An Empirical Study Of A Common Reference Scale Of Contextual Inferencing Ability For NMEE

This thesis conducts an empirical study of the common reference scale of contextual inferencing ability for National Matriculation English Examination (NMEE). At the beginning of1970s, many European scholars like Van Ek spent20years building up a Common European Framework of the Reference for Languages. In2003, the Chinese Ministry also issued Common High School English Curriculum Standard with the purpose of making a "Can-do" statement from listening, speaking, reading and writing. However, the definitions and levels of "Can-do" statements are vague, and their classifications are overlapped, posing difficulties to the item-writing and evaluation process. The definitions of the "Can-do" statements of the contextual inferencing ability for NMEE are bound to be clear so as to enhance the testing quality of contextual inferencing ability and build up a common reference scale of contextual inferencing ability, especially its dimensionalities and parameters.This study, based on CLA and Discourse Information Cognitive Processing Ability, aims to study the following questions1) How about the reliability of the descriptors of contextual inferencing ability?2) In what dimensionalities are "Can-do" statements of contextual inferencing ability for NMEE described and assessed? The objects of my study are the volumes of NMET from1987to2013and the instrument employed is SPSS. Raters’judements consistency is employed to test the reliability of these descriptors. When assorting out the dimensions of descriptors, I employ the factor analysis. As to the date collection and analysis, a lot of descriptors in the reading comprehension and expression in the volumes of1987to2013are collected to attain the basic descriptors and the frequency of each descriptor, obtaining the statistics of the descriptors. Finally, all the data collected are put into the SPSS to test their reliability and to figure out how many parameters among these descriptors.According to the data results, we found that there are35descriptors concerning the "Can-do" statements o f contextual inferencing ability for NMEE and four dimensionalities contribute to the contextual inferencing ability, among which, character and incident make a big difference, text structure subordinate. This study is expected to inspire others in the similar researches. In addition, I hope the Chinese experts could summarize our experience in language teaching and combine with China’s actual conditions to carry out a basic research work in this field of language teaching and testing. Following basic theory in the development of common reference scale and learning from the extensive experience home and abroad, we are able to carry out a phased partial theory and practice and gradually realize the ultimate goal of formulating a common reference scale.