Making task engagement visible and interpretable: A task-driven composite approach

Abstract This study investigates the effects of vocabulary support on engagement and lexical learning in paired interactive information-gap tasks. Sixty-six Japanese learners of English as a foreign language (EFL) performed the tasks either with or without access to task-relevant collocations. Engagement was operationalized through a task-driven composite approach, in which discourse-analytic indicators from transcribed interactions were summarized via principal component analysis. This resulted in three engagement patterns reflecting (a) use and negotiation of target collocations, (b) co-construction of task-relevant information and joint meaning making, and (c) task-maintenance behaviors. Bayesian regression and mediation analyses showed meaningful group differences in the engagement patterns but provided little evidence that these patterns mediated gains in productive knowledge of the target collocations, suggesting a complex relationship between task engagement and vocabulary knowledge development. Methodologically, the study illustrates how this approach can retain the visibility of engagement indicators, allowing greater interpretability of findings in relation to task design and instructional decision-making.