Linguistics After Saussure

Linguistics is a term of no great antiquity. It came into fashion in the nineteenth century when scholars began to distinguish between various possible approaches to the study of language and languages. Many, including Saussure, insisted on a distinction between traditional philology, focused on the study of literary and other texts (particularly those of earlier periods) and a more general form of inquiry which sought to study languages themselves, irrespective of whether they had produced texts of literary or cultural importance, or of whether they had produced any texts at all. Linguistics eventually emerged as the preferred term for this more general form of inquiry, of which the prospectus and methods were set out in Saussure’s posthumously published Cours de linguistique générale (Course in General Linguistics). This key treatise appeared in 1916, having been compiled from the notes taken by Saussure’s students at courses of lectures that he gave at the University of Geneva in the years 1907–11.

Linguistics, as envisaged in Saussure’s Cours, was to be a ‘science’. That idea has certainly survived down to the present day. (For discussion, see Crystal 1985, pp. 76ff.; Harris 1992.) Not many academics currently holding posts in linguistics will define their subject without invoking the term science, and some departments of linguistics even incorporate that term into their title (as if officially calling their subject a science automatically made ‘scientists’ of them). But the survival of the notion that linguistics should be a science has little to do with Saussure. Science is one of the most popular and vacuous buzzwords in modern academic culture, particularly in those areas where the ‘scientific’ status of a subject is in doubt.

Progress in describing all known languages and recording their history has not been spectacular. However one decides to count the number of languages in the world (a permanently contentious issue), the majority have still not been studied in any depth. A few, on the other hand, are disproportionately well documented. These tend to be languages with the greatest number of speakers and high cultural prestige, where practical demand for teaching materials is considerable.

The search for language universals has attracted much interest but met with mixed fortunes (Greenberg 1966; Bach and Harms 1968; Payne 1994; Croft 1994). Its problematic aspects are in part related to (i), since if there are large numbers of languages which have not been adequately investigated, it is difficult to have confidence in claims to the effect that certain features are found in all languages. There are other sources of scepticism too, which concern the a priori assumptions on which claims to universality are based, together with the difficulty of deciding whether features apparently present in more than one language are really ‘the same’.

A glance at any current glossary reveals such branches as: anthropological linguistics, applied linguistics, biolinguistics, clinical linguistics, computational linguistics, critical linguistics, educational linguistics, ethnolinguistics, neurolinguistics, pragmalinguistics, psycholinguistics, sociolinguistics and many more.

In short, one might say that linguistics has done the opposite of what Saussure hoped: it has failed to define itself. Instead, what has happened is that research which has anything to do with language at all now finds itself labelled by some compound term of which the second element is linguistics. Furthermore, the boundaries between these various branches of inquiry are often unclear.

Saussure distinguished, consistently and emphatically, between langue and langage. This distinction is not altogether easy to render into English for want of a corresponding pair of words. For Saussure, langage includes both langue and parole (Saussure 1916, p. 38). The latter is the implementation in speech of the system (langue) which the language-user brings into play in any given act of linguistic communication (parole). The priority of langue as far as (Saussurean) linguistics is concerned can be boiled down to the proposition that if any episode of human speech is to be the subject of serious scientific inquiry it must be related in the first instance to a system which must be presupposed as underlying it.

A later generation of linguists rebaptized Saussure’s distinction between langue and parole as ‘linguistic competence’ versus ‘linguistic performance’ (Fromkin and Rodman 1978, pp. 6–9 et passim). For a sceptical view, see Lakoff 1973a. But this terminological innovation makes no headway in resolving the difficulties attaching to the distinction itself. Matters are not helped by the lack of consistency between different theorists (or even sometimes in the same theorist on different occasions) regarding the way the distinction is to be drawn. Exactly what belongs to parole (performance) and what to langue (competence) remains in doubt.

Linguists who realized that discourse is in any case not just a succession of sentences, but exhibits structural coherence over a far wider syntagmatic span, sidestepped the generative framework and developed what is variously known today as ‘discourse analysis’, ‘discourse linguistics’ or ‘text linguistics’ (van Dijk 1985; Beaugrande 1994). This pays particular attention to such properties as ‘cohesion’ and ‘narrativity’, examining how sequences of events and other kinds of information are reported over longer stretches of talk or writing, up to and including book-length presentation.

Quite apart from the circularity of this theoretical maneuver, and the problem of languages that go under various language-names, such a criterion automatically excludes any speaker whose membership of the linguistic community is marginal or in dispute. This would inevitably leave the linguist with a residue of speakers languishing in linguistic limbo. Does general linguistics require the assumption that everyone speaks at least one identifiable language? If so, what are the criteria for an ‘identifiable language’? If not, how does general linguistics deal with the case of speakers who somehow manage to slip through the net?

They accepted Saussure’s narrowing down of the concept of langue in a way that relates to the passage of time. Saussure drew a basic distinction between what he called synchronic and diachronic linguistics and gave priority to the former. He would doubtless be gratified today to observe that the study of linguistic change occupies a far less prominent place in the activities of linguists than it did in the nineteenth century.

Most of Saussure’s successors accepted the ‘synchronic–diachronic’distinction, which still survives robustly in early twenty-first-century linguistics. In practice, what this means is that it is accounted a violation of principle or linguistic method to include in the same synchronic analysis evidence relating to diachronically different states. So, for example, citing Shakespearean forms would be regarded as inadmissible in support of, say, an analysis of the grammar of Dickens. Saussure is particularly severe in his strictures upon linguists who conflate synchronic and diachronic facts.

This move corresponds to the lay perception that people who may be speaking, say, ‘English’– and would describe themselves as ‘English’ speakers – nevertheless are not necessarily all speaking the same English. They may differ noticeably one from another in features of pronunciation, grammar and vocabulary. How can a ‘synchronic’ linguistic description accommodate this amount of variation?

Furthermore, although both in Saussure’s day and since then the practitioners of so-called ‘dialect geography’ have tried to delimit linguistic areas by such techniques as plotting ‘isoglosses’ (lines on maps which supposedly mark the boundaries of the geographical spread of particular linguistic features) modern transport conditions and population movements increasingly make such attempts seem futile.

A quite different strategy for trying to identify the synchronic language ‘system’ as a viable object of linguistic description has been to restrict attention to the speech of a single speaker. Each individual is envisaged as having a personal variety of speech, and this personal variety is known technically as the ‘idiolect’. Thus, in the final analysis, what is nowadays spoken as ‘English’ can be broken down, according to this view, into as many different idiolects as there are speakers (i.e. millions). Whether any two such idiolects are exactly alike is a moot point: the usual assumption is that they will be found to differ in certain features, however minimally.

The notion that each language (langue) can be thought of as represented by an ‘ideal’ speaker–listener is another way of conceptualizing languages as fixed codes. For it is hard to see how an ideal speaker–listener could be ideal without an infallible judgment as to whether a proposed form of expression were correct or not. In other words, this hypothetical figure is posited ab initio as one who already ‘knows’whether such-and-such a construction is acceptable, whether suchand- such a word is admissible, whether such-and-such a pronunciation is authentic, whether a given sentence means such-and-such or not. If the ‘ideal’ speaker– listener’s views of those matters varied from one occasion to the next, that would automatically be a disqualification for the theoretical role that such a personage is called upon to perform. It is important here to note the difference (often blurred) between ideality and typicality. A typical speaker–listener is not – and could hardly be – an ‘ideal’ speaker–listener in the sense theoretically required, although these notions are commonly conflated.

The postulation of an ‘ideal speaker–listener’ for every language tended in the 1960s and 1970s to go along with acceptance of a dogmatic distinction between the ‘surface structure’ of a language and its ‘deep structure’. Thus, for example, The dog bit the postman and The postman was bitten by the dog, although different on the ‘surface’, were seen as ‘deeply’ identical (i.e. in some – allegedly intuitive but not very clearly explicated – sense, as one and the same sentence). This is a kind of distinction Saussure never drew, and his ‘failure’ to draw it was seen by generativists as a basic inadequacy in Saussurean linguistics.

This was the outcome of Louis Hjelmslev’s interpretation of Saussure, which laid the foundation for the Danish linguistic school of glossematics. Glossematicians took the Saussurean dictum that langue is ‘form not substance’ (Saussure 1916, p. 163) to its logical conclusion, and argued that the languages now in existence and available to observation are merely historical realizations of certain systems which could equally well exist in other manifestations (not necessarily spoken and not necessarily written either). Thus ‘the task of the linguistic theoretician is not merely that of describing the actually present expression system, but of calculating what expression systems in general are possible as expression for a given content system, and vice versa’ (Hjelmslev 1961, p. 105).

Both glossematics and generativism, although claiming to improve on Saussure, are manifestly at odds with Saussure’s stipulation that the linguist, in describing la langue in any given case, should set up no more and no less categories and distinctions than those already recognized ‘consciously or unconsciously’ by its speakers (Saussure 1916, p. 195). This was the first formulation of the goal of ‘psychological reality’ in descriptive linguistics. It has proved to be a nightmare for linguists ever since.

The conclusion drawn was that linguistics could not deal with the ‘meanings’ of words except, to quote one leading behaviourist of the 1930s, in cases involving ‘some matter of which we possess scientific knowledge’ (Bloomfield 1935, p. 139). So whereas it was all right to define the meaning of the English word salt as ‘sodium chloride (NaCl)’, since science told linguists that that was what the substance called salt ‘really’ was, no such information was available for defining words like love and hate, not to mention all the other aspects of ‘the real world’ that science had not yet investigated.

The problem of ‘psychological reality’ in linguistics tended to merge with two related issues, both of them controversial in their own right. If linguistics was to be a science, it was argued, linguists must develop a methodology comparable to those of the natural sciences. Thus bias and introspection by the linguist must be banned and only objective, verifiable methods employed. This led to much argument about which so-called ‘discovery procedures’ the linguist should in practice adopt when facing a corpus of evidence to analysis. In the second place, there arose doubts not merely about whether the goal of ‘psychological reality’ was realistically attainable, but about whether linguistic descriptions described anything ‘real’ at all. Those who believed in the objective existence of linguistic structure were called ‘God’s truth’ linguists. Those who believed, on the contrary, that linguistic structure was an artifact of the linguist’s analytic methods were known as ‘hocus-pocus’ linguists.

In Saussure’s programme for linguistics there is no provision for the study of the actual contexts in which speakers communicate to one another. In other words, the assumption is that a language system (langue) remains invariant across all contexts.

It makes no difference who the actual speakers are or in what circumstances they are speaking. This somewhat implausible assumption is today championed only by partisans of what is now called ‘autonomous linguistics’, to which a majority of surviving generativists belong (Newmeyer 1994). Others, however, have realized that it makes little sense (and serves no purpose) to insist on treating languages as self-contained mental systems which bear no relation at all – except externally and fortuitously – to the lives of their speakers and the communicational purposes to which they are constantly being put.

A less radical position is that of ‘functionalists’ (Dik 1994; Martinet 1994). Functionalism is a theoretical hat with very wide brims, broad enough to shelter all those who see linguistic structure as being molded in response to communicational demands and to biomechanical factors. Thus, for instance, certain features of phonological systems would be explained by reference to properties of the human vocal tract and the communicational need to clarify auditory distinctions. The term ‘functional’ is in particular associated with the so-called ‘Prague School’, a group of neo- Saussurean linguists who in 1926 founded the Linguistic Circle of Prague and

included among their members Vilem Mathesius, Roman Jakobson and Nikolai Trubetzkoy (Vachek 1964; Fried 1972). ‘Pragmatics’ is the term that has now become general for a broad variety of studies – whether overtly functionalist or not – that place emphasis on the need to study language in relation to the actual circumstances of its use (Mey 1994). Some pragmaticians even refer to the ‘pragmatic competence’, as distinct from the ‘linguistic competence’, of speakers; but exactly how pragmatic competence can be defined, other than by reference to the specifics of particular communication situations, it is difficult to see.

Finally, Saussure’s programme for linguistics was based on a fundamental assumption about the relationship between speech and writing. In identifying la langue as the system which was the object of investigation in linguistics, Saussure made it quite clear that writing was not part of it (Saussure 1916, p. 46). In effect, he equated language with spoken language. The majority of academic linguists throughout the twentieth century followed this lead. Statements to the effect that ‘writing is not language’ are commonplace.

Current reactions by linguists to the problem of writing fall into three general classes. (1) Say nothing about it. This seems to be characteristic of most generativists, who have no discernible theory of writing at all. (2) Treat speech and writing as separate autonomous systems. This is typically the position adopted by glossematicians (Uldall 1944) and Prague-school theorists (Barnet 1972). (3) Treat speech and writing as integrated systems of communication in all literate societies. This is the position adopted by integrationists (Harris 1995).

To sum up, linguistics after Saussure expanded and diversified in ways that Saussure shows no sign of anticipating. But in so doing it lost any theoretical consensus or coherence as regards the objectives of linguistic inquiry or the methods to be pursued in attaining them. As one contemporary linguist has put it: ‘If asked point blank what the object of their science is, I assume that few professional linguists would hesitate to answer that it is “language”. But if asked what they mean by “language” serious divergences would soon appear’ (Martinet 1984). That observation is itself a comment on the extent to which linguistics has proved unable to resolve the problems that were part and parcel of the Saussurean legacy. The opening chapter of the Cours concludes with the statement that ‘the fundamental problems of general linguistics still await a solution’. It is ironic how apposite that statement still is today.

Read: The Saussurean Dichotomies

Cobley, Paul. (2001). The routledge companion to semioticsand linguistics. London and New York: Routledge.