A Comparative Analysis of AI-Scored Results in Computer-Based English Listening and Speaking Test (CELST) Across Four Cities in Guangdong, China
Qian Ouyang, Ruoxia Yu, Han Catherine Lin, Xiaobin Liu, Xiaozhen Li · 2024
Artificial Intelligence (AI) technology is instrumental in the collection and analysis of performance data in high-stakes examinations, providing exam administrators and educators with a thorough understanding of candidates' capabilities. This study leverages an AI scoring system to investigate the 2022 Computer-Based English Listening and Speaking Test (CELST) performance in Guangdong Province. It conducts a comparative analysis of English listening and speaking skills among candidates from four cities within a less developed educational region of Guangdong. By examining overall and individual item scores, and the reasons for score deductions, the study identifies performance disparities between these cities and the province. Additionally, it aims to offer targeted pedagogical recommendations to improve students' proficiency in listening and speaking, as well as effective strategies for CELST preparation.