This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The current IA2 and ATK mappings for <footer> is ROLE_FOOTER. For IA2 this makes it "Footer of a document page", and for ATK, "An object that serves as a document footer". This makes sense if the <footer>'s nearest ancestor is the <body> element in which case it is a footer for the page. But it doesn't sound right if the nearest ancestor is an <article> or <section> or other sectioning content or root element, e.g. <blockquote> or <td>, as these are not documents in themselves. The same applies to the IA2 and ATK ROLE_HEADER mappings for <header>. Wouldn't just ROLE_SECTION be more appropriate for both elements, and in the case that <body> is the nearest sectioning ancestor, the ATK mapping could be ROLE_LANDMARK? (Will IA2 be getting such a role?)
That makes sense, I'm curious though whether a footer that is not under body is an author error, and whether the browsers have to fix it.
If footer is descendant of article or section or other sectioning element that is not body, it is footer specifically for that ancestor section, as per the spec. Otherwise is footer for body, which is only other option.
I've updated the mappings for footer and and header. See also bug 27961.