Da -c^haghA < *k-kaghE or *-y-aghE (was RE: glides, etc.)

Koontz John E John.Koontz at colorado.edu
Mon Jun 23 09:49:33 UTC 2003


On Sun, 22 Jun 2003, Rankin, Robert L wrote:

(Basic stem) > Lak.   kagha    would = Omaha gaghe
(Suus stem?) > Assin. gijagha  would = Omaha gigaghe assuming it exists.
(Suus stem)  > Lak.   kichagha would = Omaha giaghe (dative stem)

Or it might be Omaha gikkaghe 'to make one's own' (suus stem).  OP
*gigaghe doesn't exist, except as the underlying form of the dative stem.

Here are the comparisons that seem to me to work best.

Category   PMV?            La             OP

Base 1     *p- kaghE       wa-kaghA       p- paghE
     2     *s^-kaghE       ya-kaghA       s^-kaghE
     3         kaghE          kaghA          gaghE

Suus 1     *wa-ki-k-kaghE  we-c^haghA       a-gi-p- paghE
     2     *ra-ki-k-kaghE  ye-c^haghA     dha-gi-s^-kaghe
     3        *ki-k-kaghE  ki-c^haghA         gi-k- kaghE

Dat  1     *wa-ki-kaghE    we-c^aghA        e-p- paghE
     2     *ra-ki-kaghE    ye-c^aghA      dhe-s^-kaghE
     3         ki-kaghE    ki-c^aghA       gi-    aghE

Although the regular datives and possessives are, as David notes, reversed
in form between Dakotan and Omaha-Ponca (and Southern Dhegiha and
IO-Winnebago show additional patterns), their datives and suus actually
match pretty well for consonant-initial stems.  The following developments
elucidate matters here:

- Dakotan loses the consonant initial pattern with the basic transitive
stem of this verb, and transfers it to the regular paradigm.

- Dakota, of course, affricates k to c^ after i, and kh to c^h, etc.

- In the first person of the basic transitive Dhegiha assimilates the
initial k of the stem to the *p (or *w or *b)  of the first person, so the
first person comes out p-paghe, instead of, say, *k-kaghe or *p-kaghe.

- In the Suus (or Possessive) stem this verb has always had *ki reduced to
k- (as *wa and *ra, earlier *ya, reduce to *p and *s^, and this *k- has
been supplemented by adding a regular *ki- to the front of it, producing a
prefixal pattern *ki-k- that appears throughout the suus forms of Dhegiha
and (less obviously) in Dakotan.

- Dakota contracts the pronouns with the underlying ki in the suus, as
Omaha-Ponca (but not Southern Dhegiha) does in the Dative, leading to we-,
ye- instead of *wa-ki-, *ya-ki and e-, dhe- instead of *a-gi-, *dha-gi-.

- OP inflects the suus and dative of consonant-stems doubly.  Withhout
going into the details, I think this arises in two slightly different
ways.  With the suus it is analogical and replaces earlier *a-gi-k-,
*dha-gi-k-, *gi-k- with a-gi-p-, dha-gi-s^-, gi-k-, however that would
work out for the particular kind of consonant stems.  The analogies are
much clearer if you look at the full set of forms.  In some cases (*p- and
*t-stems) only the second person actually changes.  With the dative it's a
result of applying the regular dative over the basic transitive.

- The contraction of *gi-gaghe to gi-aghe in the third person of the
datives looks like it might be a transfer (i.e., a contamination) from the
pattern of the dative of *ka-instrumentals, which look like they lose
initial *k throughout and loses it also in the first and second persons of
the basic transitive.  Comparisons with Crow and Hisatsa suggest that this
instrumental is really something like *raka (third person) ~ *aka (first
and second persons), however.

- Actually, though I've been taking the traditional point of view, that
both the suus and dative have *ki, only these two *ki's behave differently
in each language, there's a pretty good chance that the dative was
something like *riki ~ *iki (patterning like the *raka- instrumental)
instead of *ki.  This seems to work better than assuming two identical
morphemes being kept separate by a series of desperate morphological
expedients.  Unfortunately, I don't think *riki- is actually attested
anywhere, unlike *raka- which is.

- So, if you're still with me, to bring a long story to a sudden ending,
the source of c^h in the suus of Dakotan - I guess I should say Teton -
*k-stems (kagha 'to make', kuNza 'to decree') is probably *k-k-/*ki-__,
not *y, though the latter does (also) become c^h.  In Dhegiha the *k-k-
sequence is indistinguishable from *hk-, i.e., it become kk or hk,
depending on the language's phonetics.  The same is true, in fact, of all
*CC (or *hC) sequences, where C is a stop.  However, all *hC (where C is a
stop) become *Ch in Dakotan, cf. OP kke 'turtle' vs. Dakotan khe(ya), etc.
And thanks to the pleonastic *ki- preceding our *k-k-, that kh is after a
Ce (< *Ca-(k)i-) or ki- that affricates it to c^h.

I say this with a certain amount of deference, because the morphological
context is complex, even if the phonology is simple, and because I really
hate to contradict one (actually two) of my favorite teachers, but I think
all of the logic applied holds water, and it does eliminate an otherwise
troublesome exceptional case in the development of Dakotan epenthesis.

Note that I have accidentally cited a probable additional example of -y-
epenthesis < *r in Dakotan in the form of kheya < *hke-r-a, where -a is
the -a of s^uNk-a, cf. heya < *he-r-a (OP he), wiNyaN 'woman' < wiN-r-a
(OP miN), and a few other examples.  In this last case the historical stem
is probably *wiNh-, but the h is lost in Mississippi Valley.

> Lak.  kichagha < *ki-r-agha where prefixal [ir] > [ich] in numerous
> forms as mentioned yesterday.

Those are an interesting collection of forms - a group of forms, mostly
inalienable, hence with prefixes *wi/*ri/*i - that have c^h (< *y) where
Dhegiha has reflexes of *r.  I necessarily take them to be something
different.  My suspicion is that Dakota is right about *y, and that
Dhegiha popped those *y into the epenthetic *-y- > *-r- set irregularly,
merging them with the real *r- forms.  Since most Siouan languages
brutally merge all *y and *r into one segment more or less *r-like, this
is not too surprising.

It's worth noting that the second person pronoun has a similar problem.
We're pretty sure from Southeastern that it was *ya, and this seems
consistant with *s^- as the short form of the second person (< *y-?).
But all of the MV languages that distinguish *y and *r plop for *ra as the
second person (Dakotan ya-, not *c^ha-; Dhegiha dha-, not *z^a-, though
z^- does show up with Dh ?-stems, suggesting that Da n- there is
contamination from the *r-stems).  In this case Dakotan agrees with
Dhegiha (and, of course *ra- > ya- in Dakotan).

But notice that the contraction of *w-yi- in the A1P2 form is c^hi < *yi-
in Dakotan.  Of course, Dhegiha unhelpfully has wi- < *w-yi-, maybe via
w-ii-?  By analogy with Dakotan you'd expect *z^i-, but you don't get it.
And IO and Winnebago are no help either, since they merge *y and *r, and
have ri(i)- and niNiN-, respectively.  IO does have occasional y < *y (via
*z^, maybe, since that also becomes y), but only in a few places, and not
here.  However, Mandan comes through with miNniN- < *w-yi(N)-, for which I
have always been profoundly grateful!  Note that the nasalization of the
second person patient form comes and goes across Siouan.  I suspect it's
secondary, perhaps by analogy with the first person patient *waN- which
does seem to be nasal.

I hope this helps more than it off-puts [incorporation with fronting]

JEK



More information about the Siouan mailing list