There are several possible relationships between language and society. One is that social structure may either influence or determine linguistic structure and/or behavior. Certain evidence may be adduced to support this view: the age-grading phenomenon whereby young children speak differently from older children and, in turn, children speak differently from mature adults; studies which show that the varieties of language that speakers use reflect such matters as their regional, social, or ethnic origin and possibly even their sex (or gender); and other studies which show that particular ways of speaking, choices of words, and even rules for conversing are in fact highly determined by certain social requirements.
A second possible relationship is directly opposed to the first: linguistic structure and/or behavior may either influence or determine social structure. This is the view that is behind the Whorfian hypothesis, the claims of Bernstein, and many of those who argue that languages rather than speakers of these languages can be ‘sexist’. A third possible relationship is that the influence is bi-directional: language and society may influence each other. One variant of this approach is that this influence is dialectical in nature, a Marxist view put forward by Dittmar (1976), who argues (p. 238) that ‘speech behaviour and social behaviour are in a state of constant interaction’ and that ‘material living conditions’ are an important factor in the relationship.
A fourth possibility is to assume that there is no relationship at all between linguistic structure and social structure and that each is independent of the other. A variant of this possibility would be to say that, although there might be some such relationship, present attempts to characterize it are essentially premature, given what we know about both language and society. Actually, this variant view appears to be the one that Chomsky himself holds: he prefers to develop an asocial linguistics as a preliminary to any other kind of linguistics, such an asocial approach being, in his view, logically prior.
We must therefore be prepared to look into various aspects of the possible relationships between language and society. It will be quite obvious from doing so that correlational studies must form a significant part of sociolinguistic work. Gumperz (1971, p. 223) has observed that sociolinguistics is an attempt to find correlations between social structure and linguistic structure and to observe any changes that occur. Chambers (1995, p. xvii) echoes that view: ‘The correlation of dependent linguistic variables with independent social variables... has been the heart of sociolinguistics…’ Social structure itself may be measured by reference to such factors as social class and educational background; we can then attempt to relate verbal behavior and performance to these factors. However, as Gumperz and others have been quick to indicate, such correlational studies do not exhaust sociolinguistic investigation, nor do they always prove to be as enlightening as one might hope. It is a well-known fact that a correlation shows only a relationship between two variables; it does not show ultimate causation. To find that X and Y are related is not necessarily to discover that X causes Y (or Y causes X), for it is also quite possible that some third factor, Z, may cause both X and Y (or even that some far more subtle combination of factors is involved). We will therefore always have to exercise caution when we attempt to draw conclusions from any such relationships that we observe.
A worthwhile sociolinguistics, however, must be something more than just a simple mixing of linguistics and sociology which takes concepts and findings from the two disciplines and attempts to relate them in simple ways. Hymes (1974, p. 76) has pointed out that a mechanical amalgamation of standard linguistics and standard sociology is not likely to suffice in that adding a speechless sociology to a sociology-free linguistics may miss entirely what is important in the relationship between language and society. Specific points of connection between language and society must be discovered, and these must be related within theories that throw light on how linguistic and social structures interact.
Holmes (1992. p. 16) says that ‘the sociolinguist’s aim is to move towards a theory which provides a motivated account of the way language is used in a community, and of the choices people make when they use language.’ For example, when we observe how varied language use is we mush search for the causes. ‘Upon observing variability, we seek its social correlates. What is the purpose of this variation? What do its variants symbolize?’ (Chambers, 1995, p. 207). For Chambers these two questions ‘are the central questions of sociolinguistics.’ We may not agree with him and prefer other questions but questions we must have if we are to interesting work.
参考资料：Ronald Wardhaugh, 1998, An Introduction to Sociolinguistics, 10-11, Blackwell Publishers Ltd.