Category: Research & Readings
Keywords:

Statistics and Folklore

Despite popular and persistent statistics, we really do use more than 10% of our brains. We don’t actually eat an average of seven spiders a year, and suicide rates simply do not jump during the holidays. When statistics become sound bites, they sometimes become folklore.

languageOne such statistic I see periodically in my research is the one that reads something like this: “About half of the estimated 7,000 languages still spoken around the world will disappear by the end of this century.” This assertion is usually coupled with a statement to the effect that “every two weeks a language dies.”

I first heard this spoken by Wade Davis in a recording of his 2003 TED talk: “And the great indicator of that, of course, is language loss. When each of you in this room were born, there were 6,000 languages spoken on the planet… And of those 6,000 languages, as we sit here today in Monterray, fully half are no longer being whispered into the ears of children. They are no longer being taught to babies. Which means, effectively, unless something changes, they’re already dead… every two weeks, some elder dies and carries with them into grave the last syllables of an ancient tongue.”

You can see variations of it across the web:

A fact came out of MIT, couple of years ago. Ken Hale, who’s a linguist, said that of the 6,000 languages spoken on Earth right now, 3,000 aren’t spoken by the children. So that in one generation, we’re going to halve our cultural diversity. He went on to say that every two weeks, an elder goes to the grave carrying the last spoken word of that culture.www.ted.com/talks/phil_borges...
One language dies every 14 days. By the next century nearly half of the roughly 7,000 languages spoken on Earth will likely disappear, as communities abandon native tongues in favor of English, Mandarin, or Spanish. What is lost when a language goes silent?ngm.nationalgeographic.com
The general consensus is that there are between 6000 and 7000 languages currently spoken, and that between 50-90% of those will have become extinct by the year 2100.en.wikipedia.org/wiki/Endangered_language
It is estimated that, if nothing is done, half of 6000 plus languages spoken today will disappear by the end of this century.www.unesco.org
At least 3,000 of the world’s 6,000-7,000 languages (about 50 percent) are about to be lost.www.nsf.gov
Indeed, a widely accepted estimate is that at least 50% of the currently spoken languages in the world are endangered and likely to disappear within the 21st century.Introduction to Beliefs and Ideologies...

Variations on a theme

Sometimes the attribution goes to UNESCO, sometimes to MIT (specifically Ken Hale), or to Michael Krauss from his 1991 presentation to the Linguists Society of America. Sometimes the number of total languages is 6,000 – and sometimes 7,000. The percentage of extinction ranges between 50% and 90% – and the countdown typically ends at the end of this century (or more generally, the end of our lifetime or within a few generations). The addendum factoid is usually either a language – or an elder who is the last speaker of a language – dies every two weeks.

If we run the math in 2014, here’s what we get: 2100 – 2014 = 86 years until the end of the century. 86 years x 52 weeks = 4,472 total weeks. 4,472 / 2 = 2,236 two week periods, hence 2,236 languages dying between now and the end of the century. The difference between 2,236 and 3,000 (or 50% of 6,000) suggests a few things. Firstly, assuming we started with 6,000 languages, 764 languages have already disappeared since the original assertion was made (and it would have to have been made around 1985 for the math to work right). Secondly, if the stats are meant to be illustrative rather than accurate, we can safely leave them in the realm of folklore (along with statements like “half of all marriages end in divorce”). If we seek truth in our facts, we should probably update the stats based on current research.

It would be instructive to know the metrics behind such claims. Often times, the claims are made at the beginning of articles or blog posts about how technology or social media are saving some Indigenous language. The facts are meant to present an “establishing shot” – to paint a general picture of the current state of affairs concerning language loss in the world. But since the articles aren’t about statistical methodologies or computational linguistics, they aren’t going to delve into a detailed analysis of the claim.

Because there are several variations to the claims, a few possible sources, and some questionable mathematics to boot, I wanted to find some current statistics as well as information about how language loss is measured so that the statistics are situated within a meaningful context.

New Estimates on the Rate of Global Language Loss

In March of 2013, Karin Wiecha (who was then an intern at The Rosetta Project, an initiative of The Long Now Foundation) wrote a blog post titled New Estimates on the Rate of Global Language Loss. In it she highlights two projects (and subsequently two scales for measuring language loss): The Catalogue of Endangered Languages (ELCat) and Ethnologue.

The Catalogue of Endangered Languages (ELCat)
ELCat is an academic project funded by the US National Science Foundation. It has the goal of updating and correcting status information about the endangered languages of the world. ELCat is a 3-year project begun in 2011 (download this PDF for detailed information).

Listen to “New knowledge: Findings from the Catalogue of Endangered Languages,” a 2013 presentation by Lyle Campbell, director of the ELCat project:

Highlights from the Research

The data Lyle Campbell cites in the audio piece above and in the excerpts below are based on the ELCat’s Scale of Endangerment – a rubric for assessing the status of languages.

See ELCat’s Scale of Endangerment

Hard evidence from the Catalogue shows that 3054 languages are currently endangered (43% of all languages), based on precise criteria. This 43% is near to the oft-cited 50% (but far from the 90% worst-case) scenario of languages expected to become extinct or doomed by the century’s end.

Of all known named languages, 634 have become extinct, 141 of these (22%) in recent times (in the last 40 years). This concrete evidence demonstrates that the rate of language extinction has become much more highly accelerated in recent times, as often claimed.

All the languages of exactly 100 families, including isolates, have become extinct, from among the world’s 420 language families –24% of linguistic diversity has been lost forever. This confirms the common claim of significant loss of language diversity.

There are 335 languages with fewer than 10 speakers (11% of all endangered languages).

The very frequently repeated claim that one language goes extinct each 2 weeks is not supported by the findings; rather, the Catalogue of Endangered Languages finds that on average only 3.5 languages become extinct per year, i.e. about 1 each 4 months, true now and for any time span during the last 40 years. Though not as dramatic as the oft-cited claim, loss of 1 language every 4 months is tragic, with its irreparable damage and loss. There is no need to continue to repeat the inaccurate claim, which ultimately could have negative repercussions for our field — the number is still shocking.

Visit the Endangered Languages website (the public portal of ELCat)

endangered-languages

Ethnologue: Languages of the World
Ethnologue: Languages of the World is a comprehensive reference work cataloging all of the world’s known living languages. Since 1951, the Ethnologue has been an active research project involving hundreds of linguists and other researchers around the world. It is widely regarded to be the most comprehensive source of information of its kind.

Highlights from the Research

Details from the most recent edition of Ethnologue can be found in The world’s languages in crisis: A 20-year update, a 2013 paper by Gary F. Simons and M. Paul Lewis of SIL International (see excerpts below).

Ethnologue uses the Expanded Graded Intergenerational Disruption Scale (EGIDS) to assess the status of languages.

See Ethnologue’s EGIDS

Globally, 2,503 of the languages of the world are characterized by vigorous oral use. When the count for EGIDS level 6a is combined with the languages at higher, stronger levels (EGIDS 0–5), we see that 4,719 (63%) of the 7,480 languages in use in 1950 are still being passed on to the next generation in a sustainable way. In the discussion which follows, we refer to this group of languages as “vital” languages. In contrast, 1,480 (20%) of the languages of the world are “in trouble” (EGIDS 6b–7). In these languages the norm of complete intergenerational transmission is no longer in effect, but members of the child-bearing generation are still fully proficient in the language so that it would still be possible for a successful revitalization effort to restore intergenerational transmission. Finally, an additional 1,281 (17%) of languages are “dead or dying” (EGIDS 8a–10) since it is too late to restore natural parent-to-child transmission. The restoration of intergenerational transmission would require establishing overt language transmission mechanisms outside the home.

Among the dead and dying languages are 377 (5%) that have been identified as having lost all living speakers and ceasing to serve as a language of identity for an ethnic community (EGIDS 10) in the last six decades. The loss of linguistic diversity represented by the loss of these individual languages is even more staggering if viewed from the perspective of language families. Whalen & Simons (2012) show that with the loss of these languages, we have lost 15% of the linguistic stocks (the largest subgroups of related languages that are reconstructable) that had at least one living member in 1950.

Alarmingly, 2,384 (32%) living languages in the world are currently at some stage in the process of language loss (EGIDS 6b–9). That is more than the number of languages (2,216, 30%) that have experienced enough language development (EGIDS 0–5) to rise above the default stage of vigorous oral use (EGIDS 6a).

Visit the Ethnologue website

ethnologue

Parting shots

Following the old adage that people remember the first and and last things you tell them in a conversation, I want to repeat Lyle Campbell’s statement from above. Since the stats have been recast in the light of new research, perhaps we should avoid the old and now misleading refrains:

The very frequently repeated claim that one language goes extinct each 2 weeks is not supported by the findings; rather, the Catalogue of Endangered Languages finds that on average only 3.5 languages become extinct per year, i.e. about 1 each 4 months, true now and for any time span during the last 40 years. Though not as dramatic as the oft-cited claim, loss of 1 language every 4 months is tragic, with its irreparable damage and loss. There is no need to continue to repeat the inaccurate claim, which ultimately could have negative repercussions for our field — the number is still shocking