The aim of this research is to show that Agree is the primary operation to apply to derivation, and that when Agree is applied, Maximization must be observed. Shima(2000) argues, against Chomsky(1998, 2000), that Move is a cheaper operation than Merge. We basically follow his Preference Principle. When we consider derivation of locative inversion, we find that Maximization can reduce the number of Probings to check the same amount of features, since it prohibits partial elimination of features. Therefore, Maximization is economically favored. However, there is some overlap between the Move Preference Principle and Maximization, since they are both involved in checking an EPP feature. In order to remove the overlap, we focus on the three subcomponents of Move, that is, Agree, Pied-Piping and Merge, and argue that Agree is the primary operation because the other effects of Move are included in Maximization and because Inside Probing is more economical than Outside Probing. Finally we consider derivation of there-construction. Since, with Maximization, there-construction would be thought not to be derived, we assume that there are two types of verb be, namely, Φ_<comp> and Φ_<def> in order to remain an associate NP where it is.
抄録全体を表示