UCLA

This dissertation describes and analyzes the syntactic properties of three inter-related structures in both standard and non-standard varieties of Mandarin: yes-no question, unconditional adverbial clauses, and verb echo answers to yes-no questions.All human languages have ways to ask a yes-no question, but cross-linguistically they differ from each other in the exact strategies used. I systematically look at a type of sentence-internal question particle that is only used in forming yes-no questions in Jianghuai Mandarin, providing cross-linguistic evidence for Bhatt & Dayal’s (2020) proposal about the existence of a class of dedicated yes-no question particles in human languages. I further compare three types of yes-no questions formed with a sentence-internal question particle ha/a or/and an A-not-A string in two mutually intelligible varieties of Jianghuai Mandarin spoken in Wuhu and Nanjing. The distributional differences between ha/a and A-not-A string are understandable if they are within two separate functional projections, QuP and PolP respectively. Meanwhile, the parallel behaviors of all three types of yes-no questions are taken as evidence that they all share the same underlying structure that involve both QuP and PolP, and the variation in their surface forms is reduced to the lexicon. Building on Holmberg’s (2016) analysis of yes-no questions, I argue that the syntax of yes-no questions in Jianghuai Mandarin not only involves a PolP headed by an open-valued polarity variable [�Pol], but a QuP headed by a dedicated yes-no question particle must also be present. Furthermore, I discuss another structure where a yes-no question seems to be allowed, unconditional adverbial clauses, which have not attracted much attention in the literature on Mandarin syntax Unlike the traditional assumption, I show that there exist two distinct types of unconditional adverbial clauses in Mandarin: Headed antecedents and bare antecedents consistently differ from each other in both their internal and external syntax. Regarding the internal syntax, bare antecedents involve an impoverished structure: They are TPs without a left periphery; whereas headed antecedents involve a full-fledged CP-domain. I argue that only headed but not bare antecedents involve an embedded interrogative. In terms of the external syntax, bare antecedents are base generated at Spec douP within the consequent and then overtly move to their canonical pre-consequent position, hence they all fall under the category of central adverbial clauses. In contrast, the merger of headed antecedents is independent from dou, and hence they exhibit more flexibility in their attachment sites and varying degrees of integration with the consequent. It is argued that they can realize as any of the three types of adverbial clauses: central adverbial clauses, peripheral adverbial clauses, and overarching speech-act modifiers, each of which show distinct properties in their external syntax (and internal syntax). This provides cross-linguistic evidence for the three-way division in adverbial clauses recently proposed by Frey (2020) and Badan & Haegeman (2022). Neutral yes-no questions are typically answered by repeating the verb in the question (i.e. “verb echo answers” or VEA) in Mandarin, which is the third structure I investigate in this dissertation. Building on existing studies (Liu 2014, Simpson 2015, Holmberg 2016, Wei 2019, a.o.), I provide converging empirical evidence showing that VEAs must not only involve a base-generated verb or a VP, and there must be an underlying sentential structure. I show that Mandarin VEAs do not involve pro-drop and analyses involving pro-drop face two kinds of empirical challenges: Elements that cannot be pro-dropped are nevertheless omitted from VEAs; whereas elements that can be dropped still allow elements extracted from them to surface in VEAs. Meanwhile, I argue that a head movement approach cannot account for (i) variations in VEAs, especially various kinds of phrasal elements occurring in VEAs; and (ii) VEAs to embedded yes-no questions with matrix scope. I instead propose that Mandarin VEAs are derived through (remnant) PolP movement followed by TP ellipsis.