Mandan

[Koontz John E]

On Tue, 17 Jan 2006, Rory M Larson wrote:
> Alright, since John has removed OP from the list of polite topics for this
> forum, I've decided to learn Mandan instead.  :)

No, no!  Surely nothing I said couldn't be construed to mean that!? I
think I may have said something left-handed about my tendency to done on
about it.  Still, it is the world's most interesting language, and I
intend to continue talking about it.

> 1. The stop series is reduced to simple p, t, k, and ?.  All distinctions
> based on pre- or post-aspiration or glottalization have merged together for
> p, t, and k.  The resulting sounds are voiceless and somewhat aspirated.

In fact, I think Richard Carter had noticed a few aspirates, though they
were rare, and the historical ejectives in some cases seem to appear as
root-final glottal stops:  PMV *C?V : Ma CV(?) where (?) means a glottal
that comes and goes depending on context.

Aspirates and nonaspirates about equally common in Teton or Santee Dakotan
- where is Yuri Tambovstov when I need him - and compare to the tense vs.
lax opposition in Dhegiha or (more or less) voiceless vs. voiced
oppositions in Ioway-Otoe and Winnebago.  On the other hand, ejectives are
rare everywhere in Mississippi Valley, and unattested outside of it, as
far as I can recollect, apart from those Mandan traces.  And, in a similar
way true aspirates are pretty rare in Dhegiha, and correspond to some
odd-ball things here and there elsewhere, though mostly they merge with
either the aspirates or the non-aspirates.

Rare aspirates do occur elsewhere.  I don't recall his examples, but
Carter found them in Mandan and I believe Wes Jones pointed out a few in
Hidatsa, too.  In the causative, I think, but also in other places.  In
these languages, when aspirates have been noticed, they have been treated
as clusters.  I think this is the sensible approach, too.

It seems more or less certain that aspirates contrasted with non-aspirates
in Southeastern or Ohio Valley Siouan, too.  But they are only
indifferently recorded.  Dorsey writes dots under some of the
non-aspirates in Biloxi.

Incidentally, Mandan also has root final -h, and though it would be nice
if that corresponded to aspirates (or preaspirates) elsewhere, it seems
that it doesn't, or only very irregularly.  In fact, the -h things show up
here and there elsewhere, e.g., in Omaha haN*h*e, Dakota haN*h*epi 'night
time', for example, and in the arbitrary alternation between *-ka and
-*hka in noun and verb extensions.

The evanescent final ?, h, and also r and sometimes r? in Mandan are
described in Hollow's dissertation.  Carter pointed out to the CSD
committee that Biloxi seems to have lots of final d's matching the Mandan
r's, at least as a phenomenon.  When followed by *e this produces -di.

Comparative Siouanists argue a lot about whether any or all of these extra
bits - h, ?, r, etc. - are epenthetic.  Some hold that they are organic.
If anyone is left standing after one of these arguments, they move on to
the following vowel.  There is no consensus, but then the people
interested number about four, and some or all of them have changed their
views from time to time.

> 2. The fricatives s, s^, x, and h are all voiceless.  If there was ever a
> separate z, z^, g^ series, these have merged with the voiceless forms.

There is an artilce in IJAL c. 1970 by G. H. Matthews addressing this
interesting problem.

> 3. There are two very similar phonemes which occur in three phonetic forms.
> Both are r in intervocalic position unless the preceding consonant is a
> nasal, in which case both are n.  The two phonemes are distinct only in
> initial position, where one is n and the other is a nasal d, like the sound
> at the end of English "hand".  I assume that one of these is Siouan *r and
> the other is Siouan *y.  But surprisingly, the 'hand' instrumental prefix
> is du-, while the 'mouth' instrumental is na-, which cuts across what I am
> familiar with in MVS, such as Dakotan yu-, ya-, and OP dhi-, dha-, which I
> had thought went back to Siouan *ru-, *ra-.  Can anyone explain what is
> going on here?  And which is which with Mandan d and n, if that's even the
> correct question?

All I can remember is that Hollow reduced them to just r.  And Dick Carter
suggested it might be more complex than that, but not in print, even
unpublished print.

> 4. The only other Mandan consonants are w and m.  I don't know if either of
> these is the result of multiple Siouan phonemes collapsing together.

Hollow concluded that m and n are allophones of w and r before nasal
vowels.  Nasality spreads forward through things like w, r, and h, too.
Terry Kaufman has suggested that you can do pretty much the same thing
with Winnebago and Tutelo.  In fact, you can go far with the idea in most
Siouan languages, except Crow and Hidatsa where in the modern languages
lack nasal vowels and m and n are allophones of w and r (or d/l), but
conditioned by things like the beginning of the word or clustering.
However, Randy has show that early Crow word lists tend to have m and n in
roots that are nasalized in other Siouan languages, suggesting that the
lost of nasality is fairly recent.

> 5. Mandan has 10 vowels, 7 oral and 3 nasal:  a, E, e, i, u, o, A, and aN,
> iN, uN.  The word for 'buffalo' is ptiN, which I suppose is cognate with
> MVS *pte.  Are there any known regular vowel shifts between Mandan and MVS?

I think the conclusion in Hollow is that it's actually aeiou aN iN uN.
But there is probably also vowel length and length plus some other
environmental conditioning accounts for some allophony.  For a good time
compare the Winnebago vowels in Lipkind or the Omaha-Ponca vowels in
Dorsey with the modern lists.   I guess OP with length is post-modern OP?

> 6. The causative suffix in Mandan is -hErE, or perhaps we should say -h.rE,
> where the period indicates a weak vowel that regularly adopts the value of
> the one following it.  Thus, if -hErE is raised to /a/ grade ablaut, it
> becomes -hara.  This sort of thing seems to happen commonly in Winnebago
> where we have what would be a consonant cluster in Dakotan or Dhegihan.

Mandan and Crow and Hidatsa and Tutelo, too, I believe all show something
like "Dorsey's Law" in Winnebago.  The Crow and Hidatsa version involves
some collapsing of the vowels inside the cluster.

> At any rate, the h preceding the r seems to answer my long-standing
> puzzlement over why the causative alone in Lakhota and OP does not do
> the bl-, bdh- thing when inflected for person.

I agree.  I think that comparing the Mandan, Hidatsa, and Biloxi
causatives fully explains the rather
arbitrary situations in MVS somewhat along these following lines:  the
PS causative consisted of the main verb followed by subordinating marker
*hi followed by a causative verb *E (*a ~ *e).  The causative verb
acquires an epenthetic *y when preceded by *hi.  So the original paradigm
was something like:

A1 VERB=hi-wa-e
A2 VERB=hi-ya-e
A3 VERB=hi-ye

In most languages *y and *r merge, and in Mississippi Valley as a whole
intervocalic *y becomes *r.  So, *hiyE > *hirE most places.  Sometimes it
is treated as *hErE.  This *hire ~ *hErE is usually treated as a
prefixing stem rather than an infixing one,
but in some cases not.  Infixing forms occur in Hidatsa, I think, and
Biloxi.  I can't remember for Mandan.  Also, for example, Winnebago has:

A1  waa    < (hi)-wa-e
A2  raa    < (hi)-ya-e
A3  hi(i)  < hi(-ye)

The *hi is lost in A1 and A2, and the *re is lost in A3 while *e is merged
with the pronoun in A1 and A2.

The datives are mostly formed by adding *k- to *hire, hence Da *khiyA, OP
*khidhE, Os *kshidhe, and also Winnebago -gigi which is probably from
something liek *ki-k-hi, with an extra *ki added.  And *kh > g in
Winnebago.

> 7. Mandan uses two augments for plurality: nit for 2nd person and kErE for
> 3rd person.  Of these, nit causes preceding -E to ablaut to -a, but kErE
> does not.  Neither of these looks much like MVS *pi, or like ire, i, or E
> either.

However -kErE is usually considered to be a cognate of -ire and the
similar forms in IO and Tutelo.

That nit (Hollow's riNt) is a bit of a puzzle, though.  I wonder if it
might not be an auxiliary.

Incidentally, while the Mandan future looks a lot like the Mississippi
Valley one, Hidatsa and Crow form the future with an auxiliary, which, on
the whole, looks a lot like a remnant of Dhegiha's obligatory miNkHe,
s^niNkHe, etc., auxiliary after the future in -ttE- < *-ktE-.

> 8. The 'we' affixed pronoun is nu (nuN ?), as Bob pointed out recently.

A very complex subject.

> 9. The more assertive demand particles are gendered in their usage, as in
> other Siouan languages.  Unlike other Siouan languages, however, the gender
> is according to the person addressed, rather than according to the speaker!

This seems to happens sometimes, too, in Biloxi, and - I'll have to check
- the vocative of endearment in Omaha-Ponca varies with the sex of the
addressee.  I'm thinking about dhaN/dhe after truncated forms like
siz^iNxti < nisi z^iNga=xti 'very little offspring' or saNz^iN < saNga
z^iNga 'little younger brother' or s^pa < ttu's^pa 'grandchild'.  I've
never seen these vocatives elsewhere, but Kaw substitutes saNz^iN for
saNga for one of the sexes' little brothers, so we know that the
diminutive truncation was formerly more widespread.

I think there are some cases of dhaN/dhe with non-diminutives.

> 10. In comparing Mandan with MVS languages that I am more familiar with, I
> am amazed at the disparity between different grammatical systems in their
> degree of similarity.  Some systems, such as subject pronouns, instrumental
> prefixes, locative prefixes, reflexives, the causative, the come, go and
> arrive verbs, the conjunction ki and the postposition -ta seem to be almost
> identical in form and meaning; while other systems, such as demand
> particles, augments, positionals, question words, and all but a handful of
> the lexical vocabulary is entirely different.

While I remember seeing comparisons of these particles before, and I've
looked at them briefly, it would be interesting to hear more about your
feelings wrt these!

A lot of the unfamiliar vocabulary shows up in Crow-Hidatsa, but not all
of it.  I've always thought it might be fun to compare it with the
unfamiliar Algonquian vocabulary in Blackfoot or, at any rate, with
non-Siouan languages of the area.

Features that Mandan shares with Crow-Hidatsa include the *aku- and *aru-
prefixes and forming the dative in a serial verb construction with *k?u
'give'.  But I think on the whole that Mandan is, if not a Mississippi
Valley language then at least a parallel branch with MVS something like
"Central Siouan."  However, there is such heavy influence from
Crow-Hidatsa that it's hard to tell which.

> Do we have a good handle on where Mandan fits in the Siouan family tree?
> I've heard Mandan related variously with Crow-Hidatsa, Southeastern, and
> Winnebago.

While Crow-Hidatsa has some features like Southeastern, e.g., the verbs of
motion, or the *raka- form of the striking instrumental, I think most of
these turn out to be due to MVS innovations.  Mandan doesn't really have
much in common with Winnebago but Dorsey's Law, and IO, which is very very
similar to Winnebago, lacks that, too.  Mandan does seem to have features
in common with MVS and Crow-Hidatsa.  Among the Southeastern languages
Tutelo is especially similar to MVS.

> I'm convinced by now that it's not MVS, so the last possibility can
> probably be scratched, but beyond that I'm not sure.  (I've been
> assuming that Southeastern and MVS are closer to each other than either
> is to Crow-Hidatsa.  Is that generally accepted?)

Not really.  I think it's just that we know rather little about
Southeastern, which is rather diverse, internally, and we know a lot about
MVS which has a lot of shared innovations and a rather similar set of
retentions.  MVS and CH are clear subgroups.  People tend to accept
Biloxi-Ofo, too.  But the positions of Mandan and Tutelo are somewhat
problematic.  Both share elements with both MVS and their "neighbors" (if
Arkansas and Mississippi can be called neighbors of Virginia).

The waves of arguments about subgrouping, apart from the major
reconstructions, include papers by Voegelin, Rood, Carter and Oliverio &
Rankin.  I anticipate continued disturbances along that frontier.

> Enough ruminations for tonight.  If anybody has any insights, I'd be
> delighted to hear them!

Very interesting indeed!  It's especially neat the way your attention is
drawn to much the same things that I've noticed and David and Bob and Dick
and Wes before that, and Hu Matthews and Terry Kaufman before that.  And
you come up with the same conclusions (or pairs of possible conclusions)
from them.