Sociology and ideas about language
Conclusion
This account of some currents in sociological theory has been necessarily brief and selective. The selection was guided by the light shed in each approach on the nature of the relationship between language, social order and social action, as befits the main themes of language, structure and agency. It is often remarked that sociology has con tributed more to the socially situated study of language than linguistics has to the study of society. Attempts have been made to constitute a 'sociology of language', notably by Fishman (e.g. Fishman 1978), but as he himself observed, writing in 1992, 'the two fields are as remote from each other now as they were in the 1960s' (Fishman in Williams 1992: viii). Topics considered within the 'sociology of language', as well as in some interpretations of applied linguistics, would include bilingualism and multilingualism, bilingual and minority education, language planning, language and nationalism, language and ethnicity, and language rights.
We started with a discussion of the interactionist sociologies, including symbolic interactionism, ethnomethodology and conversation analysis. In applied linguistics, the range of approaches which fall under the heading of 'discourse analysis' includes some which develop and extend interactionist sociologies into detailed studies of language. Conversation analysis is an obvious example, but studies in the 'ethnography of speaking', associated with researchers such as Hymes and Gumperz could also be cited here. Space does not permit us to explore the differences between the various contributors to work in this area, who by no means constitute a homogeneous group. However, approaches such as these share an emphasis on the inter-subjective meanings of members of the 'speech community' to which they are said to belong, and it is, we suggested, a merit of the inter-actionist model that it recognizes an important role for human agency.
We turned our attention next to those approaches which place discourse at the heart of their explanations, and here there is obviously considerable overlap with versions of discourse analysis to which applied linguistics has contributed. If interactionism runs the risk of reducing interaction to nothing more than language, the more extreme discourse sociologies are in danger of reducing social reality to language.
In both cases (that of reducing interaction to language on the one hand and that of reducing social reality to discourse on the other), an abiding problem is how to deal with the notion of constraint, that is, with the ways in which people's conversational strategies and resources are structured. To do so adequately entails considering not only the structuring of language itself, but also issues of power and conflict. Conversation analysis has been criticized for claiming that it attends only to its linguistic data, thereby assuming a voluntaristic, consensual version of the social order:
Above all, it conveys an essentially non-critical view of the social world. The bedrock situation- or the default option- is implicitly depicted as a world of equality and participation, in which 'members' share systems of social order. Inequality is to be found in the exceptions- in institutional talk, interviews etc. Thus, traditional CA, far from being free of social presuppositions, carries them in the regular deployment of its foundational rhetoric. The warnings against being theoretical, and against using conventional sociological analyses, together with the prescription to keep to the data, can serve to protect these assumptions from analysis. (Billig 1999a: 552)
Conversation analysts have taken issue with this depiction (Schegloff 1999), and it must be noted that 'ideological' or 'critical' discourse analysis, which does seek to locate discourse within a social world of conflict and inequality, attracts criticisms of its own. For example, in relation to Critical Discourse Analysis, Stubbs expresses uncertainty about 'the extent to which cause and effect relations are being claimed, about the nature of evidence which textual traces are said to provide of social change, and so on' (Stubbs 1997: 105). As we noted, a problem with accounts which draw on a strong notion of discourse is that these often seem to leave agency behind altogether.
The theorists whose work we considered earlier have sought to hold to a distinction between language and social action, and this, we have argued, makes their work a fruitful source for exploring the relations between language and society. In different ways Habermas, Bourdieu and Bernstein have-importantly-recognized the partial autonomy of language from individuals, and its power to influence directly the nature of their social interactions.
However, as we have also indicated, their work does not fully incorporate the analysis of language itself into a social theory of language and languages. Nevertheless, it is our contention that this task needs to be predicated on the distinction between language and agency.
