m's and w's and Mitasse: Caddoan phonology question

[Rankin, Robert L]

I can add just a bit about possible areal features here.  In the 1880's
Dorsey transcribed one Kansa (Kaw) sound with the letter <m> with a
small <x> written directly beneath it.  This occurs as the initial
phoneme of all of his 1st Sg. possessive forms with the prefix wi- 'my
inalienable'.  Quite evidently it had some sort of weak nasality in
Dorsey's perception.  By the 1970's when I re-recorded all of Dorsey's
material with Mrs. Rowe and Mr. Pepper these peculiar M's with the
subscript x were fully fledged W's.  I have no recording of a bilabial
fricative or partial nasal at all in these words.  SO . . . It seems
clear that Kaw initial /w/ preceding /i/ was somewhat nasalized in the
late 19th century.  But by the last quarter of the 20th century the
nasality had totally vanished in the same vocabulary.  

We know that Caddoan and Dhegiha speakers were in contact (the Kaws and
Osages borrow 'eight' from North Caddoan), but that's about all we (or,
at least, I) know.  I do think this makes David's hypothesis more
reasonable however.

Bob

> What if, in Waco or even older Wichita, [w] and [m] had a distribution
parallel to modern Wichita [r] and [n]?  Do any other phonologists out
there think this is at all plausible?  If it is, then the name "Mittase"
might have an initial phonemic /w/.  Unfortunately, I can't go any
further than that, because /witasi/, or /wirasi/ (many English speakers
write the tapped [r] as _tt_), or other variations I can dream up still
don't match with any morphemes I know that might lead to the meaning
'white baby' or 'white child'.

What does anyone else think of the [m] = [w] speculation?