Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!www.nntp.primenet.com!nntp.primenet.com!cam-news-hub1.bbnplanet.com!howland.erols.net!netcom.com!petrich
From: petrich@netcom.com (Loren Petrich)
Subject: Re: Ruhlen's "On the Origin of Languages": a Review (I)
Message-ID: <petrichDxJyr8.7rM@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <4vvkje$guk@news.xs4all.nl> <511jpk$3k8@netsrv2.spss.com> <hubey.842308657@pegasus.montclair.edu> <5143ei$ejv@netsrv2.spss.com>
Date: Wed, 11 Sep 1996 05:18:44 GMT
Lines: 34
Sender: petrich@netcom4.netcom.com

In article <5143ei$ejv@netsrv2.spss.com>,
Mark Rosenfelder <markrose@spss.com> wrote:

	[attempted finding of cognates and estimation of probabilities...]

>Because you're not measuring the same thing as Ruhlen & Greenburg.
>Even if your calculations are correct, you're looking for phonetic and
>semantic matches.  They're not.  They broaden the phonetic and semantic
>target area such that one may expect to find matches virtually at will.

	I agree that if one spreads one's phonetic and semantic net far
enough, just about anything can be related to anything. However, they
claim that if one finds matches over several languages, then the
probability of coincidence becomes much less. Assuming sufficient 
phonetic/semantic scrambling, here is how the argument works:

	Let us suppose that the probability of a match of a word with a 
word in another language is p. Now try comparing to a third language. The 
probability of a match in all three is p^2, since the third is 
statistically independent of the other two. By mathematical induction, 
the probability of a match across a set of n languages is p^(n-1). As N 
becomes large, this probability can grow very small.

	One difficulty in applying this test occurs if some of the 
languages are known to be related; the assumption of statistical 
independence does not hold for them, and they are best treated as one 
language in this test.

	Is this reasoning good, or is there some hole I've missed?
-- 
Loren Petrich				Happiness is a fast Macintosh
petrich@netcom.com			And a fast train
My home page: http://www.webcom.com/petrich/home.html
Mirrored at: ftp://ftp.netcom.com/pub/pe/petrich/home.html


