Language Technology Methods Inspired by an Agglutinative, Free Phrase-Order Language
Gábor Prószéky, Csaba Merényi · Repository of the Academy's Library (Library of the Hungarian Academy of Sciences) · 2012
A crucial requirement for doing machine translation, machine-aided translation, information retrieval or computational information processing of texts of agglutinative languages, in general, is an efficient computational morphology.Having done the morphological processing of the word-forms of sentences, the sentence structure itself should be identified.This step is also common in all human languages.In case of agglutinative languages, like Hungarian, morphological processing provides a lot of information expressed by syntax in other, less-inflectional languages.Syntactic processing of highly inflectional languages, therefore, can use a lot of syntactic information identified by the morphological subsystem.It is needed, because the agglutinative nature of morphology results in free phrase order on the sentence level.In the first part of the paper we introduce Humor, a morphological description formalism and algorithm inspired by an agglutinative language, Hungarian.In the second part of the paper we describe a syntactic parsing formalism and algorithm inspired by Hungarian, a free-phrase-order language, and show the way how it is used in the Hungarian-English MetaMorpho machine translation system.