Robert Brady wrote:
> On Thu, 27 Jul 2000, Abdul Malik wrote:
> > How am I to encode the different forms in unicode?
>
> For the last three, you can do something like
> BENGALI LETTER WHATEVER
> BENGALI VIRAMA
> BENGALI LETTER BA
> for the -va form, and
>
> BENGALI LETTER WHATEVER
> BENGALI VIRAMA
> ZERO WIDTH JOINER
> BENGALI LETTER BA
> for the -ba form.
That's what I was going to post, but you were quicker ;-)
However, this mechanism (VIRAMA + ZWJ) should only be for forcing a
particular *visual* representation of the *same* orthographic entities:
halfConsonant + fullConsonant vs. ligature.
In this case, Abdul Malik claims that there are also *phonetic* differences
("bbo" vs. "bvo", "cvo" vs. "cbo", etc.). So why using this hack rather than
including a proper "va" letter?
Of course, I see the problem: "ba" and "va" would be identical in most
cases, and this may lead to confusion (imagine having two keys showing
identical letters: which is which?).
Nevertheless, there are several examples of characters that are exactly the
same visually, but have been kept separate on the bases that they are
logically different.
E.g., the isolate and final forms of U+06CC ARABIC LETTER FARSI YEH are
identical to U+0649 ARABIC LETTER ALEF MAKSURA, but they have been
differentiated on the basis that they have different pronunciation and
joining behavior. In fact, U+06CC also has initial and medial forms, that
are identical to the corresponding forms of letter U+064A ARABIC LETTER YEH.
_ Marco