Screen Readers and Abbreviations and Screen Reader Support for CSS 2 Aural Style Sheets / CSS 3 Speech Module

Last updated: November 17, 2008

Notes

This page tests how screen readers speak abbreviations (including acronyms) and goes on to test support for the 'aural' and 'speech' CSS media types. The tests follow a related discussion on the GAWDS discuss mailing list about how the abbreviation WHF might be read out by assistive technology software and techniques to overcome problems that arise.

NB: A full table of results is to be added. Tests will cover different versions of screen readers, with and without abbreviations set to be spoken in expanded form.

For more on aural CSS, see: Aural CSS: Support for CSS 2 Aural Style Sheets / CSS 3 Speech Module.

JAWS Quirks

• JAWS 7.10 does not read abbreviation expansions on the Web despite what is set in the verbosity settings or HTML options.
• JAWS (at least, version 8.0) does not read abbreviation expansions on the Web if you select text with the mouse, again, despite what is set in the verbosity settings or HTML options.

For further information, see my notes on screen reader quirks.

Tests

Try JAWS in normal mode and in title attribute reading mode for the following tests.

Tests 1 to 3 test basic handling of abbr and acronym. Tests 4 and 5 attempt a couple of solutions to fix how problematic abbreviations are spoken by screen readers. Tests 6 to 11 test different approaches using CSS to control "conventional" markup. Tests 6 to 11 also include additional tests which make checking the application of CSS rules easier by swapping how they are spoken.

Test 1: No markup

1. The WHF test
2. The W H F test
3. The W.H.F. test
4. The W. H. F. test
5. The laser test

Test 2: Introducing the abbr and acronym elements

1. The WHF test
2. The W H F test
3. The W.H.F. test
4. The W. H. F. test
5. The laser test

Test 3: Introducing expansions in the title attribute

1. The WHF test
2. The W H F test
3. The W.H.F. test
4. The W. H. F. test
5. The laser test

Test 4: Attempting to remove unsightly bits

1. The W H F test
2. The W.H.F. test
3. The W. H. F. test

Test 5: Another attempt to remove unsightly bits

The W H F test.

Test 6: default

1. The WHF test
2. The laser test

Test 7: aural CSS rules without media type being set; acronyms to be spoken as words and all else one letter at a time (should be same as default, but in case it is not)

1. The WHF test
2. The laser test
3. [SWAP] The WHF test
4. [SWAP] The laser test

Test 8: CSS 2 'aural' media type in head

1. The WHF test
2. The laser test
3. [SWAP] The WHF test
4. [SWAP] The laser test

Test 9: CSS 3 'speech' media type in head

1. The WHF test
2. The laser test
3. [SWAP] The WHF test
4. [SWAP] The laser test

Test 10: CSS 2 'aural' media type in external style sheet

1. The WHF test
2. The laser test
3. [SWAP] The WHF test
4. [SWAP] The laser test

Test 11: CSS 3 'speech' media type in external style sheet

1. The WHF test
2. The laser test
3. [SWAP] The WHF test
4. [SWAP] The laser test

Test 12: the Mike Cherim technique (currently under test)

1. IE

Results & Observations

Results draw only basic conclusions at the moment. Tests have been carried out with JAWS 7.10 only. Other screen readers (and perhaps other assistive technology software) to be tested.

JAWS 7.10

Running these tests using JAWS 7.10 with Firefox 1.5 and Internet Explorer 6 confirms that neither the 'aural' or the 'speech' media type has any effect. These tests include CSS applied without a specified media type, CSS included in external CSS files with media types specified and CSS applied in the document head using the media at-rule. In fact, at this time, the only known support for CSS 2 Aural Style Sheets is in the Emacspeak audio desktop and the CSS 3 Speech module is only implemented by Opera.

JAWS analyses words to determine whether they should be pronounced as a word or spoken one letter at a time. Older versions of JAWS may vary, but in these test cases, JAWS 7.10 would read WHF one letter at a time, and laser was pronounced as a word.

For problematical abbreviations (that are, for example, incorrectly spoken as a word by a screen reader), the solutions in Test 4 and 5 seem to work reasonably well. The main thing to bear in mind is that Internet Explorer 6 refuses to recognise the abbr element and will not apply your CSS. Also, it is not always correct to use acronym as not all abbreviations are acronyms. It is also worth remembering that JAWS users can get any word spoken one letter at a time using the Spell Word command (Insert+Num Pad 5 pressed twice quickly) or the left and right cursor keys.

Test 3 observations

JAWS 7.10 behaves quite well through each part of this test. There is some variation when dots are introduced.

1. WHF is read one letter at a time
2. W H F is read one letter at a time
3. W.H.F. is read one letter at a time, but with JAWS saying “dot” between each letter and the final dot being treated as a full stop
4. W. H. F. is read one letter at a time, but with longer pauses between letters as each dot is treated as a full stop

Test 4 observations

This test, where supplementary dots are introduced but hidden off-screen using CSS, produces interesting results.

1. W H F is read one letter at a time
2. W.H.F. is read one letter at a time, but with longer pauses between letters and without the dots
3. W. H. F. is read one letter at a time, but with longer pauses between letters and without the dots

Test 5 observations

W H F is read one letter at a time.

Safari 3 with VoiceOver

Some initial testing using Safari 3.1.2 with VoiceOver indicates that the speak property is not supported.

Useful References

W3C Recommendations:

• CSS 3 Speech Module (latest version)
• CSS 2, 19 Aural style sheets
• CSS 2.1, Appendix A. Aural style sheets

Related on dotjay.co.uk:

• Aural CSS: Support for CSS 2 Aural Style Sheets / CSS 3 Speech Module
• Screen readers and abbreviations (February 2008)