The integration of generative artificial intelligence into digital multimodal composing in second language classrooms: a scoping review from the perspective of tasks

With the rise of generative artificial intelligence (GenAI), integrating it into digital multimodal composing (DMC) has created innovative pedagogical opportunities for second language (L2) education. In the current research, the integration of GenAI into DMC, particularly how it facilitates task design and implementation in DMC, remains insufficiently explored. This scoping review investigates two questions. (1) In what forms is GenAI integrated into DMC tasks in L2 classrooms? (2) What specific pedagogical roles does GenAI perform in DMC task design and implementation? Using Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA), 12 empirical studies were selected and analyzed. Three forms of GenAI-assisted DMC tasks were identified, and GenAI fulfilled five functional roles. The findings further suggest that the pedagogical functions of GenAI blur the conventional boundaries of the task phases, driving a shift towards more dynamic and functionally structured language tasks. The research redefines GenAI-assisted DMC tasks as dynamic instructional practices co-constructed by teachers, students, and GenAI. The study proposes a task design framework, namely role identification–interaction and forms–instruction design, and calls for future research to validate this framework and expand GenAI applications across diverse contexts and disciplines. This review claims that GenAI should be positioned not only as a tool but as an interactive agent in meaning-making in L2 DMC tasks.