Left is right, right is not: On the constituency of the classifier phrase in Chinese

This paper argues for the left-branching constituency of the Chinese classifier phrase and demonstrates that the right-branching approach assumed by the majority of current syntactic works is not viable. The rejection of the right-branching approach entails the rejection of the “split” approach, where both left- and right-branching structures are required. In this debate, we offer a vital fresh perspective from the syntax and mathematics of complex numerals. We examine the right-branching argumentation in A. Li (2014) , which, crucially, extends Ionin & Matushansky’s (2006) non-constituent account of complex numerals, e.g. two hundred , in non-classifier languages like English to Chinese and must rely on ellipsis and a silent element YIDIAR ‘a bit’. Yet, complex numerals in Chinese, e.g. liang bai ‘200’, are in fact constituents ( He 2015 ), and the alleged YIDIAR ‘a bit’ does affect the semantics of the noun phrase and is thus by definition illicit ( Her & Tsai 2014 ; 2015 ). Other evidence comes from Chinese synchronic and diachronic syntax as well as the typology of classifier word orders. While the overall argumentation centers on Chinese, it has significant cross-linguistic implications.