Fundamental Considerations in Developing and Validating a Computer Adaptive Language Testing System

With positive prospects for the computer-based language testing (CBLT), the present research attempts to develop and validate a computer adaptive language testing (CALT) algorithm. First, the present paper is devoted to presenting how a CALT algorithm is developed through simulation and real data-based study. The paper reports fundamental considerations of an entire process of developing a CALT algorithm by accommodating such essential factors of CALT algorithm as 1) starting scheme, 2) item sampling, 3) estimation, 4) selection of an IRT model, 5) termination scheme based on item/test information function, and 6) score calibration based on item/test response time. Second, the research is intended to validate the item response time-based CALT algorithm through correlation studies on test-takers performance and the survey conducted to explore the degree of test fairness perceived by test-takers.