epenthetic glide.

[Rory M Larson]

John wrote:
> Actually, it seems likely that the -y- stage was in Proto-Mississippi
> Valley, since all of the MV languages have reflexes of *r for epenthetic
> situations.

Hmm.  I was afraid of that.  Thanks to John
for his thorough discussion of this fly in
my ointment!  However,

Bob wrote:
> As John points out, lots of western hemisphere
> languages simply use a tap or flap to separate vowels
> in hiatus.  There's no need to go through the phonetic
> [y] stage.  It is possible that Common Siouan /r/ is or
> was phonetically 'dh'.  Its reflexes turn up as a host
> of (mostly) dental liquid sounds, dh, r, n, l, t, and
> even y in sundry languages and dialects.  The important
> thing isn't so much its precise phonetic reality but
> what it contrasts with.

So perhaps I don't really need to worry about
the sound being precisely a phonetic [y].  Just
being as close in the neighborhood as [r] or [dh]
may be good enough.  All I need is for the (*r)
in *=ire to go away if the final -e is clipped.
So if (*r) represents whatever sound in the [r],
[y], [dh] range, I would propose this development
of *=ire for Winnebago and OP:

         MVS: =i(*r)e
                 |
      ---------------------
      |                   |
      |            Dh: =i(*r)e
      |                   |
      |                =i(*r)    (-e is clipped)
      |                   |
Wi: =ire             OP: =i      ((*r) goes away)


Now if we take the verb /xe/, "bury", in plain form
and the two alternate 3rd person plural forms, we
have:

  MVS:     *xe        *xa=pi         *xa=ire

   Wi:      xe         xa=wi          xa=ire

   OP:      xe         xa=bi          xa=i

I've taken the Winnebago /xe/ example (via memory)
from Lipkind.  Note that =ire conditions a-grade
ablaut in this case.


John wrote:
> This is certainly an interesting hypothesis, though ideally we'd have to
> understand the introduction of i into -ire (cf. IO -ne - pronounced with
> enye before i).

Do you see a problem with reconstructing MVS *=ire?
I think you were coming up with *=krE for PS.
Would *=kirE work?  Or do you mean that we can't
get from MVS /*=ire/ to IO /=ne/?


> Apart from that we'd have to assume that this source of i was generalized
> from the third person to the inclusive and second persons from the third,
> which is where this marker occurs in the Siouan languages that have it.

Yes, and also from 3rd plural to 3rd singular in
the declarative form, if the other ones are all
like Winnebago.  For this matter, I think we might
want to consider a hybrid explanation.  I've been
insisting that =bi and =i are separate words in
OP, while John and Bob have been equally insistent
that =i is a phonological reduction of =bi, because
they are so very similar phonologically, syntactically
and semantically.  Well, why couldn't both views be
correct?

Suppose the sequence I propose above is correct,
with *=i < *=i(*r)e coming down to proto-Dhegiha
or proto-OP as an alternate 3rd person pluralizer.
*=pi|*=bi is the general pluralizer, and is also
used for the 3rd person singular declarative.
There is presumably some subtle difference in
meaning between the two particles, else the *=i(*r)e
form would not have been retained.  But when *=i(*r)e
is reduced to *=i, the two become very confusable.
When the stop part of *=pi is elided in rapid speech,
it comes out as [i], which is interpreted as *=i(re).
Following slurred *=pi, a reconstructed *=i appears
in all the locations originally reserved to *=pi.
The process is similar to English "should have"
first being slurred to "should've" and then
reconstructed as "should of".

Now the difference between *=pi and *=i depends on
whatever sematic distinction they carry.  If this
difference is weak, then *=i is regarded as simply
a slurred variant of the more proper *=pi.  In this
case, the old *=i(*r)e particle dies out, even in
its original 3rd plural position, perhaps leaving
behind a few traces of its original presence in
fossil "diphthongs" such as /tta=i=che/.  This is
what will have happened in Osage and Kaw.

If the semantic distinction is strengthened,
however, then the original 3rd plural *=i(*r)e
element bursts its cage and becomes a completely
parallel, but separate, form that needs to be
rigorously distinguished from the *=pi form.
This is what will have happened in OP.

Rory