Upload
anita-ward
View
123
Download
5
Tags:
Embed Size (px)
DESCRIPTION
Kakari-musubi in Old Okinawan
Citation preview
THE FORMS, FUNCTIONS, AND MEANINGSOF SU AND DO FOCUS CONSTRUCTIONS
IN THE OMORO SŌSHI OF OLD OKINAWAN
A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THEUNIVERSITY OF HAWAI‘I IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
IN
EAST ASIAN LANGUAGES AND LITERATURES(JAPANESE)
DECEMBER 2009
By
Lina J. Terrell
Thesis Committee:
Leon A. Serafim, ChairpersonAlexander VovinArthur Thornhill
Keywords: Omoro sōshi, kakari musubi, emphatic constructions
We certify that we have read this thesis and that, in our opinion, it is satisfactory
in scope and quality as a thesis for the degree of Master of Arts in East Asian Languages
and Literatures (Japanese).
THESIS COMMITTEE
______________________Chairperson
______________________
______________________
ii
TABLE OF CONTENTS
TABLE OF CONTENTS............................................................................................iii
LIST OF TABLES......................................................................................................V
LIST OF FIGURES...................................................................................................vi
LIST OF ABBREVIATIONS AND SYMBOLS............................................................vii
CHAPTER 1: INTRODUCTION..................................................................................1
A. KAKARI-MUSUBI FOCUS CONSTRUCTIONS IN JAPANESE AND RYŪKYŪAN........11. Kakari particles in Modern Japanese......................................................22. Kakari-musubi in the Ryūkyūan Dialects................................................33. Kakari-musubi in Old Okinawan.............................................................4
B. IRREGULARITY IN THE KAKARI-MUSUBI OF THE OMORO SŌSHI.........................51. Endings indistinguishable from imperative form....................................62. Endings in -te...........................................................................................73. Endings in attributive form......................................................................94. Other irregularities in the kakari-musubi of the Omoro sōshi..............10
CHAPTER 2: THE OMORO SŌSHI..........................................................................11
A. BACKGROUND................................................................................................111. Manuscript lineage of the Omoro sōshi.................................................132. Print editions of the Omoro sōshi..........................................................193. Overview of scholarship on the Omoro sōshi........................................23
B. ORTHOGRAPHY OF THE OMORO SŌSHI............................................................251. Japanese historical spelling (rekishi-teki kanazukai)............................252. Kana inventory and spelling conventions in the Omoro sōshi..............28
a. Spellings showing palatalization.......................................................29b. Spellings affected by vowel raising...................................................34
i. <…u> alternating with <…o>............................................................34ii. <…i> alternating with <…e>............................................................35
c. Other sound changes affecting spelling.............................................36d. A note on dakuten and voicing in the Omoro sōshi...........................37
3. Chinese characters in the Omoro Sōshi................................................37C. PHYSICAL FORMAT OF THE SONGS.................................................................38
1. Melody names, parallel lines and refrains............................................392. Reconstructing a song............................................................................42
CHAPTER 3: KAKARI-MUSUBI IN THE OMORO SŌSHI.........................................52
A. METHODOLOGY..............................................................................................521. Description of the database...................................................................522. Eliminating non-KP lines to find the pool of KP su and do candidates 533. Reconstructing songs to further eliminate non-KP particles................574. Sorting lines by sentence-final word......................................................67
B. FINDINGS........................................................................................................70
iii
1. Endings of lines with su and its allomorphs..............................................71a. Lines with su, verbs in -e...................................................................71b. Lines with su, verb in -te...................................................................74c. Lines with su, verb in -a....................................................................77d. Lines with su, ending in a noun.........................................................78e. Lines with su, verb ending in -u, and verb ending in …u-na)...........78
2. Endings of lines with do and its allomorphs..............................................80a. Lines with do, verb in -u....................................................................80b. Lines with do: verb in -e....................................................................83c. Lines with do, verb in -te...................................................................84d. Lines with do, verb in …u-na............................................................85e. Lines with do, verb in -i.....................................................................87f. Lines with do, verb in -a....................................................................88
C. CONCLUSION...................................................................................................89
APPENDIX A: INVENTORY OF CHINESE CHARACTERS IN THE OMORO SOSHI........92
APPENDIX B: SUMMARY OF VOLUMES OF THE OMORO SŌSHI..............................93
BIBLIOGRAPHY......................................................................................................94
iv
LIST OF TABLES
Table 1. Summary of correspondences between kana and their pronunciations following sound changes in Middle Japanese........................................................................27
Table 2. Summary of possible spellings for each column affected by mergers and splits................................................................................................................................28
Table 3: Kana inventory in the OS and corresponding romanization used in this paper.. 29
Table 4. Distribution of spellings of (possible) KP su and do...........................................32
Table 5. Summary of all possible spellings of KP su and do in the OS............................53
Table 6. Pool of possible kakari particles (unreconstructed)............................................57
Table 7. Pool of kakari particles (reconstructed), excluding sentence-final particles.......66
Table 8. Count of KP su and KP do, sorted by line-final word form, in reconstructed songs......................................................................................................................70
v
LIST OF FIGURES
Figure 1. First page of photographic reproductions of the Shō manuscript in the Teihon omoro sōshi (2002)................................................................................................18
Figure 2. The first page of songs in the Omoro sōshi edited by Hokama and Saigo (1972)...............................................................................................................................21
Figure 3. The first three songs in the electronic version of the Omoro sōshi prepared by Takahashi Toshizō (2005).....................................................................................22
vi
LIST OF ABBREVIATIONS AND SYMBOLS
Abbreviations Meanings
1PS First person singular2PS Second person singularABL AblativeACC AccusativeATTR AttributiveCON ConjunctionCOND ConditionalDEV DeverbalDIM DiminutiveDV Defective verbEV EvidentialEV.KM Evidential (and kakari musubi verb)EV/IMP Evidential/ImperativeEXCL ExclamationFP Sentence-final particleGEN GenitiveHON HonorificHON.AUX Honorific auxiliaryHUM HumbleIMP ImperativeINF InfinitiveINF.AUX Inferential auxiliaryJ JapaneseKK Konkōken-shū (‘Collection of confused meanings,’ 1712CE)KM Kakari-musubiKP Kakari particleLOC LocativeM.IMP Mild imperativeMidJ Middle Japanese (Heian, Kamakura, early Muromachi periods)ModJ Modern Japanese (late Muromachi, Edo, current periods)ModOk Modern OkinawanMYS Man’yōshū (‘Anthology of a Thousand Leaves,’ 759CE)NEG NegativeNEG.IMP Negative imperativeNOM NominalizerOBJ Object markerOJ Old Japanese (before 794CE)OKDJ Okinawa kogo dai-jiten (‘Dictionary of Classical Okinawan,’ 1995)OOk Old Okinawan (before 1623CE)OS Omoro sōshi (‘Anthology of songs,’ 1531-1623CE)PASS Passive
vii
PAST Past tensePERF PerfectivePJ Proto-JaponicPN Place namePOSS PossessivePP Progressive palatalizationPRED PredicativePROG ProgressivePSNM Person namePT ParticleQP Question particleQUOT Quotative particleSHPNM Ship nameSTAT StativeSUBGER Subordinating gerundSUBJ Subject markerTENT TentativeTOP Topic markerUNC Unclear formUNK Meaning unknown
Symbols Meaningssh sh as in shirtzh j as in judgech ch as in church// indicate phonological readings[] indicate phonetic readings; in omoro, indicate reconstructed segments<> indicate romanized spellings of original kana or kanji words
viii
CHAPTER 1: INTRODUCTION
A. Kakari-musubi Focus Constructions in Japanese and RyūkyūanA well-known feature of the Japanese language spoken and written during and before the
Heian period is a set of focus structures consisting of one of five particles, plus a required
sentence-final form linked with the particle. The presence of one of the four particles zo,
ka, ya, or namu often triggers the attributive form (rentai-kei) on the main verb at the end
of the sentence, instead of the usual predicative form (shūshi-kei), lending emphasis, or
foregrounding, to the marked word or phrase. The poem in Example 1 illustrates:1
1OJ 可久婆可里 古非牟等可祢弖 之良末世婆Transcr. ka-ku bakari kwopiy-m-u tö kanete sir-amas-ebaAnalysis thus-INF PT love(INF)-TENT-PRED DV beforehand know-SUBJUNC-COND
Japanese かくばかり 恋ひむとかねて 知らませばEnglish Had I known beforehand that I would love you this much--
OJ 伊毛乎婆美受曽 安流倍久安里家留Transcr. imwo woba myi-z-u sö ar-ubey-ku ar-i-kyer-uAnalysis girl PT see-NEG-INF KP exist-DEB-INF exist-INF-RETR-ATTR
Japanese 妹をば見ずそ あるべくありけるEnglish I really ought not have seen you, my beloved. (MYS 15:3739)
Although not voiced in this poem, the particle spelled sö is the kakari particle zo, one of
the emphatic particles under consideration here. Note that the auxiliary verb complex at
the end of the poem is in the attributive -u form, rather than, in this case, the predicative -
i (which would have given ar-i-kyer-i). Several poems in the Man’yōshū without one of
the particles zo, ka, ya, or namo (later namu2), similarly end in a verb complex in the
attributive, suggesting that the link between the particles and the verb form is more of a
tendency than a rule. Japanese traditional grammar refers to the phenomenon of sentences
lacking the particle but ending in the attributive anyway as rentai-dome, or “ending in the
attributive.”
1 The transcription method of these poems is adapted from Vovin 2005. In order to simplify, the letters b, z, d, and g replace Vovin’s /Np/, /Ns/, /Nt/ and /Nk/, respectively.2 WOJ namo appears only once in the Man’yōshu.
1
A single member of the set of kakari particles, koso, is associated with the
evidential form (izen-kei) on the verb:
2OJ 左其呂毛能 乎豆久波祢呂能 夜麻乃佐吉Transcr. Sa-görömo-nö wo tukupa ne-rö-nö yama-nö sakîAnalysis 5th.month-DV-robes-GEN little Tsukuba peak-DIM-
DV
mountain-GEN ridge
Japanese さ衣の 小筑波峰ろの 山の崎English (On) the slopes of little Tsukuba’s peak (when we wear)the robes of May--
OJ 和須良許婆古曾 那乎可家奈波賣Transcr. wasura kö-ba kôsö na-wo kakê-nap-am-ëAnalysis forget(INF) come-COND KP 2PS-ACC call.out-NEG-TENT-EV
Japanese 忘らこばこそ 汝を懸けなはめEnglish If you really do forget to come, I shall not call out to you!
(MYS 14:33943)
As in Example 1, the poem’s final verb is not in the predicative (which in this case would
give kakê-nap-am-u). There does not appear to be a parallel phenomenon of “izen-dome,”
where the evidential appears without the particle koso preceding it in the line, the way the
attributive may end poems without any kakari particles.
1. Kakari particles in Modern JapaneseThese focusing constructions consisting of a particle and an expected verb-ending, known
as kakari-musubi by Japanese linguists, have not survived into the modern language,
though the particles themselves remain. Hendriks 1988 points out that two effects
compounded to make kakari-musubi lose its force. First, as noted above, the kakari-
musubi system had never been entirely consistent, with particles and their associated
verb-forms sometimes appearing independently of each other; additionally, the particles
increasingly took other functions, including as sentence-final particles. Secondly, in the
Kamakura and Muromachi periods, the attributive (rentai-kei) and predicative (shūshi-
3 Book 14 of the Man’yōshū is made up of Azuma-uta, or ‘Eastern Songs,’ supposedly composed by military men from the northeastern region of Hoshu but stationed in northern Kyushu. Many poems contain spellings and vocabulary reflecting the Eastern Japanese dialect that differed in many ways from the language of poets from Central Japan around Kyoto. Thus the spelling <kôsö>, with A and B type mid-back vowels in the same word, was possible in Eastern Old Japanese; but impossible in Central Old Japanese, which had a rule for vowel harmony requiring the word be spelled <kösö>.
2
kei) verb forms merged, with the former taking over all the functions of the latter.4 The
loss of distinction between the verb forms sped up the breakdown of the kakari-musubi
system, the inconsistencies of which may have played a part in triggering the merger of
the verb forms in the first place. Although the focus structure involving koso, linked as it
was with the evidential form¸ was not affected by the merger of the attributive and
predicative verb functions, it too faded out of use by the middle of the Edo period
(Okimori 1989) at the latest, leaving the particle able to appear without requiring a
specific verb-ending.
2. Kakari-musubi in the Ryūkyūan DialectsIn general, in the Ryūkyūan dialects of Okinawa prefecture and the Amami
Islands of southern Kagoshima prefecture, the attributive and predicative forms of verbs
have not fallen together, with the result that kakari-musubi still functions in many
Ryūkyūan dialects, including Modern Okinawan (ModOk). ModOk uses du (::J5 zo < OJ
sö/zö FOCUS PARTICLE (KP)) for emphasis, and its presence in a sentence triggers the
attributive form on the verb in final position. This particle has a variant dō that appears as
a sentence-final particle, lending its emphatic meaning to a whole statement; in these
cases the verb does not require the attributive ending. Note the change in the form of the
verb between Examples 3 and 4, of which the latter contains KP du:
4 This change explains the grammatical fact in Modern Japanese of the noun-modifying form (i.e., the attributive form) of verbs being the same as the predicative "plain" form. An exception to this replacement of the predicative by the attributive form is for the copula, as for example in na-adjectives: attributive -na in ano yūmei-na hito (that famous-be/ ATTR person, ‘that famous person’); but predicative da in Ano hito-ga yūmei da (that person-SUBJ famous be/ PRED , ‘That person is famous’). Had the copula followed the same path as other verbs, the expected form for the second example might be *Ano hito-ga yūmei-na . 5 The double colons should be read "corresponds to."
3
3
イ ャ ーガ
ヤーンジ ニントータン
‘yaa-ga yaa-nji niN-too-ta-Nyou-SUBJ house-
LOC
sleep-PROG-PAST-PRED
You were sleeping in the house.
4
イ ャ ー ガ ドゥ
ヤーンジ ニントータル6
‘yaa-ga du yaa-nji niN-too-ta-ruyou-SUBJ KP house-
LOC
sleep-PROG-PAST-ATTR(KM)
It was you sleeping in the house!
The particle su, corresponding to the Japanese particle koso, however, has not survived
except possibly as the nominalizing suffix -si/şi (Serafim & Shinzato 2009).
3. Kakari-musubi in Old OkinawanThe Old Okinawan (OOk) of the Omoro sōshi (OS, see Chapter 2, Section A
below) shows ample evidence of kakari-musubi with at least two kakari particles,
referred to hereafter as KP do (::J zo, ModOk du) and KP su (::J koso < OJ kösö < PJ
*kö#swo 'PROXIMAL DEMONSTRATIVE' + 'NOMINALIZER'7 (Serafim & Shinzato 2005)).
Particles do and su8 generally appear to trigger forms of verbs resembling the Japanese
attributive and evidential, respectively. Example 5 presents a line usually accepted as
showing the particle do (<to>) and the verb asub- (<asuf>, ‘asїb-, ‘play,’ J. asob-) in an
6 Sentences adapted from Miyara 2000, pp. 95-96. Note that here, the single left quote mark indicates a glottal stop.7 Serafim and Shinzato 2005 demonstrate that PJ *kö 'PROXIMAL DEMONSTRATIVE' and PJ *swo 'NOMINALIZER' are the origin of both the OJ compound kösö and OOk su. Thus we can presume that PJ had an alternating pair of KP, *øø-swo and *kö#swo, with an internal word boundary. This internal boundary was lost in OJ, and following a rule of internal vowel harmony, the second vowel assimilated to the first, resulting in OJ kösö. In PR, however, the second of the alternating pair either never existed or fell out of use before the word boundary disappeared, leaving only one vowel and no need for assimilation. 8 Italicized, "su" and "do" are best thought of as abstractions referring to all the possible spellings (representing allomorphs) of the OOk focus particles corresponding to Japanese koso and zo, respectively. Allomorphs are distinguished in this paper by their spellings, marked by enclosing them in diamond brackets, e.g., <siyo> (allomorph of KP su) and <tiyo> (KP do). Note also that this paper uses the conventional pronunciations used by the Omoro Sōshi Kenkyū-kai, which uses the spellings of the words in the OS as a pronunciation guide, without attempting phonological reconstructions of the words as they were actually pronounced (as done by Serafim and Shinzato 2009 and forthcoming). Thus the kakari particles are su and do rather than sï and du, and so on.
4
emphatic construction, with the verb in the attributive form (though, as in Japanese, it
belongs to a category of verbs that do not distinguish attributive and predicative verb
endings):
5imiya-kara to ore-nao-tife asuf-unow-ABL KP descend(INF)-renew-SUBGER play-ATTR(KM)From now indeed, descending (into trance) once again,(we) dance!
(90)
Example 6 does the same for particle su, with the verb mafur- (i.e., mabur-,
‘protect,’ J. mamor-) apparently an evidential form:
6MATA UFU-nusi su mafur-e waka-nusi su mafur-e
great-lord KP protect-EV?(KM) young-lord KP protect-EV?(KM)It the King whom (she) protects! It is the vital King whom (she) protects!
(40)
At the very least, these examples show that the dialect of Okinawan represented in
the Omoro sōshi preserved both types of focus structure until the 16th century when the
OS was first written down. In the case of the now-extinct particle su, of course, it could
even then have been an archaism preserved in writing and sacred singing that no longer
existed in the spoken language.
B. Irregularity in the kakari-musubi of the Omoro sōshiThere is some irregularity in the verb complexes (i.e., verb stems plus auxiliaries and
suffixes) forming the musubi portion of the kakari-musubi (KM) focus structures in the
OS. As shown in Examples 5 and 6 above, su and the allomorphs it represents usually
appear with a sentence-final verb (the musubi) in -e, possibly the evidential. Do and its
allomorphs usually appear with a musubi in -u, possibly the attributive. The similarity in
forms to the Old Japanese verbs in these cases makes it tempting to simply label the
forms as in Japanese, but the irregularities in the lines imply that there may be other
analyses for the verbs. For example, the musubi verbs in -e that end the lines in which su
occurs also appear in hundreds of lines in the OS that do not have su. Verbs in the su
musubi slot can have unexpected endings, such as a -te form that at first glance may be a
5
past tense or perfective morpheme, possibly an irregular allomorph of the evidential of
that morpheme (though it is more likely the subordinating gerund corresponding to
Modern Japanese -te). Occasionally su appears with a musubi in -u, the attributive; and in
other places, do may appear with its musubi in -e. It is the purpose of this paper to
examine these irregularities and to determine what patterns, if any, appear, and to sort
them out into clear groupings.
1. Endings indistinguishable from imperative formEndings in -e that cannot be distinguished from the imperative may simply be due to the
imperative and the evidential forms having merged in OOk, much the way the attributive
and predicative forms of verbs merged in Middle Japanese. Similarly as well, the loss of
distinctive meaning for the evidential ending may have speeded the disappearance of the
su emphatic construction, and thence the loss of the particle itself.
All the same, the verb in Example 6 above may be analyzed as appearing in
evidential, based on indirect evidence in Modern Okinawan, Old Okinawan, and Old
Japanese. In the OS, t-i-yowar-e (come-INF-HON-EV/IMP ‘come’, hereafter choware, to
reflect the conventional reading) appears several times as t-i-yowar-e-fa (i.e., t-i-yowar-e-
ba, come-INF-HON-EV-CONJGER, ‘when (it) comes’), with an ending corresponding to
what Miyara 2000 (p. 229) calls the ‘evidential suffix’ (已然辞) -(r)iba in literary Modern
Okinawan. The suffix -ba corresponds to the OJ conjunctive gerund -ba, which always
followed the evidential. Together, these facts tell us that -(r)e was the evidential ending in
OOk (-(r)i in Modern Okinawan).
However, choware appears twice as often in lines without su as lines with (145
vs. 78, respectively). For example, the refrain in omoro 490 has su (here, in the form of
its progressively palatalized allomorph <siyo>), and ends with the verb in the expected -e
form:
7aNsi-osoi siyo sedi masa-te t-i-yowar-elord-HON KP invisible.power triumph(INF)-SUBGER come-INF-HON-EV
It is the King, his spiritual power triumphant, who shall reign!(490)
6
The refrain of omoro 511 is nearly the same, but lacking the initial noun phrase plus
particle:
8weke sedi masa-te t-i-yowar-eEXCL invisible.power triumph(INF)-SUBGER come-INF-HON-EV?Oh! (his) spiritual power triumphant, let (him) reign!
(511)
Examples 7 and 8 suggest that either su is not a kakari particle at all—that is, its
appearance in a line does not affect the form of the final verb; or, that the evidential and
another ending, most likely the imperative, had fallen together and were
indistinguishable, at least by their spellings, by the time the OS was written down.9 For
this reason, from this point forward, this ending in -e in all cases is analyzed as
evidential/imperative (EV/IMP) in this paper.
Another aim of this thesis is to attempt to clarify whether the apparently evidential
and imperative endings in -e can ever be distinguished in the OS by criteria other than the
presence or absence of a kakari particle. If not, this may be evidence that the type of
emphatic construction involving KP su and a verb ending in evidential had already fallen
out of use by the time the OS was written down.
2. Endings in -teThe second set of anomalous verb endings of lines with su has long been explained by
saying that the songs of the OS are not always completely written out, so that the lines
may in fact be incomplete as they appear; that is, they show cases of ellipsis. Thus the
sentence-ending verb is simply not given, and the -te form that is seen is actually a mid-
sentence continuative form (chūshi-kei) that in ModOk has the allomorphs -ti, -di, -chi,
and -ji (::J -te, MidJ -ite, OJ -(y)i-te ‘SUBORDINATING GERUND (SUBGER)’, or ‘-ing’).
Omoro 34 provides an example of this form, which also raises several questions. In this
song, the particle su appears in each of a pair of parallel lines (Example 9 below), both
ending with the same verb complex, yor-i-ore-te (approach-INF-descend(INF)-SUBGER,
‘descending, approach’ (Hokama 2000, p. 35)):
9 Note that Modern Okinawan has the imperative suffix -ree, e.g., yami-ree ‘Stop!’ (Miyara 2000), so the endings are now distinguished not only by vowel height, but also by vowel length.
7
9MATA isi-wekeri aNsi-osoi ka-i-nade-te su yor-i-ore-te
HON-brother lord-HON grab-INF-touch-SUBGER KP descend-INF-approach-SUBGER?
It is our honored brother, our King, who is reaching up; may (the Spirit) descend and permeate (him)
MATA isi-wekeri tatam-i-kiya mi-mafu-te su yor-i-ore-teHON-brother precious-INF-child see(INF)-protect-SUBGER KP descend-INF-
approach-SUBGER?It is our honored brother, our King, who watches over (us); may (the Spirit)
descend and permeate him(34)
Elsewhere, the ending is palatalized (in Example 10, due to interplay between the
stem-final /s/ in oros- ‘descend’ and the -(i)te suffix), and written <tife>, conventionally
pronounced /che/, such as in omoro 628, where the line-final word may be read oroche:
10ITI UFU-kimi-ka u-sas-i siyo omokafa-no seti oro-tife
great-priestess-SUBJ HON-order-INF KP priestess-GEN invisible.power deposit-SUBGER?It is the Great Priestess’s order that calls down the invisible power of Omokafa
(628)
In lines without a kakari particle, the simplest conclusion is that this is, indeed,
the subordinating gerund (::OJ -ite, ModJ -te). However there is a possibility that the
ending should be analyzed differently in lines with KP su. Instead of analyzing as the
subordinating gerund -te, the ending may be an apocopated variant of -te-[r]e, that is, a
perfective-evidential, so that, for example, omoro 628 from Example 10 above should be
analyzed as follows:
11ITI UFU-kimi-ka u-sas-i siyo omokafa-no seti oro-tif-e
great-priestess-SUBJ HON-order-INF KP priestess-GEN invisible.power descend-PERF-EV?
It is by the Great Priestess’s order that the invisible power of Omokafa has descended
(628)
8
This paper aims to see how often this ending appears as the musubi candidate in lines
with KP su, and to discover whether such an analysis is ever justified.
3. Endings in attributive formA third set of anomalous lines contain su, yet end with a verb apparently in the attributive
form, instead of the expected evidential. Omoro 1107, reconstructed (see Chapter 2,
Section C.2 below), has a pair of such lines:
12MATA tiyara-moi-ka keni siyo kimi-si teta mi-tiya-ru
Chara-HON-SUBJ truly KP lord-king sun see-PAST-ATTR?It is honored Chara, truly, who saw the sun, the King
MATA omofi-kuwa-no keni siyo [kimi-si teta mi-tiya-ru]HON-child-GEN truly KP It is the honored one, truly,…
(1107)
The attributive is expected when do or one of its allomorphs appears in a line. In
Example 12, however, the particle is <siyo>, usually considered a palatalized (due to the
preceding /i/) allomorph of KP su. This is a puzzling case, but perhaps the spelling of the
particle is actually a palatalized variant of KP do, not KP su. A noticeable feature of
Modern Okinawan when compared with Modern Japanese is the swath that palatalization
and subsequent affrication of the stops /k/, /g/, /t/ and /d/ has carved through a large
portion of the phonology of its Japonic lexicon. Given the limitations of the kana
syllabary writing system, and the pronunciation overlaps between some of those kana,
any lines with what looks like KP su and a musubi in -u may be due to spelling overlaps
between possible palatalized allomorphs of do and su in the orthography of the OS. In
addition, just like so many old Japanese texts, voicing is never indicated in the Shō
manuscript of the OS, the oldest extent version. Thus a particle spelled <siyo> may have
been intended to be read as /zhu/ or /zhju/, a possible allomorph of do, making the
attributive form of the verb the correct one for that line. Chapter 2, Section B, contains
more detailed descriptions of progressive palatalization and vowel-raising in Okinawan,
and of the Ryūkyūan spelling system.
9
Conversely, the kakari particle do sometimes appears with its verb musubi in -e,
such as in Example 13 (where it is unmistakeably do):
13MATA ati-osoi ya imiya-kara do TENI-kiya SITA ito kake-te t-i-yoware
lord-HON TOP now-ABL KP heaven-POSS below thread hang(INF)-SUBGER come-
INF-HON-EV/IMP
It is from this moment that the King will hold all below heaven as on a thread, and reign
(101)
How often do lines with KP do occur with the verb in -e? And is it possible to
explain them? It is not possible to explain this particle away by saying that it is some
version of KP su, as we tried in the example before this. It is possible that lines like this
simply have no explanation at all, and are simple errors. Identifying such lines in the OS
may clarify the seeming confusion of musubi forms, and is one of the aims of this paper.
4. Other irregularities in the kakari-musubi of the Omoro sōshiFinally, even after a thorough examination of all the lines in which KP su and KP
do appear, there may remain some irregularities, with a few unexplainable cases.
Particles spelled so that they are impossible to distinguish, in songs that may be corrupted
through inaccurate transmission, or that give no final musubi to help identify the particle,
may result in some lines proving impossible to analyze.
All these confusions await explanation, or at the very least, some organization that
clearly shows where difficulties lie. This paper aims to to clarify the confusion over these
points, and to lay out in full the forms of su and do kakari-musubi in the Omoro sōshi.
10
CHAPTER 2: THE OMORO SŌSHI
The data for the study conducted in this thesis are collected exclusively from the Old
Okinawan document, the Omoro sōshi (‘Anthology of omoro’). This somewhat obscure
manuscript requires a considerable amount of introduction, thus the length of Chapter 2.
Its history, the techniques used to write down the songs (omoro) it anthologizes, its
spelling system and the linguistic sound changes reflected there, and the techniques
necessary to decode the shorthand and begin to understand the omoro, all require
explanation.
A. Background Okinawa is the largest island of the Ryūkyūan archipelago, a chain of tiny islands that
stretches between Taiwan and Japan's southwesternmost main island of Kyushu. The
Ryūkyūan islands show influence from their Pacific and Asian neighbors to the south and
west. However, the language and culture found in the archipelago parallel to a striking
degree that of the settlers who began to migrate from the Korean peninsula to the main
Japanese islands in the last centuries before the Common Era, and who eventually
established the earliest Japanese polities. Rice farming and its implements; use of sake in
religious ritual; worship of the sun; the dual governments in which a male and female
sibling pair divided temporal and spiritual authority and duties; and, most of all,
languages which clearly can be shown to be related, all point to an early common source
for the Japanese and Ryūkyūan peoples. Okinawa’s society until around the 12th century
consisted of independent small villages, their inhabitants eking out livings through
agriculture and fishing, ruled by a female shamanic priestly caste holding religious
authority alongside male secular village leaders. A period of fighting saw the gradual
unification of the villages, with three strong overlords emerging during the “Three
Kingdoms period” (三山時代, sanzan jidai). King Shō Hashi (r. 1422-39) finally succeeded
in uniting the island under one government in 1429. He had come to power in the most
prosperous of the three rivals, Chūzan, and during his reign continued to build strong
trading ties with China, Korea, Japan, and Southeast Asia.
This Chūzan kingdom, now controlling all of Okinawa, flourished for nearly five
centuries, and eventually came to include the whole archipelago. Its final capital was at
11
Shuri in the south-central part of the island. Today, on the site of the original castle that
was smashed into rubble in the Battle of Okinawa in 1945, a recent reconstruction of
Shuri Castle sits atop a hill, commanding views over the East China Sea to the west, and
the Pacific to the east. The heavy stone foundations, the red walls, and the graceful lines
of roof and beam convey an idea of the refinement cultivated by the aristocrats and royals
who lived in the old castle, a lifestyle made possible by the trade conducted across the
wide blue sea visible from every outlook. In the tradition inherited from before
unification, men and women shared authority in the kingdom, with the king and his male
government ruling the secular realm, and the highest-ranking woman, always a close
relative of the king, leading the ranks of priestesses who administered the spiritual realm.
As a conduit to the world of the gods, and with her power to confer spiritual protection
and power upon the king, the highest-ranking priestess, called Kikoe-ō-gimi in Japanese
(Chifijin in Modern Okinawan10), was probably the equal of, and in some roles may have
outranked, the king himself (Kerr 2000:110). Though this system changed and faded as
Buddhist and Confucian ideas gained strength in the kingdom (especially following the
Satsuma takeover in 1609, after which the king’s power was circumscribed), even today,
long after the kings of Chūzan have disappeared, women in their roles as noro
(shamanesses) tend to the native religious ceremonies of hearth and village throughout
the islands.
One of the greatest surviving treasures of the Ryūkyūan kingdom is the Omoro
sōshi (OS), an anthology of songs, or omoro,11 written down between 1531 and 1623.
This work, initiated under the direction of King Shō Sei ( 尚 清 , r. 1527-1555), was the
final step in consolidating a project begun under King Shō Shin ( 尚真 , r. 1477-1526) of
10 The etymological relationship between the Okinawan chifijin and the Japanese kikoe-ō-gimi is difficult to pin down, and is based mainly on the way the word is written in the OS. The possible progression (briefly) of kikowe-ofo-kimi > kitshuwi-ufu-kiN (PP, vowel raising, final nasal + i —> N) > tshitshuwi-fu-tshiN (regressive palatalization, vowel elision) > tshuwi-fi-tshiN (syllable absorption, vowel assimilation) > tshiwi-fi-tshiN (vowel assimilation) > tshi-fi-tshiN, does not agree with evidence elsewhere, such as that showing the Okinawan preserved double vowels as long vowels. In addition, the word invariably appears in the OS written as きこゑ大きみ (transcribed in this paper as kikowe-UFU-kimi). The character 大 before kimi never appears spelled out in kana, meaning there is no evidence in the OS itself of how the word was pronounced.11 Hokama Shuzen (2000b:470-475 and elsewhere) explains that the word omoro is most likely a Japanese version of the Okinawan word umui, used by the creators of the OS to lend gravitas to their document. He speculates that the word is related to the verb omou ‘think,’ in that it once referred to the action of shamanic kings and queens sending messages to the gods through their thoughts and prayers.
12
creating a government office dedicated to collecting religious songs from every corner of
the newly-unified kingdom, and further establishing the authority of the royal court at
Shuri. Omoro were the songs used by the shamanist priestesses in religious ceremonies
and rites to call for fertility and rain, and, in the case of the songs in the OS, to bless the
trade ships, military ventures, and tributary missions of the Shuri government,
specifically King Shō Shin. The collected songs of the OS give insight into the culture of
the kingdom, referring to the towns and villages of Okinawa, its neighbor islands, and
other countries, as well as heroes, priestesses, and famous singers of omoro. Fishing and
farming, the piercing rays of the rising sun, ships in full sail on the sea, women dancing
and moving their hands in graceful patterns as they chant blessings and prayers, brave
soldiers, a glorious king—all these images enrich the songs. Besides these glimpses of a
vanished world, the Omoro sōshi is also the largest available record of the language
spoken in Okinawa, especially in the capital at Shuri, before the influence of mainland
Japanese began to overtake the local dialects, a process that is largely complete today.12
12 A valiant resistance campaign against the loss of their dialects is being waged by Okinawan dialect scholars, performers of the traditional stage arts, and singers, many of whom consider it their cultural and ethnic duty to protect their heritage.
13
1. Manuscript lineage of the Omoro sōshiThe Omoro sōshi (henceforth referred to as OS) consists of a total of 155413 songs
divided among 22 volumes variously compiled between 1531 and 1623. The number
1554 includes repetitions (chōfuku, ‘duplicates’) of some songs. Not all the duplicated
songs are exactly the same, with some of the repeats missing lines, or containing different
spellings, or other small differences, so it is possible to argue that dissimilar duplicate
songs are not, in fact, repetitions, but unique songs in their own right. Hokama and
Hateruma (2002) put the count of songs at 1,295, giving us 249 duplicates.14 The
duplicate songs are valuable in their variations, because the differences give us hints
about spelling and grammar details that would not come up in identical duplicates.
In 1710, realizing there were many obscure words in this and other court
documents, and fearing to lose an important part of their heritage, officials of the Shuri
royal government began preparing a dictionary containing difficult words used in the OS
and in court ceremonial language. This dictionary is the Konkōken-shū (‘Collection of
confused meanings,’ completed in 1712; henceforth KK), itself the object of much study
as part of the effort to decipher the OS in the last century. Together, the OS and the KK
preserve the most complete record of Old Okinawan that we have today; the only other
surviving sources include later government documents, carved stone monuments, some
district government documents from the outer islands, and word lists compiled by
Chinese emissaries and Korean ship clerks. All of these texts are of great interest to
13 The Shō manuscript in fact has only 1,553 songs. Because the Tajima and the Nakayoshi manuscripts include one song at the very end that is not in the Shō, we can surmise that the Aniya manuscript had had one extra song inserted at the end of the last volume. This is reasonable, given that some songs of volume 22 appear to have been in use by the Shuri government until the kingdom was abolished by the Japanese in 1879, and the Aniya family was responsible for their performance. Aniya-descended manuscripts all show that their original must have shown other signs of editing, such as punctuation, voicing, and breath marks as well.14 Another discrepancy between the various manuscripts that complicates the count is the fact that in the Aniya-descended copies, what is song 391 in the Shō manuscript is not separate, but instead is written as the last six verses of song 390 (Hokama and Hateruma 2002). This throws off the song count, but because the Aniya texts have one extra song at the end (see previous footnote), the final song count is the same as that of the Shō manuscript. Thus the copyists might not have caught the discrepancy as they copied the texts. However, one Aniya-descended text, the Nakayoshi manuscript, has a note, possibly added by Ganaha Tomoyoshi when he compared the Nakayoshi text with the Shō manuscript in 1934, to the effect that the second mata in song 360 of the Nakayoshi text corresponds to an iti in the Shō text, meaning it marked a separate song. All the post-war print editions have 1,554 songs, with that mata shown as an iti. It is not clear whether the mistake was in the Aniya or Aniya second manuscripts, or whether it is a scribal error by one of the Meiji period copyists.
14
students of Ryūkyūan linguistics, and must also contribute to the field of Japanese
historical linguistics as early sources of Ryūkyūan, the only known sister language to
Japanese.
Over the centuries, several copies have been made of the OS, both manuscript and
print versions.15 The oldest extant copy is a manuscript made in 1710, after the original
document was destroyed along with many other treasures in a fire at Shuri Castle in 1709.
The destroyed document itself may have been a copy, since the first original may have
been lost in an earlier fire: in 1660, Shuri Castle’s central hall, the Seiden, was destroyed;
and another disastrous fire struck in 1690 (Kerr 2000:221). By all accounts, in 1710, the
royal government rushed to make at least two new copies, using as their original the
Gushikawa manuscript, which had been kept at Nakijin Castle in Okinawa's northwest
Motobu peninsula. The first of these copies was kept at Shuri Castle, and is known as the
Shō manuscript (尚家本, shōke-bon), since it belonged to the royal family, surnamed Shō;
the second copy (安仁屋本, aniya-bon) went to the Aniya family, charged with performing
the songs in ceremonies. Whether the Gushikawa manuscript was itself a copy of a
manuscript possibly lost in 1660 or 1690; or a copy of the one lost in 1709; or whether it
was, in fact, the original document itself compiled between 1531 and 1623, is impossible
to say because it disappeared after the Shō and Aniya copies were made. The Shō and
Aniya manuscripts, themselves copies of a now-lost original, itself possibly a copy, are
the originals of all the texts of the OS known in the 20th century, though not many of
those texts survived the fires of the Battle of Okinawa in 1945.
A miracle of historical circumstances preserved the Shō manuscript from that
devastation, and today it lies in the Okinawa Prefectural Museum. Its preservation may
be connected with its history: the Shō manuscript seems to have been little used by
omoro scholars of the late 19th and early 20th centuries, perhaps because it remained
among the private papers of the Shō family. Kept safe in Shuri Castle through most of the
war, and hidden by castle stewards along with several other royal treasures before the
bombardments of early April, 1945, the manuscript was looted by the U.S. troops who
entered and secured the castle that month. A Commander Carl W. Sternfelt, in his
15 The lineage of the OS manuscripts is a standard part of any treatise on the OS. Hokama Shuzen is the author of the sources used here: Hokama 1972, 2000, and Hokama and Hateruma 2002.
15
position as Chief Censor of the Tenth Army, confiscated the manuscript as well as
several, perhaps all, of the other artifacts. Sternfelt’s possession of these items came to
light after his return to the U.S. when he brought them to experts at Harvard University
and Harvard’s Yenching Institute for identification and evaluation. Sternfelt may even
have previously made a trip to Tokyo and met with Nakahara Zenchū and others in an
attempt to establish a value for the items (Hokama, in the introduction to the Teihon,
2002). Yoshizato Hiroshi, an Okinawan expatriate living in New York, hearing of this
through connections at Harvard, attempted to contact Sternfelt, without success. He
passed on the information to Sergeant William T. Davis, once stationed in Okinawa, who
came seeking sanshin lessons in the early 1950s. Yoshizato and Davis cooperated in
searching for other Okinawan treasures by appearing on radio shows. Together they
informed the State Department, and eventually U.S. Customs agents seized the treasures
still in Sternfelt’s possession.16 Sergeant Davis personally returned the items to the
Okinawan government on behalf of President Eisenhower on May 26, 1953, amidst
celebrations of the one-hundredth anniversary of Admiral Perry’s arrival in Naha harbor
(Honan 1997, Nakasone 2002). An additional note of serendipity to this story lies in the
fact that Yoshizato had acted as host to Ifa Fuyū during his visit to San Francisco in 1929,
and the two had spent many hours talking, among other things, about the Shō manuscript.
That experience may well be what enabled him to recognize the items Sternfelt was
holding, and to engineer their eventual return to the heirs of the Chūzan kingdom.
The story is not so happy for the Aniya manuscript of the Omoro sōshi and its
copies. The Aniya manuscript itself was lost in 1945. The family had buried its treasure
in hopes it would thus survive the coming conflagration, but their property in Ginowan
16 Specifically, the other items recovered in 1953 included the royal government official histories Chūzan seikan (中山世鑑, 1650) and the Chūzan seifu (中山世譜, a kanbun version of the Chūzan seikan); the Konkōken-shū (混効験集, ‘Dictionary of confused meanings’, 1712); three tama kawara (comma-shaped pendants similar to magatama) and an elaborate hair pin (kanzashi) once worn by the chief priestess, Kikoe-ōgimi; and “sixty black-and-red lacquered tablets inscribed in gold with the names of the Ryūkyūan kings” (Nakasone 2002:3). Commander Sternfelt may have had more in his possession at one time, since he is known to have sold some items, and his family auctioned off several artifacts after his death in 1975. Kishaba Shizuo, chairman of the Ryūkyū American Historical Research Society (devoted to finding and returning artifacts important to Okinawa’s history and culture), speculates that he may even have had the royal crown, a rounded cap made of velvet and gold, sewn with pearls and precious gems, and with a gold pin inserted from side to side near the top, since it had also been hidden with the other items in the Shuri Castle grounds. The crown, however, has not been recovered.
16
was commandeered as part of a U.S. military base. Decades later, when family members
finally received permission to search for what they had buried, all they found were parts
of the box it had been buried in. Aniya Masaaki, a descendent who carries on the tradition
of singing omoro, tells a story in which his teacher, Yamauchi Seihin, suspected the
Aniya manuscript may also have been found and rescued, and now lies in some archive in
the United States, but Yamauchi was never able to confirm this story.17
The Aniya manuscript was the original for several copies of the OS made in the
19th century, of which only two, plus one volume of a third, are still extant. There are no
known “first generation” copies of it in existence, and the story behind the extant
manuscripts is one of copies of copies. Still, they are valuable because they do show hints
of how the Aniya manuscript differed from the Shō.
First, a copy of the Aniya manuscript was ordered by the royal government just
before its dissolution in 1879, apparently in order to ensure the manuscript’s survival in
the upheavals of the Japanese takeover. This new copy, called the “Aniya second
manuscript” ( 安 仁 屋 副 本 aniya fuku-bon) to distinguish it from the earlier “Aniya
manuscript,” later served as the original for several copies made by the prefectural
government. It was copied for inclusion in the Ryūkyū shiryō (1888-1892), a project
sponsored by the prefectural governor, Maruoka Kanji (g. 1888-1892), to provide a
collection of Ryūkyūan study materials in response to scholarly interest from Japan
concerning the relatively unknown natural and human phenomena in the new prefecture.
Tajima Risaburō (1869-1931) may have used the Aniya second manuscript. He wrote on
the inside front cover of the last of his notebooks into which he had copied the OS that he
had used both “an old copy from the old days, and one made after the Satsuma takeover,”
implying he had access to two manuscripts, one of greater and one of lesser antiquity, 18
though whether he was referring to the Aniya manuscript made in 1710 and the Aniya
second manuscript made in the 1870s is unclear. Like the Aniya manuscript, the Aniya
second manuscript was lost by the end of World War II. In addition, the reasons for the
17 This story was related by Aniya Masaaki in 2006 to a group including this writer, at a meeting of the Omoro Kenkyū-kai in Shuri, Okinawa. Yamauchi Seihin, Aniya’s teacher, had received a message from a man in Kyushu who claimed to have information about the lost manuscript. When Yamauchi tried to make contact with him, he found that the man had died suddenly, leaving no word behind.18 http://www.lib.u-ryukyu.ac.jp/digia/tenji/iha/h7220.html#top1
17
additional copies made by the prefectural government under Maruoka, their intended
recipients, and their current whereabouts, are all today unknown.
Figure 1. First page of photographic reproductions of the Shō manuscript in the Teihon Omoro sōshi (2002), showing all of songs 1-4 and part of song 5. The songs are numbered for the reader’s convenience; numbers in parentheses indicate the page of the original.
Tajima, a middle school teacher of Japanese and pioneering scholar of Ryūkyūan
linguistics, copied the entire text of the OS from Ryūkyū shiryō into his gogaku shiryō
(‘linguistics papers’) notebooks in 1893. The Tajima manuscript is significant for two
reasons: first, Tajima was the first to replace words such as kuni and naka (‘country’ and
‘middle’), written in the Shō and Aniya manuscripts in kana, with their kanji (Chinese
character) equivalents, thus increasing the selection of kanji in comparison to the original.
Most modern print editions of the OS continue this tradition, and go even further in
18
replacing many additional kana-spelled words with kanji accompanied by the original
kana as furigana alongside. Second, the Tajima papers, which also contained his
handwritten copies of other valuable documents, such as the Konkōken-shū ( 混 効 験 集 ,
‘Collection of confused meanings,’ 1712) and the Nyōkan Omoro sōshi (女官おもろさうし
‘Women courtiers’ anthology of omoro,’ 1706-1713), are especially important because
Tajima passed them on to Ifa Fuyū (1876-1947), the revered father of Okinawan and
Ryūkyūan studies, in 1903. Iha used Tajima’s papers extensively in his own studies,
basing his 1925 print edition of the OS on them, and when he became head of the
prefectural library, he used them to make copies for the library’s Ryūkyūan materials
collection. Although most of this collection was destroyed along with the prefectural
library in 1945, the Tajima manuscript of the OS is among the remaining volumes of
Tajima’s notebooks that are now part of the Ifa Fuyū collection at the University of the
Ryūkyūs library.
The second extant Aniya-descendant manuscript, chronologically speaking,
consists of only the third volume of the OS. Known as the Kara-no-ya bon (からの舎本), it
belonged to Okakura Saburō. The inside cover states that it was a copy made from a
manuscript in Basil Hall Chamberlain’s possession, which Chamberlain himself, during
his stay in Okinawa in 1893, had had made from a manuscript belonging to then mayor of
Shuri, Nishi Tsuneo. Both the Nishi manuscript and Chamberlain’s copy have been lost,
and Tsukuba University Library now holds the Kara-no-ya manuscript.
Finally, the last pre-war manuscript of the OS still extant is the Nakayoshi
manuscript, so named because it belonged to Nakayoshi Chōsuke. The manuscript is
clearly in two different hands, because the copying work was shared by Nakayoshi
Ryōkichi (probably Nakayoshi Chōsuke’s father) and Onga Tomohiro, sometime
between 1900 and 1911. This manuscript bears notes saying it was examined in a
comparison with the Shō manuscript in 1934, by someone signing himself as Ganaha
Tomoyoshi (我那覇朝義), apparently an associate of Iha Fuyū’s assigned with the task of
finding any discrepancies between the two texts. This document also came into Iha’s
possession, and is now part of the Iha collection at the University of the Ryūkyūs library.
19
2. Print editions of the Omoro sōshiThere is a huge body of work done on the Omoro sōshi in the 20th century, not least in
the basic work of publishing annotated and indexed versions of the text to help scholars
decipher it for interpretation and analysis. After Iha Fuyū’s initial printing of the Kōtei
Omoro sōshi in 1925 (copies can still be found, but they are exceedingly rare), the
landmark editions for this purpose start with Nakahara Zenchū and Hokama Shuzen's
Kōhon omoro sōshi (Kadokawa, 1965), published along with a second volume, Omoro
sōshi jiten, sōsakuin, a dictionary and index for the OS. Although the dictionary has been
superseded by the scope and scholarship of the Okinawa kogo daijiten (Dictionary of Old
Okinawan, 1995), the index remains extremely useful as it lists every occurrence of every
word in the entire OS.
The Kōhon OS served as the basis of Hokama Shuzen and Saigō Nobutsuna's
1972 edition of the Omoro sōshi. This 1972 text is somewhat controversial as instead of
perfectly preserving the mostly kana spellings of the OS, the editors boldly inserted large
numbers of Chinese characters into the text and put the replaced original kana to the side
in small print as pronunciation (furigana) readings. As mentioned earlier, the Shō
manuscript has only minimal kanji and for only the most basic words, such as “moon,”
“mountain,” and “person,” so the mixed kana-kanji style in the 1972 edition radically
changes the appearance of the songs (see Figure 2).
Furthermore, Hokama and Saigō added line breaks that are not in the Shō
manuscript. Though the additions may sometimes make decipherment a little easier (if
one accepts their interpretations), the breaks are also often somewhat arbitrary, and do
not always add to understanding of the poems. Finally, in this edition all the songs are
annotated with translations and definitions of words, in a method that is very popular in
Japan for classical texts of all types.
20
Figure 2 The first page of songs in the Omoro sōshi edited by Hokama and Saigo (1972), showing songs 1-3 and most of song 4. The top row contains definitions and editors’ annotations. The middle and bottom rows are omoro, with each song numbered above its iti line, and the fushina in smaller font positioned to the right of the iti lines. The numbers in parentheses below the last line of some songs give the song number of chōfuku, or repetitions of the song, elsewhere in the text. Note how the added word and line breaks, as well as kanji with furigana, change the appearance of the text compared to Figure 1.
Other print editions of the OS have appeared, including a spectacular
photographic reproduction edition in 1979-80 (Hirugi-sha) made of the same kind of
paper and bound in the same fashion as the Shō manuscript; but of interest here is the one
21
Hokama Shuzen published in 2000. This is a bunko-bon-sized edition in two volumes,
which also gives the songs in the kanji-kana mix, and with the line breaks of the 1972
edition, plus updated translations and definitions. Despite the controversy regarding
orthography and line breaks, this is the main text that I rely on for meaning in this paper
because it offers the most recent translations of the songs.
きこゑ大きみがおもろ 第一 首里王府の御さうし 嘉靖十年
1 ① 一 きこゑ 大きみきや おれて あすひよわれは 1 ② てにか した たいらけて ちよわれ 1 ③ 又 とよむ せたかこか 1 ④ 又 しよりもりくすく 1 ⑤ 又 またまもりくすく
2 ① 一 きこゑ 大きみきや おれて あすひよわれは 2 ② かみ てたの まふりよわる あんしおそい 2 ③ 又 とよむ せたかこか 2 ④ 又 首里もりくすく 2 ⑤ 又 またまもりくすく
3 ① 一 きこゑ 大きみきや 世そうせち みおやせは 3 ② 千万 世 そわて ちよわれ 3 ③ 又 とよむ せたかこか 3 ④ 又 きこゑ あんしおそい 3 ⑤ 又 とよむ あんしおそい 3 ⑥ 又 首里もりくすく 3 ⑦ 又 またまもりくすく 3 ⑧ 又 大きみす まふらめ
Figure 3 The first three songs in the electronic version of the Omoro sōshi prepared by Takahashi Toshizō (2005). The three lines at the top are the volume heading; the first number indicates the song, the circled number indicates the line. Note that refrain lines are indented and have neither an iti nor a mata. Takahashi 2005 does not include fushina (melody names) given in the Shō manuscript, the notes concerning choreography to be found in Volume 9, or any information regarding repeated songs.
Finally, mention must be made of an electronic copy of the OS obtained by this
writer from Takahashi Toshizō in 2005 (see Figure 3). It has proved invaluable in
simplifying the search process necessary for finding grammatical and vocabulary details.
The Takahashi text does not follow the line breaks in the Shō manuscript, or those in any
Hokama edition; instead, main verses take up one line each, and any portions analyzed as
refrains are on separate lines. The drawback in this approach is that his decisions on
22
which portions of the originals are refrains may be incorrect. However, in large part, this
thesis accepts Takahashi’s judgment on the breaks between main verses and refrains.
3. Overview of scholarship on the Omoro sōshiWith the publication of printed, annotated editions, scholarship on the OS has taken great
strides in the last forty years. Much of the work has involved examining the songs one by
one to uncover their meaning and structure, analyzing individual words for meaning and
etymology, and trying to parse the grammar and phonology. A good example of this type
of work is Hokama's Konkōken-shū kōhon to kenkyū (混効験集校本と研究 , Variorum text
and study of the Konkōken-shū, 1970), which provides a full typeset text of the
Konkōken-shū (1712), plus page-by-page photographs of the original, and then a fully
annotated alphabetized (or rather, kana-ized) index of all the entries in the dictionary. His
Omoro-go jisho (おもろ語辞書, Dictionary of the language of omoro, 1972) is a refinement
of the earlier book, without the full text or the photographs of the earlier manuscript,
listing only those words that appear in the OS. In it, he gives further, more detailed,
annotations on each of the entries, presumably reflecting his work in preparation for the
1972 edition of the OS. Together, Hokama's OS and his editions of the Konkōken-shu
laid the groundwork for future efforts.
Besides Hokama, a group of scholars working out of the University of the
Ryūkyūs (Ryūdai) became prominent in the field of OS studies. Foremost among these is
Ikemiya Masaharu, who has written volumes on the literary and cultural aspects of the
OS. He also edited a collection of articles, Omoro sōshi seikashō (1978), written by the
members of the Omoro sōshi Kenkyū-kai, a study group formed in 1968 and devoted to
understanding the OS. In each article, reflecting the format of the study group’s
methodology during their meetings, the author takes a song from the OS, analyzes the
words, constructs a meaning, and discusses matters of note such as phonology, grammar,
or special cultural meanings in the songs. Ikemiya 1995 is a study of the Konkōken-shū
along the lines of Hokama 1970 that takes advantage of the 25 years of scholarship that
separates them. It is without the photographic reproductions, and Ikemiya incorporates
his annotations regarding the terms and their antique definitions directly into the text.
23
Tamaki 1981 is an extremely useful analysis of the structure of the songs in the
OS. He takes a look at the mechanics of their transcription, lays out a theory of how the
songs as found in the manuscript are abbreviated, and provides templates for
reconstructing the songs as they may actually have been sung. For example, the OS does
not usually repeat refrains, but instead gives them only once in the first verse, marking
the start of subsequent verses with a mata (又, ‘again’) and giving only the words unique
to each verse. The reader/researcher must be aware of this, and be ready to re-insert the
refrains where they belong into the songs to recreate them in order to get an accurate
reading.
No study of the OS can be complete without reference to Takahashi Toshizō's
authoritative Omoro sōshi no dōshi no kenkyū (Study of the verbs of the Omoro sōshi,
1991), an analysis of the forms and functions of verbs in the OS. He also produced the
companion Omoro sōshi no kokugogaku-teki kenkyū (Linguistic study of the Omoro
sōshi, 1991), an overview of the phonology and syntax of the OS. Both of these works
provide a synthesis of much of the study of the OS done prior to 1991.
An interesting addition to scholarship on the OS is the Omoro kanshō, a series of
a hundred articles providing translations and explanations of omoro songs that ran in
Gekkan Gengo between 1984 and 1993.19 Besides a Japanese translation, each song is
accompanied by an English translation done by Christopher Drake; a literary
interpretation by Higa Minoru; and a linguistic analysis by Nakamoto Masachie. This
series provides an enlightening overview of the current understanding of the OS.
Finally, the Okinawa Kogo Daijiten (1995; henceforth OKDJ), mentioned above,
is admittedly the most recent and the best of the dictionaries concentrating on the
language of the OS and in traditional performance arts such as kumiodori and ryūka.
However, care must be taken in using it, as the editors included “ghost forms.” These are
entries for words that do not actually occur in the source texts, but are standardized forms
used as umbrella terms for a variety of spellings or conjugations of a given word. The
dictionary’s index contains all the spelling and conjugation variations of words, and the
19 Although there were indeed 100 articles for the Higa, et al., series, omoro 8: 396 is presented twice, making the total songs covered only 99. In the second version of song 396, the Japanese and English translations, as well as Higa’s and Nakamoto’s articles, differ from the first.
24
reader is referred to a single main entry listed under a standardized form that sometimes
does not exist in the language. With regard to this thesis, the OKDJ has one additional
drawback. Hokama Shūzen was a member of the editorial committee that produced the
dictionary, with the result that both the main resource for translations of the omoro (i.e.,
Hokama 2000), and the main dictionary for checking meanings of words, are both the
work of the same scholar. Granted, an outside source would be useful to counterbalance
Hokama’s interpretations of the OS, but no such source really exists yet. Producing such
a source lies in the future, perhaps in the shape of an annotated English-language
translation of the OS.
B. Orthography of the Omoro sōshi Tradition says writing came to the Ryūkyūan kingdom in the 13th century, first brought
by a Japanese Buddhist missionary monk named Zenkan who decided to stay after his
ship was wrecked there around 1270. Though Zenkan seems to have been real, this story,
possibly apocryphal, probably reflects the fact that the spread of Buddhism served an
important role in bringing “arts, crafts, and ceremonial” to the kingdom, as well as
introducing Japanese literary traditions (Kerr 2000:56). At any rate, by the time the first
volume of the Omoro sōshi was written down in 1531, the Ryūkyūans had a well-
established spelling system for their language. Their character inventory was fairly
straight-forward, relying mainly on the kana of the Japanese syllabary, with the addition
of only a few basic Chinese characters. Complications arose in the application of the
rules for spelling and pronunciation of written words, since Okinawan then and now
contains sounds for which the Japanese syllabary has no representative kana, and has
undergone sound changes that resulted in variant spellings for affected words. The rules
for spelling could vary depending on the time period and the scribe, though the Omoro
sōshi itself follows a fairly consistent set of spelling rules.
1. Japanese historical spelling (rekishi-teki kanazukai)A short description of the contemporary Japanese kana spelling system can shed light on
the Ryūkyūan system. The so-called “50-sound table” (gojū-on zu) that makes up the
Japanese kana syllabary developed out of the man’yōgana writing system of Old
Japanese (OJ). By the end of the Heian period (794-1183), a series of phonological
25
mergers and splits had resulted in spelling overlaps, that is, multiple kana could represent
the same sound (a merger), or one kana could have two readings (a split), depending on
whether it was word-initial or word-medial. These shifts caused something of a
disconnect between the spoken and written languages, familiar to students of Classical
Japanese; this disconnect was largely cleared out with the Japanese spelling reforms of
1946, though it is still visible in the kana spellings of the particles wa (written with は ,
<fa>, ‘SUBJECT MARKER’), e (へ, <fe>, ‘ALLATIVE’), and o (を, <wo>, ‘OBJECT MARKER’).
The phonological splits affected those word-medial syllables beginning with /ɸ/,
the voiceless bilabial fricative, romanized here with <f->,20 so that word-medial and
word-initial spellings in <f-> came to be pronounced differently. Medial /ɸ/ merged
with /w/ when followed by a front vowel, so that after the 13th century, any word-medial
syllable /wi/ or /we/ could be spelled either as <wi, we> or as <fi, fe>. Because Old
Japanese did not distinguish the homorganic syllable /wu/ from /u/, there exists neither
man’yōgana nor kana <wu>. Instead, earlier /ɸu/ merged with /u/, and medial syllables
read [u] could be spelled as <u> or <fu>.
In addition, /w/ merged with /j/ when followed by a front vowel. That is, /we/ >
/je/ and /wi/ > /ji/ (in effect /i/, since there was no distinction between /i/ and
homorganic /ji/). Though distinguished in the man’yōgana script of Old Japanese, /e/
and /je/ had merged into /je/ before the development of the simpler kana writing system,
with the result that no separate kana ever existed for them.21 All instances of the spelling
<e> (え) were read /je/. Thus, once the merger of /we/ > /je/ was completed, the syllable
could be written as <e> or <we>, and, when word-medial, also <fe>, given the earlier
merger of medial /f-/ into /w-/.22 Similarly, /i/ could be spelled <i>, <wi>, and, medially,
<fi>. Probably through a loss of distinctive rounding, earlier /o/ and /wo/ merged into
20 Lower-case letters inside diamond brackets (< >) indicate romanizations corresponding to the original kana; upper-case letters, with or without diamond brackets, indicate Chinese characters.21 Thus the kana inventory poem Ametsuchi, included in a collection in 947, has 48 characters, but the more famous Iroha, written over a century later around 1079, has only 47. The extra character in Ametsuchi is a second <e>, in a word that exhibited variant spelling even then. 22 The pronunciation of this syllable lost its palatal element /j/ sometime during the Tokugawa period; however, interesting traces of it remain, such as in the English word for Japanese currency, yen (pronounced [(y)en] in modern standard Japanese), and alternate spellings of Yedo and Yebisu for Edo (Tokugawa period Tokyo) and Ebisu (one of the Seven Deities of Good Fortune, a district of Tokyo, and a popular beer).
26
/wo/, meaning this syllable developed three possible spellings, <o>, <wo> and (medially)
<fo>.
Another set of phonetic-level sound changes affected /s/ and /t/, depending on
their environments. Fricatives /s/ and /z/ palatalized to [sh] and [zh] before front vowels
/i/ and /e/. The stop /t/ affricated before high vowels /i/ and /u/, becoming [tshi] and [tsu];
as did /d/, becoming [dzhi] and [dzu]. Because these changes affected every instance of
/si/, /se/, /zi/, /ze/, /ti/, /di/, /tu/ and /du/ without exception, there was no need to develop
new kana for these sounds, or to develop a technique for spelling them. By late
Kamakura, the syllables represented by kana <zi> and <di> (じ,ぢ), had fallen together, as
had <zu> and <du> (ず,づ), and each pair could be used interchangeably.23
Table 1. Summary of correspondences between kana and their pronunciations following sound changes in Middle Japanese.
<f> column <w> column <ø> column <y> columninitial medial initial medial initial medial initial medial/ɸa/24
<fa>/wa/<fa>
/wa/<wa>
/wa/<wa>
/a/<a>
/a/<a>
/ja/<ya>
/ja/<ya>
/ɸi/<fi>
/i/<fi>
/i/<wi>
/i/<wi>
/i/<i>
/i/<i>
/ɸu/<fu>
/u/<fu>
/u/<u>
/u/<u>
/ju/<yu>
/ju/<yu>
/ɸe/<fe>
/je/<fe>
/je/<we>
/je/<we>
/je/<e>
/je/<e>
/je/<e>
/je/<e>
/ɸo/<fo>
/wo/<fo>
/wo/<wo>
/wo/<wo>
/wo/<o>
/wo/<o>
/jo/<yo>
/jo/<yo>
23 These are collectively known as the yotsu-gana (‘four kana’). In some Tohoku dialects, these four have fallen further together when /i/ and /u/ both centralized to /ї/, leaving us with only /dzhї/, at least in medial position.24 Slanted brackets (//)indicate phonemic-level segments, diamond brackets (<>) indicate spelling. Bolded lines group pronunciation sets by syllable initial /ɸ-, w-, ø- or y-/ (based on Serafim 1977:5).
27
Table 2: Summary of possible spellings for each column affected by mergers and splits. Sound Possible spelling
word initial word medial/ɸa/25 <fa> —/wa/ <wa> <fa> <wa>/a/ <a> <a>/ja/ <ya> <ya>/ɸi/ <fi> —/i/ <wi> <i> <fi> <wi> <i>/ɸu/ <fu> —/u/ <u> <fu> <u>/ju/ <yu> <yu>/ɸe/ <fe> —/je/ <we> <e> <fe> <we> <e>/ɸo/ <fo> —/wo/ <wo> <o> <fo> <wo> <o>/jo/ <yo> <yo>
This brief description of the Japanese spelling system (in use until 1946) does not cover
conventions for long vowels, punctuation, or affricated or palatalized syllables other than
<si>, <se>, <ti>, or <tu>.
2. Kana inventory and spelling conventions in the Omoro sōshiSo, in summary, the spelling system inherited from Japan by Ryūkyūan students of
writing did not distinguish word-initial <wi> and <i> (reading them both as /i/), <we>
and <e> (/je/), or <wo> and <o> (/wo/). Nor did it distinguish word-medial <i>, <fi>, and
<wi> (/i/); <e>, <fe> and <we> (/je/); <o>, <fo>, and <wo> (/wo/); <u> and <fu> (/u/);
or <fa> and <wa> (both /wa/). The scribes of the OS followed this system, using certain
<w->, <f-> and vowel kana interchangeably in medial and some initial syllables, and
their inventory included 47 kana, one less than the 48 (including the now little-used <wi>
and <we>, plus <N>) in use in contemporary Japan. The character <e> ( え ) does not
appear in the OS, with <we> (ゑ) appearing where <e> might otherwise be expected. In
addition, Okinawan scribes used spelling solutions similar to the solutions used in Japan
for syllables for which there are no kana. Some of these spellings were used for
25 Note that this table captures the reason why <wi> and <we> (and <wo>, with the exception of the object marker) are no longer used in the modern Japanese kana inventory: In all cases, they are replaced by <i> and <e> (and <o>), reflecting modern pronunciations.
28
borrowings from Chinese or Sino-Japanese, but they also reflected sound changes in the
language, such as palatalization and vowel-raising.
Table 3: Kana inventory in the OS and corresponding romanization used in this paper. Note that this is not a phonemic inventory of OOk.
あ a か ka さ sa た ta な na は fa ま ma や ya ら ra わ wa
い i き ki し si ち ti に ni ひ fi み mi り ri ゐ wiう u
く ku す su つ tu ぬ nu ふ fu む mu ゆ yu る ru
け ke せ se て te ね ne へ fe め me れ re ゑ weお o こ ko そ so と to の no ほ fo も mo よ yo ろ ro を wo ん N
が gaざ za だ da ば ba
ぎ gi じ zi ぢ di び bi
ぐ gu ず zu づ du ぶ bu
げ ge ぜ ze で de べ be
ご go ぞ zo ど do ぼ bo
a. Spellings showing palatalization Because the language of Old Japanese, upon which the kana syllabary was based, did not
have affricated or palatalized consonants, the orthographic system did not provide
separate kana to distinguish stops and affricates, or plain stops and palatalized stops as
separate phonemes. In Japanese, syllables such as [cha], once they developed in the
language, were spelled with digraphs in which the first kana provided the necessary
consonant, and the second the vowel sound. (In this case, <tiya>, ち や . The font size-
difference is a modern innovation.) For the same reason, neither did the Japanese
syllabary have specific kana for spelling syllables with palatalization, but scribes
developed a similar solution to spell such syllables, and Okinawan scribes adapted these
techniques in their own writing.
Perhaps the most distinctive sound change clearly visible in the spelling of the OS
is that of palatalization, specifically progressive palatalization, in which a high front
vowel /i/ causes a following stop to palatalize. Thus the frequent spelling of the subject
marker ka (i.e., ga) as <kiya>, as in omoro 1:
29
14ITI kikowe UFU-kimi-kiya orete asuf-i-yowar-e-fa
renowned great-priestess-SUBJ descend-SUB play-INF-HON-EV-CON
Kikoe-ōgimi, when you descend (into trance) and dance (1)
Just as in Japanese, where the spelling <kiyo> gives the reading /k jo/, so <kiya> should
be read /kja/ in Old Okinawan. The digraph constitutes a single syllable, with one kana
providing the consonant, and the second providing both a yod (the sound [j]) and the
vowel.26
Serafim 1977 provides a rule for this palatalization:
15ø → j / jVC__
That is to say, “a palatal element [j] in a preceding syllable causes the addition of a
palatal element [j] in the following syllable” inserted immediately after the initial
consonant (p. 15). Serafim argues that *i phonetically must have been [ji], since it
triggered palatalization.27 We can see this rule working in the case of the subject marker
ka/ga in Example 14 above, which phonetically changes from [ga] to [gja] under the
influence of the final /i/ in the second syllable of kimi (pronounced [kjimji], with an
extremely high front vowel).
This rule of progressive palatalization is regular in the OS, except in the case of
the kakari particles (as well as the subject marker ga), which do not always appear
spelled as expected in the palatalization environment. Compare the following excerpts
from different omoro, showing alternate spellings of both su and do in similar
environments:
26 The word kimi is also frequently followed by the particle spelled <ka> (or <ga>), but only in 66 out of a total of 301 tokens (counting only kimi plus <ka, ga, kiya, giya>). On the other hand, the spelling <kiya> (or <giya>) for this particle never occurs except after /i/, establishing that <ka> and <kiya>, in this environment, are in free variation, and are alternate spellings for the same word. One spelling (<kiya>) reflects a pronunciation that occurred in only one environment; the other (<ka>) is a standardized spelling that could be used anywhere. 27 This is in contrast to *e, which once it had raised to [i], still lacking the palatal element, did not trigger (regressive) palatalization (with some exceptions, discussed below). Thus in Modern Okinawan, words with *ki, *gi, *ti, and *di are all now [tshi] or [dzhi]; and we find the contrast between chiN (‘kimono’, :: J kinu) with tii (‘hand’, :: J te).
30
16MATA UFU-kimi su mafur-am-e
great-priestess KP protect-TENT-EV/IMP
It is the Great Priestess who surely will protect (the King) (3)
17[refrain] okiyaka-moi ya kimi siyo mabur-i-yowa-m-e
Ogyaka-HON TOP priestess KP protect-INF-HON.AUX-TENT-EV.KM/IMP
It is the Priestess who will surely protect the Great Ogyaka!(16)
These examples show su spelled once as <su> (Example 16) and once as <siyo>
(Example 17). Both spellings follow the honorific title kimi (c.f. J kimi ( 君 ), ‘lord,
priestess’; interpretation depends on context). Presumably, the former is a
morphophonemic (standardized) spelling, while the latter is a phonemic spelling
reflecting the pronunciation of su as [sju] or [shu] following [i].
Particle do shows a similar alternation in the following two lines, with two
spellings, one of which is phonetic, of the particle in two similar environments (and
accepting that dakuten are not consistently given in the manuscript):
18[refrain] nasaikiyo ka ikiyo-i do28 mat-i-yor-u
father29 TOP pray-INF KP wait-INF-HON.AUX-ATTR.KM
It is the fatherly King’s prayer that (we) await(1308)
19[refrain] nasafekiyo ka ikiyo-i tiyo mat-i-yor-u
father TOP pray-INF KP wait-INF-HON.AUX-ATTR.KM
It is the fatherly King’s prayer that (we) await(690)
Further analysis shows that palatalized and non-palatalized spellings of
allomorphs of both su and do have nearly perfect complementary distributions in the OS, 28 Note that any spellings giving voiced consonants (b, d, g, z) are based on information from the Aniya-descendent texts, since the Shō text gives no voicing punctuation. Both Takahashi 2005 and Hokama 2000 give the voiced spellings according to the Aniya tradition.29 The OKDJ notes that nasaikiyo (成さい子, ‘create-INF one’, i.e., one who creates) is a beautifying term for ‘father’; in the OS, it may refer to either an anji or the king.
31
except in the case of the spelling <su>. Table 4 shows how the kakari particle su, spelled
as <su>, appears in the non-palatalization environment (i.e., after the syllable-final
consonant /n/ and vowels other than [i]) more than 200 times, three times more often than
it appears following [i]. On the other hand, KP su spelled <siyo> or <siyu> appears in the
non-palatalization environment only three times,30 but 153 times following [i], a
difference of a factor of 50. Hokama 2000 translates all three of these tokens with koso,
though there are reasonable alternative readings for two of them, which could mean that
there are up to 150 times more tokens of the palatalized spelling following [i]. These
numbers show that while the spelling showing palatalization essentially never appears in
the non-palatalization environment, both spellings may appear in the palatalization
environment. This distribution of the spellings is likely a result of scribal knowledge of
the “correct” spelling of the particles. This spelling distribution is a fine example of how
the spelling system used in the OS reflects both phonemic and morphophonemic spelling,
resulting in some complexity.
Table 4: Distribution of spellings of (possible) KP su and do
Spelling Palatalization
environment
Non-palatalization
environment
su 68 205
siyo 131 2
siyu 22 1
ziyo 3 0
do 3 17
to 8 83
diyo 4 0
tiyo 8 1
In contrast to these particles, the subordinating gerund -(i)te, mentioned above as
of interest to this thesis, clearly and regularly undergoes this palatalization, and then goes
one better by also affricating. Serafim’s palatalization rule allows us to predict that this
30 Both <siyo> and <siyu> appear in the non-palatalization environment several more times, but in those cases the words cannot be interpreted as KP su.
32
suffix would always palatalize when the verb stem allowed the initial vowel /i/ (in
traditional Japanese grammar, the ren’yōkei, or infinitive, form of the stem). And indeed,
omoro 4 gives a clear example of a digraphic spelling for a palatalized -(i)te:
20
かさりうちちへみおやせKasari ut-i-tife mi-oyaseKasari(PN) defeat-INF-SUBGER HON-exist(IMP)Defeat Kasari and offer it up!
(4)
The gerund -(i)te spelled as <tife> shows that the insertion of the palatal element [j] also
caused the /t/ to affricate, or else we would find spellings such as <tefe> or <tewe>,
representing pronuniciations of /tje/ —but we never do. Often, however, the environment
for palatalization is not so explicitly clear as in Example 20. Example 21 gives an
example of the spelling <tife> that appears not to have the triggering environment:
21
いけなおろちへikena oro-tifethis.world set down-SUBGER
Setting down (on) this world(700)
We know that word medially, <fe> was read [je], thus this spelling also represents
a palatalized pronunciation. In fact, the syllable is an allomorph of original -te formed as
a result of environmentally-constrained sound changes involving the preceding /i/ and /s/.
Comparing the word orotife (‘lowering, setting down’) to its OJ counterpart, it is possible
to see how the forms relate:
22OJ örös - i - teOok oro - ø - tife
The spelled form can be explained by positing three steps, with progressive
palatalization (labeled here as PP) followed by loss of -i- and absorption of the [s]
(probably [sh]) into the resulting consonant cluster:
33
23Pronunication Process Spellingoros-i-te <orosite>*oros-i-tje PP <*orositife>*oros-tje i →øorotsh je i, s-absorption <orotife>
Thus we can confidently explain the form of this segment, and identify it as the
subordinating gerund. Additionally, we can infer that the verb stem must have contained
either or both original *s and *i whenever the gerund is spelled this way.
b. Spellings affected by vowel raising
i. <…u> alternating with <…o>
Serafim (1977, citing a 1975 manusript by Ikemiya Masaharu and the research of the
Omoro Sōshi Kenkyūkai at The University of the Ryūkyūs) and Takahashi (1991b:100-
118) separately note that *o and *u had merged by the time of the OS. Takahashi details
the evidence for this merger, showing that <u> and <o> spellings alternate in all
environments except after <t->, <d->, <s-> and <z->.31 Thus we can find orotife, the verb
from the section above, alternately spelled as <orutife>. He speculates that though *to,
*do, *so and *zo may have raised to /tu/, /du/, /su/ and /zu/ respectively, *tu, *du, *su
and *zu remained distinct as *u fronted to /ї/ in these cases (as it does in many modern
dialects). Thus we may see alternate spellings such as kuni/koni (‘country’); mi-kao/mi-
kau (‘HON-face’); se-no-kimi/se-nu-kimi (priestess name or title); kami-simo/kami-simu
(‘above-and-below’); yokaru/yukaru (‘good-ATTR’), and many others, but we rarely, if
ever, see confusions of spellings involving <tu/to>, <du/do>, <su/so> or <zu/zo>.
Thus the spellings of KP do do not show spelling confusions reflecting this vowel
raising; however, those of KP su do, but only in the palatalized spellings. The spellings
most obviously related to su and do, <su> and <to/do>, appear the most frequently, and
both can also be found following words ending in /i/. Insofar as initial /s/ and /t/ or /d/
blocked /o/ and /u/ from falling together, the particles’ non-palatalized spellings meet our
expectations: do is never spelled <tu> (which would have been read [tsї/dzї], wrong
consonant and vowel sound); and su never appears as <so> ([su], whereas the needed 31 Takahashi (1991a:117-8) lists 152 examples of variant spellings: 68 pairs where the original was <o>; 57 with original <u>; and 27 in which the original form is unclear.
34
sound was [sї]). In the case of the palatalized spellings, KP do is still never spelled with
the high vowel, appearing as <tiyo> and <diyo>, but never as <tiyu> or <diyu>. KP su,
however, does show alternation, with both <siyu> and <siyo> occurring. Omoro 93 and
665 are otherwise very different songs, but one line in each is similar. Examples 24 and
25 give the lines from each song that clearly show two alternative spellings of KP su.
Both are in the environment for palatalization:
24MATA kimi siyu mafur-i-yowar-e nusi siyu mafur-i-yowar-e
priestess KP protect-INF-HON.AUX-EV/IMP master KP protect-INF-HON.AUX-EV/IMP
It is the priestess who will protect; it is the nushi priestess who will protect (665)
25MATA kimikimi siyo mabur-e nusinusi siyo mabur-e
priestesses KP protect-EV/IMP masters KP protect-EV/IMP
It is the priestesses who will protect ; it is the nusi priestesses who will protect
(93)
The palatalized allomorphs of su do not retain a distinction in the spelling for the vowel,
since both <siyu> and <siyo> appear. By far the overwhelming majority of these
spellings of KP su occur in the environment for palatalization (only three do not, and
though Hokama 2000 analyzes them all as koso, i.e., that they are KP, the possibility
remains that they are other words entirely). Interestingly, the to-be-expected spelling
<siyu>, as a palatalized allomorph of <su>, is far outnumbered by the spelling <siyo>,
with only 22 tokens out of 164 respectively, possibly reflecting the original mid-vowel of
this particle.
ii. <…i> alternating with <…e>
On the other hand, in comparing a long list of possible spelling confusions that might be
evidence that /e/ and /i/ had merged (as they clearly have in Modern Okinawan),
Takahashi (1991b:70-99) shows that many supposed confusions actually involved
different forms of verbs, different words altogether, or words with unclear meanings,
35
except only in the case of syllables <we> and <fe>. Except for these syllables, he
concludes *e and *i were still distinct when the OS was written down.
Serafim 1977 also shows that in fact *e must have raised to /i/ in two stages, with
an initial “partial collapse of the <i:e> distinction” (p. 13) only in those syllables where
the consonant was the palatal /j/ (i.e., <we> and medial <fe>). Whether the raised vowel
was [ї] or [I], a palatal onset was enough to remove the distinction with *i ([ji]), but the
lack of “palatality” of other initial consonants prevented the remainder of syllables in the
-e column from merging with *i. Serafim gives the variety of possible spellings for the
verb corresponding to the Japanese lower bigrade verb soroe- (‘arrange’) to show that
<i>, <we>, and <fe> could represent the same syllable in the stem (Example 26):
26Spelling Number of tokens
<sorofe> 23
<sorowe> 15
<soroi> 12
He concludes that the final syllable of each spelling must have been pronounced
[ji]. The relative frequency of the spellings in Example 26 indicates either that a merger
had taken place recently enough for scribes to be aware of both pronunciations; or that
educated scribes were aware of the (historically) “correct” spelling for this verb.
At any rate, digraphic spellings such as <tiwe> and <tife> must have developed
for syllables distinguishable from /chi/, or they would not have been necessary (and these
syllables would have most often merely been spelled <ti>).
c. Other sound changes affecting spellingA relatively late process of regressive palatalization, in which stops followed by /i/
affricated, means that in Modern Okinawan, not only /ti/ and /di/ (as in Japanese), but
also /ki/ and /gi/, affricated to [chi] and [zhi]. Early evidence of this regressive affrication
may perhaps be seen in the OS in one verb that occurs hundreds of times (and is in many
of the examples in this thesis): the verb chowar-, spelled <tiyowar-> ( t-i-yowar-, come-
INF-HON-, ‘come’, ModOk g‛u-cyee-, :: ModJ irasshar-), in which the original initial k-
36
has affricated, resulting in its spelling using <ti>. Again, though general in Modern
Okinawan, in the OS, this process seems specific to this verb, since otherwise both
kikowe and kimi, words frequently appearing in the examples here, could have been
spelled with <ti> instead of <ki>, but never are (though they are in ModOk). Given how
complete this sound change is in Modern Okinawan, it is possible that it existed in (some
dialects of) spoken Old Okinawan, but had not entered into the possibly archaic (and
priestly) dialect used in the written Old Okinawan of the OS.
One final alternation appearing in the OS that is relevant to this paper is the
alternation between <r> and <t/d>. In Modern Okinawan, the dialects originating in Naha
and Itoman show this alternation, and it appears a handful of times in the OS. Thus KP do
may be spelled <ro> or <ru>, since /r/ does not block the vowel raising.
d. A note on dakuten and voicing in the Omoro sōshiIn common with most pre-WW II manuscripts, both Japanese and Ryūkyūan, the Shō
manuscript contains neither voicing punctuation (dakuten), nor any other kind of
additional information regarding word breaks or pauses. In contrast, all the Aniya-
descendant texts have both dakuten and punctuation, as well as notes regarding breathing
and meaning, indicating that the Aniya family had annotated their text to help in their
performances of the songs. Conservative print editions usually limit dakuten, if they are
included at all, to only those found in the extant Aniya-descendant manuscript copies.
However, as deciphering progresses and scholars become more sure of the meanings and
etymologies of words, additional words may be regarded as containing voiced
consonants. Iha Fuyū added voicing marks to his Kōtei Omoro sōshi (1925), and Hokama
2000 is very consistent in adding dakuten wherever he deemed them appropriate. (For
example, the subject particle is always spelled <ga> and <giya>, and never left as <ka>
or <kiya>, no matter what the manuscripts show.) The Teihon (2002), in its function as a
concordance of the differences between the extant manuscripts, includes two versions of
each verse: one as it appears in the Shō manuscript, with footnotes indicating missing or
illegible portions; and a second showing voicing (and other differences) appearing in all
the other manuscripts. The Takahashi electronic version of the OS used for this thesis
restricts itself to those dakuten found in Tajima and Nakayoshi, and this thesis reflects
37
this in its romanizations, adding in parentheses the voiced version of words only when
useful for clarity.
3. Chinese characters in the Omoro SōshiThe OS contains at least fifty Chinese characters (kanji) in the text of the songs. Note that
the capitalized spellings used in the transcriptions here are based, as far as possible, on
kana spellings for the same words (so judged because they occur in similar contexts)
found elsewhere in the text. Although such kana spellings are very likely to be readings
for the characters, they remain abstractions, since the possibility remains that they were
not read as transcribed here. If no kana spelling of a given word is available somewhere
in the text, then the given transcription is taken from known Okinawan pronunciations of
the words. If unknown, the transcription is based on Sino-Japanese. Usually the kana
spellings show the native Japonic pronunciation, but not always. In cases where there
seem to be two kana options, the least irregular option is used. For example, most sources
agree that the OS has three similar words, tosi, tose, and to, meaning ‘year.’ The first of
these is the most frequent, and the most unmistakeable, and thus has been used to
transcribe the character 年 where it appears in the text.
C. Physical format of the songsAs previously mentioned, the Omoro Sōshi is divided into twenty-two volumes, each
containing anywhere from 32 to 236 songs. Each volume has a distinct title that is
somewhat descriptive of the songs it contains; within individual volumes, songs may be
set off into groups headed by a short note naming the type of ceremony in which they
were used.
All the songs in the Shō manuscript share a similar format. Most of them are
preceded by a fushina (melody name) written in small kana; the lines of the songs are
headed with symbols to show the beginnings of the verses; and a very few songs are
annotated with notes about the kind of dance movement that should accompany the song
(also, see Figures 1 and 2 above).
The first song in the first volume of the OS provides a good example of the songs'
formats, and allows some illustration of the way individual omoro are formatted in this
paper. Were it transcribed in Roman letters, and written out horizontally from left to right
38
(instead of vertically and from right to left), omoro 1 as it is written in the Shō manuscript
looks like the following:
27 Aoriyafe-ga fusi
ITI32 kikoweUFUkimikiyaoreteasufiyowarefatenikasitatairaketetiyoware
MATA toyomusetakakokaMATA siyorimorikusukuMATA matamamorikusuku
(1)
The fushina is considerably smaller than the main text, and slightly indented
compared to the iti; the iti (一) and mata (又) symbols line up along the top margin, and
the text lines up below them. Lines that are not headed by an iti or a mata are indented so
all the song text lines are even. Note that iti and mata are always written with Chinese
characters, and that there is no gap between them and the line of text (Example 27,
written in Roman letters, does not quite reflect the layout in that detail). There are no
spaces between individual words, and a single word may be split between two lines
(though not in this omoro). The OS does employ hentaigana, or alternative kana forms,
though in the song above the only such kana is <wa> in yowarefa and choware (written
with the kana based on 王 [wa] ‘king’ rather than with わ, based on 和 [wa] ‘peace’). The
pages are not densely filled, with only five lines of text per page, plus possibly a fushina,
written very small and inserted to the right of the iti verse. Each line may contain up to
about twelve or fourteen kana.
1. Melody names, parallel lines and refrainsThe melodies of the omoro in the Omoro sōshi have not been preserved.33 In the text, the
only clues regarding melody lie in the fushina at the head of most of the songs (Ikemiya
32 Note that upper-case letters indicate Chinese characters, while lower-case indicate kana in the original. Italics indicate words or lines written smaller than the main text in the original manuscript.33 Except for five songs, passed down to Aniya Masaaki. Aniya sings omoro, but only the five he learned from his teacher, Yamauchi Seihin. Yamauchi in turn had learned them from his teacher, who had been a boy when the Shuri court was abolished, and last direct heir to the Aniya chant tradition. Aniya Masaaki does not sing with accompaniment, but a capella, which means that no record of the musical accompaniment (if any) exists.
39
1979), and in the use of the character iti (一, 'one') at the head of the first verse and mata
( 又 , 'again') for subsequent verses in each song. There does appear to be a weak
correspondence between the various fushina and the syllable counts of the songs' lines
(Valentine 1989), implying that the fushina is, in fact, the name of the melody to which
the song should be sung or by which it should be accompanied. Often, the same fushina
may be indicated for different songs, implying that a single melody would be used in a
variety of songs.34 Additionally, many of the fushina appear as lines in one or more
omoro, though the fushina for such omoro, if there is one, may be different. Most, if not
all, scholars agree that the iti and mata symbols in the OS mark where the melody
repeated. In addition, most lines with neither an iti nor a mata are generally accepted to
be either line completions that should be added to all subsequent lines in the omoro, or
are refrains which should be repeated at the end of each verse, or both (and the reader
must decide on the appropriate allocations). Supporting the idea that such lines are
usually refrains is the fact that, from song to song, these additional lines often show a
much greater degree of variability in terms of syllable count, format and content than do
main verses.
Although Valentine’s thesis showed that there were patterns of syllable counts
that corresponded to fushina, parallelism seems to be a much more important
consideration than strict syllable counts in these songs. The OS shows a very strong
tendency to use parallel structure, consisting of pairs of lines similar in structure and
syllable count that are either similar or contrastive in meaning. Two lines making up a
parallel pair may appear together in a single verse, in separate consecutive verses, or
(rarely) in alternation with a second parallel pair of lines in an A-B-A’-B’ pattern (where
A’ indicates the second part of the AA’ parallel pair). Possibly due to imperfect
transcription, or imperfect transmission of songs, a number of omoro lack the second line
of a pair that frequently appear together. Such songs may never have had the "missing"
line in them, but it seems plausible to assume they did, since it is possible that in the
34 Even today, the Okinawan classical uta-sanshin repertoire, which is descended from the ceremonial and other music played in the Ryukyuan court, allows one melody to accompany many different lyrics. Best known may be the melody Kajadefu, which may accompany dozens of songs, and last from 3 to 25 minutes, depending on the tempo (according to Norman Kaneshiro, uta-sanshin teacher at UH, in a personal communication in March, 2008).
40
abbreviated transcription method used in the Shō manuscript, lines were omitted because
the singer would know to repeat the second half of a parallel pair. It remains for future
study to determine whether such a practice was common, or can be detected in the
liturgical material from religious ceremonies still performed today, or still remembered
today even if no longer performed. Such material may provide material for the study of
the relation between song transcripts and song performances.35
In the case of omoro 1 above, the iti line and the first mata are a parallel pair
found frequently throughout the OS, as are the pair of lines headed by the second and
third mata characters. Kikowe UFU-kimi (or, in the Japanese reading kikoe ō-gimi,
corresponding to ModOk chifijin) was the title of the highest-ranking priestess in the
kingdom; toyomu setaka-ko (or toyomu sedaka-ko) is an alternate beautifying designation
( 美 称 辞 ) for the same priestess. Siyori mori kusuku (‘Shuri grove precinct’) and its
beautifying parallel ma-tama mori kusuku (‘true-jewel grove precinct’) both refer to
Shuri castle, the heart of the kingdom’s religion as well as the king’s dwelling. In both
cases, the beautifying alternate never appears except as the second half of these pairs,
though the main term may appear with other alternates.36 These two sets of parallel pairs,
among many others, appear together in dozens of songs.
The omoro are written in an abbreviated format that avoids transcribing repeated
phrases and lines more than once. In the vast majority of the songs, the refrain is given
only in the iti verse, and a phrase or line of a main verse that must be repeated is given
only the first time it appears. The reader must reconstruct, or expand, the songs (Hokama
uses the term kaidoku, “decode-and-decipher”) to their full forms by adding in these
phrases, lines, and refrains in subsequent verses. However obvious the form a
reconstruction should take may be, the reader should be cautioned as to the theoretical
nature of any such reconstruction. One difficulty in expanding the songs from their
abbreviated forms is that it is not always perfectly clear exactly which words from earlier
lines should be repeated in subsequent verses. Also, and particularly telling for this paper,
concerned as it is with kakari-musubi, lines presumably with identical verbs at the end
35 This idea thanks to Leon A. Serafim, in a personal communication, February, 2008.36 For example, in song 346, kikoe ōgimi pairs up with kikoe anji-osoi (renowned(HON) lord-HON, ‘renowned king’), reflecting how the songs often refer to the High Priestess and the King as a pair.
41
will not have the verb phrase explicity written out for all the verses. In these cases it is
vital to carefully reconstruct the verses with their verb phrases, though there may be a
risk that the reconstruction is, in fact, incorrect.
2. Reconstructing a songCrucial to any analysis or translation of omoro in the OS is the technique used in
reconstructing the songs. Hokama 2000 does not reconstruct them, and his translations
reflect that: he often gives the translations for the refrains, or repeat segments, at the end
of his translation, and once only, presumbably to save space. However, he describes the
accepted techniques for reconstructing the songs, among other explanations, in lengthy
appendices. Although a few songs in the OS do not give segments to be repeated, the vast
majority of them do.
Understanding the parallel structure and the abbreviated method of transcription
used in the OS is key to reconstructing, and then decoding, the omoro. Tamaki 1981
analyzed the patterns in songs of a variety of folk traditions in the Ryūkyūs, and found
that they all shared a characteristic tendency toward lines with parallel structures, and
then classified the omoro of the OS according to how the parallel lines patterned. This
provides a useful starting point for understanding the patterns of the songs and the way
they were written down. Consider omoro 1:1 again, as it looks in the Shō text (see also
Figure 1 above):
28Aoriyafe-ga fusi
ITI kikoweUFUkimikiyaoreteasufiyowarefatenikasitatairaketetiyoware
MATA toyomusetakakokaMATA siyorimorikusukuMATA matamamorikusuku
(1)
The first step in expanding the song is determining which words, if any, in the iti
verse should be repeated in the following mata verses. This process is often something of
a judgment call, though in the case of omoro 1 reconstruction is fairly straightforward.
42
As described in the previous section, the first words of each pair of lines make up
noun phrases that frequently appear together, and the first two have the same grammatical
marker ka/kiya (::ModOk ga, J ga, ‘SUBJECT MKR’; <kiya> is a palatalized allomorph of
<ka>) following these initial words. The second pair of lines do not show the subject
marker, possibly to keep the syllable counts even; and perhaps also because Old
Okinawan allowed “zero-marking” of “phrases that are topics, subjects, and direct and
indirect objects, where an indirect object is any phrase that may be marked with ni or
similar particles.”37 The uniformity of the two pairs of lines, each pair consisting of
parallel nouns that commonly appear together, allows us to assume that the remaining
words of the iti verse after kikowe UFUkimikiya should be repeated in each verse, making
the expanded song look like the following (with reconstructed segments in square
brackets):
29ITI kikowe UFU-kimi kiya
orete asufiyowarefateni ka sita tairaketetiyoware
MATA toyomu setaka-ko ka[orete asufiyowarefateni ka sita tairaketetiyoware]
MATA siyori mori kusuku[orete asufiyowarefateni ka sita tairaketetiyoware]
MATA ma-tama mori kusuku[orete asufiyowarefateni ka sita tairaketetiyoware]
(1)
Sometimes it is helpful to take syllable counts, or beats, into account, though
syllable count seems to be a secondary consideration in the OS, especially in comparison
to parallel phrasing. In the song above, the first line has eight beats:
ki-ko-we U-FU-ki-mi-kiya (8)37 Thanks to Leon A. Serafim for this insight in a personal communication, October 2009.
43
The subject marker spelled <kiya> is most probably pronounced as [g ja]. The first mata verse also has eight beats:
to-yo-mu se-ta-ka-ko-ka (8)
The second and third mata verses also match quite nicely. At first glance, the
second mata verse has eight syllables, just like the third one:
si-yo-ri mo-ri ku-su-ku (8)
ma-ta-ma-mo-ri ku-su-ku (8)
The spelling of <siyori> for the name of the capital as well as the king’s castle at
Shuri presents the only possible difficulty with the syllable count. If the kana <yo> adds
either a syllable or vowel length to the word, then the count agrees with that of the
following line. The problem lies in the evidence of modern Okinawan, in which Shuri has
two syllables <shu-ri> or <shu-i> (i.e., しゅり, not しゆり<si-yu-ri>). If the word was short
in Old Okinawan, this would throw off the syllable count for the second line:
siyo-ri mo-ri ku-su-ku (7)
Omoro scholars in the Omoro sōshi kenkyūkai accept the 3-beat length of
<siyori>, since some contemporary ryūka (Ryūkyūan poems) spelled it <siyuyui> and
pronounced it [shujuji].38 This of course saves the syllable count in this case.
Either technique, of counting syllables or observing parallelisms in the lines of
verse, may serve in deciding which portion of the iti verse should be repeated in
subsequent verses; however, counting syllables is the less straight-forward of the two,
because of irregularities in the numbers of syllables in lines throughout the document.
After determining what portion of the iti verse should be repeated, a decision must
be made on whether any part of this segment is a refrain; and if so, where the break
38 Leon A. Serafim, personal communication, August, 2007.
44
between the main verses and the refrain falls. For omoro 1, Takahashi 2005 (the
electronic text used for this paper) has the line orete asufiyowarefa as the second half of
the first line of each verse, and combines the remaining two lines into one line (see
Example 30, below). Hokama (1972 and 2000) and the Teihon (of which Hokama is an
editor, hence the similarity) leave the first two lines separate, but instead breaks the third
line to make the fourth line tairakete choware. Either way, each verse will have the same
words repeated, and the melody would have played over the same number of syllables.
The Takahashi version provides the most space-efficient rendition, and the expanded
song, based on the Takahashi layout, would look like the following (here given, finally,
with analysis and translation). Note that by reconstructing this omoro, the line-final
words change, and each verse has the same sentence-final word, choware.
30ITI kikowe UFU-kimi-kiya ore-te asuf-i-yowar-e-fa
renowned great-priestess-SUBJ descend-SUBGER play-INF-HON-EV-CON
Kikoe-ōgimi, when (the Spirit) descends to earth and dances
TENI-ka sita tairake-te t-i-yowar-e heaven-POSS below pacify-SUBGER come-INF-HON-IMP
O, (may the King) conquer all under heaven, and reign!
MATA toyom-u se-taka-ko-ka [ore-te asuf-i-yowar-e-faresound-ATTR spirit-high-child-SUBJ descend(INF)-SUBGER play-INF-HON-
EV-CON
Great Priestess Sedaka-ko, when (the Spirit) descends to earth and dances
TENI-ka sita tairake-te t-i-yowar-e] heaven-POSS below pacify(INF)-SUBGER come-INF-HON-IMP
O, (may the King] conquer all below heaven, and reign!
MATA siyori mori kusuku [orete asuf-i-yowar-e-faShuri grove precinct [descend(INF)-SUBGER play-INF-HON-EV-CON
To Shuri grove precinct, when (the Spirit) descends and dances
TENI-ka sita tairake-te t-i-yowar-e]heaven-POSS below pacify(INF)-SUBGER come-INF-HON-IMP
O, (may the King) conquer all below heaven, and reign!
45
MATA ma-tama mori kusuku [orete asuf-i-yowar-e-fatrue-jewel grove precinct descend(INF)-SUBGER play-INF-HON-EV-CON
To the true-jewel grove precinct, when (the Spirit) descends and dances
TENI-ka sita tairake-te t-i-yowar-e]heaven-POSS below pacify(INF)-SUBGER come-INF-HON-IMP
O, (may the King) conquer all below heaven, and reign!(1)
Here we must also note that the term “refrain” does not reflect the fact, in this
case, that the repeated words are not independent sentences or verses, but completions of
the sentences begun with the words at the start of the iti and mata verses. The verb
complex asufiyowarefa (i.e., asub-i-yowar-e-ba, play-INF-HON.AUX-EV-CON, ‘when (you)
dance’), ending as it does with a conjunctive suffix, cannot be a sentence-final verb in the
main verses. There are songs in the OS where there is no alternative, and sometimes we
must accept what appears to be an ungrammatical ending, or perhaps an incomplete
ending, to a sentence, but not here.
The omoro used in the explanation above exhibits very clearly the parallelism so
frequent in the OS. It is safe to say that every omoro displays parallelism, a characteristic
that according to Tamaki 1981, is common to all forms of Ryūkyūan religious and folk
song. Not all the songs show such perfect parallelism as omoro 1 above, possibly due to
imperfect transcription or other issues.
Powerful support for this technique of reconstructing omoro can be found among
the songs themselves. Two omoro, 592 and 1403, have duplicates (i.e., songs that are
nearly identical): 1479 and 1440, respectively. 592 and 1403 appear in abbreviated
layouts similar to that of omoro 1 above: the iti verse gives a noun phrase that parallels
with another in the first mata verse, plus a refrain segment that appears only in the iti
verse. The second and third mata verses also make up parallel noun phrases, and if our
understanding of how to reconstruct the omoro is correct, then we must add in the refrain
segment in all three mata verses. This is borne out by songs 1479 and 1440, whose
layouts are not abbreviated: both of these songs have the refrain segments explicitly
written out for every verse, and they show that the repeat segments given only the iti
verses in 592 and 1403 should indeed be repeated in all three mata verses.
46
In some songs, in addition to sets of clearly parallel lines, one line may not be a
part of a parallel pair, or two lines may be parallel to each other, but differ in some way
from other pairs in the same song. Omoro 3, for example, shows this pattern, consisting
of three pairs of parallel lines (boxed together in Example 31), plus one unpaired line.
The underlined portions in the iti verse indicate the phrases to be repeated in each verse:
31
ITI kikowe UFU-kimi kiya YO - so - u seti mi - oyas - e - fa renowned(INF) great-priestess-SUBJ world-control-ATTR power HON-offer(HUM)-EV-CONJ
When Kikoe-ōgimi offers up the power that controls the world
SEN - MAN YO sowa - te t - i - yowar - e thousand-ten.thousand world control-SUB come-INF-HONAUX-EV/IMP
May (the King) reign for myriad generations
MATA toyom-u setaka-ko-ka resound-ATTR Sedaka (lit. ‘spirit-high’, ‘of great spirit’)-child TOP
Resounding Sedaka Priestess
MATA kikowe aNsi-osoirenowned(INF) lord-HON
Renowned honored King
MATA toyom-u aNsi-osoiresound-ATTR lord-HON
Resounding honored King
MATA SHURI mori kusukuShuri grove precinctShuri grove precinct
MATA ma-tama mori kusukutrue-jewel grove precinctJewelled grove precinct
MATA UFU-kimi su mafur-am-egreat-priestess KP protect-TENT-EV/IMP
It is the Great Priestess who will surely protect (him)! (3)
The sixth mata line not only lacks a parallel line, it also differs in form from the
other lines in the song. The first six verses all have the form of NOUN.PHRASE +
47
TOPIC.MARKER + VERB.PHRASE + REFRAIN. The iti verse gives the completion of the line
after the initial noun phrase, as well as the refrain; the next five mata verses show only
the noun phrase. The assumption is that each of these should also be followed by the
topic marker,39 verb phrase, and refrain. The sixth mata verse, however, consists of a
noun phrase plus a verb complex. This is a complete sentence, and therefore, possibly,
the refrain segment given in the iti verse should not be included in this verse. Whether the
refrain line sen-man yo sowate tiyoware ‘May (you) reign for myriad generations’ should
follow the sixth mata verse is unclear. As a complete sentence, this line should be
analyzed as ending in mafurame (i.e., mabur-am-e, ‘protect-TENT-EV’), not in choware as
all the other verses do. If the refrain segment were repeated in the final, odd, verse, the
verse would consist of two complete, separate sentences.
Given the importance of parallel lines in the OS, the odd line or lines out may
sometimes be deliberate structures serving a specific purpose (when they are not simply
errors of transcription). In omoro 3, the kakari particle su may lend a hint as to the
purpose of the extra line: it is a form of emphasis, highlighting its contents with the fact
of its singularity and non-conformity with the structure of the rest of the song. Lines such
as this serve as a coda to a song, and may contain the most important words of prayer; in
this case, a request to the sun for protection of its earthly representatives, the Priestess
and the King. The verb in the last line of omoro 3 also has the expected form, though
discussion of verb forms will be kept until later.
Finally, some omoro have no clear refrain segment, or it is unclear in some other
way how the song should be reconstructed. Example 32 shows omoro 590, in which all
three verses consist of parallel sets of words. The first and third verses have parallel noun
phrases; the middle verse has parallel verb clauses.
39 Though given the “zero-marker” rule, the particle in this slot may be optional and its presence or absence may depend on the syllable count.
48
32ITI arakaki-ni owar-u ma-MONO yo-no nusi n-o ma-moN
Araki(PN)-LOC exist(HON)-ATTR true-thing world-GEN lord DV-ATTR true-thingOverlord who resides in Arakaki; Overlord who rules his domain
MATA yosa n-o kuwa fa nas-i-yowa-tife fatu n-o kuwa-fa nas-i-yowa-tife prime DV-ATTR child TOP create-INF-HON-SUBGER
first DV-ATTR child TOP create-INF-HON-SUBGER
Siring the primary child; creating the first child
MATA to-momo so n-o ikusa ya-momo so n-o ikusa ten-hundred person DV-ATTR warrior eight-hundred person DV-ATTR warriorArmy of ten thousand warriors; army of eight hundred warriors
(590)
In this omoro, perhaps a song of praise for a new-born royal heir, there is no
obvious extra segment in the first verse that should be repeated in subsequent verses. The
second verse appears incomplete, inasmuch as it appears to end in a hanging
subordinating gerund; the other two verses end in nouns, leaving us with a song made up
of incomplete sentences. Possibly such sentences are more acceptable in song than they
would be in prose or the spoken language. Certainly many lines in the OS end in that
subordinating gerund, even after all the lines have been reconstructed as far as possible.
For the purposes of this paper, reconstructing a song can change the environment
of a particle enough to affect its analysis. For example, what may initially look like a
sentence-final particle, since it falls at the end of a line of text, in fact turns out not to be,
upon expanding the song. Consider omoro 187, in its entirety:
49
33ITI sasukasa sasukasa toyom-i-war-e
Sasukasa Sasukasa resound-INF-HON-EV/IMP
Sasukasa, Sasukasa, (your glory) resounds!
we mi-MONO-yose sutu-nar-iEXCL HON-thing approach(DEV) bell-cry(NOM)Oh! The marvelous wealth-bringing handdrumbeats
MATA kimi-no aNsi ati supriestess-DV lord lord KP
Lord of priestesses, lord, indeed(187)
At first glance, this appears to be an example of su falling at the end of the song.
However, remembering the space-saving way in which the songs were recorded, the song
must be expanded before making a judgment on the status of <su>. Adding in the refrain
(shown in square brackets), the expanded version of song 187 looks like Example 34
(analysis and translation omitted):
34ITI sasukasa sasukasa toyomiware
we miMONO yose sutunariMATA kimino aNsi ati su [toyomiware
we miMONO yose sutunari](187)
Expanded in this way, the first line of the second verse is perfectly parallel to that
of the first, and the <su> falls in a non-sentence-final position. Most occurrences of <su>
that seem at first glance to be line-final can be handled in this way.
This particular omoro contains a further puzzle, the answer to which may also be
a judgment call. Note that the iti verse consists of 13 syllables/beats:
Sa-su-ka-sa sa-su-ka-sa to-yo-mi-wa-re (13)
The mata verse, on the other hand, can be counted with either 13 or 14 syllables/beats,
depending on whether aNsi is read with two or three beats:
50
1. Ki-mi no aN-si a-ti su to-yo-mi-wa-re (13)
2. Ki-mi no a-N-si a-ti su to-yo-mi-wa-re (14)
It all likelihood, the second syllable of <ati> was voiced; the Modern Okinawan
word is anji. In the Takahashi text, which reflects the voicing as indicated in the Aniya-
descendant manuscripts, the word occurs at least 700 times. Just over half are written
<ati>; just under half are written <aNsi>; and a dozen are written <adi>. It is even
possible the word has two alternatives for pronunciation, as /aji/ or /anji/ (as noted, the
latter is the accepted pronunciation today), else why the two spellings?
As they stand, to make the syllable counts in these lines match, we must posit the
two-syllable reading for <aNsi>, even if this goes against the usual policy in both
Ryūkyūan and Japanese poetic traditions. However, upon examination, the iti verse
syllable count might be expanded in one of two ways. Less likely, but still possible, is
that in order to make the syllable count match, a particle su was suppressed, but was
understood, in the iti verse immediately following the second Sasukasa, giving the line an
extra beat:
Sa-su-ka-sa Sa-su-ka-sa su to-yo-mi-wa-re (14)
This would be a convenient explanation for the form of the verb in both lines,
allowing us to read evidential for them, thanks to the presence of KP su (more on this
issue below). However, it is more likely that there is a spelling error in the iti verse: the
verb, spelled <toyomiware>, is missing a syllable. It should be spelled <toyomiyoware>,
as it is every other time it appears in the OS. That provides us with the missing syllable,
and allows us to read <aNsi> as three beats, without compromising the syllable counts of
either verse. Of course, the second explanation does not preclude the possibility of a
suppressed particle su, though such an explanation does not seem to hold for the many
other songs where an odd line has KP su but all the verses still have the same sentence-
final verbs.
51
And here, a difficulty arises upon closer analysis. The first sentence of each verse
in the expanded version ends in the same verb form. Hokama (2000) translates
toyomiware (i.e., toyomiyoware) as an imperative: todorok-i-tamae ‘resound-INF-
HON.AUX(IMP)’. If we don’t accept the idea of a “suppressed particle,” and
toyomi(yo)ware is an evidential form agreeing with the KP in the second verse, why
would it be triggered in the first verse, where there is (probably) no KP? If it is not the
evidential in the first verse, is it, in fact, imperative? Would the same form have appeared
in the second verse, or would the singer automatically insert an evidential form, due to
the presence of su? And if so, what shape would that form have? The particle cannot be
analyzed as a sentence-final particle, since it does not end the line once the song is
reconstructed, nor does it follow a verb complex. These questions and this type of pattern
will be discussed more below, in the section on the distribution of verb forms in relation
to the KPs, but for now we will accept this (overt) token of <su> as KP su, given its
appearance in the middle of the sentence, and the presence of a verb in -e at the end of the
sentence.
52
CHAPTER 3: KAKARI-MUSUBI IN THE OMORO SŌSHI
A. MethodologyIn the discussion that follows, the term “line” refers to a line of text as presented in any
one of the documents used as a source for this thesis, that may or may not also be a
sentence. A “sentence” refers to a grammatical entity that includes, at minimum, a verb
complex. Thus, a “line-final particle” is physically located at the end of a written line in a
manuscript, but grammatically may actually be in the middle of a sentence, depending on
how analysis and song reconstruction affect the shape of the line. The reverse applies as
well: a particle located after a verb, and thus possibly sentence-final, may be located in
the middle of a written line of text in the physical document, since the Shō manuscript
does not invariably return to the top of a line when beginning a new sentence.
1. Description of the databaseThe data for this thesis are based on the text of the Shō manuscript as transcribed in
Takahashi 2005. The document, converted by me into Roman letter form, was further
converted into an Excel database, initially resulting in an Excel document with three
columns of nearly 10,000 rows, including one blank row between each song. Additional
columns showed such information as volume number, volume title, fusina (if any),
numbers of any songs considered duplicates, and some notes regarding Hokama’s
translations. Finally, as analysis progressed, several columns were added showing
information such as whether a line contains a kakari particle, likely musubi candidates,
and so on, for quick reference.
A second Excel table, based on the first, was created exclusively for identifying
and categorizing the final word of each line in the OS (based mainly on Takahashi’s
analysis) after reconstruction. The second table (labeled “Line Final Words”) provided
the counts for the number of tokens in each category.
Excel allows searches for specific items using algorithms to define the search
parameters. This feature was used extensively to search for items of interest in the
column containing the song text. Excel also allows separate tables and charts that can
group information from different columns into different configurations, and count tokens.
53
Very likely a different database program specifically designed for language data
(as opposed to the numerical data for which Excel is designed) would have served better.
However, once the method of writing search algorithms was clear, the program seemed
adequate for the purposes of this thesis.
2. Eliminating non-KP lines to find the pool of KP su and do candidatesThe first step in discovering whether there is any confusion or overlap in the use of the
KP su and do in the OS was to find all the songs containing su and do emphatic
constructions. Given the spelling conventions shown in the previous chapter, this
included searching for the tokens in all their variety of possible spellings. Table 5 reprises
all eighteen logical possible spelling variations of su and do as discussed earlier:
Table 5: Summary of all possible spellings of KP su and do in the OS
su doNon-palatalized Palatalized Non-palatalized Palatalized Alternation
<o>spelling
<so>/<zo> <siyo>/<ziyo> <to>/<do> <tiyo>/<diyo> <ro>
<u>spelling
<su>/<zu> <siyu>/<ziyu> <tu>/<du> <tiyu>/<diyu> <ru>
Of course, not all of these possible spelling variations actually occur in the OS, but all
were included in the interest of being thorough.
As implied in both the OKDJ table of particles in OOk and Takahashi 1991b,
identically-spelled particles may appear in environments other than the emphatic
constructions under consideration here, and serving different, if similar (i.e., emphatic),
functions. Such particles must be culled based on criteria such as function, meaning, or
placement within a line. By far the most time-consuming task in this project was
determining which particles were NOT to be included in the final pool of those clearly
playing a role in an emphatic construction.
The process of creating a list of all the song lines (and thus songs) in the OS that
contain one (or more) of the KP spellings (Table 5) is a fairly mechanical one of
determining how many times each spelling variant occurs word-finally, eliminating those
that are only syllables, words, or at least particles not in an emphatic construction. This
step was hugely simplified by the fact that Takahashi 2005 includes word breaks, though
54
we are making an assumption in accepting these breaks as correct. Because Takahashi
based his work on that of all his predecessors in the interpretation of the OS, we will
accept that his decisions have a sound basis.
In some cases, a decision had to be taken against the analyses in Hokama 2000.
For example, some lines with KP do in which the candidate musubi initially appeared to
be nouns had to be reanalyzed as not showing any kakari particles at all. An example of
this is in omoro 1165 (Example 35):
35
Hokama includes koso in his translation of this line, but because the allomorph of
KP do spelled <tiyo> otherwise appears only in the palatalization environment (i.e.,
following a word ending in /i/), this is either not a token of KP do, or it is the palatalized
particle but carrying a completely different function, thus divorced from its original form.
The segment aku tiyo is thus obscure, making the line difficult to assess. We leave it out.
Another line that Hokama reads with KP do and which appears to end in a noun is
in omoro 1450. Here, even after reconstruction, the line- and sentence-final word is
clearly a noun, but the nature of the particle is less clear:
36 ati-osoi-teta to wa-ka teta
lord-HON-sun KP? 1PS-POSS teda ?It is the King who is our Sun
(1450)
In fact, though Hokama translates it as koso, <to> could be the quotative particle, and the
line could be translated ‘The King, who is called our Sun.’ Or, it could be the comitative
<to>, in which case the translation would be ‘The King and our Sun.’ Both of these seem
feasible translations, thus this line is also eliminated from the pool of lines with KP
particles.
55
mitume-teta aku tiyo amai stare-sun UNK KP? UNC
?It is the staring sun that … (1165)
One line with a segment that Hokama 2000 translates as koso occurs twice, once
each in a pair of duplicate songs (omoro 574 and 1459), with an apparent candidate
musubi verb in -te.
37ITI kikowe se-N-kimi-kiya
Renowned spirit-DV(ATTR)-priestess-SUBJ
Renowned Sengimi—
masikesu ma-kerafe s-i-yowa-tife masike? KP? true-build(INF) do-INF-HON-SUBGER
?(574)
To be fair, though he includes koso in the translation of the song, in the accompanying
gloss of individual vocabulary words, he does not specifiy that <su> in masikesu is a
kakari particle; he glosses it as part of the word, which he defines as kami-no zaisho
‘abode of the gods.’ The Okinawa kogo daijiten, of which Hokama was editor, also
includes the segment as part of the word. There the word is glossed as masikedu (ましけ
づ), with masike defined as kami-no zaisho, and -du noted as an unknown term. Since the
dictionary included sources other than the OS, the voicing and this spelling must be
apparent in the word as it appears elsewhere. The fact that the segment is obscure; and
that it is spelled elsewhere with a different kana, and one that is never used (at least in the
OS) to spell a particle, means that these lines cannot be included in the pool of lines with
KP.
Finally, Hokama glosses with koso six tokens of <su> that occur in parallel lines
in each of three duplicate songs. Initially these particles were included in our pool of KP
su, but a problem arises with the word preceding the particles in the first line of each pair.
By extension, the doubtful status of the first particle also casts doubt on the second.
Example 38 shows the lines from omoro 92 (the duplicates are 365 and 497), with what
Hokama labels as KP su, and a line-final verb in -yowa, a mild imperative:
56
38MATA o-gimo uti-ni yo-sira su UFU kimi-ni sina-yowa
HON-liver inside-LOC world-know? KP? great priestess-LOC bow-HON(M.IMP)(?) Bow to the Great Priestess!
nasai-kiyo-moi ati-osoi [amako yor-i-kawa-tiwe manasiya do tat-i-yor-u]maker-one-HON lord-HON [eyes approach-INF-exchange-SUBGER love KP
stand-INF-PROG-ATTR/PRED]The King, our Father, meets (her) eyes and exchanging a look, the spirit rises
(between them)
MATA ayo-ga uti-ni obowe su se-taka-ko-ni sina-yowa liver-POSS inside-LOC remember KP? spirit-high-one bow-HON(M.IMP)(?) Bow to Priestess Sedaka!
nasai-kiyo-moi ati-osoi [amako yor-i-kawa-tiwe manasiya do tat-i-yor-u]maker-man-HON lord-HON [eyes approach-INF-exchange-SUBGER love KP
stand-INF-PROG-ATTR/PRED]The King, our Father, meets (her) eyes and exchanging a look, the spirit rises
(between them)(92)
The OKDJ notes that the phrase <yosirasu> (in the first mata verse in Example 38
has been explained as meaning omou koto (‘thinking’); as shugo-sarete koso (‘(It is)
indeed being protected’); and that it can be read as sirasu, that is, shihai-sareru (‘be
controlled, ruled’); or as sirazu (‘know-not’). OKDJ’s conclusion on the word is “mishō”:
unknown (OKDJ 1995:710c). Given the difficulty in determing whether this is a particle
or part of the preceding word, these lines should probably be left out of our pool of
possible KP particles.
After making all such eliminations, the number of likely kakari particles is
considerably reduced:
57
Table 6: Pool of possible kakari particles (unreconstructed)
XLines with words ending inX
Possible KPX
so 31 --zo -- --su 486 280zu 15 --siyo 162 141ziyo 4 4siyu 37 26ziyu -- --
to 354 103do 25 21tu 92 --du 2 --tiyo 43 9diyo 9 4tiyu -- --diyu -- --ru 1014 33ro 298 10Total 2572 623
In sum, as expected, the spellings <so>, <zo>, <zu>, <ziyu>, <tu>, <du>, <tiyu>
and <diyu> do not occur in the OS as particles, even though some of them do occur as
words, in compounds if not independently. The remaining spellings may all appear as
words or segments of verb complexes, and must be examined closely to determine
whether they are particles, and if so, whether they are the kakari particles under
consideration here.
3. Reconstructing songs to further eliminate non-KP particlesAs noted earlier, analyzing the OS to identify kakari particles requires not only
understanding of the vocabulary and grammar of the omoro, but also an understanding of
the abbreviated form in which they were written to reconstruct (or “expand”) them in
making decisions about a given particle’s function and environment. Reconstructing a
song can have the effect of moving a particle that initially appeared to be line-final into
the middle of the line, or vice versa. In effect, reconstruction may substantially alter a
particle’s environment. Reconstructing a song may also result in a greater number of
58
particles than it had in its original form, since reconstructing the songs increases the
count of lines in the OS, and so also the count of lines containing kakari particles. Adding
the lines not explicitly included in the Shō manuscript back into the database used to
tabulate the sentence- or line-final words hugely increases the size of the document.
Finally, reconstructing the songs affects the shape of the lines in terms of just
what word ends a given sentence. This allows us to determine as far as possible what the
probable sentence-final word may be for each line, if different from that explicitly written
in the original. In many cases, reconstruction has the final effect of giving every line in an
omoro the same sentence-final word, especially when a refrain segment grammatically is
the completion of sentences begun in the iti and mata lines. Thus in the cases of omoro
with a KP falling in the refrain, the particle and its line-final word are assumed to recur in
every verse (if not explicitly written out, as it is in a handful of songs). In these cases,
only the explicitly written out KP and their candidate musubi (whether written out or
reconstructed) were counted, though subsequent lines are assigned the same musubi,
affecting the overall count of the number of lines ending in that word. Two-thirds of all
the KP in the OS occur in refrain lines; if the refrain lines were reconstructed in full and
spelled out, the count of particles would actually be much higher than Table 7 (below,
page 67) shows.
The final step in winnowing out non-KP is to identify the tokens that may be
kakari particles, but are not in anything resembling kakari-musubi constructions, even
after reconstructing the songs they occur in. Usually these particles appear to be filling a
sentence-final function.
By definition, a kakari-musubi construction does not have the particle in sentence-
final position.40 As we have seen, reconstructing songs becomes important in determining
whether a given occurrence of su has fallen at the end of a line of text because it belongs
at the end of the sentence, or because the method of abbreviation ended the line in that
place. Upon reconstruction, there are several cases where <su> still falls at the end of a
line of text, and there is no clear verb serving as a candidate musubi at the end of the
verse or in the refrain. For example, in song 538, the last two verses both end in <su>, but
40 With exceptions including OJ ka and type-2 kamo, which could occur sentence-finally, and also affect the form of the preceding main verb.
59
neither the refrain nor some kind of expansion provides any sort of musubi. As it appears
in the Shō text:
39ITI isikesitayoukafou yosetukerutomari MATA kanesikanetonoyo MATA isifetufakonote MATA kanafetufakonote MATA isikeyorinaotife MATA natarayorinaotife MATA kusunukifakonote MATA yamatofunekonote MATA yamatotabinobote MATA yasirotabinobote MATA kafarakaininobote MATA temotikaininobote MATA omoiguwanotamesu MATA wariganegatamesu
(538)
Reconstructing only the first iti verse (in order to look at the refrain) and the last
two mata verses (the first two and the last two lines as presented in 39 above) in their
most probable expanded forms, and inserting the word breaks, results in the following
(the particles in question are marked as unclear with “(?)”):
40ITI isike sita you kafou
Isike below world prosperity Below Isike (lie) the world’s riches
yose-tuker-u tomari bring.near(INF)-reach-ATTR harbor (This is) the harbor that brings (them)
…MATA omoi-guwa-no tame su
HON-child-GEN on.behalf.of (?)It is for our great one!
[?yose-tuker-u tomari]bring.near(INF)-reach-ATTR harbor [(This is) the harbor that brings (them)]
60
MATA warigane-ga tame suWarigane-POSS on.behalf.of (?)It is for Warigane!
[?yose-tuker-u tomari]bring.near(INF)-reach-ATTR harbor [(This is) the harbor that brings (them)]
(538)
Overall, the verses show perfect parallelism in forms and meanings, as well as a
symmetrical syllable count of eight beats per line.41 Adding in the refrain does not supply
a candidate musubi verb (though possibly a copula is understood). Without any overt
musubi candidate, we can conclude that here are two examples of su as sentence-final
particle (FP). There are nine omoro which have final pairs of lines ending in su like this
(including song 203 in Example 42 below), as well as two in which there is only one final
line ending in su. It is possible that the presence of su would have changed the refrains in
the final verses of these songs in some way that was not recorded, such that when
performed, a different verb form would have been included. However, there is no
evidence of such a change, and most likely the presence of this particle lends the line an
emphatic feeling. Falling at the end of the omoro, lines like this may serve as a climactic
coda to the song.
Similarly to omoro 538 above, 1051 also ends in a set of parallel verses with KP
su at the end of the lines, but with the addition of what appears to be an alternative refrain
for the final, if not also the penultimate, verse:
41 Vowel length does not count as an additional beat; diphthongs do. The verb ending -tife is one beat, conventionally read [tche]. -guwa (::J ko, ‘child’) is one syllable, [gwa]. Thus, for example, the verb naotife (::J naosite, ‘set right’) is three beats, /na-o-tche/ ([no-o-tche]), and omoiguwa (HON-child, ‘honored priestess’) is four beats, /o-mo-i-gwa/ ([u-mu-i-(g)gwa]). With these facts in mind, every line in this omoro has eight beats.
61
41ITI Kimi-yosi Kimi-yosi kimi-no
Priestess-HON priestess-HON priestess-SUBJ
Priestess Kimiyoshi, Kimiyoshi, you are
kerafe-yor-u kiyora yabuild(INF)-PROG-ATTR beauty EXCL
Building something so beautiful!
MATA kimi-osoi kimi-osoi kimi-noPriestess-HON priestess-HON priestess-SUBJ
Honored Priestess, honored Priestess, you are...MATA kimi-yosi-kiya ma-wasi su
Priestess-HON-POSS true-eagle KP
O! It is Kimiyoshi’s true eagle (i.e., a great ship)
MATA kami-nisiya-ka ma-wasi sudeity-hon-POSS true-eagle KP
O! It is Kaminisha’s true eagle
yo-tama kokoro yose nuk-i-yowar-eworld-jewel heart bring.near(INF) break.through-INF-HONAUX-IMP/EV
That will protect realm and treasure with all its might (1051)
A close look shows that the last mata verse has a second line, possibly a
replacement for the refrain given in the iti verse. This refrain at the end of the song ends
in a verb, unlike the first refrain, which ends in a predicate adjective. Possibly this line
needed its own refrain due to the KP su at the end of the preceding line, making a musubi
in -e necessary. At any rate, sometimes an alternative refrain is provided, and in this case
it contains a candidate musubi for the particle in the last two verses. But questions
remain: Why does this song require an alternate refrain, when there was none in omoro
538? Why does the penultimate verse not also have this refrain—should the refrain given
at the end be used for both verses? They are parallel otherwise, in structure and meaning.
The form of that verb phrase in the final line does agree with expectations concerning a
musubi for KP su, and the presence of the verb means these particular lines may show a
kakari-musubi structure. Thus on purely technical grounds, these lines remain in our
62
candidate pool; and tokens of <su> occurring in parallel structures, or only in the last
verse of an omoro, and which have no clear musubi verb, are labeled as ‘FINAL PARTICLE
(FP)’ and excluded from the pool of kakari particles under consideration here.
Many lines with either KP su or do exhibit parallelism, with two phrases both
ending in a KP, such as in song 203:
42MATA teru.kafa ka o-koto su teru.sino ka ON-koto su
shine.river TOP HON-word KP shine.?one TOP HON-word KP
It is the command of the Sun God; it is the command of the deity Terusino.
(203)
In this case, the parallel phrases make up the final verse of omoro 203 (and was probably
followed by a refrain, not shown here). In other instances, the parallel lines are relegated
one each to the last two mata verses of a single song (such as omoro 538 in Examples 39
and 40 above). In the latter cases, the format allows us to construe that the second line
repeats the melody of the first line; it also allows the possibility of the two verses being
separated by a refrain. Particles in clearly parallel structures such as in Example 42 may
have a candidate musubi verb in the refrain segment; a possible candidate musubi verb in
the expected form (t-i-yowar-e, ‘come-INF-HON.AUX-EV/IMP’) appears in the song’s
refrain. In cases where no candidate musubi was provided in a refrain segment, the
particles were analyzed as sentence-final and excluded from our pool of consideration.
Some lines with a non-final KP end in the exclamatory word, mimoN. Strictly
speaking, mi-moN (‘see(INF)-thing’ :: J mi-mono) can be considered as a noun. However,
it is used exclusively in the OS as an exclamation meaning “awe-inspiring” or
“wonderful sight.” For example, in song 703 (Example 43 below), in which every line
ends with mimoN, the singer seems to be shouting praise for the events and things
described in the song (the high priestess of Kume Island, dancing, and the King). The
second, fourth, and sixth verses, though marked with a mata, are identical and appear to
be refrains. These three lines do not have the same syllable count as the first verse,
though they apparently have the same melody. The lines that do not repeat are parallel in
63
structure and meaning, though the syllable counts have to be finessed to match. All six
lines, main and refrain, end in mimoN:
43ITI kume-no UFU-osoi-ka mi-moN
Kume Island-GEN great-HON-SUBJ see(INF)-thing (EXCL)The Great One of Kume is marvelous!
MATA we kewa-i to mi-moN O! race-NOM KP see(INF)-thing (EXCL)O! It is a race, indeed! Marvelous!
MATA kume-no se-taka-ko-ka mi-moNKume Island-GEN spirit-high-child-SUBJ see(INF)-thing (EXCL)The Priestess of Kume is marvelous!
MATA we kewa-i to mi-moNO! race-NOM KP see(INF)-thing (EXCL)O! It is a race, indeed! Marvelous!
MATA kume-no aNsi-osoi-ka mi-moNKume Island-GEN lord-HON-SUBJ see(INF)-thing (EXCL)The Lord of Kume is marvelous!
MATA we kewa-i-to mi-moNO! race-NOM KP see(INF)-thing (EXCL)O! It is a race, indeed! Marvelous!
(703)
It is not difficult to interpret the mimoN in 703 as exclamations. If we accept
mimoN as an exclamation outside the grammatical structure of these lines, then removing
the line-final mimoN from the refrains puts the particles at the end of the lines, as hinted
at in the English glosses for the lines. This makes these particles look far more like they
are filling the function of sentence-final emphatic particles (FP), and they should be
removed from the pool of KP candidates.
Two more lines, both in the same song, may also be handled this way (that is, the
word before a line-final exclamation is the musubi candidate). The main verses of omoro
382 (Example 44) have a KP su, followed by the clause (as Takahashi presents it) tama
no wau yarena. The last word, yarena, at first glance resembles an irregular negative
64
imperative. However, as with mimoN above, if we separate out the final syllable and treat
it as a sentence-final evocative particle, then the final verb is yare, i.e., ar-e,
‘exist-EV/IMP,’ and may be analyzed with the expected verb form.
44ITI YO yose mitu-mawari siyo tama n-o wau yar-e na
world bring(INF) three-strand KP jewel DV-ATTR king BE-EV/IMP PT
It is because of the three strands of wealth-bringing (jewels) that he is our jewel-like King!
kafou fa siyori oya-kuni prosperity TOP Shuri parent-countryThe blessings (accrue to) Shuri, our parent country
MATA tama n-o mitu-mawari siyo [tama n-o wau yar-e na]jewel DV-ATTR three.time-circumnavigation KP...It is because of the three strands of wealth-bringing (jewels)…
(382)
Analyzing these lines in this way moves these lines into the set of lines with KP su and
candidate musubi in -e.
Thirteen lines with KP do appear to end with nouns such as kusuku ‘precinct,’
mono ‘thing,’ kuni ‘country,’ mi-aori ‘HON-fan,’ and ma-kane ‘true metal,’ fiyasi ‘beat,
rhythm,’ and teta ‘sun.’ The fact that these lines end in nouns is alone enough to
disqualify the particles from status as kakari particles. These lines are interesting as they
provide examples of an oft-used structure in the OS, what we might call the kore-KP
structure. These lines contain the structure kore-KP-NOUNPHRASE; plus one with ore-KP-
NOUNPHRASE (which Hokama translates as sore, c.f. ModOK mesial demonstrative
pronoun uri :: ModJ sore). Omoro 899 provides an example:
65
45ITI kore ru kore fatu-nisiya
this KP this first-north.windIt is this; this is the first north wind
urako s-i-tife se-no-kimi tukai impatient do-INF-SUBGER Senokimi servantwaiting impatiently, sent for Priestess Senokimi
MATA kore ru kore okitofa this KP this horizon.windIt is this; this is the north wind from the horizon
MATA fatu-nisiya su mat-i-yo-tar-e first-north.wind KP wait-INF-HON-PERF/PROG-EV/IMP
It is the first north wind that (we) have been waiting for
MATA okitofa su mat-i-yo-tar-ehorizon.wind KP wait-INF-HON-PERF/PROG-EV/IMP
It is the north wind from the horizon that (we) have been waiting for(899)
Note that this song repeats the line-final nouns in the last two verses, both of
which have KP su (and a verb in -e), but no kore-KP construction. Presumably the refrain
segment in the iti verse should be repeated in all three mata verses.
Searching the whole document for this construction reveals it appears in a total of
26 lines. Grouping the lines by (reconstructed where possible) supposed musubi
candidate (i.e., sentence-final word) reveals they can be sorted into three groups: 1) those
thirteen ending in a noun; 2) those ending in a verb in -u (8 tokens), and thus falling into
the pattern of the majority of lines with KP do; and 3) those ending in verbs in -e or -a (3
and 2 tokens respectively). The word kore ‘this’ occurs unaccompanied by a particle in
only three lines, though one of these lines is parallel to a line with the kore-KP structure,
and may thus be lacking the particle because of scribal error, or to meet a syllable count.
The third group consisting of only five tokens may also be due to scribal errors,
but a more satisfying explanation for all three groups is to label these occurrences of do
as sentence-final particles. The phrase kore do is a complete utterance, translateable as
the cleft structure ‘It is this’ (Serafim & Shinzato, forthcoming). This reading removes
any grammatical link between the particle and the following supposed candidate musubi:
66
a verb is the predicate for a separate sentence, and thus not governed by the particle;
nouns take on either an exclamatory function, or can be read as having an “invisible”
copula. Thus the verb complexes in this group ending in -a (kore do itife toyoma ‘It is this
we are saying; let it resound!’ in both cases) and the three lines with verbs in -e are
unexceptional. Curiously, it is the lines with verbs in -u which must be regarded as
exceptions or errors, though they may not necessarily be grammatically incorrect. Of
course, labeling the particles in kore-KP constructions as sentence-final particles removes
those lines from our pool of lines with kakari-musubi constructions.
Table 7 Pool of kakari particles (reconstructed), excluding sentence-final particles
XWords ending inX
KP X, excluding FP
so 31zo --su 486 303zu 15siyo 162 136ziyo 4 4siyu 37 26ziyu --KP su 468
to 354 67do 25 17tu 92du 2tiyo 43 5diyo 9 4tiyu --diyu --ru 1014 12ro 298 4KP do 116Total 2572 584
Table 7 shows the final pool of 584 possible tokens of KP that appear in the OS as
recorded by Takahashi 2005, after excluding those spellings that, given their form,
meaning or environment after reconstruction, cannot be kakari particles. These particles
67
are in the set of songs containing the lines that we will analyze and sort, not by particle,
but by the words that fall into the musubi slot for the particles.
4. Sorting lines by sentence-final wordHaving narrowed down the selection of lines containing a kakari-musubi construction to
those containing the particles listed in Table 7, the next step is to categorize the lines
according to their sentence-final words after all possible reconstruction is complete.
For the document as a whole, reconstructed as far as possible, all sentence-final
words were identified and grouped into general categories. For example, all nouns,
including names and place names, were grouped into a single category, “nouns”. Thus all
verses ending in a noun are marked as having a noun (“N”) in the sentence-final slot.
Verb complexes were grouped according to the ending form, so for example, all
occurrences of toyome are in the larger set of verbs in -e (“vb e”). The counts of these
general categories are the figures given in Table 8 below.
A curious phenomenon appears in omoro in which a KP falls in a main verse (i.e.,
an iti or mata line). In such cases, assigning a sentence-final word may be more difficult,
since the question arises whether to assign the explicitly-written word at the end of a
main verse as the musubi candidate, or to assign the final word of the refrain as the final
word. This becomes a judgment call. Consider, for instance, omoro 208 (Example 46
below).
In 208, each main verse forms the first clause of a sentence; the refrain, written
out only once, forms a second clause; combined, each verse and its refrain may form a
compound or complex sentence. The seven main verses of 208 can be grouped as three
pairs of parallel lines, plus one additional, final, verse. The three pairs of verses plus the
final single verse end in different verb complexes (ore-wa-tife, ‘descend-HON-SUBGER’;
yar-e-fa, i.e., yareba, ‘BE-EV-CON’; ore-tar-e, ‘descend-PERF/PROG-EV/IMP’; and notate-
te, ‘pray-SUBGER’). The refrain ends in t-i-yowar-e, ‘go-INF-HON-EV/IMP.’ Although the
first two pairs of verses certainly take the refrain as a second clause of the same sentence,
the third pair, containing the KP, seem to cut the mata verse and the refrain into separate
sentences, with the main verse ending in a verb in the expected -e form. Aligning the
68
verses to the right highlights how the parallel pairs share their verbs (verb pairs are
bolded; the musubi candidates are underlined):
46ITI siyori UFU-kimi-kiya SIYORI mori ore-wa-tife
Shuri Great Priestess at Shuri, descending (into trance)—
ati-osoi siyo sedi masa-te t-i-yowar-e It is the King who will reign, spiritual power triumphant
MATA toyom-u kimi-toyomi-kiya ma-se neka-te ore-wa-tifeResounding Kimi-toyomi, praying to the true ancestors, and descending (into
trance)—[It is the King who will reign, spiritual power triumphant]
MATA amamiya-kara sufe n-o kimi yar-e-faPriestess with your power from long ago, when (you) convey (it to him,)[It is the King who will reign, spiritual power triumphant]
MATA sineriya-kara ai-tife-kimi yar-e-faPriestess (with your power) from long ago, when (you) convey (it to him,)[It is the King who will reign with his triumphant spiritual power]
MATA sasifu ITU-koro-ni mi-mafu-te su ore-tar-eThe Priestess, protecting the five human vessels, has descended(into trance)[It is the King who will reign, spiritual power triumphant]
MATA mutuki NANA-koro-ni ka-i-nate-te su ore-tar-eThe Priestess, protecting the seven human vessels, has descended (into trance)[It is the King who will reign, spiritual power triumphant]
MATA UFU-kimi-kiya ON-sause teru.kafa fa no-tate-te The Great Priestess sending her prayer up to the shining Sun, and[It is the King who will reign, spiritual power triumphant]
(208)
The translation attempts to show show that the first and second pairs of verses
form a complete sentence only with the addition of the refrain; the third pair of verses
consist of two sentences each. Thus the musubi for su in those lines is oretare, and not
choware, the final word in the refrain segment. In this song, of course, we would have
arrived at a verb ending in -e either way.
69
This phenomenon of a main verse with KP su providing its own predicate is not
consistent among all main verses with KP su, since even more frequently the refrain
segment must provide the sentence-final verb. But at least fifteen omoro have one or two
(out of many more) main verses with KP su and a verb differing from the endings of
other main verses, all of which are then followed by a refrain (which may or may not
have a KP construction). These lines fall into a class of main verses that are independent
sentences separate from the refrain line. They are categorized here according to the form
of the final word of the main verses.
For purposes of comparison, those lines without KP were also sorted, as far as
possible, by final word, in an attempt to identify patterns of distribution among them.
Some care is required here, because Takahashi’s electronic version sometimes breaks
what is grammatically a single sentence onto several lines, such as in omoro 847
(Example 47 below). As they appear here, four of these lines end in a verb plus the
subordinating gerund -tife/-te; two end in the subject marker ka (i.e., ga), and one ends in
the exclamation mimoN (column A in Example 47). In fact, this omoro consists of two
sentences, thus there are only two sentence-final words, both mimoN (column B) . Where
possible, counts were adjusted like this, but they must be taken with caution, as the
hypothetical nature of reconstructions and sentence breaks means they may be incorrect.
47 (Song 847 in column format used in the Excel database to examine data for this thesis)
No. Repeat Song line A B
847.01 ITI kikowe akesinoka N ka
847.02 sayafatake orewatife orewatife
847.03 akezumiso mesiyowatife mesiyowatife
847.04 kasanaori sasiyowatife sasiyowatife
847.05 namitotoro umitotoro osiukete osiukete
847.06 fiyakunano urafarika mimoN mimoN EXCL
847.07 MATA toyomu akesinoka N ka EXCL
As far as possible, I avoid using terms such as “evidential” and “attributive” to
name the forms of any verbs that appear in the lines under consideration, and instead use
descriptive terms such as “the verb is in -e” or “in -u”. This avoids assumptions as to the
70
meaning a certain ending gives to the verb, and focuses attention on form. At the risk of
causing some confusion, however, for the sake of brevity I have used the terms EV/IMP
(‘evidential/imperative’), ATTR (‘attributive’), PRED (‘predicative,’ more often called the
‘final’ form), and INF (‘infinitive’) in the song analyses.
B. FindingsIf Old Okinawan su corresponds to the second syllable in the Old and Middle
Japanese kakari particle koso, as Serafim and Shinzato 2005 persuades us it does, then it
may well also usually appear with a sentence-final verb in a form corresponding to an
evidential verb form, just as koso often did in Japanese until the early modern period.
Similarly, if do in the language of the OS corresponds to Japanese zo, then it should
usually appear with the sentence-final verb a form corresponding to an attributive form,
as it still does in Modern Okinawan, and once did in Japanese until at least the end of the
Heian period. At any rate, there should be a clear and distinctive pattern in the
distribution of each particle and the form of its musubi.
A single glance confirms that su and do do indeed largely follow the same pattern
as the Old Japanese kakari particles koso and zo, ka, ya, and namo (later namu), with the
verb musubi in what may be evidential and attributive form, respectively. That is, by far
the majority of lines with su have a line-final verb in -e; by far the majority of lines with
do forms have the verb in -(r)u.
71
Table 8 Count of KP su and KP do, sorted by line-final word form, in reconstructed songs
KP CategoryLines in Songs
with KPComparison: Lines in Songs
w/o KP (approximate42)
su vb e 430 1530 vb te 17 490 vb a 6 207 Noun 5 1057 vb u 2 106 vb una 1 19
KP su Total 461 3526do vb u 99 106 vb e 7 1530 vb una 3 19
vb i 2 98vb a 1 207
vb te 1 490KP do Total 113 3715
Overall, lines with kakari particles had a far more limited range of possible final
word forms than lines without one of the particles. Compared to more than 30 different
general form categories for the document as a whole, lines with KP su had six categories,
with two containing only one or two tokens; lines with KP do had six categories, with
three containing only one or two tokens after the final analysis. The two lists are similar,
with all but one of the categories for each particle also in the other list.
1. Endings of lines with su and its allomorphs
a. Lines with su , verbs in -e Obviously, overwhelmingly, in reconstructed songs KP su is linked with a verb in -e,
which includes all verbs ending in suffixes -ame, -are, -ere, -re and -e. This matches our
expectations for the particle su if it does indeed correspond to the OJ/MJ emphatic
particle koso, and if the -e ending in Old Okinawan corresponds to the evidential ending
of Old and Middle Japanese.
A typical example of su with a verb in -e is in the refrain of omoro 124, with
choware. Hokama 2000 translates the refrain using the Modern Japanese particle koso:
42 This count is approximate because only sentences in songs with no kakari particle at all were counted, to simplify the counting process. Thus, for example, in a song with one token of KP but a total of five sentences, none of the endings of the other four sentences are included in the count.
72
48ITI kikowe UFU-kimi-kiya kakura-we-ka tor-i-yowa-tife
renowned(INF) great-priestess-SUBJ gods.present-auspicious-day take-INF-HONAUX-SUBGER
Kikoe-ōgimi, choosing this auspicious day of the gods
aNsi-osoi su to-momo-sufe t-i-yowar-elord-HON KP ten-hundred-eternity come-INF-HONAUX-EV/IMP
It is our King who shall reign for a thousand years
MATA toyom-u se-taka-ko-ka [kakura-we-ka tor-i-yowa-tiferesound-ATTR spirit-high-one-SUBJ Resounding Priestess Sedaka, [choosing this auspicious day of the gods
aNsi-osoi su to-momo-sufe t-i-yowar-e]It is our King who shall reign for a thousand years
(124)
These lines make up the entirety of omoro 124; and they are precisely repeated in
omoro 6, as well (the two are marked as duplicates of each other, though 6 includes many
more verses than 124 does, and in addition one or two verses appear to be corrupted). The
presumed kakari-musubi construction is in the refrain, and since the same verb complex
ending the iti and mata lines ends in the conjunctive gerund, the refrain line is part of the
same sentence. After reconstruction, this gives two lines ending in choware. All KP that
fall in refrains in this way have been counted only once, unless the original song spelled
out the refrain lines for each verse.
It is not really clear whether the ending of the verb complex corresponds to the
Japanese evidential or imperative ending. The same refrain occurs in omoro 41 without
the particle su. However, four other songs have the words to-momo-sufe aNsiosoi su
choware (i.e., the first two words are reversed, and all four songs have su), so omoro 41
may be an exceptional case. In the case of omoro 124 at least (as well as omoro 6 and the
four with a similar phrase) choware could be analyzed as evidential.
A striking example of KP su triggering specific endings is in song 731, though it
is still possible to read the line as either evidential or imperative. In this song, the KP falls
in the second mata verse, which answers the question posed in the first mata verse. The
first mata verse shows another type of kakari-musubi, involving the question particle ka
73
and a verb in the attributive form, clearly distinguishable because it follows the
progressive/stative auxiliary -yor-:
49ITI akurena n-o tori-no kutinaka n-o tori-no weke koino
akurena DV-ATTR bird-GEN kutinaka DV-ATTR bird-GEN EXCL BIRDCALL
The Akurena bird’s and the Kutinaka bird’s (call), oh, “koino”
MATA nou mi-tife ka ofik-i-yor-u ikiya mi-tife ka ofik-i-yor-u what see-SUBGER QP follow-INF-PROG-ATTR how see-SUBGER QP follow-INF-
PROG-ATTR
What do you see as you follow? How do you see while you follow?
MATA kimi mi-tife su ofik-i-yor-e nusi mi-tife su ofik-i-yor-e priestess see-SUBGER KP follow-INF-PROG-EV/IMP priestess see-SUBGER KP
follow-INF-PROG-EV/IMP
It is watching the Priestess that you follow! It is watching the Nusi Priestess that you follow!
(731)
Finally, seven lines with KP in the OS end in verbs ending in -i. All cases involve
the original medial segment *fe, which, as discussed earlier, is precisely the syllable in
the OS that had already undergone the *e > i merger (Chapter 2, Section B.2.b.ii).
Example 49 shows the verb mukai (c.f. MidJ mukaf-, ‘face toward’) at the end of the
refrain lines, which for this omoro were all written out explicitly:
50ITI kikowe UFU-kimi-ka sei-yar-i-tomi mes-i-yowa-tife
renowned great priestess-SUBJ spirit-send-INF-HON ride(HON)-INF-HON.AUX-SUBGER
Kikoe-ōgimi, aboard the ship Sei-yari-tomi—
nami kase nakoyake-te sayafa-take kimi-kimi siyo muka-i wave wind gentle(INF)-SUBGER Seefa-shrine priestess-priestess KP greet-EV/IMP
Gentling waves and wind, it is the Priestesses at Seefa shrine who face toward (the Deity)
MATA toyom-u se-taka-ko-ka te-yor-i-tomi mes-i-yowa-tife resound-ATTR spirit-high-child-SUBJ hand-fold-INF-HON ride(HON)-INF-HON.AUX-SUBGER
Resounding Sedaka-ko, aboard the ship Te-yori-tomi—
74
nami kase nakoyake-te sayafa-take kimi-kimi siyo muka-i wave wind gentle(INF)-SUBGER Seefa-shrine priestess-priestess KP greet-EV/IMP
Gentling waves and wind, it is the Priestesses at Seefa who face toward (the Deity)(852)
Hokama glosses the verb mukai as kami-mukae (神迎え, ‘face toward the Deity’). Though
his (Modern) Japanese translation is the lower bigrade verb mukae-, the musubi candidate
in this omoro corresponds to the quadrigrade mukaf- of Old and Middle Japanese. With
the ending in -e, the final segment becomes -fe, which word-medially was pronounced
/ye/. This is exactly the segment that Takahashi (1991b) and Serafim (1977) found had
already merged with *i in the OS, resulting in a word read /mukai/. Thus we can read
mukai in omoro 852 above as ending in -e that had raised, and lines like these maybe
considered regular, at least as far as ending in a candidate musubi in an expected form.
b. Lines with su , verb in -te After verbs ending in -e, the most-frequently occurring ending of musubi candidate verbs
for KP su is that of verbs in -te. Even so, after reconstructing the lines for the likeliest
line-final words, only seventeen lines with su end in verbs in -te. The so-called “rule of
three”43 allows us to postulate that this must be a valid verb form accompanying a kakari
particle, as opposed to a mistake. All the same, this form with KP su is rare.
Several lines in this set have the verb orotife (‘bring.down-SUBGER’) as the
candidate musubi; all eight of the lines with orotife fall in only two songs, 628 and 1457,
duplicates of each other. In all these instances, the KP falls at the end of a phrase that can
be translated as ‘It is the command of [deity or priestess].’ The lines with KP su from
omoro 628 are in Example 51:
43 The ‘rule of three’ goes something like this: if a segment with an unusual form occurs only once (a hapax legomenon), it can be dismissed as an error or a fluke; if it occurs twice, it may be mere coincidence; three times, and there is probably a pattern. (With thanks to Alexander Vovin.)
75
51ITI UFU-kimi-ka u-sasi siyo omo.kafa-no seti oro-tife
great-priestess-POSS HON-order KP rising.sun-GEN power bring.down-SUBGER(?)
It is the order of the Great Priestess that brings down the spiritual power of the Sun (to earth)
aNsi-osoi-yo mafur-a tete ore-wa-tifelord-HON-OBJ protect-VOL QUOT descend-HON.AUX-SUBGER(?)Saying, “Let us protect the King,” she is falling (into trance as the spirit fills
her)
MATA se-taka-ko-ka u-sasi siyo omo.kafa-no seti oro-tife spirit-high-one-POSS HON-order KP rising.sun-GEN power send.down-
SUBGER(?)It is the order of Priestess Sedaka that brings down the spiritual power of the
Sun (to earth)
MATA teru.kafa-ka u-sasi siyo teratiN-no seti oro-tiferising.sun-POSS HON-order KP east.direction-GEN power descend-SUBGER(?)It is the order of the Sun that brings down the spiritual power of the East (to
earth)
MATA terusino-ka u-sasi siyo teratiN-no seti oro-tife dawing.sun-POSS HON-order KP east.direction-GEN power descend-SUBGER(?)It is the order of the Sun that brings down the spiritual power of the East
(to earth)(628)
Serafim’s suggestion is that the ending of oroche (i.e., <orotife>) and the other
verbs in this set could be some kind of apocopated evidential ending following the
perfective (i.e., *-i-t-ure > *-i-t-uye (after -r- drop) > *-i-twe(e) > *-i-t-e(e) > OS -ch-e(e)
‘-INF-PERF-EV’44). This is certainly plausible in this omoro, especially given that a
perfective would fit in very well in the translation of the lines: instead of the simple
present-tense descends, the line would read with the perfective has descended. This
would also go a long way in explaining the hundreds of lines (tentatively, at least 490) in
the OS that end with a verb in -te or its allomorphs, and where there are no other line-
final words available through reconstruction. All those lines could very nicely be
44 Etymology by Leon Serafim, in a personal communication, October, 2009.
76
translated with the perfective, rather than appearing to be incomplete in some way, with a
final, hanging subordinating gerund to explain. Consider, for example, the song provided
in Section C.2 above, in the discussion on reconstructing songs. It is given again in
Example 52 (33 above), where the final verb in the second verse was analyzed with a
subordinating gerund:
52 (33 above)ITI arakaki-ni owar-u ma-MONO yo-no nusi n-o ma-moN
Araki(PN)-LOC exist(HON)-ATTR true-thing world-GEN lord DV-ATTR true-thingOverlord who resides in Arakaki; Overlord who rules his domain
MATA yosa n-o kuwa fa nas-i-yowa-tife fatu n-o kuwa-fa nas-i-yowa-tife prime DV-ATTR child TOP create-INF-HON-SUBGER
first DV-ATTR child TOP create-INF-HON-SUBGER
Siring the primary child; creating the first child
MATA to-momo so n-o ikusa ya-momo so n-o ikusa ten-hundred person DV-ATTR warrior eight-hundred person DV-ATTR warriorArmy of ten thousand warriors; army of eight hundred warriors
(590)
Analyzing the line instead with a perfective gives a much cleaner reading, and removes
the troubling gerundive:
53MATA yosa n-o kuwa fa nas-i-yowa-tif-e fatu n-o kuwa-fa nas-i-yowa-tif-
e prime DV-ATTR child TOP create-INF-HON-PERF-EV/IMP
first DV-ATTR child TOP create-INF-HON-PERF-EV/IMP
The primary child has been sired; the first child has been created45
In this way we can solve one of the grammatical puzzles of the OS. However, there
remains a question: if this is an evidential form, why does it occur with KP su only
seventeen times? Perhaps it is not an evidential after all, but an imperative, thus its
frequent appearance in lines with no KP. This is the same situation for all verbs in -e in
the OS, in which the imperative and the evidential appear to have fallen together. Another
set of lines with KP su and unexpected verb forms as the candidate musubi seem to be a
45 Passive translation used here for the effect of focusing information, as in the original.
77
result of this collapse of the distinction between imperative and evidential forms, those
with verbs in the mild honorific -yowa (next section).
c. Lines with su, verb in -a Six lines (in three songs) with KP su end in verbs in -a after reconstruction. In all
but one, the verb complexes have the form [stem (+INF) +HON.AUX(M.IMP)], such as
maburiyowa (i.e., <mafuriyofa>, ‘protect-INF-HON(M.IMP)’), and can be read as mild
imperatives, such as one might give to a superior.
54ITI tomo-soi-ka nate-koro tatinate su mafur-i-yofa
Tomosoi-POSS caress-male be.intimate(INF) KP protect-INF-HON.AUX(M.IMP)Beloved of Priestess Tomosoi; it is in intimacy that you will protect (him)
yoso-o seti seti masaru wakawe-kiyou conquer-ATTR(?) power power triumph-ATTR young-oneWorld-conquering spiritual power; vital King triumphant with spiritual power
(800)
One omoro has a refrain line with KP su and the musubi candidate ending in -a, not an
imperative, but a volitional form that may be translated as an intention or suggestion:
55ITI kikowe sasukasa-ka toyom-u UFU-kimi ya
renowned sasukasa-SUBJ resound-ATTR great priestess TOP
Renowned Priestess Sasukasa, the resounding Great Priestess—
sasukasa su nasai-kiyo-moi mafur-a sasukasa KP maker-one-HON protect-VOL
It is Priestess Sasukasa who will protect the King, our Father
MATA toyom-u sasukasa-ka …resounding-ATTR sasukasaResounding Priestess Sasukasa,
(177)
This is the only instance of KP su appearing with a verb in the form of the stem + a, with
no intervening honorific auxiliary. It may be dismissed as an error; it may also be a
modification to meet a certain syllable count.
78
Given the collapse of the imperative and evidential into the form -e, the musubi
verbs in an imperative form not affected by that merger may reflect a re-analysis by
younger speakers of the forms that may accompany KP su. If a form indistinguishable
from the imperative appears at the end of a sentence with KP su, there is nothing to
prevent a new generation of speakers from generalizing the rule such that any imperative
form may follow KP su. These instances of KP su having its candidate musubi in the mild
imperative -yowa are a natural result of the partial collapse of the distinction of evidential
and imperative forms.
d. Lines with su , ending in a noun Five lines with KP su end in nouns, even after reconstruction of partial lines. Hokama
2000 translates all three lines by adding the copula de aru, obviously assuming an
“invisible” copula in the original. In fact, this is common in the songs of the OS, with
more than a thousand lines ending in nouns presumably followed by an understood
copula. Possibly the lines in question here were somewhat exclamatory in function,
announcing the presence of some important place or person.
e. Lines with su, verb ending in - u , and verb ending in … u-na ) The remaining items occur fewer than three times, thus allowing us to more easily
dismiss them as mistakes or anomalies. However, they may still be worth a closer look.
Two lines, one reconstructed, both in the same omoro, have their candidate
musubi in -u.
56MATA tiyara-moi-ka ke n-i siyo kimi-si teta mi-tiyar-u
Chara(PSNM)-HON-POSS true DV-INF KP lord-leader sun see-PERF/PROG-ATTR/PRED
Truly, it is Lord Chara who has seen our King, Kimi-si
MATA omof-i-kuwa-no ke n-i siyo [kimi-si teta mit-iy-ar-u]reign-INF-one-SUBJ true DV-INF KP Truly, it is Priestess Omoiguwa who…
(1107)
The particle is spelled <siyo>, and might be an example of KP do showing a palatalized
spelling (see Chapter 2, Section B.2.i), especially since it falls into the environment for
palatalization. This hypothesis is acceptable if we allow the possibility that the initial
79
consonant was voiced, and that OOk did not distinguish initial <zi> and <di> (like
Modern Japanese). Working against the hypothesis, however, is the fact that in both lines
here, the particles follow the phrase ge n-i (<ke n-i>, ‘true DV-INF,’ or ‘truly’). There are
no unmistakeable tokens elsewhere of KP do (e.g., ones spelled <to> or <do>) following
this phrase. KP do does occur in lines with ke n-i, but it follows other words, such as
omokasiya ‘amazing,’ or is in the sentence kore do, discussed above. There is, in fact,
only one token of ge n-i su, and it falls at the end of song 1031 in such a way that it
appears to be sentence-final, and thus has no musubi candidate. There is no way to
determine what form a verb may take in a line with ge n-i su in it.
When ge n-i appears alone, however, it may appear with the sentence-final verb in
the attributive, such as in 1107 above. Other examples include omoro 94 and 971:
57 meturasiya ge n-i a-yor-u
amazement true DV-INF exist(INF)-PROG-ATTR
Truly, there is amazement!(94)
58 itife-te ke n-i far-i-yor-u
exit-SUBGER true DV-INF run-INF-PROG-ATTR
Going out (from the harbor), truly (the ship) is running (before the wind)(971)
Thus, though ge n-i frequently appears with a verb in -e, it may also appear with the final
verb in the attributive form, such as in these examples (which are unmistakeably
attributive, because the verb complexes include the progressive extension -yor-, which
preserved the distinctive sentence-final predicative and attributive forms).
There is a degree of irregularity surrounding both ge n-i and its synonyms da n-i
and da n-i do, that has yet to receive a satisfactory explanation; what we are seeing here
may be an awareness on the part of the writer that ge n-i may take its final verb in -u,
conflicting with the awareness that it is never ge n-i do but only ge n-i su. Thus su,
spelled <siyo> because of its environment, trumps do—but does not have its musubi in
the expected form.
80
The final line with its verb in an unexpected form for agreement with KP su may
be found in song 274. At first glance the verb looks as if it ends with negative imperative
-una; however, this ending may be an attributive, plus the enclitic particle -na, giving a
hortative meaning:
59
MATA kiyaga su safezur-u-na fata su safesur-u-na near KP? twitter-ATTR-PT side KP? twitter-ATTR-PT
?It is nearby that (you) should speak! It is at (his) side that (you) should speak!(274)
Given that we have already seen a small degree of confusion in the verb endings in lines
with KP su, and an tendency to allow imperatives to fill the musubi slot for su, it does not
seem like a long leap to see this use of the hortative as a similar case. Thus this
unexpected ending may be related to the process that eventually saw the disappearance of
su-type kakari-musubi from Okinawan.
The main weakness for analyzing this ending as the hortative is the evidence in
Modern Okinawan, where the enclitic always follows the so-called mizen-kei, that is, the
verb stem plus -a. Thus, for example, keerana, ‘How about we go home.’ This form, with
-a- instead of -u- does appear in the OS, such as in omoro 96, which has omofana, ‘let’s
consider.’ This ending also appears (very rarely) with KP do (covered below), so it is
possible that the form was perceived as compatible with emphatic structures.
2. Endings of lines with do and its allomorphs
a. Lines with do , verb in -u Ninety-nine (out of 113) of the lines with KP do have a line-final verb in the expected -u.
Several of these lines have verbs with consonant-final stems, so whether the final -u is the
predicative or attributive conjugation is often impossible to say. However, 43 lines have
verbs with the stative/progressive extension -yor- (:: OJ wor-, ‘exist’), and twenty lines
give aru ‘exist’ as the musubi candidate. Because OOk preserved the -i predicative
conjugation for -yor- and ar-, a very large portion of the lines with KP do can be
definitively analyzed as ending with a verb in the attributive.
81
That Modern Okinawan retains its robust do emphatic structures may be due to
this very fact. Already in the OS we may see the rising frequency of verbs with stative
extensions -yor- and -ar-. In Modern Okinawan the plain non-past form of all verbs
incorporates the stative / progressive infix, though the precise path followed to reach the
present forms is one of the mysteries of Ryūkyūan linguistics. Thus where Japanese
constrasts non-past (e.g. kōhii-wo nomu, ‘I (will) drink coffee’) and progressive (kōhii-wo
nondeiru, ‘I am drinking coffee’), in Okinawan both are captured in the plain form (kōhii
numun, ‘I (will)drink / am drinking coffee’).46 The attributive ending is …r-u (numuru
kōhii, ‘coffee that (I) drink’; or kōhii du numuru ‘it’s coffee (that) I’m drinking’), clearly
the historical attributive ending; the mystery in Common (Modern) Okinawan lies in
explaining the form of the plain verbs, with their word-final nasal.
As for our list of verbs in -u, the few bigrade verbs that occur all appear to have
converted to final -r- stem conjugations for final and attributive forms (i.e., yokakeru,
yoseru, tatoferu, and mitiferu), though they may appear as vowel-final stems in
combinations requiring the infinitive or evidential forms. Only four verbs ending in -uru
appear in the document as a whole. Aturu (‘exist-PERF-ATTR,’ in several songs) and
mituru (‘fill-ATTR,’ in 814) appear modifying nouns, but not as a musubi in any lines with
a kakari particle. In the two other tokens, motirokiyuru (‘shine.beautifully-INF-PROG-
ATTR’) and siyuru (‘do-INF-PROG-ATTR’), the supposed attributive suffix can be more
simply analyzed as the stative/progressive auxiliary -yor-u with a spelling reflecting the
u≈o alternation, plus an ending -u.
Below (Example 60) is an omoro showing a verb with a consonant-final stem
ending in -u:
60ITI kikowe UFU-gimi-kiya
renown great-priestess-SUBJ
Kikoe-ōgimi (prays:)
amamiya-YO-no uf-u-TAMA47
46 Karimata Shigehisa, personal communication, January 2006. 47 Note that this is ub-u ‘birth-ATTR’ (as it appears in the next verse) and not ufu ‘great.’ Ubutama refers to a ‘jewel of fecundity,’ a ritual object.
82
long.ago-world-GEN birth-ATTR-jewelThe ancient jewel of fecundity—
ub-u-dama fa inor-u su-do yo kaker-ubirth-ATTR-jewel TOP pray-ATTR person-KP world
dominate-ATTR/PRED
As for the jewel, it is the one who prays (to it) who may rule the world
MATA toyom-u se-taka-ko-karesound-ATTR spirit-high-child-SUBJ
Resounding Sedaka-ko (prays:)(102)
The refrain segment in song 396 (Example 61 below), in contrast to the example
from omoro 102 above, clearly shows the attributive ending, because the verb complex
has the progressive/stative infix -yor-:
61ITI omoro ne-yakar-i ya serumu ne-yakar-i ya
song cry-rise-NOM(PSNM) TOP song cry-rise-NOM(PSNM) TOP
Omoro Singer; Serumu Singer—
mi-tife i-tiwe omo-kake to tat-i-yor-u see-SUBGER say-SUBGER face-shape KP stand-INF-PROG-ATTR
Seeing, going, it is the figure that is going(396)
For contrast, note that omoro 912, which has no KP, has the same verb, but it ends in the
distinctive predicative allowed by the presence of -yor-:
62ITI miyakiseN-ni tat-u kumo ko-kane kumo tat-i-yor-i
Nakijin-LOC rise-ATTR cloud gold-metal cloud rise-INF-PROG-PRED
The clouds that rise above Nakijin—lovely clouds rise
UFU-kimi-ni owe-tife kou-te fariy-as-e great priestess-SUBJ follow-wind pray.for-SUBGER run(INF)-HON-EV/IMP
Run (the ship) by begging the Great Priestess for tail-winds(912)
83
The remaining 14 lines with KP do end with verbs in endings other than -u, or
with noun phrases.
b. Lines with do : verb in - e Seven lines with KP do end with a verb in -e; specifically, they all end in choware
(i.e., <tiyoware>, lit., ‘go-INF-HON.AUX-EV/IMP’; usually translated here as ‘may [he]
reign’). Three of these lines have the phrase imiya-kara do (‘now-ABL KP’), such as in
omoro 113 in Example 63:
63MATA weso-niya sufe ati-osoi imiya-kara do sedi masa-te t-i-yowar-e
Eiso-HON forever lord-HON now-ABL KP power triumph-SUBGER reign-INF-HON.AUX-EV/IMP
Honored King, descendent of King Eiso: from this moment, your spiritualpower triumphant, may you reign
(113)
In his translation for this particular song, Hokama translates this line with the split
structure …zo...koto yo, which he frequently uses to handle lines with KP do. For the
others in this set, he also uses zo, but he ends his sentences with imperatives. Assuming
these lines are not the result of scribal errors, then we might surmise that when called for,
the imperative/evidential form took precedence over the predicate/attributive, and the
latter could be replaced by the former.
That said, a total of eighteen lines in the OS have the phrase imiya-kara do, and
the three in this set (i.e., the set of lines with KP do and a verb in -e) are the only ones
with their candidate musubi verbs in -e; all the rest have a verb ending in -u. Additionally,
in one of the three, omoro 736, the line may be corrupted:
64MATA ati-osoi ya imiya-kara to seti masa-te t-i-yowa-ya(re)
lord-HON TOP now-ABL KP power triumph-SUBGER reign-INF-HON.AUX-??Honored King: from this moment, your spiritual power triumphant, may you
reign(?)(736)
The Shō manuscript does not show the final syllable <re>; nor did any of the
Aniya-descendant texts. The final syllable was introduced by the authors of the 1965
84
Kōhon Omoro sōshi, Hokama and Nakahara; Takahashi added it in his electronic version
(which is why it ended up in this section, rather than in one for verbs ending in -a). The
editors of the Teihon Omoro sōshi (2002), Hokama and Hateruma, note that perhaps the
<ya> is a mistake, and due to the presence of do in the sentence, should be <ru>. All this
suggests this line, and by implication the other two with imiya-kara do and a verb ending
in -e, may be regarded as of suspect quality, and discounted as legitimate examples of the
KP do appearing with its musubi candidate ending in -e.
c. Lines with do , verb in te Section B.1.b (on lines with KP su with musubi candidates ending in supposed -te
forms) shows how this ending is a perfective-evidential ending,48 thus the analysis
provided in Example 65. Two lines with KP do with musubi ending in -t-e are in
duplicate omoro:
65MATA aNsi-osoi diyo yok-i-ya-te amako awa-tife soko-t-e
lord-HON KP go-INF-meet(INF)-SUBGER eyes join.together(INF)-SUBGER be.proud(INF)-PERF-EV/IMP
It is the King who, moving as one and locking eyes (with the priestess/deity), rejoices
(100; duplicate 490)
Since Hokama 2000 translates the particle as koso, it remained in our pool of
kakari particles by default. However, the words yokiyate (‘go-and-meeting’) and awache
(<awatife>, c.f. J awasete ‘joining together’) are both verbs indicating some mutual
action involving two people, or possibly here, the King and a god or a priestess. Thus
<diyo> could be the comitative particle to, corresponding to the Japanese comitative of
the same spelling. In fact, a few verses earlier in the same omoro, the same verb phrase
appears, preceded by a particle spelled <to>:
48 That is, original *(-i)-t-ure > *(-i)-t-uye (after -r- drop) > *(-i)-twe(e) > *(-i)-t-e(e) > OS -ch-e(e)/-t-e(e)
85
66MATA terukafa to yok-i-ya-te o-koto awas-i-yowa-tif-e
shine.river COM go-INF-meet-SUBGER HON-word exchange-INF-HON.AUX-PERF-
EV/IMP
Moving as one with the Sun Deity, (the King) exchanged solemn words(100)
Here the particle is not in the environment for palatalization. Although the verb
complex, unlike in Example 65, includes the honorific auxiliary, the verb is the same.
Additionally, Hokama does not include an emphatic particle in his translation of this line.
Reading the particle as the comitative, the song in Example 65 may be re-
analyzed as in Example 67 below, and we may in fact remove these lines from the pool of
kakari particles:
67MATA aNsi-osoi diyo yok-i-ya-te amako awa-tife soko-t-e
lord-HON COM go-INF-meet(INF)-SUBGER eyes join.together(INF)-SUBGER be.proud(INF)-PERF-EV/IMP
Moving as one with the King; joining eyes (with him, the priestess/ deity) rejoiced
(100; duplicate 490)The remaining line in the set, with a KP do and a verb ending in -t-e, has no such
verb showing mutual action:
68MATA kikowe uraosofe to kuni-no otodiya nas-i-yowa-tif-e
renowned Urasoe(PN) KP country-GEN older.sibling make-INF-HON.AUX-PERF-
EV/IMP
It is the renowned (Lord of) Urasoe whom (they) have made the Chief of theland
(477)
Unless the line can be analyzed as in Examples 65 and 66 with the particle as a
comitative meaning ‘and,’ resulting in the translation “(They) have made you (Lord of )
Urasoe and Chief of the land,” then this is an example of a line in which KP do is
accompanied by a musubi with an evidential ending.
86
d. Lines with do , verb in … u-na Three lines in two omoro with KP do have their candidate musubi ending in forms
that look like the negative imperative -una. The first is in the refrain segment of omoro
265 (which, incidentally, has a second kakari structure involving the question KP ga in
the main verses). The OKDJ gives another meaning for the ending, however, analyzing
this as ...u (attributive) plus an interrogative particle na, giving this particular song as an
example. This is, of course, the same ending seen above with a single token of KP su, in
which we took -na as an enclitic particle conveying a hortative meaning (Section 2.b.1.e).
69ITI nawotikiyo wa keyo wa no ga s-i-yor-a-i
Naojiko TOP today TOP what KP do-INF-PROG-INF.AUX-QP
I wonder what Naojiko is doing today?49
i.miki-mori diyo itiyonasiya s-i-yur-u-na sacred.wine-store KP busy do-INF-PROG-ATTR-PT
Surely it is with the store of sacred wine that (s)he is busy (265)
Hokama 2000 does not translate this song with a negative imperative, and given
the translation here, it is clear this is because the line would not really make any sense
with a negative imperative in an answer to the preceding question. Such an analysis also
fits Hokama’s translations for the two remaining lines in this set, which are in omoro
1087, given below (Example 70). This omoro has a difficult-to-translate refrain segment,
and also involves sacral wine:
70ITI uraosoi ya uraosoi ya mikii to ar-u-na sake to ar-u-na
Urasoe(PN) TOP Urasoe TOP sacred.wine KP exist-ATTR-PT wine KP exist-ATTR-PT
O, Urasoe, as for Urasoe, there certainly is sacred wine; there certainly is wine
49 Translation from Serafim and Shinzato (forthcoming).
87
tasiya tasiya kiyo ya kiyo ya yoyose-niyo-ka katife tukai UNK UNK today TOP today TOP Yoyose(PSNM)-HON-POSS location
messenger…O, today, today, (we send) a messenger to the place where Yoyose
(=King Ezo) is
MATA tokasiki ya tokasiki ya sake to ar-una miki to ar-u-na Tokashiki(PN) TOP Tokashiki TOP wine KP exist-ATTR-PT sacred.wine KP
exist-ATTR-PT
O, Tokashiki, as for Tokashiki, there certainly is wine; there certainly is sacred wine
(1087)Although the same difficulty arises, that the evidence of Modern Okinawan (and
in some songs of the OS) that this particle followed the mizen-kei, that is, the verb stem +
a, because this translation is the only one that really makes any sense, this seems to be the
best way to analyze the verb. Because these translations work in these and other songs,
we must accept Hokama’s interpretation of this verb ending.
e. Lines with do , verb in i Two lines with KP do end in verbs ending in the same form -i, but with quite
different endings in terms of function. The first line is in omoro 376, notable for also
having a sentence-final KP su (spelled <siyo>, in the iti verse, in the same slot as
question particle -i in Example 67 above50):
71ITI siyori-mori kusuku keo wa no ka s-i-yor-a siyo
Shuri-grove precinct today TOP what QP do-INF-PROG-INF.AUX FP
At the sacred grove in Shuri—what might (they) be doing today?
amafe to itiyonasiya to s-i-yor-a-i rejoice KP busy KP do-INF-PROG-INF.AUX-QP
I wonder if (they) are rejoicing, and busy, doing (their ceremonies)(376)
The second line of this set has a verb with a clear stative extension, and the KP is
spelled <ro>:
50 Serafim and Shinzato (forthcoming) note that this is an example of a KP su functioning as a question particle, having become separated from its origin as a kakari particle. Its palatalized spelling, despite the lack of a triggering high front vowel preceding it, indicate it is a fossilized form taking on a new function.
88
72ITI ak-asu kunikane ya nayof-i-tife ore-t-e
sate-NEG/ATTR Kunigane TOP flutter.gracefully-INF-SUBGER descend-PERF-
EV/IMP
Priestess Kunigane, whom we never tire of seeing51, her sleeves fluttering gracefully, has descended (into trance)
MATA kuni n-o siturai ya [nayof-i-tife ore-t-e]country DV-ATTR Siturai TOP …Siturae, nation’s priestess, [her sleeves…]
MATA kimi yo mi-tiyo ro mi-yar-i priestess PT HON-place KP see(INF)-STAT-PRED O, Priestess, it is the holy place that you are seeing!
(486)
It is not clear whether the refrain segment should be repeated in the second mata verse,
since the form of the line is so different from the preceding two. In the first two verses,
the segment before the refrain has eight syllables; the third verse as given consists of nine
syllables. The last verse is not in any way parallel to the first in meaning either, though it
does call on the priestess again. Hokama translates the particle with zo.
f. Lines with do , verb in a Four lines, three in a single song, have musubi candidates with final -a. Serafim and
Shinzato (forthcoming) label this the inferential auxiliary (IA); it lends an evocative tone
to the line.
The fourth line with KP do and a possible candidate musubi verb in -a after
reconstruction is in 1155:
73MATA a-ka otodiya sum-ar-a
1PS-POSS older.brother dwell-STAT-INF.AUX
O, our older brother (=Lord Arafae of Gushikawa) dwells (here)
MATA yositaru to sum-ar-a Yositaru(PSNM) KP dwell-STAT-INF.AUX
O, Yositaru (=Lord Arafae) dwells (here)(1155)
51 The “seeing” comes from Hokama’s explanation of ak-asu.
89
The three lines that fall in one song are in omoro 996, and strictly speaking the
verb is not sentence-final: these lines also have a sentence-final KP su (FP). The ending of
the musubi candidate ikya (<ikiya>) shows palatalization due to the /yi/ in the previous
syllable:
74MATA nisi miti-no tiyana miti ru ik-iya siyu
west road-GEN Jana(PN) road KP go-INF.AUX KP
O, it is by the westward road to Jana that (he will) go, indeed
MATA fika miti-no yaki miti ru ik-iya siyo east road-GEN Yagi(PN) road KP go-INF.AUX KP
O, it is by the eastward road to Yagi that (he will) go, indeed
…
MATA itiya yakena NAKA miti tiyo ik-iya siyo EXCL Yakena(PN) middle road KP go-INF.AUX KP
And O, it is by the middle road to Yakena that (he will) go, indeed(996)
Note also that like the particle <siyo> in Example 67 that is functioning as a question
particle, here the spelling of the sentence-final particle reflects palatalization, even
though it is not in the environment for it. This indicates that this final particle may be
fulfilling some special function, though because these lines are lacking wh-question
words, it does not appear they are question particles.
C. ConclusionThe obvious conclusion to made from the numbers shown in Table 8 is that
irregularity in the kakari-musubi system of the Omoro sōshi is very low. Once the omoro
have been reconstructed according to the principles laid out by Tamaki (1981), and with
attention paid to such matters as syllable count and parallel structure and meaning, then
the lines of the songs are far more regular, with fewer unexpected words and forms in
sentence-final slots. Separating out those segments with the same spellings as the kakari
particles and their allomorphs, but with obviously different meaning and functions, clears
out a large part of any perceived irregularity in the lines with kakari-musubi structures.
Finally, a nearly perfect patterning emerges when those kakari particles functioning as
90
sentence-final particles are set aside into their own class. The remaining irregularities are
few, and even so, most of them can be accounted for through phonological or analogical
changes. These include those verbs with stems ending in -f- that underwent a sound
change making it possible to spell them with <i>; and the apparent perception, perhaps by
younger speakers, that KP su could be accompanied by the imperative, thus allowing a
musubi in the mild imperative in some omoro.
The evidential and the imperative forms in -e cannot be distinguished in the OS,
except by the presence or absence of a kakari particle in the sentence. At some point,
younger speakers may have recognized only that KP su appeared with the imperative
verb form; this understanding then may have generalized to allow su to appear with other
types of imperative, or even hortatives.
Certain verbs occur far more often with KP su than others, such as the 78 tokens
of choware in such lines, versus the near absence of mioyase in lines with KP su,
surprising since mioyase is the single-most frequently occurring verb in the OS. The
collapse of the morphological distinction between imperative and evidential basic forms
may be the cause of the disappearance of su-type (:: OJ koso-type) kakari-musubi in
Okinawan. This may have been a process similar to that which resulted in the loss of
kakari-musubi in Japanese after the attributive and predicative forms loss their
distinction.
Conversely, already in OOk, we may see that the addition of auxiliaries -yor- and
-ar- to verb stems may be the root cause for its preservation of do-type (:: OJ zo-type)
kakari-musubi. Because OOk preserved the distinction of -i predicative and -u attributive
conjugation for these auxiliaries, it also preserved, as a side-effect, its do-type kakari-
musubi system. Thus lines with do-type KP also pattern nearly perfectly, with the musubi
most frequently appearing ending in -u.
The subordinating gerund -(i)te and the perfective-evidential ending -t-e often
cannot be distinguished either, except by sentence position. Where a verb is not the main
verb in a sentence, then usually it can be analyzed as the subordinating gerund; however,
sentence-finally, we must analyze any verb ending in <te> or its allomorphs <tiwe> and
<tife> as perfective-evidential. This offers a solution to the seeming puzzle of a high
91
prevalence of line- and sentence-final verbs in a form ending in -te, long explained away
as being due to the technique of abbreviation in writing down the songs of the OS.
Reading sentence-final verbs ending in <te> or its allomorphs as perfective plus
evidential/imperatives removes the difficulty of having so many sentences seeming to
end in hanging gerunds.
The frequency of kakari particles functioning as sentence-final particles was very
high, with about 80 tokens. Removing these lines from the pool of kakari particles sorted
by sentence-final word cleared out nearly all the perceived irregularity that motivated the
writer to conduct this study. Classifying lines with the clausal kore do as occurrences of
sentence-final particles was a key step in regularizing the analysis of lines with KP do.
Finally, in most songs, refrain segments should be regarded as completions of
sentences begun in the main verses marked by iti and mata. Thus many main verses end
in a verb in the subordinating gerund, and the sentence continues with the refrain.
Sometimes, however, the main verses may provide a main clause; in those cases, the
refrain must be considered a separate sentence from the main verses. This point is
especially relevant when a kakari particle occurs in a main verse, but no valid musubi
candidate is available in the refrain; often the musubi is actually in the main verse.
This thesis has laid out a comprehensive review of the verb forms which may be
taken as the musubi candidates for the kakari particles su and do in the Old Okinawan
represented in the Omoro sōshi. What had seemed like a fairly extensive degree of
irregularity within the kakari-musubi system in the OS has been dramatically reduced to a
double handful of lines with verbs in unexpected forms with no explanations. Future
studies might explore, for example, the enclitic particle -na, and how it may follow both
the attributive and the so-called mizen-kei form. Other future research might include an
exploration of shamanic traditions in East Asia and the Pacific, with a look at the role of
language in ritual and religious customs and beliefs. Such studies, and a wider knowledge
of the liturgical traditions of other shamanic cultures may provide more insight into
obscure references in the OS.
92
BIBLIOGRAPHY
Manuscripts of the Omoro sōshi, in chronological order [name, date of copy, parent copy, current location]:1. Original. Vol. 1: 1532. Vol. 2: 1613. Vols. 3-22: 1623. Possibly a copy of lost original.
Lost in 1709.
2. Gushikawa manuscript. By 1660. Copy of (1) above or of an older lost original. Location unknown since 1710.
3. Shō manuscript. 1710. Based on Gushikawa ms. Now in the Prefectural Museum of Okinawa.
4. Aniya manuscript. 1710. Based on Gushikawa ms. Location unknown since 1945.
5. Aniya second manuscript. 1879? Based on Aniya ms. Location unknown since 1880s.
6. Part of the Ryūkyū shiryō collection (3 copies made). 1888-1892. Possibly copied from Aniya second ms. Current locations unknown.
7. Prefectural government copies (2 or 3 copies made). 1880s? Based on Aniya second ms. Current locations unknown.
8. Tajima manuscript. 1893. Based on Ryūkyū shiryō OS. Currently in Ifa Fuyū Collection of the University of the Ryūkyūs Library.
9. Nishi manuscript. 1893? Copied from the Ryūkyū shiryō OS. Current location unknown.
10. Basil Hall Chamberlain manuscript. 1893. Copied from Nishi ms. for BHC. Current location unknown.
11. Kara-no-ya manuscript. After 1893. Copied from BHC’s ms. Volume 3 now in Tsukuba University collection; location of remaining volumes (if any) unknown.
12. Nakayoshi manuscript. 1899-1912. Based on Aniya ms. Currently in Ifa Fuyū Collection of the University of the Ryūkyūs Library.
13. Prefectural Library manuscript. Copy made while Ifa Fūyū was head of the Prefectural Library. Based on Tajima ms. Lost in 1945.
Print editions of the Omoro sōshi, in order of publication.1. Ifa Fuyū (ed.). 1925. Kōtei Omoro sōshi [Omoro sōshi, revised and updated].Tokyo:
Kyōdo kenkyū-sha.
2. Nakahara Zenchū and Hokama Shuzen (eds.). 1965. Kōhon Omoro sōshi [Omoro sōshi, variorum edition]. Tokyo: Kadokawa.
3. Hokama Shuzen and Saigō Nobutsuna (eds. and trans.). 1972. Omoro sōshi (Nihon shisō taikei 18) [Book of Omoro (Series on Japanese Thought, vol. 18)]. Tokyo: Iwanami Shoten.
93
4. Omoro sōshi [Reproduction of the Shō manuscript]. 1979-1980. Tokyo: Hirugi-sha.
5. Higa Minoru (ed.). 1993. Omoro sōshi. Tokyo: Hosei Daigaku Shuppan Kyoku.
5. Hokama Shuzen (ed.). 1993. Omoro sōshi. Tokyo: Kadokawa Shoten.
6. Hokama Shuzen (ed. and trans.). 2000. Omoro sōshi, 2 vols. Iwanami Bunko, yellow series, 142-1 & 2. Tokyo: Iwanami.
7. Hokama Shuzen and Hateruma Eikichi (eds.). 2002. Teihon Omoro sōshi. Tokyo: Kadokawa Shoten.
General BibliographyHendriks, Peter. 1998. Kakari particles and the merger of the predicative and attributive
forms in the Japanese system. In: N. Akatsuka, S. Iwasaki, & S. Strauss (eds.), Japanese/Korean Linguistics 7: 197-210. Stanford: CSLI.
Higa Minoru, Nakamoto Masachie and Christopher Drake. 1984-1993. Omoro kanshō: Ryūkyū koyō no sekai [Omoro appreciation: The world of Ryūkyūan traditional songs]. Series in Gekkan Gengo 13:10-22:10.
Hokama Shuzen and Saigō Nobutsuna (eds. and trans.). 1972. Omoro sōshi (Nihon sisō taikei 18) [Book of Omoro (Series on Japanese Thought, vol. 18)]. Tokyo: Iwanami Shoten.
Hokama Shuzen and Hateruma Eikichi (eds.). 2002. Teihon Omoro sōshi. Tokyo: Kadokawa Shoten.
Hokama Shuzen. 1970 (1978). Konkō kenshū: Kōhon to kenkyū [Konkō kenshū: Redaction and Interpretation]. Tokyo: Kadokawa Shoten.
Hokama Shuzen. 1972. Omoro-go jisho: Okinawa no kojisho Konkō kenshū [Omoro word book: The Classical Okinawan dictionary, Konkō kenshū]. Tokyo: Kadokawa Shoten.
Hokama Shuzen. 1981. Nihongo no sekai: Okinawa no kotoba. Tokyo: Chūo Kōronsha.
Honan, William H. 1997. “Hunt for Royal Treasure Leads Okinawan to a House in Massachusetts.” New York Times late edition, July 13: Section 1, p. 12.
Ifa Fuyū (ed.). 1925. Kōtei Omoro sōshi [Restored Omoro sōshi]. [@Naha?@]: Kyōdo Kenkyū-kai.
Ikemiya Masaharu. 1979. Omoro sōshi fushi-mei saku-in [Index of song names in the Omoro sōshi]. Naha, Okinawa: Nansei Insatsu Shuppan.
Ikemiya Masaharu (ed.). 1987. Omoro sōshi seikashō [Collection of essays on the Omoro sōshi]. Naha, Okinawa: Hirugi-sha.
Ikemiya Masaharu. 1982. Ryūkyū bungaku-ron no hōhō [Methods in Ryūkyūan literature]. Tokyo: San-ichi Shobō.
Ikemiya Masaharu. 1995. Ryūkyū kogo jiten Konkō kenshū no kenkyū [A study of the Classical Ryūkyūan dictionary, the Konkō kenshū]. Tokyo: Dai-ichi Shobō.
94
Karimata Shigehisa and Shimabukuro Yukiko. Ryūkyū-go no shūshi-kei: Okinawa Jana hōgen to Okinawa Agena hōgen [Sentence-final forms in Ryūkyūan: Okinawa's Jana and Agena dialects]. Ryūkyū daigaku hōbun-gakubu kiyō: Nihon tōyō bungaku ronshū [Bulletin of the College of Law and Letters, Japanese and Western Literature, University of the Ryūkyūs]
Kerr, George H. 2000 [1958]. Okinawa: The History of an Island People. Boston: Tuttle Publishing.
Mamiya, Atsushi. 1983. Omoro sōshi no kakari-musubi ni tsuite [On the kakari-musubi of the Omoro sōshi]. Okinawa Bunka 61:6-18.
Miyara Shinshō. 2000. Uchinaa-guchi kōza: Shuri-kotoba no shikumi [Course on Okinawan: The mechanics of Shuri dialect]. Naha, Okinawa: Okinawa Taimusu-sha.
Miyara Shinshō. 2002. Okinawa chūnan-bu hōgen dōshi no modariti [Modality in Okinawa's south-central dialects]. Gengo kenkyū 122:79-112.
Nakamoto Masachie, Higa Minoru, and Chris Drake. 1984-1993. Omoro kanshō: Ryūkyū koyō no sekai [Omoro appreciation: The world of the ancient songs of the Ryūkyūs]. One-hundred-article series beginning in Gekkan Gengo 13:10.
Nakahara Zenchū and Hokama Shuzen (eds.). 1965. Kōhon Omoro sōshi [Variorum text of the Omoro sōshi]. Tokyo: Kadokawa.
Nakasone, Ronald Y. (ed.). 2002a. Okinawan diaspora. Honolulu: University of Hawai‘i Press.
Nakasone, Ronald Y. 2002b. “An Impossible Impossibility.” In Okinawan Diaspora, Ronald Y. Nakasone, ed. Honolulu: University of Hawai‘i Press.
Nishimura-Valentine, Naomi. 1989. Omoro: The correlation of structure and tune in an ancient Ryukyuan song form. Unpublished M.A. thesis, Ohio State University.
Nishioka Satoshi, and Nakahara Jō. 2000. Okinawa-go no nyūmon [Introduction to Okinawan]. Tokyo: Hakusuisha.
OKDJ. See entry for Okinawa Kogo Daijiten Henshū Iinkai, 1995.
Okimori Takuya. 1989 (2001). Nihon-go shi [History of the Japanese language]. Tokyo: Ōfū.
Okinawa Bungaku Zenshū Henshū Iinkai (eds.). 1991. Okinawa bungaku zenshū, Dai 20-kan: Bungaku-shi [Complete Collection of Okinawan Literature, volume 20: History of Literature]. Tokyo: Kokusho Kankōkai.
Okinawa Kogo Daijiten Henshū Iinkai (eds.). 1995. Okinawa kogo daijiten [Dictionary of Classical Okinawan]. Tokyo: Kadokawa Shoten.
Serafim, Leon A. 1977. Front vocalism and palatalization in the Omoro Sooshi. Unpublished report to the Japanese Ministry of Education.
95
Serafim, Leon A., and Rumiko Shinzato. 2000. Reconstructing the Proto-Japonic kakari-musubi, *-ka…-(a)m-wo. Gengo kenkyū 118:81-118.
Serafim, Leon A., and Rumiko Shinzato. 2005. On the Old Japanese kakari (focus) particle koso: Its origin and structure. Gengo Kenkyū 127: 1-49.
Serafim, Leon A., and Rumiko Shinzato. 2009. Grammaticalization pathways for Japonic nominalizers; A view from the western periphery. In: Yukinori Takubo, Tomohide Kinuhata, Szymon Grzelak, and Kayo Nagai (eds.). Japanese/Korean Linguistics 16: 116-130. Stanford: CSLI.
Serafim, Leon A., and Rumiko Shinzato. Forthcoming. Synchrony and Diachrony of Okinawan Kakari Musubi in Comparative Perspective with Premodern Japanese.
Shinzato, Rumiko. 1998. Kakari-musubi revisited: Its functions and development. In: David J. Silva (ed.). Japanese/Korean Linguistics 8: 203-216. Stanford: CSLI.
Shinzato, Rumiko, and Leon A. Serafim. 2003. Kakari musubi in comparative perspective: Old Japanese ka/ya and Okinawan -ga/-i. In: Patricia Clancy (ed.). Japanese/Korean Linguistics 11, 189-202. Stanford: CSLI.
Takahashi Toshizō. 1991a. Omoro sōshi no dōshi no kenkyū. [Study of verbs in the Omoro sōshi] Tokyo: Musashino Shoin.
Takahashi Toshizō. 1991b. Omoro sōshi no kokugogaku-teki kenkyū [Linguistic study of the Omoro sōshi]. Tokyo: Musashino Shoin.
Tamaki Masami. 1981. Omoro no ka-kei [The song structure of omoro]. Ryūkyū Daigaku Hōbun-gakubu kiyō: Kokubungaku ronshū [Bulletin of the College of Law and Letters, Japanese Literature, University of the Ryūkyūs] 25:2, 66-182.
Vovin, Alexander. 2005. Descriptive and Comparative Grammar of Western Old Japanese. Folkstone, Kent: Global Oriental.
96