Need for exemptions for teaching materials, samples, multilingual documents

A year or so ago, I wrote:

<http://lists.w3.org/Archives/Public/w3c-wai-gl/2004AprJun/0592.>

>Let us all keep in mind a general exception that must be applicable 
>to all our guidelines: If the purpose of the page (section, site, 
>resource, document) is to teach the topic of accessibility itself, 
>it may violate the guidelines. Hence indeed not "all" content may be 
>accessible.
>
>It is easy to imagine giving correct and incorrect examples of 
>accessible methods. We do that already. The incorrect examples, as 
>Web content, would violate WCAG.
>
>It is also easy to imagine entire online courses where full Web 
>sites for imaginary companies are posted, with students expected to 
>take the sites and fix them. Those sites, with 
>deliberately-incorrect markup, would not be permitted without an 
>exception.
>
>I don't think that "scoping" is a good way to handle this. I suggest 
>a clear and explicit exception.

Jason White wrote in and stated that scoping would handle that just 
fine, thank you, and, as with most of my proposals, it was forgotten.

I am bringing this up again and proposing that guidelines and 
techniques be updated to include three obvious exception cases:


1. The Web content is used as a teaching example and cannot be made 
compliant without contradicting (undermining, vitiating) the matter 
being taught.

This case is known from my previous example. If you really insist, I 
can look up a few pages that instruct e.g. form markup by teaching it 
in various stages; some pages end up with invalid code because that 
stage of the elucidation cannot include all the necessary form markup 
to make the page valid.


2. The Web content is a sample for which a single natural language 
cannot be declared.

This is a case not well imagined by the Working Group (or in PDF, for 
that matter). There are many examples of Web content that is in fact 
text yet are not in a natural language. The example I'm providing 
here is type samples.

The documents linked below provide a list of all the glyphs typically 
required for a font that includes Latin, Cyrillic, and Greek 
characters. Have a look at the tagged PDF and the valid HTML.

<http://joeclark.org/dossiers/FontGlyphSynopsis-tagged.pdf>
<http://joeclark.org/dossiers/FontGlyphSynopsis-2.html>

We cannot specify a natural language because there isn't one. The PDF 
accessibility checker, in its default setting of "look for every 
error," dutifully flags this *as* an error. The HTML is valid, but it 
flunks WCAG 1 or 2 because it doesn't specify a natural language that 
cannot exist for this document.


3. The (A single unit of) Web content is provided in more than one 
natural language.

In this case, <html lang="LANGUAGECODE"> will not work. HTML, in its 
brilliance, makes it logically impossible to declare a bilingual 
document. (Really, why would CERN researchers in, of all places, 
Switzerland ever have to write in more than one language?) You can 
get around this in practical terms by marking up just the sections of 
your document as being in whatever language they actually are in, but 
WCAG needs to make such a hack explicitly legal.



In summary, we must make room for cases like these in the guidelines. 
We can't just say that scoping will take care of it, because 
essentially that forces authors to say "All our Web pages meet WCAG, 
except these two, which logically cannot meet WCAG." We know about 
these logical disqualifications up front and it is our duty to 
accommodate them.



-- 

     Joe Clark | joeclark@joeclark.org
     Accessibility <http://joeclark.org/access/>
     Expect criticism if you top-post

Received on Monday, 18 April 2005 16:50:35 UTC