A unique feature of the direct passive in Japanese
It has long been assumed that in the English passive a passive subject (such as the boy in The boy was scolded by Professor Smith) undergoes NP-movement from its original object position to get Nominative Case. Similarly, the direct passive in Japanese (e.g., Sono-otokonoko-ga Smith-sensei-ni oko-rare-ta. ‘The boy was scolded by Professor Smith.’), which has been assumed to correspond to the abovementioned English passive, has been proposed to involve NP-movement of a passive subject from the object position (Kuno 1973;Miyagawa 1989; Kubo 1990, among others). The main evidence provided by researchers which advocate such an NPmovement view for the direct passive comes from the behavior of a floating quantifier in direct passives.Miyagawa (1989) offers the mutual c-command requirement on a floating quantifier, according to which, in Japanese, a floating quantifier is allowed only when it mutually c-commands the host NP (or a trace of the host NP) which the floating quantifier modifies. Thus if a floating quantifier for a passive subject is possible in the direct passive, it has been thought to indicate that there is a trace of the passive subject in the object position, which in turn suggests that the passive subject undergoes NP-movement. Indeed, the direct passive allows a floating quantifier (e.g., Otokonoko-ga Smith-sensei-ni
oko-rare-ta. ‘Three boys were scolded by Professor Smith.’). In this way the view that the direct passive involves NP-movement similar to the English passive seems to be well motivated. However, I argue against the NP-movement view for the direct passive. I show that the evidence from floating quantifiers is not conclusive enough to conclude that a passive subject in the direct passive undergoes NP-movement. A closer inspection reveals that the mutual c-command requirement itself has to be dismissed, and furthermore, a floating quantifier does not necessarily imply that there is a trace. Therefore, theNP-movement view supported by the evidence fromfloating quantifiers is nullified. The NP-movement view has to be recast in a broader empirical investigation. I present several new pieces of empirical evidence on the direct passive, which unequivocally support the totally opposite view from the NP-movement view: a passive subject is not derived by NP-movement; rather it is base-generated in the subject position, [Spec, TP], throughout the derivation (cf. Kuroda 1979; Hoshi 1991; Kitagawa & Kuroda 1992; Matsuoka 2002). Empirical evidence comes from (i) Binding Conditions in the passive, (ii) scope interpretation (scope of negation) in the passive, (iii) Variable Binding in the passive, (iv) V-V compound (long passive) in the passive, and (v) a negative polarity item, sika-phrase, in the passive.