Roland Barthes (1964)

Elements of Semiology


Source: Elements of Semiology, 1964, publ. Hill and Wang, 1968. The first half of the book is reproduced here.


INTRODUCTION

In his Course in General Linguistics, first published in 1916, Saussure postulated the existence of a general science of signs, or Semiology, of which linguistics would form only one part. Semiology therefore aims to take in any system of signs, whatever their substance and limits; images, gestures, musical sounds, objects, and the complex associations of all these, which form the content of ritual, convention or public entertainment: these constitute, if not languages, at least systems of signification. There is no doubt that the development of mass communications confers particular relevance today upon the vast field of signifying media, just when the success of disciplines such as linguistics, information theory, formal logic and structural anthropology provide semantic analysis with new instruments. There is at present a kind of demand for semiology, stemming not from the fads of a few scholars, but from the very history of the modern world.

The fact remains that, although Saussure's ideas have made great headway, semiology remains a tentative science. The reason for this may well be simple. Saussure, followed in this by the main semiologists, thought that linguistics merely formed a part of the general science of signs. Now it is far from certain that in the social life of today there are to be found any extensive systems of signs outside human language. Semiology has so far concerned itself with codes of no more than slight interest, such as the Highway Code; the moment we go on to systems where the sociological significance is more than superficial, we are once more confronted with language. it is true that objects, images and patterns of behaviour can signify, and do so on a large scale, but never autonomously; every semiological system has its linguistic admixture. Where there is a visual substance, for example, the meaning is confirmed by being duplicated in a linguistic message (which happens in the case of the cinema, advertising, comic strips, press photography, etc.) so that at least a part of the iconic message is, in terms of structural relationship, either redundant or taken up by the linguistic system. As for collections of objects (clothes, food), they enjoy the status of systems only in so far as they pass through the relay of language, which extracts their signifiers (in the form of nomenclature) and names their signifieds (in the forms of usages or reasons): we are, much more than in former times, and despite the spread of pictorial illustration, a civilisation of the written word. Finally, and in more general terms, it appears increasingly more difficult to conceive a system of images and objects whose signifieds can exist independently of language: to perceive what a substance signifies is inevitably to fall back on the individuation of a language: there is no meaning which is not designated, and the world of signifieds is none other than that of language.

Thus, though working at the outset on nonlinguistic substances, semiology is required, sooner or later, to find language (in the ordinary sense of the term) in its path, not only as a model, but also as component, relay or signified. Even so, such language is not quite that of the linguist: it is a second-order language, with its unities no longer monemes or phonemes, but larger fragments of discourse referring to objects or episodes whose meaning underlies language, but can never exist independently of it. Semiology is therefore perhaps destined to be absorbed into a trans-linguistics, the materials of which may be myth, narrative, journalism, or on the other hand objects of our civilisation, in so far as they are spoken (through press, prospectus, interview, conversation and perhaps even the inner language, which is ruled by the laws of imagination). In fact, we must now face the possibility of inverting Saussure's declaration: linguistics is not a part of the general science of signs, even a privileged part, it is semiology which is a part of linguistics: to be precise, it is that part covering the great signifying unities of discourse. By this inversion we may expect to bring to light the unity of the research at present being done in anthropology, sociology, psychoanalysis and stylistics round the concept of signification.

Though it will doubtless be required some day to change its character, semiology must first of all, if not exactly take definite shape, at least try itself out, explore its possibilities and impossibilities. This is feasible only on the basis of preparatory investigation. And indeed it must be acknowledged in advance that such an investigation is both diffident and rash: diffident because semiological knowledge at present can be only a copy of linguistic knowledge; rash because this knowledge must be applied forthwith, at least as a project, to non-linguistic objects.

The Elements here presented have as their sole aim the extraction from linguistics of analytical concepts, which we think a priori to be sufficiently general to start semiological research on its way. In assembling them, it is not presupposed that they will remain intact during the course of research; nor that semiology will always be forced to follow the linguistic model closely.' We are merely suggesting and elucidating a terminology in the hope that it may enable an initial (albeit provisional) order to be introduced into the heterogeneous mass of significant facts. In fact what we purport to do is to furnish a principle of classification of the questions.

These elements of semiology will therefore be grouped under four main headings borrowed from structural linguistics:

I. Language and Speech.
II. Signified and Signifier.
III. Syntagm and System.
IV. Denotation and Connotation.

It will be seen that these headings appear in dichotomic form; the reader will also notice that the binary classification of concepts seems frequent in structural thoughts as if the metalanguage of the linguist reproduced, like a mirror, the binary structure of the system it is describing; and we shall point out, as the occasion arises, that it would probably be very instructive to study the pre-eminence of binary classification in the discourse of contemporary social sciences. The taxonomy of these sciences, if it were well known, would undoubtedly provide a great deal of information on what might be called the field of intellectual imagination in our time.

I. LANGUAGE (LANGUE) AND SPEECH

I.1. IN LINGUISTICS

I.1.1 In Saussure: The (dichotomic) concept of language/speech is central in Saussure and was certainly a great novelty in relation to earlier linguistics which sought to find the causes of historical changes in the evolution of pronunciation, spontaneous associations and the working of analogy, and was therefore a linguistics of the individual act. In working out this famous dichotomy, Saussure started from the multiform and heterogeneous' nature of language, which appears at first sight as an unclassifiable reality' the unity of which cannot be brought to light, since it partakes at the same time of the physical, the physiological, the mental, the individual and the social. Now this disorder disappears if, from this heterogeneous whole, is extracted a purely social object, the systematised set of conventions necessary to communication, indifferent to the material of the signals which compose it, and which is a language (langue); as opposed to which speech (parole) covers the purely individual part of language (phonation, application of the rules and contingent combinations of signs).

I.1.2. The language (langue): A language is therefore, so to speak, language minus speech: it is at the same time a social institution and a system of values. As a social institution, it is by no means an act, and it is not subject to any premeditation. It is the social part of language, the individual cannot by himself either create or modify it; it is essentially a collective contract which one must accept in its entirety if one wishes to communicate. Moreover, this social product is autonomous, like a game with its own rules, for it can be handled only after a period of learning. As a system of values, a language is made of a certain number of elements, each one of which is at the same time the equivalent of a given quantity of things and a term of a larger function, in which are found, in a differential order, other correlative values: from the point of view of the language, the sign is like a coin,' which has the value of a certain amount of goods which it allows one to buy, but also has value in relation to other coins, in a greater or lesser degree. The institutional and the systematic aspect are of course connected: it is because a language is a system of contractual values (in part arbitrary, or, more exactly, unmotivated) that it resists the modifications coming from a single individual, and is consequently a social institution.

I.1.3. Speech (parole): In contrast to the language, which is both institution and system, speech is essentially an individual act of selection and actualisation; it is made in the first place of the 'combination thanks to which the speaking subject can use the code of the language with a view to expressing his personal thought' (this extended speech could be called discourse), - and secondly by the 'psycho-physical mechanisms which allow him to exteriorise these combinations.' It is certain that phonation, for instance, cannot he confused with the language; neither the institution nor the system are altered if the individual who resorts to them speaks loudly or softly, with slow or rapid delivery, etc. The combinative aspect of speech is of course of capital importance, for it implies that speech is constituted by the recurrence of identical signs: it is because signs are repeated in successive discourses and within one and the same discourse (although they are combined in accordance with the infinite diversity of various people's speech) that each sign becomes an element of the language; and it is because speech is essentially a combinative activity that it corresponds to an individual act and not to a pure creation.

I.1.4. The dialectics of language and speech: Language and speech: each of these two terms of course achieves its full definition only in the dialectical process which unites one to the other: there is no language without speech, and no speech outside language: it is in this exchange that the real linguistic praxis is situated, as Merleau-Ponty has pointed out. And V. Brondal writes, 'A language is a purely abstract entity, a norm which stands above individuals, a set of essential types, which speech actualises in an infinite variety of ways."' Language and speech are therefore in a relation of reciprocal comprehensiveness. On the one hand, the language is 'the treasure deposited by the practice of speech, in the subjects belonging to the same community' and, since it is a collective summa of individual imprints, it must remain incomplete at the level of each isolated individual: a language does not exist perfectly except in the 'speaking mass'; one cannot handle speech except by drawing on the language. But conversely, a language is possible only starting from speech: historically, speech phenomena always precede language phenomena (it is speech which makes language evolve), and genetically, a language is constituted in the individual through his learning from the environmental speech (one does not teach grammar and vocabulary which are, broadly speaking, the language, to babies). To sum, a language is at the same time the product and the instrument of speech: their relationship is therefore a genuinely dialectical one. It will be noticed (an important fact when we come to semiological prospects) that there could not possibly be (at least according to Saussure) a linguistics of speech, since any speech, as soon as it is grasped as a process of communication, is already part of the language: the latter only can be the object of a science. This disposes of two questions at the outset: it is useless to wonder whether speech must be studied before the language: the opposite is impossible: one can only study speech straight away inasmuch as it reflects the language (inasmuch as it is 'glottic'). it is just as useless to wonder at the outset how to separate the language from speech: this is no preliminary operation, but on the contrary the very essence of linguistic and later semiological investigation: to separate the language from speech means ipso facto constituting the problematics of the meaning.

I.1.5. In Hjelmslev: Hjelmslev has not thrown over Saussure's conception of language/speech, but he has redistributed its terms in a more formal way. Within the language itself (which is still opposed to the act of speech) Hjelmslev distinguishes three planes: i) the schema, which is the language as pure form (before choosing this term Hjelmslev hesitated between system, pattern' or 'framework' for this plane):* this is Saussure's langue in the strictest sense of the word. It might mean, for instance, the French r as defined phonologically by its place in a series of oppositions; ii) the norm, which is the language as material form, after it has been defined by some degree of social realisation, but still independent of this realisation; it would mean the r in oral French, whichever way it is pronounced (but not that of written French); iii) the usage, which is the language as a set of habits prevailing in a given society: this would mean the r as it is pronounced in some regions. The relations of determination ' between speech, usage, norm and schema are varied: the norm determines usage and speech; usage determines speech but is also determined by it; the schema is determined at the same time by speech, usage and norm. Thus appear (in fact) two fundamental planes: i) the schema, the theory of which merges with that of the form" and of the linguistic institution; ii) the group norm-usage-speech, the theory of which merges with that of the substance' and of the execution. As according to Hjelmslev - norm is a pure methodical abstraction and speech a single concretion ('a transient document'), we find in the end a new dichotomy schema/usage, Which replaces the couple language/speech. This redistribution by Hjelmslev is not without interest, however: it is a radical formalisation of the concept of the language (under the name of schema) and eliminates concrete speech in favour of a more social concept: usage. This formalisation of the language and socialisation of speech enables us to put all the 'positive' and 'substantial' elements under the heading of speech, and all the differentiating ones under that of the language, and the advantage of this, as we shall see presently, is to remove one of the contradictions brought about by Saussure's distinction between the language and the speech.

I.1.6. Some problems: Whatever its usefulness and its fecundity, this distinction nevertheless brings some problems in its wake. Let us mention only three.

Here is the first: is it possible to identify the language with the code and the speech with the message? This identification is impossible according to Hjelmslev's theory. P. Guiraud refuses it for, he says, the conventions of the code are explicit, and those of the language implicit; but it is certainly acceptable in the Saussurean framework, and A. Martinet takes it up.

We encounter an analogous problem if we reflect on the relations between speech and syntagm. Speech, as we have seen, can be defined (outside the variations of intensity in the phonation) as a (varied) combination of (recurrent) signs; but at the level of the language itself, however, there already exist some fixed syntagms (Saussure cites a compound word like magnanimus). The threshold which separates the language from speech may therefore be precarious, since it is here constituted by 'a certain degree of combination'. This leads to the question of an analysis of those fixed syntagms whose nature is nevertheless linguistic (glottic) since they are treated as one by paradigmatic variation (Hjelmslev calls this analysis morpho-syntax). Saussure had noticed this phenomenon of transition: 'there is probably also a whole series of sentences which belong to the language, and which the individual no longer has to combine himself.' If these stereotypes belong to the language and no longer to speech, and if it proves true that numerous semiological systems use them to a great extent, then it is a real linguistics of the syntagm that we must expect, which will be used for all strongly stereotyped 'modes of writing'.

Finally, the third problem we shall indicate concerns the relations of the language with relevance (that is to say, with the signifying element proper in the unit). The language and relevance have sometimes been identified (by Trubetzkoy himself), thus thrusting outside the language all the non-relevant elements, that is, the combinative variants. Yet this identification raises a problem, for there are combinative variants (which therefore at first sight are a speech phenomenon) which are nevertheless imposed, that is to say, arbitrary : in French, it is required by the language that the I should be voiceless after a voiceless consonant (oncle) and voiced after a voiced consonant (ongle) without these facts leaving the realm of phonetics to belong to that of phonology. We see the theoretical consequences: must we admit that, contrary to Saussure's affirmation ('in the language there are only differences'), elements which are not differentiating can all the same belong to the language (to the institution)? Martinet thinks so; Frei attempts to extricate Saussure from the contradiction by localising the differences in subphonemes, so that, for instance, p could not be differentiating in itself, but only, in it, the consonantic, occlusive voiceless labial features, etc. We shall not here take sides on this question; from a semiological point of view, we shall only remember the necessity of accepting the existence of syntagms and variations which are not signifying and are yet 'glottic', that is, belonging to the language. This linguistics, hardly foreseen by Saussure, can assume a great importance wherever fixed syntagms (or stereotypes) are found in abundance, which is probably the case in mass-languages, and every time non-signifying variations form a second-order corpus of signifiers, which is the case in strongly connated languages : the rolled r is a mere combinative variant at the denotative level, but in the speech of the theatre, for instance, it signals a country accent and therefore is a part of a code, without which the message of 'ruralness' could not be either emitted or perceived.

I.1.7. The idiolect: To finish on the subject of language/speech in linguistics, we shall indicate two appended concepts isolated since Saussure's day. The first is that of the idiolect. This is 'the language inasmuch as it is spoken by a single individual' (Martinet), or again 'the whole set of habits of a single individual at a given moment' (Ebeling). Jakobson has questioned the interest of this notion: the language is always socialised, even at the individual level, for in speaking to somebody one always tries to speak more or less the other's language, especially as far as the vocabulary is concerned ('private property in the sphere of language does not exist') : so the idiolect would appear to be largely an illusion. We shall nevertheless retain from this notion the idea that it can be useful to designate the following realities: i) the language of the aphasic who does not understand other people and does not receive a message conforming to his own verbal patterns; this language, then, would be a pure idiolect (Jakobson); ii) the 'style' of a writer, although this is always pervaded by certain verbal patterns coming from tradition that is, from the community; iii) finally, we can openly broaden the notion, and define the idiolect as the language of a linguistic community, that is, of a group of persons who all interpret in the same way all linguistic statements: the idiolect would then correspond roughly to what we have attempted to describe elsewhere under the name of 'writing'." We can say in general that the hesitations in defining the concept of idiolect only reflect the need for an intermediate entity between speech and language (as was already proved by the usage theory in Hjelmslev), or, if you like, the need for a speech which is already institutionalised but not yet radically open to formalisation, as the language is.

I.1.8. Duplex Structures: If we agree to identify language/speech and code/message, we must here mention a second appended concept which Jakobson has elaborated under the name of duplex structures; we shall do so only briefly, for his exposition of it has been reprinted. IT We shall merely point out that under the name 'duplex structures' Jakobson studies certain special cases of the general relation code/message: two cases of circularity and two cases of overlapping. i) reported speech, or messages within a message (M/M): this is the general case of indirect styles. ii) proper names: the name signifies any person to whom this name is attributed and the circularity of the code is evident (C/C): John means a person named John; iii) cases of autonymy ('Rat is a syllable'): the word is here used as its own designation, the message overlaps the code (M/C) - this structure is important, for it covers the 'elucidating interpretations', namely, circumlocutions, synonyms and translations from one language into another; iv) the shifters are probably the most interesting double structure: the most ready example is that of the personal pronoun (I, thou) an indicial symbol which unites within itself the conventional and the existential bonds: for it is only by virtue of a conventional rule that I represents its object (so that I becomes ego in Latin, ich in German, etc.), but on the other hand, since it designates the person who utters it, it can only refer existentially to the utterance (C/M). Jakobson reminds us that personal pronouns have long been thought to be the most primitive layer of language (Humboldt), but that in his view, they point rather to a complex and adult relationship between the code and the message: the personal pronouns are the last elements to be acquired in the child's speech and the first to be lost in aphasia; they are terms of transference which are difficult to handle. The shifter theory seems as yet to have been little exploited; yet it is, a priori, very fruitful to observe the code struggling with the message, so to speak (the converse being much more commonplace); perhaps (this is only a working hypothesis) it is on this side, that of the shifters, which are, as we saw, indicial symbols according to Peirce's terminology, that we should seek the semiological definition of the messages which stand on the frontiers of language, notably certain forms of literary discourse.

I.2. SEMIOLOGICAL PROSPECTS

I.2.1. The language, speech and the social sciences. The sociological scope of the language/speech concept is obvious. The manifest affinity of the language according to Saussure and of Durkheim's conception of a collective consciousness independent of its individual manifestations has been emphasised very early on. A direct influence of Durkheim on Saussure has even been postulated, it has been alleged that Saussure had followed very closely the debate between Durkheim and Tarde and that his conception of the language came from Durkheim while that of speech was a kind of concession to Tarde's idea on the individual element. This hypothesis has lost some of its topicality because linguistics has chiefly developed, in the Saussurean idea of the language, the 'system of values' aspect, which led to acceptance of the necessity for an immanent analysis of the linguistic institution, and this immanence is inimical to sociological research.

Paradoxically, it is not therefore in the realm of sociology that the best development of the notion of language/speech will be found; it is in philosophy, with Merleau-Ponty, who was probably one of the first French philosophers to become interested in Saussure. He took up again the Saussurean distinction as an opposition between speaking speech (a signifying intention in its nascent state) and spoken speech (an 'acquired wealth' of the language which does recall Saussure's 'treasure'). He also broadened the notion by postulating that any process presupposes a system : thus there has been elaborated an opposition between event and structure which has become accepted" and whose fruitfulness in history is well known.

Saussure's notion has, of course, also been taken over and elaborated in the field of anthropology. The reference to Saussure is too explicit in the whole work of Claude Lévi-Strauss for us to need to insist on it; we shall simply remind the reader of three facts: i) That the opposition between process and system (speech and language) is found again in a concrete guise in the transition from the exchange of women to the structures of kinship; ii) that for Lévi-Strauss this opposition has an epistemological value: the study of linguistic phenomena is the domain of mechanistic (in Lévi-Strauss's sense of the word, namely, as opposed to 'statistical') and structural interpretation, and the study of speech phenomena is the domain of the theory of probabilities (macrolinguistics);" iii) finally, that the unconscious character of the language in those who draw on it for their speech, which is explicitly postulated by Saussure, is again found in one of the most original and fruitful contentions of Lévi-Strauss, which states that it is not the contents which are unconscious (this is a criticism of Jung's archetypes) but the forms, that is, the symbolical function.

This idea is akin to that of Lacan, according to whom the libido itself is articulated as a system of significations, from which there follows, or will have to follow, a new type of description of the collective field of imagination, not by means of its 'themes', as has been done until now, but by its forms and its functions. Or let us say, more broadly but more clearly: by its signifiers more than by its signifieds.

It can be seen from these brief indications how rich in extra- or meta-linguistic developments the notion language/speech is. We shall therefore postulate that there exists a general category language/speech, which embraces all the systems of signs; since there are no better ones, we shall keep the terms language and speech, even when they are applied to communications whose substance is not verbal.

I.2.2. The garment system: We saw that the separation between the language and speech represented the essential feature of linguistic analysis; it would therefore be futile to propose to apply this separation straightaway to systems of objects, images or behaviour patterns which have not yet been studied from a semantic point of view. We can merely, in the case of some of these hypothetical systems, foresee that certain classes of facts will belong to the category of the language and others to that of speech, and make it immediately clear that in the course of its application to semiology, Saussure's distinction is likely to undergo modifications which it will be precisely our task to note.

Let us take the garment system for instance; it is probably necessary to subdivide it into three different systems, according to which substance is used for communication.

In clothes as written about, that is to say described in a fashion magazine by means of articulated language, there is Practically no 'speech': the garment which is described never corresponds to an individual handling of the rules of fashion, it is a systematised set of signs and rules: it is a language in its pure state. According to the Saussurean schema, a language without speech would be impossible; what makes the fact acceptable here is, on the one hand, that the language of fashion does not emanate from the ,speaking mass' but from a group which makes the decisions and deliberately elaborates the code, and on the other hand that the abstraction inherent in any language is here materialised as written language: fashion clothes (as written about) are the language at the level of vestimentary communication and speech at the level of verbal communication.

In clothes as photographed (if we suppose, to simplify matters, that there is no duplication by verbal description), the language still issues from the fashion group, but it is no longer given in a wholly abstract form, for a photographed garment is always worn by an individual woman. What is given by the fashion photograph is a semi-formalised state of the garment system: for on the one hand, the language of fashion must here be inferred from a pseudo-real garment, and on the other, the wearer of the garment (the photographed model) is, so to speak, a normative individual, chosen for her canonic generality, and who Consequently represents a 'speech' which is fixed and devoid of all combinative freedom.

Finally in clothes as worn (or real clothes), as Trubetzkoy had suggested," we again find the classic distinction between language and speech. The language, in the garment system, is made i) by the oppositions of pieces, parts of garment and 'details', the variation of which entails a change in meaning (to wear a beret or a bowler hat does not have the same meaning); ii) by the rules which govern the association of the pieces among themselves, either on the length of the body or in depth. Speech, in the garment system, comprises all the phenomena of anomic fabrication (few are still left in our society) or of individual way of wearing (size of the garment, degree of cleanliness or wear, personal quirks, free association of pieces). As for the dialectic which unites here costume (the language) and clothing (speech), it does not resemble that of verbal language; true, clothing always draws on costume (except in the case of eccentricity, which, by the way, also has its signs), but costume, at least today, precedes clothing, since it comes from the ready-made industry, that is, from a minority group (although more anonymous than that of Haute Couture).

I.2.3. The food system: Let us now take another signifying system: food. We shall find there without difficulty Saussure's distinction. The alimentary language is made of i) rules of exclusion (alimentary taboos); ii) signifying oppositions of units, the type of which remains to be determined (for instance the type savoury/sweet); iii) rules of association, either simultaneous (at the level of a dish) or successive (at the level of a menu); iv) rituals of use which function, perhaps, as a kind of alimentary rhetoric. As for alimentary 'speech', which is very rich, it comprises all the personal (or family) variations of preparation and association (one might consider cookery within one family, which is subject to a number of habits, as an idiolect). The menu, for instance, illustrates very well this relationship between the language and speech: any menu is concocted with reference to a structure (which is both national - or regional - and social); but this structure is filled differently according to the days and the users, just as a linguistic 'form' is filled by the free variations and combinations which a speaker needs for a particular message. The relationship between the language and speech would here be fairly similar to that which is found in verbal language: broadly, it is usage, that is to say, a sort of sedimentation of many people's speech, which makes up the alimentary language; however, phenomena of individual innovation can acquire an institutional value within it. What is missing, in any case, contrary to what happened in the garment system, is the action of a deciding group: the alimentary language is evolved only from a broadly collective usage, or from a purely individual speech.

I.2.4. The car system, the furniture system: To bring to a close, somewhat arbitrarily, this question of the prospects opened up by the language/speech distinction, we shall mention a few more suggestions concerning two systems of objects, very different, it is true, but which have in common a dependence in each case on a deciding and manufacturing group: cars and furniture.

In the car system, the language is made up by a whole set of forms and details, the structure of which is established differentially by comparing the prototypes to each other (independently of the number of their 'copies'); the scope of 'speech' is very narrow because, for a given status of buyer, freedom in choosing a model is very restricted: it can involve only two or three models, and within each model, colour and fittings. But perhaps we should here exchange the notion of cars as objects for that of cars as sociological facts; we would then find in the driving of cars the variations in usage of the object which usually make up the plane of speech. For the user cannot in this instance have a direct action on the model and combine its units; his freedom of interpretation is found in the usage developed in time and within which the 'forms' issuing from the language must, in order to become actual, be relayed by certain practices.

Finally, the last system about which we should like to say a word, that of furniture, is also a semantic object: the 'language' is formed both by the oppositions of functionally identical pieces (two types of wardrobe, two types of bed, etc), each of which, according to its 'style', refers to a different meaning, and by the rules of association of the different units at the level of a room ('furnishing'); the 'speech' is here formed either by the insignificant variations which the user can introduce into one unit (by tinkering with one element, for instance), or by freedom in associating pieces of furniture together.

I.2.5. Complex systems: The most interesting systems, at least among those which belong to the province of mass-communications, are complex systems in which different substances are engaged. In cinema, television and advertising, the senses are subjected to the concerted action of a collection of images, sounds and written words. It will, therefore, be premature to decide, in their case, which facts belong to the language and which belong to speech, on the one hand as long as one has not discovered whether the 'language' of each of these complex systems is original or only compounded of the subsidiary 'languages' which have their, places in them, and on the other hand as long as these subsidiary languages have not been analysed (we know the linguistic 'language', but not that of images or that of music).

As for the Press, which can be reasonably considered as an autonomous signifying system, even if we confine ourselves to its written elements only, we are still almost entirely ignorant of a linguistic phenomenon which seems to play an essential part in it: connotation, that is, the development of a system of second-order meanings, which are so to speak parasitic on the language proper . This second order system is also a 'language', within which there develop speech-phenomena, idiolects and duplex structures. In the case of such complex or connoted systems (both characteristics are not mutually exclusive), it is therefore no longer possible to predetermine, even in global and hypothetical fashion, what belongs to the language and what belongs to speech.

I.2.6. Problems (I) - the origin of the various signifyings systems: The semiological extension of the language/speech notion brings with it some problems, which of course coincide with the points where the linguistic model can no longer be followed and must be altered. The first problem concerns the origin of the various systems, and thus touches on the very dialectics of language and speech. In the linguistic model, nothing enters the language without having been tried in speech, but conversely no speech is possible (that is, fulfils its function of communication) if it is not drawn from the 'treasure' of the language. This process is still, at least partially, found in a system like that of food, although individual innovations brought into it can become language phenomena. But in most other semiological systems, the language is elaborated not by the 'speaking mass' but by a deciding group. In this sense, it can be held that in most semiological languages, the sign is really and truly 'arbitrary"' since it is founded in artificial fashion by a unilateral decision; these in fact are fabricated languages, 'logo-techniques'. The user follows these languages, draws messages (or 'speech') from them but has no part in their elaboration. The deciding group which is at the origin of the system (and of its changes) can be more or less narrow; it can be a highly qualified technocracy (fashion, motor industry); it can also be a more diffuse and anonymous group (the production of standardised furniture, the middle reaches of ready-to-wear). If, however, this artificial character does not alter the institutional nature of the communication and preserves some amount of dialectical play between the system and usage, it is because, in the first place, although imposed on the users, the signifying 'contract' is no less observed by the great majority of them (otherwise the user is marked with a certain 'asociability': he can no longer communicate anything except his eccentricity); and because, moreover, languages elaborated as the outcome of a decision are not entirely free ('arbitrary'). They are subject to the determination of the community, at least through the following agencies: i) when new needs are born, following the development of societies (the move to semi-European clothing in contemporary African countries, the birth of new patterns of quick feeding in industrial and urban societies); ii) when economic requirements bring about the disappearance or promotion of certain materials (artificial textiles); iii) when ideology limits the invention of forms, subjects it to taboos and reduces, so to speak, the margins of the 'normal'. In a wider sense, we can say that the elaborations of deciding groups, namely the logo-techniques, are themselves only the terms of an ever-widening function, which is the collective field of imagination of the epoch: thus individual innovation is transcended by a sociological determination (from restricted groups), but these sociological determinations refer in turn to a final meaning, which is anthropological.

I.2.7. Problems (II) - the proportion between 'language' and 'speech' in the various systems: The second problem presented by the semiological extension of the language/speech notion is centred on the proportion, in the matter of volume, which can be established between the 'language' and the corresponding 'speech' in any system. In verbal language there is a very great disproportion between the language, which is a finite set of rules, and speech, which comes under the heading of these rules and is practically unlimited in its variety. It can be presumed that the food system still offers an important difference in the volume of each, since within the culinary 'forms', the modalities and combinations in interpretation are numerous. But we have seen that in the car or the furniture system the scope for combinative variations and free associations is small: there is very little margin - at least of the sort which is acknowledged by the institution itself - between the model and its 'execution': these are systems in which 'speech' is poor. In a particular system, that of written fashion, speech is even almost non-existent, so that we are dealing here, paradoxically, with a language without speech (which is possible, as we have seen, only because this language is upheld by linguistic speech).

The fact remains that if it is true that there are languages without speech or with a very restricted speech, we shall have to revise the Saussurean theory which states that a language is nothing but a system of differences (in which case, being entirely negative, it cannot be grasped outside speech). and complete the couple language/speech with a third, presignifying element, a matter or substance providing the (necessary) support of signification. In a phrase like a long or short dress, the 'dress' is only the support of a variant (long/short) which does fully belong to the garment language - a distinction which is unknown in ordinary language, in which, since the sound is considered as immediately significant, it cannot be decomposed into an inert and a semantic element. This would lead us to recognise in (non-linguistic) semiological systems three (and not two) planes: that of the matter, that of the language and that of the usage. This of course allows us to account for systems without 'execution', since the first element ensures that there is a materiality of the language; and such a modification is all the more plausible since it can be explained genetically: if, in such systems, the 'language' needs a 'matter' (and no longer a 'speech'), it is because unlike that of human language their origin is in general utilitarian, and not signifying.

II. SIGNIFIER AND SIGNIFIED

II.1. THE SIGN

The classification of signs: The signified and the signifier, in Saussurean terminology, are the components of the sign. Now this term, sign, which is found in very different vocabularies (from that of theology to that of medicine), and whose history is very rich (running from the Gospels"' to cybernetics), is for these very reasons very ambiguous; so before we come back to the Saussurean acceptance of the word, we must say a word about the notional field in which it occupies a place, albeit imprecise, as will be seen. For, according to the arbitrary choice of various authors, the sign is placed in a series of terms which have affinities and dissimilarities with it: signal, index, icon, symbol, allegory, are the chief rivals of sign. Let us first state the element which is common to all these terms: they all necessarily refer us to a relation between two relata. This feature cannot therefore be used to distinguish any of the terms in the series; to find a variation in meaning, we shall have to resort to other features, which will be expressed here in the form of an alternative (presences absence): i) the relation implies, or does not imply, the mental representation of one of the relata; ii) the relation implies, or does not imply, an analogy between the relata; iii) the link between the two relata (the stimulus and its response) is immediate or is not; iv) the relata exactly coincide or, on the contrary, one overruns the other; v) the relation implies, or does not imply, an existential connection with the user. Whether these features are positive or negative (marked or unmarked), each term in the field is differentiated from its neighbours. It must be added that the distribution of the field varies from one author to another, a fact which produces terminological contradictions; these will be easily seen at a glance from a table of the incidence of features and terms in four different authors: Hegel, Peirce, Jung and Wallon (the reference to some features, whether marked or unmarked, may be absent in some authors). We see that the terminological contradiction bears essentially on index (for Peirce, the index is existential, for Wallon, it is not) and on symbol (for Hegel and Wallon there is a relation of analogy - or of ,motivation' - between the two relata of the symbol, but not for Peirce; moreover, for Peirce, the symbol is not existential, whereas it is for Jung). But we see also that these contradictions - which in this table are read vertically - are very well explained, or rather, that they compensate each other through transfers of meaning from term to term in the same author. These transfers can here be read horizontally: for instance, the symbol is analogical in Hegel as opposed to the sign which is not; and if it is not in Peirce, it is because the icon can absorb that feature. All this means, to sum up and talk in semiological terms (this being the point of this brief analysis which reflects, like a mirror, the subject and methods of our study), that the words in the field derive their meaning only from their opposition to one another (usually in pairs), and that if these oppositions are preserved, the meaning is unambiguous. In particular, signal and index, symbol and sign, are the terms of two different functions, which can themselves be opposed-as a whole, as they do in Wallon, whose terminology is the clearest and the most complete (icon and allegory are confined to the vocabulary of Peirce and Jung). We shall therefore say, with Wallon, that the signal and the index form a group of relata devoid of mental representation, whereas in the opposite group, that of symbol and sign, this representation exists; furthermore, the signal is immediate and existential, whereas the index is not (it is only a trace); finally, that in the symbol the representation is analogical and inadequate (Christianity 'outruns' the cross), whereas in the sign the relation is unmotivated and exact (there is no analogy between the word ox and the image of an ox, which is perfectly covered by its relatum).

II.1.2. The linguistic sign: In linguistics, the notion of sign does not give rise to any competition between neighbouring terms. When he sought to designate the signifying relationship, Saussure immediately eliminated symbol (because the term implied the idea of motivation) in favour of sign which he defined as the union of a signifier and a signified (in the fashion of the recto and verso of a sheet of paper), or else of an acoustic image and a concept. Until he found the words signifier and signified, however, sign remained ambiguous, for it tended to become identified with the signifier only, which Saussure wanted at all costs to avoid; after having hesitated between sôme and same, form and idea, image and concept, Saussure settled upon signifier and signified, the union of which forms the sign. This is a paramount proposition, which one must always bear in mind, for there is a tendency to interpret sign as signifier, whereas this is a two-sided Janus-like entity. The (important) consequence is that, for Saussure, Hjelmslev and Frei at least, since the signifieds are signs among others, semantics must be a part of structural linguistics, whereas for the American mechanists the signifieds are substances which must be expelled from linguistics and left to psychology. Since Saussure, the theory of the linguistic sign has been enriched by the double articulation principle, the importance of which has been shown by Martinet, to the extent that he made it the criterion which defines language. For among linguistic signs, we must distinguish between the significant units, each one of which is endowed with one meaning (the 'words', or to be exact, the monemes') and which form the first articulation, and the distinctive units, which are part of the form but do not have a direct meaning ('the sounds', or rather the phonemes), and which constitute the second articulation. It is this double articulation which accounts for the economy of human language; for it is a powerful gearing-down which allows, for instance, American Spanish to produce, with only 2I distinctive units, 100,000 significant units.

II.1.3. Form and substance.- The sign is therefore a compound of a signifier and a signified. The plane of the signifiers constitutes the plane of expression and that of the signifieds the plane of content. Within each of these two planes, Hjelmslev has introduced a distinction which may be important for the study of the semiological (and no longer only linguistic) sign. According to him, each plane comprises two strata: form and substance; we must insist on the new definition of these two terms, for each of them has a weighty lexical past. The form is what can be described exhaustively, simply and coherently (epistemological criteria) by linguistics without resorting to any extra-linguistic premise; the substance is the whole set of aspects of linguistic phenomena which cannot be described without resorting to extra-linguistic premises. Since both strata exist on the plane of expression and the plane of content, we therefore have: i) a substance of expression: for instance the phonic, articulatory, non-functional substance which is the field of phonetics, not phonology; ii) a form of expression, made of the paradigmatic and syntactic rules (let us note that the same form can have two different substances, one phonic, the other graphic); iii) a substance of content: this includes, for instance, the emotional, ideological, or simply notional aspects of the signified, its 'positive' meaning; iv) a form of content: it is the formal organisation of the signified among themselves through the absence or presence of a semantic mark. This last notion is difficult to grasp, because of the impossibility of separating the signifiers from the signifieds in human language; but for this very reason the subdivision form/substance can be made more useful and easier to handle in semiology, in the following cases: i) when we deal with a system in which the signifieds are substantified in a substance other than that of their own system (this is, as we have seen, the case with fashion as it is written about); ii) when a system of objects includes a substance which is not immediately and functionally significant, but can be, at a certain level, simply utilitarian: the function of a dish can be to signify a situation and also to serve as food.

II.1.4. The semiological sign: This perhaps allows us to foresee the nature of the semiological sign in relation to the linguistic sign. The semiological sign is also, like its model, compounded of a signifier and a signified (the colour of a light, for instance, is an order to move on, in the Highway Code), but it differs from it at the level of its substances. Many semiological systems (objects, gestures, pictorial images) have a substance of expression whose essence is not to signify; often, they are objects of everyday use, used by society in a derivative way, to signify something: clothes are used for protection and food for nourishment even if they are also used as signs. We propose to call these semiological signs, whose origin is utilitarian and functional, sign-functions. The sign-function bears witness to a double movement, which must be taken apart. In a first stage (this analysis is purely operative and does not imply real temporality) the function becomes pervaded with meaning. This semantisation is inevitable: as soon as there is a society, every usage is converted into a sign of itself; the use of a raincoat is to give protection from the rain, but this use cannot be dissociated from the very signs of an atmospheric situation. Since our society produces only standardised, normalised objects, these objects are unavoidably realisations of a model, the speech of a language, the substances of a significant form. To rediscover a non-signifying object, one would have to imagine a utensil absolutely improvised and with no similarity to an existing model (Lévi-Strauss has shown to what extent tinkering about is itself the search for a meaning): a hypothesis which is virtually impossible to verify in any society. This universal semantisation of the usages is crucial: it expresses the fact that there is no reality except when it is intelligible, and should eventually lead to the merging of sociology with sociological But once the sign is constituted, society can very well refunctionalise it, and speak about it as if it were an object made for use: a fur-coat will be described as if it served only to protect from the cold. This recurrent functionalisation, which needs, in order to exist, a second-order language, is by no means the same as the first (and indeed purely ideal) functionalisation: for the function which is re-presented does in fact correspond to a second (disguised) semantic institutionalisation, which is of the order of connotation. The sign-function therefore has (probably) an anthropological value, since it is the very unit where the relations of the technical and the significant are woven together.

II.2. THE SIGNIFIED

II.2.1. Nature of the signified: In linguistics, the nature of the signified has given rise to discussions which have centred chiefly on its degree of 'reality'; all agree, however, on emphasising the fact that the signified is not 'a thing' but a mental representation of the 'thing'. We have seen that in the definition of the sign by Wallon, this representative character was a relevant feature of the sign and the symbol (as opposed to the index and the signal). Saussure himself has clearly marked the mental nature of the signified by calling it a concept: the signified of the word ox is not the animal ox, but its mental image (this will prove important in the subsequent discussion on the nature of the sign). These discussions, however, still bear the stamp of psychologism, so the analysis of the Stoics will perhaps be thought preferable. They carefully distinguished the phantasia logiki (the mental representation), the tinganon (the real thing) and the lekton (the utterable). The signified is neither the phantasia nor the tinganon but rather the lekton; being neither an act of consciousness, nor a real thing, it can be defined only within the signifying process, in a quasi-tautological way: it is this 'something' which is meant by the person who uses the sign. In this way we are back again to a purely functional definition: the signified is one of the two relata of the sign; the only difference which opposes it to the signified is that the latter is a mediator. The situation could not be essentially different in semiology, where objects, images, gestures, etc., inasmuch as they are significant, refer back to something which can be expressed only through them, except that the semiological signified can be taken up by the linguistic signs. One can say, for instance, that a certain sweater means long autumn walks in the woods; in this case, the signified is mediated not only by its vestimentary signifier (the sweater), but also by a fragment of speech (which greatly helps in handling it). We could give the name of isology to the phenomenon whereby language wields its signifiers and signifieds so that it is impossible to dissociate and differentiate them, in order to set aside the case of the non-isologic systems (which are inevitably complex), in which the signified can be simply juxtaposed with its signifier.

II.2.2. Classification of the linguistic signifieds: How can we classify the signifieds? We know that in semiology this operation is fundamental, since it amounts to isolating the form from the content. As far as linguistic signifiers are concerned, two sorts of classification can be conceived. The first is external, and makes use of the 'positive' (and not purely differential) content of concepts: this is the case in the methodical groupings of Hallig and Wartburg, and in the more convincing notional fields of Trier and lexicological fields of Matoré. But from a structural point of view, this classification (especially those of Hallig and Wartburg) have the defect of resting still too much on the (ideological) substance of the signifieds, and not on their form. To succeed in establishing a really formal classification, one would have to succeed in reconstituting oppositions of signifieds, and in isolating, within each one of these, a relevant commutative feature: this method has been advocated by Hjelmslev, Sørensen, Prieto and Greimas. Hjelmslev, for instance, decomposes a moneme like 'mare' into two smaller significant units: 'Horse' + 'female', and these units can be commutated and therefore used to reconstitute new monemes ('pig', + 'female' = 'sow', 'horse' + 'male' = 'stallion'); Prieto sees in 'vir' two commutable features 'homo' + 'masculus'; Sørensen reduces the lexicon of kinship to a combination of 'primitives' ('father' = male parent, 'parent' = first ascendant). None of these analyses has yet been developed . Finally, we must remind the reader that according to some linguists, the signifieds are not a part of linguistics, which is concerned only with signifiers, and that semantic classification lies outside the field of linguistics."

II.2.3. The semiological signifieds: Structural linguistics, however advanced, has not yet elaborated a semantics, that is to say a classification of the forms of the verbal signified. One may therefore easily imagine that it is at present impossible to put forward a classification of semiological signifieds, unless we choose to fall back on to known notional fields. We shall venture three observations only.

The first concerns the mode of actualisation of semiological signifieds. These can occur either isologically or not; in the latter case, they are taken up, through articulated language, either by a word (week-end) or by a group of words (long walks in the country); they are thereby easier to handle, since the analyst is not forced to impose on them his own metalanguage, but also more dangerous, since they ceaselessly refer back to the semantic classification of the language itself (which is itself unknown), and not to a classification having its bases in the system under observation. The signifieds of the fashion garment, even if they are mediated by the speech of the magazine, are not necessarily distributed like the signifieds of the language, since they do not always have the same 'length' (here a word, there a sentence). In the first case, that of the isologic systems, the signified has no materialisation other than its typical signifier; one cannot therefore handle it except by imposing on it a metalanguage. One can for instance ask some subjects about the meaning they attribute to a piece of music by submitting to them a list of verbalised signifieds (anguished, stormy, sombre, tormented, etc.);" whereas in fact all these verbal signs for a single musical signified, which ought to be designated by one single cipher, which would imply no verbal dissection and no metaphorical small change. These metalanguages, issuing from the analyst in the former case, and the system itself in the latter, are probably inevitable, and this is what still makes the analysis of the signifieds, or ideological analysis, problematical; its place within the semiological project will at least have to be defined in theory.

Our second remark concerns the extension of the semiological signifieds. The whole of the signifieds of a system (once formalised) constitutes a great function; now it is probable that from one system to the other, the great semiological functions not only communicate, but also partly overlap; the form of the signified in the garment system is probably partly the same as that of the signified in the food system, being, as they are, both articulated on the large-scale opposition of work and festivity, activity and leisure. One must therefore foresee a total ideological description, common to all the systems of a given synchrony.

Finally - and this will be our third remark - we may consider that to each system of magnifiers (lexicons) there corresponds, on the plane of the signifieds, a corpus of practices and techniques; these collections of signifieds imply on the part of system consumers (of 'readers', that is to say), different degrees of knowledge (according to differences in their 'culture'), which explains how the same 'lexie' (or large unit of reading) can be deciphered differently according to the individuals concerned, without ceasing to belong to a given 'language'. Several lexicons-and consequently several bodies of signifieds - can coexist within the same individual, determining in each one more or less 'deep' readings.

II.3. THE SIGNIFIER

II.3.1. Nature of the signaller. The nature of the signifier suggests roughly the same remarks as that of the signified: it is purely a relatum, whose definition cannot be separated from that of the signified. The only difference is that the magnifier is a mediator: some matter is necessary to it. But on the one hand it is not sufficient to it, and on the other, in semiology, the signifier can, too, be relayed by a certain matter: that of words. This materiality of the signifier makes it once more necessary to distinguish clearly matter from substance: a substance can be immaterial (in the case of the substance of the content); therefore, all one can say is that the substance of the signifier is always material (sounds, objects, images). In semiology, where we shall have to deal with mixed systems in which different kinds of matter are involved (sound and image, object and writing, etc.), it may be appropriate to collect together all the signs, inasmuch as they are home by one and the same matter, under the concept of the typical sign: the verbal sign, the graphic sign, the iconic sign, the gestural sign are all typical signs.

II.3.2. Classification of the signifiers: The clarification of the signifiers is nothing but the structuralisation proper of the system. What has to be done is to cut up the 'endless' message constituted by the whole of the messages emitted at the level of the studied corpus, into minimal significant units by means of the commutation test," then to group these units into paradigmatic classes, and finally to classify the syntagmatic relations which link these units. These operations constitute an important part of the semiological undertaking which will be dealt with in chapter 111; we anticipate the point in mentioning it here.

II.4. THE SIGNIFICATION

II.4.1. The significant correlation: The sign is a (two-faced) slice of sonority, visuality, etc. The signification can be conceived as a process; it is the act which binds the signifier and the signified, an act whose product is the sign. This distinction has, of course, only a classifying (and not phenomenological) value: firstly, because the union of signifier and signified, as we shall see, does not exhaust the semantic act, for the sign derives its value also from its surroundings; secondly, because, probably, the mind does not proceed, in the semantic process, by conjunction but by carving out. And indeed the signification (semiosis) does not unite unilateral entities, it does not conjoin two terms, for the very good reason that signifier and signified are both at once term and relation. This ambiguity makes any graphic representation of the signification somewhat clumsy, yet this operation is necessary for any semiological discourse. On this point, let us mention the following attempts:

1) Sr/Sd: In Saussure, the sign appears, in his demonstration, as the vertical extension of a situation in depth: in the language, the signified is, as it were, behind the signifier, and can be reached only through it, although, on the one hand, these excessively spatial metaphors miss the dialectical nature of the signification, and on the other hand the 'closed' character of the sign is acceptable only for the frankly discontinuous systems, such as that of the language.

2) ERC: Hjelmslev has chosen in preference a purely graphic representation: there is a relation (R) between the plane of expression (E) and the plane of content (C). This formula enables us to account economically and without metaphorical falsification, for the metalanguages or derivative systems E R (ERC).

3) S/S: Lacan, followed by Laplanche and Leclaire, uses a spatialised writing which, however, differs from Saussure's representation on two points: i) the signifier (S) is global, made up of a multi-levelled chain (metaphorical chain): signifier and signified have only a floating relationship and coincide only at certain anchorage points; ii) the line between the signifier (S) and the signified (s) has its own value (which of course it had not in Saussure): it represents the repression of the signified.

4) Sr = Sd: Finally, in non-isologic systems (that is, those in which the signifieds are materialised through another system), it is of course legitimate to extend the relation in the form of an equivalence but not of an identity.

II.4.2. The arbitrary and the motivated in linguistics: We have seen that all that could be said about the signifier is that it was a (material) mediator of the signified. What is the nature of this mediation? In linguistics, this problem has provoked some discussion, chiefly about terminology, for all is fairly clear about the main issues (this will perhaps not be the case with semiology). Starting from the fact that in human language the choice of sounds is not imposed on us by the meaning itself (the ox does not determine the sound ox, since in any case the sound is different in other languages), Saussure had spoken of an arbitrary relation between signifier and signified. Benveniste has questioned the aptness of this word: what is arbitrary is the relation between the signifier and the 'thing' which is signified (of the sound ox and the animal the ox). But, as we have seen, even for Saussure, the sign is not the 'thing', but the mental representation of the thing (concept); the association of sound and representation is the outcome of a collective training (for instance the learning of the French tongue); this association - which is the signification - is by no means arbitrary (for no French person is free to modify it), indeed it is, on the contrary, necessary. It was therefore suggested to say that in linguistics the signification is unmotivated. This lack of motivation, is, by the way, only partial (Saussure speaks of a relative analogy): from signified to signifier, there is a certain motivation in the (restricted) case of onomatopoeia, as we shall see shortly, and also every time a series of signs is created by the tongue through the imitation of a certain prototype of composition or derivation: this is the case with so-called proportional signs: pommier, poirer, abricotier, etc., once the lack of motivation in their roots and their suffix is established, show an analogy in their composition. We shall therefore say in general terms that in the language the link between signifier and signified is contractual in its principle, but that this contract is collective, inscribed in a long temporality (Saussure says that 'a language is always a legacy'), and that consequently it is, as it were, naturalised; in the same way, Levi-Strauss specified that the linguistic sign is arbitrary a priori but non-arbitrary a posteriori. This discussion leads us to keep two different terms, which will be useful during the semiological extension. We shall say that a system is arbitrary when its signs are founded not by convention, but by unilateral decision: the sign is not arbitrary in the language but it is in fashion; and we shall say that a sign is motivated when the relation between its signified and its signifier is analogical (Buyssens has put forward, as suitable terms, intrinsic semes for motivated signs, and extrinsic semes for unmotivated ones). It will therefore be possible to have systems which are arbitrary and motivated, and others which are non-arbitrary and unmotivated.

II.4.3. The arbitrary and the motivated in semiology: In linguistics, motivation is limited to the partial plane of derivation or composition; in semiology, on the contrary, it will put to us more general problems. On the one hand, it is possible that outside language systems may be found, in which motivation plays a great part. We shall then have to establish in what way analogy is compatible with the discontinuous character which up to now has seemed necessary to signification; and afterwards how paradigmatic series (that is, in which the terms are few and discrete) can be established when the signifiers are analogs: this will probably be the case of 'images', the semiology of which is, for these reasons, far from being established. On the other hand, it is highly probable that a semiological inventory will reveal the existence of impure systems, comprising either very loose motivations, or motivations pervaded, so to speak, with secondary non-motivations, as if, often, the sign lent itself to a kind of conflict between the motivated and the unmotivated. This is already to some extent the case of the most 'motivated' zone of language, that of onomatopoeia. Martinet has pointed out, that the onomatopoeic motivation was accompanied by a loss of the double articulation (ouch, which depends only on the second articulation, replaces the doubly articulated syntagm 'it hurts'); yet the onomatopoeia which expresses pain is not exactly the same in French (aie) and in Danish (au), for instance. This is because in fact motivation here submits, as it were, to phonological models which of course var with different languages: there is an impregnation of the analogical by the digital. Outside language, problematic systems, like the 'language' of the bees, show the same ambiguity: the honey-gathering dances have a vaguely analogical value; that at the entrance of the hive is frankly motivated (by the direction of the food), but the wriggly dance in a figure of eight is quite unmotivated (it refers to a distance).-" Finally, and as a last example of such ill-defined areas, certain trade-marks used in advertising consist of purely abstract' (non-analogical) shapes; they can, however, express' a certain impression (for instance one of 'power') which has a relation of affinity with the signified. The trade-mark of the Berliet lorries (a circle with a thick arrow across it) does not in any way 'copy' power - indeed, how could one 'copy' power? - and yet suggests it through a latent analogy; the same ambiguity is to be found in the signs of some ideographic writings (Chinese, for instance).

The coexistence of the analogical and the non-analogical therefore seems unquestionable, even within a single system. Yet semiology cannot be content with a description acknowledging this compromise without trying to systematise it, for it cannot admit a continuous differential since, as we shall see, meaning is articulation. These problems have not yet been studied in detail, and it would be impossible to give a general survey of them. The outline of an economy of signification (at the anthropological level) can, however, be perceived: in the language, for instance, the (relative) motivation introduces a certain order at the level of the first (significant) articulation : the 'contract' is therefore in this case underpinned by a certain naturalisation of this a priori arbitrariness which Lévi-Strauss talks about; other systems, on the contrary, can go from motivation to non-motivation: for instance the set of the ritual puppets of initiation of the Senoufo, cited by Lévi-Strauss in The Savage Mind. It is therefore probable that at the level of the most general semiology, which merges with anthropology, there comes into being a sort of circularity between the analogical and the unmotivated: there is a double tendency (each aspect being complementary to the other) to naturalise the unmotivated and to intellectualise the motivated (that is to say, to culturalise it). Finally, some authors are confident that digitalism, which is the rival of the analogical, is itself in its purest form - binarism - a 'reproduction' of certain physiological processes, if it is true that sight and hearing, in the last analysis, function by alternative selections.

II.5. VALUE

II.5.1. Value in linguistics: We have said, or at least hinted, that to treat the sign 'in itself', as the only link between signifier and signified, is a fairly arbitrary (although inevitable) abstraction. We must, to conclude, tackle the sign, no longer by way of its 'composition', but of its 'setting': this is the problem of value. Saussure did not see the importance of this notion at the outset, but even as early as his second Course in General Linguistics, he increasingly concentrated on it, and value became an essential concept for him, and eventually more important than that of signification (with which it is not co-extensive). Value bears a close relation to the notion of the language (as opposed to speech); its effect is to de-psychologise linguistics and to bring it closer to economics; it is therefore central to structural linguistics. In most sciences, Saussure observes, there is no coexistence of synchrony and diachrony: astronomy is a synchronic science (although the heavenly bodies alter); geology is a diachronic science (although it can study fixed states-); history is mainly diachronic (a succession of events), although it can linger over some 'pictures'. Yet there is a science in which these two aspects have an equal share: economics (which include economics proper, and economic history); the same applies to linguistics, Saussure goes on to say. This is because in both cases we are dealing with a system of equivalence between two different things: work and reward, a signifier and a signified (this is the phenomenon which we have up to now called signification). Yet, in linguistics as well as in economics, this equivalence is not isolated, for if we alter one of its terms, the whole system changes by degrees. For a sign (or an economic 'value') to exist, it must therefore be possible, on the one hand, to exchange dissimilar things (work and wage, signifier and signified), and on the other, to compare similar things with each other. One can exchange a five-franc note for bread, soap or a cinema ticket, but one can also compare this banknote with ten- or fifty-franc notes, etc.; in the same way, a 'word' can be 'exchanged' for an idea (that is, for something dissimilar), but it can also be compared with other words (that is, something similar): in English the word mutton derives its value only from its coexistence with sheep; the meaning is truly fixed only at the end of this double determination: signification and value. Value, therefore, is not signification; it comes, Saussure says, 'from the reciprocal situation of the pieces of the language'. It is even more important than signification: 'what quantity of idea or phonic matter a sign contains is of less import than what there is around it in the other signs':- a prophetic sentence, if one realises that it already was the foundation of Lévi-Strauss's homology and of the principle of taxonomies. Having thus carefully distinguished, with Saussure, signification and value, we immediately see that if we return to Hjemslev's strata (substance and form), the signification partakes of the substance of the content, and value, of that of its form (mutton and sheep are in a paradigmatic relation as signifieds and not, of course, as signifiers).

II.5.2. The articulation: In order to account for the double phenomenon of signification and value, Saussure used the analogy of a sheet of paper: if we cut out shapes in it, on the one hand we get various pieces (A, B, C), each of which has a value in relation to its neighbours, and, on the other, each of these pieces has a recto and a verso which have been cut out at the same time (A-A', B-B', C-C'): this is the signification. This comparison is useful because it leads us to an original conception of the production of meaning: no longer as the mere correlation of a signifier and a signified, but perhaps more essentially as an act of simultaneously cutting out two amorphous masses, two 'floating kingdoms' as Saussure says. For Saussure imagines that at the (entirely theoretical) origin of meaning, ideas and sounds form two floating, labile, continuous and parallel masses of substances; meaning intervenes when one cuts at the same time and at a single stroke into these two masses. The signs (thus produced) are therefore articuli; meaning is therefore an order with chaos on either side, but this order is essentially a division. The language is an intermediate object between sound and thought: it consists in uniting both while simultaneously decomposing them. And Saussure suggests a new simile: signifier and signified are like two superimposed layers, one of air, the other of water; when the atmospheric pressure changes, the layer of- water divides into waves-. in the same way, the signifier is divided into articuli. These images, of the sheet of paper as well as of the waves, enable us to emphasise a fact which is of the utmost importance for the future of semiological analysis: that language is the domain of articulations, and the meaning is above all a cutting-out of shapes. It follows that the future task of semiology is far less to establish lexicons of objects than to rediscover the articulations which men impose on reality; looking into the distant and perhaps ideal future, we might say that semiology and taxonomy, although they are not yet born, are perhaps meant to be merged into a new science, arthrology, namely, the science of apportionment.