CEDRIC BOECKX and FUMIKAZU NIINUMA CONDITIONS ON AGREEMENT IN JAPANESE

This paper integrates Japanese object honorification within a larger crosslinguistic context, and provides a principled explanation for an otherwise puzzling property: the fact that direct object honorification is blocked in the presence of a dative argument. Following a well-established tradition in the generative literature, we regard honorification as a case of agreement, but, unlike previous approaches, which rely on Spec-Head configurations, we show that Chomsky’s (2000) Agree mechanism suffices for (object) agreement to obtain. We argue that the blocking effect of dative elements is a reflex of a more general locality constraint, ‘defective intervention’, proposed by Chomsky 2000. The analysis also provides a compelling argument in favor of taking the 〈indirect object; direct object〉 order in Japanese as basic, and against base-generation approaches to scrambling. 1. OBJECT HONORIFICATION IN JAPANESE This paper investigates the nature of the object honorification in Japanese and its implications for principles of Universal Grammar. Honorification in Japanese is largely determined by two factors. One is sociolinguistic (the element associated with honorification must be socially superior to and respected by the speaker). The other is syntactic. We will be focusing here exclusively on defining the syntactic conditions for honorification.1 An early version of this paper was presented at the 3rd Formal Approaches to Japanese Linguistics conference at MIT (May 2001). We would like to thank Koji Sugisaki, Masashi Nomura, and very especially Nobuhiro Miyoshi for helping us sharpen the argument. Later on, one of us (Cedric Boeckx) had the chance to work closely on related issues with Youngmi Jeong. Cedric would like to thank Youngmi profusely for helping him understand the nature of agreement and Agree better, and work out the details of it in Boeckx and Jeong (2002). By incorporating the present material into Boeckx and Jeong, Cedric has benefited from comments from audiences at the University of Illinois at UrbanaChampaign, Michigan State University, Indiana University, the University of Maryland, and UC Irvine. For suggestions and comments at various points, we thank Richard Kayne, Željko Bošković, Naoki Fukui, and especially Norbert Hornstein. Finally, special thanks to Fritz Newmeyer for supporting this project, and to three very conscientious NLLT reviewers. 1 One reviewer objects to our treatment of (object) honorification as agreement on grounds that unlike agreement in, say, English, which is obligatory, honorific agreement is Natural Language & Linguistic Theory 22: 453–480, 2004. © 2004 Kluwer Academic Publishers. Printed in the Netherlands. 454 CEDRIC BOECKX AND FUMIKAZU NIINUMA We will not concentrate on the fairly well-studied case of subject honorification, in which an honorific marker on the verb is associated with the subject noun phrase, as in (1): (1) Tanaka sensei-ga Prof. Tanaka-Nom hon-o book-Acc o-yomi-ni-nat-ta read-SH-past Prof. Tanaka read the book Since Shibatani (1977), subject honorification has been treated as an instance of (abstract) subject verb agreement. Rather compelling evidence in favor of this position is to be found in Toribio (1990), Ura (2000), and Hasegawa (2002). Example (2) illustrates the phenomenon of object honorification, where the verb bears an honorific marker associated with the object noun phrase, not the subject noun phrase: optional and conditioned by extra-grammatical factors. We do not deny the importance of pragmatics in the study of honorification, but it seems to us that the nature of the constraint we discuss in this paper does not obviously lend itself to a pragmatic explanation. The fact that we are able to provide a reasonable syntactic account of an important set of sentences, in our view, argues for a syntactic characterization of honorification. (See also Hasegawa’s (2002) comprehensive overview of honorification phenomena in Japanese, where extensive arguments for a syntactic account of them are provided.) The reviewer notes that a syntactic account is inappropriate, since failing to trigger honorification leads to ‘discourse inappropriateness’ as opposed to ungrammaticality. We disagree. All we need to say to account for the difference the reviewer noted is that the honorification feature is optional. But once present agreement is obligatory, and is governed by the same mechanism of feature-checking. In other words, optionality obscures genuine ungrammaticality. Norbert Hornstein (p.c.) reminded us that the fact that object honorification with a reflexive is impossible (see (i)) provides a strong argument for treating (object) honorification as (object) agreement: (i) ∗ Tanaka sensei-ga Prof. Tanaka-Nom zibun-o self-Acc o-tasuke-si-ta help-OH-past Prof. Tanaka helped himself As Woolford (1999) extensively showed, anaphors typically resist agreement. We therefore expect anaphors to resist honorification. Note, incidentally, that other grammatical phenomena like focusand topicassignment have received comprehensive treatment in terms of syntactic mechanisms like Agree/feature-checking, even though they too depend on extra-linguistic factors. Also, in the past purely syntactic accounts of ‘optional’ agreement (sometimes associated with different linguistic registers) such as past participle agreement (see especially Kayne 1989) have led to a deepening of syntactic theory, which is also our goal in this paper. CONDITIONS ON AGREEMENT IN JAPANESE 455 (2) Taro-ga Taro-Nom Tanaka sensei-o Prof. Tanaka-Acc o-tasuke–si-ta/tasuke-ta help-OH-past/help-past Taro helped Prof. Tanaka As shown in (2), the morphological shapes of subject and object honorifics are different. The main difference pertains to the suffix associated with the verb. In the case of subject honorification, ni nar is used, whereas su or itas function as object honorifics. The morphology of honorification in Japanese is summarized in (3) (from Harada 1976, p. 504).2 (4) illustrates both types of honorifics: (3) Morphology a. HP+INF ni nar(Subject Honorification) b. HP+INF su-/itas(Object Honorification) HP = the ‘honorific prefix’, i.e. o-/go(common to both forms of honorification) INF = the infinitive form