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

Re: ALT text (re Gregg Vanderheiden) (fwd)

From: Alan J. Flavell <flavell@a5.ph.gla.ac.uk>
Date: Wed, 18 Feb 1998 11:14:42 +0000 (GMT)
To: WAI Guidelines List <w3c-wai-gl@w3.org>
Message-ID: <Pine.OSF.3.96.980218105652.24451B-100000@a5.ph.gla.ac.uk>

(I originally sent this only to the -ig list, because it was in
response to one that had been sent only to the -ig list.  However, I'm
asked in email to conduct this discussion on the -gl list, so here
goes, with minor editing)...

On Tue, 17 Feb 1998, Samsara Vagabond wrote:

(about using the ALT attribute to provide a description of the image)

> It's text,

yes

> it's an alternative,

this is where I disagree.  It's not, in general, a real alternative...

>  and it's presented when the image
> is not displayed.

yes

> Seems to fit to me. 

Superficially, maybe, but it doesn't address the "topic of discourse" 
(a rather pompous phrase, sorry, to describe a rather important
concept) 
 
> While I agree that it's better to replicate the
> function of an image when the image has an external
> function,

Not only.  If the image is conveying information (about the topic of
discourse) then the alternative text should also be conveying
information about the topic of discourse.  Providing a description of
the image is, in general, not the same thing as conveying information
about the topic of discourse.

> Succinctly put, can it be considered "accessible" to withhold either
> information 

I'm not asking you to withhold information.  You have the TITLE and
LONGDESC attributes of the IMG (or interim D-link) also (and the TITLE
attribute of the anchor link, if the IMG is within one); each has its
appropriate field of application. 

> that could be used to decide whether or not to view an
> image, or the existence of the image entirely?

It's the job of the browser to alert the reader to the fact that this
ALT text is standing-in for an image, by whatever means the browser
designer chooses.  Most browsers make that clear (it's not quite
obvious with Lynx, I admit; but the Lynx developers are very receptive
to accessibility issues: if this forum produces a viable set of
principles, I've every confidence that the Lynx developers would find
a way to realise those principles in their implementation).

> Certainly it makes reading more comfortable for people who
> have decided not to interact with images in any way, but
> at the expense, I would think, of people who may
> want to interact with SOME images, in SOME manner.

But that is exactly the way in which I work as a reader (i.e with
image loading off, and only loading the specific images I decide that
I am interested in), and I believe my authoring advice is quite
appropriate for that, even with today's browsers (for which there is
still scope for improvement). 

I should stress that I claim no specific expertise in relation
to accessibility.  But a properly marked-up WWW document is meant to
be for everybody.  Put the content-based information into the document
markup, and put the presentation-dependent machinery into browsers,
that has been the underlying principle of HTML from the start.

> Now, HTML 4.0 could change this, using the title
> attribute. So is it acceptable to assume that
> everyone reading your page has a user agent that
> is title-attribute-aware? 

TITLE isn't really new, it's been on the agenda for at least 2 years; 
LONGDESC is new.  I understand your reservations about current
browsers, but we really must not choose the wrong long-term answer
because of short-term problems.  Transitional arrangements are always
possible - but first we must have a clear idea of what we are trying
to transition _to_. 

> 1) Replace stylized text images with the equivalent text.

Sure: preferably use style sheets _instead_ of making images of text.

> 2) Replace images that have an external function (link,
> warning icon, etc.) with a word or phrase that replicates the
> function. Failing that, replace it with a word or phrase that
> describes the function.

I agree

> 3) Replace images with no external function (i.e., images
> that are there just to be images) 

I don't understand what you mean by this.  Surely no-one scatters
random images in their documents for no purpose?  Whatever their
purpose was for providing the image, that also should be the purpose
of the ALT text.  That is the genuine "alternative" in my opinion, and
that is a property of the author's document itself; it's not an
attempt to bolt-on extra accessibility - something which, to be candid
about the majority of authors, will only be done reluctantly and
without much consideration (present company, naturally, excepted!). 

Meaningful ALT text is something which even an author untutored in
accessibility needs can be expected to supply, whereas the specific
needs of different audiences are best addressed by the designers of
specialised browsers, who understand how best to _present_ the
available _information content_ in different situations, and thus
leverage the effort across the WWW, irrespective of the individual WWW
authors. 

And I still say that there's a case for empty ALT texts, in specific
circumstances.  If they cause a problem with a particular browser,
then let the browser solve the problem.  Lynx, to take an example, has
no problem with this (you just issue the "*" keycommand if you want
handles on all images, or you start up with that option as your
default if you wish). Typical graphical browsers leave you in no doubt
about the presence of images, with or without ALT="".

best regards
Received on Wednesday, 18 February 1998 06:15:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:46:57 GMT