Probabilistic tree-generating binary grammars
-
Paweł Skórzewski
and
Krzysztof Jassem
Abstract
The article presents a notion of Probabilistic Tree-generating Binary Grammar (PTgBG), a probabilistic extension of Tree-generating Binary Grammars (TgBG). TgBG is a for-malism developed to describe languages with syntactic discontinuities, designed with two practical goals in mind: (1) efficient parsing; (2) representation of results in the form of flat, easily processable phrase-structure trees. Here, we introduce rule probabilities for TgBG, thus creating PTgBG. We compare PTgBG to standard Probabilistic Context-Free Grammar (PCFG) in terms of parse tree probabilities and sentence probabilities. We conclude that PTgBG may be effectively parsed by the methods developed for PCFG, while allowing for convenient expression of some linguistic phenomena, specific for non-configurational languages.
©2014 by Walter de Gruyter Berlin/Boston
Articles in the same Issue
- Evidence for minimal pairs in Turkish Sign Language (TİD)
- On the syntax of ergativity ın Kurdish
- Targets, triggers, and directionality in non-local and local place assimilation in child and adult language
- Grammaticalization of causatives and passives and their recent development into stance markers in Korean
- Probabilistic tree-generating binary grammars
- Colour word Stroop test in Polish learners of English: No evidence that L2 proficiency matters
Articles in the same Issue
- Evidence for minimal pairs in Turkish Sign Language (TİD)
- On the syntax of ergativity ın Kurdish
- Targets, triggers, and directionality in non-local and local place assimilation in child and adult language
- Grammaticalization of causatives and passives and their recent development into stance markers in Korean
- Probabilistic tree-generating binary grammars
- Colour word Stroop test in Polish learners of English: No evidence that L2 proficiency matters
