So Many, So Few.

Back in 2017, we discussed Michael Gavin’s attempt to answer the questions: “Why is it that humans speak so many languages? And why are they so unevenly spread across the planet?” At that time he and his coworkers were investigating language diversity patterns in Australia; last year there was an update by Gavin and Marco Túlio Pacheco Coelho, “Why are so many languages spoken in some places and so few in others?

Our research team has been tackling this longstanding question by exploring language diversity patterns on the continent of North America. Prior to European contact, North America was home to speakers of around 400 languages, unevenly spread across the landscape. Some places, such as the West Coast from present-day Vancouver to southern California, had far more languages; other areas, such as northern Canada and the Mississippi delta region, appear to have had fewer languages. We drew on methods from ecology originally developed to study patterns of species diversity to investigate these patterns of language diversity. […]

Recently, our interdisciplinary research group tried to untangle which factors had the most influence on language diversity in different places. Combining ideas from linguists, ecologists, evolutionary biologists and geographers, we took a unique approach. We used statistical techniques to estimate how the effects of environmental and sociocultural factors on language diversity changed from one location to another. In our study, each location was represented by a 300 km² grid cell, as is visible in all our maps.

We found that the most important variables associated with language diversity varied from one part of North America to another.

For example, on the West Coast, we found that variability in temperature over time is a key driver linked to language diversity. This result provides some support for the idea that in areas with more stable environmental conditions, human social networks can be smaller and more languages may exist. However, in the eastern part of the continent, potential population density tends to be the factor most strongly linked to language diversity.

We also found that in some places, such as the high-language-diversity regions on the West Coast, our model could predict the number of languages present very accurately, whereas in other areas, such as the Gulf Coast of the U.S., we have limited understanding of what drove language diversification.

Maps and further details at the link; thanks, jack!


  1. David Eddyshaw says

    Nice to see work of this kind that explicitly doesn’t claim to have found the One True Answer.

  2. Yes, I felt that way too.

  3. David Marjanović says

    That’s the way scientists write about their work, at least when they don’t feel enormous pressure to advance their career.

  4. Do we actually know what was the spread of languages in America “Prior to European Contact”? How would we know: weren’t Europeans busy shooting/infecting the first peoples rather more than recording their languages?

    Wouldn’t it be in colonialists’ interests to downplay language/tribal diversity, in order to herd everybody from some particularly valuable bunch of 300 km2 grid cells into a cramped reserve?

    In fact the maps show least diversity in the places the Europeans settled first; and less in the frozen/unpopulated North: don’t need no fancy theory to explain that.

  5. AntC brings up a good point: if you look at maps such as this one-

    (Incidentally, it is quite inaccurate historically, inasmuch as it represents the locations of languages at the time of early European contact: meaning that the languages of the Pacific coast are shown on the basis of their distribution in the nineteenth century, whereas the distribution of those on the Atlantic and Gulf coasts reflects their distribution a few centuries earlier).

    -where the white parts indicate either uninhabited areas or areas whose languages (before European colonization) are unknown, the gaps in our knowledge are very substantial. It seems to me that Africa would be a much better continent for the purpose of determining what factors favor/disfavor linguistic diversity, since European colonization was far less of a demographic catastrophe for the continent and, thus, the pre-European distribution of languages is far easier to map.

  6. David Eddyshaw says

    European colonization was far less of a demographic catastrophe

    For much the same reason as the Martians lost, for all their Heat Rays and Fighting Machines.

  7. Eli Nelson says

    I don’t know how it affected the distribution of languages, but I just learned recently about a catastrophe that did affect Africa during the period of colonialization. In the late 19th century, rinderpest, which seems to have been introduced from Eurasia, killed large amounts of cattle and contributed to famine and depopulation. Its effects are also supposed to have created conditions for the spread of the tsetse fly. (“Inventing Africa“, New Scientist vol 167 issue 2251 – 12 August 2000, page 30)

  8. John Cowan says

    I wonder if the low diversity along the Gulf Coast has to do with the Missisippian empire there.

  9. Low diversity is evidence of previous linguistic expansions.

    They can occur for all kinds of reasons, not necessarily introduction of corn farming (though it likely had the largest impact).

    Woodland Algonquian expansion was linked to invention of birchbark canoes, believe it or not.

  10. Finländare says

    Wouldn’t it also make sense that the Western seaboard is (was) so diverse because it’s likely to have been inhabited much longer than the other parts of North America? Native American peoples did at first colonize* the Americas moving westwards and southwards after all.

    *Please don’t hang, draw and quarter me for using that word

  11. Well, except that it’s over 10,000 years in any case, unless I misunderstand the spread of humanity across the continent, which is plenty of time for great diversity to develop everywhere. If I’m wrong, and Native Americans didn’t get to the northeast until shortly before the Vikings showed up, go ahead and hang, draw and quarter me.

  12. January First-of-May says

    and Native Americans didn’t get to the northeast until shortly before the Vikings showed up

    How far northeast? IIRC, some extreme northeastern areas have apparently seen a cycle of previous waves of settlement dying out before the arrival of new ones; indeed, the people that the Vikings interacted with in Vinland were the remnants of one such previous wave (and died out within two or three centuries).

    That said, I don’t believe this applies to less northeastern places, such as New England (but it’s not like I’ve looked this up).

  13. That rinderpeste story is a shocker. I had known the South African part (the word is Africaans), but that’s only the tail (tale) end of it.

  14. Yeah, I just mentioned the Vikings for the sake of the (mild) snark; I was thinking primarily of New England. But I don’t know what I’m talking about, as is often the case.

  15. New England had Algonquian population which is of recent western origin (offered dates for Algonquian expansion vary from first millennium AD to as late as 12-13th century AD).

    Pre-Algonquian population (so called Maritime Archaic tradition) might have been Iroquoian speakers (or rather, Iroquoian peoples were originally part of the Maritime Archaic along with several other extinct cultures, eg the Beothuk of Newfoundland).

    Maritime Archaic is thought to have been the first human inhabitants of the area (came right after the glaciers melted).

  16. Thanks!

  17. Thanks @Etienne. I notice that map is ‘language families’. How diverse are the languages within (say) the Algic family? I see wp says “The other [than Algonquian] Algic languages are the Yurok and Wiyot of northwestern California, which, despite their geographic proximity, are not closely related.” Algonquian around 30 languages.

    The paper’s sources for their map are behind a paywall, so I can’t see whether they’re mapping individual languages or families. Are they committing some misunderstanding similar to saying there’s little language diversity across SW Europe, because they’re all Romance family?

    Did the Mississippi River delta region have lower diversity, as per the paper’s abstract? Looks pretty crowded to me from the map you give. There’s a big chunk there classified ‘unknown’, as you warn. But the paper’s map has the whole lot coloured in pastels. They didn’t do something stupid like saying all the ‘unknown’ areas speak the same language?

    (Ah, the wiki warns that the map you give is seriously out of date: its data is from 1960s-70s. The paper draws its data from 1996.)

  18. Antc: In answer to your questions: Yes, they are mapping languages families rather than individual languages (well, unless said individual language is a known or even suspected isolate). As for Algonquian, it does seem comparable to Romance or Germanic in terms of diversity. The other two Algic languages (Wiyot and Yurok) seem not to be more closely related to one another than either is to Proto-Algonquian (Despite the use of the label “Ritwan” to designate a subfamily consisting of them to the exclusion of Algonquian: historically the term was created to indicate their genetic relationship before the relationship with Algonquian was discovered: hence the name “Algonquian-Ritwan” as a synonym for “Algic”), and indeed I would not be surprised if some scholar proved that one of them was more closely related to Proto-Algonquian than to the other “Ritwan” language.

    Within Algonquian the geographically more peripheral languages/language groups (Blackfoot, Cheyenne, Arapaho, the Cree dialect continuum, the Eastern Algonquian subfamily) are sharply differentiated from one another as well as from the geographically more central Algonquian languages (thus, Cree and Ojibwe have influenced one another greatly, but it is always clear whether we are dealing with Cree-influenced Ojibwe or Ojibwe-influenced Cree. The situation is often far less clear-cut when dealing with pairs of more geographically central Algonquian languages). Some of this differentiation may be due to different Algonquian languages having been brought into contact with one another at a comparatively recent date: for instance, Blackfoot and Cree are sharply unlike one another, and part of the reason for this may be because Cree expanded into Western Canada at a very recent date (the process began just before European colonization, apparently), possibly eliminating whatever Algonquian language(s) may once have formed a dialect continuum linking Blackfoot to some other Algonquian language(s).

Speak Your Mind