Open AccessNon-transitive information flow in Japanese noun-classifier matching
Author(s) -
Roger Lévy,
David Y. Oshima
Publication year - 2003
Publication title -
proceedings of the international conference on head-driven phrase structure grammar
Language(s) - English
Resource type - Journals
ISSN - 1535-1793
DOI - 10.21248/hpsg.2003.15
Subject(s) - noun , computer science , artificial intelligence , natural language processing , determiner , classifier (uml) , transitive relation , noun phrase , linguistics , mathematics , philosophy , combinatorics
In Japanese, as in other classifier languages like Chinese and Malay, numeralsdo not directly quantize nouns, but first combine with a classifier to form ameasure phrase (MP; cf. Aikhenvald 2000). From the perspective ofconstraint-based approaches to syntax/semantics, the mutual selectiverestriction between classifiers and nouns can be stated in terms ofinformation-sharing and featural identity, to some extent parallel to thetreatment of gender/number agreement (between determiner and noun, forinstance) (cf. Pollard and Sag 1994; Kathol 1999). There are, however, datathat challenge this line of approach to noun-classifier matching. Wedemonstrate in this paper that it is possible that a single noun is associatedwith different types of classifier, and show why they are problematic forunification-based approaches, similar to the situation with case syncretism inEuropean languages (Ingria 1990 and others). Later in the paper, we argue thatinformation-sharing between noun, predicate and classifier is not completelytransitive, and present a formal analysis which models multiple selectionalrequirements with sets.