On Thu, 9 Nov 2000 23:19:27 -0800, Marcus Smith <[log in to unmask]> wrote:

>Basilius wrote:
>>Last time I had the opportunity to check Starostin's Altaic databases
>>on my own computer, they were approaching 3,000 Proto-Altaic roots
>>attested in no less than *three subfamilies* each (the subfamilies
>>being: Turkic, Mongolian, Tungus-Manchu, Korean, Japanese-Ryukyuan).
>The main problem being that Starostin is willing to accept more semantic
>and phonetic leeway than most credible Indo-Europeanists. But that's a
>methodological issue that could be quibbled over for years.

I can't agree. Don't know what you call 'leeway', but for me, an
average entry in his database looks no less reliable than an entry in
some Walde-Pokorny. I think it's simply the hypnosis of old tradition:
everybody knows about I-E, and nearly everybody forgets all those minor
(and major) phonetic irregularities, shaky hypotheses on semantic
development, etc. that one has to assume for nearly every root.

I bet I can take the worst of Starostin's roots (not marked as doubtful
by Starostin himself)) and immediately find a well-known I-E comparison
that would look much less convincing if you weren't aware that it belongs
to a well established language family.

And the database represents in fact an unpublished material. Do you know
any critique of it already made public? I mean that the generalisation
about what Starostin & Co are willing to accept must have been pronounced
by somebody knowing the material, whom I guess you were citing?