[Author Login]
[Home]
ROA:746
Title:Comparing two Optimality-Theoretic Learning Algorithms for Latin stress
Authors:Diana Apoussidou, Paul Boersma
Comment:
Length:14
Abstract:This paper compares the performance of two formal Optimality-Theoretic learning algorithms in modelling the acquisition of Latin stress from overt language data: Error-Driven Constraint Demotion (EDCD; Tesar 1995) and the Gradual Learning Algorithm (GLA; Boersma 1997). We present computer simulations of learners who are trained on several kinds of overt Latin stress patterns: a case with main stress only, three cases with overtly available secondary stress, and a case in which the learners are free to invent their own secondary stress patterns. Several of these cases turn out to be learnable with the GLA, none with EDCD.
Type:Paper/tech report
Area/Keywords:Learnability, Language Acquisition
Article:Version 1