[Author Login]
[Home]
ROA:1349
Title:Calibration of constraint promotion does not help with learning variation in stochastic OT
Authors: Giorgio Magri, Benjamin Storme
Comment:to appear in Linguistic Inquiry
Length:24 pages + 11 pages of supplementary material
Abstract:The Calibrated error-driven ranking algorithm (CEDRA; Magri 2012) is shown to fail on two test cases of phonologically conditioned variation from Boersma and Hayes (2001). The failure of CEDRA raises a serious unsolved challenge for learnability research in stochastic OT, because CEDRA itself was proposed to repair a learnability problem (Pater 2008) encountered by the original GLA. This result is supported by both simulation results and a detailed analysis whereby a few constraints and a few candidates at the time are recursively 'peeled off' until we are left with a 'core' small enough that the behavior of the learner is easy to interpret.
Type:Paper/tech report
Area/Keywords:Stochastic OT; learnability analysis
Article:Version 1