Phonological Rule Induction: An Architectural Solution

Abstract : Acquiring phonological rules is hard, especially when they do not describe generalizations that hold for all surface forms. We believe it can be made easier by adopting a more cognitively natural architecture for phonological processing. We briefly review the structure of M3P, our connectionist Many Maps Model of Phonology, in which extrinsic rule ordering is virtually eliminated, and iterative processes are handled by a parallel clustering mechanism. We then describe a program for inducing phonological rules from examples. Our examples, drawn from Yawelmani, involve several complex rule interactions. The parallel nature of M3P rule application greatly simplifies the description of these phenomena, and makes a computational model of rule acquisition feasible. (KR)