By John Gardner
Electronic text has been accessible in principle for
decades. I used several brands of screen
reader with DOS computers, switched to Linux for a few years, and finally began
using Windows in the mid-1990’s. But I
was never able to read a mainstream math equation until 2004 when the
MathPlayer plug-in to Internet Explorer was enhanced to permit audio access to
the equation. Until that time, the only
way that a blind person could read mainstream math was to read the source file
of a document composed in one of the page-setting family of TeX languages. Generations of blind people either had to
learn TeX and somehow obtain a copy of the original source from the author, depend on sighted helpers to “make math
accessible”, or just not do math. TeX is
accessible, but it is hard to argue that it is really very usable when, for
example, the equation for the solution to the quadratic equation (fraction with
numerator –b plus/minus square root(b squared – 4ac) and denominator 2a) is
written using LaTeX (the most popular flavor of TeX):
$x=\frac{-b\pm\sqrt{b^2-4ac}}{2a}$. Imagine trying to solve equations for algebra homework using this notation!
$x=\frac{-b\pm\sqrt{b^2-4ac}}{2a}$. Imagine trying to solve equations for algebra homework using this notation!
With the MathPlayer enhancement, it became possible for me
and other blind people to read math in mainstream web documents provided they
were expressed in the new math language MathML.
Unfortunately MathML was not displayed visually by many web browsers,
(including Internet Explorer unless MathPlayer was installed). In those early days of MathML, pages coded in
that language just crashed when displayed on most web browsers. Not surprisingly, few authors used MathML. Instead, they continued to display math
equations as images – which were of course inaccessible to me.
Today, MathML is becoming fairly popular, because equations
can be displayed on any web browser with a new web technology, MathJAX, developed by a consortium of scientific organizations. Most authors now want to use MathML, because
MathML equations are much prettier and more readable than equation images. This is good for blind people, because more
and more scientific literature is becoming directly readable. Life is still far from perfect however,
because MathPlayer makes a math equation accessible as a string to be spoken. For example, the equation 1+1=2 is displayed
by MathPlayer as “one plus one equals two”.
The equation can be captured and copied or displayed on a braille
display as exactly that string of text, not as “1+1=2”. This is access, but it sure isn't excellent
access. A Braille reader, or somebody
who wants to have a written record would be far better served by the compact
math notation and not the spoken string.
This not-excellent access is due to limitations of present
day screen readers and not from the MathML.
So further development should eventually provide excellent access to
MathML. Since the introduction of
MathPlayer accessibility, I have often said that “accessibility of math for
reading is a solved problem in principle, but writing math is not”. In one important sense, that second statement
is dead wrong. People who are fluent in
math Braille can not only read math readily but can also write math and
develop algebraic solutions with moderate facility. MathML is translatable to math Braille, and
there are several applications that do that, although all have sufficiently
serious bugs/errors that responsible math transcribers still proof most documents
and fix errors before releasing it to the blind user. Computer translating math Braille to standard
notation is not really feasible, partly because there are ambiguities that
computers cannot easily sort out, but mostly because back translation of all braille
math codes are extraordinarily sensitive to author errors. In practice, a transcriber is necessary for
Braille math to be converted to standard notation. This is not likely to change unless one uses
math Braille codes that are not “context dependent’. So at present, Braille math is not really a
feasible way to access mainstream math, especially for writing.
At the present time, TeX is still the most widely used
method for writing mainstream math by blind people. Triangle, a linear notation developed in my
Oregon State University group, the European Lambda editor,
and the ChattyInfty.
Applications are used by a few people, but none of these fully meets needs. So about a year and half ago I got tired of this situation and decided to develop a blind-friendly interface to standard math applications. My goal was to allow really usable access for both reading and writing. I call it the LEAN method, and I will describe it in detail in a later blog post. For the moment, let me say that it is a descendant of Triangle and Lambda, using a compact linear notation and a lot of very useful shortcuts for doing lots of math manipulation, particularly with fractions. The first implementation is to MSWord with the MathType editor, the authoring software used by 75% of people in the world who publish scientific documents. For the first time, blind people will be able to read and author scientific Word documents almost as easily as sighted authors can. LEAN Math has just gone into beta testing, and I expect it to be available commercially within a few months. It will be free to individuals and available at modest cost to institutions.
