My feeeling is that half-width kanas behave like Latin letters and do not even have to follow the ideographic composition square to line up with them (unlike standard kanas). So effectively their line breaking behavior is very different.
Those "half-width letters" are in fact similar to linear jamos (not composed into syllabic squares) in the Korean script, and to Bopomofo letters. And may be we could add the CJK key letters (radicals used for example in IDS) to this list, or Yi radicals. They are harmonized to be used along with other alphabetic scripts. In fact they may even not be really "half-width" but proportional. They are also used with non-ideographic punctuation (notably the ASCII punctuation) and standard SPACE (U+0020). If rendered in vertical lines, they could be either rotated (just like Latin letters), or not (aligned horizontallly like letters in columns of crosswords, but they may also have proportional height, like in Latin/Greek/Cyrillic where it is sometimes needed for example with capital letters with stacked accents, or when using sized spaces) So IMHO, those "half-width" letters are in fact to be considered as another separate script, for typographic purpose. They are "unified" with non-halfwidth letters, only for collation with minor differences (plain-text searching and sorting). 2015-04-28 4:20 GMT+02:00 Makoto Kato <m_k...@ga2.so-net.ne.jp>: > Hi. > > http://www.unicode.org/reports/tr14/proposed.html#ID defines Ideographic > (ID). Although full-width katakana is included in ID, half-width > katakana (U+FF66 and U+FF71-U+FF9D) isn't. Why? > > Also, Conditional Japanese Starter (CJ, > http://www.unicode.org/reports/tr14/proposed.html#CJ) considers > half-width variants such as half-width katakana letter small a. > > > -- Makoto >