AI-peer integrated feedback in second language writing classes: exploring students’ engagement and writing performance

Abstract

While a considerable body of research has focused on examining the efficacy of peer feedback or AI-generated feedback in L2 writing, limited research has investigated the effect of the combination of AI and peer feedback on students’ engagement and their writing performance. To address this gap, the present study adopted a mixed-methods design. A 10-week feedback intervention was implemented with 122 Chinese EFL students over a 12-week study period: the experimental group ( n  = 61) received AI-peer integrated feedback, whereas the control group ( n  = 61) engaged in conventional peer feedback. To assess their engagement and writing performance, all participants completed pre- and post-tests, five writing assignments, and engagement questionnaires. Additionally, 16 students from each group participated in semi-structured interviews and kept reflective journals, providing rich qualitative data on their engagement patterns. The results indicated that the experimental group exhibited significantly higher levels of behavioral, affective, and cognitive engagement than the control group. Moreover, they demonstrated greater improvements across four writing subscales, namely task achievement, coherence and cohesion, grammatical range and accuracy, and lexical resource, albeit with varying magnitudes of gain across these dimensions. These findings suggest that AI-peer integrated feedback represents a promising pedagogical approach for fostering L2 writing development, shedding light on the potential for human-AI collaboration in language education.