Investigating the effectiveness of corpus and AI integration in developing Chinese EFL university students’ speaking proficiency and attitudes

This study aims to construct a Corpus and AI-aided self-directed English-speaking training programme, evaluate the effectiveness of this approach, and assess English learners' attitudes toward the application of this self-directed English-speaking training approach. 31 English as a Foreign Language (EFL) learners participated in a pre-test, a five-session Corpus and AI-aided English-speaking training, a post-test, a self-reflection, a survey, and a follow-up interview. Participants' overall speaking performances and performances on the four subskills (fluency and coherence [FC], lexical resource [LR], grammatical range and accuracy [GA], and pronunciation [PN]) improved after the training with the greatest progress in LR, while the least progress in PN. Participants agreed that the integration of spoken corpus helped them to identify the linguistic features which may influence English oral proficiency and the use of AI tools, including ChatGPT and Murf, in English speaking training was effective. Participants agreed that the integration of spoken corpus helped identify the linguistic features which may influence English oral proficiency and the use of AI tools was effective. Interactive AI tools also helped create an interactive self-directed speaking learning environment. The majority of participants expressed positive attitudes toward the application of Corpus and AI-aided speaking training approach and showed willingness to interact with the interactive AI tools even though the tools have some limitations.