Study On The Cloze Test In DHU-Ⅱ

Cloze Procedure, invented' by W. L. Taylor in 1953 as a means of assessing readability of a text, has been broadly used in all levels of English examinations since 1970s as it is believed to be able to reflect test takers' overall language proficiency. Therefore, in 2002 when our university decided to organize DHU-II, a syllabus-based English test which is intended to evaluate the English proficiency of the non-English major students of our university, cloze was included as one of its test components. In order to find a cloze format which can serve our purpose, we conduct this study to look deeper into cloze testing.The study aims to better understand cloze testing as a good indicator of general linguistic ability and to decide, of the two cloze formats, SRD (Standard Rational Deletion Cloze) and FLG (a modification of the Standard Rational Deletion Cloze with the first letter of each deleted word given), which one should be used in DHU-II.By carrying out an experiment and conducting the questionnaire surveys, we study in great detail these two cloze formats, SRD and FLG andcollect data on their validity, reliability, practicality as well as backwash effects.The study reveals that compared with FLG, SRD is always problematic in its scoring. On the other hand, FLG, being reliable and valid, is much more practical in scoring which is a necessity in DHU-II as there are about 3000 students taking the test every time. It also has a better washback effect. Therefore, we think it is worthwhile to adopt FLG Cloze in DHU-II. But we also suggest that before a FLG cloze test is formally conducted, it is better to carry out a pre-test to ensure that each cloze item will measure effectively what it is intended to measure and it is quite desirable to increase the number of the cloze items from ten to twenty to achieve higher reliability.