[Blindmath] Proposal for making math more accessible

Mon Jun 6 04:34:52 UTC 2011

I fear this will turn into a TeX vs MathML or TeX vs Word or some other
battle.  But I'll be intrepid and try and correct a few misstatements...

For people in some scientific fields, TeX is indeed used by most everyone in
that field for scientific documents.  In my field of computer science, TeX
is (or at least was) universally used -- I wrote my thesis using it.
However, the reality is that those fields do not represent the majority of
authors of STM documents  Design Science did a survey of STM publishers
about 10 years ago and found that about 70% of the documents submitted to
the publishers were Word docs that used MathType for the math.  I'm reminded
of a quote by Pauline Kael who, after Nixon won the presidential election by
a large majority, said something like 'I live in a rather special world. I
only know one person who voted for Nixon.'

Years ago, I was part of the working group that put forth the MathML
standard.  We had long debates about whether to use a TeX-like syntax or
come up with something that uses an XML syntax so that the tools of the web
(XSL, XPath, CSS, ...) could manipulate it like other XML.  Ultimately, we
decided that being a first class web citizen was more important than using
some variant of a language that was already widely used in a different
context.  We wanted to avoid the situation of math being "special" and
requiring separate tools to parse, access, link to, etc.  Hmm, maybe I
should say we wanted to avoid having the situation where math was "separate
but equal" :-)

MathML was never intended to be an authoring language and the MathML working
group never saw MathML as competition to TeX.  There are in fact many
TeX-to-MathML converters available.  Just as you would rarely directly edit
HTML these days, or for that matter PDF, there are tools that convert TeX
and many other formats (such as MathType or Mathematica) to MathML or PDF.
MathJax (mathjax.org) is one of those tools that does that conversion
internally when it does it's layout in a web page.  MathJax should make it
obvious that math on the web is not a TeX vs MathML debate -- either works
fine with MathJaX and indeed MathPlayer 3 (still two weeks away :-( ) will
make either accessible.

Paul:  you mentioned that images are often the output of TeX.  That is true
for some TeX to HTML convertors but not true for TeX to PDF converters
(which I believe is the most common format for publishing with TeX).  In
PDF, the characters are drawn using the same commands as text is drawn
with.  The difference however is that math is not linear and one can't
simply put the math back together by concatenating the characters as is done
with text in PDF.  I am working with one of the authors of pdftex to try and
get pdftex to put out tagging for math (along with text).  TeX's compiler
was not designed with this in mind and so it is taking quite a bit of work
to do that.  I do have a prototype that works with some of his hand-crafted
output that speaks math in acrobat documents just as MathPlayer does in HTML
documents.  So perhaps some day, you will be able to open the same PDF
document as your sighted colleague and listen to and navigate the document.

Accessibility, math or otherwise, doesn't come for free.  The web wasn't
accessible when the first browsers were built.  Standards committees and
software vendors have worked hard for years to try and come up with
documents that are universally accessible, and with every new technology,
the struggle continues.  Unfortunately, there is no magic bullet for
accessibility -- it takes real work to get it.

Neil Soiffer
Senior Scientist
Design Science, Inc.
www.dessci.com
~ Makers of MathType, MathFlow, MathPlayer, MathDaisy, Equation Editor ~

On Sun, Jun 5, 2011 at 5:11 PM, Paul Wright <paulrite at math.umd.edu> wrote:

> Hi Ken,
>
> You raise some good issues, as did Kevin Chao in his e-mail yesterday
> evening.  I don't know about Octave, but Maple absolutely will give you
> LaTeX output if you ask for it.  MATLAB, which I use extensively, will
> give you LaTeX-formatted output on demand.
>
> My main point is that there is already a standard in the world of
> serious mathematics, science, and technology for writing math equations
> in ASCII code.  The standard is TeX, often seen through its variants.
> (For example, from the website of MacKichan Software, the makers of
> Scientific WorkPlace, "In Scientific WorkPlace, you can typeset complex
> technical documents with LaTeX, the industry standard for mathematics
> typesetting. Because of its superior precision and quality, publishers
> and writers of scientific material use LaTeX extensively.")  Trying to
> use anything other than TeX automatically cuts you off from the vast
> majority of the technical writing done today.
>
> The biggest difficulty in trying to make STEM documents accessible is to
> get everyone in the technical writing world to use one standard, and
> that problem is already solved.  I know that for more basic mathematics,
> such as 3x+5=12, Microsoft Word is used by some individuals, but it is
> never used by publishers or serious technical writers.  Furthermore, I
> don't think basic algebra equations are what we are really discussing,
> and for the more complicated stuff, Word is just not used.
>
> The second difficulty is that this standard, TeX, was designed for
> converting the ASCII code people write into pretty images, which are not
> accessible.  However, in the earlier '90s, T. V. Raman developed AsTeR,
> and later Emacspeak, which converts LaTeX documents into speech in an
> intelligible way.  So, the second problem is also solved.
>
> What remains?  Well, the documents generated using TeX that are
> distributed currently obliterate the ASCII code used to generate the
> equations, and replace them by images.  I am no expert in this area, but
> I don't think it would be that hard to modify the process so that these
> images retain tags with the original ASCII code.  The software makers
> would have to do this, but, once done, the end users would not even know
> the difference.  Everyone could make accessible documents without even
> knowing it!
>
> Finally, a plug-in for screen reading software programs, based on AsTeR
> would be needed to decode the embedded tags into speech.  Again, this is
> not a trivial issue, but I don't think it is too hard.
>
> I'm afraid that efforts to make STEM documents accessible that don't use
> TeX as their starting point are unfortunately destined to leave us in a
> separate and very unequal world.
>
> Best,
>
> Paul
>
>
>
> On Fri, 03 Jun 2011 21:11 -0400, "Ken Perry" <kperry at blinksoft.com>
> wrote:
> > The only problem I see with this idea is it assumes people use Tex,  with
> > mathml, and other xml document types I think this needs to be wrote more
> > as
> > a math standard for publishing so that the output includes speech output
> > as
> > well as any other accessibility needs.  For example what does Maple,
> > Octave,
> > and others save their formulas in I don't think it's Tex.  Also Microsoft
> > Word saves its stuff in some XML format that is now docx
> >
> > Ken
> >
> > -----Original Message-----
> > From: blindmath-bounces at nfbnet.org [mailto:blindmath-bounces at nfbnet.org]
> > On
> > Behalf Of Paul Wright
> > Sent: Friday, June 03, 2011 5:08 PM
> > To: blindmath at nfbnet.org
> > Subject: [Blindmath] Proprosal for making math more accessible
> >
> > Hi Everyone,
> >
> > Sorry for getting in a little late, but I wanted to respond to the
> > discussion from last week regarding making math more accessible.
> >
> > I have limited experience with MathML, and so I can't critique it
> > directly, but it seems to me that there might be better ways of at least
> > substantially addressing the current difficulties.  Furthermore, I agree
> > with Ken:  True accessibility means that the math documents everyone
> > creates are automatically accessible.  Otherwise, we are trying to live
> > in a "separate but equal" universe, and I personally don't want to be
> > there.
> >
> > Here are the relevant points to what I have been thinking about:
> >
> > 1.  Virtually all serious mathematical typesetting is done using some
> > variant of TeX.  Implementations such as LaTeX are not hard to use, and
> > these days you can plug TeX equations into many standard word processing
> > documents. TeX is the basis for everything from advanced physics papers
> > to most of the math textbooks.  If you want serious math written by the
> > general population to be accessible, you have to deal with TeX based
> > code.  Furthermore, TeX is open-source and free.
> >
> > 2.  TeX was written to convert ASCII code into beautiful math equations
> > on paper and screens.  However, the problem of converting the same code
> > into speech has also (more or less) already been solved by Emacspeak. My
> > understanding is that this is also open-source.  However, Emacsspeak
> > must deal with the original .tex document, because math equations
> > usually end up as images in things like PDF files.  This is highly
> > inconvenient, because most people don't distribute their source files,
> > which can look quite ugly and come in many pieces.
> >
> > 3.  Here is what I would like to see happen:  TeX implementations need
> > to be slightly rewritten so that, when they generate, say, a PDF file,
> > they include the source code of each equation as a tag.  Then a plug-in
> > for screen reading software, based on AsTeR (the basis for Emacspeak),
> > could be made.  When installed, whenever your screen reading software
> > came to an equation, it would switch to AsTeR and read the equation in a
> > pleasant way.
> >
> > I know that I am leaving out some details, but to me this seems like the
> > simplest and cheapest way to make most of the math documents being
> > generated today accessible.
> >
> > Any thoughts?
> >
> > Best,
> >
> > Paul
> > -----------------------------------
> >   Paul Wright
> >   Department of Mathematics
> >   University of Maryland
> >   http://www.math.umd.edu/~paulrite
> >
> >
> > _______________________________________________
> > 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/kperry%40blinksof
> > t.com
> >
> >
>
> -----------------------------------
>  Paul Wright
>  Department of Mathematics
>  University of Maryland
>  http://www.math.umd.edu/~paulrite
>
>
> _______________________________________________
> 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/neils%40dessci.com
>