It seems to be a general feature of natural language that the elements of a sentence can be pronounced in one position, while at the same time serving a function in another part of the structure of a sentence. Linguistic theories in the Transformational Generative Grammar tradition have tried to provide a unified explanation for this observation by proposing analyses that involve movement of constituents.
The subject of this dissertation is parsing sentences using a grammar formalism that allows for movement of constituents. The recognizers presented in this dissertation will recognize languages generated by Minimalist Grammars as defined in [Sta97]. Minimalist Grammars are simple formal grammars based on the Minimalist approach to linguistic theory ([Cho95]). Minimalist Grammars are derivational and feature-driven: phrases are derived by applying transformational operations to lexical items and intermediate structures, and the applicability of the transformational operations is determined by the syntactic features of the structures involved.
We will show that the set of languages generated by Multiple Context-Free Grammars ([SHF91]) is included in the set of languages generated by Minimalist Grammars. Together with the reverse result in [Mic98], this completes the proof that Multiple Context-Free Grammars and Minimalist Grammars are weakly equivalent.
Minimalist Grammars can move material from positions arbitrarily deep inside a sentence. Michaelis' equivalence result ([Mic98]) permits a representation of Minimalist Grammars in which the operations of the grammar can be characterized in such a way that they are strictly local. We will use this representation to derive several polynomial time recognition algorithms for Minimalist Languages, including an Earley-style recognizer ([Ear68]). Unlike known recognizers for Multiple Context-Free Languages, the recognizers in this dissertation can handle grammars with empty strings which are not immediately derived from the start symbol of the grammar. Moreover, the recognizers will parse a sentence from left to right.
The research reported in this dissertation contributes to a better understanding of Minimalist Grammars and related formalisms. Also, because Minimalist Grammars lends themselves very well to expressing current proposals about the kind of structures found in natural language, the recognizers described in this dissertation can be used to rigorously explore the computational consequences of psycholinguistic theories of human sentence processing.
Cited By
- Kobele G Importing montagovian dynamics into minimalism Proceedings of the 7th international conference on Logical Aspects of Computational Linguistics, (103-118)
- Stabler E Top-down recognizers for MCFGs and MGs Proceedings of the 2nd Workshop on Cognitive Modeling and Computational Linguistics, (39-48)
- Kobele G Without remnant movement, MGs are context-free Proceedings of the 10th and 11th Biennial conference on The mathematics of language, (160-173)
- Harkema H A recognizer for minimalist languages New developments in parsing technology, (251-268)
Recommendations
Minimalist Grammar Transition-Based Parsing
Logical Aspects of Computational Linguistics. Celebrating 20 Years of LACL (1996–2016)AbstractCurrent chart-based parsers of Minimalist Grammars exhibit prohibitively high polynomial complexity that makes them unusable in practice. This paper presents a transition-based parser for Minimalist Grammars that approximately searches through the ...
A recognizer for minimalist languages
New developments in parsing technologyMinimalist Grammars are a rigorous formalization of the kind of grammars proposed in the linguistic framework of Chomsky's Minimalist Program. One notable property of Minimalist Grammars is that they allow constituents to move during the derivation of a ...
Guided parsing of range concatenation languages
ACL '01: Proceedings of the 39th Annual Meeting on Association for Computational LinguisticsThe theoretical study of the range concatenation grammar [RCG] formalism has revealed many attractive properties which may be used in NLP. In particular, range concatenation languages [RCL] can be parsed in polynomial time and many classical grammatical ...