[math-fun] lojban
One possibility for the future is that some highly logical language like lojban will be employed by computer systems like "google translate." (Actually GT presently is unaware of lojban's existence.) See, if the developers of such systems had brains, then they would proceed by the "gcc principle" -- translating language X into a canonical form (such as lojban) then translate to language Y. The work to build N languages into the system would then be 2N. As opposed to the bad design of translating X to Y, which would be (N-1)*N work to build it -- much larger. This also would aid human-computer communication -- the computer could hope to understand what a human lojban speaker is saying, whereas that is very difficult for English and simply is not going to happen anytime soon. Investigating... Wikipedia claims "Nick Nicholas, an Australian linguist, was the first fluent Lojban speaker." A December 1996 web post claimed at that time he was the only one in the world. Lojban was developed during 1987-1997 (building on "loglan" invented 1955). The definition of Lojban was published in a book then frozen during 1997-2002. As of 2013 there are probably fewer than 100 fluent Lojban speakers and perhaps 1000 non-fluent ones. NN wrote book "Lojban For Beginners" coauthored by Robin Turner. Online book by him & Jack Cowan: http://www.academia.edu/1278376/What_Is_Lojban QUOTES: "Lojban is designed to be used by people in communication with each other, and possibly in the future with computers... Lojban has an unambiguous grammar. Lojban has phonetic spelling, and sounds can be divided into words in only one way. Lojban is simple compared to natural languages; it is easy to learn. Lojban’s 1350 root words can be easily combined to form a vocabulary of millions of words. Lojban is regular; the rules of the language don’t have exceptions. Lojban attempts to remove restrictions on creative and clear thought and communication." "There are also linguists interested in Lojban’s potential as an intermediate language in computer-aided translation of natural languages; and Lojban is of interest as a potential stepping-stone for students learning other languages." "Because there are no idioms to shorten expressions, a Lojban text can be longer than the corresponding colloquial English text. The unambiguous linguistic structures that result are a major benefit that makes this worthwhile" (That last does not sound good. Really vocabulary in both human and computer languages should be designed according to principles of the Huffman code; more-used words and expressions being shorter...) "Lojban is actually much simpler than natural languages. It is only slightly more complex in its grammar than the current generation of computer languages (such as C++ and Perl)" "Because Lojban’s grammar is simple, it is easier to learn than other languages. Using flashcard-like techniques, a working vocabulary including the complete set of 1350 root words can take 8-12 weeks of study at 1 hour per day. It is by no means uncommon for people who embark on learning Lojban to be able to write grammatical Lojban within a few days, and to hold at least a limited conversation within a few weeks. Natural languages, especially English, take several years to learn to a comparable level of skill." "Lojban has an unambiguous grammar (proven by computer analysis of a formal grammar), pronunciation, and morphology (word forms). In practice this means that the person who reads or hears a Lojban sentence is never in doubt as to what words it contains or what roles they play in the sentence." "Lojban is culturally fully neutral. Its vocabulary was built algorithmically using today’s six most widely spoken languages: Chinese,Hindi, English, Russian, Spanish, and Arabic."
Other properties claimed for the "lojban" artificial human language: 1. No homonyms. In English "it's" and "its" and "heard" and "herd" sound the same. In lojban that never happens. 2. Also true for multiword phrases. E.g. in English, "cargo shipment" and "car go shipment" and "cargo ship meant" all sound same... In lojban that never happens. Supposedly from the syllable sequence you always can deduce a unique word sequence. This suggests lojban well suited for computer continuous-speech recognition. 3. English contains many ambiguities where it is possible to parse a sentence in more than one valid way-- having different meanings. They give as an example "Time flies like an arrow." Which is the verb -- time, flies, or like? Other examples are where it is unclear which words modifies which others. In lojban there is always exactly one valid parse. Indeed, a computer program was written, where you input your alleged lojban sentence, and computer outputs "that was a valid lojban sentence (or not)" along with some information about the sentence if it was valid. As far as I know, it has not been possible to construct such a computer program for any other human language. This not only would make lojban well suited for computer-human communication, but also for scientific writing where ambiguity is often a big problem. 4. But speakers still can say ambiguous things in lojban by use of "ellipses" where information is intentionally omitted. At least one linguist has complained she found lojban to be TOO unambiguous and logical, which was a pain. Nevertheless the fact at least one fluent lojban speaker exists proves this obstacle is not insurmountable. So, interesting. In other news, it turns out various mathematicians have been involved in the artificial language kooky-world, including Hildegard von Bingen (1098-1179) http://en.wikipedia.org/wiki/Hildegard_of_Bingen Isaac Newton (sketched a new artificial language in an notebook) Hans Freudenthal (1905-1990; published an artificial language designed for communication with space aliens) Andries Evert Brouwer (who actually was a co-author of mine once) speaks Esperanto with his family... Esperanto is the most widely-spoken artificial language (the second apparently is "Klingon" created for the TV show "Star trek"!) and there are at least a few people whose first language is Esperanto. (Such as, I guess, Brouwer's children, if he has any.) Probably fewer than 1000, though. Esperanto supposedly was going to cause World Peace. Apparently it is not regarded as all that well designed (indeed its creator later endorsed "Ido" which was intended to fix the most obvious flaws in Esperanto) but it still looked clear to me on naive inspection that Esperanto was more sensible than any actual language.
Yet another famous mathematician who created an artificial language: Giuseppe Peano (1858-1932). His was merely an improved version of Latin. ------ Having reviewed the history of this, my conclusions: 1. If you think human-computer communication is desirable/feasible via a new language, then something with the unambiguity properties of lojban, BUT it needs to be greatly simplified, would be desirable. (Get rid of, for example, the "emotional tags" feature of lojban. The full grammar of lojban is described in a 600-page book, which is far too much; and about half the lojban sentences even on the lojban.org web site are incorrect as one may verify using their own "jboski" lojban-to-english translator and syntax verifier program!) Is this possible? I do not know. The techniques the lojbanists developed do seem sufficiently powerful to achieve their claimed properties. But the structure of many real-world sentences is simple so that for them those techniques are overkill and waste time and space. I doubt any language can or should be popular if it is (say) 200% inefficient. I think a good language design would have to involve a lot of computer simulation and optimization; mere unaided human designers would not be able to do a good job and/or be sure they had. 2. Esperanto is by far the most popular one, with about 1-3 million who know it at some level and about 1-2 thousand who speak it as first language; and it seems clear such a language would help humanity immensely if it really became universal; but there seems to be zero hope of that. 3. Freudenthal's "Lincos" language designed for communication with space aliens seems of some interest as providing a mathematical model of how it can be that an entity like a human baby, could learn language. -- Warren D. Smith http://RangeVoting.org <-- add your endorsement (by clicking "endorse" as 1st step)
If people are intrigued by this corner of human endeavor, I strongly recommend Arika Okrent's <i>In the Land of Invented Languages</i>. Okrent is trained as a linguist and knows what she's talking about. (There is no relationship between her and Marc Okrand, the linguist who invented the Klingon language.) On Sun, Jan 19, 2014 at 12:42 PM, Warren D Smith <warren.wds@gmail.com>wrote:
Yet another famous mathematician who created an artificial language: Giuseppe Peano (1858-1932). His was merely an improved version of Latin.
------
Having reviewed the history of this, my conclusions:
1. If you think human-computer communication is desirable/feasible via a new language, then something with the unambiguity properties of lojban, BUT it needs to be greatly simplified, would be desirable. (Get rid of, for example, the "emotional tags" feature of lojban. The full grammar of lojban is described in a 600-page book, which is far too much; and about half the lojban sentences even on the lojban.org web site are incorrect as one may verify using their own "jboski" lojban-to-english translator and syntax verifier program!) Is this possible? I do not know. The techniques the lojbanists developed do seem sufficiently powerful to achieve their claimed properties. But the structure of many real-world sentences is simple so that for them those techniques are overkill and waste time and space. I doubt any language can or should be popular if it is (say) 200% inefficient. I think a good language design would have to involve a lot of computer simulation and optimization; mere unaided human designers would not be able to do a good job and/or be sure they had.
2. Esperanto is by far the most popular one, with about 1-3 million who know it at some level and about 1-2 thousand who speak it as first language; and it seems clear such a language would help humanity immensely if it really became universal; but there seems to be zero hope of that.
3. Freudenthal's "Lincos" language designed for communication with space aliens seems of some interest as providing a mathematical model of how it can be that an entity like a human baby, could learn language.
-- Warren D. Smith http://RangeVoting.org <-- add your endorsement (by clicking "endorse" as 1st step)
_______________________________________________ math-fun mailing list math-fun@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/math-fun
Steven Pinker "The Language Instinct" I found a compulsively readable account explaining how natural languages come to be their actual untidy selves --- and by extension, why artificial languages are unfit for purpose outside whatever specialised domain for which they have been designed. WFL On 1/19/14, Allan Wechsler <acwacw@gmail.com> wrote:
If people are intrigued by this corner of human endeavor, I strongly recommend Arika Okrent's <i>In the Land of Invented Languages</i>. Okrent is trained as a linguist and knows what she's talking about. (There is no relationship between her and Marc Okrand, the linguist who invented the Klingon language.)
On Sun, Jan 19, 2014 at 12:42 PM, Warren D Smith <warren.wds@gmail.com>wrote:
Yet another famous mathematician who created an artificial language: Giuseppe Peano (1858-1932). His was merely an improved version of Latin.
------
Having reviewed the history of this, my conclusions:
1. If you think human-computer communication is desirable/feasible via a new language, then something with the unambiguity properties of lojban, BUT it needs to be greatly simplified, would be desirable. (Get rid of, for example, the "emotional tags" feature of lojban. The full grammar of lojban is described in a 600-page book, which is far too much; and about half the lojban sentences even on the lojban.org web site are incorrect as one may verify using their own "jboski" lojban-to-english translator and syntax verifier program!) Is this possible? I do not know. The techniques the lojbanists developed do seem sufficiently powerful to achieve their claimed properties. But the structure of many real-world sentences is simple so that for them those techniques are overkill and waste time and space. I doubt any language can or should be popular if it is (say) 200% inefficient. I think a good language design would have to involve a lot of computer simulation and optimization; mere unaided human designers would not be able to do a good job and/or be sure they had.
2. Esperanto is by far the most popular one, with about 1-3 million who know it at some level and about 1-2 thousand who speak it as first language; and it seems clear such a language would help humanity immensely if it really became universal; but there seems to be zero hope of that.
3. Freudenthal's "Lincos" language designed for communication with space aliens seems of some interest as providing a mathematical model of how it can be that an entity like a human baby, could learn language.
-- Warren D. Smith http://RangeVoting.org <-- add your endorsement (by clicking "endorse" as 1st step)
_______________________________________________ math-fun mailing list math-fun@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/math-fun
_______________________________________________ math-fun mailing list math-fun@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/math-fun
Do you know if Arika is related to Daniel Okrent, who used to write the Ethicist column in the NY Times? --Dan On 2014-01-19, at 3:01 PM, Allan Wechsler wrote:
If people are intrigued by this corner of human endeavor, I strongly recommend Arika Okrent's <i>In the Land of Invented Languages</i>. Okrent is trained as a linguist and knows what she's talking about. (There is no relationship between her and Marc Okrand, the linguist who invented the Klingon language.)
Wikipedia says she's his niece. On Sun, Jan 19, 2014 at 6:26 PM, Dan Asimov <dasimov@earthlink.net> wrote:
Do you know if Arika is related to Daniel Okrent, who used to write the Ethicist column in the NY Times?
--Dan
On 2014-01-19, at 3:01 PM, Allan Wechsler wrote:
If people are intrigued by this corner of human endeavor, I strongly recommend Arika Okrent's <i>In the Land of Invented Languages</i>. Okrent is trained as a linguist and knows what she's talking about. (There is no relationship between her and Marc Okrand, the linguist who invented the Klingon language.)
_______________________________________________ math-fun mailing list math-fun@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/math-fun
participants (4)
-
Allan Wechsler -
Dan Asimov -
Fred Lunnon -
Warren D Smith