Can we predict if a kanji has a reading that ends in long う

Question

Can we predict if a kanji has a reading that ends in long う? By "long う" I mean that the preceding mora also is in a u-row syllable such as ふう, くう, つう.

Consider these. I will list the kanji followed by its Mandarin Chinese pronunciation followed by its Japanese pronunciation:

• 風 fēng→fuu　（ふう）
• 空 kōng→kuu　（くう）
• 龍 lóng→ryuu　（りゅう）
• 通 tōng→tsuu　（つう）
• 痛 tòng→tsuu　（つう）
• 勇 yǒng→yuu　（ゆう）
• 中 zhōng→chuu/juu　（ちゅう・じゅう）
• 終 zhōng→shuu　（しゅう）

It seems that it starting consonant is maintained (except zh which seems to be able to go to ch, j, sh) while ng maps to the long う sound. It also seems that the vowel sound in Mandarin is limited to e and o for the penultimate mora in Japanese to also be in the u-row.

And if the vowel were a (e.g. 方 fāng→ほう, 放 fàng→ほう, 講 jiǎng→こう, 郎 láng→ろう, 浪 làng→ろう, 様 yàng→よう, 障 zhàng→しょう) then it seems to take a おう ending instead of うう.

In summary this is my hypothesis:

• The ng ending in Chinese corresponds to a final う mora in Japanese.
• If the vowel is a in Chinese it corresponds to a penultimate o-row mora in Japanese.
• If the vowel is e or o in Chinese it corresponds to a penultimate u-row mora in Japanese.

Restating the question, what are the sounds that allow me to predict if a kanji reading ends in a う sound with the penultimate mora in the u-row in Japanese?

Answer 1

• The ng ending in Chinese corresponds to a final う mora in Japanese.

Generally, except becoming い after a front vowel え, Mandarin -ng it almost always corresponds to final う, and -n to ん.

But there's a very few words which have undergone irregular development in Modern Chinese break the rule: 肯 (こう but kěn), 貞 (てい but zhēn), 馨 (けい but xīn) etc.; they're all used to be -ng group in Middle Chinese.

• If the vowel is a in Chinese it corresponds to a penultimate o-row mora in Japanese.

If you narrows down to the -ng group characters and 漢音・呉音 reading, yes. More recently introduced words after the establishment of moraic nasal were transcribed with ん regardless of -n or -ng: 行灯 (あんどん xíngdēng) etc.

• If the vowel is e or o in Chinese it corresponds to a penultimate u-row mora in Japanese.

I wish it were that clear-cut, but things don't seem to be so easy. I found a brand-new Ph.D. thesis on comparison of Mandarin and Sino-Japanese pronunciations of N2 kanji.

Conclusion first, you can only have probablistic guess whether a kanji has -おう or -うう ending when its rime in Mandarin is -eng or -ong (pp. 108, 118, 225, 230). The major reason is the merge of MC 東韻一等 and 東韻三等 characters in ModC. The main source of ModJ -うう kanji in 漢音 is 東韻三等 (pp. 35, 215), but since Mandarin (as well as most Chinese dialects) has lost their distinction, we have no clue to tell them from -おう kanji today.

To add to it, we must consider some other factors:

• 呉音 intervention

  There are small, but considerably common kanji almost always read in 呉音. In your examples, readings of 空, 龍, 通, 痛 and 勇 are actually 呉音 (漢音 are こう, りょう, とう, とう and よう, respectively). As in the paper above, the correspondence between MC rime groups and 呉音 is much more chaotic if not irregular.

  ^{(Also don't forget 重: じゅう in 呉音, ちょう in 漢音)}

• Rime anomaly

  Despite the paper I cited, there are some readings inexplicable with MC rime groups. For example, 風 (fēng, 東韻三等平声) has 漢音 ふう vs. 鳳 (fèng, 東韻三等去声) has 漢音 ほう. Does perhaps the tone affect the vowel? I have no idea.

• Checked tone

  This may be another topic, but some of -うう ending kanji come from MC closed syllable groups e.g. 入 (にゅう, rù), 急 (きゅう, jí). So you must have knowledge of Cantonese, or any dialect retains -p ending. 急 is particularly difficult to guess when you only know either of Mandarin and Cantonese.

My observation is that -うう ending -e/ong rhyme kanji are relatively rare and better remembered one by one. Or you could learn from that paper, which has a number of informative charts.

Answer 2

There is no widespread direct connection between on-yomi and Mandarin pronunciation. Both systems have diverged and developed from previous varieties of Chinese.

You may find a general rule that applies for a few groups of Kanji, but I think there is very little if anything that would be an all-inclusive pattern.