Desktop version

Home arrow History

  • Increase font
  • Decrease font

<<   CONTENTS   >>

Moving to be a host

The status of the copy of Y in (17) and of its landing site requires further elucidations. First, given properties of the head movement operation, being a copy of a head cannot be hastily taken to give rise to syntactic behaviour and interpretive properties similar to those exhibited by copies of phrasal objects. Nor would it be an intuitively satisfying result, either: chains arising during head movement are not interpreted in a manner analogous to chains of arguments—that was one of arguments in favour of the PF theory of head movement in Chomsky (2001):

There are some reasons to suspect that a substantial core of head-raising processes (...) may fall within the phonological component. One reason is the expectation of (near-) uniformity of LF interface representations (...). The interpretive burden is reduced if, say, verbs are interpreted the same way whether they remain in situ or raise to T or C, the distinctions that have received much attention since Pollock 1989. As expected under (1), verbs are not interpreted differently in English versus Romance, or Mainland Scandinavian versus Icelandic, or embedded versus root structures. More generally, semantic effects of head raising in the core inflectional system are slight or nonexistent, as contrasted with XP-movement, with effects that are substantial and systematic. That would follow insofar as head raising is not part of narrow syntax. (Chomsky 2001: 38)

Retaining head movement as a syntactic process (at least for core cases) does not affect the fact that interpretive consequences of head movement are entirely different than those expected if it created a discontinuous object analogous to chains of phrasal displacement; movement of a head interpreted as a modal operator does not put the expression in its scope, thereby changing semantic properties of the entire expression by scoping over it. And a closer inspection of structures created by head movement on the proposal of Chomsky (2015b) reveals that already applying the notion of ‘occurrence’, in terms of which chains are analyzed, creates difficulties—as it might be expected if the head at the landing site becomes adjoined to the moved head. Adjunction does not create a new occurrence of the host, nor has it ever done so (true enough, it used to be supposed to create a new segment thereof, but this cannot count as an occurrence in the sense required by the theory of chains). In the case of vP phases, displacement of R to the phase head results, on the analysis in Chomsky (2015b) in a structure along the lines of (18).

Epstein, Kitahara, and Seely (2016) comment on structures as in (18) as follows:

Note that we do not consider the nature of ‘occurrence’ with respect to pair-Merge, and interesting questions arise. For example, if there is only one occurrence of R in (2), as claimed, then what is the element represented as “R” in (R, v*>. The little v* is not an occurrence of R (since (R, v*> was not created by set-Merge). So, what is it? Note further that if there is only one occurrence of R in (2), then the following situation arises: Derivationally R was internally Merged to v*, but representationally there is only one occurrence of R in the resulting structure; so this is movement (derivationally) but nonmovement (representationally). (Epstein, Kitahara, and Seely 2016: 91 n. 10)

Although it has an air of paradox to it, the picture is pretty much as it should be expected to be (except that ‘(R, v*>’ in the quote above should be rather ‘(v*, R>’—it is v* that is adjoined to R, not the other way round): it is still the ‘copy’ of R that is interpreted, although its role undergoes modifications by entering into an interaction with v*. The adjunction structure created by head movement—as in the analysis in Chomsky (2015b) for cases of (v*, R>—behaves in several respects as it might be expected on the part of an adjunction structure: it does not create a new occurrence of the host, the adjoined structure being most plausibly interpreted in a way partly analogous to adverbial adjuncts in the phrasal domain, i. e. in an operator-like way (partly analogous, because adverbial adjuncts are most plausibly seen as endotypical functors, as in Thomason and Stalnaker (1973) and related work, whereas an adjoined v*-head is interpreted at the featural level, such features operating exclusively on the type-specification of R); differences are not surprising given the difference in the complexity of objects in both cases and the featural level involved in the case of head-head movement. The mechanism of head movement formulated in (16) is obligatorily applied in structures in which Y is unable to be the source of label, as in the case of roots; and it is restricted to cases in which both participants stand in an independently required featural and structural relationship. It stands to reason that cases other than root raising—by hypothesis occurring universally—and at least partly T-to-C raising (with qualifications mentioned also in Chomsky (2015b: 15)) might in general need explanations in terms of post-syntactic processes unless the required relationship between heads is present due to a lexically determined property of a higher head other than a phase head, as it might be the case with the notorious problem of labeling in a subclass of {R, CP} structures for which Chomsky (2015b) assumes labeling by C—available if the LI heading CP raises to R, the latter belonging to a lexically specified class of roots allowing such operations due to its featural specification. If a head undergoes successive displacements, the structure created in this way becomes complex in a way preserving scopal dependencies among such operator-like elements: were it the case that after movement of X to Y the structure were created as [1] instead, adding a new layer of adjunction:


Such conceptualization of head movement preserves the intuition behind such notions as ‘extended projections’ of Grimshaw (2005) or ‘electric current’ running through members of a head-chain (the metaphor of Zwart (2001), who requires that valuation of features be involved), although in a different way and with different properties: there is no mechanism of projection, the operation of head movement affects obligatorily only the lowest part of the structure, viz. the root and the phase head with which it enters into a featural transaction; and both the status of the copy of Y in (17) is distinct from the more standard take on head movement, and labeling properties of both Y and {{X}, Y} are different than on the projection-based account. In contrast to X in (17), neither copy of Y may be supposed to qualify as the head of a chain (for reasons which were made clear in connection with adjunction structures), but the crucial difference seems to be that its internal merge resulting in a non c-commanding position (a property present also when head movement is modeled with the help of ordered pairs, as well as when it involved adjunction of Y to X) makes the base position retain its visibility for labeling purposes without losing the information that the occurrence in {{X}, Y} appears now in a structure which is acceptable for the labeling algorithm qua an adjunction structure. It is thus not necessary to exclude such copies from the set of occurrences, attributing exclusively to simple applications of merge the power to create ones; in both contexts the occurrence of Y is well-defined. It must be noted, though, that the account of the effects of head movement given in section 1.3.3 covers interaction at the level of features which are seen in the C-I component as metalinguistic instructions; if an LI is interpreted as an element of the object language, the picture becomes much more complicated. Since the details of the syntactic theory of head movement remain to a significant degree unclear, it would seem premature to attempt to deal with such cases here, given that the working hypothesis concerns the influence of narrow syntax on the C-I component, not the other way around.

  • [1] X}, Y}, and subsequent movement to Z led to {{Z}, X}, the interpretive result would correspond to ‘Markovian’ adjuncts postulated in the realm of phrasal adjunction (see e. g. Irurtzun and Gallego (2007) with further references); rather, the structure is {{Z}, {{X}, Y
<<   CONTENTS   >>

Related topics