'Heart'

[Koontz John E]

Incidentally, naN'aNde 'heart' is NOT a *R word.  It belongs to a small
class of body part terms that have *y > c^h in Dakotan, but *r/_VN > n
elsewhere in Mississippi Valley Siouan.  So the set here is Lakota c^haNl
~ c^haNte', OP naN'aNde, Wi naNaNc^ (I seem to recall this alternant) ~
naNaNc^ge'.  This particular term normally as a -ge < *-ka extension in
independent form in Winnebago.  The alternants in Dakotan and perhaps
Winnebago show that this stem is "consonant final" in some sense.

Actually, since only Dakotan and Dhegiha distinguish *y and *r, it's kind
of hard to say what happens in Chiwere, Winnebago, Mandan, etc., where
these merge.

I suspect that the variation here has to do with different treatments of
forms with/without inalienable possessor pronominals prefixed, though body
part terms are not usually inalienable in modern MV languages.
Hypothetically, for example, suppose the stem was *aNt-, and *i-aNt-
'his/her heart' > *yaNt- in PreDakotan, but in PreDhegiha *aNt- ~
*i-r-aNt- with an epenthetic r was perhaps reanalyzed as a *raNt-.

There are some complications with carrying this to other branches of
Siouan, and I don't want to insist that this is the particular explanation
involved.  For example, in Crow-Hidatsa there is a paradigmatic opposition
between body parts (normally possessed inalienably) that seem to have an
organic initial i and those that don't.  I don't recall the details.  I
seem to recall that this doesn't line exactly with the *y : *r sets, but I
wonder if it might be connected.

Anyway, it might be interesting to see if any of the other *y : *r sets
might involve length, even if there aren't minimal pairs.

JEK