A Comparative Study On The Measuring Capacity Of Three Types Of Cloze Test

As a convenient and effective test method in measuring testees' a wide range of language competence,the cloze test is a test method extensively used in many large-scale language tests in China.However,little attention is given to the study of the cloze test in China.Moreover,fewer researchers have touched upon the comparative study of the three types of cloze test, Multiple-choice,Open cloze and Banked cloze.Therefore,little is known about the measuring capacity of the three cloze formats,let alone the appropriate use of cloze format to elicit the reliable and valid measurement results from the students.This paper conducts a comparative study on the measuring capacity of the three types of cloze test where the comparison variables of difficulty,discriminability,reliability and validity were calculated and analyzed based on the respective statistic techniques on language testing cited from the book A guide to language testing:development,evaluation and research proposed by Grant Henning.And the construct validity is obtained on the basis of the introspective technique proposed by McDonough and the study of"four level"context of cloze stated by Bachman.The present paper aims to answer the following three questions:1)Are test-takers' performances different when different test formats of cloze are utilized?2)Are there any variances in difficulties,discriminabilities,validities and reliabilities of multiple-choice cloze, open cloze and banked cloze?3)Which kind of cloze test can yield the most effective and reliable measurement of testees' overall language competence?To answer the above questions,the carefully designed cloze tests including three cloze formats were administered to 30 junior students from Qufu Normal University.Their scores of test are input into computer for statistical analysis with Excel and SPSS(version 14.0).Then a questionnaire and a retrospective report on the three cloze formats were conducted,aiming to inquire the face validity and construct validity of the cloze test.The findings from the research are drawn as follows:Firstly,the test-takers have significant difference,on the results of the three types of cloze test,that the test takers perform much better on the multiple-choice cloze while having the worst performance on open cloze.Secondly,based on the comparison of the variables of difficulty,discriminability,reliability and validity,it can be concluded that both open cloze and banked cloze have the desirable measuring capacity with strong discriminability power,high reliability and validity,while the most frequently used multiple-choice cloze does not perform a satisfied result on all of the comparison variables. Thirdly,open cloze yield the most effective and reliable results in measuring the test-takers' overall language proficiency.Therefore,based on the above results,the researcher suggests that test designers should choose appropriate cloze type in accordance with their different test purpose and use and multiple-choice cloze should not be used alone in a large-scale proficiency or selection test; instead,it can produce better results when combined with other testing instruments,especially the open cloze format.The research draws conducive implications about the language testing.However,due to constraints of the experiment conditions,it applies a small size of sample,does little inquiry on the washback effect of the cloze test and covers only the variable of cloze format,all of which need to be further improved.