Abstract
Among different types of rating scales in scoring speaking performance, the EBB (Empirically derived, Binary-choice, Boundary-definition) scale is claimed to be easy to use and highly reliable (Turner & Upshur, 1996; 2002). However, it has been questioned whether the EBB scale can be applied to other tasks. Thus, in this study, an EBB scale was compared with an analytic scale in terms of validity, reliability, and practicality. Fifty-two EFL learners were asked to read and retell four stories in a semi-direct Story Retelling Speaking Test (SRST). Their performances were scored using these two rating scales, and then the scores were compared by using generalizability theory, a multitrait-multimethod approach, and a questionnaire delivered to the raters. As a result, the EBB scale, which consists of four criteria, was found to be more generalizable (i.e., reliable) than those of the analytic scale and generally assessed the intended constructs. However, the present EBB scale turned out to be less practical than the analytic scale due to its binary format and because it had more levels in each criterion. Further revisions seeking a better scale for the SRST are suggested.
