| Based on the theoretical researches on historical development of language testing as well as some key concepts in the field of language testing, the author finds that a good test is the one that exploits merits of the four main approaches to testing, especially of the most advanced one—communicative testing, and besides, a good test can reach high validity, high reliability, beneficial backwash, appropriate practicality and certain discrimination as well.The research reported in this thesis is concerned with testing of reading for College English Test (CET). It is found that reading is a communicative and interactive process in nature, and CET, as an achievement-proficiency test, calls for communicative testing; thereby it is necessary that the reading part of CET should take a communicative testing approach. The author proposes five features for a communicative reading test: 1) choosing a variety of genres and topics; 2) establishing real-life purposes; 3) testing different reading skills; 4) adopting diverse testing techniques; and 5) keeping or mimicking authentic formats. However, according to an analysis of CET test papers from 1995 to 2002, the author finds that there is a significant gap between the CET reading test and the communicative reading test. Thereby, the reading tasks of the CET need improvement and reform. This viewpoint is confirmed through questionnaire surveys and interviews among 300 college students and some college English teachers. It is found that the current CET reading tests do exert certain harmful backwash effect upon learning and teaching of college English. To verify whether the communicative reading tasks can be more beneficial to candidates, the author carried out an experiment among students of two classes. The experimental class was regularly given communicative reading tasks, whereas the control class was given regular reading exercises that were typical of the current CET reading tasks. After this kind of distinctive reading training for two semesters, the students from the experimental class, compared with those from the control class, demonstrated a better reading ability in addition to a higher motivation in English learning through reading.Then the author proposes that re-consideration should be given to the CET reading test. First, in the selection of the reading materials, text genres can be enriched, and some authentic formats of texts, such as titles, typesetting and illustrations, can be kept. Second, in the design of the reading tasks, real-life reading purposes are proposed to be established especially for some practical genres, and testing techniques should be diversified. Thus, CET can make a greater contribution to the advance of the English teaching and learning in China. |