Re: languages of pre-I.E. Europe and onwards

Hallo!

On Fri, 23 Jan 2009 08:23:17 +0000, R A Brown wrote:

> > ["evidence" for IE-Etruscan relationship]
>
> Personally, I remain unconvinced. I am very much in agreement with Jörg
> that one needs at least about 100 cognate sets displaying _regular_
> sound correspondences.
Good matches between inflectional paradigms can, however, make good
for a dearth of matching lexemes, but only if

1. the match involves entire *paradigms* (or substantial subsets
   of the attested paradigms) rather than a few individual
   morphemes;
2. there are corresponding allomorphies in both of the languages
   compared;
3. the sound correspondences are regular.

This was, indeed, the way the Indo-European family was proven in
the early 19th century.  However, it was soon bolstered by many
hundred lexical cognates.

Also, it is difficult to reconstruct regular sound correspondences
from paradigmatic morphological correspondences; you probably always
needs at least a few dozen lexical cognates in addition to the
matching morphology to support the assumptions drawn from the
morphology.

On the other hand, even hundreds of seemingly matching lexemes may
give false results, if one language has massively borrowed from
another within a generation or two: in such a case the borrowed words
will all have the same phonological history and thus appear to match.
The best way to weed out such errors is to apply the apparent regular
correspondences to morphology - if the morphologies refuse to match,
you are certainly dealing with massive borrowing.

... brought to you by the Weeping Elf

Reply