Modeling mediation in the dynamic assessment of listening ability from the cognitive diagnostic perspective

The distinguishing feature of dynamic assessment (DA) is the dialectical integration of assessment and instruction. However, how to design the targeted instruction or mediation has been relatively underexplored. To address this gap, this study proposes the attribute‐based mediation model (AMM), an English‐as‐a‐foreign‐language listening mediation model based on item attributes (constructs) derived from cognitive diagnostic assessment (CDA). CDA provides theoretical and methodological support for the accurate correspondence of items and attributes via Q‐matrix development and validation. The CDA‐identified attributes can inform the construction of targeted mediation models in DA, within which the mediation will serve as the subsequent remedial instruction. In this way, diagnosis and instruction were bridged through the integration of CDA and DA. The results of three one‐on‐one DA sessions showed that the AMM has the potential to diagnose learners’ listening ability at the attribute level and to stimulate microgenetic development. This study is significant in linking the psychometric‐based validity of a test and the DA‐based validity of development. It also provides insights into future mediation design and a possible way of merging DA and diagnostic language assessment.