Desktop version

Home arrow History arrow A Minimalist View on the Syntax–Semantics Relationship: Turning the Mind into a Snowflake

Source

Loneliness of head movement

The phenomenon of head movement re-enters the stage of strictly syntactic derivations after a period in which it seemed possibly eliminable from the theory of syntax; it reappears as a syntactic process which is part and parcel of the labeling procedure, playing therefore a role of its own in the determination of interpretive properties of syntactic objects. The basic structure of phasal chunks in the framework of Chomsky (2013c, 2015b), in accordance with the earlier practice to use ‘C and T as surrogates for richer systems’ (Chomsky 2000a: 143 n. 31), thereby leaving out issues arising in the context of the cartographic enterprise both at the edges and within phase complements (see Cinque (1999, 2010), Rizzi (1997), Rizzi and Cinque (2010, 2016) for expositions of the framework and Shlonsky (2010), Boeckx (2015a: 55-58) for relevant discussion of issues arising for the integration of cartographic research into the mainstream minimalism)—although the mechanism of feature inheritance, connecting phasal heads with heads of their complements, restricts the space of possibilities in a radical way, unless enriched with additional assumptions (see Branigan (2016) for a proposal to employ multiple feature inheritance to account for the appearance and behaviour of cartographic hierarchies)—requires presence of alternating phasal and nonphasal heads in the structural skeleton of syntactic objects, with phasal heads transmitting their features downwards (a mechanism activated in the presence of unvalued features). The basic configurations for which ideally no labeling problems arise are {X, YP} structures, in which the search for the source of a label determines the lexical item as responsible for interpretive properties of the structure as a whole: ‘In a structure SO = {H, XP}, where H is a head and XP its complement, minimal search will assign the structure the category H’ (Chomsky 2013b: 66). Thus, in the clausal case, the bottom of the structure visible for the labeling algorithm is supposed in Chomsky (2013b) to contain a V0 head, providing the label required for such syntactic objects to exhibit the behaviour of verbal phrases. The picture becomes murky once the hypothesis of a-categorial roots is combined with the labeling theory. Crucially for labeling purposes, nonphasal heads, roots which occupy the bottom of the syntactic structure in particular, are assumed to be unable to provide a label for complex syntactic objects in which they are visible as lexical items destined in principle to determine the label of the whole.

The replacement of LI’s bearing categorial labels at the bottom of the structure with roots affects therefore significantly labeling, and hence also interpretive issues, with several issues touched upon already in sections 1.3.2-1.3.5, but the mechanics of head raising, with which we are now concerned, put aside. Roots, following the line of research which takes a constructivist, exoskeletal stand on their properties (variously detailed in the radically exoskeletal proposals of Borer (2004, 2005a,b, 2013a,b), the Distributed Morphology school of Marantz (1997, 2000, 2013), Embick and Noyer (2007), Embick and Marantz (2008), Embick

  • (2015)), are supposed in the labeling framework to require entering into a close relationship with a categorizing head and to be unable to label a syntactic structure by themselves. The connection with labeling is a consequence of taking labels to be properties of syntactic objects necessary for interpretive processes in the C-I component: to repeat the main line of the discussion in chapter 1, intuitively, an interpretively deficient LI, undetermined with regard to its cat- egorial properties, cannot be supposed to provide information for interpretive component(s) concerning the interpretation of a syntactic object of which it is the most prominent element. This weakness of roots, detailed in section 1.3, explains why they need a categorizing head to appear with, in accordance with the Categorization Assumption (cp. Embick and Noyer (2007: 296)); in the system of Chomsky (2015b), this process requires head movement of the root to the category specifying head, be it v* or n—A-movement which, in contrast with all theories which seek to unify different flavours of displacement, is entirely unlike internal merge of other kinds of syntactic objects, involving internal pair-merge, with the internally merged object heading the structure.
  • (9)

The theory thus makes head movement an operation sui generis, although parts of the story have been proposed and explored already earlier—both the categorization requirement, on which a root must be merged with another head, the possibility that a displaced object takes the role of the head of the structure (for a discussion of the ‘reprojecting’ head movement see e. g. Suranyi (2008)), and availability of internal pair-merge (see Richards (2009)), are parts of various theoretical proposals; yet being displaced so as to become a host for an adjunct—which is what becomes of the former head—is a novelty, explicitly commented on by Chomsky (2015b) as reversing the standard picture:

More generally, the conventional theory of head-raising seems to have the story backwards: the host should be affixed to the raised element. There is no conceptual argument against this: head-raising is a unique operation, with special properties, and there is no conceptual preference for one or the other form of amalgamation, hence no reason to reject what seems to be accurate empirically. Accordingly, raising of R to v* yields an amalgam with v* adjoined to R, and the affix is invisible to the labeling algorithm. Note that although R cannot label, the amalgam [R-v*] can. (Chomsky 2015b: 12)

The movement operation so defined exhibits properties unexpected on standard accounts of the phenomenon: in particular, burying the head at the landing site in an adjunction structure, it makes it invisible for syntactic processes (not for interpretive processes, though—the C-I component is able to interpret adjuncts quite generally, after all, and thus should be able to cope with structures like (v*, R)). The ordering of operations proposed in Chomsky (2015b: 14) reflects this fact, locating head movement towards the end of the derivation at the phase level, adapted in (7) in section 1.3.1 to a simple transitive case and repeated below as (10) for convenience:

(10) a. form R-IA by EM;

b. IM of IA in a (EPP);

c. Merge v*, reaching the phase level;

d. Inheritance;

e. Labelling; a is labelled (ф, ф);

f. R raises to v* forming R with v* affixed, hence invisible, so phasehood is activated on the copy of R, and IA remains in situ, at the edge;

g. Transfer of the copy of IA.

The ordering in (10) involves a tight relationship between steps (10d), (10e), and (10f): feature inheritance must precede labeling to enable features to participate in the process; and raising of R, beside making v* invisible for the labeling algorithm, activates immediately phasehood on the copy of R. Crucially, for the latter to be able to function as a derived phase head, it must receive features from their original bearer, v* in this case. The idea behind packing both raising of R and phasehood activation on its copy in a single step—in other words, making them components of a single operation, not two separate ones—seems to be that phasehood is not a property that can disappear and reappear indepedently on its own, after an arbitrarily long sequence of syntactic operations, as if it were kept apart in the working memory to be inserted/activated ad libitum. IA is displaced minimally, violating thereby constraints against too local movement proposed in much recent research (see e. g. Grohmann (2003, 2011), Abels (2012) for representative discussions of the issue with further references), in accordance with the general assumption of the freedom of merge—not adopted in several proposals with regard to availability of internal merge (see e. g. recently Boskovic (2016) for an antilocality-based approach within the general framework involving labeling), but proven important in a solution of the EPP-related paradox of Chomsky

(2013c) proposed in Chomsky (2015b) and for a unification of EPP-effects across phases (see also Epstein, Kitahara, and Seely (2016: 90 n. 6)). It should be recalled in passing that the mechanism enabling such raising is also discussed in Chomsky (2008):

Hence v* should transmit its Agree feature to V, and probe of an object with structural Case by v* should be able to raise it to Spec-V, a step-by-step analogue to raising to Spec- T by C. That would yield the intriguing but puzzling conclusions about raising of objects to Spec-V, particularly in ECM constructions, but perhaps generally. The evidence is compelling, but has been unclear why such rules should exist: why should objects raise to Spec-V at all, an operation that is even more odd because its effects are not visible, given that V then raises to v*? These strange facts fall into place automatically if the properties of the C-phase hold of phases generally. (Chomsky 2008: 148)

The differences between the web of assumptions of Chomsky (2008) and the framework of Chomsky (2015b) are telling: both admit the possibility that raising to a position below the v* level extends beyond ECM constructions, yet Chomsky (2015b) no longer requires that there be any motivation for the process, internal merge itself being free to take place (and thus avoiding the issue of countercyclic movement to the EPP position in both v* and C phases) and resulting structures being evaluated with respect to interpretability via an application of the labeling algorithm. It should be also noticed that the framework of Chomsky (2015b) effectively abandons the earlier idea of categorization resulting from merger of a categorizer and a root at the very beginning of the derivation, as in (11a)—an assumption stemming from the Distributed Morphology tradition, and adopted in Chomsky (2013c): ‘Another long-standing problem has to do with head-head constructions, the first step in a derivation. If the Marantz- Borer conception is adopted, these will be of the form f-root, where f is one of the functional elements determining category’ (Chomsky 2013c: 47)—and makes categorization part and parcel of head movement, thereby tying the universal nature of head raising in the first phase to the requirement of categorization of roots: {K, R} structures mentioned in (Chomsky 2015b: 8) are rather to be considered (R, K), arising in virtue of head movement, as in (11b).

 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >

Related topics