Desktop version

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

Avoiding substitution

Adjunction in head-head structures is taken to result in an ordered pair on the account in Chomsky (2015b), in accordance with the general choice for modeling adjunction structures; it requires furthermore that tampering with a structure be allowed to the extent that the head of a structure {X, [1]} (where{WP} may = {ZP}, may Ф {ZP} or may be null) be replaced with (X, Y) to result in {(X, Y), {Y, ZP}} to take into account the effect of head movement.

This analysis of head movement requires additional elucidations, lest it lead to abandoning NTC, with the hypothesis that no substitutional operations are authorized within the bounds of narrow syntax as its corollary. A relaxation of the ban on substitutional operations might be possibly argued for with regard to parts of syntactic objects which are unlabeled yet, labeling having the ‘freezing’ effect, but without further restrictions it would open the road for other substitutional countercyclic operations, freely modifying objects populating the extended edge of a phase (including countercyclic merge into an ‘empty position, postulated inter alia to account for effects studied by Lebeaux (1988, 2000)), and would lead to postulating a new kind of merge, as Chomsky (2013c: 40 n. 20) notes. A radical solution would be to deny entirely that head movement modifies the object at the landing site and to affirm that ‘the replacement of Y0 in an SO with a Y0max category is beyond the generative power of Merge,’ as Narita and Fukui (2016: 15) do, proposing instead that head movement involves external merge and creates another syntactic object, related to the original one in a doubly-peaked manner, along the lines of (12).

(12)

Two-peaked structures—beside raising non-trivial issues of interpretive unification of such syntactic objects in a correct way—does not lead to effects summarized in the quote from Chomsky (2015b: 12) above, the interaction of the head complex with the main structural spine for the purpose of labeling in the first place. Certain features of the structure in (12) might be mirrored, on the other hand, if—starting with head movement structures being represented as ordered pairs, i. e. as adjunction structures—one raises the question of their relationship to phrasal adjunctions, for which a simpler {{XP}, YP} structure seems preferable, as discussed in section 2.1. Creation of adjunction structures in the head movement case is more complex than in the case of ‘external pair-merge' One way of assimilating head movement to phrasal adjunctions sticking rigidly to the ban on changing structures already created would involve more than an embedded subroutine of internal self-merge—to get an {{X}, Y} via head movement, without altering the input structure at that, one should enrich the procedure in (3):

  • (13) Operations in head adjunction, 1st version
  • 1. Take an SO = {X, [1]}, where{WP} may = {ZP}, may Ф {ZP} or may be null;
  • 2. take two LI’s X and Y;
  • 3. self-merge X;
  • 4. merge the output of 3 and Y;
  • 5. merge the output of 4 and the SO in 1.

(14)

A structure generated in this way is reminiscent of (12) in that neither involves tampering with the existing structure; it also incorporates to some extent the conception of head movement as ‘reprojecting’ (mentioned already above), on which it would be the case that Y would undergo internal merge, as in (15).

(15)

Differences abound, though; the raised head Y in (15) does not enter into a relationship with X as envisaged on the theory of Chomsky (2015b), thereby leaving the change of status of X after movement of Y without sufficient explanation; furthermore, ‘reprojection’ of Y in (15) would mean in the current framework providing a label for the entire structure in (15)—whereas it is assumed to be too weak to label a structure by itself, standing in need of (i) feature transmission to label a lower stratum of the structure, and (ii) head raising for categorization reasons in the case of roots (the exact status of T-raising remaining to be clarified, see Chomsky (2015b: 15)). The structure in (14), on the other hand, built with the help of nested applications of merge, makes X inactive much as it does so in the case of phrasal adjuncts, turning in this case both the base position and the landing site buried in an adjunction structure into copies, not participating in labeling. It does have virtues similar to those of (12), leaving the structure existing before head movement unaffected, but at the same time it does not fully derive labeling properties arising in virtue of head movement: the copy of Y would be expected to be inactive as much as the copy of X, their status would thus be the same, contrary to what is hypothesized to happen in R-to-V raising; it is also unclear why the complex created by head movement in (14) does not lead to a halting problem for the labeling algorithm.

  • [1] WP}, {Y, ZP
  • [2] XP}, YP} structure seems preferable, as discussed in section 2.1. Creation of adjunction structures in the head movement case is more complex than in the case of ‘external pair-merge' One way of assimilating head movement to phrasal adjunctions sticking rigidly to the ban on changing structures already created would involve more than an embedded subroutine of internal self-merge—to get an {{X}, Y} via head movement, without altering the input structure at that, one should enrich the procedure in (3):

    • (13) Operations in head adjunction, 1st version
    • 1. Take an SO = {X, {{WP}, {Y, ZP
 
Source
< Prev   CONTENTS   Source   Next >

Related topics