Syntax of Classifiers and Measure Words in Chinese and Beyond

Abstract

This thesis deals with classifiers and measure words in Chinese, focusing specifically on Mandarin, Cantonese and Northern Wu. It provides a typological account of the variation observed in these languages and offers both a synchronic and a diachronic approach to syntactic variation. Numeral classifiers are known to occur obligatorily in quantified noun phrases in Chinese (Chierchia 1998: 354-355). Different accounts have been proposed in the literature for these: some researchers propose a left-branching structure in which the classifier merges with the numeral (Greenberg 2013 [1974]; Her 2017), while others argue for a right-branching structure in which the classifier merges directly with the head noun (Cheng & Sybesma 1999; Simpson 2005). In addition, it has been observed that Chinese languages have measure words such as shao 'spoonful', and it has been proposed that they are syntactically identical to classifiers (Her 2017). In this thesis, I propose that classifiers should be distinguished from measure words, and that both structures are needed for a complete analysis of the classifiers and measure words in Mandarin, Cantonese and Northern Wu. I hypothesize that classifiers and measure words can be syntactically different, and that they tend to undergo different types of diachronic changes, which may account for their synchronic syntactic differences. These hypotheses are tested against data collected from native speakers of Northern Wu and Cantonese with a grammaticality judgement task; the historical data come from previous descriptive works and corpora of Old and Middle Chinese. The findings include a systematic syntactic difference between classifiers and measure words in Cantonese and Northern Wu (which is not found in Mandarin), as well as several types of diachronic changes that can be subsumed under grammaticalization. Based on those data, I argue that classifiers differ syntactically and categorially from measure words in Cantonese and Northern Wu, and that they are syntactically distinct from Mandarin classifiers. They are associated with different structures: classifiers and measure words in Mandarin as well as Cantonese and Northern Wu measure words occur in a left-branching structure, while classifiers in Cantonese and Northern Wu are used in a right-branching structure. These differences between the three Chinese languages are attributable to the diachronic changes that they underwent, and I propose a correlation between the syntactic behaviour of classifiers and their level of grammaticalization. This thesis on Chinese classifiers provides original work in that synchronic syntactic properties are directly linked with diachronic change and in proposing a universal left-branching structure for measure words and language-specific structures for classifiers. My theory on Chinese classifiers has several major theoretical consequences. First, classifiers and measure words are associated with different functional heads. Second, the classifiers that occur in right-branching structures in quantified noun phrases are expected to show other signs of grammaticalization (e.g. phonological reduction). Third, numeral classifiers and number morphology are not in complementary distribution (Bisang 2012), and their co-occurrence is costly but not intrinsically prohibited. Last, some properties of classifiers could be analyzed as emergent rather than fixed (Hopper 1998), and they explain why classifiers undergo some changes diachronically and bear certain vestigial similarities to measure words despite more systematic differences.