A Corpus-Based Discourse Analysis of Chabot-Learner Interactions in Second Language Acquisition

This study explores how intermediate English learners interact with AI-powered chatbots like ChatGPT during open-ended conversations. Using a corpus-based approach, we analyzed the structure and quality of dialogues between ۳۰ Iranian learners and the chatbot, focusing on how learners take turns, self-repair, and respond to the chatbot’s replies coherently. Learners completed various real-life conversation tasks. These tasks included discussions, problem-solving, and information exchange. Results showed that most learners actively participated in the dialogue, initiated turns, and sometimes corrected themselves. The chatbot mostly responded clearly and smoothly, though it rarely offered direct corrections. Overall, the findings suggest that chatbots can support meaningful language practice, increase learner confidence, and encourage self-monitoring. However, for more language development, chatbots may still need better ways to offer feedback and handle complex conversations.

کلیدواژه ها:

chatbot ، corpus ، turn-taking ، self-repair ، coherence ، task