaning in words like slack, slouch, slush, sludge, slime, slosh, etc. Several studies by Dwight Bolinger dating from the late 1940s and early 1950s take up and develop Firth’s idea, giving added impetus to Benjamin Lee Whorf’s contention that ‘the patternment aspect of language always overrides and controls the lexation or name-giving aspect.’ However, these efforts concentrate on evidence that is at the periphery of language, on what is very much of a piece with the phenomenon known as onomatopoeia (sound imitation), of which Saussure himself was not unmindful (together with a long list of predecessors stretching back into antiquity).
A series of studies by Roman Jakobson from the 1950s and 1960s put the doctrine of the arbitrariness of the linguistic sign in a fundamentally new perspective. Where previous investigators had left unchanged the recognition of the sign’s basic arbitrariness, Jakobson succeeded in uncovering facts of language structure (primarily, Russian) that demonstrated the extensive patterns of similarity and difference in the phonic shape of grammatical morphemes corresponding to relations of similarity and difference in their meanings. In a path-breaking article, ‘Quest for the Essence of Language’ (1965), he cited several examples of such correspondences; for instance, the relationship between singular and plural forms in all languages of the world: where the plural is formed by adding a morpheme, the singular is never distinguished from the plural by an additional morpheme. Moreover, the plural tends generally to be longer than the singular, reproducing the numeral increment by an increase in the length of the form. As pointed out by Jakobson, syntax resorts to a mimetic (imitative) representation of the order of events, with regard to time or to rank, when it records the progression of Caesar’s acts by Veni, vidi, vici, or reflects the unequal status of the subjects in a coordinated sequence like ‘the President and the Secretary of State attended the meeting.’ The mirroring of content relations in relations of linguistic expression can be seen en gros in the relationship between lexical and grammatical morphemes in all languages. A pervasive pattern dictates that the semantically more restricted class of grammatical affixes be expressed by the smaller class of sounds—vowels; and the semantically less restricted class of lexical roots be expressed by the correspondingly larger class of sounds—consonants. English is a good example of this phenomenon: only two consonants, s and t and their combination -st, occur among the productive inflectional morphemes. Russian, with an inventory of twenty-four obstruents (true consonants), limits their use in the system of inflectional suffixes to just four. Moreover, corresponding to the opposition ‘plural’ vs. ‘singular,’ Russian nouns display a relatively greater vs. lesser number of segments (sounds) in case desinences (endings) implementing these two numbers. Regardless of the specific shape of the desinences, each plural desinence contains one more segment than the corresponding singular.
Jakobson’s discussion of such correspondences in ‘Quest for the Essence of Language’ represents a major achievement in the search for principles of organization in the structure of language. His recognition that a system of sound units may diagram relations in the corresponding system of meaning units establishes in a most concrete way that the content system of language is indeed a structure, not just a purely additive code like an alphabet or Morse code. The use of the word ‘diagram’ here is not fortuitous, for one of the important methodological advances Jakobson made in this programmatic essay was to couch his strictly linguistic analysis in terms of the semeiotic, or theory of signs, of the American philosopher-scientist Charles Sanders Peirce, who gives ‘diagram’ a precise definition: a species of sign in which the relations of the parts of a thing are represented by analogous relations in parts of the sign itself. The main aspect of this definition of diagrams is the representation of relations by relations. For linguistics, this means the reflection of the relations at the content level (the level of meaning) in relations at the expression level (the level of sounds).
Peirce investigated semeiotic over a span of nearly fifty years (from around 1867 till his death in 1914), taking the name from Locke’s Essay Concerning Human Understanding, where the Greek semeiotike is adopted by Locke to mean ‘the doctrine of signs, the most usual whereof being words.’ Peirce, with whose work the study of signs may be said to have received the most thorough philosophical grounding and the richest source of insights for application to diverse fields uninvestigated by Peirce himself, carefully defined semeiotic as ‘the doctrine of the essential nature and fundamental varieties of possible semiosis.’ He was equally careful to define precisely what he meant by the word semeiosis, calling it ‘an action, or influence, which is, or involves a cooperation of three subjects, such as a sign, its object, and its interpretant, this tri-relative influence not being in any way resolvable into actions between pairs.’ The great bulk of what Peirce actually wrote on semeiotic remained unpublished and survived in manuscript form, some of it finally appearing posthumously in the 8-volume edition of his Collected Papers, most of it still awaiting publication to this day.
Saussure does not appear to have been aware of Peirce’s founding of semeiotic, and in his programmatic pronouncements concerning the status of language and linguistics he called for a general science of signs, which he gave the provisional name semiologie. Saussure was convinced that linguistics stood in direct need of this more general discipline for the proper conduct of inquiry into the nature of language as a sign system. Like Locke, moreover, Saussure accorded language pride of place among human semeiotic systems. Later in the century, the outstanding Danish linguist and theorist of language Louis Hjelmslev was to capture this preeminent status of human speech when he termed it a ‘pass-key language,’ i.e. a semeiotic system which can encompass any conceivable matter, the content substance of language being capable of including the content substance of all other human sign systems.
Saussure’s interpretation of sign structure, particularly of the linguistic sign, stressed the indissoluble linkage of the two components which he called signifiant ‘signifier’ and signifié ‘signified.’ This conception of sign and of its two components appears to constitute a wholesale adoption of a semeiotic theory with roots in Stoic logic and medieval philosophy of language. The Stoics regarded sign as an articulated whole consisting of the signifier (semainon) and the signified (semainomenon)—the former defined as perceptible (aistheton), the latter as intelligible (noeton). They systematically distinguished the relation between the signifier and the signified (which Saussure termed signification) from denotation (tynchanon) or reference, much as Peirce himself did when he called the former the sign’s depth and the latter its breadth. The medieval adaptation of the Stoic doctrine, particularly by St. Augustine, utilized Latinized equivalents of Greek terminology: signum (sign), signans (signifier), and signatum (signified). Indeed, medieval logic and the conception of sign of the Schoolmen (Duns Scotus, John of Salisbury, Thomas of Erfurt, and Peter Abelard among others) was a continual wellspring of inspiration and insight for Peirce throughout his life, during which he acquired a thorough knowledge of Scholasticism.
Peirce was not a linguist in the modern sense. He did, however, have many penetrating thoughts on the structure of language which can be found interspersed at numerous points in his writings, particularly on the general topic of semeiotic. When, in 1903, Peirce paused to take stock of the development of the theory of signs from antiquity to the twentieth century, he lamented the great void that followed upon the successes of medieval logic and attributed this neglect to the ‘barbarous rage’ which had engulfed ‘the marvelous acuteness of the Schoolmen,’ to the centuries-long detriment of the study of semeiotic. Indeed, Peirce was firmly convinced that had the Middle Ages been followed by periods of achievement of the same high order, such fields as linguistics, for which semeiotic forms a necessary foundation, would be ‘in a decidedly more advanced condition than there is much promise that they will have reached at the end of 1950.’
It is tempting to speculate what course the history of linguistics in the twentieth century would have taken had Peirce’s seminal writings on semeiotic not remained largely unpublished, hence unknown to Saussure (or to Baudouin de Courtenay). Saussure laid out a program of research in linguistics which subsumed the latter under the general science of signs while explicitly recognizing it as the most important subdiscipline of the wider study. In the middle of the century and a decade after the original Danish publication of Prolegomena to a Theory of Language, its author Louis Hjelmslev, as unaware as Saussure before him of Peirce’s contributions to the foundations of semeiotic, could still call the field ‘practically uncharted territory.’

The first linguist to become aware of Peirce’s relevance to the advancement of linguistic theory was Roman Jakobson. It is in the early 1950s that mention of Peirce and brief allusions to his theory of signs crop up in Jakobson’s articles and public appearances. The publication of ‘Quest for the Essence of Language’ in Diogenes (1965) marks a milestone in the history of linguistics: while programmatic in purport, it is the first attempt to ground the essential questions of language structure in an explicitly Peircean mode. Jakobson concentrates almost exclusively on Peirce’s most famous trichotomy of signs, that of icon, index, and symbol, by which Peirce meant to characterize the mutable relationship (‘ground’) between the sign and its object. Recasting the sign constituents as signans and signatum to conform to the Saussurian aspect under which Jakobson confronts semiology and semeiotic, he aimed thereby at a kind of trial amalgamation of the European structuralist tradition and the semeiotic legacy of the American founder of pragmatism.
There are some definite points of tangency between Saussure and Peirce. Thus, for instance, Saussure originally used the term symbol in the same sense (later abandoned) as Peirce, namely a sign in which the connection between signans and signatum ‘consists in its being a rule,’ and whose interpretation depends on a convention. Saussure singled out the concept of opposition as the basis for ‘the entire mechanism of language.’ Peirce, whose scope embraced not just linguistic signs but anything that could be interpreted as a sign because of its action, considered opposition to be the essential dyadic relation. Indeed, for Peirce ‘a thing without oppositions ipso facto does not exist,’ hence it is the study of oppositions which underlies the understanding of the mode of being of things.
There is, however, a capital difference between Saussure and Peirce that is not brought out in Jakobson’s comparative discussion. Saussure, as a linguist and founder of structuralism, took his cue from what he perceived to be the structure of language; he therefore emphasized the dyadic nature of the sign, its two-sided or dichotomous character as an entity. For Peirce, however, the sign is triadic. Semeiosis takes place when the three constituents—sign, object, and interpretant—cooperate in a ‘trirelative influence’ that brings the sign into relation with its object, on one hand, and with its 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.’ The role of the interpretant in Peirce’s conception of semeiotic is obviously central; there is nothing strictly comparable in Saussure, with the possible exception of his idea of valeur linguistique (linguistic value). In one of the shorter definitions Peirce gives of the interpretant, he calls it ‘the proper significate outcome of a sign.’ The whole of pragmatism and, therefore, the entire tangled question of the ‘meaning’ of an intellectual concept is bound up, for Peirce, with the study of interpretants. Indeed, he devoted much of his thought and writing to elaborating a typology of interpretants in the context of what he came to call ‘pragmaticism,’ in order to dissociate it from the pragmatism of William James.
The question as to how meaning comes about in language thus receives a subtle, ramified, and appropriately complex treatment in the thought of Peirce; in this respect no conceivable construal of Saussure’s ideas about signs allows semiology to rival semeiotic in depth or breadth, whatever the object of analysis. The extraneous obstacles which prevented linguists (among others) from reaping the benefits of Peirce’s life-long study of signs have for the most part now been removed, and the investigation of language as a semeiotic system ought no longer be regarded, to echo Hjelmslev, as the charting of unknown territory.
Yet the recent history of linguistics can hardly be said to reflect the rapprochement with the wider study of semeiotic that Saussure’s program, and those of Hjelmslev and Jakobson after him, invited. The multi-year hegemony of transformational-generative grammar, particularly in the United States, has had the practical effect of keeping inquiry into language as a system of signs off the agenda of linguistic theory. One of the (unintended) impediments that TG grammar has erected to the pursuit of theory along lines suggested by European structuralists and semeioticians has been its emphasis on language as an activity rather than as a work, to echo the Humboldtian terms borrowed from Greek, energeia ‘activity’ and ergon ‘work’ respectively, in which the question has occasionally been cast by TG proponents themselves.
One of the main rallying points of TG grammar at its inception was an essentially negative one, a reaction to what was perceived as the sterile emphasis on taxonomy for its own sake on the part of the so-called American structuralists (Bloomfield and his followers). Linguistic theory was proclaimed to be advanced directly in the measure of its accounting for the creative or generative capacities of human users of language. Language was not to be viewed as a corpus of ready-made formulae and patterns that speakers of a language learn by rote, but was to be grasped as the activity in which members of a speech community create and re-create speech.
The existence and influence of linguists of all schools on both sides of the Atlantic, whose methodological inclinations predisposed them to regard linguistics as a set of prescriptions for transforming a corpus of texts into a grammar of the language in question, certainly tended to contribute to an impasse in the advancement of theory. This emphasis on language as ergon may, indeed, have led to the opposite swing of the pendulum, away from inquiry into the structure of the building-blocks of language (in traditional structural linguistics) and toward its productive potential (in TG grammar), as a means of widening the compass of theory to include usage which had not yet been realized. In effect, this new preoccupation with energeia to the neglect of ergon meant a narrowing of the concept of patterning in language, since the ‘norms of usage’ (to reflect the terms propagated by Eugenio Coseriu), comprising what is historically realized and codified in a given language, did not share to the same extent in the dynamic aspect of the ‘functional system’ of the language. The identification of codedness with unproductiveness, ultimately a tendency to undervalue the patternment of inherited linguistic material in the newly-discovered interests of accounting for the creative possibilities of language use, led to the almost exclusive preoccupation with syntax and syntactic novelty that has continued to characterize the theory and the practice of TG grammarians. Not surprisingly, to the extent that this conception has contributed anything towards moving the enterprise of linguistic theory forward, its successes have chiefly been limited to the investigation of syntactic problems.
The relative freedom enjoyed by the language user in constructing sentences has obscured the complementary restrictedness in selecting the material of which sentences are made. Along the hierarchical scale of units in language, from distinctive features at the phonological level on the bottom to paradigms of sentence and discourse types at the syntactic level on the top, there is a middle ground that is constituted by words and their forms, or morphology. In the tradition of European philology and its structuralist continuation, the study of morphology occupied a position of theoretical and practical importance. In a quite direct sense, recognized from the very beginnings of systematic inquiry into language structure, words are the building blocks of language, and it is their relatively set modes of internal construction that have fostered the perception of language as a work—an ergon—in short, a made object. Since the primitive element in the make-up of words is the smallest meaningful unit, or morpheme, the relative fixity of the patterning of language has mainly to do with the fixed ways in which morphemes—grammatical and lexical— combine to constitute words and their forms.
While it is clear that language in actual use allows for the production of syntactic arrangements that are novel and the manipulation of meanings that result in semantic innovations (e.g. figural speech), at the core of language structure there is a stock of words and forms that, in their ensemble, are very much akin to a work of art. Because language shares with music, literature, and dance an unfolding along the temporal axis that is missing from the plastic arts, it is difficult to speak of it as an object, in the way that made objects—artifacts—are spoken of, due to the immediate simultaneous presence of a physical whole and all of its parts. Works of literature are closer to being physical objects, despite their dependence on time, than languages, and it is indicative of their shared statuses and characteristics that the word ‘work’ is applied to temporal as well as to atemporal manifestations of art.
Besides the organic connection between literature and language that results from the former being constituted by the latter, the manner in which literature is studied has much to contribute to the proper understanding of the structure of language. The traditional meeting ground of language and literary texts is philology, and although the scientific investigation of language for its own sake may be reckoned to have a millennial history, it is the study of language as an instrument of culture—mainly literature—that has dominated the history of humanistic inquiry throughout the literate world. The preeminence of philology is particularly marked in nineteenth-century Germany, where it was singled out as the paragon example of a Geisteswissenschaft, a ‘science of man’ in contrast to Naturwissenschaft, or ‘natural science.’ The philosopher Wilhelm Dilthey took these terms and developed their conceptual purport by linking them with a methodological dichotomy that has had a great influence ever since, that of Erklären ‘explanation’ and Verstehen ‘understanding,’ introduced originally by the German historian-philosopher Johann Droysen. On this view, explanation is the aim of the natural sciences, whereas the sciences of man (alias ‘history’) aspire to understanding. Although ordinary usage tends not to differentiate between the words ‘explain’ and ‘understand,’ since practically every explanation contributes to our understanding of things, the effect of this distinction is the inclusion of intentionality within the compass of understanding, a consideration that generally finds no place in scientific explanation. Explanation in the natural sciences concentrates on the observation and prediction of events; understanding in the Geisteswissenschaften strives to encompass the goals and purposes of an agent, the meaning of signs, and the significance of social institutions or practices.
This nineteenth-century antipositivistic espousal of understanding as a methodology for the sciences of man came to be known under the name of hermeneutics, meaning the art of interpretation. With roots in the systematic exegesis of the Bible—a decidedly philological enterprise— hermeneutics was associated particularly with studies in the philosophy of history and the beginnings of sociology as a systematic discipline. Hermeneutics declined after the passage from the scholarly scene of its great German originators and lay dormant in European intellectual life until around the middle of this century, when it revived, particularly through the efforts of the German philosophers Hans-Georg Gadamer, Jurgen Habermas, and Karl-Otto Apel.
Perhaps the most well-known feature of hermeneutic analysis is the ‘hermeneutic circle,’ by which is meant that the analyst always starts with some pre-knowledge (Vorverständnis), from which he works outward by a series of explicative steps, ultimately to loop back upon the starting point, thereby completing the circle. What is important here is that the circle is not vicious in the strict logical sense of circularity; it is more precisely a spiral, consisting of organically successive complementary links that enable the analyst to arrive at a grounded and articulated knowledge by a series of mutually-reinforcing interpretations.
Since language is the product of historical accretion, and hermeneutic analysis takes history as a kind of paradigm object of interpretation, language is thus particularly suited to study through hermeneutic method. The pursuit of philological analysis in the wider sense promoted by such nineteenth-century hermeneuticians as August Boeckh (in his famous Encyclopaedia) is fundamentally a process of ‘re-cognition’ or re-learning (on the model of Greek anagignoskein ‘know again, read’). The net result amounts to ‘knowledge of what is known,’ an increase in understanding or generalized knowledge by reconstruction reminiscent of Plato’s concept of anamnesis. In the case of language, understanding linguistic structure on this view means the analytical interpretation of the sense of accomplished cognition as embodied in grammatical facts. It is a recovery or a reconstruction of the coherence which enables facts to subsist as such.
The introduction of coherence may make hermeneutical analysis appear to be directed merely at uncovering the system underlying the facts, but this is not so. Even the most workmanlike investigation of linguistic structure aims at revealing the system of relations assumed to be immanent in the data. This usually results in a description which is internally consistent and in full compliance with the admonition of Ockham’s razor; such as one might, for instance, find in a good grammar book of a language analyzed in that manner, serving the ends of pedagogy and general information. A truly interpretative analysis, however, aspires to an explanatory understanding that goes beyond the cataloguing of linguistic units and the rules of their combination. Its ultimate goal is a re-cognition of the cognized relations embodied by the facts.
This is a task that structuralism, for all its programmatic ambitiousness, has never seriously addressed. It has contented itself with a fundamentally non-hermeneutic approach to linguistic theory, choosing to follow the causal or Galilean model of explanation customary in the natural sciences, rather than the teleological or Aristotelian model of the human sciences. In its adherence to the mathematical ideal-type of a science, linguistics has generally allied itself with the strong positivistic strain that has characterized the methodologies of all academic disciplines, not exempting the humanities.
A reoriented structuralism is not, however, incompatible with hermeneutic analysis. In the case of language, the first step to be taken in this direction is the recognition that language is a hermeneutic object. What this means is that, to the extent language is capable of objectification, it is made up of a network of inferences, akin to the explanatory hypotheses of a scientific theory. Inherent in the dichotomic structure of the sign—the linking of the signans and the signatum—is a generalization of the type ‘If A then B’ which for linguistic signs in particular implies a kind of rule of the form ‘If content A then expression B.’ The relation between sign and object, between signans and signatum, is thus fundamentally an illative one (‘A ergo B’), a circumstance masked by the scholastic formula Aliquid stat pro aliquo so often cited in support of the substitutive role of the signans (expression) in relation to the signatum (content).
Without limiting himself to language, Peirce is quite emphatic in his advocacy of illation as the fundamental relation of logic, hence of semeiotic. The relation between content and expression—and correspondingly between the signatum and the signans of the individual sign—is equivalent in form to the relation between a protasis (‘If. . .’) and an apodosis (‘then …’). ‘The copula of equality,’ says Peirce, ‘ought to be regarded as merely derivative.’ Moreover, the relation is asymmetric and transitive, hence dynamic and unidirectional. On this view, the structure of language is a system of inferences whereby content entities are assigned to expression entities through a series of interpretative translations. It follows that at the heart of this system are the interpretants, the constituents of sign structure that enable linkages of signantia and signata to make sense.
It is through the notion of sense that semeiotic and hermeneutic converge, nowhere more clearly than in the structure of language. If we accept as axiomatic (following Jakobson) the notions that ‘all linguistic phenomena … act always and solely as signs’; and, furthermore, that ‘any linguistic item . . . partakes—each in its own way—in [sic] the cardinal, viz. semantic, tasks of language and must be interpreted with respect to its significative value‘ [emphasis added], then we ought reasonably to expect interpretation to occupy the central position in the structure and theory of grammar. This is precisely the point at which the crucial importance of the fit between the theory of the object and the structure of that object transpires. The role allotted to language as a structure—to its very nature and function as a hermeneutic object—demands that the methods of inquiry into language underlying linguistic theory faithfully reflect the principles of organization of language itself.
The essence of hermeneutic is involved in Peirce’s definition of meaning as ‘the translation of a sign into another system of signs’ [emphasis added]. Translation is, after all, tantamount to the intercession of an interpretant. A more direct apprehension of the intimate connection between semeiotic and hermeneutic is provided in the Preface to a series of unpublished ‘Essays on Meaning’ which Peirce drafted in 1909. In discussing what part of Logic should study the ‘different sorts of Meanings of signs,’ Peirce adduces as a model Aristotle’s De Interpretatione, originally Peri hermeneias (On Interpretation), and suggests that this study ‘might be called hermeneutic, the science of interpretations or Meanings. Or it might be called Universal Grammar, the grammar of signs in general.’ Immediately thereafter, Peirce defines a sign as ‘anything which represents something else, its Object, to any mind that can Interpret it so.’ The convergence of semeiotic and hermeneutic via the nature of Sign is thus complete.
What remains to be determined is the precise method by which linguistics is to exploit these insights. Jakobson calls Peirce’s notion of interpretant ‘one of the most ingenious findings and effective devices received from Peirce by semiotics in general and by the analysis of grammatical and lexical meanings in particular.’ Given that the ‘essence of language’ is to be found in the inherent organization of grammar as a system of patterned relationships between sounds and meanings, precisely how are we to proceed in uncovering these relationships? A programmatic subsumption of all linguistic analysis under the rubric of meaning or hermeneutic must be augmented by a method which allows access to the structure of meaning.
The habitual colligation of signata with signantia in ready-made linguistic entities of varying breadth and depth (from distinctive features to whole utterances) tends to obscure a pivotal disjunction between the content system and the expression system of language. Although each system forms a structure, the kind of linguistic sign which constitutes the content system is categorically distinct from the kind of sign which constitutes the expression system. The sounds of language that organize themselves into a relational system called a phonology are made up of ultimate units, variously called ‘distinctive features,’ ‘diacritic categories,’ or ‘diacritic paradigms.’ What is of special importance is their status as signs with a purely diacritic function. The diacritic signs of a phonological system have the requisite semeiotic structure, being comprised by a sign vehicle or signans realized as a (relational) sound property, and a meaning or signatum, namely its diacritic function. Now, although diacritic signs contract paradigms (oppositions) and combine into syntagms which are simultaneous (i.e. phonemes) or sequential (clusters, syllables, words, etc.) in ways quite analogous to non-diacritic signs, they differ from other linguistic signs in one cardinal respect. Each diacritic sign has the same unique signatum: ‘otherness’ or ‘alterity,’ i.e. pure opposition (to all other diacritic signs).
The category to which the signs of the content system belong is fundamentally different from that of the diacritic signs, in that nondiacritic signs always have their own positive signata. The signata of content signs may consist of single content elements or of syntagms of content elements; on the basis of this division into unitary and complex signata, content signs are correspondingly divided into asynthetic and synthetic signs.
There exist content signs whose signantia have a direct phonic manifestation (e.g. the different intonation contours associated with the opposition ‘interrogative’ vs. ‘declarative’ in many languages), but content signs must typically resort to being represented by complexes of diacritic signs, each with its own signans but devoid of a positive signatum. Content signs for the most part have no material signans; diacritic signs have no individual, positively definable signatum, their signata being strictly synonymous (‘otherness’). Content signs, therefore, form oppositions and an entire system of oppositions strictly on the basis of their signata, whereas diacritic signs are opposed and comprise an entire system of oppositions strictly on the basis of their signantia.
The inherent asymmetry between the two articulations of language has a fundamental bearing on the investigation of linguistic structure and on the theory of grammar. Expression and content cannot be compared directly: the structure of language is such that purely diacritic signs possessing no meaning except otherness are used to constitute the material manifestation of content signs (more precisely, their signantia), which do possess a substantive meaning. How is this fundamental disjunction overcome by the structure of language? How is it that grammar actually presents itself as a patterned, coherent arrangement of sounds and meanings?
The answers to these crucial questions form the subject of this book. In anticipation of their greatly amplified treatment in the chapters to follow, it suffices to say here that these questions have never before been posed with the framework for definitive solutions in mind. The full implementation of the requirement of a thorough-going, unified theoretical approach to the problem of form and meaning was manifestly on the agenda of the early European structuralists, particularly the three leading Russian members of the Prague Linguistic Circle, Nikolaj Trubetzkoy, Roman Jakobson, and Sergej Karcevskij. But it remained largely a programmatic desideratum rather than an explicit achievement of structuralism, and the subsequent history of linguistics cannot be said to have made significant advances toward the solution of this all-important problem. (This is true no matter how broadly or narrowly one defines the scope of structural linguistics.)
Contemporary linguistic practice with regard to method has, irrespective of particular doctrines or persuasions, been chiefly oriented towards the description of languages (synchrony, the writing of descriptive grammars) and language states (diachrony), with a pedagogical aim in view more often than not. A concomitant result of this orientation has been the preoccupation of linguists with rule formulation, in concord with the prevailing conception of language as ‘rule-governed behavior,’ and the presumption that advances in theory are to be identified with the construction of formalisms of maximal generality and abstractness.
With the rise to near-hegemony of TG grammar has come the ascendancy of rules of grammar, not as prescriptive devices but as a means of capturing the systematic (regulative) norms inherent in linguistic behavior. Rules as statements of regularities appear to serve ends that can be considered preliminary to the task of understanding grammar and making sense of it. The hermeneutic treatment of linguistic facts, on the other hand, encompasses a notion of rule that strives to represent the ‘trirelative’ bond between signans, signatum, and interpretant. The semeiotic relation between the three elements of a sign must obtain in order for semeiosis to occur: the patterned correspondence between sets of signata and conjugate sets of signantia in language owes its coherence to the sets of interpretants that inhere in every semeiosis. ‘Rule’ in a semeiotic sense, therefore, is neither causal nor predictive, as it is in the natural sciences. ‘Rule’ from the hermeneutic standpoint is inherent in ‘interpretant,’ but the latter concept is much wider and more productive of understanding, given that the object of study is language (or language-like in form).
The displacement of the concept of rules as descriptive devices from the theory of grammar and its supersedure by the semeiotic notion of interpretant reintroduces questions about the patterned relation between expression and content. We know that the disjunction between expression-form and content-form (to use Hjelmslev’s terminology) is overcome by language, which is able to do so because the interpretants on both sides of the sign situation, being directly comparable in kind and function, bridge the hiatus between signantia without signata and signata without signantia. Interpretants are the agents of mediation between sign and object, as Peirce himself realized when he equated mediation with representation. The interpretant of the expression sign—of the phonological signantia—can be compared directly to the interpretant of the content sign—the grammatical and lexical signata. Any coherence that emanates from the bond between coordinated signata and signantia in the form of linguistic entities and collocations of entities is, therefore, to be found in the patterned relationship of interpretants.
But what exactly is the interpretant of a linguistic sign? We have ascertained that every sign of language, being the kind of sign that has a ‘tri-relative’ structure, must have an interpretant. If it is interpretants that mediate between content-form and expression-form, they themselves must be ontologically unitary, whether the domain of their reference is sound or sense. Is there a dimension of language structure which matches the function of the interpretant as agent of mediation in semeiosis, imparting form to meaning?
An answer to these questions can perhaps be traced to the traditions of European structuralism, specifically those of the Prague Linguistic Circle and its most illustrious members, Trubetzkoy and Jakobson. In 1930 Trubetzkoy discovered that the terms of a phonological opposition are not merely polar in phonetic implementation, but that their ‘intrinsic content’ is ‘contraposed.’ He identified the unequal evaluation of the terms of a phonological opposition with the presence (or maximum) versus absence (or minimum) of a ‘positive mark’ and called this conceptual superstructure of the phonological sign ‘markedness.’ In 1932 Jakobson extended the scope of markedness by applying it to oppositions in grammar, specifically the morphological categories of the Russian verb, and recognized explicitly the inherent asymmetry of markedness relations. Trubetzkoy designated the term of a phonological opposition characterized by the presence (or maximum degree) of a physical (phonetic) quality or mark as the ‘marked’ member of the opposition, and the term characterized by the absence (or minimal degree) of that quality or mark as the ‘unmarked’ member. Jakobson’s extension of markedness to grammar (and lexis) brought out the fact that members of grammatical and lexical paradigms are not defined individually by their absolute referential scope; rather that whole paradigms, both dichotomic and graded, diagram differences in referential substance with the ‘skewed projection’ dictated by the asymmetry of such paradigms. The marked term of an opposition has a narrowed referential scope, while the unmarked term is broader in the scope of its application to the field of reference. One part of the referential field must be represented by the unmarked term of an opposition, but the remaining part may be represented by either the marked or the unmarked term. For instance, in the grammatical representation of time, the substantive opposition anteriority vs. non-anteriority to the speech event is rendered by the formal tense opposition ‘past’ vs. ‘non-past,’ such that non-anteriority is unambiguously signaled by the non-past—the unmarked member of the opposition—whereas anteriority may be signaled either by the past tense—the marked term of the opposition—or by the non-past (here, the so-called praesens historicum). In this example it is clear that the contradictory opposites of the referential category of time are so rendered grammatically that one member of the opposition of tense includes the other member. This broader scope of the unmarked member is similarly reflected in lexical oppositions such as English man vs. woman, where the former is the generic (unmarked) designation of humankind, while the latter is reserved for the designation of only a subset of the referential field.
It can be seen that the concept of markedness facilitates the unitary conception of the structure of phonology, grammar, and lexis—in short, of language. This unitary conception is at the center of the long-standing supposition (stemming from the work of Jakobson and Hjelmslev in the 1930s) that different levels of language structure are governed by identical principles of organization, which is to say that the levels are isomorphous. The isomorphism is mirrored in part by the formal identity of the definition of markedness as it pertains to the diverse elements of both the expression system and the content system. Despite differences stemming from the disparity in focus—the phonic level of signantia in the case of expression and the semantic level of signata in the case of content—all instances of the marked term share a narrowed specification and a circumscription of scope vis-à-vis their unmarked counterparts. In phonology, the marked term of an opposition constrains or narrows a certain (negative or positive) relational sound property which is relatively (and polarly) unconstrained and uncircumscribed in the corresponding unmarked term. In grammar and lexis, the narrowed definition affects a conceptual item, delimiting the referential scope of the marked term vis-à-vis the relatively unnarrowed scope of the unmarked term.
The Prague School concept of markedness is largely confined to linguistics and the study of language structure, despite some inklings as to its applicability to other areas of human behavior. Its fundamental role as a semeiotic universal is adumbrated somewhat more sweepingly (if inchoately) by Saussure’s famous dictum that ‘language is a system of pure values.’ Unfortunately, Saussure failed to integrate his notion of linguistic value with his sign theory, and the subsequent development of European structuralism, both before and after the Second World War, does not include a ramified appreciation of the relationship between markedness and value. (An awareness of Peirce and of his semeiotic would no doubt have facilitated the progress in understanding grammar that is now finally emerging, owing to the wider dissemination of Peirce’s philosophical writings).
Indeed, the idea can now be advanced with some confidence that markedness is a species of interpretant, fully compatible in its own way with the system of interpretants established by Peirce (see chapter 1 herein). One of Peirce’s (many) definitions of sign 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.’ If by signification is meant the action of a sign whereby the interpretant is brought into relation with the object of the sign, it is understandable why Peirce saw the sign’s ‘essential significant character [as] the character of causing the interpretation of its object.’ The being of the sign, therefore, consists in its causing an interpretation; in other words, in causing an evaluation of the relationship between sign and object.
The evaluative or axiological dimension of the sign’s connection with a system of interpretants is implied by Peirce’s discussion of semeiotic but has not been clearly perceived. The interpretants of linguistic signs are values—markedness values. While markedness is subject to grading, degrees of markedness are expressible exclusively in terms of just two values, ‘marked’ and ‘unmarked,’ which imply each other but are fundamentally asymmetric. The asymmetry of the linguistic sign in its paradigmatic dimension of markedness emerges in its syntagmatic dimension as ranking or hierarchy. Thus the relation between signans and signatum which gives rise to signification always comports some measure of significance or value. If this were not true, sign relations would not conform to a pattern because there would be no overarching principle of order.
‘Every single constituent of any linguistic system,’ writes Jakobson, ‘is built on an opposition of two logical contradictories: the presence of an attribute (“markedness”) in contraposition to its absence (“unmarkedness”). The entire network of language displays a hierarchical arrangement that within each level of the system follows the same dichotomous principle of marked terms superposed on the corresponding unmarked terms.’ While this formulation faithfully reflects the original Prague School understanding of markedness as being associated with attributes or marks, and thereby with substance, the status of markedness as an interpretant points to its proper place in the form of language. Markedness is a matter of conceptual complexity and as such is to a significant extent independent of the substance of language. Conceptual complexity is tantamount to (grades of) value. Thus every linguistic opposition, besides consisting of a signans/signatum duple, has an evaluative superstructure defined by the two polar values, marked and unmarked. These values constitute inherent semeiotic definientia of a given opposition’s terms. The scope of markedness as the dominant principle of conceptualization is not limited to language: it inheres in the patterning of all human semeiotic systems, hence in all of human culture. Its asymmetric character, moreover, is clearly rooted in biological and neurophysiological isomorphisms, namely the structure of the genetic code and the lateralization of the brain. The mental capacity of human beings is defined by the universal principle that there is no conceptualization without evaluation: the integration of concepts into paradigms and syntagms necessarily involves grading and ranking, i.e. markedness.
The concept of markedness can be advanced materially beyond its programmatic proclamation as the universal semeiotic principle underlying the organization of linguistic structure. One of the chief tasks set out in the chapters to follow is the aggrandizement of the Prague School notion of markedness to embrace discoveries of its nature and workings made in the last thirty-five years or so. The access to the principles of organization governing linguistic structure provided by markedness also affords a way of returning to the question posed at the very beginning of this section, that of physei vs. thesei.
Jakobson’s identification of the ‘essence’ of language with the fact that the system of signantia may diagram relations in the corresponding system of signata, and that these expression/content mapping relations pervade the entirety of language, establish language in part as what mathematicians call an ‘automorphism,’ i.e. as a structure defined by relations of symmetry between its parts. To the extent that such symmetry or congruence is manifest in language, it is an affirmation of the ‘naturalist’ (physei) position. Language conforms to nature by the fact that it diagrammatizes content in expression.
The veracity of the physei-as-diagrammatization position is directly ascertainable from a consideration of the relations in a synchronic grammar. This does not mean, however, that mapping relations are irrelevant to the problem of language change, more specifically to the assumption that change is to a large extent motivated rather than arbitrary, just as relations in the structure of the linguistic sign itself. Such relations of semeiotic congruence are undoubtedly involved in the teleology of function characteristic of linguistic change, although they are typically covert and not accessible to direct observation by the grammarian while in statu nascendi. Indeed, covert patterns of correspondence determine tendencies of development, so that the drift of a language can be explicated as a gradual actualization in its surface forms of virtual patterns, patterns that are established over time as part of the linguistic competence of speakers. It is these patterns that constitute the functional system or the productive center of language, in contradistinction to its norms or unproductive periphery, and determine which deviations from the received grammar will be accepted, which rejected. The dynamics of language follows a trajectory of maximizing the patterns of diagrammatization and minimizing or ultimately eliminating those that are devoid of such semeiotic basis. This understanding of the telos of linguistic change brings synchrony and diachrony into an inalienable structural relation: change is thereby conceived as an aspect of continuity.
The presumption underlying all contemporary inquiry into language—that it is a system—also entails the search for patterns of coherence among linguistic facts. The semeiotic perspective on language structure consonant with Peirce’s fundamental discoveries of the nature of signs is informed by three cardinal interconnected tenets. First, there exist semeiotic universals—principles of organization—which govern the patterning of linguistic data. Second, the patterning is coherent, which is to say that the genuinely structured or motivated sets of facts—the functional system or structure sensu stricto, as distinct from the norm-governed adstructure—are explicable and to be understood as cohesions or correlations between expression-form and content-form. Third, the patterning of form/meaning correlations owes its coherence to a mediating interpretative component of semeiosis or ‘structural cement’ that binds the facts together and allows them to subsist systematically alongside each other. This component, corresponding in all essential details to Peirce’s interpretant, is markedness.
Why are certain specific expressions associated with certain specific contents? This utterly basic question has, remarkably, never been posed in the history of linguistics, perhaps because it seemed absurd to ask why a fact can be a fact. But that is precisely what needs to be inquired into so as to arrive at a truly explanatory theory of grammar, a theory of language facts that satisfies the requirements of the hermeneutic understanding of a hermeneutic object.
The semeiotic values that enable sounds and meanings to cohere in a pattern are markedness values. The search for principles of organization, for coherence in language structure, is thus an investigation of the ways in which markedness values arrange themselves in language, giving this most important of all forms of semeiosis its status as a system.
No linguistic entity is without its markedness value, since every linguistic entity participates in a network of oppositions whose nature and significance are directly determined by markedness. Language is a system of signs, a semeiotic; therefore all such entities are signs and contribute as parts to the whole that is a semeiotic system. While heretofore such stock items of linguistic description as stems and suffixes, including their positional shapes or alternants, have been looked upon simply as artifacts which facilitate an economical, internally consistent statement of distributional facts, now these entities must be viewed as having semeiotic values—markedness values—which vary coherently and uniformly with contexts and the values of contexts.
There is thus at the core of structure a coherence of facts, which resides in the patterned cooccurrence of contexts and units accompanied by a coordination of their markedness values. The circularity inherent, furthermore, in manifestations of coherence must not be viewed as a defect: quite the contrary, it is of the very nature of language as a hermeneutic object. To conceive of facts as cohering with other facts, as contexts do with units, is to recognize circularity as a definiens of coherence. The search for ‘independent motivation’ in linguistic explanations is actually a distorting imitation of the Galilean mode of the natural sciences. The notion of coherence consonant with the Aristotelian mode appropriate to hermeneutic conceptualization entails circularity as a virtue, owing to its immanence in the structure of language. Both the theory of grammar and the method of analysis leading to the proper understanding of linguistic facts cannot dispense with circularity for the simple reason that it is of the essence of language.
Linguistic facts must be recognized for what they are, the actual variations of language rather than the ‘underlying forms’ or ‘deep structures’ posited by contemporary practitioners. A theory of grammar which places the matter of the sense of grammatical alternation at the center of its agenda considers variations of form associated with variations of meaning to be its proper explananda. It substitutes for the question ‘How does one get from deep to surface structure?’ the question ‘Why are the surface facts of grammar as they are?’ Seeking the answer to such a radical question presupposes the belief that ‘surface’ variations—the actual stuff of language—do not vary haphazardly but organize themselves into a semeiotic, a system of signs. Surface variants and alternations are thus seen not as mere agglomerations of data to be systematized by the linguist’s intervention and appeal to formalisms at a putatively deeper (hence ‘truer’) level of linguistic reality, but as entering directly into a pattern of semeiotic relations with each other.
Transposing the theoretical enterprise of linguistics to another dimension, away from the mechanistic and scientistic impasse in which it has been mired in the last quarter-century, means formulating a theory of grammar that puts fundamentally different questions to its data and frames them in a fundamentally different mode, one defined by the nature of language as a hermeneutic object. The replacement of causal explanation by hermeneutic understanding as the province of theory entails the jettisoning of conceptions of language structure and linguistic method that result in the prevailing self-confinement to goals that are fundamentally (if unwittingly) non-explanatory. The pursuit by TG grammarians of a ‘complete scientific description of the language’ corresponding to ‘a fluent speaker’s knowledge of his language’ expressed in the construction of rule formalisms is, therefore, fundamentally irrelevant for linguistic theory: a theory of grammar is not a theory of knowledge but a theory of habit, in the sense imparted to the word and the concept by Peirce’s pragmaticism (see chapter 1). Explanation aspiring to hermeneutic understanding must focus on why the data of language cohere as signs, not on mechanisms by which grammatical forms can be derived by the judicious choice and application of rules (ad hoc or not). This requirement once and for all removes predictability-via-rules from the agenda of theory.
The entire recent history of linguistics demonstrates with great clarity the feasibility of forcing data into a proliferating number of mutually compatible formalized configurations or notational variants. It is characteristic that these frameworks, and the schools with their adherents that they represent, take no cognizance of the principle (laid down by Jakobson among others) that all linguistic entities participate above all in the semantic tasks of language and must be interpreted in terms of their significative value. It is obvious, on the other hand, that even the interests of a purely descriptive linguistics are ill-served by an attitude toward language that ignores its status as a semeiotic.
The grammarian writing a description of a particular language must accept the burden of showing how the various grammatical rules he formulates stand with respect to their semeiotic function. Since linguistic rules are such that one entity or structure is transformed into another entity or structure in a given context, they thereby purport to act as an interpretant which gives a means of representation (signans) to an object of representation (signatum). If this is so, the grammar writer cannot limit his task to formulating rules that merely register generalizations about the distribution of entities in texts, or which transform structures of one kind into structures of the same kind without any change in information or function. A linguistic description which lays claim to being the faithful account of a speaker’s language competence cannot evade the responsibility of explicitly characterizing the semeiotic status—the ‘significative value’— of all the primitive elements and all the effects wrought on them by the rules formulated to encompass them.
It is a plain fact that the mainstream of linguistic practice has failed to conceive its tasks in terms of this responsibility. In so doing, linguists have ignored the fundamental truth that language is a semeiotic.