Using AI Chatbots as Speaking Partners: Benefits, Limitations, and Learner Perceptions

The integration of artificial intelligence chatbots as conversational partners in second language speaking instruction has gained considerable attention in recent years, driven by advances in natural language processing and automatic speech recognition technologies. This systematic review examines the pedagogical affordances, technical and instructional limitations, and learner perceptions of AI chatbots for speaking practice in English as a Foreign Language contexts. Drawing on 30 empirical studies and systematic reviews published between 2018 and 2025, this paper synthesizes evidence regarding the benefits of chatbot-mediated speaking practice, including increased practice opportunities, reduced speaking anxiety, enhanced learner engagement, and personalized learning experiences. However, the review also identifies persistent technical constraints---particularly speech recognition errors, limited contextual coherence, and inadequate pragmatic competence---alongside pedagogical challenges such as variable feedback quality, curriculum misalignment, and insufficient teacher preparation. Learner perceptions reveal a complex landscape: while students appreciate the accessibility, convenience, and low-pressure environment chatbots provide, they express frustration with unnatural interactions, repetitive responses, and limited corrective feedback. The paper situates these findings within established second language acquisition frameworks, including the Interaction Hypothesis, Affective Filter Hypothesis, and Sociocultural Theory, and discusses implications for pedagogical practice, teacher professional development, and future research directions. This review concludes that AI chatbots represent a valuable supplementary tool for speaking practice when thoughtfully integrated into curricula, but cannot replace human interaction in developing comprehensive communicative competence.