Irregular "to eat" in Dakotan

[Koontz John E]

On Wed, 15 Aug 2001, Constantine Xmelnitski wrote:

I surveyed the r-stem pattern and also 'to eat' which has additional
irregularities, in:

Koontz, John E. 1983.  Siouan Syncopating *r-Stems.  pp. 11-23,
Na'p<a-macron>o [Plains Cree for 'Man'] Vol. 13, October 1983, Special
Issue:  Proceedings of the Second Siouan Languages Conference, 1982.  Ed.
by Mary C. Marino.

(MM says na'p<a-macron>o is actualy Plains Cree for 'male'.)

> I wonder what is the cause of irregular paradigm of Dakotan transitive
> verb yu'tA. L/D: 1s: wa'te, 2s: ya'te, 12: uNyu'tapi > uN'tapi,
> yul-/yun-/yud- According to Shannon West, Assiniboine's "yuda" behaves
> [at least in some dialects?] as if a regular Y-verb: mnu'da - nu'da -
> uNyu'dabi.

Winnebago has alternative inflectional patterns for this stem, either
haa'c^/raa'c^/ruu'c^ (cf. Dakotan wa'te/ya'te/yu'tA) or
duu'c^/s^uruc^/ruu'c^ (cf. Assiniboine mnu'da/nu'da/yu'da).  IO has
haj^i'/raj^i'/ru'j^e, which is more or less along the first pattern,
though the accentuation and final vowel are not quite an exact fit.

In the paper I treated this stem as being monomorphemic, i.e., with the
initial *ru (> yu in Dakotan) being only coincidentally identical with the
hand instrumental.  Clearly, though, that coincidence would be an
operative factor in the evolution of the verb.  The Dhegiha cognate of
this stem would be *dhute (cf. Osage dhuce, spelled thidse by LaFlesche),
which has a gloss 'to scoop from a hollow place' in which the hand
instrumental seems interpretation to be involved. Omaha-Ponca seems to
lack the *dhute form - it would be expected to occur as dhide.

There is a potential cognate of Proto-Siouan *rut(e) 'to eat' in Catawban
exhibited in the Woccon utterance rendered Noccoo Eraute? 'Have you got
anything to eat' in Lawson's word list (cited by Carter in a 1980 article
on the Woccon data).

Frankly, I'm not sure why the PS verb *rut(e) has first and second persons
based on a stem allomorph *t(e) in Dakotan and Winnebago and Ioway-Otoe.
I'm inclined to think that this pattern is old, and that occurences of the
more normal pattern in Dhegiha and elsewhere are cases of regularizations.
Of course, the pattern "regularized" to is that of the *ru- instrumental,
which follows the *r-stem syncopating conjugation as opposed to thre
"regular" pattern used with cluster and fricative initials.

If the Dakotan pattern for 'eat' is old, then one possibility is that the
stem was originally *ot(e) or *ut(e).  In that case the first and second
persons could be explained as involving a vowel contraction (notice that
they are normally accented in violation of the second syllable tendency),
while the third person has an epenthetic *r.  This *r would be conditioned
when suitable vowel-final morpheme preceded the stem.  One that would make
the most sense for me formally would be a hypothetical third person
pronominal *i, but there is no consistent evidence for this in the active
paradigms, though third person *i is reflected in possesive paradigms.

It may also be worth noticing that Osage has an opposition of dhache 'to
eat' vs. idhache 'to eat one thing with another'.  Allan Taylor tells me a
semantic distinction of 'eat one thing' vs. 'eat two things together'
occurs in some Algonquian languages (I think Atsina is one).  If *i were
regularly used to derive such stems in Siouan, then a frequent opposition
of *ut(e) vs. i-rut(e) might lead to a revision of the root from *ute to
*rute, affecting only the third person, though subsequently the first and
second persons might be modified to fit the normal pattern for *r-stems.

> Besudes Dakotan stuff I have only Osage and Winnebago dictionaries, so
> I could found only these possible cognates: Osage: dha'ce (bdha'ce -
> sta'ce - oNdha'ca i), "to eat" (vt?) noN'bdhe (awanoN'bdhe -
> wadhanoN'bdhe - oNwoN'noNbdha i) to eat, to consume (vt?) wanoN'bdhe,
> to eat, to dine (vi?)

Rankin explained these sets to me back in '82 as:  Dhegiha *dhathE 'to
eat' cf. Dakotan yathA 'to chew'; Dhegiha *naNbdhE 'to eat' cf.  Dakotan
naNpc^ha 'to swallow'.  He also pointed out the *dhute 'scoop up' set to
me and argued that it did not historically involve an instrumental.  The
PS stems seem to be *ra-the', *raN'pye (or *naN'pye?), *(r)u't(e).