ie 'speak' again.

[R. Rankin]

> Well, unless we have large groups of verbs in which the "locatives" seem
> functionless and the apparent underlying stem can't be plausibly
> etymologized, my inclination would be to leave it a mental note to be
> alert for such cases and go on treating locative-like vowels as locatives
> as the working hypothesis.

The are lots of those, and i- is the biggest class.

> Actually, i'e 'to speak' and uhaN' 'to cook, to boil' are the two main
> examples that some immediately to mind.  I tend to think that the final -e
> of i'e is the final -(h)e of 'to say', though this is only in the nature,
> again, of a working hypothesis.

*e:he only reduces to e(:) in Dhegiha as far as I know.  Maybe not even 100%
there.  Unless ie 'speak' retains the /h/ in one or another of the languages, I
can't consider ie a reduction of i + he.  And i- still doesn't make sense here
even if the e < he.  Boas and Deloria mention that there are lots of
unanalyzable i- prefixes.  I haven't looked at the o's and a's.