Re: HTML Action Item 54 - ...draft text for HTML 5 spec to require producers/authors to include @alt on img elements.

On May 12, 2008, at 5:57 AM, Steven Faulkner wrote:

> Hi Maciej,
>
>> This proposal does not cover the use case
>> where HTML generated by a tool does not have a textual alternative
>> available.
>
> The current draft of item54 does not provide explicit coverage, as  
> stated:
>
> 'Advice has been sought, is needed, and is pending from PFWG regarding
> the separate issue of what an authoring or publishing tool should
> insert, in a case where no alt has been provided by the author, but
> the image is known to be "critical content".'
>
> source: http://www.paciellogroup.com/blog/misc/uc/
>
> But, the proposal does implicitly cover the use case (the content  
> will be
> invalid, because the author is either using a broken authoring tool,
> or doesn't care about the structure of the data).

Defining any content produced by the use case to be non-conforming is  
not handling the use case, at least for purposes of document  
conformance.

>
>
>
>
>> It also requires redundant text in many cases where the current  
>> spec would
>> call for empty alt.
>
> The current spec conflicts with WCGA 2.0 (candidate recommendation)
> advice on such images.
>
> As stated in the proposal;
> "The aim of this draft is therefore to comply with WCAG 2.0, Guideline
> 1.1. Text Alternatives:"
>
> source: http://www.paciellogroup.com/blog/misc/uc/
>
>> Is there any reason to believe that redundant text description of  
>> an image
>> that recapitulates the text is helpful, rather than harmful, to  
>> users who
>> use textual alternatives?
>
>
> potential benficiaries of this information include, blind users,
> visually impaired users, cognitively impaired users and user with
> intellectual disabilities. each of the user groups have different
> needs and AT that they use will consume and present the information in
> a variety of ways, users are also provided with a wide variety of
> options on how to navigate and access content.

I believe my question is valid regardless of the target audiences and  
their many options.

> So your simplistic notion of redundancy does not hold up.
>
> examples:
> A screen reader user may initially hear the short summary of the image
> content, then access then access the longer description via a key
> stroke.
>
> The screen magnifier/reader user who can see the image, but not
> understand its contents due to it being to blurred, will have the alt
> text announced, to give them some idea of the image content.
>
> The cognitively impaired user accessing the page using a talking
> browser, likewise will have the benefit of a short image label,
> although they may not be able to understand the content of the image
> from its visual rendering.
>
> for all of these users the presence of the short alt text provides an
> alternative labelling mechanism that can be presented to the user via
> their AT.
>
> For the magnifier user, the short text alternative may or may not be
> provided visually. The association between the aria-describedby text
> and image could be rendered visually in a number of of ways.
>
> Different AT's currently provide different mechanisms to navigate and
> access content. For example screen readers such as JAWS provides
> methods to navigate a page via graphics (using the G key), in this
> case, not providing the alt text means that the image would be
> skipped.

Thank you for the explanation of how different users may interact with  
assistive technologies. However, that does not answer my question. In  
the part of the proposal I cited, the suggested markup structure was  
more or less like this:

<p id="info">Complete explanation of a chart with full details.</p>
<img src="chart.png" alt="Explanation of chart." aria- 
describedby="info">

A user reading this document with a screen reader would hear something  
like this, as they navigate:

"Complete explanation of a chart with full details."
"Image. Explanation of chart. Long description available"

Then, if the user chooses to hear the long description:

"Complete explanation of a chart with full details."

Can you explain why this is better than hearing "Complete explanation  
of a chart with full details" once, and then skipping the image (which  
is what would happen if the image had alt="" and no aria-describedby  
attribute)? I tried it with VoiceOver on Mac OS X, and the repetition  
does not seem helpful.

>> Is there reason to believe that screen reader users like to hear  
>> things two
>> or three times? I have not done any studies but this is surprising  
>> to my
>> intuition. I would have concluded that using alt="" to present the  
>> screen
>> reader user (or other users of aural or text-only media) with the
>> information only once is best. It may be that this surprising  
>> conclusion is
>> correct but I would like to hear some justification.
>
>
> As stated previously
>
> "The aim of this draft is therefore to comply with WCAG 2.0, Guideline
> 1.1. Text Alternatives:"
>
> The advice in WCAG 2.0 has been developed over a long period of public
> consultation. If you feel that your view of what constitutes best
> advice on how to provide appropriate text alternatives is superior to
> WCAG,  I urge you to correspond with the WAI WCAG working group. I am
> sure that they would be more than happy to explain how they arrived at
> the advice provided in WCAG 2.0.


That sounds like an appeal to authority, not a justification. Even the  
best exerts make mistakes, or fail to think through certain edge  
cases. Or sometimes, they can reach correct conclusions that non- 
experts find surprising, in which case an explanation can be of great  
benefit.

Regards,
Maciej

Received on Monday, 12 May 2008 15:17:57 UTC