Introduction

The use of participant observation and social network analysis (SNA) were popularized in sociolinguistics by James and Lesley Milroy in the 1980s with their application of various measurements of network strength to data collected from three working class communities in Belfast. The technique was introduced as a method for studying sociolinguistic variation between individuals who were not discernible in terms of socio-economic class. Labov’s (2006 [1st edn. 1966]) model of language variation and change attempted to correlate linguistic variation with ‘global’ social categories such as social class, age and sex. Eckert (2005) describes studies which employ these methods as ‘first wave’. These studies typically show regular and replicable patterns of linguistic variation where often the use of vernacular variants strongly correlates with low socio-economic status. However, this approach is unable to explain the variation that continues to exist within larger social categories. Second wave studies (e.g. Rickford 1986; Milroy 1987a) employ ethnographic methods of data collection and SNA in an attempt to better understand the patterning of linguistic variation in a local context. Although highly innovative in the 1980s, the use of these techniques, especially SNA, has received heavy criticism (see e.g. Murray 1993). However, the techniques of SNA have advanced greatly in other disciplines and now incorporate more sophisticated mathematics (e.g. clique analyses based on graph theory) and more detailed methods of data collection. Dodsworth and Hume suggest that ‘linguists could construct more useful measures of network integration and investigate many more qualities (both quantitative and qualitative) of social network data’ (2005: 290). This chapter discusses one attempt to do just that.