WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Alt text (was VIKI - text transcodeing)

for

From: John Foliot - Stanford Online Accessibility Program
Date: Jan 19, 2007 5:40PM

Keith Parks wrote:
> John,
>
> One thing I forgot to ask, going back to your original reply, was
> about the [square brackets] around ALT text. You said you presumed
> that the text transcoder added the brackets for the same reason as
> you do. What's that reason?

One of the simple test I use to illustrate to people about "screen readers"
is to visit a site (their site, my site?) using Lynx. The web is such a
powerful visual medium for so many people that "seeing" what visually
impaired people get (raw, hopefully semantically formatted text) to me is
almost more useful than me trying to demo a screen reader - for one, I am a
punter in that area (as any experienced screen reader user will attest,
learning the software is a huge learning curve). Seeing text only, without
the benefit of style or images kinda brings it home, graphically. Weird
perhaps, but it seems to work.

A common page, and one that I demo'd often was the Bio, or Profile page.
These semi-standard pages often look like this:

<h1>John Foliot</h1>

<p><img src="path to file" alt="John Foliot"> John has been at Stanford
since... </p>


Now, viewing this in Lynx, you get:

John Foliot
John Foliot John has been at Stanford since...

...which "reads" funny. Contrast that with my modified alt text 'norm':

John Foliot
[Photo - John Foliot] John has been at Stanford since...

Now, this is still pretty visual, as depending on the verbosity settings,
often the square brackets are ignored by the screen reader, but since
Universal Accessibility is more than just web pages for blind people
<grin>... And by adding the bit of info that it is a Photo, most screen
readers will now voice "Image: Photo <slight pause> John Foliot <pause> John
has been..."

I tried it out on a few daily AT users, and they all were "cool" with it...
Not a roaring hurrah, but subtle and useful was the general feedback I got
(granted, it was a small pool of users)

Getting back to the initial question, I thus presume that Phil's readable
interface (which includes square brackets around the alt text) is for
essentially the same reason - the visual display - it "breaks out" the alt
text from other onscreen text.

>
> One thing I've struggled with (well, at least had second thought
> about) is how far to "editorialize" in ALT text, when I started
> trying to have the ALT text describe the *purpose* as well as the
> appearance of an image.
>
> For instance, since we're both in the edu realm, you've probably
> dealt with many of the "faculty together with a student" photos. A
> descriptive ALT text could be "Photo: a student talking with a
> faculty member." But thinking about *why* the photo is there, what it
> is really supposed to do is say "Students enjoy personal interaction
> with faculty members." That's the story that the picture is supposed
> to tell. Without having to actually *say* it in the copy.

Well, here, you and I agree, although others may not always. But I, like
you, have always asked "what is the *reason* for the photo, what is the
message it is intending to send", and then try and incorporate it into the
alt text. It's not always easy to do so succinctly.

>
> So is it valid to say that in the ALT text? Aside from the question
> of precisely how to word it, does that go beyond the "text
> equivalent" function?

Well, I would argue that in many ways, by providing the editorial "insight"
you *are* providing the "equivalent". Telling the user that it is a photo
of a building really has no practical use does it? Saying that it is the
Chancellor's Office, and that it is an example of the California Mission
style *is* of use, as it then "paints the picture" for the user.

>
> Same for simple campus architectural shots. Rather than "photo: the
> Library building", what I *hope* the photo is saying is more like
> "It's another beautiful day in San Diego... the sky is blue, the palm
> trees are swaying, and students are enjoying our nice climate." etc.
> (pictures being worth 1000 words, and all.)

Right, but you also need to be careful to not add *too much* info in your
alternative text. If the idea of blue sky and swaying palm trees is
integral to the message, it should be in the actual text - this is where
creative writing comes into play. For above, an alt text of "[Photo -
Students lounging outside the library building, enjoying our typical
temperate climate]" gives a sense of purpose to the photo, without it being
those 1,000 words.

>
> And how about things like ethnicity? We try hard to have our
> publications reflect the diversity of our campus (how's *that* for a
> canned phrase? ;^) Yet it would seem awfully weird to have and ALT
> tag of "A caucasian female student and a hispanic male student talk
> with an asian female instructor." If that's important "content" of
> the image, shouldn't it be part of the description?

Well... This is a loaded question, as it also calls into play political
correctness - a sensitive issue. My thoughts here are that unless the
container page/site is specifically about racial diversity or similar, I
personally would probably side-step the issue. (Do you also feature photos
of students/faculty in wheel chairs, with assistive dogs, etc.?) If it is
germane to identify a person by an attribute (ethnicity, disability, etc.),
then it needs to be handled with sensitivity, and I would consult heavily on
the appropriateness of any text, but that's just conservative ol' me
talking... I suppose if it were necessary to "identify", I would also seek
to personalize the characters, and include their "difference" into the text
that way: [Photo - Wheelchair bound Keith successfully manages to navigate
the barrier free campus], [Photo - Claudius Smith, an African-American, and
Cindy Lopez, a Mexican-American discuss racial diversity issues on campus].
In other words, identify their "difference" only if it completes the story.

All just opinion of course...

JF