Chatbot-Human Interaction and Its Effects on EFL Students’ L2 Speaking Performance and Anxiety

Abstract

Recently, chatbot interactions have been used for oral communication practice in the field of foreign language education. Some existing studies have highlighted the use of chatbots in relation to specific L2 skills, yet unfortunately, the user experience component of chatbot interaction has not been empirically researched. This study investigates the effect of chatbot-human interaction with the chatbot Replika on L2 speaking performance and speaking anxiety. The participants in the current study were 90 EFL students from a state university in Turkey. They used the application Replika for twelve weeks practicing targeted subjects outside the class. Each student was assessed with the L2 speaking anxiety scale both before and after the intervention. Also, an open-ended questionnaire was administered to collect their perceptions of using chatbots for L2 speaking practice. The interactions with the chatbot were screen taped for a randomly chosen speaking task: Well-being. The findings reveal that there were negative perceptions and attitudes toward the chatbot interaction. Students reported facing difficulties in being understood precisely, which might have contributed to higher anxiety in L2 speaking. Notably, student performance with Replika was significantly better than their face-to-face peer interactions. Overall, the results show that although chatbot interaction is a novel way to provide speaking practice for students, the actual interaction with a chatbot might not be a reliable way to lessen their anxiety with L2 speaking.