- From: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
- Date: Thu, 19 Jan 2012 13:45:06 +0100
- To: Martin J. Dürst <duerst@it.aoyama.ac.jp>
- Cc: fantasai <fantasai.lists@inkedblade.net>, public-i18n-cjk@w3.org, 'WWW International' <www-international@w3.org>
"Martin J. Dürst", Thu, 19 Jan 2012 18:00:32 +0900: > On 2012/01/19 16:05, Leif Halvard Silli wrote: >> "Martin J. Dürst", Thu, 19 Jan 2012 10:50:52 +0900: >> <ruby> >> <rb>W</rb><rp>[</rp><rt>World</rt><rp>]</rp> >> <rb>W</rb><rp>[</rp><rt>Wide</rt><rp>]</rp> >> <rb>W</rb><rp>[</rp><rt>Web</rt><rp>]</rp> >> </ruby> >> >> A screen reader would read the above as: W - World, W - Wide, W - Web. > > Why? A really dumb one might read the parentheses, too. Not really. Screenreaders *should* not read elements which are set to {display:none;} - which is the case for <rt>. However there apparently are some readers that, at least previously, need that you, in addition, use {visibility:hidden;} - may be those can be described as 'really dumb' ... > A more > intelligent might read only WWW or only World Wide Web, with an > acoustic indication that there's something else that can be listened > too. At least NVDA and VoiceOver do not [unless they 'special cases' East Asian text - I did not test]. However, HTML5 includes ARIA attributes, and in my change proposal to include rb, I described how it could be done: <http://www.w3.org/html/wg/wiki/IncludeRB#Use_cases_for_rb>. The demo should work in any browser that supports aria-hidden='true', such as VoiceOver: <http://malform.no/messages/issue-172/aria.html>. [But using aria-hidden='true' might not be the optimal solution.] >> So it seems to me that Ruby Annotation has some 'built-in' thoughts >> about what's a word and what not. > > At least for Japanese screen readers, it would be very important that > they read the ruby, not the base text. That's basically what I tried to do in my demo above. Btw: HTML5 has not defined any specific ARIA roles - or any other ARIA features - for <ruby>, <rb>, <rt> or <rp>. It seems to me that, in order to get accessibility technology to read ruby like you suggest above, then HTML5 should define some ARIA roles/featuers for these elements. I think bug should be filed about that. > There are definitely > exceptions, but in roughly 90% or more of the cases, ruby indicate > pronunciation. Even the above example can be seen this way. > > I guess that indicating which parts need to be spoken could be > indicated quite easily by an audio stylesheet (and if necessary > classes to distinguish pronunciation guides from e.g. additional text > to make a point or jokes). The screen reader could then pick up from > there. Audio stylesheets [@media speech] are not too well supported - the focus appears to be on other accessibility features. > I'm not sure to what extent that screen readers do this currently, > though. I had a Japanese IBM Home Page Reader working for an older > machine, but it doesn't work any more on the current machine, and I > didn't get aroud yet to install a new one. The open source NVDA client is really lightweight and very simple to install in Windows XP and above [a bit unlike e.g. the demo version of Jaws, which is a bit more involved to use]. However -- Leif H Silli
Received on Thursday, 19 January 2012 12:45:58 UTC