L2 Student Use of ChatGPT's Content-focused Feedback in Text Revision: Behavioural, Cognitive, and Affective Engagement

While emerging literature suggests ChatGPT’s superiority to existing automated written evaluation systems in giving content-focused feedback, there remains a paucity of scholarship exploring how individuals utilise it in text revision. Against this backdrop, this study examined how six mainland Chinese undergraduate EFL students engaged with ChatGPT’s content-focused feedback when revising their argumentative essays. Employing a three-dimensional framework of learner engagement, the study scrutinised the behavioural, cognitive, and affective dimensions of engagement. Multiple data sources were collected, including screen recordings of students’ revision process, original and revised essays, ChatGPT chatlogs, stimulated recalls, and semi-structured interviews. Findings showed that behaviourally, students substituted a majority of ChatGPT’s suggestions at the content level due to concerns about academic dishonesty. Cognitively, students noticed and comprehended the received comments, applying extensive cognitive and metacognitive strategies to make informed decisions on revision operations. This suggests a relatively high level of cognitive engagement. The affective engagement was primarily negative, with participants pointing out the limitations of the feedback and their preference for human feedback when it comes to essay content. Pedagogical implications of developing students’ effective engagement with ChatGPT’s content-focused feedback are discussed.