8 - Modeling Listener Responses

Summary

This chapter demonstrates how to more fully integrate an interactional perspective into the definition, categorization, and quantification of a variable firmly rooted in the organization of discourse: listener responses, often called backchannels (Yngve, 1970).Building on an interactional analysis of approximately 5,200 vocalized listener responses in 17 hours of dyadic conversations, the study proposes (1) an interaction–based definition of listener responses, (2) an interactionally accountable way of quantifying, and (3) a coding scheme for different actions listeners can do, taking into account the discourse as it unfolds (see also Schegloff 1993 on quantification and conversation analysis). Given that the listener responds to their interlocutor at any given point, the envelope of variation needs to be defined based on aspects of the interlocutor’s linguistic production (similar to Duncan & Niederehe 1974). This chapter considers listener responses to an ongoing (multi–unit) turn relative to the number of words in this turn. Zero–inflated Poisson regression models confirm that women listening to women produce the highest number of listener responses, and men listening to men the lowest. This finding is also reflected in the distribution of action types, with women in all–female dyads producing the highest rates of complex responses.