The perception of ChatGPT-based feedback in Hindi writing classes: A case study of Korean university students majoring in Hindi

Abstract

This study explores how Korean university students majoring in Hindi perceive and use ChatGPT-based feedback in an intensive Hindi writing course. Using a mixed-methods design, the study surveyed 24 students and interviewed 8 participants who completed a 32-hour program at Hankuk University of Foreign Studies in June 2025. Learners rated AI feedback as highly effective in grammar (M = 4.33), spelling (M = 4.21), and sentence structure (M = 3.92), while cultural appropriateness received the lowest score (M = 2.79). Reported advantages included immediacy (95.8%), repeatability (79.2%), and reduced anxiety (66.7%), whereas key limitations were insufficient cultural understanding (83.3%) and restricted creativity (70.8%). Interview data supported these findings, indicating that students relied on AI for quick, low-stakes revisions but sought instructor feedback for tasks involving cultural nuance and emotional expression. The results suggest that while ChatGPT facilitates self-regulated learning in form-focused writing, it remains limited in meaning-making and identity-related dimensions. Therefore, a hybrid model combining AI tools and human instructors is recommended to support both linguistic accuracy and culturally appropriate expression. This study offers practical implications for integrating AI into foreign language learning and highlights the evolving role of educators in the age of generative AI.