Loading ...
Sorry, an error occurred while loading the content.
 

Re: THEORY: Models of language spread.

Expand Messages
  • BPJ
    Þe Ynglisc tung is already forcom, behwerved wiþ a cunnyng Normannisc bastard-laad crafted for werld lordscip, forþij nobody can tell hweþer its a Þeedisc
    Message 1 of 8 , Feb 1, 2013
      Þe Ynglisc tung is already forcom, behwerved wiþ a cunnyng Normannisc
      bastard-laad crafted for werld lordscip, forþij nobody can tell hweþer its
      a Þeedisc reerd or a Romanisc.

      Den fredagen den 1:e februari 2013 skrev Peter Collier:

      > Perhaps the term should be changed to Lingua Angla...
      >
      > -----Original Message-----
      > From: Constructed Languages List [mailto:CONLANG@...<javascript:;>]
      > On
      > Behalf Of George Corley
      > Sent: 01 February 2013 19:34
      > To: CONLANG@... <javascript:;>
      > Subject: Re: THEORY: Models of language spread.
      >
      > [snip]
      >
      > I imagine that when it falls it will follow a similar course to Latin --
      > lingering around after the Anglophone powers have retreated as a lingua
      > franca of elites before eventually dying.
      >
    • Peter Collier
      Whither did you come by that? ... From: Constructed Languages List [mailto:CONLANG@LISTSERV.BROWN.EDU] On Behalf Of BPJ Sent: 01 February 2013 21:11 To:
      Message 2 of 8 , Feb 1, 2013
        Whither did you come by that?


        -----Original Message-----
        From: Constructed Languages List [mailto:CONLANG@...] On Behalf Of BPJ
        Sent: 01 February 2013 21:11
        To: CONLANG@...
        Subject: Re: THEORY: Models of language spread.

        Þe Ynglisc tung is already forcom, behwerved wiþ a cunnyng Normannisc bastard-laad crafted for werld lordscip, forþij nobody can tell hweþer its a Þeedisc reerd or a Romanisc.

        Den fredagen den 1:e februari 2013 skrev Peter Collier:

        > Perhaps the term should be changed to Lingua Angla...
        >
        > -----Original Message-----
        > From: Constructed Languages List
        > [mailto:CONLANG@...<javascript:;>]
        > On
        > Behalf Of George Corley
        > Sent: 01 February 2013 19:34
        > To: CONLANG@... <javascript:;>
        > Subject: Re: THEORY: Models of language spread.
        >
        > [snip]
        >
        > I imagine that when it falls it will follow a similar course to Latin
        > -- lingering around after the Anglophone powers have retreated as a
        > lingua franca of elites before eventually dying.
        >
      • Alex Fink
        Meant to comment sooner. Life. Actually, some years ago while I was playing around with NetLogo , I made a language
        Message 3 of 8 , Feb 1, 2013
          Meant to comment sooner. Life.

          Actually, some years ago while I was playing around with NetLogo <http://ccl.northwestern.edu/netlogo/>, I made a language spread simulation or two. They were probably around as simple-minded as the ones you describe below. I'd have to pore over my past self's undocumented code to remember what I actually did, at this level of detail, but I could attempt this, or throw the files up somewhere, if there was interest.

          Something I should like to do is read through these papers and see if they give support to a claim I often make (based on mathematical intuition) when long-range comparativists are about, as follows.
          Suppose that monogenesis is true, and that history is completely uniformitarian. Draw out the phylogenetic tree of all modern-day languages, trimming away any branches which have gone completely extinct (and ignoring horizontal transfer and mixed languages and whatnot). Then, as you go back in time in one language's history, the nodes should become spaced *exponentially* further apart in time from each other; i.e., since only nodes are reconstructible, comparative reconstruction should get exponentially harder the further back you go. Therefore there really shouldn't be much we can discern left.

          (In fact, history is not uniformitarian, but I think it's mostly in the direction of making deeper reconstructions than we have now harder than they might otherwise have been. The sweeping historical expansion of the PIE speakers has handed a plethora of descendants to us on a silver platter, making PIE's reconstruction anomalously easy, and in so doing probably erased a bunch of what could've become useful evidence for PIE's sisters, making the next stage more harderer by comparison.)

          On Wed, 30 Jan 2013 15:48:05 +0100, Leonardo Castro <leolucas1980@...> wrote:

          >I would like to hear some comments on how realistic do you think that
          >the assumptions of computational simulation of language spread are:

          Well, as you already mentioned, the fact that none of them actually account for sociolinguistic factors is an unrealistic element of all of them. I wonder how you'd modify them to track that? I suppose, to a first approximation, each culture should have values for how well it esteems the other cultures which which it comes in contact, which drift over time in some fashion. And these should in turn affect the patterns with which that culture's members are bilingual, and (therethrough?) the patterns in which its language is influenced by the other languages. Given a feature model like [2] or [3], the features themselves could be set up to have different susceptibilities to cultural replacement etc.

          >-> Viviane model [1]:

          "Fitness" leading a group to occupy sites seems to be the core element of this one (and its successor), so the big question would be what "fitness" is supposed to represent. I suppose it's something like economic might?

          >* probablility of language mutation (creating a new language labelled
          >with another unused integer) inversely proportional to its fitness.

          That is particularly bizarre. You can't actually retard language change by being rich. I wonder if Viviane was thinking of language standardisation here -- if so, the better outcome to simulate would be some sort of diglossia.

          >-> Modified Vivane model [2]:
          >* if the bit strings of the languages of two sites are the same, they
          >are the same language (so I presume that the original language can be
          >reobtained from one of its mutated versions);

          Not a bad assumption. In fact, the closer two languages are, the likelier one is to displace the other: you know, if you speak, say, Aramaic it'd be easier for you to switch to pick up Arabic than Nuuchahnulth.

          But 16 bits are too few for this to be really realistic, I think, in number of characters and especially in number of their values (at least for lexical-like characters). (Both Schulze's F and Q.)

          >-> Schulze model [3]:

          About as good as anything simple I can imagine, really, if you don't try to account for any of these social factors.

          Alex
        • Leonardo Castro
          ... Interesting! I have never heard about it, but it looks very nice. When I played Ikariam, I imagined that maybe MMORPG s could be used to perform
          Message 4 of 8 , Feb 4, 2013
            2013/2/1 Alex Fink <000024@...>:
            > Meant to comment sooner. Life.
            >
            > Actually, some years ago while I was playing around with NetLogo <http://ccl.northwestern.edu/netlogo/>, I made a language spread simulation or two.

            Interesting! I have never heard about it, but it looks very nice.

            When I played Ikariam, I imagined that maybe MMORPG's could be used to
            perform simulation, and maybe some conclusions could be made based on
            the already existing games.

            > They were probably around as simple-minded as the ones you describe below. I'd have to pore over my past self's undocumented code to remember what I actually did, at this level of detail, but I could attempt this, or throw the files up somewhere, if there was interest.
            >
            > Something I should like to do is read through these papers and see if they give support to a claim I often make (based on mathematical intuition) when long-range comparativists are about, as follows.

            It seems that the graph of "number of language" vs. "number of
            speakers" for language families is the real data some of these
            simulations try to reproduce:

            http://www.sciencedirect.com/science/article/pii/S0378437106010740

            Maybe the paper above is not available from non-universitary computer,
            but googling "Viviane Oliveira languages simulation" (without quotes),
            you can see some graphs as well.

            > Suppose that monogenesis is true, and that history is completely uniformitarian. Draw out the phylogenetic tree of all modern-day languages, trimming away any branches which have gone completely extinct (and ignoring horizontal transfer and mixed languages and whatnot). Then, as you go back in time in one language's history, the nodes should become spaced *exponentially* further apart in time from each other; i.e., since only nodes are reconstructible, comparative reconstruction should get exponentially harder the further back you go. Therefore there really shouldn't be much we can discern left.

            Good point!

            [...]

            > On Wed, 30 Jan 2013 15:48:05 +0100, Leonardo Castro <leolucas1980@...> wrote:
            >
            >>I would like to hear some comments on how realistic do you think that
            >>the assumptions of computational simulation of language spread are:
            >
            > Well, as you already mentioned, the fact that none of them actually account for sociolinguistic factors is an unrealistic element of all of them. I wonder how you'd modify them to track that? I suppose, to a first approximation, each culture should have values for how well it esteems the other cultures which which it comes in contact, which drift over time in some fashion. And these should in turn affect the patterns with which that culture's members are bilingual, and (therethrough?) the patterns in which its language is influenced by the other languages. Given a feature model like [2] or [3], the features themselves could be set up to have different susceptibilities to cultural replacement etc.
            >
            >>-> Viviane model [1]:
            >
            > "Fitness" leading a group to occupy sites seems to be the core element of this one (and its successor), so the big question would be what "fitness" is supposed to represent. I suppose it's something like economic might?

            I guess this simulation doesn't distinguish what's cause from what's
            consequence. You might be powerful because you're lucky to have
            resources, or you have resources because you're powerful and can take
            them.

            >>* probablility of language mutation (creating a new language labelled
            >>with another unused integer) inversely proportional to its fitness.
            >
            > That is particularly bizarre. You can't actually retard language change by being rich. I wonder if Viviane was thinking of language standardisation here -- if so, the better outcome to simulate would be some sort of diglossia.

            Maybe they assume that once a language expands itself to a certain
            extension, some kind of language control arises naturally.

            >>-> Modified Vivane model [2]:
            >>* if the bit strings of the languages of two sites are the same, they
            >>are the same language (so I presume that the original language can be
            >>reobtained from one of its mutated versions);
            >
            > Not a bad assumption. In fact, the closer two languages are, the likelier one is to displace the other: you know, if you speak, say, Aramaic it'd be easier for you to switch to pick up Arabic than Nuuchahnulth.

            I remember that someone commented about Arabic being successful to
            spread only in regions where Afro-Asiatic languages were already
            spoken.
          Your message has been successfully submitted and would be delivered to recipients shortly.