Compositional Mechanisms of Japanese Numeral Classifiers

Copyright 2012 by Miho Mano 26th Pacific Asia Conference on Language,Information and Computation pages 620–625 Compositional Mechanisms of Japanese Numeral Classifiers Miho Mano Naruto University of Education 748 Nakashima, Takashima, Naruto-cho Naruto-shi, Tokushima, JAPAN mmano@naruto-u.ac.jp Abstract This paper suggests that Generative Lexicon Theory (Pustejovsky, 1995, 2006, 2011) offers a new analysis of numeral classifiers, focusing on Japanese having various kinds of classifiers. It is often said that classifiers agree with quantified nouns, that is, the nouns have to match the semantic requirements of the classifiers. This paper examines their lexical structures and compositional mechanisms. Though Huang and Ahrens (2003) explain the compositional mechanisms between the classifiers and the quantified nouns using “coercion” instead of the agreement, this paper indicates that other mechanisms including Type Matching (Pustejovsky, 2011) also occur in Japanese depending on the type required by the classifier and the source type of the quantified noun, following Mano and Yonezawa’s (to appear) suggestion. 1 Introduction Japanese has various counters, called josuushi in Japanese, including so-called “numeral classifiers (Aikhenvald, 2000),” as well as other East Asian languages such as Chinese, Indonesian, Korean, and Thai. In Japanese, nouns cannot be directly modified by numerals but must be quantified by counters, as shown in (1). (1) a. *ni-{inu/kuruma} (Japanese) 2-dog/car ‘two dogs/cars’ b. ni-hiki-no inu/ ni-dai-no kuruma 2-CL-GEN dog/ 2-CL-GEN car1 ‘two dogs/two cars’ The counters are morphemes used together with numerals, and each of them has semantic restrictions on its objects. For example, a classifier -匹 hiki in (1b) requires its objects to be nonhuman animals, and -台 dai mainly selects for machines2. Most of the previous studies on classifiers assume that the modified nouns agree with the classifiers, because the classifiers can only count nouns which have particular meanings. Many studies have been done on the semantic restrictions of numeral classifiers (concerning Japanese, see Matsumoto (1991, 1993); Downing (1996), Iida (1999), and Nishimitsu and Mizuguchi (2004).). Huang and Ahrens (2003), however, suggest that the classifiers do not simply agree with the quantified nouns but coerce particular meanings to them, focusing on the numeral classifiers of Mandarin Chinese. This paper shares the view that classifiers can coerce the nouns to refer to particular types, but as Mano and Yonezawa (to appear) point out, it seems that they may agree with the nouns without changing their source types. There are some contradictory examples in Japanese, however. Taking chuusha ‘injection,’ for example, which is a polysemous noun that means a 1 The abbreviations used in this paper are as follows: ACC= accusative case, CL=classifier, GEN=genitive case, NOM= nominative case, phys=physical object, PRES=Present tense, PROG=progressive, PST=past tense, TOP=topic marker 2 See Matsumoto (1991, 1993) and Iida (1999) for more detailed restrictions of -dai. 620 physical object ‘syringe’ and also an event ‘injection.’ It can be counted by the classifiers, -本 hon and -回 kai. -Hon requires one dimensional (i.e. long and thin) physical objects (phys), and -kai is a classifier for events. Given that classifiers coerce the quantified nouns to be required types, ni-hon-no …

Content truncated to 500 words