RE: Clarification of WCAG intent and meaning of techniques [Re: WCAG considering amending F65 to NOT fail missing ALT text if title or aria-label is present]

Hi Steve,

thanks for pointing this out. I'd like to comment and be more precise on your statement

>> For the case of alt it has not been agreed that this is the case.
Using MS Inspect32 or Paciello Group Aviewer (, on the image example from in Firefox 22 and IE 11, the MSAA accName  and UIAutomation CurrentName is properly mapped with the content of the alt attribute. So, by principle, nothing hinders AT to use this information. And I think screen readers do that already, so we're fine here. I will not discuss here if aria-label is or is not mapped with a concurrent alt definition in place for images.

The mapping of alt with <a> elements is another story (like described in and I think aria-label could be a valuable alternative here (like described in

Also, you classify browsers also as "AT" flavors whereas I favor the term "User Agent" here.  I think typical AT is still more focused on particular needs for accessibility then browsers are. However, you are right that this "display-in-place" feature is yet missing for aria-label in browsers, so as of today "alt" will cover a broader scope.

-          Stefan

From: Steve Faulkner []
Sent: Donnerstag, 28. November 2013 11:24
To: Schnabel, Stefan
Cc: Michael Cooper; David MacDonald; Janina Sajka; HTML Accessibility Task Force; WCAG WG;; Gregg Vanderheiden;
Subject: Re: Clarification of WCAG intent and meaning of techniques [Re: WCAG considering amending F65 to NOT fail missing ALT text if title or aria-label is present]

Hi Stefan, this only works for criteria that are solely contingent upon accessibility APIs exposing information to AT.
For the case of alt it has not been agreed that this is the case.
Ramon, for example brought up the case of a low vision user who turns off images in the browser because the information in the images is difficult to perceive, but the alt text exposed as text by the browser is not. This involves no AT.
In this case what is required for all of the suggested alternatives aria-label etc is that they are displayed in place of an image when an image is not displayed. This is currently not the case. If we can interest browser implementers exposing aria-label as text in this case then we have a practical alternative to alt.



HTML 5.1<>

On 28 November 2013 09:58, Schnabel, Stefan <<>> wrote:
In my opinion, (to say it again) successful and comprehensive mapping to a given platform accessibility API based on clear and transparent rules is the key, no matter what Assistive Technology does with it afterwards. Given that, (platform-dependent) automatic testing of APIs (not only markup) can exclude the effect of current AT implementations (as "prove of concept" that support works) and serve as a solid base.

"Successful" means, for instance,  that e.g. in the windows MSAA world both accessible Name and accessible Description for images can be obtained by using the appropriate host language AND ARIA metadata attributes. Yes there are possible contradictions, clashes etc. and therefore necessary precedence rules to be taken into account to yield a successful and reasonable mapping, but it is not a "black or white" discussion.

In this sense,

<img src="../images/giraffe.jpg" aria-labelledby="123"/>
<p id="123"> Giraffe grazing on tree branches</p>

serves as a perfect replacement for *alt* since both will be mapped to the same accessible Name property by the user agent. In addition, visible text is associated in a meaningful and reasonable way with the image.

Furthermore, it needs to be decided FINALLY on an API level what has to have precedence: accessible Name or accessible Description. If the former, then

<img src="../images/giraffe.jpg" title="Giraffe grazing on tree branches"/>

would violate this (no *alt* and therefore no accessible Name mapping). If there is no precedence, then example is fine. It can be that such decisions have been already made. If so, they have not hit the developers mind since I often hear from "rebel-minded" developers that giving a title is perfect enough. I know that this is wrong regarding the neglected status of the *alt* attribute but I cannot say that they are completely wrong since they DID something for accessibility. So following the idea of James Craig, yes, a markup rule validation but not an accessibility validation. Same holds true for the aria-label usage in these cases.

The most important task is  to extract the use cases where both host language AND ARIA metadata attributes work together in a supplementary standard-conformant and meaningful way and recommend them. This includes also usage examples of e.g. *aria-label* in all cases where no *alt* attribute is defined for the element. Of course this task should include bad usage examples, too. I see efforts addressing exactly this already underway (e.g. for labelling of landmarks, which is a good example).

Finally, I don't understand why screen readers still do not make full usage of ARIA attributes in all their modes (Virtual or not). I also cannot comprehend that these tools do not follow a minimum common rule set how to pick up an concatenate the API and DOM info on a given platform.

In my opinion, they should serve people's needs by extracting (as configurable default) as much info as possible from a given context.

Example discussions for contradicting content like for

<img src="../images/giraffe.jpg" alt="Giraffe grazing on tree branches" aria-label="A nice giraffe"/>

would be obsolete if there would be a requirement in a W3C spec for screen readers to gather ALL info no matter which mode they are and not rely on API mapping alone (which would be "Giraffe grazing on tree branches" according to alt precedence mapping rules). This is no big issue for them since they already always query the DOM nodes in parallel. Personalization of speech output must refine the behavior here, but the default should be really "include all info provided for this particular context".

-          Stefan

From: Michael Cooper [<>]
Sent: Mittwoch, 27. November 2013 21:03
To: David MacDonald; 'Janina Sajka'; 'HTML Accessibility Task Force'; WCAG WG;<>
Cc: 'Gregg Vanderheiden';<>
Subject: Clarification of WCAG intent and meaning of techniques [Re: WCAG considering amending F65 to NOT fail missing ALT text if title or aria-label is present]

I thought it would be helpful to clarify what the WCAG Working Group is considering and its reasons for asking the question that started this thread.

To recap background on the structure of WCAG 2.0 and its supporting materials: The Web Content Accessibility Guidelines 2.0 defines normative requirements for accessible content. This means that an author wishing to claim conformance to WCAG 2.0 must ensure their content meets the Success Criteria. The WCAG 2.0 Success Criteria are abstract, though, so they are supported by informative resources including the WCAG 2.0 Techniques. Informative resources provide interpretive guidance from the WCAG Working Group, and Techniques suggest how the Success Criteria might be met in particular technologies. But these are only informative suggestions, and it is possible to meet or fail to meet the Success Criteria without matching the patterns suggested in the techniques.

The same is true of Failure Techniques. These describe authoring practices that might cause a page to fail the Success Criteria. But these, too, are informative, and it is possible to fail the Success Criteria in other ways, or to use patterns that match a failure technique but do not in fact cause the page to fail because another technique was used that caused the page to pass. In fact, as time passes, the Working Group has become more and more concerned that some of the failure techniques no longer describe situations that always fail, due to changes in technology. The group is now trying to identify and update some of these techniques.

This brings me to the technique at question in this thread, "F65: Failure of Success Criterion 1.1.1 due to omitting the alt attribute...". This technique says it is a failure of WCAG 2.0 if the alt attribute is omitted on certain elements, because text alternatives would not be available. Since the time that technique was written, new approaches to associating text alternatives have become available, for instance aria-label and aria-labelledby among others. While there are concerns about how widely implemented these are and if they fully fulfill the role of alt, it seemed to the Working Group that it would be worth updating the failure to reflect new technology, so it would not indicate that content fails when it may not.

There are many considerations to making a change like this. Although failure techniques are only informative, the goal is to make them as accurate as possible to the technologies of the time. But as we all know, there is not consensus in the accessibility community about whether newer technologies are an acceptable replacement for alt, for both technical and strategic reasons. The WCAG WG did not want to make a change to the technique that is not in line with the consensus of the HTML Accessibility Task Force which has worked most closely within W3C on this issue. This is the reason that the group brought the question to this forum.

The purpose of all this background is to clarify the intentions and plans of the WCAG Working Group. The goal is to coordinate and make sure the group takes an action that is accepted in the accessibility community, rather than make one decision within the WCAG Working Group that it knows might not be accepted by many other stakeholders. The group wants to understand the variety of perspectives and does not intend to make a hasty or one-sided change. It seems, though that some update to the failure technique is needed, although it could turn out after this exploration that the consensus is not to change the technique. Even after the group makes a proposed change, if any, to the technique, there will be further rounds of review - in context of a draft of the technique at that point. And in the end, the technique is still non-normative - other organizations are free to disagree with it and follow other approaches to meet WCAG 2.0.

I hope this helps people understand the intent of the message that started this thread and what the WCAG Working Group does and does not plan to do.

On 22/11/2013 6:27 PM, David MacDonald wrote:

On behalf of the WCAG working group, I have an action item to solicit

responses from the wider community regarding a proposed amendment to WCAG

failure technique F65 regarding missing ALT. Currently; if an <img> element

is missing from an ALT attribute the page fails WCAG SC 1.1.1 Level A. Some

are proposing that we allow authors to use the aria-label, aria-labelledby,

and title attributes INSTEAD of ALT.

So under the amended failure technique NONE of the following would fail


<img src="../images/giraffe.jpg" title="Giraffe grazing on tree branches"/>

<img src="../images/giraffe.jpg" aria-label="Giraffe grazing on tree


<img src="../images/giraffe.jpg" aria-labelledby="123"/>

<p id="123"> Giraffe grazing on tree branches</p>

As you can imagine there are strong opinions all around on this so I

suggested we get a sense of what other groups such as the HTML5 A11y TF and

PF think.

Those in favour of the change provide the following rational:

--These alternatives on the img element work in assistive technology

--The aria spec says these attributes should get an accessible NAME in the


--They say it's easy to teach beginner programmers to just always use an

aria label on everything, rather than requiring a label on form fields and

alt on images

--They feel as a failure F65 is very strong if fails a page for missing ALT,

especially if other things work, and they would like to soften it to allow

other things that work.

--html 5 allows a <figure><legend> combination instead of alt, so they feel

WCAG will have to change F65 anyway to allow a figure with a legend, and

that helps open the door to this discussion

Those in favour of the status quo (which fails missing alt text) provide the

following rational:

--aria-label, labelledby and title, are not really suitable attributes for

img alternative text because they implies a label or title, rather than an

alternate text, so it is not a semantic equivalent

--title is not well supported

--some feel that the aria spec is not in any way suggesting these as

replacements to ALT.

--aria instructs authors to use native html where possible, and they could

not come up with viable use cases of omitting alt text

--there are hundreds of millions of dollars invested in current evaluation

tools, and methodologies, and this would represent a major departure from

one of the most basic accessibility convention, that is almost as old as the

web and is the "rock star" of accessibility

--it could cost a lot of money to change guidance to developers etc..., and

muddy the waters on a very efficient current evaluation mechanism

--when the figure/legend is supported by AT we can amend F65 but that is a

different issue and the semantics of this construct are OK for text

alternatives, rather than the label/labelledby/title options

--it may cause some confidence problems to WCAG legislation, because it

represents a strong loosening to a fundamental Success Criteria, an

unnecessary change that doesn't help the cause of accessibility, but just

complicates things

--ALT is better supported and the text appears when images are turned off.

--initial twitter feedback from the community is strongly against changing

this failure

There are probably other reasons on both sides which we hope to hear ... but

these should start it off. Please give your opinions and reasons.

Current technique here:

Proposed failure here (see test procedure)


David MacDonald

CanAdapt Solutions Inc.

Tel:  613.235.4902<tel:613.235.4902><>

  Adapting the web to all users

            Including those with disabilities


Michael Cooper
Web Accessibility Specialist
World Wide Web Consortium, Web Accessibility Initiative
Information Page<>

Received on Thursday, 28 November 2013 11:37:50 UTC