W3C home > Mailing lists > Public > www-international@w3.org > October to December 2010

RE: [css3-ruby] [css3-text] Position values and before/after definitions in LR vertical writing mode

From: Ishii Koji <kojiishi@gluesoft.co.jp>
Date: Thu, 30 Sep 2010 22:07:25 -0400
To: Stephen Zilles <szilles@adobe.com>, Richard Ishida <ishida@w3.org>, 'fantasai' <fantasai.lists@inkedblade.net>, "www-style@w3.org" <www-style@w3.org>, 'WWW International' <www-international@w3.org>
CC: "btmnk0825@gmail.com" <btmnk0825@gmail.com>
Message-ID: <A592E245B36A8949BDB0A302B375FB4E0A385D6956@MAILR001.mail.lan>
Thank you for writing this up. This is really helpful.

First, I completely understand the needs for "auto" value, and this should be considered carefully before the spec is finalized. Your opinion is greatly appreciated. But this is a little different topic; as you pointed out, regardless of whether we have auto or not, we need a way to explicitly specify which side emphasis marks, ruby, or underlines are drawn.

There will not be the right answer anyway, because we're trying to name something that doesn't exist in English. So either answer could be true.

The same thing actually happened for Japanese and Chinese. Underline in Japanese vertical writing is drawn on right as you might know. In Japanese, it's called "傍線", which means "side-line", so neither "under" nor "over" is the correct translations. We chose to name it "overline", because "over" is correct if you look at alphabet orientations in vertical text. "before" is also a right answer because the baseline direction matches to block progression in these scripts. As a result, we've had inconsistency where underline position is under|over while ruby/emphasis mark position is before|after. But this didn't get much attention up until now.

Now that we're facing a new case where "over" defined in the context as we did for Japanese and Chinese does not match to the block progression.

Here're pros/cons as I think.

* over|under

Pros:
* Can resolve inconsistency with underline position before ruby/emphasis marks specs are finalized.
* Can introduce a new directional concept that is based on baseline. There may be more features like this in future.
* Since underline spec is finalized, it makes sense to use the terminologies to indicate positions based on baseline, rather than changing underline spec.
* The way we name it is consistent with the way we named underline for Japanese and Chinese.
* This is consistent with which direction super/subscripts are drawn, although super/sub uses different terminology than under/over, so this argument can be a little weak.

* before|after

Pros:
* Do not have to introduce new directional concept and terminologies, although underline is still an exception.
* Can keep backward compatibility with old WD.

Do you have any additions to the list? I may underestimate pros for before|after.


Note that I cc'ed a Mongolian friend of mine who's studying Mongolian word processor.


Regards,
Koji Ishii

-----Original Message-----
From: Stephen Zilles [mailto:szilles@adobe.com]
Sent: Friday, October 01, 2010 9:43 AM
To: Ishii Koji; Richard Ishida; 'fantasai'; www-style@w3.org; 'WWW International'
Subject: RE: [css3-ruby] [css3-text] Position values and before/after definitions in LR vertical writing mode

It seems to me that several things are being confused.

The advantage of "start" "end" "before" and "after" is that the depend only on the "writing-mode" and not on the script or language being used. For that reason, these "coordinate reference" may not be the best references for describing where emphasis marks go. They are, on the otherhand, clear.

Consider Mongolian in its traditional script (rather than in Cyrillic script). This traditional script is derived from a semitic (Syriac from Aramaic) script that was written right to left horizontally. The script was adapted and shifted 90 degrees to be written top to bottom (a la Chinese) by the Uighurs. The Mongol (and Manchu) scripts were adapted from the Uighur script. Now, since the script was originally a right to left script, rotating it 90 degrees meant that the block progression direction became left to right.

So, why is this important? Well, at least historically, an emphasis mark on the right of a vertical line of Mongolian script would be an "underline" (and not an "overline". Whether a modern writer of Mongolian would see this as being the case, I do not know. I know that with the adoption of the Cyrillic script for Mongolian, the language came to be written left to right, which means that the "original" letter forms of the Mongolian script are writing "upside down" in current horizontal practice.

Why this diatribe? Well, although "under" and "over" have clear meanings for horizontal writing, they do not have such clear meanings for vertical writing.

That is why I would argue the following:
(1) There seems to be a favored place for emphasis marks in east asian languages. This place tends to be, at least, language specific, and is, perhaps, even country or area specific. As has been noted, in horizontal writing, the PRC uses dots below the text and Japan uses dots above the text. Similarly, in vertical writing, PRC uses dots to the left of the line and Japan normally prefers dots to the right of the line. (These preferences are at least consistent with the idea of a right hand 90 degree rotation to get vertical text.) This means, however, that there is no good default position for the emphasis marks other than "auto". Here, "auto" means look at the language and position the emphasis marks accordingly.
(2) If "auto" is the default/initial value for emphasis position, then anyone setting an explicit position must know what they are doing. If they know what they are doing, which is more useful to them, (a) coordinate references that are fixed with respect to writing-mode or (b) coordinate references that are relative to a notion of "baseline".

The catch comes with defining the notion of "baseline relative". Consider that there are two ways to put rotated Hebrew or Arabic text in a vertical line. (The same applies for Latin script texts, but the point is easier to see with a right to left text.) One way, the way used for Latin script texts, is to rotate the letter forms 90 degrees to the right and then the Hebrew/Arabic text runs from bottom to top (to preserve its right to left-ness. The other way is to rotate the Hebrew/Arabic glyphs 90 degrees to the left and have the text run from top to bottom. Depending on which way the text is rotated, "under" is on the left (for the first way) and on the right (for the second way). What is "under" supposed to mean in this case? Since "before" and "after" do not depend on which why the rotated text is rotated, they have a clear, unambiguous meaning no matter which way the rotation goes.

Furthermore, if the text in which the Hebrew/Arabic is being embedded (in rotated form) is Japanese (or Chinese) then the author may prefer to have all the emphasis marks appear on the same side, independently of which way the embedded text was rotated. This also suggests using "before" and "after" is the better way to go.

Since positioning of the emphasis dots in Chinese ("after" in both horizontal and vertical text) and Japanese ("before" in both horizontal and vertical text) this suggest the emphasis position can be specified explicitly independent of writing-mode.

As I understand things, it is concerns about Mongolian scripts that cast doubt on this solution. I have attached an example (from a text book teaching the Mongolian script using the Cyrillic script) which does show underlining on the "right" or "after" side of the Mongolian text (as noted above, I do not know whether that is over or under). I do not have, unfortunately, any examples of Mongolian script in horizontal usage so I cannot say where the emphasis should be in that case. (Nor do I claim to have any expertise in Mongolian.)

What I do believe is that this is a complex question to answer and that we need to consider all the factors influencing a decision, including rotated text.

Steve Zilles


> -----Original Message-----
> From: Ishii Koji [mailto:kojiishi@gluesoft.co.jp]
> Sent: Wednesday, September 29, 2010 6:08 PM
> To: Stephen Zilles; Richard Ishida; 'fantasai'; www-style@w3.org; 'WWW
> International'
> Subject: RE: [css3-ruby] [css3-text] Position values and before/after
> definitions in LR vertical writing mode
>
> Well, if the issue is only about Ruby, I can agree with you. It's too
> much work compared to what you get. But unfortunately it's not.
>
> If you draw alphabets, they are rotated clockwise by 90 degree. If you
> underline them, the line is on left. If you superscript them, they
> move to right. All these indicate that, at the character level, right
> is the direction for "before", and this conflicts with the block progression.
>
> If underline is on left, naming that side as "before" is logically
> incorrect, right?
>
> Originally we had only one direction, right. We renamed it to
> "physical direction" had split it to page progression and block
> progression because we understand that in RTL and in vertical writing
> mode, they are actually different.
>
> And now we're looking at a new case where block progression differs
> from character progression. So the proposal looks to me that we need
> another level of the definition in the logical directions.
>
> start|end|before|after the block
> over|under the character
>
> well, if you have find words, I'm fine with them. But as long as
> target object is different, we need different naming system. It could be:
>
> before-char|after-char
>
> A little lengthy, and a little misleading from English point of view
> though.
>
> I hope we won't find more new cases that require yet another level of
> logical direction though :)
>
>
> Regards,
> Koji Ishii
>
> -----Original Message-----
> From: Stephen Zilles [mailto:szilles@adobe.com]
> Sent: Thursday, September 30, 2010 4:16 AM
> To: Ishii Koji; Richard Ishida; 'fantasai'; www-style@w3.org; 'WWW
> International'
> Subject: RE: [css3-ruby] [css3-text] Position values and before/after
> definitions in LR vertical writing mode
>
> I believe that the issue is not related to the baseline; In vertical
> text the baseline is often down the center. The problem with Mongolian
> seems to be that it is asserted that "ruby" would be placed on the
> righthand side of a line, which, since Mongolian is a TB-LR language would be the "after"
> edge of the line. This would conflict with "before" being the default
> position.
>
> I do not have any examples of Mongolian text with annotations. All of
> my examples lack such. I would like to see examples to have a better
> idea of the problem.
>
> If it is necessary to introduce new terms for annotation positions, I
> would certainly prefer "over/under" to "above/below" because, at
> least, the "over/under" pair relate to "overlines" and "underlines"
>
> I think that making the default value of the ruby position be "auto"
> is a much better way to handle the language differences
> "automatically", however.
>
> Steve Zilles
>
>
> > -----Original Message-----
> > From: Ishii Koji [mailto:kojiishi@gluesoft.co.jp]
> > Sent: Wednesday, September 29, 2010 11:39 AM
> > To: Stephen Zilles; Richard Ishida; 'fantasai'; www-style@w3.org;
> > 'WWW International'
> > Subject: RE: [css3-ruby] [css3-text] Position values and
> > before/after definitions in LR vertical writing mode
> >
> > The issue is that, in Mongolian, block progression does not match to
> > the baseline.
> >
> > In that case, margin-before is left, but "before" value of ruby
> > position should be right because it's based on baseline, not on
> > block
> progression.
> >
> > I agree with fantasai that we should come up with a new pair of
> > words that indicates directions against baseline. And I would vote
> > "over/under" than "above/below", as the consistency with the
> > underline
> makes sense to me.
> >
> >
> > Regards,
> > Koji Ishii
> >
> > -----Original Message-----
> > From: www-style-request@w3.org [mailto:www-style-request@w3.org] On
> > Behalf Of Stephen Zilles
> > Sent: Tuesday, September 28, 2010 1:36 AM
> > To: Richard Ishida; 'fantasai'; www-style@w3.org; 'WWW International'
> > Subject: RE: [css3-ruby] [css3-text] Position values and
> > before/after definitions in LR vertical writing mode
> >
> > I agree with Richard, especially since Ruby seems more common on
> > vertical text.
> >
> > Steve Zilles
> >
> >
> > > -----Original Message-----
> > > From: www-international-request@w3.org [mailto:www-international-
> > > request@w3.org] On Behalf Of Richard Ishida
> > > Sent: Monday, September 27, 2010 6:11 AM
> > > To: 'fantasai'; www-style@w3.org; 'WWW International'
> > > Subject: RE: [css3-ruby] [css3-text] Position values and
> > > before/after definitions in LR vertical writing mode
> > >
> > > My first instinct is to question whether there is an issue here.
> > > Is it a problem that ruby text labelled 'before' will appear to
> > > the left of vertical mongolian text?  Before and after refer to
> > > the position relative to the block progression, in my mind. It's
> > > not about top of line coincidence.
> > >
> > > I think that above and below are confusing, since they suggest
> > > physical locations that are not appropriate for vertical text.
> > >
> > > (Note that the CSS Ruby module says " vertical-ideographic layout
> > > mode, the ruby appears on the right side of the base" - mongolian
> > > isn't
> > ideographic.
> > > That could certainly be made clearer with a note.)
> > >
> > > RI
> > >
> > > ============
> > > Richard Ishida
> > > Internationalization Lead
> > > W3C (World Wide Web Consortium)
> > >
> > > http://www.w3.org/International/
> > > http://rishida.net/
> > >
> > >
> > >
> > >
> > > > -----Original Message-----
> > > > From: www-international-request@w3.org
> > > > [mailto:www-international- request@w3.org] On Behalf Of fantasai
> > > > Sent: 26 September 2010 15:15
> > > > To: www-style@w3.org; 'WWW International'
> > > > Subject: [css3-ruby] [css3-text] Position values and
> > > > before/after
> > > definitions in
> > > > LR vertical writing mode
> > > >
> > > > In top-to-bottom horizontal writing mode (English, most other
> > > > scripts)
> > > and
> > > in
> > > > right-to-left vertical writing mode (CJK), the "before" side and
> > > > the
> > > effective
> > > > top of the line (wrt vertical alignment, glyph rotation, etc)
> coincide.
> > > >
> > > > See
> > > >
> > > http://fantasai.inkedblade.net/style/discuss/vertical-text/diagram
> > > s/
> > > te
> > > xt-
> > > flo
> > > w-
> > > > vectors-tb.png
> > > >
> > > http://fantasai.inkedblade.net/style/discuss/vertical-text/diagram
> > > s/
> > > te
> > > xt-
> > > flo
> > > w-
> > > > vectors-rl.png
> > > >
> > > > But in left-to-right vertical writing mode (Mongolian), the "before"
> > > > side
> > > and
> > > > the effective top of the line do not coincide. See the
> > > > illustration
> > here:
> > > >    http://fantasai.inkedblade.net/style/discuss/vertical-
> > > > text/diagrams/mongolian-vectors.jpg
> > > >
> > > > The "before" side of a line is to the left. The "top" (ascender)
> > > > side of
> > > a
> > > line
> > > > is to the right.
> > > >
> > > > If ruby-position and text-underline-position use "before" to
> > > > mean "on the right side of the line" in vertical text, then we
> > > > have a problem where "before"
> > > > means
> > > > different sides of an item depending on what property is involved.
> > > >
> > > > Either the definitions should be updated to depend on whether
> > > > the block flow is right-to-left or left-to-right, or the
> > > > keywords should be changed to something else to avoid a conflict
> > > > in meaning. I suggest the latter, since I
> > > suspect
> > > that
> > > > the current definitions are the ones that are typographically
> relevant.
> > > >
> > > > However, I haven't encountered any really good pairs of keywords.
> > > > (Koji
> > > and I
> > > > are using "above" and "below" for now.)
> > > >
> > > > Richard, do you have any thoughts on this?
> > > >
> > > > ~fantasai
> > > >
> > > >
> > > > No virus found in this incoming message.
> > > > Checked by AVG - www.avg.com
> > > > Version: 9.0.856 / Virus Database: 271.1.1/3159 - Release Date:
> > > > 09/25/10
> > > > 18:45:00
> > >
> >
Received on Friday, 1 October 2010 02:06:07 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 1 October 2010 02:06:08 GMT