Diatopic variation in digital space

Abstract

The availability of large amounts of social media text offers tremendous potential for studies of diatopic variation. A case in point is the linguistic geography of Texas, which is at present insufficiently described in traditional dialectological research. This paper summarises previous work on diatopic variation in Texas English on the basis of Twitter and presents an approach that foregrounds functional interpretability over a maximally clear geographical signal. In a multi-dimensional analysis based on 45 linguistic features in over 3 million tweets from across the state, two dimensions of variation are identified that pattern in geographically meaningful ways. The first of these relates to creative uses of typography and distinguishes urban centres from the rest of the state. The second dimension encompasses characteristics of interpersonal, spoken discourse and shows an East-West geographical divide. While the linguistic features of relevance for the dimensions are not generally considered in dialectological research, their geographic patterning reflects major tendencies attested in the literature on diatopic variation in Texas.[1] [1]I am grateful to Alex Rosenfeld for sharing his data with me. This work was initially presented at a panel on Twitter in sociolinguistic research at NWAV 49, organised by Stef Grondelaers and Jane Stuart-Smith. I would like to thank both of them for giving me this opportunity and the attendees of the panel, especially Lars Hinrichs and Alex Rosenfeld, for fruitful discussion. Finally, my gratitude goes to Erling Strudsholm and Anita Berit Hansen for their invitation to participate on the Coseriu Symposium and their patience in organising this special issue.