I'm building a world where I'm tracking the languages and people throughout time. So I'm starting from a single point with a single language and then expanding that out as time progresses. My basic presumption based on what I've seen is that every x (in the case I'm working with it's 1500 years, but in the real world it seems to be 500 years) number of generations a language has diverged enough to become considered 2 different languages...

My question is what is the physical distance between 2 languages that have diverged on average? Or how much difference should there be between band 0 (language standard group) and band 1 (group that has direct contact with standard group) and at what "band" does Language 1 become Language 2?

To be clear, I know it has to do with interactions between groups, but I need to map it in physical space... As my map is set out now each band is roughly 12.5 km away from any other band

  • 5
    Language change is not a factor of distance alone. Dialect/language density ie per sq km varies greatly, eg in Switzerland or the Caucasus there are multiple language families inside a very small region whereas Russian or American accents are often indistinguishable from the Atlantic to the Pacific. Topography is a factor. If there is continued interaction then the language is maintained, at least as diglossia. Substrata and superstrata are another factor. Any realistic model will be very complex. Jan 30, 2017 at 7:09
  • @A.M.Bittlingmayer Well, in present day we have mass media that maintains dialects and such better, but I'm working with hunter-gatherers (at least presently) that interact with 18 other bands in a hex pattern around them so that the band 1 move to the left will interact with a few different other bands than the original center band. i'm trying to figure out how much change between these bands... and when it's a different language... is it just when they are no longer part of the same trade network... that's doesn't seem right. I'm just ignoring land barriers right now ^.^
    – Durakken
    Jan 30, 2017 at 9:50
  • 1
    Over time your described scenario will essentially have the very maximum autochtonous language per sq km observable on earth, greater than that of modern en.wikipedia.org/wiki/Languages_of_Papua_New_Guinea. 1500 years is enough for significant divergence, 500 not quite without admixture. Jan 30, 2017 at 10:45
  • @A.M.Bittlingmayer I don't agree with that 500 years is not enough time for significant divergence simply because that looks to be the average time before a language has changed enough to diverge/turn into a new language from everything I have seen. The reason 1500 years is chosen here is the speakers live 3 times longer thus it takes more time for changes to happen generationally. If I apply this to my question... there should be something like 25 "intermediates" between one language 'origin' and another.
    – Durakken
    Jan 30, 2017 at 10:56
  • 3
    A great deal depends on political, economic, and social variables. If the contact leads to a lot of crossbreeding, especially permanent bonds, then there will be more unpredictable changes in the language with the most immigrants. If it leads to conquest, one language may be more advantageous to know, or knowing both may be advantageous, so again there are unpredictable changes. If it leads to trade and occasional contact, then trade details will predominate and may spread over a large trading area. Etc. "Any realistic model will be very complex." as @A.M.Bittlingmayer points out.
    – jlawler
    Jan 30, 2017 at 13:12

2 Answers 2


There are many factors involving the rate of linguistic change.

Daniel Nettle argues in his paper Is the rate of linguistic change constant that small speech communities have a higher rate of linguistic change than large ones (and for the purpose of the paper, 50 000 speakers is already large).

Note that physical distance is not the only thing that separates speech communities from each other, and that there may be many small speech communities with very different languages in some small area (notorious such areas are the Caucasus mountains and New Guinea).

Geographical features like mountain ridges are very efficient obstacles promoting linguistic diversification. Rivers separate their banks, but connect their ends, so the shape of a language areal is not approximately a circle (or hexagon).

  • I argue that size doesn't matter, the only thing that matters is social cohesion. The reason why one can observe a correlation between total population and rate of change is that you can't have an internally highly isolated community of a million people (until we colonize other solar systems).
    – user6726
    Jan 30, 2017 at 16:29
  • 1
    @user6726: Coherence is implied by the word "speech community", I think. When a group of speakers of one language is split into two parts without communication which each other, they will constitute two speech communities. At least, this is my understanding of the term speech community. Jan 30, 2017 at 19:21
  • I disagree with the implication that the coherence of a (speech) community is uniform across (speech) communities.
    – user6726
    Jan 30, 2017 at 22:56

It all depends on how realistic you want your constructed world to be. Even if you choose a hex grid as the medium for your map, this needn't force you to map every aspect of the world following that pattern. That's worldbuilding advice, anyway. As for the linguistics part: there's no fixed distance or chronological separation (or any other simple-to-measure factor) that can predict that one language will split into two. There are, as explained in other answers, many factors that tend to make languages appear or disappear, and none is very easy to measure.

Small bands of humans distributed uniformly, at relatively short average distances from each other, and separated by nothing but flat terrain, will most probably speak the same language (if there aren't many) or a bunch of dialects of the same language, each of them intelligible to their immediate neighbours, or maybe dialects of two or three different, but closely related languages (if the extremes points of the map are far from each other). It's a matter of neverending discussion when a dialect turns into a language.

This, however, is not a realistic setting, or even close to realistic. Unless there's a reason not to, I would (if I were you) throw in some mountain ridges, rivers and lakes, deserts, and other obstacles (or enablers) of human movement. Then you could stop thinking about exact distances or timeframes and use those geographic accidents, along with whatever fictional history you wanted to include in your world, to figure out where the splits between languages or dialects should occur. You could use the history of actual languages as a guide.

  • It isnt that my map doesn't have geological barriers, but that I'm trying to get a grasp on the situation without them so that I can then modify where needed... Th best explanation of how the continent is laid out is humans live a max of 300 years, 10,000 of them pop into existence in the middle of the american great plains all speaking the same language... they then expand from there as hunter-gatherers... Do all that are in the great plain speak the same language? Do all that make it to the desert speak the same? Canada? Florida? Where/When/how much does the language change?
    – Durakken
    Jan 31, 2017 at 8:40
  • Maybe using a simplified model could work. If all humans pop into existence in the same place and then spread around, you can assume that when a group crosses a geographical barrier it will eventually develop a new language. How much time depends on whether contact is maintained. For comparison, consider that the Romance languages took about 1,000 years to develop as separate entities from Vulgar Latin, over an area that goes from Portugal to France to Italy.
    – pablodf76
    Jan 31, 2017 at 10:35

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.