Re[2]: [tied] Re: Germanic Scythians?

At 3:08:20 PM on Tuesday, March 25, 2003, P&G wrote:

>> Producing the same change in several branches
>> independently? It doesn't matter much what triggered the
>> change; as long as it's independent in the several
>> branches, elementary probability theory will tell you
>> that it's unlikely.

> And yet elementary linguistics tells you that in well
> attested cases, it has happened. Greek, I-I and Italic all
> seem to have maintained laryngeals at a time when they
> were separate, and all lost them in parallel (though not
> identical) ways.

Maintenance of an existing feature is a different issue.
Whether the loss in parallel ways is surprising depends on
the likelihood of the loss in the first place and the extent
and nature of the parallelism. To exaggerate considerably
in the interests of making the point clear, .9^3 is still
.729, but .1^3 is only .001. Obviously there's a continuum
of likelihoods in different cases.

> Grassman's Law appears to be independent in Greek and Skt.
> and so on and so on....

But this is only two branches, and it isn't on its face an
unlikely development. Certainly the situation here is very
different in degree from the one that Torsten's proposal
would require.

Brian