EFL students' self-directed learning of conversation skills with AI chatbots

This study investigates the potential of AI chatbots as practice partners in the context of EFL secondary school students’ self-directed learning of conversation skills. Using the Dialogflow CX Console at Google Cloud Platform, we designed two sets of AI chatbots based on two target dialogues. A total of 112 eighth-graders at a public middle school in South Korea were divided into two experimental groups: a chatbot group and a worksheet group. The two groups practiced the target dialogues in self-directed activities with different tools, i.e., chatbots versus worksheets, and evaluated the quality of the chatbot- or worksheet-based self-directed activities in a questionnaire. In addition, their improvements in conversation skills and self-directedness were measured by a discourse completion task and another questionnaire in pre- and posttests. Results showed that the chatbot activities were evaluated to be better than the worksheet activities, especially in terms of entertainment. We also found that the chatbot group showed greater improvements both in conversation skills and self-directedness than the worksheet group. Based on these findings and qualitative analyses of dialogue records, implications for using AI chatbots in the self-directed learning of English are discussed.