From: Piotr Gasiorowski
Message: 2239
Date: 2000-04-28
----- Original Message -----From: Glen GordonSent: Thursday, April 27, 2000 11:23 PMSubject: [cybalist] Linguistic Mathematics?Hi, Glen,I've expressed my aversion to glottochronology a long time ago on this list, and just don't feel like discussing it all again, but I hasten to oblige the sceptics. The margin of error in glottochronological dating is such that it has no real value for dating IE splits. The method is not based on any independently established principles of language differentiation but relies on empirically determined formulas which often sadly fail to demonstrate their applicability to a NEW group of languages when there is some external evidence for measuring the time-depth. I wonder how glottochronological calculations could ever be verified with respect to language families with little of no recorded history. For very distantly related languages the accuracy of glottochronology reduces to nil because of the usual "cognate or lookalike" controversies and the statistical problems regarding the interpretation of having just one or two matches.There is an elegant arithmetic formula (invented by the Polish mathematicial Sierpinski), which yields the n-th prime for any n. Ain't that wonderful? The only snag is that the formula contains a real number (defined as the sum of a certain series) which can only be determined with the desired accuracy if you already know all the primes from p(1) to p(n). Sierpinski showed that such a number exists but didn't show how to arrive at it without knowing the primes in advance. This mathematical joke reminds me of glottochronology.Piotr
Glen wrote:
And why are we pretending that glottochronology is a useful tool in
linguistics?? I get trashed for making some specific booboos on IE while
general booboos (like embracing glottochronology as a real science) go
unabated. Not to mention that this whole topic barely has to do with IE at
all despite moderation.