U.S. President and cloud/sky: more.

[Koontz John E]

On Wed, 14 Feb 2001, RLR wrote:
> And re 'sky, cloud(s)':
>
> Quapaw has maNghe 'sky'; moxpi 'cloud(s)'
> Kansa has maNghe 'sky, "the upper world"' and maxpu" or moxpu"
> 'cloud(s)'.
>
> All of this leads me to speculate that the mysterious Chiwere form for
> 'sky', maNghu, is from earlier *maNxwu, which would be the outcome in
> Chiwere of the same form that gives Kansa/Osage maNxpu". My recollection
> is that the p > w in this context in CH. This would give Quapaw and
> Omaha-Ponca maNxpi regularly, leaving only the Dakotan /i/ unexplained.

Here's my take on this.

Note that Dorsey, in 1885 article 'On the Comparative Phonolpogy of Four
Siouan Languages', gives as set 92 'sky':

Da ma-qpi-ya; Po (OP) ma-qpi; Ks ma-qpu"; Os ma-qpu"; Ch. ma-xu (I),
ma-xue (Ot); Wi (Ho) maxi-da.

His q is modern x; his x is modern gh (gamma).  He tends in this to give
Winnebago -ra as -da.  The -ra (or =ra) is identified as the definite
article and has functions in marking relative clauses.  I'm not sure if it
is properly a "definite" article, but it is probably not wrong to call it
an article in some sense, and it is not an indefinite article.

Summarizing the forms supplied in which maNx- occurs:

Language   Te        OP       Ks       Os        IO         Wi
'sky'      -         maN'ghe  maNghe   maNghe    -          -
'cloud'    maxpi'ya  maN'xpi  maNxpu"  maNxp[i?] maN'ghu(e) maNaNxi'(wi)

In Teton, compare maNxpi'xpiya 'scatering clouds' (reduplicated); also
maNxpi'ohaNzi 'a shadow caused by a cloud', maNxpiyohaNzizi 'the passage
of clouds, at intervlas obstructing the sunlight'.
Winnebago maN(aN)xi=ra is just maNaNxi' with =ra.
Winnebago maNaNxioc^o is maNaNxi' + oco 'be blue in/there'.
IO "mo"xpi" attributed in Good Tracks to Maximillien, sure looks like
Omaha-Ponca to me.
Os maNxpi per LaFlesche could be maNxpu", given LaFlesche's difficulties
with u".
IO maNghue is clearly a spelling variant of maNghuwe.
IO maNghu(we) in the maNxuwe variant reduplicates as maNghughuwe.

In languages that lack a form in the first row, the form in the second row
takes the interpretation 'sky; cloud(s)'.

'Cloud(s)' is also rendered 'cloudy' in translations.

Reconstructions.

Form 1 is *maN'ghe.  For the moment I'll avoid the waN vs maN issue.
Form 2 is *maN'xpi.

The developments of form 1 don't seem to pose any problems.  Note that
those languages that lack it have homophones, mainly *maNgh- 'field'.

The second form acquires a formant -ya in Dakotan.  This disappears in
some compounds, though not maNxpiyatho 'blue sky; Arapahoe'.  And it's
difficult to argue this convincingly as the only compound I have is
V-initial, making a V1 + V2 => V2 contraction argument possible.

However, compare the flat absence of any such extension elsewhere, except
Winnebago, to which I'll return, and the existence of other root + ya
nouns that lose -ya in compounds, e.g., wiN(yaN) 'woman', khe(ya)
'turtle', he(ya) 'louse', iN(yaN) 'stone' (cf. B&D, p. 71).  Also the
pattern of insertion of -ya after many CV nouns before certain
postpositions, e.g., with thi- 'dwelling', mniN- 'water', ble- 'lake',
c^haN- 'woods', xe- 'away from camp; mountains?' (cf. B&D, p. 144).  A
similar pattern occurs with appended a or ablaut in Omaha-Ponca with
certain postpositions, e.g., ttiatta 'in the house'.  The Dakotan
postpositions are -ta, -taya, -taNhaN.  The OP ones are -di, -tta, -ttadi,
ttahaN, sometimes others.

In OP, the forms remains intact.

In Ks and posibly Os for reasons as yet unclear final i becomes u".  This
might be the influence of the preceding p, but there are other cases of
unexpected u" for *i and unexpected i for *u, and OP and Qu both merge *u
and *i, while Ks and Os shift *u to u".

In IO, *maNxpi > *maNxwi regularly, and this is reinterpreted as
maNxu(w)e, presumbly via maNxui.  Compare IO xume, xumi 'stinky' with OP
xwiN 'phew!'.  For the developments of *xp, cf. Te xpe'c^a(ka) 'faint,
exhausted', Os xpeka 'languid, drooping', IO xwe'ge 'week, feeble,
drousy'.  The final change is the loss of -(w)e in some usages.

Wi maNaNxi'wi is the regular development of maN'xpi.  But -wi is one of
the plural enclitics, cf. Dakotan -pi, OP -bi ~ -i, etc., so it is often
deleted through a reanalysis process, yielding maNaNxi'.  And the enclitic
article =ra can be added to this reduced form.  Whether or not this =ra is
to be connected with the Dakotan -ya extensions and insertions, the
Omaha-Ponca -(y)a insertions, and perhaps the readines of IO to delete an
apparently extraneous -(w)e at the end of a word is, again, a separate
issue.

Yet another issue is whether *maN'xpi 'cloud(s), cloudy' is the plural of
*maN'gh(e) 'sky'.  Clearly the concepts of 'sky' and 'cloud(s)' connect
well today in many cases and I don't see any real semantic difficulty in
treating 'cloud(s)' as 'pieces of sky, skies'.  However, whatever the
nature and origin of the final -e of *maNghe, it is true that the
Mississippi Valley plural marker pi (and, as far as I know, all Siouan
plural markers) is notorious for preferring to mutate -e to -a.  Also, Bob
Rankin has argued powerfully that we should probably see -pi as -api even
in synchronic cases, so perhaps the plural of *maN'ghe should be expected
to be *maNghapi.  (How do you * a *-ed form to indicate that it doesn't
occur?)

Note that the CSD draft reports cognates for form 1 'sky' in Crow-Hidatsa
and Southeastern.  The CH forms seem a bit difficult.  Form 2 'cloud(s)'
(albeit the Dakotan form is included under 'sky') seems to be restricted
to Mississippi Valley.  The 'rounding is secondary in Dhegiha' approach is
adopted in 'cloud(s)'.

John Koontz