ie 'speak' again.

[R. Rankin]

> . . . but I don't know of any verbs that start with vowels that don't
> get handled as locatives or preverbs.

Right.  Nor do I.  But I don't think we can say that this means that every such
verb actually has/had a locative (or instrumental) prefix.  It's just that
there's been massive analogical leveling and concomitant extension of the
infixing patterns.  This eliminates infixation as evidence for the presence of
locatives/instrumentals, and we'll have to rely on other criteria.

> The verb 'to say' seems to be just
> e(e)' in the third person, but this is suspect of being contracted from
> *e...he.

Right.  It is underlying e:he since it is conjugated e:phe, e:$e, etc. (where
Omaha/Ponca eliminates the p).

> The verb does also seem to have infixed derivation, as in the case of
> i'gie 'speak against someone', and the i is accented, but apart from that,
> and a suggestion that somebody who can should try various sorts of
> complements with it (in various languages), I don't know how to counter an
> argument like this.  It's certainly not unreasonable to consider the
> possibility that i'...e is reformulated from *...ie.

Yes, that's the whole point.  I remember John's remarking on the lack of
V-initial conjugations in Siouan back when we were working on the comparative
dict. project.  This explains why, at least for the three most common vowels, i,
a, o.  They all require infixes (as if, but not because, they were locative).
Intial nasal vowels tend to get an epenthetic *r or *w, depending on the initial
vowel.  As we know this leaves doublets like Dakotan yuNka and waNka 'be lying'
Verbs like 'ask' and 'be sitting' fall into this class.

Bob