Re: ALT vs. TITLE usage in WAI logo example

On Fri, 29 Dec 2000, Al Gilman wrote:

> At 09:25 AM 2000-12-29 -0500, Leonard R. Kasday wrote:
> >Hmmm.  Speaking of small problems with the logo, presently the image has
> >
> >ALT="Level Double-A conformance icon, W3C-WAI Web Content Accessibility 
> >Guidelines 1.0"
> >
> >Following the principle that ALT should be an equivalent for what the 
> >sighted person sees, would it not be better to have
> >
> >ALT="W3C: WAI triple A, WCAG 1.0."

When the icon is used for its purpose, I'd say that this is along the
right lines as far as the information content of the ALT text is
concerned, though it may have been pruned back too hard.

My problem with the original is not that it might be too long, but
that it makes what seems to me a pointless reference to a graphical
token (the icon) rather than to the factual situation (AA-conformance
etc.) that is the central point of the action here.

Unless the token were genuinely essential (hypothetically: "no
accessibility claims are genuine without the authorized W3C icon"),
then calling attention to the icon rather than to the claim which it
represents seems to me to be a pointless distraction for text-mode
readers.

> >The longer description is an explanation that in principle is seen by 
> >everyone, so it should be the TITLE
> >
> >TITLE="Level Double-A conformance icon, W3C-WAI Web Content Accessibility 
> >Guidelines 1.0"

This seems to fit my understanding of the appropriate usage of TITLE
and ALT attributes.

However, there isn't a formulaic answer to this question: it depends
on context.  On a page whose purpose is to exhibit a selection of WAI
icons and to discuss their appropriate usage, the optimal ALT text
could be very different.  The page itself might not be AA-conformant,
and it would be quite wrong to use, without an appropriate
explanation, an ALT text that had been designed specifically for
asserting that it was so.

> AG:
> 
> I have to admit to a violent negative reaction to this suggestion.
> 
> Both of the above spellings for ALT text satisfy the sense of 'equivalent' as
> used in checkpoint 1.1. 

One has to make up one's mind whether the intention is to supply the
information, or to give the text-mode reader a running description of
a visual presentation.

I'm aware that some blind readers have taken the view that a web page
is essentially visual and that they wish to be given a verbal
description of that visual presentation.  For some subset of web
pages, I concede that this is a reasonable description of the
situation; but I suggest that it would be a mistake to extend this
approach to all web pages - for many of which, a detailed account of
its visual presentation would be pointless, irrelevant, and could
seriously distract from the potentially complex information content
for which the page had been made.

As a sighted reader, I've commented from time to time that the radio
can have better pictures than TV.  But this isn't because the radio
drama is laboriously reading out the stage directions to us, which is
effectively what seems to be happening in some unsatisfactory ALT
texts.

As a general principle and subject to appropriate exceptions, I would
much rather make pages which adapt themselves gracefully to the
browsing situation in which they find themselves, and do their
respective best to convey their information content to readers in each
kind of browsing situation, without the potential discourtesy of - in
effect - saying "by rights you should be reading this visually on a
graphical browser, rush off and get one right away".

Sometimes you have no choice, if the information itself was
essentially visual.  But that's no reason to turn everything into a
visual experience, with only third-rate concessions for anyone else. I
take it we all pretty-much agree on that principle, though, otherwise
we wouldn't be here.

> It is a loose equivalence and they both meet this
> loose criterion.  It is an equivalence of sense; not of spelling.  And at the
> more superficial level that you are evaluating 'equivalence' in the above, the
> answer is _no_, the ALT should not be 'equivalent' to the visual presentation
> _in those terms_.  It should be _different_ in those terms so as to be both a)
> equivalent at the deeper level and b) effective in the alternative
> presentation.  The ALT text as understood when heard should be equivalent to
> what the sighted user understands on seeing the visual logo.  "What the
> sighted
> user _sees_" is too medium-specific to be the base for equivalence
> comparison. 
> "What the sighted user _learns_" is closer.

You are indeed looking for equivalent _information content_, as
opposed to a description of a visual experience.
 
> The fact is that telegraphic diction such as exemplified by the suggested
> alternative ALT above works in a visual environment where the user can random
> scan over the component elements, and does not work well in a serial pass over
> the content such as in synthetic speech.  The more expanded version is more
> likely to be comprehensible when read aloud than is the series of cryptic
> blurps.

You're saying that ALT="W3C: WAI triple A, WCAG 1.0." , just standing
there alone without context, is incomprehensible, is that the issue?  

But I don't see how that shortcoming is remedied by providine what
was, in effect, a description of a visual experience ("triple-A
conformance icon" or whatever).  Again, one runs the risk of confusing
the drama with the stage directions.

ALT="This page conforms to W3C WAI triple A, WCAG 1.0." , perhaps?
With whatever kind of delimiter or break which might be needed to
offset it from any running text.

best regards and season's greetings

Received on Saturday, 30 December 2000 11:52:16 UTC