Re: aria-label usage in table cells

"Advisory" in this context means "not required to understand the 
content, but might add some degree of usability". That is, you cannot 
rely on the title attribute to convey important information. If I 
understood well your case, you are giving this information in the table 
headers, so you are not relying on @title, isn't it?

Regarding the use of @title when you cannot use a label, it refers only 
to form controls. In this case, when a form control has no associated 
label, the @title is mapped to the accName property in the accessibility 
API, so it is read by the screen reader exactly the same as the label. 
It could also used by voice recognition software, and of course 
presented to mouse users.

For keyboard-only sighted users it would certainly not work, but 
remember that the technique is intented only for situations where a 
visible label is present, thus allowing the sighted user to identify the 
field (or components of the field). For example, imagine that you have a 
"birthdate" label (probably within a <legend>) above three separate 
<select> fields (day, month, year). For design reasons you might not be 
able to use visible "day", "month" and "year" labels, but the 
"birthdate" is assumed to be enough for sighted users.


Devarshi wrote:

> I am with you on comparable access, but who is at fault here -- 
> developer, browser, business, or spec? Isn't the title attribute 
> supposed to be exposed to sighted keyboard only users?  Spec says "This 
> attribute offers advisory information about the element for which it is 
> set."
> A case in point, and it may be extended to large data tables where 
> column / row headers are not visible to all user groups - H65: Using the 
> title attribute to identify form controls when the label element cannot 
> be used ( - Question: can a 
> project team using standard code claim compliance because the title 
> attribute serves screen reader and mouse users and possibly other user 
> groups , but not sighted keyboard only users? Would it be advisable to 
> fail that page even though the team used standard code (or at least tried)?
> -Devarshi

Received on Friday, 20 June 2014 19:47:40 UTC