The large-scale, authentic assessment of listening and speaking as interactive communication: Issues in reliability

In this article we discuss reliability issues in the large-scale assessment of speech communication through authentic or performance techniques used recently in Saskatchewan. New performance-based approaches enable educators to evaluate the integrated, interpersonal communication skills of large populations of students, thereby modeling best professional practice. We conclude, however, that decentralized teacher rating approaches do not yet demonstrate sufficient reliability to enable public officials to use the results for high-stakes testing purposes.