Investigating student translators’ human–GenAI interaction patterns: an exploratory study using cluster analysis and epistemic network analysis

This study investigates how student translators interact with Generative AI (GenAI) during translation tasks. To identify distinct interaction patterns, we first categorised participants into three clusters based on their engagement. The interactions of these clusters were then analysed through the Community of Inquiry (CoI) framework, and the clusters were labelled as reflective deep engagers, pragmatic global adjusters, and minimalist result-seekers. Reflective deep engagers paired high cognitive engagement and emotional regulation with recursive inquiry, achieving a relatively high mean and consistently stable translation performance. Pragmatic global adjusters used a linear, functional approach with limited exploration, yielding intermediate and more variable performance. Minimalist result-seekers showed fragmented cognitive engagement and minimal communication, resulting in inconsistent outcomes and a higher risk of poor translation quality. These findings highlight the importance of fostering cognitive reflection and emotional engagement in translator training to fully harness GenAI's potential.