AI Voice-based Chatbots as a Conversational Partner in Practicing English Speaking: A Systematic Literature Review and the Ways Forward
TL;DR
A consistent trend in research design in this field with the predominance of quasi-experimental mixed-methods research is revealed, particularly in EFL contexts where the need to be proficient in speaking skills is substantial, yet the opportunities to communicate with a competent interlocutor are limited.
Abstract
The research on AI-mediated language learning has been growing across different educational contexts. Previous systematic reviews have summarized the existing studies on the affordances and challenges of implementing generative AI tools in language education. Some of these reviews covered a range of AI technologies, or Automatic Speech Recognition tools, which relied on pre-programmed speech. However, the reviews of studies focusing on the integration of AI voice-based chatbots (hereafter, AVCs) in speaking skills remain limited. This study reviewed 28 empirical studies published in English between 2020 and 2025, identified from Scopus, Web of Science, Google Scholar, using the PRISMA flowchart and predefined inclusion/ exclusion criteria. It aims to examine the trends of research designs in AI-mediated speaking instruction, speaking tasks between chatbots and learners, the impacts, challenges, and mediating factors of AVCs in speaking instruction. The study reveals a consistent trend in research design in this field with the predominance of quasi-experimental mixed-methods research, particularly in EFL contexts where the need to be proficient in speaking skills is substantial, yet the opportunities to communicate with a competent interlocutor are limited. The results also highlight positive impacts of AVCs on facilitating aspects of speaking proficiency, including fluency, pronunciation, vocabulary, and grammar. Several challenges related to technological issues, user experience, and users’ affective and behavioral engagement were identified in the selected sources. Mediating factors include learners’ language proficiency, affective factors, chatbot-related features, and pedagogical support. The article concludes with some implications and recommendations for diversifying and strengthening the evidence base, particularly in the teaching and learning of speaking skills. This review protocol (IDESR000187) was registered with the International Database of Education Systematic Reviews.
