ABSTRACT

 Objectives: Speaking proficiency remains a substantial obstacle for Chinese English as a Foreign Language (EFL) learners, particularly in areas with limited educational resources. This study explores the incorporation of ChatGPT, an artificial intelligence-based large language model, into the Production-Oriented Approach (POA) to improve the oral English competence of university students in Anhui Province, China. Methodology: Employing a mixed-methods approach, data were gathered from 240 students and 8 instructors to evaluate students' perceptions of ChatGPT's perceived ease of use and perceived usefulness throughout the motivating, enabling, and assessing stages of the POA. Additionally, the study assessed improvements in key speaking components, including fluency, accuracy, and pronunciation. Findings: The findings reveal notable advancements in speaking proficiency, with progress observed in fluency and accuracy, facilitated by personalised, interactive feedback and structured support. Interview data from instructors also indicated that ChatGPT contributed to lowering learners’ anxiety and enhancing their autonomy by offering a low-pressure conversational setting. These results suggest that the integration of AI-driven tools such as ChatGPT within established pedagogical models can effectively mitigate speaking challenges in EFL education, offering viable and scalable strategies for similar under-resourced educational environments.