I was in the process of compiling pitch accent information for Sino-Japanese words from the ＮＨＫ日本語発音アクセント辞典〜新版 for some self-study, when I noticed some interesting patterns.
The biggest thing that I noticed is that the downstep is on the first mora most of the time.
乙 òtu 町 tyòu 子 sì 有 yùu 刀 tòu 訃 hù 上 zyòu
However, all of the words I could find ending in ち had the downstep on that.
一 itì 吉 kitì 七 sitì 質 sitì 八 hatì 鉢 hatì 罰 batì
To my knowledge, words ending in ち only occur in 吳音, with the exception of ｢罰｣, which has a 慣用音 reading.
I also found that 吳音 words ending in く had the downstep on that.
菊 kikù 軸 dikù
Although this only seemed to occur in those beginning with い-syllables.
億 òku 獄 gòku
There were also 吳音 words with no downstep at all.
客 kyaku 逆 gyaku 酌 syaku
But this was only guaranteed if the first mora contained a diphthong.
角 kàku; kaku 学 gaku; gàku 作 saku; sàku
And this doesn’t occur in 漢音 at all.
脚 kyàku 略 ryàku 局 kyòku
This is by no means a comprehensive analysis of all SJ words, particularly since I’m literally writing this up on the same day I noticed these patterns, just so that I don’t forget about them. My pool of data is currently limited to around 150 or so words. ^_^;
I’ve restricted myself to readings from the 常用漢字 list, as I feel it’s the best way to avoid the problem Miyake (2004) describes whilst lacking access to more comprehensive resources:
Many Go-on readings were lost after the importation of Kan-on . . . Yet we find Go-on readings for every single character in SJ dictionaries (kanwa jiten). Lexicographers have invented what I call ‘pseudo-Go-on’ on the basis of fanqie (Miller 1967: 106, Tôdô 1980: 168) Go-on readings in today’s SJ dictionaries are a mix of genuine borrowings . . . and what Miller (1967: 106) calls “linguistic ghosts”. It is often difficult if not impossible to sort out which readings are real and which readings are imaginary. I have yet to see two SJ dictionaries whose Go-on readings match completely.
. . .
Tôdô (1980: 168) himself gives some good advice about how to avoid pseudo-Go-on and pseudo-Kan-on: “To use Kan-on and Go-on as data . . . it is best for us to limit ourselves to old materials with glosses for character readings or specific vocabulary items which have been transmitted such as 東西 tousai ‘east and west,’ 兄弟 kyaudai ‘brothers,’ and 緑青 rokusyau ‘green rust on copper.’”
I’ve also sourced which readings are 吳音 and 漢音 from Wiktionary, which I’m aware isn’t exactly the best source, but again I’m limited by my access to better resources.
On the basis of these observations, I want to propose these hypotheses:
- The default downstep of Sino-Japanese words occurs on the first mora.
- The downstep of Sino-Japanese words ending with ち occurs on the ち.
- The downstep of 吳音 readings beginning with い-syllables and ending with く occurs on the く.
- There is no downstep in 吳音 readings ending with く with a diphthong in the first mora.
｢角｣, ｢学｣, and ｢作｣ have identical 吳音 and 漢音 readings. Initially, hypothesis №3 was extended to all syllables and ending with く, but this was contradicted by ｢億｣ and ｢獄｣. Because these three words are all very frequent, I conjecture that the no-downstep readings are the result of reanalysis, as two-character 音-reading words overwhelmingly have no downstep, or “nativization” into the pitch-accent rules of 訓 readings.
I think what’s most striking to me is the distinction between the 吳音 ｢客｣ /kyaku/ and the 漢音 ｢脚｣ /kyàku/. If there are further pitch accent rules to be discovered, this could help in differentiating between 吳音, 漢音, and their pseudo-counterpart readings.
However, my big question is this:
Are there any resources — particularly English-language resources — that detail what the rules of pitch accent in Sino-Japanese words are? ^_^