Re: UNS: RE: UNS: WCAG considering amending F65 to NOT fail missing ALT text if title or aria-label is present

Hi all,

to clarify how screen readers use aria-describedby, or more in general, the thing called AccDescription as opposed to AccName:
When you browse a page in JAWS, NVDA, or any other screen reader with a virtual buffer concept, the accessible name of an image will be picked up, put in the buffer, along with the semantic information that this is a graphic, or image.
What will *not* happen, is that accDescription will be picked up. AccDescription will *only* be picked up on a *focused* element when tabbing to it and speaking the focused item. So a link that has both text and title, where the title becomes the accDescription, will be spoken as the text plus the indicator for a link, followed by the bit provided by AccDescription.
If you provide a description on an image via aria-describedby, users will never be notified of it because images, unless they are part of a link construct, will never receive focus by default. What's in an AccDescription for an image will simply be ignored inside the virtual buffer.

Other screen readers such as VoiceOver on Mac and iOS may pick this up and report it, just because VoiceOver's concept is to put its own version of the focus on anything. Firefox for Android and Firefox OS will probably do the same for images, since on touch screen devices, the lines between what is focused/focusable or not are very weak compared to the desktop.

Again: If aria-describedby is being used on an image, chances are very high that screen readers on Windows, while the user browses a page, will simply ignore whatever is provided though that mechanism.

Marco

On Nov 26, 2013, at 3:35 PM, Sailesh Panchang <spanchang02@yahoo.com> wrote:

> David / Joshue,
> This solution about aria-describedby  has been in the survey comments for two weeks now and I had drawn attention to it during the call too: 
> "iii. On  reflection, I think it is always better to use aria-describedby on the alt (instead of aria-labelledby). The intent is to have available text on 
> the page announced without need for duplicating and also to avoid pitfalls of not assigning an alt or using alt="". So the alt can be say, a single word and the more descriptive text elsewhere on the page. Then both will be read. For instance the alt can be "Logo:" and the aria-describedby can be the org's name say at the top of the page. In the star rating example (on aria-labelledby example for image), the alt can be "Rating:" and the other text can be "3 /5" etc. So if we adopt this approach, we should not advocate use of aria-labelledby for images." 
> 
> Using aria-describedby in this way helps to enhance accessibility / user experience. It does not distort  the way different user groups identify the image if aria-labelledby were used .... a point I made earlier too and is illustrated in David's code sample:
> <img scr="eiffelTower.jpg" alt="Eiffel tower" labelledby="123">
> <p id="123"> The tower stands 324 metres (1,063 ft.) tall</p>
> 
> Is this not a concern? Or is the focus only on ARIA supported AT users?
> Again the focus of ARIA is on rich Internet apps / content. The algorithm's focus is clearly on  interactive elements as per the examples.
> Thanks and regards,
> Sailesh
> 
> 
> 
> 
> --------------------------------------------
> On Tue, 11/26/13, Joshue O Connor <joshue.oconnor@cfit.ie> wrote:
> 
> Subject: Re: UNS: RE: UNS: WCAG considering amending F65 to NOT fail missing   ALT text if title or aria-label is present
> To: "David MacDonald" <david100@sympatico.ca>
> Cc: "'HTML Accessibility Task Force'" <public-html-a11y@w3.org>, "'WCAG WG'" <w3c-wai-gl@w3.org>
> Date: Tuesday, November 26, 2013, 6:30 AM
> 
> David MacDonald wrote:
>> There are several fixes, here are two:
>> 1) use aria-describedby insread of aria-labelledby
> because it maps to the
>> accDescription, not the accName, so the alt gets the
> accName and the
>> describedby gets accDescription
>> 2) leave off the aria-labelledby, so the user will hear
> the ALT text, and
>> encounter the supplementary text in the paragraph after
> the image.
> 
> Thanks, and if you find a suitable spot in one of our wikis
> that'd be great (if not already there somewhere). Also maybe
> also overviews of what accName/accDescription are - most
> designers don't think on this level at all.
> 
>> If the algorithm does not shift from OR to AND, then we
> will need to
>> introduce WCAG failures for the presence of both
> ALT+aria-label, and
>> ALT+aria-labelledby in an image.
> 
> <chair hat off> Maybe, maybe not. We really have to
> discuss these things on a case by case basis. Accessible
> development techniques and UA support are always changing,
> so some caution is needed - (this is part of the reason we
> are having this unforeseen @alt debate, no?). Ideally, the
> standard should define the behaviour but as we know in
> practice that's just not always the case.
> 
> Thanks
> 
> Josh
> 
> 
> 

Received on Tuesday, 26 November 2013 17:53:11 UTC