In the former half of the upper level Japanese class, for which the authors are in charge, eight instructors are teaching four classes, managed under the same syllabus. The report assignment occupied forty-five percent of the class activities, therefore, the fair and equal evaluation by several instructors are necessary. For these reasons, the researchers created the referenceable rubric for both instructors and students and adjusted each evaluation point. As the result, we conclude that differences in inter-rater reliability is caused from differing interpretations of descriptions and vagueness in some parts of the rubric. Further the rubric's reliability can be iteratively improved upon throughout its use and the vagueness removed and clarified.
