This piece has been prompted by
- Thread: Why is WCAG 2.0 criterian 2.4.6. only level AA?
- CommonLook - Defining "Heading" in HTML and PDF
When does SC 2.4.6 (AA) apply?
Consider 2 Web pages of say, ABC Company: a Products page and a Services Page. If both pages have an h1 heading (apart from page title) that says "ABC Company" and not "Products we offer" and "Services from ABC Co.", then that violates SC 2.4.6 because the heading does not describe the topic. (I assume every product and service is marked up as an h2 with related details and links on each of the 2 pages respectively.)
Alternatively, setting page title (TITLE element) along the lines of "ABC Co - Products" and "ABC Co - Services" respectively for the 2 pages would ensure compliance with SC 2.4.2.
When does SC 2.4.10 (AAA) apply?
The "Understanding SC 2.4.10" text is quite relevant and useful in this context, and I encourage you to read it. As stated there, it may be unusual to use headings in a personal letter, so SC 2.4.10 does not apply to a letter; neither will SC 2.4.6 because there are no headings in the letter. But because text in the letter is laid out as paragraphs, the letter should comply with SC 1.3.1.
Now consider a research paper with sections like: Abstract, Objective, Background, Methodology, Results, and Conclusion. It is possible these are not styled or marked up as headings. They could be numbered 1, 2, 3, etc. Every section may have paragraphs like 3.1, 3.2, etc. This too is a structure that meets SC 1.3.1.
The research paper complies with SC 2.4.10 as headings are indeed used to organize content though they are not marked up or styled differently from the rest of the research text.
If the paper had no section markers (like Abstract, Objective, Background, etc.) at all and the entire research text were placed as sequential paragraphs, it would fail SC 2.4.10.; and SC 2.4.6 would be inapplicable because it has no headings, like the letter.
Now both together: SC 2.4.6 and SC 2.4.10
As the section headers in the above research paper are not marked up as headings, it seems SC 2.4.6 does not apply to the paper. Certainly marking them up as headings would be good for accessibility and usability for all, but that is at the author's discretion. That's why it is necessary to separate SC 2.4.6 from SC 2.4.10. Indeed, if they are marked up as headings, then both SC 2.4.6 and SC 2.4.10 apply.
But here is a poser: if section headings (that are not marked up as headings, as in above research paper) are used to organize content, then doesn't one have to test if these headings describe the topic of the content that follows? Does test of "description of topic" apply only when section headings are marked up as headings?
So it appears for now that if section headings are used to organize content, and these are indeed marked up as headings, then both SC 2.4.10 and SC 2.4.6 apply. If they are not marked up as headings, then only SC 2.4.10 applies.
If section headings are styled differently from the text but not marked up as headings then it is a failure of SC 1.3.1 (A).
Meaning of "heading"
I do not think there is a difference in the meaning of the term "headings" as used in HTML4 or HTML 5 or the PDF specs. Essentially they all refer to section headers.
My view is that one should not do too much hair splitting. Headings are headings, like headings in a newspaper (some big, some small) or headings in a text book for chapters and sections within a chapter. Certainly conveying structure and hierarchy is the main point, but one should use discretion and might occasionally need to mark up something else on the page as a heading; e.g., a promotional offer like "Offer valid upto May 14", because it is important text that needs to be brought to the reader's attention for usability reasons. Note, this is not a section heading and may not fit the hierarchy of headings on the page. The purpose is to facilitate human reading; if headings are overdone on a page, it could defeat the purpose of headings.
Refer to Web Content Accessibility Guidelines (WCAG) 2.0 for WCAG 2 related references above.
By Sailesh Panchang, Deque Systems