Option to toggle reporting of landmark types

[bhavyashah]

Is your feature request related to a problem? Please describe.

There are tons of different landmark types: main, banner, navigation, complementary, content info, region, etc.

1. Most users don’t understand the difference in the purpose each landmark type serves semantically. If they don’t understand it, they can’t care about it, and thus deem it noise.
2. Often, it is clear what a landmark is based on its text. For instance, if you are reading this issue on GitHub, notice how the last landmark on the page is a “Footer navigation landmark Footer navigation.”
3. Issues like #10420 show that verbosity is a problem on the web, especially when it comes to landmarks. Plus, words like “content info” and “complementary” and “navigation” are gargantuan.
4. Most landmarks are reported as “{type} landmark” (like “navigation landmark”), but not all. For instance, NVDA only says “region” for region landmarks. This leads to a user not knowing to navigate by regions using “d” on the web. This was me when I was struggling to navigate by regions to move cell by cell in Google Colab.

Describe the solution you'd like

Include a toggle in Document Formatting settings for reporting of landmark types. When disabled, all landmarks are reported simply as “landmark.” I'd go as far as to say that by default, disable it.

Describe alternatives you've considered

For the fourth problem, a more direct fix would be to report regions as “region landmark.” Alternatively, we could change the Input Help message of “d” from “moves to the next landmark” to “moves to the next landmark or region” (consistent with the Document Formatting setting labelled “Landmarks and regions”).

Additional context

N/A

[seanbudd]

We don't see the clear benefit of this middle state, why not just turn off landmark reporting altogether?

[bhavyashah]

I explore a new page using “h” for headings and Ctrl+up/down arrows for paragraph navigation. As I do so, NVDA reads out elements on the page, including landmarks. This builds up my mental model of the page, highlighting the areas of the page the author designated as “landmarks”, much like how I flag landmarks when mental mapping a physical space. Most concretely, knowing what and where the landmarks are can cue me to move through that page quicker with “d.” E.g., landmark navigation to move between cells in Google Colab. I think this is a common user flow. According to WebAIM’s most recent survey on screen reader user preferences, only 3.7% users report landmark navigation as their first move for finding information on a webpage. However, 83.9% of users still do navigate by landmarks. The way I interpret this is that most people don’t press “d” to map out a new page, but once they come across a landmark through things like arrowing around, they may use “d” to jump to it if it corresponds to a part of the page they care about.

[SaschaCowley]

@bhavyashah I'm still not sure I understand the desire to report landmarks but not their type. Landmark type presents semantic information.