Representing emergent phonology with the phonological attractor model:
Making child language data behave

Ellen D. Schmidt, Brent Nicholas & Lise Menn

Abstract for paper presented at
Child Phonology Conference
Purdue University
2-3 June, 2008

Traditional phonological analysis tools (rewrite-rules, templates, OT, etc.) were not designed to handle child data; none of them can tell the whole story of phonological development. Problems with these methods have been noted in the literature, and solutions, when offered, have not been adequate. We present further typical (Ferguson & Farwell, 1975) and atypical data (Schmidt, unpublished) that clearly require alternative modes of analysis, and offer a new exemplar-based attractor model, building on Vihman & Croft (2007) with Nicholas' visualization of the relationships between adult and child forms. This model combines strengths of previous attempts to understand phonological development and provides a major improvement in psycholinguistic plausibility. It incorporates frequency information, the notion that doing/thinking actually changes the brain and the phonological system, and the importance of what is salient/relevant to the child, and it provides a formal representation of associations between elements in the child's phonology and the impact they exert upon one another. We show how to apply the model to both typical and atypical acquisition data.

Ellen Schmidt is a PhD student in the Department of Linguistics at the University of Colorado at Boulder. She can be reached at: Ellen.Schmidt@Colorado.EDU

Brent Nicholas is a PhD student at the University of Colorado at Boulder. Lise Menn is a Professor Emeritus of Linguistics at the University of Colorado at Boulder.