Re: Exceptional rules

From: Richard Wordingham
Message: 50916
Date: 2007-12-15

--- In cybalist@yahoogroups.com, "Grzegorz Jagodzinski"
<grzegorj2000@...> wrote:

> From: Richard Wordingham

> > But we must remember that admitting 'exceptional' rules reduces the
> > quality of an explanation, as does introducing ad hoc rules.

> So, you believe that a false explanation can have better quality
than a true
> one?

Most etymologies have a risk of being wrong, ranging from negligible
to high.

> <snip>For example, let the
> question be if the Polish "pokrzywa" (nettle) is related to Czech
kopr^iva.

The fact remains that it is more likely that Polish "pokrzywa" is
unrelated to Czech "kopr^iva" than that Polish "koprzywa" is unrelated
to Czech "kopr^iva". At least in this case the rest of the word makes
the relationship unlikely to be coincidence, and the recorded history
also helps. The recorded history also makes quantifying the level of
confidence difficult.

> OK, and what if we have not other sources? Cretan adeuphos, Modern Greek
> aderphós, Attic adelphós all mean "brother". We cannot formulate a
strict
> rule for l > u in Cretan or for l > r in Modern Greek. Should we believe
> that all these three forms are similar in form and meaning only by
chance?

The Cretan and standard Modern Greek forms actually provide evidence
for one another being cognate to the Attic form - unless Cretan loses
/r/ before consonants, as some northern Greek dialects do. But do you
dispute that we would be even more confident of the relationship if
the forms were both */Delfós/? (I do believe your 'ph' means /f/ in
the modern forms, and 'd' means 'dh', i.e. the voiced fricative. I'm
not sure what to make of the reported Cretan 'eu' - /aDeffos/?)
Actually, doesn't the initial /a/ of the modern forms argue for at
least contamination with a reborrowing from Attic?

We can see more extreme forms of this in Afroasiatic etymology - see
Militarev's paper on consonant additions, which I referenced for
Patrick and Arnaud's spat on Semitic roots. While we may see
derivation in that case, the effect is the same - increased
uncertainty as to cognacy.

The worst case is probably Nostratic, where a good many of the alleged
correspondences have to be wrong even if the Nostratic hypothesis is
ocrrect - but which is probably an answer that can at best be answered
with a confidence level.

Richard.