I'm no expert, but the process of using ARIA in HTML feels very much like trying to employ "semantic" HTML5 tags, except that ARIA provides a much richer vocabulary. It can be good inline documentation, too: when you come back later and wonder, "What is this div for?", seeing a role="presentation" attribute is a handy thing.
It starts to feel like ARIA takes over the responsibility of semantics and HTML becomes just a scaffold, even moreso than it has been in its relationship with CSS (tags that serve no purpose except as hooks for styling). Which I appreciate, because "semantic HTML" has always felt limited, awkward and a bit pointless, apart from the small part of the effort that has known SEO implications (h1 elements and so forth).
http://www.w3.org/TR/wai-aria/roles#role_definitions
http://www.w3.org/TR/2013/WD-aria-in-html-20131003/#recommen...
Then you can test with the screenreader NVDA (if you have access to Windows or a Windows VM):
http://www.nvaccess.org/
I'm no expert, but the process of using ARIA in HTML feels very much like trying to employ "semantic" HTML5 tags, except that ARIA provides a much richer vocabulary. It can be good inline documentation, too: when you come back later and wonder, "What is this div for?", seeing a role="presentation" attribute is a handy thing.
It starts to feel like ARIA takes over the responsibility of semantics and HTML becomes just a scaffold, even moreso than it has been in its relationship with CSS (tags that serve no purpose except as hooks for styling). Which I appreciate, because "semantic HTML" has always felt limited, awkward and a bit pointless, apart from the small part of the effort that has known SEO implications (h1 elements and so forth).