No. This means that if a word is attested
*only* in one subgroup of a larger genetic group, then (if no external
evidence exists) we cannot legitimately reconstruct it for that larger taxon.
For example, if a word occurs *only* in English, Dutch and Frisian, the most
parsimoniuos solution is not to reconstruct it for Proto-Germanic and
regard it as an Ingveonic innovation. However, if there is, say, a Gothic
cognate, or if evident IE cognates exist outside Germanic, then a Proto-Germanic
reconstruction is legal even if North Germanic doesn't have the word (in such
cases we assume that it has been lost in the group that lacks it). We are
interested in optimal working hypotheses (this is what science is
about), not in the ultimate truth of the universe. Of course it may so
happen that our methodologically optimal assumption is wrong, but how can we
know that if there is no evidence? I've no idea what kind of Occam's Razor you
use, but I'm surprised at your surprised tone.
Piotr
----- Original Message -----
Sent: Thursday, February 08, 2001 3:00 PM
Subject: [tied] Re: Oesysla/Eysysla etc.
Do we? That's not the Occam's razor I know. I suppose
that means that if a word in a West Germanic language doesn't exist in Old
Norse, then it's not in PGmc.
either?