[Blindmath] What does "support braille math in a screenreader" mean technically?

[Michael Whapples]

Hello,
Susan, your first comment is something I have been thinking about for 
some time, Braille translation, particularly for maths, up to now has 
really not been perfected in quite a number of years of people working 
at it. Therefore, is the call simply for Braille support a "dead end", 
will it ever be perfected?

There probably is a place for Braille, but I am more thinking of when 
using a screen reader where interaction with the computer is possible 
then might there be better, more interactive ways of accessing the 
maths? Also might more interactive ways of exploring the maths be a 
simpler problem to solve?

While I haven't got a full idea together yet for what this more 
interactive reading would be, knowing how people process equations would 
be useful (any links), here are some thoughts which may help understand 
what I mean.

MathPlayer, while being good in accessing the mathml, I think we can all 
agree really does not use the computer to its full to give the best user 
experience (some of the limitations are external to design science and 
there is nothing which they can do without help from other companies). 
One area where MathPlayer is weak, is easy navigation of the equation, 
currently you must wade through the words of what the equation is. There 
may be times where you don't need fine details of the equation but the 
general form of the equation. May be parts of the equation could be 
presented to the user in a more concise form (an example might be the 
systm shown in the presentation "Spoken Mathematics Using Prosody: 
Earcons and Spearcons" as presented at ICCHP 2010). May be certain 
elements (eg. fractions) could be container objects (like voiceover on 
the Mac handling tables, or NVDA's object navigation), so you can either 
skip over the fraction should it be of little interest at that time but 
should you need more detail then you could descend into the fraction to 
look at detail.

Also by using a computer there might be a more natural way of 
communicating the mathematics (speech is one example) and so it probably 
will be relevant to more people, therefore it would have a higher value 
to the screen reader vendors, so having a greater chance of being 
implemented.

Michael Whapples
On -10/01/37 20:59, Susan Jolly wrote:
> Despite the Nemeth code having been a US standard for something like 
> 40 years, there is no transcribing software that can currently convert 
> print math fully accurately to Nemeth.  (By print math I mean 
> electronic math represented as either LaTeX or presentation MathML.) 
> This problem persists despite there having been a significant amount 
> of work addressing this issue. The same is true for other math codes 
> but since I know more details about Nemeth I will stick mainly to that 
> code here. So it seems to me that it is more than a bit unlikely that 
> screenreader developers would have the resources to accomplish 
> something that many others have not succeeded at.
>
> I know there is currently a French project attempting to put support 
> for several math codes, including Nemeth, into OpenOffice.  I don't 
> know how that is progressing.
>
> Please remember that the Nemeth code is a complete code; it has rules 
> for both text and math.  The rules for text are quite similar to the 
> rules for EBAE (English Braille American Edition) but have minor 
> changes as necessary for compatibility with the representation of 
> math.  The reason this fact is significant to this discussion is that 
> it is my understanding that screenreaders or display drivers that 
> convert to braille text in real time use the EBAE rules for text, not 
> the Nemeth rules.
>
> Officially here in the US, materials transcribed according to the 
> Nemeth code must consistently follow all the rules of that code.  
> However, my guess is that many Nemeth readers would be quite happy if 
> the text portions of an HTML document were to be transcribed according 
> to EBAE while MathML islands were transcribed  according to the Nemeth 
> rules for math.  I just noticed that the latest version of DBT 
> (Duxbury's software) has something similar in that it allows for 
> Nemeth math to be used with various non-English choices for text.
>
> There are others on this list who understand the details of MathType 
> and of screenreaders so this next may be a bit simplistic.  But it 
> seems to me that the minimum you'd want is for a screenreader to 
> simply recognize MathML islands and to at least have the capability to 
> pass their contents through a user-supplied filter on the way to the 
> display.
>
> Finally let me point out that there are some serious formatting issues 
> to be addressed when targetting braille math to a braille display. I'm 
> referring to both simple issues such as where to break a line and more 
> complex issues involving planar layouts.
>
> I'm hoping for feedback here.
>
> SusanJ
>
>
>