W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 2002

Re: 4.1 reproposed

From: Charles McCathieNevile <charles@w3.org>
Date: Mon, 16 Sep 2002 15:49:43 -0400 (EDT)
To: Avi Arditti <aardit@voa.gov>
cc: <w3c-wai-gl@w3.org>
Message-ID: <Pine.LNX.4.30.0209161519140.12468-100000@tux.w3.org>

Hi Avi,

I agree with you about the success criteria. I don't think that any of the
proposed success criteria (except "there is a claim that you meet this
checkpoint") are testable in a way that will produce consistent results and
provide a benefit.

Your checklist items start to show the way, however. I don't think there is
much testable value in "does the syntax elicit the appropriate response" - it
seems to be a way of allowing for the claim that such work as e e cumings
poetry is accessible. Fine poetry and worthy of being published it is (at
least some of it). But that doesn't make it accessible.

I think if we aim for things like "any verb form used is not the same word
form as a noun", we will be getting closer.

"No word is used which is not the same in all major dialects of the language,
unless there is no common word". That cuts out "jumper", a thing Australians
wear in winter which has nothing to do with what most americans think of as a
jumper. "Pullover" I think is common to different versions of english, and I
am fairly sure that "sweater" is. On the other hand, "sidewalk" and
"footpath" make sense because only the one word is used in each of the US and
Australia. (Hmm. Footpath may have a different sense in the US, but there is
no common word).

"The first sentence of a paragraph summarises the paragraph." This means that
you can read just the first sentences and understand what is being said.
Perhaps this works only for an argument, where reading all the first
sentences means you miss the expanded reasoning, but have the tenets of the

Then there are things that really are specific to language. In english,
"there are no verbs in the passive impersonal form". "There are no dependent
clauses which themselves contain dependent clauses". (I am guessing here -
maybe that is relevant to all languages, or maybe it isn't that helpful).

For english "An automatic translation (such as babelfish, voila, etc) to
french and then from french to english does not distort the sense of the
document". This seems to me a good test. The results will be the
same more than for the test "someone responsible said this was good writing".
And it can be done in the real world.

(I tried that test on the last two paragraphs one at a time. The first
paragraph below went to spanish and back. The second went to french and back.

Then there are the things that really are specific to the language. In
English, "there are no verbs in the passive impersonal form". "there are no
dependent clauses that they themselves contain the dependent clauses". (I am
conjecturing here - that is excellent all the languages, or perhaps it is not
perhaps that beneficial).

For English "a machine translation (such as the babelfish, veiled it, etc.)
with the French and then the French-English one does not twist the direction
of the document". This seems with me a good test. The results will be the
same ones more than for the test "somebody says responsible this was good
writing". And it can be made in the true world.

After all, we need to be able to apply our tests to our own work <grin/>)

So I think we are still not there. But I think it is feasible, and that the
tools we need exist - we are just looking for the right way of using them,
and of explaining how to.



On Mon, 16 Sep 2002, Avi Arditti wrote:

>My action item: "repropose 4.1 based on communications today [Sept. 5].
>ask people to focus on success criteria instead of checkpoint text."
>I've read the 1.0 core techniques
>(http://www.w3.org/TR/WCAG10-CORE-TECHS/#comprehension), as Wendy
>suggested, and studied the latest internal draft of 2.0.
>These success criteria remain a terrific struggle. I'm still not happy
>with them. Suggestions (of the collegial sort) gratefully accepted. I
>hope to come back with more ideas from the plain-language conference in
>Toronto at the end of the month.
>To restate the obvious, 4.1 has got to be delimited to address the
>central weakness: testability. So I've started a checklist as an idea,
>and in a style that could apply across language systems.  (I've written
>them in a way suggested by a friend at the U.S. Department of
>Education.)  If we proceed with this checklist, I could add items from
>the long list that's been put together. Also, there's the techniques
>list which could contain language-specific ideas and references. One
>more thing: because of 3.1, I've focused on other-than-structural
>elements.  So here goes ...
>Checkpoint 4.1 Strive to write clearly [I know, focus on the criteria
>first -- but any thoughts on this wording, the verb "strive" borrowed
>from the 1.0 core techniques?]
>You will have successfully met Checkpoint 4.2 at the Minimum Level if:
>1. Portions of new content (especially directions, commands and options)
>are written clearly to the extent those responsible consider
>You will have successfully met Checkpoint 4.2 at Level 2 if:
>1.Portions of new content meet at least several items on the following
>2.A conformance claim associated with the content asserts conformance to
>this checkpoint at level 2.
>You will have successfully met Checkpoint 4.2 at Level 3 if::
>1.Significant portions of new content meet applicable elements of the
>2. Or, significant portions of new content meet elements of an
>independently established set of guidelines for clear writing.
>3. Remaining content is reviewed and changed as desired by those
>4.A conformance claim associated with the content asserts conformance to
>this checkpoint at level 3.
>Note: This checkpoint deals with the strength of writing to convey
>information -- in other words, to help users understand. It is
>consistent with the aims of a global movement to promote clarity in the
>communications that affect people's lives.
>Clear writing benefits everyone, but especially:
>    -- those with cognitive disablities
>    -- those whose ability to parse text is limited by screen readers or
>other assistive technology
>    -- those whose native language is different from the language of the
>Clear writing does not have to mean simple writing. It is writing that
>is appropriate for the purpose and the audience. The goal is not to
>limit creativity or the scope of content. The goal is to encourage a
>reduction in needless complexity.
>How clear is clear enough? The answer will always vary by audience,
>subject and context.  So too, ideas and techniques to communicate in
>"plain language" vary from language to language. Ultimately, users may
>be the ones who, through their actions and choices in viewing sites,
>decide the answer.
>1. When content gives directions or commands, does the text make clear
>what the user must do?
>2. When a term of address is used, it is clear when the user is being
>3. Would the writing style reasonably be considered clear by the
>standards of the language and culture (public or professional) in which
>the content is written?
>4. Overall, is the syntax appropriate for eliciting the desired action
>or response?
>5. Overall, is the vocabulary appropriate for eliciting the desired
>action or response?
>6. Would the verb forms generally be considered easy to understand for
>the intended purpose of the content?
>7. Is a controlled language used?

Charles McCathieNevile  http://www.w3.org/People/Charles  tel: +61 409 134 136
SWAD-E http://www.w3.org/2001/sw/Europe ------------ WAI http://www.w3.org/WAI
 21 Mitchell street, FOOTSCRAY Vic 3011, Australia  fax(fr): +33 4 92 38 78 22
 W3C, 2004 Route des Lucioles, 06902 Sophia Antipolis Cedex, France
Received on Monday, 16 September 2002 15:50:18 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:20 GMT