epenthetic glide.

[Koontz John E]

On Sat, 21 Jun 2003, Rankin, Robert L wrote:
> I think it is Proto-Siouan *y that becomes ch in Dakotan, zh in Dhegiha,
> etc.  Proto-Siouan *r becomes Dakotan y~l~n.

True, although, it might be more precise to say that *r becomes Dakotan y
and Dhegiha dh (with special developments in some dialects, i.e., d in
Quapaw and y in Kaw).  It's *R that becomes Dakotan l ~ d ~ n and Dhegiha
n (OP), t ~ c (merger with *t) (Os), and so on.  But *r develops as *R in
certain clusters, i.e., s^r in Dakotan and Dhegiha and *pr (or *wr or *br)
in Dakotan, but not Dhegiha (except in nouns).  Dakotan, of course, loses
s^ in *s^R < *s^r in the second persons of *r-stems, and OP does between
the 1880s and today.  So, you find in 'to go':

PMV     La      OP     Os
*p-re   ble     bdhe   bre (bdhe, but with dh like an r in context)
*s^-re  le     (s^)ne  sce
*re     ye      dhe    dhe

And you can compare these with

*pre    ble     ne     ce    'lake'
*Rez^e  lez^e   nez^e  cez^e 'urine'

The observation that *R behaves like *r in some clusters is presumably
what led Kaufman to reconstruct *?r for the *R correspondence, where ? (an
apostrophe - glottal stop) is something like the old Bloomfield theta in
Algonquian, as is R, really.  The *R correspondence was first noticed by
Dorsey, but I think thereafter neglected until Kaufman.  I first noticed
it in Dorsey's Comparative Phonology, in spite of having earlier seen
Kaufman's *?r, and originally thought it might be a development of an
allophone of *t (matched by *W for *p).

> Epenthesis at different times has yielded different results.  Older
> epentheses seem to yield reflexes of *r while new ones yield reflexes
> [y] and [w] both.  Also, we need to remember that in Dakotan, the *r
> that is inserted after possessive i- shows y-like reflexes.  We
> explain a number of strange Dakotan ch's like 'ice' in this manner as
> I recall.  I'm in Bloomington and don't have my notes with me though.
> But I think 'kettle' is another.  These are the result of *ir
> sequences as I recall though.

These are not the sets I was thinking of.  Here we have

         La        OP
'ice'    c^hagha   nughe
'pot'    ch^egha   neghe

Here Dakotan is the odd man out, with c^h < *y in sets where others like
OP show *R (or *wr, cf. Cr bilaxa and buluxa).  The 'ice' set in
particular tends to accumulate nasal vowel variants instead of *o.  I've
sometimes wondered if Teton legha 'glittering; might not be a better match
for the 'pot' set.  The Dakotan forms simply look like they don't belong,
but you have to wonder.

The sets I was thinking of were like:

'heart'  c^haNte   naNde

I was arguing that these derived from original forms like *i-yaNt-
'(one's) heart', which in PMV dialects underlying Dhegiha was interpreted
as **i-y-aNte and underwent rhotacism to *i-r-aNte.  Subsequently, with
widespread loss of inalienable inflection of body parts, some PMV dialects
had *yaNt-e, while others had *raNt-e.  We can suspect that the *-e is
separable here, because some dialects evidently had *yaNt-ka or *raNt-ka
(cf. Winnebago naNaNc^ge).

Forms like 'ice' and 'pot' could be handled similarly, assuming
inalienable senses like '(one's) pot' or '(it's) congealed skin' (or
water), but the hitch here is that these forms do not consistently show
nasal vowels, so there is nothing to explain OP n, for example.  And, in
fact, the 'heart' forms have reflexes of *r/ _VN, while 'ice' and 'pot
have reflexes of *R/ _V(oral).  They just look similar in OP.

JEK