Japanese embedded questions are nominal: Evidence from quantificational variability effect

Abstract

Abstract The exhaustivity of an embedded interrogative sentence can be altered by the presence of an adverb in the matrix clause. This phenomenon, known as Quantificational Variability Effect (QVE), manifests itself in a peculiar way in Japanese. A QVE-inducing adverb can take the form of a numeral classifier that agrees with the embedded Wh-phrase. While a QVE-inducing numeral classifier appears to be associated with an embedded wh-phrase, it is not clear how such an association can be established. I argue that Japanese embedded questions are implicitly nominalized in the fashion similar to the internally-headed relative clause construction, and that the nominalized embedded questions are treated as concealed questions. The proposed analysis gives a very simple account for the puzzling QVE construction, as the floated quantifier structure with a concealed-question-denoting NP is commonplace. The paper examines a variety of phenomena, such as doubly headed relative clause structure and selectional restrictions on QVE, which support the nominal structure of Japanese embedded questions.