W3C home > Mailing lists > Public > public-html@w3.org > July 2008

Re: Translation control in HTML5

From: Julian Reschke <julian.reschke@gmx.de>
Date: Thu, 31 Jul 2008 23:30:02 +0200
Message-ID: <48922EDA.1020403@gmx.de>
To: Chris Wilson <Chris.Wilson@microsoft.com>
CC: "public-html@w3.org" <public-html@w3.org>, Chris Wendt <Chris.Wendt@microsoft.com>

Chris Wilson wrote:
> I wanted to suggest a feature for HTML5 around autotranslation control.  
> Chris Wendt of Microsoft actually came up with this idea, so I’ve cc’ed 
> him on this mail.
> Situation:
> There are a number of online and offline automated translation services, 
> which can be used on any HTML document. These automated services 
> typically translate the text content of all elements and a few selected 
> attributes, such as the title and alt attributes. Examples of online 
> automated translation services: http://translate.google.com, 
> http://babelfish.yahoo.com, http://translator.live.com. All of these 
> services allow user to enter a URL and translate any accessible web page.
> Problem:
> The translation services translate all elements, including the ones that 
> need to be left untranslated. The document author has no option to 
> control the behavior of the translation service.

Yes, that's a problem.

> ...
> Suggested solution:
> Google has a <meta> name/value that their translation service respects, 
> but it acts on a document level only:
> <meta name="google" value="notranslate">

I was looking for that when Google Translate came out, but of course 
that's not fine-grained enough.

> ...
> In HTML 5, this could be done with a new attribute “translate”, valid on 
> all elements. Values “yes” and “no”. Default is “yes”.  By default 
> attributes are not translatable, alt and title remaining as exceptions. 
> HTML will not introduce new translatable attributes.
> The precedence for this feature comes from the ITS (Internationalization 
> Tag Set) in http://www.w3.org/TR/its/, which in section 6.2 specifies an 
> its:translate="no" attribute and a rule for determining non-translatable 
> content in an XML document. This solves the problem for XHTML content, 
> but (obviously) not for HTML.
> ...

That would work, but it may make pages with technical content very 
chatty. It probably would be good to attach that property implicitly to 
elements like <code>.

Attaching it to style information would work even better, but it would 
conflate semantics with styling, right?

BR, Julian
Received on Thursday, 31 July 2008 21:30:47 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:36 UTC