PS *th => Da h (Re: Archaic A1 p- in Dakotan.)

[Koontz John E]

> > > 1sg	hi-b-u
> > > 2sg	hi-l-u
> > > 3sg	hi(y)u (where -y- is just an epenthetic glide)
> > > 1du  u~-hi(y)u
> > >
> > > There are alternative forms wa-hi-b-u, ya-hi-l-u, etc.
> Koontz:
> > Given the usual pattern of serial motion verbs in Dhegiha, I'm inclined
> > to wonder if the middle variant (cf. wa-hi-b-lu) isn't the most
> > conservative,
>
> ... I agree both verbs might well be conjugated, but then I'd
> expect conservative 1sg *phibu, ...

Oops, as I was thinking about the *W stuff I remembered that PS *re (RE?)
+ *hi => PS *thi => Da hi, while PS *hi => Da i.  That is, the h here (in
hiyu)  is from Taylor's *rh (which I've written *th), which was one of the
main points of his article on motion verbs.  So, this Da hi is the one
that would be inflected regularly ab initio, cf. Dhegiha athi 'I arrive
here', dhathi 'you arrive there', a'thi=i 'he arrives there', aNfathi=i
'we arrive here'.  I'm afraid we students of Dhegiha find Dakota verbs of
motion very confusing, what with all the unusual phonology that converges
there.

Of course, this isn't really a critical point in the context of the
discussion, since I'd readily concede that if it were i < *hi, then the
regular inflection here would have to be secondary.  That is, for my point
to
hold, i in a putative *iyu would have to be being regularized along with
i, without losing the double inflection *waibu, *yailu, etc.  Of course,
in the event, this is just a thought experiment.  The form is hiyu < PS
*thi=u.

To see Bob's point where it applies, consider iyaya, in which the i is
from *hi, but isn't inflected.  I'd expect that formerly this was
inflected *phibleble, *hilele, *iyeye, *uNkiyeye=pi, though I don't know
of any such forms being attested, so this is also something of a thought
experiment.

JEK