Japanese Honorification in an HPSG Framework

18163

Authors

nan

Abstract

We present a solution for the representation of Japanese honorificational information in the HPSG framework. Basically, there are three dimensions of honorification. We show that a treatment is necessary that involves both the syntactic and the contextual level of information. The Japanese grammar is part of a machine translation system. 1

Keywords

nan

Full Content

JAPANESE HONORIFICATION IN AN HPSG FRAMEWORK Melanie Siegel Department of Computational Linguistics University of the Saarland Postfach 151150 D-66041 Saarbriicken, Germany siegel@dfki.de Abstract We present a solution for the representation of Japanese honorificational information in the HPSG framework. Basically, there are three dimensions of honorification. We show that a treatment is necessary that involves both the syntactic and the contextual level of information. The Japanese grammar is part of a machine translation system. 1 Introduction The Verbmobil system is a machine translation system for German, English and Japanese dialogues. The szenario is a special sort of task-oriented dialogues: appointment scheduling' . Spoken language encodes references to the social relation of the dialogue partners. The utterances can express social distance between addressee and speaker and third persons, who are mentioned. Honorifics can even express respect concerning entities of the world. Consider the following examples from Japanese, German and French: ( � wann I ) when� haben have� Sie you� Zeit time est-ce que� vous� avez� du� temps (2) qwvennd do� you� have� the� time (3) itsu when� go-tsugoo yoroshii� deshoo HON-conditions� Argcdm�hii� good�dh COP�lea QUE The semantic content of these utterances is: 'When does it suit you?'. But there is an additional pragmatic content: The speaker expresses social distance concerning the addressee. This is expressed by the honorific pronouns Sie and vows in German and French. In the Japanese example it is expressed by the following attributes: 'See [Wahlster1997] for further reference. • The honorific prefix go in front of tsugoo • The honorific adjective yoroshii • The honorific copula deshoo A Japanese utterance with the same semantic content in � for example — a family context could be: (4\ (4) itsu� jikan� _ga� aru when� time� NOM� have� QUE Information about honorification is - on the one hand - necessary for the description of syntactic phenomena like honorific agreement or relative sentences and - on the other hand - necessary for correct translation. In order to understand the whole meaning of the Japanese utterances it is important to represent the different honorific attributes in the analysis structure. The information can be used to resolve zero pronominalization and topicalized structures. It is even more important for the adequate generation of the Japanese utterances. In other investigations on zero pronoun resolution in task-oriented dialogues (cf. [Siege11997]) we calculated that 23.9% of the zero pronominals can be solved using lexical pragmatic restrictions about honorification. 2 Honorific Forms in Japanese Honorifics in Japanese express the social relation of familiarity or distance between speaker, addressee and third persons. Consider the situation where the speaker asks if s/he returned a book to the addressee, 'did I return the book to you?'. In a familiar context (s)he would say: (5) hon book AKK kaeshita returned QUE In a more formal situation with more social distance between speaker and addressee the utterance would be: (6) hgoontonok� Ngdm� o-kaeshi-shimashita hon-returned-hon� ka QUE The book as belonging to the addressee is prefixed by the honorific go. The predicate gets the …

Content truncated to 500 words

Labeled Text

We present a solution for the representation of Japanese honorificational information in the HPSG framework. Basically, there are three dimensions of honorification. We show that a treatment is necessary that involves both the syntactic and the contextual level of information. The Japanese grammar is part of a machine translation system. 1

Publication Details

Paper ID:
18163
Year:
2000
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan