‘Butterfly effects’ in dyadic dynamics: a complex dynamic systems theory perspective on task-based interaction

While a substantial body of research on L2 oral performance has focused on aggregated outcomes, this study adopts a Complex Dynamic Systems Theory (CDST) framework to investigate the co-constructed processes inherent in dyadic interaction. Specifically, we explore the butterfly effect, wherein minor variations in initial conditions can lead to divergent interactional trajectories. A complex dynamic system comprises many interacting components, where micro-level interactions among components change their properties and, over time, create new properties at a macroscopic level (van Geert, 2011). From this perspective, participants are continuously influenced by their interlocutors through iterative processes of turn-taking, and a macro-pattern of discourse emerges as a result of dynamic co-regulation between the participants (Larsen-Freeman & Cameron, 2008). To investigate dyadic dynamics as a complex dynamic system, we conceptualized planning time (3 minutes vs. none) as the initial condition for a discussion task and re-analyzed the task-based interaction performed by 32 L2 learners in dyads (Nitta & Nakatsuhara, 2014). We employed a microgenetic approach, analyzing turn-taking dynamics and using Conversation Analysis to examine the qualitative features of the conversations to identify emergent interactional patterns under each condition. The analysis revealed a primary pattern observed in the majority of dyads: the non-planning condition fostered a highly collaborative mode of interaction, whereas the planning condition promoted a shift toward a sequence of monologues. However, a deviant pattern, identified in a minority of dyads, was characterized by consistent collaboration across both conditions. These findings suggest that while initial conditions like planning time are highly influential, their effects are not deterministic. We conclude that process-oriented, microgenetic methods are invaluable for understanding the nuanced dynamics of L2 interaction, and that CDST offers a promising theoretical lens for advancing both language learning research and pedagogical practice.