THE IMPACT OF AI-ASSISTED FEEDBACK ON ESL STUDENTS’ GRAMMAR LEARNING: A PRE- AND POST-TEST STUDY

Abstract

This research study investigates the impact of AI assisted diagnostic feedbacks (using ChatGPT) on the grammar identification skills of English as a second language learners (aged 8-9) in grade 3.Recognizing the challenges faced by ESL learners in accurately identification of grammatical elements (nouns, adverbs and verb forms) and the practical limitations of providing individualized real time feedback in traditional classrooms, this study addresses the critical gap by comparing AI generated feedback with traditional assessment. The researchers have conducted mixed method research using Quasi- experimental research. The study involves selecting 50 students using non probability sampling. Single group pre-test/intervention/post-test design was employed over a two-week period. Initial performance on 15-item identification resulted in a mean score of M=7.20(48% accuracy). During the intervention, students received real time, diagnostic AI feedback on their errors, along with corrective explanations and reinforced learning. Post-test results demonstrated a significant improvement, with the mean increasing to M = 12.54 (84% accuracy) and an average gain of +5.34 points. Statistical analysis shows the significance of this improvement (t = 13.32, p < .05), with a large effect size (Cohen’s d = 1.44). Qualitative findings revealed that AI feedback was especially effective in clarifying complex grammatical categories such as manner adverbs, verb forms, and uncountable nouns. The study concludes that AI-assisted feedback can serve as a highly effective tool for enhancing foundational grammar skills among young ESL learners. However, concerns about excessive reliance on AI and lack of emotional nuance still exist.