> > And it would starkly illustrate
> > the fact that an appropriate character encoding does not
> > necessarily directly reflect the phonological structure of
> > a language as represented by that script.
>
> "Not necessarily" is the operative word. The question is whether that
> failure to reflect is tolerable. At present, three possibilities have
> been kicked about:
>
> 1) Encode the vowel signs as combining characters, and therefore after
> the base characters over/under which they appear, logical order be
> damned.
>
> 2) Encode the vowel signs as base characters explicitly ligatured with
> ZWJ to the characters over/under which they appear, in
> logical order.
>
> 3) Encode the vowel signs as base characters implicitly ligatured by
> the font ligaturing table to the base characters
> over/under which they
> appear, in logical order. This alternative requires the use of
> distinct ligaturing tables (perhaps distinct fonts) for different
> modes of use.
How about:
4) Encode the vowel signs as combining characters, after
the base characters they logical follow. Consider them as
"double" [width] combining characters, that happen to
have no "ink" above/below the character they apply to,
but (like double width combining characters) have ink
over/under the glyph for the base character that follows.
(I know Ken wouldn't like it.)
If these vowels may occur on word initial bases, there may
be a problem (esp. at the beginning of lines/paragraphs)...
/kent k
