[Author Login]
Title:Uncovering structure hand in hand: Joint Robust Interpretive Parsing in Optimality Theory
Authors:Tamas Biro
Comment:Acta Linguistica Academica Vol. 64 (2017) 2, 191-212. DOI: 10.1556/2062.2017.64.2.2
Abstract:Most linguistic theories postulate structures with covert information, not directly recoverable from utterances. Hence, learners have to interpret their data before drawing conclusions. Within the framework of Optimality Theory (OT), Tesar & Smolensky (1998) proposed Robust Interpretive Parsing (RIP), suggesting the learners rely on their still imperfect grammars to interpret the learning data. I introduce an alternative, more cautious approach, Joint Robust Interpretive Parsing (JRIP). The learner entertains a population of several grammars, which join forces to interpret the learning data. A standard metrical phonology grammar is employed to demonstrates that JRIP performs significantly better than RIP.
Type:Paper/tech report
Area/Keywords:Learning algorithms, Robust Interpretive Parsing, genetic algorithms, hidden structure, metrical stress
Article:Version 1