In the beginning, there was one people, perhaps no more than 2,000 strong, who had acquired an amazing gift, the faculty for complex language.
Favored by the blessings of speech, their numbers grew, and from their cradle in the northeast of Africa, they spread far and wide throughout the continent. One small band, expert in the making of boats, sailed to Asia, where some of their descendants turned westward, ousting the Neanderthal people of Europe and others east toward Siberia and the Americas.
These epic explorations began some 50,000 years ago and by the time the whole world was occupied, the one people had become many. Differing in creed, culture and even appearance, because their hair and skin had adapted to the world's many climates in which they now lived, they no longer recognized one another as the children of one family. Speaking 5,000 languages, they had long forgotten the ancient mother tongue that had both united and yet dispersed this little band of cousins to the four corners of the earth.
Peter DaSilva for The New York Times
Dr. Greenberg has grouped most of the world's languages into a small number of clusters based on their similarities.
Many deride attempts to reconstruct the family tree of languages beyond the most obvious groupings like the Romance languages and Indo-European. Their argument is that language changes too fast for its roots to be traced back further than a few thousand years. If any single language ever existed, most linguists say, it is irretrievably lost.
But one scholar in particular, Dr. Joseph H. Greenberg of Stanford University, has defied this ardent pessimism. In the course of a long career, he has classified most of the world's languages into just a handful of major groups.
Though it remains unclear how these superfamilies may be related to one another, he has identified words and concepts that seem common to them all and could be echoes of a mother tongue.
And this month, at the age of 84, Dr. Greenberg is publishing the first of two volumes on Eurasiatic, his proposed superfamily that includes a swath of languages spoken from Portugal to Japan.
Like the biologist E. O. Wilson, Dr. Greenberg is that rare breed of academic, a synthesizer who derives patterns from the work of many specialists, an exercise the specialists do not always welcome.
But though biologists came to acknowledge the pioneering value of Dr. Wilson's work, linguists have reached no such consensus on that of Dr. Greenberg.
Will he one day be recognized as having done for language what Linnaeus did for biology, as his Stanford colleague and associate Dr. Merritt Ruhlen believes, or is his work more fit, as one critic has urged, to be "shouted down"?
Dr. Greenberg is by no means an outcast from his profession. He is one of the very few linguists who are members of the National Academy of Sciences, the country's most exclusive scientific club. His work on language typology (universal patterns of word order) is highly regarded. Somewhat puzzlingly, his fellow linguists generally accept his work on the relationships among African languages but furiously dispute his ordering of American Indian languages, even though both classifications were achieved with the same method.
Dr. Greenberg's work is of considerable interest to population geneticists trying to reconstruct the path of early human migrations by means of genetic patterning in different peoples.
Although genes and languages are not bequeathed in the same way, both proceed in a series of population splits.
"We have found a lot of significant correspondences between what he says and what we see genetically," said Dr. Luca Cavalli-Sforza, a leading population geneticist at Stanford. In his view, the majority of linguists are not interested in the evolution of language. They "have attacked Greenberg cruelly, and I think frankly there is some jealousy behind it because he has been so successful," Dr. Cavalli-Sforza said.
In a windowless office lined with grammars and dictionaries of languages from all over the world, Joseph Greenberg fishes in the plastic shopping bag that is serving as his briefcase. He pulls out one of the handwritten notebooks that are the key to his method of discovering language relationships. Down the left hand margin is a list of the languages being compared. Along the top are names of the vocabulary words likely to yield similarities.
His method, which he calls mass or multilateral comparison, is to compare many languages simultaneously on the basis of 300 core words in the hope that they will sort themselves into clusters representative of their historical development. Many linguists believe such an exercise is futile because words change too quickly to preserve any ancestry older than 5,000 years or so.
"They sell their own subject short," Dr. Greenberg says. "Certain items in language are extremely stable, like personal pronouns or parts of the human body."
Born in Brooklyn in 1915, he was interested in language almost from birth. His father spoke Yiddish and his mother's family German. "I was brought up to believe Yiddish was an inferior language because my father's relatives got invited to the house as seldom as possible," he said. Hebrew school exposed him to a fourth language. He had a good enough ear that an alternative career as a professional pianist beckoned.
But anthropology won out. After doctoral studies at Northwestern, he did fieldwork on the pagan cults of the Hausa-speaking people of northern Nigeria before deciding that his true interest lay in linguistics.
At the time, there was no agreement on the history of African languages. "So I started in a simple-minded way," Dr. Greenberg said. "I took common words in a number of languages and saw if the languages fell into groups." He found that he could reduce all the continent's languages first to 14 and later to 4 major clusters.
In a 1955 article, he described these as Afro-Asiatic, which includes the Semitic languages of Arabic and Hebrew, as well as ancient Egyptian, and is spread across Northern Africa; Nilo-Saharan, a group of languages spoken in Central Africa and the Sudan; Khoisan, which includes the click languages of the south; and Niger-Kordofanian, a superfamily that includes everything in between, including the pervasive Bantu languages.
After a decade of controversy, Dr. Greenberg's African classification became widely accepted. "But then a lot of people said I had gotten the correct results with the wrong method," he said.
Method is the formal issue that divides Dr. Greenberg from his critics. They say that the only way to prove that a group of languages is related is by establishing regular rules governing how words change as one language morphs into another.
The 'p' sounds in ancestral Indo-European, for example, change predictably into 'f' in German and English. Mere similarities between the words in different languages, like those on which Dr. Greenberg relies, fall far short of proof, his critics say, because the similarities could arise from chance or borrowing.
Because of the looseness of sound and meaning that Dr. Greenberg allows in claiming similarities, his data "do not rise above the level of chance," said Dr. Sarah Thomason, a linguist at the University of Michigan.
Dr. Brian D. Joseph of Ohio State University, who studies Nostratic, a proposed language superfamily similar to Euroasiatic, described Dr. Greenberg as "a romantic" for believing his methods could retrieve long lost languages.
Dr. Lyle Campbell, of the University of Canterbury in New Zealand and the author of a textbook on historical linguistics, said that rigorous proof was necessary because languages changed so fast, and that Dr. Greenberg's methods were "woefully inadequate."
To Dr. Greenberg and his colleague Dr. Ruhlen, the critics' requirement for establishing regular rules of sound change defies both common sense and history. The sound regularities in Indo-European, they say, were not detected until after the languages had been grouped by inductive methods similar to Dr. Greenberg's. The insistence on demonstrating sound-change regularities, in their view, has thwarted any further reconstruction of language families.
"It's a misguided perfectionism that is so perfect they have had no result," Dr. Ruhlen said. His and Dr. Greenberg's aim is to establish the probable links from which the full history of human language can be inferred.
"The ultimate goal," Dr. Greenberg said in concluding his 1987 book "Language in the Americas" (Stanford University Press), "is a comprehensive classification of what is very likely a single language family. The implications of such a classification for the origin and history of our species would, of course, be very great."
Because the Americas have been inhabited only recently, at least as compared with Africa, it would be surprising to find a larger number of language groups, and Dr. Greenberg decided there were only three, even though other linguists posit 100 or so independent stocks.
Amerind is the vast superfamily to which, in his view, most native languages of North and South America belong.
The other two clusters are Na-Dene, a group of languages spoken mostly in Alaska and northeast Canada, and Eskimo-Aleut, spoken across northern Alaska and Canada.
One striking feature that unites the Amerindian languages of both Americas, in Dr. Greenberg's view, is the use of words starting in 'n' to mean I/mine/we/ours and words beginning in 'm' to mean thou/thine/you/ yours. Not every language shows this pattern, but almost every Amerindian language family has one or more languages that have it, suggesting that all are derived from an original language in which first and second person pronouns started this way.
In the course of classifying the languages of the Americas, Dr. Greenberg realized that their major families were related to languages on the Eurasian continent, as would be expected if the Americas had been inhabited by people migrating through Siberia. Na-Dene, for example, is related to an isolated Siberian language known as Ket.
To help with the American classification, Dr. Greenberg started making lists of words in languages of the Eurasian land mass, particularly personal pronouns and interrogative pronouns.
"I began to see when I lined these up that there is a whole group of languages through northern Asia. I must have noticed this 20 years ago. But I realized what scorn the idea would provoke and put off detailed study of it until I had finished the American languages book," he said.
Thirteen years later, Dr. Greenberg has now classified most of the languages of Europe and Asia into the superfamily he calls Eurasiatic. Its seven living components are Indo-European (examples are English, Russian, Greek, Iranian, Hindu); Uralic (Hungarian, Finnish); Altaic (Turkish, Mongolian); the Korean-Japanese-Ainu group; Eskimo-Aleut; and two Siberian families known as Gilyak and Chukotian.
His concept of Eurasiatic was derived independently but overlaps with the proposed Nostratic superfamily, the theory of which has been developed in the last 30 years by Russian linguists.
At first sight it may seem hard to believe that languages as different as English and Japanese, say, share any commonalities. But in his new book on the grammar of Eurasiatic (a second volume on vocabulary is in progress), Dr. Greenberg has found many elements that he argues knit the major Eurasian language families into a single group.
Words beginning in 'm,' for example, are found in every Eurasiatic family to designate the first person (English: me; Finnish: minÃa; proto-Altaic: min; Old Japanese: mi). Every branch of Eurasiatic, Dr. Greenberg says, uses n-words to designate a negative, from the no/not of English to the -nai ending that makes Japanese verbs negative.
Every branch uses 'k' sounds to indicate a question. In Indo-European, many Latin interrogatives begin qu-, as in quid pro quo. In Finnish, -ko is added to a verb to indicate a question. In Japanese the same role is played by -ka. The word for 'who?' is kim in Turkish, kin in Aleut.
If Dr. Greenberg's Eurasiatic proposal is at first no more favorably received than his Amerindian classification, he will not be surprised.
"A fair part of my publications is just polemics," he says, with an air of resignation.
Meanwhile, Dr. Ruhlen believes that if the Eurasiatic grouping is accepted, the world's 5,000 languages can be seen to fall into just 12 superfamilies.
How these in turn might be related to a single mother tongue remains to be seen. But several years ago, Dr. Greenberg identified a possible global etymology derived from the universal human habit of holding up a single finger to denote one.
In the Nilo-Saharan languages the word tok, tek or dik means one.
The stem tik means finger in Amerind, one in Sino-Tibetan, 'index finger' in Eskimo and 'middle finger' in Aleut.
And an Indo-European stem deik, meaning to point, is the origin of daktulos, digitus, and doigt -- Greek, Latin and French for finger -- as well as the English word digital.
No one has pointed more clearly at the one language than Joseph Greenberg.