The history of discourse analysis is beset by a vagueness around the homonym ‘discourse’. The term names a large territory, located somewhere between two ‘markers’, which might, generally speaking, be something like ‘providing accounts of connected stretches of language in use’ and ‘uncovering salient social, political, psychological features in text-like entities’. In sociolinguistics, by and large, the major emphasis has been on understanding the link between (environments of) language use and (features of) the language used . In such work, ‘the social’ and its meanings and effects are foregrounded: who speaks, to whom, when, with what purposes, in what ways. These factors and the purposes leave traces – whether the details of pronunciation in Labov’s early work or the regularity of use of a certain range of linguistic resources, leading to the development of the notion of ‘codes’ in Bernstein’s theory.In more linguistically rather than sociolinguistically or sociologically oriented approaches, the emphasis has been on seeing whether regularities of a ‘formal’ kind could be discerned in ‘stretches’ of speech and writing ‘above’ the sentence, somewhat akin to those that linguistics had been able to establish in relation to the sentence in the 1970s – whether in mid-century American structuralism or in Chomskyan conceptions of the organization of language at or below the level of the sentence.