Order of verb affixes

[Koontz John E]

Just one more follow up here.  The relative order of pronominals and
locatives in MV Siouan languages differs extensively.  The Dakota pattern
is only generically similar to what occurs in Dhegiha or Chiwere or
Winnebago.  For example, aN(g)-, the "agent" inclusive pronominal in OP,
precedes the a-locative and u-locative, but follows the i-locative.  The
surface forms are aNg-a-, aNg-u- and aN-dh-aN- < *i-r-aN-.  On the other
hand wa-, the "patient" inclusive pronominal, precedes the a-locative and
i-locative, and follows the u-locative:  wa'-, we'-, uwa'-.

This oddity is mitigated, but in no way simplfied, when one realizes that
the wa following u is actually secondary.  Historically *wa-o- > *wo- and
that form is reduced to u'- in OP and it is often difficult to distinguish
this u'- from u- not incorporating a hidden wa- once the inflecting starts
and the dust rises.  But some verbs - I'm not quite able to characterize
which except by listing them - insert a "new" or "pleonastic" wa after u.
I think this usually occurs with human objects.

In addition, I've noticed that in OP things like reflexives sometimes get
inserted not in some template-determined location, but simply at the front
of whatever is "already" (lexically) present.  I'd have to track down the
examples, though I once used them in a SACC paper.

I can't remember if we used these examples in the 'Word' paper, but they
certainly entered into our thinking for it.

In a lot of Siouan morphological situations it looks to me like "point of
insertion" is less determined by a mental slot map than by a sort of
enzyme-like consideration.  If an X is to be added, we look at the shape
of plug the X has and then stick it in the first (or all) sockets where
that plug would fit, and failing such a socket we pile it in front of
the form.  I'm not sure this metaphor helps all that much, since the
nature of what constitutes a socket is not made particularly clear.