Winnebago *p > w (long)

[Rory M Larson]

Thanks for the great overview, John!  I'll be a while trying to
absorb that!

So for single-consonant, syllable-initial situations,
we have:  ?

MVS      Da       OP        Os        IO        Winn

*p       p        b         p                    w

*t       t        d         t                    j^

*k       k        g         k                    g

*hp      ph       pp        hp                   ph

*ht      th       tt        ht                   th

*hk      kh       kk        hk                   kh

*ph      ph       ph        ph                   w

*th      h        th        th                   j^

*kh      kh       kh        kh                   g

Is this paradigm correct?


One minor correction (I think):

> *pro-ka 'male' > -dok 'augmentative'
    [...]
> With *-a after velars this loss via *-e occurs with
> clear *-Ca forms like those in *-ka, e.g., the *pro-ka
> example, which is just nu in OP (no *-ka), but is -dok
> in Winnebago.

Actually, 'male' in OP is nu'ga, which matches *pro-ka.
OP nu means 'man'.  Based partly on this pair, I've been
coming to the view that *-ka functions as a qualitative
generalizer.  I would suppose that it acted to abstract
the quality of the base that it attached to, making
either a stative verb, as in this case, or a noun as an
oblique reference to an entity that characteristically
manifested that quality.  Does this seem likely to you?

Rory







                      Koontz John E
                      <John.Koontz at colorad        To:       Siouan List <siouan at lists.colorado.edu>
                      o.edu>                      cc:
                      Sent by:                    Subject:  Winnebago *p > w (long)
                      owner-siouan at lists.c
                      olorado.edu


                      10/16/2003 02:27 AM
                      Please respond to
                      siouan






This gathers together several threads in my reasoning about Proto-Siouan,
or mainly Proto-Mississippi Valley, phonology and morphology as they
underpin modern Mississippi Valley.  Almost nothing here is new to those
who have been listening to my chatter on the subject over the years will
notice, but the issue of how to deal with the conundrum of final -p in
Winnebago struck me forcibly recently, leading me to synthesize several of
my arguments here.

Naturally, most of the basic assumptions and phonology here trace
historically to Wolff, Matthews, and Kaufman, separately, and occasionally
even to Dorsey, as well as to Rankin, Carter, Jones, and Rood,
collectively and severally, not to mention Rood and Taylor, who
individually introduced me to all this and Taylor specifically relative to
*th, discussed by him as *rh.  I may be unique in my willingness to merge
all cases of *wr and *pr and/or *br in at least Proto-Mississippi Valley,
distinguishing them only by context, in my treatment of final *-e and
*-[velar]a in Winnebago-Chiwere, and in my tendency to treat final
-e/-a/-ka in PMV as morphemes.  I think, however, that most recent
students have considered most of these angles before me without
necessarily coming to the same final conclusions or mentioning all the
possibilities out loud or in print.

One of the interesting things about Winnebago phonology is that *hC
(preaspirates *hp/ht/hk) merge to such an extent with *C (inaspirates
*p/*t/*k).  The merged *ht and *t both come out c^, of course, while hp/p
> p, and hk/k > k.  Unmerged *C come out w, j^, g.  For example, *hpe(e)t-
'fire' > peec^, *se(e)p- 'black' > seep, *pe(e)thaN 'to fold' > weej^aN,
*pras-ka 'flat' > paras (*-ka not attested in Winnebago).

These mergers occur because Winnebago eliminates the contrast of
preaspiration vs. non-aspiration in favor of voicelessness vs. voice.  As
a result, it is syllable initial *C, mainly, remains distinct, as the
voiced series w/j^/g.

All cases of d come from *R (or *pr), e.g., *Reek- 'MoBro' > hi-dek ~
dee(k)-ga, *pro-ka 'male' > -dok 'augmentative'.  Interestingly, the
scarce true aspirates become *w/j^/g, too, so, for example, *thi becomes
OP thi 'arrive here', but Winnebago j^ii.  (Note for Dakotanists - *th > h
in Dakotan, hence pehaN and hi for the verbs above.)

The initial *hC sequences - some people like to think of such things as
clusters, some don't - appear as p/c^/k, as does *C (where retained) in
*Cr or *C when word final final, as in the examples in the first
paragraph.  Word final instances of -C occur quite frequently, as
Winnebago seems to have lost all final *-e.  And there are quite a few of
these, since final *-a after velars seems to have become -e as well.
Thus the proto-forms - nouns and verbs - in final *C-, which appear with
-Ca or -Ce elsewhere, are just -C in Winnebago, per the examples above
again.

With *-a after velars this loss via *-e occurs with clear *-Ca forms like
those in *-ka, e.g., the *pro-ka example, which is just nu in OP (no
*-ka), but is -dok in Winnebago.  The exception is cases where there is a
sequence *-C-ka, which manifests as *-Cke.  So *yaNt- 'heart', which
manifests sometimes as if it had been *yaN(aN)t-e and sometimes as if
*yaN(aN)t-ka (different classs of absolute marker?), appears in Winnebago
as both naNaNc^ and naNaNc^ge.  The same merger of post-velaric -a and -e
is also attested in Ioway-Otoe.

As the -Cke examples show, too, *C (when t or k) is voiceless in CC
clusters, though *p becomes w, cf. *yaN(aN)tpa 'ear' > naNaNc^awa.  The
final -C forms do seem to voice when a resonant-initial enclitic is added
to them, cf., -xjuk 'pulverized' (an instrumental stem), but xjuug-re 'be
crushed'.  Miner generally writes *C as voiced in SC clusters, again cf.
-xjuk < *-xtuk-, or *ska 'white' > sgaa.  With *p this voicing means w, as
in *s^paN 'soft' > s^aN(aN)waN (length not expected here, but indicated by
Miner).

Probably the way to think of this is that *C is voiced in initial position
and when it preceeds a resonant (before resonant-initial enclitics), and
that *hC consistently loses preaspiration.  (I think all initial voiceless
stops are aspirated, and I think final ones probably are not, but merely
unreleased.)  This doesn't handle cases like *tp or *Sp, however.

One very interesting thing about this system of changes though, is that
final *-p actually does appear as -p ~ -b#r-.  This is problematic because
in Ioway-Otoe, where these final *-p are not final, since final *-e has
not been lost, they appear as -w-, e.g., *sep- 'black' is <th>e(e)we.

The IO behavior suggests that all initial and final-medial *p > w in both
Winnebago and Ioway-Otoe.  In fact, since it is generally assumed that
these two languages share a common ancestor later than Proto-Mississippi
Valley, we could say this happened in Proto-Winnebago-Chiwere.

But if it did happen that way, then when final *-e was lost, forms like
Proto-Winebago-Chiwere (or Pre-Winnebago) *seewe (cf. IO (th>e(e)we) must
have become seep, the form attested in Winnebago, with nominal reversal of
*p > *w.  What seems more likely is that Pre-Winnebago speakers were
somehow aware that *w in this context was really or alternatively *p (or
*b), perhaps because there were still at that time alternating forms like
*seewe ~ *seep (or *seeb) in different contexts, not unlike the
contemporary cases of (?)  sapa ~ sap (or sab), *(wV-)ra(a)p- 'beaver' >
c^hapa ~ c^hap (or c^hab) in Dakotan.

Of course, such alternates are virtually eliminated in modern Winnebago,
which is all seep, and modern Ioway-Otoe, which is all <th>ewe.  A
possible exception to this is Winnebago was^c^iNiNk 'rabbit', but
was^c^iNiNge-ga 'the Rabbit', in contrast with hi-dek, but dee(k)-ga
'uncle'.  The -e- certainly looks like a relict *-e.  Dhegiha tends to
follow the Ioway-Otoe model, though we see probable reflexes of the *-C-e
~ *-C alternation in cases like *khet- 'shoulder' > OP iNkhede, but
iNkhe-sabe 'black-shouldered' (epithet of the buffalo preserved in a clan
name), with no -d(e).  I believe the Dhegiha compounds with truncated
first stems reflect phonetic loss of -C in relict *-C forms, since
compounds are -C alternate contexts in Dakotan.

In recalling the Winnebago-Chiwere shift of *p to *w, it might be
appropriate to notice that something similar, occurring in a different and
earlier context might explain the *b- (or *p-) ~ *wa- alternation in the
Proto-Siouan first person.  That is, perhaps we see here reflexes of
**ba-, which becomes *wa-, except in contexts where it was reduced to *b-
before certain consonants.

(We might also wonder if *wa- (perhaps from **Ba-) occurs where it does
(especially before *R and *CC-, including *hC- and *Ch- and *C?-) because
epenthetic *a was needed there to simplify to *baCC- the **bCC cluster
that would otherwise result.  However, it is not necessary here to explain
how we come to have the vocalic alternation that we find in *b- ~ *wa-.)

John E. Koontz
http://spot.colorado.edu/~koontz