[accessibility] Testing PDF tagging with a pdf generated using pdflatex

[Ross Moore]

Hi Boris, and others.

On 17/03/2018, at 7:30, "Boris Veytsman" <borisv at lk.net> wrote:

> RM> From: Ross Moore <ross.moore at mq.edu.au>
> RM> Date: Fri, 16 Mar 2018 20:15:16 +0000
> 
> 
> RM> The big question here is  ‘What should that alternative text be?'
> RM> Just reading the characters separately is not sufficient.
> RM> How about the LaTeX source, complete with backslashes ?
> RM> Some people say that this is desirable.
> RM> What do you think?
> 
> My thoughts the following.  Reading aloud means translating
> two-dimensional math into one-dimensional text.  This is exactly the
> inverse of the problem of translating one-dimensional code into
> two-dimensional math.  I strongly suspect that solutions to both of
> these problems are correlated, and conversion of math to TeX
> representation is the best choice.  Anecdotically many blind
> professional mathematicians are very fine with spoken TeX
> 
> I would think that some slightly preprocessed TeX (e.g. with \cos read
> as co-sain rather then backslash-si-oh-es) is the ideal.  In the first
> version we can drop the preprocessing.

There are many instances of this kind of thing.
Another is when the math expression is a single letter, $a$ say.

A screen reader will probably say  'uh'  as in the English word,
whereas we would want to hear it phonetically as 'ay'.
This could be put into alternative text, but maybe better is   'variable-ay'
to be precise about how the symbol is being used.

You and I can have our own ideas on this kind of thing, but I'd like to hear directly from visually-impaired mathematicians and students about what they would like to
  1. have read to them;  and
  2. what should be presented through a Braille keyboard.
Should it be the same?
Can it be different, or does Adobe's API for Assistive Technology make that decision?


With your  \cos  example, someone with a Braille keyboard would surely be mystified by receiving the phonetic spelling. 
Besides, a screen reader might well say  'slash cos'  (with the spoken word as the lettuce).

I feel that we cannot optimize for the spoken form without making it much harder for a Braille reader. And conversely. That's why we need input from those who use the technologies.
My understanding is that screen-readers and Braille keyboards are often used together.

I know that this is an argument against 'variable-ay'.

But then it should also sound reasonable with a low-tech screen reader, such as Adobe's Read Out Loud.

So what is the best compromise?


Another possibility is to use a Unicode character as a delimiter to start and end mathematics.
That could alert the reader that what comes is not ordinary words.
There is such a character in Braille-math, isn't there?
But what would JAWS do with this, in an  /Alt  replacement?
And other screen readers?


> 
> -- 
> Good luck
> 
> -Boris
> 
> The best thing about growing older is that it takes such a long time.


All the best,

   Ross


Cheers,

   Ross