W3C home > Mailing lists > Public > public-html@w3.org > June 2007

Re: More on <CAPTION> element etc

From: Sander Tekelenburg <st@isoc.nl>
Date: Fri, 22 Jun 2007 13:03:05 +0200
Message-Id: <p06240631c2a15c966be1@[]>
To: public-html@w3.org

At 19:54 +0100 UTC, on 2007-06-21, Joshue O Connor wrote:

> Sander Tekelenburg wrote:
>> My impression is that the explanation is in the name, "screen reader".
>> you'd think a talking browser would act as you say, but it seems that it
>>is a
>> relatively recent thing that screen readers look at the actual HTML [...]
> A screen reader is not talking browser

Yes, that's exactly the point I was trying to make.

Btw, earlier you said that you found that when a caption is authored below a
table, it isn't presented before it. It would still be useful to know which
tool(s) you have that experience with.


At 22:42 +0100 UTC, on 2007-06-21, Joshue O Connor wrote:

> Maurice Carey wrote:
>> Does anyone know, when a screen reader encounters a longdesc, does it
>> navigate to the long description or does it pull it up behind the scenes and
>> read it out loud in context?
> No, it doesn't do either.

Exactly which screen reader are you referring to here?


>> To provide a more intelligent presentation of content, such software can
>> receive some interpretation of the HTML from the host OS, which it in turn
>> gets from the GUA.
> Most Screen readers use the Off Screen Model (OSM)

The term OSM appears to refer to a data storage model used by screen readers.
I was referring to where that data has to come from -- a step earlier in the
process. See <http://en.wikipedia.org/wiki/Screen_reader#Off-screen_models>.

> apart from Dolphins
> Supernova which interacts directly with the DOM.

That seems a much better approach. But as I understand it, Supernova still
needs a GUI browser to provide it with that DOM. So there's still a layer
inbetween, where data can get lost.


> Again most use the OSM. A screen reader will also work in various modes
> and for example only interact directly with the web page itself in what
> JAWS refers to as 'Forms mode'. This is to navigate forms, enter text
> into form fields etc.

Exactly in what sense does it 'interact' with the HTML itself? AFAIK Jaws
still needs IE or FIrefox.

> It is usually in virtual mode, which is the screen
> reader using the OSM, which is in effect a virtualisation of the HTML
> page the author creates. The better the HTML, the better the virtualisation.

But that 'virtualised HTML' still has to come from a GUI browser which is
outputting for an entirely different envronment -- the greatest HTML can
easily get mangled to something quite poor there.

Sander Tekelenburg
The Web Repair Initiative: <http://webrepair.org/>
Received on Friday, 22 June 2007 11:06:30 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:22 UTC