PIE and Uralic

[Ante Aikio]

On Thu, 27 Jan 2000 X99Lynx at aol.com wrote:

> Now I have somewhat earlier dates here for the existence of proto-Uralic
> (Dolukhanov 1996) -- 10,000 - 7,000BP -- citing (Hajdu 1975).  I also have it
> extending all the way to the Black Sea and eastern Caucasus to the Urals.  Do
> you have newer data?

I didn't have a chance to get a copy of Hajdu 1975 now, so I'll
possibly get back on this (as well as some other comments/questions
in yor mail) later in more detail. However, the area from Volga to Urals
is the most common view these days, at least among Finnish
researchers. I'm interested in finding out on what basis Hajdu assumes
that the PU homeland reached south to the Black Sea and Caucasus - I have
seen no evidence that would support such a view. As for dates, 7000BP is
a sensible estimate and quite compatible with my "4000BC or earlier",
while 10000BP seems -very- early.

(I wrote:)
> <<There are also IE loan words with a narrower distribution within Uralic,
> which actually seem to derive from "early daughter languages" which still had
> laryngals (see my previous mail to the list). The example words I put forward
> in my first mail must be older loans, since they go back to proto-U.>>

(Steve Long replied:)
> How you identify the difference between PIE and the IE daughters in this
> region - I think - would be interesting.  The specific attributes of
> daughters in that region might be of value to IEists, especially if they
> retain laryngals.  I am not sure what IE daughter language is supposed to
> have been spoken in the Ukraine at such an early time.

("Ukraine" seems to be a misunderstanding - the loans I originally
referred to appear only in the Baltic Sea / Scandinavia area, i.e. Finnic
and Saamic, and were probably adopted from the Indo-European battle axe
culture in that region around 3000-2000BC.)

They can be identified with phonological and distributional criteria. The
earliest loans show PU *k and *x as substituents of IE laryngals, and they
have wide distribution in Uralic. The newer loans show the PU retroflex
sibilant *S as a substituent of PIE *H, and have a more restricted
distribution. The introduction of a new sound substitution pattern (IE *H
> U *S instead of earlier IE *H > U *x/*k) seems to be connected with the
fact that PU *x disappeared as an independent phoneme in all U language
branches, and these developments probably took place at quite an early
date. Some of the later loans also show other post-PU characteristics
(e.g., labial vowels in non-initial syllables, see the examples below).

The following serve as examples of later loans. All appear -only- in
Finnic, except number 2 which also has cognates in Saamic and Mari. All
etymologies derive (once again) from Jorma Koivulehto. (PU *S > Finnish h
is a regular development).

1) Finnish rehto 'row (of constructions of one type or other)' (< *reSto)
   < PIE / Pre-Germanic *rH-tó- (> Germ. *radha- 'row etc.')

2) Finn. lehti 'leaf' (< *leSti)
   < PIE / Pre-Germ. *bhlH-tó- (> Germ. *bladha- id.)

3) Finn. rohto 'herb, medical plant' (< *roSto)
   < PIE / Pre-Germ. *ghróH-to- (> Germ. *gro:tha- 'herb, plant etc.')

4) Finn. puhdas 'clean' (< *puStas)
   < PIE *puH-to-s (> Old Indic pu:táh id.)

5) Finn. tahdas 'dough' (< *taStas)
   < PIE *taH-i-s-to-s (> Cymric toes, Russian testo id.)
   (Pre-Finnic *taStas instead of *taSistas because the latter was
   excluded on phonotactic grounds.)

We have thus two chronologically distinct layers of IE loans in Uralic
that show different laryngal reflexes. With a bit more speculative
approach on the data, it seems possible to add a third layer
(pre-Uralic). The loans that on distributional criteria seem to be the
oldest ones show two kinds of substitutives fro IE *H in word-initial and
postconsonantal position: PU *k on the hand  and zero on the other
(e.g. PIE *Hokw- 'see' > PU *koki- and PIE *kwelH- > PU *kulki-, but PIE
*Hag´- 'drive' > PU *aja- and PIE *pelH- 'frightening  etc.' > PU
*peli-). It is hardly a coincidence that PU *x can be reconstructed only
for preconsonantal and intervocalic position, but it seems to have been
phonotactically excluded from postconsonatal position and word
initium. Thus, it seems possible that these are pre-Uralic loans where PIE
*H actually was substituted with Pre-U *x, and subsequently a sound change
*x > zero in the aforementioned postions took place in PU. If this is the
case, we have at least the following Pre-Uralic loans:

PU *peli- 'fear' < Pre-U ?*pelxi- < PIE *pelH-
PU *puna- 'plait' < Pre-U ?*punxa- < PIE (zero grade) *pnH-
PU *pura- 'drill' < Pre-U ?*purxa- < PIE (zero g.) *bhrH-
PU *aja- 'drive' < Pre-U ?*xaja- < PIE *Hag´-
PU *käliw- 'brother/sister-in-law' (-w- is a suffix) < Pre-U ?*kälxiw-
  < PIE *ghlHi-

When the change *x > zero had already happened, PU had to substitute PIE
*H in these positions otherwise and thus we have e.g. PIE *kwelH- > PU
*kulki- (and not ?*kulxi- > *kuli-).

The above reconstructions are of course tentative, and no doubt other
possible explanations for the correspondence IE *H = U zero exist,
too. However, it seems at least possible that there were not only PIE
loans in PU, but also (Pre-)IE loans in Pre-U.

(Steve Long wrote:)
> The vocabulary you mention is both quite basic (it would be at 4000+BC) and I
> believe you said quite extensive - I forget, hundreds of words?  Consider
> that the retained presence of hundreds of basic cognates from PIE in a
> language that was attested when? - well, for 6000+ years would probably
> entitle Uralic to a special place on the Swadesh chart - possibly better than
> some IE languages.

There are not hundreds of them. The reconstructed PU vocabulary amounts
to only approximately 300-350 items, and among these there there are 35-50
PIE loans (the number depending on how critically one views the
etymologies, 35 reflecting a very critical attitude).

(I wrote:)
> <<As for the assumption of "Urverwandtschaft" between U and IE, we are
> obviously dealing with loaning here. The loan explanation has more
> explanatory power.... This is not the case if one assumes common genetic
> origin: it is not possible to demonstrate regular sound correspondences
> between the items.>>

(Steve Long replied:)
> Okay, so please follow me here.  If p-U and PIE are both at their place of
> origin "4000BC or before" one just north or south of the other - one would
> think they would have been in some way related or there would have been
> contact between the parents of these two unrelated languages - suggesting
> that perhaps one was intrusive.  (Can a language be intrusive in its own
> homeland?)  Since you find these PIE elements as definite loans happening at
> a definite time - might that not suggest that PIE entered the area - or the
> loans would have happened earlier - in the proto-proto-period?

I can only say that on linguistic grounds, there is very little we can say
about Pre-U and Pre-IE. Consequently, it is hard to identify Pre-IE loan
words in Pre-U since we cannot reconstruct these languages. This creates a
tempting, but possibly false picture that the IE and U groups came into
contact when they still spoke their relatively uniform
proto-languages. But we can't exclude the possibility that there was
contact between the predecessors of the proto-languages even in very
remote past - we simply cannot say, because without reconstructed
pre-proto-languages the loan words may have chaged beyond our
recognition. But even if we knew that there was contact between Pre-U and
Pre-IE, this of course would not make it any more likely that the
languages are ultimately related.

(Steve Long wrote:)
> As Jens mentioned in his elegantly fair message: <<They may of course also be
> older than PIE; since it is apparently only lexical material, there is little
> to tell us anything about the time depth of the donor forms.>>

Yes; what I presented above only strengthens the Pre-IE > Pre-U loaning
hypothesis. But then again, this implies that speakers of Pre-IE and Pre-U
were neighbors, and subsequently also PU and PIE were.

  Ante Aikio