E-mail List Archives

Re: Alternate text for images having caption adjacent


From: Jared Smith
Date: May 30, 2014 3:55PM

On Fri, May 30, 2014 at 3:17 PM, Denis Boudreau wrote:

> Luckily, WCAG 2.0 provides a definition of "pure decoration”

There's a conversation going on right now in the working group
(http://www.w3.org/WAI/GL/track/issues/21) about the confusing nature
of this definition and that it doesn't match reality or Steve's HTML5
definitions and examples at

> we can’t exactly use the alt text value of an image to describe
> how the person physically looks. That would be awkward!

Not only awkward, but it would be a misuse of the alt attribute if the
description of what they look like is not an alternative to the
content of the image.

> Therefore, the alt text ends up repeating the person’s name and
> consequently, provides no real additional information that the
> adjacent text doesn’t already provide. Which, I assume, is why
> Jared was saying it’s not really important.

The information certainly is important. But it shouldn't be conveyed
twice. Conveying it once in the caption is sufficient.

> If a screen reader user has been told there’s a picture of the
> person on the page and they don’t know that person's name,
> they might want to search the page through images to find it.

If a screen reader user can't search through the images on a page,
this is a screen reader deficiency. The burden should not be placed on
authors to say, "Hey screen reader, here's an image". We certainly
have no requirement to identify forms or tables, for example, with
text that indicates they are present. Screen readers do this based on
the semantics of the page.

If an image with alt="" has a programmatic association to content
(adjacent text, aria-label or description, <figcaption>, etc.), then
assistive technologies should utilize this when presenting an image
list or while navigating images.

> Relying on null alt text takes that functionality or ability away.

No, screen reader deficiencies don't make that functionality
available. I think it's a slippery slope if we start creatively
interpreting accessibility guidelines and HTML specifications to fill
gaps in screen reader behavior.

> 2. The ability to actually locate the picture, and download it (for whatever purpose) is another thing we take away from these users when we choose to go with null alt text.

As above, this is a user agent issue, not an authoring issue.
Certainly the alt attribute is NOT defined as "an indicator of the
presence of an image" - that's what <img> is for.

I would, however, argue that if an image is of such importance that
the user would likely have a need to download it, then it probably
necessitates at least a brief alternative as Whitney and I suggested

It continues to fascinate me how complex and subject to interpretation
alternative text is despite being the foundational technique of our
entire field. Thanks for engaging in the dialogue!