The Role of Prompt Language in ChatGPT-Assisted Revision: Effects on Korean EFL Argumentative Writing

This study investigates how Korean EFL university students revised their argumentative essays using ChatGPT, with a focus on how prompt language (English vs. Korean) and specificity influenced the nature and uptake of AI-generated feedback. Drawing on analyses of revised texts and prompts students used, the findings reveal that English prompts more frequently led to global-level changes, such as improved organization, argument clarity, and academic tone. In contrast, Korean prompts typically resulted in sentence-level edits, including lexical refinement and fluency improvement. Across both languages, prompt specificity emerged as a key factor: students who provided clear, goal-oriented instructions received more relevant and actionable feedback. While learners generally valued ChatGPT’s support in enhancing expression and correcting grammar, some expressed concerns about loss of authorial voice and over-dependence on AI tools. These results underscore the need to develop students’ prompt literacy and critical awareness when integrating AI into academic writing instruction.