Wh Movements out of complement subject and complement VP
The derivation (1), which violates the SCC, misgenerates (2a) if SEE represents the finite form of the verb, and (2b) if SEE represents the nonfinite form.
(1)
(2)
The explanation for (2a) was given in section 2. We turn now to (2b). The structural analysis of (2b) to which conditions on proper binding apply is (3).
(3)
By the Opacity Principle, whoi and whatj properly bind their respective traces in S2. Nonetheless, the binding of ti by whoi is improper with respect to another condition on binding that concerns structures of obligatory control.20 In short, the subject of a wh infinitive construction is a position of obligatory control, and structures whose analysis is (4) are structures of obligatory control (see Chomsky and Lasnik (1977) for discussion).
(4)
(5)
In (5a), who is interpreted as the object of see; and John, as the subject of see. That this interpretation must hold for all constructions containing the structure (4) is illustrated by (5b,c). (5b) shows that a lexically filled NP may not occur in the subject position of a wh-infinitive. (5c) illustrates the same point if who is analyzed as the complement subject. Yet even if who is analyzed as being in the embedded COMP—thereby binding a trace in the complement subject, as in (6)–(5c) is still ungrammatical.
(6)
In this way, (5) demonstrates that the subject of a wh-infinitive may not contain lexical material or its derivational reflex, a bound trace. The interpretation of these subjects is accomplished by a rule of obligatory control. It follows from this fact about (4) that (3) is ill-formed and therefore that (2b) is a case of misgeneration.
It is assumed here that the above-mentioned fact about (4) holds universally and therefore need not be stipulated as part of any language-particular grammar. Given that this fact is part of Universal Grammar (UG), nothing more is required to account for how (2b) follows from general principles. Of course it may be that this fact about (4) can be derived from other independently motivated principles of UG. However, such considerations are methodological rather than empirical, and as such do not affect the derivation of the SCC in question. This is not to say that the issue is unimportant—see appendix 2 for some discussion.
