Heidi Harley: Numeral Quantifiers in the l-node hypothesis (reply to Mark Volpe)

[Martha McGinnis]

Hi all --

I did have a guess about NQs and NumP, that
I haven't pursued but I thought I'd throw
it out for musing:

seems to me that num and NQs might be instantiations
of the head that nominalizes roots -- the
nominal equivalent of vP, in other words,
nP. Here's the kind of thoughts I've had
about it:

as I work along I find that, with regard to
v, I more and more think that, while sometimes
the root which gets stuck onto v comes
from the phrase that is v's complement (as
in H&K denominal verbs, e.g.), other times
the root that shows up in v is a Manner
element that just gloms in there for free,
to realize the v head itself (as in
e.g. resultative constructions, way-constructions,
etc.)

In Persian, I have recently discovered (duh!)
there are essentially no verbs as we
normally think of them, but rather just
about 80-100 'light' verbs that combine with
non-verbal elements to make real verbs.
Seems like we can really 'see' the light-verb
structure there. But nonetheless I
still feel strongly that there aren't 80-100
UG-provided "v" heads, per se; rather,
there's just 2, 3, or at most, 4
(configurationally defined). The Persian
light verbs are just the few Manner elements
that get to show up in v in Persian.

Along the same lines, then, classifier
systems are just systems where there is
hardly any incorporation into n. The classifiers
that you see are like the persian light
verbs, drawn from a medium-size class of
'manner' elements that can realize n.

Semantically the analogy seems pretty good,
I think; I'm not sure how the syntax works
out, but it does seem clear that we need
some candidates for 'n', cross-lingusitically,
and classifier systems look to me like a
good bet. In any case, I don't think that
it's a problem that we can't do a nice
feature-based analysis of Persian light
verbs, nor do I think it's a problem for
classifiers, if the guess turns out to the
right. I do think v and n do have just a few
'real' values ([+/-change], [+/-external],
if you want to do v in terms of features,
which I'm not crazy about); and similarly
n could have 'Num' -style features. But
the fact that both v and n have multiple
overt realizations of each set of possible
feature values, I think, is just something
we're going to have to grit our teeth and
accept.

Or maybe there really is a set of UG-provided
light verbs (and classifiers). Who knows?

keep me posted,

:) hh

---------------------------------------------------------------------
Heidi Harley
Department of Linguistics
Douglass 200E
University of Arizona
Tucson, AZ 85721
Ph: (520) 626-3554
Fax: (520) 626-9014
hharley at u.arizona.edu