E-mail List Archives

RE: Length of Alt Tags

for

From: Jukka Korpela
Date: Dec 11, 2002 2:01AM


Jim Thatcher wrote:

> The issue can be phrased as what would you like to listen to
> ... when you must listen to everything?

I agree with the idea that alt texts should be written
primarily with speech (or Braille) presentation in mind.
The reason is that in such use, the alt texts are indispensable.
A person who surfs around with image loading switched off
just for efficiency can switch them on, when needed. Someone
who uses Lynx on a Unix box can probably use a graphic browser
as well. And so on.

So the criterion is good. But on a sufficiently advanced
browser, you don't need to listen to everything. You can
skip to next paragraph if the current one sounds irrelevant.
This, by the way, is a good reason to put a content-rich
image inside <p>...</p> markup: we can expect (or hope)
that browsers that are able to operate paragraph-wise will
then treat the alt text as paragraph content.

If the image is so essential that it tells part of the
very story of the page, like a piechart that shows the
results of a study that is the starting point of the
discussion on the page, then I think the alt text should
contain the equivalent information, if possible at all.
People who do not see the image should get the corresponding
text immediately, not via a link. But of course there are
situations where the information content of an image cannot
be feasibly presented in plain text of reasonable length;
then a link is needed.

> I think both 64 and 50 characters are on the long
> end especially when the alt text replaces a few characters of
> text in the image.

Certainly. If the image is just text in image format, the
alt text should contain either that text or a modified
version thereof (e.g., with abbreviations expanded, so that
speech synthesis has a better starting point).

My recommendation of 50 character maximum is partly based
on technical considerations, and it's conditional ("if feasible"),
mainly relating to situations where you could formulate the
message in different ways, shorter and longer.

The key principle is: as short as possible, but not shorter.
Various rules of thumb are just meant to help in achieving that.
Most alt texts can be written in a straightforward manner,
as soon as the author has understood the _idea_ (which has
many essential sides). But sometimes one needs to think
about it, to select the best alt text for a specific case;
and the rules of thumb say that you should think twice before
exceeding a limit like 50 or 64 characters.

(When the principles are understood, most alt texts will
be relatively short, often empty, sometimes a single character,
often a word or two, and in some cases fairly long, like a
short paragraph. Actually it's rarely a matter of writing
something in the length range 50 to 64 characters or so;
it's usually either considerably shorter or much longer.)

> And long descriptions are required if you are
> providing a text equivalent for a chart or graph.

I think this requirement has often been as more absolute
than it really is. In particular, accessibility checkers
may give the impression that the longdesc attribute is
needed for any image with real content (other than text).
What WCAG 1.0 really says about it is this:

For complex content (e.g., a chart) where the "alt" text does not provide a
complete text equivalent, provide an additional description using, for
example, "longdesc" with IMG or FRAME, a link inside an OBJECT element, or a
description link.
Source: http://www.w3.org/TR/WCAG10/#gl-provide-equivalents

So longdesc is never required. Even for complex content, alt
text could provide a complete equivalent. Even if it does not,
longdesc is just an example of a technique. In fact, in my
honest opinion all the examples mentioned there should be
regarded as secondary at most. A simple explicit link, with
easily understandable link text, is best, in most cases.

--
Jukka Korpela, senior adviser
TIEKE Finnish Information Society Development Centre
http://www.tieke.fi/ My phone +358 9 4763 0397


----
To subscribe, unsubscribe, or view list archives,
visit http://www.webaim.org/discussion/