First Person Agents in (h)a- < *wa

[Koontz John E]

I had an interesting exchange with Mark Donohue, who is working on a north
Australian languagfe called Palu'e.  In the course of it, I noticed the
following:

       A1 Regular             A2 Syncopating (*stop-stems)

PMV    *wa-                   *h-C... (?), *p-C... (?), *w-C... (?)

Da      wa-                    wa-C... (regularized)

OP       a-                    p-C... (ppa- : ba- :: A1 : A3)

Os       a-                    h-C... (hpa- : pa- :: A1 : A3)

IO      ha-                    C[h]... (p[h]a- : wa :: A1 : A3)

Wi      (h)a-                  C... (pa- : wa- :: A1 : A3)

In particular, it never before occurred to me to wonder if the change of
regular *wa- to (h)a- was connected somehow with the way that *w-C...
appears as a preaspirate.   The IO aspirates or voiceless stops and the
Winnebago initial voiceless stops are reflexes of *hC, just as the OP
tense stops and Osage preaspirates are.

We know that Dakotan probably had preaspirates here, too, but I won't go
into the logic here.  It involves the behavior of k-stems with ki-.

For what it is worth, a shift of initial p to f or h is not that unusual,
and in other contexts *wa- behaves rather like *pa-, e.g., the behavior of
*wa with *r-stems, where *wa-r... appears as *p-r... in Dakotan and
Dhegiha, but as *R... in Winnebago and IO.

Note that initial *p- is fairly rare in Siouan.  It is mainly restricted
to certain verb stems, mostly formed with *pV-instrumentals, and to *pe
'who' and a few highly irregular sets like 'hill' that smell a lot like
loan words.  The verbs in question were all irregular (syncopating).

For the present, I'm not sure if there's anything in this parallel,
really, but I've wondered for a long time why *wa A1 seems to lose its
initial w so easily, when *wa(a)- 'idefinite object' sticks to its w like
glue (outside of Biloxi-Ofo).