Background: Although both peer and AI-generated feedback are increasingly used in EFL writing instruction, little is known about how they differ in terms of cognitive depth and interpersonal delivery. Existing studies often overlook the mechanisms through which feedback operates, limiting instructors’ ability to design balanced feedback ecosystems. Purpose: To provide a comparative analysis of peer and ChatGPT-4o feedback on undergraduate EFL academic writing, drawing on Hattie and Timperley’s (2007) cognitive model and Hyland and Hyland’s (2006) interpersonal feedback strategies. It aims to identify how each source targets task, process, and self-regulation levels, and how their rhetorical styles shape learner engagement. Method: Thirty Iranian undergraduate EFL students participated in a qualitative classroom-based study in which each essay received both peer and ChatGPT-4o feedback. A total of 430 peer comments and 224 ChatGPT feedback units were coded deductively using validated analytical frameworks. Inter-rater reliability for 20% of the dataset yielded substantial agreement (κ = .82). Results: Peer feedback was subjective, socially expressive, and frequently mitigated, with comments focusing on sentence-level issues but also including self-regulation prompts (15%) that encouraged reflection and decision-making. ChatGPT-4o provided predominantly task-level feedback (94%), characterized by structured, consistent, and objective guidance on grammar, organization, and coherence. However, its feedback exhibited minimal interpersonal variation and rarely promoted metacognitive engagement. Conclusion: The findings indicate that peer and AI-generated feedback serve complementary pedagogical functions: AI offers technical accuracy and consistency, while peer feedback contributes emotional support and occasional reflective prompts. A hybrid feedback model that integrates both sources may therefore enhance the revision process in EFL writing instruction.
Building similarity graph...
Analyzing shared references across papers
Loading...
Hossein Bozorgian
Hossein Mohammad Rahimi
SHILAP Revista de lepidopterología
Building similarity graph...
Analyzing shared references across papers
Loading...
Bozorgian et al. (Mon,) studied this question.
www.synapsesocial.com/papers/69a91d6dd6127c7a504c029c — DOI: https://doi.org/10.17323/jle.2025.27195