From LOJBAN%CUVMB.bitnet@YaleVM.YCC.YALE.EDU Sat Mar 6 22:59:04 2010 Date: Sat, 30 Oct 1993 04:12:15 -0400 Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group X-To: conlang@diku.dk, lojban@cuvmb.cc.columbia.edu Subject: Lojban X-From-Space-Date: Sat Oct 30 00:12:15 1993 X-From-Space-Address: @YaleVM.YCC.YALE.EDU:LOJBAN@CUVMB.BITNET Message-ID: UC> > Not quite that frame of mind. We used several English synonyms and l UC> > computer decide which one matched up with phonemes from other langua UC> > best. In this case, you are correct that "view" was the word chosen UC> > that may or may not have been because of similarity with European la UC> > words. We tried "see" as well. UC> UC> Did you also choose from synonyms in the same way for other UC> languages? UC> (esp. Chinese) Yes and no. Since I am a native English speaker, I have a much broader knowledge of my language's vocabulary, and the tolerable limits of memory hook than I do for other languages. On the other hand, for other languages, I sometimes was limited in synonym ability simply because the dictionary did not list multiple possibilities that I could check on. Also, different people did different languages, and one person was very prone to only trying for a single word for each concept especially in Hindi, which he could barely read. (Arabic was a particular mixed bag, since in addition to synonyms, you have to deal with various infix vowels which affect recognition scores but often make little difference in the concept as seen by an Arabic speaker - but of course we never knew which forms the algorithm would prefer, so Arabic almost certainly got shortchanged in the word making. Chinese and Spanish probably had the best representation among non-English, with Russian not too far behind, though suffering from misunderstandings of Russian morphology and phonetics. So, Chinese is particularly well-covered, especially if you are finding the synonyms we used relevant. Probably not as well covered as English, but there were probably two synonyms or more for manyif not most Chinese contributions. It is nice to know that you have chosen exactly the technique we would prefer people to use (identify the language component, then use it as a memory hook, rather than trying to go the other way). Your testimony is the most powerful statement for our approach I have had in a long time (the recognition score technique has been under almost relentless attack by American English speakers who would probably prefer the words to be more recognizable. lojbab Cc: u7911005@[140.113.4.17] cowan