a phonetic mystery

[Koontz John E]

On Tue, 20 Mar 2001, ROOD DAVID S wrote:
> 	I would like confirmation of the following from others who listen
> to Lakhota, but I hear a contrast between mni 'water', with an oral vowel,
> and mniN kte 'I will go' with a nasal "i".  If that's the case, isn't
> there a problem with deriving the m-n of 'water' from *w-r+nasal?  I still
> think we have been far too sloppy in our listening to Lak. vowels after
> nasal consonants, but it would take a dedicated study to straighten it
> out, preferably one that uses instrumental phonetics and a variety of
> speakers from different places and age groups.  Until we do this, however,
> I think all our *w & *r reconstructions are suspect.

I agree that the resconstructions of the sonorants oral and/or nasal could
use some work.  I was spouting current wisdom without including any
caveats.  I am puzzled by a number of cases where nasalization doesn't
occur or does occur.

I wonder, however, here if the issue isn't where the nasality comes from.
The impression I've always gotten from you is that there isn't much
perceptible nasality in vowels following nasal sonorants in Lakota.  I
tend to interpret this as a sort of lexical shift of nasality into
sonorants or sonorant clusters where this was possible.  (So, OK, where
does that leave wiNyaN et al., neglecting historical explanations!)
Otherwise, nasality resides in the vowel.  But in the future the nasality
is introduced with the enclitic and is lexically still tied to the vowel,
even though it does affect the sonorants of the stem.  I guess this
presumes Bob's view that ablaut involves an enclitic's initial vowel
replacing a base's final vowel, and not a change in the base's final
vowel.  (The issue being what entity owns the vowel.)

If you want another interesting case, consider the fact that in Dhegiha
nasality of the root vowel seems to nasalize the sonorants of nouns, but
not of verbs, e.g., ni(N) 'water', but bdhaN 'have an odor'.  And the
nouns lose the initial labial element, too, for that matter, while verbs
don't.

In fact, looking across Mississippi Valley, either there are effectively
four different environments for "*pr" - noun initial, inflected verb
initial, verb stem initial, medial - or one is driven to dividing the sets
up into *wr, *pr, etc.  No matter which way you slice it, however, you
have to assume a lot of analogical influence.

Sets illustrating the four *pr (or *wr or *war):

gloss     La    OP       Wi

lake      ble   ne       dee
water     mni   ni       nii

A1 + r    bl... bdh...   d...

flat      blas- bdhas-   paras
smell     -mna  (-)bdhaN paNnaN

three     yamni dhabdhiN daaniN

Nasal inflected verbs are somewhat complex cases - we've just seen one,
anyway, so I'll omit them.  'Three' is the only medial instance I know of.

Note that *pr pretty much behaves as *R when it simplifies, which is one
of the arguments, originating with Kaufman I think, for considering *R as
a sort of cluster.

Also, in IO, *pr behaves differently in ran(~)i 'three' and grerabriN
'eight', though I tend suspect this might involve, say, a loan from
Dhegiha.