Discontinuous Grammar

This book presents a new model of human language,
human parsing, and human language learning. It
provides a detailed
formal account of a
wide range of linguistic phenomena, including
discontinuous word
order, secondary dependencies, parasitic gaps,
elliptic coordinations,
punctuation, and important aspects of morphology and
discourse. The
book argues for a probabilistic model of language,
and proposes a
psycholinguistically motivated non-projective
dependency parsing
algorithm that can take advantage of the
probabilistic model.

The book falls within the tradition of dependency
grammar, but
differs from dependency theories such as Word
Grammar, Functional Generative Description, and
Meaning Text Theory by building on a purely
probabilistic conception of human language,
and by providing a unified account of computational
linguistics, psycholinguistics, and linguistics.
The
theory has been used as the basis for the 100,000
word Danish
Dependency Treebank and the Copenhagen Dependency
Treebank for
Danish-English.

The book is targeted towards researchers and students
working within
computational linguistics, psycholinguistics, and
linguistics.

Autorentext

Matthias Buch-Kromann is an associate professor in computationallinguistics at the Copenhagen Business School. His researchinterests include dependency treebanks,parsing, language modelling, and machine translation. Hehabilitated with the present book, earning a dr.ling.merc. degreefrom the Copenhagen Business School in 2006.

Klappentext

This book presents a new model of human language,human parsing, and human language learning. Itprovides a detailedformal account of awide range of linguistic phenomena, includingdiscontinuous wordorder, secondary dependencies, parasitic gaps,elliptic coordinations,punctuation, and important aspects of morphology anddiscourse. Thebook argues for a probabilistic model of language,and proposes apsycholinguistically motivated non-projectivedependency parsingalgorithm that can take advantage of theprobabilistic model.The book falls within the tradition of dependencygrammar, butdiffers from dependency theories such as WordGrammar, Functional Generative Description, andMeaning Text Theory by building on a purelyprobabilistic conception of human language,and by providing a unified account of computationallinguistics, psycholinguistics, and linguistics.Thetheory has been used as the basis for the 100,000word DanishDependency Treebank and the Copenhagen DependencyTreebank forDanish-English.The book is targeted towards researchers and studentsworking withincomputational linguistics, psycholinguistics, andlinguistics.