*CVC Stems as *CVVC Stems (RE: Kaw and Osage 'stand')

[Koontz John E]

On Wed, 3 Nov 2004, Corey Telfer wrote:
> At the conference this summer I thought somebody (or even a few people) said
> that the irregular /CVC/ roots in Lakota developed from long vowels in Proto-
> Siouan.  Is this true, or is it just my wishful thinking?  If it is true, is
> there a published reference I can cite?

I don't recall the particular instance in question - maybe something Bob
said? - but the Dakotan CVC roots correspond generally to CV'Ce roots in
Dhegiha, and these are also the roots that tend to exhibit [CV'?(V)Ce]
pronunciations when said slowly and carefully, as opposed to more rapid
[CV'Ce].  (See John Boyle's bibliography for Bob Rankin's paper on this
phenomenon, which may not mention vowel length.)  I think Kathy Shea may
have suggested to me that she thinks that these are artifacts of /CV:'Ce/
form and presumably this forms part of her dissertaion in progress.

This ties in which the notion that PMV has second mora accent, i.e., these
stems may be PS or PMV *CVV'C(e) stems.

The *(e) manifests in various ways in various MV languages:  as e' ~ a ~ 0
in Dakotan, as e in Dhegiha and Ioway-Otoe, and as 0 in Winnebago.  The
Dhegiha and IO stems sometimes show fossilized CVC or CV allomorphs, e.g.,
Da ha'za 'berry' : OP hazi 'grape, grape plant' < *has=hu 'berry stem' :
IO has(j^e) 'strawberry' : Wi haa's 'berry; fruit', or Da s^uN'ka 'dog,
horse' ~ (tha)s^uN'ke 'his particular horse'~ s^uNg(wiNyela) 'mare' : OP
s^aN'ge 'horse' ~ s^aN'(ttaNga) 'wolf' : IO suN'<ng>e ~ suN'<ny>e 'dog,
horse' :  Wi s^uNuN'k 'dog, horse'.  And sometimes Winnebago shows a final
vowel in certain contexts only, e.g., OP maNs^tiN'ge 'rabbit' : Wi
was^j^iNk 'rabbit' ~ was^j^iNge'(ga) 'the Rabbit'.  Or IO or Winnebago
might show an alternative to (e), e.g., Da c^haNte' 'heart' ~
c^haNl(wa's^te) 'be pleasant' : OP naN(aN)'de 'heart' : IO naN(aN)'hc^e <
*naNaNkte < *raNaN't(ka) : Wi naNaN'c^ < *raNaN't(e) ~ naNaNc^ge' <
*raNaN't(ka).  In my kinship term paper I pointed out that sometimes you
get *CV-<glide>-e/a forms, too, where epenthetic glide can be *y (*r
between vowels) or *w.

There are various ways to analyze the (e) ~ (ka) elements, either
morphologically or both phonologically and morphologically.  I don't see
how a purely phonological analysis is present with *ka in the picture.
My own preference is to see the final elements *(<glide>)e ~ *(<glide>)a ~
*ka as morphemes with nominalizing force, and there are various ways to
see that in historical terms.  I have a paper (unpublished) on that, which
John would list, and I have commented extensively - well, ad nauseam,
really - on the possibilities on this list.  I don't know that I have
mentioned the length component of the notion before.  That stems (no pun
intended) from Kathy as far as I know.  She and I discuss length and
accent from time to time, but I'm pretty sure this was her suggestion or,
rather, observation.

I guess you would say that Da CVC stems develop from long vowel stems if
the analysis is that PMV *CVVCe => PreDa *CVC (a phonological
development), and that the various ablaut alternants or *a-extennsions or
*ka extensions develop secondarily in Dakotan and the various other
languages.  Otherwise you would just say that CVC stems are reflexes of
PMV *CVVC stems, and that other patterns of PMV stems, including at least
*CV(V) and *CVCV exist.

Either way you could argue that Da forms like CVCa reflect *CVV'Ca, while
the variants in CVCe' and compounds accented CVC(CV'...) reflect *CVC with
shortening of the stem internal vowel from *CVVC /__ "certain morphemes."
Shortening leads to the following vowel (the new second mora) taking the
accent.