Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!news.duke.edu!godot.cc.duq.edu!newsfeed.pitt.edu!gatech!swrinde!howland.reston.ans.net!ix.netcom.com!netcom.com!alderson
From: alderson@netcom.com (Richard M. Alderson III)
Subject: Re: List of Mandan & Welsh Points of Resemblance
In-Reply-To: philip@storcomp.demon.co.uk's message of Wed, 8 Feb 1995 01:05:14 +0000
Message-ID: <aldersonD3p4o5.2o9@netcom.com>
Reply-To: alderson@netcom.com
Fcc: /u52/alderson/postings
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <D3ECzH.BzG@news.cis.umn.edu> <3gtb93$269@hecate.umd.edu>
	<792205514snz@storcomp.demon.co.uk>
Date: Wed, 8 Feb 1995 19:26:28 GMT
Lines: 73

In article <792205514snz@storcomp.demon.co.uk> philip@storcomp.demon.co.uk
(Phil Hunt) writes:

>In article <3gtb93$269@hecate.umd.edu> jmu@hamlet.umd.edu "James Unger"
>writes:

>>Even "look-alike" lists like this have some value for comparative
>>linguistics, for they demonstrate, in a backhanded way, what a really good
>>list of matches ought to look like, viz. a lot better than this in terms of
>>interlocking, exceptionless phonemic correspondences and semantic fits.  If
>>two languages that have such a tiny probability of being related as Mandan
>>and Welsh can yield a list such as this, then we must obviously demand much
>>better in cases in which the probability is, by general consensus,
>>considerably higher.

>One way to do it scientifically would be to have a fixed list of a few hundred
>common words that are likely to stay unchanged - words as body parts, family
>relationships, numbers etc, and measure how much difference there is in (1)
>languages we know are related, (2) any 2 languages taken at random. (There
>would have to be an algorithm for measuring how different 2 phonetic
>representations are form each other, and perhaps it could also try to find
>regular sound shifts).

As has been noted on this newsgroup, numbers are not "likely to stay unchanged"
--the Indo-European experience turns out to be the unusual state of affairs.

Family relationships unfortunately have the tendency to be influenced by so-
called "nursery words"--the early vocalizations of children learning to speak
are taken up into adult speech, amplified, and brought into the lexicon.  More
distant relationships are not particularly stable:  The reconstructed kinship
system of PIE is very different from that of modern English, for example.

It is possible that certain body part names are resistant to change, but I view
this as circular:  The original hypothesis was that some may be resistant, with
a suggestion of which ones those were, and behold! those lexical items seem to
be resistant to change.

>If this was done in a consistent way for a large number of pairs of languages,
>it would be possible to say how close 2 languages are before coincidence
>become evidence of a relationship.

The problem is that *pairwise* comparison of languages is the wrong way to go
about studies of genetic relationship.  Adding a third language to the study
acts as a check on coincidental similarities; ideally a fourth is available as
well.

>Does anyone know if this has actually been done?

It's been attempted.  The results aren't particularly convincing, in the
opinions of a large number of linguists.

>>Conversely, it would be enlightening to see how Greenberg or some
>>Nostraticist would explain how this list is, in some clear and definite way,
>>inferior to their lists of alleged cognates.

Not to defend Greenberg, but his lists are at least of much greater length.
The Nostraticists work within the generally paradigms of the comparative method
(as used by Algonquianists, for example); the issue for critics of Nostratic
seems to be with the sparsity of good material from families other than Indo-
European rather than with methodology.

>I once read of someone involved with Nostratic or one of these other
>superfamilies who reckonned that two words were related because they sounded
>similar and had similar meanings (one meant "milk", the other "neck") -
>apparently the semantic shift was milk->breasts->upper torso->neck.

That was a Proto-World etymology, adequately criticized by Hock in a paper
presented at the 1993 Chicago Linguistic Society.
-- 
Rich Alderson		[Tolkien quote temporarily removed in favour of
alderson@netcom.com	 proselytizing comment below --rma]

Please support the creation of the humanities hierarchy of newsgroups!
