Tongue-twisted long ago

The new analysis, reported in the most recent edition of the journal Nature, retraces the same ground as a controversial discipline called ‘‘glottochronology’’ — calculating dates of language divergence by examining cognates, or words that are similar in two or more languages.

‘‘It gave some obviously inaccurate results, and linguists gave up on it,’’ said evolutionary biologist Russell D. Gray, lead researcher of the new study. ‘‘I was interested in using more sophisticated computational methods to get at dates of divergence.’’

The study’s conclusions, however, have placed Gray and co-author Quentin D. Atkinson, both of New Zealand’s University of Auckland, squarely in the middle of one of the most contentious debates in modern linguistics: whether Indo-European originated with Anatolian farmers in Turkey 8,500 or 10,000 years ago; or, as many scholars believe, with Kurgan horsemen on the Asian steppe 6,000 years ago; or elsewhere.

‘‘The problem you face among Indo-Europeanists is that (building a root language) is hard,’’ said Ives Goddard, the Smithsonian Institution’s senior linguist. ‘‘Really smart people have been working on this for a very long time, and nobody believes you can get good results with just a word list anymore. Indo-Europeanists think this is junk science.’’

The quest to reconstruct an Indo-European mother tongue began in 1786, when the British scholar and linguist Sir William Jones noticed marked similarities between Sanskrit and ancient Greek and Latin.

Since then, scholars have expanded the Indo-European family to include scores of languages past and present, including those such as English, Spanish and Russian, which together are spoken today on every continent except Antarctica.

Gray and Atkinson analysed 87 languages in their study. Over time, historical linguists developed a painstaking ‘‘comparative method’’ to reconstruct the mother tongue by examining not only the words from Indo-European languages but also their grammar and morphology — how words are put together. ‘‘What the comparative method gives us is a partial reconstruction of the ancestor language — that’s all,’’ said Indo-Europeanist Donald Ringe of the University of Pennsylvania. ‘‘It tells you nothing about where and when the ancestor language was spoken, or any of the other historical questions that people in general want to ask.’’

But in the ’50s, linguist Morris Swadesh took a stab at it. First, he put together a core vocabulary of 100 to 200 ‘‘basic’’ words and used it to compare languages, rating them as more closely or less closely related depending on their percentage of cognates. ‘‘This was fairly controversial by itself,’’ said computational biologist David B. Searls, who wrote a Nature commentary to accompany Gray and Atkinson’s report. ‘‘But then Swadesh pushed his luck by developing glottochronology. We know that languages tend to undergo word substitution over time, but if the rate of change is constant, Swadesh thought he could date languages. This was really controversial.’’ And quickly discarded by a majority of linguists, who demonstrated that rates of change can vary greatly. Also, conclusions reached through glottochronology often did not square with the archaeological record.

Gray, however, said his analysis overcomes these problems because it uses sophisticated statistical techniques ‘‘that can infer family trees, genealogies and even dates of change’’ with a rigour never contemplated by Swadesh or his detractors. Results are recalibrated by inserting the dates of known events and recomputing.

The team concluded that Indo-European originated 7,800 to 9,800 years ago, a result that closely supports the Anatolian hypothesis, which holds that Indo-European began to spread with agriculture as farming moved over the world from Turkey beginning 8,500 to 10,000 years ago.

The difficulty, however, is that most linguists discard the Anatolian hypothesis in favour of the view that Kurgan horsemen from the Asian steppe spoke Indo-European and spread the language during raids 6,000 years ago. Scholars note that Indo-European has words such as ‘‘wheel’’, a concept that did not exist in prehistoric Anatolia.

More important, however, most linguists ‘‘have a good seat-of-the-pants guess on how old Indo-European is, and if you made it 3,000 years earlier, we would notice that,’’ Goddard said. ‘‘But nothing happens during that time gap. If we’re going to use computational techniques, we can’t ignore everything we know.’’ (LA Times-Washington Post)