W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 1999

Re: PRI - 9 LANG Attribute

From: Jason White <jasonw@ariel.ucs.unimelb.edu.au>
Date: Tue, 27 Apr 1999 11:19:10 +1000 (AEST)
To: Web Content Accessibility Guidelines <w3c-wai-gl@w3.org>
Message-ID: <Pine.SUN.3.95.990427110245.1532B-100000@ariel.ucs.unimelb.EDU.AU>
On Mon, 26 Apr 1999 thatch@us.ibm.com wrote:

> 1) I believe that satisfying this requirement will remove no barriers, let
> alone remove significant barriers.
To the contrary, it will remove significant barriers if there is more than
one language being used in the document, and the document is being read
via text to speech software or translated into braille, as I pointed out
in my original message. If the LANG attribute is not supported, and/or not
provided in the document, the user has no choice but to manually adjust
the speech synthesizer or insert controls for the braille translator in
order to generate accessible output. This would impose an onorous burden
on the user. Thus, failure to supply the LANG attribute amounts to a
serious barrier to accessibility, as does the failure on the part of user
agents and assistive technologies to implement it.
> 2) It is inconceivable to me why the lang attribute internal to the page
> should be priority 2, while the lang attribute on the html element is
> priority 3.

The reasoning is quite straightforward: the base language of the document
can be supplied not only with the LANG attribute, but also, as I
understand it, in HTTP headers; hence the LANG attribute on the HTML
element is not the only available solution. Moreover, a user who is
working principally in one or two languages could readily select a
reasonable default and switch between the different languages when reading
various web pages.

The situation becomes more complex when language changes occur within the
document, even inside sentences. I frequently work with such material and
have had direct experience of the problems associated with producing
braille copies thereof: one has to edit the markup and insert the
language-related codes manually, before sending the text to a braille
translator. I have already indicated that essentially the same problem
arises in the case of text to speech conversion where a multilingual
speech synthesizer is available.

Unless the user agent can accurately ascertain the language of every word
in the document, the only means of overcoming these difficulties is for
the author to specify language changes explicitly in the markup. Prior
discussion of this issue within the WAI context has indicated that such
exacting language identification is difficult, and could in any case be
better carried out by an authoring tool than by the user agent.

In conclusion I would again reiterate the importance of leaving the
priority status of this checkpoint unchanged.
Received on Monday, 26 April 1999 21:19:16 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:29 UTC