On Wed, 28 Feb 2024 at 20:53, Emanuel Han via ntg-context < ntg-context@ntg.nl> wrote:
Thank you all for your suggestions and contributions to the wiki.
I don't intend to nag, but when looking at what ConTeXt is producing, I need to state that the result is still far away from a properly typeset Japanese text.
So the nihongo script which comes with ConTeXt handles *line breaks / line wrapping*. But the line break rules defined in it need a rework, because they don't follow the standards. The standards are documented here: https://www.w3.org/TR/jlreq/#possibilities_for_linebreaking_between_characte... , and all affected characters are listed here: https://www.w3.org/TR/jlreq/tables/table_en3.pdf
We have different rules, depending what kind of character is surpassing the text width (or is in its last position).
Rule 1:
Before closing brackets, closing quotation marks, iteration marks, the Prolonged sound mark and small Kana, line breaking is prohibited.
’”）〕］｝〉》」』】ヽヾゝゞ々ーぁぃぅぇぉァィゥェォっゃゅょッャュョ etc.
The actual programmed behaviour by the nihongo script is that, if in the position which exceeds the line width, these characters jump to the next line and take the previous character with them. If they're in the last position of the line, they stay where they are. This behaviour is correct.
Rule 2:
After opening Brackets and opening quotation marks, line breaking is prohibited (but not before).
‘“（〔［｛〈《「『【
The actual programmed behaviour by the nihongo script is that these characters jump to the next line and take the previous character with them. This behaviour is wrong. They should jump to the next line without taking the previous character with them, just like any regular character. The difference to a regular character is that they jump already when still within the line length, and they're in the last position of the line. The correct behaviour can be seen in LibreOffice Writer in action.
Rule 3:
Comma (tōten), full width comma, full stop
、，。
The actual programmed behaviour by the nihongo script is that, if in the position which exceeds the line width, these characters jump to the next line and take the previous character with them. This behaviour is wrong. They have to be put back to the end of the previous line, but beyond the specified line length. (JIS Z 8125) (Search for "Line adjustment by hanging punctuation" under https://www.w3.org/TR/jlreq/ ) If they're in the last position of the line, they stay where they are. The correct behaviour can be seen in LibreOffice Writer in action.
Rules 4, 5, ...:
Combinations of inseparable characters... (see https://www.w3.org/TR/jlreq/#possibilities_for_linebreaking_between_characte... ) and eventually more, which I didn't test.
It might be useful to define three scripts nihongo_loose, nihongo_strict and nihongo_very_strict which each implement one of the 3 cases described here: https://www.w3.org/TR/jlreq/#addendum_a
According the *line gap* (Otared uses \setupwhitespace[big], which is exceeding common line gaps), I'd like to quote from https://www.w3.org/TR/jlreq/ :
*It is common that the line gap for the kihon-hanmen is set to a value between half-em spacing and the one em spacing of the character frame used for the kihon-hanmen. Half-em spacing can be chosen in cases where the line length is short, but one em spacing or close to it is more appropriate when the line length is longer than 35 characters.*
I like the standard line gap which is provided by ConTeXt, which is equivalent to \setupwhitespace*[0pt]*. Even when using ruby, it works well. I found the best voffset for ruby to be -1.7ex.
The *line adjustment* provided by ConTeXt by default is not meeting the needs for Japanese (and Chinese) text, which follow a grid pattern. Especially the last line of a paragraph is squeezed, which is "hurting the eye".
When characters need to jump to the next line due to previously discussed line breaking rules, ConTeXt seems to apply "Line adjustment by inter-character spacing expansion", which is a valid method according to https://www.w3.org/TR/jlreq/#line_adjustment , although "Line adjustment by inter-character spacing reduction" is preferred.
The last point which ConTeXt is missing, when talking about Japanese typesetting, is vertical writing.
I know, this is a lot of work. Hopefully we can achieve with joint efforts to make ConTeXt Japanese ready.
If I happen to have made false statements, please accept my apology. I tried to be of help as far as I could. I grew up in Japan and know more or less how typeset text should look like.
Emanuel
It would be nice if you can put your notes above into https://wiki.contextgarden.net/Chinese_Japanese_and_Korean or in general improve/maintain that page (e.g. the links about the fonts are broken at the moment) Perhaps with Jeong Dal ? Just to say, a few days ago I have seen https://ken-lunde.medium.com/genuine-han-unification-redux-3912b561ecae (only webp images, so a bit tricky to make a pdf) -- luigi