TITLE: Learning novel sound sequences: A computational model and its application to atypical language development
SPEAKER: Gray Jones (School of Social Sciences, Nottingham Trent University)

ABSTRACT:

Nonword repetition is used to examine phonological knowledge in both typically-developing children and children with Specific Language Impairment (SLI).  However, the underlying mechanisms involved in nonword repetition are still poorly understood  in particular because there is an interaction between phonological working memory and long-term lexical phonological knowledge.  I will present a computational model that specifies how phonological working memory and long-term lexical phonological knowledge combine in the learning of novel sound sequences.  First, I will show how the model is able to simulate typically-developing children’s repetition performance.  Second, I will detail a repetition study that specifically examines the interaction between working memory and long-term knowledge by varying the lexicality of the nonwords.  Children with SLI show a deficit for low but not high lexicality nonwords, consistent with a phonological processing deficit where lexical phonological knowledge is used as an aid to repetition performance.  Third, I will detail how impairments to the model can be made and compared against the performance of the children with SLI.  The SLI study shows how manipulations to lexicality can tease apart phonological processing and working memory accounts of the SLI deficit.  The computational model shows how a parsimonious account of the learning of novel sound sequences can help us to understand the SLI deficit.