VietTESOL International Convention 2026

This empirical study explores the pedagogical potential of Generative AI (GenAI) as an immediate feedback tool to enhance the linguistic complexity and professional relevance of writing outputs among low-proficiency technical students after an instructional delay. The study focused on the "City Description" writing task from Unit 5 of the English File Pre-intermediate curriculum, a standard topic often perceived as redundant by engineering majors.
Adopting a quasi-experimental pre- and post-test design, the author examined the writing artifacts of 46 engineering students (Experimental Group, n=21; Control Group, n=25). To measure delayed language retention, the intervention was administered seven weeks post-instruction. The CG followed traditional drafting, whereas the EG was trained to engineer specific prompts directing ChatGPT/Gemini to provide formative feedback on comparatives, superlatives, and urban infrastructure vocabulary. A double-blind scoring procedure using a standardized analytic rubric was implemented to ensure evaluative objectivity. The results demonstrated that the AI-scaffolded final drafts exhibited a statistically significant improvement in structural accuracy and professional tone compared to the CG ($p < 0.05$). The visual representation on this poster contrasts the raw and refined student artifacts, details the prompt-engineering scaffolding process, and discusses the implications of utilizing GenAI to transition non-English majors from exam-driven learning to functional, career-oriented language application.