E-mail List Archives

Search E-mail List Archives for

Number of posts in this thread: 2 (In chronological order)

From: Patrick H. Lauke
Date: Mon, Oct 06 2014 5:07PM
Subject: Does a failure of WCAG 2 3.1.1 automatically mean a failure of 3.1.2?
No previous message | Next message →

WCAG 2 3.1.2 SC states that "The human language of each passage or
phrase in the content can be programmatically determined [...]". If a
page already fails 3.1.1, failing to provide the "default human
language"and thus making it impossible to programmatically determine it
for the page as a whole, would I be right in thinking that this also
makes it impossible to programmatically determine the language for any
passage.phrase?

And conversely, even if there are no language changes in the current
document, if a page passes 3.1.1, would it also automatically pass 3.1.2
as if you can determine the language for the whole, you can also
programmatically determine it for any passage/phrase?

(Context here is that, unless there are actual language changes, some
WCAG 2 reports mark 3.1.2 as non applicable, rather than pass/fail...but
I'd contend that it should be assessed based on the above logic).

Thoughts?

P
--
Patrick H. Lauke

www.splintered.co.uk | https://github.com/patrickhlauke
http://flickr.com/photos/redux/ | http://redux.deviantart.com
twitter: @patrick_h_lauke | skype: patrick_h_lauke

From: Jukka K. Korpela
Date: Tue, Oct 07 2014 3:26PM
Subject: Re: Does a failure of WCAG 2 3.1.1 automatically mean a failure of 3.1.2?
← Previous message | No next message

2014-10-07 2:07, Patrick H. Lauke wrote:

> WCAG 2 3.1.2 SC states that "The human language of each passage or
> phrase in the content can be programmatically determined [...]". If a
> page already fails 3.1.1, failing to provide the "default human
> language"and thus making it impossible to programmatically determine it
> for the page as a whole, would I be right in thinking that this also
> makes it impossible to programmatically determine the language for any
> passage.phrase?

3.1.1 is a level A requirement, intended to be simple to fulfill,
basically just one lang attribute when using HTML, whereas 3.1.2 is
level AA and may require much more work (though only for content that
mixes different languages). So I think the question is rather theoretic.
But the answer would be that it depends - on the assumed "programmatic
determination" mechanisms.

Consider a structurally trivial HTML document where the body consists of
only one h1 element followed by some p elements, each carrying a lang
attribute. This would satisfy 3.1.2. Does it also satisfy 3.1.1 if there
is no lang attribute on the body or html element? It depends on whether
we can expect a user agent to infer the "default human language". What
if the elements have different lang attributes?

> And conversely, even if there are no language changes in the current
> document, if a page passes 3.1.1, would it also automatically pass 3.1.2
> as if you can determine the language for the whole, you can also
> programmatically determine it for any passage/phrase?

Yes, it would be unnatural to think otherwise. The lang attribute has
been defined so that it specifies the language of content unless
overridden by lang attributes for descendant elements.

> (Context here is that, unless there are actual language changes, some
> WCAG 2 reports mark 3.1.2 as non applicable, rather than pass/fail...but
> I'd contend that it should be assessed based on the above logic).

Logically, a rule that requires that something be made for all language
changes must be considered as fulfilled if there are no language changes.

The concept "programmatically determined" is vague. In practice it seems
to be a generalization of the lang attribute and similar metadata. But
the definition says: "determined by software from author-supplied data
provided in a way that different user agents, including assistive
technologies, can extract and present this information to users in
different modalities". This is probably mean to mean metadata, not data.
But read literally, it could often be considered as satisfied even in
the absence of lang attributes or other metadata.

After all, current state of the art makes it possible to determine the
human language from the text itself, in a reasonably reliable way.
Google does this in a massive scale and actually ignores lang attributes
as too unreliable. And the textual content is "author-supplied data", is
it not?

Whether assistive software actually performs language recognition from
the content itself is a different issue. The definition says "can",
instead of referring to actual behavior.

Yucca