Extended Abstract

Phonological opacity as local optimization in Gradient Symbolic Computation

Authors
  • Anna Mai (University of California, San Diego)
  • Eric Bakovic (University of California, San Diego)
  • Matt Goldrick (Northwestern University)

Abstract

We present a novel approach to counterbleeding rule interactions in Yokuts (Californian) using Gradient Symbolic Computation (GSC). GSC, a dynamical systems model, optimizes two constraint sets: a set specifying a Harmonic Grammar (HG) and a set of quantization constraints preferring discrete symbolic states. During optimization, quantization strength gradually increases, increasing the relative harmony of discrete symbolic vs. intermediate blend states. The output of the system therefore reflects the dynamics of optimization, not simply grammatical harmony. With appropriate dynamics, relatively high harmony intermediate states can trap optimization near less globally harmonic but locally optimal symbolic candidates; this can model Yokuts counterbleeding.

Keywords: phonology, opacity, GSC

How to Cite:

Mai, A., Bakovic, E. & Goldrick, M., (2018) “Phonological opacity as local optimization in Gradient Symbolic Computation”, Society for Computation in Linguistics 1(1), 219-220. doi: https://doi.org/10.7275/R5416V7W

Downloads:
Download PDF

105 Views

32 Downloads

Published on
01 Jan 2018