'nine' again

[David Costa]

>> Hope this helps.  I'm not sure why 'nine' would be borrowed.  As Haas and
>> others have noticed, the word for 'seven' was borrowed from Creek into
>> Cherokee, but seven is the special number for the Cherokee, so that sort of
>> makes sense (four is the special number in the Muskogean languages).

> Oddly enough a Siouan term for 'seven' was borrowed into Miami-Illinois.

Not to be nitpicky, but it was 'eight', actually. :-)

As a matter of fact, 'nine' is borrowed quite often. The reason this whole
discussion began is because Rich Rhodes and I are cowriting a big paper on
the history of number words in Algonquian. A couple of larger points have
become quite clear from this paper which we didn't quite predict going into
it: (1) the notion that lower numbers (1-10) are resistant to borrowing has
to be thrown right out the window when doing comparative work on North
American languages. Numbers and various innovations among numbers
(especially 5-10) are *constantly* passed around among Algonquian languages.
And point (2), 'nine' is by far the least stable number among the Algonquian
languages. It's the most prone to borrowing and the most prone to innovation
with new neologisms getting passed around for it. Several widely separated
Algonquian languages have independently innovated phrases translating as
'missing one' or 'one left' for the number, though with wildly varying
material for the 'missing' or 'left' part (sometimes even for the 'one'
part).

That said, it *does* look as though a word can be reconstructed in
Proto-Algonquian for 'nine', PA */$a:nka/ ('$' = s-hacek; no known
etymology). An unsettlingly large number of the languages lack it (all the
Eastern languages, for starters), but the geographic spread on it is such
that it's probably not really feasible to attribute it to borrowing: it's
shared by Ojibwe-Potawatomi, Fox, Cree, Cheyenne, and, perhaps most
compellingly, one Arapahoan dialect. Granted, Shawnee screwed up the initial
consonant of the word, but other words have been reconstructed for
Proto-Algonquian on much flimsier evidence. (If it weren't for the Arapahoan
form, I'd lean towards calling it a very early loan from god-knows-where.)

So, then, this gets us back to the main question: if this word is
reconstructible in Proto-Algonquian, what's it doing in Siouan? And,
moreover, are the resemblant forms in Western Muskogean and Tunica related
in any way? For the latter, the question is whether Choctaw /cakka:li/~
Chickasaw /cakka?li/ is connected to Shawnee /ca:kat0wi/ '9' or /ca:ka/ '90'
in any way, or, secondarily, to Proto-Algonquian */$a:nka/. My current hunch
is that the Proto-Algonquian form and the Western Muskogean form(s) are
unrelated and just coincidentally similar, BUT that the Shawnee form changed
its initial consonant of 'under the influence' of the Western Muskogean
form. So it's sort of a quasi-loan.

One thing that might sway me away from thinking that the Proto-Algonquian
form and the Muskogean forms are just coincidentally similar is if a
scenario could be shown for how Western Muskogean could borrow a word with
s-hacek and eventually turn it into c-hacek. If this is an old loan within
Muskogean, and if there is an old sound law within the correct Muskogean
languages taking 'sh' to 'ch', then perhaps Muskogean did borrow it (note I
don't say from where). But the geography necessary for such a scenario isn't
very easy to visualize. At least not for me.

As far as the Siouan forms are concerned, the message Bob has given me
several times is that the word just cannot be reconstructed for Proto-
Siouan. So Chiwere, Dhegiha, and Tutelo probably got it from Algonquian (tho
Biloxi got it from Western Muskogean). As Bob has pointed out, this would
explain why the sibilants don't match among the Siouan forms. Tho I leave it
to the Siouanists to explain why Siouan seems to have borrowed it with an
initial */kV-/ syllable tacked on, which no Algonquian language has.

Thanks for listening. Now back to your regularly scheduled programming. :-)

Dave Costa