Locative Postpositions

[Robert L. Rankin]

[In the demonstrative particle]
> It's hard to say whether it's an underlying glottal stop or an
> epenthetic one.  Whatever it is (and I haven't heard the relevant
> examples myself), it caused Dorsey to write two e's in a row.
> Dorsey's not terribly reliable on glottal stops.

"Epenthetic" means 100% predictable in purely phonetic terms; none of
the trendy '70's  "grammatical" or "social" environments.  In that
generally accepted sense it cannot be epenthetic.  Dhegiha languages
generally lack the rampant glottal stop additions of Dakotan, where many
if not most words that begin with a vowel have a [?].  I think that most
Siouan languages lack this productivity for glottal stop.

These recurring, non-predictable glottal stops are found in a handful of
words including ?u:N 'do, be', ?iN 'wear about the shoulders', ?e:
'general demonstrative', ya?iN 'think', ?o: 'shoot at and hit, wound',
?iN- 'stone, rock' and a few others.  Some are nouns, some verbs.  Most
seem to be word-initial, which is suspicious of course.  There are a few
others I think.  Very messy, but there they are.

If one seems to have a semi-predictable segment, i.e., one with apparent
grammatical or lexical conditioning it can mean several things.  (1) it
can mean that the segment *used to be* epenthetic but that recent
changes have obscured part of the conditioning (Dakotan epenthesis or
Omaha *x? > *k? > ?, etc.).  (2) it can mean there have been borrowings
(including Labovian dialect borrowing) that may have muddied the waters.
(3) or it may be wishful thinking on the part of the linguist who sees
some partial but accidental pattern.

Pan-Siouan glottal stops are peculiar enough that I'm perfectly willing
to entertain (1) or (2) above as possibilities in Proto-Siouan.  But
they are also general enough (existing in several daughter languages
spread across a lot of geography) and unpredictable enough that they
cannot be epenthetic in any synchronic phonology.  (The same argument
holds for verb-final, unaccented ablauting -e, which is found from
Montana to Virginia).  These facts merely promote the argument about
possible conditioning factors back about 3000 years and make them more
difficult to un/re-cover and justify, of course.  But that's how we have
to do it if we want to get to the bottom of the problem.

Bob