中国英语能力等级量表之听力能力子量表的构建

发布时间：2018-01-14 04:13

本文关键词：中国英语能力等级量表之听力能力子量表的构建　出处：《浙江大学》2017年硕士论文　论文类型：学位论文

【摘要】：中国英语学习者众、英语考试繁多,为了应对全球化带来的挑战、深化我国的教育改革,建设标准统一的外语能力测评体系迫在眉睫。制定中国英语能力等级量表是我国"外语能力测评体系"建设中首要和最为基础的环节。听力能力子量表是中国英语能力等级量表中不可或缺的一个部分。本文系统介绍了听力能力子量表的构建,包括理论基础、框架构建、描述语库的建立和描述语的效度验证和量表化等,同时以现有的部分数据(5、6级别)为例,采用IRT模型之一的多层面Rasch模型进行了简单的实证分析。本研究首先构建了基于运用的听力能力模型,接着通过文献法和采样法建起了听力描述语库并对库中描述语进行了多轮修改,最后通过直觉法、定性法、定量法相结合的方法对听力描述语进行量表化。基于现有数据的Rasch分析显示,这两个级别共239条描述语质量较好,模型拟合度较高,难度和区分度都较为合理。但同时也暴露了一些问题,如这两个级别描述语中有个别分级不合理,有待全部数据采集、分析后作进一步调整。
[Abstract]:There are many English learners in China and there are many English tests. In order to meet the challenges brought by globalization, we should deepen the educational reform in our country. It is urgent to build a unified foreign language proficiency assessment system. The establishment of the China English proficiency scale is a "foreign language proficiency assessment system" in China. The listening ability subscale is an indispensable part of the Chinese English proficiency scale. This paper systematically introduces the construction of the listening ability subscale. It includes theoretical basis, frame construction, establishment of description library, validity verification and measurement of description language, and so on. At the same time, it takes some existing data as an example. A simple empirical analysis is carried out by using a multi-level Rasch model, one of the IRT models. Firstly, a listening ability model based on application is constructed in this study. Then, the listening description library is constructed by literature and sampling methods, and the description in the library is modified in several rounds. Finally, the qualitative method and intuitionistic method are adopted. The Rasch analysis based on the available data shows that the quality of the two levels of 239 descriptors is better and the model fit is higher. The degree of difficulty and differentiation are both reasonable, but some problems are also exposed, such as the unreasonable individual classification in these two level descriptors, which need to be collected from all the data and further adjusted after analysis.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：H319.3

【参考文献】