I'm currently developing my proficiency in both Mandarin Chinese and Japanese as part of my program of study.
In Chinese, because so many characters have semantic-phonetic decomposition, predicting an unknown character's Chinese pinyin is somewhat reliable, ignoring tonal and slight nuclear changes.
However, using semantic-phonetic decomposition as a mnemonic tool for Japanese onyomi, I'm finding, is a bit harder. This is in part due to the go-on, kan-on, tousou-on, etc. distinction, in addition to historical sounds changes in Japanese.
The general problem I'm tackling is predicting a given Kanji's on-yomi (singular or plural) based on the on-yomi of Kanji with identical phonetic component. While this is difficult, I'm wondering if it might be possible to understand differences in on-yomi between character with identical phonetic component by looking look at the given character's readings in different Chinese dialects. The assumption, perhaps wrong, is that different Chinese dialects will be closest to different historical forms of Chinese, such that dialect X will best predict a Kanji's kan-on reading, dialect Y will best predict its go-on readings, etc.
I'm leaving the number of on-yomi families indefinite here, as this post and other sources indicate that, when one considers both regional and chronological differences in the borrowing history, the situation would be more complicated.
While this is certainly roundabout approach for Japanese learning, I find it interesting enough as a phonology project, and hopefully a very good mnemonic aid, to be worth pursuing.
Given this program, in which modern Chinese languages am I most likely to find readings closest to a given Kanji's on-yomi?
Academic references for Kanji history or dictionaries that link on-yomi to their historical Chinese origins would also be appreciated.