RE: null alt-text, spacers, etc. etc.

Bruce,

You guessed right that the folks in the survey were experienced, as I later 
came to realize (I was new to the email lists when I asked these 
things).  In the survey there were 11 respondents, although not everyone 
answered all questions.  Eight described themselves explicitly or 
implicitly as blind.   Another 3 respondents were sighted folks with 
experience in the field.

I absolutely agree that it would be useful to have a survey that took into 
account user experience and LONGDESC.  I did that survey in ancient times 
before LONGDESC.

I think we still disagree on bullet. Are you saying that bullets should 
just be ignored?   I don't want to argue too much on philisophical 
grounds.  The bottom line is what do actual users prefer.  However, I do 
want to point out that the answer we get will depend on the examples we 
give.  For example, you have an example

quote
And there is a bulleted list.  January, no activity.  February, we had a
rally.  March, our annual dinner..."
end quote

In that example, the names of the months, january, february, march, etc. 
serve as the list markers so "bullet" is not necessary.  Similarly, if each 
bullet item is just one word like

. Prepare.
. Act.
. Evaluate.
adding any word is extra baggage.

However, if we consider an example where each bullet item is two or more 
sentences, the bullets become more useful  and I bet there would be more 
positive responses.

There's also your rule of thumb, which I like, about reading the page as 
you would read it aloud to a person who was blind... or part of your radio 
audience... or preparing dinner as you read the paper to her...  Do you 
think even you would start saying "bullet" if you were reading something 
where all items are several sentences long?

So as for what we say for bullets, horizontal rules, etc, the full answer 
is "it depends".

Same goes for horizontal rules.  They can be eye candy, e.g. if they simply 
start or end the page.  But in my experience they are as functional as a 
paragraph tag.   If we disagree on this it's probably because we are 
thinking of a different corpus of examples... you're thinking of eye candy, 
I'm thinking of places where they are useful.  Perhaps div renders them 
obsolete... I haven't thought about that... for for the present at least do 
any speech browsers or screenreader/visual browser combos speak them at all?

I guess my feeling is that right now most web sites out there are 
disasters... no ALT text at all, and we're better keeping the rules simple 
even if it means that users will hear "bullet" or "new section" more than 
would be optimal.  So we give them a list of default words, and that's the 
default rule.   Then we say for people who really want to think more deeply 
about optimizing their page, they can use null or blank alt text for 
bullets or rules that really really truly serve no explicit or subtle 
implicit function whatsoever.

It's like if a person has an image labeled "button for contact 
information".  I'd rather they omit the words "button for".  But I'm going 
to downplay that comment and only suggest they follow it for new buttons, 
and polish up their ALT text as last priority.

Len


At 08:39 AM 1/6/00 -0500, Bruce Bailey wrote:
>On Wednesday, January 05, 2000 5:19 PM, Leonard R. Kasday
>[SMTP:kasday@acm.org] wrote:
> > Bruce, I've got to take issue with your statement that
> >> Some of your examples for alt content for bullets and lines are counter
> >> indicated.  There is no other way to say this but that
><Q>alt="bullet"</Q>
> >> and <Q>alt="horizontal rule"</Q> are just WRONG.  Perhaps these should
> >> even be treated as suspicious?  ALTernative text should capture the
> >> intent/function and not be solely pseudo-descriptive.  Using a value of
> >> "---" for ALT is infinitely better than using "bar".
> >
> > This question keeps coming up.  When I did a "survey" a few years ago
> > "bullet" got the highest ratings.  See
> > http://astro.temple.edu/~kasday/web_access/surveys/alt-text.html
> > I put "survey" in quotes because there weren't many responses.  Also the
> > difference in preference between "bullet" and an asterisk is not
> > statistically significant.
>
>If I read it right, their were only ten respondents?  If these ten were all
>sophisticated and experienced web surfers who just happen to be blind, I
>would say that you had something that was really representative.  I
>regularly work with folks with a variety of computer experience and a
>variety of physical and sensory abilities.  My clear impression is that
>folks like Curtis Chung and David Poehlman strongly prefer accessible sites
>over "text-only parallel versions" and brief alt text over verbose
>descriptive alt text.  Newbies (and this includes some folks with political
>influence, but limited computer experience) have the opposite bent.  Since
>the newbies will quickly gain experience, I don't think we should be
>designing (nor creating rules) for them -- except for maybe special purpose
>sites.
>
>A formal survey (that took measure of user experience) would be very
>valuable.  Has the NFB or ACB (or a non-USA equivalent) posted a short
>reference on writing alt text?  The question becomes more interesting as
>support for longdesc improves.
>
> > Now we can take a more philosophical approach and argue that "bullet"
> > describes appearance rather than function.  However, this is inconsistent
> > with recommending "----"  which also describes appearance, using ASCII
>art
> > no less.
>
>Your survey, even with the small sample, did a nice job of capturing the
>reasons people preferred one style over another.  One good reason for not
>using alphanumeric for decorative IMGs is that "read punctuation" is a
>toggle feature available in virtually all screenreaders.
>
>I would argue that "----" is a more function than ASCII art.  72 hyphens
>would be for appearance over function.  Alt like "divider" or "end of page"
>for a horizontal rule just seems silly to me.  Maybe a single dash is more
>in keeping with my argument!
>
> > Furthermore, even though "bullet" describes appearance, it has acquired
>the
> > meaning of  "list item" in at least some subcultures.  In the AT&T
> > subculture, for example, we would say things like
> > "I have a comment about the third bullet"
> > In fact, the Cambridge Dictionary
> > http://www.cup.cam.ac.uk/elt/dictionary/
> > has the following definition:
> > "A bullet is also a symbol, often a small black circle, used in printing
> > and sometimes in writing either to show the beginning of or to separate
> > items in a list. "
> > Note that it says "often a small black circle". So they are implying that
> > appearance may vary.  So they are defining it semantically, not by
> > appearance.  I therefore think it's a good choice philosophically as
>well.
> >
> > I move therefore that we recommend "bullet".
>
>I respect your reasoning and opinion, but I disagree entirely!  I use the
>rule of thumb that ALT represents how I would read a page out loud.  If you
>were reading a web page to a blind individual, you would say something
>like:
>"And there is a bulleted list.  January, no activity.  February, we had a
>rally.  March, our annual dinner..."
>You would NOT say:
>"And there is a list.  Bullet, January, no activity.  Bullet, February, we
>had a rally.  Bullet, March, our annual dinner..."
>The first is what is what HT markup L denotes with <UL> <LI> ...
>
>With a horizontal rule, you would probably skip over it entirely, or maybe
>just pause.  For a long document (that uses <HR> or graphical bars) between
>section, a human reader might include verbiage like "end of section" -- but
>this is what HTML does with <DIV> and horizontal rules (of any type) are
>really just eye candy.
>
>More complex images follow the same "philosophy" for ALT text.  A human
>reader might say:
>"And there is a photo of people at the dinner."
>If the blind person asked for more details, the human reader might say:
>"There are twelve people, seven men and five women.  They are all older and
>big grins on their faces.  They are standing around a table at what appears
>to be a catered dinner at a restaurant."
>This again mirrors what an HTML author provides first with ALT text and
>then with LONGDESC (or maybe a D-link).
>
>I would love to see some consensus on these points.
>-- Bruce

-------
Leonard R. Kasday, Ph.D.
Institute on Disabilities/UAP, and
Department of Electrical Engineering
Temple University
423 Ritter Annex, Philadelphia, PA 19122

kasday@acm.org
http://astro.temple.edu/~kasday

(215) 204-2247 (voice)
(800) 750-7428 (TTY)

Received on Thursday, 6 January 2000 12:06:00 UTC