ALT for text?

[Dan Jacobson]

On my computer
/usr/share/doc/w3-recs/RECS/html4/struct/objects.html#adef-alt
says one can use ALT= for images, but how about for text: if they
can't see the Chinese character, give them the romanization of it.
Or must one produce a .GIF to enjoy this, instead of putting the raw
character (2 or 3 bytes) into the page?

[Alan J. Flavell]

On Fri, 10 Feb 2006, Dan Jacobson wrote:

On my computer
/usr/share/doc/w3-recs/RECS/html4/struct/objects.html#adef-alt
says one can use ALT= for images, but how about for text:
That's easy: ALT= is for *alternative* text, when the images can't be
seen. If you want to provide *supplementary* information, that's what
the title= attribute is for.
if they can't see the Chinese character,
Well, we can *see* them, even if we can't read them. Assuming we've
installed an adequate range of fonts, anyway.
give them the romanization of it.
<span lang="whatever" title="some romanised text">some chinese
text</span>

You'd probably want to style it somehow, to alert the reader to the
possibility of supplementary information.

Not that it would make much difference! I remember two colleagues
whose spoken Chinese was mutually unintelligible, holding a
conversation by writing Chinese on the table napkins at lunch.
Or must one produce a .GIF to enjoy this, instead of putting the raw
character (2 or 3 bytes) into the page?

Oh no, at least not in theory; but the average browser's provision for
viewing the title= value could be rated as less than ideal for this
situation.

good luck

--

Most folks would think a Referer header is something you smoke.
-- Bruce Tomlin in a.s.r

[Jukka K. Korpela]

Alan J. Flavell wrote:

if they can't see the Chinese character,
Well, we can *see* them, even if we can't read them.

Or maybe not. We might be using a speech synthesizer or a Braille device
that does not grok Chinese characters but could render a transcription
in Latin letters at least by presenting the letters. Or...
Assuming we've
installed an adequate range of fonts, anyway.
The assumption often fails. Maybe the system just hasn't got a font with
Chinese characters. Besides, no existing font is sufficient for
rendering _all_ Chinese characters. Even if someone created a font with
all characters that exist in Unicode at present, they keep adding new
characters to Unicode.
<span lang="whatever" title="some romanised text">some chinese
text</span>

You'd probably want to style it somehow, to alert the reader to the
possibility of supplementary information.

Or maybe even explicitly describe this feature, even though it means
discussing the techniques - without knowing whether they work for the
user, so the text should be sufficiently conditional.

It occurred to me that we might use Ruby markup in cases like this. When
supported, the ruby annotation would appear above the annotated text,
and you could additionally use title="..." (because the annotation
appears by default in so small print that it is difficult to read).

Ruby markup is supported to a limited degree by IE 6. See some tests at
http://www.cs.tut.fi/~jkorpela/html/ruby.html
(I didn't find a way to make the annotation text appear closer to the
annotated text. The effect on line length is now rather disturbing.)

[Andreas Prilop]

On Thu, 9 Feb 2006, Alan J. Flavell wrote:

<span lang="whatever" title="some romanised text">some chinese
text</span>
You'd probably want to style it somehow, to alert the reader to the
possibility of supplementary information.

Or write DFN instead of semantic-free SPAN since romanized text
may be considered some sort of definition.

--
All free men, wherever they may live, are citizens of Denmark.
And therefore, as a free man, I take pride in the words "Jeg er dansker!"

[Alan J. Flavell]

On Fri, 10 Feb 2006, Jukka K. Korpela wrote:

Assuming we've installed an adequate range of fonts, anyway.
The assumption often fails. Maybe the system just hasn't got a font
with Chinese characters. Besides, no existing font is sufficient for
rendering _all_ Chinese characters.

I noticed, when I'm using linux Mozilla, that when it hasn't got a
glyph to display, it displays a box with the Unicode character number
in it (which could then be looked-up in the Unicode specification...)
It occurred to me that we might use Ruby markup in cases like this.

Right - it does seem to be meant for just this kind of purpose.

http://www.w3.org/International/tutorials/ruby/

(If anyone's confused about this - this is *not* about the
program scripting language called Ruby).

I've made no secret of the fact that I know very little about CJK
writing systems, so I hadn't looked much at ruby. But prompted by
your example, I've taken a look and it does seem rather good.

(Are we allowed to use this to annotate English-language document with
English-language annotations? Just a throwaway thought...)

As they rightly say, it's not valid in HTML/4 nor XHTML/1.0, so it's
stretching things a bit to include it in current documents; but it's
been designed for graceful fallback, and seems indeed to behave well,
despite making the document fail validation.

I suppose you wouldn't care to make it technically valid by using
XHTML/1.1 (and, for obvious reasons, deciding not to follow the W3C's
"SHOULD" about not serving it as text/html) ?

If I try to view the W3C's XHTML/1.1 test
http://www.w3.org/International/test...-markup-1.html
in IE6, I get an alert telling me "Your current security settings do
not allow this file to be downloaded." ;-)

I couldn't resist trying the Mozilla extension mentioned at W3C,
namely http://piro.sakura.ne.jp/xul/_rubysupport.html.en

And indeed it seems to work with your document.

cheers

[Harlan Messinger]

Jukka K. Korpela wrote:

Alan J. Flavell wrote:

if they can't see the Chinese character,
Well, we can *see* them, even if we can't read them.

Or maybe not. We might be using a speech synthesizer or a Braille device
that does not grok Chinese characters but could render a transcription
in Latin letters at least by presenting the letters. Or...

Isn't the bottom line that the browser sees text data, subjects it to
whatever algorithm it uses for displaying such, and ultimately doesn't
really know whether what it's showing the user is legible text or not?
If that's correct, then the browser won't know when to show alternative
content.

Assuming we've installed an adequate range of fonts, anyway.

The assumption often fails. Maybe the system just hasn't got a font with
Chinese characters. Besides, no existing font is sufficient for
rendering _all_ Chinese characters.

I'm picturing a situation where a browser would render those Chinese
characters that it could AND would show pinyin for those characters that
it couldn't render. What a mess!

[Alan J. Flavell]

On Fri, 10 Feb 2006, Alan J. Flavell wrote:

If I try to view the W3C's XHTML/1.1 test
http://www.w3.org/International/test...-markup-1.html [...]
I couldn't resist trying the Mozilla extension mentioned at W3C,
namely http://piro.sakura.ne.jp/xul/_rubysupport.html.en

Just a word of warning about this - it seemed to have wrecked
Mozilla's PDF plugin. Every subsequent attempt to view a PDF file had
got me a hung instance of Acrobat Reader, and had also hung all my
Mozilla windows until I brutally killed the Acrobat Reader process.

After de-installing the Ruby plug-in, PDF viewing is working again
without problems. Well, the author /does/ have a disclaimer,
http://piro.sakura.ne.jp/xul/_rubysu...ml.en#problems ,
so I've no cause to grumble, but I thought that - having posted
the above mention - I'd better add this caveat.

cheers