The Course begins (after a brief historical survey) by addressing the key issue that makes linguistics as a discipline methodologically problematic: the lack of a clear definition of its object of study.1 To say simply that a linguist studies “language” is not to address the problem at all, since a variety of other disciplines also deal with language. Language figures, one way or another, within the framework of anthropology, sociology, psychology, acoustics, physiology, history, philosophy, philology, cultural studies—a deceptive richness that in fact leaves language without an epistemological home of its own: “In fact, the whole world is occupied with it to a greater or lesser extent; but—a paradoxical consequence of the interest it has attracted—there exists no other field that germinates more absurd ideas, prejudices, mirages, and fictions” (CLG, 21–22/7). Barely disguising his irritation under the terse rhetorical veneer of his lectures, Saussure declares it the “first and foremost” duty of a linguist to “denounce” all these untruths, making them “dissipate as completely as possible” (CLG, 21–22/7).
Saussure’s epistemological critique of the foundations of linguistics in his lecture courses reflected pressing concerns he had had since the early 1890s. In the now famous note beginning with the words “Unde exoriar?” (Whence to begin?)2 and elsewhere,3 Saussure persistently reiterated his thesis that a snapshot of language could be taken from an indeterminate plurality of possible angles. Consequently, he spoke of the “grand necessity” of separating linguistics from other sciences dealing with language.4 We read in the notes that any idea concerning language can be subjected to “twenty kinds of analysis” (ELG, 232/164); this point is echoed in the introductory section of the Course, which emphasized that linguistics should be “carefully distinguished” from the way language is treated by other disciplines (CLG, 21/6).
Saussure’s position was consistent with the antipositivist revolution in contemporary epistemology. The spirit of the times, with its rising skepticism about positivist “facts,” which only recently had seemed irrefutable, coupled with the quest for an explicit methodology, can be perceived in Saussure’s notes when he remarks, for instance, that “the past of linguistics consists of a general doubt about its role, its place, and its value, accompanied by a colossal acquisition of facts.”5 This is what distinguished Saussure from virtually all of his colleagues—“pure” linguists whose “exceedingly amusing” lack of awareness of the cognitive abyss over which they trod, as if they were in possession of the truth itself (ELG, 116/81), was a constant source of consternation for him.
Proponents of the new trend in epistemology and philosophy of science emphasized that any object of study always appears to be a mental construct that rests on certain postulates, whether consciously formulated or tacitly implied. By explicitly outlining the categorical boundaries of its subject, a discipline is prevented from drowning in the intellectual muddle (the Saussurean “twenty kinds of analysis”) that occurs when such boundaries remain vague.
Even before major modernist philosophical schools, such as neo-Kantianism, empiriocriticism, phenomenology, and the Bergsonian concept of duration, emerged at the turn of the century, the new trend could be seen in the natural sciences, notably in the works of Hermann von Helmholtz. In a popular summation of his studies of optics, Helmholtz (1996b [1895]) proclaimed that there is no such thing as the “natural” appearance of objects. What we perceive as visual snapshots of objects are in fact signs whose apprehension involves certain skills. As if anticipating one of Saussure’s central postulates about language, Helmholtz points to a “striking analogy” between visual perception and “another system of signs which is not given by nature but arbitrarily chosen”—namely, “the words of our mother tongue” (Helmholtz 1996b [1895]:274–276). In another work, Helmholtz applied the same principle to sensory experience in general: “our sensations are, as regards their quality, only signs of external objects, and in no sense images” (343).
Saussure’s use of the metaphor of “vision”—that is, of the need for linguistics to learn how to identify its object among the plurality of projections in which it appears—might have been directly adopted from Helmholtz’s optical studies. Saussure’s identification with the methodological premises of the “new physics” (and also, perhaps, the “new chemistry” of Ostwald and Mendeleev) was just the flip side of his scornful rejection of the old-fashioned scientific claims of the Neogrammarians (the Junggrammatiker). The irony of the situation was that while the Neogrammarians envisioned language as a positively given object that could be described by the methods of the (positivist) natural sciences, the new trend in science sought an analogy with the “arbitrary signs” of language for explaining the phenomena of nature in their relation to cognizing consciousness.
If the “fundamental truth” about language cannot be simply pointed out, the only way to “retrieve a firm terrain” for linguistics lies in a “comparative critique” of different points of view on language (ELG, 199/137). This “comparative critique” is essentially a reductive task: it constructs an object by subtracting from it what lies beyond its postulated boundaries. Edmund Husserl’s Ideas of Pure Phenomenology and Phenomenological Philosophy appeared in 1913, the year Saussure died. Yet, setting aside the question of any direct influence, one can detect a pronounced parallel between Husserl’s phenomenological reduction and Saussure’s negative epistemological strategy—a parallel that attests to the extent of Saussure’s awareness of the philosophical culture of his time in general. Like Husserl, Saussure proceeds by taking away all of the substantial features of language, both “objective” and “subjective”: on the one hand, its material fabric (sounds), and on the other, the perceptions language evokes in the mind of a speaker. Demonstrable as they are when one observes how people speak, the substantial material and psychological features of language are not transcendental, in the sense that they do not exist as a priori (in Saussure’s term, veritable) dimensions of language outside the manifold conditions of its usage which, if taken into the picture, “open the door immediately to thousands of contestations” (ELG, 81/55).6
Saussure calls the transcendental object that emerges out of this reductionist critique la langue, to distinguish it from le langage, i.e., speaking activity in general whose various dimensions could be studied from a number of perspectives. La langue is a construct in which language is reduced to its inalienable features, the ones that belong to it unconditionally. It is a product of an immanent vision that considers language for its own sake, ignoring what language may be used for or how it is related to various human faculties and institutions.
While on the one hand postulating a priori features defines an object in its own right, by the same token it also points to its place among other similarly constructed objects. When, however, after constructing “the true and unique” (i.e., a priori given) object of linguistics, Saussure turned to defining its place among other disciplines, he encountered insurmountable difficulties that undermined the results of his epistemological labor. For la langue, as it was postulated, showed itself to have certain features that set it apart from all the known phenomena, natural and social alike.
The fundamental units of language, its inalienable property, are signs. A sign is a bipolar phenomenon: it couples a material representation with a content that is represented. That in itself, of course, has been common knowledge since Aristotle and St. Augustine. What Saussure realized was that for the sign as a linguistic phenomenon, its substantial physical shape and content are irrelevant. As a matter of principle, any physical shape can be coupled with any domain of meaning to form a sign. How many distinct signs cover a certain domain of content in a given language, and how the forms of those signs are related to each other, is a matter that rests solely on the composition of that language, which is shaped by blind convention. What matters for a sign is the bare fact of interconnectedness between some physical shape and some content and not the nature of what is connected.
That the connection between the form and the meaning of a sign is a matter of social convention—a contrat social between speakers of a language—was an idea introduced by French philosophers of language of the Enlightenment (to whom Saussure usually referred generically as les philosophes), particularly by Rousseau and Condillac, and applied to linguistics proper by Whitney. Saussure praised these thinkers on numerous occasions for emphasizing the social factor in language. However, he found missing from the argument of his predecessors a crucially important point, which ensued from the peculiarity of the double nature of signs.
When stripped of any substantial outward properties, the two sides of a sign turn out to be inseparable; they are “united in an absolutely indissoluble way” (ELG, 73/48); both sides exist only vis-à-vis each other, correlatively (CLG, 99/67). A sign has nothing to show but the fact of correlativity between its two sides. Its content is contingent on representation, while the representation is contingent on the content it represents (CLG, 144–145/101–102). The moment one side is separated from the other, it simply ceases to exist as a phenomenon of language. If we listen to speech without recognizing its meaning, what we hear is a continuum of sounds; speakers’ ability to apportion this continuum into distinct units rests on their realization of the meaning attached to them. Yet while the continuum of speech owes its segmentation to the meaning carried by distinct sound combinations, the continuum of meaning turns into tangible concepts by virtue of its representation through distinct physical entities. Meaning is like hydrogen filling a balloon: it is what makes the balloon fly, but without the balloon, it would dissipate without a trace (ELG, 115/78). “The two chaoses yield an order by uniting. Nothing could be more futile than to try to put them in order separately” (ELG, 51–52/32).
Again and again, Saussure comes out with thunderous denunciations of the traditional view—he calls it “nomenclaturism”—according to which words are nothing more than the labels pasted on certain ideas, assigned to a content that is already “there.” Saussure finds the idea “most vulgar” (ELG, 106/70). He does not mean, of course, that no thinking is possible outside language. But “without resorting to signs, we would be incapable of distinguishing two ideas in a manner that is clear and consistent” (CLG, 155/110). Human thought would be “nothing but an amorphous and indistinct mass” but for the fact that it is apportioned into distinct entities of meaning by being represented by distinct material carriers (CLG, 154/109).
Saussure rejected the traditional terms “form” and “meaning” for precisely the reason that they implied the possibility of observing the two sides of a sign as separate phenomena, apart from their mutual relations. In early notes, he suggested composite terms instead: “vocal figure” (la figure vocale), i.e., a configuration of sounds shaped by its correlation with a meaning, and “form-sense” (forme-sens), i.e., a configuration of meaning shaped by its correlation with a sound (ELG, 17/3). It was only in the second, revisory part of his last lecture course that Saussure abandoned awkward composite denominations in favor of the terms signifier and signified (signifiant and signifié), whose etymological kinship reflected their interdependence; emerging at the very last moment, they nevertheless eventually became standard items of Saussurean terminology—one of his terminological signatures—thanks to the published Course.
Thus, although the dual configuration of signs is based on a social convention, it is a convention of a peculiar kind. Conventions of social behavior, while a matter of social acceptance, are never totally unrelated to the primary experiences of those who accept them. That one is supposed to fall prostrate nine times when facing the Chinese emperor is, of course, a conventional act, as far as the number of the bows is concerned; but the gesture of prostrating oneself as an expression of submission and awe has a certain experiential ground (CLG, 101/68). Compared with this, the linguistic “gesture” of moving one’s organs of articulation in a certain way to express submission and awe has nothing to do with the experience of submission and awe. Language as a social institution is not merely a convention but the empty shell of convention; in this sense, it can be called a “pure institution” (ELG, 211/147).
It is in this sense that the Course proclaimed the inner structure of signs to be “arbitrary.” Speakers accept the signs of their language as they are, without asking for any logical or empirical justification of their dual configurations. True, the configurations themselves may change as time passes. These changes, however, are also arbitrary. A particular change can be justified after the fact: a sound change may eliminate an awkward sound combination; a new word, or a new meaning of a word, may emerge to fill a conspicuous gap in naming certain phenomena. But language development at large knows no teleology.7 The inner structure of a language at any particular point in time is a matter of pure “chance” (hazard); it reflects nothing but the blind vicissitudes of the tradition that maintained the transmission of the language over millennia, a process in which it shed and acquired features unpredictably (ELG, 262/188).
Just when Saussure succeeded in postulating the constitutive property of language, he came to realize the peculiar predicament faced by the discipline due to the unique character of its subject. Although Helmholtz declared all sensations to be a result of learning rather than direct experience, he did not deny the existence of an inherent connection between an object and its representation: “an image must, in certain respects, be analogous to the original object.”8 Even though a natural phenomenon has to be mentally construed in order to be properly cognized, its primary perception, at least, is given directly in experience; there are limits, set by the primary experiential features of an object, that its mental presentation does not overstep.
Saussure agrees: he is convinced that all descriptive disciplines except linguistics have in their possession primary “facts,” either material or social, that are given, at least as an initial approximation, in experience. Zoology is occupied with living organisms, physics with physical bodies, chemistry with chemical elements, history with historical events and actors, sociology and anthropology with social institutions and cultural rites. How these facts are viewed, how their properties are explained, depends, of course, on their conceptualization. But the primary apportioning of experience into entities to be conceptualized is more or less evident from the beginning.
In this respect, language is dramatically different. The “facts” that speakers experience are inherently heterogeneous: “neither sounds nor ideas” (ELG, 250/178), but rather, “signs-ideas” as symbiotic unities (ELG, 20/5). A particular apportioning of meaning among signs in one language may prove totally alien to another language; as a matter of fact, symbiotic entities constituting signs never repeat themselves precisely within the frameworks of different languages. We would search in vain for a common principle that could serve as a basis for comparison for all the varieties of sound-cum-meaning combinations to be observed in various languages.
A scholar dealing with metals, minerals, or biological species can rely on a primary set of observable phenomena as his starting point; however, the configuration of signs in one or another language is “strange and perplexing”—pure “play” that finds no support among perceptible phenomena (CLG, 149/105). An anatomist does not dissect a cadaver indiscriminately, he follows the natural joints of the body; yet nothing indicates in advance to a linguist where he must “cut” an amorphous “band” of sounds or a cloud-like conglomeration of meanings flowing in speech (ELG, 257–258/184). How can one classify “objects” consisting of two totally heterogeneous components that are brought together in a fashion so capriciously idiosyncratic as to defy any common logic? Imagine, Saussure remarks in frustration, a scientist given the task of sorting out “species” that present themselves as a cross between a horse and a “bizarre” iron plaque or a sheep whose body is extended with a brass ornament; no doubt he would reject such a task as absurd. Meanwhile, linguists occupy themselves with such absurd “species” with unperturbed assurance, as if they were only natural (ELG, 18/3).
A symbiotic conflation of the material and the spiritual makes language into an object for which any epistemological analogy turns out to be imprecise. In a passage in the Course, Saussure speaks of the “mysterious” phenomenon of the “thought-sound” (pensée-son), which is neither the “materialization of thought” nor the “spiritualization of sound” (CLG, 156/111). It is concrete and abstract at the same time: concrete, because it always deals with particularly apportioned entities of sound and meaning; abstract, because it deals not with the sound and the meaning as such but with their interplay.
According to Saussure, linguistics, together with studies of other systems of conventional signs—among which he lists certain social rites, forms of politeness, military signals, games, and so forth—should form a peculiar domain of study that belongs to neither the natural nor the social sciences.9 He proposed calling this new discipline semiology (CLG, 33/15).10 This decision, however, does not remedy the fact that a linguist cannot rely on established methods that are valid for other disciplines. Saussure’s postulate of the arbitrariness of the dual configuration of signs excluded the possibility of anchoring language either in natural laws (as claimed by the Neogrammarians) or in universal logical categories (as would be done later by generative grammar). The whole field of “semiology,” of which linguistics obviously constituted by far the most important part, had to be invented anew as a domain of cognitive activity belonging to a class of its own.
It was in this sense that Saussure spoke of the “fatality” of the science of language (ELG, 227/159). The conclusion he arrived at is aporetic: it is the very process of the cognitive construction of language that makes it a phenomenon that defies any firm theoretical basis for such construction. This leaves language totally at the mercy of the point of view from which one views it. “Far from the case of objects preceding a point of view, one can say that it is the point of view that creates an object [of language], while nothing tells us in advance that one of the ways of considering the facts in question should precede or would be superior to the others” (CLG, 23/8). Paradoxically, language “in itself and for itself” turns out to be a phenomenon about which one has to ask oneself whether there is anything there but “our points of view multiplying without limit” (ELG, 67/44).
Arbitrariness: The Categorical and Empirical Dimension
Saussure’s understanding of language as a phenomenon of a peculiar kind rests on the core assumption that the relation between the two sides of a sign is arbitrary, i.e., not grounded in any natural order of things or logical pattern. The absence of any rationale, save blind convention, for any particular signifier and signified to be coupled in a sign makes its duality irreducible. The infinite variety of signifier-signified constellations that can be observed in languages never coalesces around an ultimate general principle; the structure of a sign system as a whole remains essentially unpredictable as it emerges from one case or another, one language or another, one historical moment to the next. Any emerging patterns prove, under scrutiny, to be merely local patches of orderliness that never become a general rule. The idea of the ultimate unruliness in the manner by which signifiers and signifieds are tied together by convention stands at the core of the radical novelty of Saussure’s view of language, setting it apart from the conventionalist approach of his predecessors.
Yet for all the enormous weight Saussure gave to the concept of arbitrariness,11 he never expounded it in a definitive and unambiguous way. Paradoxically, it was perhaps Saussure’s eagerness to explain all the consequences of arbitrariness, which indeed are formidable, that compelled him to skip over the concept’s definition in haste, leaving it, both in his notes and in the lectures, barely outlined, often by pointing to random examples that proved more confusing than helpful. In his typical manner of resorting to emphasis whenever he sensed holes in his representation of an issue, Saussure admonished his listeners that the consequences of arbitrariness are “innumerable”; however, not all of them could be made immediately evident: large exploratory “detours” were needed for the “primordial” importance of the principle to be appreciated (CLG, 100/68). If he had hoped that the concept could be wrested from the danger of trivialization solely by virtue of those “detours,” it was long before such hopes were realized.
It is easy to mistake arbitrariness for something no more significant than a scholastic paraphrase of Shakespeare’s “What’s in a name? That which we call a rose / By any other name would smell as sweet.”12 One need hardly be a linguist to realize that the sound shape of the word rose has nothing to do with the object it “names.”13 The situation recalls what Saussure once said about language in general: it seems obvious to everyone what “language” is, yet it is precisely the deceptive clarity of the quotidian perception that plunges the matter into a fatal confusion. Saussure himself gave “aid and comfort” to this kind of trivialization of the issue by illustrating arbitrariness—perhaps on the spur of the moment in a lecture—with a seemingly obvious yet in fact misleading example: namely, that there is no reason why a certain animal species (bull) needs to be represented either by the signifier b-ö-f (boeuf) or by o-k-s (Ochs[e]); this just happens by convention “on either side of the border” (CLG, 100/68). As Jakobson rightly pointed out, this passage in the book succumbs to the “nomenclaturist” understanding of the issue that Saussure himself never tired of fighting: it implies that there is a ready concept waiting to be given one or another “name” by convention.14 To claim that “the same” animal is named differently on either side of the Franco-German border obfuscates the fact that the signifieds represented by boeuf and Ochse only partially overlap; their overall semantic spaces are as incompatible as their sound shapes.15
This lack of precision in defining the concept opened the way to the criticism that, on several points, it seemed to contradict empirical observations of linguistic data.
First, arbitrariness as a universal principle seems to be undermined by the existence of so-called iconic signs, such as interjections and onomatopoetic words, in which the physical shape of the signifier is homomorphous to certain features of the concept it represents. Saussure tried to refute the menace from this corner by asserting that such cases are few and that they are all “semi-conventional” anyway (CLG, 102/69).16 This rather feeble defense could only provoke accusations that he “exaggerated considerably the role of arbitrariness and underestimated sound symbolism.”17 As later studies—notably by Jakobson, who consistently expressed hostility to the Saussurean idea of arbitrariness—suggested, iconicity in language seems insignificant only to those who adhere to an extremely narrow definition of the issue. A more inclusive approach not only reveals the much wider presence of iconicity in the vocabulary,18 but shows its manifestation in numerous morphological and syntactic patterns whose structure reflects certain features of the concept they express. Crossing into the domain of the linguistic ideas of Russian futurist “transrational poetry” (Khlebnikov 1987 [1913]), with its intense quest for universal sound symbolism, Jakobson suggested that it is iconicity rather than arbitrariness that underwrites language as a universal principle (Jakobson and Waugh 1987:chapter 4).
But even setting aside the contentious issue of iconicity, a second formidable empirical challenge to the universality of arbitrariness comes from a widespread phenomenon that Humboldt called the “inner form.” As Humboldt pointed out, in many cases the form of a word represents its meaning not directly but by referring to another meaning. In such cases, the coupling of a signifier and a signified acquires a rationale because of its reference to another sign. We recognize that the signifier teacher represents the meaning “he/she who teaches” because of its relatedness to the verb teach and the agentive suffix -er; the meaning of housekeeper emerges out of its reference to keeper (in its turn referring to keep and -er) and house; or as Saussure himself pointed out, while there is no reason (at least, as far as modern French is concerned) why the sound combinations deux and vingt should signify “2” and “20,” respectively, the relation of vingt-deux to its meaning “22” is motivated by its reference to the primary signs from which it is derived. In words with an “inner form,” the relation between the signifier and signified is constructed rather than arbitrarily declared.
Saussure’s response to this problem, once again, was not convincing. To cover cases such as vingt-deux, he introduced the notion of “relative arbitrariness.” He even suggested that some languages, notably Chinese, due to the sparsity of its derivational patterns, may feature relative arbitrariness to a lesser degree (and, by the same token, show a stronger propensity for “absolute arbitrariness”) than languages such as Indo-European and Finno-Ugric, while Semitic languages occupy an intermediary position between these polar types (ELG, 301–302/210). Introducing “relative” arbitrariness invalidates the concept as a universal principle, turning it into an empirical feature that various languages possess to different degrees.19
Perhaps the most serious objection to the concept of arbitrariness arises from the issue of “reality,” whose implicit exclusion as something irrelevant to the sign led to Saussure being accused more than once of championing a modern version of medieval realism (Furton 1995:48). Ogden and Richards (1923) considered Saussure’s avoidance of the notion of the symbol (whose traditional interpretation presumes a real-world phenomenon represented in it) as a sign of his theoretical “naïveté.”
By far the most persuasive (“beautiful,” in Jakobson’s estimation) critique along this line came from Benveniste (1966a). It stemmed from Saussure’s own thesis of speakers’ blind acceptance of the convention that holds a signifier and a signified together. As Benveniste pointed out, if speakers accept their language unquestioningly as a phenomenon whose features need neither logical reasoning nor empirical justification, “le lien entre signifié et signifiant n’est pas arbitraire, il est nécessaire.” Benveniste saw in Saussure’s emphasis on arbitrariness a token of his adherence to the modernist “philosophical relativism” of the turn of the twentieth century.
Perhaps as a result of this critique, scholars in the last half-century mostly felt reticent about this aspect of Saussure’s theory. One can detect a defensive note in Joseph’s (2004:69) assertion that championing arbitrariness was Saussure’s way of taking a stand against the proliferation of artificial languages, on the one hand, and “wrong-headed” concepts of language as a carrier of collective mentality, with its nationalist and racist implications, on the other. To many, the principle of absolute arbitrariness signified an excessive relativist rigidity that contradicted the general “spirit” of Saussure’s teaching.
A notable exception to this trend was a work by the literary theorist Jonathan Culler (1986 [1976]). Contrary to the predominant tendency to marginalize or trivialize the issue, Culler argued that arbitrariness constituted the very core of Saussure’s approach to language because of its inalienable connection to the principle of the duality of the linguistic sign. He refuted the view of arbitrariness as a heuristic instrument of excessive systemic rigidity; far from that, Culler claimed, arbitrariness, understood as the direct consequence of the unbreakable link between the signifier and the signified, means the total volatility of the sign, its openness to incessant and multiple challenges that make it unable to sustain whatever state it finds itself in at any given moment.
Culler’s is a brilliant theoretical insight; what remains to be clarified is the specifically linguistic side of the problem, namely, how to maintain the principle of total freedom, signified by arbitrariness, in the face of the many instances of iconically or etymologically “motivated,” i.e., not fully free, semiotic units that can be observed in any language. One can agree with Gadet (1987:40–41) that arbitrariness “in a banal sense” does not explain the double nature of the linguistic sign. The question remains: how to interpret the relevant linguistic data beyond the “banal”?
To appreciate the full scope of Saussure’s concept, it is necessary to distinguish two different levels on which arbitrariness can be understood: on the one hand, as a founding principle of philosophy of language and, on the other, as an empirical feature that can be observed in vocabulary. The identification of the arbitrary with the unmotivated misses this distinction and, as a result, paves the way for the idea that there can be different kinds and degrees of motivation and arbitrariness. Meanwhile, the universality of arbitrariness as the constitutive principle of language is not contradicted by the empirical fact that in certain words the signifier-signified relation appears to have a rationale of one kind or another.
The difference between the postulated principle and the observable data becomes reconciled when we note that even in cases where the relation between a form and a meaning could be inferred from a derivational pattern, this inference alone is never sufficient to predict the exact character of the sign as it is known to speakers. True, the composition of a derivative sign can be inferred from its relation to the base. But there exists an infinite variety of potential paths of inference by which a sign could be constructed out of its semiotic antecedent; there is no way to predict which of those infinite possibilities will be employed by language in one particular case or another. Although we understand how a derivative sign has been constructed, we understand this only after the fact of our knowledge that it has actually happened this way and no other.
Looking at words with clear derivational patterns, one can easily be lured to the conclusion that, even if one had no knowledge of the value assigned to them by convention, one would still be able to grasp it by inference. This conclusion, however, is illusory. Both the form and the meaning of housekeeper stem from house and keeper in a way that seems perfectly logical. Imagine, however, someone trying to infer the meaning of housekeeper from its ingredients, without knowing that meaning for a fact. Does housekeeper mean a person who guards the house, or owns it, or retains it temporarily, or occupies it as a squatter—or perhaps not a person but a device that prevents the house from collapsing? The only way to know that our educated guesses are wrong is to know the meaning of housekeeper as determined by convention—that is, to treat it as arbitrary.
Even in such seemingly trivial cases as that of twenty-two, the situation is not as simple as it looks at first glance. How can one know beforehand whether the signifier for “22” should be twenty-two, or twenty-and-two, or two-and-twenty (German zwei und zwanzig); or perhaps it should be altogether different, say, two dozen minus two?20 The assortment of possible alternatives is potentially infinite since it knows no logical limit.
Language seems to invite speakers to reason with it—to build explanations based on patterns that, at first glance, seem logical and quite transparent. Such partial inferential patterns, seemingly suggesting an ultimate logical organization of signs, may be forthcoming for all eternity without ever coalescing into a coherent universal order. They are doomed ever to remain isolated pockets of order whose extent remains undetermined. In the end, any rational procedure capitulates in the face of blind convention.
Empirical knowledge about the world is of no more help in grasping the value of a sign than are logical patterns of inference. Again, it seems obvious that the words of a language reflect the experiences of its speakers; trivial examples to this effect, such as the proliferation of words describing various kinds of snow in the languages of inhabitants of the far North, abound in popular linguistic literature. But it would be futile to try to predict how many such words would emerge in one northern language or another and what semantic space would be allotted to each of them. When one encounters the Finnish word ahava, which means, approximately, “a piercingly cold damp wind in early spring,” one can easily picture the natural conditions that make this word a useful item of the Finnish vocabulary. But it would be manifestly absurd to try to infer the whole repertory of words in Finnish related to bad weather from observations of the habitat of its speakers.
To sum up, arbitrariness does not mean that the relation between form and meaning must be totally fortuitous in all empirically observable cases. What it means is that, although an external rationale, of one kind of another, for the given character of a sign does exist as an empirical feature of the vocabulary, such a rationale is never sufficient either for determining the value of signs or for predicting the direction of its further development. In the final analysis, the status and the destiny of every sign, be it “iconic” or purely “symbolic,” “relatively motivated” or “simple,” is a fact of convention.
It is in this sense that arbitrariness can be understood as a truly universal, a priori given property of signs. It is a foundational principle of language, its constitutionally given transcendental feature, rather than an empirical yardstick more or less applicable to various domains of the vocabulary. The absoluteness of arbitrariness is grounded in the fact that the signifier and the signified, as they are defined by being linked together in a sign, do not exist outside that linkage. Strictly speaking, no justification of the “relation” between the two sides of a sign is possible outside the bare fact of its existence at a given moment in a given language; they are nothing but “sides” of a bipolar phenomenon—not phenomena in their own right that could be “related” to each other.
Negative Implication: The Irrelevance of Substance
Saussure’s insight had overwhelming methodological consequences whose scope no one could appreciate better than himself. Since neither the signifier nor the signified of a sign can exist outside their arbitrary interconnectedness, it is impossible to speak of either of them in positive, substantial terms. The moment we focus on either side of a sign in an attempt to describe it, the phenomenon of the sign is lost. Saussure’s idea undermined the traditional understanding of the sign as something standing for something else, suggested by Aristotle, canonized by medieval scholasticism and universally accepted ever since.21 Under Saussure’s approach, neither that “something” (i.e., the sign’s form) nor the “something else” (the meaning) exist prior to and outside of their encountering each other in a sign. Defining the meaning of a sign by referring to its content does not render “a linguistic fact in its essence and all its scope” (CLG, 162/116). By the same token, the classification of its speech sounds, useful as that may be as an empirical tool, cannot be taken for a linguistic description of the signifier. Saussure even suggested that the excessively precise acoustic or articulatory description of sounds is counterproductive, as far as linguistics is concerned, since it puts too strong an emphasis on substance.
If the sign literally cannot be “grasped,” either from its material or its semantic side, how can one approach la langue as a system of signs? The key to the problem lies in the notion of the “system.” Unsustainable on their own, the signs of a given language exist only together, in mutual relations. It is a “grand illusion,” Saussure forewarns, to think that one can describe a sign as a single instance of union between a form and a meaning, since that would mean referring to them in substantial terms; signs can be reached only in a system in which they are positioned vis-à-vis each other (CLG, 157/112).
The total dependence of the form and meaning of signs on their intrasystemic relations is captured in one of Saussure’s key terms: the “value” (valeur), in the notes sometimes also called the “differential value” (valeur différente). The “value” of a sign is what distinguishes it from other signs. It characterizes signs as “oppositive, relative and negative entities” (CLG, 164/117). The word in itself is an empty shell whose values, on the side of both its signifier and its signified, are determined by its oppositions to or distinctions from what it is not (ELG, 74/50). Two signs “are not different, they are only distinct; there is nothing between them but an opposition” (CLG, 167/119). A sign has neither “form” nor “meaning” in a traditional sense; it does not possess a secure semiotic space allotted to it but, rather, occupies as much space as other signs in the same language allow it to have. The meaning of supplice is defined not by its relation to certain experiences, historical realities, images, and so forth, Saussure explains, but solely by its differentiation from martyre, tourment, torture, affres, agonie, and other related terms. The Course seconds the point by stating that the meaning of redouter, craindre, and avoir peur are determined by nothing but their mutual opposition; it is impossible to understand the meaning of any of them without referring to “its competitors” (CLG, 160/114). (One cannot help sensing, in those ostensibly random examples, the depth of the depression into which Saussure was plunged by his own discovery.)
According to Saussure, a typical mistake which les philosophes shared with common users of language consists in treating language as a “vehicle of thought.” The point is that language is not about whatever referential content it may evoke in the minds of its speakers. The primary characteristic of language is negative: language is a pure (i.e., defined negatively in oppositional terms) form; it consists of differences and nothing but differences (ELG, 264/189)—a crucial point that is aphoristically formulated in the Course: “Dans la langue il n’y a que des différences sans termes positifs” (CLG, 166/118).22
Taking a jab at Humboldt’s vision of language as the carrier of the collective experience of a nation, Saussure speaks with pointed sarcasm about those who think that all the richness contained in the meaning of a word “could not have happened otherwise than as the fruit of reflection, of experience, of a profound philosophy accumulated in the depth of the language by generations of those who used it.” What happens in and with a language is just the opposite: “no sign is delimited by the sum of positive ideas on which it is focused at the given moment; it is never delimited otherwise than negatively, by the simultaneous presence of other signs” (ELG, 78/52).
The lack of any substantial quality makes the integrity of a word, as it is laid down in a dictionary, “imaginary” (ELG, 83/56). The different meanings of a “polysemous” word emerge from the different oppositions in which this word finds itself in a language. For instance, the word moon has one signification in opposition to sun, another in opposition to day or year (as a time measurement related to an astronomical cycle), yet another in opposition to stars as a source of light in the night, and so on. There are no positive substantial features that would “coordinate” all those meanings; they all arise in a random way, negatively, i.e., each by being differentiated from something else (ELG, 88/60).
I believe that Saussure goes further than any twentieth-century theoretician of meaning in exploring the depth of negativity in language. In particular, Saussure’s understanding of polysemy as “an allegorical picture” whose various components are conjoined in an arbitrary fashion (ELG, 112/76) was more radical than Wittgenstein’s idea of family resemblances. The latter, while rejecting the idea of a common denominator or “invariant” of meaning, still considered the relations between particular meanings in positive terms, as a matter of their “familial” similarities and overlaps.
The immanent character of the system of semiotic differences in each language makes it impossible for signs in different languages to have identical semantic values, no matter how close they appear to be with respect to their overt referential content. Projecting the signs of one language onto the signs of another never yields a perfect fit. Each language presents a labyrinthine maze of semiotic spaces held together by no other principle than their mutual support through a mutual delimitation “in a complicated interplay resulting in an eventual balance” (ELG, 66/43).
The purely oppositive character of linguistic values renders void the question of the “origin of language” with which eighteenth-century philosophy of language was much preoccupied. Saussure’s vision of language as a network of pure differences makes the question of its origin logically unsustainable. Language cannot have originated from the absolute unity of an imagined “first sign.” It would be vain to attempt to construe some absolutely unified point of departure from which language ostensibly started to evolve—the kind of speculation of which the age of Enlightenment was so fond. Language begins when something is distinguished from something else—which means that language can never “begin” as a single step from nothing. No fact of language is elementary (ELG, 20/5); from the very first moment in the existence of language there has already been a differentiation.
By the same token, language evolution has nothing to do with “development” in a conventional sense. The signs of a language are “always adequate” to what they express (ELG, 102/67) because they express nothing but mutually defined values. Saussure once suggested that if there were a language that had just two distinct signs, it could still accommodate everything in the world; in such a language, a certain plurality of objects would be identified with sign A, while another plurality would receive their identity from sign B.
The radicalism of Saussure’s negativity was always a sticking point even for those who were otherwise positively disposed toward his theory. As has been pointed out more than once, the world as perceived by human consciousness is not an empty space waiting to be filled by language in some fashion or other. Thought prior to the sign is not totally amorphous: it already presents experience organized into “proto-concepts” (Mazzone 2004). Such a critique misses Saussure’s point, which is not to deny that perception has any experiential basis or that consciousness is able to organize perceptions into broader categories. Neither does the Saussurean approach deny any link between those perceptions, on the one hand, and patterns of their semiotic representation in language, on the other. What the principle of negativity means is that this link is never consistent. Any threads of connection that can be discerned between our experience of the world and its reflection in our language are doomed ever to remain no more than spurious and coincidental. Whatever people’s “proto-concepts” about the world may be, the structure of their language cannot be derived from them. In their perception of the world, people rely on prototypes grounded in experiential features of phenomena. But once this reflection of the world by consciousness enters the realm of signs, it undergoes incessant arbitrary reconfigurations, making it impossible to rely on the experiential basis when dealing with language. Whatever we think in connection with or alongside language, in the last count the language remains subject to no rules but the fact of its existence.
Whether Saussure’s commentators adhered to his concept of the arbitrariness of signs or tried to contest it, the idea itself, based primarily on the way it was presented in the Course, was almost universally understood as an emphatic assertion of the necessity, or even compulsion, for speakers to use language as it is. This understanding was largely responsible for the strong determinist bent in structural linguistics and, by the same token, for the subsequent critique of its (and, along with it, of the Course’s) deterministic “authoritarianism.”
Indeed, the Course paints a striking picture of a speaker deprived of any tools of reasoning or worldly experience when facing the conventions of his language, to whose inexorable arbitrariness he has no choice but to succumb. The speaker faces the imperative of the inner structure of the language, which has no other justification but that it is the way it is: “A signifier chosen by la langue cannot be replaced by another.… No society knows or has ever known its language otherwise than as a product inherited from preceding generations and taken as it is” (CLG, 104–105/71–72).
Saussure’s notes, however, offer a picture that at first sight seems to be diametrically opposed to the one presented in the Course. Here, the principle of freedom, which the Course seemed to deny completely, emerges with the utmost emphasis. Not only does language change, but it cannot exist otherwise than in a state of spontaneous change; it evolves inexorably, like a stream running down a mountain (ELG, 94/73). La langue as a semiotic system is not a ship at anchor, Saussure remarks, but a ship at sea: “The moment it touches the sea, it is vain to think that one could predict its course by projecting it from one’s knowledge of the material out of which it has been made, and of its interior construction” (ELG, 289/202).
It is fascinating to see how Saussure applies Durkheim’s concept of “social facts” to opposing ends in the Course and in the notes. In the former, language is presented as a “social act”23 because it compels the speaker to conform; any individual effort, either by a single speaker or by a collective, succumbs to the imperative of social convention: “No individual, even if he willed it, could modify in any way whatsoever the choice that has been made; and what is more, the community itself cannot control so much as a single word; it is bound to the existing language” (CLG, 104/71). But in the notes, the seemingly helpless, conformist “community” becomes an uncontrollable “mass” of speakers. Here Saussure sounds more like Bakhtin: “From the instant at which a symbol becomes a symbol, which is to say, from the instant at which it becomes immersed in the social mass that establishes its value at any given moment, its identity can never be fixed.” The “social collectivity” is heterogeneous—language evolves “between many” speakers (ELG, 290/203); all the changes are produced in a chaotic and improvised fashion, in the process of spoken exchange (ELG, 95/64–65).
How can one assess this glaring contradiction? The habitual explanatory path taken by “Saussurology” is to assign the superior value to the notes while blaming the book’s editors for distorting or misunderstanding Saussure’s thought in their “vulgate.” And yet, Saussure himself was aware of the contradiction. The Course speaks about the two main principles according to which signs exist, calling them “immutability” and “mutability.” Typically for the book (which in this respect probably reflects the rhetoric of Saussure’s lectures), there is a vague protestation to the effect that these two principles are not as contradictory as they look and can be reconciled (CLG, 108–109/74), which, however, is left hanging in the air without any further elaboration. The fact of the matter is that the immutability and mutability of signs represent two complementary facets of the most general a priori feature of language, which is its arbitrary constitution. Both principles are logical consequences of the fact that, insofar as they are considered within the inner arrangement of language, signs turn out to be devoid of any substantial support.
The “immutability” of signs stems from their inner emptiness, which makes them unsustainable otherwise than as a system in mutual relations. Speakers cannot willfully target any single sign for adjustment, since there is no such thing as a solitary sign to be targeted; each time a speaker deals with a particular entity he faces a whole network of oppositions. The “social contract” between speakers concerning their language has a wholesale character. One cannot perform “surgery” on a language by purposefully assigning a new meaning to a particular word without causing incalculable side effects to ripple through the system as a whole. The same principle applies to the sign’s material representation: any tampering with one signifier would affect other signifiers by tipping the balance of their distinction.
And yet, in an astonishing paradox, it is precisely the lack of substance that makes the sign, immutable when looked at from one point of view, also mutable, and moreover, unsustainable. The lack of any solid ground behind the sign’s value makes it vulnerable to arbitrary change. One cannot defend a sign’s integrity on the ground that it corresponds well to a certain logical pattern or certain quanta of worldly experience. Since the convention that supports the sign is “pure”—i.e., empty, groundless—it turns out to be unenforceable; any mutation goes, insofar as it is spontaneously accepted and adopted by speakers in the process of language use.24
Culler was the first to note that the idea of oppositive values, predicated on the principle of arbitrariness, makes the system of such values open to uncontrollable and unpredictable changes.25 Having plunged into an open “sea” of usages, with its infinite variety of voices and intentional vectors, the system of purely oppositional values has nothing to shield it against the pressures that come at it from all sides in a chaotic and volatile fashion. Nothing prevents it from being pushed in any direction at any moment.
For all their apparent polarity, the thesis foregrounded in the book, that nobody and nothing can institute a change in la langue, and that in the notes, that nobody and nothing can control the inexorable process of its change, stem from the same fundamental vision of the elemental nature of language.
The notes that depict language as moving precipitously, like a stream rushing downhill, totally at the mercy of the “mass” of heterogeneous social forces, in fact elucidate the meaning of the emphatic assertions in the Course of the powerlessness of both the individual and the community to change anything in their language. Speakers are indeed powerless, but only in the capacity of purposeful, teleologically minded agents. Language does not yield to any deliberate attempt to amend it in a certain “desirable” direction, because no particular entity in it can be exempted from the whole for a targeted operation. A purposeful “language policy” may achieve certain local results, but, in the long run, consciously instituted changes always bring with them unintended consequences, leaving language as unyielding to deliberate control as ever. A “fundamental error” of les philosophes of the eighteenth century was their failure to understand that, as a social phenomenon—precisely because it is a social phenomenon—language cannot be either constructed or revised in a rational manner (ELG, 94/64).
However, when speakers act unconsciously, just by using language on the spur of the moment, they provoke imperceptible and uncontrollable changes—in fact, cannot help inflicting them. Each time a speaker applies a certain expression to the unique circumstances of a given moment of speech, he opens the way to inflecting its meaning. To say “adopting an idea,” by analogy with “adopting a child,” or fleur de la noblesse by analogy with fleur du pommier, is to affect not just a single utterance but the whole state of linguistic values (CLG, 151/107). Likewise, each time a speaker produces a given expression with a particular unique combination of pitch curve, tempo, dynamic, and vocal timbre—responding, with the help of these tools, to the demands of the moment—he affects the balance of established signifiers.
The principles of the immutability and mutability of signs coexist, and moreover, are complementary, two different perspectives from which language can be viewed. To the uninvolved observer who posits himself outside language, it looks like a stream of uninterrupted and uncontrollable transmutations. No single instance of language use is ever an exact repetition of a previous one, with regard to both its signified and its signifier. Defenseless against the incessant and spurious forces that press on it, because of its inner emptiness, language proves incapable of maintaining itself in this volatile environment. Far from the solid geometry of a crystal lattice, its network of oppositions behaves like a “gaseous” entity, no more capable of resisting outside forces than a cloud at the mercy of the wind.
When, on the other hand, the observer acts in the capacity of a speaker, his consciousness of his language moves together with the movement of the language itself. From this inside perspective of the speaker—or, if we use Saussure’s expression, from the point of view of la langue en lui-même—language appears immutable. No matter how many changes in a language could be noted by an outside observer, its speakers always face it as a “state”; whatever spontaneous change arises at a given moment causes the whole system to adapt to it, thus remaining in perfect harmony with itself. As Saussure pointed out on several occasions in his notes, the differences between “old,” “middle,” and “modern” French, or the idea of French as an offspring of Latin, are no more than abstractions constructed by linguists (in their capacity as outside observers). As far as speakers were concerned, they could never have known—without an outside prompt—that their language had ceased to be “old French” and become “middle” or that the language they spoke was not “Latin” anymore but “French.” Had there been a speaker whose life span comprised two thousand years, he might have thought of himself as still speaking the language of Cicero—with perhaps a few “generational” changes in it—while in fact speaking modern French or Italian.
Saussure’s depiction of a speaker whose every choice is being made for him by his language addresses the speaker’s linguistic consciousness, not his linguistic behavior. The latter, on the contrary, involves a constant meddling with available linguistic resources, to which language always yields, most often imperceptibly.
To part with any external support, from empirical “common sense” to universal categories of reason, while dealing with language in its purest essence, did not prove easy. As Saussure notes, we tend to assign “precipitously” some positive existence to what is in fact purely negative and purely differential. And yet, Saussure adds, “perhaps, I admit it, we are called to recognize that without this fiction, the spirit would find itself literally incapable of commanding this sum of differences, in which no single shred of a positive and firm signpost appears at any moment” (ELG, 65/42).
The lack of any substantial anchor makes language fleeting and volatile. It never stands still—the moment you have grasped it in a certain state, another state emerges that refutes the previous one. For instance, take any notion of a grammatical category, “for example genitive,” Saussure remarks (apparently recollecting his experience with classifying cases of the “genitive absolute” in his dissertation). The notion of the genitive is “completely ungraspable” (insaisissable), “a word literally devoid of any sense,” since the meaning of genitive is always “extending, from one moment to another, one page to another, one line to another.” To speak about the genitive as a category implies that there is an idea “superior to signs, exterior with regard to signs, independent of signs” (ELG, 55/34). Language has no firm “signposts,” as the conventions of grammatical description suggest; in reality, there is nothing but a continuum in which the relational network remains ever in flux.
The fundamental error of both the structuralist and the generativist approaches was that they gave ontological value to such “precipitous” acts. Structural patterns and algorithmic rules imposed on language become a second reality, acquiring a life of their own. By proceeding in this way, twentieth-century theoretical linguistics continued to revel in what Saussure perceived as the codification of “phantoms.” It was that complacent “illusion” Saussure rebelled against and desperately sought to escape—even though each escape attempt left him empty-handed, peering into a void.
A close reading of the Course reveals the same anxieties lurking behind its aphoristic facade, occasionally erupting onto the surface. In “ces gants sont bon marché,” is bon marché an adjective?—Saussure demands of his audience. Apparently it is not, since it does not correspond to many features of a “normal adjective”; on the other hand, simply declaring marché a “noun” would mean ignoring constructions like this one (CLG, 152–153/108). If one reads the Course as a theoretical “grand narrative,” it is easy to overlook this passage as an incidental aside. Yet the discussion of this example leads Saussure to a conclusion that reverberates with the central concern of his notes: “Thus, linguistics labors without respite over concepts forged by grammarians.… And if they are phantoms, with what realities could they be matched? In order to escape from illusions, it is necessary to become convinced from the start that concrete entities of la langue are not presented to our observation by themselves; that one strives to grasp them, and by doing this, one comes into contact with reality” (CLG, 152–153/108).
When Saussure was writing the initial corpus of his notes in the mid-1890s, he was still hopeful that eventually he could succeed, through an ardent labor of inferential thinking, in getting past all the “phantoms” and touching the essential core of language, thus establishing a solid objective foundation for linguistic studies; hence his notorious hesitation to give any finalized shape to his thought, a step whose arbitrary relativity—that is, the fact that it could always be reshaped in innumerable ways—he immediately realized once he tried to take it. But in the second half of the 1900s Saussure came to realize the epistemological impossibility of this goal. Language is not an abstraction; nothing in language exists outside and beyond representation. Yet its palpable reality remains perpetually elusive as a result of the absence of any general ground on which one could establish reliable connections between the representational surfaces and the values they represent—no foundation, that is, except the negative one of their arbitrariness. The principle of negativity, based on the arbitrary duality of linguistic signs, becomes the “irreducible” fact of language (ELG, 39/22). La langue is not merely an inner knowledge (“competence”) stored in the minds of its speakers, as Saussurean linguistics often claims. It represents no empirical reality at all, even an ideational one. Rather, la langue as a pure (negative) form is a notion of the metaphysical state of language.26 Consequently, la parole cannot be understood as the “manifestation” of la langue in the outward reality of speech; it is separated from la langue by a metaphysical abyss.
Yet discovering arbitrariness, despite its devastating methodological consequences, was not a purely negative outcome. By wresting language from any empirical or logical order, Saussure’s principle of arbitrariness revealed itself as the principle of freedom—a linguistic manifestation of the metaphysical principle of free will that people exercise in their capacity as speakers. Challenging arbitrariness through examples of more or less extended local patterns of reason in the composition of its signs is as futile as challenging the principle of free will through examples of human behavior that are more or less obviously caused by logical reasoning or empirical necessity. In both cases the essence of the matter is not the possibility of a local causality but the ultimate impossibility of deriving human behavior at large from any comprehensive causality pattern.
The fundamental freedom of language, grounded in arbitrariness, shows itself in the unlimited diversity of forms different languages, or even one language in the course of its development, may assume. No logical or empirical restrictions exist that could determine the playground of values that is language or set limits to its transformations. For Saussure, the structural diversity of languages is more than an empirically known fact—it is, rather, “primordial reality,” reflecting the very essence of language. The question “whence to begin?” is a question without a definitive answer when one approaches “the slippery substance” of language (ELG, 281/197). The truth about language lies not in ultimate generalizations but, on the contrary, in the unceasing comparative analysis of its diverse manifestations.27
This makes language a fundamentally “human” phenomenon in the most general sense that its metaphysical essence reflects the nature of those who speak it. The principle of the linguistic “free will” of speakers leaves language on its own amidst all the phenomena of the world subject to physical and/or logical causality. The Kantian subject faces a world that is shaped by a priori categories he cannot transcend. Yet the subject is endowed with the gift of “genius,” which makes him never tired of challenging the boundaries of a world circumscribed by pure reason. It is this gift (or curse, as some Romantics would say) of free creativity that sets him apart from the rest of the world. In a striking parallel, the Saussurean langue emerges as endowed with a quality that sets it apart from all other phenomena of the world. Like human consciousness itself, it is “conventional, nay arbitrary, totally devoid of a natural rapport with objects, absolutely free and lawless in rapport with itself” (ELG, 202/140).