Re: [html4all] some reflections on @alt usage (and even sometimes @aria-labelledby...)

** summary:

Don't dismiss @aria-labelledby from consideration as an alternative
to @alt in the "where to find the text alternate" algorithm.

User text should be Web text, not attribute-value-text.

Details below.

On 28 Apr 2008, at 6:15 AM, Gez Lemon wrote:

> 2008/4/28 Steven Faulkner <>:
>>  It should be noted that some of the suggestions outlined by John and
>>  myself were previously raised by Al and discussed on a thread  
>> back in
>>  February [1]:
>>  [1] what's machinable [was: Re: ALT issue redux]
> Just to add to my position on using aria-describedby and
> aria-labelledby in response to Al's original message. I think
> aria-describedby could be considered an appropriate alternative to
> longdesc when the description exists in the same page, but obviously
> not when the description is in another document. I can also see there
> may be some edge-cases when aria-labelledby may be considered a
> suitable replacement for alt. These are edge-cases, as ARIA was
> designed to bridge the accessibility gaps with the current structure
> we have. Providing alt text for images wasn't one of those gaps, as
> there is already a suitable attribute.

The attribute is not always suitable, because of the limitations
that attribute values put on the text.

If the short text alternate contains the name of a Japanese person
which should properly be marked up with Ruby, or an acronym for
an obscure concept, etc. the use of @labelledby with reference to
an element containing hypertext would be materially better than
using @alt.

This is definitely a minority case.

But I have to mention it because staying hung up on what is simple
to teach, taking a very "lowest level language" definition of text
for the text alternate rather than the Web standard of hypertext,
*is injurious* to access by people with cognitive disabilities in
its overall effect.

On the flip side I hear from our script authors that coining a
new element with a non-conflicting ID is a lot of work.  So there
are real costs to outsourcing the text alternate to an element,
if the text string isn't already there in the accompanying text

Some of us have come the conclusion that in general, content that
is to be presented to the user should be in element content and
not attribute values.  I don't know how widespread this belief is,
but the above examples give some of the motivation for suggesting
this architectural pattern.


This is where I may lose some of my audience...

Labeling is a more generic design pattern than alternatives to
sensory-dependent content.  Universal design would suggest that
we use the generic design pattern where it suffices.

Using label markup is arguably the most mnemonic for authors
in the case of images that iconically represent user action


Controls, Input: If it is a control or accepts user input, then it  
has a name that describes its purpose. (See also Guideline 4.1.)




> Cheers,
> Gez
> -- 
> _____________________________
> Supplement your vitamins

Received on Monday, 28 April 2008 14:47:18 UTC