W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > July to September 2001

Re: Bengali on the web?

From: David Woolley <david@djwhome.demon.co.uk>
Date: Mon, 27 Aug 2001 10:01:00 +0100 (BST)
Message-Id: <200108270901.f7R910t20721@djwhome.demon.co.uk>
To: w3c-wai-ig@w3.org
Cc: mdshreza@hotmail.com
mdshreza@hotmail.com wrote:

> My country name is Bangladesh and my language is bengali(bn). My
> question is
> if i want to show my webpages in bengali what i have to do.I want to
> show my

Using the Bengali script is a difficult one because Bengali isn't an
interesting language for browser writers as, presumably, the people
most likely to have computers are likely to be fluent in English, and
because you can't just treat it like the latin alphabet with very 
few ligatures.

Thai is about the only Indic script that seems to get supported by
international software developers.

I presume that the Word 2000 Unicode font does cover Bengali, although,
not having tried it, I'm not sure how much support from the application 
is required (Unicode Bengali is in terms of logical characters, not 
glyphs, except for the initial/final vowel distinctions).  There are some
free TrueType fonts (e.g. itxbeng), consisting of glyphs.  These are
invalidly coded (the Symbol font problem), from a web point of view,
and require additional software to position them correctly, relatively
to each other.

> pages in client browser who has no bengali font installed on his local
> computer.Please inform me details.

This comes under a generalisation of accessibility, which I think is
valid, but some on the list prefer a narrower definition which excludes
things for which money spent on a browser will provide a solution.

If you really have no control over the browsers, I would have to suggest
that this is a case where imaging each line of text is the only practical
solution at the moment; I would use a romanised text as the alt attribute
(the choice of language and coding for text as graphics is more complex
when the graphic is a link to another language environment, and you
wouldn't use romanised forms for presentational material, with the body
text in Bengali characters).

Some other Indian++ languages are suitable for downloadable fonts,
I've seen this for Telugu, but this puts restrictions on the browser,
and I don't think it will work satisfactorily for Bengali (Devanagari,
Gurmukhi, and Gujarati).  The Telugu example used the Symbol hack##,
so is not something that should be particularly encouraged.

If PDF is an option, I would go for that, as it allows embeddable fonts
and has the fine positioning capabilities (even if they may make cut
and paste difficult) needed to present the text properly.

Proper support for North Indian languages in browsers will almost
certainly come with proper fonts, that allow the use of standard character
codes, used properly, so I don't see that you are going to lack the font
and still have a browser capable of the language.

One other option, although if the sticking point is lack of fonts rather
than lack of support for them, the necessary download is probably out
of the question, is SVG.  It has a more powerful font mechanism for
non-European scripts, although this may not work well for small sizes.

Handling these languages in the UK is, if anything, more difficult,
as the likely places for the browser would be public libraries and the
homes of younger relatives whose main working language is English and
feel no need to access their ancestral languages on the web.  The Indian
language versions of a recent publicity campaign about renumbering of the
telephone system were Word documents (all versions were Word) containing
scanned images of each page (I can't remember if they were handwritten).
(Note that I've not worked in this area, its just one were there is an
obvious need.)

(My local council, Brent, which has large numbers of people whose first
language isn't English, chickens out on their web page, and has a note
suggesting the use of Babel Fish, but that will give you a romanised
version, if it supports Bengali at all.)

One thing that you can control is what the server does, and if there
are English versions of the page, I would hope you would do proper
language negotiation.  There is, however, a problem with language 
negotiation at the present, in that it doesn't cater for users with
the language but without the capability to display it in the correct
character set.  A standardised way of requesting romanised Bengali (if
you set your language to hi for Google, you get romanised Hindi, not
Devanagari), Unicode Bengali, text as graphics Bengali or Arabic script
Bengali, would be desirable for this sort of language where workarounds
for lack of browser capability may be needed.

> I need this solution within a very short time.If u please dont mind to

Asking at the last minute is never a good idea, and I don't think there
is a clean solution.

What would help though, is some background on the application, the
audience, where the browsers would be, and the general support of Bengali
on PCs.

## The Symbol hack is selection of a font which appears completely
different from the characters sent; it breaks when the browser really
doesn't support fonts or doesn't have the right one.

++ Not intended to represent any political unit.
Received on Monday, 27 August 2001 09:13:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:13:56 GMT