Characterizing communication networks in a web-based classroom: Cognitive styles and linguistic behavior of self-organizing groups in online discussions

In this study, we explore the cognitive style profiles and linguistic patterns of self-organizing groups within a web-based graduate education course to determine how cognitive preferences and individual behaviors influence the patterns of information exchange and the formation of communication hierarchies in an online classroom. Network analysis was performed on communication data collected from 1131 student messages posted in 19 asynchronous online discussion forums to determine centrality, clique membership, and core-periphery structure in the communication networks. The social network data were examined in relation to the students' cognitive style profiles, which were assessed using the Kirton Adaption-Innovation Inventory (KAI) (32 and 33). The cognitive style composition of small cliques (dyadic and triadic) was found to be highly heterogeneous, often with large cognitive gaps between clique members, which suggests that web-based environments may mask cognitive style differences that have been shown to create conflict in face-to-face interactions. In addition, the cognitive style mean of the students in the core of the network was found to be significantly more adaptive than that of the periphery group. To further characterize the nature of the communicative interactions, automated linguistic analysis was used to analyze the students' writing samples. Interestingly, students in the core of the social network demonstrated a significantly higher usage of several language features associated with individuals who actively promote enhanced group performance and cohesion. For our sample, the linguistic behaviors of students in the core of the social network, coupled with their more adaptive cognitive style preferences, suggest that these students may inherently place greater value on fostering group cohesion than those in the periphery.