Instrumental/locative.

[Rankin, Robert L]

My thanks to everyone who has responded on this question.  My immediate
question was indeed the less important (but not insignificant) problem of
how to communicate the "truth" about Siouan to an audience of general
linguists.  And in a single 3 or 4 letter abbreviation.....

I've decided to settle on INSTR for the more specific yu-, ya-, ka-, etc.
sets, both inner and outer, and use LOC for the more general i-.  David,
Regina, etc. seem to go along with this, and I think I agree that this is
the easiest way to handle it.

The larger and far more interesting/important matter of the "real" identity
of i- is a can of worms I hadn't intended to open just now.  My
understanding is that there are at least TWO distinct prefixes historically,
and probably synchronically, in each Siouan language.  One is instrumental
i- and the other is locative i-.  The locative generally signals 'movement
toward'.  One has a long vowel and/or is inherently accented.  The other is
short/unaccented underlyingly. But obviously accent and
lengthening/shortening rules in particular languages along with speakers'
reanalyses can be expected to mess with this nice scenario.

As for the semantic bleaching that affects the i-instrumental, this is a
process that routinely affects derivational morphemes in every language.
What we always end up with is an affix with a CONTINUUM of
meanings/functions that were originally contextually defined.  The context
is often lost or blurred and the morpheme moves along the continuum that
exists between polysemy and homophony, often ending up in the latter camp.

One of the points I tried to make in the paper I'm revising is that, because
of the very nature of morphosyntactic change, it will sometimes not be
possible to make a principled distinction between auxiliaries, enclitics and
suffixes in Siouan post-verbal morphology.  There will always be cases in
which assignment is somewhat arbitrary.

For example there are Dakotan enclitics that betray their (former?) AUX
status in that they are still marked for number.  And in Dhegiha there are
enclitics (?) that are still inflected for person -- but only 2nd person.
Then there are the various evidential-like constructions built out of /ehe/
'say' that seem to be both verbs and enclitics.  I think everyone is
familiar with the sort of phenomenon I'm talking about.

This does not, of course, solve the problem of the 2/several i-'s, but it
suggests why the problem is thorny, and perhaps insoluble on principled
grounds. Like everything else in linguistics that is interesting, it implies
a continuum rather than a set of discrete categories.

Again, many thanks for the input.

Bob