The claim that a language dies every two weeks is widely accepted. It has been repeated in the media, in the scientific literature and even in reports produced by the United Nations.
To determine whether the estimate is supported by facts, Université de Montréal demography and population sciences professor Michaël Boissoneault traced its origins and tested it against the latest data and models.
What is the basis for the claim?
“It goes back to the American linguist Michael Krauss,” who asserted in a 1992 article that 3,000 languages may be lost in the 21st century, as transmission between generations weakens, Boissoneault said. “At that time, there was growing awareness of the scale of language extinction, as more comprehensive language inventories were being compiled.”
According to Krauss, in the early 1990s roughly half of the world’s 6,000 languages were no longer being passed on to children. “He didn’t discuss the rate of decline, or even the disappearance of languages as such, but that statistic gave rise to the one we hear all the time,” said Boissoneault.
In 2000, British linguist David Crystal cited Krauss’ estimate in the introduction to his book Language Death.
He made a calculation based on the number of known languages at the time: if half of them die out over the coming century, that would come to an average of one language every two weeks over the entire period,” said Boissoneault. “But people took it as an estimate of the current rate.” This number was then widely cited, much to Crystal’s surprise.
Around 2020, researchers began to question the claim. Better counts of the world’s languages were becoming available. (Krauss’s forecast had been based on the languages known to him rather than the total number of languages in the world, which was imprecise at the time.)
“Two research teams independently compiled the existing data and concluded that a language was disappearing every six weeks or 13 weeks,” said Boissoneault.
Good news? Not exactly. “We’ve realized that while these estimates are based on real data, they have their own biases,” Boissoneault said.
In short, no one knows exactly how quickly languages are disappearing.
“If we count only confirmed extinctions, we’re omitting the loss of languages that were never catalogued and those that have yet to be confirmed,” Boissoneault pointed out. Therefore, the most recent projections are likely to be underestimates.
“As for the number of languages that are no longer being passed on to children, estimates vary but it appears to be closer to 20 per cent than 50 per cent,” he added. It is believed that a total of between 6,500 and 7,000 languages are spoken today.
Boissoneault became interested in this issue while he and his research team were looking into the demographic trajectories of languages—particularly those on the verge of extinction.
“Our hypothesis is that if we track languages over time, we will be able to apply a statistical model that corrects for unconfirmed and uncatalogued extinctions, allowing us to measure the rate of language extinction more accurately,” he said.
Boissoneault is now working on a database project. “The idea is to determine, beginning with data from the 1950s, when a language first became known, and then track changes in the number of speakers over time,” he said.
Over this past summer, he and his students examined the demographic trends of some 200 languages in North America. “We’re compiling all publications that mention the number of speakers of endangered languages,” he explained. “We have many data points, and they can be somewhat contradictory.”
Though estimates of the rate of language extinction are trending downward, the fact remains that languages are dying out, year after year., he noted.
“I think it’s important to correct certain misconceptions, and to do that, we need to have more comprehensive data and to analyze them with the right tools."
