Higher education has gone through unprecedented emergencies and challenges in changing our instruction and assessment from the traditional face-to-face format to online. American universities and colleges are no exception. The pandemic has had a lasting impact on American higher education. With this background in mind, this paper will start by discussing how the pandemic has transformed our assessment and evaluation practices at higher education in the United States (U.S.) in general, and Japanese language testing in particular. Among the various assessment tools available for teaching Japanese, testing remains the most widely used assessment procedure in the U.S. This paper will raise issues concerning Japanese proficiency standards, criterion-referenced testing for classroom assessments, and grading. It will also discuss the issues related to some of the most influential Japanese proficiency tests in the U.S., i.e., the ACTFL OPI/OPIc, Advanced Placement (AP) exams, and the Nihongo Nōryoku Shiken (Japanese Language Proficiency Test [JLPT]). The paper will demonstrate that more research will be needed on Japanese language assessment and testing from multiple perspectives, such as validity, impact, practicality, and institutional support. It will conclude by providing a future research agenda for assessing Japanese in higher education.

This study examines a classroom-based second-language speaking test that uses picture description tasks and a rating scale of functional adequacy (FA) in terms of measurement quality. We used four tasks and FA scores from 36371 Japanese learners of English at two public senior high schools and 44 undergraduate and graduate students. Each examinee was evaluated by two or three raters. We used many-facet Rasch measurement to stack longitudinal data and analyzed the test scores in detail. We found that examinees, tasks, raters, and the FA rating scale worked well, with some concerns. For example, some examinees showed serious underfit, and large differences in rater severity were apparent. The bias analysis suggested that the percentages of biased patterns were small between examinees and tasks, between examinees and raters, and between tasks and raters. Based on our results, implications for assessment are provided, such as the need to conduct rigorous rater training.