W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2001

Re: alt to text

From: Alan J. Flavell <flavell@a5.ph.gla.ac.uk>
Date: Thu, 3 May 2001 22:54:48 +0100 (BST)
To: Anne Pemberton <apembert@erols.com>
cc: WAI Guidelines List <w3c-wai-gl@w3.org>
Message-ID: <Pine.OSF.4.21-pb.0105032201440.28497-100000@a5.ph.gla.ac.uk>


You have asked me directly, and I shall therefore offer you my
best answer in the circumstances.  But as I said before, it's my
opinion that the best choice depends on the author's purpose.

To my way of thinking, what an author is providing in a web page is
not a pre-programmed presentation, designed for a particular kind of
user (that isn't feasible, as there are groups of users who have
mutually contradictory requirements): what the author is providing is
content, in this case in textual form, to be interpreted by whatever
"assistive technology" is being used.  I put that phrase in quotes,
since for some users the "assistive technology" might be nothing more
than an ordinary text-mode browser such as w3m or Lynx.

On Wed, 2 May 2001, Anne Pemberton wrote:

> In the case of the "Picture of George
> Washington", the title, which followed the picture on the same line was
> "George Washington in the French and Indian War" ... the purpose of the
> picture was to illustrate the "who" of the content ...

Let me try to set out my train of thought.

The picture is of G.W, and your title, which you say is part of the
normal text, already says so.  So, there seems no need to repeat that.  
The purpose of the picture is to visually identify the person; but
we're dealing with a situation where such a graphical presentation is
unavailable, and the text already has identified the person.  
Naturally, if you want to supply additional information about what the
picture contains, you have longdesc/D-link for doing so, but that
isn't the immediate issue here.

Since I understand you to say that there was no particular content in
the picture which you wanted to bring to the reader's attention, aside
from reinforcing the identification of G.W., then if it had not been
for your title/caption, I would have recommended considering an empty
ALT text.  Had the picture, on the other hand, contained important
information or been of particular artistic interest or merit, a
different recommendation would have seemed appropriate.

(As an aside, I have the feeling that there are benefits in making
picture captions be a complete sentence, so that they can stand alone
as if they were side-notes, without seeming to dangle.)

As it is, that dangling title would seem to call for some kind of
marker to justify its presence.  ALT="[Picture]" is hardly ideal, but
I guess serviceable in such a situation.

> In the case of the
> "Illustration of an ideal web page", the picture is followed by Checkpoint
> 3.2 ... The function is to illustrate the componants of a ideally
> accessible web page. 

Now this is harder, isn't it, because if the picture contains vital
evidence to support the checkpoint, then merely announcing to the
reader that this is a picture of vital evidence, without telling them
what that evidence was, would be pure frustration.  Does the text
already contain that evidence?  If not, then a long description would
seem essential - if the evidence can be summed up in a few pithy words
then the ALT text could state it directly - butI guess that's not
feasible in this context.

Let me illustrate what I mean by taking a different example.  Suppose
the picture were a plot of an endangered species, then I have seen ALT
texts such as "Plot of population against time".

Well...?  The text-mode reader wants to know what happened!  The ALT
text should rather say "Plot shows population fell dramatically
through 1980's until fishing moratorium imposed", or whatever it is
that the plot is showing.

To some extent I'd submit that the choice may depend on whether the
illustration is vital to your purpose, or was merely meant to be
illustrative or reinforcing of some main theme.  In the former case
then you may have no alternative than to hit the reader with the fact
that there was a complex picture here, and get them to read the
supporting long description (the notorious "thousand words") to get
the best out of it.

In the latter case, that approach runs the risk of distracting the
text-mode reader with much detail over something that was only
intended to be a visual side-note reinforcing the main theme; then it
would seem better to wrap up the same message briefly in a piece of
text, slip it in ALT text as if it were a side-note, and pass on,
without bothering the reader with the fact that there was a picture
that they can't see.

This is my personal opinion, nothing more. 

As I say, in my estimation it's up to the author to consider what the
purpose of their image was, and then to purpose some text "to serve as
content" (that quote from the HTML4 specification again).  In other
words, to be a best-value alternative (equivalent, replacement,
"substitute") for the purpose of the image.

best regards  (and do feel entirely free to pick these suggestions apart).
Received on Thursday, 3 May 2001 17:55:14 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:10 GMT