a11y-discov-vocab
a11y-discov-vocab copied to clipboard
w3c
Definition update for describedMath
The current defintion for describedMath is:
4.2.2.5 describedMath Textual descriptions of math equations are included, whether in the alt attribute for image-based equations, using the alttext attribute for [MathML] equations, or by other means.
I believe we no longer recommend that MathML's alttext attribute is appropriate to be used, and that describedMath should only be present if there are images of math which includes alt text description.
@mattgarrish, @GeorgeKerscher @gregoriopellegrino Thoughts?
You can take out alttext but we've discussed the property previously and it isn't only applicable to images. There's value in describing complex mathml equations for users as mathml doesn't make all equations readable. Not everyone is using AT and not all AT can effectively read mathml markup.
I agree that alttext should be removed.
I am trying to think where somebody would add descriptions of MathML. In textbooks, this is part of the textbook that is providing the training. Will there be cases where MathML is followed by a link to an extended description? For example, "This is a matrix withfour lines with four sub-expressions on each line."
The description methods would likely be the same as for any other html content that needs describing. The daisy precedent for a math description is prodnote but that doesn't translate well.
What's the use case for having described math be applied to images of math, though?
It's kind of moving us back in that direction of describing the content instead of saying much new about the accessibility.
mathOnVisual would tell you there's visual math, and wcag conformance and/or alt text and extended descriptions tell you the images have alternative text and descriptions, so it seems kind of duplicative.
If describing equations independent of their markup isn't a big enough case to be covered by this property, maybe we should be looking at retiring it?
I see your point @mattgarrish not sure how many people would infer that when there is "mathOnVisual" (which 99% of the time the publish wont include) but has altText or longDescriptions folks would know that there is images of math which are described.
It may be somewhat duplicative but you would have to infer that assuming both mathOnVisual and altText / longDescriptions are present.
IMO, I think we should keep describedMath for the case of easily searching for accessible math content.
E.g. I can search for "MathML" or "describedMath"
But are the images really "described" if all they have is some basic alt text that tries to represent the equation in ascii text? What if the user doesn't use AT and can't access the alt text?
We're trying to fit the description into a use case where it may not belong, just to try and say that some math is images, which isn't an accessibility feature at all.
Well as an example Bookshare has scanned over 100K books in our collection using "PageAI" and created alt text descriptions of the math images found which would be the same text spoken using MathJax. making this accessible (although not ideal) to AT users.
Ultimately MathML will be the standard but in the interim if I am looking for accessible math if my reading system can deal with MathML then I will search for MathML, but if I am on a reading system that doesn't yet support mathML, I would love to be able to at least do a metadata search for describedMath, since this will at least give me something.
Anyways my 2cents.
Oh and BTW Bookshare does plan on injecting real MathML back into those books instead of or in addition to the images of math with alt descriptions once our own Reading systems fully support MathML.
Ya, but how is "described math" as alt text any different from putting in ascii text math instead of an image? We don't treat the latter specially, so what makes it necessary to point out when the only difference is that there's also an image (and you may not actually be able to access the "description")?
That's what's bothering me about this term. If it's not to provide an extended description of equations, independent of how they're encoded, then what are we defining beyond what we already know - that images have text alternatives?
Or to a flip it a bit, should there be 'describedCharts', 'describedChemistry', 'describedMusic', etc. to match all the other "onvisual" things that can have alt text and descriptions? (My answer is no, which is why I don't like this one case for math images.)
We really need to try and figure out this "describedMath" mess.
I'm looking at how to apply these for the epub guide, and longDescription mentions it's also for describing math:
Descriptions are provided for image-based visual content and/or complex structures such as tables, mathematics, diagrams, and charts.
(Aren't diagrams and charts always image-based visual content in practice?)
That makes sense since an image description is an image description regardless of what the image is of. So what is describedMath but a repetition of longDescription and alternativeText for calling out a specific type of content?
Even say we try to limit "describedMath" to images that are described, do you have to set all three features or are regular image descriptions and alt text exclusive of those used for math?
And to pile on the ambiguities, if latex is used as the alternative text, do you say the publication has alternativeText, describedMath, latex, or all three?