-
1. LF-Effects of Head Movement
- The MIT Press
- Chapter
- Additional Information
1 LF-Effects of Head Movement The claim that head movement lacks semantic e¤ects can be seen as a justification for placing this operation exclusively in the PF-interface(s), seeing it as purely an aspect of the interface between narrow syntax and the morphological and phonological operations that ultimately derive phonetic/phonological representations. I do not wish to deny that there may be cases of PF head movement (subject procliticization in French— for example, Kayne 1983, Rizzi and Roberts 1989, and section 3.5.1 below on Romance subject clitics generally; see also chapter 4, note 20, on English pseudogapping). But, in order to show that head movement is not exclusively a PF-operation, we need to find one instance where head movement has an e¤ect on the semantic component—that is, where it interacts with what are most plausibly thought of as LF operations. Given the standard assumptions concerning the organization of the grammar (essentially any version of the ‘‘T-’’ or ‘‘Y-model’’), this will be enough to deny the strong assertion that it is impossible for head movement to be part of the narrow syntax (in fact, a stronger assertion than that actually made by Chomsky, quoted in the introduction), although of course it leaves open the possibility that not all instances of head movement are part of the narrow syntax. A number of proposals that head movement a¤ects LF have been made (see for example Cinque 1999, 102–103, 184 n. 8; Zwart 2001). In this section, I will present two cases where head movement appears to affect LF, one of my own and one due to Lechner (2005). Together, the two cases indicate that head movement can interact with polarity licensing and the determination of scope and reconstruction e¤ects, properties we would normally attribute to LF-sensitive operations. I will also show that a remnant-movement account of these phenomena faces serious problems. Finally, I will argue, following Matushansky (2006), that there may be a good reason why many instances of head movement lack clear semantic e¤ects. In fact, in this respect, head movement is similar to A-movement. This leads to the general conclusion that head movement does not di¤er fundamentally from A-movement in its relationship to LF. Since A-movement is standardly assumed to be part of narrow syntax , head movement should also be taken to be part of narrow syntax on the basis of the evidence presented here. 1.1 Head Movement and Licensing Polarity Items Here I will argue that negative polarity any is licensed in subject position of interrogatives by subject-auxiliary inversion. Assuming that licensing of elements of this kind is an LF matter, and assuming that subjectauxiliary inversion is head movement (it is standardly seen as T-to-C movement; see section 1.3 for a brief discussion of the remnantmovement alternative), we have here an instance of head movement affecting LF. Given the standard T-/Y-model of the grammar, then, the proposal that no head movement is narrow-syntactic is refuted. Polarity items such as anyone, anything, and so on are dependent expressions that cannot appear without a special element to license them in the same syntactic domain, as the following examples illustrate: (1) a. Did you see anyone? (Questions) b. If you see anyone, let me know. (Conditionals) c. John is taller than anyone. (Comparatives) d. I didn’t see anyone. (Negation) e. *I saw anyone. (Not licensed) To go into the precise syntactic and semantic details of how polarity items are licensed would take us too far afield here (for an overview of the relevant literature, see Horn and Kato 2000, 9–11). The important point for our purposes is that polarity any has to be asymmetrically ccommanded by the element that licenses it, as examples of the following type show:1 (2) a. *Anyone didn’t see John. John didn’t see anyone. b. Nobody said anything about it. *Anything about it, nobody said. c. *That nobody trusts him bothers anyone. d. *Few professors and any students were at the party. 8 Chapter 1 [3.87.209.162] Project MUSE (2024-03-28 22:53 GMT) The c-command condition was originally seen as an S-structure condition (see for example McCloskey 1996). That it cannot be a condition on underlying representations is shown by the fact that movement a¤ects it, as the pair in (2b) illustrates (and cf. *Anyone wasn’t seen...