CONVERSATIONAL AI FOR SPOKEN ENGLISH PRACTICE:

Abstract

As more human interaction is exploding with AI machines, a humongous gate has opened for second language (L2) speaking practice, yet the pedagogical realities of implementing AI chatbots in actual classrooms remain underexplored. This case study investigates the integration of a customized AI chatbot into spoken English instruction for 82 male elementary and intermediate learners at a Syrian university. Drawing on Jeremy Harmer's (2015) pedagogical framework for speaking instruction and contemporary second language acquisition theory, the study examines both learner outcomes and pedagogical implications of AI chatbot use over an eight-week period. As the classroom teacher, I implemented random AI chatbots as a supplementary speaking practice tool, observing and documenting learner engagement, progress, and challenges.

Findings reveal a rough 30% overall improvement in speaking proficiency, with a clear distinction between learner levels: intermediate students demonstrated some gains, communicating effectively with the AI chatbot and benefiting from independent practice. However, elementary students struggled significantly to communicate with the AI chatbot, requiring extensive teacher guidance and mediation to benefit from the tool. The study proposes pedagogical guidelines for responsible AI integration, emphasizing differentiated approaches based on proficiency level, structured teacher mediation for lower-level learners, and the complementary roles of AI and human instruction. Ethical considerations including data security and the risk of replacing rather than augmenting human instruction are addressed. The study concludes with practical recommendations for language educators seeking to integrate AI chatbots into their teaching practice.