Kanji: When Phonetic & Reading Don't Jive

[lexico]

�C [umi; G. kai; K. kai] "sea" has phonetic 每 [G. mai; K. bai] "always; (graph) woman with head-dress."
/kai/ and /mai~bai/ are quite different in the initials, although they rhyme.
What is the cause of this discrepancy?
Possible histories of �C readings are;

1. old reading /*hmai/ characters diverging into two classes of /hai/ and /mai/ characters.

2. old character reading of 澥 (< �� [G. ke, ge; K. kai]) "sea of Bohai" being inherited by �C
when it replaced the older Warring States character ��澥 > �݊C.

In the second case, the Japanese interpretive reading /umi/ may be preserving the Archaic reading of �C.
Even that Archaic word may have been borrowed into the Chinese language
from the language of an unknown sea-faring people, non-Chinese in origin.

Have you seen a kanji with a suspicious phonetic or reading?

[lexico]

About 80% of all kanji, styled semantic-phonetic compounds �`�ߎ�, are supposedly made of two components.
For example, �] [e, kawa; kou] "river" has two components, the phonetic of which denotes the reading of the compound kanji �].

semantic component 氵[mizu; sui] "water, (graph) stream."
phonetic component �H [takumi; G. ku; K. kou] "craft, craftsman, (graph) anvil."

In this example, the /kou/ reading of �H explains the /kou/ reading of �].

Another example; �� [kawa; G. ga; K. ka] "river" has

semantic component 氵[mizu] "water"
phonetic component �� [gushi; ka] "spit, wick, (graph) hand-ax."

Here the /ka/ reading of �� explains the /ka/ reading of ��.

This is by no means a perfect system. Traditional classifications have in some cases failed to properly identify the phonetic aspect of some kanjis. I've found a couple of counter-examples, but anyone who has started learning one's kanjis can discover more, I believe. Therefore in this thread, I'd like to invite all of you to contribute any odd kanjis that you think is hard to explain by the book. Any kanji is welcome, by the way. :-)

[lexico]

All the books that I've read tell me the kanji �� [���‚�����; G. bi, K. mi] is not a phonetic compound. In fact they all say that it is a semantic compound ���ӎ�, with such flashy etymologies as "(the graph of) a big white sheep which is beautiful," or "(the graph of) a sheep on the shoulders of a man picked for an offering to the shrine." While these analyses are interesting in themselves, they leave out any possibility of explaining the sound. The sound has to be taken in blind faith. But must we believe this long-held theory without examination?

The failure to see any phonetic component in the kanji �� stems from the fact that the reading /you/ of �r [�Ђ‚�; you, G. zyou, K. shyou] "sheep, goat," is nothing close to /bi~mi/.

But let's take a look at a spin-off of �r, 羋 [ ? ; G. bi, K. mi] "the bleating of a sheep, (graph) a puff of bleating coming out of the sheep's mouth." These two characters are very similar in strokes, and hence difficult to tell apart. I have one piece of evidence to support this hypothesis. ���S��, "Commentary to the Book of Rivers," of the early 6th century Northern Wei dynasty lists a certain X�m�� as an alias for 湄��. In this example, the �m character is an obvious corruption of the now-forgotten character 氵+ 羋. Hence by analogy, we can induce that �� followed a similar route of corruption or reduction resulting in one less character stroke.

There should be plenty more kanjis like this lying out there, waiting for your pick. They're for the grabs! :)
Any one that you've noticed strange or mysterious?

[lexico]

phonetic component �� [gushi; ka] "spit, wick, (graph) hand-ax."�� [gushi; ka] "spit, wick" should have said instead

1-1. "good, permissible, passive prepositional particle; a phonetic loan from �� 'skewer, spit, wick.'"
1-2. "skewer;" synonymous with 串 [�P�: kushi, tsuranoku, wugatsu; ���: G. ken; K. kan] "skewer, shishkebab."

2. �� [gushi; ka] is itself a semantic-phonetic compound composed of

2-1. sematic component �� [�P�: kuchi < kuti, ana; ���: G. ku; K. kou] "mouth"
2-2. phonetic component 丂 [�P�: ?; �P�: kou ?] "(onomatopeoia of striking a tree) 'kyok-kyok'; (graph) woodcutter's ax" as in ���ؒ���[sic.]; ���� here is a corruption of 丂丂.