Proto-Steppe Numerals

From: Glen Gordon
Message: 3239
Date: 2000-08-18

Piotr:
> Oh, you apparently missed my very direct comments concerning your
> >numeral list:
>
> http://www.egroups.com/message/cybalist/3159

Now where did that come from? I apparently was unaware of this page
otherwise I would be certainly putting my opiniative two-cents worth in :)
Frankly, by asking these questions, you don't know what kind of genie you
let lose from the bottle - I apologize for the extreme length of this
post...

> (1) Unless you want to propose a ternary system of counting at some
> >point, why should the Japanese word for 'nine' have anything to do >with
>the Proto-Steppe word for 'three'?

Eeeeeexcellent question. This is a great time for JAPANESE LESSON #1!
Japanese is known to have a _binary_ counting system as can be seen in Old
Japanese which I can only at the moment quote from heart (so be good little
boys and girls and verify what gLenNy is saying at your local library):


ENGLISH (like duh!) OLD JAPANESE
-----------------------------------------
one/two hito"-tu/futa-tu
three/six mi-tu/mu-tu
four/eight yo"-tu/ya-tu

Here we can see clearly that the vowels alternate in a premeditated way (o"
with a, i with u) in a manner that is totally Altaic-vowel-harmony-like.
Obviously, odd numbers above three don't get to be part this happy binary
clique. Thus we have the following numbers alone, cold, helpless. In a sad
sense, they are truely "odd", like me :(

five itu-tu
seven nana-tu
nine ko"ko"no"-tu

Now it stands to reason that if there are any numbers harking back to an
earlier stage like Altaic, only _half_ of the binary set could possibly be
inheirited, unless of course, there is some sort of evidence for this binary
opposition outside Japanese (and there isn't). The question is which ones
are likely to be inheirited and which are to be derived from their numerical
counterpart? This is the way I see it:

one *bitu -> hito"-tu (> futa-tu "2")
three *gur- -> ko"ko"no"-tu "nine"
four *du"r- -> yo"-tu (> ya-tu "8")
five *hitu -> itu-tu
six *nu -> mu-tu (> mi-tu "3")
seven *na -> nana-tu
ten *tuhan -> to"wo"

As for /kokonotsu/ "nine", the word literally means "three times three" and
is nothing strange given the fact that higher numbers in Steppe languages
tend to be compound words anyway (cf. Finnish kakdeksa"n/ykdeksa"n
"eight/nine" < kaksi/yksi "two/one"). Why, even my reconstructed "nine" for
ProtoSteppe is itself a compound word meaning "one away from ten"! To boot,
"nine", because it is an odd number, could never be formed using the above
binary process. Duh! :)

Piotr:
>Needless to point out, Classical Japanese mi 'three' doesn't match >your
>protoform.

Sorry, irrelevant. The Japanese numeral /mitsu/ is a derivation from "six".
It was "six" that was inheirited from Altaic, not "three" (except as part of
the compound word for "nine", ko"ko"no"-).

Piotr:
>For Altaic, you offer *kul/*gur. I don't know what your extra-Japanese
> >evidence for *kul- is, but *gur- is definitely Mongolic rather than
> >common Altaic.

Yes, Mongolian /ghurban/, for example. The first element of these
reconstructive pairs (like *kul/*gur) is meant to represent Altaic with
Korean (Altaic I), versus post-Korean Altaic (Altaic II). Of course, in this
case, Korean doesn't show this numeral but rather /set/.

>Likewise, you sell Finno-Ugric forms as Proto-Uralic. It's a well->known
>fact that common numeral protosystems (even in the 1-5 range) >are only
>reconstructable for the subfamilies of Uralic and Altaic, and >yet you
>boldly erect a Proto-Steppe reconstruction, complete with the >decads.

Yes. First, I may seem "bold" to you only because you don't quite understand
yet the methodology I use. My view is that it is better to reconstruct
_anything_ rather than nothing at all. In my manner of thinking,
extrapolation and evolution are key elements to advancing knowledge in
theoretical studies and what I reconstruct is not meant to "sell" anything
but rather is a proposal fully replaceable with a better theory in the
future, made possibly by people like you. Have one?

Second, it is evident why certain subfamilies of Uralic and Altaic should
retain the numeral system and others not. In the Uralic family, Samoyedic is
the most ancient branch and yet, it resides so far up north that one can see
very clearly why a standard numeral system would not be retained here. In
this instance, only the other branch, FinnoUgric, can be expected to offer
data about the ancient numerical system. In the Altaic family, the speakers
of these languages have been historically as well as pre-historically
nomadic. Their lifestyle doesn't lend itself either to retaining standard
numerical systems, so we see bits and clips of the original system across
this family.

However, the stems for "three" and "four" are too clearly related and
phonetically regular in sound correspondance between Mongolic and
FinnoUgrian to be mere coincidence:

Mongolic FinnoUgrian (as found in EncBritt)
*gur- *kolme
*du"r- *nelja"

Two things to help put this pair further into perspective:
----------------------------------------------------------
It must be noted that Uralic avoided Steppe's consonant-final stems (cf.
*nil "four") by added a terminating vowel - This is the reason for the
overwhelming number of disyllabic terms (also *wete "water" < *wit:). Thus
the word *nelja" was originally athematic like many other Uralic words.
Palatalisation of *l naturally occured from the neighbouring vowel *i (>
*e).

It must also be noted that Altaic II has a strange sound change that turns
many initial nasals of Altaic I into stop counterparts. Thus *m becomes *b
and *n becomes *d, further Altaic I *r- helped to fill back in some of the
initial nasal gap by becoming Altaic II *n- (as in the word for "seven").
Since there is no initial velar nasal in Steppe, there is no velar
counterpart to this phenomenon. (cf. Turkish ben "I" < Altaic I *mun)

Finally, *-mu is often attached to monosyllabic numerals between three and
ten (hence *kolme).

Now, in the end, we must either view these two numerals as borrowings or as
inheirited words from an earlier stage. As far as I know, Mongolic and
FinnoUgric were not in contact with each other, hence they are likelier
inheirited terms. Further, *du"r- "four" is fully attested across Altaic.

> (2) How did the original word for 'five' come to mean 'four' in
> >Proto-Indo-Tyrrhenian and 'six' in Uralic? BTW, my Tyrrhenian is not
> >very strong, but Etruscan huth is usually interpreted as 'six' rather
> >than 'four' (pace McCallister's online glossary) on combinatorial
> >evidence (for example, the arrangement of numerals on Etruscan dice
> >implies that HUTH + THU = 7).

Hmm, well the contraversy lives on. The word /huth/ is still not agreed on,
AFAIK. Some say "four" and some say "six". I go with "four" since it matches
IE's numeral well and because, this way, six is /s'a/ which can be explained
as a borrowed numeral from a Semitoid language. Other numerals prove clearly
that there was a Semitoid influence on the Etruscan number system (semph
"seven", s'ar "ten").

Now all this would mean that IndoTyr *kWetWe shifted the meaning of this
word from "five" to "four". If so, why? Well, I have an idea... but I can't
be sure whether it's a reasonable one or not yet.

Five to four??
--------------
You see, if IE *penkWe (through which we also get "fist" from) is an ancient
word derived from an IndoTyr **penkWe as it regularly should, we might not
think anything of it other than a possible differing term for "five".

The question is, if this word existed in IT, how was it formed? Inheirited,
borrowed, compound word? Curiously, SinoTibetan has a similar word for
"five", *bNa (Cantonese ng, Mandarin wu), but surely IT and ST were not side
by side!

Why mention this? Well, I'm coming to the interesting conclusion that
IndoTyrrhenian and _Pre-NWC_ were side-by-side. Unfortunately, the term
reconstructed in the later NWC stage would seem to be *sxW@, which is not
what we imaginative comparative linguists are hoping for.

Of course, I'm not one to give up. Upon an unrelated closer examination
(fueled in part by Guillaume), I've become convinced that NWC, Na-Dene and
ST share relatively close commonalities and must have derived from a parent
language (SinoDene) located in C Asia prior to the arrival of ProtoSteppe.
As IndoTyrrhenian spread out from the Steppe linguistic area, Pre-NWC was
ahead of it to the west, after having spread out from its own SinoDene
language area just before Steppe's arrival.

There are common numerals to be found between NWC and SinoTibetan...
including *p@...@ "five" (N=velar nasal). The velar nasal, which was probably
unvoiced, regularly becomes a laryngeal in NWC. The curious initial *s- that
Starostin has difficulty explaining in NWC *sxW@ is the result of
assimilation via *s-initial "three" which spread like wildfire to "four",
then "five". This occured in a similar fashion in SinoT where we find a
spread of *b- from the numeral *bNa to others like "four", "nine" or "ten"
in some ST languages.

Along with the seemingly strong areal influence on phonology between NWC and
IndoTyrrhenian, a Pre-NWC loanword would be just what I need to help clinch
things. And the phonetics work so well too :)

Pre-NWC *p@...@ -------> IndoTyr *penkWe
(Note: IndoTyr *e is a _schwa_)

> (3) Why are your sound correspondences so capricious? E.g., ITyrrh.
> >*kWetWe > IE *kWetwores, but ITyrrh. *kWel > IE *tr�jes (via
>*tWer�i->ec). Why doesn't your *t: in *kit:u/*kWut:u show up as
>(traditional) >IE *d?

Okay. This is a bugger to explain :)

About *kit:u/*kWut:u
--------------------

First, IndoTyr *kWetWe would normally have become IE **kWetu. Evidently,
there is a plural suffix secondarily attached here. Thus we should expect
MidIE *kWetwer-ec... but how do we explain the intrusional *-r-?

The likeliest idea to come to mind so far is that this form *kWetwer-ec is a
_LATE_ MidIE term which was spoken _after_ the *-n > *-r sound change. Thus
we might suspect Old IE/Early MidIE *kWetWen(-ec). Now, for some funny
reason, Old IE likes to add *-n to many monosyllabic roots... like for
instance, the word for "water" which would originally have been Early OIE
*we:t: but certainly ended up Late OIE *we:t:en, to be later changed to Late
MIE *we:t:er. The inanimate *-n is mentioned in my Steppe grammar where it
originally was a singular topic marker found also attached to pronouns.
Perhaps it later acquired a collective singular meaning (and hence an
inanimate gender).

In Etruscan, everything appears kosher. The final vowel is not present but
we can't be certain whether Etruscan omitted it or not. Etruscan scribes
seem to like to omit unstressed vowels alot.

Finally, why it is that *kWut:u becomes IndoTyr *kWetWe instead of *kWet:We
is up for grabs. Suggestions? Is it some sort of typological restriction
preventing voiceless aspirate stops (*kW) from coexisting with tense ones
(*t:)?...


*kWul- "three"
--------------
This numeral is giving me some severe headaches in IndoTyr. It seems
possible that it may be attested in Etruscan as /ci/ but as for IE, we seem
to have this odd numeral *treies. Again, as with "four", the word is given a
superfluous plural suffix. Thus, we might logically project back a MidIE
*t(W)erei-ec.

This isn't what we should expect. A form *kWul- should regularly become
IndoTyr *kWel-. On the other hand, let's remind ourselves that numbers
aren't alone but part of a system, surrounded by other numbers, and
sometimes, phonetics can get mingled amongst each other.

The IndoTyr word for "two" appears to be *t:Wei (Steppe *t:ui). If we were
to suppose for a moment that *t:Wei had influenced **kWel to produce a
poetic **kWel-ei (and mind you, Steppe *-i is itself a plural suffix), we
get farely close already to the eventual form of MidIE *t(W)erei. Perhaps,
further dental-initial assimilation could have brought about a pleasant
numeric triplet in Old IE *[t:u / t:Wei / tWelei].

If so, the addition of a plural suffix *-ec by the time of the penultimate
accent changes would first produce *tWeleiec with accent on the second
syllable, followed by expected vowel reduction to produce an awkward
**tleies. From here, it's easy to explain why *-l- should become *-r-.

It's a thought anyways. Other suggestions?

> (4) What is the contrast (if any) between *kit:u and *kWut:u?

Variants of the same numeral, that's all. The form *kit:u is attested in
Altaic while *kWut:u is shared between IndoTyr and Uralic. I'm still not
sure whether Nostratic has *hut?u or *hit?u (Coptic fto "four", Hausa huD'u)
or whether I'm imagining things :)

> (5) What's your evidence for *ri, *ra, and *munri (the most >mysterious
>of them all, for which I can't see a shred of support in >the attested
>"Steppe" languages [or perhaps Nivkh is meant to be your >sole witness, but
>why?])?

So many questions! Wonderful. This mystery has an exotic explanation. First,
these three numerals oddly seem to go back tens of thousands of years (along
with many of the lower ones too). I know that this sounds counter-intuitive
given that many think these numerals shouldn't be recoverable from such an
ancient time.

I reconstruct Nostratic *rut, *ra(-mu) and *munri as well as the parent
DeneCaucasian numerals *rutL, *sul-rit "three from" and *mn-rit "two from".
You're probably frightened by these reconstructions right now so I'll step
back a moment and just talk about Steppe.

The evidence in Steppe is mostly from Altaic (as what Starostin presents for
the same numeral online where, after making some really stooopid errors, he
reconstructs Altaic *n'u "6" and *nad[i] "7"):

Altaic I *ru
Altaic II *nu "six"
-------------------
Old Japanese mu-
Tungusic *n'uNgun
Negidal: n�uNun
Manzhu: niNgun
Nanai: n�uNgu~
Ulcha: n�uNgu(n)
Orok: nuNgu(n)
Zhurzhen: niuN-z^u (641)
Even: n�uN'n
Evenki: n�uNun

(/N/ = velar nasal)

Altaic I *ra
Altaic II *na "seven"
---------------------
Old Japanese *nana-
Tungusic *nadan
Negidal: nadan
Manzhu: nadan
Nanai: nada~
Ulcha: nada(n)
Orok: nada(n)
Zhurzhen: nadan
Even: nad'n
Evenki: nadan
Mongolic *dal- (?)

Starostin attempts to include Turkic in his reconstruction for "seven" - I
couldn't possibly describe how utterly stupid that suggestion is, so I
won't.

The word for "eight" has no attestion at all :( and is just a suggestion
based on some other Nostratic languages which seem to suggest that DC *mnrit
was inheirited into Nostratic as *munri. This requires LOTS of explaining in
itself so I'll respond to that in a seperate posting if anyone asks me.

I hope I answered everything so far. Now tell me, what do you have on Nivkh,
perchance?


- gLeN

________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com