The Mass-Count Distinction and the Syntax of Classifiers in Japanese

Abstract

In this thesis, I investigate the properties of classifiers and the grammaticized mass-count distinction in Japanese from a generative perspective (Chomsky 1981, 1995 and subsequent works), adopting the analyses of Distributed Morphology (Halle and Marantz 1993 and subsequent works) and the Exo-Skeletal model (Borer 2005a). I propose that, in addition to the mass-count distinction being manifested at the conceptual level, the mass-count distinction is grammaticized in Japanese. I do this by examining the individuation and number properties in various constructions with and without classifiers in the language (bare nouns, plurals, and numeral classifier constructions). The idea is that, in addition to classifiers, several residual ways to encode a grammaticized mass-count distinction are present in Japanese. Bare nouns and plurals also involve Ind(ividuation) Phrases, headed by a null Ind head or plural markers. Additionally, I show that these constructions are polysemous in nature, and encompass individuating and non-individuating varieties. I introduce a list of criteria that helps sort out which elements in the grammar manifest the grammaticized mass-count distinction from those that do not. First, I propose that, in Japanese, there are two types of number neutral bare nouns. I show that bare nouns in Japanese are not restricted to narrow scope (in contrast to those in other languages: e.g., Rullmann and You 2006 for Mandarin). Based on observations from relative scope, but also from kind reference and telicity (Carlson 1977, Borer 2005b), I argue that Japanese bare nouns that take wide scope (which I call DP bare nouns) involve individuation, whereas Japanese bare nouns that take narrow scope (which I call nP bare nouns) lack IndPs. (See, e.g., Dayal 2011, Espinal and McNally 2011 for a similar heterogeneous view in other languages). Second, I show that plural making in Japanese exhibits diverse properties, and can be mapped across nominal functional heads (cf. Mathieu 2014; also see Wiltschko 2008). Individuating plurals, including -tati (when used with proper names) and reduplication (Sudo 2017), are inclusive plurals, and pattern with bare plurals in English. Non-individuating plurals, on the other hand, including -tati (when used with common nouns) are exclusive plurals. This contributes to the recent literature on the inclusive-exclusive contrast seen in plural interpretation (e.g., Mathieu 2012). I point out that the observations in plurals in Japanese are not fully explained by an analysis of the head-modifier distinction of plurals proposed by Wiltschko (2008). Third, I show that there are structurally two types of numeral classifier constructions in Japanese, each with different properties of individuation (see Cheng and Sybesma 1999 and Li 2013 for similar claims in Mandarin). Classifiers have been commonly classified in terms of what type of nouns they are used with: sortal classifiers are mainly used with count-y objects, while mensural classifiers are used with either count-y or mass-y nouns. Therefore, it is the type of classifiers that is often assumed to manifest or reflect the grammaticalized mass-count distinction in Japanese and other languages (e.g., Cheng and Sybesma 1999). I instead propose that each of these two types in Japanese allows either a count or mass reading. In other words, the grammaticalized mass-count distinction in Japanese is not distinguished by types of nouns or classifiers, but by the structural positions of classifiers. Regardless of the types of classifiers, classifiers become individuators when they are at IndPs.