Locative Postpositions

[Koontz John E]

On Wed, 27 Oct 1999, ROOD DAVID S wrote:
> Here are my thoughts on the locatives, seen strictly from a synchronic
> Lakhota perspective.  I would be happy to have corrections or
> countertheories from others, and I apologize to those who are looking for
> specific diachronic information, because there won't be much here.

Actually, I think a diachronic explanation was perhaps the last thing Alan
wanted!  It was I who got things onto that track, it being the first thing
I generally think of.

> 	There is a contrast between postpositional or "inflectional"
> (suffixal)) -ta and verbal -tu; the verb e'tu means 'to be in/at a
> particular place', and I am not convinced that this stressed "e'" is
> the "neutral" demonstrative root John has talked about; the latter is
> probably extant in ekta 'to/toward', etaN 'from' etc.  I think the
> stressed "e" is the verb root 'be', found in e.g. the so-called
> personal pronouns (miye', niye', unki'yepi) and the definite existential
> verb e' seen in sentences like "He' Robert e'" 'that's Robert'. The -l/-n
> marker is, in my opinion, the reduced form of e'tu, not the -ta
> postposition, but I can'nt give you any good arguments for that
> assertion.

I guess I don't distinguish between the two usages in looking at
Omaha-Ponca.  This may be a deficiency in my approach, but I think it's
mostly because Omaha-Ponca doesn't make the verb :  postposition
distinction morphologically by truncating some of the postpisitions.
Certainly e and e=di (cf. e-tu) are both used verbally more often that
e=tta (cf. e-k-ta), though I'm not positive at the moment whether I recall
a verbal example of e=tta at all.  I think I have found other e=POST
sequences that behave verbally, though mostly they all occur
postpositionally (including e=di, there being no analog of el in
Omaha-Ponca).

The existence of the truncated subordinate/postpositional forms in -l is a
major difference between Dakotan and Dhegiha and has clearly had a major
influence on Dakotan grammar or perhaps only the analysis of it, because
it can be reinterpreted in terms of this verbal/non-verbal distinction.

The verb : postposition distinction seems to depend on whether the form
occurs in predicate position in main clauses, where it is naturally taken
as verbal, or not, where it would normally seem postpositional.  I suppose
that the non-predicative instances could just as easily be considered
participial (subordinate verbal), if etu is to be considered verbal. But
if e is the verbal root, then don't all the demonstratives have to be seen
as verbal, or at least some instances of he, too?  I thought tu occurred
with all the demonstratives.  Hetu would be ambiguous, but if gatu occurs
then there's no e present there.

Identifying some e + thing sequences as verbs and some as (demonstratives
plus) postpositions does seem a rather awkward business, though it's easy
to see where it comes from.  Even if a language lacks the untruncated :
truncated forms with their different distributions, there are sometimes
semantic problems.  I'm always bothered in Omaha-Ponca by finding edi
'(be) there, at that' in the same class with enaN '(be) so many, that
many'.  Postposition seems a poor characterization of the latter.  Perhaps
the verbal analysis is the more general and flexible, even if the
postpositional one leads to simpler, more natural translations in most
cases.  I suppose the fallacy I've been slipping into is assuming that one
must categorize based on the least marked translation.

In any event, if we consider forms like etu as verbs, I wonder if the
verbal root isn't the "postpositional" component (tu/l, etc., in Dakotan;
=di, naN, etc., in Omaha-Ponca), not the e, which is simply an
(obligatory)  incorporated demonstrative.

It's true that the e alone can also appear to behave verbally, but I
assume that these sentences are historically 'he/she/it is' clefts
analogous with the first, second, and inclusive person pronominals:  He
Robert e 'that one Robert it/he (is)', with no equational verb verb
present.  My French isn't entirely satisfactory, but this is more or less
analogous to things like Celui, c'est Robert, except for word order and
except for lacking the "est."

We do have to ask ourselves whether the possibility of such an analysis
for the origin of the situation in Dakotan should cause us to adopt it
over the one that David likes, which clearly works, too, if we're not
overly concerned to reduce the number of "e" entities.  A lot would depend
on data that I don't control.  I would have to say that the possibility of
the two alternative analyses only exists in Dakotan, not in Dhegiha, and
primarily because of the pattern of reduction of components like tu to -l,
etc., in various contexts in Dakotan.

I've also been wondering if calling e a neutral demonstrative or 'the
aforesaid' as I usually do might not be just an awkard way to say 'third
person pronominal'.  Of course, when it appears in sequences like miye, as
I assume it does, that's a bit awkward synchronically, though presumably
these are etymologically something like 'it is I'.

JEK