[Author Login]
Title:Phonotactic learning without a priori constraints: Arabic root cooccurrence restrictions revisited
Authors:John Alderete
Comment:To appear in proceedings of CLS 48; supersedes ROA-1055.
Abstract:In this article, we develop a connectionist model of learning phonotactics and apply it to the problem of learning root cooccurrence restrictions in Arabic. Two types of connectionist networks are developed: a multilayer network with a hidden layer and a single layer network with recurrent connections. They are both shown to classify Arabic words and nonwords in ways that are qualitatively parallel to psycholinguistic study of Arabic. In these networks, units and connections act like soft constraints in the computation of acceptability scores. Statistical analysis of the multilayer network behavior shows that the functions computed by the hidden layer nodes correspond well with OCP-Place constraints familiar from generative phonology. Because these constraints are malleable and can change gradually over time, the networks learn phonotactic generalizations without requiring the prior existence of the exact constraints responsible for phonotactics, a fact that sets this model apart from many phonotactic learners.
Type:Paper/tech report
Area/Keywords:phonology, computational learning, subsymbolic learning, connectionism, parallel distributed processing, Optimality Theory, Arabic, cooccurrence restrictions, dissimilation, nature vs. nurture, gradience, exceptions
Article:Version 1