*pr vs. *wr (RE: pomme de terre)

[Rankin, Robert L]

On Sun, 3 Feb 2002, Rankin, Robert L wrote:
> I think the *pr cluster is really *wr.  Dakotan [b] is from */w/, not
> *p.

>It's true that [b] (sometimes w or m) is from *W, and that we have every
reason to suspect that in the first persons of *r verbs and others that
syncopate the first person was originally *wa, reduced to *w, but I
don't think it's actually possible to distinguish *w and *p in *wr and *pr
clusters, and, given the phonetics of the *<?>r clusters, I think *pr is
probably the more reasonable value.

I don't think there are any *pr clusters; they can all be derived from *wC.
The only one I can recall as even a possibility is in 'flat', which is a
widespread form with similar terms in Muskogean and so probably a borrowing.

Siouan /pC/ clusters outside of the 1st persons of verbs come overwhelmingly
from prefixes like *wa- or *wi- (the latter for animates, foodstuffs and a
few other terms). Considering the morphology, the *w emerges clearly as the
front-runner.

>The following table summarizes my impression of the *pr/*wr sets.  The
actual distinctions follow the environment, including morphological
class of root.  I've pondered this data from time to and not come up with
any way to divide it into *pr and *wr cases except possibly by looking
outside Mississippi Valley,...

Well, I think that's the first thing to do!  Go where the evidence is.  As
for MVS, I tried to show in my Siouan Conf. paper last year that [b] is
still an allophone of /w/ even synchronically.  It's been phonologized in a
few dialects, but mostly not.

>Within, any divvying up
>seems a matter of arbitrary fiat.  So, I prefer to say that we can't
distinguish them and to reconstruct *pr, since we also have *kr, though
only two sets that might be *tr (or *tw), with the proviso that *w + r
certain behaves as *pr and may explain most cases of it.

>In certain environments (varying with the dialect/branch) *pr behaves as
*pR (Dakotan) or simplifies to *R (Dhegiha, IO and Wi).  Where Dhegiha,
IO, and Wi retain the labial, the reflex of *r is proper for *r, not *R.
Nasality seems to keep roots in the *r category in all dialects, but the
labial is mostly lost in sync with the loss in the corresponding oral
cluster.  We're not sure about *R vs. *r vs. *n before nasal vowels, so
it's hard to say whether Dakotan mni 'water' for, say *priN 'water' is
*R
or *r or even *n behavior.

>Pre-PMV   PMV             Vwl  Sa   Te   OP    OS     IO       Wi
          *pr (verb root) oral md   bl   bdh   br     br       pVrV
                          nas  mn   mn   bdh   br     br       pVrV
          *pr (medially)  oral md   bl   n     t ~ c  d ~ j^   d
                          nas  mn   mn   bdh   br     n ~ n~*  n
**w-r     *p-r (inflctn)  oral m-d  b-l  b-dh  b-r    d ~ j^   d
                          nas  m-n  m-n  b-dh  b-r    n ~ n~   n
**wV-r    *pr (noun root) oral md   bl   n     t ~ c  d ~ j^   d
                          nas  mn   mn   n     n      n ~ n~   n

* => br in '8', otherwise n~ in '3' and 'bean'

>x ~ y is represents examples before back/low and front/high vowels, for
which, in some cases, the relevant examples do not occur.

>Santee md (of Riggs) is actually bd, as far as I know, in most places.
I'm not sure about the dialect of the Santee reservation in Nebraska.
Yankton has bd, too.  Stoney and Assiniboine have mn - at least that's
the
simplified version for non-students - but Stoney simplifies this to m in
verb inflection.

E>xamples:

Inflection - *r stems, of which there are examples with nasal root
vowels,
though not cognate forms that I recall of hand, though 'have' is fairly
reliable.

>Medial - with nasal following vowels, 'three' and 'eight' (where derived
from 'three') and 'bean', which is essentially regular, though it looks
to be a loan word.  With following oral vowel, 'woman's older brother',
which looks like a compound of 'house' and 'male'.

>Verb root initials - things like 'flat' (oral) and 'have an odor'
(nasal)

>Noun root initials - things like 'lake' and 'male' and 'potato' (oral)
and
'water' (nasal)

While I think it's important to chart out the sound correspondences as John
has done so thoroughly, in the final analysis it's necessary to figure out
what the morphemes are that are being represented in these frozen prefixes
of MVS.  Once you see that they go back to wa-s and wi-s, the phonological
source of the sets and their phonetic trajectories become clear.

The questionable tw or tr clusters seem to have kw origins outside of Siouan
probably, but this brings up a point.  There are numerous Dakotan clusters
that, so far as I know, no one has ever figured out.  These include sw, s^w,
xw, gm (some from kwVN) among others.

Bob