Predicting the receptive lexical Competence of Language Learners through a personalised and adaptive Model integrating a meaning-based Approach
This project is situated in the intersection of Computer-Assisted Language Learning and Natural Language Processing and aims at modeling the receptive lexical competence of French and Dutch as a foreign language learners. Recently, a number of computational linguistic models have been elaborated to automatically predict a language learner's individual lexical competence. However, these studies present two shortcomings: (1) they adopt a form-based approach to the prediction of vocabulary knowledge and (2) they are mostly non-personalized. Furthermore, they are generally focused on production.
In order to tackle these shortcomings, we propose a novel perspective on the automatic prediction of L2 vocabulary knowledge. On the one hand, we aim to empirically evaluate a learner's receptive lexicon by adopting a meaning-based approach to L2 vocabulary knowledge. To this end, we will use a learner corpus where each lexical unit is semantically disambiguated and annotated by learners in a reading context. On the other hand, we will examine the impact of personal learner characteristics (i.e. the learner's native language, his age and personal interests) on vocabulary knowledge prediction. To this end, we will elaborate a computational learner model that integrates these personal characteristics and which adapts to the individual learner.