Alec Marantz: Gender in DM (response to Bobaljik posting)

[Martha McGinnis]

Suppose it's the case the when you make a verb from a root, the choice of
vocabulary item for the little v node creating the verb never depends on
the gender of the root in the language.

Suppose, then, the gender that we see on adjectives and verbs in agreement
with a noun isn't directly a property of the root but rather of the little
n node that creates the noun from the root.

Suppose, then, that we need to assume a phase-like cyclicity to the
derivation such that little n causes the LF and PF interpretation of the
constituent it heads, prior to the merger of the nP into a larger structure.

Then, the root information that's relevant to the gender that triggers
agreement will have an opportunity to check/determine/constrain the gender
features on little n prior to within-DP agreement, even though the syntax
doesn't deal with specific roots but only unspecified root nodes.

Actually, I'm not sure how many of these suppositions/suggestions are
necessarily linked.  However, the notion of a cyclic derivation (by phase)
does allow for root-specific information to influence syntactic derivation
to the extent that the features of concern can be argued to be (syntactic,
in this case uninterpretable) features of nodes that trigger the end of the
cycle.

--Alec Marantz