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

Re: Screen-reader behaviour

From: Sander Tekelenburg <st@isoc.nl>
Date: Fri, 31 Aug 2007 03:48:11 +0200
Message-Id: <p0624064ec2fd183a51d8@[192.168.0.101]>
To: <public-html@w3.org>, wai-xtech@w3.org

[Note: Reply-To set to <public-html@w3.org>, as suggested by Chaals, so as
not to flood <wai-xtech@w3.org>]

At 12:43 -0400 UTC, on 2007-08-30, Gregory J. Rosmaita wrote:

[... About Jaws 8]

> in virtual cursor mode, for example, i can advance by
> paragraph by pressing the p key -- and can move backwards using the
> SHIFT key in addition to the p key (note that in large documents, it
> is often impossible to actually move the virtual cursor back using the
> SHIFT modifier)

Without an indication that this is due to the nature of HTML, I can only
consider this a bug in Jaws (and thus out of scope for the HTML WG). Or maybe
you mean something else; "impossible" could mean anything ;) Does Jaws not
respond to Shift-p, does it jump to the wrong spot, does it crash? What is
"impossible"?

[...]

> from JAWS 8's help files [...]

My jaw dropped...

> C) the use of the INSERT key as a modifier key, means that commands
> which use UI chrome keys (such as "Enter Forms Mode using INSERT+F5)
> often leads the command being misinterpreted by the UA as a "refresh"

How can this be acceptable to users?

[...]

> D) the longer the document, the less likely the gross navigational
> commands are to work as intended -- one can often move forward
> using H to go to the next header, but shift+H won't always get you
> back to the previous header

Is this a literal quote from the Jaws documentation? This is pathetic. How
can such  behaviour be acceptable? In an eighth versionof an app even...

> E) it is often impossible to get contextual information by querying
> the element which has focus -- on a hyperlink, for example, INSERT+TAB
> will not read the title text if JAWS is set to read hyperlink text,
> and vice-versa, which is problematic, for many well intentioned
> hyperlink titles -- when spoken in place of hyperlink text -- make the
> flow of a document disorienting and hard to understand...

We need to know the reason for these limitations. Where are the Jaws
developers? If these limitations are due to HTML, we might be able to do
something useful. But they may just as well be due to Windows, IE, or just
lack of Jaws' developers skill.

[...]

> Display Element Information: SHIFT+INSERT+F1

What does "information" mean here?

[...]

> Create Custom Label: CTRL+INSERT+TAB

What does this do exactly?

[...]

> Refresh JAWS Virtual Mode: INSERT+ESC
>   * updates JAWS' screen shot of the document instance loaded into JAWS'
>      virtual mode buffer; it is often necessary to use this command when
>      switching from BrowserTab to BrowserTab, otherwise all the
>      information about the document instance you have just left will be
>      retained in JAWS' virtual memory (this is an example of the OSM in
>      action, the OSM being the "off-screen model" which JAWS scrapes
>      from the initial rendering of the page, so it can work off of the OSM
>      generated snapshot; when the page refreshes, the OSM does not
>      always refresh, nor does it always refresh when a user switches
>      between document instances, so it is one of the most invoked keys
>      in virtual mode

Why does the OSM "not always refresh"?

[...]

> Move to Next Division: Z

Does this refer to the <div> element?

[...]

> JAWS has a "virtual HTML features" list, containing 20 items  [...]
>
> note that there is neither a "class" list nor an "id" list

No, but in Jaws 4 already you could configure it to read or not read title
attributes. Does it not offer settings to do something useful with class
names by now?


Thanks, Gregory. This is useful. At the same time it raises a lot of
questions though. We need the developers of Jaws (and other such software) to
participate. Their silence is deafening :(


-- 
Sander Tekelenburg
The Web Repair Initiative: <http://webrepair.org/>
Received on Friday, 31 August 2007 01:53:17 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:48 UTC