Key points are not available for this paper at this time.
Purpose: Responding to pedagogical calls for reducing teacher feedback burden and empowering student feedback agency and considering the complex and dynamic nature of writing development, this study aims to longitudinally investigate the efficacy of exemplar-based feedback (EF) and Generative AI-generated feedback (AF), compared to traditional teacher feedback (TF), in driving L2 writing development, with gender as a critical yet underexplored predictor.Design/methodology/approach: A 12-week longitudinal study was conducted with 181 Chinese L2 learners, assigned to EF, AF, or TF conditions. Using latent growth curve analysis (LGCA), developmental trajectories in writing performance were modeled to examine initial proficiency, growth rates, and the predictive role of gender across feedback types.Findings: Results revealed that the overall writing proficiency displayed a significant upward trajectory, gender critically predicted the writing developmental patterns, with female students exhibiting higher initial level and accelerated growth rate than males and that the writing progress varied significantly by feedback resource, with AF yielding the most rapid growth, followed by EF, both surpassing TF in long-term effectiveness.Originality: This research pioneers in the use of LGCA to model longitudinal L2 writing development under different feedback modalities. By identifying AF and EF as more effective and sustainable alternatives to TF for sustainable L2 writing growth, it offers practical strategies for leveraging GAI tools and exemplars to create scalable and efficient feedback ecosystems. This synergy optimizes feedback pedagogical efficiency, alleviates instructor feedback burden and advances learner autonomy in L2 writing education.
Zhu et al. (Fri,) studied this question.