The Semantic-Pragmatics Interface and Island Constraints in Chinese

This thesis is about strong island effects and intervention effects. Strong island effects are contexts where operator-variable dependencies cannot be established. The paradigmatic cases of strong island violations in Chinese occur in why-questions. This thesis explores a basic contrast: why-questions fail to be interpreted in strong island contexts, as opposed to other wh-questions. This contrast is illustrated in (1a) and (1b): (1) a. ;The main questions that my account of strong islands addresses are the following: • Is it true that only why-questions induce strong island violations, while others don't? • If Chinese strong island violations are indeed tied to why-questions, what is special about this question type that leads to strong island violations? • What is the nature of strong island violations in why-questions? Are they syntactic, semantic, pragmatic or a combination?;This thesis develops a semantic account for strong islands, and the core idea can be summarized as follows. What sets apart the reason adverb why from other wh-interrogative phrases is that why is ontologically different. Why modifies propositions, relating a proposition to a set of reasons, rather than corresponding to a part of the proposition. This proposition-level operation exhibits a main clause phenomenon, meaning that a why-question should only occur as a root clause (main clause). Based on this observation, I conclude that no why-questions may be embedded. In this view, the island-creating contexts cause interpretation problems simply because they are embedded clauses. There is nothing special about these island domains per se. Indeed, I provide evidence that a why-question cannot even embed as a complement clause. This theory predicts that if we can find another type of question that similarly modifies the proposition level, island effects should arise there, too. In this thesis, I find one such example in A-not-A questions. I argue that A-not-A questions are yes-no questions that relate a proposition to its truth values. As predicted, island effects occur in A-not-A questions.;Intervention effects arise when scope-taking elements linearly precede an interrogative phrase. This constraint resembles strong island violations, in that it also applies to why-questions and A-not-A questions, yet fails to apply to other wh-questions. In this thesis, I show that intervention exhibits variability: (i) monotone increasing quantifiers as well as non-monotonic quantifiers do not obey the intervention constraint; (ii) conversely, monotone decreasing quantifiers and focus-sensitive expressions are subject to the constraint. Based on the proposal that why-questions and A-not-A questions involve interrogative phrases that are proposition-level modifiers, my thesis proposes that scope-taking elements that take precedence over the interrogative phrases need to be topics. This proposal correctly predicts the variability in intervention effects.