11.4.9.15:39: UCHINAAGUCHI PART 18: FIRE(W)ORKS

All of the following words contain Chinese morphemes:

The Okinawan words were borrowed from Japanese. Which of those borrowings are older? Newer? Why?

11.4.8.23:59: UCHINAAGUCHI PART 17: WISHING AND WORSHIPPING

Last night, I dealt with unexpected initial w- in Okinawan. A few words lack an expected medial -w-. For instance, the Middle Chinese morpheme 頑 *ŋwæn 'stubborn' appears as both expected gwan and unexpected gan (Sakihara 2006: 48, 52)

頑固 O gwankuu : J ganko < gwanko 'stubborn'

頑丈 O ganjuu (not gwan ...!) : J ganjou < gwanzyau 'strong'

Here's an apparent doublet involving both initial and medial w:

wuganju ~ ʔugwanju 'place of worship'

I suspect that these two forms have different etymologies. Both share -ju < *-N-su (su is 'place' from Chn 所).

The wugan- of wuganju may be from wugam-i 'to worship' (cf. J ogam- < woŋgam-). The spelling (御)拝所 implies this etymology, as 拝 is 'worship' in Japanese (and presumably Okinawan). (But 御- is ʔu-, not wu-! See below.)

The ʔugwan of ʔugwanju is from ʔugwan 'prayer' < ʔu- (native honorific prefix 御) + gwan < the Middle Chinese morpheme 願 *ŋuanh 'wish'. (Cf. Japanese o-gan.) ʔugwan has a variant ugan (Sakihara 2006: 190) without -w-. Sakihara also lists ʔugan 'small shrine' which may be short for *ʔugan-ju 'place of worship' (does this variant exist?).

Are the w-less forms (頑 gan, 願 gan) from Japanese? Or are they the results of medial w-loss under Japanese influence (since Japanese has no gw)?

Medial -w- is missing from loanwords rather than native vocabulary like mingwi 'muddiness' (see part 9) which has no variant *mingi. This distribution suggests borrowing from a source that underwent medial -w-loss. But the vowels of ganjuu and ugan have undergone raising absent in Japanese. Have J ganjou and ogan been consciously Okinawanized after w-loss in Japanese? The O honorific prefix ʔu- is cognate to J o- and an obvious substitute for it. But why change jou [dʑoo] to juu rather than joo?

11.4.7.23:49: UCHINAAGUCHI PART 16: WUN-DERING WHAT'S GOING ON

From last night: Can you guess the Okinawan cognates of these Japanese words?

If the Japanese forms began with a-, i-, u-, ue-, y-, w-, then the Okinawan cognates have ʔa-, ʔi-, ʔu-, ʔwi-, y-, w-:

J ari : O ai 'ant' (-r- > zero between a vowel and I)

J wara : O wara 'straw'

J yari : O yai 'spear'

J ito 'thread' : O ichu 'silk' (other cognates*)

J usagi : O usaji 'rabbit' (gi > ji)

J uer- : O ʔwir-  'to plant'

J yumi : O yumi 'bow'

J yoko : O yuku 'side'

However, if the Japanese form begins with e- or o-, the Okinawan form is unpredictable:

In some cases, the Okinawan form preserves a nonglide vs. glide distinction lost in Japanese:

J ebi : O ʔibi 'shrimp' (cf. Old Japanese embi)

J eda : O yida 'branch' (cf. Old Japanese yenda)

J otos- : O ʔutus- 'to drop' (cf. Old Japanese ətəs-)

J osamer- : O wusamir- 'to control' (cf. Old Japanese wosamer-)

J ougi : O ʔooji 'fan' (cf. Old Japanese apuŋgi)

J ouji : O wooji 'prince' (cf. Old Japanese *wau < Middle Chinese 王 *wuaŋ 'king')

But in others, I don't know why Okinawan has ʔ- instead of w-, or vice versa. Both

J ourai : O wooree 'traffic'

J oukan : O ʔookwan ~ wookwan 'main road'

have the same initial morpheme (cf. Old Japanese *wau < Middle Chinese 往 *wuaŋʔ 'to go')

(Thanks to Sven Osterkamp for finding O wookwan with the expected w-)

The title refers to this puzzling set of words:

I would have predicted that both 'favor' and 'fortune' would be O un.

Could O wun 'favor' be borrowed from a Japanese dialect in which *o- > wo-?

Similarly, could O wun 'fortune' be borrowed from a Japanese dialect in which *u- > wu-? Or even a Chinese (early Mandarin?) dialect in which 'fortune' had initial w-? (Cf. Hphags-pa Chinese ɦwin.) Proto-Ryukyuan did not borrow directly from Middle Chinese, so O w- is not directly from MC w-.

*4.8.4:20: O ichu 'silk' is the expected cognate (semantics aside) of J ito 'thread'. Sakihara (2006: 67, 68, 74) lists three other cognates:

itu 'silk' - note the lack of palatalization of t after i - is this a semi-Okinawanized loan from Japanese? Or is this a descendant of a native word *eto whereas these other forms are from an *ito borrowed from Japanese?

iichu 'silk' - with vowel length that could not be predicted on the basis of Japanese

iichuu 'thread' and Gushichan dialect ichuu 'thread' - again with unpredictable vowel length

11.4.6.23:59: UCHINAAGUCHI PART 15: BEFORE I WUN-DER ...

The title will make more sense in part 16 which I originally intended to write tonight. This part will prepare you for the wun-derment ahead.

Examine the following Okinawan-Japanese cognate pairs:

O ʔati : J ate 'goal'

O wata : J wata 'entrails'

O Yamatu : J Yamato 'Japan'

O ʔicha : J ita 'board'

O ʔirab- : J erab- 'to select'

O yii : J e 'picture'

O ʔudi : J ude 'arm'

O ʔuduruk- : J odorok- 'to be surprised'

O ʔookwan : J oukan 'main road'

O ʔwii : J ue 'top'

O wudui : J odori 'dance'

O woo : J ou 'king'

O yuka : J yuka 'floor'

O yuru : J yoru 'night'

I previously didn't write initial glottal stop before vowels in Okinawan but I'm going to start now.

Japanese does not have yi, ye, wi, wu, we, wo, or initial ʔw-.

Can you guess the Okinawan cognates of these Japanese words?

ari 'ant'

wara 'straw'

yari 'spear'

ito 'thread'

ebi 'shrimp'

eda 'branch'

usagi 'rabbit'

uer- 'to plant'

otos- 'to drop'

osamer- 'to control'

ougi 'fan'

ourai 'traffic'

ouji 'prince'

yumi 'bow'

yoko 'side'

'No' is an acceptable answer in some cases if you can explain why. Answers tomorrow.

11.4.5.23:49: UCHINAAGUCHI PART 14: OONE AND TWOO

From a Japanese perspective, vowel length in Okinawan is partly predictable. Vowel length usually matches, but Okinawan monosyllables are usually long: e.g.,

O yaa : J ya 'house'

O kii : J ki 'tree', ke 'hair'

O muu : J mo 'seaweed'

Exceptions are function words: e.g.,

O ga (not gaa) 'nominative/genitive case marker' : J ga 'nominative case marker'

O nu (not nuu) 'nominative/genitive case marker' : J no 'genitive case marker'

O tu (not tuu) : J to 'and'

However, Okinawan often has unexpected long vowels in polysyllabic words: e.g.,

O tiichi : J hitotsu (not hitootsu) 'one'

O taachi : J futatsu (not futaatsu) 'two'

Is this length primary or secondary in 'one' and 'two'?

fI briefly thought that tii- 'one' and taa- 'two' were lengthened because they had become monosyllabic roots like mii- 'three', yuu- 'four', muu- 'six', and yaa- 'eight' after they lost their first syllables:

*pitə- > *pite- > *ɸi̥te- > ɸu̥te > ti > tii

*puta- > ɸu̥ta- > ta- > taa- (attested as monosyllabic  た <ta> [taa]? in Omoro soushi)

However, some Sakishima forms are disyllabic with -tii- and -taa- (Uchima & Arakaki 2000: 442 as reproduced in Vovin 2010: 349, 353). Was the length of -tii- and -taa- in those disyllables

(a) due to influence from the north?

(b) by analogy with adjacent numerals?

(c) inherited from Proto-Ryukyuan (and even Proto-Japonic)?

The loss of the first syllables may be due to analogy with adjacent monosyllabic mii 'three' and yuu 'four' and perhaps also lengthening, if it ever occurred: long *-tee- and *-taa- would be more salient than short *ɸV- which would add no informational value after  *ɸi̥ and *ɸu̥ merged. In other words, if 'one' were XYY- and 'two' were XZZ-, why not drop the X- that they share? YY- would still be distinct from ZZ-.

11.4.4.23:58: UCHINAAGUCHI PART 13: A WEIRD 'ONE'

I was surprised by the Okinawan word for 'one' when I first saw it years ago. Applying the rules in this series to Japanese hitotsu 'one' would result in hichuchi, yet the actual word is disyllabic tiichi. The loss of the first syllable is not surprising given chui 'one person' corresponding to J hitori, but the front vowel -ii- is.

This oddity is not recent. Omoro soushi (16th c. AD) contains an Old Ryukyuan form that Vovin (2010: 350) described as "intriguing": ふてつ <futetsu> 'one' coincidentally resembling Japanese ふたつ <futatsu> 'two' rather than ひとつ <hitotsu>. <futetsu> would regularly develop into a hypothetical Okinawan *futichi. I suspect that <futetsu> represented [ɸ(u̥)tɪɪtsz̩] with a voiceless vowel in the first syllable or even a cluster. Perhaps *[ɸi̥] (of an earlier *ɸi̥tɪɪtsz̩ 'one') and *[ɸu̥] had merged into [ɸu̥] which was reduced to *[ɸ] and then nothing.

In Haedong chegukki (1471), 'one' is transcribed in hangul as putyəytsʌ, possibly representing [ɸu̥teetsz̩]. Middle Korean had no [e(e)], so there was no means to directly write [e(e)] in hangul. I assume yəy is an approximation of [e(e)]. I doubt it stood for [yəy], as there is no other evidence suggesting that Old Ryukyuan had [ə] without surrounding [y]. If OR lacked [ə], it probably also lacked [yəy]. (4.5.00:31: The reverse is not necessarily true: modern Korean has [ə] but not [yəy] which has become (y)e.)

Could these forms for 'one' be the result of irregular palatal assimilation?

*pitə- > *pite- > *ɸi̥te- > ɸu̥te- > ɸu̥tɪ- > tɪ- > ti-

I have ignored length above. Why is the later front vowel e  ~ ɪ ~ i  of 'one' long? (4.5.2:07: Analogy with the long vowels of 'two' through 'four'?)

A more Japanese-like form ひとつ <fitotsu> [ɸitʊtsz̩]? is attested in later Ryukyuan poetry and may be ancestral to written Okinawan fituchi. The absence of the shift of -t- > -ch- after i normally suggests an earlier preceding *e. Could such an *e have resulted from irregular height assimilation?

*pitə- > *petə- > *peto- > fitu-?

But I know of no obsolete spelling like へとつ <fetotsu> confirming an earlier e in the first syllable.

The Omoro soushi has the expected ancestor of chui 'one person': ひとり <fitori> ([ɸitʊri]?) rather than *ふてり <futeri>.

11.4.3.23:27: SOLELY SUPPORTING SLOW PROFITS (OR ONE PERSON LEFT INSULTS)

Last night, I mentioned the unusual word pindari 'one person' from Nihon shoki. The normal Old Japanese word for 'one person' is pitəri < *pitə-tari.

I proposed that pindari could be from an alternate reduction of *pitə-tari, but there is no obvious reason why pindari has -nd- instead of -t-.

In the Oxford corpus of Old Japanese, the word appears as

毘襄利 pita-ri

spelled phonetically with Late Middle Chinese

毘 *pɦi 'to support, adjoin'

襄 *sɨaŋ 'to assist'

利 *li 'profit'

襄 is an error for 儾 LMC *ndaŋ 'slow', a phonogram for OJ nda. (Why does an *s-graph 襄 share a phonetic with an *nd-graph 儾?*) So I reject Oxford's pita-ri (which I would have analyzed as pi-tari if the second graph had stood for *ta).

John Bentley goes further and rejects the interpretation of pindari as 'one person'. He interprets the Middle Chinese tones of the graphs 毘儾利 as indicating a low-high-high pitch pattern in OJ which doesn't match the low-high-low pattern of pitəri 'one person'. Although the LHH pattern matches that of OJ pindari 'left', the preceding line (see below) concerns the emisi 'barbarians' and 'left' in Old Japanese has positive connotations (unlike English sinister!), so pindari must mean something else. John translates pindari as the gerund of a verb pindar- 'insult' cognate to Okinawan hijaruu < *pidar ... 'awkwardness' and Ishigaki pïdarun 'to hurt someone's feelings'. Following John, I could translate the first two lines of the Nihon shoki poem containing pindari as

愛瀰詩烏

emisi wo

barbarians ACC

毗儴利

pindar-i

insult-GER

'Insulting the barbarians ...'

Next: A weird 'one' in Old Ryukyuan.

*In Old Chinese, the phonetic 囊 was *snaŋ which could be phonetic in *n-graphs like 儾. (There are no early attestations of 儾, but its phonetic 囊 OC *naŋ has such attestations and in turn contains 囊 OC *snaŋ as its phonetic.) In the northwestern dialect of Late Middle Chinese known to the Japanese, OC *sn- became *s- but OC *n- became *nd-.