Abstract

More than a decade ago, McCarthy (1998) noted that the role of lexical patterns in written texts had been the object of detailed attention, especially within the study of lexical cohesion (Halliday and Hasan, 1976; Hasan, 1984). Similarly, the significance of multiple ties between words in written texts had been meticulously recorded by Hoey (1991). This, McCarthy asserted at the time, was not matched by anything like the same amount of research into lexical patterning in everyday spoken language. The present chapter can report some considerable progress since then, especially in light of the increased number of spoken corpus-based studies using large amounts of data and of growing interest in the study of collocation and chunking, which have in turn contributed to the methodology and findings of discourse analysis, as we demonstrate below. Chunking in particular has been examined in terms of its role in spoken interaction. In this chapter, we consider how the study of lexis using large amounts of spoken data can underpin the insights into lexical patterning already observed by keen-eyed discourse and conversation analysts in one-off extracts and can provide empirical support from a wide range of occurrences for statements concerning the regularity and recurrence of particular lexical phenomena at the level of discourse (Stubbs, 2001). The first question we address is whether there are differences between the spoken and written lexicon as a whole and what implications any differences might have for an understanding of spoken discourse. We then focus on how lexical patterns manifest themselves within and across speaker turns and their contribution to the unfolding discourse. We base our evidence on everyday, informal, spoken data, mostly social conversations, for it is there, we would argue, that patterns of negotiation and social convergence at the lexical level are most fruitfully observed. We take as uncontroversial the claim that the use of corpus data can offer considerable enhancements to discourse analysis, as demonstrated for example in Bublitz's (1988) use of corpus data in the study of cooperative conversations and, more recently, by Thornbury (2010), who argues that corpus linguistics, with its emphasis on the study of co-text, can powerfully supplement the discourse analyst's investigation of context, as well as providing large numbers of examples of given phenomena.