From: "Michael Everson" <[EMAIL PROTECTED]> > At 22:37 +0100 2004-01-03, Philippe Verdy wrote: > > >Note that a fundamental property of character identity is its most common > >classification as a vowel, consonnant, or semi-vowel. > > That isn't true. The letter "v" is a vowel in Cherokee, a consonant > in Czech, and (often) a semivowel in Danish.
Also: what are you demonstrating here? Only that distinct languages have their own view on the classification of letters. But it's a fact that this classification is a property of each character used in each language, even if the classification can vary between languages. If simply I accepted and followed your argument, then the classification as letters, digits or symbols, or as uppercase/lowercase/titlecase, or as letters/ligatures would also be irrelevant in Unicode... How will you then reveal a character identity with Unicode? What you say there, if accepted universally, would make Unicode completely unusable in practice, and impossible to unify without separating each language encoded with these scripts. So your argument is better militating for a better documentation of language support and usage for each script, possibly with additional properties that may be missing in Unicode... Thanks we have other sources than Unicode to find these additional properties, notably the other related standards and repertoires from which characters were imported and unified (here I include the mappings of non-Unicode charactere sets onto Unicode-ISO/IEC10646 code points).
