LINGUIST List 6.83

Fri 20 Jan 1995

Disc: Comparative Method

Editor for this issue: <>


  1. , Comparative Linguistics
  2. Dan Everett, Re: 6.66 Comparative Method

Message 1: Comparative Linguistics

Date: Wed, 18 Jan 95 18:08:52 ESComparative Linguistics
From: <>
Subject: Comparative Linguistics

Re: Poser's recent posting, I would like to ask: How many
forms are required to be shared between two or more languages
for us to be able to tell that they are related? Meillet,who
should have known, seems to have felt that ten forms was
plenty in some cases he discussed, although he did not
specify a minimum.

Alexis MR
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Re: 6.66 Comparative Method

Date: Wed, 18 Jan 1995 07:50:19 Re: 6.66 Comparative Method
From: Dan Everett <>
Subject: Re: 6.66 Comparative Method

Bill Poser notes that the data used by Greenberg for some of his
classifications is less than overwhelming. In my check of Greenberg's
classification of the Mura family, more than half (19) of the
approximately 30 forms he lists are wrong. Of the remaining forms, only
four are (in my opinion - and how, in the absence of independent evidence
can we say whether my opinion is right or wrong?) really similar (in their
recorded or "phonological" forms) to other Chibchan-Paezan forms (where G
puts Muran). One of those four, ?ii wood, might be discounted as a
coincidence, since it is also like other, unrelated families' forms for
wood (e.g. Hokan and Tucanoan). The remaining forms are four segments or
fewer in length (and Muran languages had small phonemic inventories;
Piraha, the only surviving member has 7 consonants and three vowels;
Greenberg's forms don't include tones), which could easily be the result
of chance with such small sets of phonemes. This reduces his entire basis
of classification of Muran to ONE word, the word for 'flower', which he
lists as ?iobai (no tones given, again), but which is actually ?ao'bai'
('=high tone).

For his work on Arawan and Chapakuran, Bob Dixon and I have agreed that
there is NOT ONE proto form for Arawan that supports Greenberg's
comparison with Arawakan. Moreover, I have found no convincing
similarities in lexical items between Chapakuran and Arawakan. The *only*
similarity I have noticed is that Arawan, Arawakan, and Chapakuran all
have gender (Chapakuran has three genders, the others have two). But
so do two other close-by languages - Spanish and Portuguese.

Hardly overwhelming.

Dan Everett
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue