Artificial Intelligence and Teacher Feedback in Second Language Writing: A Comparative Review and Future Directions

Abstract

The rapid integration of artificial intelligence, particularly large language models such as ChatGPT, has profoundly reshaped feedback practices in second language (L2) writing instruction. This paper presents a systematic and comprehensive review of contemporary research on AI-generated and teacher-provided feedback in L2 writing contexts, focusing on their respective functions, comparative effectiveness, and interactive relationships. The review clarifies three core thematic areas: the complementary value of dual feedback modes, student engagement with AI-mediated feedback, and optimized prompt design for AI-based writing tools. Findings indicate that AI feedback demonstrates notable advantages over traditional teacher feedback in terms of efficiency, coverage, and capacity for positive reinforcement. However, teacher feedback remains indispensable for delivering personalized guidance, addressing nuanced rhetorical issues, and fostering learner self-regulation and agency. Drawing on these insights, this paper advocates for a hybrid AI-teacher feedback framework that strategically integrates the strengths of both modalities to maximize instructional outcomes. Furthermore, the review identifies critical gaps in the existing literature and proposes several promising future research directions, including learner-centered perspectives, diversified feedback modalities encompassing multimodal and peer-assisted approaches, and the adoption of mixed-methods research designs. By bridging theoretical foundations with practical implications, this review aims to inform educators, researchers, and policymakers about the evolving landscape of feedback in L2 writing and to guide the development of more effective, equitable, and technologically informed pedagogical practices.