Distributed agency in AI-assisted L2 writing

Abstract

This longitudinal qualitative study examines distributed agency between human writers and generative AI in the context of L2 writing. Grounded in Bandura’s theory of agency, the study analyzes students’ written texts, reflective accounts, and AI interaction logs collected from Taiwanese university students. The findings indicate that human-AI distributed agency shapes the enactment of L2 writing across intentionality, forethought, self-reactiveness, and self-reflectiveness. Moreover, distributed agency both supports and constrains learners’ engagement, depending on how it is exercised.