The Treatment of Japanese Focus Particles Based on Lexical-Functional Grammar

The treatment of Japanese focus particles based on Lexical-Functional Grammar Tomoko OHKUMA Corporate Research Group Fuji Xerox Co., Ltd. 430 Sakai, Nakai, Ashigarakamigun Kanagawa, Japan Ohkuma.Tomoko@fujixerox.co.jp Hiroki YOSHIMURA Corporate Research Group Fuji Xerox Co., Ltd. 430 Sakai, Nakai, Ashigarakamigun Kanagawa, Japan Hiroki.Yoshimuragfujixerox.co.jp Hiroshi MASUICHI Corporate Research Group Fuji Xerox Co., Ltd. 430 Sakai, Nakai, Ashigarakamigun Kanagawa, Japan Hiroshi.Masuichi@fujixerox.co.jp Yasunari HARADA School of Law, Waseda University 1-6-1 Nishi-waseda, Shinjuku, Tokyo, Japan harada0waseda.jp Abstract The purpose of this paper is to discuss treatment of 'focus particles' in Japanese sen- tences to be incorporated in a Japanese parsing system based on the Lexical-Functional Grammar(LFG) formalism. Focus particle can follow nouns, quantifiers, verbs, other particles (postpositions) and auxiliary verbs. Thus, it is necessary for a large-scale grammar to treat focus particles properly. Furthermore, there are syntactic and se- mantic ambiguities caused by the particles. We propose phrase structure rules and lexical entry constraints which cover focus particles in various positions and account for the ambiguities. 1 Introduction As has been observed in many studies, various functional relations are expressed by particles in Japanese. In particular, particles such as `kurar , `bakarr , 'clake', `nomr, `hodo' and 'made' specify focus in sentences. It has been observed that these particles mark contrastive focus within sentences (Numata et al., 2000). It is important for a Japanese parsing system to be able to treat these focus particles properly, because focus particles bear different syntactic functions depending on where they appear in a given Japanese construction. In this paper, we propose phrase structure rules and lexical entry constraints to treat Japanese focus particles on the basis of the Lexical-Functional Grammar (LFG) formalism (Kaplan and Bresnan, 1982; Dalrymple, 2001). The proposed rules and constraints for treating focus particles have been incorporated into a Japanese LFG system we are currently developing and have been used extensively for parsing real-world Japanese text. 2 Japanese LFG system Our Japanese LFG parsing system development aims at broad coverage and deep analysis (Ma- suichi and Ohkuma, 2003; Masuichi et al., 2003). The lexical entries for the words in the input are created by the output of the Chasen mor- phological analyzer (Matsumoto et al., 1999). In addition to these lexical entries, we have other lexicon files for verbs, adjectives, adjectival nouns and function words. Lexical entries for verbs, adjectives and adjectival nouns were written based on the case frame information in the Japanese IPAL dictionary (IPA, 1987) and have been manually enhanced. 448 Lexical entries for verbs consist of 10,387 entries and 41,115 functional annotations for 2,366 verbs. Lexical entries for adjectives and for adjectival nouns consist of 947 entries and 2,197 functional annotations for 369 words in total. Lexical entries of focus particles are included in our 'core lexical entries'. Core lexical entries include those for basic words such as auxiliary verbs, postpositional particles and so forth, plus syntactically important nouns such as `toki' (time) and `aida' (interval) and consist of 1,252 entries and 1,913 functional annotations for 675 words. Phrase structure rules and …

Content truncated to 500 words