MVS 'eight' (Re: Behind the 8-ball)

[Koontz John E]

On Tue, 27 Apr 2004, Koontz John E wrote:

A few corrections and additional comments.

> 'three'    la'ani(N)

Here I mean to represent an accented long vowel as deduced by Oliverio.
Maybe I should have written aa'.  Anyway, as usual ' is not a glottal
stop.

> Biloxi and Ofo have
>            Bi           Of
> 'two'      noN'pa       nuN'pha

With Biloxi and Ofo I'm just adapting Dorsey as adapted by Swanton and
Swanton, respectively, though I'm aiming at a hypothetical phonemicization
of the data.

> In Mississippi Valley we find:
>            Te            OP            IO            Wi
> 'three'    ya'mni(N)     dha'bdhiN     da(a)'<ny>i   daani'
> 'seven'    s^ako'wiN     ppe'naNba     sa(a)'hmaN    s^aagoo'wiN
> 'eight'    s^aglo'gha    ppe'dhabdhiN  greeraa'briN  harumaN'k
>
> The reconstructions supported by these forms are:
>
> 'seven'    *s^aakoowiN ~ *s^aakwaN  (Winnebago looks like a Dakota loan)

I'm assuming that we should expect something more like the IO form in Wi,
maybe *s^aagamaN or *s^aagamiN.  The attested Wi form is exactly the Da
form rendered in Wi orthography, while the IO differs considerably in
detail and looks like it comes from PreIO *s^aakwaN, which would lead to
Wi *s^aagamaN if it were also the Proto-Winnebago-Chiwere form.

> *Entirely* Speculatively finger gestures u*p*ward or near the head might
> have been opposed to ones elsewhere to indicate the upper five of the
> decade.

> IO has an indiosyncratic formation with *kre + 'three'.  What is
> interesting here is that the form of 'three' exemplified is not the form
> of 'three' found in the simple form.  IO and Winnebago agree in having
> reflexes of *RaaniN for 'three'.  The simplification of the medial cluster
> *pr to *R ...

Oops - very obscure.  The initial *R here is required by initial IO d ~ j^
: Wi d (often written t in the sources).  A real (initial) *t becomes
(initial)  j^ in Winnebago, so this is *RaaniN, not *taaniN, because
Winnebago has daaniN', not *j^aaniN'.  IO doesn't distinguish *t and *R,
but Winnebago does.  However, this *R is apparently irregular; some sort
of reanalysis or analogy is indicated.  The rest of Siouan seems pretty
convinced that PSi had *rapriN 'three' (cf. Da yamni(N) or OP dhabdhiN).
IO and Wi seem to arbitrarily convert the initial *r to *R.  I actually
referred to *R in connection with the medial *pr.  Initial *pr in nouns
does become *R in IO and Winnebago, cf. Teton ble 'lake', but IO j^e(e)'
and Wi tee'.  If the following vowel is nasal, *R is realized as n (or *r
before a nasal vowel), as in *priN 'water', cf. Teton mniN', IO
<ny>iN(iN)' and Wi niNiN'.  In 'three' in IO and Winnebago this reduction
of *pr to *R or actually to *r before a nasal vowel has occurred medially.
In referring to *pr to *R I was thinking of of the way that IO and
Winnebago have medial n where Teton has mn and OP has bdh.

I mentioned 'bean' as another example in which *pr appears as *r before a
nasal vowel (i.e., a surface n).  I have now thought of a problem case in
which *pr appears as unsimplified *pr medially in IO and Winnebago.
Awkwardly enough it's another number!  It's *kyepraN 'ten':

       Te              OP           IO        Wi
'ten'  (wi)kc^emna(N)  gdheb(dh)aN  grebraN   kerepaNnaN'(iz^aN)

In these forms Teton wi and Wi -iz^aN are both 'one' multipliers.
(Probably - I'd expect wiN- in Teton.)  OP has gdhebdhaN in Say's list
from the early 1800s, but has gdhebaN everywhere in both dialects today.
This is an irregular and arbitrary simplification of the form that seems
to have caught on.

The important thing to note is that IO has gre(e)'*br*aN and Winnebago has
kere*paNn*aN.  In this (numeral) form (also) medial *prVN does not reduce
to nVN.  One could argue that at last some medial *pr do not reduce and
that *kreeraapriN 'eight' was somehow exempted from whatever processes
converted PSi *raapriN 'three' to *RaaniN in Proto-WC.  This implies, of
course, that *kreeraapriN existed in Proto-WC or that *raapriN 'three'
survived into Pre-IO where it led to the production of *kreeraapriN
'eight' and then changed to *RaaniN 'three' in IO and Wi independently.

Anyway, we have an exception, though not a nice simple solution.  It looks
lik we'd expect IO *gre(e)'naN 'ten', but we don't find it.  (Or Wi
*kerenaN'.)

Notice that the IO initial gree- in 'eight' could actually be from *kye,
like the initial syllable in 'ten'.

> Whenever I'm out on this limb, I always ask myself, if greeraa'briN is a
> loan, what might it have replaced?  Presumably something within the range
> of 'eight' forms we know from other Siouan languages, and one of those
> possibilities there is a form based on *hpa or *hpe + 'three', which, if
> IO 'three' is any basis on which to judge, would have been something like
> *hpaRaaniN or *hpeRaaniN.

Or *phadaaniN or *phedaaniN to put things in something more like
contemporary IO form and less like Proto-Siouan.  These are modelled on
Tutelo/Ofo and on Dhegiha, respectively, of course.

Other possible antecedents might be *gre(e)daaniN (modelled on IO) or
*saagroxa (modelled on Dakotan) or *arumaN<ng>e ~ *arumaN<ny>e (modelled
on Winnebago), to appeal to other attested forms of 'eight' in Mississippi
Valley Siouan.

The attraction in appealing to Ioway-Otoe or possibly Michigamea (if a
Siouan language) over Tutelo is entirely geographical.  Ioway-Otoe was
spoken next door to Miami-Illinois at contact, while Michigamea was spoken
by some outlying or soon to be portions of the Illinois confederacy at
contact - people who later were merged into the MI linguistic population.
Tutelo, on the other hand was spoken east of the Appalachians in Piedmont
Virginia, while Ofo was spoken in Arkansas.  No contact of either with MI
in the historical period is documented and no intimate contact in that
period seems possible on general grounds of adjacency, unless perhaps
between Michigamea and Ofo.

We do, of course, hypothesize that Tutelo and Ofo were earlier spoken
somewhere near the Ohio, or at least that their antecedents were once
spoken near(er) the main body of Siouan languages.  The latter is an
absolutely minimal and necessary assumption, in fact.  Siouan languages
arise from earlier Siouan languages.  They never come into existence
spontaneously.  But we know far less about when and where contact between
MI (or early Algonquian) and Tutelo (or Ofo) (or early Southeastern
Siouan) was possible than about when and where contact between MI and IO
or MI and Michigamea was possible.  In fact we can only hypothesize that
contact between MI and Southeastern Siouan was ever possible.  The strong
Tutelo character of MI 'eight' form is actually part of the evidence
supporting the surprising hypothesis of MI-Tutelo contact, rather than the
Tutelo origin of MI 'eight' being rendered more plausible by the
likelihood of MI-Tutelo contact.

That's my case for a Mississippi Valley source of MI paraare.  I admit
it's not particularly strong and I don't want to suggest that we should so
more than consider it as a footnote to Bob's Tutelo explanation.