I want to make it clear from the start that I'm not any kind of expert in linguistics or history, which is why I'm asking this question here. (Perhaps it's because of my physics background that I'm wondering about "characteristic timescales.")

I've heard that Polynesian languages, which spread throughout the Pacific Ocean starting 3000 years ago, are very similar to one another. In Wikipedia, for instance, it says

Still today, Polynesian languages show strong similarity, particularly cognate words in their vocabulary; this includes culturally important words such as tapu, ariki, motu, fenua, kava, and tapa as well as *sawaiki, the mythical homeland for some of the cultures.

Meanwhile, (pre-European) languages in North and South America are described as being extremely diverse. From ayahuasca.com,

Now consider the following languages: Salinan, Shasta, Yuki, Maidu, Pomo, Yokuts, Esselen, Washo, Karuk, Chimariko. All are from present-day California, but what is notable about these languages is that each one of them represents a unique isolate, like Basque. That is ten isolates in California alone — each utterly unique, representing an entirely unique language stock all by itself, and each as different from any other language as English is from Arabic or Vietnamese is from Zulu.

People first spread across the Americas at least 13,000 years ago (pre-Clovis discoveries could push that back farther).

I was wondering if these two cases of people spreading across large distances could be taken as a way to measure the rate of language drift: 3000 years is too short to develop into fundamentally different languages ("isolates"), but 13,000 years is long enough. So the characteristic timescale is somewhere between them, like ~10,000 years. (The tilde means "more than 1000, less than 100,000.") I'd like to know if this is something that has been written about, possibly rejected, that I could go read up on.

(In preparing this question, I came across a point that could weaken it: the original migrations into the Americas could have involved more than one language group, but it still seems hard for me to believe that the number of isolates in modern times came from that many independent migrations into the continent. Also, I know that human situations are different in different times and places, unlike drifting particles, but it would be nice to know if something universal can be said about this. Like, each generation is willing to change the language a certain amount, but not much more or much less, and if there are cultural variations, they average out in the large sweep of time.)

  • Does 'pre-European' here mean 'pre-Columbian' or 'pre-Viking' or something like either? I ask because 'pre-European' is a term I've never heard before and without specifics, it seems to me to suggest more than a date in history; perhaps a linguistic influence… Dec 26, 2021 at 1:55
  • I guess I meant pre-either, but the Vikings didn't make much impact—probably no languages changed. I wasn't intentionally avoiding a common term, just picked the wrong one. I just wanted to be clear that I wasn't talking about Spanish, Portuguese, English, French, ... Dec 27, 2021 at 4:21

3 Answers 3


It's certainly not the case that languages change at anything like a uniform rate, as user6726's answer points out. Nevertheless, I think the answer to your question is a qualified yes.

There's one universal fact we do know about the rate of language change: it's never zero (except in the case of a dead language). All languages are constantly changing, and changes accumulate over time. The rate of change varies based on the social histories of the language communities, but it's still broadly true that two languages that split 3,000 years ago are much more likely to be perceptibly similar to each other than two languages that split 13,000 years ago.

Evidence for this is the fact that almost all known language families are believed to have originated more recently than 13,000 years ago. The only possible exception I know of is Afro-Asiatic; there's a wide range of estimates for the date of Proto-Afro-Asiatic, some of which go that far back or more. But, even given that they are related, it's not at all the case that all the Afro-Asiatic languages are obviously similar to each other: it's probably fair to say that Hebrew is as different from Hausa "as English is from Arabic or Vietnamese is from Zulu" (to use the comparisons in your quote). We know they're related because of painstaking comparative work and because several languages in the family have been attested in writing for millennia, which obviously makes comparison and reconstruction easier; if this weren't the case, and if the languages had as few speakers and had been as scantily studied as most indigenous American languages, we might know no more about their relationships than we do about those of the languages of California.

On the other hand, there are very many groups of languages that are known to be related within a 3,000-year time frame. If time didn't figure into the equation at all then there should be comparable numbers of known language families or subfamilies in the 13,000- and 3,000-year ranges.

So to your hypothesis that

3000 years is too short to develop into fundamentally different languages ("isolates"), but 13,000 years is long enough

-- the second part is clearly true. There are no languages that split 13,000 years ago but are nevertheless similar in the way that the Polynesian languages are similar. The first part is too strong: there pretty certainly exist languages that were related 3,000 years ago but which are no longer as similar as the Polynesian languages. This is because radical linguistic changes can happen quite rapidly. But we can definitely say that 3,000 years is not necessarily enough time to obscure genetic relatedness, while 13,000 years almost always is.

  • 3
    Cool! Thank you for the detail. It's useful to know that 13,000 years also exceeds the splitting of other language families (except possibly Afro-Asiatic), the effect of familiarity on language grouping (that Californian languages might be more related than is currently known, but lack of centuries of written samples or even current speakers could prevent the discovery of a grouping), and that Polynesian languages are surprisingly similar, even for their 3000 year vintage. These are the kinds of details I was hoping to learn. Dec 24, 2021 at 3:56
  • 1
    And given the fact that people have been talking for as long as a million years, it means that recent language families just represent a few recent groupings. The fact that the comparative method, like Carbon-14 dating, has a limited time-depth, means we'll never know how language developed in its earlier stages because there is no data.
    – jlawler
    Dec 24, 2021 at 15:12
  • 1
    @jlawler: a million years is way outside currently accepted estimates. 200,000 years is more like it.
    – TonyK
    Dec 25, 2021 at 0:40
  • 1
    Proto-Polynesian is usually put at about 2,000 years. It seems the the original 'Polynesians' arrived in the 'Polynesian Triangle' around 3,000 years ago, but the spread of daughter populations from Proto-Polynesian (and consequent development of daughter languages) began around 2,000 BP. Dec 25, 2021 at 8:53
  • 1
    Maybe Trans New Guinea happened 10KBP, whereas Pama-Nyungan happened much more recently in Australia. The differences between the language variance in those two linguistic areas continue to amaze me.
    – jlawler
    Dec 25, 2021 at 18:23

The quick answer is "no", what's more important IMO is why. The way in which a language changes over time depends on so many factors that there isn't anything meaningful that can be said about the rate of language change. Social facts are so important that I do not see that linguistic properties of languages are particularly relevant to language change: it's all in the social factors. For example, the geographical distribution of English, Spanish and French is due to specific well-known social factors, ones not applicable to Dutch, Portuguese and Swahili. The spread of Russian vs. Ukrainian has a simple social explanation. We have some idea about the relevant social factors surrounding pre-modern spread of Bantu or Polynesian languages.

The situation with North American language isolates is not necessarily about the actual "isolatedness" of the languages, it may be a consequence of not having sufficient information to assert relatedness, for example, Yokuts may be related to Costanoan and Miwok, but this is not yet convincingly established. You may ask, what will it take to establish relatedness – more language descriptions and comparative work, which is increasingly more impossible (Utian languages are mostly extinct).

There is a thought experiment that tells us that an infinite amount of time will not lead to language-split. Assume a group of English speakers who settle on an asteroid and are stuck in outer space forever. They survive for eons on a 50 sq. mile rock with no contact with the rest of the universe. Social forces will keep that language relatively uniform over the millenia – they won't suddenly start speaking different languages because 3,000 years have elapsed. What will matter is whether people continue to talk to each other.

  • 1
    +1 One of the bad parts of the biological/language evolution metaphor is that, while there is a rate at which mutations accumulate in biosystems, there is no such rate for language mutations. The fact that humans may (and usually do) have several languages but only one genotype is responsible. That's the source of many of the overenthusiastic articles about language history that get published in biological journals these days; the biologists don't understand how complex language is and how fast and slow it can change.
    – jlawler
    Dec 23, 2021 at 23:11
  • 2
    All good points; it's worth keeping in mind that language change is fundamentally different from biological change, in that human decisions are involved. However, even on an asteroid, wouldn't some change happen just because of errors, like the game of Telephone? Despite being an English speaker surrounded by English speakers, I was well into my 20's before I noticed that I had some private vocabulary and spelling that didn't square with standard English (unintentionally), and groups of kids delight in intentionally introducing new words, not because their lifestyle differs from their parents. Dec 24, 2021 at 0:37
  • 1
    @Jim Pivarski, " language change is fundamentally different from biological change, in that human decisions are involved". Spot on. And those decision may be made to obfuscate (or communicate and obfuscate simultaneously), as in a thieves' cant or rhyming slang. Later people who aren't "in the know" start to use some of the cant, so people who belong to the underworld modify their language. Dec 24, 2021 at 7:35
  • 1
    And it's all human decisions (and unchangeable human differences) that make individuals recognizable on the telephone. We tolerate unbelievable amounts of variation in language, while still believing we're all speaking "the same" language. And so Norma Loquendi is never what she used to be, but never grows old.
    – jlawler
    Dec 24, 2021 at 15:17
  • 1
    @jlawler I think that there is a phenomenon similar to mutations, it's just that the "error correction" mechanisms are different, and language evolution is driven more by the "error correction" than the mutation rate. For instance, there's a strong correlation between a verb being irregular and it being common, because the more common it is, the more "correction" there is when children try to conjugate it regularly. Dec 26, 2021 at 18:31

@Jim Pivarski, I'm not a linguist, either. Here's a few point to ponder.

  1. Different Polynesian cultures stayed in contact across the Pacific. According to Navigating the Stars, by Witi Ihimaera, voyages continued from Aotearoa New Zealand (the last country colonized by the Polynesians) back to other parts of Polynesia. So contact wasn't lost.
  2. Can you read Beowulf? I can't, except with a parallel text. I've heard people from other cultures claim that they can read texts from before the 10th century (Greek for instance); that's with a bit of effort but not, I suspect, comparable to learning Old English. Greece and England were both invaded several times (1066 was the last instance for England), but the invaders assimilated, unlike the Turks in Greece.
  3. I used to live in the Solomon Islands, where there are 80 indigenous languages--not dialects, languages. Vanuatu has 138 languages, and Papua New Guinea over 800. So, for some reason, there is a lot of pressure for Melanesian languages to diverge.

As others have said, there is no constant rate for languages to diverge.

  • 1
    I've heard about the voyages Polynesians take between islands, to exchange special objects and generally keep in touch. But the Americans were in contact, too (within each continent, at least). On #2, no, I can't read Beowulf, or even Chaucer (and struggle with Shakespeare), but I know English speakers who can, and they're generally better at foreign languages, too. There's some plasticity that some people seem to have more of than others. These things contribute to variation, but I was thinking that if enough people and enough generations are involved, all those variations get averaged over. Dec 24, 2021 at 6:21
  • @Jim Pivarski, Here is something that I half remember from a talk that Roger Keesing gave in Honiara back in 1983. If I remember correctly there was a taboo on mentioning the name of a "bigman" within the community where he was influential, so a different word would have to be found. The taboo also applied to words that sounded similar. This was one of the pressure on language that I mentioned.. Dec 24, 2021 at 6:37
  • 1
    About your second paragraph, in the case of arabic, most dialect speakers, even with very little exposure to classical arabic can understand arabic texts from even as far as the fifth century CE. But then again, arabic is a very prescriptivist language, which greatly slows down the rate of language change, even in the local dialects (except for the dialects for north west africa). Dec 24, 2021 at 10:54
  • 1
    In 1986 Jeff Marck proposed that an overnight voyage is a reasonable distance to maintain regular contact between islands within Polynesia, but more than that and contact becomes very irregular (though it did occasionally occur). Drawing circles of the 'overnight voyage' distance around islands/atolls/etc gives a mapping very similar to the distribution of distinct (ie non-mutually intelligible) languages, at least within Micronesia and Polynesia. Dec 25, 2021 at 9:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.