I'm joining this discussion a little late, and I hope I have understood all the posts on this thread.
(I). If Bangla users prefer display the khandaTa as the halant form, and Ta Virama as the half form; from the OpenType font perspective here is what can be done. In the lookup for half forms, make sure Ta Halant -> glyph.TaHalant For constructing ligatures of conjuncts like Ta Halant Ta; make sure the pre-base substitution lookup contains a substitution glyph.TaHalant Ta -> taTa.Ligature Then, in the lookup for halant forms, make sure there is a rule that does: Ta Halant -> khandaTa The substitution in the Halant form lookup will be applied only when the halant form is invoked eg: Ta halant space (as in the end of a word); or Ta Halant ZWNJ (when a syllable break is forced). Thus for cases like kutsit, utkarsha, utkal, etc; the ZWNJ will force the vowel.I to be rendered as needed. i.e. khandaTa ZWNJ vowel.I C2. In Malayalam, if C1 Chillu C2 do not ligate, and C2 is the base consonant; then the sequence C1 Chillu C2 vowel.E will be displayed as: C1 Chillu vowelE C2. This vowel sign re-ordering is a standard feature for this script, it is not done as an exception for one or more characters-- which was being asked for in Bengali. (II). Occurrence of Ta Virama. Although I haven't been able to locate the Ta Virama displayed in a Bengali dictionary, but I see it being used for a few words in an Assamese dictionary. Eg. TaT: (Ta Ta Halant) Leisure, rest TaTHeRaA: (Ta Ta Halant Ha vowel.E Ra Vowel.Aa) to be confounded TaTKaR: (Ta Ta Halant Ka Ra) to perceive The dictionary is printed using metal-type and it's possible that the typesetter ran out of khandaTa's for this page. But, the use of Ta Virama is still possible even if it is just to fill a space where a khandaTa should have been used instead. (III). This thread also re-raised the question of the use of Virama with vowels. There seemed to be agreement that doing so was illogical. Does Unicode continue to stand by the following recommendation in the FAQ? Vowel_A_zophola_AA = 0985 09CD 09AF 09BE ( a- halant ya -aa ) In Uniscribe we have in place a slightly different shaping structure, which might be better. Which is: Vowel_A_zophola_AA = 0985 ZWJ 09AF 09BE (a ZWJ ya aa). I'd appreciate any thoughts. Thanks, -apurva -----Original Message----- From: Andy White [mailto:[EMAIL PROTECTED]] Sent: Monday, November 18, 2002 12:21 PM To: 'Unicode Mailing List' Subject: RE: Errors in the Indic FAQ A brief reply to Marco Marco wrote: > In some Indic scripts (e.g., Devanagari), left-side matras reorder > around the whole consonant cluster; in some other scripts (e.g., > Tamil, Malayalam), they reorder around the base consonant only: > > Devanagari: Ta Virama ZWNJ Ta MatraI -> MatraI Ta+Virama Ta > > Tamil: Ta Virama (ZWNJ) Ta MatraI -> Ta+Virama > MatraI Ta > > (Notice that ZWNJ is redundant in Tamil, as the rendering would be > identical without it.) > > My assumption is that Bengali, in this respect, behaves with Tamil and > Malayalam. No in general Bengali behaves like Devanagri in this respect, Only where KhandaTa is concerned is Malayalam logic true. > ... The purpose of ZWJ and ZWNJ us one of the few things in Indic > Unicode which is quite clear. It is not clear. > A sequence of consonant+Virama+ZWJ always shows a half form > glyph (such as a the Half-Ta in Devanagari or the Khanda Ta > in Bengali)... >...What's wrong in saying that it is a half form [i.e. Khanda Ta] Yes a consonant+Virama+ZWJ shows a half form but what makes you think that a half Ta should look like a KhandaTa? Why should the Bengali script not be allowed to have a Half Ta? In some fonts the Bengali half Ta is drawn as a smaller raised Ta whilst khandaTa is given as a separate glyph. Remember that Khanda Ta is considered a separate character by Bengalis. > ... ... Andy
