Impact of ChatGPT on metacognition and motivation in EFL academic writing

Abstract

Using a convergent mixed-methodology approach with a within-subject classroom design, the study examines the impact of ChatGPT on EFL learners’ metacognitive strategies and motivation in academic writing. Participants were 120 English junior majors (CEFR level B1/B2) who engaged in the survey, completing the “with vs. without” AI writing tasks. Pre- and post-task surveys and reflections measured changes in self-regulatory behaviours including planning, monitoring, revising, and affective factors such as self-efficacy, task value, and enjoyment. Findings reveal divergent outcomes. Specifically, ChatGPT significantly enhanced planning and revision strategies, especially among B2-level students obtaining large effect sizes (Cohen’s d ≈ 0.85–1.06). Nonetheless, there was a slight decline in monitoring among B1 students due to excessive cognitive reliance on AI. Higher-level students also gained motivational improvements with moderate-to-large effects for self-efficacy (d = 1.19), task value (d = 1.27), and enjoyment (d = 1.23). Plus, gains in self-efficacy were strongly correlated with motivation gains (r = .60–.67, p < .001). Overall, ChatGPT appears to function as a cognitive scaffold and an affective enhancer in EFL academic writing. The study proposes the S.M.A.R.T. AI framework to integrate AI critically and responsibly in writing instruction.

Ähnliche Arbeiten