Computational Analysis of Social Positioning in Collaborative Learning Interactions - Penstein-Rosé

In this talk I will present recent work on computational modeling of social positioning in conversational interactions, or leadership taking in collaborative learning interactions more specifically. Basic concepts of power and social distance from psychology explain social processes operating in conversational interactions.  This theoretical foundation explains how we gain influence within interactions through our manipulation of horizontal and vertical social distance through a variety of types of signaling, which may or may not be linguistic. Research in sociolinguistics illustrates and describes how these signals are encoded in language at various levels, from pronunciation, to lexical choice, to rhetorical strategies at the discourse level. Thus, this work describes how social processes are reflected through patterns of language variation.  If we can understand the connection between social processes and language by means of these signals, we can structure computational models of language interactions more effectively. Models that incorporate these structures should be capable of making predictions about the presence of social processes from interaction data and can then be used to monitor collaborative learning processes in real time.  Two specific operationalizations of leadership will be addressed conceptually as well as computationally, which are referred to as Authoritativeness and Transactivity. In addition, Rosé will show and comment on a short Wiseman clip from the documentary High School I from the point of view of social positioning, as it is related to authority.