Verb-Second
Overt displacement of a wh-phrase is often accompanied by head movement, resulting in the verb-second configuration. In this section, I would like to explore the possibility that the pied-piper feature is involved in this well-known phenomenon.
The schematic configuration of verb-second is shown in (1), where XP is raised from within TP as an operator.
(1)
It is true that verb movement to C is not limited to cases where an overt operator is raised to Spec,CP. It is also found in yes/no questions, for ex ample, leading some researchers to posit a null operator in Spec,CP even for yes/no questions. There is a sense, however, in which overt phrasal movement induces verb movement to C, given the rather systematic pat tern displayed by wh-questions in Germanic. The question is how.
The idea I would like to pursue is that the pied-piper feature that accompanies overt phrasal movement to Spec,CP is responsible for verb movement to C. Note first that the raised finite verb marks the right edge of the phrase in Spec,CP, since verb-second arises in the head-initial structural representation. Compare the verb-second configuration in (2a) with the structure in (2b), where the pied-piper feature (abbreviated as F(PP)) is realized as part of the raised operator.
When the pied-piper feature is realized on the right edge of the displaced operator, its positioning is strikingly similar to that of the raised verb. It is, therefore, not unreasonable to assume that the pied-piper feature of the fronted operator induces verb movement as well.
To implement the idea that verb-second is due to the pied-piper feature, we need to have a theory of head movement in general. Chomsky (2000, 2001) has suggested that head movement is not part of the narrow syntactic computation but belongs to morphophonology, on the grounds that it differs from phrasal movement in significant respects. The properties listed in (3) certainly set head movement apart from phrasal movement.
(3)
Zwart (2001), on the other hand, argues that the agreement system lies behind head movement, while recognizing that it has morphophonological aspects as well. My claim is that head movement is morphophonological but has a syntactic origin (namely, the pied-piper feature), at least in some cases.
The pied-piper feature must be hooked to the phonological material. Now suppose that when Agree takes place, the pied-piper feature is copied onto the head that contains the probe. This is an extension of the idea suggested by Chomsky (1998) that Agree involves feature copying. Feature copying converts the structure in (4a) into the one in (4b) after the head H agrees with the feature F.
(4)
For empirical support from switch reference and negative concord, see Watanabe 2000, 2004a. My proposal is that the pied-piper feature is also copied onto the probing head when Agree takes place. It thus looks as if the pied-piper feature itself entered into the checking relation with the EPP feature. Once it becomes part of the head that contains the probe, the pied-piper feature asks for phonological material. It copies the phono logical content of the closest head when the head where the copied pied piper feature is located does not have phonological content itself. Here, the closest head means the head of its sister. This copying of phonological material results in head movement. Since it takes the form of copying the phonological content of a head onto the next higher head, the properties in (3) follow straightforwardly.
Notice that this theory also predicts V-to-T movement when the subject is raised to Spec,TP. Obviously, this is a wrong result for English sentences that do not contain an auxiliary. Essentially the same problem arises in embedded questions in Germanic in general. As illustrated by the Danish examples in (5), there is no verb-second effect in indirect questions.
(5)
But the theory I am proposing incorrectly requires verb raising to C in (5b), unless it is modified in some way.
Now I would like to claim that head movement needs a catalyst in addition to the copied pied-piper feature. More generally, I propose that head movement is induced when the request for phonological material is combined with a catalyst. It does not take place otherwise. We should in fact allow something other than the copied pied-piper feature to ask for phonological material, because the verb raising that creates verb-initial sentences (including yes/no questions) takes place without an accompanying phrasal movement. The copied pied-piper feature cannot be held re sponsible for verb raising in such cases. Now, the role of a catalyst is to fulfill the request for phonological material only in its presence. The failure of verb raising in (5b), then, can be accounted for quite easily. In contrast to the matrix C, the embedded interrogative C does not contain a catalyst and thus fails to induce head movement, even though the pied piper feature is copied. In other words, the request for phonological material is aborted in the absence of a catalyst.
The catalyst replaces Chomsky’s (1993) strong V-feature. The difference is that while the strong V-feature is assumed to trigger overt head movement by itself, a catalyst works in collaboration with a request for phonological material such as the pied-piper feature.
Although at first sight the theory looks as stipulative as the theory of strong V-features, it actually succeeds in capturing the empirical properties of head movement in an optimal way. The notion of a catalyst accounts for the robust generalization (e.g., Roberts 1993; Vikner 1997; Rohrbacher 1999) that rich verbal morphology is linked to V-to-T movement. In other words, the catalyst for V-to-T movement is the property marked on the T head that is realized as rich inflection. At the same time, Jonas (2002) observes that Shetland English retained V-to-T movement until quite recently despite the absence of rich inflection. For these cases, the brute force specification of the catalyst is needed, which is triggered by syntactic cues in parameter setting, as Jonas suggests. The same parameter setting triggered by syntactic cues probably applies to the catalyst on the matrix interrogative C in Germanic in general as well. Thus, empirical evidence points to the conclusion that the presence of a catalyst is triggered in parameter setting by inflectional morphology or by syntactic cues. In this sense, the category of catalysts is a mixed bag, indicative of morphological arbitrariness, which is incompatible with the notion of strong features. The notion of catalysts is intended to capture the case-by-case nature of head movement as copying of phono logical material.
The idea that head movement is copying of phonological material onto a head captures the peculiarities listed in (3). Copying of phonological material takes place in phrasal movement as well. The pied-piper feature connects phrasal movement to head movement in terms of the request for phonological material.
