IN their archeological digs through the strata of human language, linguists have long been fascinated by the seeming similarities between the English words "fist," "finger" and "five." The motif is repeated by the Dutch, who say "vuist," "vinger" and "vijf," and the Germans, who say "faust," "finger" and "funf." Traces of the pattern can even be found as far away as the Slavic languages like Russian.
Conceivably, sometime in the distant past, before these languages split from the mother tongue, there was a close connection among the words for a hand and its fingers and the number five. But did the mathematical abstraction come from the word for fist, or, as some linguists have proposed, was it the other way around? The answer could provide a window into the development of the ancient mind.
In a paper now being prepared for publication in a book next year, Dr. Alexis Manaster Ramer, a linguist at Wayne State University in Detroit, argues that the mystery may now be solved: fist came before five. But more important than his conclusion is the method by which it was derived.
It is widely accepted that English, Dutch, German and Russian are each branches of the vast Indo-European language family, which includes the Germanic, Slavic, Romance, Celtic, Baltic, Indo-Iranian and other languages -- all descendants of more ancient languages like Greek, Latin and Sanskrit. Digging down another level, linguists have reconstructed the even earlier tongue from which all these languages are descended. They call it proto-Indo-European, or PIE for short.
But in a move sure to be hotly disputed by mainstream linguists, Dr. Manaster Ramer contends that to find the root of the fist-five connection one must look beyond the Indo-European family and examine two separate language groups: Uralic, which includes Finnish, Estonian and Hungarian, and Altaic, said to include Turkish and Mongolian languages. All three families, he contends, contain echoes of a lost ancient language called Nostratic.
If Dr. Manaster Ramer is right, his discovery will provide ammunition for a small group of linguists who make the controversial claim that Indo-European, Uralic, Altaic and other language families like Afro-Asiatic, which includes Arabic and Hebrew, the Kartvelian languages of the South Caucasus and the Dravidian languages concentrated in southern India, all are descendants of Nostratic, which was spoken more than 12,000 years ago.
Most language experts remain highly skeptical of the Nostratic hypothesis, which enjoyed so much publicity in the late 1980's and early 1990's that it is sometimes described as the linguists' version of cold fusion. "It would be terrific if it's true, but we don't want to jump to conclusions," said Dr. Brian Joseph, a linguist at Ohio State University in Columbus. Dr. Joseph and Dr. Joe Salmons of Purdue University in West Lafayette, Ind., are editing the book, " Nostratic: Evidence and Status" (John Benjamins), in which the analysis of the five-fist connection will appear.
But Dr. Joseph believes that while the Nostratic debate remains as heated as ever, it has reached a higher level of sophistication, with both sides offering more precise arguments and careful scholarship. "Mainstream linguists who in the past had dismissed Nostratic are now willing to examine it on an objective and scientific basis," he said. While he and Dr. Salmons both count themselves as skeptics, they hope their book will be a milestone in linguistic scholarship. "Even if the more mainstream linguists decide to reject Nostratic, " Dr. Joseph said, "at least the evidence will be laid out in a fair and balanced way."
It is not that most linguists find implausible the idea that all languages may ultimately have derived from an ancient ur-language spoken millenniums ago. After all, analysis of mitochondrial DNA from the cells of various ethnic groups strongly supports the notion that all humans come from the same genetic stock. If this small group of original humans spoke a single language, then all present-day languages are descended from it. The hypothetical Nostratic is not the ur-language but might be one of its major branches. However, critics of the Nostratic hypothesis have long argued that it is unprovable -- any similarities between languages as distant as the Altaic and Indo-European would have been washed out long ago. They dismiss the parallels unearthed by the Nostraticists as coincidences.
As recently as the early 1990's, most evidence for an ancient language relied on work done in the 1960's by Soviet scholars, who popularized the word Nostratic, meaning "our language." But now a second wave of research is revitalizing the field. Veterans of the Nostratic program like Dr. Vitaly Shevoroshkin of the University of Michigan in Ann Arbor and Dr. Aaron Dogopolsky of the University of Haifa in Israel continue to come up with new evidence, as do younger scholars like Dr. Manaster Ramer.
In a book published last year, "The Nostratic Macrofamily: A Study in Distant Linguistic Relationship" (Mouton de Gruyter), two independent scholars, Allan Bomhard and John Kerns, compiled some 600 Nostratic roots with counterparts (what the linguists call cognates) in languages said to be descended from Nostratic. On another front, Dr. Joseph Greenberg, a retired Stanford University linguist, is in the midst of a two-volume study of his own version of the Nostratic hypothesis: "Indo-European and Its Closest Relatives: The Eurasiatic Language Family." Dr. Greenberg's Eurasiatic overlaps with Nostratic but also includes other languages like Japanese and Eskimo-Aleut.
In an unpublished manuscript of yet another forthcoming book, "Indo-European and the Nostratic Hypothesis," Mr. Bomhard concludes that the evidence for the common ancestral language is "massive and persuasive." "As the 20th century draws to a close, it is simply no longer reasonable to hold to the view that Indo-European is a language isolate," he writes. "Indo-European has relatives and these must now be taken into consideration."
One of the most vehement critics of Nostratic, Dr. Donald Ringe, a linguist at the University of Pennsylvania, recently surprised himself by finding statistical evidence that resemblances between Uralic and Indo-European may indeed be due to more than chance. Dr. Ringe expected his analysis, which will also be published in the book edited by Dr. Joseph and Dr. Salmon, to undermine the Nostratic hypothesis.
But Dr. Ringe is quick to point out that a connection between Indo-European and the Uralic languages like Hungarian and Finnish is the least controversial claim of the Nostraticists. He remains as dubious as ever that statistically significant connections can be found between Indo-European and more distant languages.
In a paper called " ' Nostratic' and the Factor of Chance," published in the current issue of the journal Diachronica, Dr. Ringe examined a list of 205 cognates that the Russian linguist Vladislav Illich-Svitych found among six language families commonly said to have descended from Nostratic. He concluded that the similarities are indistinguishable from those that would have arisen by chance. As a test of his analytical technique, Dr. Ringe applied the same method to two Indo-European languages, which are known to be related, and found that the similarities there are indeed statistically significant.
"It is time to tighten up standards of evidence in historical linguistics," he concluded in his paper. "If we enforce rigor, the truth will enforce itself."
But some linguists believe Dr. Ringe is misinterpreting his own statistics. Dr. Manaster Ramer argues that Dr. Ringe, who has accused the Nostraticists of "innumeracy," is himself engaging in "pseudomathematics."
"To use mathematics in any science, including linguistics, you have to understand the meaning of the mathematics and not just learn to manipulate formulas," he said. Dr. Manaster Ramer believes that the probability distribution that Dr. Ringe found for Nostratic is exactly what would be expected in languages that split apart long ago and developed independently. The true test of whether languages are related is not statistical comparisons, he insists, but the tools of historical linguistic analysis. If one can find answers in Uralic and Altaic to puzzles in Indo-European, like the five-fist connection, he says, that strengthens the argument for an ancestral Nostratic tongue.
Historical linguists start with two languages they suspect are related, then search for potential cognates -- words like the Italian "luce" ("light") and "pace" ("peace"), which appear in Spanish as "luz" and "paz." Then, by deriving rules for how sounds mutate over time, they try to reconstruct the ancient roots.
In actual practice, the correspondences between related words are usually far more convoluted and opaque to superficial examination. English and Armenian both are believed to descend from proto-Indo-European. But it takes a great deal of linguistic manipulation to show how the Armenian word for two, "erku," is related to its English counterpart. To add to the confusion, words that seem similar can turn out to be unrelated. Linguists consider it coincidental that the German word for "awl" happens to be "ahle," or that the Aztec word for "well" is "huel." For that matter, the English word "ear," referring to the fleshy flaps on either side of the head, has been found to be historically unrelated to an "ear" of corn.
There are other mirages that can create the illusion of a deep historical wellspring. Baby words like "papa" and "mama" are common across languages probably because the labial consonants -- those made with the lips -- are among the first that children learn. Onomatopoeic words like "clash" or "meow" also tend to turn up independently in unrelated languages. And of course languages borrow words from one another all the time. A Japanese office worker can log off her "konpyuutaa" and head for "Makudonarudo" to grab a "hanbaagaa" and a steaming cup of "hotto kohii" for lunch.
To avoid being misled by such specious similarities, linguists try to concentrate on basic words -- numbers, parts of the body -- likely to have been embedded in a language from the start. As reconstructed by linguistic archeologists, the ancient Indo-European word for five was "penkwe," which became "pente" in Greek, "quinque" in Latin and "panca" in Sanskrit. One can immediately see surface similarities between "penkwe" and the Indo-European roots for fist, "pnkwstis" and finger "penkweros." But though the resonances ring, the source of the connection has remained obscure.
Finding few clues within Indo-European itself, Dr. Manaster Ramer looked farther afield. Linguists examining Finnish, Hungarian and Estonian had reconstructed an ancient Uralic root, "peyngo," meaning fist or palm of the hand. And from Turkish, Mongolian and related languages, linguists had reconstructed the corresponding word in Altaic: "p'aynga." (The accent is a sign that there were two different p sounds in the language.)
Working backward from Uralic and Altaic, Dr. Manaster Ramer reconstructed a hypothetical Nostratic antecedent, "payngo." Then, using what he believed to be the rules by which Nostratic mutated into proto-Indo-European, he showed how the Nostratic word for fist could have spawned the Indo-European word for five.
In another attempt to show that the Indo-European languages descended from Nostratic, Dr. Manaster Ramer analyzed the word "stink," which came into English from the hypothetical Germanic root stinkwan. Linguists find this word interesting because it appears to have no counterparts in other Indo-European languages. Dr. Manaster Ramer argues that it could have derived from a hypothetical Nostratic word, "stunga."
In his own work, Mr. Bomhard points to evidence that the first-person pronoun "me" and variations like "mi," "ma," "mo" and "mea" appear in PIE and in the reconstructed protolanguages Kartvelian, Afro-Asiatic, Uralic, Altaic and the extinct language Sumerian. Mr. Bomhard believes that ancestral Indo-Europeans said "bor" for "to bore" or "to pierce"; the Afro-Asiatics said "bar," the Altaics said "bur," the Sumerians "bur," and the Dravidians "pur," while the Uralics said "pura" for borer or auger. And while Indo-Europeans said "pes" or "pos" for penis, speakers of Altaic said "pusu" for "to squirt out" or "to pour" and the Sumerians said "pes" not only for sperm and semen but also for descendant, offspring and son.
Most linguists are leery of reading too much significance into reconstructions that are based on reconstructions. Are the Nostraticists excavating into the past or building a house of cards?
"The bottom line is that the evidence isn't good enough," Dr. Ringe said. "In particular, neither Manaster Ramer nor anyone else has demonstrated that the similarities they've found between the various recognized language families are due to anything other than chance."
With such different ideas about how Nostratic scholarship should proceed, it is unlikely that either Dr. Ringe or Dr. Manaster Ramer will come around to the other's point of view. In the meantime, linguists watching from the sidelines say there is a huge amount of work to be done before Nostratic can confidently be verified or rejected.
"I think there is a new appreciation of the level of sophistication one needs to approach the problem," said Dr. Brent Vine, a Princeton University classicist. "So much is now known about all the different language families involved that no one person can seriously claim to have the kind of control needed for Nostratic research. What is really needed is a team effort."
GRAPHIC: Chart: "Say What?"
The six language families shown as branches on this tree are believed to have originated in the Nostratic language (tree trunk), one of several major branches of a single hypothetical ancient "mother tongue" from which all languages are believed to be derived.
Map/Diagram: "Modern Children of Nostratic? "
One scheme of current language group distribution. (Source: Scientific American)