Companion Terms for 7 and 8 (Re: 'eight' some more)

[Koontz John E]

On Wed, 5 May 2004, R. Rankin wrote:
> Apart from the fact that, for me, appeal to such hypothetical evidence
> just isn't comparative linguistics, ...

I think you're right there.  It's not comparative linguistics, as there's
nothing to compare.  I do think it's a sort of second level deductive
process that might qualify as historical, if not comparative, linguistics.
It is, of course, informed by comparative linguistics.  None of this would
have occurred to me if I hadn't been bothered by -raabriN in Chiwere
'eight'.  Things would remain there, of course, if it weren't for the IO
form providing a possible glimpse of what might have been.  I am, if not
comparing, at least correlating the MI 'eight' form and the deduced gap
atr "eight" in the IO numeral set.

In a related sense you are comparing the MI 'eight' form with the Tutelo
'eight' form, though with the considerable advantage an two concrete
forms.  Tutelo p(h)alaaniN is certainly a reasonably good fit for MI
paraare ~ palaani and the 'six' forms David brought up are suggestively
similar, too, but a certain amount of geographical hypothesizing is
necessary to bring Tutelo and MI into conjunction in an appropriate time
frame, even though they're not so astoundingly far apart that we start
wondering about coincidence.  Tutelo 'eight' is the best surviving fit for
MI 'eight', and it's clear that MI 'eight' is of Siouan origin, but it
seems a bit of a leap to assume that because Tutelo is the best fit and a
reasonable fit it is "the fit."

The Tutelo form looks good, but was Tutelo in the right place at the right
time?  Actually, another Southeastern language would probably do as well
as far as the form of 'eight', and perhaps with less strain.  The combined
evidence of Tutelo and Ofo suggest that Biloxi's pattern of 'eight' is the
divergent, innovated form, and that if there were other Southeastern
languages now extinct, they would have had an 'eight' consistent with
Tutelo and/or Ofo.  If Tutelo was located in West Virginia about the time
of contact it could have been a Siouan language located there because at
some point a largish population of Pre-Tutelo speakers moved there from
elsewhere, but it is at least as likely that it reprented a local variety
of a family with a more or less contiguous distribution from the
Trans-Mississippi to West Virginia.  If Algonquian expanded into the
middle of this area then wherever MI came into existence in this process
it might have been in the presence of a Southeastern Siouan substratum and
it would be very likely that any Southeastern language would have had a a
Tutelo and/or Ofo-like form of 'eight'.

What I did instead of taking this route was to notice that Mississippi
Valley languages also have, in the case of Dhegiha, 'eight' forms not
unlike the Southeastern 'eight' forms, and that, unlike MI-Tutelo contact,
which rests entirely on 'eight' and a very general proximity, there is
considerable historical evidence of direct contact and even merger of MI
and MV groups.  So I asked myself what it would take to get the 'eight'
form from MV insteadof SE and came up with the previously observed problem
of the IO 'eight' form's shape, plus the observable fact that IO and
Winnebago have forms of 'three' that are pretty good matches for the
'three' part of the MI 'eight' term, this form of 'three' being more or
less precisely the form of 'three' so strangely missing in that IO 'eight'
form.  The lingistic logic is admitedly much more complex than the simple
similarity of the Tutelo attested form, but I think that the geography and
history of the MV-source hypothesis are much simpler, to the point of
being essentially a matter of record.

> I'm certainly at a loss to explain how a transparent hypothetical older
> form **phe:-ra:niN would be replaced by an equally or
> not-quite-as-transparent borrowed replacement kre:-ra:briN.

This is one of the mysteries of historical linguistics.  Yet we know that
Siouan (and Algonquian) languages repeatedly replaced perfectly good, and
in some cases, perfectly transparent numerals with others, either borrowed
or locally produced.  While I would certainly like to know why this
happened in every case (and we have some general ideas, certainly) to some
extent it's enough to know that the pattern is attested to occur.  So, one
might hand the question off by saying that the reasonswere not unlike
those which explain why MI took up a Siouan 'eight' form.

My guess would be, in both cases, that a more or less bilingual or
multi-lingual population ended up sorting several numeral systems into one
of mixed origin.  In fact, I think that the nature of the Siouan numerals
suggests that multiple systems within one language were the norm in the
past.

> But that argument doesn't work for replacing our hypothetical
> **phe:ra:niN.

In the particular case of replacing one transparent formulation with
another the very transparency is probably a factor.  If 'eight' is clearly
'three more' then any formulation for 'three more' is equivalent, and if I
like the sound of 'thereto three' better and others follow me, the formula
has changed, or, at least, an additional formula has come into being.  In
the hypothetical IO case there was a population in which a *phe or *pha
series and a *gree series of formulations coexisted, complicated by some
speakers using *raabriN for 'three' and others using *daa'niN, presumably
as parts of coherent series of lower-digit numerals with different MV
dialect origins.  Other forms may have existed as well, some less
transparent, just as hki'etopa and hpedhabriN coexist(ed) in Osage, but
without any conception that only one numeral system was possible at a
time.

> I assume the Osages (and Kaws) borrowed a Wichita word for 'eight'
> precisely because it resembled the Siouan term for 'four' with a prefix
> they could folk-etymologize.

I'd forgotten this aspect of the Osage form!