TED Conversations

Michael Libres Uy

This conversation is closed. Start a new conversation
or join one »

The viability of implementing a logically constructed Language. Namely Lojban.

We are aware of how profound of an effect language has on its speakers, but we still have yet to implement a constructed language that is constructed on the basis of logic rather than history. Lojban is a language constructed on the basis of predict logic by James Cooke Brown and the Lojban institute and can be found here: http://www.lojban.org/tiki/Lojban
This will be a flexible discussion, if you are wondering if your comment is relevant. Post it!

Lojban has a number of features which make it unique:

Lojban is designed to be used by people in communication with each other, and possibly in the future with computers.
Lojban is designed to be culturally neutral.
Lojban has an unambiguous grammar, which is based on the principles of logic.
Lojban has phonetic spelling, and unambiguous resolution of sounds into words.
Lojban is simple compared to natural languages; it is easy to learn.
Lojban's 1300 root words can be easily combined to form a vocabulary of millions of words.
Lojban is regular; the rules of the language are without exception.
Lojban attempts to remove restrictions on creative and clear thought and communication.
Lojban has a variety of uses, ranging from the creative to the scientific, from the theoretical to the practical. and Here we will discuss the ideas of how to implement the language, the problems of doing so, and all the for and against notions regarding language in general.

progress indicator
  • May 17 2013: To clarify a bit.
    1. The language created by James Cooke Browne is Loglan, which is still supported by The Loglan Institute. Lojban is a revised version created by the Logical Language Group, Robert LeChevalier, Founder and President. The two have diverged significantly within the same broad framework, but both have active followers.
    2. The any non-trivial claim that language has a profound effect on speakers is contentious (the Sapir Whorf Hypothesis, itself never clearly formulated). The suggestion that a language based on the principles of Logic (whatever they may be) will say something significant about this question was the original purpose behind Loglan, but the experiments were never carried through.
    3. Both Loglan and Lojban exist, the one for 58 years in various forms, the other 24 or so, and both are. in some sense, based on the principles of logic, so the issue of viability seemed settled in the affirmative, unless there are further details that are not mentioned.
    4. The characterizations of Lojban are basically correct but perhaps misleading (and are, for the most part, not unique).
    a. Lojban has developed its own culture (related largely to Western computer scientists) and is hardly neutral.
    b. Lojban grammar is unambiguous by fiat and the grammar that guarantees lack of ambiguity is not proven to tie in with principles of Logic.
    c. No one reports Lojban to be easy to lean (this may be partly due to old-fashioned learning tools, now being replaced):its vocabulary is totally unfamiliar (despite claims to the contrary), its basic sentence structure is different from that of familiar languages, lack both prepositions and cases, which latter accounts for its regularity -- there is nothing to change.
    d. Lojban culture goes for precision and the requirement to get things just right has proven to be a hindrance rather than a help for creativity -- constatnly going back to check delays going forward.
    But it still functions.
  • May 14 2013: Just how can a predicate be incomplete? It is what it is (a set of some sort, say). A two-place "go" predicate ("go to", say) may not have everything you might want for your "go" predicate, but it may have all that is needed in a particular case and so, for that case, it is complete. We're into strategy here: is it better to have minimal predicates and expand them at need, or maximal ones leave or cover up pieces you don't need on particular occasions? In fact, Lojban uses both strategies but favors the latter a little because it fudges on the unneeded places: it leaves them blank rather than filling them with what FOPL requires, bound variables of some sort. This results in a small and insignificant ambiguity, but you can be sure that, were that problem taken to be serious and so all insignificant places on any occasion had to be filled, the first system would be far and away the standard one.
  • May 14 2013: Usability is a goal as well. It is ultimately a human language. The logical "core" forms a rather small, albeit omnipresent, part of the language. But usability is secondary to the other goals. If the language is made hard to learn because of the other goals, that's not considered a problem, whereas I think it would be for Esperanto and know for certain it would be for Toki Pona.
  • May 14 2013: You don't get a smaller vocabulary with multiple words, even when inflection is neglected. Quite the opposite. Think of "go": you have the verb "go", you have "to" and "from" that are required to fully use the verb. Then accessing those places requires new nouns altogether, superficially unrelated to "go", like "destination" and "origin", and the latter isn't even unambiguously related to going. English's crippled relative clause system arguably helps here (consider "a place to which someone goes"), but a sentence with more than one relative clause quickly becomes unacceptably awkward, especially in speech. (Note: I concede that "route" and "vehicle" are probably not as intrinsically tied to going as Lojban's gismu list makes them out to be, which is why I don't include them in my argument.)
    • May 14 2013: Well, it probably depends on how you do it. If words are multivalent enough, you can make one do a lot of work (look at toki pona), so "to" can serve as destination and goal and a number of other such notion, while "go" may cover a variety of motions. And so on. A survey of natural languages shows many ways of solving these problems (and conlangs add many more), any of which (well, almost) is compatible with logical structure.
      • May 14 2013: They are compatible with *propositional* logical structure, I agree. With *predicate* logic structure, having sufficiently incomplete predicates starts to seem more and more strange, since the statements stop being actual predications.
  • May 19 2013: Let me try to what I conceive to be your idea of making up a universal language based on logical thinking as well as combining the phonetics of most spoken languages of the world. I agree with your goal and wish your success. Even though I also learned about Esperanto many years ago, but it seems having gradually faded away.
    Here are a few of my speculation of what you are trying to do.
    One example of the sentence " 请坐“ in Chinese seems to be simplest and logical. So this approach should be chosen over others like "Asseyez-vous, s'il vous plait" (in French) seems redundant as well as not too logical to place the word "please" in the back of sit-down. Also the English "sit down" is also a little redundant, because if you are talking with a guest who is standing face to face with you, he ,of course, is to sit down, instead of up.
    Also, in lot of Romance languages, the adjective is frequently located after the noun it is to refer to. Example, "mercado libre" (free market) when it is spoken by Spanish speakers, they don't even give it a thought whether the noun preceded the adjective while the latter is the modifier on the former.
    And the conjugation of the verbs according to the pronouns, in Latin/Romance languages, are not absolutely necessary in the logical sense.
    The basic change of a language, by phonics, into a simplified one, actually have already happened in Korea in their own alphabets, and in Vietnam, by the French alphabets. And also in Japanese in tortuously slow metamorphosis.
    Please tell me how I am doing.
  • thumb
    May 14 2013: how is that better than esperanto?
    • May 14 2013: Esperanto is really meant to be an auxiliary language and not much else, to my knowledge. Consequently it is designed to make importing words easy, to be easily recognizable to speakers of Romance languages, and to have a grammar which resembles that of Romance (and to some extent Slavic) languages, with some of the awkward irregularity removed. Lojban has an entirely different set of goals; in particular it was not designed to ultimately become an auxiliary language. So I wouldn't say it's better or worse, I'd say you're comparing the wrong things. In Lojban I'd say {na'i} which is something like "I can't respond to that because it is ill-formed".
      • May 14 2013: Nice use of {na'i} (I'm proud of that one). It's not that Lojban couldn't gbe an auxlang, at least of the Interglossa sort, for science and cataloging and the like, it is just that no particular effort was made in that direction (although Brown always claimed that his predictes would be easy to learn by most human speakers -- not tested).
      • thumb
        May 14 2013: you did not really answer the question. esperanto is a language built from scratch, not just a patch of existing structures. slavic languages might be an influence, but certainly not a design goal. i don't really understand what an "auxiliary language" would be, but esperanto was designed to be a language for everyone to communicate with. i see no difference in the design goals and target audience of the two. so the question is still open: why lojban over esperanto? why would it have bigger success? why should it have bigger success?
        • May 14 2013: The fact that Esperanto resembles various natural languages isn't a design goal, but it is a means to the overall design goal of serving as an auxiliary language. Since you mentioned you're not familiar with the term, an auxiliary language is a language explicitly designed for communication between individuals that don't share another common language. They are generally designed to be easier to learn than natural languages as well, so that one can easily pick up one language and then communicate with everyone else that has done the same. Languages not designed for this purpose can be said to serve as auxiliary languages without technically being auxiliary languages; for example one could argue that English does this today.

          Lojban doesn't have this in its goals. Before anything else its goal is to be grammatically unambiguous. (Technically it also needs to do so in a nontrivial way, that is, in a way so that the grammar actually tells you something about the meaning of the sentence. Otherwise it could just have a single grammatical class and then every sentence would have a trivial parse tree.) If making this work makes the language sufficiently complicated that it is never widespread, that's acceptable, if perhaps unfortunate.

          As for your last questions, it really depends what you mean by "success". If you mean widespread adoption, I can't really answer that, because I don't consider that to really be a core goal. It'd be great, but isn't really the point.
      • thumb
        May 14 2013: oh, i see. so practical usability is not even a design goal of lojban? in that case, i don't get who the target audience is. for the most people, logic is not fun. for computers or science, it is still too much language like, why don't we have numbers as words? like 22 would mean "tree". it makes little difference. so it seems to be aimed to be learnable by a human, but what for?
        • May 14 2013: Well, "practical usability" depends on what you want to use it for. Loglan, Lojban's ancestor, was designed to test the Sapir Whorf Hypothesis. That required that a number of test subjects become reasonably fluent in the language over a range of topics. This goal is no longer active and the perfection of the language has become central. At the same time, to be called a language, in some evaluative sense of the term, it must be capable of dealing relatively efficiently with a range of topics and so the core group of Lojbanists also includes many who are working on translations or original texts at a variety of level and in different fields. Not that anyone is expected to learn Lojban to read these but just to show that Lojban is capable of doing what natural languages do. This includes being spoken in face to face or at least instantaneous interplay as well as writing (which cuts out numbers as words).
  • May 14 2013: Comments so far are rather irrelevant. The issue is not math versus the world nor phonemic spelling, nor is it merely familiar words or usefulness for ordinary folks. It is about what a logical language would do for the human mind. Brown had rather expansive dreams in that respect, starting with a dubious version of a dubious hypothesis and proceeding with a language which did not test even that. But the question remains: would a language which unambiguously presented the logical structure of any discourse in it lead to more reasonable discourse. We know already that simply learning logic does not help that much and that there are other factors -- factual and emotional, for example -- which enter into reasonableness and about which Lojban and like languages have nothing to say. Of course, we are probably not too clear about what reasonableness is, but getting a language like a cleaned up Lojban, which meets on part of a possible definition, would be a useful step.
  • thumb
    May 13 2013: 2 + 3 = 5... except if 2 are seeds that could germinate and 3 are rocks that cover or crush the seeds... This is not how math works. Why should a language... Will my search ever end to find a reason why the i in the word find is a long sound and the e in search is short..? Me, we, be, he, my, no, so- simplest two letters, vowel at the end, sound is long... except they just forgot about to and do when they made the rule, along with words like see and pie..?

    The word eat is even part of the word sweat, but sweat is not sweet and should be swet like wet.

    2 + 3 = 5... You teach the rule and should be able to read the language, not just make up rules to test the memory.

    All of any music made, or will be made is done with letters a-g... why do we need 26, and are they in that order just to make that song rhyme..?

    How many tools are older than the written word... is too big to fail a reason for not fixing it..?
  • thumb
    May 12 2013: Lojban. Such a great idea and so useless.
    Yes, logic and unambiguous grammar is cool and I see so many benefits for computer-speak and communication, but why, oh why, would you use so many strange words.
    English is international. It is used in so many parts of the internet, why not start from here? (Let's call it EngLog)
    Make it easier to learn and to understand. Then have several levels of proficiency, which introduce more and more logical stuff. EngLog1, EngLog2...

    EngLog1: Just English without grammatical irregularity.
    Example: I be Ricard. You understand EngLog1. Everybody, who read EngLog1 and speak English, can understand it.

    EngLog2: The same, but use stuff in front of verbs to indicate time:
    < for past tense (read: have).
    > for future tense (read will).
    ! for "not/negative/don't" (read: not).
    Example: I < be in Russia 2 years ago, but I !speak Russian. Maybe I > learn it some day.

    EngLog3: As EngLog2, but with parenthesis for structuring.
    Example: If (you be programmer & you !be tired) you >understand(EngLog3 and EngLog4).

    EngLog4: Use object-notation to indicate attributes and possession.
    If (you.car be electric) you.investment be good.
    If (you < ! buy electric car) you.car >be too expensive soon, because oil.price >be higher.

    EngLog5: Maybe something like attribute-selection and logical quantifiers? I don't know.
    Example: I hope: Forall people[age < 3] education >be modern, if politicians >! be stubborn.

    Hehe. Thinking about artificial language rules is is actually quite fun. :-)
    But my point is: If you take something known and morph it more people will understand and use it.
    At the same time it is easier for people, who learn EngLogXY to learn true English. Thus it provides incentives to learn EngLog in the first place.
    I think Lojban is too radical, too alien to understand and learn.

    EngLogXY on the other hand is just an "intrinsic structure"
    So one could learn GermanLog or FrenchLog and then switch to EngLog (new words, but known grammar)
    • May 14 2013: Several of your ideas are difficult to make audiovisually isomorphic, and by the time you get to your later rules, the system looks more foreign than Lojban to me, honestly.

      The more important problem, however, is that a predicate-driven language needs to have "full" predicates, that is, ones which encode full sentences by themselves. For example, in English "go" requires many prepositions to use fully; a logical language really needs a relation "go" that relates goers, destinations, and origins.
      • May 14 2013: I agree that EngLog looks pretty hopeless, but that is largely problems with English orthography (a misnomer if ever). The dividing of predicate up over several words offs some advantages, chiefly in terms of a smaller vocabulary (questionable) and added ease in ignoring what isn't relevant. "He goes to the store from the school along Route A on foot" is neither more or less logical that {ko'a klama lo zarci lo ckule la A lo [foot - no dictionary at hand]}, but saying just "He's going to the store" is relatively simpler.