Wh- Subjects in English and the Vacuous Movement Hypothesis

In this squib I consider the Vacuous Movement Hypothesis (henceforth VMH), the notion that in English local overt wh-movement takes place except for subjects (George 1980, Chomsky 1986). There is considerable evidence that a wh-subject does not move locally to [Spec, CP] in English. However, the notion that overt wh-movement in English involves feature licensing/clausal typing with C (Rizzi 1996, Cheng 1991) implies that even in the case of wh-subjects, movement to the domain of C must still occur. Furthermore, wh-islands involving a wh-subject in the embedded clause have raised problems for the VMH under the classical treatment of wh-islands that attributes them to Subjacency. I propose to reconcile the evidence for and against the VMH via a simplification of the feature-checking system advanced in Chomsky 1995 and a treatment of overt movement that separates a feature chain (CH_FF) from a category chain (CH_CAT). The proposal resolves the discrepancies observed with English wh-subjects in a conceptually desirable way.

1 A Peculiar Asymmetry

George (1980) and Chomsky (1986) have hypothesized that local overt wh-movement takes place except for subjects (the VMH). For nonsubject questions the fact that movement occurs is clear from visible displacement, as well as from the inversion of subject and auxiliary in these interrogative structures.

(1)

1. a. [_CP What₂ has₁ [_IP John t₁ bought t₂]]?

2. b. [_CP How₂ has₁ [_IP Mary t₁ fixed the car t₂]]?

However, in subject questions it is not clear from the surface string whether wh-movement and inversion have applied.

(2) Who has fixed the car?

Traditionally, two competing analyses have been available for sentences like (2). According to the VMH, the wh-subject remains in [Spec, IP] and Aux does not undergo inversion.

(3) [_CP C [_IP who has fixed the car]]

Alternatively, the wh-subject moves to [Spec, CP] and Aux undergoes inversion.

(4) [_CP who₂ has₁ [_IP t₂ t₁ fixed the car]]

It is easy to show that both structures yield the same linear ordering.

Unlike the structure in (3), the structure in (4) places subject questions in line with nonsubject questions like those in (1), giving a unified structural analysis for both types of questions.¹

However, evidence from topicalization suggests that (3) may be the appropriate structural analysis. Lasnik and Saito (1992) present evidence that local topicalization of subjects is impossible in English.

(5)

1. a. John, I like t.

2. b. *John, t left.

3. c. John thinks that Bill, Mary likes t.

4. d. *John thinks that Bill, t likes Mary.

I assume, contrary to Lasnik and Saito's original implementation, that topicalization involves movement of a DP into [Spec, CP] on a par with wh-movement (the embedded clause in (5c) would perhaps contain a CP-recursion structure in the sense of Authier 1992 and Browning 1996).

(6) [_CP John₁ C [_IP I like t₁]]

If we assume this analysis for topicalization, then the deviance of (5b) and (5d) confirms an analysis like (7), where the subject is forced to remain in [Spec, IP].

(7) [_CP C [_IP John left]]

The evidence would thus seem to confirm the VMH and the structural analyses in (3) for subject questions and (7) for subject "topics."

However, considerations based on clausal typing and feature licensing have raised doubts about the VMH. For example, Cheng (1991) questions theVMHon the basis of her Clausal Typing Hypothesis, according to which either a wh-element or a particle must be present in CP to type a wh-question. This means that in English subject questions the wh-subject is required to be in [Spec, CP] for clausal typing. Rizzi's (1996) Wh-Criterion, stated in (8), also forces a wh-subject to raise locally to [Spec, CP] in order to license [+ wh] C.

(8)

1. a. A wh-operator must be in a specifier-head configuration with X⁰ .

            [+ wh]

2. b. An X⁰ must be in a specifier-head configuration

          [+ wh]

   with a wh-operator.

Such considerations require us to adopt a...