MLS4UNIX Mongolian Language Support For UNIX by Oliver Corff (Nov. 7 1993) UNIX systems are notorious for their treatment of 8-bit characters frequently used in PC environments. This issue directly affects the cross-boundary performance of systems relying on the 8th bit such as MLS which was originally designed for PCs. It became thus necessary to develop an extended (or prefixed) transliteration of Mongolian suitable for data transfer in UNIX systems while maintaining a minimum of in- telligibility. The main task of this extension is to render the vowel gender of a Mongol word. Indigenous Mongol words contain either front (resp. female) or back (resp. male) vowels but rarely a mixture of both. The latter is usually the case with contractions of verb roots and formerly free compounds. Only foreign loans contain free mixtures of male and female vowels. Originally, MLS uses diacritics (a trema) on top of the vowels to mark them as female. Male vowels are unmarked. MLS4UNIX follows the marking scheme but marks the whole word by prefixing it by an underscore. Furthermore, marked "a" (ASC #132) is replaced by "e". If a purely Mongol word con- tains one ore more "e"s, then the underscore is omitted, e.g. "newtruulex". However, "erx _qoloo" requires a prefix in front of "qoloo". In words of unclear vowel structure, vowels are immediately prefixed by the caret (^) the validity of which does not stretch beyond that vowel: "Shag- darsueraen" becomes "Shagdars^uren". Long vowels are treated as single symbols. (I've permitted myself to replace the um- laut symbols in the example by ae and ue resp.) It is confessed that the appearance of this system seems to be a bit clumsy. However it should be kept in mind that the main purpose of this convention is to transport data over system bounderies while ignoring ASCII/BINARY switches of FTPs, KERMITs and other protocols.