Exploring the effectiveness of ChatGPT-based feedback compared with teacher feedback and self-feedback: Evidence from Chinese to English translation
Project Overview
The document assesses the role of generative AI, specifically ChatGPT, in enhancing education by focusing on its application in providing feedback to advanced ESL/EFL learners engaged in Chinese to English translation. It contrasts the effectiveness of ChatGPT-based feedback with traditional teacher feedback (TF) and self-feedback (SF). The findings reveal that while ChatGPT offers immediate and tailored feedback and excels in enhancing lexical capabilities, it falls short compared to TF and SF in improving overall translation quality and syntax skills. This suggests that, although ChatGPT can serve as a valuable supplementary tool in the learning process, traditional feedback mechanisms remain crucial for achieving comprehensive language proficiency. The study underscores the necessity of integrating generative AI in educational settings while maintaining a balance with conventional teaching methods to optimize learning outcomes.
Key Applications
ChatGPT-based feedback for ESL/EFL learners
Context: Advanced ESL/EFL learners, specifically Master of Translation and Interpretation (MTI) students in China
Implementation: Students received feedback from ChatGPT after submitting their translation drafts, which was compared to teacher feedback and self-feedback.
Outcomes: ChatGPT-based feedback improved lexical capabilities but was less effective in overall translation quality compared to TF and SF.
Challenges: ChatGPT struggled with syntax-related skills and could not consistently identify errors, especially in culturally sensitive translations.
Implementation Barriers
Effectiveness barrier
ChatGPT-based feedback was less effective than traditional feedback methods in improving overall translation quality and syntax-related skills.
Proposed Solutions: Integrating ChatGPT as a supplementary tool alongside traditional feedback methods.
Consistency and cultural sensitivity barrier
ChatGPT's feedback could be inconsistent across different student translations and often failed to detect errors in culturally sensitive translations, leading to unreliable guidance and inaccuracies.
Proposed Solutions: Improving ChatGPT's training data and feedback algorithms to enhance consistency, and incorporating more culturally aware training data to refine the model's understanding of cultural nuances.
Project Team
Siyi Cao
Researcher
Linping Zhong
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Siyi Cao, Linping Zhong
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai