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

Re: How to "grow" a conlang.

Expand Messages
  • Gary Shannon
    It seems to me that the conlang will be as alike or as different as you care to make it and I would think that anyone with any reasonable amount of conlang
    Message 1 of 15 , Jan 6, 2013
    • 0 Attachment
      It seems to me that the conlang will be as alike or as different as you
      care to make it and I would think that anyone with any reasonable amount of
      conlang experience would be able to avoid that pitfall.

      Obviously you need to "translate" in the sense that you use the original
      sentence to tell you what happened, not to tell you how to talk about what
      happened.

      As for using Chinese, if I don't know any Chinese then I'm going to have to
      translate the Chinese into English and then into my conlang. Unless you're
      suggesting that the Chinese version can't be translated into English. Or
      perhaps an anotated sentence such as "You (plural) saw that we
      (inclusive)..." and so on.

      --gary

      On Sun, Jan 6, 2013 at 6:04 PM, Rich Harrison <rick@...> wrote:

      > > To create and evolve a conlang begin by translating 100 reasonably easy
      > > sentences.
      >
      > I have some qualms about relying on translation. Doesn't it give the
      > source language too much influence over the structure of the developing
      > conlang? i.e. won't the conlang wind up having the same syntactic and
      > lexical strengths and weaknesses that the source language has?
      >
      > I would suggest that half of these first 100 sentences should come from
      > languages drastically different from the conlanger's native language. For
      > example, include some Mandarin sentences that use the aspectual particles
      > "zhe" and "le." And two Japanese sentences that are identical except that
      > one uses "kimi" and the other uses "omae" as a second person
      > pronoun-like-word. Etc.
      >
    • Nikolay Ivankov
      I don t know, how can this work for me. I m always starting from two different ends - pure grammar and pure phonology, and then start matching them up by
      Message 2 of 15 , Jan 7, 2013
      • 0 Attachment
        I don't know, how can this work for me.

        I'm always starting from two different ends - pure grammar and pure
        phonology, and then start matching them up by trying to create a lexicon.
        I've never got far in creating lexicon, because, as an artlanging fan, I
        love exceptions. And making exceptions suggests going at least one step
        back, to a protolang where they still were rules. In case of Yanyarin it's
        four steps back, but now I know precisely, why my words sound like I want
        them to sound. As I'm still young, I can let myself not bother too much
        about creating a conlang quickly, and focus on satisfying my tastes.

        Well, that's how it works for me, at least )

        Kolya


        On Mon, Jan 7, 2013 at 1:50 AM, Gary Shannon <fiziwig@...> wrote:

        > Here's a random thought that occurred to me this afternoon:
        >
        > To create and evolve a conlang begin by translating 100 reasonably easy
        > sentences.
        >
        > DO NOT write down your lexicon.
        > DO NOT write down how your grammar works, or even any notes about your
        > grammar.
        > DO keep EVERYTHING in your head, and only those 100 sentences are committed
        > to writing.
        > DO NOT gloss the translations, only keep the conlang translation and the
        > English original, since a gloss constitutes notes on how the grammar works.
        > Such notes are forbidden.
        >
        > Each day, translate 10 new sentences. You may refer to your previous 100
        > sentences to refresh your memory about lexicon or grammar, but you may not
        > refer to anything else.
        >
        > After translating the day's 10 sentences, discard the oldest 10 sentence
        > from the top of the list.
        >
        > If you know you once coined a particular word for an earlier translation
        > but it's not in the most recent 100 sentences, if you can't remember it,
        > coin a new word from scratch. Use it or lose it.
        >
        > You MAY keep older translations archived for later followup to trace the
        > history of the language after the experiment has been completed, but YOU
        > MAY NOT refer to any sentence more than ten days old when you are writing
        > each day's new sentences. Nor may you peek at older sentences to refresh
        > your memory.
        >
        > You may recycle an English original for re-translation only after the older
        > translation has been discarded from the list and is no longer visible,
        > preferably after it has been gone for a few days or a week.
        >
        > This procedure allows the conlang to be fluid and flexible. Nothing is ever
        > "cast in stone", and when something changes there's no reason to have to go
        > back and revise earlier documents. There are no earlier documents, except
        > those archived purely for historical reasons, and those aren't expected to
        > be just like the more recent versions. This eliminates inertia and
        > discourages the persistence of ugly constructions just because it would be
        > too hard to go back and change them.
        >
        > This could also be done as a collaborative conlang where the most recent
        > 100 sentences are visible on the web page and the next sentence to be
        > translated is visible only in English. Each participant would be limited to
        > making some number of translations per day, and a maximum of ten new
        > sentences would be exposed each day. That way one participant couldn't log
        > in and re-write the whole conlang in one sitting, and no matter how many
        > people participated, it would never happen that you find all 100 sentences
        > are new since yesterday and the grammar you were familiar with is gone.
        > Each time a participant translated the next sentence, the oldest sentence
        > would disappear, and only the most recent 100 would ever show.
        >
        > For the collaborative version you would be allowed to keep notes on what
        > you _think_ the grammar and lexicon are like, but you are not allowed to
        > collude with other participants to turn the grammar or lexicon in a
        > particular direction. Each participant must be a free agent.
        >
        > An alternative version would start with some smaller number of sentences
        > visible, say 20. Each new translation would cause the oldest to be
        > discarded so that no more than 20 would show. The grammar and vocabulary
        > would start out being very volatile. Then, at certain intervals, say once
        > every 10 new translations, the number of sentences allowed to show would
        > grow by 1. After 100 translations, 30 sentences would show. Once 1000
        > sentence had been translated the visible corpus would contain 120
        > sentences. 1000 sentences after that it would have 220, and so on. As more
        > sentences are translated, eventually the grammar and lexicon begins to
        > stabilize and "mature", so that a longer reference corpus could be somewhat
        > stable, but still allow for gradual evolution.
        >
        > If you are familiar with "simulated annealing"
        > http://en.wikipedia.org/wiki/Simulated_annealing this is similar in that
        > the early (small) corpus allows radical mutation (analogous to "high
        > temperature" in S.A.) and a larger corpus has more "mass" and a lower
        > temperature in that it resists mutation (has begun to crystallize). Just as
        > the aggregate corpus of the English Language (highly crystallized) resists
        > mutation of the that language.
        >
        > In reading the description of simulated annealing just think "small corpus
        > = hot = mutable", "large corpus = cold = frozen (or nearly so)".
        >
        > --gary
        >
      • Jim Henry
        ... I would suggest basing the corpus not exclusively on translated sentences, but largely on originally composed sentences. And, of course, when translating,
        Message 3 of 15 , Jan 7, 2013
        • 0 Attachment
          On Sun, Jan 6, 2013 at 9:04 PM, Rich Harrison <rick@...> wrote:
          >> To create and evolve a conlang begin by translating 100 reasonably easy
          >> sentences.
          >
          > I have some qualms about relying on translation. Doesn't it give the source language too much influence over the structure of the developing conlang? i.e.

          I would suggest basing the corpus not exclusively on translated
          sentences, but largely on originally composed sentences. And, of
          course, when translating, translate from different sources as you
          suggest.

          If collaborative participants are contributing original sentences as
          well as/instead of translations, I would suggest that they provide
          translations of their conlang sentence into more than one natlang
          where possible. That might encourage participants who are speakers of
          more than one native language to participate. And perhaps if Alice
          has contributed a conlang sentence with a translation into English,
          Bob could come along and add a translation of that sentence into
          Welsh, and Carol could add a translation of it into Hixkaryana, in
          addition to new sentences of their own with one or more natlang
          translations. People viewing the site could choose to see only
          conlang sentences with translations into a natlang they know, or all
          sentences but only the translations into a natlang they know,

          I've done something similar to this, though without the discarding of
          old sentences and without an absolute prohibition of interlinear
          glosses. My artlang Lusanja consisted of a corpus of original
          sentences, each with a translation into gzb or English or occasionally
          another language; though with a few of the most complex sentences I
          had an interlinear gloss of at least part of the sentence as well.

          I like Gary's idea of gradually growing the corpus over time while
          still discarding the oldest sentences. 20 or even 100 sentences seems
          a little too small a corpus, but might be good to start with if we can
          grow it gradually (though I'd suggest growing the corpus at maybe a
          third the rate at which participants add new sentences, rather than
          1/10 the rate).

          --
          Jim Henry
          http://www.pobox.com/~jimhenry/
          http://www.jimhenrymedicaltrust.org
        • Gary Shannon
          ... I see two problems with that approach. First, that puts the burden on the participant to think of something original to say. For some people that can
          Message 4 of 15 , Jan 7, 2013
          • 0 Attachment
            On Mon, Jan 7, 2013 at 7:26 AM, Jim Henry <jimhenry1973@...> wrote:
            >
            > I would suggest basing the corpus not exclusively on translated
            > sentences, but largely on originally composed sentences. And, of
            > course, when translating, translate from different sources as you
            > suggest.

            I see two problems with that approach.

            First, that puts the burden on the participant to think of something
            original to say. For some people that can discourage them from
            participation. I've seen it happen with other collaborative projects,
            and in language classes where the teacher tells the student "Say
            something in Spanish." Errk! Brain freeze.

            Second, with other collaborative projects I've done where original
            sentences were used some participants seemed to enjoy interjecting
            wildly improbable sentences. After someone has translated sentences
            like "John gave the book to Mary." and "I have three more cookies than
            you have.", the perverse participant would come along with something
            like "It is a tricky problem to find the particular calibration in
            timing that would be appropriate to stem the acceleration in risk
            premiums created by falling incomes without prematurely aborting the
            decline in the inflation-generated risk premiums."

            They would introduce lexicon items that nobody else would use in a
            million years, and constructions that could hardly be untangled. They
            would coin ridiculously complex terms from organic chemistry, or
            jargon specific to some branch of botany or internal medicine. Why, I
            can't imagine. The result was that the whole collaborative project
            fell apart under the weigh of a handful of show-offs.


            [---snip---]
            And perhaps if Alice
            > has contributed a conlang sentence with a translation into English,
            > Bob could come along and add a translation of that sentence into
            > Welsh, and Carol could add a translation of it into Hixkaryana, in
            > addition to new sentences of their own with one or more natlang
            > translations. People viewing the site could choose to see only
            > conlang sentences with translations into a natlang they know, or all
            > sentences but only the translations into a natlang they know,

            That's a lot of time and effort devoted to a sentence that is
            ephemeral and will disappear from the corpus in a couple more days!

            One alternative would be to have some fixed number of sentences, say
            1000, covering as many different kinds of concepts and constructions
            as possible, and present those sentences in English, Spanish, French,
            Italian, Russian, Japanese, Korean, Marathi, Swahili, etc., etc.

            The whole database of 1000 sentences could be pre-translated into
            perhaps the top 10 languages. Then as new sentences are needed for the
            corpus they are taken from this database and presented in all those
            various languages (or in whichever single language a given participant
            preferred). As the corpus grows, fewer sentences of the 1000 are left
            hidden in reserve, until finally, all 1000 sentences are visible in
            the current corpus and at that point the project is done, and the
            language is "finished".

            [---snip---]
            >
            > I like Gary's idea of gradually growing the corpus over time while
            > still discarding the oldest sentences. 20 or even 100 sentences seems
            > a little too small a corpus, but might be good to start with if we can
            > grow it gradually (though I'd suggest growing the corpus at maybe a
            > third the rate at which participants add new sentences, rather than
            > 1/10 the rate).

            I'm not sure what the rate of corpus growth should be. I jut picked a
            number off the top of my head. That's a parameter that would have to
            be fine tuned.

            > Jim Henry

            ---------------------------------

            On Mon, Jan 7, 2013 at 1:39 AM, Nikolay Ivankov <lukevilent@...> wrote:
            >
            > I don't know, how can this work for me.
            >
            > I'm always starting from two different ends - pure grammar and pure
            > phonology, and then start matching them up by trying to create a lexicon.
            > I've never got far in creating lexicon, because, as an artlanging fan, I
            > love exceptions. And making exceptions suggests going at least one step
            > back, to a protolang where they still were rules. In case of Yanyarin it's
            > four steps back, but now I know precisely, why my words sound like I want
            > them to sound. As I'm still young, I can let myself not bother too much
            > about creating a conlang quickly, and focus on satisfying my tastes.
            >
            > Well, that's how it works for me, at least )
            >
            > Kolya

            Obviously the method I'm suggesting is very different from the method
            you prefer. My goal is not to _plan_ a conlang. Not one single user of
            any natlang in the world ever sat down to plan the grammar or
            phonology of their native language. Natlangs are "used into existence"
            by people who do no planning and know nothing about linguistics.

            This project might not even appeal to conlangers for the very reason
            that it is NOT about planning a language, or any aspect of that
            language, but is about using the language to do some short, easy
            translations, keeping everything in your head for maximum fluidity or
            instability.

            Nothing is written down for the very reason that once it's written
            down it becomes too rigid and inflexible. Language evolves when people
            forget how something "is supposed to be" done, or how it was done in
            years gone by. English lost case endings when people forgot how to use
            case endings and started using word order instead. "Went" became the
            past tense of "go" when people forgot that it used to be the past
            tense of "wend". For a conlang to evolve stuff has to be forgotten,
            discarded and replaced constantly. The mistake I made in my previous
            collaborative conlang projects was that nothing was ever forgotten.
            The whole corpus was always there like an anchor holding back all
            growth and evolution.

            This project is entirely based on the requirements that the language
            be unplanned, that it be undocumented, and that it be kept maximally
            fluid and unstable.

            Don't forget that English took great leaps forward in evolution during
            a period when few people could read or write and had to keep it all in
            their heads. There was no massive corpus that fought back against
            change, and when the kids "got it wrong" they created a new and better
            language by doing so. (Yes, I know better than to claim one language
            is better than another. I'm saying this for dramatic effect. Call it
            poetic license in support of a cause!)

            --gary
          • Alex Fink
            ... I think the benefits of original compositions outweigh these. Particularly in the lexical domain, as Rick pointed out: if your first ten English sentences
            Message 5 of 15 , Jan 7, 2013
            • 0 Attachment
              On Mon, 7 Jan 2013 09:54:56 -0800, Gary Shannon <fiziwig@...> wrote:

              >On Mon, Jan 7, 2013 at 7:26 AM, Jim Henry <jimhenry1973@...> wrote:
              >>
              >> I would suggest basing the corpus not exclusively on translated
              >> sentences, but largely on originally composed sentences. And, of
              >> course, when translating, translate from different sources as you
              >> suggest.
              >
              >I see two problems with that approach.

              I think the benefits of original compositions outweigh these. Particularly in the lexical domain, as Rick pointed out: if your first ten English sentences are about a "girl", then dollars to donuts the language being constructed will end up with a term which is basically a translation equivalent for "girl". Because what can participants do but try to look up what was done earlier, and then all too often notice a word glossed "girl" and stop there? I kinda suspect that no corpus-based method can avoid that difficulty, unless you had some explicit auxiliary tool to make it _easy_ (indeed _default_) to look up not an English gloss, but all sentences containing words in that region of semantic space that might or might not be glossed with the same English word.

              As for syntax, well, that seems to be a little less of a problem in practice, but I suspect that's (1) because making a properly different lexicon is an order of magnitude more work than making a properly different syntax; and (2) because the material being translated is often syntactically impoverished. For whatever reason, people preparing language instructional materials tend to come up with sentences of this sort of over-controlled form, which are often just not very representative of real language use; you get pragmatically unnatural ones which scarcely anyone would ever need to say. These have been called _postillion sentences_ (Crystal's article in the WP bibliography is good):
              http://en.wikipedia.org/wiki/My_postillion_has_been_struck_by_lightning

              For example, we had this discussion awhile back about the fact that explicit noun phrases (aot pronouns) are statistically rarish as subjects of transitives (and how that's not obvious as a fact): in actual discourses, subjects have a definite tendency already-introduced topics. But the postillionate tradition ignores that, and throws out hosts of sentences like "The boy gave the cookie to the girl in the kitchen yesterday". Working from that probably biasses your conlang not to have naturalistic-useful topic-tracking strategies.

              Really, it'd be nice if the contributors could write _dialogues_ of various sorts in the course of construction, though one line at a time, to retain that part of the "easy to contribute" feel, probably. A lot of conlangs don't give a lot of consideration to phenomena larger than the sentence, discourse-structuring phenomena and pragmatics and illocutionary force and all that. So

              >First, that puts the burden on the participant to think of something
              >original to say. For some people that can discourage them from
              >participation. I've seen it happen with other collaborative projects,
              >and in language classes where the teacher tells the student "Say
              >something in Spanish." Errk! Brain freeze.

              Maybe prompts (ignorable, for those who have better ideas) would be a good middle ground. "In this language, how would you describe what your house looks like / order a coffee / explain how to fix a bicycle / try to get your boss to give you next Monday off / etc. etc."; they could of course be set more or less explicit, and suggesting shorter or longer responses.

              >Second, with other collaborative projects I've done where original
              >sentences were used some participants seemed to enjoy interjecting
              >wildly improbable sentences. After someone has translated sentences
              >like "John gave the book to Mary." and "I have three more cookies than
              >you have.", the perverse participant would come along with something
              >like "It is a tricky problem to find the particular calibration in
              >timing that would be appropriate to stem the acceleration in risk
              >premiums created by falling incomes without prematurely aborting the
              >decline in the inflation-generated risk premiums."
              >
              >They would introduce lexicon items that nobody else would use in a
              >million years, and constructions that could hardly be untangled. They
              >would coin ridiculously complex terms from organic chemistry, or
              >jargon specific to some branch of botany or internal medicine. Why, I
              >can't imagine. The result was that the whole collaborative project
              >fell apart under the weigh of a handful of show-offs.

              You'll excuse me if I react a bit allergically to allegations of "showing off". For my part, and for what I understand of other conlangers I know well, I just like to have a nerdy sort of fun with language, and composing intricate examples that use the rules in unexpected combinations is the most developed expression of this sort of nerdy fun. In the few collaborative projects I've been in myself I've never acted for the sake of showing off, and if I'd been accused of it it would've smacked stingingly of anti-intellectualism, even if the intent behind it could've been a reasonable one, such as your "hey, try to stick to words someone else might ever reuse; every word you introduce is someone else's word consigned to oblivion".

              The only collablang I know well of enough standing to make a statement about, Kalusa, wasn't killed by anything like this at all. It was killed because its core following got few enough that it became vulnerable to explosions of essentially personal drama, one of which proved fatal. Perhaps underlying this was the fact that it was hard to get into unless you were there from the start (witness late contributions that weren't even sentences), so there could be only attrition, not reinforcement, of the ranks. Would this rolling window idea help that?

              Beyond which, to be honest, I am _bored_ of the subject matter of material like McGuffey's, boys and girls playing with their balls in the park; similarly with that of many first courses in foreign languages, all books and teachers and desks and classrooms. I'd rather not be stifled with it.

              >On Mon, Jan 7, 2013 at 1:39 AM, Nikolay Ivankov <lukevilent@...> wrote:
              >>
              >> I don't know, how can this work for me.
              >>
              >> I'm always starting from two different ends - pure grammar and pure
              >> phonology, and then start matching them up by trying to create a lexicon.
              >> I've never got far in creating lexicon, because, as an artlanging fan, I
              >> love exceptions. And making exceptions suggests going at least one step
              >> back, to a protolang where they still were rules. In case of Yanyarin it's
              >> four steps back, but now I know precisely, why my words sound like I want
              >> them to sound. As I'm still young, I can let myself not bother too much
              >> about creating a conlang quickly, and focus on satisfying my tastes.
              >>
              >> Well, that's how it works for me, at least )
              >>
              >> Kolya
              >
              >Obviously the method I'm suggesting is very different from the method
              >you prefer. My goal is not to _plan_ a conlang. Not one single user of
              >any natlang in the world ever sat down to plan the grammar or
              >phonology of their native language. Natlangs are "used into existence"
              >by people who do no planning and know nothing about linguistics.
              >
              >This project might not even appeal to conlangers for the very reason
              >that it is NOT about planning a language, or any aspect of that
              >language, but is about using the language to do some short, easy
              >translations, keeping everything in your head for maximum fluidity or
              >instability.

              I'd qualify a bit: the _experiment_ may be of interest, but I'd expect the resulting conlang qua language will not be very interesting to conlangers, because the more you rely on spontaneity the more you'll end up with holes being filled exactly the way they are filled in the contributor's L1. That's what's behind the sort of influences you see in real-world language contact situations, after all. Or, given that you are encouraging people to invent things, you'd have a good chance of ending up with (the same grammar but) new lexis -- again, that's the most common difference between registers, especially registers constructed to exclude outsiders (which gives an impetus for _intentional_ change, something otherwise generally lacking).

              >Nothing is written down for the very reason that once it's written
              >down it becomes too rigid and inflexible. Language evolves when people
              >forget how something "is supposed to be" done, or how it was done in
              >years gone by. [...]
              >Don't forget that English took great leaps forward in evolution during
              >a period when few people could read or write and had to keep it all in
              >their heads. There was no massive corpus that fought back against
              >change

              Language evolves when people simplify paradigms or constructions which previous evolution has made too complex, or when they outright ditch a complex construction for a simple one that's simultaneously available, or when the way a different language known to most of the speakers does it ousts the original one, etc... but people don't just "forget" a word and make up from whole cloth a new one to replace it, for who else would understand them? And there *is* a massive corpus of sorts: the many many more than 100 sentences (far more than you can expect a project participant to take up!) that had been heard before by each of the many other members of the speech community, who if they are competent with the old form have no particular reason to ditch it for the new one (though a whim of fashion might win the day for it).

              No-one learns language through writing, as it seems to me they'd need to for writing to have a properly chilling effect on language change; that's what we mean when we say speech has _primacy_. What writing tends to bring with it is the establishment of a formal standard and the consequent downgrading in social class of the spoken form, which continues to evolve naturally as it always does -- though then there are inevitable influences between the two that muddy the waters, and we end up with the remarkable variety of patterns of diglossia.

              Alex
            • neo gu
              ... Coincidently, I ve been working on a grammar that allows at most one noun phrase per clause (counting oblique phrases as separate clauses). Making up short
              Message 6 of 15 , Jan 7, 2013
              • 0 Attachment
                On Mon, 7 Jan 2013 14:52:21 -0500, Alex Fink <000024@...> wrote:
                >
                >As for syntax, well, that seems to be a little less of a problem in practice, but I suspect that's (1) because making a properly different lexicon is an order of magnitude more work than making a properly different syntax; and (2) because the material being translated is often syntactically impoverished. For whatever reason, people preparing language instructional materials tend to come up with sentences of this sort of over-controlled form, which are often just not very representative of real language use; you get pragmatically unnatural ones which scarcely anyone would ever need to say. These have been called _postillion sentences_ (Crystal's article in the WP bibliography is good):
                > http://en.wikipedia.org/wiki/My_postillion_has_been_struck_by_lightning
                >
                >For example, we had this discussion awhile back about the fact that explicit noun phrases (aot pronouns) are statistically rarish as subjects of transitives (and how that's not obvious as a fact): in actual discourses, subjects have a definite tendency already-introduced topics. But the postillionate tradition ignores that, and throws out hosts of sentences like "The boy gave the cookie to the girl in the kitchen yesterday". Working from that probably biasses your conlang not to have naturalistic-useful topic-tracking strategies.
                >
                >Really, it'd be nice if the contributors could write _dialogues_ of various sorts in the course of construction, though one line at a time, to retain that part of the "easy to contribute" feel, probably. A lot of conlangs don't give a lot of consideration to phenomena larger than the sentence, discourse-structuring phenomena and pragmatics and illocutionary force and all that. So
                >
                >Alex

                Coincidently, I've been working on a grammar that allows at most one noun phrase per clause (counting oblique phrases as separate clauses). Making up short examples is definitely a challenge -- I frequently have to resort to putting a topic clause in front of the example, probably to an unnatural extent.
              • Gary Shannon
                ... [---snip---] ... I think many conlangers would be capable of things like woman-child or putting a diminutive ending on woman , or deciding that the
                Message 7 of 15 , Jan 7, 2013
                • 0 Attachment
                  On Mon, Jan 7, 2013 at 11:52 AM, Alex Fink <000024@...> wrote:
                  > On Mon, 7 Jan 2013 09:54:56 -0800, Gary Shannon <fiziwig@...> wrote:
                  >
                  >>On Mon, Jan 7, 2013 at 7:26 AM, Jim Henry <jimhenry1973@...> wrote:
                  >>>
                  [---snip---]
                  >
                  > I think the benefits of original compositions outweigh these. Particularly
                  > in the lexical domain, as Rick pointed out: if your first ten English
                  > sentences are about a "girl", then dollars to donuts the language being
                  > constructed will end up with a term which is basically a translation
                  > equivalent for "girl".

                  I think many conlangers would be capable of things like "woman-child"
                  or putting a diminutive ending on "woman", or deciding that the "girl"
                  is a peasant serving wench.

                  > Because what can participants do but try to look up what was done earlier,
                  > and then all too often notice a word glossed "girl" and stop there?

                  What is the "gloss" you speak of? The rules expressly forbid glosses.

                  [---snip---]


                  > Maybe prompts (ignorable, for those who have better ideas) would be a
                  > good middle ground. "In this language, how would you describe what
                  > your house looks like / order a coffee / explain how to fix a bicycle
                  > / try to get your boss to give you next Monday off / etc. etc."; they
                  > could of course be set more or less explicit, and suggesting shorter
                  > or longer responses.
                  >
                  Or define a specific domain, like a rural town with farms and farmer's
                  markets. The conlang would be purpose-built for buying and selling
                  produce and animals. That's a pretty limited domain, but when you want
                  to model a complex process doesn't it make sense to restrict the
                  model? Look at "Blocks World"
                  http://en.wikipedia.org/wiki/Blocks_world

                  [---snip---]

                  >
                  > You'll excuse me if I react a bit allergically to allegations of "showing off".
                  > For my part, and for what I understand of other conlangers I know well,
                  > I just like to have a nerdy sort of fun with language, and composing
                  > intricate examples that use the rules in unexpected combinations is
                  > the most developed expression of this sort of nerdy fun.

                  Granted, but if the specific objective of an experiment is to evolve
                  ordinary day-to-day discourse in a conlang then intricate examples are
                  not appropriate to the experiment.

                  [---snip---]

                  > The only collablang I know well of enough standing to make a statement
                  > about, Kalusa, wasn't killed by anything like this at all. It was killed
                  > because its core following got few enough that it became vulnerable to
                  > explosions of essentially personal drama, one of which proved fatal.
                  > Perhaps underlying this was the fact that it was hard to get into unless
                  > you were there from the start [---snip---] Would this rolling window idea
                  > help that?

                  I think it very well could. A newcomer to the project only has 20 or
                  30 sentences to refer to. And each newcomer could contribute
                  innovations that old-timers hadn't thought of. They would be like the
                  kids that make up new slang to the eternal consternation of the old
                  fogies who complain that kids don't know how to speak properly.

                  > Beyond which, to be honest, I am _bored_ of the subject matter of
                  > material like McGuffey's, boys and girls playing with their balls in the
                  > park; similarly with that of many first courses in foreign languages,
                  > all books and teachers and desks and classrooms. I'd rather not be
                  > stifled with it.

                  Participation would, of course, be voluntary. ;-)

                  [---snip---]

                  [---snip---]
                  > ... but people don't just "forget" a word and make up from whole cloth a new
                  > one to replace it, for who else would understand them? And there *is*
                  > a massive corpus of sorts: the many many more than 100 sentences (far
                  > more than you can expect a project participant to take up!) that had been
                  > heard before by each of the many other members of the speech community,
                  > who if they are competent with the old form have no particular reason to ditch
                  > it for the new one (though a whim of fashion might win the day for it).

                  Good point. But this thought experiment is really mean to to be
                  nothing more than an over-simplified model of language evolution. It's
                  a toy. It's just for fun. Whether the language would be interesting to
                  conlangers, as you mentioned (in the snipped portion) is irrelevant.
                  And I guess what I'm suggesting is not really conlanging at all. Who
                  know WHAT it is? I don't know.

                  But at any rate, I think, as a solo project, it still might be fun to
                  try. For me, anyway.
                  It might even be interesting to start with a picture. Something from a
                  children's book, like a farmyard scene with lots of different animals
                  in it. The first 100 sentences could be describing or talking about
                  anything in the picture, but could not mention anything not pictured.

                  Then a second picture is added. Maybe the kitchen inside the farmer's
                  house, and another 100 sentences created. No English, No Spanish, no
                  models, just a pictured domain. Then a picture of a railroad train in
                  the station with passengers coming and going. Then, 100 sentence later
                  add a picture of a factory or whatever, enlarging the domain
                  gradually, starting with something as restricted as "Blocks World" and
                  gradually expanding.

                  Maybe it wouldn't work as a collaborative project. That was really
                  just an afterthought. But as a solo project I think it would be fun.

                  --gary
                • Randy Frueh
                  I like the basic idea. It has given me a few ideas for a project of my own. I like the rolling window of sentences to work with but I think I d also keep a
                  Message 8 of 15 , Jan 7, 2013
                  • 0 Attachment
                    I like the basic idea. It has given me a few ideas for a project of my own.
                    I like the rolling window of sentences to work with but I think I'd also
                    keep a sort of "conceptual index". No direct translations in the index,
                    just base concepts and coined terms that have to do with them. I am going
                    to work on this as a group project so I will have people to speak to
                    regularly.
                    This creates the possibility for a rolling window of sentence
                    constructions. Using the language a certain minimum amount and then
                    reviewing what constructions don't come up. Creative wordplay would be
                    encouraged and any changes to the grammar could be made so long as you are
                    still understandable. I think that by encouraging change in word order,
                    pronunciation, even tempo to suit peronal taste would evolve it pretty
                    naturally. Any change would have to be accepted or at least understood by
                    the majority.
                    I guess I'm not restricting notes for this so much as trying to create
                    fluency in a short time.
                    I've always felt odd when I create a language and end up speaking it like
                    an immigrant with no exposure to their new land.
                    On Jan 7, 2013 4:06 PM, "Gary Shannon" <fiziwig@...> wrote:

                    > On Mon, Jan 7, 2013 at 11:52 AM, Alex Fink <000024@...> wrote:
                    > > On Mon, 7 Jan 2013 09:54:56 -0800, Gary Shannon <fiziwig@...>
                    > wrote:
                    > >
                    > >>On Mon, Jan 7, 2013 at 7:26 AM, Jim Henry <jimhenry1973@...>
                    > wrote:
                    > >>>
                    > [---snip---]
                    > >
                    > > I think the benefits of original compositions outweigh these.
                    > Particularly
                    > > in the lexical domain, as Rick pointed out: if your first ten English
                    > > sentences are about a "girl", then dollars to donuts the language being
                    > > constructed will end up with a term which is basically a translation
                    > > equivalent for "girl".
                    >
                    > I think many conlangers would be capable of things like "woman-child"
                    > or putting a diminutive ending on "woman", or deciding that the "girl"
                    > is a peasant serving wench.
                    >
                    > > Because what can participants do but try to look up what was done
                    > earlier,
                    > > and then all too often notice a word glossed "girl" and stop there?
                    >
                    > What is the "gloss" you speak of? The rules expressly forbid glosses.
                    >
                    > [---snip---]
                    >
                    >
                    > > Maybe prompts (ignorable, for those who have better ideas) would be a
                    > > good middle ground. "In this language, how would you describe what
                    > > your house looks like / order a coffee / explain how to fix a bicycle
                    > > / try to get your boss to give you next Monday off / etc. etc."; they
                    > > could of course be set more or less explicit, and suggesting shorter
                    > > or longer responses.
                    > >
                    > Or define a specific domain, like a rural town with farms and farmer's
                    > markets. The conlang would be purpose-built for buying and selling
                    > produce and animals. That's a pretty limited domain, but when you want
                    > to model a complex process doesn't it make sense to restrict the
                    > model? Look at "Blocks World"
                    > http://en.wikipedia.org/wiki/Blocks_world
                    >
                    > [---snip---]
                    >
                    > >
                    > > You'll excuse me if I react a bit allergically to allegations of
                    > "showing off".
                    > > For my part, and for what I understand of other conlangers I know well,
                    > > I just like to have a nerdy sort of fun with language, and composing
                    > > intricate examples that use the rules in unexpected combinations is
                    > > the most developed expression of this sort of nerdy fun.
                    >
                    > Granted, but if the specific objective of an experiment is to evolve
                    > ordinary day-to-day discourse in a conlang then intricate examples are
                    > not appropriate to the experiment.
                    >
                    > [---snip---]
                    >
                    > > The only collablang I know well of enough standing to make a statement
                    > > about, Kalusa, wasn't killed by anything like this at all. It was killed
                    > > because its core following got few enough that it became vulnerable to
                    > > explosions of essentially personal drama, one of which proved fatal.
                    > > Perhaps underlying this was the fact that it was hard to get into unless
                    > > you were there from the start [---snip---] Would this rolling window
                    > idea
                    > > help that?
                    >
                    > I think it very well could. A newcomer to the project only has 20 or
                    > 30 sentences to refer to. And each newcomer could contribute
                    > innovations that old-timers hadn't thought of. They would be like the
                    > kids that make up new slang to the eternal consternation of the old
                    > fogies who complain that kids don't know how to speak properly.
                    >
                    > > Beyond which, to be honest, I am _bored_ of the subject matter of
                    > > material like McGuffey's, boys and girls playing with their balls in the
                    > > park; similarly with that of many first courses in foreign languages,
                    > > all books and teachers and desks and classrooms. I'd rather not be
                    > > stifled with it.
                    >
                    > Participation would, of course, be voluntary. ;-)
                    >
                    > [---snip---]
                    >
                    > [---snip---]
                    > > ... but people don't just "forget" a word and make up from whole cloth a
                    > new
                    > > one to replace it, for who else would understand them? And there *is*
                    > > a massive corpus of sorts: the many many more than 100 sentences (far
                    > > more than you can expect a project participant to take up!) that had been
                    > > heard before by each of the many other members of the speech community,
                    > > who if they are competent with the old form have no particular reason to
                    > ditch
                    > > it for the new one (though a whim of fashion might win the day for it).
                    >
                    > Good point. But this thought experiment is really mean to to be
                    > nothing more than an over-simplified model of language evolution. It's
                    > a toy. It's just for fun. Whether the language would be interesting to
                    > conlangers, as you mentioned (in the snipped portion) is irrelevant.
                    > And I guess what I'm suggesting is not really conlanging at all. Who
                    > know WHAT it is? I don't know.
                    >
                    > But at any rate, I think, as a solo project, it still might be fun to
                    > try. For me, anyway.
                    > It might even be interesting to start with a picture. Something from a
                    > children's book, like a farmyard scene with lots of different animals
                    > in it. The first 100 sentences could be describing or talking about
                    > anything in the picture, but could not mention anything not pictured.
                    >
                    > Then a second picture is added. Maybe the kitchen inside the farmer's
                    > house, and another 100 sentences created. No English, No Spanish, no
                    > models, just a pictured domain. Then a picture of a railroad train in
                    > the station with passengers coming and going. Then, 100 sentence later
                    > add a picture of a factory or whatever, enlarging the domain
                    > gradually, starting with something as restricted as "Blocks World" and
                    > gradually expanding.
                    >
                    > Maybe it wouldn't work as a collaborative project. That was really
                    > just an afterthought. But as a solo project I think it would be fun.
                    >
                    > --gary
                    >
                  • Roger Mills
                    Much  to agree with in Alex s msg., but I just wanted to bring this up (which everyone has probably heard of at sometime or other,,,, The assignment: form a
                    Message 9 of 15 , Jan 7, 2013
                    • 0 Attachment
                      Much  to agree with in Alex's msg., but I just wanted to bring this up (which everyone has probably heard of at sometime or other,,,,

                      The assignment: form a sentence that includes the Deity, the nobility, motherhood and mystery. Scroll down  a bit for the answer,,,,,,,,,,,
































                      "My God, the countess is pregnant, and noone knows by whom!"
                    • Nikolay Ivankov
                      ... Well, I see no reason why couldn t the presunption of having a language with no or little corpus be allpied to a planned conlang. In fact, that s
                      Message 10 of 15 , Jan 8, 2013
                      • 0 Attachment
                        > On Mon, Jan 7, 2013 at 1:39 AM, Nikolay Ivankov <lukevilent@...>
                        > wrote:
                        > >
                        > > I don't know, how can this work for me.
                        > >
                        > > I'm always starting from two different ends - pure grammar and pure
                        > > phonology, and then start matching them up by trying to create a
                        > lexicon.
                        > > I've never got far in creating lexicon, because, as an artlanging
                        > fan, I
                        > > love exceptions. And making exceptions suggests going at least one
                        > step
                        > > back, to a protolang where they still were rules. In case of Yanyarin
                        > it's
                        > > four steps back, but now I know precisely, why my words sound like I
                        > want
                        > > them to sound. As I'm still young, I can let myself not bother too
                        > much
                        > > about creating a conlang quickly, and focus on satisfying my tastes.
                        > >
                        > > Well, that's how it works for me, at least )
                        > >
                        > > Kolya
                        >
                        > Obviously the method I'm suggesting is very different from the method
                        > you prefer. My goal is not to _plan_ a conlang. Not one single user of
                        > any natlang in the world ever sat down to plan the grammar or
                        > phonology of their native language. Natlangs are "used into existence"
                        > by people who do no planning and know nothing about linguistics.
                        >
                        > This project might not even appeal to conlangers for the very reason
                        > that it is NOT about planning a language, or any aspect of that
                        > language, but is about using the language to do some short, easy
                        > translations, keeping everything in your head for maximum fluidity or
                        > instability.
                        >
                        > Nothing is written down for the very reason that once it's written
                        > down it becomes too rigid and inflexible. Language evolves when people
                        > forget how something "is supposed to be" done, or how it was done in
                        > years gone by. English lost case endings when people forgot how to use
                        > case endings and started using word order instead. "Went" became the
                        > past tense of "go" when people forgot that it used to be the past
                        > tense of "wend". For a conlang to evolve stuff has to be forgotten,
                        > discarded and replaced constantly. The mistake I made in my previous
                        > collaborative conlang projects was that nothing was ever forgotten.
                        > The whole corpus was always there like an anchor holding back all
                        > growth and evolution.
                        >
                        > This project is entirely based on the requirements that the language
                        > be unplanned, that it be undocumented, and that it be kept maximally
                        > fluid and unstable



                        > Don't forget that English took great leaps forward in evolution during
                        > a period when few people could read or write and had to keep it all in
                        > their heads. There was no massive corpus that fought back against
                        > change, and when the kids "got it wrong" they created a new and better
                        > language by doing so. (Yes, I know better than to claim one language
                        > is better than another. I'm saying this for dramatic effect. Call it
                        > poetic license in support of a cause!)
                        >
                        > --gary


                        Well, I see no reason why couldn't the presunption of having a language
                        with no or little corpus be allpied to a "planned" conlang. In fact, that's
                        precisely how my concultures look like, and that's presicely how and why I
                        design my conlang's evolution. The things go forgotten and reanalized. In
                        fact,in constructing a grammar, I also always building a bridge between a
                        protolang - which is in effect a language with no structure - and a neolang
                        which has some features I want. And then, starting with a lang with little
                        known structure, I'm trying to guess how I, as an "illiterate speaker",

                        - first: may have put the words together so that my vis-a-vis was able to
                        understand me (a pidgin)

                        - second: how I could have "mistaken" and what could I have "forgotten" so
                        that the language evolve towards the neolang.

                        This means, in particular, that my neolangs are NOT fixed at all. Once I
                        end up with a nice "mistake" at some point of development, it may reshape
                        the grammar of neolang completely. And so yes, I do rewrite, reshape and
                        reorder my grammars

                        That's basically how Yanyarin has obtained its multiple declinations -
                        people just made classifiers and compounds to cope with a rapid sound
                        erosion (again, not planned by this illiterate stone age culture), and
                        since the compounds were of different orign (say, sometimes one of the
                        words was nominative, sometimes genitive, sometimes locative) and the
                        compounding occurred at least twice, the language ended up with so complex
                        pattern (only to be forgotten in futulang). But that didn't affected a
                        purpose - a polysynthetic language with overly rich vowel inventory.

                        That's basically how Sivarian language got its gender distinction by first
                        syllable - it used to be a definite article). That's also why do verbnouns
                        agree in gender with subject (originally it was Nom vs. Acc in a tripartite
                        lang, but I have to think of what it looks like to a layman, not what could
                        it have been to a linguist). But that again reshaped, but not dismissed the
                        goal - to make a language with very little but still substantial number of
                        verbs.

                        And finally I can get back to a being having an idea of how languages look
                        like, but now not I only know that the past form of "go" is "went", but
                        also why it is so.

                        Kolya
                      • Gary Shannon
                        That s very interesting. Is different from my idea, but the results are the same. You get can get a very rich a detailed conlang that seems natural. --gary ...
                        Message 11 of 15 , Jan 8, 2013
                        • 0 Attachment
                          That's very interesting. Is different from my idea, but the results
                          are the same. You get can get a very rich a detailed conlang that
                          seems natural.

                          --gary

                          On Tue, Jan 8, 2013 at 3:07 AM, Nikolay Ivankov <lukevilent@...> wrote:

                          >
                          > Well, I see no reason why couldn't the presunption of having a language
                          > with no or little corpus be allpied to a "planned" conlang. In fact,
                          > that's
                          > precisely how my concultures look like, and that's presicely how and why I
                          > design my conlang's evolution. The things go forgotten and reanalized. In
                          > fact,in constructing a grammar, I also always building a bridge between a
                          > protolang - which is in effect a language with no structure - and a
                          > neolang
                          > which has some features I want. And then, starting with a lang with little
                          > known structure, I'm trying to guess how I, as an "illiterate speaker",
                          [---snip---]
                        • Nikolay Ivankov
                          ... I call this English Park Conlanging - cooperating with the Nature while trying to understand Her principles. As to the understanding, I still suck ...
                          Message 12 of 15 , Jan 8, 2013
                          • 0 Attachment
                            On Tue, Jan 8, 2013 at 5:48 PM, Gary Shannon <fiziwig@...> wrote:

                            > That's very interesting. Is different from my idea, but the results
                            > are the same. You get can get a very rich a detailed conlang that
                            > seems natural.
                            >

                            I call this "English Park Conlanging" - cooperating with the Nature while
                            trying to understand Her principles. As to the understanding, I still suck
                            :P

                            --gary
                            >
                            > On Tue, Jan 8, 2013 at 3:07 AM, Nikolay Ivankov <lukevilent@...>
                            > wrote:
                            >
                            > >
                            > > Well, I see no reason why couldn't the presunption of having a language
                            > > with no or little corpus be allpied to a "planned" conlang. In fact,
                            > > that's
                            > > precisely how my concultures look like, and that's presicely how and why
                            > I
                            > > design my conlang's evolution. The things go forgotten and reanalized. In
                            > > fact,in constructing a grammar, I also always building a bridge between a
                            > > protolang - which is in effect a language with no structure - and a
                            > > neolang
                            > > which has some features I want. And then, starting with a lang with
                            > little
                            > > known structure, I'm trying to guess how I, as an "illiterate speaker",
                            > [---snip---]
                            >

                            Kolya
                          • Leonardo Castro
                            ... Interesting! Maybe related to this, Spanish-Portuguese past forms of the verbs to go ( ir ) and to be ( ser ) are the same: yo/eu fui/fui tu
                            Message 13 of 15 , Jan 9, 2013
                            • 0 Attachment
                              2013/1/7 Gary Shannon <fiziwig@...>:
                              > Nothing is written down for the very reason that once it's written
                              > down it becomes too rigid and inflexible. Language evolves when people
                              > forget how something "is supposed to be" done, or how it was done in
                              > years gone by. English lost case endings when people forgot how to use
                              > case endings and started using word order instead. "Went" became the
                              > past tense of "go" when people forgot that it used to be the past
                              > tense of "wend".

                              Interesting! Maybe related to this, Spanish-Portuguese past forms of
                              the verbs "to go" ("ir") and "to be" ("ser") are the same:

                              yo/eu fui/fui
                              tu fuiste/foste
                              él/ele fue/foi
                              nosostros/nós fuimos/fomos
                              vosotros/vós fuisteis/fostes
                              ellos/eles fueron/foram
                            Your message has been successfully submitted and would be delivered to recipients shortly.