Differing Only by Language.

Via Far Outliers, a passage from A Death in the Rainforest: How a Language and a Way of Life Came to an End in Papua New Guinea, by Don Kulick (Algonquin Books, 2019):

As far back as anyone in Gapun has been able to remember, though, Tayap has never had more than, at most, about 150 speakers: the entire population of Tayap speakers, when the language was at its peak, would have fit into a single New York City subway car. Tiny as that count is, such a small language was not unusual for Papua New Guinea. Most languages spoken in the country have fewer than three thousand speakers. And linguists estimate that about 35 percent of the languages (which means about 350 of them) have never had more than about five hundred speakers.

Contrary to received wisdom, and common sense, this constellation of tiny languages was not the result of isolation; it didn’t arise because villages were separated from one another by mountain barriers or impenetrable jungle walls. Quite the opposite: throughout Papua New Guinea, the areas that have the highest degree of linguistic diversity (that is, the most languages) are the ones where people can get around relatively easily, by paddling a canoe along rivers and creeks, for example. The areas where travel is more difficult, for example in the mountains that run like a jagged spine across the center of the country, is where the largest languages are found (the biggest being a language called Enga, with over two hundred thousand speakers).

The conclusion that linguists have drawn from this counterintuitive distribution of languages is that people in Papua New Guinea have used language as a way of differentiating themselves from one another. Whereas other people throughout the world have come to use religion or food habits or clothing styles to distinguish themselves as a specific group of people in relation to outsiders, Papua New Guineans came to achieve similar results through language. People wanted to be different from their neighbors, and the way they made themselves different was to diverge linguistically.

Large swathes of neighboring groups throughout the mainland share similar traditional beliefs about what happens after one dies; they think related things about sorcery, initiation rituals, and ancestor worship; they have roughly similar myths about how they all originated; and before white colonists started coming to the country in the mid-1800s, they all dressed fairly similarly (and they all do still dress similarly, given the severely limited variety of manufactured clothing available to them today—mostly T-shirts and cloth shorts for men, and for women, baggy, Mother Hubbard–style “meri blouses” introduced by missionaries to promote modesty and cover up brazenly exposed breasts). Neighboring peoples hunt the same pigs and cassowaries that inhabit the rainforest; and they all eat sago, or taro or sweet potato—whichever of those staples their land is capable of growing.

In terms of the languages they speak, though, Papua New Guineans are very different from one another.

We discussed Kulick in 2019.


  1. David Eddyshaw says

    This reminds me of Halliday’s concept of “Anti-languages”:


    The reference to Bangime there is also interesting, though Bangime differs from Dogon languages in much more than vocabulary alone, so the analogy is not all that persuasive. Moreover, the Bangande, in line with their belief that they are, in fact, Dogon, seem to be under the impression that the actual Dogon understand Bangime, so if Bangime in fact functions as an anti-language, that would be nothing to do with the intentions of the speakers themselves.

  2. David Marjanović says

    Wikipedia: “An Enga-based pidgin is used by speakers of Arafundi languages.”

  3. I don’t remember if it was Papua New Guinea, but it was somewhere in the general vicinity, there were tribes who had a taboo on saying the name of someone who had died. And not just their name, but any word that sounded like it as well. So whenever someone died, they would have to come up with substitutes for a certain number of words in their language.

    Some anthropologist came in and collected a word list. It was a few decades until the next anthropologist showed up, only to discover that a substantial part of the first word list had gone obsolete.

    It’s a process that could only operate in a small group of people, those who actually knew the dead person. So a population of 100 or so would be about right.

    I really know very little about PNG, so I don’t know if this happens there. But it demonstrates that there are mechanisms that could generate a large number of extremely divergent small languages in a relatively short time.

  4. Stu Clayton says

    People wanted to be different from their neighbors, and the way they made themselves different was to diverge linguistically.

    Well, the Schinarites had more gumption than to start off like that, but had it imposed on them after the Tower of Babel incident. The Big Bible Bully wasn’t taking any risks. In diversity is weakness.

    1. Mose 6-7: # Und der Herr sprach: Siehe, es ist einerlei Volk und einerlei Sprache unter ihnen allen und dies ist der Anfang ihres Tuns; nun wird ihnen nichts mehr verwehrt werden können von allem, was sie sich vorgenommen haben zu tun. Wohlauf, lasst uns herniederfahren und dort ihre Sprache verwirren, dass keiner des andern Sprache verstehe! #

    “Lasst uns herniederfahren” is great. “Get the Cadillac, I’m gonna have to drive down to the city and sort these peons out”.

  5. The modern nation-state is the implacable enemy of diversity.

  6. *raises fist of anarchism*

  7. January First-of-May says

    I don’t remember if it was Papua New Guinea, but it was somewhere in the general vicinity, there were tribes who had a taboo on saying the name of someone who had died. And not just their name, but any word that sounded like it as well. So whenever someone died, they would have to come up with substitutes for a certain number of words in their language.

    I’ve also heard very similar stories about Australia, the Amazon, and one of the Greenlandic dialects (East, IIRC). I’m not sure which language it actually happened in, if any, but it does sound like the kind of thing that could have been independently derived in multiple places.

  8. That’s been discussed here before, but I forget where.

  9. Tonkawa has been proposed as a case in point.

  10. marie-lucie says

    Tabooing name of dead person leads to language change

    I think this is more prevalent than in just one small tribe.

    But what is changed is part of the vocabulary, not the morphology or grammar of the language in question, so in classifying the local languages, or comparing them for possible historical work, the vocabularies are relatively useless, what is important is the structural skeleton.

    The result is comparable to the addition of new slang words to a language, enabling people with reasons to avoid being overheard to carry on a conversation next to others speaking normally, without being really understood. The hearers recognize that the language is their own, because pronouns, auxiliary verbs, and some common nouns, verbs, etc are mostly there, but they have little idea of the conversation topic as the keywords have been replaced. If the new words get adopted by the general population, they have to be replaced by new slang in the vocabulary of the secretive group.

    The “new” words do not have to be completely invented: words from a different language not widely known among the population will do too.

  11. David Eddyshaw says

    That’s been discussed here before, but I forget where.


    I believe it is indeed the reason why East and West Greenlandic are surprisingly divergent, considering that the latest Inuit settlement in Greenland seems to go back only about a millennium (several previous populations having died out, including, of course, the Norse.)

    IIRC, Dixon invokes this as one mechanism whereby Australian languages have maintained their distinctness despite (in his view) long periods of equilibrium during which they have come to resemble each other more and more closely by contact.

  12. David Eddyshaw says

    Dixon’s claim is referenced – and criticised – here (section 4)


  13. marie-lucie says

    Tonkawa ‘it is said’

    It is not surprising that a morpheme meaning “it is said” or “they said” should be extremely frequent. Such a morpheme is typical of story-telling in a number of cultures, since the teller is reporting events and conversations which they did not witness but learned of through hearing others tell of them. Historical recollections from people actually known and vouched for by the speaker (such as long dead great-grandparents) are identified as such, not by the impersonal “it is said”.

    Imagine the impact in some circles of adding “it is said” to every sentence of the Bible! But Plato’s story of Atlantis, for instance, does not just imply “it is said” but cites several knowledgeable persons from previous generations in order to vouch for the veracity of the story.

  14. J.W. Brewer says

    The opposite of the modern nation-state when it comes to tolerating linguistic diversity is not really the raised fist of anarchism but the pre-modern illiberal dynastic empire on the Hapsburg or Ottoman model. Pay your taxes and don’t revolt and you can talk amongst yourselves however you like. It’s not like the autocrat wants to spend his money compelling your children to become literate in either his language or yours. Of course, there was regime-to-regime variation and by the 19th century the Czarist regime was pursuing heavy-handed Russification of some but by no means all of its varied subject peoples. I think the rule of thumb there was that if your L1 was non-Slavic you were much more likely to be left to your own devices (unless it was Lithuanian).

  15. I second marie-lucie about constant churning of vocabulary as people move about acquiring new neighbors who speak different languages, while structural features slowly converge thanks to translationese, the constant need to translate within the current repertoire of village languages. These small communities are rarely isolated and rarely stay in one place for many generations, even though they may move about in the same general area. Typical small villages of 200 or so people are not self-sufficient in critical resources, including spouses. (Linguistic isoglosses common run down marriage beds in each household.) Much traditional trade was between in-laws, and villages sometimes disperse to their in-laws when disaster strikes.

    Persistent translationese, for instance, has caused many non-Oceanic languages in PNG to innovate inclusive/exclusive distinctions in their 1st person pronouns (yumi vs. mipela in Tok Pisin) that is so typical of Oceanic languages elsewhere. (Some Oceanic languages in PNG have lost that distinction.) The New Guinea mainland is the only place you can find Oceanic languages with verb-final word order, thanks to long spells of communication with non-Oceanic neighbors who speak verb-final languages.

    The first article I published after doing fieldwork in a small New Guinea village was on multilingualism and language mixture in that village, and my fieldwork led to a dissertation on word-order change in PNG Austronesian languages rather than a description of just one language.

  16. Was it in PNG that a language was described where the nouns all had the opposite gender to what was expected based on the closest related languages, and the explanation was supposed to be that the speakers had got together and said “Let’s flip all our noun genders just to be different”? (I guess it can’t have been, since I don’t think PNG languages have binary gender systems.)

  17. Bathrobe, Hat, J.W. Brewer: In between the extremes of the textbook modern Nation-States (democratic or not) which cause linguistic uniformization (both in terms of causing a single standardized form of the “national” language to spread at the expense of dialects/non-standard varieties and in terms of eliminating minority languages) and pre-modern empires which typically do not, there are plenty of intermediate cases. Norway, for example, still exhibits an impressive degree of dialect diversity, and Switzerland combines stable multilingualism with well-preserved dialect diversity (well, outside its French-speaking areas).

    I have often thought that one of the tragedies of many former European colonies, in sub-Saharan Africa especially, is that their intelligentsias took the local European colonizers’ textbook nation-state model as the only existing model for “modernization”, when a Norwegian- or Swiss-type language policy would probably have been much more appropriate for many if not most of these newly independent countries (David Eddyshaw: Thoughts?)

    Marie-Lucie: Regarding your comment-

    “But what is changed is part of the vocabulary, not the morphology or grammar of the language in question, so in classifying the local languages, or comparing them for possible historical work, the vocabularies are relatively useless, what is important is the structural skeleton”

    -I partly disagree. Large-scale vocabulary replacement can lead to major grammatical changes. Let us imagine, for example, a form of English suddenly coining or borrowing massive amounts of vocabulary. Let us further bear in mind that borrowings or neologisms follow the more common (“regular”) morphological patterns (i.e. plurals in -s, verbs will be conjugated as weak verbs). If the new words are numerous enough and include enough basic (common) vocabulary, irregular nominal plurals and strong verbs are far more liable (than they would be in a more lexically conservative form of English) to shrink as classes or even to disappear altogether.

    This is not wholly hypothetical, incidentally. I do know of a few real languages (some Romani and Modern Nahuatl varieties, for instance) where a morphological pattern which was once marginal and reserved for certain marked categories of words, such as borrowings, subsequently, as a direct result of massive borrowing, became a major if not the dominant morphological pattern.

    As a result, the structural skeleton gets profoundly transformed as well. Since older, synchronically irregular forms are often good indicators of genetic relatedness, massive renewal of vocabulary is liable to make language classification more difficult even if the vocabulary is not directly taken into account by scholars examining the matter.

    Finally, on language differentiation being used to maintain social distance from neighboring groups: This has been observed in less exotic places:


  18. Tahitian is another language famous for taboo replacement of common words, which happened to have served as parts of unpronounceable names. For example, the word pape ‘water’ (as in Pape’ete ‘water from a basket’) replaced the common Polynesian wai, presumably because of some such prohibition.

    Etienne, do you have handy the references for the Romani and Nahuatl cases you mention?

  19. Y: My sources are: 1-For Spanish influence upon Modern Nahuatl, Chapters 5 and 6 of this book:


    and 2-For Romani, this article:


  20. marie-lucie says

    Etienne, OK, my statement may be too restrictive. It is of course influenced by my research focus on Penutian languages, a group which officially does not exist! Most work on them has emphasized lexical-phonological comparison, concluding that resemblances are due to “massive borrowing”, while I emphasize morphology, including “shared retentions” even in widely separated families.

  21. David Marjanović says

    I once read somewhere that when Baden and Württemberg were merged in the 1950s (shortly after the establishment of the Federal Republic of Germany), the former border between them began to become a dialect boundary as people tried to figure out which words (I don’t know about other features) were more characteristic of their side or the other side and began to speak accordingly.

  22. TR: Many Non-Austronesian (NAN) languages in PNG do distinguish gender in their pronouns (reflecting the noun classes to which they refer). Austronesian languages that do that almost certainly innovated it from long contact with NAN-speaking neighbors.

    Many PNG languages also have a tok hait (hidden talk), either rearranging syllables or using circumlocutions like ‘ant eggs’ for ‘rice’, ‘sword people’ for ‘Japanese’ during the war, ‘treetop canoe’ for ‘airplane’, ‘underwater canoe’ for ‘submarine’ and so on, so as not to pronounce widely borrowed and understood words. In very multilingual communities, it takes a bit of work to make your vocabulary unintelligible to most.

  23. Nitpick: the Northern parts of Baden and Württemberg were already merged into the state of Württemberg-Baden (American zone) in 1945. This state was merged with the states of Baden (actually Southern Baden and called Südbaden until 1946) and Württemberg-Hohenzollern (both French zone) into Baden-Württemberg in 1952.

  24. J.W. Brewer says

    Not that dialect boundaries would perfectly track political boundaries (wherever they may have been drawn at a particular historical moment), but pre-merger Württemberg was mostly Schwäbisch speakers while pre-merger Baden was mostly Niederalemannisch speakers, innit? From the perspective of the rest of the Teutonosphere those dialects are more similar than dissimilar of course, but they’re not identical. (I think Hohenzollen-Sigmaringen was more Swabian, but I’m not sure about that even though one of my great-great-grandfathers came from there. To be fair, he died approximately 80 years before I started learning German myself, which was in turn several years before I became aware of the existence of the Swabian variant.)

  25. @JWB: Basically correct. Hohenzollern is Schwäbisch. The border between Alemannisch and Schwäbisch runs along the Black Forest. The South of Württemberg near the Bodensee is also Alemannic, and Northern Baden and Württemberg are both East Franconian.

  26. There was a BBC documentary a few years ago demonstrating that the line of the Danelaw is still to some extent a dialect boundary today, more than 1000 years later. The dialects have converged quite a lot, but what differences remain still happen at the exact same boundary.

    I’ll briefly mention the small California town of Boonville, which, back around the end of the 19th century, decided to create its own language, Boontling, just to be different. It was a bad time to start up your own language, because soon the Bucky Walter (telephone) and other modern technology arrived to connect them to the outside world. But if they had had a few centuries of continued isolation…

    As discussed above, Boontling didn’t innovate in terms of grammar, just in vocabulary.

  27. In Australia, news reports on tragic or controversial Aboriginal deaths are all too common. On the public broadcasters, these are preceded by a warning that they say “the name of someone who has recently passed”; on the commercial channels not so — at least not in 2004.

  28. a warning that they say “the name of someone who has recently passed”

    After the mass shootings at the mosques in Christchurch NZ ~2 years ago, the Prime Minister vowed the name of the killer would never pass her lips (no oxygen of publicity). Most of the population, and local media have followed suit.

    International media not so much. And then local media carry syndicated pieces from offshore, which it seems don’t get proof-read …

  29. the structural skeleton gets profoundly transformed as well

    Uzbek and Tajik have borrowed quite a lot of morphology, afaik.
    Uzbek has the prefix na-, as in nafaqat ‘not only’, and prefixes are not a Turkic thing. The loss of vowel harmony in literary Uzbek and many dialects has also come about under the influence of Tajik.
    Tajik, on the other hand, has borrowed the question particle -mi.

  30. @mollymooly: On public tv channels in Australia, it’s common to see the warning “this story may contain images of deceased persons”.

    There is a taboo in Aboriginal Australia against mentioning the deceased. There are a couple of puzzling things though:

    – It doesnt seem to apply to deceased non-Aborigines.
    – Some figures from the past are mentioned and talked about by Aborigines, eg. Yagan – there is even a square in the centre of Perth named after him.

  31. marie-lucie says

    zyxt: The rules of Aboriginal spirituality would probably not apply to non-Aborigines, in this case as in many others. And the rules probably apply to recent deaths, not historical ones.

  32. ə de vivre says

    I recall one of the speakers from that summer series on Caucasian languages (which I heard about here; thanks Hat!) talking about the high language density in the Caucasian highlands. There, again, despite the terrain, the communities weren’t that isolated. But the environment could support a network of scattered small villages where the main types of contact were trade and intermarriage (i.e., small trickles of contact rather than large-scale movement). Wives married into other villages where they were isolated from their previous language community (if it was a different one), so their kids grew up speaking the local language. The end result was frequent multilingualism without much pressure for language replacement (aside from the occasional unusual circumstance and, of course, the present Russifying situation).

  33. David Marjanović says

    – It doesnt seem to apply to deceased non-Aborigines.

    I would guess the taboo is on mentioning dead friends or relatives. (And indeed there are reports of people breaking down and crying, i.e. having their trauma triggered, when such are accidentally mentioned.) If the fact that they’re dead is the first thing you learn about someone, you’re not likely to become very emotional…

  34. The problem is not emotions, but calling a spirit of a community member who is not quite departed. It can harm you. What do aboriginals care about non-aboriginal communities being infested with their departed?

  35. marie-lucie says

    D.O. Just what I was thinking, but you said it better.

  36. David Eddyshaw says

    I would imagine that the bar on mentioning the name of the dead in Australian cultures operated in a context where everybody knew everybody, pretty much. There is probably no real traditional model for applying it in a world composed largely of strangers, like our own.

  37. This interview with Erin Shay, linguist studying the Chadic languages in Cameroon might be of interest (from the Brill blog):

    Life of a Linguist. Erin Shay on Chadic Languages and Fieldwork in Cameroon

  38. John Cowan says

    And indeed there are reports of people breaking down and crying, i.e. having their trauma triggered, when such are accidentally mentioned.

    Somewhat similarly, there is an Asimov story (“Sunset and Evening Star”) in which the French protagonist and his American partner are planning the design of a Lunar base (for a movie). There is a heated dispute about where to place it: the Frenchman wants to use the crater Bahyee (as the story calls it), to which the American objects stubbornly and unreasonably. (Bahyee is remarkable because it is right at the Moon’s visible edge, and one of the few places on the Moon where the Earth rises and sets due to libration.) The mystery is resolved when the partner’s wife’s lover is found to be named Bailey.

  39. Trond Engen says

    David M.: I would guess the taboo is on mentioning dead friends or relatives. (And indeed there are reports of people breaking down and crying, i.e. having their trauma triggered, when such are accidentally mentioned.)

    D.O.: The problem is not emotions, but calling a spirit of a community member who is not quite departed. It can harm you.

    Wouldn’t those emotions be a sign that the spirit’s been called?

  40. Sure, which is why you want to avoid it.

Speak Your Mind