短语境阅读测试法的效度研究

Validation of Shortcontext Technique in Testing Reading Proficiency

在中国，阅读一直是英语教学的重点，也是各类英语考试中比重较大的部分，长期以来测试阅读的方式主要是要求考生读一篇文章，然后回答四至五个问题，通常是多项选择题，这种传统的阅读测试法已越来越多地受到语言测试专家的质疑。本文作者对另一种阅读测试法—短语境阅读测试法（ｓｈｏｒｔｃｏｎｔｅｘｔｔｅｃｈｎｉｑｕｅ）进行了实验，并作了效度研究，结果证明，这是一种行之有效的阅读测试法，在许多方面优于传统的阅读测试法。将短语境阅读测试法应用于计算机适应性考试，不仅提高了阅读理解测试的效率，同时也提高了测试的精确度。

ssessment of ESL / EFL reading proficiency has generally followed the procedures used for testing the reading competence of native speakers of English. These procedures reflect a pattern developed during the first quarter of the twentieth century and are usually organized around separate components including both reading and vocabulary comprehension. The reading comprehension component consists of short prose passages followed by sets of comprehension questions that test the examinee′s word knowledge, literal comprehension, ability to reason and make inferences, ability to follow the structure of the passage, ability to recognize the mood and literary techniques of the writer, etc. The vocabulary component, on the other hand, attempts to measure the quality and depth of a subject′s mastery of vocabulary.In constructing such measures the basic assumption has been that reading is made up of a number of skills. However, with no taxonomy of affective and cognitive subskills acceptable to the majority of experts, the precise number and type of skills that are tested have been dependent on the test writer′s choice of a model of the reading process.Over the past years, traditional reading tests have been subject to mounting criticism of various kinds, such as lack of construct validity, failure to sample relevant realworld reading behaviors, lack of variety, etc. This paper discusses some of the criticisms and the ways in which an alternative approach  the shortcontext technique  avoids these defects. The shortcontext technique is a method of testing reading comprehension through the use of brief stimulus materials  perhaps only one, two or three sentences  followed by one or two questions testing comprehension of overall meaning. Such a technique, in my opinion, avoids the overly detailed study of a passage to find small details and does not require highly inferential questions which may measure intelligence rather than control of basic reading skills and strategies. In addition, since a greater number of stimulus materials can be included, the variety of the contexts can be greater so as to reduce the chance that the contexts will be weighted to favor some examinees over others. Moreover, the limited number of comprehension questions per passage eliminates the possibility of the testee finding clues to the right answer from other test items. In the present study an attempt is made to examine the validity of the shortcontext technique in the testing of reading proficiency and to apply the technique to computerized adaptive testing for reading comprehension. Results have shown sufficient evidence substantiating the shortcontext technique as an authentic measure of reading ability and that it can be easily fitted into an adaptive procedure for the test of reading comprehension. Combination of the shortcontext technique with the traditional reading comprehension test may bring about higher proficiency and precision in computerized adaptive testing for reading comprehension.