外语测试去污染信度估计研究——基于语料库视角  

On Estimating Parallel Reliability of Foreign Language Tests Based on Unpolluted Evidence——From the Perspective of Corpora

作　　者：詹先君[1] 刘玲[1] 

机构地区：[1]安徽科技学院外国语学院,安徽凤阳233100

出　　处：《西安外国语大学学报》2014年第1期79-82,共4页Journal of Xi’an International Studies University

基　　金：安徽省教育厅重点教研项目"应用型高校外语教师教育评价能力发展研究"(项目编号:2012jyxm402);安徽科技学院校级教研项目"基于支架理论的应用型大学英语口语教学设计研究"(项目编号:x2012074)的阶段性成果

摘　　要：既有的外语测试信度估计方法大多基于测试分数,属测后估计,信度证据容易受到无关因素的污染,估计结果并不可靠;而基于语料库的信度估计方法则是基于测试文本,属测前估计,信度证据不受其它因素污染,估计结果较为可靠;通过对我国权威性外语测试考研英语阅读理解试题进行横向和纵向复本信度估计,能够证明这种基于语料库的去污染信度估计方法的可靠性和可行性。本研究的价值在于从语料库语言学视角为外语测试的信度估计开拓新途径。The present reliability estimating methods, using the test score as reliability evidence which is easily polluted by irrele- vant factors, is a post-test estimation and thus can hardly obtain reliable results; the reliability estimating method from the perspec- tive of corpora, using the test text as reliability evidence which is unpolluted, is pre-test estimation and can get reliable results. This paper, through estimating the parallel reliability horizontally and vertically by employing corpora linguistic method, verifies that reli- ability estimation using the corpora linguistic method is valid and practical. The value of this study is to blaze new paths for reliabili- ty estimation of foreign language tests from the corpora linguistic perspective.

关 键 词：外语测试 去污染 信度估计 

分 类 号：H319[语言文字—英语]