Enhancing ESL/EFL Writing Through GenAI-SRL instruction: Effects on Self-Regulated Learning, Cognitive Processing, and Writing Performance

• In Event: AI in Higher Education: Balancing Support, Ethics, and Agency

Abstract

Amidst debates on whether Generative AI (GenAI) fosters dependency or develops skills, this study tested a novel GenAI-Self-Regulated Learning (SRL) instructional model designed to cultivate self-regulation and skill transfer in ESL/EFL writing. Through a randomized mixed-methods experiment with 56 university learners, process data from human-AI interactions revealed that the experimental group employed SRL strategies more effectively, transitioning from shallow to deep cognitive processing. Epistemic network analysis further demonstrated stronger connections between their SRL strategies and higher-order thinking. Importantly, the experimental group significantly outperformed the control group on a post-test conducted without GenAI assistance. This research validates a pedagogical pathway where structured instruction transforms GenAI from a crutch into a cognitive partner, fostering genuine, transferable competence.

Authors

• Jing Wang, Zhejiang University

• Tianni Chen, Zhejiang University

• Yuqin Yang, Central China Normal University