W3C home > Mailing lists > Public > www-international@w3.org > October to December 1998

Re: Transliteration

From: Martin J. Duerst <duerst@w3.org>
Date: Fri, 16 Oct 1998 15:57:44 +0900
Message-Id: <199810160908.SAA18547@sh.w3.mag.keio.ac.jp>
To: Keld J|rn Simonsen <keld@dkuug.dk>
Cc: carrasco@dragoman.org, Harald.Alvestrand@maxware.no, www-international@w3.org
M.T. Carrasco Benitez wrote:

> > > "Codes for language transformation"
> > > http://dragoman.org/winter/lanco.html

> > >Is there a need to label languages transformations ?
> > >e.g. "this is a Greek text transliterated into English"

Harald Alvestrand wrote:

> > My answer hasn't changed:
> >
> > 1) I don't see a compelling need. Others might.

Braille conversion, which can easily be seen as a form
of transliteration, seems to need this, or at least
subtags to indicate various grades of contraction and
different transliterations for different fields (e.g. Math,...).


> > 2) This is orthogonal to the language code, and trying to extend 1766
> >    to cover this case may be actively harmful.
> >    If needed, it should be a separate label.

Yes, the problem is that currently, language tags are more or less
hierarchical. Adding the extensions that Tomas is proposing would
take away quite a bit of this hierarchy.


Hello Keld,

At 19:20 98/10/15 +0200, Keld J|rn Simonsen wrote:

> If we use locales, and the locale naming scheme of ISO/IEC 15897

I think not many people are familliar with the ISO standards you
mention. Can you give us a bit of background or a reference (URI)?


> then that can be done within that scheme. I believe RFC 1766
> allows for locale names to be referenced.

The way I understand it, RFC 1766 language codes work in parallel
with locale names in as far as they can contain two-letter language
codes followed by two-letter country codes. But these two codes are
separated by a "-". And RFC 1766 language codes don't include
any character encoding related parameters.

Would you use the character encoding to indicate to which script
something is transliterated, i.e. use iso-8859-8 to say something
is transliterated to Greek? That is interesting, but rather limited.
Or is there another mechanism?


> ISO/IEC FCD 14652 has provisions for language to language simple
> tranliteration specifications.

What do you mean by "simple"? Are these just tags, similar to
what Tomas is proposing, or is it a mechanism to actually specify
what should happen in the transliteration (i.e. a Cyrillic C
goes to a Latin S,...)?


Regards,   Martin.
Received on Friday, 16 October 1998 05:11:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 19:16:53 GMT