How do screen readers read HTML entities?
Home / Site Building & Testing / How do screen readers read HTML entities?
How would JAWS, Window-Eyes, and Home Page Reader read the following line?
<p>a − b > c.</p>
I assume it reads > as "greater than", but is the − read as "minus"?
In general, how do the screen readers read entities such as: °, ×, ÷, ±, ½, ². I have a maths site that uses these entities.
_________________
Nick
Share: 
I've not managed to test this much yet, but I'd presume that JAWS is pretty good with these sorts of entities. My test with Window-Eyes doesn't read any of them so your equation sounds like "uh bee see".
I'll get some more tests done.
_________________
Jon Gibbins :// blogs at dotjay.co.uk, works with Analog.
Last edited by dotjay on 20 Feb 2008 10:12 pm; edited 1 time in total
Share:
Thanks Jon. That's not good news about Window-Eyes. I'm surprised it didn't even read the "greater than" entity.
Since posting the above, I've obtained a copy of IBM Home Page Reader. HPR reads all of the above-mentioned entities perfectly, and even reads the decimal numeric equivalent of − -- ampersand hash 8722; -- correctly.
I'd be interested to know whether JAWS can read −, ampersand hash 8722;, and the other entities correctly.
Of course, this is just the minutae of making maths accessible. More daunting is the whole question of how to mark up mathematical expressions, which may contain many terms, with complicated nested parentheses, and specialized symbols, such as the integral sign, or matrices. Text is linear in nature, while mathematical equations are two dimensional. Ironically, despite its origins in the scientific community, HTML was not designed to mark up mathematical content, and the current generation of screen readers is not designed to render it. I'd like to use MathML, but until IE supports it "out of the box" that's impractical.
_________________
Nick
Share:
| coplanar wrote: |
| Thanks Jon. That's not good news about Window-Eyes. I'm surprised it didn't even read the "greater than" entity. |
Me too - but I have to admit that I don't often use Window-Eyes. My JAWS install is currently stuck in 40-min mode, so no testing in that for me at the moment.
| coplanar wrote: |
| HPR reads all of the above-mentioned entities perfectly, and even reads the decimal numeric equivalent of − -- ampersand hash 8722; -- correctly. |
That's useful to know. I've just downloaded HPR myself, so will try a few more tests later.
| coplanar wrote: |
| I'd be interested to know whether JAWS can read −, ampersand hash 8722;, and the other entities correctly. |
I'll hazard a guess and say that I think it does, but I'll test as soon as JAWS is back online for me.
| coplanar wrote: |
| More daunting is the whole question of how to mark up mathematical expressions, which may contain many terms, with complicated nested parentheses, and specialized symbols, such as the integral sign, or matrices. |
Have you thought of using MathML to mark up and then transforming it until browsers support it more?
_________________
Jon Gibbins :// blogs at dotjay.co.uk, works with Analog.
Share:
| coplanar wrote: |
|
Of course, this is just the minutae of making maths accessible. More daunting is the whole question of how to mark up mathematical expressions, which may contain many terms, with complicated nested parentheses, and specialized symbols, such as the integral sign, or matrices. Text is linear in nature, while mathematical equations are two dimensional. Ironically, despite its origins in the scientific community, HTML was not designed to mark up mathematical content, and the current generation of screen readers is not designed to render it. I'd like to use MathML, but until IE supports it "out of the box" that's impractical. |
There's also the question of how to make charts, graphs and plots of scientific data accessible too. Longdesc might work for that, but some charts can contain an awful lot of information eg. the stellar spectra from the Sloan Digital Sky Survey on this tutorial: http://outreach.atnf.csiro.au/...
_________________
Jim O'Donnell
work: Royal Observatory Greenwich
play: eatyourgreens
Share:
| dotjay wrote: |
| Have you thought of using MathML to mark up and then transforming it until browsers support it more? |
I hadn't thought of that, to be honest! Do you mean using XSLT? Is this possible even if the target browser does not have a plug-in (such as MathPlayer or Techexplorer) installed? Can the transformation be done on the server?
_________________
Nick
Share:
Yup - it's not something I've done much with, but you should be able to provide the equations in MathML, but transform them on the server-side for the meantime. With a bit of sniffing, you could even serve XHTML with MathML for those browsers that support it.
Or, this (admittedly a bit old) page seems to have a solution that uses XSLT stylesheets to transform MathML:
http://www.w3.org/...
_________________
Jon Gibbins :// blogs at dotjay.co.uk, works with Analog.
Share:
| eatyourgreens wrote: |
|
There's also the question of how to make charts, graphs and plots of scientific data accessible too. Longdesc might work for that, but some charts can contain an awful lot of information eg. the stellar spectra from the Sloan Digital Sky Survey on this tutorial: http://outreach.atnf.csiro.au/... |
Since a chart is just a graphical representation of tabular data it would seem to me that a way to make complex charts accessible is to use a data table.
_________________
Jonathan Worent
The early bird gets the worm, but the second mouse gets the cheese.
Share:
| jw_developer wrote: | ||
Since a chart is just a graphical representation of tabular data it would seem to me that a way to make complex charts accessible is to use a data table. |
That might work well for simple charts, but it would be difficult for complex data. Have a look at the stellar spectra in the tutorial that I linked too. Here's a link to one of them:
http://outreach.atnf.csiro.au/...
There's quite a lot going on there - you couldn't convey it simply by listing the hundreds or thousands of data points in the spectrum.
Scatter plots would also be tricky, where you want to show a correlation between two factors to within a certain degree of error. Someone reading the chart needs to be able to agree or disagree with your conclusions based on your experimental data so a simple "this shows x is proportional to y to within a 90% level of confidence" glosses over quite a lot. Listing all the points in a table, on the other hand, doesn't really convey the information conveyed in the diagram. A table gives you the properties of the individual points. A diagram gives you a view of the dataset as a whole and the relationship between the points. Admittedly, a table would allow you to put the points into your own analysis software, which could be useful, and I would be suspicious of any experiment which didn't publish the data such that you could perform an independent analysis.
Just playing devils advocate, and pointing out that alt text for maths is difficult.
_________________
Jim O'Donnell
work: Royal Observatory Greenwich
play: eatyourgreens
Share:
| dotjay wrote: |
| I've not manged to test this much yet, but I'd presume that JAWS is pretty good with these sorts of entities. My test with Window-Eyes doesn't read any of them so your equation sounds like "uh bee see".
I'll get some more tests done. |
I've run some tests with JAWS, Home Page Reader and Window-Eyes. My results can be found here on my site:
Screen Readers and HTML Character Entities
As I suspected, JAWS isn't all that bad, but doesn't know the entity for minus, so the best you can do I guess is use a dash I suppose.
Also, I think there's a newer version of Window-Eyes that might do better than the version I tested with - perhaps I'll see at some point.
_________________
Jon Gibbins :// blogs at dotjay.co.uk, works with Analog.
Share:
All of the result links produce 404 errors._________________
My CV type thing and my Life of Ben (Blog). Nigel Peck's Accessify Forum Requirements.
Share:
| Cerbera wrote: |
| All of the result links produce 404 errors. |
Cheers, Ben. Missed a hash symbol - the results are on the page.
_________________
Jon Gibbins :// blogs at dotjay.co.uk, works with Analog.
Share:
I hadn't scrolled that far down the page; I've noticed the table now. 
Its's a pretty thorough study which shows that there seem to be two schools of thought amongst screen reader implementations:
- Read out the name of the entity.
- Don't read out the entity at all.
_________________
My CV type thing and my Life of Ben (Blog). Nigel Peck's Accessify Forum Requirements.
Share:
Whether the source document had a character represented as an entity reference, or a character reference, or the character directly, shouldn't have any affect on how the application treats the character. If it does, the application is probably broken.
That is, the following snippets of HTML are exactly equivalent and should be treated exactly the same by the application:
2 − 1
2 − 1
2 − 1
2 − 1
_________________
Simon Pieters
Share:
I assume the transportation layer would just send the Unicode character, which the screen reader would then decide how to pronounce?_________________
My CV type thing and my Life of Ben (Blog). Nigel Peck's Accessify Forum Requirements.
Share:
All times are GMT
You cannot post new topics in this forumYou cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum