Re: Speakers in Dialogues (Was: what is dt?)

Jeremy Keith writes:

> Stephen Stewart wrote:
> 
> > > I concur completely with both your exclamation point and your  
> > > question mark. It's nutty advice that will be ignored by authors.
> >
> > How is it "nutty" and why will it be ignored by authors?
> 
> It is nutty because it suggests that a semantically empty element (which 
> is how the <b> element is now defined) is somehow suitable for marking up 
> the semantics of a person being cited in a dialogue ...

John Foliot recently pointed to an example of dialogue on a web page:

  http://www.cssquirrel.com/comicscripts/script36.htm

That doesn't use any particular mark-up to convey which are speakers to
readers; that can be inferred by being the text at the start of each
bullet point before the colon.  Users don't seem to be inconvenienced by
browsers not knowing that those parts of the text are the speakers (any
more than they don't know which words in the text are adverbs, or which
have their roots in Latin).

As it happens the speakers are marked up, with <span class=speaker>.
<span> is definitely a 'semantically empty element'; a user agent
without specific knowledge of what class=speaker denotes on that
particular page knows nothing about those bits of text.

Using <b class=speaker> instead would at least mark it as "a span of
text to be stylistically offset from the normal prose without conveying
any extra importance" -- user-agents would know the text is in some way
'special'.

But I think using <span> (or not using anything at all -- which for the
above page would leave it with identical rendering) is fine.  Ditto for
the choice of using <li>-s rather than <p>-s.

Perhaps the spec could soften its encouragement to authors over
conversations, merely mentioning <p> and <b> as one plausible way of
doing it.

> ... while obstinately refusing the accept that the existing practice
> of marking up cited people with the <cite> element is a viable option.

As defined by HTML5, a user agent can treat the contents of a <cite>
element as being the title of a work; if <cite> is expanded to do two
distinct things (both titles of works and conversation speakers) then
<cite> effectively becomes a semantically empty element two: a user
agent can't know which of the two meanings is intended, so can't presume
either of them.

Smylers

Received on Tuesday, 15 September 2009 22:30:45 UTC