Need differentiator between "no alt text provided" and "no alt text necessary"

Please keep me in the CC replies b/c I'm not on public-html. This is  
the tail end of a thread from the PFWG.

Begin forwarded message:

> James Craig wrote:
>
>> Simon Pieters wrote:
>>
>>> So do you think HTML5 should go back to support the case where an  
>>> image needs a label (or textual equivalent) but lacks one?
>>
>> I think so, yes.

…snip…

>>> How do you envision ATs to differentiate [these cases]?
>>
>> The language is for any software (including search engines), not  
>> just UA/AT. If a search engine (or a future implementation of AT)  
>> has the capability, it may try to use optical character recognition  
>> on images that are determined to be meaningful but without  
>> alternative text.

Or to change the navigation mechanism based on that difference.

For example, at least one screen reader has a preference to allow you  
to navigate no images, all images, or only images with a description.  
An image with a role of presentation should not be treated as an image  
at all, so it should no be navigated to, even if the user setting is  
to navigate all images. However, if the user's preference is to  
navigate only to images with descriptions, then there would be no  
navigational difference between a presentational image and a  
meaningful image that just lacks appropriate alternative text.

>>> Why is <img alt="" noalt> better than just <img>?
>>
>> As long as there is a clear differentiation between these three  
>> states, I'm okay with whatever the markup looks like.
>>
>> 1. Presentational image (no alternative text necessary)
>> 2. Meaningful image (no alternative text provided)
>> 3. Meaningful image (alternative text provided)
>
> Now we're getting somewhere. :-) You should express the above points  
> on public-html or somewhere public that gets on the editors' radar.

So, to the public-html group:

The reason there needs to be a differentiator is because each host  
language (in this case, HTML 5) needs a way to determine when an image  
should use the "presentation" role [1] and when it should use the  
default "img" role so the user agent can convey the role to an  
assistive technology API. An author or an authoring tool may have that  
information, but there is currently no way to convey it in the language.

1. http://www.w3.org/WAI/PF/aria/#presentation

Received on Monday, 2 February 2009 22:57:47 UTC