The Effectiveness of ChatGPT in Developing Academic Writing Skills for Architecture Students

Abstract

In the context of ongoing digital transformation in higher education, the integration of generative artificial intelligence into foreign language teaching has gained increasing scholarly and pedagogical interest. This study examines the effectiveness of ChatGPT as a virtual teaching assistant in supporting the development of English academic writing skills among non-English-major students at Hanoi Architectural University. Adopting a quasi-experimental research design, the study involved 60 first-year students who were assigned to an experimental group receiving ChatGPT-supported writing feedback and a control group following conventional writing instruction without AI support. Data were collected through pre- and post-writing tests evaluated using an analytic scoring rubric, supplemented by learning journals to capture students’ writing experiences.
The findings indicate that although both groups showed improvement over the intervention period, the experimental group demonstrated significantly greater gains in vocabulary use, grammatical accuracy, and textual coherence. The intervention yielded a large effect size, suggesting a strong pedagogical impact. Qualitative evidence further reveals enhanced learner confidence, improved awareness of recurrent language errors, and a noticeable shift toward iterative writing and revision practices. Overall, the results suggest that ChatGPT, when pedagogically guided, can serve as an effective supportive tool in university-level English academic writing instruction.