RE: The alt="" attribute

Ian Hickson wrote:

> So. We need another option.
> Are there cases where the image is lacking good alt text that
> wouldn't be covered by one of the following?:
>  - title="" attribute on the <img> itself
>  - <legend> of the <figure> that contains the <img>
>  - heading of the section that contains the <img>
> F. We could say that for these "key content without alt text" cases, we 
> have the alt="" attribute omitted, but there must be at least one of the 
> above, and the first of the above that is present must include sufficient 
> information to orient the user.
> I've put proposal F into the spec.
> I've based this on as much objective data as possible, as described above.

> If people disagree with this, I would like to encourage them to please
> provide actual data to back up their opinion. 

<olive branch>

Thank you for the detailed note regarding the @alt situation.  It appears
that you have succinctly summarized the issues in most of it's facets;
certainly enough to be able to condense discussion.  I would like to offer a
few observations and comments however, in the spirit of trying to move
forward with what appears to be a workable proposal.

</olive branch>

The first comment concerns "MUST" in the RFC2119 sense, as identified in the
HTML 5 draft (2.2 Conformance requirements).  Since <img> contains @alt
under most scenarios foreseen, allowing it to be omitted "sometimes" is to
my mind problematic.  As well, WCAG 1 experience has taught us that
conditional (interpretive) statements in Recommendations and Guidelines are
difficult and open to abuse ("Until user-agents...") and so they should be
discouraged at all times. For these reasons I believe that insisting that
@alt always be present ("MUST") should be part of the specification.

You could insist alt="" remain (satisfying "MUST"), knowing that the
alternative method(s) proposed will over-ride the empty value of alt="";
however keeping the @alt attribute as a mandatory attribute for <img>
maintains a certain "purity" of code conformance. Granted, the empty @alt
does nothing for accessibility directly, but it removes the ability to be
subjective here and from a "political" perspective does not back-slide on
HTML 4.01 and any ground fought for and won with regards to accessibility
and mandatory @alt: it is consistent with WAI direction and guidance.  

While I have no hard data to back this up, reading RFC2119
[] is pretty straightforward and it is
clear on it's definition of "MUST"; and with close to a decade of outreach
and education in the web accessibility field, the "always" statement
regarding @alt "feels right" and is consistent with the message that we (I)
have been delivering all these years.


The second comment concerns the alternative delivery methods proposed: you
cite @title, <legend> or heading as being the carriers that provide the
contextual information.  I would suggest that @longdesc also be included in
that list (the usefulness of @longdesc not yet being disproven ) as well as
@caption, and I am curious why other methods (such as perhaps @role, or the
ARIA labeledby/describedby) could not also be added to the list - providing
as many options as possible to content authors allows for a more granular
solution and makes it "easier" for content authors to be conformant -
clearly a stated goal for HTML5.

At this time the data surrounding @longdesc is not sufficient to refute it's
usefulness[], only to illustrate that to date it
has been under-used or mis-used: with the release of HTML5 and this
"improved" way of thinking about images and their need for being properly
described, it could/would be an opportune time for the web accessibility
community to focus on education surrounding the proper implementation of
@longdesc (A List Apart, Smashing Magazine, numerous blogs, etc.). Let's not
throw the baby out with the bathwater. 

Current support for @role and labeledby/describedby is not mainstream enough
to "prove" the viability of their proposed inclusion, however if the Working
Group was interested in pursuing this line of option I am sure that test
cases could be developed to provide some measurable data.  Given that HTML5
is already adding a number of new elements and attributes, this line of
pursuit is not without precedence.



Received on Tuesday, 26 August 2008 20:25:38 UTC