论英语口语考试的评分误差

口语考试的评分是评分员基于评分标准对语言产出的认知处理过程,处理的目的就是解释考生之间的分数差异(score vari-ance)。用于解释分数差异的变量包括构念相关变量(construct-rele-vant variables)和构念不相关变量(construct-irrelevant variables)。如果构念不相关变量发生作用,那么评分就产生误差。考试误差可区分为系统性误差(systematic error)和随机性误差(randomerror)。随机性误差是评分误差控制的重点内容。口语考试评分误差的主要表现形式包括评分员的个性差异、回归均值趋势和假正态分布。我们可以通过分数差异分布和回归系数等统计手段验证口语考试评分误差的大小程度。本文还讨论了口语考试评分误差控制的目标、原则和方法。评估误差控制的目的就是最大化构念相关变量的作用,最小化构念不相关变量的影响作用;这就要求评分员在评分过程中坚持一致性、完整性和独立性三条基本原则;在手段的使用方面,口语考试的评分误差控制主要包括管理手段、技术手段和统计手段等。

On the Scoring Errors and Their Control in the English Speaking Tests

Rating an English speaking test is a psychological process, in which the raters match the rating criteria with each testee's language output to score the testee's performance in the English speaking test. The purpose of the rating process is to explain the score variance among testees, and the variables, which may contribute to such explanation, include construct-irrelevant variables and construct-irrelevant ones. The construct-irrelevant variables are the main sources of the scoring errors, which consist of systematic errors and random errors.Random errors are the focus of the scoring error control. In practice,the scoring errors are represented in the forms of individual difference among raters,mean score tendency and false normal distribution. Two hypotheses are proposed in the present paper to test the existence of the scoring errors:Score Difference Distribution Hypothesis and Regression Index Hypothesis.The present paper also deals with the purpose, principles and approaches concerning the scoring error control.The purpose is to maximize the effect of the construct-relevant variables and minimize the effect of the construct-irrelevant variables. For the purpose of the scoring error control, three principles are established for the raters:consistency, completeness and independence. Management, technology and statistics are the three commonly-used approaches to control the scoring errors in the English speaking tests.