E-mail List Archives
Thread: MathML Usage for Styling
Number of posts in this thread: 5 (In chronological order)
From: Jamous, JP
Date: Wed, Jul 27 2016 12:32PM
Subject: MathML Usage for Styling
No previous message | Next message →
Hi folks,
I just joined this fine group and was wondering about WCAG's take on MathML for styling. W3 suggest in chapter 6 of HTML5 not to use MathML symbols for styling, but there is nothing related to that in WCAG.
http://www.w3.org/TR/MathML2/chapter6.html,
I have been advocating heavily against the usage of MathML symbols by my company's web developers and Copy Writers. However, nothing has been done about it. The issue is that those symbols are read by JAWS and VoiceOver. Some are ones I recognize from college days and others do not make any sense like "SINGLE LEFT-POINTING ANGLE QUOTATION MARK" and "SINGLE RIGHT-POINTING ANGLE QUOTATION MARK".
I was able to find those symbols in JAWS 17 in the following file:
C:ProgramDataFreedom ScientificJAWS17.0SETTINGSSymbolDescriptionsSymbolDescriptions.enu.txt
I turned them off by commenting the symbols that I don't care about, but that doesn't work for web users. Every time the MathML property "›" is used, it causes JAWS to read the link label followed by SINGLE RIGHT-POINTING ANGLE QUOTATION MARK. This can be so annoying and verbose considering the heavy usage that we have of it on the site.
Has any of you heard any updates to WCAG 2.0 in regards to MathML? Any feedback is highly appreciated.
From: Olaf Drümmer
Date: Thu, Jul 28 2016 1:31PM
Subject: Re: MathML Usage for Styling
← Previous message | Next message →
- there is not such a thing as MathML characters.
- MathML is a set of specific tags for mathematical content
- Unicode covers quite a number of mathematical symbols (several thousand I was told the other day by an expert during ICCHP 2016, my own research revealed at least around 2000…)
- accessibility can be cumbersome; if a sighted person sees that "SINGLE RIGHT-POINTING ANGLE QUOTATION MARK" - should a non-sighted or print-dsiabled or … person have access to the fact that it is a "SINGLE RIGHT-POINTING ANGLE QUOTATION MARK" character or not? In other words: shall everything be accessible or not?
- accessibility can be wonderful and burdensome in the exact same moment
Olaf
> On 27.07.2016, at 20:32, Jamous, JP < = EMAIL ADDRESS REMOVED = > wrote:
>
> Hi folks,
>
> I just joined this fine group and was wondering about WCAG's take on MathML for styling. W3 suggest in chapter 6 of HTML5 not to use MathML symbols for styling, but there is nothing related to that in WCAG.
> http://www.w3.org/TR/MathML2/chapter6.html,
>
> I have been advocating heavily against the usage of MathML symbols by my company's web developers and Copy Writers. However, nothing has been done about it. The issue is that those symbols are read by JAWS and VoiceOver. Some are ones I recognize from college days and others do not make any sense like "SINGLE LEFT-POINTING ANGLE QUOTATION MARK" and "SINGLE RIGHT-POINTING ANGLE QUOTATION MARK".
>
> I was able to find those symbols in JAWS 17 in the following file:
> C:ProgramDataFreedom ScientificJAWS17.0SETTINGSSymbolDescriptionsSymbolDescriptions.enu.txt
>
> I turned them off by commenting the symbols that I don't care about, but that doesn't work for web users. Every time the MathML property "›" is used, it causes JAWS to read the link label followed by SINGLE RIGHT-POINTING ANGLE QUOTATION MARK. This can be so annoying and verbose considering the heavy usage that we have of it on the site.
>
> Has any of you heard any updates to WCAG 2.0 in regards to MathML? Any feedback is highly appreciated.
> > > >
From: James A.
Date: Mon, Aug 01 2016 4:07AM
Subject: Re: MathML Usage for Styling
← Previous message | Next message →
The lack of standards and advice on best practice with how symbols are being styled/marked up is definitely an issue at the moment. While stand-alone equations maybe marked up using mathml (or similar) I have come across numerous examples where they are unintelligible to screen-reader/TTS users.
Below is a description of the experience of using NVDA with a paragraph from a scientific paper that I used to demonstrate the issue. This contains 3 equations within the text each marked up differently:
Equ 1: html - NVDA attempts to read this phonetically and it is not possible to perceive the contents correctly
Equ 2: mathml - NVDA first reads it phonetically then reads correctly using the mathml
Equ 3: image of equation with n alt-tag "View the MathML source" followed by a hidden container containing the MathML of the equation. NVDA reads the alt-tag followed by the equation read correctly using the mathml
In addition that paragraph contains <em>C<sub>j</sub> </em> which should be read as C subscript j but is reads as C j and <em>topN</em> ⩽ <em>N</em> which is read as top N N missing out the "less than or equal to" (screen readers seem to only read the few most common html entities).
The standalone equations are read correctly using a hidden class containing the mathml. I get a similar results when using Read&Write Gold except Equ 2 is not read aloud but this is read correctly if mathjax is turned on (although that stops NVDA reading the equations...).
This is just 1 paragraph on 1 paper...
I have conversations with STEM publishers who are keen to improve accessibility. They are in a catch-22 situation as the content they are publishing meets the current standards but they would like more guidance as you can see they are using many different approaches. The main culprit seems to be the use of html entities and mark-up when the content should be semantically tagged as maths.
Regards
Abi
=====================================Dr Abi James
Research Fellow
Accessibility Team, WAIS, ECS
University of Southampton
Web: https://access.ecs.soton.ac.uk/
Email: = EMAIL ADDRESS REMOVED =
From: Mike Warner
Date: Tue, Aug 02 2016 2:13PM
Subject: Re: MathML Usage for Styling
← Previous message | Next message →
Thanks for the test results, Abi. We're working on this, too. We aren't
using complex math yet, though, so we may end up with an aria-hidden div
with the math written normally and a hidden mathml div for screen readers.
The image and hidden div combo sounds like an interesting solution for when
we progress beyond simple things like fractions and the like, even though
students can't copy the math to paste it anywhere.
Mike Warner
Director of IT Services
MindEdge, Inc.
-----------------
Message: 5
Date: Mon, 1 Aug 2016 10:07:30 +0000
From: "James A." < = EMAIL ADDRESS REMOVED = >
To: Olaf Drümmer < = EMAIL ADDRESS REMOVED = >, "WebAIM Discussion List"
< = EMAIL ADDRESS REMOVED = >
Subject: Re: [WebAIM] MathML Usage for Styling
Message-ID:
< = EMAIL ADDRESS REMOVED = >
Content-Type: text/plain; charset="utf-8"
The lack of standards and advice on best practice with how symbols are
being styled/marked up is definitely an issue at the moment. While
stand-alone equations maybe marked up using mathml (or similar) I have come
across numerous examples where they are unintelligible to screen-reader/TTS
users.
Below is a description of the experience of using NVDA with a paragraph
from a scientific paper that I used to demonstrate the issue. This contains
3 equations within the text each marked up differently:
Equ 1: html - NVDA attempts to read this phonetically and it is not
possible to perceive the contents correctly
Equ 2: mathml - NVDA first reads it phonetically then reads correctly
using the mathml
Equ 3: image of equation with n alt-tag "View the MathML source"
followed by a hidden container containing the MathML of the equation. NVDA
reads the alt-tag followed by the equation read correctly using the mathml
In addition that paragraph contains <em>C<sub>j</sub> </em> which
should be read as C subscript j but is reads as C j and
<em>topN</em> ⩽ <em>N</em> which is read as top N N missing out
the "less than or equal to" (screen readers seem to only read the few most
common html entities).
The standalone equations are read correctly using a hidden class
containing the mathml. I get a similar results when using Read&Write Gold
except Equ 2 is not read aloud but this is read correctly if mathjax is
turned on (although that stops NVDA reading the equations...).
This is just 1 paragraph on 1 paper...
I have conversations with STEM publishers who are keen to improve
accessibility. They are in a catch-22 situation as the content they are
publishing meets the current standards but they would like more guidance as
you can see they are using many different approaches. The main culprit
seems to be the use of html entities and mark-up when the content should be
semantically tagged as maths.
Regards
Abi
===================================== Dr Abi James
Research Fellow
Accessibility Team, WAIS, ECS
University of Southampton
Web: https://access.ecs.soton.ac.uk/
Email: = EMAIL ADDRESS REMOVED =
Mike Warner
Director of IT Services
MindEdge, Inc.
From: Olaf Drümmer
Date: Tue, Aug 02 2016 2:24PM
Subject: Re: MathML Usage for Styling
← Previous message | No next message
Unless your target audience is screen reader users, hiding MathML such it will only be found by screen readers (or their users) could be considered questionable. People using screen readers are only a subset of those people with disabilities that may have to rely on assistive tooling.
Olaf
> On 02.08.2016, at 22:13, Mike Warner < = EMAIL ADDRESS REMOVED = > wrote:
>
> Thanks for the test results, Abi. We're working on this, too. We aren't
> using complex math yet, though, so we may end up with an aria-hidden div
> with the math written normally and a hidden mathml div for screen readers.
> The image and hidden div combo sounds like an interesting solution for when
> we progress beyond simple things like fractions and the like, even though
> students can't copy the math to paste it anywhere.