Re: <q> and the Chicago Manual of Style

2008/10/30 Robert J Burns <rob@robburns.com>
>
> [...] Even with those narrow constraints it is clear that quotations have
> diverse presentational styles that require an element such as the q element
> to provide authors and users the benefits of the separation of concerns.
>

Glad you think so :)


> So basically the Chicago Manual of Style says to style quotations in an
> English language document according to typically English language
> rules. This means a CSS rule like:
> html:lang(en)  q  {
>      quotes: '"' '"' "'" "'"
> }
>
> q:before { content: open-quote }
> q:after  { content: close-quote }
>

Fortunately, some browsers are already implementing rules along these lines
by default, though I think there is probably some way to go before sensible
defaults have been implemented, by all major browsers, for more than just a
handful of languages.


> It may also make sense for UAs to apply quotation styles to a document more
> based on the locale of the user than the root language of the document. In
> this way, the default UA stylesheet could be drawn from the software's
> localized strings so that regardless of the language of the document, the
> user saw quotations in a style they were most familiar with. Of course users
> could also make use of user stylesheets with the important keyword. I'm not
> saying this is how it must happen, but I think it is not entirely clear
> whether the presentation of quotations is:
>
> 1) a locale-dependent presentation (the locale/culture of the user /
> consumer)
> 2) a document language dependent presentation (the Chicago Manual of Style
> implication though the CMS is wholly focussed on English language documents
> presented to English language — largely American even — users / readers);
> 3) the context of the quoted foreign language within the immediate
> surrounding language (even if this is different than the language of the
> document on the whole)
>

Authors must have the final say in the presentation, I think; otherwise,
they are disempowered.

Authors who don't require a custom style will probably be happy for quotes
to render according to the most popular convention* for the language they
are marking up. The CSS rules you've given above may well represent the most
popular convention in English, for instance, and might indeed be a
reasonable default for this language. It it these conventions that I think
should be expressed in the default styling for <q> recommended by HTML 5.

I am not aware of any popular conventions that require the use of local
quote marks for foreign texts; in other words, French documents viewed in
English should retain French styling, and so on mutatis mutandis for all the
other cases I'm aware of. However, there may be cases that I'm unfamiliar
with which would require locale-based styling. This is something I hope will
become clear** in due course.


> Aside from the proper quotation marks to use in presentation involving
> quotation marks,  the other issues raised by Benjamin Hawkes-Lewis [1] could
> equally be addressed by CSS enhancements.
>

I think you're correct, though I'm not sure *all* of the issues Benjamin
raised would need to be handled by a default set of rules for rendering <q>.


> a) kerning of trailing ::after quotation marks over trailing punctuation
>

I'm not convinced this is necessary (though I may be open to persuasion).

b) omission of terminating quotation marks (e.g., French and Russian)
>

CSS 2.1 probably would need to be extended if it were to be able to handle
this without author intervention. Not knowing how prevalent this convention
is in those languages, I can't yet judge whether this would be needed for a
default <q> style sheet, but I rather hope it would become clear**. (NB. For
anyone reading this who may be concerned that the kind of "handling" I
mentioned earlier in this paragraph must be heuristic, let me say now that I
think it need only be deterministic.)

c) line marker quotation marks (e.g., French) which may already be handled
> by proposed CSS line markers
> d) block presentation of lengthy inline quotations
>
>  • for issue (a), the simple syntax of:
>         q:after  { content: kerned-close-quote }
>
> might indicate to use the specified close quotation mark, but kerned out
> (to the right for ltr) over trailing punctuation.
>
>  • for issue (b):
>       html:lang(ru) q:after,
>       html:lang(fr) q:after  { content: collapsed-close-quote }
>
> could indicate that repeated close quotation marks should be collapsed into
> a single quotation mark (perhaps "content: kerned-collapsed-close-quote"
> also)
>
> issue (c)
>
> lang(fr) > q::line-marker { quotes: '' '' '«\2009' '' }
>
> as Benjamin suggested.
>
>  • for issue (d)
> html {long-quote: 100 words} /* or alternately {long-quote: 500 chars }
>  though this introduces new units to CSS as well as a new property */
>
> and then a pseudo-class
>
> q:long-quote {display: block}
> q:long-quote:before {   content: no-open-quote}
> q:long-quote:before {   content: no-close-quote}
>

Those all seem to be promising ways of handling the issues.

However, I'm concerned not to put the cart before the horse. I think that
it's worth first of all seeing if the sort of proposals I've been advocating
(and which in some cases others, including yourself, have been advocating) -
that is, keep <q>, and have HTML 5 specify (or link to a specification
defining) default presentation for the <q> element, preferably in an
i8n-sensitive manner - have enough support that they will be implemented. If
so, the next step is to identify what those rules should be. The third step
would be to express them. If this expression happens (partly) in CSS, then
the existing CSS rules may turn out to be inadequate, in which case CSS
extensions like those you've proposed above may need to be agreed before the
drafting of the default <q> style rules can be completed.

So, to use another cliché, let's not jump the gun :)


> [1]: <http://lists.w3.org/Archives/Public/www-style/2006Sep/0141.html>
>

Regards,

Sam

*I realise the phrase "the most popular convention" suggests a quantitative
comparison of conventions. While I think this might be useful if it really
isn't clear**, for a given language, what the most popular convention is, I
hope that in many cases either it will be fairly obvious what the most
popular convention is or else there will be a national standard that
expresses what the convention ought to be. Either of these, I think, would
be adequate. (For comparison, <li> elements in <ul> elements typically
render black dots because this is a popular convention for indicating
unordered list items, but it wouldn't have been the end of the world if
they'd typically rendered black squares instead, because this is also a
popular convention, and in any case, the user can override the default
presentation if needs be.)

**Clear, that is, to the group (which I hope will be determined by the HTML
5 WG in the not-too-distant future!) tasked with: (1) identifying the most
popular convention for each language in scope and (2) documenting those
conventions for the benefit of UA vendors - or, at the very least,
referencing other documents which express them (for the latter case, ISBN
2743304820 would be an example of a document that could be referenced).