Talking to the Future: The Impact of Conversational AI on Adolescent EFL Speaking Proficiency

Abstract

Speaking is one of the most demanding dimensions of English as a Foreign Language (EFL) learning, yet opportunities for sustained oral production in secondary classrooms are frequently limited by instructional time, class size, and learners' speaking anxiety. To address these challenges, this study investigated the effectiveness of Langua, an artificial intelligence-based conversational tool, in developing the English-speaking skills of tenth-grade students. Employing a quantitative quasi-experimental pretest-posttest control-group design, the research examined 280 students across four secondary educational institutions in Riobamba, Ecuador. Each institution included an experimental group ( ) that integrated Langua into regular instruction for 10 weeks, and a control group ( ) that followed conventional instruction. Speaking proficiency was assessed using the Speaking component of the Cambridge A2 Key (KET) examination. Analysis of covariance (ANCOVA), controlling for baseline performance and institutional differences, revealed a statistically significant effect of the treatment ( ), with the experimental group demonstrating a substantially higher mean gain ( points) than the control group ( points). These findings indicate that conversational AI, when pedagogically integrated alongside teacher guidance, can serve as an effective complementary environment for repeated oral practice. Furthermore, this research extends the emerging literature on AI-assisted language learning by providing empirical evidence from an underrepresented adolescent population and Latin American educational context.