The influence of large language models as collaborative dialogue partners on EFL English oral proficiency and foreign language anxiety

Abstract

Listen

Translate article

Collaborative dialogue is an important method for improving English as a Foreign Language (EFL) oral proficiency and reducing foreign language anxiety, but it is often influenced by the proficiency level of peers and whether they are native English speakers. With the support of artificial intelligence and automatic speech recognition language synthesis technology, Large Language Models (LLMs) have the potential to enhance EFL oral proficiency and reduce foreign language anxiety. However, there is limited empirical research on the impact of Large Language Models on EFL oral proficiency and foreign language anxiety. This study recruited 83 freshmen from China to participate in a 4-week LLM-assisted English oral proficiency learning activity, employing a randomized controlled trial. Participants’ oral proficiency, foreign language anxiety, willingness to communicate in English, and self-efficacy were assessed before and after the intervention. The results indicated that the LLM (GPT-4) significantly enhanced their oral proficiency. Furthermore, it significantly increased participants’ willingness to communicate orally in English and their self-efficacy, while significantly reducing foreign language anxiety. Semi-structured interview analysis revealed learners’ positive attitudes towards using LLM and their willingness to continue using them in the future but also highlighted occasional deviations from topics and feedback delays due to network speed during conversations, which influenced participants’ experiences to some extent.