Why computational models are better than verbal theories: the case of nonword repetition (pages 298–310)
Gary Jones, Fernand Gobet, Daniel Freudenthal, Sarah E. Watson and Julian M. Pine
Article first published online: 15 NOV 2013 | DOI: 10.1111/desc.12111
Tests of nonword repetition (NWR) have often been used to examine children's phonological knowledge and word learning abilities. However, theories of NWR primarily explain performance either in terms of phonological working memory or long-term knowledge, with little consideration of how these processes interact. One theoretical account that focuses specifically on the interaction between short-term and long-term memory is the chunking hypothesis. Testing the chunking hypothesis across three sets of nonwords that varied in wordlikeness showed that its predictions were only partly borne out. Only when implementing the chunking hypothesis as a computational model were all predictions supported. The research shows how any theory that involves long-term knowledge must be able to properly estimate not only the long-term knowledge involved but also how that knowledge interacts with temporary memory.