Issues of Rating Scales in Speaking Performance Assessment

Abstract

As with any other area of language assessment, the fundamental issues to be considered in a speaking assessment are: (a) whether or not the test is used as intended, and (b) what its consequences may be (Bachman & Purpura, in press). To ensure that the uses and consequences of a speaking test are fair, the operational definition of speaking ability in the testing context should be examined, since the definition of speaking ability varies with respect to the targeted use and the decisions made. One way to elicit the construct of speaking ability for a certain context is through a scoring rubric which informs test users what a test aims to measure (Luoma, 2004). However, a scoring rubric can affect the speaking assessment, as there may be an interaction effect between the rating criteria and examinees’ performance (Luoma, 2004; McNamara, 1996). Different interpretations of the construct may cause biased effects on test-takers’ performance, leading to unfairness in scoring and test use. Thus, careful examination of how rating scales interact with speaking performance needs to be considered to determine the fairness of the speaking assessment.