GIORGIO AGAMBEN - Potentialities: Collected Essays in Philosophy

PART ONE
Language

§ 4 Philosophy and Linguistics
I
To undertake a philosophical review of a work of linguistics poses a problem of legitimization. The history of the relations between philosophy and the science of language (taking this term in the large sense, such that it includes the technē grammatikē of the ancients and the grammatica of the medievals) is so rich in exchanges, crossings, and accidents that any attempt to distinguish the two with precision appears both necessary and impossible. Not only does the ancient tradition attribute to Plato and Aristotle the origin of grammar, but further, from the beginning, logical categories and grammatical categories have been so tightly interlaced that they appear inseparable. The Stoics, whose linguistic theory had such decisive importance for the history of the study of language, thus considered phōnē (in the grammatical sense of phōnē enarthros, "articulated voice") as the arkhē and foundation of dialectics. And in Aristotle Categories it was already impossible to understand what was indicated by the concept of legomena kata mēdemian symplokēn without taking into account the necessarily grammatical part of speech (meros tou logou) that it implies. In the same treatise, moreover, the determination of pure Being (protē ousia) is inseparable from the meaning of the deictic pronoun and the proper name, in accordance with a parallelism that characterizes the entire history of ontology (it suffices to think of the importance of the pronoun and the proper name, and more generally of grammatical categories, in the treatment of the problem of supreme Being in medieval the ology, or of the impossibility of distinguishing between logic and grammar in a Scholastic treatise de modis significandi).
The project proposed by Heidegger in a crucial passage of Being and Time--"to liberate grammar from logic"--cannot, therefore, be easily accomplished. Language would have to be simultaneously liberated from grammar (a program formulated, more or less consciously and according to different modalities, throughout the history of Western thought). And this would presuppose a critique of the interpretation of language implicit in the most elementary grammatical categories: the concepts of articulation (arthron), letter (gramma), and part of speech. Such is the significance of these categories, which the Greeks already clearly defined in their reflection on language and which, strictly speaking, are neither logical nor grammatical but rather what renders possible every logic, every grammar, and perhaps even every epistēmē in general.
II
Forms of thought find their first exteriorization in man's language, where they are so to speak deposited. . . . One finds the intervention of language in everything that becomes his interiority, in his representation in general, in everything that he makes his own. Everything with which he forms his language and by which he expresses himself in language contains a more or less concealed, mixed or explicit category. Thus he naturally thinks according to his logic; or, rather, his logic constitutes his very nature. But if one wanted to oppose nature in general to the spiritual, as something belonging to the physical world, one would have to say that logic constitutes the supernatural, penetrating into all of man's attitudes toward nature, his feelings, intuitions, desires, needs, impulses; and one would have to say that man is what humanizes them.
This passage from the preface to the second edition of Hegel Science of Logic clearly expresses one of the enduring subjects of the philosophical tradition: the intertwining of thought and language and the task it implies for thinking. In our time, this task was decisively reformulated in a different way by Alexandre Kojève when he defined philosophy as the discourse "that can speak of everything, on the condition that it also speak of the fact that it does so." If this definition is correct, the so-called "linguistic turn" by which contemporary philosophy and its interest in lan guage (in the large sense) have been defined risks stating merely a trivial truth. The fact is that the term "language," to take up Aristotle's phrase, "is said in many ways," and only an elucidation of what philosophy and linguistics respectively understand by this term can lead to a useful consideration of their relationship. That there is an interlacement between philosophy and the study of language does not necessarily mean that philosophy and linguistics have the same object. Heidegger's observations that "the Being of the being that linguistics takes for its object remains hidden" and that philosophical reflection, for its part, should give up "the philosophy of language" to ask itself above all "what mode of Being should be attributed to language" (in other words, if language has the mode of Being of a worldly object or not)--these observations have lost none of their currency today. As something "said in many ways" (pollakhōs legomenon), the very concept of language is caught in a vague homonymy and often remains imprecise, both in the field of linguistics and in that of philosophical research.
III
Milner's book presents itself as an "introduction to the science of language." It is the work of a linguist who is also a thinker of great originality. While his two recent books ( L'amour de la langue and Les noms indistincts) are among the most important contemporary French contributions to the study of language, references to them are rare. This is perhaps because Milner's enterprise, as he describes it in his Introduction, aims at being "resolutely scientific," in the sense that it undertakes to examine and maintain "the hypothesis according to which linguistics is a science, just as a natural science may be a science." 1
It is not by accident that this introduction to "a" science of language appears at a time when the glorious season of linguistics seems a thing of the past. With the exhaustion of the project of comparative grammar and the decline of the no less brilliant, if perhaps less significant, project of generative grammar, linguistics today is no longer the "foremost" human science, as it was clearly thought to be only two decades ago. The prestige of the human sciences in general is now in a period of decline. The project of a "general science of the human," which reached its apex at the end of the 1960s, dissolved with the political project of the same years. The severe prose of the world of the 1980s tolerates only positive sciences and, alongside them, a philosophy that is more and more oblivious of its destination.
One could think that a book such as this Introduction, which wishes to be wholly consecrated to the foundation of a positive science of language, could not help clarify the relationship between philosophy and language. But precisely the contrary is the case, for in more than one point Milner's Introduction contributes decisively to the clarification of the concept of language and its homonyms. This review cannot, of course, take account of the book in its entirety (a task to which only a linguist would be adequate); it will, instead, concentrate on some of the points to which we have already alluded. In discussing them, I propose to show how this book, while maintaining itself inside the science of language, allows for a precise determination of the relationship between philosophy and linguistics as well as their respective tasks.
IV
A first point is to be found in Part I of the book, which is devoted to the epistemological status of language and concerns the identification of the very object of the science of language. While Milner does not mean "to propose a theory of knowledge" (p. 23 ), it would be difficult to find a work of epistemology that contained such a clear and original presentation of the concept of Galilean science. According to Milner, the mathematization characteristic of Galilean science has as its basis not (as is usually thought) quantification but "literalization," by which Milner means that "one uses symbols that can and must be taken completely literally, without regard to what they may designate," and that "one uses these symbols solely in accordance with their own rules." "The possibility of full communication . . . rests on the fact that, once the rules for the use of the letters are learned, everyone will use them in the same way" (p. 14 ). Literalization therefore implies "the irreducible difference between restriction and the substance of restricted beings." "What is then taken from mathematics is the dimension of restriction, which applies to beings whose objective reference (substance) can certainly be determined, but does not have to be when one uses restraint itself. It then follows that one can use beings without 'seeing' what they designate, and one then correctly speaks of blind use" (pp. 91 - 92 ).
Immediately afterward, Milner lists a series of "primitive facts" that function as irreducible limits, which linguistics must confront and beyond which it cannot venture. In the first place there is the factum loquendi, whose sole content is the existence of language, the fact that there are speaking beings:
The usual name for this brute fact is language. One may note that it presupposes only one thing: that there are speaking beings. In this sense, to speak of language is simply to speak of the fact that speaking beings exist. Nevertheless, to speak of this fact in an interesting manner, it will be necessary to call the existence of speaking beings into question. But this is precisely what linguistics cannot do; for linguistics, this existence can be neither deduced nor explained in general. It is thus possible to understand the sense in which linguistics does not have language as its object: language is its axiom.
This does not at all mean that one cannot consider this existence in itself, questioning its conditions of possibility. It is only that one then finds a question of the following kind: "Why is there language rather than no language at all?" And this is a properly metaphysical question. (p. 41 )
The second "primitive fact," which must be clearly distinguished from the first, is the factum linguae:
It suffices to establish that beings speak to conclude that language exists. The question as to the properties of what they say is not pertinent at this level. Linguistics cannot remain here; it must therefore admit more than the single, massive existence of language. Linguistics admits that speaking beings speak languages.
To say that the effectuations of language are languages is to suppose at least that the set of linguistic productions merits being designated by a common name. It is, moreover, to suppose that they are distributed, like the different realms of nature, in classes and subclasses, each class generally corresponding to what one calls a species in nature. It is, finally, to suppose that one can say what a particular language is. Briefly, it is supposed (1) that one can distinguish a language from nonlanguage and (2) that one can distinguish one language from another language. It is therefore necessary to reason in terms of properties; one must, in other words, distinguish the properties of a language from the properties of nonlanguage and the properties of one language from those of another language. (p. 43 )
This implies not only that languages are diverse while belonging to a homogeneous class (what Milner calls the factum linguarum), but also and above all that languages are describable in terms of properties. Mil ner calls this fact the factum grammaticae, and for him it is the constitutive and characteristic fact of linguistics.
The clarity of this definition makes it the only one to untangle the ambiguity inherent in the term "language" and to distinguish with precision the object of philosophy from the object of linguistics. If the object of linguistics is language (understood as shorthand for the factum linguae, the factum linguarum, and the factum grammaticae), philosophy is instead concerned with the factum loquendi, which linguistics must simply presuppose. Philosophy is the attempt to expose this presupposition, to become conscious of the meaning of the fact that human beings speak. It is possible to see how it is the factum grammaticae that marks the difference between philosophy and linguistics: philosophy is concerned with the pure existence of language, independent of its real properties (transcendental properties, which belong to philosophical reflection, do not go beyond the field of pure existence), while linguistics is concerned with language insofar as it is describable in terms of real properties, insofar as it has (or, rather, is) a grammar.
Hence the exclusion from philosophy of speculations on the origin of language, which traditionally belong to the patrimony of the philosophy of language. As Milner observes, hypotheses on the origin of language are nothing other than "the fictional form of the limit between 'language does not exist' and 'language exists,' insofar as this limit is presented as a passage. What is supposed to appear in this fictional passage are essential and defining properties: those properties without which one cannot say that there is language" (p. 42 ). Philosophy's attempts to identify the real properties defining the essence of language are doomed to failure precisely because they illegitimately step beyond their own boundaries into the territory of science. For philosophy, there is not and there cannot be an essence of language (or, consequently, a philosophical grammar), since the task of philosophy is exhausted in the presentation of the existence of language. Here one encounters the boundary separating the field of epistēmē from that of first philosophy. In its relation to language, philosophy can only remain faithful to its originary vocation as the science of pure existence. If science in the strict sense is the discipline that knows the properties of beings (or of beings insofar as they possesses real, describable properties), philosophy (as first philosophy) is the science that contemplates beings insofar as they exist (on hē on, on haplōs), that is, independent of their real properties.
V
But the relationship between philosophy and language (and hence between philosophy and linguistics) is in fact more complex. In the face of an epistēmē, philosophy can only assert its proper vocation as a science of pure existence through a particular experience of language. The pure existence (without any properties other than transcendental ones) that constitutes the sole object of philosophy is something to which philosophy has no access other than through reflection on the factum loquendi and the construction of an experience in which this factum is thematically at issue. Only the experience of the pure existence of language allows thought to consider the pure existence of the world.
Hence--from Plato to Wittgenstein--the striking relation of philosophy to language, which is one of both defiance and disavowal, "philology" and "misology." Hence also the proximity of and distance between philosophy and the science of language. Both refer back to the same place, whose existence one discipline must contemplate and the other presuppose for the establishment of grammatical categories. Both lack particular instruments and firm ground for the realization of their goals; both must experience language without having at their disposal (as do the other sciences with respect to their objects) any external observation post. One could thus say of philosophy what Milner says of linguistics--that it is "an experimental science without an observation post" (p. 128 ), a science that has the example as its proper mode of experimentation. The questions that philosophy poses (like the fictions it sometimes employs) do not demand any information as their answer (nor do they have any narrative value). They hold, instead, as examples, in the sense in which Milner defines examples for linguistics. 2 Despite the refinement of its logical technique, philosophy, like linguistics, must ultimately keep to natural language. If linguistics, according to Milner's phrase, is a scientia infima--which "gives itself the most minimal object conceivable" and of which it is true that "whatever a theory's degree of mathematical formalization, the final instance will always be a proposition stated in natural language" (p. 130 )--it is from a still more minimal place, namely from the pure existence of language, that philosophy must depart.
Do the two sciences, at once so close and so far apart with respect to their object, touch at any point? Is there a place in linguistics in which the existence of language can be said to emerge as such?
VI
A place of this kind can be found in the third chapter of the second part of Milner's book, which is called "Restricted Theory of Terms." These forty pages constitute an exemplary analysis of one of the most complex parts of linguistic theory (one of its fundamental claims about this field is, as Milner states, that if "positions" concern syntax, "linguistic entities" can be said to be "of two kinds: terms and positions," p. 409).
From its beginnings, the Greek reflection on language assigned a fundamental place to the distinction between onoma (name or term) and logos (speech or proposition). According to a tradition that originated with the Stoa, the event of nomination (appellatio, nominum impositio) is conceptually and genetically distinct from actual discourse. In Antisthenes, this grammatical distinction is linked to the problem of the unsayability of pure existence, in the sense that primal and simple elements can have no defining discourse but only names. A proposition cannot say what the name has named (as Wittgenstein would write in proposition 3.221 of his Tractatus: "I can only name objects. . . . I can only speak of them. I cannot assert them").
In the Categories, Aristotle distinguishes the deictic pronoun and the proper name, which signify a pure existence (protē ousia), from other names, which always designate qualities. And Plato, who uses the anaphora auto to designate the Idea, does not allow language any possibility of directly designating pure existence without properties (hence the asthenia of the logoi in the philosophical excursus of the Seventh Letter).
Another philosophical problem is tied to the domain of names (and hence to the theory of terms), namely, the problem of self-reference (of the name of the name). This problem has given birth to a series of paradoxes, the most famous being what one could call "the White Knight's paradox," referring to an episode in Through the Looking-Glass. Can the name of an object be itself named without thereby losing its character as a name and becoming a named object? Is it, in other words, possible for a name to refer to itself in its existence as a name (nomen nominans and not nomen nominatum)? In proposition 4.126 of his Tractatus, Wittgenstein implicitly gives a negative answer to the question. Carnap, by contrast, maintained that a name can perfectly well be named, by means of the use of quotation marks; but Reach refuted him in a famous article. 3
Once again, Milner's precise awareness of the problems at issue allows him to order complex material. He does so in a mere ten theorems, with a clarity unparalleled in the history of linguistics. To begin with, he abandons the "contextual principle" (usually attributed to Frege) according to which it is not possible to determine the properties of a linguistic term without reference to its discursive context. The first theorem of the "Restricted Theory of Terms" thus reads as follows: "It is possible to establish the properties of a term without reference to its use" (for example, in recognizing its lexical sense, which constitutes the fundamental principle of dictionaries). But what is a linguistic term considered in itself? What is the onoma of Greek linguistic theory?
According to Milner, a term is nothing other than the set of its distinctive properties, which Milner defines by the three traits: (1) belonging to a category; (2) phonological form; (3) lexical meaning (or virtual reference). None of these three properties (not even phonological form, which we are used to identifying with the term itself, as when we say, for example, "cat is a one-syllable word") in itself constitutes a linguistic term. And if, in this sense, linguistic individuals are not substantial realities but only "packets of properties" (p. 330), it will not be possible to name a term other than by an indirect procedure:
The procedure is well known: it is the operation of quotation by which one says table to designate the linguistic individual, table. . . . Let us be more precise: what designates the linguistic individual table is in fact the phonological concatenation t⋀a⋀b⋀l⋀e. It goes without saying that in using the phonological concatenation t⋀a⋀b⋀l⋀e, we mean the lexeme table with all its lexical properties: its meaning, its categorial belonging, and, of course, its phonological form. In other words, one uses one of its identifying properties to take down in shorthand the set of identifying properties that constitute the individual. (pp. 330-31)
The problem is that of linguistic entities and their names. Here Milner takes his point of departure from Saul Kripke's thesis on the proper name, according to which the proper name is not shorthand for a series of identifying properties:
Let us recall his demonstration: the mere fact that on the basis of the proper name Aristotle and a predicate P, one can construct a proposition such as "Aristotle is P" and its counterfactual "Aristotle is not-P," proves that the proper name Aristotle is not shorthand for a packet of identifying predicates. It is thus crucial that if the proposition "Aristotle liked cats" is held to be fac tual, " Aristotle did not like cats" be held to be counterfactual. Let us consider the terms of a language: a proposition such as "table does not have the phonological form of table" is clearly a contradiction in adjecto and not a counterfactual. The same holds, despite appearances, for propositions such as "table is not a noun" or even "in French, table is not feminine." (p. 331)
Kripke's thesis therefore cannot apply to linguistic terms, and Milner can then state a new theorem: "The linguistic term has no proper name" (p. 332). With this theorem, whose importance cannot be overestimated, Milner introduces into linguistics the principle of the impossibility of metalanguage, which is a fact without precedent in the history of linguistics. It is precisely by means of the anonymity and insubstantiality of linguistic Being that philosophy was able to conceive of something like pure existence, that is, a singularity without real properties. If the linguistic term were not anonymous, if we always already had names for the name, we would always already encounter things with their real propertics; there would never be a point at which our power of naming (or of the attribution of properties) would come to a halt. This stopping point cannot be constituted by a nonlinguistic being, since language can name everything, its naming power knowing no limits (the nonlinguistic, in this sense, is nothing other than a presupposition of language). But language cannot name itself as naming; the only thing for which names are truly lacking is the name. It is this anonymity of the name that in Plato allows for the appearance of the Idea (which is designated not by another name but simply by means of the syntagma name-auto, the Idea of a thing thus having the form of "the thing itself," to pragma auto). It is only because the term rose is anonymous, because rose is not the name of the name rose, that in uttering "a rose" I can make l'absente de tous bouquets, that is, the rose itself, appear. And it is only the anonymity of linguistic Being that gives meaning to the metaphysical thesis according to which existence is not a real property, or, in other words, the position of the transcendental. If one considers the matter, the fact that "being" (ens) is not a real predicate, that it--like the other transcendental predicates (unum, verum, bonum, etc.)--belongs to all predication without thereby adding any real property to it, can only mean that predicated Being is not itself namable, as is implicit in Milner's theorem. Being said is, in this sense, the archi-transcendental that allows for the possibility of all predication; but precisely for this reason it cannot apply to the name. Milner's theorem is in reality also a theorem concerning the transcendental; neither the name of the name nor the named name are names, and what maintains itself in relation to this anonymity of the name is pure existence. (Here one recognizes Heidegger's central thesis on language: Being can emerge only where the word is lacking, but the word is lacking only at the point at which one wants to say it.)
VII
Another point at which the existence of language as such seems to emerge within linguistics is the problem of the predisposition to language and its innateness (a thesis maintained in particular by the school of Cambridge). Milner very clearly illustrates the difficulties and contradictions to which this thesis inevitably gives rise. 4 The claim that "language is innate" cannot concern individual languages, which are wholly acquired by individuals according to the linguistic environment in which they find themselves; it can only concern language in general. But what does it mean to speak of a predisposition to "language in general"?
Let us recall that no one can suppose that a speaking being speaks French innately. Those who reason in terms of innateness suppose only the following: a speaking being speaks innately, and "to speak" is "to be capable of speaking a language in general." And this is language. Of course, it has been maintained that this "disposition to language" is not empty (and that, in other words, language has properties). But the content of this disposition is a disposition to any language or any type of language. If the disposition to language is not empty, then it is necessary that there exist properties common to many languages, if not all. Consequently, the supposition of a disposition to language necessarily meets up with the question of universal grammar. (p. 227 )
Yet the expressions "language in general" and "universal grammar" risk being meaningless:
Language can only ever be observed in a particular language. In anthropology, it always appears possible to separate clearly and sufficiently the innate part of behavior from its acquired part. In linguistics, this point of departure is never simple; more precisely, it concerns theory and not observation. Let us suppose that it can be shown that in all languages, certain properties can be found, which in each of them are always combined with particular properties. Theoretical reflection must certainly give a distinct representation of the universal and the particular, but observation only ever encounters a state in which the two are combined. (p. 232 )
This disposition to language in general (or to any language) is, in truth, something like the famous tabula rasa of the potential intellect of Aristotelian philosophy, which is itself not an actual intelligible but is nevertheless capable of being any intelligible whatsoever. What is to be found in all these general notions, while remaining unthought, is nothing other than the factum loquendi, the pure existence of language grasped as a universal linguistic essence. The innateness of language in general in the form of a universal grammar is, in short, only a shadow of this factum loquendi with which the science of language cannot reckon. That there is language, that human beings speak, is not a real property that could be determined as a universal grammar in which all languages would participate. Here we can observe the mechanism by which Aristotelian protē ousia, which is pure singular existence, becomes the sub-stantia underlying all categories. Thought that seeks to grasp the factum loquendi, language as pure existence without properties, is always about to become a kind of grammar.
It does not come as a surprise, then, that the different projects that, throughout the history of Western culture, sought to construct a pure experience of the existence of language (that is, of language without real properties) often ended by being substantialized in the form of a (more or less universal) grammar. At the beginning of Romance culture--at the basis of the project of Provençal love poetry and in Dante--there lay the attempt (which is philosophical and not simply poetic) to grasp the pure existence of language by means of the figure of a woman who was held to be the supreme love object and through whom the mother tongue was explicitly opposed to grammar. However one understands the properties that Dante assigned to his "vulgar" language (illustriousness, nobility, etc.), they are certainly not grammatical properties; they seem, instead, to constitute an equivalent to the transcendentia of medieval logic, being just as empty of real content as they are. But it is also thus that both Provençal lyric poetry and Dante, in historical circumstances that cannot be examined here, ultimately led to the construction of a grammar. Provençal poetry ended with the Lays d'amors, that is, with a monumental grammar of the Provençal language, in which the laws of language were assimilated to the rules of love; and Dante's project of an "illustrious vernacular" ended, albeit at the price of betrayal and contradiction, in the attempt to construct the grammar of a national language.
When, on the other hand, these projects appeared in Western culture in an authentically philosophical form (examples in our century can be found in both Benjamin's "pure language" and the late Heidegger's die Sage), what is at issue each time is not the phantasm of a universal language (or grammar) but an experience whose object is the factum loquendi, the pure existence of language.
One could make analogous observations (though the register would be different) regarding the science of language. The different attempts to construct a universal language or grammar (from the lingua matrix of seventeenth-century philology, to the universal language and the characteristica that interested Leibniz, to certain aspects of the reconstruction of Indo-European) register the need to take account in some way of the factum loquendi but end up simply showing language's excess with respect to science.
But how, then, is it possible to bear witness legitimately, through knowledge, to the pure existence of language?
VIII
The preceding observations should give an idea of the complexity of the relationship of philosophy and linguistics insofar as it is implicitly shown by Milner Introduction. As a scientia infima, linguistics certainly has the fundamental position attributed to it by medieval classifications, which placed grammar first among the seven disciplines of the School. If language is the condition of all learning, grammar--which renders a science of language possible--is the science that conditions all others. And it is easy to see that a science in the modern sense is possible only if language possesses certain recognizable properties; if language were without such properties, or if they had not remained the same, no knowledge would be possible. But there is more to be said. It is only because linguistics presupposes the factum loquendi, presupposes the existence of language, that other sciences can presuppose the existence of something that, in turn, underlies the objects whose properties they describe. Pure existence corresponds to the pure existence of language, and to contemplate one is to contemplate the other. The "literalization" effected by gram mar--"literalization" in the sense we have seen, insofar as it implies the irreducible difference between restriction and the substance of restricted beings--then constitutes the fundamental literalization determining all others. In this sense, it is surely significant that the grammarians of antiquity held as the principle of their knowledge not the pure voice but the "written voice," phōnē engrammatos, vox quae scribi potest. The principle of the science of language (and hence of every epistēmē) is grammaticalization, the literalization of the voice. What is at issue in this literalization is the existence of language as presupposition, the transformation of the factum loquendi into a presupposition that must remain unthought.
How, we must then ask ourselves, is the science of language itself marked by the existence of language as the presupposition at issue in literalization? In his preceding books (not only in L'amour de la langue but also in De la syntaxe à l'interprétation), Milner brought to light the points at which something in language exceeds language as an object of knowledge, whether it be in the theory of the subject of enunciation or in the grammar of insults.
If there is something in the Introduction that bears the trace or scar of this presupposition and excess, it is the theme of contingency, which traverses the whole book. In Milner's conception, Galilean (or literalized) science is destined to contingency. What clearly distinguishes it from classical science is that its object could have been otherwise than it is; the properties that belong to it are certain and constant but not necessary. The disorder that contingency introduces into the world is nevertheless balanced by a principle that is more or less present in all knowledge and that was clearly formulated by Aristotle. This principle, which is usually called the "principle of conditioned necessity," states that if all potentiality is potentiality of a thing and its contrary, and if every being could have been different, nevertheless, in the instant in which it actually is, it cannot be otherwise. As Milner wrote in a text on Lacan, "in the instant of a flash, each point of each referent of each proposition of science appears as if it could be infinitely different, from an infinite number of viewpoints; in the final instant, the letter fixes it as it is and as incapable of being otherwise." 5 It follows that contingency is contained in a barrier that always necessarily inscribes its expression in the form of a past: something could have been otherwise than it is. This temporal articulation in fact conditions Western science's entire representation of possibility (and this is as true of linguistics as of all other disciplines).
This said, is it possible to grasp contingency otherwise than as "something that could have been"? Is it possible, in other words, to call into question the principle of conditioned necessity, to attest to the very existence of potentiality, the actuality of contingency? Is it possible, in short, to attempt to say what seems impossible to say, that is: that something is otherwise than it is?
This appears to be precisely the task of coming philosophy: to redefine the entire domain of categories and modality so as to consider no longer the presupposition of Being and potentiality, but their exposition. This is the direction in which Milner's most recent work seems to move. And if there is a linguist today who is capable of grasping language's point of excess with respect to science (as Saussure and Benveniste did in their time), it is surely the author of this Introduction.

Notes:
1.
Jean-Claude Milner, Introduction à une science du langage ( Paris: Seuil, 1990), p. 10. All page citations included in the text of this chapter refer to this edition.

2.
Ibid., pp. 109-26. See also Milner article "L'exemple et la fiction," in Transparence et opacité, ed. T Papp and P Pica ( Paris: Le Cerf, 1988), pp. 145-81.

3.
K. Reach, "The Name Relation and the Logical Antinomies," Journal of Symbolic Logic 3 ( 1938): 97-111.

4.
See Milner, Introduction, pp. 216-36.

5.
Jean-Claude Milner, "Lacan et la science," lecture given in May 1990, on the occasion of the conference "Lacan et la Philosophie."