W3C home > Mailing lists > Public > www-html@w3.org > July 2005

Unstyled Content (was Re: Draft suggestion: "normal" needed)

From: Edward Lass <elass@goer.state.ny.us>
Date: Fri, 15 Jul 2005 14:17:33 -0400
Message-Id: <s2d7c59b.083@mail.goer.state.ny.us>
To: <www-html@w3.org>
Cc: <Mike-GBHXonline <mikecherim@gbhxonline.com>

Is there, or will there be, a requirement in the XHTML 2.0 WD or any
future XHTML specifications that conformant user agents must have at
least CSS Level 1 support?

There seems to be a recurring, and I think legitimate, concern about
the default presentation of semantic markup. The typical response is
that semantic markup isn't concerned with presentation, but the web
community looks to the W3C specifications for cross-vendor
compatibility. And if styling is not mandatory with XHTML, then who's
responsible for determining default presentation? If it's the browser
makers, wasn't that what we were trying to avoid in the first place?

- Ed.

>>> Al Gilman <Alfred.S.Gilman@IEEE.org> 7/15/2005 1:06:57 PM >>>

Laurens, thanks for getting this onto the list.

Let's look at the example.

>><p><i>A good example is needed</i>, he thought.</p>
>>/A good example is needed,/ he thought.

To me, this is not best cast as application for <em>.  I would say a
semantic --> presentation encoding of this rhetorical episode would be

<p><q class="unSpoken thought">A good example is needed</q>, he

In the default visual presentation, this could be allocated an italic 
with a selector keyed to both the element type and [at least one of] 
the class marks.

In Braille, I don't know for sure, but I do know that they have 
different tranliterations
for italics based on the reason for the italics.  Some are retained 
and some are
suppressed.  This example might be mapped to quotes by a
human transcriber.  [I speculate.]

A TTY interface lacking italic font would use quotes before and after,
and the
reader would find this entirely appropriate and natural, not sense any

A TTS stylesheet would probably preserve the same voice but diminish
the speech rate and the pitch range or modulation index.

The latter explains why <em> is a bad model for "a semantic
replacement for <i>." Phrases are set off with quotes or italics any
time that there is some sort of difference in the rhetorical
world-graph: quotes, asides, idioms, etc. Quotes and italics are used
very broadly to set something off from its background. But the intent
is as often recessive as protrusive. So "replace <i> with <em>" is a
semantically broken rule and has caused people to use <em> visually
as a euphemism for <i> without regard for emphasis connotations.

The use of 'class' need not be un-semantic.  What it takes is some form
publishing interpretation guides for the classes that your are using, 
in a way that is accessible to the Semantic Web.

It should be understood that <em> as it is currently understood is a 
binary flag
and that nested <em> elements can be styled differently, but most
and authors wouldn't know what to do with them.

For this example I claim that <q> is a semantic fit for mental asides
an overt dialog with 'class' being used to indicate that the language
is unvoiced, only perceivable by the uttering persona in the dramatic
(or dialog world-model).

But beyond that, the misfit between <em> and <strong> and the many
uses of intalics and quotes in classical typographical style leaves us
with a
challenge.  We need a <change> semantic and the best we have in
in HTML is to use <span> with changing 'class' attributes that are as
as we can make them so that stylists in different delivery contexts
these stylesheets without access to the original content-text author]
have a reasonable shot at defining appropriate presentation in the 
new situation.

The existing <em> and <q> etc. elements do fit some of these cases,
should be used where they fit.  But this example is a <q> and not an
and not all the rhetorical modulations that get styled with italics and
fit an existing element.

I realize that trying to come up with a natural-langauge-neutral
interlingua will result in culture wars.  On the other hand, I am sure
the Computational  Linguists are far ahead of us on this in their
models for the production of prosody.


'normal' is here a presentation effect.  Your semantics for something
to be set off within an italicized or othewise set-off span is

<context>Blah <different>blah <different>blah</different>blah 

"revert to normal presentation" is just in this case a presentation
effect to
convey the "yet different" or innner 'different' semantics when you
run out of stylistically appropriate alternatiives for modulation.  In
production you would probably not have run out of presentation aspects
to tweak.

So the real problem is nested <different> sections.  Since you *can*
write context-aware selectors, this is not a fundamental problem.

You are left with an authoring challenge to cast the rhetorical
for why this section is different, why it is to be distinguished in
presentation.  Don't call that 'normal' when it is "yet different."

But I am with you in resisting casting this example to <em>.  It would
nice to reserve <em> for cases that should be afforded a protruding or
attention-grabbing presentation variation, as opposed to a recessive

I would find it hard to lay down a hard and fast rule for when to use
<em class="rationale"> vs. <span class="rationale"> among the various
rhetorical uses of some sort of <different> indication.

But here the actual rhetorical situation of an unspoken aside, a
is a good-enough match to <q> and a less-good match to <em> IMHO.


At 4:41 PM +0200 7/15/05, Laurens Holst wrote:
>Mike-GBHXonline schreef:
>>Hello Laurens,
>>  I'm absolutely right on target with this, and I managed to 
>>convince some very big names in the web standards movement of this 
>>need so please hear me out here. I will try to explain this in a 
>>more understandable way:
>>  Of course I can use CSS to lend a visual cue to <em> within <i>, 
>>that's not the issue. But tell me, what does one see when CSS is 
>>unsupported or has styles disabled, or they are using their own 
>>style sheet? That's right. We've lost value and meaning.
>Neither <i> nor <normal> have or would have meaning, they only 
>convey a presentation, not semantics. So who are you to critisise :).
>Anyways, I don't agree - emphasis within emphasis could indicate 
>more emphasis, if there is an appropriate case. There is nothing in 
>the specification that says this cannot be. If you want specific 
>visual representation for that, you should so that with CSS, there 
>is nothing wrong with that.
>Existing user agents might mark up nested <em>s with either normal 
>style, oblique style, or just italics style making it 
>indistinguishable to the eye of the user (the latter of which most 
>probably do, but that doesn't make it right). In any case, the 
>semantics, and therefore the ability to express it, are still there.
>>In the print world a thought is to be written as italic text. Not 
>>as a quote, but simply italicized. This is the/ standard/ in the 
>>print world! Thus, when an author wants to provide emphasis within 
>>that body of italicized text, he or she make that word or group of 
>>words non-italic. Normal. Not bold, they would only do that if the 
>>meaning was strong emphasis. Look at some novels and you will see 
>>exactly what I mean. I'm talking about a/ standard/ universally 
>>accepted and widely used.
>That may be so, but nevertheless it is NOT semantic. An element such 
>as <thought> would be appropriate.
>Also, it may be the standard in the print world, but if I want to be 
>the odd duck and style my thoughts in a non-standard way (e.g. bold 
>with a yellow background), I should not have to change the markup 
>for that.
>So, to put something like that on the web, a story or screenplay, 
>per se, with some "thought" content we italicize. So far so good, we 
>can write <i></i> and that is valid and semantically correct (not 
><em>). This is true of those with assistive technologies, visual 
>browsers, the works. With styles on or not... no matter.
>I'm not saying that you should use <em> when it is not emphasis. 
>Sorry if I misunderstood you.
>In HTML, use <span class="thought"> then. That is just as 
>meaningless as <i> is, except that doesn't convey a specific 
>><p><i>A good example is needed</i>, he thought.</p>
>>/A good example is needed,/ he thought.
>Right, I gather that <em> is not what you wanted there :). Anyways, 
>HTML does not have an element for 'thought', to my knowledge, so it 
>is impossible.
>With regard to your statement "But tell me, what does one see when 
>CSS is unsupported or has styles disabled, or they are using their 
>own style sheet?" earlier, the answer to that is: nothing, because 
>HTML does not define markup for 'thought'. Simple.
>Using <i> for that purpose is just abuse. And suggesting to 
>introduce <normal> of all things (!) is even worse, instead you 
>should address the actual problem: find or search a replacement for 
><i> itself, to mark up thoughts. And mark up emphasis inside just as 
>usual with <em>.
>>So, now let's emphasize a word within that body. Not strong 
>>emphasis, but light emphasis, instead. Strong emphasis is supplied 
>>to us already. So, visually we must /not/ italicize that word -- 
>>thus putting <em></em> makes it drop from sight for the sighted 
>>visitor. Meaning and value: gone! To those with assistive 
>>technologies, they get it. They hear the emphasis.
>Well then:
>.thought { font-style: italic; }
>.thought em { font-style: normal; }
>>However, if there was a reverse or normal emphasis, we could use 
>>that and thus the sighted and not sighted could get the emphasis 
>>out of the thought. So, I suggest a new element, let's call it 
>><nem> instead of <normal>. It's more accurate, perhaps.
>>  <p><i>A <nem>great</nem> example is needed</i>, he thought.</p>
>>/A /great/ example is needed,/ he thought.
>>  *Proposed Element and Rule:* /<nem>/ - /Normal emphasis/. To be 
>>used as a method of providing light emphasis within a body of 
>>italicized text. Typically used by writers to indicate passages of 
>>thought or reflection.
>That's great. But there is no single reason that you cannot use <em> 
>for that. You are not looking at the cause of the problem.
>>I cannot do that with CSS and retain value to all visitors. Think 
>>of a table's "colspan" if that works for you. And simply not 
>>italicizing the emphasized word to show emphasis doesn't do it for 
>>the non-sighted.
><em> does not necessarily have to be styled italic...
>>Look, I'm not wasting my time, nor am I wasting yours. If there 
>>wasn't a real need here, a loophole within the spec, then I 
>>wouldn't be asking for this. But that's not the case. And please... 
>>nested <em</em>s?! What on earth are you thinking. First of all 
>>you're telling me to use <em> not for emphasis but the style text 
>>italic. Please, spare me. I should use <i>, not <em>, and as 
>>explained, if one puts <em> with the <i>, the unsighted get it but 
>>the sighted do not.
>If you are going to address me in this way, I do not think I want to 
>have this conversation.
>I made a simple mistake in that I confused your 'thought' markup 
>with simple emphasis. But then again, you did the same, by confusing 
>emphasis with either <i>, <normal> and <nem>. And by confusing 
>thought with <i>.
>And you weren't exactly explicit about that in your original 
>message. I do not think it is odd that talk of <i>, <normal>, and a 
>message sent in wild colours and fonts, causes me to think that it 
>is sent by someone who is still stuck in the era of tables and 
>spacer images, and needs some explanation about semantics. And not 
>unrightly so, it seems.
>>Please reconsider this.
>I am not pulling the strings here.
>>As I said, I've managed to make some of the biggest names in the 
>>web standards movement realize that this /is/ a legitimate need. 
>>Quite simply because it is -- there are simply no options available 
>>in order to do this correctly so /all /visitors get the meaning or 
>>value. The element /is/ needed, of this I am 100% certain.
>I recognise the need, but do not think your solution is a good one.
>>I do suggest passing this email around to those who decide and this 
>>email likely explains it in more depth.
>I will forward it to the www-html mailinglist.
>>Don't make me have to start quoting Tim Berners Lee on the 
>>universality of the world wide web for you.
>Calm down.
>Ushiko-san! Kimi wa doushite, Ushiko-san nan da!!
Received on Friday, 15 July 2005 18:25:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:16:03 GMT