Usage-based grammars and sign languages: Evidence from Auslan, BSL and NZSL Adam Schembri, La Trobe University, Melbourne, Australia Trevor Johnston, Macquarie University, Sydney, Australia 1. Introduction Hollmann and Siewierska (2011) noted that scholars in cognitive linguistics have argued for some time (e.g. Langacker, 1999) that language should be analysed as both a cognitive and a social phenomenon, and not solely as resulting from processes within the mind of the individual speaker. Despite this there has been very little work that has explored the relationships between cognitive linguistics and sociolinguistics. With the publication of two edited volumes by Kristiansen and Dirven (2008) and by Geeraerts, Kristiansen and Peirsman (2010), the field of ‘cognitive sociolinguistics’ has, however, begun to emerge from these inter-connecting perspectives on language. In particular, much of the work draws on insights from variationist sociolinguistics (see Hollman & Siewerksa, 2011, for a summary). This partly reflects the somewhat under-appreciated fact that work within variationist linguistics often has implications for an understanding of language in general, as the multivariate approach to the study of corpus data can be used to test claims made within any linguistic theory (Walker, 2010). For example, recent variationist analyses of the complementiser that in Canadian English (Torres Cacoullos & Walker, 2009) and of definite article reduction in the Lancashire dialect of British English (Hollmann & Siewierksa, 2011) provides support for usage-based accounts of grammar. In this paper, we discuss a number of findings on variation in specific phonological, morphological and syntactic features in Auslan (Australian Sign Language), New Zealand Sign Language (NZSL) and British Sign Language (BSL) that