Awahawi / Amahami

[Koontz John E]

On Wed, 21 Jul 1999, Robert L. Rankin wrote:
> Another interesting (and undiscussed) alternation of b/m and d/l/n is
> found in Dakotan where underlying (and reconstructed) p, t, k end up
> syllable final due to compounding or reduplication.  Dakotan does not
> license such obstruents syllable finally, so they become their
> corresponding resonants, which may be either oral or nasal depending on
> the preceding vowel.
>
> So you get sapa 'be black' but sab-sapa 'black redup.'  Then:
>            nu~pa 'be two'  but num-nu~pa 'two redup.' with an [m].

Something like this happens in Ioway-Otoe, where at least some dialects
have haNma < *haNpa 'corn' and s^uN<ng>e < *s^uNke 'dog', if I recall the
forms correctly.  This is without truncation.

And, of course, the alternation between regular *wa and syncopating *p [b]
~ *m in the first person is also comparable.  The second person
alternation of *ya with *s^ ~ *z^ is similar in a general way, but is
clearly not an *r ~ *t [d] alternation.

> These [b, d, g] have (I think mistakenly) been analyzed by a whole
> string of linguists as variants of p, t, k syllable finally, but what
> happens to the dental series in Lakota (a clear resonant, [l]) and what
> happens after a nasal V (m, n, ng) make it clear to me that the [b, d,
> g] are what Keren Rice has called obstruent sonorants.  And since
> sonorants are voiced in their unmarked state, this explains why we get
> voiced stops syllable-finally, and we resolve what many linguists
> considered a paradox.

This might be a reasonable place to point out that the simple stops are
very rare in initial position in Siouan languages, being virtually
nonexistent there in nouns (but, e.g., Da pa'ha, but Dh ppahe' 'hill' -
not a regular correspondence).  They are also fairly rare in verbs, though
some of the morphemes involved (inner instrumentals in *p or *k) are
fairly common.  Most of the examples in verbs participate in either the
syncopating conjugation or the g-lenition conjugation where these exist.
These examples are, of course, post-inflectional.  In pre-inflectional
position (in preverbs) simple stops are again virtually nonexistent and *R
and *W rule the roost in the outer instrumentals.  Simple stops do occur
in numerals, and are common in medial/final position.

Since the (unclustered) *R and *W sets are close to being restricted to
precisely the environments where *t and *p (also *k) don't occur, I've
suggested in the past that *R and *t (and *p and *W) might once have been
allophones, though I've pretty much had to abandon this in the face of
evidence from Rankin that these reflect clusters.  I think Kaufman's view
when he was looking at Siouan data was somewhat similar (to that of Rankin
et al.).  Of course, it's also possible that the "missing" initial simple
stops are represented by some *w and *r.  Initial *p (verb
post-inflectional slot) certainly becomes w in Winnebago and Ioway-Otoe.
However, initial *t and *k are more conservative, manifesting as stops or
affricates.

It's also possible that the simple stops in initial position are in
alternation with the preaspirate stops there, under accentual
conditioning.  Preaspirates are about as rare in medial position as simple
stops in initial position.  This approach has been considered by Dick
Carter and Bob Rankin, I think.  I believe it proved not to quite work,
but hasn't been totally rejected either.

It appears that simple stops behave somewhat unusually throughout Siouan.