Order of verb affixes

[Koontz John E]

On Wed, 10 Nov 2004, Rory M Larson wrote:
> I don't know about IO, but to the best of my knowledge pretty
> much all verbs that begin with u- in OP are formed from the
> locative prefix u- < *o-, meaning 'in', 'into' or 'in the context'
> of something, plus the root.

I think the idea is that once there was a Proto-Siouan prefix *o- 'in'
with, hypothetically, the Proto-Siouan morphosyntax (morpheme order):

INCLUSIVE-o-PRO-STEM

then any verb with a stem starting with *o for other reasons was in danger
of being reinterpreted as having that o-locative prefix present, even if
the morphosyntax was also different, maybe something like:

INCLUSIVE-PRO-o...

Imagine that 'cook' was historically just a simple root *ohaN in which of
was just an "organic" part of the root.  Naturally, once it was reanalyzed
as being o 'in' + haN 'cook, boil', it would be easy to take the o 'in'
part as referring to the cooking vessel, even if it didn't originally.

I can think of several kinds of evidence for this having happened:

1) Some Siouan language retaining the original morphosyntax or something
otherwise differing in behavior from the locative pattern in that
language.

2) The discovery of forms from outside Siouan that argued for a
non-locative origin, e.g., cognates from Catawba or further afield with a
shape that argued for PS *ohaN being unitary.  They might not start with
o, of course, but they should be regularly cognate with *ohaN and not
merely *haN.

3) We might wonder about verbs with locative *o that have no semantic need
of a locative.  Of course, forms like this are not a very strong argument.
Maybe we just don't understand their logic!

An example parallel with this involves Eskimoan final -p, -t, -k, and -q
in nouns.  These are the only final stops in most Eastern Eskimoan
languages.  I think there are some c^ finals in Western Eskimo.  A -p is
the regular ergative case ending, and -t and -k are the plural and dual
endings for the absolutive and ergative cases.  Final -q is apparently
free of underlying meaning in nouns, but there is a strong tendency to
interpret it as the absolutive singular, and to add it as such where it
was not historically present or delete it where it belongs but seems to
break the pattern.

In addition, nouns that underlyingly end in p, k or t are somewhat
problematic because the unmarked absolutive form tends to look like an
ergative, dual, or plural.  These nouns tend to be manhandled in some way
to avoid the obscuring of the paradigmatic pattern - deletion of the final
consonant, addition of additional elements meaningful or more or less
arbitrary, etc.  I could probably track down examples if these are
required, though it has been a long time since I was at all familiar with
the data.  I think the examples I noticed occurred with forms cognate with
Greenlandic kimmik 'dog' and nanuq 'bear'.  I don't remember any nouns
ending in p or t off hand, but there may be some.

These latter cases, where final p, k or t look like they do something that
they don't and get handled as such willy-nilly are the parallels for any
Siouan cases where initial *o not a locative has been made into a locative
to preserve the pattern.