GROWING EVIDENCE FROM COGNITIVE SCIENCE suggests that knowledge is represented via a network in cognition (Hudson 2001: 1). This has led to a re-analysis of the modular view of language in the mind (Chomsky 1986). Consequently, many cognitive linguists now propose that linguistic knowledge is organised in the mind in the same way as other, more general aspects of cognition (see e.g. Langacker 1987; Goldberg 1995: 5). This proposal has enormous consequences for linguistics because it blurs many of the distinctions that were traditionally made in linguistic theory. For example, it implies that cognitive linguistic theories no longer recognise a clear dichotomy between linguistic and non-linguistic concepts. Social, cultural and linguistic knowledge are thought to be unavoidably entwined in cognition (Langacker 1994: 31-33). Therefore, theories that come under the cognitive linguistics umbrella2 claim not to deny the importance of the social aspects of language use; they claim not to treat social influences on language as secondary or less important. However, cognitive linguists have given little consideration to the ways in which a network model of cognition can incorporate the enormous amounts of research that have been generated in the field of sociolinguistics. Furthermore, sociolinguists have paid little attention to the rapidly expanding theories of cognitive linguistics, despite often facing criticisms that sociolinguistics is a mainly empirical subject that lacks any central theory (e.g. see Spolsky 1997:7-8; Chomsky 1979: 57). This paper will begin to bridge this gap by highlighting one apparent area of crossover between the disciplines of cognitive linguistics and sociolinguistics: their respective treatment of networks. By exploring the links between a cognitive network model and a social network model, this paper demonstrates that there are, in fact, some remarkable similarities between both frameworks. Moreover, the proposed explanation for these similarities is that social networks may, in fact, exist in the mind of the individual. If this is the case, then social networks must exist as part of the larger cognitive network. The paper is organised in four parts. Section one exemplifies how variation in sociophonetic data can be represented in a cognitive network model. Section two examines the social network structure of the group of speakers from which the data were obtained and highlights the parallels that exist between the social network model and the cognitive network model. Section three explores the possibility that social networks exist in the mind of the individual and section four is a discussion of the implications of this approach.