A computer-based dynamic assessment to predict language learners’ current knowledge

Abstract

In the domain of second language learning, assessment is commonly divided into non-dynamic and dynamic assessment. Both methods differ by the way they have to predict the learner’s future performance, which can be characterised by either a past-to-present or present-to-future model, respectively. While the past-to-present model tends to infer the learners’ future achievements from past and present performance, the present-to-future model intends to predict the learners’ potential development from the amount of assistance they require to complete the task. However, determining the learners’ grammatical knowledge when assessing their performance is not without complications, for the reason that one cannot be sure whether the incorrect forms in their written texts are due to errors of competence, i.e. errors, or performance issues, i.e. mistakes.

Firmly grounded in sociocultural theory as well as in a dynamic assessment perspective, this paper proposes a theoretical model to predict, from free written texts, the current grammatical knowledge of learners of French. Following a brief discussion on the differences between non-dynamic and dynamic assessment, as well as the main approaches to distinguishing between errors and mistakes, the process of how to apply the novel model in educational settings is approached, and concrete examples drawn from the performance of one learner of French are provided.

Paper presentation

Thouësny, S. (2010). “A computer-based dynamic assessment to predict language learners’ current knowledge“. Paper presentation. First International Language Conference on Translation, Technology and Autonomy in Language Teaching and Learning, 10-DEC-10 – 11-DEC-10, Galway, Ireland.