Re: phonetic markup

I think I've got it. 

While we could fake it with SAMPA text in the 'title' attribute
of a SPAN element and use the 'class' attribute to annotate that
this was SAMPA, that would be a stretch.  OBJECT offers us a
better solution.

The 'data' attribute on an OBJECT admits either inline data or an
URL.  So you could inline SAMPA text as the value of OBJECT.data
and put the plain text of the material the SAMPA pronounces as
the content of the OBJECT.

The OBJECT element lets you give 'type' and 'lang' qualifiers for
these 'data.'

The trick is to get browsers to recognize that if they can't
translate the SAMPA to sound, they must consider the OBJECT.data
unimplementable, and so they will present the text content of the
object instead.  This must be done with browser behavior rules
keyed off 'type' and/or 'lang.'

We need to research the MIME registration status of SAMPA.  It
could be registered as a type text/sampa or as a language
i-sampa.

That way a pure-audio browser, or the audio channel of a
multimedia browser, can elect to vocalize the phonetics and the
visual presentation can ignore the phonetics and present the
text.

-- Al

Received on Saturday, 22 November 1997 23:48:26 UTC