Can categorization and generalization difficulties explain word learning characteristics in Developmental Language Disorders?

Lexical and semantic difficulties often go along with Developmental Language Disorders (DLD), but the underlying deficits remain misunderstood. This study aims at determining if children with DLD can identify a classification rule and generalize it. Firstly, we will observe this phenomenon according to the principles of Bayesian theories of cognition. These theories place inductive inferences as a central mechanism of the learning processes and are in line with the idea of a multiple-level generalization process. Secondly, we will consider the idea of an effect of the complexity in line with a limitation of processing resources in children with DLD. Our design includes two parts: a word-learning task and a generalization task. In the word-learning task, twenty schooled-aged children with DLD will be exposed to different categories perceptually or relationally defined. Furthermore, the number of features determining each category will increase progressively to see how children can learn subdivisions of each category. We will use eye tracking during the progressive learning in order to analyse if children with DLD explore the features in the same way than their typically developing (TD) peers. Our assumptions are threefold: first, we assume that children with DLD will need more exemplars to master a rule compared to age-matched TD peers. Secondly, we assume that these difficulties will be more important when the categories are relationally defined and/or when the number of features will increase. Finally, we suppose that the generalization process will be less efficient in DLD children than in their TD peers.