Remarks on WCAG 1.0 checkpoint 3.1 discussion

One of the basic features of the Web Content Accessibility Guidelines is
that they strive to specify what must be provided in order to attain a
high degree of accessibility. Constraints on the freedom of choice
available to content developers are only imposed where necessary to avoid
design practices that are detrimental to the underlying objectives of the
guidelines. Thus, for example, if the content were the same and kept up to
date properly, it would be entirely possible, to take an extreme case, to
create two versions of the same content, one of which is richly structured
with semantic markup, with the other being purely presentational. The
underlying requirement is that the structural markup must exist, not that
it must be provided to the exclusion of all parallel versions of the same
content.

Let us now apply these principles to the checkpoint 3.1 (WCAG 1.0)
situation that is under discussion here. Does the use of text in images
detract from the developer's capacity to ensure that proper logical
structure and semantics are available? Indeed it can, but only under
certain conditions. Thus, as Charles pointed out in last week's meeting,
if an entire paragraph of text is represented in an HTML document as an
image with an ALT attribute, this is problematic as the structural
divisions within the text can not be defined in the markup (an ALT
attribute is one long character string). However, it has been argued that
a short fragment of inline text need not create any problems in this
respect, if it contains no internal semantic distinctions that require
special treatment in markup. Furthermore, even where the use of text in
images does create access issues, there are alternatives available,
including:

1. The application of language features such as the OBJECT element in HTML
or the SMIL SWITCH element, which allow complete alternatives, including
block-level markup and references to alternative versions of content, to
be included, for selection by the user agent.

2. The construction of alternative versions of the content, to be selected
either directly by the user or via a negotiation protocol. Thus, an SVG
version could be provided alongside a bitmap image (in fact the latter
could be generated from the former).

3. The inclusion of decorative images in style sheets rather than as
actual content.

Thus I have argued, first, that the use of text in images may or may not
constitute an access problem. Where it does, there are various options
available (more so in newer markup languages than in HTML 3.2 for example)
which enable the access difficulty to be overcome. Which solution is used
will depend on the objectives of the content developer and the extent to
which backward compatibility is regarded as essential. Is there any reason
why the guidelines should prescribe one solution over the others, given
that they are all acceptable from the access standpoint?
To what extent should this working group attempt to prescribe which
technologies should be deployed, in cases where "accessible" solutions are
available for each of the possible choices?

Received on Tuesday, 24 October 2000 20:07:27 UTC