Every few weeks, a language dies
Linguists estimate that on average a language goes extinct every two weeks. Since spoken languages do not have archaeological records, it is impossible to have a true count of languages that have ever existed. At the low end, scholars estimate that roughly 31,000 languages have existed1 across modern human history. Today, reference catalogs like Ethnologue or Glottolog document approximately 7,100 living languages. Between 1950 and the present, hundreds of languages ceased being spoken and 50% to 90% of current living languages could go extinct or fall silent by 2100.
I wrote about how relentless advances in travel and communication have slowly elided cultural distinctions and moved us to a homogenized world. Languages are not just alternate lexicons for the same universal concepts. Languages shape the way we think and the way we see the world. Every language encodes a distinct mental model and classification of reality. A language is said to be dead when no one speaks it any more. With the death of a language the world loses a unique perspective, we become less interesting as a species. If stories and shared myths allow accumulated wisdom to transcend biological mortality, the death of a language deprives us of a few rungs up that ladder.
In highly diverse regions such as the Indian subcontinent, Nigeria or Papua New Guinea2, what are designated as single languages often cover dozens or hundreds of mutually distinct regional dialects (as a famous linguist once said, a language is nothing but a dialect with an army and a navy). Nobody is even counting the loss of those dialects. There is a loss of beauty when those dialects all fade into one sanitized TV variant of that language. I mourn the fact that I have not fully inherited all the unique idioms, metaphors and ancestral mythologies shaped over hundreds of generations and particular to the dialect of Malayalam3 I grew up with, much less passed them on to my kids. Did you know that higher road density and increased years of formal non-bilingual schooling are the top predictors of language endangerment?
This dire sounding situation does not have to doom us to fatalism. This recent announcement points a way out: the Gates Foundation and a global coalition of partners have launched a joint commitment to expand AI access to approximately 3.4 billion speakers of under-represented and low-resource languages over the next five years. To bridge this linguistic divide, the initiative focuses on creating open-source language data infrastructure, developing shared benchmarks, building locally relevant tools, and applying responsible data practices. In the midst of general uneasiness about AI, ventures such as these point to unexplored but critical problems.
If languages and linguistics interest you, I highly recommend Lingthusiasm., a podcast Gretchen McCulloch and Lauren Gawne. In a culture that seems increasingly suspicious of expertise, listening to two researchers who remain genuinely enthusiastic about their discipline is a breath of fresh air. Give this recent episode a listen if you want to get a different aspect of the loss of languages the world over.
1 https://catdir.loc.gov/catdir/samples/cam032/99053220.pdf
2 The Greater Pacific area comprises Japan, Taiwan (China), the Philippines, insular Malaysia, Indonesia, Papua New Guinea, the Solomon Islands,Vanuatu, Fiji, Micronesia, Polynesia, Australia and New Zealand, as well as Hawaii. Over 2,000 living languages, about a third of all those in the world, are located in this area. Papua New Guinea has nearly 500 of them.
3 Fun fact - Malayalam branched off from Tamil, the world’s oldest living language, around the 9th century. It started off as a western dialect of Middle Tamil. Heavy influence of Sanskrit driven by Brahmanical culture and geography shaped the modern version. Malayalam dialects are different enough that the Thalassery dialect that I grew up with has phrases and words unintelligible to some of my friends from other parts of Kerala. Some other languages such as Greek and Chinese can make similar claims to being the oldest but Tamil has the distinction that you can still read the ancient texts with relative ease, which is not something you can say about any other language.
