W3C home > Mailing lists > Public > public-html@w3.org > June 2008

Re: Proposal: ARIA-ROLE & CSS definition integration

From: Pat Hayes <phayes@ihmc.us>
Date: Mon, 9 Jun 2008 15:04:28 -0500
Message-Id: <p06230935c473383c0c06@[]>
To: Al Gilman <Alfred.S.Gilman@IEEE.org>
Cc: Justin James <j_james@mindspring.com>, James Craig <jcraig@apple.com>, <public-html@w3.org>, "'W3C WAI-XTECH'" <wai-xtech@w3.org>, <www-tag@w3.org>
At 2:26 PM -0400 6/9/08, Al Gilman wrote:
>[radically reducing distribution; please feel free to copy to anyone who
>needs a copy]

[I put it back on the public lists as I think its important to try to 
converge on terminology.]

>On 9 Jun 2008, at 11:16 AM, Pat Hayes wrote:
>>At 1:03 AM -0400 6/9/08, Justin James wrote:
>>>James -
>>>>  I think you may have missed the point of my JavaScript example. It was
>>>>  just one way you could insert ARIA semantics into the DOM by flagging
>>>>  an HTML class or id, or for that matter, by any CSS selector. The
>>>>  example was not my recommendation for how it should always be done.
>>>Ah, gotcha, thanks!
>>>>  One thing you could do is to help ensure that all of the ARIA
>>>>  semantics get rolled into HTML 5.
>>>I fully support this, and I will be looking out for it. I think that ARIA is
>>>too important to be *not* rolled 100% into HTML. First, it eliminates many
>>>of my gripes with HTML as a presentation layer for application development
>>>(HTTP is still wholly inadequate for the task...), by finally (15 years too
>>>late) providing a mechanism for AT systems to "get" HTML. Secondly, it
>>>provides a way to get really darned close to the semantic Web ideal.
>>As someone reading all this from the sidelines, this direction of 
>>discussion seems to me to have gone into left field.
>>Can anyone briefly explain what is meant by "ARIA 
>><em>semantics</em>" (my emphasis),
>"This object has the focus," "this option is selected," etc.
>See, for example

OK, thanks. Just for the record, that isn't the way that the word is 
used in Semantic Web discussions.

>>what CSS has got to do with semantics (in any sense),
>Layman 'sense' of the symbol 'semantics':
>3 a: the meaning or relationship of meanings of a sign or set of 
>signs; especially : connotative meaning

That's sufficiently general and bland to apply to almost anything. 
But you don't mean semantics which supports a notion of inference 
which can be performed by a machine, right? And in any case, I 
wouldn't have said that text color was semantic even in this sense.

>>and what either of these is likely to do for the semantic Web?
>Dunno.  "the semantic Web ideal" may refer to "making the web
>more semantic (i.e. a less error-prone representation of meaning)."

No, that's not what it means. See http://www.w3.org/2001/sw/. And 
that's not what 'semantic' means, either: 'more semantic' doesn't 
mean 'more precise'. Imprecise languages have semantics too.

>In that case, the beneficiary is the Web, not the Semantic Web.
>>Isnt CSS entirely about, well, graphic style?
>And what is the style about?  Connotative meaning, quite often.

I think not, most of the time, in fact: but certainly not in any 
sense of 'meaning' used in the context of talking about the SWeb.

>Web 0.1: this text is clickable, is a hyperlink.
>Web 2.0: this <div>is a drop-down menu, but currently
>>I see nothing even slightly semantic in the question of whether 
>>some text items should be rendered in large red characters (say).
>Graphic artists and GUI designers have been violating that, with
>results that they like, to the point that it is standards operating practice.
>Given the frequency with which facts like "this form field requires
>an answer" or "the answer in this form field is invalid" is entrusted
>to presentation properties such as a red color, there is meaning in forms,
>that all users need, that is encoded in presentation properties in the
>GUI look and feel.

Meaning to human readers, maybe. But that's not what the SWeb is 
about. Also, I wonder how many of these graphic devices are anything 
close to being 'standard', in fact. I bet most GUI designers would 
rather resent being told what colors to use in order to conform to a 
W3C directive. (BTW, 'requires an answer' is most often encoded by an 
asterisk, in my experience.)

>>If this is semantics, then we must be talking about entirely 
>>different notions of "semantic Web".
>Different?  Probably yes.  Entirely?  That's a matter of 

Well, actually its a matter of public record: the term "semantic web" 
was introduced by the W3C, after all, and they have a reasonably 
clear meaning for it.

>>What is your "semantic Web ideal" ?
>Here's mine:
>That the semiotics of the Web be more productive.  That the
>media used on the web encode more kinds of meaning that
>the speaker and hearer can beneficially share, in encodings/representations
>that are efficient for the speaker and effective (low misunderstanding
>rate) for the hearer.

Interesting, but nothing at all to do with the SWeb project/goals. 
You are still talking about human/human communication here. The SWeb 
goal is to perform work without human intervention or communication 
being necessary. Perhaps 'inference web' would have been a better 
term, but we are stuck with 'semantic web' now.

How about calling your goal the 'semiotic web'? (Half serious 
suggestion: it might catch on and be useful. The world needs both 
semiotics and semantics.)


Thanks for the feedback, as it reveals a profound difference between 
usages of the S-word. We will all get very confused if we don't keep 
these differences in mind.


IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
http://www.ihmc.us/users/phayes      phayesAT-SIGNihmc.us
Received on Monday, 9 June 2008 20:05:14 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:34 UTC