Quiz: Tabindex disorder
Dennis in Web Axe Episode 24 mentions”
If the tab orders are different than the actual flow of the content on your page, it’s going to be really disorienting for the (screen reader) user.
Really? We know that this sort of disorder will be apparent in regular browsers. That happens with screen readers? Please try a test case where I have intentionally made the tabindex values different that their order in the source. Use your favorite screen reader to tab from link to link and tell us what sequence you hear.
September 7th, 2006 at 10:02 am
JAWS 7.10 with both Firefox 1.5.0.6 and Internet Exporer 6 reads the list of links in the order 1-4 if you read straight through them in reading mode, but are ordered 1,4,3,2 when tabbing through the page.
Jon
September 13th, 2006 at 11:21 pm
For JAWS 7.0 and 5.0 I got the same results - when allowing the reader to read through the page, the links are read in source order and tabindex makes no difference; but when tabbing through the page manually (using either of JAWS’ cursor modes) the links are followed in tabindex order.
For Window Eyes 5.0 (IE6) and 5.5 (IE6 and Firefox 1.5) the results were slightly different from JAWS (though all consistent with each other). When allowing the reader to read through the page, the links are read in source order; when tabbing through the page manually with the reader in MSAA Mode (or “browse on” - the normal mode used for reading pages), it’s just the same - links are followed in source order; however when the reader is not in MSAA Mode (or “browse off”, the mode used when completing forms) it’s a different story - the links are followed in tabindex order.
In Hal 6.5 and Connect Outloud 2.0, the links are always followed in source order, irrespective of mode or interaction - tabindex makes no difference.
I would conclude from this that using a non-natural tabindex does indeed create the potential for confusion, and therefore that tabindex should not be used at all.
Though I’ll be interested to investigate whether using tabindex=”0″ (effectively, an “auto” setting), or the “-1″ value of the mozilla/IBM scripting initiative, will make any difference anywhere, as opposed to elements with no tabindex.
September 14th, 2006 at 4:12 am
One thing that sticks out a mile is this “disorienting for the (screen reader) user”. Once again it has been assumed that the only people who navigate a website via the keyboard is a screen reader user. This is not the case, I know many people who prefer to use the keyboard, or are in fact forced to because they are unable to use a mouse. If the tabbing index does not follow the flow of the page then it becomes very disorientating. You have to visually try and ‘guess’ where you are.
Your example is really too small to show this, but after you tab through the list I still had to try and find where I had then jumped to! Furthermore the lack of style associated made things worse.
September 15th, 2006 at 2:44 pm
Rich: Hear here!
It was Dennis who said “disorienting for the (screen reader) user.” I agree with you. There are many people beyond screen reader users who use the keyboard and will be equally disoriented in a more confusing case.
Yeah, I intentionally avoided any style that could give more hints.
Thanks for stopping by!
September 23rd, 2006 at 6:05 pm
Hi guys. It seems my quote from the Web Axe podcast was a little misleading. It seems a screenreader will “say” your test page above correctly, but a user with an input device that relies on a keyboard or soft keyboard would be confused by the inconsistency in the order of the list items and the tabindexes.