Siouan initials.

[Koontz John E]

On Thu, 22 Jul 1999, Robert L. Rankin wrote:
> Verbs would almost all have prefixes also, but the syllabification problem
> there is much harder to decipher because there were so many different
> possible prefixes.  Lexicalization of aspiration, once it was generated in
> any allomorph, seems to be quite general though.
>
> Instrumentals are an interesting set of exceptions that need to be
> accounted for systematically.  None has aspiration.  Siebert (1945) showed
> that the instrumentals are cognate with Catawba verb roots, so at some
> point we're dealing with compound or serial verbs.

Stop initial verbs (in simple *p/t/k) tend to arise from instrumentals if
they are in *p (*pa 'push' and *pi 'press') or *k (*ka 'strike' or
'wind/current').  There are also a few simple stems in *p/t/k.  Most of
the *t-stems are forms meaning 'see', though Dakotan has some stems of
other meanings that always occur with *k-.  Leaving aside the special case
of *ka, these forms are interesting in the context of conditioned
preaspiration in that they don't do it.  Instead they get shortened (or
non-lengthened) pronominals.  Among these the first person in *p then gets
*hC from *p 'first person' plus *C (the initial), but otherwise no
preaspiration.

So instead of

1  *w(a)-CV'...  =>  *wa-hCV'...
2  *y(a)-CV'...      *ya-hCV'...
3       *CV...          *CV...'   => *hCV... (by analogical leveling)

you get

1  *w(a)-CV'... =>   *p-CV'...
2  *y(a)-CV'...     *s^-CV'...
3       *CV...'        *CV...'

Of the two developments, the second (above) is normal if the syllable
after the pronominal is a prefix (or first coverb), but sometimes occurs
with simple verb roots, too.

The first pattern of development is normal if the syllable after
the verb is a root (or second coverb), though sometimes simple roots
prefer the second treatment instead.

Simple roots following a prefix (second coverbs) tend strongly to prefer
the first pattern:

Prefixation:    *PRO-CV'-CV... =>  PRO(short)-CV'-(h)CV...
                     CV-CV'... =>             CV-hCV'...

Here we'd presumably want to argue that the first line was originally
?PRO(short)-CV'-CV... with the latter C being preaspirated later by
analogy with the treatment when no PRO is involved.  (Hence I parenthesize
the h.)

All of this amounts to saying that there are two different environments,
perhaps accentually conditioned at one time, and now attested only in the
two different patterns of development.  In that case, of course, I
probably shouldn't show the accentual pattern as the same in both cases,
but it isn't clear what the two alternatives would have been.  It is in
fact striking that both patterns of development are usually explained as
conditioned by second syllable accent.

Perhaps we need to fall back on something about the boundary or lack of
one between the two components - e.g., the mutilated remnants of the
second coverb's pronominals sandwiched in between two coverbs, but not
between a pronominal and a prefixed root.  However, this line of reasoning
suggests there is something additional sandwiched between the regular
pronominals and those simple roots that become preaspirated.  Of course,
there is at least that -a-.  Maybe these are forms that have doubled
pronominals, the inner one of the short kind.  This would mean that all
regular verbs with preaspirated initials are really
PRO(long)-PRO(short)-root stems.  Presumably PRO(long)-root(initial
cluster) where cluster isn't a preaspirate is another context conditioning
PRO(long).  This would suggest PRO(long) arise epenthetically.

JEK