W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 2002

Fwd: RE: update on hebrew and arabic

From: Martin Duerst <duerst@w3.org>
Date: Wed, 04 Sep 2002 14:14:40 +0900
Message-Id: <4.2.0.58.J.20020904140838.03d57528@localhost>
To: w3c-wai-gl@w3.org, w3c-i18n-ig@w3.org

Dear WAI and I18N groups,

I'm forwarding this from some internal discussion:

>From: Martin Duerst <duerst@w3.org>

>Lisa Seeman has come up with some very good leads, in particular
>http://www.cs.technion.ac.il/~erelsgl/bxi/hmntx/teud_tokna.html#English,
>where the software seems to be open source.
>
>Rates such as 95% sound quite good; that may be about the rate
>at which the results are 'bearable' (unfortunately,
>many people in the accessibility area are used to 'bear' one
>or the other problem). Ordinary Hebrew text sometimes uses vowel
>points for disambiguation; it would be interesting to know
>whether these cases correspond in any way to the cases where
>the programs have difficulties.
>
>Another interesting thing to investigate would be to
>see what Hebrew/Arabic Braille do. If they use some
>equivalents of diacritic vowels, then this means that
>we have the same problem with braille transcription.
>If they don't, then this means that many blind people
>may be used to figure out the vowels themselves, which
>may mean that lower correctness rates, or even just
>reading with dummy vowels (like classic Egyptian texts :-)
>could work.
>
>Please note that English also has ambiguities (read/read),
>and similar for many other languages. How would English be
>handled in the present guidelines?
>
>As for Arabic having many more possibilities and therefore
>being more difficult, I think there is a misunderstanding.
>The number of diacritic vowels is much smaller in Arabic
>(3) than in Hebrew (about 10). There are other ways to
>create more different word forms (pre/suffixes, long
>vowels that are written like consonants), and that may
>have been counted. I think the problem for Arabic should
>be similar in complexity, or even easier.
>
>Earlier, we have considered the following solutions:
>
>1) Pushing fully vocalized texts.
>2) Asking the Unicode consortium for a second series
>    of 'hidden' vowel signs, not displayed usually.
>3) Pushing annotations (<ruby>, something like alt,...)
>4) Using technology, and hoping that it is advanced
>    enough and will advance further.
>
>Given Lisa's findings, and the problems we would have
>with getting any of 1) to 3) accepted in practice,
>things seem to point to 4). If the new guidelines
>can be written in a way that encourages both research
>and business to follow that direction, that would
>probably be best.
>
>Regards,    Martin.
Received on Wednesday, 4 September 2002 04:57:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:20 GMT