in: Classroom assessment for language teaching: challenges, choices, and consequences, Tahnee Bucher and Eddy White, Editor, Cambridge Scholars Publishing, Newcastle Upon Tyne, pp.60-100, 2020
Although classroom assessment (CA) is considered low-stakes, assessments designed, executed and assessed by class teachers are used to make important educational decisions, and adopting a thorough approach to CA is vital to achieve positive educational outcomes. This study documents the development of a text discussion task intended to be used in an English for Academic Purposes course. Students are expected to discuss an assigned text in small groups. The process of construct definition, task design, and rating scale development are explained providing a replicable model for designing similar assessments. Through ground analysis of the transcriptions of four text discussions administrated over the course of the term, skills, subskills and language functions operationalized through the task were investigated, and features that describe and distinguish levels of performance were identified. Additionally, the role of group dynamics in shaping individual’s and group’s performance was explored. The findings underpinned the co-constructed nature of group performance and highlighted the role of interactional competence in group discussions. They also revealed construct-irrelevant factors influencing test takers’ performance posing a threat to the validity, reliability and fairness of the test method. Implementations of the findings are discussed as well as recommendations for dealing with validity, reliability and fairness concerns.