Incorporation

[Koontz John E]

> > In a purely infixing pattern OP insets not just 1st and 2nd, but also
> > the inclusive, cf. mu=a'se, mu'=dhase, mu=aN'se 'I/you/we cut with a
> > shot'.
> 
> Some may not recognize that mu- here is an "outer" instrumental.  So I
> personally don't regard this as infixing in any sense.  For me a true
> infix must come between two non-grammatical parts of the stem and
> locatives and instrumentals wouldn't count.  They're just prefixes in the
> usual prefix order.  The ma- of 'walk' and 'steal' isn't.

I apologize for the terminology.  It's true that I know of no "infixing" 
stem that inserts the affix into a morpheme in OP (or other Siouan).  It's
always within a stem, but between morphemes.  I've been referring to stems
of the form [all pronominals]-stem as prefixing and stems of the form
stem_part_1-[all pronominals]-stem_part_2 as infixing, in talking to
myself, but this is just a expedient, and I'd be willing to consider
something purer if it's equally convenient.  Come to think of it, I
sometimes call these simple and preverb stems, but preverb could be
understood to include locatives (see below). 

There's a third class of stems, of course, that has the form [inclusive
pronominals]-stem_part_1-[other pronominals]-stem_part_2.  At least this
pattern holds pretty well in Dakotan.  The stem_part_1 is generally a
locative, and I call these locative stems, even when stem_part_1 is really
the maN of maN=...dhiN 'walk' or maN=...dhaN 'steal'.  This terminology is
clearly also somewhat bastardized.  

Of course, the relative order of inclusive > {1, 2} is retained in all
three classes and in terms of slots there's always an inclusive slot
preceding the slot for first and second persons.  This conception has to
be modified to account for various exceptional behaviors, like extraction
of the a and aN first persons, etc.

There's also a compound pattern of the form [inclusive pronominal]-[first
and second pronominals]-stem_part_1-[first and second person
pronominals]-stem_part_2.  That is, there are two independently inflected
parts of the stem, but only the first of these takes an inclusive
pronominals.  The standard OP example is gaN=dha 'to want', which is
kkaN=bdha, s^kaN=s^na, aNgaN=dha for 'I/you/we want'. 

To take this a little bit further now, in my work with OP I've come to the
conclusion that prefixing and infixing stems are really more or less the
same thing, except that the latter have a chunk of invariant preverbal
stuff that sits in front of a component "main" stem that obeys all the
prefixing rules.  In short the preverb is certainly an incorporand in at
least an abstract sense, The compounds stems are also just basically two
component prefixing stems in a row, except for only taking the inclusive
with the first.  

If you don't allow for this hierarchical structuring of components in
handling the infixing (or preverb) and compound stems you can quickly tie
yourself in knots.  In this respect I think that Siouan languages differ
from languages like Quechua or Turkish.  It's impossible to define simple
linear position class descriptions of the verbs.  Maybe I'm just ingorant
of Quechua and Turkish, of course!

This is particularly evident when one considers what happens when infixing
stems are subjected to further derivation.  In OP, anyway, for the most
part the preverb and main stem collapse into a single prefixed stem.  So
when kki-, the reflexive, is added to a verb in mu=..., the kki goes
before the mu= (usually), and the pronominals "now" forget about the slot
after mu=... and congregate in front of the kki-.  So you'd get akkimase,
dhakkimase, aNkkimase for 'I/you/we cut myself', not ma=akkise, etc., or
some other intermediate possibility.  Similar things happen when locative
are added to stems with outer instrumentals or other preverbs.  There is
some variation in applying the rules. 

I think that examples like these show that stem structures in OP are
hierarchical, and only the outermost layer or two of derivation are
relevant to pronominalization, which is controlled by the morphosyntactic
stem classes defined by the morphemes in these outermost layers, not by
rules in terms of mingled series of derivational and pronominal morphemes.
Thus, a standard Siouan grammar's rule like outer_instrumental > inclusive
> locative > first/second > dative > inner_instrumental > root isn't
really satisfactory.  Of course, such a rule works fairly well for the
simple component locative and prefixing stems, so chop off the outer
instrumental and rewrite the dative + inner instrumental + root sequence
as stem and you are back in business. 

I sincerely apologize to Bob for building this on the accidentally
tendered straw man of his comment about outer instrumentals preceding
pronominals.  He wasn't, of course, setting out to argue for a strict
system of positional classes at all.  I think he only meant to observe
(correctly) that the order of instrumentals and pronominals is
morphologically determined and intermorphemic, and not based on some sort
of phonological template, contrary to what my entirely sloppy use of the
term infixing might suggest.