A Study of Rating Criteria for Speaking Assessment: Using Empirical Data of EFL Learners

The purpose of this study is to help establish an efficient speaking assessment tool for learners of English as a Foreign Language (EFL) by examining a degree of relevance of rating criteria employed and raters' perceptions of criteria for speaking. This study analyzed 11 native speaker raters' assessment on speaking performances of 22 Korean secondary school students participating in an English speech competition. Ratings were conducted based on the following 6 criteria: presentation, content, memorization/timing, fluency, grammar, and pronunciation. Findings exhibit first, relatively high correlations were found between the criteria and the final scores of students' performances except pronunciation. The most influential criteria in contributing to the students' overall speaking ability, however, were fluency and content. Also, there was a correlation between the final scores and scores the students obtained on a TOSEL proficiency test, adding improved validity of the criteria. Finally, raters' perceptions tend to be reflected in their ratings as the results of the questionnaire show fluency, content and grammar, pronunciation were perceived as important features. The results from the two analyses suggest that the criteria employed here are considered valid, while fluency and content may be more significant in determining learners' overall speaking ability. Finally, pronunciation did not turn out to be a strong indicator in any of the student performances. The reason for this could be explained in part by lack of detailed and explicit standards for pronunciation.