Language Teaching Research Quarterly

The recent surge in the popularity of Large Language Models (LLM) for language assessment underscores the growing significance of cost-effective language evaluation in our increasingly digitalized society. This paper posits that the application of computational psychometrics can enable the incorporation of technology into language assessment, enhancing test accessibility for learners while simultaneously elevating the precision of language proficiency evaluation. In this context, computational psychometrics is defined as a fusion of theory-based psychometrics and data-driven methodologies drawn from machine learning, artificial intelligence, natural language processing, and data science. This amalgamation offers a more robust and adaptable framework for analyzing intricate data, particularly within the contemporary landscape of learner-centric assessment. The paper concludes by emphasizing that the integration of computational psychometrics into language assessment opens up promising avenues for future research and practical applications, heralding an era of innovation in this field.