Syntax, Semantics, and Pragmatics of Accusative-Quotative Constructions in Japanese

Abstract

This dissertation is a data-driven exploration of the syntax, semantics, and pragmatics of the Japanese accusative-quotative construction (also known as the "subject to object raising" construction, or "exceptional case marking" construction (ECM), or "prolepsis" construction). An example: Hanako wa Tarou o baka da to omotte iru 'Hanako believes Tarou to be an idiot.' A special type of statement of propositional attitude, it exhibits epistemic specificity, and its construal involves more than one context of interpretation. Accusative subjects are interpreted as specific with respect to the beliefs of the agent of attitude (ruling out unambiguously non-specific indefinite noun phrases). A semantic and pragmatic constraint covers this observation: Accusative-quotative constructions cannot embed predications denoting existential assertions as evaluated under the domain of the belief operator generated by the matrix verb. The constraint accounts for the lack of certain scope ambiguities involving existential quantification, the inadmissability of weak cardinal floating quantifiers hosted by accusative subjects, the impossibility of embedding specificational pseudoclefts in accusative-quotative complements, and the impossibility of construing wh-accusative subjects with questions embedded in accusative-quotative complements. Furthermore, because predications with past or future tense reference and predications referring to events or temporary states depend on eventualities for their interpretations, they can only be licensed in accusative-quotative complements either through generic quantification over eventualities, or by referring to discourse antecedent eventualities supplied in the context. Such reference obviates the need for existential quantification over eventualities.

One semantic/pragmatic constraint provides a unified and more observationally adequate account than analyses that refer to the distinction between tensed and un-tensed predicates, or between stage-level and individual-level predicates, or between thetic and categorical judgments, or between referentially transparent and opaque contexts.

That scope ambiguity between an accusative subject and a complement argument is attested (provided that neither element involves existential quantitification) suggests that "reconstruction" occurs. This supports a raising analysis over a prolepsis analysis. Accusative subjects are never thematic arguments of matrix verbs in accusative-quotative constructions.