[BlindMath] JAWS appears to have broken electronic math accessibility

Wed Jun 9 17:24:53 UTC 2021

Hey there,

hmm, I did not use Jaws seriously for more than year, so I don't quite
remember the details, but wasn't this always the way Jaws shaped its
MathML presentation engine?

I mean, the function was added in Jaws 17? Or 16? Not sure right now,
but I have feeling it was always like this. May be except for brackets,
here I'm not really sure, but I think the rest did not change, feel free
to correct me if I'm wrong.

i personally like the Jaws approach. I don't know MathSpeak or this Dr.
Nemeth, but when I'm reading your examples of how it's supposed to
sound, I feel like it would be one of those tools I'd throw out of
window under the first circumstance. :)

Like, it's really nice to have formally correct language, but what's
really the point if it makes things hard to read?

Unpractical technology is almost like no technology at all.

In your examples, most of the time i hear that something starts or ends,
with the really relevant information being overwhelmed by this
positional stuff.

Not even speaking about the lenght of the resulting message.

On the other hand, with the Jaws approach, I can focus on the relevant
parts without being disturbed by details. Like yes, these details are
pretty important, but I'm going to check out the formula in tree mode
anyway, so what's not spoken will be revealed sooner or later.

And as long as I know the layout, I no longer need to hear it, my brain
automatically maps relevant parts to their appropriate positions.

In other words, at least in my opinion, speaking math notation should be
about... well... speaking it, like you'd normally read it to a sighted
audience, not about data serialization.

Serialization is for computers, speech is for humans.

But thanks anyway for sharing. I'm working on math accessibility for
Linux, and i should definitely give this a second thought.

Best regards

Rastislav

Dňa 9. 6. 2021 o 17:45 Brian Richwine via BlindMath napísal(a):
> Hi,
>
> Does anyone have any insight into what Freedom Scientific is doing in
> developing MathML support in JAWS?
>
> It used to be that we could create accessible HTML math content using
> MathJax and have it reliably spoken accurately and unambiguously by
> screen-reading software. However, JAWS and its apparently flawed MathML
> support have made accessible electronic math unreliable and thus unusable.
>
> The default MathJax settings add visually hidden MathML into the page for
> AT that can read it (not the speech text that MathJax can provide). So, now
> that JAWS “speaks” MathML JAWS users end up getting the speech text that
> JAWS generates instead of what MathJax generates. Unfortunately, it appears
> that the speech text for MathML generated by JAWS is very flawed and is
> incapable of accurately and unambiguously speaking even basic algebraic
> constructions.
>
> The following link contains pairs of simple distinct MathML equations as
> examples of basic math that JAWS does not read unambiguously (sighted users
> can view it in Firefox to see the MathML rendered):
> https://brichwin.pages.iu.edu/tests/jaws_Reading_MathML_ambiguously-2021-06-07.html
> I have included the JAWS speech history and the MathSpeak Grammar verbose
> speech text generated by MathJax for each example. You will notice that
> JAWS fails to read basic mathematical relationships distinctly
> (unambiguously) whereas MathJax does generate unambiguous and accurate
> speech text.
>
> We have filed several issues on the Freedom Scientific / Vispero Standards
> Support github:
>
> ·         JAWS is not reading MathML in a clear, unambiguous manner:
> https://github.com/FreedomScientific/VFO-standards-support/issues/539
>
> ·         JAWS No Longer Announcing the Plus Operator when reading MathML
> with Verbosity of Punctuation=None:
> https://github.com/FreedomScientific/VFO-standards-support/issues/538
>
> ·         MathML equations at beginning of list items are not read
> correctly:
> https://github.com/FreedomScientific/VFO-standards-support/issues/503
>
> I am concerned because:
>
> ·         Non-visual users will not know that JAWS is incorrectly speaking
> mathematics.
>
> ·         People believe putting MathJax, KaTeX, or straight MathML on an
> HTML page makes the mathematics accessible.
>
> o   Increasingly electronic mathematics content is being created and relied
> upon. Especially in the education arena.  Students will be disadvantaged.
>
> o   I am beginning to hear people blaming MathJax for the JAWS failings.
>
> ·         JAWS appears to render its own flavor of speech text vs. using an
> established mature and common grammar (Mathspeak, Clearspeak, etc.). If
> this is the case, the different speech rendering could confuse listeners
> and add unnecessary cognitive burden as a barrier to comprehension and
> learning.
>
> ·         I cannot find any JAWS settings to disable or otherwise configure
> the JAWS MathML support. Typical configuration options would include
> selecting a speech grammar, the grammar’s verbosity level, explicivity,
> semantic interpretation, etc.
>
> Does anyone know how to disable the JAWS MathML support? It would be great
> if we could turn it off until they make it reliable. Can we get Freedom
> Scientific / Vispero to treat MathSupport as a beta feature that is off by
> default until it is robust enough not to disadvantage non-visual users?
>
> Very interested to hear the thoughts of others!
>
>     Brian
> _______________________________________________
> BlindMath mailing list
> BlindMath at nfbnet.org
> http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
> To unsubscribe, change your list options or get your account info for BlindMath:
> http://nfbnet.org/mailman/options/blindmath_nfbnet.org/rastislav.kish%40protonmail.com
> BlindMath Gems can be found at <http://www.blindscience.org/blindmath-gems-home>