Merritt Ruhlen
Merritt Ruhlen is a linguist known for his work on classification of language families. He received his
Ph.D. from
Stanford University in 1973 with a dissertation on the generative analysis of
Romanian morphology. He subsequently worked for several years as a research assistant on the
Stanford Universals Project directed by
Joseph Greenberg and
Charles Ferguson. Since 1976 he has been employed primarily as an independent linguist. He is currently a lecturer in Anthropological Sciences and Human Biology at Stanford and a co-director of the
Santa Fe Institute Program on the Evolution of Human Languages.
Ruhlen is an extremely controversial figure in the
linguistics community due to his vocal support of the
Proto World Hypothesis, which claims that all the extant languages of the world can be traced back to a single
proto-language, and that parts of this proto-language can be accurately reconstructed. Most mainstream
historical linguists reject Ruhlen's assumptions and methodology, holding that is is impossible reconstruct a language spoken at least 30,000 years ago, possibly more than 100,000 years ago.
The majority of criticisms of Ruhlen center around his use of
mass comparison, which instead of using common historical linguistic methods of comparison, involves comparing the
lexicons of however many languages one is investigating and examining them for words in two or more languages which appear similar phonologically and have a similar meaning. Historical linguists argue the results of mass comparison could easily be cases of simple coincidence rather than pointing to a common ancestor. The accepted method of determining relationships between languages involved finding regular phonological correspondences between vocabulary items. This mainstream method is in principle
falsifiable: if one linguist suggests that correspondences between languages A and B suggest descent from a common ancestor, another linguist can search for examples which violate these correspondences, thus falsifying the hypothesis. This is difficult or impossible to do with the results of mass comparison, however, which makes most linguists skeptical.
Another problem with mass comparison is that it does not provide any way to distinguish between similarities due to common descent and those due to borrowing. Ruhlen maintains that borrowing of so-called basic vocabulary is rare.
Linguists are also wary of Ruhlen's arguments because of errors which have been uncovered in the data he presents, most frequently, mistranslations, citing a form when the form in the parent language is known and is more distant phonologically and semantically from the proposed Proto-World root, and separating cited words with arbitrary morpheme boundaries.
Ruhlen and his followers reply that the sheer volume of the correspondences which their mass comparisons have turned up is far too large to possibly be due to chance. They insist that even if many of the results were chance similarities, it is beyond belief that there could be so many similarities. They have not undertaken formal statistical analyses, however, and their claims have been contested in statistical studies such as Kessler's (2001).
Perhaps the strongest evidence supporting Ruhlen can be found in the work of the geneticist
Luigi Luca Cavalli-Sforza, who has studied the genes in human populations throughout the world and constructed a
phylogenetic tree, a structure similar in many respects to traditional trees of language families, showing where in the "tree" given genetic groups separated. The results are widely (though not universally) accepted as matching up remarkably well with Ruhlen's proposed structure of the languages and language families of the world. This tree has been criticized by some linguists and anthropologists on several grounds: that it makes selective use of languages and populations (omitting the very numerous Sino-Tibetan speakers of northern China, for example); that it assumes the truth of such linguistic groups as
Austric and
Amerind that are unproven or highly controversial; and that several of the population groups listed are defined not by their genes but by their languages—making the correlation irrelevant to a comparison of genetic and linguistic branching and
tautological as well (Bateman 1990, Trask 1996).
*
