[Blindmath] Questions about complex formulas, tagging PDFs, and users' expectations

[Michael Whapples]

Hello,
Here are my views:
1. There are standards various people/groups have come up with for 
speaking maths, how "standard" these really are I don't know, the main 
thing is make it clear what you mean (IE. don't say "x over 2 y", where 
is the y, do you mean the LaTeX \frac{x}{2}y or \frac{x}{2y}, either 
specify where the fraction begins and ends or add brackets eg. "x over 
open bracket 2 y close bracket" or "open bracket x over 2 close bracket 
y" depending on what you meant). However unless you intend to make a 
spoken version of the articles (IE. an actual audio recording) then this 
shouldn't concern you much as its the responsibility of the reader 
software to do this for you.

2. The more important question I think. I have to say for accessibility 
of maths forget PDF for the moment, I think there is still some work 
(both technically and in convincing some important players in this, eg. 
screen reader manufacturers). The main options I would recommend are:
* produce a mathml version of the document, this is meant to be the 
preferred form of maths on the web although take up does seem to be 
slow. Screen readers (at least in windows) can access this using 
mathplayer from design science or possibly Braille copies can be 
produced using liblouisxml. If you produce the document originally in 
LaTeX then look round for some LaTeX to mathml translation software, two 
I found reasonable are tex4ht and ttm. Mathml results may not be the 
very best if you convert a document from LaTeX, but in my experience 
they were usable.
* Publish the document in HTML form with images for the equations and 
setting the equation images' alt-tags to the LaTeX source for the 
equation. This can be done automatically from a LaTeX document with some 
software, it may be harder to find as the task is harder to define in a 
search. One piece of software I know of which does this is plasTeX 
(http://plastex.sf.net). The disadvantage of doing this is that the 
browser considers the equation as an image so cannot do any alterations 
(IE. it is not able to use different fonts or different colours, etc) 
where as mathml would allow or technically allows the browser to do 
this. Another down side is that anyone unable to see the image needs to 
know LaTeX to make use of it, but may be this isn't such an issue as 
LaTeX for writing an equation isn't too hard and generally I would 
recommend (I think quite a number of others here would also) learning 
LaTeX if a visually impaired person wants to make serious use of maths.
* The final suggestion I would make is to provide the source LaTeX as 
well as a "ready to read format" eg. PDF and then users who don't like 
the final document format can go back to the source. This I could 
imagine could cause some concern for some authors/publishers and may 
also not be helpful to most people as there might be a lot of extra 
formatting LaTeX in the source not really needed if you just want the 
document meaning. Also readers needing the LaTeX source must know LaTeX, 
this is made more complex with the additional document formatting 
commands, newcommands, etc, for someone learning LaTeX it can be hard to 
know what commands might be important for understanding the document and 
what can safely be ignored. I consider this option as a bit of a last 
resort, try and do one of the other two first.

Hope this gives you some ideas and some useful pointers.

Michael Whapples
On 29/10/09 23:27, joseph.dalaker at census.gov wrote:
> Good evening
>
> I'm a statistician at the U.S. Census Bureau.  My work group has some
> statistical papers with lots of complex formulas (with Greek letters,
> indexed variables, letters with circumflex marks, subscripts and
> superscripts, etc.) that we want to publish on the web, and also ensure
> that the formulas can be read with a screen reader.  Everyone in my work
> group (including myself) is sighted.
>
> We traditionally have put up papers as PDFs, but I haven't had much luck
> getting Adobe to recognize how to read these formulas out loud.
>
> I have 2 questions about what to do next -- the first is about users'
> expectations, the second is technical.
>
> 1)  If we're stuck tagging the formulas by hand as figures, does the blind
> community have a commonly-held standard that is expected for reading
> complex formulas aloud?
>        I found this reprint of an article by Abraham Nemeth that describes a
> protocol for reading formulas aloud.
> http://people.rit.edu/easi/easisem/talkmath.htm
>
>        That Dr. Nemeth's system looked clear and elegant to me, coupled with
> the fact that Nemeth code is also used in Braille for math, makes it sound
> plausible to me like the above protocol was likely to be "the" standard
> protocol that the blind community would expect for math formulas, but I
> want to check that assumption.  I heard of an organization that reads and
> records textbooks, that I'm pretty sure does not use the above protocol
> verbatim.
>
>        Is there a "right" way to read formulas?
>
> 2) Following on the heels of 1), if we were to find software that could tag
> complex math formulas automatically, are there software packages that you
> think provide good results and are easy to use?
>
>        Or failing that, are there software features that you would recommend
> that we look for?   And what websites do a good job of tagging complex
> formulas?
>
> Thanks for any input you can provide.
>
> --Joe
>
>
> Joseph Dalaker
> Statistician
>
> Government Organization and Special Programs Branch
> Governments Division
> U.S. Census Bureau
>
>
>
> _______________________________________________
> Blindmath mailing list
> Blindmath at nfbnet.org
> http://www.nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
> To unsubscribe, change your list options or get your account info for Blindmath:
> http://www.nfbnet.org/mailman/options/blindmath_nfbnet.org/mwhapples%40aim.com
>