Teacher feedback VS AI-assisted peer feedback in L2 writing: A quasi-experimental study in a Chinese University

Despite the growing interest in AI-assisted peer feedback, few studies have systematically compared it with teacher feedback within a sociocultural framework. Consequently, the dynamic evolution of these two modes, their effects on micro-linguistic features, and the possibility of integrated feedback models remain largely unexplored. To address these gaps, this study, grounded in Vygotsky’s theories of scaffolding and the Zone of Proximal Development (ZPD), adopted an 8-week quasi-experimental design to conduct a multi-dimensional comparison of teacher feedback and AI-assisted peer feedback among 61 university students (244 writing texts). The innovation lies in revealing, for the first time from a “heterogeneous scaffolds” perspective, the essential differences and complementary mechanisms between the two feedback types within a sociocultural framework. The main findings are: (1) Teacher feedback demonstrated a characteristic of “comprehensive coverage and dynamic adjustment,” whereas AI-assisted peer feedback exhibited features of “sustained focus and efficient guidance.” (2) Teacher feedback produced significant short-term improvement but insufficient subsequent momentum, while AI-assisted peer feedback displayed a stable “learning-curve effect.” (3) AI-assisted peer feedback enhanced lexical diversity only in the initial task, and neither approach significantly improved syntactic complexity. In terms of practical contributions, this study proposes concrete pathways for optimizing teacher feedback and improving AI-assisted peer feedback, and constructs a hybrid “AI-Peer-Teacher” integrated feedback model, offering an actionable reform proposal for English writing instruction in application-oriented universities.