Japanese text does not use any kind of spaces between characters to define word breaks or whatsoever.
Thus WordWrap never manages to break a japanese (or other east asian) text.
In my opinion, the WordWrap node needs an additional inlet (“Force line break”), that breaks words when they are longer than one line. This would help for long words and would be sufficient to break japanese text (yes, I know that you should cannot break japanese text at any point, but these cases are rare).
Of course, we could also work with glyphinfo (which does not return any size information for jp characters…?) and define a CharacterBreak Node, but that would be too much overhead I think.