Assessing the Efficacy of AI-Powered Speaking Apps in China's Mainland Middle Schools: A Mixed-Methods Study on Oral Accuracy and Fluency

Abstract

English plays an increasingly crucial role in cross - cultural communication and education, especially for secondary students in China's Mainland, where oral proficiency has drawn significant attention due to high - stakes examinations like zhongkao. Conventional English language teaching methods face challenges such as public speaking anxiety, delayed feedback, and limited practice opportunities. AI - powered speaking tools emerge as a potential solution, integrating real - time feedback and adaptive learning technologies. For example, students can receive feedback or obtain suggestions instantly following their completion of speaking exercises. This mixed - methods study aims to evaluate the effectiveness of these tools in enhancing oral accuracy, such as pronunciation, and fluency among middle school students. A cohort of 32 ninth - grade students preparing for zhongkao participated in a structured questionnaire survey. The results show that 56% of the students considered AI tools beneficial for pronunciation refinement and fluency enhancement, although improvements in grammatical precision and lexical diversity were less evident. The main advantages highlighted by participants include technological affordances like instantaneous feedback from speech recognition algorithms and pedagogical flexibility like self - paced practice outside class. However, barriers such as technical limitations (device compatibility and network instability) and economic constraints (high subscription costs) also exist, which may impede sustained engagement.