LINGUIST List 5.1446

Wed 14 Dec 1994

Disc: Comparative Method

Editor for this issue: <>


  1. , comparative method
  2. Karl Teeter, Why Indo-European

Message 1: comparative method

Date: Dec 2 1994
From: <>
Subject: comparative method

) I recently asked whether anyone could produce examples in which
) proposed genetic relationships were rejected on the basis of
) perceived limitations of the comparative method, suggesting that
) in fact there are no such examples and that this issue is a
) red-herring.
) Alexis-Manaster Ramer responds that Johanna Nichols' book
) contains an example of this type on p.6. I find no such example.
) While she takes the position that in general the comparative
) method only gets us back so far and that other methods must be
) used if inferences about the remote past are to be drawn, she
) does not anywhere in the book dismiss a remote proposal on the
) basis of its temporal depth.
) Lloyd Anderson says that Ives Goddard took such a position at
) the Albuquerque LSA Institute. Not having been present I can
) neither confirm nor deny this, but it is not consistent with
) Goddard's other statements. In any case, even from Anderson's
) description it appears that Goddard merely offered an
) EXPLANATION of why there are no generally accepted remote
) relationships and of why we are not likely to discover more.
) This is a far cry from dogmatically refusing to consider a
) proposal on the basis of perceived limitations of the
) comparative method. As Anderson admits to have been the case in
) Albuquerque, Goddard's published statements on Amerind contain
) explicit criticism of the data and methods - they do not rely on
) temporal limitations to reject the proposal out of hand.
) It thus appears, as I thought, that there are no real examples
) of perceived limitations of the comparative method being used
) as the basis for rejecting proposals of genetic affiliation.
) There is some confusion about ad hominem arguments.
) To answer Alexis' query, the reason I characterized the
) claim that proposals have been rejected because of perceived
) limits of the comparative method as ad hominem is that it
) goes only to the motivation of the critics, not to their
) arguments.
) The point about ad hominem arguments is this. Recent proposals
) of remote genetic affiliation have been criticized on two
) grounds:
) (a) data (mostly in the case of Greenberg's Amerind)
) (b) method
) To a considerable extent, proponents of these remote
) relationships, both here on LINGUIST and elsewhere, have not
) defended the questioned data or methods, but have instead
) indulged in ad hominem attacks on the critics. (Probably the
) most ridiculous and offensive is Merritt Ruhlen's claim (The
) Stanford Daily, 10 February 1993) that the skeptics are closet
) racists who do not want English to be related to the languages
) of other races.) Insofar as the criticisms of the data and
) methods are correct, the motivations of the critics are utterly
) irrelevant.
) Lloyd Anderson argues that motivation is relevant under the
) credibility-of-witnesses exception to the irrelevance of
) ad hominem arguments. He appears to think that the critics
) are relying on their own authority and that their credibility
) is therefore a legitimate issue. However, the critiques do
) not rely on the authority of the critics. The methodological
) arguments are fully laid out in the literature - no one is
) asking for the critics' judgments to be taken on faith. Some
) critiques of the data rely on the critics knowledge of the languages,
) but to a large extent the data is publicly available and the
) problems with it are explicitly laid out. The validity of the
) criticism thus does not depend on the credibility of the critics,
) whence the motivation of the critics is not probative.
) Anderson raises a number of additional issues, several of which
) warrant a response.
) Anderson disputes the 12 KYBP figure I used for Amerind. The figure is
) Greenberg's, not mine.
) Anderson claims that:
) ...there does come up repeatedly the innuendo or invited
) inference that if it has not been discovered by competent
) specialists already, it cannot be.
) In fact, I do not hold this view and find nothing in my posting
) to suggest it. The "invited inference" is a fantasy of Mr. Anderson's.
) The confusion of "...what a particular set of methods or rules
) of evidence can achieve" with "what historical-comparative method
) can achieve" is Mr. Anderson's. The discussion here is of "the
) comparative method", a particular set of techniques, and its
) limitations. Both I and other critics are fully aware of the
) possibility of using other techniques.
) Mr. Anderson believes that critics refuse even to consider
) proposals of remote relationship. This is hardly the case.
) Indeed, the very furor that has erupted over Amerind attests
) to the attention they have been willing to give to such proposals.
) What governs the attitude of critics is not potential remoteness
) but the quality of the data and the argumentation.
) There is a peculiar teleology in Mr. Anderson's views of what is
) legitimate "consideration". He says that a rule requiring
) matches in three consonants would prevent us from even
) considering the relationship between Siouan and Yuchi, whence he
) concludes that the rule must be wrong. How do we know a priori,
) as Mr. Anderson appears to, that Siouan and Yuchi are related
) and that this relationship is necessarily discoverable? If
) (probably contrary to fact), matches of three consonants are
) necessary to exclude chance, and if this precludes a
) demonstration that Siouan and Yuchi are related, then so be it.
) I do not see why we should be unwilling, in that case, to
) conclude either that they are not related or that, if they are,
) the relationship is not demonstrable. Mr. Anderson seems to want
) to design our methods and criteria to suit the conclusions he
) wishes to draw.
) Mr. Anderson suggests that critics have dishonestly claimed
) that Greenberg's data is error-ridden when they could not back
) up such claims. Whatever difficulty he may have had in obtaining
) lists of errors in LIA from his correspondants, the fact is that
) quite a few have been published, as any reader of IJAL, the
) leading journal in this area, can attest. In addition to the
) errors discussed in the review by Robert Rankin that he cites,
) some are mentioned in Lyle Campbell's review in Language as well
) as in other reviews. Berman (1992), Kimball (1992) and Poser
) (1992) contain detailed critiques of the data for several
) languages. The errors are in many cases very significant. For
) example, Kimball argues that when errors are corrected there is
) NO Muskogean evidence for the "Amerind" pronominal pattern that
) is Greenberg's piece de resistance. These critiques of the data
) in LIA are to my knowledge unrebutted.
) In sum, there is no reason to believe that the criticism of
) proposals of remote relationship such as Amerind stems from
) perceived limitations of the comparative method. This red-herring
) should not distract us from the real issues of data and method.
) Berman, Howard (1992) ``A Comment on the Yurok and Kalapuya Data
) in Greenberg's Language in the Americas,''
) International Journal of American Linguistics
) 58.2.230-233.
) Campbell, Lyle (1988)
) Review of Language in the Americas.
) Language
) 64.591-615.
) Kimball, Geoffrey D. (1992)
) ``A Critique of Muskogean,"Gulf", and Yukian Material in
) Language in the Americas,''
) International Journal of American Linguistics
) 58.447-501.
) Poser, William J. (1992)
) ``The Salinan and Yurumangui Data in Language in the Americas,''
) International Journal of American Linguistics
) 58.2.202-229.
) Rankin, Robert (1992)
) Review of Language in the Americas.
) International Journal of American Linguistics
) 58.3.324-351.

 Bill Poser, First Nations Studies, University of Northern British Columbia,
 3333 University Way, Prince George, British Columbia, V2N 4Z9, Canada
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Why Indo-European

Date: Sat, 10 Dec 1994 11:16:24 Why Indo-European
From: Karl Teeter <>
Subject: Why Indo-European

 A recent (11/14) communication by Alexis Manaster Ramer asks why
 critics of such hypotheses as "Nostratic, Amerind, or even Altaic" "keep
 changing the criteria for demonstrable language relatedness in ways
 calculated to make Indo-European the sole standard or model of a genuine
 language family". He goes on to point out that IE is atyical because it is
 a "very young family".

 Young it may be, and yet it is the oldest family I know of which is
 comparably well-established, and that which has been the subject of most
 studies of the comparative method, and these seem to me good and sufficient
 reasons to take IE as a known case which we may then compare with other cases
 we would like to know better. Alexis cites two things as examples of his
 claim, "the insistence on morphological paradigms being reconstructed before
 we accept the validity of a proposed family, and the insistence on
 `complicated' and `unnatural' sound changes (such as *dw-) rk)".

Both of these examples are familiar enough (they stem from the
writings of Meillet, the most important theoretician of the
comparative method. In effect, Meillet claims that we can tell real
protolanguages by the fact that we can write grammars of them, at
least in principle. Where morphological comparisons come in is in his
discussion of how to evaluate similarities as probative of genetic
relationship. He generally cites the paradigm of the IE verb `to be'
to show this. Thus German has ist/sind where Latin has est/sunt,
and this single morphologically structured similarity is worth hundreds of
vocabulary look-alikes. As to dw) rk, this is indeed another favorite Meilet
 example, but his point in citing this (Greek dwo vs. Armenian erku `two') is
 essentially only to show that perfectly simple and natural sound laws, which
 are operable here, regularly work together to produce dissimilar forms, so
 items which correspond genetically quite commonly do not look alike.

 Why should Indo-European, then, continue to be taken as the standard or norm
 of a genuine language family? Quite simply, because it is the most obvious
 established case.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue