[YG Conlang Archives] > [Latejami group] > messages [Date Index] [Thread Index] >
The experiment using "na" resulted in both brevity and lack of monotony. However, in a machine translation interlingua, neither of these features are important. Much more important is regularity. Thus, the use of "na" SOLELY to preserve brevity and euphony would be a bad choice. Because of this, of the two solutions I mentioned, I definitely prefer the second (i.e., the one which requires at least one suffix to terminate each word). The results may be lengthier and more monotonous, but it is technically the best solution. However, there may be a third solution. I spent a little more time reading sentences using the current scheme, and I realized that the difficulty is not caused by the presence of semi-vowels - it's caused by having too many CSV's in a short span of words. I discovered that if I just converted most CSV's to CVSV's, the problem disappeared. In other words if I replace a lot of CSV's with CVSV's (especially among the CyV's), I'll solve the problem without needing to change the existing morphology at all. Specifically, I'll change almost all "wi" to "iwe", "ya" to "aya", "ye" to "ewa", "yo" to "oya", and "yu" to "awa". Here are some examples: Old: Bweko dwise vapwe. New: Bweko diwese vapwe. English: The doctor liked the cat. Gyote ligogyose to pa bocay to. Goyate ligogoyase to pa bocay to. Some students asked me for some water. Pyo zeko pa buti pyome veca pa. Poya zeko pa buti poyame veca pa. I'll think about the truck after I cook. Jutay sya bweko Lajonse veco tona dasye datoy. Jutay saya bweko Lajonse veco tona dasewa datoy. John doesn't like the kitchen because of the low ceiling. Sya fawma vyocumpa Lajonse disye dweku. Saya fawma voyacumpa Lajonse disewa dweku. It's true that John closed the two doors. And the counts for Lessons 1-6 are: Ladekwa bytes = 7036, and English bytes = 8360 Let me know what you think. Regards, Rick Morneau http://www.eskimo.com/~ram