From: Gautam Sengupta (gsghyd@yahoo.com)
Date: Sat Apr 22 2006 - 12:08:58 CST
Dear Dr. Ganesan,
> The ad hoc use of ZWJs in non-Devanagari scripts creates lot
> of confusion.
I fully support and endorse this view. In fact it creates a lot of
problems in Devangari as well, where it is used to encode semantic
differences as well as glyph variations. The manner in which PR-37
was resolved by the UTC was most unfortunate. Mercifully Bangla no
longer needs to encode Khanda-Ta with ta|virama|zwj. Given that, I
don't see why Malayalam chillus should be encoded with ZWJ either.
Does anybody understand the logic behind this dispensaton? I
certainly don't.
Best
Gautam Sengupta
--- "N. Ganesan" <naa.ganesan@gmail.com> wrote:
> On 4/18/06, Nagarjuna Venna <vnagarjuna@gmail.com> wrote:
> > Here is a small analysis and a plausible solution for Telugu
> script.
> >
> > Virama is basically used in two places in the current standard
> for Telugu:
> >
> > 1. In a sequence of the form <C1, virama, C2> to generate
> conjuncts. The
> > only thing that the virama is doing here is to tell the renderer
> to generate
> > the vattu from of the consonant C2. Hence, I referred to it as an
> escape
> > character.
> >
> > 2. In a sequence of the form <C1, virama, ZWNJ>, IMHO, this is a
> farce
> > because you have one escape character (ZWNJ) overriding the
> behavior of
> > another escape character.
> >
> > If you consider a conjunct like 'rju' in arjuna, linguistic
> analysis would
> > probably tell you that this syllable is made of the sequence
> <Consonant RA,
> > virama, Consonant JA, Vowel U>. Analyzing Telugu script will tell
> you that
> > it is wirtten using the sequence <Consonant RA, Vattu JA, Vowel
> sign U>. As
> > you can see, the model in ISCII is a hybrid - <Consonant RA,
> virama,
> > Consonant JA, Vowel Sign U>.
> >
> > A plausible solution for Telugu like scripts for the above two
> scenarios is:
> >
> > 1. Encode conjuncts as in the written form <C1, V1>. In the
> general case,
> > <C1, V1, V2,.....VS1> where the V are vattulu and VS is the Vowel
> sign.
> > 2. Encode suppression of inherent vowel (pollu in Telugu) as <C1,
> virama>.
> >
> > It is some times stated that the vattu form of RA, for example,
> is simply a
> > glyph of variation of consonat RA. I believe this claim is bogus
> because
> > vattulu are to consonants what vowel signs are to vowels -
> secondary forms
> > of the basic alphabet that are integral to the writing system.
> (If I were
> > encoding glyph variations of vattu form of RA, I would be
> encoding one code
> > point for the semi-circle glyph, one for the L shaped glyph that
> binds to
> > the left and one more for the mirror form that binds to the
> right.)
> >
> > A scheme like this would work very well for Kannada and most
> likely
> > Malayalam too. Devanagari is a different story because Devanagari
> uses a
> > very different writing system.
> >
>
> The ad hoc use of ZWJs in non-Devanagari scripts creates lot
> of confusion. In fact, cillus of Malayalam are not half-forms
> at all, and to use ZWJs on them, they are claimed so (eg., pr-37).
>
> It is possible that leaving the functionality of ZWJ to Devanagari,
> a C2-conjoining script joiner (let us call it VJ, vattu jr.) will
> help
> issues related to Telugu, Kannada, Oriya. It will help
> Malayalam too, and also (yet to be encoded) Tamil Grantha
> script used by Tamils to write Sanskrit. Perhaps parts
> of Burmese script issues. It may be useful to study
> Tibetan script consonat joiner mechanisms for a solution
> to C2-conjoining scripts of India - Telugu, Kannada, Malayalam,
> Tamil Grantha, Oriya as ultimately their origins go to
> Pallavan era.
>
> Thanks
> N. Ganesan
>
> > Thanks,
> > nagarjuna
> >
>
>
>
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
This archive was generated by hypermail 2.1.5 : Sat Apr 22 2006 - 12:11:40 CST