On the Wire
Linguistics In the News
No current headlines.
On the Wire
Science In the News

Access Your PC from Anywhere
Abstract
Marcus's Computational Account of Some Constraints on Language
By Steve Hoenisch
Last updated on August 11, 2004
Copyright 1996-2006 www.Criticism.Com
Table of Contents
1 Citation
2 Determinism Hypothesis
3 Related

1 Citation

From: Marcus, M. 1986. A Computational Account of Some Constraints on Language. In Readings in Natural Language Processing, Grosz, Jones and Webber, eds. 1986. Los Altos, Calif.: Morgan Kaufmann.

Top

2 Determinism Hypothesis

The purpose of Marcus's paper is to present arguments showing that the constraints imposed by the structure of the grammar interpreter called PARSIFAL are such that grammar rules cannot parse sentences violating either the Specified Subject Constraint or the Subjacency Principle, a result that stems from a component of the grammar interpreter motivated by what Marcus calls the Determinism Hypothesis. This hypothesis "claims that natural language can be parsed by a computationally simply mechanism that uses neither backtracking nor pseudo-parallelism, and in which all grammatical structure created by the parser is `indelible' in that it must all be output as part of the structural analysis of the parser's input" (89). The structure of PARSIFAL, Marcus says, is based upon the hypothesis that a parser for natural language does not need to simulate a nondeterministic machine.
PARSIFAL uses two major data structures, one of which is "a pushdown stack of incomplete constituents called the active node stack" (90; emphasis in original). The other structure, which distinguishes the parser from others, is "a small three-place constituent buffer which contains constituents which are complete, but whose higher level grammatical function is as yet uncertain" (90; emphasis in original). The structure and operation of the parser is motivated by several properties that Marcus maintains a nondeterministic parser should include and which PARSIFAL embodies: the parser should be partially data-driven, able to "reflect expectations that follow from the partial structures built up during the parsing process" (91), and have a constrained look-ahead capacity.
After outlining the grammar interpreter, Marcus, working within the Chomskian framework that postulates traces, demonstrates how the SSC and the Subjacency Principle "fall out" from the formulation of PASSIVE. Marcus shows that in the context of a grammar for the parser, a solution emerges for constructions of passivization and, without adding complexity to the grammar, raising. Marcus demonstrates these claims by successfully applying the parser to examples of passivization and raising. In doing so, he develops a computational account of NP-movement (but not WH-movement) in accordance with the SSC and Subjacency Principle.
However, as Marcus himself acknowledges, he does not present strong, complete evidence for the Determinism Hypothesis. He concludes only that the NP-movement subcases of the constraints follow from the grammar interpreter, a result that provides some evidence for the Determinism Hypothesis but does not confirm it outright. Accordingly, the paper provides clear evidence for the parser's power in handling NP-movement in accordance with Chomsky's constraints but leaves undetermined whether the parser can also handle cases of, for instance, WH-movement. The result is a strong contribution to the development of computational models based on the generative grammar of Chomsky.
Top
Google       

Criticism.com Web

cover artFoundations of Statistical Natural Language Processing
The MIT Press
New $60.83
cover artFoundations of Language : Brain, Meaning, Grammar, Evolution
Oxford University Press
New $19.95
cover artSpeech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics and Speech Recognition
Prentice Hall
New $83.28
cover artThe Oxford Handbook of Computational Linguistics (Oxford Handbooks)
Oxford University Press, USA
New $45.00
cover artConstructions at Work : The Nature of Generalization in Language
Oxford University Press, USA
New $29.95
cover artComputational Linguistics and Formal Semantics (Studies in Natural Language Processing)
Cambridge University Press
New $39.16
cover artRepresentation and Inference for Natural Language : A First Course in Computational Semantics (Center for the Study of Language and Information - Lecture Notes)
Center for the Study of Language and Inf
New $30.00
cover artComputational Linguistics (Studies in Natural Language Processing)
Cambridge University Press
New $26.99
cover artFinite State Morphology
Center for the Study of Language and Inf
New $34.76
cover artBeyond Communities of Practice : Language Power and Social Context (Learning in Doing: Social, Cognitive & Computational Perspectives)
Cambridge University Press
New $24.99
(Prices May Change)
Privacy Information

cover artNatural Language Information Retrieval (Text, Speech and Language Technology)
Springer
New $151.00
cover artLanguage As a Cognitive Process: Syntax
Addison-Wesley
New
cover artCorpus Linguistics : Investigating Language Structure and Use (Cambridge Approaches to Linguistics)
Cambridge University Press
New $31.99
cover artMathematical Models of Spoken Language
John Wiley & Sons
New $120.00
cover artModels of Language Acquisition : Inductive and Deductive Approaches (Oxford Linguistics)
Oxford University Press, USA
New $33.35
cover artMemory-Based Language Processing (Studies in Natural Language Processing)
Cambridge University Press
New $75.00
cover artData-Oriented Parsing (Center for the Study of Language and Information - Lecture Notes)
Center for the Study of Language and Inf
New $35.00
cover artComputer-Assisted Language Learning : Context and Conceptualization
Oxford University Press, USA
New $45.00
cover artConstraint-Based Grammar Formalisms: Parsing and Type Inference for Natural and Computer Languages
The MIT Press
New $45.00
cover artWordNet: An Electronic Lexical Database (Language, Speech, and Communication)
The MIT Press
New $58.19
(Prices May Change)
Privacy Information