ABSTRACT
Social media refers to the communication generated on platforms that enable users to send text messages and multimedia content to a group of other users. Despite its growing importance, social media has received scant attention in our field due in part to the kinds of text used in social media posing challenges for corpus methods and techniques. In the study reported in this chapter, a multi-dimensional (MD) analysis was carried out to detect the major dimensions underlying the variation across platforms, user groups and individual users in social media. MD analysis is a framework for corpus analysis that uses multivariate statistical analysis to identify correlations among linguistic features across the texts in a corpus. Two dimensions were identified: dim. 1: Formal, informational, argumentative discourse; dim. 2: Informal, interactive, speaker-oriented discourse. Each dimension corresponds to a set of linguistic features commonly occurring across the texts. After scoring the texts for each dimension, the variation accounted for by platform, user groups and individual groups was calculated. The findings suggest that the variation seems to be driven primarily by the individual users rather than by the platform or the types of user.
