Revisiting the two double-nominative constructions in Japanese

Abstract

In Kuno and Johnson (2004), it was proposed that the double subject construction in Japanese (e.g., John-ga otoosan-ga sinde simatta ‘John - his father has died’) has a complex sentence structure of the type of [_S1 NP₁-ga [_S2 NP₂-ga Verbal]]. This structure was intended to account for the unacceptability of Yamada-sensei-ga inu-ga o-nakunari-ni natte simatta ‘Professor Yamada - his dog has died’ (Shibatani 1976): Subject Honorific Marking is clause-bound, and therefore, it cannot apply to the embedded clause verbal with the main-clause subject NP1-ga as target of deference. It turns out that this analysis produces an apparent paradox with respect to the licensing of negative polarity items (NPIs), which is also a clause-bound rule: an NPI can show up in the main-clause subject position with a negative verbal in the embedded clause, as in Taroo-sika otoosan-ga hukeikai-ni syussekisi-nakatta ‘Taro was the only one whose father attended the teacher-parent meeting.’ That is, it appears that while Subject Honorific Marking calls for a complex-sentence structure for the double-subject construction, NPI Licensing calls for a simplex-sentence structure. This paper is intended to resolve this paradox on the basis of (i) the dual role the verbal in the [_S1 ... [_S2... Verbal]] construction plays, and (ii) the properties of NPI Licensing, including the notion of “clause-boundedness” as it applies to the rule, that are very different from those of Subject Honorific Marking.