Well, notice more subtleties in the article you shared.
I think it would depend on how many equivalences/cognates you have, whether the theory is worth pursuing or not. Some relationships seem tenuous to me. Read Babel, by Abraham Abhesera (mentioned earlier in the thread) for more possible explanations of why these equivalences exist. Noy saying that there couldn't be a common ancestor, of course, just that it may be more complex than it is made out to be.
Identifying cognates has helped linguists categorise modern languages into families that evolved from the same protolanguage. English, Swedish and Farsi are all part of the Indo-European language family that is thought to descend from proto-Indo-European, whereas Finnish and Hungarian are thought to descend from proto-Uralic.
But could most of the protolanguages on the Eurasian continent and even across the Bering Strait into Alaska be part of a superfamily that descended from a more ancient mother tongue – proto-Eurasian (see map)?
This idea has been knocking around for almost a century, says Mark Pagel at the University of Reading in the UK. “It’s kind of an obvious idea – Eurasia is this contiguous landmass, and similar proposals have been made about the protolanguages of Australia and North America.”
However, the existence of superfamilies has, until now, been extremely hard to prove. Changes in languages pile up over time, so the further back you go, the fewer cognates there are to compare. What’s more, similarities can arise purely by chance. Many researchers do not believe languages can be traced back further than about 8000 years.
“That’s the argument that there’s a limit on the time for which we can go back in historical linguistics,” says Quentin Atkinson from the University of Auckland in New Zealand.
But now Pagel, Atkinson and colleagues have smashed through that limit, travelling further back in time than linguists have ever done before. They say that not only do their results suggest that the Eurasian superfamily exists, but that they have also been able to mathematically deduce its evolution.
The key to the breakthrough was the team’s discovery in 2007 that the form of frequently used words evolves at a much slower pace than less common words. These conserved words, the team reasoned, were more likely to retain traces of their ancestry than words that evolved rapidly.
They then used this idea to predict which words in a database of Eurasian protolanguages were cognates, removing words that looked and sounded similar but were not high-frequency words. This eliminated fluke similarities.
The team then ran a list of these cognates – such as those meaning “I”, “mother”, “hand” and “fire” – through a statistical model that deduced the relationship between the words based on how quickly they changed with time. This gave a tree of the Eurasian superfamily whose common ancestor can be traced back 15,000 years.
“We found it remarkable that we got this 15,000-year result because it coincides beautifully with the retreat of the ice sheets after the last glacial maximum,” says Pagel. “One realistic scenario is that the tree represents the expansion of human populations as the climate improved and more people could be supported.”
Luisa Miceli, a historical linguist at the University of Western Australia in Perth, agrees that the result proves that the languages are closely connected, but she suggests they might be neighbours rather than siblings – with the words borrowed from one another rather than being true cognates.
However, she says the interesting questions about what factors were associated with the spread of these languages do not depend on whether they were part of a true superfamily or simply sets of languages that were in contact.
I think it would depend on how many equivalences/cognates you have, whether the theory is worth pursuing or not. Some relationships seem tenuous to me. Read Babel, by Abraham Abhesera (mentioned earlier in the thread) for more possible explanations of why these equivalences exist. Noy saying that there couldn't be a common ancestor, of course, just that it may be more complex than it is made out to be.