I just published a new web accessibility article on my blog:

Marijke Luttekes

Thanks again to @jscholes for sharing his wisdom with me!

Curtis Wilcox

@mahryekuh
Screen readers have shortcuts for navigating both by heading and by landmark (using NVDA they're H and D, respectively), what's the advantage of adding a visually hidden heading to a landmark?

@jscholes

Marijke Luttekes

@cwilcox808 I'll wait to see if James has time to elaborate since he is excellent at nuanced answers to such questions.

@jscholes

James Scholes

@cwilcox808 In the most recent WebAIM Screen Reader Survey, when asked "How often do you navigate by landmarks/regions in your screen reader?", 36.7 percent of respondents said seldom or never (only 17.9 percent said "Whenever they're available").

By contrast, 71.6 percent said that when "trying to find information on a lengthy web page", their primary method would be to "Navigate through the headings", as opposed to the tiny 3.7 percent that would first reach for landmark navigation instead.

Source:

WebAIM: Screen Reader User Survey #10 Results

(webaim.org)

@mahryekuh

Christoph Heim

@mahryekuh Hi Marijke, I have one question about this approach. Doesn’t you mess up the over all heading hierarchy? At least from a seo perspective?

Curtis Wilcox

@jscholes
I think the survey's Finding Information question is put in such a way that it's expected landmarks wouldn't be the first choice; on a lengthy page, most information will be in the main landmark so that doesn't narrow the search down much.

More relevant is the 36.7% choosing Seldom or Never vs. 31.8% choosing Often or Whenever they're available for the Landmark/Region question.

So is the redundancy everyone experiences considered a small price to pay for those who do appreciate the landmarks?

@mahryekuh

Marijke Luttekes

@christoph_heim Hi!

What I know about it is listed in the section "Wait, I thought we weren't supposed to skip heading levels?"

Personally, I don't care about SEO, especially if it harms accessibility. I also assume that search engines will know how to deal with this, especially when the most critical heading (H1) is in the main content.

Google reported in the past that they want to boost websites with good content and accessibility, but who knows? ‍️

James Scholes

@cwilcox808 If `aria-labelledby` is used, a screen reader can pick up on the fact that the first element inside a container provides that container's name and suppress the duplication accordingly. Only NVDA offers this quality of life improvement at present as far as I know, and I would encourage other screen reader vendors to adopt it. @mahryekuh

Curtis Wilcox

@jscholes @mahryekuh
Thanks.

My original question was not entirely genuine, I've recommended similar designs at times. I asked because I didn't feel the article explained why to use it and because I was curious if there were considerations I wasn't aware of.

James Scholes

@cwilcox808 All good. I'm also not trying to dismiss your concerns; some screen readers are seriously bad offenders in terms of how they convey container boundaries, names, roles, and first/last elements.

I've built pages that work absolutely flawlessly with NVDA, giving me all the concise context and navigability I could ever want. Then when I open them on my phone with VoiceOver, it's verbosity soup.

It's tricky (and possibly a good candidate for a series of articles in itself) to decide where to draw lines in your implementation between flexibility and verbosity (particularly as the latter can change underneath you). @mahryekuh

Marijke Luttekes

@jscholes Thank you, James! Screen reader nitty gritty is definitely beyond my expertise.

@cwilcox808

Adrian Roselli

@mahryekuh @jscholes
I am a bit confused why the headings would be hidden from sighted users (including sighted screen reader users).

If the content warrants a heading, then everyone would typically benefit.

If it’s only used to make regions navigable, then you risk arbitrary heading structures (and repeated navigation points when using heading *and* landmark navigation).

If that, then why not non-heading visually-hidden arbitrary plain text instead? You only need it for the accName.

Adrian Roselli

@mahryekuh @jscholes Not meant as criticism, I just don’t understand the use case.

Marijke Luttekes

@aardrian Does the thread I’ll link below answer your questions? Otherwise I’ll let James field them again.

(The redundancy in some screen readers is a real issue, though)

James Scholes (@[email protected])

@[email protected] In the most recent WebAIM Screen Reader Survey, when asked "How often do you navigate by landmarks/regions in your screen reader?", 36.7 percent of respondents said seldom or never (only 17.9 percent said "Whenever they're available"). By contrast, 71.6 percent said that when "trying to find information on a lengthy web page", their primary method would be to "Navigate through the headings", as opposed to the tiny 3.7 percent that would first reach for landmark navigation instead. Source: https://webaim.org/projects/screenreadersurvey10/ @[email protected]

The Dragon's Cave (dragonscave.space)

@jscholes

Adrian Roselli

@mahryekuh @jscholes
I had not seen that thread, thanks. It doesn’t answer my question, though. And James has already spent a lot of time in that thread and in another forum explaining things in the last 24 hours so I hate to put him out again.

Marijke Luttekes

@aardrian The goal was explicitly to add headings so screen reader users can use the heading navigation tool instead of switching between or combining heading and landmark navigation.

Sighted users already have visual clues about where a primary navigation or footer is.

I referred to James because he and his colleagues made me aware of this technique and requested it in a project, whereas I only have second-hand experience. But while he enjoys rabbit holes, no explanation is owed.

@jscholes