Devanagari [was Re: Any web-published rebuttals to criticism?]

From: Sandeep Sibal (
Date: Thu Jan 09 1997 - 03:54:03 EST

> Is there a list of "rebuttals," clairifications/acknowledgement of
> weaknesses, or a FAQ pertaining to the most common misunderstandings
> about Unicode on the web?

I would like to second this! It would be nice if some of the
contentious issues were catalogued somewhere. Minutes of meetings
etc., would be valuable as well. I would suspect that many of the
thorniest issues arise from languages/scripts that are not Latin
based, and experts in these areas would probably have their
criticisms expressed (if at all electronically) in a medium
that is not English. Pages such as the one you point out are rare.

For an outsider like myself, who has a strong interest in a
particular language (Devanagari in this case), it has been an
uphill battle to understand the logic behind Unicode's
encoding of the script.

In Unicode's encoding of Devanagari, "half-consonants" have no
characters associated with them, and neither do consonant
combines. Half-consonants in Unicode would be derived by
appending an "implicit-vowel remover" to the full-consonant.
This is a very odd construct to say the least. In the way
the language is written, a glyph of a half-consonant
(in general) is not constructed by a combine of the glyph
representing the full consonant with any other glyph. The
use of the "remover" is therefore not script driven.
It appears something like a modifier one would use in a
keyboard mapping scheme. The reason for the use of such
a construct might come from the fact that Unicode's
encoding is based on the ISCII std., whose major objective
was to encode all of Devanagari (and more) in a 7-bit code,
and to map it onto a regular keyboard. Perhaps this is
historic baggage that got tagged along. There are less
than 50 half-consonants, so I'm not talking about
exponential complexity here.

The fact that there are no character positions assigned
to these half-consonants is particularly odd considering
that a whole slew of Latin and Arabic ligatures appear to
be present. Mind you, I have not even begun discussing the
issue of ligatures in the Devanagari script, which (unlike
many of the Western ligatures) can look very different from
the individual consonants/vowels that comprise them. From my
perspective, Unicode appears inconsistent and heavily
biased. It is possible that the apparent inconsistencies are
due to the fact that the Devanagri encoding was established
by folks who spent little time/energy on it, and the encoding,
subsequently, was not critically analysed by others.


> --
> Adrian Havill <> MAIN: +81 (06) 536-6731
> S.E., System Planning & Production Division FAX: +81 (06) 536-6741

Sandeep Sibal
Phone: (908) 949-6277

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:33 EDT