From: Michael Everson (everson@evertype.com)
Date: Sat Mar 01 2003 - 16:06:14 EST
At 18:05 +0000 2003-03-01, Andy White wrote:
>No one has ever written a Virama after a vowel, never ever. You will
>not find it in any printed matter, except in the Indic FAQ, and in
>articles discussing the FAQ.
>
>*You will however find a Ya-phalaa after a vowel*.
Ya-phalaa can be found after a vowel, and after a consonant. It is a
conjunct form of YA, and is produced by the sequence VIRAMA + YA.
> > >The virama is meant to strip a consonant of its inherent vowel "a".
>> >As a side effect of this it combines two consonants when it sits
>> >between then. It is not an exclusive combining mark so that it can
>> >combine a vowel (how can you strip a vowel of its inherent
>> >vowel?)and a consonsnt as unicode FAQ asks its to be.
>>
>> It may not strip an inherent vowel, but it does attach itself to an
>> independent vowel.
>
>You are muddled here. Dr Mitra is talking about 'Virama'. You seem to be
>talking about Ya-phalaa.
No, I said what I meant. I will say it again. The virama when
attached to a consonant kills the inherent vowel. It is also used to
alter the glyph shape of the preceding consonant, or the following
consonant, or both. In the case of TA + VIRAMA + YA, the YA takes the
squiggly ya-phalaa form. Unusually, the VIRAMA + YA can follow an
independent vowel, in which case there is no "vowel killing" per se,
but in which the the YA takes the squiggly ya-phalaa form. Same
behaviour, same encoding.
>I must cut in here. I am concerned that you seem to think that Bengalis
>write 'pratyeka' as PA + VIRAMA + RA + TA + VIRAMA + YA + E + KA.
>It was not until the advent of ISCII and Unicode that anyone thought of
>japhalaa as being the same as VIRAMA + YA. Ask a Bengali teacher what
>the equivalent to 'Virama-Ya' is, and you will most likely get a blank
>or puzzled face.
Fine. So give them a single key on their input methods if they cannot
abide typing VIRAMA + YA, so that when they press the key the
sequence VIRAMA+YA is entered. Hide the encoding from them if you
must. But don't ask us to change the encoding, which is historically
accurate and works just fine.
> > In Bengali, however,
> > it is not pronounced [pratje:ka] as it is in Sanskrit; it is
>> pronounced [prottek]. Nevertheless, the orthographic syllable is tya,
>> and is so written. It is also the case that in initial syllables,
>> when this ya is used, the sound formed is [ę]; cf. byaakaran.a
>> 'grammar', pronounced [bękOron].
>>
>> Although you may think it illogical, the fact of the matter is that
>> at some stage in Bengali written history, noticing that byaa was
>> pronounced [bę] encouraged an extension of this, so that by applying
>> it to an independent vowel an initial [ę] could be pronounced in
>> initial position. It is Bengalis who invented this practice, not us.
>> We just encoded it.
>
>This notion is wrong. Bengalis invented the placing of 'Yaphalaa' after
>'A' not 'virama'. And you are proposing to encode it in an semantically
>incorrect way.
And Ya-phalaa is encoded as VIRAMA + YA.
Look, this is pretty simple. BYAA (BA + VIRAMA + YA + AA) is
pronounced [bę]. AYAA (A + VIRAMA + YA + AA) is pronounced [ę]. We
encode both in the same way.
>Viramas and vowels should mix in the Unicode encoding scheme. That is
>why we have Vowel Signs.
>E.g. The syllable 'KU' is a combination of 'Ka' and 'Full Vowel U'. It
>is encoded as 'Ka' + 'VowelSignU' is Unicode. It could also be said that
>'KU' can be encoded as 'Ka'+'Virama'+'Full_Vowel_U' - but that is in
>line with the Unicode Indic encoding scheme.
This isn't the same. YA is a consonant, not a vowel sign, and it is
affected by the preceding VIRAMA. As it happens, the special shape
that consonant gets affects the pronunciation in a peculiar way (so
that the sound [ja] isn't heard) but that is incidental.
>(Incidentally that is also why the example in TUS3.0, Chapter 9, page
>214, Figure 9-3 ("Conjunct Formations"), example (4) is also wrong as
>previously discussed. That example 'conjunct' is correctly encoded as
>Ra+'VowelSignVocalicR' in Unicode and not as shown.
Yes, we know that that figure is an error, and have verified it with
Monier-Williams' Sanskrit dictionary, and it will be corrected.
> > I am sure that Unicode and ISCII data can be exchanged with regard to
>> this matter.
>
>Not very easily, the exchange of data will require a 'four-character
>ahead lookup'
One understands that this is not all that difficult.
-- Michael Everson * * Everson Typography * * http://www.evertype.com
This archive was generated by hypermail 2.1.5 : Sat Mar 01 2003 - 16:34:03 EST