The rhythmic effects of rendaku and hiatal elision in Western Old Japanese poetry

John Kupchik Kyoto University / JSPS Postdoctoral Fellow

25 January 2013, NINJAL International Conference on Phonetics and Phonology (ICPP 2013)

This talk will discuss Morpheme-Based Rendaku (Synchronic) in contrast to hiatal elision in a select corpus of Western Old Japanese (WOJ) poetry.

Why compare these? They are the only two post-lexical processes

available in the language to modify the rhythm of a line of verse from the input to the output.

Both involve the synchronic elision of a vowel.

Discovering how they interact with one another may give new insight into the overall phonology and rhythm of the language.

We may find meaningful differences in WOJ in comparison with Eastern Old Japanese.

[V] = elided vowel -N-= morpheme that is the underlying source

of a rendaku output ö = /ə/

Italic font is used to transliterate phonograms, while standard font is used to transliterate logograms

The language of the capital Nara and Asuka in the 7 and 8th centuries CE.

Phonology: 13 consonant phonemes: /p, t, k, mb, nd, ŋg, s, nz, m, n, r, w, j/.

8 vowel phonemes: /i, ɨ, u, e, ə, əj, o, a/ (Miyake 2003).

Syllable structures: V, CV, CVj. Syntax and morphology: SOV, suffixing &

prefixing, modifier-head, mainly Nom-Acc

Man’yōshū Books 5 and 15: Primarily phonographic, mostly tanka (5 lines, 31 syllables).

Similar forms from other Man’yōshū books are given as examples when relevant (there are over 4,000 WOJ poems in the Man’yōshū).

Lines of verse usually consist of sequences of 5 or 7 syllables.

Hypermetrical lines occur in both tanka and chōka, hypometrical usually only in chōka and in the early books.

RENDAKU: When a word that normally has an unvoiced onset occurs in the non-initial position of a phonological word or phrase, its onset becomes voiced.

E.g. asagapo ‘morning glory flower’ (asa ‘morning’, kapo ‘face’).

A small number of words in OJ have a voiced obstruent onset that is unrelated to rendaku: götö- ‘be.like’, bisi-bisi ‘coughing’.

What is the source of the Rendaku output?

TYPE 1: Process-based (reduplicative): e.g. kuni-guni ‘provinces’ no derivable

morpheme from rendaku output.

TYPE 2: Morpheme-based – Diachronic: e.g. opobune ‘great boat’, kugani ‘gold’ morpheme is derivable historically, but not

synchronically (i.e. *opo nö pune and *ko nö kane are unattested in Old Japanese)

TYPE 3: Morpheme-based – Synchronic (MBR):

Rendaku form is only used when needed to conform to the rhythm of a single line of verse.

Examples:

(1) 5.829 (Line 3) 佐久良婆那 sakura-N-bana sakura-GEN-blossom ‘Sakura blossoms’

6 syllable input, 5 syllable output.

Compare: (2) 10.1887 (Lines 4-5) 開有櫻之 sak-er-u sakura-nö 7 syllable input bloom-PROG-ADN sakura-GEN

花乃可見 pana-nö mi-y-ubey-ku 7 syllable input blossom-GEN see-PASS-DEB-INF ‘The sakura blossoms that are blooming ought to be

seen.’

COMMON: Locative -ni, Genitive -nö, Infinitive copula ni,

Adnominal copula nö. UNCOMMON: mi syllable: yama miti ~ yama-di ‘mountain road’ Possessive -ga: kapo-ga pana ~ kapo-bana

‘Japanese bindweed flower’ Vowels that elide in the MBR process: /i/, /ə/, /a/ (rarely)

Occurs synchronically, when needed, between two word forms when their adjacent vowels create a hiatus.

Used to adjust underlying hypermetrical lines to metrically licit lines in the output.

(3) 10.2059 (Line 5)

率滂出 iza kög-i ide-m-u well row-INF go.out-TENT-ADN ‘Well, I should row out.’ 7 syllables in input and output.

COMPARE: (4) 15.3709 (Line 3) 己藝弖奈牟 kög-i [i]de-n-am-u row-INF go.out-PERF-TENT-AND ‘I should row out.’ 6 syllable input, 5 syllable output.

(5) 15.3625 (Line 12) 左宿等布毛能乎 sa-n-u tö [i]p-u monöwo PREF-sleep-FIN QUOT say-ADN although ‘Although you say we should sleep

together…’ 8 syllable input, 7 syllable output.

Compare: (6) 5.800 (Line 13) 由久智布比等波 yuk-u t[ö] ip-u pitö pa go-FIN QUOT say-ADN person TOP ‘Who, as they say, goes.’ 8 syllable input, 7 syllable output. Author is

Yamanoue no Okura (Asuka dialect).

COMPARE: (7) 19.4295 (Line 4) 別等伊倍婆 wakar-u tö ip-e-ba separate-FIN QUOT say-EV-CONJ ‘Although you say we should part…’ 7 syllables in input and output.

114 Poems (#793-906), 104 are tanka. Composed between 724-733 CE. Various authors, but most poems are by

Yamanoue no Okura and some are by Ōtomo no Tabito.

Compiled by Yamanoue no Okura.

The chōka contain some hypometrical lines (either 4 or 6 syllables). Any poems with hypometrical lines were excluded from this study.

The tanka usually consist of lines of 5 or 7 syllables, with no hypometrical lines.

A hypermetrical line without a hiatus is attested in a chōka (poem 897, line 12).

(8) 5.855 (Lines 1 and 2) 麻都良河波 matura-N-gapa 5 syllable output (6 PN-GEN-river syllable input)

可波能世 比可利 kapa-nö se pikar-i 7 syllables river-GEN rapids shine-FIN ‘The river rapids of Matura river are shining.’

Contrast: (9) 5.857 (Lines 2-3) 末都良能加波尓 7 syllables

matura-nö kapa-ni (input/output) PN-GEN river-LOC 和可由都流 5 syllables waka-yu tur-u young-sweetfish angle -ADN ‘Angle young sweetfish at Matura river.’

(10) 5.827 (Line 2) 許奴礼我久利弖 könure-N-gakur-i-te tree.tip-LOC-hide-INF-SUB ‘Hiding in the tree tips.’ 8 syllable input, 7 syllable output.

(11) 5.794 (Line 29) 伊弊社可利伊摩須 ipe-N-zakar-i-imas-u home-LOC-be.far.from-INF-HON-FIN ‘You are far from home.’ 9 syllable input, 8 syllable output.

(11) 5.837 (Line 4)

和何弊能曽能尓

wa-ga [i]pe-nö sönö-ni V2 elision of [ai]>[a]

1.S-POSS house-GEN garden-LOC ‘In the garden of my house.’ 8 syllable input, 7 syllable output. Author: Ōmichi of the Shi[ki] clan, an Accountant.

(12) 5.841 (Lines 2-4) 和企弊能曽能尓

wa-g[a]-ipe-nö sönö-ni V1 elision of [ai]>[i]

1.S-POSS-home-GEN garden-GEN ‘In the garden of my house.’ 8 syllable input, 7 syllable output. Author: Oyu of the Taka clan, Clerk of Tsushima

province.

(13) 13.3272 (Line 11) 己之家尚乎 wa-ga ipe sura-wo 1.S-POSS home RPT-ACC ‘Even my home.’ 7 syllable input and output.

Only one example suggests that MBR is preferred to hiatal elision when both are available. Therefore, the result is inconclusive.

Furthermore, a high frequency collocation that tends to undergo elision takes precedence over MBR.

208 poems (#3578-3785), 201 are tanka. Composed 736-744 CE. Various authors, including ~1/3 by Nakatomi

no Yakamori or his wife Sano no Otogami no Wotome.

The compiler is unknown.

Regular 5-7 rhythm with no hypometric lines. Many hypermetrical lines, all but one

contains a hiatus (the exception is line 5 in poem 3763 which has 8 syllables).

(14) 15.3632 (Line 3) 波麻藝欲伎 pama-N-giyo-ki beach-GEN-be.clean.ADN ‘The beach is clean.’ 6 syllable input, 5 syllable output.

Compare: (15) 7.1239 (Line 5) 濱之浄奚久

pama-nö kiyo-keku beach-GEN be.clean-ADN.NML ‘The beach is clean.’ 7 syllable input and output.

(16) 15.3723 (Line 2) 夜麻治古延牟等 yama-di koye-m-u tö mountain-road cross-TENT-FIN QUOT ‘(My lord who) is going to cross over the

mountain road.’ 8 syllable input, 7 syllable output.

COMPARE: (17) 15.3728 (Line 4) 許能山道波 könö yama miti pa this mountain road TOP ‘This mountain road.’ 7 syllable input and output.

(18) 15.3665 (Line 4) 安左宜理其問理 asa-(N?)-giri-N-gömor-i morning-(GEN?)-fog-LOC-hide-INF ‘Hiding in the morning fog.’ 8~9 syllable input, 7 syllable output.

(19) 15.3655 (Line 2) 安伎豆吉奴良之 aki-N-duk-i-n-urasi autumn-GEN-approach-INF-PERF-SUPP ‘Autumn seems to be coming to a close.’ 8 syllable input, 7 syllable output.

No attested examples.

(20) 15.3584 (Lines 1-5) 和可礼奈波 wakare-n-aba part.INF-PERF-COND

宇良我奈之家武 MBR (ura-ni >ura-N) ura-N-ganasi-kem-u heart-LOC-be.sad-TENT-FIN 安我許呂母 a-ga körömö 1.S-POSS garment

之多尓乎伎麻勢 sita-ni wo ki-[i]mas-e Elision of [ii] below-LOC EPT wear.INF-HON-IMP 多太尓安布麻弖尓 No elision of [ia] tada ni ap-u-made-ni direct COP.INF meet-ATTR-TERM-LOC

‘If we part, I would be sad in my heart. Wear my

garment beneath your clothes until we meet directly.’

WOJ-Book 5 WOJ-Book 15

EOJ dialects

MBR attested? YES YES YES

Hiatal elision attested?

YES

YES YES

Hiatal elision: which vowel elides?

V1~ V2, fluctuation with same input

V1, V2 V2 default (V1 only happens with [ua] and [ia] input)

MBR preferred to hiatal elision?

INCONCLUSIVE ??? YES

Kupchik (2012) argues MBR in EOJ is a more primary grammatical process, while hiatal elision appears to be secondary and possibly optional.

There are many examples of hypermetrical lines with unrealized hiatal elision, but no examples of hypermetrical lines with unrealized MBR (true in both WOJ and EOJ).

Kinoshita, M. (ed.) 2001. Man’yōshū CD-ROM-han [Man’yōshū: The CD-ROM edition]. Tokyo: Hanawa shobō.

Kupchik, J. 2012. Morpheme-based rendaku as a rhythmic stabilizer in Eastern Old Japanese poetry. In Acta Linguistica Asiatica (2) 1: 9-22.

Miyake, M. 2003. Old Japanese: A phonetic reconstruction. London: Routledge.

Vovin, A. 2009. Man’yōshū Book 5. Folkestone: Global Oriental. Vovin, A. 2010. Man’yōshū Book 15. Folkestone: Global Oriental.

Contact: [email protected]