W3C home > Mailing lists > Public > public-pfwg-comments@w3.org > January to March 2008

Re: Comment on WAI-ARIA Role

From: Aaron M Leventhal <aleventh@us.ibm.com>
Date: Mon, 25 Feb 2008 11:21:06 +0100
To: "Neil Soiffer" <Neils@dessci.com>
Cc: public-pfwg-comments@w3.org, Richard Schwerdtfeger <schwer@us.ibm.com>, w3c-wai-pf@w3.org, w3c-wai-pf-request@w3.org
Message-ID: <OF013289F6.2C81E74B-ONC12573FA.0038D3A3-C12573FA.0038FFAD@us.ibm.com>
If we do this, we'll need to use the object attribute xml-roles to expose, 
because there is no role for math or equations in ATK/AT-SPI.
There's a ROLE_EQUATION in MSAA but I was planning to use that as a 
container for MathML equations, which could have multiple objects.

- Aaron

"Neil Soiffer" <Neils@dessci.com> 
Sent by: w3c-wai-pf-request@w3.org
02/25/2008 10:34 AM

Richard Schwerdtfeger/Austin/IBM@IBMUS
public-pfwg-comments@w3.org, w3c-wai-pf@w3.org, w3c-wai-pf-request@w3.org
Re: Comment on WAI-ARIA Role

We have been doing some experiments, but don't have a product for it. 
However, MathType 6 does allow you to copy images from sources like 
wikipedia which contain TeX and paste them into MathType.  It recognizes 
that the TeX and does the conversion to its internal format.  So, while 
not a web browser solution, there is a proof of concept out there.  We 
have gotten very positive feedback on this feature (it was added to our 
latest release last fall).

Neil Soiffer
Senior Scientist
Design Science, Inc.
~ Makers of Equation Editor, MathType, MathPlayer and MathFlow ~ 

On Sun, Feb 24, 2008 at 2:23 PM, Richard Schwerdtfeger <schwer@us.ibm.com> 

Hi Neil, 

Thank you for the request. WAI-ARIA 1.0 is very close to last call and we 
are trying to lock it down. Also, one of our key implementations, Firefox 
3, is close to being locked down. So, we will discuss this but at this 
point it is very unlikely we can get this into 1.0. Also, if we were to 
add a new role there are a number of people waiting in the wings that 
would like to be considered and this would open up the flood gates. 

We have talked about the follow-on for 1.0 and we have been talking about 
extensibility for that. Is there a math handler that can do math 
interpretation from an image? 

I have to ask though:  Would it not be better to create a taxonomy for 
Math than simply calling something a Math object type? 


Rich Schwerdtfeger
Distinguished Engineer, SWG Accessibility Architect/Strategist
Chair, IBM Accessibility Architecture Review  Board
blog: http://www.ibm.com/developerworks/blogs/page/schwer 

"Neil Soiffer" <Neils@dessci.com> 
Sent by: w3c-wai-pf-request@w3.org 
02/21/2008 01:25 AM 

w3c-wai-pf@w3.org, public-pfwg-comments@w3.org 

Comment on WAI-ARIA Role

I work for Design Science, which is a W3C member.  I am a member of the 
MathML Working Group and have been since its inception.  I am also chair 
of the DAISY/Math Working Group.  That WG developed a spec that added 
MathML to DAISY; it was approved about a year ago.  Our company developed 
MathPlayer, the standard "plug-in" for IE to display MathML both visually 
and aurally.  MathPlayer works with JAWS, Window-Eyes, TextHELP, and other 
AT to make math encoded with MathML accessible.  It is capable of 
generating speech, synchronized highlighting, and braille from the MathML. 

I have a request for an additional "role" to further support math 
accessibility.  First, some context... 
Because of browser implementation issues, MathML adoption has not been as 
widespread as it should otherwise be.  A lot of people/sites, including 
wikipedia, still use images for math.  However, the images often include 
alt text or embedded comments that could be used to make the image 
accessible.  An idea that Design Science has been thinking about is 
writing JavaScript that grabs the alt text or embedded comments and 
provides some level of accessibility.  For example, if the image contains 
embedded MathML and MathPlayer was installed, then the javascript could 
rewrite the page as one containing MathML and MathPlayer could take over 
display and interaction with AT.  If MathPlayer wasn't installed (for 
example, someone is running Firefox on Linux), the JavaScript could still 
do the translation of MathML to speech and braille.  It wouldn't be able 
to magnify or sync highlight.  If TeX or some other known format was 
found, the JavaScript could translate it to MathML and then make it 

The same JavaScript accessibility issues arise with this idea as with 
other JavaScript -- AT doesn't know about the JavaScript and doesn't know 
that the image (or div/span for JSMath) is really math.  If it did know, 
then it could call on our interfaces (or eventually some standard expert 
handler interface) and get an appropriate string to speak or string to 
send to the braille display (math has it own braille codes and they are 
not identical to what is spoken). 
Now for the suggestion...  ARIA should have a way to say 'this element is 
actually math'.  My non-expert suggestion for how to do this would be to 
add a "math" role to the list of known roles.  'math' would be similar to 
'grid' in that it represents document structure and contains other 
elements.  Ideally, an extensible mechanism would be desirable, but math 
is an obvious case that maps onto an existing W3C standard and would 
enhance existing pages that don't use MathML.

As a real life example from Wikipedia [
http://en.wikipedia.org/wiki/Nth_root], the first image in the XHTML could 
potentially be made accessible if AT knew about it by adding a role 
attribute as 
<img class="tex" alt="\sqrt[n]{a}" role="math" src="
[Note:  the alt text is present on the Wikipedia page.  The TeX was used 
to generate the image in the first place] 
Adding role="math" says that the alt text or comments in the image 
contains information that can be used for accessibility purposes.  It 
alerts AT software so that they can call on some expert math handler to 
get information about how to handle the element. 
Neil Soiffer
Senior Scientist
Design Science, Inc.
~ Makers of Equation Editor, MathType, MathPlayer and MathFlow ~ 
Received on Monday, 25 February 2008 10:22:55 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:45:56 UTC