In school we learn that science proceeds logically from one experiment to the next, leaving in its wake a complete and certain body of knowledge.
But science isn’t like that. It twists and turns, careens and tumbles and gets stuck in deep, sticky mudholes. And sometimes, science backtracks.
That’s happened in cosmology recently, as observations of the universe’s accelerating expansion have forced theorists to go back and restore a notion — the cosmological constant — that Einstein abandoned by the scientific roadside eight decades ago. Now, it looks like something similar may happen in the study of how people arrived in the Americas.
In 1987, a maverick linguist named Joseph Greenberg published a book called Language in the Americas. In it, he used a very unfashionable method to classify the New World’s indigenous tongues. While his colleagues meticulously compared pairs or small groups of languages side by side, analyzing grammar, phonetics and meaning in great detail, Greenberg took a Reader’s Digest approach. He made lists of a few hundred common words that any language should have — pronouns like I and me and we, nouns for natural objects, like water, and verbs like die. Then he made lists of those words from hundreds of Native American languages and started grouping those lists according to phonetic similarity.
It was a pragmatic approach to a problem that more traditional research had failed to solve: How many migrations from Asia to North America produced the diverse collection of people who inhabited the Americas when Europeans arrived, and when did that migration (or migrations) occur?
Archaeological evidence suggested a single migration perhaps 15,000 years ago. But if every group in the Americas, from the Alaskan Inuit to the Yaghan of southern Chile, diverged from one group of people speaking a single tongue, how could their languages have diverged so quickly into the sprawling patchwork of families that traditional linguists describe?
Greenberg argued that they didn’t. There were three language families in the Americas, he argued, not three dozen. The dominant one, Amerind, was spoken over all of Central and South America, and most of North America as well. Another, Na-Dene, was spoken by groups in western Canada and along the northern Pacific coast, along with the Navajo and Apache of the U.S. Southwest. The third, Eskimo-Aleut, encompassed the languages of people living in the Arctic and the Aleutian Islands of Alaska. By Greenberg’s estimate, these three groups had arrived in migrations dating back more than 11,000 (Amerind), 9,000 (Na-Dene) and 4,000 years (Eskimo-Aleut).
Predictably, Greenberg’s linguistic critics were livid. But their objections mostly came down to basic principles about how languages evolve. Not much could be done to determine who had the more valid point of view.
Then geneticists began using DNA in the male sex chromosome, handed down from fathers to their sons, and in mitochondria, inherited through the maternal line, to trace the ancestry of groups around the world.
At first, the mitochondrial studies seemed to support Greenberg’s scenario. But by the turn of the century, further mitochondrial studies pointed to a single migration of no more than a few thousand people about 15,000 years ago.
Now genetics, a field that’s zipping forward as if it were a Lamborghini on a lonesome desert highway driven by a man who just robbed a bank, has provided newer tools. Those tools can sample hundreds of thousands of points of common variation throughout the genome, not just in the all-female and all-male lines probed by the mitochondrial and Y chromosome studies. A large international team has used these tools to analyze samples contributed by 493 people from 52 Native American populations. Reporting their results online July 11 in Nature, these researchers conclude that the populations fall into three clusters corresponding to the ones originally proposed by Greenberg.
Why did earlier studies using mitochondrial and Y chromosome DNA fail to resolve these three migrations? It was probably because after arriving in North America, the Eskimo-Aleut and Na-Dene migrations soon mixed with the original Amerind settlers of the continent. In genetic terms, the Nature paper estimates, today’s Eskimo-Aleut are 57 percent Amerind; the figure for Na-Dene is a whopping 90 percent.
This is by no means the final word. The long and contentious history of research into the peopling of the Americas is not about to end just because of one study. It will take much more research by geneticists and researchers in many other fields to test the claims of this latest paper and carry the research farther down the road. Which may very well lead into a blind canyon, back out the way it came, down a bumpy wash and up a series of steep switchbacks to a place nobody ever thought existed.