Earticle

The purpose of this study is to investigate middle school English teachers' rater reliability in English speaking tests operated at schools. Eight raters who participated in this study all work at middle schools. Three of them are native speakers of English and have at least one year long educational experience at Korean middle schools, and five raters are middle school Korean English teachers with different educational experiences. The Korean English teachers are divided into two groups of ‘high’ English speaking competence and ‘medium’. All of them rated eighteen test takers' utterances, who are second year at a middle school, and their rating patterns were analyzed with FACETS, a computerized program of many-faceted Rasch model. This study also interviewed the raters to make it possible to identify the causes of rater variability and rater errors in the procedures of rating, and the actual conditions of English speaking tests at a middle school