WebAIM - Web Accessibility In Mind

E-mail List Archives

RE: accesskey assignments (esp Canadian CLF)

for

From: John Foliot - bytown internet
Date: Dec 11, 2002 7:49AM

Hi Emma,

As you have discovered, there are issues with CLF's recommended ACCESSKEY
implementation. I brought this to the attention of the CLF Access Working
Group earlier this year, and as you have noted they are now discouraging the
assignment of the 3 accesskeys to the "skip nav" block altogether. While I
cannot define Canadian Government policy, the WATS* testing service (of
which I am a member) will strongly encourage this best practice.
(* http://www.cio-dpi.gc.ca/clf-upe/1/wats/wats_e.asp)

If you think about it, the first 3 links a speech browser will always hear
are:
"Skip to main content" (this should be first)
"Skip to secondary content" (this should be second)
"Francais" (or "English" depending on the site/page) (This is the first link
at the beginning of the common menu bar)

Upon ariving at any given site, the user will probably "listen" to all of
the links presented, to "acquaint" themselves with the navigation scheme.
After hearing this scheme 2 or 3 times however, the user will probably want
to get straight to the "heart" of the page, thus skipping all links ("Skip
to Main Content"). Since CLF limits sites to a maximum of 15 "buttons" in
the main nav bar however, larger sites also use site specific navigation on
the left hand margin, so providing a quick access to these links ("Skip to
Secondary Navigation") will by-pass the 10 to 15 links in the common nav bar
and get you there, thus it's secondary position. The final recommended
"internal" navigation then is simply "Return To Top", which takes you to the
First Link - "Skip All Navigation" (and the circle begins again...).

As far as ACCESSKEYS are concerned, as you have probably discovered there
really are no safe combinations; you will always be "offending" one
technology or the other. As the attachment you included in your email notes
(and thank you by the way for that list... I will "borrow" it), with JAWs
classic laptop included in the test lab there are only about 3 or 4 keys
"unclaimed", and even then they may not be universally available. I ran a
small test during the summer
(http://www.bytowninternet.com/examples/accesskeys.html) and determined that
there were perhaps 4 keys available, including "", "/", and "]". However,
Jukka Korpela noted that these keys are not "standard" on some European
keyboards, but in fact those charactes are invoked using keystroke
combinations. However, if you take JAWs classic laptop out of the mix, you
*are* left with a few alpha keys, including "Z", which might be a good
letter to assign to the "Back To Top" function. At least that way you can
quickly return to the beginning, and then link to the "Main Content"...

One thing of interest is the following passed to me by my associate (as well
as posted to the webaim mailing list) back when I was doing my testing:

> I've only just retested this behaviour I've noticed before. In IE 5+ PC,
> Netscape 6+ PC, and Moz PC I can use access keys that are already
> assigned to intrinsic commands for the menus -- they just need to be
> invoked differently. Sorry, but my Mac and Linux box are both out of
> commission right now, so I'd like to hear from other testers.
>
> Test page at: http://www.furtherahead.com/accesskey.php, which contains
> three links (Foo, Bar, and Edit, with F, B and E as accesskeys
> respectively.
>
> When an accesskey is defined and appears to conflict:
> If I hold down ALT and then press F, I get the access key behaviour.
> If I press ALT, release it, and then press F, I invoke the File menu.
>
> When an access key is NOT defined, both key combinations (press and hold
> ALT then F, and press and release ALT then F invoke the File menu)

> In essence, if I know the difference in the keystrokes, it doesn't
> matter if there is conflict with intrinsic functionality (although the
> presence of the accesskey might actually confuse a user that is
> expecting the menu to be invoked)

One thing that struck me was your comment, "But it doesn't really meet all
of our needs." Does this mean you were looking at adding *more* accesskeys?
I once saw a site where the developer had added over a dozen accesskeys to
each document/page. Even he couldn't remember all of them for each page.
Too many acceskeys is troublesome and impractical... who can be bothered to
remember that many, especially on a page which may be infrequently reviewed
by any given user? Three is probably more than enough ACCESSKEYs for any
given page, and then they should be "standard" through out the site.

So, while adding ACCESSKEYs can aid in navigation, they also present
problems and thus, on balance, perhaps are not worth the effort. (This
seems to be the current thinking within the CLF circles I talk to, but again
I must caution that I do not speak for the Government.)

I would be curious to hear what others on the list think: ACCESSKEYs yea or
nay?

JF





>