Status of PS Glottal Stop

[Robert L. Rankin]

> Just in case there's any question about it, I should make it clear
> that I'm not denying the existence of a distinct paradigmatic pattern
> for so-called *?-stems (Proto-Siouan glotal stop stems) - a pattern
> distinct from *r-stems or regulars.  I'm just saying that, given the
> data, I wonder if the class in question might not be *V-stems,
> Proto-Siouan stems beginning with a vowel, rather than *?-stems.



> It's the same stems, and the same distinctness of pattern.  Only the
> reconstructed stem initial changes, and even then only in my manic
> phase.

> It isn't epenthesis now, of course, but I'm using the term in an
> ostensive way.  At some point, one presumes, it was a phonetically
> regular epenthesis, process, but it's been grammaticalized.  It's
> messy, of course, not to know what the exact conditions fo the
> epenthesis were, and frustrating to think that they may be
> fundamentally unknowable, but it does seem to require some sort of
> explanation why reflexes of *? are so shy about manifesting themselves
> in stems that we suppose to have started with *?.

Part of this, I've no doubt, has to do with phonological typology.
Laryngeals are simply more prone to being lost than buccal articulations.
We see this in many languages.  In Siouan both /h/ and /?/ are included
and are demonstrably subject to frequent loss.  This alone could account
for "defective phonemes" or the like.

The fact that just the same verbs fall into our R, ?, H, W, etc. classes
in all the Siouan languages that have irregularities means that, if [?]
was ever epenthetic at all, it had been lexicalized already in
Proto-Siouan times.  We have very few alternations upon which to base
further reconstructive hypotheses.

> Taking some steps in the direction of characterizing when epenthetic
> *[?] occurred, as opposed to epenthetic *r or *w (*[y] or *[w]?), it
> seems to be at the PRO + STEM-INITIAL boundary, where STEM-INITIAL is
> defined to refer to the initial of the part of the stem that the
> second person precedes, i.e., not locatives or "outer" components.  I
> suppose *? also occurs word initially (before vowels).  Conversely,
> epenthetic *r (or *w occurs) preceding or between locatives, or
> between locative and STEM-INITIAL, maybe between stem(-final) and
> enclitic or stem & stem.

I still have trouble calling it epenthesis if we have to appeal to
boundaries, which are essentially morphosyntactic, not phonological,
constructs.  And we'd expect w/r/? all to occur with a single lexical verb
stem under different affixal conditions, wouldn't we?

> The complications occur when the epenthetic *? or *y or *w is taken as
> part of the stem and introduced into forms where it doesn't belong,
> especially if subsequently (or concurrently?) the epenthesis process
> is lost, or the *y is rhoticized (which evidently happened in PS).

I think you've offered good evidence for considering our so-called W-stems
to have an epenthetic [w].  There are two convincing cases *waNke 'to be
lying down' and *i-waNxe 'to ask'.  Both have variants with reflexes of *r
instead of *w:  Dakota waNka vs. Lakota yuNka 'lying down' and Dhegiha
*i-waNxe varying with *i-raNxe 'ask' where there are reflexes of w~r in
each case.  These were, at some much earlier time, vowel-initial stems
probably.  But then how can R or ? stems also have been V-initial stems?

*Y has completely distinct reflexes.  You're assuming, if I understood
correctly, that proto-Siouan *r, (Dakotan y and l/d/n, Chiwere r or l,
Dhegiha dh, l, r, etc., etc.) was once [y].  But there is a separate *y
(Dakotan ch, Chiwere y or zh, Dhegiha zh, Biloxi y, Ofo ch, Tutelo y) that
competes for that slot.  Nor does Catawba support an assumption that *r
goes back to a yet-earlier *y.  Why not just r?  It is the unmarked and
often epenthetic glide in many South American languages (which seems to
show that a tap is just as good a transition as a [y] or [w]).

Bob