Akmajian (1977) argues from the standpoint of autonomous syntax that participial perception verb complements (henceforth: PPVC's) and infinitival perception verb complements (henceforth: IPVC's) derive from single NP constituents with heads as in (1) and nonconstituents as in (2), respectively:
(1)[S [NP we] [AUX past] [VP [V see] [NP [NP the moon] [VP rising over the mountain]]]]
(2)[S [NP we] [AUX past] [VP [V see] [NP the moon] [VP rise over the mountain]]]
Along with this, he rejects the view that what distinguishes PPVC's from IPVC's is the progressive aspect.
The purpose of this paper is twofold. One is to argue, as counterarguments to Akmajian's analysis, that the two complements come from a common sentential complement and that the contrast between them is one of nonprogressive vs. progressive aspect. The other is to show that an analysis of the progressive made in K. Inoue (1979, 82) within the framework of 'localistic' theory will shed new light on the account of syntactic and semantic behavior of PPVC's.
In Sec. 1 I present a number of arguments which indicate that the relation between IPVC's and PPVC's corresponds to that holding between nonprogressive and progressive constructions. In Sec. 2 I give arguments for the underlying structure of two PVC's being a sentential complement and suggest, by drawing attention to the fact that cornplementizerless sentences cannot occur in focus position, that Akmajian's argument for (1) on the basis of the constituent structure tests such as Pseudo-CIeft, Cleft, etc. does not hold and that we can assign IPVC's as an S complement with a null complementizer. Moreover, I point out a case of PPVC which likewise fails in Akmajian's tests and thus suggests that PPVC's have the same underlying structure. Sec. 3 is devoted to the discussion as to how to derive PPVC's from a sentential complement, i.e. how to account for the facts like Number Agreement which Akmajian claims constitute evidence for the head-complement structure of PPVC's. There are two ways conceivable to resolve this seemingly paradoxical situation: One is Raising, and the other, `Pseudo-Modifier Creation', a transformation proposed by Declerck (1981b.). I support the latter over the former, because it can account for the facts Akmajian's arguments are based upon as well as the above-mentioned exceptional case of PPVC. Sec. 4 is concerned with an analysis of the progressive from a localistic' point of view and its relation to PPVC's. It is shown that the two sources I postulate for the progressive in K. Inoue (1979, 82) are reflected in PPVC's.
From the above, we come to a conclusion that as far as PVC's are concerned, the hypothesis of 'autonomous syntax' does not hold.