Exploring ESL Learners’ Peer Interactions in GenAI-Supported Collaborative Multimodal PowerPoint Composing

Using a multiple case-study approach, this study investigated the dynamics of peer interactions that took place through Microsoft Copilot among six pairs ( n  = 12) of ESL learners, focusing on their interaction patterns, use of semiotic and multimodal resources, and functions of languaging. Our analyses illustrated that four pairs displayed collaborative orientation, one pair demonstrated an expert-novice pattern, and one pair exhibited a dominant-passive pattern. It was also found that while the collaborative group students focused more on text search and revision, the expert-novice group and the dominant-passive group allocated more time to image generation. Furthermore, in terms of languaging functions, our results revealed that all six pairs of students spent most of their time negotiating writing content. Our study has implications for technology-enhanced multimodal writing pedagogy in a generative artificial intelligence machine-in-the-loop setting.