OV to VO und Verwandtes

[DAVID GIL]

Frans' analogy between OV>VO and dual>no-dual is a
fascinating one.

Without taking issue with any of the points Frans' makes,
I think it is important to note, also, one difference
between the two changes, which may weaken the analogy
somewhat.  The dual>no-dual change is, of course, as
Frans recognizes, one particular instance of the well-
known generally if not always unidirectional course of
diachronic change whereby free forms become bound forms
which then erode and ultimately disappear.  What this
entails -- and again, Frans suggests this in his last
paragraph -- is that the mechanisms of dual loss (ie.
attrition of morphology) are qualitatively different
from the mechanisms of dual acquisition (the wearing
down of free forms to bound forms).  In contrast,
word-order-change, in principle, involves changes
*within* a single module, or grammatical domain, or
what have you, and thus has a much more symmetric
feel to it.

Therefore, to me at least, it would be more surprising
to find a predominance of OV>VO over VO>OV, than a
predominance of dual>no-dual over no-dual>dual.

(But I have to stop and wonder, at this point: is it
really the case that there have recently been more
losses than acquisitions of duals?  Maybe it's just
that morphologists, who would be the ones to document
the losses, have done their homework better than
synacticians, who would be the ones to document the
acquisitions.  Interestingly, this very list was
recently the venue of a discussion involving, primarily,
Grev Corbett, Alan Dench, Edith Moravcsik and myself,
which dealt with the problem of distinguishing quantification
from number marking -- I think some of the forms we were
discussing were in fact number words on the verge of
becoming number markers.  So maybe there are more
acquisitions of duals taking place out there than we
are aware of.)

But let's say the assumption is true:  there is more
dual>no-dual than no-dual-dual.  This makes one stop and
ask: isn't the same true for LOTS of grammatical
categories.  If we take Frans' two initial assumptions
and replace "dual" by "plural", or "feminine", or
"instrumental", then won't the two assumptions remain
true, and hence, mutatis mutandis, all of Frans' other
points?  Which I find rather puzzling, since this in
turn leads to the rather unexpected conclusion that
the further back you go in time, the more morphologically
complex languages were in general.  Surely this can't be
true, right?

David Gil