Improving EFL speaking performance among undergraduate students with an AI-powered mobile app in after-class assignments: an empirical investigation

Abstract

English speaking represents one of the most challenging competencies for EFL learners, mainly due to the limited opportunities for authentic practice, especially within monolingual contexts like China. However, the ubiquitous personal mobile devices (smartphones) and the advent of AI-powered mobile apps equipped with automatic speech recognition, natural language processing, and text-to-speech present novel solutions to overcome these hurdles. This study investigates the effects of an AI-powered mobile application (Liulishuo) on Chinese undergraduate EFL students’ speaking performance, using features such as automatic feedback, process-oriented monitoring, and tailored instructions. Employing a quasi-experimental design, this 10-week study was conducted at a Chinese university with two groups of participants. The control group (n = 31) engaged with WeChat for after-class assignments, whereas the experimental group (n = 32) supplemented WeChat use with Liulishuo, diversifying their educational tools. Results indicated that participants in the experimental group significantly outperformed those in the control group regarding overall speaking performance. More specifically, notable improvements were observed in pronunciation and fluency, while vocabulary and grammar enhancements were not statistically significant. These findings underscore the capacity of AI-powered mobile apps to enhance EFL speaking performance, signaling novel pathways for the evolution of language education.