A computational approach to expose conversation dynamics in engineering design activities

Abstract

In this paper, we apply a computational approach to expose conversation dynamics in design conversations. We apply this approach to a case study comprising of two sessions, with 300 speaker turns respectively. The resulting analysis of the variance in speaker score (based on speaker turns and word count) allows the researcher to expose the conversation dynamics throughout the session and creates the foundation for more in-depth analysis of design team interaction. Our main interest are these design team interactions, more specifically design conversations. These conversations are made up of information content (semantics) and temporal patterns of information exchange (dynamics). Many design researchers have focused on semantics (e.g. Schön, 1993; Goldschmidt, 2014; Dorst & Cross, 2001), while the dynamics of design conversations has not received the same attention in the past. We believe that there is a mutual interaction between semantics and dynamics, which must be taken into account when studying design activities. The dynamics of a design team is governed by the combined contribution of the individual participants, for example a single person dominating the conversation, or the whole team interacting in equal proportions. In this work, we explore the concept of conversation dynamics.