I wrote: > Vowel marks appearing to the left of the > consonants are pronounced before them; those to the right, after them.
This leads to an interesting, if so far theoretical, Unicode question: how to encode abjads and abugidas that have vowel signs which are pronounced *before* the base consonant. Two Unicode principles, logical order and base-before-combining, are thus put into conflict. In (Feanorian) Tengwar itself, the reading order is actually language-dependent: thus "Quenya" (a Quenya word) is written QU-e-N-y-a (where caps are base, smalls are combining), but "Sindarin" (a Sindarin word) would be "S-N-i-D-R-a-N-i", if written with base-before-combining, or "S-i-N-D-a-R-i-N" if written with logical order, in which case the default grapheme clusters have to be broken up using complex rendering code in order to get i over N and a over R. The problem could be sidestepped with a grapheme-cluster encoding such as is used for Ethiopic, but the feel is very different: Ethiopic vowel signs are normally treated as part of the letter, whereas Tengwar vowel signs are more like typical abjad signs: partly optional indications of "colorings" to the fundamental consonant structure. Unicode tribal elders are invited to mention which of the two conflicting principles they reckon to be the more important. -- John Cowan http://www.ccil.org/~cowan [EMAIL PROTECTED] Please leave your values | Check your assumptions. In fact, at the front desk. | check your assumptions at the door. --sign in Paris hotel | --Miles Vorkosigan