X-Digest-Num: 151 Message-ID: <44114.151.899.959273824@eGroups.com> Date: Fri, 28 May 1999 19:40:00 +0200 (MET DST) From: Max Kubierschky Subject: Re: misc practical questions X-Yahoo-Message-Num: 899 Content-Length: 1607 Lines: 38 > The limitation is in the natural langauge side of the interface. If there > is an engine that could generate smooth text from coarse keyword glosses, > we would be very close in Lojban to English, since the parser/glosser can > output coarse keyword glosses. The major limitation here remains in the > lexicon - too few words are defined well enough for an MT program to > translate them. As far as i understand, a "coarse keyword gloss" is some isomorphic transcription of lojban, that is more readable to non lojbanists. True? Have problems similar to the keyword gloss -> english problems yet been attacked (with or without success) in computer linguistics? > Only to the level of the parser/glosser that I mentioned, which would be > "low quality" by any standard. Anything higher would take serious work, > probably funded research or graduate-studies research work, because the > person-months of work needed to produce the lexicon and reasonably fast and > sophisticated code are not going to be easily found among volunteers > working occasionally in their spare time. No doubt that the amount of work needed is high. But how high is it compared to translating tons of linux manuals to a target language by hand (which is actually done on a non fundned basis for some languages)? I don't think that there is lack of resources on the programming side of the problem (considering the tremendous amount of free software on this planet). So the question reduces once again to: How hard is it as a linguistic problem? --- Max P.S: thank you for the quick and accurate answer.