This thesis analyzes a new approach to the application of computers in language testing-- Computer-Adaptive Testing (CAT). Computer-based English Language Testing (CBELT) and Computer-Adaptive Testing (CAT) are defined, and their relationship explained--CAT is a subcategory of CBELT, but more precise and efficient. Chapter One also introduces the development of CAT and the different approaches of its critics and proponents in testing field. Chapter Two describes the basic principles of assessment embodied in the CAT-- Item Response Theory (IRT) and Item Banking, and explores their application, relationship, advantages and disadvantages. Although still in its infancy, CAT is expanding very rapidly and has changed markedly over the past few years. Chapter Three introduces the application of CAT, paying particular attention to the recent research at Capital Normal University in Beijing. Chapter Four examines the strengths and weaknesses of CAT, especially the research findings in the earlier chapter. It highli
ghts the potential, the success, and pitfalls of this new educational medium, which are peculiar to the CAT as opposed to traditional paper-and-pencil tests, and concludes that with the decrease of computer cost, the sophistication of software and the increase in computer literacy, more advantages of CAT will surface, and some objections will begin to diminish. Chapter Five attempts to forecast some of the directions the future research on computers in language testing might take, the role of computer in language testing in this new century, and suggests the adoption of a combined English test including CAT and paper-and-pencil items. It also discusses the issues developers need to consider when designing L2 CATs in China. Chapter Seven concludes the revolutionary nature of the adoption and development of computers in language education.