IE "Urheimat" and evidence from Uralic linguistics

[Robert Whiting]

On Sat, 5 Feb 2000, Hans Holm wrote:

> RW>Either two (or more) languages are related or they are not.  This is
> RW>the basic hypothesis of historical linguistics.
> .. is it? Nice to learn indeed.

Well, better late than never ...

>  You will find defenders for or against relationships between any two
> languages. You can argue about the degrees and ways, about significance or
> chance resemblances. But you can't prove unrelatedness. We had that
> discussion already. Perhaps You should read Anttila 89:320..

Indeed, one cannot prove unrelatedness, but without a historical record,
one can not prove relatedness either.  One can only amass enough evidence
to show that it is inconceiveable that certain languages are not related.
But if you have actually read Anttila 89:320, you will know that he says
there "'Related' in linguistics means 'relatable'."  So when I say that
languages are either related or they are not, that means that languages
can either be shown to be related ('are relatable') or they cannot.  This
simply means that if languages cannot be shown to be related, they must be
considered unrelated.  Now it is not impossible that all languages are
ultimately related, which, if true, would mean that there is no such thing
as unrelated languages and the "or not" becomes meaningless.  But since
there is no way, with present methodology, to prove this pro or con, one
cannot deny the possibility of unrelated languages.  So when a linguist
says that certain languages are unrelated this means only that there is no
(or insufficient) evidence to show relatedness.

And if you want to use Anttila 89 as a source then you should read the
heading of the chapter that you are quoting from, which says in part:
"... typological classification is never perfect or absolute.  This
contrasts with the absolute nature of genealogical classification."
(ibid. 310)

So there may be distantly related language, but there are no slightly
related languages.  Languages are either related or they are not.  Perhaps
you should read Anttila 89:300.  Start with the part where he says:
"'Related' is a technical term, exactly like the equivalent 'cognate',
meaning that the items were once identical."  This is the criterion of
genetic relatedness in historical linguistics.  If you find this
impenetrable, the same concept is explained is slightly different
terms in Anttila 89:318:  "Those languages that represent outcomes of
one and the same proto-language are grouped into a family."  This means
that related languages (those that form a family) are variant outcomes
of a single language.

> RW>This is totally irrelevant

> You mean /you/ do not see the point.

Yes, that is what irrelevant means.  Perhaps someone who does see the
point can explain it to me.

> You might have missed that my ancestor example referred not to languages
> but to speakers.

Which is precisely what makes it irrelevant.  The genetic relatedness
(or lack of it) of the speakers of a language has no bearing on the
genetic relatedness of languages.  The ability to learn and use language
(the language acquisition device if you prefer) is genetic and inherited,
but there is no genetic disposition to learn any particular language.
Any normal human child, regardless of its genetic background, placed in
any linguistic environment will learn the language(s) of that environment.

> RW>Languages do not need a mommy language
> .. really?

> RW>But two genetically related languages have only one common ancestor,

> .. let me take an Indo-European example, e.g. Italian and French.
> Superficially seen they only have one ancestor: Latin. But this is only
> the dominant ancestor.
> If we look at e.g. French it has a lot of strata which can be called its
> fathers: The languages spoken by the pre-celtic cultures, the Celts
> themselves; later all the Germanic invaders not perfectly succeeding in
> learning the current states of that language. Anttila (and others) cite
> languages where you cannot even decide whether they should be named after
> their mother or their father.

Ah, well, it is a wise child that knows his own father.

Substratum, adstratum, and superstratum languages can certainly influence
the outcome of a language's development, but that does not necessarily
make them genetically related.  Latin had considerable influence on
English, but that does not mean that English is descended from
Latin; Latin had considerable influence on Basque, but that does not
mean that Basque is related to Latin.  Sumerian had considerable influence
on Akkadian, but that does not mean that Akkadian is related to Sumerian.
My professors in grad school had considerable influence on my later life,
but that does not mean that they are related to me.

Convergence (and even mixture) do not make for genetic relationships.

Bob Whiting
whiting at cc.helsinki.fi