WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Best practices for identifying forms and their fields

for

From: Denis Boudreau
Date: May 6, 2013 1:48PM

Hi Jared,


On 2013-05-06, at 2:26 PM, Jared Smith < <EMAIL REMOVED> > wrote:

> On Mon, May 6, 2013 at 12:07 PM, Denis Boudreau wrote:
>
>> I think it would be advisable to point out that for improved consistency and support with assistive technologies and user agents, relying on the aria-label attribute rather than the title attribute would be a good idea.
>
> Are there modern user agents that do not read the title attribute if a label is missing? I'm not aware of any.


Maybe my information is a bit outdated, but last time I checked (has it been two years already?!), the title attribute was only partially supported, or at the very least, not always read by default by screen readers, I just looked it up and found this post by Terrill Thompson that shows title support was still not reliable back when Jaws was on v12. Has much changed since then? http://terrillthompson.com/tests/forms_title_not_label.html


> But I can think of lots of user agents that don't support aria-label, including almost everything over a year or so old.

I would have said two years, but I agree with you. I do see much better chances at interest in improving support for aria and html5 goodies than for old HTML attributes that are way over 10 years of age. Unless it gets totally refactored in the course of the HTML5 overhauls.


> I agree with you in principle - I've never been entirely comfortable
> with using title as a replacement for a label. At a minimum it's an
> abuse of the spirit of HTML's definition of title as being for
> "advisory information". If it's inaccessible without the title, it
> sounds like more than advisory to me. I do note, however, that title
> has been redefined in HTML5 to include labels for interactive
> controls.

Right, but if we're worried about support for aria on technologies that are a few years old, then I'd be very reluctant to rely on this for now either.


> If one can get the same behavior and support from a native HTML
> attribute, I'd think it should be used over an ARIA attribute.

I do agree with you that we should always prioritize native HTML over aria, and that's certainly true in cases where native html is reliably supported. Which explains why I was advocating for the use of aria-label in this case - I don't think it is reliably supported, but I'd be happy to be shown otherwise. Maybe I need to check that out again. Things move pretty fast and it's easy to miss some of the moving parts.

Given proper support (if it's indeed the case), then I too would advise to use the title attribute first and pach with aria when all else fails. That makes sense. The reason why I prefer aria-label right now is I was confident that support was more consistent. But now that belief has been shaken. <smile>

/Denis