Patterns in Constituency of Num-CL-N Sequences in Sinitic Languages: What Tone Sandhi Can Reveal

Abstract

This thesis studies the surface syntactic structure of Numeral-Classifier-Noun (Num-CL-N) sequences in Sinitic languages. Without considering the phonology-syntax interface, it is difficult to concretely assert that these sequences are left-branching [[Num CL] N] or right- branching [Num [CL N]]. Tone sandhi in Sinitic languages is a process that is limited by specific syntactic domains, thus observed tone sandhi in speech production should be able to clarify the surface syntactic structure. Recent work based on Mandarin’s third tone sandhi has proposed that the surface structure for these sequences is left-branching (Allassonnière-Tang et al., 2021), however, it is currently unclear whether this is true only in Mandarin, or a general fact for Sinitic languages. This thesis starts by replicating these results for Mandarin, then extends the methodology to examine the same construction in Shanghainese and Fuzhounese to contribute to the general discussion of phonology-syntax interface in Sinitic languages. I find that Fuzhounese, Shanghainese, and Beijing Mandarin all exemplify tone patterns consistent with left-branching models and propose that these may provide an insight into a universal of Sinitic languages, or even more strongly, a universal of the numeral classifier itself.