Abstract

An examination of Ian Parker's definitions of discourse reveals them to be non-distinctive and of limited utility. It is argued that discourse analysis should be integrated with, rather than set against, social psychology. Discourse analysts should attend to the issues of the representativeness and generality of their evidence, should be wary of attributing causality to discourse, and should consider the advantages of systematically investigating, rather than asserting, the social consequences of the use of different discourses