RE: Footnote discussions

Thanks! I had misunderstood. I had thought @role was restricted to a specific vocabulary. I am probably confusing it with the ARIA side of things. And yes, I'm very aware of the negatives of overloading @class (and would prefer @role for what @role really means!) but it is really common in publishing because it is seen as the only way to convey that kind of publisher- or publication-specific vocabulary. (Plus the other thing is the difficulty in getting publishers/editors/production staff to think of such things semantically rather than positionally. They just say "but it IS a footnote! And this one IS a marginal note!")

There have been discussions between the DPUB IG, IDPF, and PF about expanding the semantics for @role. I will defer to Tzviya, who has led the charge on that, and apologize to her if I have stirred something up I shouldn't have.

From: ahby@aptest.com [mailto:ahby@aptest.com] On Behalf Of Shane McCarron
Sent: Friday, February 06, 2015 3:21 PM
To: Bill Kasdorf
Cc: Matt Garrish; Liam R E Quin; Siegman, Tzviya - Hoboken; Dave Cramer; David MacDonald; Robert Sanderson; George Kerscher; public-digipub-ig@w3.org
Subject: Re: Footnote discussions

A couple of things about role:

1) Remember that it takes multiple values!  That means you can convey information to ATs (the first recognized role is what an AT will use) AND convey other sorts of information to other processors.

2) I am unaware of debate about structural semantics for role (and I am editor in the Role Attribute spec).  Can you provide some pointers?

3) While you can easily use class to convey semantics AND control layout, there are risks associated with that.  Embedded content might end up getting styled in unexpected ways, for example.   Obviously you can do whatever you want, but in general I recommend to my customers that they avoid overloading class in that way.  After all, it's what role is for.


On Fri, Feb 6, 2015 at 1:57 PM, Bill Kasdorf <bkasdorf@apexcovantage.com<mailto:bkasdorf@apexcovantage.com>> wrote:
Well, this was exactly my point. They need to be distinguished semantically, not positionally.

The reasons I made the comment about @role="note" @class="[whatever]" are:

1. There is lots of debate about the proposed new structural semantics for @role, and I'm coming around to agreeing that structurally these are all just notes.
2. It is convenient to make the content semantic distinction in the @class attribute because usually that also drives a rendering distinction. I realize that that is not purely what @class is about, but it is really practical for publishers. And the reason for the "[whatever]" is that there are any number of distinctions that need to be made, there is no way to create a CV that will work for that.

-----Original Message-----
From: Matt Garrish [mailto:matt.garrish@bell.net<mailto:matt.garrish@bell.net>]
Sent: Friday, February 06, 2015 2:30 PM
To: Bill Kasdorf; Liam R E Quin
Cc: Siegman, Tzviya - Hoboken; Shane McCarron; Dave Cramer; David MacDonald; Robert Sanderson; George Kerscher; public-digipub-ig@w3.org<mailto:public-digipub-ig@w3.org>
Subject: Re: Footnote discussions

> needing to be distinguished

For print layout, sure. But with more information, instead of simple positional statements, both print and digital can be smarter. Distinguish author and importance and your formatting software can generate the correct layout based on internal workflow rules. Persist content division on the web, where only certain types of notes are allowed to live in certain documents, and we hobble digital as a mere reflection of print.

It feels wrong to me (if you can't guess!) that in a digital world publishers should have a say in where notes appear and how. They should only be providing the context for rendering the notes and leave it to the user and their reading system to determine the most appropriate presentation for them. I don't even see this as an "accessibility" issue so much as a simple usability issue for everyone. We all benefit from better control of our reading experiences.

Matt

-----Original Message-----
From: Bill Kasdorf
Sent: Friday, February 06, 2015 2:02 PM
To: Matt Garrish ; Liam R E Quin
Cc: Siegman, Tzviya - Hoboken ; Shane McCarron ; Dave Cramer ; David MacDonald ; Robert Sanderson ; George Kerscher ; public-digipub-ig@w3.org<mailto:public-digipub-ig@w3.org>
Subject: RE: Footnote discussions

Agreed. But just to point out: for the publisher, it really isn't as much a matter of importance as that they've just got six different _kinds_ of notes, equally important but needing to be distinguished. (They only think of them in terms of position now; and yes, a degree of importance certainly influences which "kind" they call a footnote and which "kind" they call an
endnote.) That’s why I think we're talking more content semantics than structural semantics here.

-----Original Message-----
From: Matt Garrish [mailto:matt.garrish@bell.net<mailto:matt.garrish@bell.net>]
Sent: Friday, February 06, 2015 1:02 PM
To: Bill Kasdorf; Liam R E Quin
Cc: Siegman, Tzviya - Hoboken; Shane McCarron; Dave Cramer; David MacDonald; Robert Sanderson; George Kerscher; public-digipub-ig@w3.org<mailto:public-digipub-ig@w3.org>
Subject: Re: Footnote discussions

The distinctions you mention below between editorial and authorial notes ties into the OA metadata I mentioned. Same with translators and any other commentators in an annotated work. Who authored the notes is important to know, and a needed semantic to do notes "right".

I don't know that a corollary can be drawn between the author and the note position for all books, though (leaving out marginalia). Position facilitates reading, but that also varies on preference and type of work. In a general non-fiction work, I probably don't want to see footnotes as I'm reading. But in a scientific paper, I likely wouldn't want to flip back and forth to find out additional context to what I'm reading.

If position itself is important, what I think we're missing is a question of importance. Other than a note being easier/faster to reference at the bottom of a page -- which is true, but can be a toggleable feature in digital -- what makes the note so important that it has to be positioned at the foot of the page instead of the end of the section/work? I believe such semantics are likely to be unearthed (criticality of understanding v. ancillary information), but we need to dig deeper into why we use notes the way we do.

It seems like we intuitively know what we intend in publishing when we make these positional determinations, but to move away from the two dimensions of print we need to better surface the why's and have the markup carry that information.

In this dream world of mine, a formatting tool could use the semantics to correctly reposition the notes for print, too, without reliance on classes.

Matt

-----Original Message-----
From: Bill Kasdorf
Sent: Friday, February 06, 2015 12:24 PM
To: Liam R E Quin ; Matt Garrish
Cc: Siegman, Tzviya - Hoboken ; Shane McCarron ; Dave Cramer ; David MacDonald ; Robert Sanderson ; George Kerscher ; public-digipub-ig@w3.org<mailto:public-digipub-ig@w3.org>
Subject: RE: Footnote discussions

An e-mail I just got from a client I'm working with (one of the world's largest scholarly publishers, this is not an outlier) contained something pertinent to this discussion:

"On a given textual edition, there might be a combination of any of the
following:
--authorial footnotes;
--authorial endnotes;
--authorial marginal notes;
--editorial footnotes;
--editorial endnotes;
--editorial marginal notes (these are rare, but do occur – e.g., Richardson, Early Works, p. 71)."

Clearly those are semantic distinctions. They are positioned differently for a reason. The reason is what matters, not the position.

However, I think the fundamental question wrt markup is whether these are _structural_ semantic distinctions or _content_ semantic distinctions.

It's quite arguable that they're the latter. To the publisher, as long as they have a way of making that distinction in their markup, they're probably fine.

So what this might lead one to conclude would be @role="note" and @class="[whatever]".

--Bill K

-----Original Message-----
From: Liam R E Quin [mailto:liam@w3.org<mailto:liam@w3.org>]
Sent: Monday, February 02, 2015 7:41 PM
To: Matt Garrish
Cc: Siegman, Tzviya - Hoboken; Shane McCarron; Dave Cramer; David MacDonald; Robert Sanderson; Bill Kasdorf; George Kerscher; public-digipub-ig@w3.org<mailto:public-digipub-ig@w3.org>
Subject: Re: Footnote discussions

On Mon, 2 Feb 2015 16:03:02 -0500 Matt Garrish <matt.garrish@bell.net<mailto:matt.garrish@bell.net>>
wrote:

[...]
> It’s not to suggest that the distinctions aren’t real in publishing
> and don’t have meaning, but aren’t they all just variations in kind
> when we take print workflows out of the equation? [...]. Accept that
> they’re all annotations -- and that styling/layout can be so fluid in
> digital that they could be pop-ups, margin notes, bottom of page notes
> or end notes all depending on the capabilities of the user agent and
> the desire of the reader


To pull some threads together...

The nature of a footnote is that it does not require any actuation -- in print you can see it by looking at the bottom of the page, and in an e-reader one can easily imagine a dedicated footnote area anywhere on the screen.

The nature of marginalia is that it can be seen without any actuation, and also that one can use marginalia as a way t find related content.

The nature of an end note is that it does require actuation (even if in print it's just keeping two bookmarks and turning a bunch of pages, or scanning down to the end of a section).

So there is a difference in authorial intent.

That many Web designers are not familiar with the demands of producing and processing complex texts doesn't mean the distinctions are not important -- these texts are a huge challenge to put on the Web today.

I think it's useful to look at a second level of footnotes as an annotation.
For example, a critical edition of a text wthat had footnotes with dagger, star, obelisk for numbering ill often add another level off footnotes with 1, 2, 3, or even Greek callouts, as here [Member-only link, sorry; I'll try and make them public in the next few days but I have also attached it to this message] https://www.w3.org/Style/XSL/Group/2008/06/footnote-examples/pages/Hearne-Works-VolIV-447/


Here, note (1) is from the first edition, and note (*) is a clarification in this second edition (a footnote to a footnote in fact). The margin note helps people to find a passage of signifcance (in the opinion of the editor) and the 2 at upper right is a page/folio number from the original manuscript.

This is not a complex example of notes in the humanities.

Speaking from a markup point of view, and from searching and editing, one might reasonably want different markup for these different cases, perhaps yes using the work of the annotations WG - assuming they accept complex/rich text annotations that contain markup and might themselves contain footnotes.

And from an accessibility perspective, a footnote as a link isn't bad if you can reliably go back to exactly the right place; I' might also want to be able to have a sort of table of contents of a section containing only the marginalia, so as to be able to go straight to a passage of interest, and the markup would need to support this. Endnotes are in a sense already links.

The current CSS draft in GCPM does not I think handle multiple levels of footnotes (as on this sample page). I have many more recent examples but used this one because it's out of copyright. In one of the other samples in the same set there are multilingual examples, with Greek and Hebrew. An interlinear gloss is another example where I think the term "annotations"
may be OK or may seem to trivialize the importance of the gloss - the gloss is actually often the main text of the document. So in markup one might want to make the gloss - the translation written between the lines of the original - be the main text and the "original" be an annotation, that-which-is-translated. Another example would be Japanese Ruby. However, in processing the text one typically needs to be able to ignore the ruby annotations.

It might be that from an HTML perspective note class="footnote" -- Matt, they are called footnotes, it's not about the placement, deal with it eh?
:-) :-) -- would work, perhaps with ARIA roles to indicate rendering and actuation intent.

> then what becomes more interesting is the semantics of the information
> carried in the annotation, who has provided it (author, translator), etc.
> to facilitate intelligent rendering.

I think this probably goes back to HTML class and ARIA.

> The multi-reference problem would have to be overcome, although maybe
> it wouldn’t be so much of an issue if you aren’t “going somewhere” so
> much as “something opens” (i.e., that can be closed to the current
> spot). Or, if you have the choice, you could pick the rendering option
> that best works for you.

In complex texts it's not uncommon to have a single note referred to multiple times in a chapter - it should only appear once on each applicable page in print of course, but can make the UI harder for "going back".

> [...] Forcing people to read notes where they occur removes the
> ability to explore the notes after reading the text (something I’m
> fond of doing, as I hate losing the narrative thread). But perhaps
> this is a UA problem to solve if we leave rendering to the UA.

I think it probably is. However, Web browsers haven't been very innovative in hypertext so far, and ideas like content narrowing (show only...) have to be done in JavaScript today -- and seem not as common as one might have thought when the Web started.


> But that leads to the question do we really want another linking method?
> Maybe I’m not as informed on the @rel/@role attributes, but couldn’t
> one of them carry the semantic that the <a> references a note? I get
> the urge for a dedicated reference, but most of the appeal of HTML (to
> me) is its minimalism, and EPUB’s stuck with <a> without problem. It
> would be nice not to have explicit references at all, but as Dave
> Cramer noted, inlining leads to messy content models plus duplication
> for multi-references (and on to maintenance headaches...).

The problem isn't simple. So it's a case of whether the markup helps, or whether we have to solve the problems by overloading the markup.

Liam (just in from clearing snow here in Eastern Ontario)

--
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/ Pictures from old books: http://fromoldbooks.org/




--
Shane McCarron
Managing Director, Applied Testing and Technology, Inc.