On megalocomparison

JA Matisoff - Language, 1990 - JSTOR
JA Matisoff
Language, 1990JSTOR
There is no denying that G's central thesis-that all the native language families of the
Western hemisphere (except for Eskimo-Aleut and Na-Dene) are genetically related,
descending from a common ancestor,'Proto-Amerind', spoken about 11,000 years ago (335)-
has a certain romantic sweep to it. Such a far-reaching claim would seem to require pretty
convincing evidence to back it up. Yet G expresses his disdain for the conventional sort of
historical com-parison, so'obsessionally'preoccupied with sound correspondences and …
There is no denying that G's central thesis-that all the native language families of the Western hemisphere (except for Eskimo-Aleut and Na-Dene) are genetically related, descending from a common ancestor,'Proto-Amerind', spoken about 11,000 years ago (335)-has a certain romantic sweep to it. Such a far-reaching claim would seem to require pretty convincing evidence to back it up. Yet G expresses his disdain for the conventional sort of historical com-parison, so'obsessionally'preoccupied with sound correspondences and asterisked reconstructions. 2 Instead he values the kind of evidence provided by the more'powerful'method of'mass comparison', whereby wordlists and grammatical paradigms from many languages are grokked simultaneously to see whether any root-words or affixes look alike. Languages which share a number of resemblant vocabulary items are then deemed to be genetically related, unless these forms are obvious loans (22-23). So powerful is this method, G believes, that it yields valid results even with the worst data, and to any desired time-depth:'The method of multilateral comparison is so powerful that it will give reliable results even with the poorest of materials. Incorrect material should have merely a randomizing effect'(29). Errors do not greatly concern G:'Although I have exercised great care, it would be miraculous if, in handling such a vast amount of material, there were no errors of fact or interpretation'(1989: 112). To this G's critic Adelaar (a Quechua specialist) rejoins that'This is highly questionable if one looks at the quality of the data G presents where the number of erroneous forms probably exceeds
*[Editor's note: Because of the importance to historical linguistics of the issues raised by Joseph Greenberg, his supporters, and his critics, the editor decided to solicit a discussion of these issues by a scholar whose qualifications included extensive experience in comparative linguistic research and a presumed lack of prior bias either for or against Greenberg's views. The following essay is the result.]
JSTOR