[155]
§1 The language as thought organised in sound
In order to realise that the language itself can be nothing other than a system of pure values, one need only consider the two elements which are involved in the way it functions: ideas and sounds.
Psychologically, setting aside its expression in words, our thought is simply a vague, shapeless mass. Philosophers and linguists have always agreed that were it not for signs, we should be incapable of differentiating any two ideas in a clear and constant way. In itself, thought is like a swirling cloud, where no shape is intrinsically determinate. No ideas are established in advance, and nothing is distinct, before the introduction of linguistic structure.
But do sounds, which lie outside this nebulous world of thought, in themselves constitute entities established in advance? No more than ideas do. The substance of sound is no more fixed or rigid than that of thought. It does not offer a ready-made mould, with shapes that thought must inevitably conform to. It is a malleable material which can be fashioned into separate parts in order to supply the signals which thought has need of. So we can envisage the linguistic phenomenon in its entirety – the language, that is – as a series of adjoining subdivisions simultaneously imprinted both on the plane of vague, amorphous thought (A), and on the equally featureless plane of sound (B). This can be represented very approximately as in the following sketch (top of p. 111).
The characteristic role of a language in relation to thought is not to supply the material phonetic means by which ideas may be expressed. It is to act as intermediary between thought and sound, in such a way that the combination of both necessarily produces a mutually complementary delimitation of units. Thought, chaotic by nature, is made precise by this process of segmentation. But what happens is neither a transformation of thoughts into matter, nor a transformation of sounds into ideas. What takes place, is a somewhat mysterious process by which ‘thought-sound’ evolves divisions, and a language takes shape with its linguistic units in between those two amorphous masses. One might think of it as being like air in contact with water: changes in atmospheric pressure break up the surface of the water into series of divisions, i.e. waves. The correlation between thought and sound, and the union of the two, is like that.
Linguistic structure might be described as the domain of articulations, taking this term in the sense defined earlier (p. [26]). Every linguistic sign is a part or member, an articulus, where an idea is fixed in a sound, and a sound becomes the sign of an idea.
A language might also be compared to a sheet of paper. Thought is one side of the sheet and sound the reverse side. Just as it is impossible to take a pair of scissors and cut one side of paper without at the same time cutting the other, so it is impossible in a language to isolate sound from thought, or thought from sound. To separate the two for theoretical purposes takes us into either pure psychology or pure phonetics, not linguistics.
Linguistics, then, operates along this margin, where sound and thought meet. The contact between them gives rise to a form, not a substance.
These observations clarify our earlier remarks about the arbitrary nature of the linguistic sign (p. [100]). Not only are the two areas which are linguistically linked vague and amorphous in themselves, but the process which selects one particular sound-sequence to correspond to one particular idea is entirely arbitrary. If this were not so, the notion of value would lose something. For it would involve a certain element of imposition from the outside world. But in fact values remain entirely a matter of internal relations, and that is why the link between idea and sound is intrinsically arbitrary.
In turn, the arbitrary nature of the sign enables us to understand more easily why it needs social activity to create a linguistic system. A community is necessary in order to establish values. Values have no other rationale than usage and general agreement. An individual acting alone, is incapable of establishing a value.
Furthermore, the notion of value, thus defined, shows us that it is a great mistake to consider a sign as nothing more than the combination of a certain sound and a certain concept. To think of a sign as nothing more would be to isolate it from the system to which it belongs. It would be to suppose that a start could be made with individual signs, and a system constructed by putting them together. On the contrary, the system as a united whole is the starting point, from which it becomes possible, by a process of analysis, to identify its constituent elements.
[158]
To develop this idea, we shall look at it first from the point of view of the signification or concept (§2), then from that of the signal (§3), and finally from that of the sign as a whole (§4).
Since we cannot have direct access to concrete entities and linguistic units, we shall take words as examples. Although, as previously noted (p. [147]), words do not answer exactly to our definition of linguistic units, they will be adequate to give a rough idea, and will obviate the necessity for talking in abstract terms. So we will treat them for present purposes as specimens supposedly equivalent to the actual signs of a synchronic system. The principles which will emerge may be taken as valid for linguistic entities in general.
§2 Linguistic value: Conceptual aspects
The value of a word is mainly or primarily thought of in terms of its capacity for representing a certain idea. That is indeed an aspect of linguistic value. But in that case, does its linguistic value differ from what is called its meaning? Are value and meaning synonymous terms? Not in our view, although it is easy to confuse them. For the subtlety of the distinction, rather than any analogy between the two terms, invites confusion.
Value, in its conceptual aspect, is doubtless part of meaning. It is by no means easy, indeed, to draw the distinction in view of this interconnexion. Yet it must be drawn, if a language is not to be reduced to a mere nomenclature (cf. p. [97]).
Let us first consider meaning, as usually understood, in the light of our previous analysis (p. [99]). As the arrows in the diagram indicate, a meaning is simply the counterpart of a sound pattern. The relevant relation is one between a sound pattern and a concept, within the limits of the word, which is for this purpose treated as a self-contained unit, existing independently.
[159]
The paradoxical part of it is this. On the one hand, the concept appears to be just the counterpart of a sound pattern, as one constituent part of a linguistic sign. On the other hand, this linguistic sign itself, as the link uniting the two constituent elements, likewise has counterparts. These are the other signs in the language.
A language is a system in which all the elements fit together, and in which the value of any one element depends on the simultaneous coexistence of all the others. It may be represented as follows.
So how does it come about that value, as defined, can be equated with meaning, i.e. with the counterpart of the sound pattern? For it looks impossible to assimilate the relations represented here by horizontal arrows to those other relations represented in the previous diagram by vertical arrows. In other words, to go back to our comparison with the sheet of paper (p. [157]), it is difficult to see how the relation between different shapes cut out (call them A, B, C, D, etc.) can fail to be different from the relation between one side of any given shape and its reverse side (A/A´, B/B´, etc.).
In answering this question, it is relevant to point out that even in non-linguistic cases values of any kind seem to be governed by a paradoxical principle. Values always involve:
(1) something dissimilar which can be exchanged for the item whose value is under consideration, and
(2) similar things which can be compared with the item whose value is under consideration.
These two features are necessary for the existence of any value. To determine the value of a five-franc coin, for instance, what must be known is: (1) that the coin can be exchanged for a certain quantity of something different, e.g. bread, and (2) that its value can be compared with another value in the same system, e.g. that of a one-franc coin, or of a coin belonging to another system (e.g. a dollar). Similarly, a word can be substituted for something dissimilar: an idea. At the same time, it can be compared to something of like nature: another word. Its value is therefore not determined merely by that concept or meaning for which it is a token. It must also be assessed against comparable values, by contrast with other words. The content of a word is determined in the final analysis not by what it contains but by what exists outside it. As an element in a system, the word has not only a meaning but also – above all – a value. And that is something quite different.
[160]
A few examples will show that this is indeed the case. The French word mouton may have the same meaning as the English word sheep; but it does not have the same value. There are various reasons for this, but in particular the fact that the English word for the meat of this animal, as prepared and served for a meal, is not sheep but mutton. The difference in value between sheep and mouton hinges on the fact that in English there is also another word mutton for the meat, whereas mouton in French covers both.
In a given language, all the words which express neighbouring ideas help define one another’s meaning. Each of a set of synonyms like redouter (‘to dread’), craindre (‘to fear’), avoir peur (‘to be afraid’) has its particular value only because they stand in contrast with one another. If redouter (‘to dread’) did not exist, its content would be shared out among its competitors. On the other hand, words are also enriched by contact with other words. For instance, the new element introduced into décrépit (as in un vieillard décrépit, cf. p. [119]) is a result of the coexistence of décrépi (as in un mur décrépi). So the value of any given word is determined by what other words there are in that particular area of the vocabulary. That is true even of a word like soleil (‘sun’). No word has a value that can be identified independently of what else there is in its vicinity. There are languages, for example, in which it is impossible to say the equivalent of s’asseoir au soleil (‘to sit in the sun’).
[161]
The above remarks apply not only to words but to all linguistic elements, including grammatical entities. The value of a French plural, for instance, does not match that of a Sanskrit plural, even though they often mean the same. This is because in Sanskrit, in addition to singular and plural, there is a third category of grammatical number. In Sanskrit the equivalents of expressions like mes yeux (‘my eyes’), mes oreilles (‘my ears’), mes bras (‘my arms’), mes jambes (‘my legs’) would be neither in the singular nor in the plural but in the dual. It would thus be inaccurate to attribute the same value to the Sanskrit plural as to the French plural, because Sanskrit cannot use the plural in all the cases where it has to be used in French. Its value thus does indeed depend on what else there is in its vicinity.
If words had the job of representing concepts fixed in advance, one would be able to find exact equivalents for them as between one language and another. But this is not the case. French uses the same verb louer (‘hire, rent’) both for granting and for taking a lease, whereas German has two separate verbs, mieten and vermieten: so there is no exact correspondence between the values in question. The German verbs schätzen (‘to value’) and urteilen (‘to judge’) have meanings which answer roughly to those of the French verbs estimer and juger: but in various respects there is no one-to-one correspondence.
Flexion offers some particularly striking examples. The distinctions of tense which are so familiar to us are unknown in certain languages. The Hebrew verb does not even mark the fundamental difference between past, present and future. Proto-Germanic has no separate verb form for the future: it is sometimes said that it uses the present tense for this purpose, but that is misleading because the value of a present tense is not the same in Germanic as in those languages which have future tense forms in addition to present tense forms. The Slavic languages regularly distinguish two verbal aspects: the perfective aspect represents an action as a whole, as a single point, taking no development into account, whereas the imperfective aspect represents the same action in the process of development, taking place in time. These categories are difficult for a Frenchman, because his language does not recognise them. If they were predetermined categories, there would be no such difficulty. In all these cases what we find, instead of ideas given in advance, are values emanating from a linguistic system. If we say that these values correspond to certain concepts, it must be understood that the concepts in question are purely differential. That is to say they are concepts defined not positively, in terms of their content, but negatively by contrast with other items in the same system. What characterises each most exactly is being whatever the others are not.
[162]
The full significance of our diagram representing the linguistic sign should now be apparent. This means that in French, the concept ‘juger’ (‘to judge’) is linked to the sound pattern juger. So the diagram represents what the word means. But it must not be supposed that the concept in question has any kind of priority. On the contrary, that particular concept is simply a value which emerges from relations with other values of a similar kind. If those other values disappeared, this meaning too would vanish. If I say, simply, that a certain word means this or that – going no further than identifying the concept associated with that particular sound pattern – then what I am saying may in some respects be accurate, and succeed in giving a correct picture. But I fail inevitably to capture the real linguistic fact, either in its basic essentials or in its full scope.
[163]
§3 Linguistic value: Material aspects
Just as the conceptual part of linguistic value is determined solely by relations and differences with other signs in the language, so the same is true of its material part. The sound of a word is not in itself important, but the phonetic contrasts which allow us to distinguish that word from any other. That is what carries the meaning.
This may seem surprising. But how could it possibly be otherwise? No particular configuration of sound is more aptly suited to express a given message than any other such configuration. So it is clearly the case – indeed, it must be the case – that no linguistic item can ever be based, ultimately, upon anything other than its non-coincidence with the rest. Here the terms arbitrary and differential designate two correlative properties.
The processes of linguistic change amply demonstrate this correlation. It is precisely because two signs a and b are never grasped as such by our linguistic consciousness, but only the difference between a and b, that each sign remains free to change in accordance with laws quite unconnected with their signifying function. The Czech genitive plural žen (cf. p. [123]) has no positive case maker. Yet the contrast žena vs. žen works just as well as žena vs. ženЬ, which preceded it. The reason is that all that matters is the difference between the signs: žena functions effectively simply because it is different.
Another example which brings out even more clearly the systematic nature of such contrasts is the following. In Greek éphēn is an imperfect and éstēn an aorist, even though their morphological formation is identical. The former belongs to the present indicative system of phēmi (‘I say’), whereas there is no present form *stēmi. It is the relation between phēmi and éphēn which corresponds to the relation between present and imperfect (cf. deíknūmi – edeíknūn). These signs thus function not according to their intrinsic value but in virtue of their relative position.
[164]
In any case, it is impossible that sound, as a material element, should in itself be part of the language. Sound is merely something ancillary, a material the language uses. All conventional values have the characteristic of being distinct from the tangible element which serves as their vehicle. It is not the metal in a coin which determines its value. A crown piece nominally worth five francs contains only half that sum in silver. Its value varies somewhat according to the effigy it bears. It is worth rather more or rather less on different sides of a political frontier. Considerations of the same order are even more pertinent to linguistic signals. Linguistic signals are not in essence phonetic. They are not physical in any way. They are constituted solely by differences which distinguish one such sound pattern from another.
This fundamental principle applies to every material element used by a language, even the basic speech sounds. Each language constructs its words out of some fixed number of phonetic units, each one clearly distinct from the others. What characterises those units is not, as might be thought, the specific positive properties of each; but simply the fact that they cannot be mistaken for one another. Speech sounds are first and foremost entities which are contrastive, relative and negative.1
What proves this is the latitude speakers are allowed in pronunciation, provided they distinguish one sound from another. In French, for instance, the fact that r is usually pronounced as a uvular consonant does not prevent many speakers from pronouncing it as an apical trill. It makes no difference to the French language, which requires only that r should be distinct from other consonants. There is no necessity that it be pronounced always in exactly the same way. I can even pronounce a French r like the German ch in Bach, doch, etc.; whereas I could not in German substitute r for ch because German, unlike French, distinguishes between r and ch. Likewise in Russian, there is no latitude of pronounciation for t in the direction of t’ (i.e. palatalised t), because the result would be to confuse two sounds distinguished by the language (cf. govorit’, ‘to speak’ vs. govorit ‘he speaks’). But a Russian is more at liberty to aspirate a t, because th is not a separate sound in the Russian system.
[165]
An identical state of affairs is to be found in that other system of signs, writing. Writing offers a useful comparison, which throws light upon the whole question. We find that:
1. The signs used in writing are arbitrary. The letter t, for instance, has no connexion with the sound it denotes.
2. The values of the letters are purely negative and differential. So the same individual may write t in such variant forms as:
The one essential thing is that his t should be distinct from his l, his d, etc.
3. Values in writing are solely based on contrasts within a fixed system, having a determinate number of letters. This feature, although not the same as 2 above, is closely connected with it; for both 2 and 3 follow from 1. Since the written sign is arbitrary, its form is of little importance; or rather, is of importance only within certain limits imposed by the system.
[166]
4. The actual mode of inscription is irrelevant, because it does not affect the system. (This also follows from 1.) Whether I write in black or white, in incised characters or in relief, with a pen or a chisel – none of that is of any importance for the meaning.
Everything we have said so far comes down to this. In the language itself, there are only differences. Even more important than that is the fact that, although in general a difference presupposes positive terms between which the difference holds, in a language there are only differences, and no positive terms. Whether we take the signification or the signal, the language includes neither ideas nor sounds existing prior to the linguistic system, but only conceptual and phonetic differences arising out of that system. In a sign, what matters more than any idea or sound associated with it is what other signs surround it. The proof of this lies in the fact that the value of a sign may change without affecting either meaning or sound, simply because some neighbouring sign has undergone a change (cf. p. [160]).
But to say that in a language everything is negative holds only for signification and signal considered separately. The moment we consider the sign as a whole, we encounter something which is positive in its own domain. A linguistic system is a series of phonetic differences matched with a series of conceptual differences. But this matching of a certain number of auditory signals and a similar number of items carved out from the mass of thought gives rise to a system of values. It is this system which provides the operative bond between phonic and mental elements within each sign. Although signification and signal are each, in isolation, purely differential and negative, their combination is a fact of a positive nature. It is, indeed, the only order of facts linguistic structure comprises. For the essential function of a language as an institution is precisely to maintain these series of differences in parallel.
[167]
Certain diachronic developments are most revealing in this respect. We find countless cases where a change in the signal brings with it a change in the idea expressed. Time and again we observe that in principle the number of ideas distinguished matches the number of distinct signals available. When two words merge through phonetic change (e.g. décrépit from Latin decrepitus, and décrépi from Latin crispus, cf. p. [119]), the ideas tend to merge as well, however dissimilar they may be. What happens when one word gives birth to two alternative pronunciations (e.g. French chaise and chaire, both from Latin cathedra)? Inevitably the phonetic difference which has emerged will tend to acquire significance, although perhaps not always immediately or always successfully. Conversely, any difference in ideas distinguished by the mind will seek expression in different linguistic signals; whereas two ideas the mind no longer differentiates will tend to find expression in the same signal.
The moment we compare one sign with another as positive combinations, the term difference should be dropped. It is no longer appropriate. It is a term which is suitable only for comparisons between sound patterns (e.g. père vs. mère), or between ideas (e.g. ‘father’ vs. ‘mother’). Two signs, each comprising a signification and a signal, are not different from each other, but only distinct. They are simply in opposition to each other. The entire mechanism of language1, which we shall consider below, is based on oppositions of this kind and upon the phonetic and conceptual differences they involve.
What is true of values is also true of units (cf. p. [154]). A unit is a segment of a spoken sequence which corresponds to a certain concept. Both are purely differential in nature.
Applied to units, the principle of differentiation may be formulated as follows. The characteristics of the unit merge with the unit itself. In a language, as in every other semiological system, what distinguishes a sign is what constitutes it, nothing more. Difference is what makes characteristics, just as it makes values and units.
[168]
Another consequence, and a rather surprising one, of the same principle is this. What is usually called a ‘grammatical fact’ corresponds in the final analysis to our definition of a unit. For there is always an opposition of terms involved. What is special is that the opposition happens to be particularly important, e.g. German plural formations of the type Nacht vs. Nächte (cf. p. [120] ff.). Each of the items which contrast grammatically (the singular form without the umlaut and without the final -e, contrasting with a plural form having both) is itself the product of the operation of oppositions within the system. In isolation, Nacht and Nächte are nothing: the opposition between them is everything. In other words, one might express the relation Nacht vs. Nächte by an algebraic formula a/b, where a and b are not simple terms, but each represents a complex of relations. The language is, so to speak, an algebra which has only complex terms. Some of the oppositions it includes are more important than others. But ‘units’ and ‘grammatical facts’ are only different names for different aspects of the same general fact: the operation of linguistic oppositions. So much so that it would be perfectly possible to tackle the problem of units by beginning with grammatical facts. Starting from an opposition like Nacht vs. Nächte, one would inquire what are the units involved. Are they just two words? Or are they whole series of similar words? Or are they just a and ä? Or are they all singulars and all plurals?
[169]
Units and grammatical facts would not merge in this way if a linguistic sign was constituted by anything apart from differences.But linguistic structure being what it is, however one approaches it, nothing is simple. Always and everywhere one finds this same complex equilibrium of terms holding one another in mutual juxtaposition. In other words, the language itself is a form, not a substance (cf. p. [157]). The importance of this truth cannot be overemphasised. For all our mistakes of terminology, all our incorrect ways of designating things belonging to the language originate in our unwittingly supposing that we are dealing with a substance when we deal with linguistic phenomena.
Notes
1 When this passage is compared with the detailed account of speech sounds given earlier (p. [63] ff.), it is evident that the published text of the Cours lacks any careful and consistently drawn distinction between phonetic and phonological units. The speech sounds discussed on p. [63] ff. are clearly language-neutral elements, characterised in physiological terms whereas the speech sounds discussed here are defined contrastively in the context of particular languages. Cf. p. [180] fn. (Translator’s note.)
1 The term used in the text here is langage, but the chapter referred to is entitled ‘Mécanisme de la langue’. (Translator’s note)