Cognitive Diagnostic Assessment (CDA) distinguishes itself with finer-grained attributes such as knowledge structures and processing skills underlying learners' responses. Dynamic Assessment (DA) can integrate assessment and mediation into one unified activity, making it possible to track how learners' potential abilities emerge and develop through graduated mediation. In order to be more valid in diagnosing and mediating EFL listening, the mediation moves are advised to match the task attributes. The combination of the above two holds a great potential in promoting students' language ability. Therefore, this study attempts to construct the Attribute-based Mediation Model (AMM) for English listening.
"Cognitive Diagnostic Assessment (CDA) distinguishes itself in identifying learners' cognitive strengths and weaknesses with finer-grained attributes such as knowledge structures and processing skills underlying learners' responses. Dynamic Assessment (DA) can integrate assessment and mediation into one unified activity，making it possible to track how learners' potential abilities emerge and develop through graduated mediation (Poehner, 2015). In order to be more valid in diagnosing and mediating EFL listening, however, the mediation moves are advised to match the task constructs/attributes. The combination of the above two holds a great potential in promoting students' language ability. Therefore, this study attempts to construct the Attribute-based Mediation Model (AMM) for English listening, and verify its effectiveness.The EFL listening cognitive attributes for mediation were first determined and the diagnostic test was constructed through learners' verbal protocols and experts' judgment. Second, a four-step attribute-based mediation model (AMM) was established based on the interactionist DA. For model validation, three rounds of one-to-one dynamic assessment experiments were conducted on 8 lower-intermediate freshmen every other week. Afterwards, the subjects and the mediators were interviewed for their perceptions.The results showed that: 1) the actual scores (prior to mediation) of the experimental group were significantly higher than those of the control group despite the increasing test difficulty. 2) the most benefited attributes were grammar, followed by inference, while the least effective ones were main idea and detail. 3) 4 out of the 8 students acknowledged the effectiveness of AMM; the mediators indicated its good feasibility while pointing out the possible negative emotions resulting from frequent mediations. AMM can provide a reference for computerized DA, and the effectiveness of its large-scale application requires further validation."
