WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: PDF Reading order question

for

From: Duff Johnson
Date: Aug 3, 2024 7:36PM

> On Aug 3, 2024, at 11:54, Philip Kiff < <EMAIL REMOVED> > wrote:

> Thanks for the clarification/correction about how both WCAG and PDF/UA target "semantic intention" Duff. The fine detail differences between WCAG and PDF/UA can be hard to pin down sometimes (!).

I’m not sure there’s much of a difference per se. PDF/UA focuses on how semantic intent is encoded in PDF whereas WCAG - appropriately - doesn’t get into these weeds, but approaches the subject in principle.

> To expand on what I was getting at about two orders: I didn't mean the visual order/content order vs. the tag order/logical order.
>
> Maybe I misunderstood what you were getting at, but what *I* meant was that there are (at least) two legitimate "semantic" orders available in a document represented by Jon's structure. One follows the visual hierarchy based on typical English reading patterns, and another follows a subjective model of a theoretical document structure. Both are meaningful and valid?

If I understand your point correctly it’s certainly a valid observation about the nature of “typical documents”, and not specific to PDF.

The author’s layout choices are often meaningful but that doesn't imply that they represent an alternative (or competitor) to the author’s intended logical content order (which the author represents via the tag tree).

> WCAG makes clear in SC 1.3.2 that there can be multiple "correct" meaningful reading orders, and all that is required is that one correct order needs to be provided. In a PDF document, pretty much any order that follows the visual hierarchy of the printed/visual version of the document, starting at the top-left, and then proceeding to read the content the way regular English readers proceed from left to right and down to the bottom-right corner of the document is a "correct" meaningful, semantic order in my opinion.

Well, it works fine until / unless the author’s intended ordering disagrees with the left-to-right, top-to-bottom generalization.

Software can always, of course, do what it wants irrespective of the author’s intent, but there is only one tag tree in a PDF file, and thus, a PDF file can only define a single logical content ordering.

Duff.


>
> On 2024-08-02 9:03 p.m., Duff Johnson wrote:
>> While I agree with the practical advice you've offered I want to offer a clarification here:
>>
>>> On Aug 2, 2024, at 19:24, Philip Kiff< <EMAIL REMOVED> > wrote:
>> <snip>
>>
>>> I don't think there is a standard or guideline that provides strict rules in such cases. Technically, I think the WCAG tends to favour the author's "semantic" intention, while PDF/UA tends to favour something they call "logical" order.
>> It’s certainly true that there are no “strict rules in such cases” because semantic intent varies with the situation, the author, the intended audience, and so on.
>>
>> Both WCAG and PDF/UA target “semantic intention”. It’s simply that in PDF/UA, semantic intent is encoded by way of the order resulting from traversal of the tag-tree, known as “logical content order”.
>>
>>> But I know some folks who remediate PDFs believe that the reading order should match the visual/physical order as closely as possible because I think they consider the placement on the page to be part of the significant semantic meaning.
>> I think the reason is that these remediators are trying to provide an acceptable experience for users whose software doesn’t understand tagged PDF.
>>
>>> And they aren't wrong. So that doesn't help deciding. There are two competing orders.
>> </snip>
>>
>> There are not “competing” orders in PDF. There are multiple orderings for various purposes. Reading order is intended to establish the visual appearance while logical content order establishes the semantic intent.
>>
>> The fact that some software still (25 years after the introduction of tagged PDF) cannot use the tags and relies on the visual ordering of content is a different problem.
>>
>> Duff.
>>
> > > >