The goal of the thesis is to strengthen empirical arguments for the existence of multiple dominance structures in human language, along the lines of Citko (2000 2003, 2005). Citko (2006) argues that determiner sharing constructions also involve multiple dominance structures. However, her analysis fails to explain the availability of object determiner sharing in English. In Chapter 2, I show that her problem is solved by adopting the hypothesis that English has overt object shift to the lower position than the base-generated position of the subject, argued by Johnson (1991) and Koizumi (1993). I also reconsider the so-called anti-reconstruction effects in determiner sharing constructions, which are apparently problematic for the multiple dominance analysis. I also argue that linearization considerations cannot be triggers for movement, contrary to Moro (1997, 2000). In Chapter 3, I critically examine two recent treatments of parasitic gap constructions, Nunes (2001) and Nissenbaum (2000), and analyze them in terms of multiple dominance. The purpose of the chapter is to give an explanation for why the antecedent of a parasitic gap should be NP, which has not been given any principle explanation in the literature.
In Chapter 4, I show that he difference between Japanese and English with respect to the directionality of gapping, which is not expected under ellipsis approaches, is captured under the proposed multiple dominance approach, extending Fukui and Takano's (1998) mechanism of linearization. In Chapter 5, I provide a multiple dominance-based analysis of Korean backward object control constructions, where the embedded subject marked with nominative is shared between the matrix verb and the embedded v '. The shared subject moves to [Spec, TP] due to the EPP, which makes the multiple dominance structure linearizable, which resolves an empirical problem for Monahan's (2003, 2005) analysis on quantifier case-matching effects. Furthermore, the proposed analysis explains some differences between Korean and Japanese concerning backward control, which provides a further argument against Moro's (1997, 2000) proposal.
