Social categories and theoretical descriptions
aJ Social categories as theoretical descriptions
The claim that the social categories that we employ do not present themselves directly to us means that such categories are fashioned out of theoretical networks, and are devised for particular purposes. The positing of social categories is not the exclusive preserve of researchers: the social categories that we all use in our everyday interactions are often matters of unexamined convention and tacit agreement. These categories may conceal the assumptions about social reality on which they rest. Any uncritical use of them in social research will import these hidden assumptions, which makes problematic any efforts to define, for research purposes, social categories like 'adolescents', or 'the Latino community', or 'lower working-class speakers' in such a way that they can function as 'independent variables'.
Some theoretical descriptions of these phenomena locate individual speakers at the center of a nest of concentric circles. Speakers then voluntaristically proclaim their identification with certain social groups rather than others, subject to the pressures exerted on them by immediate peer groups- the gangs to which they belong, the tight social networks of which they are members (Milroy and Milroy 1992) and the more remote contextual influences such as social class. Chambers (1995) summarizes aspects of this idea as follows: '... the difference between social networks and social classes as norm enforcement mechanisms has to do with their proximity to the individual, or the immediacy of their influence' (p.68), and '... the difference between class and network seems to be one of degree rather than kind' (p.71).
The problem with such accounts, from our perspective, is the linear model on which they are implicitly based. In this, the linguistic practices adopted by actual speakers are regarded as deviations or deflections from the probabilistic patterns identified by the correlations of independent variables. Actual speech that does not conform to the pattern predicted by the correlated variables becomes a remaindered leftover, to be explained by factors exogenous to the variables originally considered, such as local networks, political attachments and so on.
