Newsgroups: sci.lang,soc.culture.nordic,soc.culture.turkish,soc.culture.mongolian,soc.culture.magyar,soc.culture.german
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!news-out.communique.net!mr.net!newsfeeds.sol.net!uwm.edu!cs.utexas.edu!howland.erols.net!feed1.news.erols.com!news.idt.net!enews.sgi.com!news.sgi.com!sdd.hp.com!vixen.cso.uiuc.edu!uchinews!not-for-mail
From: deb5@midway.uchicago.edu (Daniel von Brighoff)
Subject: Re: Sun Language Theory? (was Re: Finnish related to Turkish?)
X-Nntp-Posting-Host: ellis-nfs.uchicago.edu
Message-ID: <E3zGDH.CB8@midway.uchicago.edu>
Sender: news@midway.uchicago.edu (News Administrator)
Organization: The University of Chicago
References: <5a7qak$8fc@nyheter.chalmers.se> <5aotll$6t4@cpmt.cyberport.net> <E3rH8M.59q@midway.uchicago.edu> <32daa7f9.9106304@news.exodus.net>
Date: Tue, 14 Jan 1997 05:24:53 GMT
Lines: 49

In article <32daa7f9.9106304@news.exodus.net>,
Bill Vaughan <bill@osisoft.com> wrote:
>deb5@midway.uchicago.edu (Daniel von Brighoff) wrote:
>
>>You're missing a crucial element of language reconstruction:  They have to
>>be *regular* similarities, the results of *exceptionless* sound changes.
>>Anybody can find resemblances between sets of words; finding patterns of  
>>correspondences is considerably tougher. ...
>
>AFAIK not all scholars would agree. Certainly Greenberg et al believe
>or believed that multilateral comparison of word lists was a valid
>technique for detecting relationships between languages. Working out
>the precise _lautlehre_ (sound change rules) is principally important
>for bilateral comparisons, or so I have read.

	Don't even get me started on Greenberg, et al.  I don't see how
throwing out two centuries of hard-won experience in comparative
reconstruction and replacing it with "eye comparisons" (what *are* his
systematic rules for determining whether two or more words are cognate in
a mass-comparison list?) advances our knowledge of language origins one
bit.

>> ... Why is Mr. Chong starting with
>>modern forms anyway?  If he really wants to prove a relationship, why
>>doesn't he compare Sumerian to reconstructed Proto-Finno-Ugric and
>>Proto-Altaic?
>
>Perhaps because, as you point out, there are numerous speakers of
>Finnish, Turkish, etc. who can point out errors in the data, but there
>are relatively few native speakers of Proto-Finno-Ugric and
>Proto-Altaic around. Working from reconstructions is second-generation
>research -- one step further removed from the raw data, with all the
>ambiguities, errors, and occasional prejudices that are accidentally
>embedded in the reconstruction. It is akin to what the statisticians
>call "data-mining" -- very useful, if you are very careful.

	Yes, but working in ignorance of data that, in some cases,
represent a lifetime of painstaking reconstruction just strikes me as
foolhardy.  By all means, double check the data sets to see if your
predecessors knew what they were talking about, by why rediscover Grimm's
Law?  Who (besides Greenberg et ilk) would think of trying to prove a
relationship between the modern European languages and, say, those of the
Ancient Middle East without any reference to Proto-Indo-European as
presented by Pokorny et al.?

-- 
	 Daniel "Da" von Brighoff    /\          Dilettanten
	(deb5@midway.uchicago.edu)  /__\         erhebt Euch
				   /____\      gegen die Kunst!
