W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > October to December 1998

Re: Redundant alt text??

From: Alan J. Flavell <flavell@a5.ph.gla.ac.uk>
Date: Wed, 18 Nov 1998 13:15:38 +0000 (GMT)
To: Greg Gay <g.gay@utoronto.ca>
cc: WAI Guidelines List <w3c-wai-gl@w3.org>
Message-ID: <Pine.OSF.3.96.981118130055.14808A-100000@a5.ph.gla.ac.uk>
On Tue, 17 Nov 1998, Greg Gay wrote:

> The utility of ALT text in linked images is questionable. Spending a couple
> hours with a JAWS user, it became immediately clear that having to listen to
> everything twice was a major annoyance to her. She strongly recomended not
> including ALT text with linked images and image maps, but rather to include a
> text link below without the ALT text, or include the ALT text without the text
> links. One or the other, but not both.

That all makes sense to me.  However, it's quite common on non-trivial
pages to have navigation links in more than one place, for example in
the margin at the top as well as in a border at the foot.  It might be
acceptable to use ALT-ed images alone in the one place, and text alone
in the other?

> In support of using text links, JAWS does not handle ALT text well in
> Netscape.

Netscape does not handle ALT text well, period.  If the image is
rather small, and its size is specified in the IMG tag, then it can
happen that NS suppresses the display of the ALT text entirely.  Is
the problem that you're describing maybe a consequence of that?  While
a possible workaround is to omit the size specification, this
naturally is not without other consequences. 

MSIE4 has an option (which inexplicably is turned off by default) to
display the whole ALT text even when it wouldn't fit the stated height
x width (when images aren't loaded, I mean).  Opera also has helpful
options in this area, for example.

There is a sense in which the ALT text has a clearly stated purpose in
the HTML spec (and, of course, in the WAI recommendations), namely to
act as _alternative_ content in the absence of images being displayed,
and there comes a limit to the extent to which content authors can be
expected to work-around the shortcomings of client software that fails
to implement that purpose appropriately.  It's a matter for discussion
just where that limit should be placed, of course.
Received on Wednesday, 18 November 1998 08:16:01 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:28 UTC