Investigations

Language families v. genetic relatives
When the Oakland, California, school board proposed recognizing African-American Vernacular English, or Ebonics, in its classrooms in 1996, it began a firestorm of controversy—largely because the proposal cited studies that "have demonstrated that African Language Systems are genetically based and not a dialect of English." Although the school board quickly put out a clarification saying the phrase "genetically based" was "used according to the standard dictionary definition 'has origins in'" and "not used to refer to human biology," the damage had been done. The public had a confused notion of how linguists metaphorically refer to languages as "genetically related" or belonging to "language families."

"A lay person looks at this kind of thing and says, 'Genetic, oh,' and it becomes an explosive issue," explains Michael Silverstein, the Charles S. Grey distinguished service professor in anthropology, linguistics, psychology, and the Committee on General Studies in the Humanities. "People transfer through this metaphor and come to think people are genetically determined to speak an inferior dialect" (even though, Silverstein notes, black English is not inferior but merely a separate dialect from standard English).

It's easy to see how the metaphor could confuse nonlinguists, admits Silverstein, who studies the structure and history of North American indigenous and Australian languages. After all, humans generally inherit language from their family, the same group from which they inherit their genes. And linguists' diagrams of language family trees "look very much like our biological taxonomies." But a person's particular language is a function of social group and not biology, Silverstein argues, and what linguists call families are determined through similar word forms and meanings across entire languages.

That's why it irked Silverstein when Joseph Greenberg, the late Stanford linguist, claimed in a 1987 book to prove the unity of almost all the languages of the Western Hemisphere, and, in 2000, the languages of Europe and Asia, based on partial similarities of many words. For example, Greenberg noted, the English word no and the Japanese word-ending -nai both begin with an n-sound and designate a negative. And many native North and South American languages use words starting with n-sounds to mean I/mine/we/ours and words beginning with m-sounds to mean thou/thine/you/yours. Using such comparisons, Greenberg distilled 60 or 70 language families in the Western Hemisphere into one family, which he called "Amerind."

As Greenberg's proposals—which are based on partial-word similarities, much like DNA studies compare the degrees of similarity in genetic material—were touted in the press, geneticists proclaimed they could trace mitochondrial DNA in three major lines of indigenous people of the Western Hemisphere. The two ideas were used to confirm each other, compelling Silverstein—who speaks quickly yet with extremely clear diction-to address the topic at a May Skokie Public Library presentation, the October Humanities Open House, and even in his June American Anthropologist obituary of Greenberg.

Although exciting to the general public, Silverstein says, Greenberg's ideas "are based on nothing more sophisticated than probabilistically weak comparisons of look-alikes of bits and pieces of word forms. These proposals are speculative guesses on the basis of the wrong kind of data." Rather than word similarities, he argues, language families are based on "historical particularities of specific languages that point to common historical origins."

Consider the English two, French deux, German zwei, and Russian dva/dvje—which sound similar and mean the same thing. Although they all look alike, other evidence shows that only English and German come from the same language family: Germanic. French comes from Italic and Russian from Balto-Slavic. All four have roots in the larger language family Indo-European, but from there their branches head in different directions.

Many language subfamilies, such as the Romance languages, spread extensively through vast military conquests. In the period just before the large colonial immigration to modern imperial centers, Europe had only four basic language families: Indo-European, Finno-Ugric, Turkic, and Basque. Conversely, what is now the state of California had about 18 indigenous families when the Europeans arrived.

Sometimes the line of a language family seems counterintuitive. Take the English verb forms break, broke, and broken and the English words fracture, fragile, fragment, and frangible. The latter look as if they were French, which comes from Latin, and might suggest that English is a Romance language. But the correct comparison, the one that "reveals where English really came from," Silverstein says, is with the German brechen, brach, (ge)brochen. The irregular verb forms of the English break correspond point for point with German grammar.

The Germanic family's br- corresponds to the Italic family's fr-, not because German or French are intimates in a small family, but because they are related in the larger Indo-European family at a more remote time. The lesson: linguistic materials that look alike often are not closely related, and materials don't have to look alike to demonstrate family relationship.

Unlike DNA studies, therefore, language families are not classified by how much linguistic materials resemble each other, but whether scholars can find matching form and meaning elements across languages. Some language resemblances may reflect sheer chance, language borrowing, or functionally driven independent reinventions such as onomatopoetic words (the words for blow, burp, and fart, Silverstein points out, tend to begin with b- or ph- sounds in many languages).

Then there's the matter of time. While scientists can trace the human species back about a million years, linguists have traced language back only 7,000-10,000 years. One can't simply fill in speculation for that gap, Silverstein says. Though Silverstein respects Greenberg's ambitious goals (to classify every human language) and expanse of work—his Greenberg obituary was highly laudatory—he believes his colleague's analysis lacks scientific basis.

And as shown by the Oakland Ebonics example, linking language acquisition with biological genetics has questionable cultural implications. "Why did this have such an extraordinary buzz among nonscientists?" Silverstein wonders. He believes people may be "grasping for a biological anchor for the existence of culturally and linguistically different groups" as an easy solution to the inequities in the world. They could argue ethnic, racial, or linguistic superiority, or that people have fundamental differences that can never be bridged. And that, he says, is the real danger of the genetic metaphor.

— A.B.