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

Human lingo or ...

Expand Messages
  • Mike Adams
    If you was to program a computer to create a language on its own? Could you or ?? How would you? Mike (resending some of my poems I sent accidently via my cell
    Message 1 of 7 , Feb 24, 2014
      If you was to program a computer to create a language on its own?

      Could you or ?? How would you?

      Mike

      (resending some of my poems I sent accidently via my cell and my gmx
      account).
    • Ralph L. DeCarli
      I think the problem is that computers don t actually think so they don t have a reason to communicate. A computer could easily generate random pronounceable
      Message 2 of 7 , Feb 24, 2014
        I think the problem is that computers don't actually 'think' so they don't
        have a reason to communicate. A computer could easily generate random
        pronounceable 'words' and assign random meanings to those words. The
        resultant 'language' would probably be unlearnable by anyone without
        an eidetic
        memory, but that isn't a problem for a computer.

        Alternately, a computer could be furnished with a knowledge base of sound
        changes, grammatical rules, syntactic rules, etc. The computer could pick
        random rules and generate words which would be assigned meanings according
        to the rules chosen. Random exceptions could be added to simulate
        naturalism.

        It is unlikely that this would actually create a naturalistic language on
        any given trial, but eventually a 'language' might be generated that was
        less artificial sounding than the previous attempts.

        Did you have a purpose in mind for this language? Constructing the
        knowledge base would be a nontrivial task.

        Ralph


        On Mon, Feb 24, 2014 at 3:28 AM, Mike Adams <abrigon@...> wrote:

        > If you was to program a computer to create a language on its own?
        >
        > Could you or ?? How would you?
        >
        > Mike
        >
        > (resending some of my poems I sent accidently via my cell and my gmx
        > account).
      • R A Brown
        ... Exactly! They have no more reason to communicate with us humans than any other machine. I haven t noticed my TV trying to chat with me when it s on
        Message 3 of 7 , Feb 25, 2014
          On 25/02/2014 03:33, Ralph L. DeCarli wrote:
          > I think the problem is that computers don't actually
          > 'think' so they don't have a reason to communicate.

          Exactly! They have no more reason to communicate with us
          humans than any other machine. I haven't noticed my TV
          trying to chat with me when it's on standby, or my lawn
          mower suggesting that it's maybe about time I cut the
          grass ;)

          OK, I know some machines have been small computer programs
          in them that allow it to give verbal warnings. some cars.
          e.g. can tell you they are running low on fuel. But try
          chatting back and asking if it saw a service station back
          there or even just ask if it's liking the weather toady, or
          the scenery etc.

          Machines do not actually think - even tho they can be made
          to appear to have some limited "intelligence."

          > A computer could easily generate random pronounceable
          > 'words' and assign random meanings to those words.

          As indeed Mark Line did with his Classical Yiklamu. The
          vocabulary was computer generated by a program that mapped
          pseudo-random words, permissible according to Yiklamu
          phonology, to each of the 70000 (more or less) entries in
          the WordNet thesaurus.

          I know others have used computer programs to generate
          vocabulary - tho perhaps on not such a large scale. for
          those who find creating vocabulary boring, this is a method;
          but I would suggest not creating 70K items!

          > The resultant 'language' would probably be unlearnable
          > by anyone without an eidetic memory,

          Well, it would just be a whole load of lexical items - not
          much use for communication without at least some basic rules
          of syntax. OK, these could be programmed into the computer
          so it could generate pseudo-random strings of words that
          conformed to the syntax - but there's no communication going on.

          [snip]

          > Did you have a purpose in mind for this language?
          > Constructing the knowledge base would be a nontrivial
          > task.

          Nontrivial is putting it mildly ;)

          > On Mon, Feb 24, 2014 at 3:28 AM, Mike Adams wrote:
          >
          >> If you was to program a computer to create a language
          >> on its own?
          >>
          >> Could you or ??

          See above

          >> How would you?

          See also above.

          But more to the point "Why would you?"

          --
          Ray
          ==================================
          http://www.carolandray.plus.com
          ==================================
          "Ein Kopf, der auf seine eigene Kosten denkt,
          wird immer Eingriffe in die Sprache thun."
          [J.G. Hamann, 1760]
          "A mind that thinks at its own expense
          will always interfere with language".
        • Andrew Schade
          As was already stated, the difficulty resides mostly in the creation of the knowledge base required to develop the language, in that you d have to teach the
          Message 4 of 7 , Feb 25, 2014
            As was already stated, the difficulty resides mostly in the creation of the
            knowledge base required to develop the language, in that you'd have to
            teach the program to be able to recognise related words, so that it would
            (at least in certain cases) relate their generated conlang versions.
            The heuristics (sp) required to keep the language naturalistic would be
            very intensive, complicated, and a general pain.
            All in all. I doubt that it would be a worthwhile excercise. Much more
            usefull would be a computer tool that would aid in the creation of conlangs.

            That raises some interesting questions. What would you all, as (probably)
            more experienced conlangers than I want to see in a tool for creating
            conlangs?
            -Schade
            On Feb 24, 2014 11:33 PM, "Ralph L. DeCarli" <omnivore@...>
            wrote:

            > I think the problem is that computers don't actually 'think' so they don't
            > have a reason to communicate. A computer could easily generate random
            > pronounceable 'words' and assign random meanings to those words. The
            > resultant 'language' would probably be unlearnable by anyone without
            > an eidetic
            > memory, but that isn't a problem for a computer.
            >
            > Alternately, a computer could be furnished with a knowledge base of sound
            > changes, grammatical rules, syntactic rules, etc. The computer could pick
            > random rules and generate words which would be assigned meanings according
            > to the rules chosen. Random exceptions could be added to simulate
            > naturalism.
            >
            > It is unlikely that this would actually create a naturalistic language on
            > any given trial, but eventually a 'language' might be generated that was
            > less artificial sounding than the previous attempts.
            >
            > Did you have a purpose in mind for this language? Constructing the
            > knowledge base would be a nontrivial task.
            >
            > Ralph
            >
            >
            > On Mon, Feb 24, 2014 at 3:28 AM, Mike Adams <abrigon@...> wrote:
            >
            > > If you was to program a computer to create a language on its own?
            > >
            > > Could you or ?? How would you?
            > >
            > > Mike
            > >
            > > (resending some of my poems I sent accidently via my cell and my gmx
            > > account).
            >
          • MorphemeAddict
            I agree in principal that computers basically don t have a reason to communicate with other computers in a language , but at least one project has already
            Message 5 of 7 , Feb 25, 2014
              I agree in principal that computers basically don't have a reason to
              communicate with other computers in a 'language', but at least one project
              has already been done in which autonomous robots ('lingodroids') create a
              language to describe to each other the geometry of their surroundings.
              Somehow the beeps of the robots in the video are 'translated' to human
              morphemes.
              http://www.bbc.co.uk/news/technology-13510988

              There is also ROILA, a project to create a language specifically for
              interaction between humans and robots.
              http://en.wikipedia.org/wiki/Robot_Interaction_Language

              stevo


              On Mon, Feb 24, 2014 at 10:33 PM, Ralph L. DeCarli
              <omnivore@...>wrote:

              > I think the problem is that computers don't actually 'think' so they don't
              > have a reason to communicate. A computer could easily generate random
              > pronounceable 'words' and assign random meanings to those words. The
              > resultant 'language' would probably be unlearnable by anyone without
              > an eidetic
              > memory, but that isn't a problem for a computer.
              >
              > Alternately, a computer could be furnished with a knowledge base of sound
              > changes, grammatical rules, syntactic rules, etc. The computer could pick
              > random rules and generate words which would be assigned meanings according
              > to the rules chosen. Random exceptions could be added to simulate
              > naturalism.
              >
              > It is unlikely that this would actually create a naturalistic language on
              > any given trial, but eventually a 'language' might be generated that was
              > less artificial sounding than the previous attempts.
              >
              > Did you have a purpose in mind for this language? Constructing the
              > knowledge base would be a nontrivial task.
              >
              > Ralph
              >
              >
              > On Mon, Feb 24, 2014 at 3:28 AM, Mike Adams <abrigon@...> wrote:
              >
              > > If you was to program a computer to create a language on its own?
              > >
              > > Could you or ?? How would you?
              > >
              > > Mike
              > >
              > > (resending some of my poems I sent accidently via my cell and my gmx
              > > account).
              >
            • Alex Fink
              ... Ah, I nearly forgot my obligatory mention that I m working on a program to create conlangs. (At least in theory. It s been ages since I ve had any time
              Message 6 of 7 , Mar 4, 2014
                On Tue, 25 Feb 2014 06:18:08 -0500, Andrew Schade <schade.andrew.l@...> wrote:

                >As was already stated, the difficulty resides mostly in the creation of the
                >knowledge base required to develop the language, in that you'd have to
                >teach the program to be able to recognise related words, so that it would
                >(at least in certain cases) relate their generated conlang versions.
                >The heuristics (sp) required to keep the language naturalistic would be
                >very intensive, complicated, and a general pain.
                >All in all. I doubt that it would be a worthwhile excercise. Much more
                >usefull would be a computer tool that would aid in the creation of conlangs.

                Ah, I nearly forgot my obligatory mention that I'm working on a program to create conlangs. (At least in theory. It's been ages since I've had any time for it.) So far I have a little component for generating the set of categories expressed in the morphology, and a component for phonologies. Those are here:
                https://github.com/alexfink/random_language

                What I've been planning to do next is ... a better phonology component, one which allows for actually interesting underlying patterns and not just picking a selection of contrastive features off a fixed menu of 30. I have some thoughts about morphosyntax as well; and since the last time I had a reason to mention this here, David Alfter of Trier expressed interest to me in making a script-generating component. But the lexicon is entirely intimidating: as you observe, doing it pèrfectly is more or less tantamount to artificial general intelligence. I'm hopeful that, without going that far, there is some way to cobble together semantic networks and different choices of semantic features etc. that can at least give non-relexy results in semantic fields it is taught about.

                Alex
              • Leonardo Castro
                ... Very interesting! The online random phonology generator is great! http://000024.org/cgi-bin/gleb.cgi I can hardly wait for the random text generator. Or at
                Message 7 of 7 , Mar 11, 2014
                  2014-03-05 2:06 GMT+01:00 Alex Fink <000024@...>:
                  > On Tue, 25 Feb 2014 06:18:08 -0500, Andrew Schade <schade.andrew.l@...> wrote:
                  >
                  >>As was already stated, the difficulty resides mostly in the creation of the
                  >>knowledge base required to develop the language, in that you'd have to
                  >>teach the program to be able to recognise related words, so that it would
                  >>(at least in certain cases) relate their generated conlang versions.
                  >>The heuristics (sp) required to keep the language naturalistic would be
                  >>very intensive, complicated, and a general pain.
                  >>All in all. I doubt that it would be a worthwhile excercise. Much more
                  >>usefull would be a computer tool that would aid in the creation of conlangs.
                  >
                  > Ah, I nearly forgot my obligatory mention that I'm working on a program to create conlangs. (At least in theory. It's been ages since I've had any time for it.) So far I have a little component for generating the set of categories expressed in the morphology, and a component for phonologies. Those are here:
                  > https://github.com/alexfink/random_language

                  Very interesting!

                  The online random phonology generator is great!

                  http://000024.org/cgi-bin/gleb.cgi

                  I can hardly wait for the random text generator. Or at least a
                  pseudo-text that would pass a "static Turing test".
                Your message has been successfully submitted and would be delivered to recipients shortly.