18.104.22.168:32: UCHINAAGUCHI PART 37: SPRINGFIELD
In part 36, I mentioned Japanese una- 'sea' which I first encountered in the word 海原 unabara 'expanse of the sea' (lit. sea-field). I guessed the meaning of unabara from context and the Chinese character spelling 海原 'sea-field'. The only other modern una-word I know of is 海路 unaji 'sea-road' which has a variant umiji with umi, the normal word for 'sea'. There are some earlier una-words in Old Japanese. Are there any in Ryukyuan languages? Sakihara (2006) has none.
The second half of unabara, -bara < genitive + *para 'field', does have Ryukyuan cognates with non-a final vowels: e.g.,
Shuri/Naha Okinawan haru 'farm' (sounds like haru 'spring'!)
Nakijin phuroo, pharoo; -baru(u) (in compounds; borrowed from Shuri?*)
Miyako -pari ~ -bari < ?*-e < ??*-a-i (in compounds; but is there any other evidence for an -a word for 'field' in Ryukyuan?)
(Is there an Amami cognate?)
I can't think of any other Japonic word with such variation in vowels. Could these forms reflect various attempts to borrow a foreign word with a final consonant like par? At first, the obvious potential source is Korean pəl 'field', but this word is first attested in the 18th century, long after it would have been borrowed into Japonic. Korean phəri 'field' dating from the 16th century
- is older, but not old enough
- ends in a vowel that shouldn't correspond to -a, -u, or -oo in Japonic
- has an initial aspirate ph- that probably comes from *HVp- or *pVH-. There is no trace of a back consonant corresponding to earlier Korean *H in the Japonic forms.
For further arguments against a Korean connection for Japonic *parV 'field', see Vovin (2010: 105-106).
Other Japonic words for 'field' have Korean connections:
Japanese hatake < *patakai and Proto-Ryukyuan *patake (Thorpe 1983: 285) 'dry field'cf. Korean path < *patVkV 'dry field'Japanese no, Miyako nuu 'wild field' - but Shuri/Naha and Nakijin moo with m-!
cf. Korean non 'paddy field'
One might expect Japonic and Korean to share a lot of agricultural terminology, but they don't.
*Nakijin Yanbaru 'northern part of Okinawa Island' is at least segmentally identical to Shuri/Naha Yanbaru. The kanji spelling 山原 implies that yan- < yama 'forest'. This would be an unusual case of a low vowel being lost after a nasal.
22.214.171.124:59: UCHINAAGUCHI PART 36: HORSE *ʔON-THE OCEAN
Back in part 31, I proposed that *i/uN- became Okinawan ʔn- whereas *ʔe/oN- didn't. Sven Osterkamp reminded me that Thomas Pellard (2008:146) observed different reflexes in some Ryukyuan languages for Proto-Ryukyuan *um- and *om-. Below are two columns of Pellard's table plus four rows I have added at the bottom:
|Yonaguni||mma||unnaga < ?*umi-n ...|
|Okinawan||ʔnma||ʔnmi ~ ʔumi|
In Okinawan, *um- and *om- may merge into ʔn-, so my proposal is wrong for Okinawan.
In Yoron (whose speakers are closer to Japan than those of the other Ryukyuan languages in the table) and Japanese, *um- and *om- have merged into um-.In part 34, I mentioned Okinawan ʔonma 'horse' (familiar). If Pellard is right, the mid vowel in this word cannot be an archaic retention, since he reconstructed Proto-Ryukyuan *uma rather than *oma.
4.30.00:20: However, could I say that Okinawan ʔum- always goes back to PR *om-, even if O ʔn- is ambiguous?
|Proto-Ryukyuan||*um- or *om-||*om-|
The only Okinawan ʔum-root without any cognates listed in Pellard's paper is ʔumir- 'to bury', cognate to J umer-. Its Nakijin cognate is ʔubeer-, perhaps implying Proto-Ryukuan *omer-.
4.30.1:08: Proto-Japonic *uma 'horse' is probably somehow connected to Middle Korean mʌr, Manchu and Classical Mongolian morin, Old Chinese *mraʔ, Old Tibetan rmang < *mr-, etc. but its u- doesn't correspond to a vowel in other languages. Could *uma be an attempt to imitate a foreign (Chinese?) *rm-?
Another Proto-Japonic word of possible Chinese origin is 'plum': O ʔmi, J ume < Middle Chinese 梅 *məy. Nakijin ʔumii implies Proto-Ryukyuan *ome < Proto-Japonic *omai, assuming the word isn't a loan into Proto-Ryukyuan from some early southern Japanese dialect. There is no Chinese-internal evidence to reconstruct any presyllable or prefix like *o- for 'plum', though my own theory of Old Chinese phonology requires a presyllable with some sort of nonhigh vowel (which I write as *ʌ) to block the raising of *ə:
Could my *Cʌ- be rewritten as *ʔo-?
*Cʌ-mə > *məɰ > *məybut *(Cɯ-)mə > *mɨə > *muə > *mu (with high vowel presyllable or no presyllable)
4.30.1:32: I am not familiar with Yonaguni, so I don't know if its unnaga 'sea' can be derived from *umi-n .. However, I think Okinawan ʔun- in
ʔun-dumi < ʔumi + tumi 'taboo on fishing' (lit. 'sea-stop')
ʔun-jami < ʔumi + gami (< genitive nasal + kami 'god' with palatalization after *i) 'sea god festival'
is from *ʔumi < *omi.
Yesterday morning, I wondered if Japanese una- 'sea' (in compounds) could be from *omi-na- with *na- being a plural or locative suffix (as defined by Vovin 2006: 103, 153):
una- < ?*o/unna- < *omi-na-
(Don't know if *mi-reduction pre- or postdated the raising of *u.)
Cf. Japanese and Proto-Ryukyuan for 'woman' in which *me (with a mid vowel!) was reduced to n or even zero before *n:
J onna < womina < ?*womena (cf. me 'female'; also Proto-Ryukyuan *me 'female' [Thorpe 1983: 304])PR *wona- < ?*wonna (Thorpe 1983: 350) < ?*womena
Does the first syllable of O winago < lit. wenago contain an irregular reduction of *wome?
126.96.36.199:21: UCHINAAGUCHI PART 35: EATING VOWEL LENGTH
Nakijin (Okinawan: Nachijin) is spoken about 60 km north of Naha (Okinawan: Naafa), whose dialect is the basis of Sakihara's 2006 Okinawan-English Wordbook. Despite its geographical proximity to Naha Okinawan (hereafter 'Naha'), Nakijin has a number of distinct characteristics: e.g., glottalization and phonemic aspiration, traits absent from both Naha and standard Japanese. I may discuss those characteristics later. For now, I just want to look at vowel length.
Nakijin has the same five vowels as Naha:
As in Naha, all five vowels can be long or short. However, Nakijin vowel length does not necessarily match Naha vowel length: e.g. (nonmatching length in bold; Curry 2004: 38, 43),
|nose||phanaa||hana||hana < pana|
|bucket||hukˀi(i)||wuuki||oke < woke|
In my last two posts, I found that Naha short e and o tend to be in closed syllables in the first syllables of words, and I suspect this pattern also applies to syllables in other positions. (Short e in a word-medial open syllable in ʔAmerika above is in a loanword.)
Nakijin short e and o also tend to be in closed syllables: e.g.,
haa-sen 'be bright' (Naha ʔaka-san 'red', J aka-rui 'bright'; the Nakijin long vowel aa is from *aka)
kˀen 'to eat' (citation form)
Notice that 'to eat' has long vowels in open syllables of other forms (Curry 2004: 67):
kˀeeru (linked form)
I regard the short vowel as secondary since Curry derived the long vowel ee is from *ai:
kˀee < *kˀwee < *kˀwai < *kˀurai (cf. Middle Japanese kuraɸ-i < *kurap-i)
Examples of short mid vowels in open syllables from Curry 2004 and the online Nakijin Dialect Dictionary are mostly onomatopoetic apart from a few nouns:
ei ~ yai 'spear' (Naha yai, J yari)
ei 'noise made when lifting heavy objects'
eisaa 'Bon dance' (loan? cf. Naha (y)eisaa)
kˀekkˀegekkˀee 'cry of a bantam'
choroochoroo 'sound of flowing water'
dohaa 'bank of a river'
goroo 'loose fit'
ʔohooʔohoo 'sound of coughing'
Conversely, I can't find any examples of long mid vowels in Nakijin closed syllables.
188.8.131.52:59: UCHINAAGUCHI PART 34: M-O-RE PALI PARALL-E-LS?
Pali o(o) is like its front counterpart e(e):
- long in open syllables and short in closed syllables
- partly derived from an earlier VCV sequence (ava) and diphthong (au)
Is Okinawan short o like its front counterpart e which is mostly in closed syllables?
Sakihara (2006) contains the following words with short o in the first syllable:
chon-chon 'sound of liquid dropping'
chondaraa 'begging puppeteer' (< J 京太郎 *kyaũtaraũ + O -aa)
chotchon-gwaa 'bush warbler' (i.e., the bird that makes the noise chotchon)
don 'sound of a drum'
gomu-kan 'slingshot' (< J gomu 'rubber' < gum)
gon-gon 'sound of walking briskly'
nonkaa 'easygoing person' (< J nonki + O -aa)
no-yashi 'field palm' (< J no 'field' [cf. native O moo] + yashi 'palm' < Chn 椰子)
ʔOmoro (sooshi) 'Omoro soushi' (< J reading of conservative O spelling おもろさうし, now pronounced ʔUmuru sooshi)
ʔonma 'horse' (informal; normally ʔnma)
ʔotoo 'dad' (< J otou)
pon 'sound of object falling into water'
ton-ton-mii 'a stone-throwing game'; 'goby fish'
yon-naa ~ yoon-naa 'slowly' (cf. yoon 'lightly')
yos-saa 'even if'
Short o is always in closed syllables except in J borrowings (gomu, no, ʔomoro, ʔotoo).
The -on of chondaraa is from an earlier diphthong *aũ (cf. Skt au > Pali o, but I don't know of any cases of *awa > O o parallel to Skt ava > Pali o).
The -on of yon-naa may be a shortening of the long -oon in yoon.
Nearly all other native instances of -on are in onomatopoeia.
I don't understand how ʔonma developed from *ʔuma. Vowel lowering is unusual in Okinawan.
I also don't understand the derivation of yos-saa (the etymological hyphen is from Sakihara 2006: 222 and is presumably the addition of an editor rather than Sakihara himself).
184.108.40.206:45: UCHINAAGUCHI PART 33: PALI PARALL-E-LS
Last night, I derived Okinawan fensa 'peregrine falcon' from *payambusa. I was inspired by the correspondence
O fee- ~ hee- : J haya- < *paya- 'fast'
This fusion of *aya into *e(e) is similar to the reduction of Sanskrit aya to e [e(e)] in Pali. One might expect Okinawan to have no instances of aya, but it does: e.g.,
O haya-mir- : J haya-me- 'to hasten' (as noted by Sven Osterkamp)
O sayumi ~ sayun 'newly woven textile'
O dayaa ~ daraa ~ daruu 'languid person' < daru- 'languid'
O ayamar- : J ayamar- 'error'
O aya : J aya 'stripes'
O kaya : J kaya 'thatch'
O sayaka : J sayaka 'clearly'
O kayaku : J kayaku 'explosives' (< Chn 火藥)
I would expect loanwords borrowed after aya-reduction to contain unreduced aya.
kayaku is not only a loanword but also has an aya across morpheme boundaries: ka 'fire' + yaku 'medicine'.
Although aya, kaya, and sayaka might be loanwords, the others either have no known Japanese cognates or have verb paradigms with indigenous characteristics.There is one potential case of *aya becoming ii instead of e(e):
O shii ~ saya : J saya 'scabbard'
Is O saya a borrowing from Japanese postdating *aya-reduction?
We have already seen correspondences implying the reduction of *ae, *ai to ee in parts 5 and 20:
O mee : J mae < Old Japanese mape 'front'
O ʔeeda : J aida < Old Japanese apinda 'space'
This process also has a Pali parallel: Sanskrit ai > Pali ee.
I thought O (ʔi)mensee- 'to be' (honorific) might be cognate to Japanese mair- < *mawir-. Sven confirmed that and pointed out that ʔi- is an exalting prefix.
Last night, I found that Okinawan short e was almost exclusively in closed syllables, at least at the beginnings of words. This is also like Pali, in which short e is in closed syllables and long ee is in open syllables. However, in Pali, short and long e are in complementary distribution, whereas they are not in complementary distribution in Okinawan:
Okinawan short e + no coda:Okinawan long ee + coda:
geren-geren 'mentally deranged person' (浦添小湾 Urasoe Kowan dialect)
(ʔi-)mensee-n 'to be' (honorific)
keen : J kai 'number of times' (< Chn 回)
keenkeen (also kenken) 'sound of a Buddhist gong of the Nenbutsu sect'
Next: Is Okinawan short o like Pali o?
220.127.116.11:30: UCHINAAGUCHI PART 32: ʔWEE OUT IN SPACE
I think I've covered all the alternations from ʔwee, er, way back in part 21 except for one: Okinawan ʔ- ~ ʔw- corresponding to (Old) Japanese zero as in
O ʔeeda ~ O ʔweeda : J aida < Old Japanese apinda 'space'
Why does the O ʔw-variant exist at all? Is it an Okinawan innovation? I can't find anything like it in other languages in the online Ryukyuan database:
Miyako (no cognate found)
Is ʔweeda a hypercorrection with an initial ʔw- that is impossible in Japanese? Why does the ʔ- ~ ʔw- alternation only occur before long ee? There's nothing about a mid front vowel that makes it more friendly to a labial glide. After two weeks, I still have no idea what's going on.
Below the various ʔee- ~ ʔwee- words in Sakihara (2006: 208-209) are a few words with short ʔwe- that do not alternate with ʔe-. Here are all the words with initial Ce- in Sakihara, excluding obvious compounds:
O banshiruu ~ benshiruu : J banjirou 'guava' (< Chn 蕃石榴)
O fensa : J hayabusa < Old Japanese payambusa 'peregrine falcon'
O gennoo 'hammer'
O geren-geren 'mentally deranged person' (浦添小湾 Urasoe Kowan dialect)
O kekkwa : J kekka 'result' (< Chn 結果; Urasoe Kowan dialect)
O (ʔi)mensee- 'to be' (honorific)
O nen-nen ~ nin-nin : J nenne 'sleep'
O penki : J penki 'paint' (< Dutch)
O senba : J senba 'threshing tool'
O s(h)ensuruu 'mayfly'
O ten-busaa 'person with a protruding navel'
O ten-busu 'protruding navel'
O ʔwe-nchu 'person above' : J ue 'above' + no (genitive) + hito 'person'
O ʔwendaa 'gentle person'
O ʔwendas- 'gentle'
O Yeigo ~ Eigo 'English' (< Chn 英語)
O yeisaa ~ eisaa 'group Bon dance'
In all but one case (kekkwa), short e is followed by a sonorant, usually a nasal.
In all but one case (geren-geren), short e is in a closed syllable (assuming ʔwe-nchu is syllabified as [ʔwen-chu]).
Some words are of foreign origin.
Ben for 蕃 is puzzling since 蕃 has no readings with an e-like vowel in Mandarin or Middle Chinese.
Could ʔwe-nchu contain a ʔwe- borrowed from Japanese ue? I doubt it's an archaism that escaped the raising in the native word ʔwii 'above'.
fensa may be from an earlier *feensa < *payambusa with irregular shortening. Could other instances of en be from *een?
18.104.22.168:54: UCHINAAGUCHI PART 31: ROADBLOCKS
Last night, I asked why not all nasal + i/u- or ʔi/u- before nasals turned into (ʔ)n- in Okinawan. What blocked the shift to (ʔ)n-? Why do these words exist? Why aren't they nku, njumi, nchi, nra, ʔnmi, and ʔnnu?
O niku : J niku 'meat' (< Chn 肉)
O nujumi : J nozomi 'wish'
O michi : J michi 'road'
O mura : J mura 'village'
O ʔimi : J imi 'meaning' (< Chn 意 + 味)
O ʔunu 'that' (no J cognate)
'Meat' and 'meaning' are borrowings that might postdate the shift.
I don't know of any Okinawan words that begin with (ʔ)nr-. Perhaps the shift is blocked before r- to avoid that sequence. Note, however, that Sakhiara (2006: 131) does list
O ndas- : J nuras- 'to wet'
O ndir- : J nure- 'to become wet'
with nd- presumably from *nr-.
O nind- 'to sleep' even has medial -nd- corresponding to J -mur- in nemur- 'id.'
'Wish' made me wonder if the shift didn't apply to original mid vowels:
*Ni/u > n-
*Ne > Ni-
*No > Nu- (e.g., 'wish')*i/uN- > ʔn-
*eN- > ʔin-
*oN- > ʔun-
Vovin (2010: 68) reconstructed Proto-Ryukyuan *ʔo- 'that'. If my hypothesis is correct, I would not expect *ʔo-no 'that' to become *ʔnnu. Moreover, *ʔo-no is bimorphemic, and perhaps the shift does not apply to monosyllabic morphemes.
O nji 'thorn' initially seems to pose a problem for the mid vowel hypothesis if it corresponds to J nogi 'beard of a plant'. However, Thorpe (1983: 340) reconstructed 'thorn' as *nige in Proto-Ryukyuan: cf. Yamatoma and Inou nigi.
O michi 'road' is also a problem because it goes back to Proto-Ryukyuan *miti, not *meti. Bentley (2000: 429) linked the Old Japanese cognate miti to Paekche 彌知 *mete 'village', but the semantics don't match. Moreover, I would expect such a Koreanic word to be borrowed into Proto-Japonic as *mete, not *miti.