RE: AXAPI Role mappings for DPub and Graphics AAMs

> EPUB requires the use of a table of contents that links to sections of the
documents

 

There is a required table of contents, but authors typically do not
reference every section as it makes navigation problematic for sighted
readers due to indenting and line wrapping.

 

> EPUB requires the use of class="acknowledgements"

 

No, the class attribute has no purpose beyond styling in EPUB.

 

> EPUB requires the use of epub:type="acknowledgements"

 

Semantics are not enforced. Elements are not required to have them. The
application is expected to be done sensibly.

 

But the bigger point is that the semantics do not get conveyed to assistive
technologies, even though it was hoped from the outset that reading systems
would be able to expose the information in the epub:type attribute without
placing an extra production burden. It's not happened. iBooks does not
expose this information for navigation, for example; you still can't even do
basic navigation from paragraph to paragraph. As books move onto the web,
the functionality of a reading system is expected to diminish (a thin layer
in the browser), which means more reliance on traditional AT. We're hoping
it will give readers a better experience than traditional standalone apps,
as they'll be able to use the technologies they prefer.

 

It's also not the case that these are just reading system hooks. Although no
use is ruled out, the specification doesn't define any reading system
requirements for any of the semantics. Of all the semantics we have in epub
(a much bigger list), only footnotes get special treatment that I'm aware
of, and authors have to account for lack of support as this again is not
spec-defined or mandated behaviour that can be counted on. I can't really
follow what the document structures are thought to be for if not to improve
navigation, to be honest. The motivator for the attribute and semantics was
accessibility and harmonization with publisher production processes more
than anything.

 

As we seek better alignment with the web, it's clear that we need to realign
with ARIA for accessibility, but there is nothing simple about the current
ARIA roles for publishers. Unless you are already on the inside, a "region"
or "group" of a publication is confusing and unintuitive. They also don't
fit well with their workflows. Publishers will never mistake what the
publishing roles are for or misapply them, but I guarantee you they'll make
a mess of that pair. Authoring simplicity is always in the eye of the
beholder.

 

Publications are also an immersive experience, so you don't often find them
in the clutter of a standard web page. They're also broken across many
documents. Take away the publishing landmarks and there's nothing left. This
move isn't going to surface needed information, only take it away. Why force
the user out of the page to a table of contents -- and likely lose their
playback position, which isn't always easy to return to -- if they just want
to quickly determine what is available in the current document? Why make
them jump to every heading with the same result?

 

But this disagreement seems rooted in a misunderstanding of the nature of
publications relative to web pages. Go to your library and grab a handful of
books and you'll find the structures we've included as the primary headings.
These are the major navigation points, and they're no less important than
the major regions of a web site. This isn't an academic exercise, but the
culmination of twenty years of making publications accessible through DAISY
talking books and now EPUB.

 

Rich's point about diminishing returns was also twisted out of context. The
issue those of us working on this proposal are aware of is that there are
diminishing returns once you move into every sub-component that might occur
below these primary headings, just as you limit web page landmarks to the
major regions. There are also nuances that don't necessitate new headings,
even if the heading would be more precise. None of us disagree with this.

 

As Rich already noted earlier, I'd hope that these roles could also lead to
new and more innovative ways to make publications accessible to all readers.
We're all trying to get the best experience, even if we have different
perspectives on what is useful.

 

Matt