Perceptions and Effectiveness of AI-Assisted Written Corrective Feedback: A Case Study of Chinese EFL University Students

Abstract

This study explores Chinese university students’ attitudes toward AI-assisted written corrective feedback (AI-WCF) and their perceived improvement in argumentative writing. Adopting a one-group retrospective pretest–posttest design, the study collected data from 89 first-year undergraduates through two questionnaires: one on attitudes toward AI-WCF, and the other on self-assessed progress across four writing dimensions—unity, support, cohesion and coherence, and language use. Over a 12-week College English course, students engaged in structured writing tasks supported by AI feedback. Results showed generally positive attitudes, with strong endorsement of AI-WCF’s usefulness and future applicability. Paired-samples t-tests revealed significant perceived gains across all subskills (p < .001), with large effect sizes (d = 0.77–1.57), particularly in thematic clarity, organization, and lexical precision. Nonetheless, students expressed reservations about fully integrating AI-WCF into formal instruction, citing its limited rhetorical depth. Findings highlight the importance of differentiated support and suggest integrating AI tools within scaffolded feedback processes to enhance learner engagement and autonomy.