Abstract:

This thesis consists of two parts. The first part is a theoretical description of grammatical relations in Japanese (Chapter 3). The second part addresses three theoretical issues of two constructions discussed in the first part of the thesis, a reflexive construction (Chapter 4 and Chapter 5) and a pseudo-raising construction (Chapter 6).

In Chapter 3, grammatical relations of the Japanese language will be detailed. Though there are several studies which examined grammatical relations in Japanese in the past, all of them are neither comprehensive nor conclusive. This thesis attempts to give a more comprehensive and fresh look at the grammatical relations of Japanese. More than twenty constructions, including less discussed oblique 'subject' constructions, are examined in terms of Role and Reference Grammar (RRG) which has developed a fine-grained system to analyze grammatical relations.

In Chapter 4, a long-standing issue of the antilocality effect seen in the Japanese reflexive constructions is addressed. It is first shown that the antilocality effect of the Japanese reflexives has nothing to do with the notion of lexical reflexivity advocated in the literature. It is demonstrated that the antilocality of the Japanese reflexive is due to a peculiar subcategorization nature of the 'antilocal verbs'. That is, what we call 'antilocal verbs' do not allow metonymy and require NPs of certain meaning (quale). It is concluded that the antilocality effect is not specific to reflexive constructions and, accordingly, there is no need to posit a constraint on the behavior of the reflexive itself. For a formal treatment, RRG representation and qualia structure are employed.

In Chapter 5, an interpretive issue of the reflexive construction which has not been discussed before in the literature will be addressed. Two things will be demonstrated. First, the Japanese lexical reflexives (i.e. morphologically reflexive-marked) behave exactly as expected from the universal principle of Condition R (Lidz 2000, 2001). Second, more importantly, it will be pointed out that the Japanese syntactic reflexives (i.e. unmarked verb + reflexive) do not show the behaviors expected from Condition R. It will be argued that the unexpected behavior of the Japanese reflexive is due to the interaction of two types of focus structure: focus by intensifier and focus by construction.

In Chapter 6, a construction hitherto classified as 'raising' will be examined. It will be claimed that the alleged 'raising' construction should be analyzed as a control construction. It is shown that the data given for a 'raising' analysis do not constitute evidence exclusively for a 'raising' analysis and there are many more pieces of evidence against a 'raising' analysis than reported before. Furthermore, the data which were presented against a control analysis turn out not to pose a genuine problem. The chapter will be concluded proposing a possible logical structure representation that subsumes the peculiarities of the construction as control.