Accepting fewer etymologies

[Richard M. Alderson III]

On 21 Sep 1999, Steve Long wrote:

>But doesn't the elimination of some potentially critical sound rule put into
>question the validity of any "cognate" you find?  If you don't have Verner's
>Rule, doesn't Grimm's Law look quite faulty and therefore unusable?  Doesn't
>that lead you to finding something less ambitious than those "laws" to decribe
>your system?  And can't that possibly lead you to using your narrowly
>"derived" sound laws to find a list of "inherently plausible" cognates that
>were never really there?

Perhaps, yes, no, and no.

You see, if a set of correspondences shows a (possible) pattern of exceptions,
we do not, as you seem to think, throw out the set, but rather, seek to find
the explanation.  But it is only by looking at a subset of the data that the
first rule can be hypothesized for testing, to learn that a second rule may be
necessary.

Your concern over "plausible semantics" is laudable, but misguided:  What is
meant is that first we see what can be determined by looking only at near or
exact matches in the semantics of otherwise phonetically plausible potential
cognates.  By restricting ourselves only to such, we have a much better chance
of seeing *any* pattern at all.

It is only after we have a (relatively) secure set of correspondences, based on
what we choose to call "plausible semantics", that we can legitimately look for
less immediately plausible cognate sets by noting phonetically plausible
matches.

								Rich Alderson