أطراس
Volume 7, Numéro 2, Pages 80-97
2026-07-15
Effects Of Ai-powered Feedback On English Writing Development: A Comprehensive Systematic Review And Meta-analytic Synthesis
Authors : Rahamtalla Mohammed Taha Bashier .
Abstract
Background: The recent proliferation of AI-based technology, in particular, generative AI applications, like ChatGPT, or automated writing assessment (AWE) tools, like Grammarly, Pigai, and Criterion, have both changed how writing is taught in education. Nevertheless, the synthesized evidence on their effectiveness in comparison with traditional sources of feedback, in a variety of research environments, is limited. Objectives: This research is a systematic review of the existing empirical studies investigating the relevance of AI-based feedback systems in influencing the outcome of English writing activities among EFL/ESL students, in terms of the quality of the writing outcome, feedback literacy development, and self-regulated learning. Methods: We used Scopus, Web of science, ERIC, and PsycINFO to identify the publications published between January 2018 and December 2024. The inclusion criteria were peer reviewed experimental, quasi-experimental, or rigorous mixed-method studies of adult learners in institutions of higher learning. Results: Meta-analytic results from Zhai and Ma (2023) demonstrate that the overall effect of AWE on writing quality is large (g = 0.861, 95% CI [0.65, 1.07]), with medium-sized effects on language quality (g = 0.61) and content quality (g = 0.57). A recent meta-analysis by Sun and Zhou (2024) of 41 studies (N =)4,813 participants) revealed no significant difference between AI and human feedback (g = 0.25, 95% CI [-0.11, 0.60]). Deng et al. (2024) reported moderate but statistically significant impacts on academic performance (g = 0.53). Notably, hybrid feedback systems involving AI and teacher feedback yielded superior results compared to AI-only conditions. Conclusions: AI-based feedback is an encouraging addition to traditional writing teaching, the impact of which depends on the proficiency level of learners, the type of feedback and the duration of the intervention, as well as the pedagogical strategy. Future studies need to focus on long-term learning results, the formation of feedback literacy and the models of human-AI collaboration that are the most effective.
Keywords
AI feedback; automated writing evaluation; generative AI; ChatGPT; EFL writing; feedback literacy; self-regulated learning; meta-analysis
Les articles similaires
Abdaoui Mounya
.
pages 890-901.
Chettouh Farid
.
Hadji Abderrahmene
.
Bougandoura Fares
.
pages 253-277.
Zidani Soraya
.
Bahloul Amel
.
pages 341-358.
Elherrag Salah Eddine
.
Traoré Youssouf
.
Khaled Meghit Boumediene
.
pages 93-110.
