Good article, but misses one very interesting detail.<p>E.g. in the example with 司る (tsukasadoru "be in charge"): the article says they "gave" the phrase a kanji. I would however assume that it happened the <i>other way</i>: the kanji was approximated with two Japanese words.<p>What's the difference? Let's go back to when kanji was adopted. The article notes Japanese writers approximated sounds with Chinese kanji readings, but there's another overlooked part: they also approximated Chinese text with Japanese words.<p>That is, traditionally they would often write in classical Chinese, but read it out loud in Japanese. Indeed, they developed a system[0] that let them retrofit an <i>entire</i> language, with a completely different sentence structure, phonetics, etc. into their own. Or, in short: they could read Chinese in Japanese.<p>This is likely where 司る comes from; some classical text using 司 in a way that was at some point best approximated by the Japanese word tsukasadoru in that context.<p>[0]: Example from <a href="https://en.m.wikipedia.org/wiki/Kanbun" rel="nofollow">https://en.m.wikipedia.org/wiki/Kanbun</a> (abridged):<p>> 楚人有(下)鬻(二)盾與(一)(レ)矛者(上)<p>> [...] the word 有 'existed' marked with 下 'bottom' is shifted to the location marked by 上 'top'. Likewise, the word 鬻 'sell' marked with 二 'two' is shifted to the location marked by 一 'one'. The レ 'reverse' mark indicates that the order of the adjacent characters must be reversed.<p>> Following these kanbun instructions step by step transforms the sentence so it has the typical Japanese subject–object–verb argument order.<p>> Next, Japanese function words and conjugations can be added with okurigana, [...]<p>> The completed kundoku translation reads as a well-formed Japanese sentence with kun'yomi:<p>> 楚（そ）人に盾と矛とを鬻（ひさ）ぐ者有り<p>Obviously, the system comes with limitations; it's more of a system to analyze classical Chinese text than a way to magically translate it into Japanese. Still, I find it the most fascinating part of the language, because you can view it as a sort of "machine translation" from a millennium before computers existed, simply by abusing the fact that they used the same sort-of-semantic alphabet.<p>This is also where the "many readings of a single word" property of kanji comes from. Modern Japanese writing is the fusion of the phonetic and semantic interpretation of kanji - kana being the simplification of phonetic forms, and kanji's weird readings being derived from kanbun-kundoku.