[SynSem] Alec Marantz (NYU)

Syntax Beyond Merge: Replacing Head Movement with “Mirroring”

A question arises for contemporary syntactic theory that derive syntactic structure from the
recursive combination (“Merger”) of constituents: How much of syntactic well-formedness
should be computed at the point of Merger in a derivation and how much should be determined
from constraints or rules computed over the resulting Merged tree structures? I am investigating
an approach to syntax in which syntactic well-formedness involves strictly local conditions on
linear projections of Merged trees, following the lead of T. Graf’s work within Tier-based Strictly
Local grammars. Within generative grammar, principles that involve “c-command” relations
between constituents already refer to a linear path through a tree; I generalize the use of such
paths to two linear projections from the hiearchical tree structure. The first involves just phrases
and the heads that combine with phrases; this is the c-command projection. The second
involves heads that Merge with other heads; this is the head projection. Linearization of
constituents involves both projections, with the c-command projection ordered by Kayne’s
“Linear Correspondence Algorithm” (higher is lefter) and the head projection ordered by a
version of Brody’s “Mirror” (higher is righter).

This talk will explore the empirical consequences of this approach for the English auxiliary
system, for subject-aux inversion and verb second, for the explanation of Holmberg’s
generalization, and for explaining various Final-Over-Final condition effects within words. On
this last point, note that, although “unlockable” is famously ambiguous, “unlockability” can only
mean the property of being able to be unlocked, a Final-Over-Final effect explained by the
theory.