For the past 17 days I had a fantastic holiday in Japan, going from Tōkyō (+Nikkō), Takayama, Shirakawago, Kanazawa, Naoshima, Hiroshima, Kōya-san to Kyōto (+Nara). It was a fantastic trip, and a long time dream of mine to finally visit Japan. I studied Japanese for years when I was young (as earlier blog posts on this very old blog can attest to); I've kind of lost track, although I was pleasantly surprised how much I could still read, understand and make myself understood.
Different from year ago though, I am now a much more experienced historical linguist, and a lot of neat stuff about Japanese fell into place for me which I never figured out when I was actually studying it. I'm sure all of this is common knowledge, but I thought it'd be fun to share.
Sino-Japanese final *p
Middle Chinese, from which Japanese got most of its massive amount of Chinese loanwords, had final stops in its syllables (as several dialects still do today). Namely *p, *t, and *k. These The reflexes of *t and *k have fairly transparently developed into bisyllabic realizations in Japanese, since Japanese can't have word-final *t and *k hence you get: tokubetsu /toku-betu/ 'special' from something like *dʰək-bʰyɛt something still obvious from the Korean realization of these signs teug-byeol. However I never quite realized what the outcome of final *p was until recently. Middle Chinese final *p was simply borrowed into Japanese as *pu just like final *k and *t were borrowed as *ku and *tu. However, Japanese *p at some point lenites to *ɸ and quite late to h (hence Dutch koffie being borrowed as kōhī, precding the *ɸ> h shift) and intervocalically to *w. This is why you get hara'field' but the name fujiwara from *para and *puzi-para respectively. Subsequently *w was also lost in all positions execept before *a. This loss of *w, however, lead to a whole bunch of new Vowel+Vowel sequences. In verbs this vowel+vowel sequence was usually preserved (probably due to analogical levelling), e.g. *ap-u'to meet' became a-u; but I learned this holiday that in other positions it collapsed. And this is actually a very important source for the palatalization and long vowels that we find in Japanese words today
*apuki> awuki> auki> ōki 'Japanese bead tree'
*opoki > owoki > ooki > ōki 'big'
*ipu> iwu> iu> yū'to speak' (one of the few verbs whose verbal form was not analogically leveled; although it's spelling is iu)
*kepu> *kewu> keu> kyō'today'
And that realization made me understand what happened to *p final Sino-Japanese words! A word like Middle Chinese *zhip 'ten' would be borowed as *zipu, yielding ziwu > ziu > zyū (i.e. jū) the modern word for 'ten'. Words of this type are subsequently not easily distinguished from words that have a final long ū that came from a Middle Chinese word with final *ŋ which also yields vowel length, e.g. Middle Chinese 重 *djʰioŋ 'heavy' which yielded zyū in Japanese as well.
However, forms that have a final *p in Middle Chinese have a funny morphophonological feature that is missing from words that have a long yū from Chinese *-jo/uŋ; Namely: they turn into *iC- sequences in compounds such as: zip-pun'ten minutes' but zyū-ni hun '12 minutes'. This behaviour makes perfect sence if it developed from final *p:
*zipu ni pun > *ziu ni hun > zyū ni hun
*ziC-pun > zippun
This behaviour of causing gemination before geminatable consonants of historical final stops is shared with *k and *t final Sino-Japanese words; For example Tekken'Iron fist' from *tet(u) + *ken and rakkan'optimism' from *rak(u) + *kan.
Hence, whenever one sees alternation in a Sino-Japanese word between *yū and *iC in compounds, you can be sure it comes from a Chinese syllable that ended in *-ip; I would predict that you should also see *yō/*eC; *ō/*aC and *ō/oC *ū/uC alternations pointing to Middle Chinese *ep, *ap, and *op syllables, but I do not know of any clear examples off the top of my head.
This behaviour is actually rather difficult to undertstand if in the earliest Sino-Japanese loans the final *t, *k and *p already had a final vowel, which leads me to think that they may have existed in Japanese without this final vowel for quite some time. There appears to be some direct evidence for syllable-final *t these days. Christian sources transcribe, e.g. hatsunetsu suru'to get a fever' as fotnet suru. But it seems to me that this must have been true for *k and *p at some point too.
Sino-Japanese *n-w > nn assimilation
I used to always be puzzled by words like 観音kannon 'Avalokiteshvara' lit. 'Perceives sound' and 天王 tennō 'Heavenly king (a Buddhist Deity)' lit. 'Heaven-king', which for some reason had a long n while the second word is supposed to start with a vowel on'sound' as in on-gaku'music' and ō-zyo 'Princess' lit. 'King-woman'. As it turns out, words such as these simply used to have an initial glide *w which was assimilated hence *kan-won > kannon and *ten-waŋ (cf. the common Chinese last name wang) > tennaŋ> tennō. But as initial *w was subsequently lost except before *a this is no longer readily transparent.
A prediction would then be that we should also fina *wa with n-wa > nna alternation, and I was happy to find examples of that in Japan! 天和tenna (also tenwa) 'Tenna era (1681.9.29-1684.2.21)' with assimilation of the *w of wa as we also find it in hei-wa'peace'.
Middle Chinese recitation in Buddhist chants
On Kōya-san we were lucky enough to attend a morning ceremony that involved recitation in highly Japanesified Sanskrit and Classical Chinese. Especially the Classical Chinese was very striking. Classical Chinese is simply read with the Sino-Japanese reading of the characters, giving it a very distinctly Japanese sound. But rhythmically something interesting happens. The recitation has a strict one syllable per count system, with an interesting exception historical syllables that end in a stop *t or *k, are bisyllabic in Japanese, hence an original sentence like *mu sik syo ko mi sok ho now now has two syllables too many, the solution is to pronounce siki and soku in the space that the other syllables are pronounced, hence becoming twise as fast which can be heard here. Whether that should be seen as evidence of original final *k in Sino-Japanese or whether this is just an adaptation to deal with the Classical Chinese style of pronunciation with the issue of bisyllabic pronunications, I'm not sure; but I thought it was really cool!
I have a couple more interesting observations but these three are a nice single 'topic': issues of Sino-Japanese words. I'll do another blog post soon with several other interesting things I ran into.