The Language Instinct: How the Mind Creates Language (43 page)

Most of the other human languages on earth can also be grouped into phyla descending from ancient tribes of astoundingly successful farmers, conquerers, explorers, or nomads. Not all of Europe is Indo-European. Finnish, Hungarian, and Estonian are Uralic languages, which together with Lappish, Samoyed, and other languages are the remnants of a vast nation based in central Russia about 7,000 years ago. Altaic is generally thought to include the main languages of Turkey, Mongolia, the Islamic republics of the former USSR, and much of central Asia and Siberia. The earliest ancestors are uncertain, but later ones include a sixth-century empire as well as the Mongolian empire of Genghis Khan and the Manchu dynasty. Basque is an orphan, presumably from an island of aboriginal Europeans that resisted the Indo-European tidal wave.

Afro-Asiatic (or Hamito-Semitic), including Arabic, Hebrew, Maltese, Berber, and many Ethiopian and Egyptian languages, dominates Saharan Africa and much of the Middle East. The rest of Africa is divided among three groups. Khoisan includes the IKung and other groups (formerly called “Hottentots” and “Bushmen”), whose ancestors once occupied most of sub-Saharan Africa. The Niger-Congo phylum includes the Bantu family, spoken by farmers from western Africa who pushed the Khoisan into their current small enclaves in southern and southeastern Africa. The third phylum, Nilo-Saharan, occupies three large patches in the southern Saharan region.

In Asia, Dravidian languages such as Tamil dominate southern India and are found in pockets to the north. Dravidian speakers must therefore be the descendants of a people who occupied most of the Indian subcontinent before the incursion of the Indo-Europeans. Some 40 languages between the Black Sea and the Caspian Sea belong to the family called Caucasian (not to be confused with the informal racial term for the typically light-skinned people of Europe and Asia). Sino-Tibetan includes Chinese, Burmese, and Tibetan. Austronesian, having nothing to do with Australia (
Austr
- means “south”), includes the languages of Madagascar off the coast of Africa, Indonesia, Malaysia, the Philippines, New Zealand (Maori), Micronesia, Melanesia, and Polynesia, all the way to Hawaii—the record of people with extraordinary wanderlust and seafaring skill. Vietnamese and Khmer (the language of Cambodia) fall into Austro-Asiatic. The 200 aboriginal languages of Australia belong to a family of their own, and the 800 of New Guinea belong to a family as well, or perhaps to a small number of families. Japanese and Korean look like linguistic orphans, though a few linguists lump one or both with Altaic.

What about the Americas? Joseph Greenberg, whom we met earlier as the founder of the study of language universals, also classifies languages into phyla. He played a large role in unifying the 1,500 African languages into their four groups. Recently he has claimed that the 200 language stocks of native Americans can be grouped into only three phyla, each descending from a group of migrants who came over the Bering land bridge from Asia beginning 12,000 years ago or earlier. The Eskimos and Aleuts were the most recent immigrants. They were preceded by the Na-Dene, who occupied most of Alaska and northwestern Canada and embrace some of the languages of the American Southwest such as Navajo and Apache. This much is widely accepted. But Greenberg has also proposed that all the other languages, from Hudson Bay to Tierra del Fuego, belong to a single phylum, Amerind. The sweeping idea that America was settled by only three migrations has received some support from recent studies by Cavalli-Sforza and others of modern natives’ genes and tooth patterns, which fall into groups corresponding roughly to the three language phyla.

At this point we enter a territory of fierce controversy but potentially large rewards. Greenberg’s hypothesis has been furiously attacked by other scholars of American languages. Comparative linguistics is an impeccably precise domain of scholarship, where radical divergences between related languages over centuries or a few millennia can with great confidence be traced back step by step to a common ancestor. Linguists raised in this tradition are appalled by Greenberg’s unorthodox method of lumping together dozens of languages based on rough similarities in vocabulary, rather than carefully tracing sound-changes and reconstructing proto-languages. As an experimental psycholinguist who deals with the noisy data of reaction times and speech errors, I have no problem with Greenberg’s use of many loose correspondences, or even with the fact that some of his data contain random errors. What bothers me more is his reliance on gut feelings of similarity rather than on actual statistics that control for the number of correspondences that might be expected by chance. A charitable observer can always spot similarities in large vocabulary lists, but that does not imply that they descended from a common lexical ancestor. It could be a coincidence, like the fact that the word for “blow” is
pneu
in Greek and
pniw
in Klamath (an American Indian language spoken in Oregon), or the fact that the word for “dog” in the Australian aboriginal language Mbabaram happens to be
dog
. (Another serious problem, which Greenberg’s critics do point out, is that languages can resemble each other because of lateral borrowing rather than vertical inheritance, as in the recent exchanges that led to
her negligées
and
le weekend
.)

The odd absence of statistics also leaves in limbo a set of even more ambitious, exciting, and controversial hypotheses about language families and the prehistoric peoplings of continents that they would represent. Greenberg and his associate Merritt Ruhlen are joined by a school of Russian linguists (Sergei Starostin, Aharon Dogopolsky, Vitaly Shevoroshkin, and Vladislav Illich-Svitych) who lump languages aggressively and seek to reconstruct the very ancient language that would have been the progenitor of each lump. They discern similarities among the proto-languages of Indo-European, Afro-Asiatic, Dravidian, Altaic, Uralic, and Eskimo-Aleut, as well as the orphans Japanese and Korean and a few miscellaneous language groups, reflecting a common ancestor proto-proto-language they call Nostratic. For example, the reconstructed Proto-Indo-European word for mulberry,
mor
, is similar to Proto-Altaic
mü@@@@
“berry,” Proto-Uralic
marja
“berry,” and Proto-Kartvelian (Georgian)
marcaw
“strawberry.” The Nostraticists would have them all evolve from the hypothetical Nostratic root
marja
. Similarly, Proto-Indo-European
melg
“to milk” resembles Proto-Uralic
malge
“breast” and Arabic
mlg
“to suckle.” Nostratic would have been spoken by a hunter-gatherer population, for there are no names of domesticated species among the 1,600 words the linguists claim to have reconstructed. The Nostratic hunter-gatherers would have occupied all of Europe, northern Africa, and northern, northeastern, western, and southern Asia, perhaps 15,000 years ago, from an origin in the Middle East.

And various lumpers from this school have suggested other audacious superphyla and super-superphyla. One comprises Amerind and Nostratic. Another, Sino-Caucasian, comprises Sino-Tibetan, Caucasian, and maybe Basque and Na-Dene. Lumping the lumps, Starostin has suggested that Sino-Caucasian can be connected to Amerind-Nostratic, forming a proto-proto-proto language that has been called
SCAN
, covering continental Eurasia and the Americas. Austric would embrace Austronesian, Austro-Asiatic, and various minor languages in China and Thailand. In Africa, some see similarities between Niger-Congo and Nilo-Saharan that warrant a Congo-Saharan group. If one were to accept all of these mergers—and some are barely distinguishable from wishful thinking—all human languages would fall into only six groups:
SCAN
in Eurasia, the Americas, and northern Africa; Khoisan and Congo-Saharan in sub-Saharan Africa; Austric in Southeast Asia and the Indian and Pacific Oceans; Australian; and New Guinean.

Ancestral stocks of this geographic magnitude would have to correspond to the major expansions of the human species, and Cavalli-Sforza and Ruhlen have argued that they do. Cavalli-Sforza examined minor variations in the genes of hundreds of people representing a full spectrum of racial and ethnic groups. He claims that by lumping together sets of people who have similar genes, and then lumping the lumps, a genetic family tree of humankind can be constructed. The first bifurcation splits the sub-Saharan Africans off from everyone else. The adjoining branch in turn splits into two, one embracing Europeans, northeast Asians (including Japanese and Koreans), and American Indians, the other containing southeast Asians and Pacific Islanders on one sub-branch, and aboriginal Australians and New Guineans on another. The correspondences with the hypothetical language superphyla are reasonably clear, though not perfect. One interesting parallel is that what most people think of as the Mongoloid or Oriental race on the basis of superficial facial features and skin coloring may have no biological reality. In Cavalli-Sforza’s genetic family tree, northeast Asians such as Siberians, Japanese, and Koreans are more similar to Europeans than to southeast Asians such as Chinese and Thai. Strikingly, this non-obvious racial grouping corresponds to the non-obvious linguistic grouping of Japanese, Korean, and Altaic with Indo-European in Nostratic, separate from the Sino-Tibetan family in which Chinese is found.

The branches of the hypothetical genetic/linguistic family tree can be taken to depict the history of
Homo sapiens sapiens
, from the African population in which mitochondrial Eve was thought to evolve 200,000 years ago, to the migrations out of Africa 100,000 years ago through the Middle East to Europe and Asia, and from there, in the past 50,000 years, to Australia, the islands of the Indian and Pacific Oceans, and the Americas. Unfortunately, the genetic and migrational family trees are almost as controversial as the linguistic one, and any part of this interesting story could unravel in the next few years.

A correlation between language families and human genetic groupings does
not
, by the way, mean that there are genes that make it easier for some kinds of people to learn some kinds of languages. This folk myth is pervasive, like the claim of some French speakers that only those with Gallic blood can truly master the gender system, or the insistence of my Hebrew teacher that the assimilated Jewish students in his college classes innately outperformed their Gentile classmates. As far as the language instinct is concerned, the correlation between genes and languages is a coincidence. People store genes in their gonads and pass them to their children through their genitals; they store grammars in their brains and pass them to their children through their mouths. Gonads and brains are attached to each other in bodies, so when bodies move, genes and grammars move together. That is the only reason that geneticists find any correlation between the two. We know that the connection is easily severed, thanks to the genetic experiments called immigration and conquest, in which children get their grammars from the brains of people other than their parents. Needless to say, the children of immigrants learn a language, even one separated from their parents’ language by the deepest historical roots, without any disadvantage compared to age-mates who come from long lineages of the language’s speakers. Correlations between genes and languages are thus so crude that they are measurable only at the level of superphyla and aboriginal races. In the past few centuries, colonization and immigration have completely scrambled the original correlations between the superphyla and the inhabitants of the different continents; native English speakers, to take the most obvious example, include virtually every racial subgroup on earth. Well before that, Europeans interbred with their neighbors and conquered each other often enough that there is almost no correlation between genes and language families within Europe (though the ancestors of the non-Indo-European Lapps, Maltese, and Basques left a few genetic mementos). For similar reasons, well-accepted language phyla can contain strange genetic bedfellows, like the black Ethiopians and white Arabs in the Afro-Asiatic phylum, and the white Lapps and Oriental Samoyeds in Uralic.

Moving from the highly speculative to the borderline flaky, Shevoroshkin, Ruhlen, and others have been trying to reconstruct words ancestral to the six superphyla—the vocabulary of the language of African Eve, “Proto-World.” Ruhlen has posited 31 roots, such as
tik
“one” which would have evolved into Proto-Indo-European
deik
“to point” and then Latin
digit
“finger,” Nilo-Saharan
dik
“one,” Eskimo
tik
“index finger,” Kede
tong
“arm,” Proto-Afro-Asiatic
tak
“one,” and Proto-Austro-Asiatic
ktig
“arm or hand.” Though I am willing to be patient with Nostratic and similar hypotheses pending the work of a good statistician with a free afternoon, I find the Proto-World hypothesis especially suspect. (Comparative linguists are speechless.) It is not that I doubt that language evolved only once, one of the assumptions behind the search for the ultimate mother tongue. It’s just that one can trace words back only so far. It is like the man who claimed to be selling Abraham Lincoln’s ax—he explained that over the years the head had to be replaced twice and the handle three times. Most linguists believe that after 10,000 years no traces of a language remain in its descendants. This makes it extremely doubtful that anyone will find extant traces of the most recent ancestor of all contemporary languages, or that that ancestor would in turn retain traces of the language of the first modern humans, who lived some 200,000 years ago.