How conversation analysis informs the design of social, mobile audio spaces
Conversation analysis is informing the design of technology that extends traditional audio spaces to meet the needs of social, mobile groups. To date, audio spaces have provided little support for the dynamic participant regroupings characteristic of social groups. Most technologies either require that participants explicitly specify with whom they wish to talk, or they simply present all participants as if they were in a single conversation. One prototype system our group has developed identifies conversations as they emerge by monitoring participant behavior. The system includes a learning algorithm that identifies conversations using two fundamental features of the organization of turn-taking: 1) one party talks at a time, and 2) the minimization of gap and overlap in transitions from one turn to a next. Once the conversations have been identified, the system adjusts audio dynamically to support multiple, simultaneous conversations. Subsequent CA analysis has revealed two other phenomena that are currently being explored for integration into the system: the patterned alternation of overlaps and gaps (more-than one and less-than-one) as a systematic product of the turn-taking organization, and variations of schism-inducing turns that implicate the emergence of new conversations. The data corpus consists of seven hours of audio recordings of mundane, copresent, multi-party (8-10 people) conversational interaction. The participants are friends or acquaintances ranging from 14 to 24 years old. The analysis is twofold. First, the data are coded to signal the initiation and subsequent lapse of a conversational floor; many floors occur in overlap. This coding enables machine learning and also provides a graphic overview that facilitates subsequent detailed analysis. Second, spates of talk containing phenomena of interest are transcribed, analyzed and then comparatively analyzed within its collection.
Szymanski, M. H. ; Aoki, P. M. ; Thornton, J. D. ; Plurkowski, L. How conversation analysis informs the design of social, mobile audio spaces. International Conference on Conversation Analysis; 2006 May 10-14; Helsinki; Finland.