W3C home > Mailing lists > Public > public-i18n-geo@w3.org > June 2004

RE: Translation of i18n resources [hreflang]

From: Richard Ishida <ishida@w3.org>
Date: Tue, 29 Jun 2004 20:30:04 +0100
To: "'Masayasu Ishikawa'" <mimasa@w3.org>
Cc: <public-i18n-geo@w3.org>, <karl@w3.org>
Message-Id: <20040629193004.15B004F29F@homer.w3.org>

Hi Masayasu,

The points below are (as usual) all useful. To be honest, I was having
similar thoughts as I wrote the page. This discussion also provides useful
input to proposed text for our techniques documents, and could be easily
made into an FAQ.

Our intent (not yet realised, but easily implemented) was to use CSS to flag
links to pages in a different language, so that the user was aware before
they went there that they may not be able to read the text. This would be
achieved by automatically adding [<hreflang-value>] after the link, eg. "
...see my document [en]".

Note that the necessary CSS doesn't work on IE. So this was never going to
be a perfect solution anyway.

Whether done by CSS or with text, it's difficult to decide what's best.  

On the one hand, there is usability value in warning users before they
travel a link that they may not be able to read the page at the other end. A
user who makes that mistake several times may ignore links that do lead to
readable text (especially if they are on a dialup connection).

On the other hand, there are the dangers that such information becomes out
of date or is inappropriate.

I agree that if one uses this approach, one should:
-	make a best effort to ascertain whether the link is content
negotiated, 
-	check from time to time whether new translations are now available,
and change the link and hreflang information if needed

I would like to add here that one should only use these 'flags' if one is
absolutely sure that the linked resource is unlikely to be translated in the
future, but:
-	it's sometimes hard to be sure
-	inconsistent application of flags might be worse than none at all
(the user may come to mistrust the flags)

Thoughts and suggestions welcome from all those copied.

RI


PS: I copied Karl, because I know he has an interest in this topic.



> From: Masayasu Ishikawa [mailto:mimasa@w3.org] 
> Sent: 29 June 2004 09:53
...

> - "Linking to other pages" - "Please add hreflang information to all
>   links that point to a document in another language than that of
>   your translation."  I'm less sure about this.  I used to use the
>   hreflang attribute more often, but a linked resource may be using
>   language negotiation, or may adopt language negotiation later.
> 
>   For example, "FAQ: Apache language negotiation set up" [2] includes
>   links to Apache's documentation, which was only available in English
>   in old times but is now available in multiple languages.  Although
>   Apache version 2.0 documentation is currently not available 
> in French,
>   some day they might provide a French version as well.  When 
> I translate
>   this document into French now, what should I do for a link to Apache
>   version 2.0 documentation?  The linked document is indeed in another
>   language(s) than that of my translation at the moment, but I cannot
>   say which, and in the future the situation might change.
> 
>   So personally I don't use the hrelang attribute these days 
> unless I'm
>   absolutely sure that the linked resource is indeed written in that
>   particular language and is unlikely to change in the future (e.g.
>   a link to the specific language variant of language-negotiated
>   resource). 
> 
>   Also what should I do when a linked resource is a multilingual
>   resource?  Do I have to add hreflang="mul" or something?
> 
>   Anyway, if you advise translators to add hreflang information to
>   *all* links to a document in another language, you should elaborate
>   the correct usage of it.
> 
> [1] http://www.w3.org/International/questions/qa-controls
> [2] http://www.w3.org/International/questions/qa-apache-lang-neg
> 
> Regards,
> --
> Masayasu Ishikawa / mimasa@w3.org
> W3C - World Wide Web Consortium
> 
Received on Tuesday, 29 June 2004 15:30:06 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:12:38 GMT