Monday's recordings

[Koontz John E]

On Tue, 3 Oct 2006, Rory M Larson wrote:
> Well, inadvertant to the list or not, I had been thinking about posting on
> this anyway.  I had suggested a couple of years ago that Omaha might
> recognize a distinction between /e/ and /E/, based largely on the minimal
> pair: he, 'horn', vs. hE, 'louse', where to my ear the latter sounded more
> like the /E/ in "pet".  The speakers are apparently able to distinguish
> these two terms, at least some of the time.  ...

> Last night, we had our elder speaker pronounce these terms for us again.
> Most of the time there seemed to be little or no difference.  When she
> pronounced them very carefully, though, it seemed to me that there was a
> difference, but not necessarily exactly what I had thought before.  For he,
> 'horn', the vowel came out sharp, clear, and short, rather like the vowel
> in si, 'foot'.  For hE, 'louse', the vowel seemed to linger and change a
> bit, as a diphthong.  It could be construed as starting with /he/ and
> gliding toward the center, as he[A], with the final part of that glide very
> brief and optional.

How about kke 'turtle'?  Dakotan has kheya.  But OP tte :  Dakotan pte, as
far as I can recall.  It would also be interesting to know if there was
any trace of this pattern in the rest of Dhegiha, though nothing would
preclude OP being a rule unto itself.

How about accentuation?  Any difference in various he-compounds?  I don't
see one in the forms you folks have been citing.  I've often wondered
about the tendency of the "generic compounds" with tte- or tta- plus body
parts, etc., to appear with accent on the body part.  I'd expect accent on
the "beast" part consistently.  The exceptional "accents" in question are
in Dorsey.

The rule I am thinking should account for accentuation would be something
like V => V' / #C__ = C..., or, in English, accent occurs on the second,
or first and only syllable of the first element in a compound, so
*tta'=he, *tte'=he, etc., though I think that one or more of these is an
exception.

Perhaps the rule is more like accent occurs on the second mora of a
compound (like any other word), so that if the first element is short, the
accent should appear on the second element, but if the first element is
long, accent occurs on it. This rule might be somewhat irregular in
application, due to analogy.  But it should tend to result in he 'louse'
being accented, while he 'horn' was not, when they were first elements of
a compound.   Or perhaps something else is at work in the exceptions I've
seen, and 'horn' vs. 'louse' have nothing to do with them.

If you are right about hearing "more stuff" in he 'louse' it should show
up in a sonogram, I would think.

> So now I'm starting to think that there is probably no phonemic difference
> between /e/ and /E/, but that there is a difference between these two
> words, such that 'louse' could be spelled he'a, with the final vowel
> de-emphasized almost to nothing, and only rarely even perceptible.  When it
> can be heard, the shift from /e/ to /A/ passes through the range of /E/,
> which is why my English ear could parse it that way.

It is like ea in weahide or eawa... and so on where there ea is
consistently noted?


> If this interpretation is correct, it would appear that the Omaha word for
> 'louse' is a straight match for the Dakotan term, he'ya.  If so, I wonder
> if we can get this straight from MVS, with *he'a remaining in both language
> groups, but with epenthetic /y/ being recognized in the spelling for
> Dakotan, and the final syllable being almost but not quite dropped in
> Omaha?

The thing to know about Dakotan heya, kheya, etc., is that they are only
bisyllabic as independent forms.  In compounds they are he-, khe-.  Other
words in this set are wiNyaN ~ wi(N)- 'female' and iNyaN ~ iN- 'stone'.
For the short form, check out Dakotan 'nit', which I recall as
hez^aN'z^a=la.  (Dakotan mostly shift accent to the second syllable in
tight compounds with monosyllabic first elements.)

Outside of Dakotan, modulo your OP material, there is no trace of the -ya
per se in any of these forms.  For example, Winnebago has hee (all
monosylalbles are long).  But the "definite" in Winnebago is forms with
=ra, which is a good match for Dakotan -ya, e.g., hee=ra '(the) louse'.
I'd say it was cognate, myself.

I've tried to explain these various little -(*r)a and -(*r)e extensions
that appear in various contexts in various ways over the years.  For the
moment I see them as a system of absolute markers in Proto-Siouan, though
they seem to act as a sort of generic (*a) : specific (*e) pair in
languages where both occur.  For example, -e occurs with body parts,
kinterms, tha-possessives, and concrete deverbal nouns in Dakotan, while
-a occurs in generic body part compounds, animal species terms, and
non-concrete deverbal nouns.

I argue that the -ya in Dakotan heya 'louse' and the -ye in Dakotan
s^ahiye=la 'Cheyenne' or the -we in c^huNwe 'elder sister' are just *a and
*e with epenthetic glides separating them from preceding vowels, and
parallel to the -a in forms like s^uNk-a 'dog, horse' or the -e in forms
like thas^unk-e 'his (personal) mount'.  Most of the Siouan languages tend
to eliminate all (or nearly all) -a in favor of -e, cf. OP s^aNg-e or IO
suNny-e.  (It seems unlikely that Da -a : OP -e is a regular vowel
correspondence.)

In a few cases nouns in final *h, e.g., maybe *wiNh- 'female' and *haNh-
'night' retain the h before a vowel in sporadic contexts, e.g., OP
haNhe'wac^hi 'night dancer' and Dakotan haNhe'=tu 'nighttime' or Tutelo
mi(N)he 'woman'.

Biloxi is particularly prone to retaining 0 ~ di after vowels (< 0 ~
*-r-e) and Mandan has a system of -r- ~ -h-, etc. finals that appears
after vowel final roots when various vowel-initial suffixes are added,
notably -e, which is sometimes called a definitizer (Kennard?) and
sometimes seems to be considered as a sort of absolutive marker - a suffix
added to a noun to make a form that can stand alone without a compounding
or paradigmatic marker following it (Hollow).

> Or would one language have borrowed it from another in more recent
> times?

Conceivably.  Our readiness to allow some "slack" in the regularity of
correspondences, plus a certain "early days" status of Siouan comparisons
has made us a bit insensitive to the possibility of "inter-dialect" loans.

However, I'd have to say that 'louse' is not a good candidate for a loan
word.  You do get them from other folks, but they are not much of a
novelty, anywhere you go.

> Would there be any constraints with what we should expect for the
> development of epenthetic /y/ in MVS?

Well, actually there is a problem there.  If it were PMV *hera, you'd get
Dakotan heya and OP hedha.  If it were PMV *heya, you'd get Dakotan
*hec^ha and OP hez^a.

> How would this compare with the 'speech' term, i'e / i'ye ?  Do other
> MVS languages tell us anything?

This is essentially *ie.  OP (and Dakotan?) workers occasionally write -y-
between two vowels where there seems to be nothing historically.

I think you'd have to claim it was PMV *hea.  But the evidence for that
would be essentially your OP recordings.

Most of the ea sequences that are firmly attested in OP arise from
paradigmatic situations in which we have -a-wa- for 'us' combined with and
infixed i-locative or gi-dative (*-a-(g)i-wa-), or from *wa + i + a where
i and a are locatives.  Without the wa- you get idha < *i-r-a.  Adding wa-
seems to compress things a bit.  In the same way *wa-i-o- yields wiu-
while *i-o yields udhu < *idhu < *i-r-o.  There are a few where *e
'demonstrative;  definite;  third person' is proclitic to an initial a- or
aN, e.g., eaN 'how' or to a -a- extension before a postposition, e.g.,
eat(t/h)a (?).