Word Order and Clause Structure in Early Old Japanese

要旨

The purpose of this paper is to show that there is a striking difference in word order between Modern Japanese and Early Old Japanese. Early Old Japanese lacks the canonical transitive pattern [Subject-ga Object-o V]. The basic word order in Early Old Japanese is [Subject-ga/no Object-Ø V], in which the subject is marked by the genitive ga or no and a morphologically unmarked object must appear immediately adjacent to the verb. When the object is marked by wo, it is obligatorily moved over a subject, resulting in [Object-wo Subject-ga/no V]. Following Miyagawa (1989) and Miyagawa and Ekida (2003), I argue that a bare object is assigned abstract case under the strict adjacency requirement, but that wo in Early Old Japanese does not function as an accusative case. The particle wo differs crucially from the case particle o in Modern Japanese in that it marks not only the direct object of a transitive verb, but all kinds of internal arguments of both transitive and intransitive verbs. Furthermore, wo conveys a definite interpretation. An element marked by wo moves to a particular structural position, namely Spec(vP) or Spec(CP), where it is assigned definite/topic interpretations.