This paper deals with the position of subjects (specifically, ga-marked subjects) in Japanese in terms of the Minimalist Program (Chomsky 2000, 2001, 2003). Based on the "Purely EPP Eliminates Reconstruction" generalization (Nevins and Anand 2003), the paper examines scope freezing effects in Japanese in contrast with scope ambiguities in English, thereby arguing for the "raising" approach, the one maintaining that in Japanese the subject originates within vP and moves to Spec-T (Ueda 1990, 1993, among others). In addition, the paper deals with a question why subject raising in Japanese differs from that in English, and shows the difference to be accounted for by proposing a parametric constraint on Move in terms of the functional-lexical dichotomy.