From Tool to Partner: ChatGPT-Supported Instruction for English Majors’ Academic Writing in Diverse EFL Contexts

Abstract

Within Vietnamese higher education, English majors need advanced writing skills in technology-mediated and diverse English as a foreign language learning environments. This study examined whether ChatGPT could support second-year English majors in developing academic writing. Drawing on Vygotsky’s Sociocultural Theory, the Zone of Proximal Development and scaffolding, and Krashen’s Input Hypothesis, the intervention integrated technology-supported instruction and formative evaluation. Over seven weeks, 73 students participated in a quasi-experimental design, with 38 receiving ChatGPT-supported instruction and 35 following conventional writing classes. Both groups focused on advanced language use and APA in-text citation. Pre- and post-test writing scores and semi-structured interviews with 12 students examined perceptions of ChatGPT-assisted instruction. Results showed that the ChatGPT group made significant gains in academic writing (Sig. 2-tailed < .05, large effect size Cohen’s d = 1.27 and 1.03). Interview findings indicated that students perceived ChatGPT as useful for thesis development, argument quality, in-text citation, and easy to integrate into writing automously. However, concerns about over-reliance and critical thinking highlighted the need for careful pedagogical scaffolding. Overall, findings demonstrate ChatGPT’s potential to support academic writing’ learner autonomy, and engagement while emphasizing instructor guidance, critical reflection, and purposeful AI integration across diverse EFL learning contexts.