[tied] Re: OE *picga

From: Richard Wordingham
Message: 16572
Date: 2002-10-31

--- In cybalist@..., "tgpedersen" <tgpedersen@...> wrote:
> And note that the spotted, "besprenkelt" meaning is just one the
> branches that the almighty Austronesian/AfroAsiatic/IE *p/bH-r/l-
> word has sprouted
> http://www.angelfire.com/rant/tgpedersen/Opr.html
> namely "split, half" > "smash" > "pulverize, bespatter"
> "variegated"
> which means you can't contain *h1/w-epros to IE either, nor
> Considering that Manansal's/my list contains less than 100 roots of
> coincidence, isn't it surprising how much mileage I can get out of
> it? Every time a central idea in IE comes up I can find a match for
> it on the list? Unless one wants to consider the alternative: there
> was contact?

The p/bH-r/l- pattern matches a good many roots, as Piotr has already
pointed out. As an experiment, why not take every 50th root in
Pokorny and see what proportion of them you can match up with the
list. You may need to record purely phonetic and phonetic plus
semantic matches separately, but I fear not. Keep your matches; we
may want to discuss them, Piotr permitting. As you've looked at
www.zompist.com , you can see where this argument is going.

Incidentally, my 'dog' survey got bogged down on the issue of
independence. I got my sample set up to 60 odd, but it's difficult
to get more independent samples. (Turkish is the only Altaic
language with a Swadesh word list for the Rosetta project!) My
sampling rules eventually picked up Yidiny 'gudaga', a cognate of
Mbabaram 'dOg', but I have a statistical/linguistic problem - is it
related to Anindilyakwa wuruwad.e? If it is, how do I process them?
Discard the later sample?