Probabilistic learning in developmental language disorder

University Of Arizona, Tucson AZ

Investigators

Linked publications & trials

Abstract

Project Summary/Abstract Currently, we lack an understanding of why grammatical deficits, particularly, are the primary deficit in developmental language disorder in children, and we also lack effective clinical treatment for these deficits. Recent studies have suggested that grammar may be impaired because of its statistical properties, which may be difficult for children with this disorder to learn or attend to. However, we do not yet understand what exactly about statistical input is difficult, and therefore face a gap in determining theoretical mechanisms that can explain the profile of developmental language disorder. In the proposed research, we explore the hypothesis that manipulations in the statistical properties of linguistic input may facilitate grammar learning in children with developmental language disorder. We will test this hypothesis through the following aims: We will compare learning for deterministic versus probabilistic statistical information to determine which type is more easily learned by children with developmental language disorder. We will compare learning for dependencies at different distributions to determine if children with developmental language disorder benefit from certain statistical structures. We will address these aims through two studies: an artificial grammar learning task and a sentence processing task with eye-tracking. These studies are innovative because they use tasks from basic research on language acquisition and processing to isolate aspects of linguistic input that could improve learning in developmental language disorder. Identifying the characteristics of linguistic information that are particularly problematic or relatively helpful for children with developmental language disorder as they learn grammar can help us understand where and how grammatical deficits in this population arise. Findings will be impactful because they may lead to more effective treatment through control of variables that can be simply managed in clinical settings, e.g. how often a word appears in one grammatical structure compared with another. The training provided in this proposal will provide theoretical and technical training in intervention research, working with individuals with developmental language disorder, and using real-time technologies to study language.