Hallo!

On Mon, 26 Jan 2009 08:30:40 +0000, R A Brown wrote:

> Just a brief reply to David's post - Jörg has already responded with 
> many of the points I would have made    :)
> 
> Alex Fink wrote:
> > On Sun, 25 Jan 2009 16:54:32 +0000, David McCann 
> > <[log in to unmask]> 
> > wrote:
> > 
> >> Etruscan is spoken in Europe. All other European languages are either
> >> Nostratic or Dene-Caucasian (that should be good for a few more
> >> comments :-). 
> 
> The main comment is obviously that there is no widely accepted proof of 
> that statement. As Jörg pointed out in an earlier post, the Nostratic 
> theorists are in (at least) two camps, only one of which can be right - 
> and both may well be wrong.

Indeed.  In a court of law, if two witnesses contradict each other,
and there is no reason to consider one of them more reliable than
the other, both will be doubted.

>       As for Dene-Caucasian .... from what I have  
> seen of it, there is more evidence for the existence of Santa Claus  ;)

Well put.  I consider Dene-Caucasian much less likely still than
Nostratic - what do the Basques have to do with the Chinese, or
with the Navajos?  Any common ancestor of those languages would
have to have been spoken tens of thousands of years ago.

> Our knowledge of neolithic Europe before the spread of IE languages is 
> virtually non-existent. Personally, I would be surprised if were 
> dominated by just two super-families.

I'd be so, too.  We don't know how many waves of immigration
there were in Europe before Indo-European and Uralic bulldozed
it all away.

>       Is the non-IE lexicon of Insular  
> Celtic to be attributed to Nostratic or Dene-Caucasian??

Ah, the mythological non-IE lexicon of Insular Celtic.  I still
cannot find a list anywhere, yet the etymologists must have come
up with a number of words which they don't know where they came
from.  May have to do with the matter that most Indo-Europeanists
and Celticists react on the words "pre-Celtic substratum" in a
similar way as astronomers react on the word "horoscope".
The Semitic substratum nonsense has left an intellectual minefield,
it seems, and nobody wants to touch the matter at all anymore.
 
> IIRC in an earlier mail Jörg has speculated that there might have been 
> at least 25 different language groups in neolithic Europe. We just don't 
> know.

Indeed.  I arrived at the "25 families" guess by taking the approximate
number of indigenous language families in the USA & Canada and halving
it (because Europe is about half the size).  However, the linguistic
diversity patterns may be different, and the actual number may have
been lower - or even higher!  It of course also depends on what one
defines as a "language family".

The proponents of a highly homogenous linguistic landscape of pre-IE
Europe argue with parsimony and Ockham's Razor, but the uniformitarian
principle speaks against that.  They misapply Ockham's Razor, as they
usually fail to come up with a convincing mechanism which accounts for
such linguistic homogenity in a region that had had human inhabitants
without stately organization for almost 30,000 years.  (Vennemann
"solves" this problem by moving the first human colonization of
northern Europe to a mere 10,000 years ago - which is nonsense.)

> >> The odds are that Etruscan is going to be one or the
> >> other. The one thing we certainly know is that it won't be Khoisan.
> 
> I agree that Etruscan won't be Khosian - but that gets us nowhere. To 
> speak of the odds that Etruscan belongs to one of Nostratic or 
> Dene-Caucasian is meaningless while the odds that either of these two 
> super-families existence is IMO not great.

Right.

> [snip]
> 
> I won't comment further on Jörg's reply, which I am in agreement with, 
> except this:
> 
> >> >       and ancient population movements can be 
> >> > reconstructed by molecular biology.
> > 
> > Be careful.  The idea of grafting language family tree on genetic
> > trees, as done by Cavalli-Sforza and his followers, is generally
> > met with suspicion, as genetic relationship does not necessarily
> > imply linguistic relationship, and vice versa.
> 
> Yes indeed - *be careful*. Genetic relationship does *not* necessarily 
> imply linguistic relationship, and vice versa.

Indeed.  Where is your disagreement with me?

>       An obvious example of the  
> latter is Saami & Suomi (Finnish). No one doubts the two languages are 
> fairly closely related but they spoken by two rather different ethnic 
> groups.

Yes.  Also, archaeological cultural horizons cannot be equated
with linguistic filiations.  A good example is the Corded Ware
culture, which some people think was Indo-European speaking.
However, the Corded Ware horizon extends from the Rhine to the
Volga, and includes what probably was the homeland of Proto-
Finno-Ugric (at least according to the opinion of most Uralicists).
So it cannot have been IE-speaking in its entirely, but also not
FU in its entirety (only crackpots will think that there ever was
FU in Germany).  After all, modern Europe would form, from the
viewpoint of future archaeologists, a single cultural horizon, and
yet only the inscriptions on the objects will reveal that there
were dozens of different languages in the area.

On Mon, 26 Jan 2009 10:37:53 -0500, John Vertical wrote:

> They did get Dene-Yeniseian (Deniseian?) together tho, didn't they?

Heard of that, yes, and apparently the evidence is not all that
bad.  But I think it is still controversial, even if Wikipedia
(which is usually rather conservative about such matters) accepts it.

> But anything more will probably be longer in coming. :)

Indeed.

> Nobody requires the continent to have ever been dominated by two
> proto-languages. They could have been no bigger players than anyone else
> back in the day and just the only main lineages to survive (accepting for
> the while that they're even valid).

Indeed.  As I said above, those who propose as few families as possible
misapply Ockham's Razor.

> Population genetics is probably valid for some very general conclusions
> about the interrelations of the world's languages.

Not really.

>       If we go to sufficient
> time-depth, it should be a safe bet to assume eg. that all current languages
> of Eurasia are closer related to each other than to Khoisan, with the
> possible exception of Afro-Asiatic and other things in the Middle East. And
> something similar should be possible for most of "Amerind". (Not necessarily
> every one tho; I wouldn't be surprized either if it eventually turns out
> there exists let's say, for a random example, Ainu-Haida-Aymara along the
> Pacific coast.)

It is usually reasonable to assume that the nearest relatives of
language X are found in its neighbourhood.  For instance, it is
more likely that the nearest relative of Indo-European is Uralic,
rather than, e.g., Wakashan.  But this assumption can easily fail,
because some groups of people apparently engaged in large-scale
migrations, and languages evidently spread even faster.  The real
reason to consider IE and Uralic to be related to each other would
be a sufficient quantity of regular sound correspondences between
lexemes and grammatical morphemes.  If such correspondences were
demonstrated to exist between IE and Wakashan, one would be forced
to accept that they were related, despite the vast distance by which
they are separated.

> But *within* a continent? Most dates I've seen for macro-families are
> staggeringly young, only a handful of millennia beyond the oldest
> reconstructed proto-languages.

Indeed.

>       Should we really expect major expansions such
> as IE, or perhaps as a more accurate comparision (since this would almost
> certainly be a pre-agricultural period) Algic to have wiped everything else
> away in just a few thousand years? I don't think so.

A few days ago, I finished reading _The Rise and Fall of Languages_
by R. M. W. Dixon, in which he agues that massive expansions such as
IE and Austronesian are the exception rather than the rule.  Basically,
his assumption is this:

There are times of linguistic equilibrium, when many languages co-exist
with each other on more or less equal footing.  In such times, language
splits and displacement of languages occasionally occur, but are not
common, and linguistic evolution is convergent, resulting in language
areas (groups of languages typologically similar to each other, and
often also containing common loanwords, but not necessarily sharing
a common ancestor).

Then there are - much less often - times of punctuation, when one or a
few languages happen to get into a situation where they can massively
expand, e. g. by elite dominance or some economic "edge" to them (such
as farmers vs. hunter-gatherers).  The result is divergent linguistic
evolution, wherein a single language breaks up into many daughter
languages and thus gives birth to a language family.

If we apply this "punctuated equilibrium" model to Europe, we get these
punctuation events:

1. The first peopling of Europe by _Homo sapiens_, ca. 40,000 to 30,000
   years ago.

2. The end of the last ice age, when northern Europe became habitable
   and central Europe became much more hospitable than before, ca.
   12,000 years ago.  This probably entailed massive northward
   migrations of humans, including those who set first foot on
   Scandinavia.

3. The neolithic revolution, ca. 9000 to 6000 years ago.  This is
   generally assumed to have been demic (i. e., involving massive
   migrations) north of the Alps, but cultural (without massive
   migrations) in the Mediterranean area.  One would expect a single
   language family to spread north of the Alps, but only a slight
   reduction of linguistic diversity in the Mediterranean (and indeed
   we find quite a few apparently unrelated language groups there
   in ancient times).

4. The spread of Indo-European and Uralic, ca. 4000-500 BC.

5. The Greek colonization of Italy and Asia Minor, ca. 1000-500 BC.

6. The Roman Empire, ca. 300 BC - 400 AD.

7. The Völkerwanderung, ca. 400-900 AD.

8. The rise of modern transportation, communication and the nation-state
   ideology, ca. 1800-2000 AD.

There may have been more punctuations which are no longer recoverable.
Between these punctuations, there would have been phases of linguistic
equilibrium when few languages split and few were displaced.  The oldest
linguistic relationships in Europe would date back to the first peopling
of Europe and thus so old that they would be practically irrecoverable.

>       There's probably
> something to eg. Nostratic that is in the reach of comparativ linguistics,
> but beyond that, things like initial consonants associated with pronouns can
> be extremely stable (things like /m/ especially so); a "macro-Nostratic"
> could easily be simultaneously a valid family (in that the last common
> ancestor would not be the last common ancestor of most modern-day language
> families) and some 40-50,000 years old.

Possible, but probably unrecoverable.

> The alternativ would be that the apparent similarities are just apparent,
> and the closest relativ of IE is something like Sumerian-Austronesian.

One would expect the closest relatives of a language family to be
found rather nearby, which speaks for an IE-Uralic relationship,
but as I said before, regular sound correspondences are what counts.


>       (I am 
> reasonably sure large-scale groupings must at least exist, even if I don't
> expect them necessarily to be resolvable without a time machine.)

Yes.

>       But in the 
> meantime, the current proposals are the best clue we have to go on.

I would not say that Nostratic or Dene-Caucasian are impossible, but
the evidence is simply insufficient.  I indeed feel that IE, Uralic
and probably a few others are related - but I cannot prove it.

... brought to you by the Weeping Elf