The present study contributes empirical evidence to the growing literature comparing human teacher feedback and AI‑generated feedback in the university context. A group of 16 Year 1 students were invited to complete two English paraphrasing tasks in class (without access to electronic devices). They were asked to revise two poorly paraphrased texts following writing instructions given by a human teacher and generated by AI respectively. The study analyses instruction‑by‑instruction uptake and paraphrasing behaviour of the two sets of answers. Results show that human teacher feedback demonstrated stronger learner success in addressing conceptual and interpretive instructions, including numerical interpretation, categorical accuracy, and meaning level reformulation. On the other hand, AI‑generated feedback revealed high effectiveness in lexical and mechanical corrections, including tone normalization and spelling refinement. However, students showed weaker performance on sentence completeness, syntactic integration, and abstract meaning expression. In terms of quality of paraphrasing (degree of rewriting), no significant differences were observed, indicating that the task type influenced where students struggled rather than overall level of achievement.

Authors:
Pui Lun Daisy Chow, The Chinese University of Hong Kong, Hong Kong

About the Presenter(s)
Dr. Chow Pui Lun is currently a Lecturer at English Language Teaching Unit, Faculty of Arts, The Chinese University of Hong Kong, Hong Kong.

See this presentation on the full schedule – Monday Schedule

Bookmark