A Study On The Validity Of Rational Cloze In NMET

The National English Test of the College Entrance Examination has made improvements in the type and content,which replaces traditional multiple-choice grammar test by rational cloze.So far,rational cloze has been in existence for 4 years.Is it better than multiple-choice grammar test? Does it have high validity? How can teachers and students improve grammar teaching or learning? We'll discuss all the questions.Based on validity theory and three sources of validation in Standards for Educational and Psychological Testing,the paper studies the validity of rational cloze of NMET from content of the test,the response process of the candidates and the internal structure of the questions.First of all,the paper analyzes the characteristics of contents and test points for the 2014-2017 rational cloze.It collects data from topics,styles,lengths,readabilities,and spatial distances.Moreover,we compare the test point levels and focus factors of rational cloze in 2014-2017 national tests according to Li Xiaoju's test validity theory.The results show that the topics of rational cloze are various by covering all themes of Curriculum Standards.There are narration,argumentation and exposition in genre.The degree of readability is about 65,so the difficulty is suitable for senior students in high school.However,the distances between gaps are quite different.For example,the spatial distance in 2015 is so short that affects the difficulty of the test.In terms of test points,the test level of four-year rational cloze increases every year.The test points at the sentence level have increased significantly,while the test sites at the phrase level have gradually decreased.The focus is mainly on grammatical factor,and the questions on the meaning factor have increased.Secondly,the validity evidence is collected from the reaction process by the candidate's verbal reports and strategy lists of rational cloze.Comparing candidate's test-taking processes with expected behaviors,we find that 97.7% candidates get the right answer by using expected knowledge and skills.According to the questionnaires of 3 groups with different level,81% candidates choose expected strategies during the test.However,the strategies of pre-test and post-test are quite different from expected strategies,so we'll analyze the reasons for the differences.Finally,the paper uses structural equation software AMOS 24.0 to build three structural models based on three hypotheses of grammatical competence.By comparing fit indexes of each model,we find that three-factor model with “form-meaning-use” is the best one,which can represent the internal structure of rational cloze.In a word,the research shows that the validity of 2014-2017 NMET rational cloze is good.It can examine the grammatical competence effectively under the framework of discourse.But some test points are still limited to the word or phrase level,so the function of discourse has not been made full use of.The study not only can give suggestions for teachers and students to improve grammar teaching and learning strategies,but also provide reference for test designers to improve the validity of rational cloze.