Sociolinguistics or sociology of language

1 Introduction

After half a century of structuralism dominating linguistics,^[1] we are delighted to see that at last linguists, sociologists and historians are revisiting the issue of the articulation between language, discourse and social organization and between language and ideology. It was not until about twenty years ago that researchers claiming to be Marxists returned to the Marr-Stalin polemic, e.g. Marcellesi and Gardin devoted a chapter to it.

Among the recent publications, we have selected two books which attempt to take stock of this issue. The publication by Robin (1973) particularly addresses historians, and the problems they face in the analysis of documents. The book by Marcellesi and Gardin (1974) is a more eclectic presentation of the field of current sociolinguistics. In both works, we find a certain amount of research that we will not return to here, especially since it is not always relevant to the issue at hand, which is particularly the case in the book by Marcellesi and Gardin. We will focus our reading on the passages in the books where the authors explain their positions.

The general question is the same, namely, what is the place of language and discursive productions in relation to ideological superstructures? While both authors refer to Marxism, some differences between the two publications should be noted. First of all, they do not refer to the same texts: in Robin’s text, explicit reference is made to Foucault’s work and to his research on enunciation; however, she does not refer to more general debates, such as the Sapir-Whorf hypothesis, the Marr-Stalin polemic, the Labov-Bernstein opposition, or the work by Volochinov, which, for Marcellesi and Gardin, are at the basis of the problems of sociolinguistics.

Yet, the essential difference lies in the fact that Robin relates to an Althusserian epistemological framework, in which Marcellesi and Gardin do not position themselves. The theoretical aim for Robin is the constitution of a scientific object that is articulated at the intersection of two fields of knowledge: The field of history, i.e. the science of social formations, permeated by class struggle, and that of linguistics, the science of language. This aim is coated with the concept of discursive practice, as it can be defined from Althusser’s text (1970: 3–38) on the Ideological State Apparatus and its further developments, notably by Pêcheux (cf. 1969 and 1975). Without going into all these developments in detail, we would like to highlight how the fundamental concepts are linked together to constitute this theory of articulation for Robin.

A social formation, at a given moment in its history, is constituted by the interweaving of various modes of economic production, which are organized according to the dominant mode of production. It is therefore the material organization, i.e. the economic relations of exploitation, which fundamentally, or ultimately, structures social formation. Nevertheless, the dominance of a specific mode of production over a social formation is not reduced to economic relations, as it passes through other forms of domination that define the whole ensemble of social relations. For instance, the State is the fundamental form of domination and it is the expression of the ruling class (Robin 1973: 98), but it cannot be taken as a direct expression of the dominant mode of production. On the contrary, the State is always “out of sync”, with economic domination. This gap is constituted by what Althusser called the Ideological State Apparatus (ISA), which contains the religious ISA, the juridical ISA, educational ISA, cultural ISA, etc. This theoretical concept thus covers diversified sets of material practices: gestural, ritual and discursive practices, which manifest ideological representations, i.e. the way in which social actors, who are assigned a social position, experience their relationship to their conditions of existence. As such, ideological representations can be described as material objects whose combinations constitute ideological formations. These ideological formations include, among others, discursive formations. This is the basis for a theory of articulation. A general theory of ideology opens up the possibility of describing discursive formations and explaining discursive practices. Discursive formations are only one component of the ideological formations whose theory constitutes, strictly speaking, the articulation between the field of history and that of language.

Thus, in theory at least, linguistics and sociology, or history, cannot, in Robin’s perspective, be simply juxtaposed. In this sense, it seems to provide a more satisfactory theoretical framework than the one traditionally proposed by sociolinguistics, i.e. a juxtaposition of the two domains linguistics and sociology, and a simple parallelization of the results obtained by the two disciplines’ own methods.

2 Sociolinguistics and the notion of covariance

Cohen, the initiator of sociolinguistic research in France, distinguishes, from 1956 onwards, an external approach – i.e. the study of bilingualism, dialectology and linguistic colonization – from an internal approach (Cohen 1956; 1971). A truly historical analysis would undoubtedly not allow such a distinction to be maintained, as we shall see below. As for the internal approach, Cohen describes it as follows:

Is it possible to establish structural relationships between certain social facts and certain linguistic facts (…). Language, with its fundamentally unified general functioning, varies indefinitely in accordance with social divisions; each social group has its own individuality

(Cohen (1971: 15; 59), own translation MZ and MF).^[2]

According to Fishman (1971: 219), “the sociology of language – descriptive sociology of language – seeks to answer the question ‘who speaks (or writes) what language (or what language variety) to whom and when and to what end?’” Fishman’s definition of the field is similar to that proposed by Cohen, who states that it is a matter of relating linguistic “facts” to social “facts” and of setting up a multidisciplinary enterprise.

What we emphasize in these different works is the identity of the approach, which we could summarize as follows: the study of the covariance between elements from two fields that are considered distinct from each other. This is also the approach adopted by Marcellesi and Gardin, already formulated by Marcellesi in Langue Française No. 9 (1971: 19, own translation MZ and MF): “The aim of sociolinguistics is to highlight the systematic character of the covariance of linguistic and social structures and eventually to establish a cause-effect correlation.”

3 The missing questions

It is undoubtedly the absence of a certain number of questions concerning language and discourse that explains why Robin, like Marcellesi and Gardin, despite their divergences, maintain the division between the linguistic and the social. These questions, however, will allow us to show how language can be conceived as a constitutive element of social formation rather than only as material with traces of social organization that exists elsewhere than in language.

4 Language activity

The absence of these questions is particularly striking in Marcellesi and Gardin’s conclusion, where they put forward several notions, e.g. language, linguistic activity, linguistic consciousness, and world. We will try to analyze the links between these notions.

Linguistic activity appears to be understood by the authors as an activity of speakers, made possible by the human linguistic faculty; a social activity and a producer of discourse, oral or written, individual or collective (in this case, they speak of “collective speakers”). These products of linguistic activity are never homogeneous: “mixtures of languages but also of levels and varieties of languages” (Marcellesi and Gardin 1974: 249, own translation MZ and MF).

Since the authors’ aim is to “sketch out a theory of linguistic activity and its reflection” (Marcellesi and Gardin 1974: 241, own translation MZ and MF), we investigated the text focusing on the following question: what is the reflection of linguistic activity?

1. “Linguistic awareness is the reflection and knowledge – necessarily approximate, partial and progressive – of the functioning of linguistic activity and its versatile character” (Marcellesi and Gardin 1974: 249, own translation MZ and MF).

2. Grammar, always partially adequate, reflects linguistic activity.

3. Language is the reflection of linguistic activity.

4. The reflection of linguistic activity can be seen as “knowledge, more or less ideologised and empirical, or scientific” (Marcellesi and Gardin 1974: 251, own translation MZ and MF).

If we follow these points, linguistic activity is reflected both in language, grammar, and linguistic awareness. It seems to us that the authors are confusing several analytical levels here. Moreover, by whom is this more or less ideologised knowledge produced, which they see as the reflection of linguistic activity? Can we legitimately equate the knowledge that speakers have of their own language with the knowledge that linguists and grammarians develop? Certainly, from a strictly descriptive point of view, we could confuse them: this knowledge is the product of a reflection on language in terms of language. It is an activity common to all speakers, inherent to the faculty of language; one can only speak about language while using language, unlike other semiotic systems. This work is constantly present in language practices. In this sense, speakers develop a permanent grammatical activity in their practice.^[5] However, the analogy must end there. In terms of social implications, these two practices must be carefully distinguished. The grammarian and the linguist are professionals (and get paid); they produce work and hold their discourses on language from a dominant social position. The normative activity of grammarians, like the descriptive activity of linguists, has socially a legislating function. The (salaried) reflection on language can even, in certain historical situations, be explicitly linked to the constitution of an economic and legal order. This was the case with the Jacobin language policy of the French Revolution, as we have seen above (cf. Balibar and Laporte 1974).

Our point is that the linguistic knowledge generated by linguistics is never without (obvious or non-obvious) social implications. In this sense, we reject Marcellesi and Gardin’s division into ideological and scientific knowledge.

Today, we can see these implications of how linguistics is applied in schools: development of new, functional, structural, generative grammars, reports of the commission for the reform of French teaching, etc. Each time, reference is made to the achievements of modern linguistics, from Saussure to Chomsky, to support not only a pedagogy but also a language policy.

While these social implications might be less obvious in scientific linguistic texts, we should not conclude that they do not exist. On the contrary, we should find the means to bring them to light through critical analysis. We would like to illustrate this point with a few examples, but we are aware that only an in-depth analysis of the history of linguistics could be of demonstrative value. Structural linguistics broke away from the normativity of grammarians. By substituting description for prescription, linguists had the plan to construct a scientific object and thus to escape ideology. This construction, however, could only be made by establishing the language/speech dichotomy and by making language (or the competence of an ideal speaker-listener for the Chomskyans) the exclusive object of linguistics. Reduced to being only a means of access to the language system, speech data had only an instrumental place, i.e. that of a language sample. At worst, speech facts have taken the place previously occupied by examples in traditional grammars; at best, they constitute the corpus of linguists. In so doing, the language/speech distinction has foreclosed any possibility to question the reality of language practices in linguistics. The formal apparatus of dominant linguistics has been built at this price, which is one thing for some linguists to accept, but it is quite another for them not to see the ideological consequences. Language practices, the “dark continent”^[6] of linguists, a domain of complexity, even of subjectivity, could not be thought of as a domain of analysis in its own right. The consequences of this exclusion go beyond linguistics; the almost non-existence, up to now, of work on the “languages of people” has serious consequences, particularly in the field of education. School is supposed to “teach children to speak”, but functions in total ignorance of the language practices within. How do teachers speak? How do children speak when they enter school, when they are there? One could blame the institution, but one could also blame the linguists who were not interested in the speakers’ language production, and hence objectively reinforced the institution’s shortcomings and its social consequences.

Another example is the attitude of structural linguists towards grammarians. The superb contempt of these linguists for grammarians, their clear-cut assertions that language has never bent to the prescriptions of grammarians, are based on “scientific” arguments, i.e. on laws of the internal economy of language or autonomous development. This is obviously not wrong, and there is no question of going backwards and declaring that the French language was shaped from scratch by normative grammarians. But by systematically ignoring the work of grammarians in the name of formal laws, linguists have delayed the emergence of questions about the function of this production. As shown by Balibar and Laporte (1974) and to a certain extent Chevalier (1968), normative grammars are part of the process of political, economic and ideological empowerment of the rising bourgeoisie. As such, they have had multiple social effects, including on the functioning of the language itself, which current research, focused on formalisation, completely obscures.

These few remarks cannot constitute a coherent critique of the scientific illusion in linguistics. Nevertheless, when setting the task of describing language practices, one cannot avoid questioning the fundamental concepts of dominant linguistics and criticizing its historical development. This criticism, absent in Robin’s work, does not seem to us to be much more present in Marcellesi and Gardin’s abstract definition of linguistic activity.

5 For a sociology of language: language practices

We have tried to bring together the approaches of these two publications, while being aware of the reductive and biased nature of this undertaking. We have consciously oriented our reading around the following question: in what terms should we frame the field of the sociology of language? We propose this term – rather than the term sociolinguistics, which usually refers to current work that focuses on covariance, as mentioned earlier – to mark a decentering of the issue, i.e. mainly to avoid the theory of reflection. Whether it is language or linguistic practices that are conceived as a “reflection of the world”, what is at stake in this notion is that language is not considered as a constitutive factor of a given society and, as such, dependent on the organization of social formation. R. Balibar and D. Laporte rightly state the following: “… ‘French’, which is at the same time a reason for and an object of standardization, is not a material that is always-already there and ready to function, but (…) on the contrary, it constitutes an enormous set of practices whose construction is historically a result of class struggle” (Balibar and Laporte 1974: 103, own translation MZ and MF). We do not want to play with words: we cannot simply take a term, already heavy with history and meaning, to mean something else. By sociology of language, we imply that the two fields should no longer be separated, but that language should be seen as a constitutive element – both a means and an agent – of a social formation.^[7]

It should be noted that the term sociology of language is also used by J. P. Faye, and not, as with Cohen or Fishman, as a synonym for sociolinguistics. The sociology of language, or more precisely of languages, which constitutes for Faye (1972a) the essential part of his critique of the narrative economy (économie narrative), sees itself as an empirical science; in its analysis, it aims to link the field of production of language to the field of the struggle – or war – of social classes in, by showing that the first is constitutive of the second, and is not merely its reflection or trace.

We will not focus in detail on Faye’s (1972b) formulations and results from his research on the constitution of totalitarian languages. However, we share some of his concerns that are absent in both Robin and Marcellesi and Gardin:

1. The connection between language and social formation cannot be resolved theoretically but concretely in an exploration of the materials of language and their circulation under specific conditions.

2. The circulation of statements allows us to follow both the constitution and transformation of the various languages as well as their social implications.

3. The sociology of languages cannot be reduced to an analysis of materials in terms of apparatus (i.e. revisiting precise ideological positions and therefore a choice of materials that are necessarily calibrated, whether by writing or by the device of experiments). Thus, for instance, a research project on the Russian revolution would have a method comparable to his work on totalitarian languages, as Faye (1972b) writes: “(the work) would only be possible if the complete archives of the weeks of Petrograd could be compiled: this would presuppose the collection of all the letters and notes, of all the messages without distinction issued by the principal authors of the Petrograd Soviet and the leaders of its various parties – and of the recording, stenographic or otherwise, of all the speeches delivered, of all the oral improvisations.” (Faye 1972b: 116, own translation MZ and MF).

6 An attempt to define language practices

We will use an example to show that the impasse pointed out by Robin is perhaps linked to the fact of remaining locked in the study of the linguistic elements that constitute an utterance (syntactic or lexical). During a survey on an assembly line, a sociologist^[8] was confronted with language phenomena that were not his initial focus: the workers developed a specific form of language exchange: i.e. either the complete absence of exchange (they never spoke during work), or extremely brief exchanges, interspersed with long silences (eight spoke for 10 min a day, eight spoke for longer). It is to be noted that during the breaks, the workers displayed the same language “behavior”. If we had recorded and transcribed these language productions, we would probably have found elements of syntax specific to the oral production,^[9] possibly lexical elements determined by the social origin of the speakers; but we would have completely missed what any observer would have spotted: i.e. the flow – pauses and silences – which was the basis for a particular language practice.

We will therefore not use the terms “linguistic practices” or “linguistic activity” to avoid confusion between the activity of linguists and the everyday language practice of all speakers. We want to try to understand a set of socially hierarchized language practices, specific to a given social formation; a set in which every speaker is embedded without even being aware of it.

We put forward the idea of a language formation, understood as a regulated set of language practices, which organizes these practices according to power relations into dominant and dominated practices. In this sense, it is a matter of describing power relations that structure language practices within a language formation. We thus argue that there are power relations between language practices. As such, our suggestion goes beyond the perspective by Robin or Marcellesi and Gardin, in suggesting that language carries a trace of power relations existing elsewhere.

We propose these notions of language practice and language education on the basis of research in three primary school first-year classes (Cours Préparatoires) in Paris. For a year, we observed and recorded language exchanges and language practices in the classroom with a strong interest in social class. We did not use a precise experimental protocol, but we were regularly present and were integrated into the class activities. Based on these observations and on discussions with the teachers, the following hypothesis was born: the school apparatus regulates language in a very precise sense, which, it seems to us, differs somewhat from the descriptive literature commonly accepted in the field of socio-pedagogy. The children’s oral language practices (e.g. how they speak to each other) do not initially show relevant differences, which would clearly indicate class differences. However, this particular oral language practice is gradually eliminated by the school, first by teaching writing/written language practice (i.e. through the threefold submission of children to a discipline of gesture and body, the discipline of spelling, and the discipline of written “style”). From then on, oral language practice no longer has any recognised existence and is totally subordinated to written – the only institutionally recognised practice. And, in fact, differences in class membership are clearly revealed in the diversified submission to the norms of written language practice. From our point of view, the school apparatus is therefore essentially characterized by one dominant language practice, i.e. the written word, and the elimination of a language practice common to all children, i.e. the spoken word.

The task we assign to the sociology of language is therefore to describe language formation in a given social formation, i.e. a set of language practices linked together in relations of domination, and therefore contradictory. Any study carried out in this sense must be historically situated. Whether language formation in France under the Ancien Régime or language formations in 1975, they need to be specifically described to analyse their relationship. The studies by Balibar and Laporte (1974) paved the way for such research.

Since language practices cannot be identified in individual terms, the interview situation, which is widely used in sociolinguistics, can hardly be used from a methodological point of view. It does not allow us to understand the hierarchy of practices nor the power relations that are part of them.

As Labov showed, all speakers do in fact have at their disposal all the linguistic means (of a language), and that, in this sense, it is wrong to speak of ‘linguistic handicaps’.^[10] However, the social organization, by regulating exchanges, imposes where one speaks, where one does not speak, and how one speaks. It is not in artificial situations (tests, experimental set-ups) and individual situations (interviews) that we will be able to understand language phenomena but in real, historically significant social situations. This is why we have chosen to work on the preparation class because we see it as a key moment in the structuring of language practices.

Yet, it would also be interesting to work on historical situations, in which practices become “disrupted”, e.g. in cases of strikes, where the balance of power reveals a different structuring by being (temporarily) reversed. The LIP workers [workers of the LIP watch factory who launched a big strike in Besançon] made numerous remarks^[11] about the profound changes in their language practice over the course of their struggle. Their language practice changed from silence to speech, from consent to debate and controversy, from the empty and rare words of everyday life to the common search for a language capable of expressing the meaning of a struggle while transforming it. All of this calls for a deepening of the notion of language practice that can capture linguistic organisation within the very fabric of social organisation.