Dies ist eine Übersichtsseite mit Metadaten zu dieser wissenschaftlichen Arbeit. Der vollständige Artikel ist beim Verlag verfügbar.
CHATGPT’S AND HUMANS’ WRITTEN CORRECTIVE FEEDBACK IN EFL/ESL WRITING: A SYSTEMATIC REVIEW
0
Zitationen
1
Autoren
2026
Jahr
Abstract
Written corrective feedback (WCF) is critical in EFL/ESL writing classrooms. This study aims to review existing research studies on the quality of WCF generated by ChatGPT and humans and to examine how WCF generated by ChatGPT can support WCF that humans give to students’ written work in higher education EFL/ESL writing contexts. To meet these goals, the researcher examined 23 studies published in 18 peer-reviewed journals from 2023-2024. The analysis results highlighted potential strengths and weaknesses of WCF generated by ChatGPT and humans. WCF generated by ChatGPT could review various aspects of writing and provide extensive descriptive feedback, yet it sometimes might experience system fatigue that affects the WCF quality. ChatGPT could also provide inaccurate WCF, give WCF that lacked human nuances, and were greatly influenced by how prompts were presented to it. Meanwhile, humans could provide aspects of WCF not accounted for ChatGPT, but their WCF might involve subjective judgment and be unclear. With these findings in mind, the researcher identified three phases of how ChatGPT can work harmoniously with humans to provide quality WCF for students in writing classrooms. Those phases were making various efforts to enhance WCF generated by ChatGPT in students’ writing in support of the feedback provided by lecturers, enhancing students’ in-depth understanding that not all ChatGPT feedback is accurate, and combining ChatGPT-based WCF, specifically on earlier drafts of students’ writing. Pedagogical implications, which highlighted the essence of working synergistically with ChatGPT and mastering prompts to generate the most helpful WCF, for EFL/ESL lecturers in writing classrooms who wish to support their feedback with ChatGPT to provide quality WCF to their students’ written work were presented. The researcher then proposed recommendations for future systematic review studies, expanding the research contexts, the scope of reviewed materials, and the research method of the reviewed articles, to address gaps and expand on the insights presented in this study.
Ähnliche Arbeiten
Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI
2019 · 8.231 Zit.
Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead
2019 · 8.084 Zit.
High-performance medicine: the convergence of human and artificial intelligence
2018 · 7.444 Zit.
Proceedings of the 19th International Joint Conference on Artificial Intelligence
2005 · 5.776 Zit.
Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)
2018 · 5.423 Zit.