Identifying vocabulary and verbosity level attached to roles

[gautierchomel]

Adding semantic information to rendered content for non-visual reading experiences is common. This allows users to access semantic information that is usually visually conveyed. For example, when a table starts, the audio reader can hear "table."

A list of suggested output for DPUB ARIA is maintained by Daisy transition to epub Working Group and a Naming role guidance is provided by WAI ARIA.

Because high verbosity can become annoying and disrupt the reading experience, it needs to be user-controlled. This means we should also figure out settings, rules, and verbosity levels in an understandable and usable way.

[gautierchomel]

For reference, here are role labels applied by NVDA

[HadrienGardeur]

Thanks @gautierchomel, that's useful to identify how specific elements should be contextualized but also to check which role is missing compared to NVDA.

I'm wondering about start/end variations though, is this completely generic or could be be specific to the role?

For example Daisy suggests "chapter" for the chapter role. Should we use "start of chapter" and "end of chapter"? We might get into i18n issues similar to what we've seen with the display guide, where this approach wouldn't work with all languages and/or roles.

I'm also worried that the output could be less than ideal in some cases. For example with a page break, we can probably do better than outputting:

• start of page break
• 10
• end of page break

Unlike screen readers, this contextualization could be used by a much wider range of users, which is why I'm tempted to spend additional efforts on making them feel right in the middle of an audio playback.

[HadrienGardeur]

For now, we have 72 roles in Guided Navigation vs 163 in NVDA.

Their coverage is more extensive but also much more Web-oriented with many roles that mostly make sense for applications and/or documents.

I think it's worth opening an issue dedicated to cherry picking some of these roles as well.