Re: summary on shearing style for ruby

I think when we made InDesign 斜体 we consciously decided ruby were not to be transformed because they are annotations of the text and harder to read I’d made 斜体 as well.
Also, people sometimes think 斜体 is the same as shearing but it is not.

-Nat
________________________________
From: 木田泰夫 <kida@mac.com>
Sent: Monday, July 20, 2020 5:37:42 PM
To: W3C JLReq TF <public-i18n-japanese@w3.org>
Subject: Re: summary on shearing style for ruby

Thank you for Shinono-san for sending out this detailed report.


Fuqiao said:
> I wonder if there is plan to write this into a
> future version of jlreq, or are layout requirements for subtitles
> considered out of scope for jlreq?

yes, it is out of scope of the current JLReq. It seems subtitles are closer to design elements than composition of multiple lines (Illustrator vs InDesign in Adobe language? :).

With that said I see benefits of maintaining important information related Japanese text layout and making them discoverable. Will discuss in TF.

best,

- kida

> 2020/07/20 23:57、Fuqiao Xue <xfq.free@gmail.com>のメール:
>
> Hi Atsushi, all,
>
> Thanks for the summary! I wonder if there is plan to write this into a
> future version of jlreq, or are layout requirements for subtitles
> considered out of scope for jlreq?
>
> Fuqiao
>
> 2020年7月20日(月) 16:59 Atsushi Shimono (W3C Team) <atsushi@w3.org>:
>>
>>  hi all
>>
>>  This is a summary of off-list conversations on shearing style for ruby, relates to feature
>> request from timed-text [1].
>>  Japanese text follows. Please redirect discussions (in Japanese, esp.) to public-i18n-japanese.
>>
>> *1 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fw3c%2Fi18n-activity%2Fissues%2F775&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=of1PmvX1B4Q%2FSOSRTcmADuPBlpVE6GKdE17Sx0hWqVk%3D&amp;reserved=0
>>
>>
>> Background:
>>  In subtitle, there is an order [2] to apply shear style to lines other than voice on screen.
>> Within this order, there are two ways in Japanese subtitle as an image in [3], middle panel
>> shows shearing by block of ruby base text and annotation, bottom panel shows shearing individually.
>> In current CSS spec, shearing on ruby will be applied as bottom panel.
>>  In timed-text ML (TTML) spec, in addition to general italic feature (fontShear), shearing
>> over entire one line (lineShear [5]) is provided to enable such feature.
>>
>> +2 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpartnerhelp.netflixstudios.com%2Fhc%2Fen-us%2Farticles%2F215767517-Japanese-Timed-Text-Style-Guide&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=eUtc5kW13JWAzMmDieENAS5hCGatkbPq00GXmTLhmFo%3D&amp;reserved=0
>> *3 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2FJunTajima%2Fstatus%2F1281439682513920000&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=z8%2B7%2BLYrguQEVvW0pQ3sjZDdewolBgxuplYy1Y%2FCEaA%3D&amp;reserved=0
>> *4 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fnetflixtechblog.com%2Fimplementing-japanese-subtitles-on-netflix-c165fbe61989&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=nX3avvp8YjiW8ANFpxC273afC%2BZxpEpYHVGjABU7Rmw%3D&amp;reserved=0
>> *5 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.w3.org%2FTR%2F2020%2FCR-ttml2-20200128%2F%23style-attribute-lineShear&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637308886866774618&amp;sdata=vCAPmTWyOw7xn%2FpRcbKaw0Bd5yKmZN3U3UUHHEN77wI%3D&amp;reserved=0
>>
>>
>> Use case:
>>  Looking around some subtitle authoring tools in Japanese, one manual [6] shows possible
>> option for this, that an image in p43 shows an option labelled as "move ruby annotation
>> part to match with ruby base text when sheared". (Not sure on how wide this option is used.)
>> In Netflix [4], all examples show shearing individually.
>>  I could not found any line for ruby+italic in ARIB STD-862.
>>
>> *6 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fcanvass.co.jp%2Fwp-content%2Fuploads%2F2019%2F01%2Fsstg1pro_manual.pdf&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=f9vGD1%2BzqjvrgDWydr%2FKDBW%2FZILgUMjXwme6CJjrzm8%3D&amp;reserved=0
>>
>>
>> Wider input:
>>  Tajima-san kindly posted on this issue as [3], and got following feedbacks. (see Japanese
>> part for details)
>>
>>
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2Fworks014%2Fstatus%2F1281445686530027520&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=0rQ14uEtj8eN6bQcZ%2BljtObIwj73w6JfutP1Lk5qcO4%3D&amp;reserved=0
>>> he points to like middle example, wants to see as one block
>>
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2Fpictex%2Fstatus%2F1281451158612008960&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=UYSEjCjh26NqoAl2j%2BJ07rt6ywYsVs3GBDAPf46LBXo%3D&amp;reserved=0
>>> In Japanese, italic does not have semantic meaning and just a design, so usually ruby is
>> not applied there. Some lawless area like advertisement could have, and could be moved as
>> a part of balancing.
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2Fpictex%2Fstatus%2F1281453325414576128&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=bjNNuQVuKJ8FR%2FmJCd%2F8TcQj0vC4JZR4V3rwCNGuJps%3D&amp;reserved=0
>>> In subtitles, ruby+italic could happen, but no meaningful difference between middle and
>> bottom. Also such case should be quite rare.
>>
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2FNagahisa_design%2Fstatus%2F1281468860701368320&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=kNdG7gixDvpshPnrogDxEd%2B5Dq7KhFzeUMtsfdlzLrs%3D&amp;reserved=0
>>> Considering Japanese reads ruby part as one line (or ruby annotation part is a part of line
>> of ruby base text), middle panel seems natural.
>>
>>
>> (Note: no conclusion from this conversation, just confirmed former ones like italic is minor
>> in Japanese and especially ruby+italic itself is not used commonly, and see no reason for
>> further research as Japanese and/or CSS point of view)
>>
>>
>> -------- Japanese
>>
>> 背景:字幕では[2]のように登場人物以外の発話などについては斜体にするという運用が行われているよ
>> うです。この中で、ルビ付きの部分について、[3]にある画像のように中段のルビ文字を含めて全体をブ
>> ロックとして斜めにする運用と、下段のようにルビ文字・親文字それぞれについて斜体にする運用([4]の
>> fig.12,13,18のような感じ)の両方があるそうですが、現状のCSSでは斜体にした場合は下段のようになり
>> ます。
>>  字幕用のtimed-text ML (TTML)では、これも含めた要求の実現のためにitalic (fontShear)以外に行全
>> 体を斜体にするスタイル(lineShear [5])を追加しています。
>>
>> +2 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpartnerhelp.netflixstudios.com%2Fhc%2Fen-us%2Farticles%2F215767517-Japanese-Timed-Text-Style-Guide&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866774618&amp;sdata=eUtc5kW13JWAzMmDieENAS5hCGatkbPq00GXmTLhmFo%3D&amp;reserved=0
>> *3 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2FJunTajima%2Fstatus%2F1281439682513920000&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=srRFO1WDBXFYlwJoYB6hAyKw1m0w5NFYTOZ8wAMAhAo%3D&amp;reserved=0
>> *4 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fnetflixtechblog.com%2Fimplementing-japanese-subtitles-on-netflix-c165fbe61989&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=ZuS8M0Ap1gQu%2B%2BG5MGgqmPBKqNlozfMno0hOfuGIBSk%3D&amp;reserved=0
>> *5 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.w3.org%2FTR%2F2020%2FCR-ttml2-20200128%2F%23style-attribute-lineShear&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637308886866784571&amp;sdata=X9qUe4VxDE%2FU5MFGcVEVMqjo3Kp%2FQW%2F3Mgoriq8NXfo%3D&amp;reserved=0
>>
>>
>> 利用状況:日本国内の字幕のオーサリングツールについて市販品をザクっと見た限りでは、とある製品
>> のマニュアル[6]のp.43に"斜体時に本文に合わせて移動する"というオプションがあるので、これが該当
>> かなと思われました。どれくらい利用されているのかなどについては不明ですが。前出のNetflix [4]で
>> は全部ルビ文字・親文字のそれぞれを斜体にしているキャプチャーが出ています。
>>  ARIB STD-B62では特にruby+italicの場合についての記述は見当たらず、です。
>>
>> *6 https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fcanvass.co.jp%2Fwp-content%2Fuploads%2F2019%2F01%2Fsstg1pro_manual.pdf&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=RpyPfOE0Q4qZPWy1jEa078mEwWthX1%2Fnvn9ECAYpZAA%3D&amp;reserved=0
>>
>> より広い見解:田嶋さんに[3]のようにポストをしていただいた限りでは以下のようなコメントがありま
>> した。
>>
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2Fworks014%2Fstatus%2F1281445686530027520&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=hFXlOkw5zaNiwPmJ9ho3zvCUplW%2BTy%2BPPbnS74lFL18%3D&amp;reserved=0
>>> なんでやねんDTP/おぢん / @works014
>>> 好みとしては真ん中のセンターが揃っているのがベターかと…ひとかたまりとして考えたい(目に入ってくる)…
>>
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2Fpictex%2Fstatus%2F1281451158612008960&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=KK5%2Fh7ynOvU0YJUm9Dqs8xeak0vG3WjGD1EJi1qGrTY%3D&amp;reserved=0
>>> 深沢英次 / @pictex
>>> 和文の場合、欧文と違って斜体にセマンティックな意味が無く、斜体は単に見た目のデザイン処理として扱われます。だから通常はそこにルビを付けたりはしませんね。ただ広告などでは「なんでもあり」なのでデザイナーの趣味の範疇ですが、ルビの前後位置をバランス調節することはあり得ると思います。
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2Fpictex%2Fstatus%2F1281453325414576128&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=BPlBkTP1ppcV9ekmneGLL9wGmAy69ZEE2xYeK2osPkY%3D&amp;reserved=0
>>> なるほど。字幕なら「外国語」だったり「引用の読み上げ」などを斜体で表記することはありそうですね。無理矢理ルビを入れるケースもあるかな。そういう使い方であれば、真ん中と下に有意な差は無い、と考えていいと思います。そもそもがレアケースですし。
>>
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2FNagahisa_design%2Fstatus%2F1281468860701368320&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=umJtUjhjQdtzsWZ%2BwLyd8rNjrWxKIkvkZBdA24Y%2FZfY%3D&amp;reserved=0
>>> 長久雅行/nagahisa / @Nagahisa_design
>>> 通常は、この図の真ん中の例のように組むのが自然に見えると思われますし、そのように組む場合が多いのではと思います。
>>> ●このことは、われわれ日本人がルビもその行の一部、2行ではなく1行として、見ていることとイコールだと思います。
>> https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Ftwitter.com%2FNagahisa_design%2Fstatus%2F1281469345852358656&amp;data=02%7C01%7Cnmccully%40adobe.com%7C21b1fd3ec1594dc143f708d82d0e54b4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C1%7C637308886866784571&amp;sdata=Ec0j6aiq1tUTe0XpAbzY0GIGhzjTQmCM08CjWVozs7Y%3D&amp;reserved=0
>>> ルビではなく複数行に渡る文章にすべて斜体がかかっている場合、次の行を斜体に応じた角度でずらしていくのではなく、垂直になるように組むのが普通です(当たり前ですが)。そのような考え方で行くと、図の下のように組むべきだ、となる気がしますが、ルビとその行は「1行」と考えるべきだと。
>>
>>
>> (なお、従来のitalicはマイナー機能だし、しかもrubyに掛けるようなことはまずしないだろうし、block
>> として傾けたい要求は非常にニッチなものである、という解釈&これ以上の追加調査に意味はなさそう以
>> 上に、特に結論はありません。)
>>

Received on Tuesday, 21 July 2020 00:49:53 UTC