1

Tense, Aspect, and Modality in the Creation of Narrative Structure:

Early Heian Japanese Translations of Sinitic Buddhist Texts

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree

Doctor of Philosophy in the Graduate School of The Ohio State University

By

John Adikes Bundschuh, M.A.

Graduate Program in East Asian Languages and Literatures

The Ohio State University

2021

Dissertation Committee

Charles Quinn, Advisor

Naomi Fukumori

Brian Joseph

2

Copyright by

John Adikes Bundschuh

2021

ii

Abstract

Kundokubun, the linguistic variety that arose from transposing and reciting Sinitic

texts in Japanese, is as old as the act of reading itself in Japan. The religious and political

classes who learned, copied, and propagated Buddhist sutras during the Heian period (794–

1185 CE) used kundokubun when reciting them in Japanese. These sutras are presented as

first-hand accounts narrated by someone who witnessed the Buddha addressing and

conversing with a host of assembled followers. Although most of these sutras originated in

India, they arrived in Japan in their Sinitic renditions. However, in translating these texts

into Japanese, the monks had to read between the lines, both figuratively and literally.

Figuratively, because Chinese does not express the same range of grammatical categories

found in Japanese, such as those we find in the latter’s complex agglutinative predicate

morphology. To effectively communicate in Japanese, the translators had to add tense,

aspect, modality, honorifics, and other markers to predicates and case particles to nouns.

Literally, because in order to preserve their translations in writing they used diacritic

markings between, and occasionally on, the source text’s Chinese characters to denote the

appropriate Japanese morphosyntax and occasionally phonology.

This dissertation examines morphological marking in Early Heian Japanese

renditions of Buddhist texts to explain how tense, aspect, and modality create narrative

frames in kundokubun discourse. It utilizes rubrics and techniques of narrative studies and

iii

linguistic analysis to show how Japanese monks created inspirational narratives in

kundokubun through the act of translation during the early Heian period.

In contrast with the acclaimed vividness of more vernacular wabun tales,

kundokubun has commonly been defined as a more formalized register of Japanese, due to

its abundance of calques, which is a consequence of its Sinitic source texts. Thus, while

the narrative functions of tense, aspect, and modality auxiliaries have been studied in Heian

period wabun texts, there has yet to be a study that relates the findings of such studies to

narrative structure in kundokubun. While informed by an appreciation for the pragmatic

roles of these auxiliaries in secular wabun tales, this study further determines the narrative

functions of six auxiliaries—ki, keri, tu, nu, ari, and tari—in the genre of early Heian

kundokubun texts.

This research sheds new light on a crucial facet of the introduction of Buddhism to

Japan—the rhetorical role of Japanese grammar in shaping and presenting the stories that

would bridge two cultures at a turning point in Japan’s history, not long after the technology

of writing had arrived from China.

iv

Acknowledgements

This dissertation came to fruition thanks to the support that I have had the great

fortune of receiving. First and foremost, my advisor, Charles Quinn, has my deepest thanks.

The guidance he has generously given has led me to expand my understanding of

premodern Japanese grammar from simply a set of linguistic rules to a complex system of

registers, genres, and styles. He provided my first introductions to both kundokubun texts

and the relation between grammar and narrative structure. His keen advice over the years

has improved both my writing and linguistic analyses and I am especially thankful for the

time and support he has given me in the final months of writing this dissertation. I am

deeply indebted to Naomi Fukumori for introducing me to premodern Japanese literature

and the role Sinitic texts played in the intellectual and social discourses of the Heian period.

She also instilled in me an appreciation for close readings and illuminated how

argumentative structures in Buddhist texts came to be used in vernacular stories. I would

also like to express my sincere gratitude to Brian Joseph for providing invaluable historical

linguistics training and kindly allowing me to join Sanskrit courses and reading groups

throughout my time at Ohio State so that I may better access Buddhist source materials. I

could not have asked for a more supportive committee, and I am immensely thankful for

their guidance and mentorship throughout this experience. This dissertation has been

greatly improved due to their perceptive critiques and suggestions, and any remaining flaws

in the text are my own.

v

I have also greatly benefited from the guidance and support of other faculty

members of the Department of East Asian Languages and Literatures—particularly

Mineharu Nakayama, Mari Noda, Etsuyo Yuasa, Shelley Quinn, Richard Torrance,

Marjorie Chan, Zhiguo Xie, and Meow Hui Goh—and of the Department of Linguistics at

Ohio State. Fellow graduate students from both departments enriched my graduate

experience in Ohio as well.

Special thanks also go to Satoshi Kinsui of Osaka University, where I conducted

the bulk of my data collection and analysis, for kindly accepting me as a visiting student

and deepening my understanding of Japanese historical linguistics. My thanks extend to

the other faculty and graduate students who frequented the Japanese literature and

historical linguistics research room of the Osaka University Graduate School of Letters for

their comradery and invitations to classes, conferences, and symposia.

My undergraduate and Masters studies at Tulane University set me on the path that

led me to where I am today. Judith Maxwell introduced me to linguistics and first suggested

graduate studies in the field. She and other Linguistics faculty, including Olanikė Ola Orie,

Nathalie Dajko, and Rebecca Starr, have my thanks for the foundations of linguistic

analysis I acquired at Tulane. Finally, my ability to access Japanese texts developed under

the careful instruction of Michael Wood, who also has my appreciation for first suggesting

I apply to Ohio State when I began to consider pursuing a Ph.D. in Japanese.

I would also like to thank the institutions and foundations that have generously

supported my studies and research. My first year at OSU was funded by The Ohio State

University Fellowship and my second by the U.S. Department of Education and The Ohio

State University East Asian Studies Center Foreign Language and Area Studies

vi

Fellowship. The Blakemore Freeman Foundation graciously funded my year at the Inter-

University Center for Japanese Language Studies in Yokohama. I was able to gather data

at the Japanese National Diet Library the following summer with the support of the Toshiba

International Foundation. My research at Osaka University was funded by a U.S.

Department of Education Fulbright-Hays Doctoral Dissertation Research Abroad

Fellowship. In addition, The Ohio State University Department of East Asian Languages

and Literatures funded two years of graduate teaching associateships, which not only

funded my studies and research, but instilled in me the joy of teaching.

Finally, I would like to thank my family. My parents and sister have closely read

and made excellent suggestions on countless drafts over the years. My children, Keigo and

Ayane, have provided joy and inspiration. Above all, I am grateful to my wife, Yuri, for

moving with me from Japan to Columbus with a newborn baby six years ago, packing up

everything for five subsequent moves since then, and taking on undue burdens over the

past year to allow me to focus on writing. This dissertation would not have been completed

without her love and support.

vii

Vita

2009 ...................................... B.A. Linguistics & Asian Studies, Tulane University

2010 ...................................... M.A. Linguistics, Tulane University

2010–2015 ............................ Assistant Language Teacher, Hikone, Shiga, Japan

2015–2016............................. University Fellow, The Ohio State University

2016–2017............................. Foreign Language and Area Studies (FLAS) Fellow,

The Ohio State University

2017–2018 ............................ Graduate Teaching Associate, The Ohio State University

2018–2019............................. Blakemore Freeman Fellow, Inter-University Center for

Japanese Language Studies

2019–2020............................. Fulbright-Hays Doctoral Dissertation Research Abroad

Fellow, Osaka University Graduate School of Letters

2020–2021............................. Graduate Teaching Associate, The Ohio State University

2021–2022............................. Visiting Assistant Professor, Swarthmore College

Publications

Bundschuh, John. 2021. “Evidence in Heian Buddhist Kundokugo Narration.” In

Matthew Mewhinney, ed. Evidence, Transmission, and Inheritance in Japanese

Literature and Media (Proceedings of the Association for Japanese Literary

Studies Vol. 20, 2019): 143–156.

Bundschuh, John. 2021. “Please Remind Me: A History of the Japanese Particle of

Recollection Kke.” In Patrick Farrell, ed. Proceedings of the Linguistic Society

of America 6(1): 421–432.

Bundschuh, John. 2020. “Rethinking Japan’s Earliest Written Narratives: Early-

Heian Kundokugo Translations of Chinese Buddhist Texts.” In Naoya Unoda,

ed. Anthology of Transborder Cultural Studies 3: 43–63.

viii

Bundschuh, John. 2018. “Japanese Particle i: A Study in Early Middle Japanese.” In

Zhiguo Xie, ed. Buckeye East Asian Linguistics 3: 11–20.

Fields of Study

Major Field: East Asian Languages and Literatures

Area of Specialization: Japanese Linguistics

ix

Table of Contents

Abstract ............................................................................................................................... ii

Acknowledgements ............................................................................................................ iv

Vita .................................................................................................................................... vii

List of Tables ................................................................................................................... xiii

List of Figures ................................................................................................................... xv

List of Abbreviations ....................................................................................................... xvi

Chapter 1. Introduction ....................................................................................................... 1

1.1 Sinitic Buddhist Texts in Early Japan ....................................................................... 2

1.2 Kundokubun: The Language of Rendering Sinitic Texts in Japanese ....................... 3

1.3 An Introduction to Kunten: The Glossings Associated with Kundokubun ............... 6

1.4 Grammatical Features of Kundokubun ...................................................................... 9

1.5 A Brief History of Kundokubun Research in Japan ................................................ 16

Chapter 2. Tense, Aspect, and Modality Auxiliaries in Early Middle Japanese Wabun .. 22

2.1 Auxiliaries nu and tu in Early Middle Japanese Wabun ......................................... 26

2.2 Auxiliaries ari and tari in Early Middle Japanese Wabun ...................................... 27

2.3 Auxiliaries ki and keri in Early Middle Japanese Wabun ....................................... 30

2.4 Summary of TAM Auxiliaries in Early Middle Japanese Wabun ............................ 34

Chapter 3. Framework and Methodologies ....................................................................... 36

3.1 Analytical Framework ............................................................................................. 36

3.2 Methodology ........................................................................................................... 39

3.2.1 Transitivity........................................................................................................ 40

x

3.2.2 Lexical Aspect .................................................................................................. 40

3.2.3 Discourse Function ........................................................................................... 41

3.3 Data Sources ............................................................................................................ 42

Chapter 4. Close Analyses of Selected Early Heian Sutra Narratives .............................. 45

4.1 The Narration of the Saidaiji Temple Golden Light Sutra ...................................... 45

4.1.1 Transitivity in GLS Narration ........................................................................... 50

4.1.2 Lexical Aspect in GLS Narration ...................................................................... 52

4.1.3 Discourse Function in GLS Narration .............................................................. 54

4.1.4 Conclusions Regarding GLS Narration ............................................................ 58

4.2 An Embedded Parable in the Saidaiji Temple Golden Light Sutra ......................... 61

4.3 The Narration of the Tōdaiji Temple Ten Wheels Sutra ......................................... 68

4.3.1 Transitivity in TWS Narration ........................................................................... 77

4.3.2 Lexical Aspect in TWS Narration ..................................................................... 82

4.3.3 Discourse Function in TWS Narration .............................................................. 88

4.3.4 Conclusions Regarding TWS Narration ............................................................ 92

4.4 An Embedded Parable in the Tōdaiji Temple Ten Wheels Sutra ............................ 93

4.5 The Narration of the Immeasurable Meanings Sutra ............................................ 100

4.5.1 Transitivity in IMS Narration ......................................................................... 107

4.5.2 Lexical Aspect in IMS Narration .................................................................... 111

4.5.3 Discourse Function in IMS Narration ............................................................. 114

4.5.4 Conclusions Regarding IMS Narration ........................................................... 116

Chapter 5. The Relation Between TAM Marking and Narrative Structure in Kundokubun

......................................................................................................................................... 121

5.1 Auxiliaries nu and tu in Heian-period Kundokubun Buddhist Texts .................... 122

5.1.1 Transitivity of Sentence-final Perfectives ...................................................... 123

5.1.2 Lexical Aspect of Sentence-final Perfectives ................................................. 126

5.1.3 Discourse Function of Sentence-final Perfectives .......................................... 128

5.1.4 Conclusions Regarding nu and tu in Kundokubun Narration ......................... 130

5.2 Auxiliaries ari and tari in Heian-period Kundokubun Buddhist Texts ................. 132

xi

5.2.1 Transitivity of Sentence-final Statives ........................................................... 133

5.2.2 Lexical Aspect of Sentence-final Statives ...................................................... 134

5.2.3 Discourse Function of Sentence-final Statives ............................................... 136

5.2.4 Conclusions Regarding ari and tari in Kundokubun Narration ...................... 138

5.3 Auxiliaries ki and keri in Heian-period Kundokubun Buddhist Texts .................. 139

5.3.1 Transitivity of Sentence-final Established Fact Modals ................................. 140

5.3.2 Lexical Aspect of Sentence-final Established Fact Modals ........................... 143

5.3.3 Discourse Function of Sentence-final Established Fact Modals .................... 144

5.3.4 Conclusions Regarding ki and keri in Kundokubun Narration ....................... 148

Chapter 6. Kundokubun and Linguistic Variation .......................................................... 152

6.1 Framing in Kundokubun and Wabun Texts ........................................................... 152

6.1.1 Outermost Narration ....................................................................................... 153

6.1.2 Embedded Narration ....................................................................................... 162

6.1.3 Conclusions regarding ki and keri in Wabun and Kundokubun Narration ..... 164

6.2 Variation within Early Heian Kundokubun ........................................................... 166

6.2.1 A Comparison of Transitivity and Sentence-Final ki ..................................... 172

6.2.2 A Comparison of Lexical Aspect and Sentence-Final ki ................................ 173

6.2.3 A Comparison of Discourse Function and Sentence-Final ki......................... 175

6.2.4 Conclusions Regarding Differences Between the Sutras ............................... 182

Chapter 7. Conclusions ................................................................................................... 186

References ....................................................................................................................... 194

Appendix A. Overview of the Tense, Aspect, and Modality Auxiliaries in Early Heian

Kundokubun .................................................................................................................... 207

Endoactive Perfective Auxiliary Nu ............................................................................ 207

Exoactive Perfective Auxiliary Tu .............................................................................. 208

Stative Auxiliary Ari ................................................................................................... 210

Stative Auxiliary Tari.................................................................................................. 211

Externally Established Fact Modal Auxiliary Keri ..................................................... 211

Established Past Fact Modal Auxiliary Ki .................................................................. 213

xii

Weaving Stories with the TAM Auxiliaries .................................................................. 214

Appendix B. The Kundokubun Texts.............................................................................. 216

The Saidaiji Golden Light Sutra ................................................................................. 218

The Tōdaiji Golden Light Sutra .................................................................................. 219

The Ten Wheels Sutra ................................................................................................. 219

The Immeasurable Meanings Sutra ............................................................................ 220

xiii

List of Tables

Table 1 Suzuki’s EMJ Tense-Aspect Paradigm (2009: 163) ............................................. 25

Table 2 TAM Marking Totals in GLS Narration ................................................................ 47

Table 3 The Transitivity of TAM-marked Predicates in GLS Narration ........................... 50

Table 4 The Lexical Aspect of TAM-marked Predicates in GLS Narration ...................... 52

Table 5 The Discourse Function of TAM-marked Predicates in GLS Narration ............... 55

Table 6 Summary of Findings Regarding GLS Narration ................................................ 58

Table 7 TAM Marking Totals in TWS Narration ................................................................ 70

Table 8 The Transitivity of TAM-marked Predicates in TWS Narration ........................... 77

Table 9 The Lexical Aspect of TAM-marked Predicates in TWS Narration ...................... 82

Table 10 The Discourse Function of TAM-marked Predicates in TWS Narration ............. 88

Table 11 Summary of Findings Regarding TWS Narration .............................................. 92

Table 12 Sentence-Final TAM auxiliaries in IMS Narration ............................................ 102

Table 13 The Transitivity of TAM-marked Predicates in IMS Narration ........................ 107

Table 14 The Lexical Aspect of TAM-marked Predicates in IMS Narration ................... 111

Table 15 The Discourse Function of TAM-marked Predicates in IMS Narration ............ 114

Table 16 Summary of Findings Regarding IMS Narration ............................................. 117

Table 17 Transitivity of Sentence-final Perfectives........................................................ 123

Table 18 Lexical Aspect of Sentence-final Perfectives .................................................. 126

Table 19 Discourse Function of Sentence-final Perfectives ........................................... 128

Table 20 Overview of Sentence-final Perfectives .......................................................... 130

Table 21 Transitivity of Sentence-final Statives ............................................................. 133

Table 22 Lexical Aspect of Sentence-final Statives ....................................................... 134

Table 23 Discourse Function of Sentence-final Statives ................................................ 137

Table 24 Overview of Sentence-final Statives ................................................................ 138

Table 25 Transitivity of Sentence-final Established Fact Modals .................................. 140

Table 26 Lexical Aspect of Sentence-final Established Fact Modals............................. 143

Table 27 Discourse Function of Sentence-final Established Fact Modals ..................... 145

Table 28 Overview of Sentence-final Established Fact Modals ..................................... 149

Table 29 Overview of Saidaiji Sentence-final Elements ................................................ 169

Table 30 Overview of Tōdaiji Sentence-final Elements ................................................. 170

Table 31 Correspondences between Saidaiji ki and Transitivity .................................... 172

Table 32 Correspondences between Tōdaiji ki and Transitivity ..................................... 172

Table 33 Correspondences between Saidaiji ki and Lexical Aspect............................... 173

xiv

Table 34 Correspondences between Tōdaiji ki and Lexical Aspect ............................... 174

Table 35 Correspondences between Saidaiji ki and Discourse Function ....................... 175

Table 36 Correspondences between Tōdaiji ki and Discourse Function ........................ 175

Table 37 All sentence-final predicates and their narrative discourse functions of the

Saidaiji tale of Jalavāhana narration ............................................................................... 177

Table 38 All sentence-final predicates and their narrative discourse functions of the

Tōdaiji tale of Jalavāhana narration ................................................................................ 180

Table 39 EMJ TAM Auxiliaries in Buddhist Kundokubun Narration ................................ 188

xv

List of Figures

Figure 1 Heian Period Kunten Glossing Scheme ................................................................ 6

Figure 2 Framing in Buddhist Narratives ......................................................................... 19

xvi

List of Abbreviations

ACC accusative case

ABL ablative case

CAUS causative

COM complementizer

CONCL conclusive form

COND conditional

CONJ conjectural

COP copula

ED embedded dialogue

EN embedded narration

EMJ Early Middle Japanese

FUT future

GEN genitive case

GER gerund

GLS Golden Light Sutra

HON honorific

HUM humble

IMS Immeasurable Meanings Sutra

INST instrumental case

J. Japanese

LOC locative case

N noun

NEG negative

NKBZ Nihon koten bungaku zenshū (Abe et al. 1994)

NMZ nominalizer

NOM nominative case

NTR intransitive

OD outermost dialogue

ON outermost narration

PASS passive

PL plural

Q interrogative particle

S-F sentence-final

xvii

Skt. Sanskrit

TAM tense, aspect, and modality

TOP topic

TR transitive

TWS Ten Wheels Sutra

1

Chapter 1. Introduction

This thesis examines the relationship between the grammatical marking of tense,

aspect, and modality (TAM) and narrative structure in early Heian period translations of

Buddhist texts. The time span most often associated with the Heian period is 794–1185 CE,

with the early part of the period being the first century, through ca. 900 CE.1

This dissertation comprises seven chapters. Chapter one serves as an introduction

to kundokubun, the register of Early Middle Japanese (EMJ) used when rendering Sinitic

texts. Chapter two presents a linguistic overview of the TAM auxiliaries examined in this

dissertation with regards to their use in wabun, or more vernacular registers not associated

with the act of translating Sinitic texts. Chapter three introduces the frameworks for

analysis used in this study before outlining the methodologies used herein. Chapter four

presents close analyses of both outermost narratives and select parables in representative

early Heian Japanese renditions of Buddhist sutras. 2 Chapter five summarizes the

semantics and narrative functions of the TAM auxiliaries in the kundokubun texts. In chapter

six I discuss linguistic variation between wabun and kundokubun narration and between

1 Kobayashi (2012a: xii) uses the reigns of Japanese emperors to define the early part of the Heian period.

He takes the early Heian period to be from the beginning of the Enryaku (延󠄂曆) era through the end of the

Shōtai (昌泰) era, so 782–901 CE.

2 A sermon or discourse between the Buddha and his followers (see Buswell & Lopez 2014: 875).

2

contemporary renditions of the same Sinitic source text. Chapter seven concludes this

dissertation.

1.1 Sinitic Buddhist Texts in Early Japan

In this chapter I introduce the linguistic variety of Early Heian Japanese used to

read Sinitic texts aloud in translation. Although this dissertation focuses on the language

itself, I begin with a brief history of the religious and textual tradition for which it was

employed.

Depending on the chronology, Buddhism began in India in the 6th or 5th century

BCE with the life of the historical Buddha, called either Gautama or Śākyamuni.3 The

Buddhist tradition that first made its way to Japan, Mahāyāna ‘great vehicle,’ emerged in

the first century BCE. The Mahāyāna tradition came to Japan via China and Korea. The first

Buddhist monastery was established in China in 67 CE, during the Han dynasty. Buddhism

was formally introduced into the Korean peninsula three hundred years later, in 372 CE.

Buddhism gained political backing in Japan by Prince Shōtoku (聖徳太子

Shōtoku Taishi) around the turn of the 7th century CE after arriving from the Korean

kingdom of Paekche in 552 CE. During the 7th century the Sinitic renditions of the texts

examined in this dissertation were translated by Xuanzang (玄奘, J. Genjō, 602–664 CE)

3 See “buddhavarṣa” in Buswell & Lopez (2014: 156) for more on Buddhist chronologies and “Śākyamuni”

(Buswell & Lopez 2014: 741) for more on the historical Buddha.

3

and Yijing (義淨, J. Gijō, 635–713 CE), both Chinese Buddhist practitioners who travelled

to India to study and gather Sanskrit texts.4

Due to the official sponsorship of Buddhism by the Japanese state during the Nara

period (ca. 8th century CE), Sinitic renditions by Xuanzang and Yijing were imported and

recited in temples throughout Japan (to the extent that the archipelago was in control of the

Yamato court) within a century of their transportation and import from India. Although

official recitations of sutras in Japan were (and still are to this day) done in a localized

pronunciation of Middle Chinese,5 they were, like most all Sinitic texts in early Japan, also

recited and discussed in the Japanese vernacular.6 I discuss this variety of the Japanese

vernacular for the remainder of this chapter.

1.2 Kundokubun: The Language of Rendering Sinitic Texts in Japanese

Kundokubun (訓読文) is a linguistic register—or type—of early spoken Japanese

adapted for reading texts written in Sinitic script aloud as Japanese or quoting Sinitic

texts in Japanese. I follow Steininger (2017) in using the term “Sinitic script” to refer to

writing using only sinographs, or Chinese characters—kanji in modern Japanese and

hanzi in modern Mandarin Chinese. “Sinitic texts” refer to texts written in Sinitic script,

regardless of their origin.

4 See Buswell & Lopez (2014: xix–xxiv) for a more in-depth and international timeline of Buddhism.

5 Called “Sinitic(-derived) pronunciation” in Steininger (2017).

6 See Lurie (2011) for a detailed study on the vernacular reading of Sinitic texts in the Nara period.

4

The focus of this study is the genre of Japanese used to read and quote Sinitic

Buddhist texts. The term kundokubun ‘vernacular reading register’7 comes from the

expression kanbun kundoku (漢文訓読 ‘Sinitic writing, vernacular reading’). Using this

term invites the apt comparison of kundokubun with wabun (和文 ‘(native) Japanese

register’)8 later in the dissertation. Kundokubun also contrasts with the terms for

‘vernacular-reading vocabulary,’ kundokugo (訓読語) or kundoku goi (訓読語彙). This

distinction is made explicitly in Tsukishima Hiroshi’s (1982) monograph Heian jidaigo

shinron (平安時代語新論 ‘A New Treatise on the Language of the Heian Period’),

which eloquently advocates the inclusion of vernacular gloss materials (訓点資料

kunten shiryō) in our understanding of the Japanese language of the Heian period.

However, other scholars, such as Kobayashi (2011) and Matsumoto (2017), use the term

kundokugo when referring to both the type of language (i.e., register) and the vocabulary

therein.9 Furthermore, many inquiries into kundokubun examine the glossing system

used to add Japanese morphemes to Sinitic texts. These glosses are called kunten (訓点

7 I follow Biber & Conrad (2019) in distinguishing register, genre, and style. The register perspective refers

to the aggregate of linguistic features that characterize a variety, or type, of language; the genre perspective

attends to structural or textual conventions, such as the use of repetitive hymns in a sutra; and the style

perspective focuses on an individual’s use of a linguistic register or its use during a specific time period.

8 Depending on the scholar or school of scholarship, this term is used to refer either to (a) the register

generally used in Classical Japanese prose texts that use fewer Chinese borrowings, or to (b) the written genre

that uses more of the hiragana (平仮名) and katakana (片仮名) syllabaries rather than kanji (漢字). I refer

to the former of these, i.e., (a), when using the term wabun in this dissertation.

9 I previously preferred the term kundokugo (e.g., Bundschuh 2020; Bundschuh 2021), until learning from

Kinsui Satoshi (p.c.) that, when discussing narrative (and, thus, register), kundokubun would be more

accurate. (Kanbun) Kundokutai ((漢文)訓読体) is another term used for this register, but it is most often

used to demarcate a written translation genre rather than a spoken linguistic register (cf. Ueda 2008).

5

‘vernacular glosses’) and the written language is thus often called kuntengo (訓点語

‘vernacular gloss language’). Many studies of Sinitic texts rendered in Japanese focus on

these markings, and use the term kuntengo throughout (e.g., Otsubo 1961). Kundokubun,

kundokugo, and kuntengo are thus terms that different scholars may use to refer to the

same variety of Japanese, with the former two referring to spoken language and the latter,

written. However, to other scholars, the latter two—kundokugo and kuntengo—also

suggest a focus on lexical items over morphosyntax and pragmatics, which is why they

are not used in this dissertation.10

The kundokubun examined in this dissertation could also be called Buddhist

Hybrid Japanese. The Middle Chinese texts from which it derives have been referred to

as Buddhist Hybrid Chinese (Shi 2014) and Buddhist Hybrid Sinitic (Mair 1994), many

of which are translations of Buddhist Hybrid Sanskrit (Edgerton 1936) texts. As noted

above, kundokubun gains its moniker by the method of reading written source material,

which suits the Japanese of Buddhism, Confucianism, and translations of literary

Chinese. “Buddhist Hybrid Japanese” is a practical term to use when discussing the

variety of Japanese employed in Buddhist discourse due to it being strongly influenced

by Buddhist Hybrid Chinese vocabulary and Buddhist Hybrid Sanskrit discourse

strategies. However, when examining the data from a comparative lens, this thesis

focuses on the kundokubun-wabun dichotomy, so the term kundokubun is be used instead

of Buddhist Hybrid Japanese for its duration.

10 See Whitman et al. (2010) for more on English translations of technical terminology regarding

kundokubun and its source texts.

6

1.3 An Introduction to Kunten: The Glossings Associated with Kundokubun11

The following figure represents the most common scheme of glossing Sinitic texts

in Heian-period Japan. However, this only represents one of over two hundred glossing

schemes used in Japan during the Heian period (Nakada 1979, Alberizzi 2014).

Figure 1 Heian Period Kunten Glossing Scheme

(Adapted from Alberizzi 2014: 6)

11 This section is largely adapted from a section of Bundschuh (2020).

7

Although this figure displays a system of dots around the kanji, there are additional

readings indicated by slanted, vertical, and bent lines in the same locations. These markings

are often together called wokototen (ヲコト点 ), after the glosses indicated with the

uppermost two dots on the right side of the kanji, which represent the accusative marker

wo and the nominalizer koto.12 While the term wokototen refers specifically to this system

of glossing, kunten includes both these and other reading marks, such as furigana (振り仮

名 ‘reading glosses/ruby’) that translators wrote around sinographs in the source text to,

for example, indicate the phonology of a borrowed Middle Chinese morpheme or a

Japanese semantic equivalent to use when reading the text aloud in kundokubun.

Beginning in the Heian period, kunten-marked texts also employed precursors to

modern furigana pronunciation glosses and okurigana (送り仮名 ), the kana (仮名 ,

‘Japanese phonological orthography’) that indicated the derivational and/or inflectional

morphology that followed a word’s base, written as kanji. Bjark Frellesvig’s (2010) A

History of the Japanese Language presents an accurate and succinct account of how Sinitic

texts were read as Japanese following kundokubun conventions. Frellesvig presents the

following example describing the reading method and his romanization scheme as follows:

12 Some scholars refer to these glosses as okototen due to the fact that in today’s Japanese the initial /w/ is

no longer pronounced (e.g., Frellesvig 2010). I follow Whitman et al. (2010) in retaining the initial /w/.

8

We have noted in CAPITALS readings not indicated in the kunten text [...]

Everything in lower case [...] is directly represented in the kunten text. We use

boldface for glosses giving grammatical information not represented in the

Chinese text, and bold italics for such information given by okoto-ten, and we

underline words which have been transposed, generally by a move to the right as

instructed by numbers, but note also that the order of 欲 and 上 has been

reversed without any overt instruction. Knowing and following the

conventions used in this tradition of kanbun-kundoku allows a reader

to render the Chinese text into Japanese [. . .]

後 時 佛

NOTI no TOKI ni POTOke,

after COP.ADN time DAT Buddha

天 上 欲

AME ni NOBOra-mu to possu

heaven DAT ascend-CONJ.CONCL COMP want.CONCL

‘Later, the Buddha wanted to ascend to heaven.’

(Frellesvig 2010: 261–262. Kunten-marked text adapted from p. 261)

The wokototen gloss representing ni, which is often identified more generally as locative

but which Frellesvig more specifically identifies as dative, marks the upper left corner of

toki and ame. This represents the glossing scheme quoted above from Alberizzi 2014.

Furthermore, Frellesvig’s romanization of today’s ha-gyō (ハ行, ‘mora beginning with

/h/’) as pa-gyō represents the general consensus among historical linguists regarding the

phonology of early Japanese (see 2010: 204–205). Due to this dissertation’s focus on the

morphology and narrative structure of the linguistic register kundokubun rather than its

associated orthography, the only bold and underlined parts in romanizations for its

duration are predicates in which we find the tense, aspect, and modality morphemes

under discussion.

9

1.4 Grammatical Features of Kundokubun

Now that kundokubun and its associated orthography have been introduced, we now

turn to the grammatical features of the register. As noted, kundokubun is a variety of early

Japanese used to read Sinitic texts aloud. Regarding the difference between kundokubun

and wabun, Guest (2013: 25–26) writes:

… the formation of semi-standardized tropes of equivalence (or calques) between

written characters and Japanese words helped to shape kundoku renderings as a

distinctive style [...] that was not expected to conform exactly to any other style of

classical Japanese.

Steininger (2017: 143) describes the process as such:

Rather than providing a naturalistic translation, kundoku cleaves closely to the

original text. The sentence produced does not aim for an idiomatic construction

according to colloquial speech patterns, but represents the meaning through a

limited, formalized Japanese register while maintaining the structure of the original

as much as possible.

Both Guest (2013) and Steininger (2017) focus their discussions on the practice of

‘reading’ Sinitic texts aloud ‘as Japanese,’ i.e., kundoku, rather than the register of Japanese

employed in the act, kundokubun, which is the primary object of study in this dissertation.

The historical periods of the Japanese language during which kundokubun was a

prominent variety of Japanese were Old Japanese (8th century CE), Early Middle Japanese

(9th–12th century CE), Late Middle Japanese (13th–16th century CE), and Early Modern

Japanese (17th–late 19th century CE). This dissertation focuses on the kundokubun spoken

when reading Sinitic texts as Japanese during the first century of the Heian period. This

was the beginning of vernacular glossing in Japan and, thus, there was no standardization.

In other words, there was no established tradition of correspondences for vernacularization.

10

However, by mid-Heian, or the 10th century CE, more deference was shown to previous

kundoku strategies, which led to greater homogeneity (Kobayashi 2011: 192). Although

these early examples of kundokubun maintain some grammatical structures associated with

Old Japanese, they are also the earliest examples of what is thought to resemble the spoken

language of the political elites during this time, Early Middle Japanese (EMJ).

Like the Japanese of today, EMJ is morphologically primarily an agglutinating

language with inflecting predicates and nouns with postpositional particles denoting case

marking. Furthermore, it generally follows a Subject-Object-Verb word order, with the

sentence-final position both heading the predicate and interacting with the discourse.

Traditional grammars of EMJ describe the inflectional morphology of predicators

(verbs, adjectives, copulas) with the following paradigm: mizenkei (未然形) ‘irrealis

form,’ ren’yōkei (連用形 ) ‘continuative form, nonfinite form,’ shūshikei (終止形 )

‘sentence-final form, finite form,’ rentaikei (連体形) ‘adnominal form, attributive form,’

izenkei (已然形) ‘realis form,’ and meireikei (命令形) ‘command form.’ Below is an

example of the paradigm for the quadrigrade verb, kiku ‘hear, listen,’ in which I present

the standard paradigm’s six forms with my functional characteristics added.13

13 I recognize that the traditional paradigm described here is often viewed as anachronistic and too beholden

to Japanese orthography. The example verb given here, which I call kiku, is more accurately described using

its stem kik- (see Frellesvig 2010: 228 for the inflected verb forms of EMJ). I have elected to use the traditional

paradigm in this dissertation to more closely match existing literature on kundokubun written in Japanese.

11

Quadrigrade Verb kiku (聞く14 ‘hear’) EMJ Paradigm

Japanese term Translation Example Functional label

mizenkei ‘unrealized form’ kika-ba ‘irrealis stem’

‘if (they) hear’

ren’yōkei ‘continuative form’ kiki ‘participle’

‘hearing; (having heard)’

shūshikei ‘conclusive form’ kiku same

‘(they) hear’

rentaikei ‘adnominal form’ kiku pito same

‘person who hears’

izenkei ‘realized form’ kike ‘realis form’

‘given (their) hearing’

meireikei ‘command form’ kike ‘imperative’

‘Hear/listen!’

Kiku is a representative quadrigrade verb because its stem has four potential phonological

shapes in its paradigm: (1) kika, (2) kiki, (3) kiku, and (4) kike, as we see above. The other

verb types in Early Middle Japanese kundokubun are upper bigrade, lower bigrade, upper

monograde, s-irregular, r-irregular, k-irregular, and n-irregular. Below I provide a

paradigm of each verb class found in kundokubun narratives.15

14 There are numerous potential orthographical representations of this and other verbs described below. I

have selected a commonly seen kanji to represent each in early Heian kundokubun.

15 There is also a lower monograde verb class in EMJ that consists of only the verb keru (蹴る ‘kick’), but

this verb class is not found in kundokubun. K-irregular and n-irregular verbs are rare in kundokubun texts

(Otsubo 1981: 153) and not found in the narratives examined in this dissertation.

12

Upper Bigrade Verb otu (落つ ‘fall’) EMJ Paradigm

Functional label Example Translation

irrealis stem oti-ba ‘if (they) fall’

participle oti ‘falling; (having fell)’

conclusive form otu ‘(they) fall’

adnominal form oturu pito ‘person who falls’

realis form oture ‘given (their) falling’

imperative otiyo ‘Fall!’

Lower Bigrade Verb u (得 ‘gain, acquire’) EMJ Paradigm

Functional label Example Translation

irrealis stem e-ba ‘if (they) gain (it)’

participle e ‘gaining (it); (having gained it)’

conclusive form u ‘(they) gain (it)’

adnominal form uru pito ‘person who gains (it)’

realis form ure ‘given (their) gaining (it)’

imperative uyo ‘Gain (it)!’

Upper Monograde Verb miru (見る ‘see’) EMJ Paradigm

Functional label Example Translation

irrealis stem mi-ba ‘if (they) see’

participle mi ‘seeing; (having seen)’

conclusive form miru ‘(they) see’

adnominal form miru pito ‘person who sees’

realis form mire ‘given (their) seeing’

imperative miyo ‘See!’

13

S-irregular Verb su (為 ‘do’) EMJ Paradigm

Functional label Example Translation

irrealis stem se-ba ‘if (they) do (it)’

participle si ‘doing (it); (having done it)’

conclusive form su ‘(they) do (it)’

adnominal form suru pito ‘person who does (it)’

realis form sure ‘given (their) doing (it)’

imperative seyo ‘Do (it)!’

R-irregular Verb ari (有り ‘be/have’) EMJ Paradigm

Functional label Example Translation

irrealis stem ara-ba ‘if (they) are there/have (it)’

participle ari ‘being (there)/having (it);

(having been there it/had it)’

conclusive form ari ‘(they) are (there)/have (it)’

adnominal form aru pito ‘person who is (there)/has (it)’

realis form are ‘given (their) being (there)/

having (it)’

imperative are ‘Be (there)!/Have (it)!’

Upper bigrade verbs are so named for the two stem forms in the paradigm, which end in

(1) i and (2) u. Lower bigrade stems end in either (1) e or (2) u. The stem of upper

monograde verbs always ends in i. The distinction between “upper” and “lower” is due to

i coming before e in the standard ordering of the Japanese syllabary. The inflectional forms

of adjective and copular predicators follow the same paradigmatic structure. The participial

form is of particular importance to the present study, as this form is the base that tense and

aspect morphemes governed in EMJ. In English scholarship it is also referred to as the

“infinitive” (e.g., Frellesvig 2010), and was commonly repurposed as an ‘-ing’ noun.

14

Although one may call this a “gerund” in the traditional Western sense, the gerund in EMJ

is constructed by adding the suffix te to the participle—the gerund of kiku ‘hear’ was kiki

-te (hear-GER ‘having heard’). In EMJ kundokubun both the participle and gerund head

subordinate clauses, with the former more strongly connected to the subsequent clause.16

As described above, the morphemes that are the focus of this study are all

auxiliaries that govern the participle.17 On this point it bears noting that in traditional

Classical Japanese grammars, the auxiliary ari is taken to govern the realis form, rather

than the participle, due to opacity in its form. This ad-hoc reanalysis continues to be taught,

but it is historically inaccurate. To illustrate, when the participle of the verb kiku ‘hear,

listen’ is governed by ari, they combine as such:

kiki ‘hearing’ + ari ‘be’ > kikeri ‘has heard’

The participle kiki is concealed when it is governed by ari in its resulting form, kikeri, due

to crasis, or sandhi. When crasis reduced the four syllables of our example きき＋あり

(participle kiki + AUX ari) to the three of きけり(kikeri), the discrete morpheme boundary

was lost. Subsequent reanalysis relocated the morpheme boundary by construing the third

syllable as the auxiliary “ri”, and the first two syllables as the verb’s inflected form, kike—

which by EMJ was segmentally identical with the realis form.18

16 In other words, we more often find participial clauses modifying a gerund than vice-versa.

17 This dissertation follows Martin (1987) and Frellesvig (2010) in using the term “auxiliary” when

discussing inflecting suffixes, which mostly overlap with the term jodōshi (助動詞) in Japanese.

18 Onset glides before front vowels in Old Japanese were lost by EMJ. Historically, the combination of kiki

and auxiliary ari was kikyeri. The realis form and imperative were distinguished in Old Japanese by an onset

glide as well (kike and kikye, respectively), which has prompted some school grammarians in Japan to teach

auxiliary “ri” governed the imperative.

15

kikeri < kike-ri

Thus, even today high school students in Japan learning Classical Japanese grammar are

taught that an auxiliary ri governs the realis stem (although it is the only auxiliary that

seems to do so). However, ari is clearly an auxiliary in the TAM system, which governs the

participle of the predicate paradigm. While many analyses of EMJ TAM markers

acknowledge this auxiliary is a form of ari, they follow the traditional nomenclature and

label it ri in their analysis (e.g., Ogawa 1983, Suzuki 2009).

Although kundokubun shares grammatical features with its contemporary varieties

of Japanese, such as agglutinative morphology and SOV sentence structure, it utilizes

predominantly Chinese-derived vocabulary. This is primarily due to it being the result of

translating texts full of philosophical and religious concepts new to Japan.19

Finally, the romanization used when citing kundokubun throughout this dissertation

follows the general consensus regarding early Heian phonology (see Frellesvig 2010: 176

for an overview). For example, while the character つ today is read tsu, at the time it was

tu, and the character は , although ha today, was pa. Regarding how to read Sinitic

characters, the word 清淨 ‘pure,’ although read shōjō in Buddhist contexts today, was

read syauzyau during the early Heian period. I use the Shin kangorin (新漢語林) and the

19 Although this dissertation focuses on Buddhist sutras, the Confucian Analects and other Sinitic texts

related to governance were also recited in kundokubun during the Heian period (see Steininger 2017).

However, Confucian scholars of the Heian period and onwards generally minimized their use of TAM

auxiliaries in their Japanese renditions, leading to a more marked variant of kundokubun than that seen in

Buddhist texts. For example, see Otsubo (1977) for a comparison for the use of ki and keri in Confucian and

Buddhist kundokubun texts.

16

6th edition of the Kōjien (広辞苑) dictionaries to determine the historical phonology and

produce romanizations throughout this dissertation.

1.5 A Brief History of Kundokubun Research in Japan20

Since the research of Oya Tōru (大矢透) and Yoshizawa Yoshinori (吉澤義則),

scholars have recognized the importance of Sinitic texts marked with Japanese reading

glosses in understanding the history of the Japanese language. Oya (1909) was the first

scholar to examine kunten texts as a source of data for Japanese language research in his

search for the origins and history of the kana syllabaries. Yoshizawa (1919) is credited as

being the first to analyze wokototen glosses for their grammatical significance in

rendering Sinitic texts in Japanese. Oya and Yoshizawa subsequently built upon each

other’s research to establish the documentation and analysis of kundokubun as a sub-

discipline of Japanese linguistics (国語学 kokugogaku) by producing numerous works

and inspiring generations of scholars, including Kasuga Masaji (春日政治), Nakada

Norio (中田祝夫), and Otsubo Heiji (大坪併治), to take up the lexical, grammatical,

phonological, and orthographical analysis of kunten texts (Tsukishima 2001: 4–5).

Kasuga’s (1985a) work documenting a complete Golden Light Sutra glossed at

Saidaiji temple in the early Heian period remains the standard for kundokubun linguistic

analysis and the rendering of kunten-marked Sinitic texts in kakikudashibun (書き下し

文), or a transliteration of the glossed Sinitic script into Japanese script. Kasuga also is

20 This section is largely adapted from Bundschuh (2020).

17

the first scholar to note the use of past established fact auxiliary ki primarily at the

beginning and conclusion of kundokubun sutras.

Nakada (1979) expanded the diachronic scope of kundokubun research, which had

been focused on the early Heian period, by examining texts from throughout the Heian

and Kamakura (1185–1333 CE) periods. By comparing texts across genres and

generations, he found patterns in wokototen marking and determined that the wokototen

glossing schema displayed in section 1.2 above was both one of the earliest and most

common patterns of the grammatical glossing of Sinitic texts in Japan.

Otsubo is recognized for his in-depth analyses of Heian period kundokubun

grammar, and published, to my knowledge, the first article on the relation between

grammar and narrative structure in kundokubun texts. He notes that throughout the Heian

period, Buddhist kundokubun narration begins a story with ki marking, but shifts to bare

verbs and aspectual forms such as ari and tari before concluding narratives with ki. On

the other hand, he found that secular, or Confucian, Sinitic texts lack grammatical past

tense marking via ki in their kundokubun renditions even when narrating ‘tales of long

ago’ (Otsubo 1977). While Otsubo recognized the use of what we would call the

historical present (a well-known narrative technique discussed below), he does not

attempt to explain its use throughout Buddhist kundokubun texts.

Recently Kobayashi Yoshinori (小林芳規), from 2011 to 2019, published a nine-

volume series titled Heian jidai no bussho ni motozuku kanbun kundokushi no kenkyū (平

安時代の仏書に基づく漢文訓読史の研究 ‘Research on the history of kanbun kundoku

based on Heian period Buddhist texts’). It contains rich synchronic analyses of early,

18

mid, and late Heian kundokubun (Kobayashi 2012a, 2012b, and 2013, respectively) and

discusses the linguistic shifts found in the register throughout the period (specifically

Kobayashi 2017). Kobayashi demonstrates that, by the end of the Heian period, most

Buddhist translators of Sinitic texts could no longer internalize the linguistic nuances of

texts as a whole and thus came to focus on lexeme-by-lexeme translation without

attempting to produce natural Japanese discourse or even sentence structures (Kobayashi

2012a: 326–340).

Furthermore, Kobayashi finds the following specific changes between an early

and late Heian rendition of the Collected Accounts of the Diamond Sutra (J. 金剛般若經

集驗記 Kongō hannyagyō shūgenki): (1) Reduction of sentence complexity—Multi-

clause sentences came to be rendered as multiple single-clause sentences; (2) Reduction

of TAM marking and other morphemes not found in the Sinitic source test; (3) Sinographs

related to case originally rendered differently depending on the context came to be

translated the same way regardless of the surrounding predicates, which led to

grammatical errors in the late Heian rendition; (4) An increase in Sinitic phonological

readings (J. 字音 jion) and codified Japanese renderings (J. 和訓 wakun) regardless of

context; (5) A decrease and codification of syntactic agreement between adverbs and

sentence-final predicates (J. 呼應語 koōgo);21 (6) The shift in tendency from rendering

21 For example, in the codified kundokubun used in Japan today, there is a syntactic rule that sentences that

begin with masani (当に ‘truly’) must end with the modal auxiliary beshi (‘should/must’). In the early Heian

texts examined in this dissertation, we also find the conditional auxiliary mu governing predicates with

masani in the verb phrase.

19

couplets (J. 対句 taiku) as two separate sentences into a single one (Kobayashi 2012a:

339–340).

Matsumoto Mitsutaka (松本光隆) is the first scholar to examine the shifting

narrative perspectives of these texts (Matsumoto 2011a). He proposes the following

multi-layered formation of Buddhist narratives:

Figure 2 Framing in Buddhist Narratives

(Adapted from Matsumoto 2011a: 215)

The outermost layer contains only the first line, “Thus I have heard.”22 Of the entire

sutra, this is the only line in which the narrator describes their own actions in the first

person. Every subsequent line is a depiction of what he heard.23 Counting inward from

22 Written 如是我聞 in Sinitic script, a translation of Sanskrit Evaṃ mayā śrūtaṃ ‘Thus was heard by me,’

and often rendered as something along the lines of ko no gotoki ware kikitamapeki in EMJ.

23 Katō Kōji (加藤浩司) cites a Lotus Sutra commentary, written ca. 700 CE in Japan and credited to Shōtoku

Taishi, in arguing that the prevailing view regarding the narrator of the sutra is that it was Ānanda (J. 阿難

Anan), who was present and thus witnessed the dialogues between the Buddha and his followers depicted

throughout the text. Katō thus argues that ki is a witnessed, or first-hand evidential, past marker in Japanese

Buddhist texts (Katō 1998: 13).

20

the outermost frame, the second layer consists of the narrator describing the scene in

which the Buddha has conversations regarding the dharma24 with the members of a

gathered assembly. I refer to these first two layers as the “outermost narration”

throughout this study, as they are where the narrator presents the events in their own

voice. The conversation itself is then the third layer in. However, often in these didactic

conversations, the Buddha becomes the narrator of parables. This narration by the

Buddha comprises the 4th layer. Finally, the characters in the tales themselves talk of past

events in their conversations, and these form the innermost layer in Matsumoto’s

depiction of the narrative frames that form Buddhist kundokubun texts. Furthermore, he

posits that the translators were aware of this layered narration and used different

linguistic styles in their rendering of each (Matsumoto 2011a: 215).

Matsumoto (2011b) also compares the differences in grammatical marking in

verse （J. 偈 ge, Skt. gāthā）, quotations outside of verse, and narration outside of verse

in the 25th chapter of a late-Heian (ca. late 11th century CE) rendition of the Lotus Sutra.25

He inspects both auxiliaries and flectives26, and finds verse the most varied in suffixation,

followed by quotations and finally by narration. In Matsumoto (2017) he recognizes that

there is an abundance of the past established fact auxiliary ki (more properly introduced

in section 2.3 below) in the first chapter of the Lotus Sutra narration in comparison to

24 The term dharma generally means ‘Buddhist law’ or ‘Buddhist teachings.’ See Buswell & Lopez (2014:

242–243) for more on its polysemy.

25 Glossed at the temple complex on Mt. Kōya and still stored in in its Ryūkōin treasure house. See Otsubo

(1968) for a full kundokubun rendition of the text.

26 Suffixes that do not inflect (Frellesvig 2010: 51).

21

subsequent chapters’ narration and embedded quotations throughout the sutra. He posits

that this is due to ki grounding the discourse in the past to establish a base from which the

narrative can develop (something this study demonstrates for earlier kundokubun texts).

However, he finds that the narrative-grammatical analysis contradicts his narrative-

dialogue framing methodology—in that taking the outermost narrative to be discourse

that is evidentially grounded to some extent undermines his argument that dialogue is

closer to spoken language—and thus only devotes a few pages to the idea in his text

(Matsumoto 2017: 486–491).

In sum, Matsumoto (2011a, 2011b, 2017) recognizes the potential for a narrative

analysis of kundokubun texts, but primarily focuses on the differences in the grammar

and lexicon between narration and quotations. Building upon Otsubo (1977) and

Matsumoto’s (2017) narrative analyses, the present study examines how sentence-level

grammatical choices can provide a lens through which the broader narrative style in

Buddhist kundokubun narrative comes into focus. This seems a promising approach by

which to reveal the nature of the narrative stance of both the primary narrator and the

characters sharing parables within each text’s story world.

Next, I introduce the tense, aspect, and modality auxiliaries analyzed in this study

and how they are used Early Middle Japanese wabun ‘(native) Japanese register,’ which,

in comparison to kundokubun, has a rich tradition of linguistic and literary analysis.

22

Chapter 2. Tense, Aspect, and Modality Auxiliaries in Early Middle Japanese Wabun

This chapter provides an overview of the tense, aspect, and modal (TAM) auxiliaries

examined in this dissertation—nu, tu, ari, tari, ki, and keri—in Early Middle Japanese

(EMJ) wabun, the register more strongly associated with “native” Japanese poetry and

prose. Examining these auxiliaries in wabun allows us to form an understanding of how

they were used in less-marked contexts without the mediation of translation associated with

kundokubun.

To begin, Yamaguchi (1986) finds the EMJ auxiliaries ki, keri, tu, and nu to fulfill a

continuum of tense, aspectual, and modal roles depending on the context of their use.27

These four auxiliaries, in addition to ari and tari, all govern the participle, or ‘continuative

form,’ of a predicate. Thus, most research on the grammatical expression of time in EMJ

focuses on these six auxiliaries—nu, tu, ari, tari, ki, and keri (e.g., Ogawa 1983, Suzuki

1992, Sandness 1999, Suzuki 2009).28 The paradigms of these auxiliaries are as follows:

27 Contrasting with Bybee et al. (1994), Yamaguchi (1986) argues that these TAM markers are on a

grammaticalization path of modality to aspect to tense. While this is intriguing, he supplies no evidence, and

thus this dissertation remains agnostic regarding his views of the auxiliaries’ historical development.

28 There are other auxiliaries associated with tense and modality in EMJ, such as the conjectural mu and the

subjunctive masi (see Frellesvig 2010: 64, 78 for their semantics). They are not included in this study’s

analysis because they govern the irrealis stem and are not used in matrix clauses in early Heian kundokubun

narration.

23

nu’s Paradigm

irrealis stem -na-

participle -ni

conclusive form -nu

adnominal form -nuru

realis form -nure

imperative -ne29

tu’s Paradigm

irrealis stem -te-

participle -te

conclusive form -tu

adnominal form -turu

realis form -ture

imperative -teyo

ari’s Paradigm

irrealis stem -ara-

participle -ari

conclusive form -ari

adnominal form -aru

realis form -are

imperative -are

29 Frellesvig (2010: 61) does not include an imperative form in nu’s paradigm, but Nakada (1958: 11)

assumes two in his kakikudashi (Japanese script transliteration), such as in the following line spoken by

followers to the Buddha.

唯リ願フ、為に說（き）たまひ（ね）。

pitori negapu tame ni toki-tamapi-ne

one wish purpose ABL expound-HON-NU

‘We make just one wish, for that purpose please expound (the sutra).’

24

tari’s Paradigm

irrealis stem -tara-

participle -tari

conclusive form -tari

adnominal form -taru

realis form -tare

imperative -tare

ki’s Paradigm

irrealis stem -se-

participle N/A

conclusive form -ki

adnominal form -si

realis form -sika

imperative N/A

keri’s Paradigm

irrealis stem N/A

participle N/A

conclusive form -keri

adnominal form -keru

realis form -kere

imperative N/A

Following Suzuki (2009), I analyze the TAM auxiliaries as pairs: nu and tu, ari and tari,

and ki and keri. Suzuki’s morphological analysis of the grammatical expression of time in

Early Middle Japanese is an authoritative account on the subject. Following a word-and-

paradigm approach, rather than using the terms “morpheme” or “auxiliary,” Suzuki refers

to forms that share grammatical meaning. For example, while this dissertation refers to

kiki-keri as the verb kiku in its participial form governed by the auxiliary keri, Suzuki refers

25

to this as the keri-form of the verb kiku.30 Below is Suzuki’s account of the tense-aspect

paradigm (his characterization) of Early Middle Japanese.

Table 1 Suzuki’s EMJ Tense-Aspect Paradigm (2009: 163)

Tense⇩ Aspect⇒ Perfective Imperfective Perfect

Non-past tu, nu bare tari, ari

Past teki, niki ki tariki, ariki

The forms in the past perfective and past perfect are composed of the participial forms of

tu, nu, tari, and ari, respectively, governed by ki. Suzuki acknowledges that (since it is

grounded in a present) perfect should be distinguished from aspect, but says he follows

recent trends in including it as such in the paradigm. Furthermore, he explicitly notes that

while keri does not fit cleanly into the tense-aspect system, its meaning is related to the

expression of time, and thus belongs in the analysis.

It must be noted that Suzuki is concerned primarily with the analysis of time

expressions that are realized morphologically in the verb phrase, and he limits his data to

avoid syntactic and pragmatic factors that affect meaning. He clearly states that his data

are drawn from dialogue rather than narration, and, furthermore, he selects only verbs in

their declarative, conclusive form. The analysis within this dissertation regarding the TAM

auxiliaries’ meaning in kundokubun primarily looks to Suzuki (2009) and Quinn (1987) to

form contrasts and comparisons with their meanings in contemporary Heian vernacular

30 Although I agree that the word-and-paradigm has its theoretical advantages, using the terminology of

concatenative verb morphology is of great utility when analyzing both kundokubun and wabun Early Middle

Japanese, which display relatively clear-cut agglutinative predicate morphology.

26

literature but also considers the analyses of Stinchecum (1985), Ogawa (1983), and

Kumakura (1980) when analyzing the relation between the TAM markers and narrative

structure.

2.1 Auxiliaries nu and tu in Early Middle Japanese Wabun

Auxiliaries nu and tu likely developed from phonological reductions with aphesis

(loss of an initial vowel) of the verbs inu (往ぬ ‘leave’) and utu (棄つ ‘discard’),

respectively (Oda 2015: 126).

Suzuki (2009) contrasts tu and nu forms with unmarked verbs (“bare forms”) as

marking perfective aspect (完成相 kanseisō). Auxiliary tu is an exoactive perfective

(Quinn 1987), governing predicates that show greater transitivity on the transitivity scale

established by Hopper and Thompson (1980). Auxiliary nu, on the other hand, is an

endoactive perfective (Quinn 1987) that governs predicates with lower transitivity, i.e.,

unaccusatives and unergatives.31 Suzuki makes a similar distinction, while also noting that

tu is much more likely to be used to mark events that occurred before the utterance time,

in contrast with nu that shows no such preference. Frellesvig, also, refers to both as

perfectives that differ in their relation to the predicate’s transitivity. He describes their two

shared functions as their perfective aspect and the assertion, or affirmation, of the predicate

(Frellesvig 2010: 66–67).

31 See Quinn (1987: 93–96) for a discussion regarding the distinctions between the prototypically

“autonomous realization” of endoactive nu and the “caused realization” of exoactive tu.

27

There is an abundance of research on the narrative functions of these perfective

auxiliaries, i.e., how they are used by a narrator when constructing a story, in vernacular

wabun Heian texts. Kumakura, for example, argues “tu and nu signify the realized point in

time, punctuating the flux of time wherever a change occurs in the course of events” (1980:

74). Watase (2013: 190, 90) highlights the use of both perfectives when narrators shift

scenes and tu’s potential to mark recent past tense—in evidential terms, a maximally high-

evidence past (Quinn 2018).

Throughout this study I follow scholars such as Quinn (1987) and Frellesvig (2010)

in referring to these auxiliaries as perfectives.

2.2 Auxiliaries ari and tari in Early Middle Japanese Wabun32

The auxiliaries ari and tari are often treated as a pair in studies of EMJ morphology

due to their shared semantics, as demonstrated in studies below. In pre-EMJ texts, auxiliary

ari was more productive, with tari generally limited to the periphrastic construction te ari

(the gerund formant te followed by the stative verb ari). This construction lost its

morpheme boundary and became the tari analyzed in this study after undergoing

phonological reduction (te ari > teari > tari). However, by the middle of the Heian period,

tari became a predominant stative due to its greater morphological versatility (Suzuki

2009: 290, Oda 2015: 137).

32 Portions of this subsection are to be published in Bundschuh (2021b).

28

As an auxiliary, ari is limited to governing quadrigrade and s-irregular class verbs

(introduced in section 1.4). Thus, with atumaru ‘x gathers’ we get atumareri (< participle

atumari + ari) and with su ‘do,’ we get seri (< participle si + ari). Auxiliary tari, on the

other hand, is attested governing participles from all verb classes in the Heian period,

including those governed by ari. For example, in the same 883 CE kundokubun rendition

of the Ten Wheel Sutra, discussed below, we find atumari-tari from the quadrigrade verb

atumaru ‘gather’ and si-tari from the s-irregular verb su ‘do,’ in addition to sugi-tari from

the upper bigrade verb sugu ‘exceed’ and e-tari from the lower brigade verb u ‘gain, attain.’

Auxiliary ari is much more prevalent than tari in Early Heian texts, with the latter still in

its periphrastic form te ari in some earlier cases. However, although the distinctions

between their morphology are well known and widely accepted, many researchers continue

to analyze them as semantic equivalents.

For example, the Jidaibetsu kokugo daijiten jōdai hen (時代別国語大辞典上代編

‘A periodized dictionary of Japanese: Old Japanese edition’), an authoritative dictionary of

Old (and Early Middle) Japanese, defines both ari and tari as resultative, or stative,

auxiliaries that govern the participle (Omodaka 1967: 449–50, 810). Suzuki (2009)

describes both as ‘perfects’ (J. パーフェクト pāfekuto) that express both the result of a

change and evidentially informed epistemic modality at the end of the verb phrase. Ijima

(2011) defines them as resultative statives (J. 結果存在 kekka sonzai) with their most

common textual function being scene depiction. Watase (2013) also describes them as

statives that depict an unmoving scene before our eyes. Most recently, Oda (2015)

29

recapitulates prominent analyses and concludes there is no significant semantic difference

between the auxiliaries.

I follow Frellesvig (2010) in referring to both ari and tari as statives. Here I quote

his discussion on their respective semantics in EMJ in full:

… there is no agreement among scholars about whether the two auxiliaries

through the period of coexistence were (free or conditioned) variants or whether

they expressed different categories, apart from the fact that whatever distinction

they may have expressed was neutralized with vowel base verbs (where only

[tari] was used). It is clear that [tari] interacted differently with other aspect

auxiliaries than [ari] did, and that [tari] at some point acquired a function of

perfect (understood as referring to the ‘continuing present relevance of a past

situation’ (Comrie 1976: 52)), i.e. with some temporal reference, whereas [ari]

remained a simple, atemporal stative. (Frellesvig 2010: 239)

He goes on to cite Takeuchi’s (1987) defense of making a categorical distinction between

these auxiliaries.

Takeuchi (1987) surveys both Heian and Kamakura wabun texts and finds a

semantic difference in ari and tari’s relation to temporal deixis. Here I quote her

conclusion in full:

… the distinction between [ari] and [tari] hinges primarily on the latter’s

specifying that the state in question is true for only a limited interval, while [ari]

conveys no such specification. This means that a state of an [ari] predicate may or

may not be true for, or have relevance for, an interval of time preceding or

following that of the aspect locus.33 [A tari] predicate, on the other hand,

specifies that the state is true for the aspect locus and for an interval ulterior to it,

as well as perhaps another interval preceding it. The precise specification must

often be inferred from the context. Thus, [tari] is the more specific, that is the

more limited of the two. (Takeuchi 1987: 166–7, emphasis in original)

33 The point of reference when viewing an event.

30

Takeuchi subsequently labels tari a ‘limited perfect’ and ari a ‘non-limited perfect.’ In

other words, tari is temporally deictic, whereas ari is not.34

2.3 Auxiliaries ki and keri in Early Middle Japanese Wabun

Frellesvig (2010: 65) describes ki and keri as “simple past” and “modal past,”

respectively. He goes on to clarify ki as indexing direct experience and historical

accounts and keri as indexing hearsay, sudden realization, emphatic or exclamatory force,

and perfect aspect (Frellesvig 2010: 74).

At first glance, the definitions for both ki and keri in the aforementioned

authoritative dictionary of early Japanese, the Jidai betsu kokugo daijiten jōdai hen, do

not seem to fall far from Frellesvig’s analyses. Auxiliary ki indexes a personally

experienced past fact or is used when expressing a past fact regarded as certain

(Omodaka 1967: 236). Auxiliary keri, on the other hand, indexes confirmations of past or

present facts, realizations of past or present facts, or is used when expressing

unexperienced or hearsay facts. Furthermore, the entry for keri includes the following

regarding ki: “Contrasting with ki, which expresses a past fact as a past fact, keri’s true

nature is interpreting this in [relation to] the present scene and expressing it” (Omodaka

34 These findings are supported by Otsubo’s (1981) survey of Heian period kundokubun. He finds tari could

be governed by mu, ki, and rarely by tu, which, respectively, can mark future, past, and recent past tense,

depending on the context. On the other hand, Otsubo finds examples of ari governed by negation and modal

auxiliaries zu, zi, and besi, in addition to the three related to tense (mu, ki, and tu) which may govern tari.

However, as explained in chapter 3, this study focuses on TAM auxiliaries in the sentence-final position and

the semantic nuances distinguishing the stative auxiliaries are best examined in non-sentence-final position.

I therefore refer to both as stative auxiliaries throughout this study (see Bundschuh 2021b for an analysis of

their subtly differing semantics in Heian period kundokubun).

31

1967: 283, translation and emphasis mine). This definition of keri, which includes the act

of interpretation in conjunction with expression, resembles Takeoka’s (1963) anata naru

ba (‘other place = peripheral information’) analysis, in which keri introduces information

relevant to the present scene that was, until so introduced, inactive (cf. Chafe 1999).

Hosoe (1932) is considered the first scholar to draw parallels between the Turkish

evidentials di and miş and Japanese ki and keri. He argues di and ki index direct

experience, and miş and keri, indirect experience.35 Suzuki (2009: 375) recognizes the

evidential and epistemic indexicality of keri but finds no connection between this

grammatical category and ki. However, he acknowledges that the two auxiliaries have

semantic overlap to the extent that ether can be used predicate past situations. He also

argues ki is the grammatical marker of past tense in EMJ, while keri marks indirect

evidentiality or mirativity.

Kumakura’s (1980) dissertation investigates narrative time in the longest EMJ

wabun narrative text, Genji monogatari (源氏物語 ‘The Tale of Genji’). He similarly

argues that ki is a simple past-marker and keri is used in narrative to indicate “‘the

situation is that [such and such happened]’; ‘what I relate to you is that [it so happened

that . . .]’; ‘what I have realized is that . . .’; or some such expression” (Kumakura 1980:

6). This translation strategy is reflected in Stinchecum’s (1985) monograph on the

narrative voice in the latter half of the same wabun tale. She argues that keri “implies the

presence of a speaking subject, i.e., a narrator; it implies further that the narrator is

35 See Shinzato (1991) for a more developed argument, in English, for Hosoe’s comparison.

32

making a statement of judgment, something like, ‘I’m telling you that it is so that…’ and

also a reference to a recollected event” (Stinchecum 1985: 12). Both scholars seem

focused on the perfect nature of keri, in that they stress that it involves present

recollection of a past situation.

Suzuki (1999) gives separate analyses for the use of ki and keri in quotations and

in narration. He describes the employment of ki in Genji monogatari as such: in both

quotations and narration, it is used primarily to express past tense events in some way

experienced by the speaker. In quotations, it is never used to qualify one’s own actions,

but rather what one saw or experienced. He defines this use of ki in quotations as

‘sensory reminiscence’ (J. 目睹回想 mokuto kaisō) (Suzuki 1999: 85–87).

In the narrative discourse of the Genji he finds only four non-stative verbs

governed by ki, three of which depict speech acts. In addition, all uses of ki in narrative

passages mark predicates that are no longer the case in the historical present of the

narration, contrasting with above perfect definitions of ki. He defines ki as a ‘relative

tense’ (J. 相対的テンス sōtaiteki tensu) marker, giving its predicate a past tense relative

to the general historical present of the narration (Suzuki 1999: 96–98). In other words,

according to Suzuki, auxiliary ki’s narrative function in wabun is to present facts that are

no longer the case.

His discussions of keri in Genji monogatari cover much more ground, reflecting

the constellation of meaning (Janda & Joseph 1999) of the polysemous definitions given

above. He finds four uses of keri in speech quoted in the text: (1) ‘hearsay’ (J. 伝聞

denbun); (2) ‘recognition of facts’ (J. 認識 ninshiki); (3) ‘realizations of facts’ (J. 気付

33

き kizuki); and (4) ‘interpretation of facts’ (J. 解釈 kaishaku). All four of these, he adds,

emphasize the current relevance of what the speaker is saying (Suzuki 1999: 87–94).

In narrative discourse keri’s constellation of meanings expands further. Suzuki

finds the narrator using keri to mark: (1) ‘relations between condition and consequence’

(J. 条件–帰結関係 jōken kiketsu kankei); (2) ‘contrast with other events’ (J. 対比

taihi); (3) contents (J. 内容 naiyō); (4) ‘transitions’ (J. 切替 kirikae); and (5) ‘other’ (J.

その他 sono hoka) (Suzuki 1999: 98–108). Suzuki translates most of these uses of keri

as nominalized predicates (i.e., referentialized and, thus, deictically grounded in the

discourse) in today’s Japanese.36 In EMJ wabun narratives, however, keri marks breaks of

paragraph-like sections of text as well, as Suzuki notes of use (4) ‘transitions.’ Such keri-

marked predicates are often ones that move the story forward.37 The fifth use of keri,

which Suzuki calls ‘other,’ includes uses that are ambiguous among the preceding four.

Near the conclusion of his section on ki and keri, he notes keri’s use in framing the

narrative (J. テキストの分割と構造化 tekisuto no bunkatsu to kōzōka) (Suzuki 1999:

108).

As we can see from the above analyses of ki and keri, their semantics and

pragmatics differ far more greatly than those of the perfective auxiliaries nu and tu and

36 This nominalization in the modern rendering suggests keri-marked sentence-final predicates in EMJ secular

wabun narrative do not depict events that move the story forward because such clauses need to interact with

the discourse and thus, not be subordinated in nominalized ones. The discourse functions of such events

depict new actions that progress in time are analyzed as Complications under the Labovian (1972) and Nara

(2011) framework discussed in section 3.2.3 below.

37 The discourse function of such paragraph-final sentences is coded as a Result below. See section 3.2.3.

34

the stative auxiliaries ari and tari. However, for the duration of this study I follow the

trend of analyzing them as a pair. I refer to these modal auxiliaries as the past established

fact auxiliary ki and the externally established fact auxiliary keri.38 Next, I present a brief

summary of the TAM auxiliaries in EMJ wabun.

2.4 Summary of TAM Auxiliaries in Early Middle Japanese Wabun

In summary, although Oda (2015) finds nu to be primarily inceptive, he, Quinn

(1987), and Suzuki (2009) all define both nu and tu as perfectives. Quinn (1987) describes

nu as an endoactive perfective and tu as an exoactive perfective, falling on different ends

of Hopper and Thompson’s transitivity spectrum (1980: 252). Ijima (2011) focuses on the

textual function of tu and describes it as signifying a sense of presence for events occurring

in succession and both he and Fujii (2013) argue the textual function of nu as that of

marking scene changes and scene conclusions, respectively.

Both ari and tari have been described as perfects (Suzuki 2009) and statives along

with other semantically related effects (Frellesvig 2010). Regarding their narrative

function, Ijima (2011), defining them as resultatives, argues they are utilized for scene

descriptions and, similarly, Watase (2013), who, like Frellesvig (2010), considers them

statives, finds them most often used for concrete, visual depictions of scenes.

38 Charles Quinn (p.c.) suggested these functional labels, which the website cjp.asc.ohio-state.edu (Quinn

2021) utilizes. See Quinn (1987) and Shinzato (1991) for details on their respective semantics and epistemic

modality.

35

Finally, both Suzuki (2009) and Oda (2015) describe ki as the grammatical marker

of past tense in Classical Japanese, whereas keri marks “reported past,” “narrative past,”

and “mirativity” (Oda 2015: 153). Quinn (1987) demonstrates a modal contrast between

the two with ki marking speaker-grounded established facts and keri externally established

facts, i.e., information whose factuality the speaker deems to be already established, outside

his/her ken; keri thus “bolsters” a predication with a grounding beyond the speaker. These

serve as framing devices for personally grounded vs. externally established information,

respectively.

The analyses of the TAM auxiliaries’ sentence-level semantics and narrative

functions in the Early Middle Japanese wabun register discussed throughout this chapter

are referenced throughout this study as we consider the use of nu, tu, ari, tari, ki, and keri

in EMJ Buddhist narratives. We now turn to the frameworks from which I draw my

methodologies to analyze these TAM markers in their relation to narrative discourse in early

Heian kundokubun.

36

Chapter 3. Framework and Methodologies

In this chapter I provide an overview of the research frameworks that have

informed the methodologies I use in my investigations into the relationship between the

Early Middle Japanese auxiliaries introduced in chapter 2 and the narrative structure of

kundokubun Buddhist texts. After discussing relevant scholarly literature, I introduce my

methodology and data sources.

3.1 Analytical Framework

This research builds upon the traditions of linguistic narrative analysis pioneered

by William Labov’s work in the late 1960s and early 1970s that brought the linguistic

analysis of spoken stories into narrative studies (Labov 1972). The linguistic and cognitive

analyses of the “Pear Stories” by Chafe and others (Chafe 1999) show, among other things,

why narrators chose certain TAM markers in constructing spoken narratives. The present

study extends this descriptive and theoretical tradition by expanding our understanding of

the relationship of morphological marking on individual predicates to larger narrative

structures in a distant but historically significant textual tradition.

Western linguists who investigate kundokubun still generally do so through an

exclusively comparative lens; they have contrasted the heavily borrowed Chinese lexicon

and conservative style with the classics of Heian period belles-lettres, focusing their

37

analysis on its role in the history of the Japanese language, but bypassing the facets of its

narrative structure (e.g., Frellesvig 2010, Alberizzi 2015). From a literary perspective,

investigations of narrative in Heian-period texts have included nuanced accounts of the

pragmatic use of TAM marking. However, using the tools of literary criticism, the authors

devote the majority of their attention to the linguistic expression of internal states of the

characters depicted (through interior monologue and the like), rather than to the larger

narrative structure itself (e.g., Kumakura 1980, Stinchecum 1985). The present study builds

on and complements these literary and historical studies by focusing on what purposes TAM

marking serves in structuring kundokubun Buddhist narratives. The heuristics of narrative

analysis uniquely lend themselves to understanding the overlapping framing strategies of

a narrator quoting the Buddha who is, in turn, quoting others as he relates his parables. In

this study I show how certain features in the Japanese expression of time were adapted in

the process of translation from Chinese to reproduce Buddhist narratives in ways that could

effectively engage a Japanese audience.

While not specifically concerned with the linguistic genre of Buddhist kundokubun,

several researchers have undertaken analyses of Buddhist practice in Japan, primarily from

the perspective of comparative literary studies. Kornicki (2014), for example, explores how

Buddhist texts were vernacularized in several earlier Asian societies, including Japan.

Lowe (2017) presents a detailed study of the ritualized practice of sutra copying with

careful analyses of colophons written by the transcribers in pre-Heian Japan. The

kundokubun glossings examined in this study are on such copies of Sinitic sutras

reproduced in Japan.

38

Other scholars have described the study, reproduction, and repurposing of Sintic

literature among the Heian elite. Guest (2013) investigates kanbun primers and concludes

that “the centrality of kundoku reading and the way this emphasis on logographic reading

promoted the adaptation of elements from the classical Chinese tradition into ‘vernacular’

(kana-based) contexts” (Guest 2013: 203). Steininger (2017) examines official state

documents written in Sinitic script during the period and argues that, similar to Lurie

(2011)’s findings that Sinitic texts were read aloud in kundokubun in pre-Heian Japan, texts

of the Chinese cannon were primarily read via kundoku (‘vernacular reading’) throughout

the Heian period as well. Fukumori (2001) explores how mid-Heian women used their

knowledge of the Chinese canon to further their standing in court and notes that this

Chinese knowledge included “the reading and writing of sutras” (Fukumori 2001: 207).

These studies suggest a living tradition of composition as performance within the

Heian court. By comparison, much research, fine-grained and grammatical, remains to be

done on how monks created and maintained narrative structure in the documents that they

studied, copied, and preserved. Sutra translation in Japan provides a unique vantage point

for theorizing such concerns because of its place in the “transnational” tradition of

performing classical Buddhist verse, which extended far outside its original contexts.

Nara (2011) investigates tense-switching phenomena in a modern Japanese novel

from a discourse-based perspective. He surveys researchers with differing views that argue

tense-switching in Japanese is due to the sentences’ transitivity, lexical aspect, narrative

structure, and vividness. Focusing on the past/perfective morpheme ta in modern Japanese,

Nara codes every sentence of Natsume Sōseki’s 1906 novel Botchan for its location in the

39

text, lexical aspect, transitivity, tense, location within the episodic structure, and discourse

function. He concludes that “transitivity is not as significant a player in influencing tense

choice in this text, while the other two variables—aspect and narrative metafunction—

appear to play a larger role in motivating tense choice” (Nara 2011: 288).

The six tense, aspect, and modality auxiliaries of interest in this study—nu, tu, ari,

tari, ki, and keri—all have contexts in which they are rendered in modern Japanese as ta,

the morpheme examined in Nara’s study. Although my analysis of the internal structuring

of narrative in the sutras is mindful of the analyses of the TAM auxiliaries in contemporary

(Heian-period) Japanese stories, discussed in the previous chapter, the heuristic that Nara

brought to bear on Sōseki’s novel—analyzing the relationship between the grammatical

marking of time and transitivity, lexical aspect, and discourse function—presents a novel

approach to understanding this relationship in early Heian kundokubun Buddhist texts.

3.2 Methodology

My research primarily follows the methodology of Nara (2011), which I have just

surveyed. Every sentence in the corpus that ends in one of the six TAM auxiliaries is coded

for transitivity, lexical aspect, and narrative discourse function. Tokens of these auxiliaries

in non-final predicates are not included in this analysis because their embedding limits their

direct interaction with the discourse. Furthermore, I only count TAM markers that head a

sentence-final predicate. Thus the stative auxiliary ari in ara-zu (ARI-NEG) and the

perfective auxiliary nu in ni-ki (NU-KI) are beyond the focus of this study, although they

are mentioned in discussions below. Including transitivity and lexical aspect in the analysis

40

allows us to determine whether they relate consistently to how these auxiliaries are used

narratively. The remainder of this subsection describes the coding process.

3.2.1 Transitivity

Nara (2011) gives each finite predicate a transitivity score of 0 to 3 based on Hopper

and Thompson’s ten transitivity parameters (1980: 252): number of participants, kinesis,

telicity, punctuality, volitionality, affirmation, mode, agency, affectedness of object, and

individuation of object, using aggregates of points calculated from each. In Nara’s study,

this leads to a score of 0 for sentences in quotations, and a score of 1 for low, 2 for medium,

and 3 for high transitivity. To properly account for the many narratives embedded in

quotations in Buddhist sutras, I include the transitivity of direct quotations in the present

study of sentence-final predicates. I score intransitive predicates (1), transitive predicates

that do not result in change in their direct object (2), and both ditransitive predicates

(causatives) and transitive predicates that result in a change in their direct object (3). For

example, nari-nu (なりぬ ‘become-NU’) is scored (1), mi-tu (見つ ‘see-TU’) is scored (2),

and e-sime-tu (得令めつ ‘gain-CAUS-TU’) and yabure-tari (破れたり ‘break-TARI’) are

scored (3).

3.2.2 Lexical Aspect

Lexical aspect in this study is determined by the predicate’s stativity (whether it is

static or dynamic), durativity (whether it occurs or remains over an interval or not), and

telicity (whether there is an endpoint or not), i.e., using the features of the Kenny-Vendler

41

system (Kenny 1963, Vendler 1967). Following Nara’s (2011) coding, each sentence-final

predicate is determined to be one of the following: (1) a State (static, durative, atelic), (2)

an Activity (dynamic, durative, atelic), (3) an Accomplishment (dynamic, durative, telic),

or (4) an Achievement (dynamic, instantaneous, telic). For the purposes of this study, the

lexical aspect of finite predicates is coded with their full morphology in context, including

relevant adjuncts. In other words, although ipu (言ふ ‘speak’) is an Activity as a bare verb

without context, when it is governed by the perfective auxiliary tu after someone speaks it

becomes an Accomplishment due its telicity, or having an endpoint.39 We can illustrate

this with examples utilizing past established fact ki, exoactive perfective tu, and endoactive

perfective nu: ari-ki (ありき be/have-KI), State; kango.si-ki (歡娛しき delight-KI), Activity;

kiki-tu ( 聞 きつ hear-TU), Accomplishment; and nari-nu ( 成 りぬ become-NU),

Achievement.

3.2.3 Discourse Function

Nara (2011) was influenced by Labov’s (1972) six “elements of narrative

structure.” However, instead of coding Quotations cited by the narrator as Evaluations (one

of Labov’s functions), he gives them their own category, leading to seven “discourse

function types” that he uses in his analysis of narrative in Japanese. The Abstract outlines

the story; the Orientation establishes the characters and setting; Quotations are when a

39 Verbs governed by the participial forms of the perfectives nu and tu, ni and te, respectively, followed by

another TAM auxiliary generally also gain telicity, unless that context or additional morphosyntax prevent

such an interpretation.

42

character is speaking in a narrative and Semi-direct Quotations are predicates outside of

direct quotations governed by the complementizer particle to; 40 Complications, or

Complicating Actions, are events that move the story forward; Evaluations describe a scene

or make an evaluative comment regarding its development (and thus pause the plot);

Results bring closure to the scene; and the Coda shifts the focus from the story to the time

in which the narrative is presented. Regarding Results, any clause that depicts an event

directly preceding what can be considered to be a new paragraph in a sutra’s narrative, such

as those beginning with so no toki ni (尒時に ‘at that time’), gain this classification.

Following Nara’s (and, in part, Labov’s) categorization, I identify and code each sentence-

final predicate for its function in or as (1) Abstract, (2) Orientation, (3) Quotation, (4)

Complication, (5) Evaluation, (6) Result, and (7) Coda.

3.3 Data Sources

Data were drawn from the following Buddhist texts rendered into kundokubun via

gloss during the first half of the Heian period: an early-9th-century CE complete rendering

of the Golden Light Sutra (J. 金光明最勝王経 Konkōmyō saishō ōkyō Skt.

Suvarṇaprabhāsottama sūtra) glossed at Saidaiji temple (Kasuga 1985a); a mid-9th-

century CE rendering of the 5th and 9th scrolls of the Golden Light Sutra glossed at Tōdaiji

temple (Tabuchi 1987); a late-9th-century CE rendering of eight scrolls of the Ten Wheels

40 There are numerous narratives within quotations in Buddhist texts, the majority of which are depicted as

spoken by the Buddha in response to a question regarding Buddhist practice or cosmology. Once an

embedded narrative begins, the lines therein are scored for their own discourse function, with character lines

scored as Quotations (3).

43

Sutra (J. 地蔵十輪経 Jizō jūrin kyō41) (Nakada 1958, Nakada 1980); and an early-10th-

century CE complete rendering of the Immeasurable Meanings Sutra (J. 無量義経

Muryōgi kyō42) (Kabutogi & Nakada 1979). The cited researchers have deciphered the

original glossed texts and rendered them into kakikudashibun (書き下し文 ), or a

transliteration of the glossed Sinitic script into Japanese script (introduced in section 1.5

and referred to as kakikudashi for the remainder of this study).

The reason for using such secondary texts is three-fold. First, access to extant copies

of Heian-period sutras is extremely limited. Most are regarded as holy relics at temples

where they are stored and gaining access to one source text can take a great deal of time to

build rapport with ranking monks (Kobayashi 2011). Second, even with direct access to

the source text, one must decipher the kunten glosses to determine the proper Japanese

rendering. Third, we later researchers of these texts owe it to the texts, their tradition, and

researchers before us, to inform ourselves of what they have to teach us.

I limit the present study to early Heian kundokubun texts because by the latter half

of the Heian period glossing practices became more standardized due to monks paying

deference to established kundoku (訓読 ‘vernacular reading’) strategies, which led to

greater codification and homogeneity (Kobayashi 2011: 192). See appendix B for more on

41 There is no extant Sanskrit version of this sutra (Buswell & Lopez 2014: 448).

42 Not only is there no extant Sanskrit version of this sutra, but because the monk credited as its translator

into Sinitic is not associated with any other texts, there are those who suspect it of being an apocryphon, i.e.,

a Chinese composition (Buswell & Lopez 2014: 1001–1002).

44

early Heian kundokubun data sources. In the following chapter I present findings on the

distribution and functions of the EMJ TAM markers in narratives found in these texts.

45

Chapter 4. Close Analyses of Selected Early Heian Sutra Narratives

We now turn to the kundokubun texts themselves. In this chapter I examine the

relationship between transitivity, lexical aspect, and discourse function of sentences in

which we find a sentence-final TAM auxiliary in sutras rendered in Japanese in the first half

of the Heian period (ca. 9th–10th centuries CE). In the first two sections I examine a

rendering of the Golden Light Sutra’s (金光明最勝王経 Konkōmyō saishō ōkyō)

outermost narrative (i.e., spoken by the matrix narrator rather than embedded in the

dialogue) and an embedded narrative depicted as being told by the Buddha. Then in the

third and fourth sections I turn to the outermost and a select embedded narrative in the Ten

Wheels Sutra (地蔵十輪経 Jizō jūrin kyō). Finally, in the fifth section I survey the

outermost narrative of the Immeasurable Meanings Sutra (無量義経 Muryōgi kyō).

4.1 The Narration of the Saidaiji Temple Golden Light Sutra43

I investigate a kundokubun rendition of the Golden Light Sutra (金光明最勝王経

Konkōmyō saishō ōkyō) that was translated into Japanese from Chinese via gloss at Saidaiji

43 A slightly different version of this subsection will be published in Bundschuh (2021a).

46

Temple ca. 830 CE (Okimori 2001), and for the remainder of this subsection referred to as

the GLS.

The sutra was likely completed in Sanskrit around the turn of the 5th century CE.

The Indian monk Dharmakṣema (J. 曇無讖 Donmusen) first translated it into Chinese for

the kingdom of Northern Liang dynasty. However, his translation did not include all the

chapters and was thus incomplete. In 703 CE, the Chinese Buddhist monk Yijing (J. 義淨

Gijō), who had spent many years studying and copying sutras in India, produced a complete

Chinese translation, which is the source text for the kundokubun rendition we examine

below. This translation was quickly imported into Japan and recited in temples throughout

the kingdom (国分寺 kokubunji) by imperial degree in 741 CE to project centralized

authority and protect the state (Kasuga 1985b: 6–7; Buswell & Lopez 2014: 247–284, 877,

1028).

Returning to the text itself, there are 425 sentence-final predicates outside of

quotations in the sutra’s outermost narrative, which make up less than 13% of the over

3,500 lines throughout the text. Although not sentence-final, I include pre-quotative

adverbial phrases that end in verbs of speaking nominalized by aku, such as ipaku (曰く

‘that which is said’), in this count in order to overcome coding difficulties because they

form an endpoint to the narration before a quote. One of the well-studied features of

kundokubun is the use of “speech-act” predicates both before and after quotes. In many

kundokubun sutras, including the GLS, there are numerous extended embedded quotations,

which lead to sentences with only two clauses of outermost narration spanning numerous

lines and pages. For example, the sentence that begins 世尊即彼の天及諸の大衆の為

47

に、伽他を說（き）て曰（は）ク (seson sunapati ka no ten oyobi moro no daisyu no tame ni

gada wo tokite ipaku ‘The Lord then, for the sake of all those devas, said in verse’) (Kasuga

185a) on page 166 contains 70 lines of an embedded verse before the sentence concludes

with とのたまふ (to notamapu ‘he humbly said’) on page 172.

For the duration of section 4.1, all references to the “narrative” of the GLS

specifically refer to the outermost one rather than any embedded ones. As mentioned

above, data were gathered from Kasuga 1985a.

The table below presents the total times the TAM auxiliaries are used throughout the

narration of the GLS, how many of those are sentence-final, and their sentence-final usage

percentage, rounded to the nearest percent.

Table 2 TAM Marking Totals in GLS Narration

TAM Narrative Total Narrative Sentence-Final Total Sentence-Final %

ki 23 18 78%

keri 2 0 0%

nu 54 49 91%

tu 22 20 91%

ari 40 23 58%

tari 19 9 47%

As we can see, ki, nu, and tu are primarily used sentence-finally, i.e., finitely, and thus

strongly interact with the discourse, as its “leading edge”. Stative perfectives ari and tari

are both used sentence-internally and sentence-finally to a relatively even degree. On the

other hand, as it is only used sentence-internally in the narrative, keri has no direct

interaction with the discourse. The following example of keri is representative.

48

(1) 身心踊悅して、未曾有にいましケリと歎したてまつる。 (Kasuga 1985a: 19)44

sinsin yuetu.si-te mizou ni imasi-keri

body.mind dance.with.joy-GER unprecedented COP be.HON-KERI

to tan.si-tatematuru

COM be.moved-HUM

‘His body and soul dancing with joy, he marvels [saying] he realizes this is

unprecedented.’

Here keri is clearly a grammatical marker of mirativity45 , serving one of its primary

functions in early kundokubun (Kasuga 1985b, Otsubo 1977). As keri is only used by the

narrator in direct and indirect quotations, it is not included in the subsequent analyses.

There are some cases of two TAM auxiliaries governing the same predicate. The

following is the only example of both nu and tari, ni-tari, in the narrative of the GLS.

(2) 二乘の所行の境界を逾（え）にたり。 (Kasuga 1985a: 9)

nizyau no syogyau no kyaugai wo

koe-ni-tari

two-vehicle46 GEN congition GEN sphere ACC

overcome-NI-TARI

‘They have completely overcome the sphere of cognition practiced by the two-

vehicle adherents.’

This use matches Suzuki’s (2009: 370) definition of ni-tari—a perfect with first-hand

evidential qualities that most commonly governs verbs of motion.

44 See the introductory remarks (凡例 hanrei) in Kasuga (1985a) for orthographical conventions. Emphasis

in all examples mine. I use periods in the Japanese between nouns and the verbalizer su and in the

metalanguage when I use multiple words or abbreviations to render a single Japanese element.

45 Mirativity is a grammatical category that includes “sudden discovery” and “new information” (Aikhenvald

2012).

46 This is a pejorative term in Mahāyāna Buddhism applied to followers of so-called Hīnayāna, or ‘lesser

vehicle,’ Buddhism (Buswell & Lopez 2014: 286).

49

While focusing on the textual functions of ki and keri in Early Middle Japanese

wabun, Fujii (2016: 254) notes that nu followed by ki, ni-ki, only heads narrative

Complications and Results in kundokubun and is an expression used to create narrative

frames.47 The following example demonstrates how this pattern is used in the GLS.

(3) 妙幢菩薩は佛の足を礼したてまつり已（り）て、（從）座ヨリして（而）起（ち）て、其

の本處に還（り）にキ。 (Kasuga 1985a: 19)

meudau bosatu pa potoke no asi wo

Ruciraketu48 bodhisattva TOP Buddha GEN feet ACC

rei.si-tatematuri-wopari-te za yori.si-te tati-te so no

bow-HUM-finish-GER seat be.from-GER stand-GER that GEN

ponsyo ni kaperi-ni-ki

original.place LOC return-NI-KI

‘The bodhisattva Ruciraketu finished paying obeisance to the Buddha’s feet, stood

up from his seat, and returned to his original place.’

I consider Fujii’s analysis further in this section’s conclusion below. In cases when there

are multiple TAM auxiliaries governing the sentence-final predicate, such as in examples

(2) and (3) above, only the outermost auxiliary, that which interacts directly with the

discourse, is coded. We now turn to how the TAM auxiliaries relate to transitivity, lexical

aspect, and discourse structure in the outermost narration of the GLS.

47 Fujii (2016) does not cite Labov (1972), but appears to be influenced by his framework. He uses not only

the terms tenkaibu (展開部 ‘development part’) and shūkyokubu (終局部 ‘final part’), which I interpret as

Complication and Result, respectively, but he also uses hyōgobu (評語部 ‘evaluation part’), Evaluation, in

a similar way to how it is defined in Labov (1972).

48 Soothill & Hodous (2014: 234).

50

4.1.1 Transitivity in GLS Narration

The table below depicts the relation between TAM marker and transitivity scores.49

There were no TAM-marked predicates of high transitivity in the outermost narrative of the

GLS.

Table 3 The Transitivity of TAM-marked Predicates in GLS Narration

TAM low (%) medium (%)

ki 17 (94%) 1 (6%)

nu 36 (73%) 13 (27%)

tu 1 (5%) 19 (95%)

ari 20 (87%) 3 (13%)

tari 5 (56%) 4 (44%)

Auxiliaries ki, ari, and, as expected from its endoactive semantics, nu, display a clear

proclivity to mark low-transitivity predicates, whereas tari seems evenly split. On the other

hand, tu, being exoactive, primarily governs predicates of medium-transitivity, most

commonly u (得 ‘gain, acquire’).50 The following is its only case of governing a low-

transitivity predicate.

49 As discussed in section 3.2.1 above, intransitive predicates are marked as having low transitivity, transitive

predicates that do not result in any external change are marked medium, and transitive predicates that cause

an external change to the scene are marked as having high transitivity.

50 It does, however, govern high-transitivity predicates in embedded quoted narration.

51

(4) 尒時釋迦牟尼如來、是（の）經を說（き）たまふ時に、（於）十方世界に有ラルゝ無

量百千萬億の諸の菩薩衆い、各（從）本土ヨリ、鷲峯山に詣（り）つ。 (Kasuga

1985a: 200)

so no toki Syakamuni nyorai ko no kyau wo

that GEN time Śākyamuni51 tathāgata52 this GEN sutra ACC

toki-tamapu toki ni zipupau sekai ni ara-ruru

expound-HON time LOC ten.directions world LOC exist-PASS

muryau pyakusen mannoku no moro no

immeasurable hundred.thousand trillion GEN all GEN

bosatu syu i onoono pondo yori

bodhisattva53 PL NOM each homeland from

zyupusen ni makari-tu

Vulture.Peak54 LOC come-TU

‘At that time when the Tathāgata Śākyamuni preached this sutra all the

immeasurable hundreds of thousands of trillions of bodhisattvas of the worlds of

the ten directions came to Vulture Peak from each of their homelands.’

One possible explanation for this use of tu is the notion of volition, which generally

increases with transitivity. Bodhisattvas coming to Vulture Peak when the Buddha

preaches the GLS would fall under the category of high volition.55

51 The historical Buddha (see Buswell & Lopez 2014: 741–742).

52 This is a common epithet for the Buddha from a past passive participial form of Skt. tathā-√gam- or tathā-

ā√gam-, meaning ‘one who has thus gone/come’ (see Buswell & Lopez 2014: 897).

53 A being who resolves to generate aspirations of enlightenment in others and follows the path toward the

achievement of Buddhahood through the development of the six perfections of giving, morality, patience,

effort, concentration, and wisdom (Buswell & Lopez 2014: 131, 134, 137–138, 624).

54 Skt. Gṛdhrakūṭaparvata, where the Buddha preached many of the sutras (Buswell & Lopez 2014: 327).

55 Auxiliary tu could also be indexing a recent past from the narrator’s perspective—the bodhisattvas may

have recently arrived.

52

4.1.2 Lexical Aspect in GLS Narration

The following table displays the relationship between TAM morphemes and lexical

aspect in the narration of the GLS.

Table 4 The Lexical Aspect of TAM-marked Predicates in GLS Narration

TAM State (%) Activity (%) Accomplishment (%) Achievement (%)

ki 2 (11%) 10 (56%) 5 (28%) 1 (6%)56

nu 0 (0%) 0 (0%) 16 (33%) 33 (67%)

tu 0 (0%) 0 (0%) 3 (15%) 17 (85%)

ari 24 (100%) 0 (0%) 0 (0%) 0 (0%)

tari 9 (100%) 0 (0%) 0(0%) 0 (0%)

Auxiliary ki is the only TAM morpheme that interacts with the narrative discourse, i.e., is

found sentence-finally, that is not restricted by lexical aspect, which is to be expected as it

is an indicator of tense and modality. However, the majority of its uses occur with

predicates that depict an activity, such as the following.

(5) 尒時に無量無邊恒沙の大衆い佛の説を聞（きたま）へ已（り）て、皆大に歡喜して

信受し奉行しキ。 (Kasuga 1985a: 208)

so no toki ni muryau mupen gauzya no

that GEN time LOC immeasurable immense innumerable GEN

daisyu i potoke no setu wo kiki-tamape-wopari-te

group NOM Buddha GEN explanation ACC listen-HUM-finish-GER

mina dai ni kwangi.si-te sinzyu.si bugau.si-ki

all great COP be.overjoyed-GER accepting.truth enact-KI

‘At that time the immeasurable, immense, innumerable group, having heard the

Buddha’s explanation, all in great delight, accepted and carried it out.’

56 Percentages occasionally add up to 101 or 99 rather than 100 due to rounding.

53

The auxiliaries nu and tu only govern telic predicates, which reflects their perfectivity.

Their preferences for Achievements over Accomplishments reflects the fact that in Early

Middle Japanese predicate morphology, certain specific contextual features are required

for a verb to participate in representing the latter. In other words, all Accomplishments are

the joint constructions of verbs that by themselves express Activities and some other

feature that adds an end point to the Activity. An endpoint was commonly provided with

an auxiliary that indicated one, such as the following examples with the perfective

auxiliaries.

(6) 悉ク皆雲のゴトク集（り）ヌ。 (Kasuga 1985a: 4)

kotogotoku mina kumo no gotoku atumari-nu

all everyone cloud GEN like gather-NU

‘They all gathered like clouds.’

(7) 尒時如意寶光耀天女、（於）大衆の中にあり、深法を說（き）たまふを聞（き）ツ。

(Kasuga 1985a: 89)

so no toki nyoipou kwau’eu tennyo

that GEN time wish-fulfilling.gem57 shining celestial.nymph58

daisyu no naka ni ari zinbapu wo

gathering GEN in LOC be profound.truth59 ACC

toki-tamapu wo kiki-tu

expound-HON ACC hear-TU

‘Then the shining celestial nymph of the wish-fulfilling gem, being among the

crowd, heard the Buddha’s preaching of the profound truth.’

57 Skt. cintāmaṇi (Buswell & Lopez 2014: 193).

58 Skt. apsaras (Buswell & Lopez 2014: 60).

59 (Soothill & Hodous 2014: 356).

54

Both atumaru (‘gather’) and kiku (‘hear’) by themselves express atelic Activities, but in

the above examples they are depicting narrative events, explicitly presented as completed,

i.e., having an endpoint, by the two perfective auxiliaries.

The auxiliaries ari and tari, both aspectually stative, naturally only head State

predicates. They both present the state resulting from the change that the base verb

represents. Example (2) above and the following example demonstrate this.

(8) 衆の寶の羅網をモチテ其の上を莊嚴せり。 (Kasuga 1985a: 187)

syu no takara no ramau wo moti-te so no upe

group GEN jewel GEN net ACC hold-GER that GEN above

wo syaugon.se-ri

ACC adorn-ARI

‘They have adorned his head with many jeweled nets.’

The English translations I have given in these two examples reflect the resultative nature

of these two statives. As previously mentioned, their difference is more morphological than

semantic, in that ari is generally limited to governing quadrigrade and s-irregular verbs

whereas tari, in the early Heian period, most commonly governs monograde and bigrade

verbs. The older ari came to be replaced by the innovative tari due to the latter’s potential

to govern any verb class (see section 2.2 above). We now turn to the relationship between

the TAM auxiliaries and discourse function in the GLS’s outermost narration.

4.1.3 Discourse Function in GLS Narration

The following table displays the relation between the TAM morphemes and this

study’s narrative discourse functions.

55

Table 5 The Discourse Function of TAM-marked Predicates in GLS Narration60

TAM Abstract (%) Orientation (%) Complication (%) Evaluation (%) Result (%)

ki 1 (6%) 2 (11%) 1 (6%) 0 (0%) 14 (78%)

nu 0 (0%) 0 (0%) 46 (94%) 0 (0%) 3 (6%)

tu 0 (0%) 0 (0%) 20 (100%) 0 (0%) 0 (0%)

ari 0 (0%) 14 (58%) 0 (%) 10 (42%) 0 (0%)

tari 0 (0%) 6 (67%) 0 (%) 3 (33%) 0 (0%)

The first few lines of the GLS are provided in example (9) below to begin our discussion.

(9a) 是（の）如キことを我レ聞きたまへキ。

ko no gotoki koto wo ware kiki-tamape-ki

this GEN like NML ACC I hear-HUM-KI

‘Thus I have humbly heard.’61

(9b) 一時薄伽梵、王舍城鷲峰山の頂に、（於）最も清淨にして甚深なる法界の諸佛

（の）（之）境たる、如來の所居に在（し）キ。

itizi bagabon wausyazyau zyubusen no itadaki ni

once bhagavat62 Rajgir63 Vulture.Peak GEN peak LOC

motomo syauzyau.ni.si-te zintan naru popukai no

most be.profound-GER deep COP dharma.realm GEN

syobutu no sakai taru nyorai no syokyo ni

all.Buddhas gen region COP tathāgata GEN abode LOC

zai.si-ki

dwell-KI

60 Percentages throughout this study occasionally add up to 101 or 99 rather than 100 due to rounding.

61 The expression ko no gotoki koto can be more literally translated as ‘a thing/matter such as this,’ but this

translation does not properly represent this Buddhist context. The Chinese source from which the Japanese

is rendered, 如是, is a translation of the Sanskrit expression evaṃ, which is most often translated as ‘thus’

in the opening line of sutras.

62 This is a common epithet of a buddha, which generally means ‘lord’ (see Buswell & Lopez 2014: 108).

63 Skt. Rājagṛha. An ancient capital city in India near Vulture Peak (see Buswell & Lopez 2014: 694).

56

‘Once on the top of Vulture Peak of Rajgir the lord dwelled the abode of the

Buddhas in the sphere of religion in the profound Buddha-region.’

(9c) 與には大苾芻の衆九万八千人ありキ。

tomo ni pa daipitisu no syu kuman patisen

together LOC TOP great.monk GEN gathering 90,000 8,000

nin ari-ki

people be-KI

‘Alongside him was a gathering of 98,000 great monks.’

(9d) 皆是レ阿羅漢なり。 (Kasuga 1985a: 1)

mina kore arakan nari

everyone these arhat64 COP

‘All of these are arhats.’65

(9e) 能ク善ク調伏せり。

yoku yoku zyaubuku.se-ri

successfully well discipline-ARI

‘They have successfully properly disciplined themselves.’

(9f) 大象王の如し。

daizauwau no gotosi

great.elephant.king66 GEN like

‘They are like Śākyamuni.’

(9g) 諸の漏已に除し、復煩悩無し。

moromoro no rui ni zyo.si mata bonnau nasi

all GEN desire LOC remove also affliction67 be.NEG

64 Skt. term meaning ‘worthy one.’ This term refers to Buddhist saints (see Buswell & Lopez 2014: 62).

65 Although using the past tense here in English (‘All of these were arhats.’) would be more natural in a

literary translation, throughout this dissertation I endeavor to have my English renditions reflect the

kundokubun predicate morphology as closely as possible.

66 This is a term for the historical Buddha Śākyamuni (Soothill & Hodous 2014: 391).

67 Skt. kleśa (Buswell & Lopez 2014: 438).

57

‘Removing all desires, they also have no afflictions.’

(9h) 心善（く）解脱し、恵善ク解脱せり。

kokoro yoku gedatu.si e yoku gedatu.se-ri

heart properly enlighten mind properly enlighten-ARI

‘Their hearts being properly enlightened, their minds have been properly

enlightened.’

(9i) 所作已に畢へたり。 (Kasuga 1985a: 1)

syosa sude.ni wope-tari

action already end-TARI

‘Actions have already ceased.’

An analysis of this narrative passage also serves to introduce the discourse functions. The

first sentence (9a) is best categorized as the Abstract, defining the entirety of the rest of the

sutra as something heard by the narrator. The following two lines, presenting the location

and those in attendance, both begin the Orientation and also contain the only two predicates

governed by ki in it. However, the Orientation continues with stative predicates unmarked

by ki, including verbs governed by ari and tari. Auxiliary ki is most often used to mark

Results, such as in (3) and (5) above. The one instance of its use in a Complication, to

notamapiki (トノタマヒキ ‘honorifically said that’ (Kasuga 1985a: 163)), comes after a

multi-page quotation, and thus, perhaps, is pragmatically employed to reorient the listener

and reframe the narrative scene.

Examples (6) and (7) above of nu and tu, respectively, also display their use in

Complications. These two perfectives are frequently used in GLS narration to present

dynamic events that, as Complications, move the story forward.

58

4.1.4 Conclusions Regarding GLS Narration

In section 4.1 I examined the transitivity, lexical aspect, and narrative function of

finite predicates throughout the outermost narrative of the GLS governed by ki, keri, nu, tu,

ari, and tari. Although keri does not occur sentence-finally in kundokubun narration, the

other five morphemes have clearly delineated functions. Table 6 summarizes the above

findings.

Table 6 Summary of Findings Regarding GLS Narration

TAM Total Transitivity Lexical Aspect Discourse Function

ki 18 low (94%) unrestricted Result (78%)

nu 49 low (73%) Achiev. (67%) Complication (94%)

tu 20 medium (95%) Achiev. (85%) Complication (100%)

ari 24 low (87%) State (100%) Orientation (58%)

tari 9 unrestricted State (100%) Orientation (67%)

The outermost narration of the GLS shows a preponderance of low-transitivity predicates.

Perfective auxiliary nu’s semantics generally limit it to these, whereas perfective tu heads

predicates at the other end of Hopper and Thompson’s transitivity scale (1980: 252).

However, the semantics of both of these perfective auxiliaries guarantee the telicity of their

predicates, most of which being Achievements, and in kundokubun narration telic

predicates are most commonly used in Complications. Returning to Fujii’s (2016)

comment on the framing function of ni-ki in kundokubun discourse, the data gathered here

demonstrate that perfective nu has no particular affinity with the onsets and conclusions of

narratives. Its single use in a Result is one also governed by ki, which I introduced at the

beginning of this subsection but reproduce here.

59

(10) 妙幢菩薩は佛の足を礼したてまつり已（り）て、（從）座ヨリして（而）起（ち）て、其

の本處に還（り）にキ。 (Kasuga 1985a: 19)

meudau bosatu pa potoke no asi wo

Ruciraketu bodhisattva TOP Buddha GEN feet ACC

rei.si-tatematuri-wopari-te za yori.si-te tati-te so no

bow-HUM-finish-GER seat be.from-GER stand-GER that GEN

ponsyo ni kaperi-ni-ki

original.place LOC return-NI-KI

‘The bodhisattva Ruciraketu finished paying obeisance to the Buddha’s feet, stood

up from his seat, and returned to his original place.’

This Result, stated at the end of a scene, happens to involve the finality of the bodhisattva

returning to his seat, but most Results in GLS narration are depictions of the assembly

reaction to the words of the Buddha, such as the following example introduced above.

(11) 尒時に無量無邊恒沙の大衆い佛の説を聞（きたま）へ已（り）て、皆大に歡喜して

信受し奉行しキ。 (Kasuga 1985a: 208)

so no toki ni muryau mupen gauzya no

that GEN time LOC immeasurable immense innumerable GEN

daisyu i potoke no setu wo kiki-tamape-wopari-te

group NOM Buddha GEN explanation ACC listen-HUM-finish-GER

mina dai ni kwangi.si-te sinzyu.si bugau.si-ki

all great COP be.overjoyed-GER accepting.truth enact-KI

‘At that time the immeasurable, immense, innumerable group, having heard the

Buddha’s explanation, all in great delight, accepted and carried it out.’

Sentences similar to this, i.e., predicates classified aspectually as Activities where the

group responds with joy after having heard the words of the Buddha, are the most common

form of Result in the overarching narration of the GLS. Auxiliary nu’s oft-remarked use as

a framing device is really due to its semantics as an endoactive perfective—many narrative

passages end with someone departing from the scene. Thus, contrasting with Fujii (2016),

60

I find it to be the predicate’s sentence-level semantics that leads to the ni (perfective nu’s

participle) in ni-ki predicates rather than a particular framing function.

The auxiliary ki, on the other hand, clearly displays a narrative function in the

kundokubun rendition of the Golden Light Sutra’s discourse. Every case of its use in the

outermost narrative is to frame the discourse, whether at the very beginning for the Abstract

and first two lines of the Orientation; between scenes to reinforce the sutra’s modality; or

at the conclusion to bring the focus back to the time of speaking. There is a preponderance

of low-transitivity predicates at these peripheries, which is why ki is so often found

governing such predicates in the GLS narration rather than having such a proclivity due to

its semantics. Narrators in early Heian kundokubun texts use this auxiliary to mark the

factuality of the sutra itself.

Auxiliaries ari and tari, being statives, are natural candidates for governing

predicates that lay out and depict the scene in which the action will occur. These scenes

are limited to Orientations and Evaluations, and those are, indeed, where we find every

instance of ari and tari interacting with the discourse in kundokubun narration.

In conclusion, the Japanese translator who glossed the Saidaiji Golden Light Sutra

used Early Middle Japanese kundokubun tense, aspect, and modality auxiliaries to move

the narrative along, vividly display and update the scenes, and frame the religious discourse

as an irrefutable truth.

61

4.2 An Embedded Parable in the Saidaiji Temple Golden Light Sutra68

In this section, we investigate the longest embedded narrative in the Saidaiji

Golden Light Sutra, the tale of Jalavāhana (Skt. meaning ‘flowing water,’ J. 流水

Rusui), which spans the 24th and 25th chapters and depicts the deeds of Jalavāhana and the

origin of the ten thousand divine sons (天子 tensi). With this section and 4.4 below, we

step away from the more quantitative discussion of transitivity, lexical aspect, and

narrative function, to develop a more qualitative analysis regarding grounding

(Fleischman 1985, Hopper 1987) and, by extension, framing, for selected embedded

parables. Out of the 267 sentences in the tale, 32 are embedded quotes depicted as being

spoken by the characters. In other words, 235 of the total 267 lines, or 88%, compose the

embedded narration discussed in this subsection.

This embedded story is depicted as being narrated by the Buddha to the

Bodhisattva Samuccayā, the ‘Goddess of the Bodhi Tree’ (菩提樹神 bodai zyusin). The

tale progresses as follows, with the translation’s tensing consistent with the kundokubun

text.

Part I: Long ago, there was a kingdom where the populace was afflicted by

various diseases. There was a great doctor, Jaṭiṃdhara, who had a son named Jalavāhana.

Jalavāhana knew his father was too old to travel the land and save the people, so he asked

Jaṭiṃdhara to teach him. Jalavāhana becomes a great doctor and travels the kingdom

healing the sick.

68 The findings in this section have largely been published in Bundschuh (2020).

62

Part II: Jalavāhana had two sons who grew up to be honorable men. They travel

together through some woods and happen upon a drying lake filled with ten thousand fish

approaching death. As Jalavāhana feels great sorrow for these creatures, a tree goddess

appears asking him to save them. He asks the king for elephants to carry water to the

lake, and the king accepts the request. Once he saves the fish, he teaches them the

dharma.

Part III: The ten thousand fish pass away and are reborn as divine sons. They

realize Jalavāhana teaching them the dharma led to their wondrous rebirth, and, with

great joy, light up the sky and rain countless flowers upon a sleeping Jalavāhana. The

next morning, the king asks his ministers why the sky was alight, and they suggest asking

Jalavāhana. The king summons him and Jalavāhana assumes the fish have died and been

reborn as heavenly beings. To confirm this, the king sends a minister along with

Jalavāhana’s sons to inspect the lake, and they find the fish had indeed passed away. The

king rejoiced upon hearing the news of this glorious rebirth. The Buddha concludes the

tale by informing Samuccayā that he was Jalavāhana in a past life, she was the king in a

past life, and the other characters in the story were also past lives of Buddhist deities and

bodhisattvas.

The following displays the TAM marking of every sentence in the tale of

Jalavāhana outside of (double-)embedded quotations, separated into the three parts

described above. As a reminder, ari and tari are stative auxiliaries, ki is a past established

fact auxiliary, and nu and tu are perfective auxiliaries.

Part I: ari-ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, ki, (TAM-unmarked), nu,

(TAM-unmarked), ki, nu, nu, ki, (TAM-unmarked), ni-ki, ki, ni-ki, ki, ki, (TAM-

63

unmarked), ki, ari-ki

Part II: ki, ki, ki, ki, nu, (TAM-unmarked), (TAM-unmarked), (TAM-unmarked),

(TAM-unmarked), tu, (TAM-unmarked), (TAM-unmarked), (TAM-unmarked), (TAM-

unmarked), (TAM-unmarked), (TAM-unmarked), (TAM-unmarked), nu, (TAM-

unmarked), (TAM-unmarked), tu, (TAM-unmarked), (TAM-unmarked), (TAM-

unmarked), (TAM-unmarked), (TAM-unmarked), (TAM-unmarked), (TAM-

unmarked), ni-ki, (TAM-unmarked), (TAM-unmarked), (TAM-unmarked), (TAM-

unmarked), (TAM-unmarked), nu, (TAM-unmarked), (TAM-unmarked), ni-ki, ki, ki,

ki, (TAM-unmarked), ki, ni-ki

Part III: tu, nu, ki, nu, tari, ari, (TAM-unmarked), nu, ni-ki, (TAM-unmarked),

(TAM-unmarked), ki, (TAM-unmarked), (TAM-unmarked), (TAM-unmarked), nu,

(TAM-unmarked), (TAM-unmarked), (TAM-unmarked), (TAM-unmarked), ki

At first glance, it is clear the narrative is framed by ki-marked sentences. However, when

we compare the three parts, we find a preponderance of past established fact ki in the first

part of the tale. The reason for this is twofold. First, Part I is essentially the Orientation of

the parable. It not only grounds the narrative in the past, but it also establishes Jalavāhana

as a healer who cares for all living beings. Second, as we discuss in section 5.3 below,

Buddhist kundokubun texts have a proclivity to ground the discourse at the beginning

with ki throughout the Orientation, then taper its use until the Result. This is also seen in

the outermost sutra narration’s grammatical marking, with the first chapter of the sutra

having the most ki marking as it establishes the setting.

When we look at Part I, we also see that it is framed by ari-ki—what Suzuki

(2009) calls past perfects (see Table 1 above). Below are the first and final lines that

frame the part.

64

(12) 過去の無量不可思議阿僧企耶の劫に、尒時に佛有して（於）世に出現（し）たま

へリキ。 (Kasuga 1985a: 174)

kako no muryau pukasigi asaugiya no kopu ni

past GEN immeasurable unimaginable incalculable69 GEN aeon LOC

so no toki ni potoke imasi-te yo ni

that GEN time LOC Buddha be.HON-GER world LOC

syutugen.si-tamape-ri-ki

appear-HON-ARI-KI

‘Immeasurable, unimaginable, incalculable eons ago a Buddha had appeared in

the world.’

(13) 是（の）如ク稱歎すること城邑に周遍せりキ。 (Kasuga 1985a: 178)

ko no gotoku syautan.suru koto zyauopu ni

this GEN like praise NML city LOC

syupen.se-ri-ki

spread-ARI-KI

‘Thus [his] praise had spread throughout the city.’

Predicates with stative auxiliary ari primarily give background information, and past

established fact ki grounds the narrative in a real past. Part I of the tale, thus framed as by

ari-ki, can be seen as the background to the rest of the story.

This backgrounding function of resultatives is reflected in the use of the tari and

ari we find in Part III in the following successive sentences.

69 阿僧企耶 asaugiya is a calque of Skt. asaṃkhya ‘incalculable’ (see Buswell & Lopez 2014: 67).

65

(14) 時に長者子は高樓の上に在（り）て安隱にして（而）睡（り）たり。 (Kasuga 1985a:

183)

toki ni tyauzyasi pa kauru no upe ni ari-te

time LOC merchant.son TOP tower GEN above LOC be-GER

annon.ni.si-te nemuri-tari

be.peaceful-GER fall.asleep-TARI

‘Then the merchant’s son [Jalavāhana], being on top of a tower, has peacefully

fallen asleep.’

(15) 曼陁羅花摩訶曼陁羅華を雨（り）て、積レルこと、（于）膝マでに至せり。 (Kasuga

1985a: 183)

mandarage makamandarage wo puri-te tumore-ru

white.lotus large.white.lotus70 ACC rain-GER accumulate-ARI

koto piza made ni itase-ri

NML knees to LOC bring-ARI

‘They have brought the rained accumulation of various white lotuses to reach his

knees.’

The State depicted in (14) lays out the scene before the divine sons rain flowers upon

Jalavāhana and (15) describes the resulting flowers’ accumulation, which is presently

manifest. Finally, in kundokubun we find the perfectives tu and nu marking events that

move the narrative forward as both realized and complete and, by extension, foreground

the events their predicates express and draw the listener into the action. They are

sprinkled among the “(TAM-unmarked)” sentences unsuffixed for tense, aspect, and

modality, which depict events as happening before our eyes. Perfectives contain this

immediacy as well, but mark sudden changes—both scene shifts, marking verbs of

movement such as yogiru (‘go across’) and character’s changes, marking verbs of

70 Skt. mandārava mahāmandārava (Soothill & Hodous 2014: 436).

66

attainment or transformation such as naru (‘become’). The end of Part II contains an

established-past-fact perfective, ni-ki, to ground the narrative in the past and foreground

the scene shift as Jalavāhana returns home after teaching the dharma to the fish.

(16) 尒時長者子流水と及其の二（り）の子とい、彼の池の魚の為に、水を施し食を施

し、并（せ）て法を說キ已（り）て、俱共に家に還（り）にキ。 (Kasuga 1985a: 183)

so no toki tyauzyasi rusui to oyobi so no

that GEN time merchant.son Jalavāhana with and that GEN

putari no ko to i ka no ike no uo no

two GEN child COM NOM this GEN lake GEN fish GEN

tame.ni midu wo podokosi ziki wo podokosi

for water ACC give food ACC give

apase-te papu wo toki-wopari-te tomo ni

bring.together-GER dharmaACC expound-finish-ger together LOC

ipe ni kaperi-ni-ki

house LOC return-NU-KI

‘Then the merchant’s son Jalavāhana and his two children, having given water,

given food, and together finished preaching the dharma for the sake of the fish of

that lake, together returned home.’

The story continues as follows with perfectives marking the tale’s progression at the

beginning of Part III, which describes the reincarnation of the fish into divine sons.

(17) 是の長者子流水い、復（於）後の時に、聚會有（り）て衆の伎樂を設（け）たるに

因（り）て、酒に醉（ひ）て（而）臥（し）つ。 (Kasuga 1985a: 183)

ko no tyauzyasi rusui i mata noti no toki

this GEN merchant.son Jalavāhana NOM also later GEN time

ni zyue ari-te moromoro no gigaku wo

LOC gathering have-GER various GEN entertainment ACC

mauke-taru ni.yori-te sake ni wepi-te pusi-tu

hold-TARI due.to alcohol by get.drunk-GER fall.asleep-TU

‘This merchant’s son Jalavāhana, furthermore at a later time, due to having a

gathering and having held a group masked performance, became drunk on alcohol

67

and fell asleep.’

(18) 時に十千の魚同時に命過して三十三天に生レヌ。 (Kasuga 1985a: 183)

toki ni zipusen no uo dauzi ni inoti

time LOC 10,000 GEN fish same.time LOC life

sugosi-te sanzipusanten ni umare-nu

pass-GER heaven71 LOC be.born-NU

‘Then the ten thousand fish passed away at the same time and were [re]born in

heaven.’

(19) 是（の）如キ念を起（し）しク、 [what they thought] とオモフトキニ、便相ヒ謂（ひ）

て曰（は）ク、 [what they said] といひキ。 (Kasuga 1985a: 183)

ko no gotoki nen wo okosi-siku [what they thought]

this GEN like thought ACC give.rise-KI.NML

to omopu toki ni sunapati api-ipi-te ipaku

COM think time LOC then together-say-GER say.NML

[what they said] to ipi-ki

COM say-KI

‘Once they gave rise to the following thought, [(what they thought)], they then

together said [(what they said)].’

In these three examples, we find, in order, (17) Jalavāhana falling asleep after a party,

marked with tu, and (18) the rebirth of the divine sons, marked by nu, before (19) the

episode is grounded in the past with ki after a long quotation discussing the reason for

their fortuitous rebirth.

In this section I have demonstrated how the tense, aspect, and modality auxiliaries

were used over the course of an embedded narrative using the GLS tale of Jalavāhana as

an example. I present a quantitative analysis of the TAM markers in the tale in section 6.2

71 Skt. trayastriṃśa, the heaven of rebirth (see Buswell & Lopez 2014: 921–922).

68

below when I compare the Saidaiji temple translation of the GLS to a near-contemporary

rendered at Tōdaiji temple. But next we turn to a different Tōdaiji text, the Ten Wheel

Sutra.

4.3 The Narration of the Tōdaiji Temple Ten Wheels Sutra

The Kṣitigarbha Ten Wheels Sutra (地蔵十輪経 Jizōjūrin kyō), which I refer to as

the Ten Wheels Sutra throughout this study, consists of a dialogue between the Buddha and

the bodhisattva Kṣitigarbha, Skt. ‘Earth Store’ (known as Jizō in Japan), who rescues those

reborn in one of the Buddhist hells. According to the preface of the Sinitic rendition, it was

translated into Chinese by the famous monk Xuanzang (玄奘 J. Genjō), who was “one of

the two most influential and prolific translators of Indian Buddhist texts into Chinese”

(Buswell & Lopez 2014: 1015).

The text examined here is a kundokubun rendition glossed at Tōdaiji temple in 883

CE on an 8th-century-CE copy of Xuanzang’s Sinitic translation, which includes the preface

detailing the importance of the sutra and the circumstances that led to him (re-)translating

it. We know the year it was glossed due to the text’s shikigo (識語), which is a “colophon

recording the date and circumstances of glossing” (Whitman et al. 2010). Although

originally ten scrolls, only eight remain today. Scrolls 1, 2, 4, 8, 9, and 10 are housed in

the Tōdaiji temple library and scrolls 5 and 7 are housed in Shōsō’in (the Tōdaiji temple

repository, or treasure house), while scrolls 3 and 6 are lost. The Tōdaiji temple library data

were gathered from Nakada (1958) and the Shōsō’in scrolls are found in Nakada (1980).

69

Although most early Heian glossings were composed as notes during a single

lecture, the glossings on this text were written over at least three sittings, as evidenced by

a notation at the end of the fifth scroll.72 This careful, repeated glossing suggests the monk

who produced the kundokubun rendition was aware that such glosses were becoming

necessary for others to access the Sinitic source text (Nakada 1979: 776–779). In this

subsection, I examine the tense, lexical aspect, and discourse function of sentence (i.e.,

matrix clause) predicates governed by the TAM auxiliaries ki, keri, nu, tu, ari, and tari

deployed by the outermost narrator of the Ten Wheels Sutra, which I refer to as the TWS

for the duration of this chapter.

Within this section, whenever I use the terms “narrator” and “narration,” I

specifically mean that of the outermost narrative (defined and discussed in section 1.5

above), where we find the portrayal of the conversation between the Buddha and his

followers, outside of such embedded quotations as occur. There are 299 lines in this

outermost narration out of 2443 total lines in the sutra. In other words, the matrix narrative

takes up approximately 12% of the total lines of the sutra, the rest of the lines being

dialogue. As with the GLS narration investigated above, nominalized speech-act predicates

that precede quotations such as ipaku (‘that which was said’) and notamapaku (‘that which

was honorably said’) are included in the total line count. This allows us to count the

72 The notation reads 三交了, which could be interpreted as mitu kuwape-woparu ‘finished adding three

times,’ or, more likely according to Nakada (1979: 778), mitu kangape-woparu ‘finished considering three

times,’ with the character 交 ‘add’ being a contraction of 校 ‘consider.’

70

sentences found in dialogue before the post-quotation sentence-final speech-act

predicates.73

The table below gives the total times the TAM auxiliaries are used throughout the

outermost narration of the TWS, how many of those are sentence-final, and their sentence-

final usage percentage, rounded to the nearest percent.

Table 7 TAM Marking Totals in TWS Narration

TAM Total Sentence-Final Total Sentence-Final %

ki 29 19 65%

keri 0 0 0%

nu 13 11 85%

tu 21 16 76%

ari 31 16 52%

tari 17 5 29%

Among the TAM markers, the perfectives nu and tu are the most likely to be found sentence-

finally and thus interact with the discourse, by adding to it. Sentence finally, the narrator

predominantly employs nu and tu at 85% and 76%, respectively, followed by ki at 65%.

The stative ari is more evenly distributed, appearing sentence-finally 52% of the time.

Auxiliary tari is most often used sentence-internally, modifying nouns in its adnominal

73 One could, for example, count the following as a single line of the outermost narrative.

佛言は（く）、“[15 lines of quotation]” （と）のたまふ (Nakada 1958: 79)

Potoke notamapaku [quotation] to notamapu

Buddha that.which.he.honorably.says COM honorably.says

‘The Buddha honorably said, “[15 lines of quotation].”’

For the purposes of this study, this counts as 17 lines—the prequotative nominalized speech-act predicate,

the 15 lines of the quotation, and the speech-act predicate ending the sentence.

71

form, and is found in sentence-final position only 29% of the time. Auxiliary keri never

appears in the outermost narrative, except when the narrator quotes someone else’s use of

it, i.e., in embedded dialogue or narration. Before focusing on sentence-final predicates

below, we first look at some examples of how these TAM markers are used sentence-

internally, for reference.

Below are the first and final examples of sentence-internal ki in the TWS.

(20) 一雲に普ク洽シて（而）卉木各茂リ、（自）鶴林に色を變へ慧の日光を寢メシ（とき

より）、達學電のゴトクに謝して、（以）肩を息メ、眞人長ク往（き）て（而）慮を寂メ

たり。 (Nakada 1958: 1)74

itiun ni amaneku uruposi-te kiboku onoono

one.cloud ABL completely spread-GER vegetation all

sigeri kakurin ni iro wo kape we

growing.thick Crane.grove75 LOC color ACC changing enlightenment

no pi pikari wo yasume-si toki yori datugaku

GEN sun light ACC give.a.rest-KI time since successful.study

inaduma no gotoku ni sya.si-te kata wo pisome

lightning GEN like COP disappear-GER shoulder ACC resting

sinnin nagaku yuki-te mune wo pisome-tari

true.person76 long go-GER thoughts ACC quiet-TARI

‘Since the time that from a single cloud all the vegetation spread and grew thick,

the color in the Crane grove changed, and the sun of enlightenment set [its] light to

74 See Nakada (1958: 11) and Nakada (1980: 170–171) for orthographical conventions. Nakada assumes an

embedded ki (in its adnominal form si) in the line previous to this one (照（らししとき）に terasi-si toki ni

‘when [the morning sun] shone’), in (41) below. I do not include it here due to lack of evidence that the

glossed LOC particle (に ni) on the Sinitic character 照 was preceded by the auxiliary ki in this case.

75 Where the historical Buddha died, “when the trees burst into white blossom resembling a flock of white

cranes” (Soothill & Hodous 2014: 486).

76 This is another term for Skt. arhat (Soothill & Hodous 2014: 331).

72

rest, successful study disappeared like lightning, resting their shoulders, those who

embody truth have long ago quieted their thoughts.’

(21) 時（に）、薄伽梵是の經を說き已（り）たまひシカば、（於）衆會の中の [. . .] 一切

の大衆、佛の所說を聞（き）たまへて、皆大に歡喜し信受し奉行しき。 (Nakada

1958: 145–146)

toki ni bagabon ko no kyau wo

time LOC bhagavat this GEN sutra ACC

toki-wopari-tamapi-sika-ba syu’e no naka no [. . .]77

expound-finish-HON-KI-COND assembly GEN among GEN

issai no daisyu potoke no syosetu wo kiki-tamape-te

all GEN group Buddha GEN words ACC hear-HUM-GER

mina dai ni kwangi.si sinzyu.si bugyau.si-ki

all great COP being.overjoyed accepting.truth enact-KI

‘Then, since the lord had finished expounding this sutra, all those [. . .] within the

group, having humbly heard the Buddha’s doctrine, were greatly overjoyed,

accepted the truth, and dutifully carried it out.’

We find ki in its adnominal form si in (20) because it modifies the noun toki (‘time’). This

is marking the resting of the sunlight as a fact established prior to the scene, with sentence-

final tari heading the resulting state of quietness that has held true since that time. The

narrator of the translator’s Preface (i.e., Xuanzang) is using this metaphor to begin a

discussion on correct Buddhist practice, which has become difficult since the death of the

historical Buddha. In (21), on the other hand, we find ki in its realis form governed by the

conditional suffix ba at the head this subordinate clause. This thus marks the condition, or

circumstances, given which the following consequence occurred, i.e., upon the Buddha’s

having finished expounding the sutra, his assembled followers carried out the doctrine.

77 The transcription has elided a long list of bodhisattvas and other great beings here.

73

This example ends with TAM marking as well, with ki capping the final line of the sutra, a

pattern that is discussed further in section 5.3 below.

As mentioned above, the perfective auxiliaries are predominantly used sentence-

finally. Although the endoactive perfective auxiliary ni has two non-sentence-final tokens

in the outermost narration of the TWS, only one was clearly written by the early Heian

period glosser, the one given below.

(22) 尒（の）時（に）、世尊、是の頌を說き已（り）たまひヌ（る）トキに、（於）衆会の中に

无量百千の衆生有り。 (Nakada 1980: 205)

so no toki ni seson78 ko no zyu wo

that GEN time LOC lord this GEN verse79 ACC

toki-wopari-tamapi-nuru toki ni syue no naka ni

expound-finish-HON-NU time LOC gathering GEN in LOC

muryau hyakusen no syuzyau ari

immeasurable hundred.thousand GEN sentient.beings be

‘At that time, when the Lord had finished expounding this verse, there were

immeasurable hundreds of thousands of sentient beings in the gathering.’

The only other time nu is found in a subordinate clause in the TWS kakikudashi80 rendered

by Nakada (1958, 1980), it represents an assumption he made due to the predicate being in

its participial form and followed by particle wo, which suggests that an adnominally

inflected auxiliary follows that participle. Since perfective nu takes participles as its

78 Skt. bhagavat, often rendered as ‘Lord’ in English (Buswell & Lopez 2014: 108).

79 Skt. gāthā, the hymns in sutras (Soothill & Hodous 2014: 418).

80 A written-out transcription according to the kunten (‘glossing’) markup of the text (introduced in section

1.5).

74

complement and is also appropriate for a recently realized event, Nakada makes an

adnominal nuru explicit. The token is displayed below.

(23) 地藏菩薩摩訶薩の无量の稱讚の功德（を）成就セ（る）ことを說（き）たまひ（ぬ

る）を聞（き）たまへて [. . .] (Nakada 1958: 29)

dizau bosatu makasatu no muryau no

Kṣitigarbha bodhisattva mahāsattva81 GEN immeasurable GEN

syausan no kudoku wo zyauzyu.se-ru koto wo

praise GEN good.deeds ACC fulfill-ARI NML ACC

toki-tamapi-nuru wo kiki-tamape-te [. . .]

expound-HON-NU ACC hear-HUM-GER

‘Having humbly heard the expounding [by the Buddha] of the great bodhisattva

Kṣitigarbha’s having fulfilled the immeasurable praise-worthy good deeds, [. . .]’

The parentheses in the kakikudashi mean these morae are the educated guess of the scholar

producing the more accessible text and are not found in the original glossing. They clearly

follow the language’s morphotactics, or constraints on which inflected forms precede (i.e.,

can be the complements of) which auxiliaries. In other words, what is written 說（き）たま

ひ（ぬる） in Nakada (1958) might be romanized as to(ki)tamapi(nuru) and broken down

morphologically more carefully as expound(.participle)-HON.participle(-NU.adnominal)

using only what we have available from the source text. Because the toki-tamapi predicate

is followed by wo kiki-tamape-te (‘having humbly heard’), we can assume that (a) what

was heard was aspectually realized and (b) referred to deictically (i.e., in the adnominal

form). The first example of sentence-internal auxiliary nu in (22) above is from the seventh

81 The Sanskrit term mahāsattva ‘great being’ is “a standard epithet of an advanced bodhisattva” (Buswell

& Lopez 2014: 508).

75

scroll of the sutra, to which Nakada had not yet had access when he produced his 1958 text.

His assumption that the perfective in (23) was nu is confirmed by more fully marked tokens

such as what we see in (24) below, from earlier in the sutra—of which Nakada would of

course have been aware.

(24) 世尊（い）是（の）地藏菩薩の諸の功德を說き已（り）たまひヌ。 (Nakada 1958: 8)

seson i ko no dizau bosatu no moro no

lord NOM this GEN Kṣitigarbha bodhisattva GEN all GEN

kudoku wo toki-wopari-tamapi-nu

good.deeds ACC expound-finish-HON-NU

‘The lord finished expounding on all the good deeds of this bodhisattva

Kṣitigarbha.’

As I discuss in more detail below, “speech-act” verbs, such as toku ‘expound,’ are more

likely governed by perfective tu in kundokubun. The use of nu in the tokens cited above

(and elsewhere) may be due to the Sinitic character 已 in the predicate, which was

strongly associated with auxiliary nu.82 I am therefore hesitant to include the assumed nu

in (23) in my analysis and do not discuss it further.

The exoactive perfective auxiliary tu, on the other hand, is clearly found sentence-

internally, although it is never found modifying nouns. Below are two examples where we

find tu in non-sentence-final positions in the TWS.

(25) 光明に照（らさるる）ガ故に、願に隨（ひ）て皆得ツ（といふ）ことをミ、 (Nakada

1958: 5)

82 In early Heian texts, the Sinitic character 已 was read as the auxiliaries nu, tu, and tari depending on

the context, but by mid Heian, other than a few tari examples, it was primarily read as the auxiliary nu or the

verb woparu (‘to finish/be complete’) (Kobayashi 2017: 35). By late Heian most Sinitic characters came to

be read as stand-alone words, and, accordingly, 已 generally lost its auxiliary nu reading and was

predominantly rendered as the independent verb woparu (Kobayashi 2013: 1194).

76

kaumyau ni terasa-ruru ga yuwe ni gwan ni

shining.light ABL shine-PASS GEN reason LOC wish LOC

sitagapi-te mina e-tu to ipu koto wo mi

follow-GER everyone gain-TU COM say NML ACC see

‘having seen [him] saying that everyone, following their wishes, gains [it] due to

being bathed in shining light, . . .’

(26) 尒（の）時（に）、一切の諸來の大衆（い）、既に地藏菩薩摩訶薩を見已（り）て、

皆、希奇（なる）ことを獲、未曾有（なる）ことを得てき。 (Nakada 1958: 10)

so no toki ni issai no syorai no daisyu

that GEN time LOC all GEN those.who.came GEN group

i sude.ni dizau bosatu makasatu wo mi-wopari-te

NOM already Kṣitigarbha bodhisattva mahāsattva ACC see-finish-GER

mina kiki naru koto wo e mizou naru koto

everyone rare COP thing ACC gain unprecedented COP thing

wo e-te-ki

ACC gain-TU-KI

‘At that time, having already finished seeing the great bodhisattva Kṣitigarbha, the

entire group who had come all gained something that was rare and gained

something that was unprecedented.’

The first example above is part of a description of the gathering listening to the Buddha’s

sermon near the beginning of the TWS. The rest of the sentence depicts them seeing

wonderful music filling the heavens. While in (25) we do find tu in its sentence-final, or

finite, form, it is followed by the complementizer particle to, which means it is, narrative-

wise, subordinate—operative at a level below that of the ongoing discourse. The only

example of tu in its participial form te in the outermost TWS narration is that in (26).

Although part of the finite matrix predicate, it is governed by ki. This combination, te-ki,

is defined as a ‘past perfective’ (完成相の過去形 kanseisō no kakokei) by Suzuki (2009:

77

398) (see Table 1 above). Auxiliary tu is used here due to its affinity with the transitive

telic, or change-of-state, verb u ‘gain/acquire (it)’ that it governs, and ki was selected to

end the sentence due to this sentence’s position/role in the narrative structure, i.e.,

concluding an episode. Evidence for this interpretation is presented in the present and

following chapters, below.

For the remainder of my analysis of TAM marking in the TWS in section 4.3, I focus

exclusively on the outermost narration, i.e., all lines outside of any quotations.

4.3.1 Transitivity in TWS Narration

The table below depicts the relationship between TAM markers and the transitivity

scores of sentence-final predicates in the outermost narrative of the Ten Wheel Sutra. Just

as in the GLS analyzed above, there were no TAM-marked predicates of high transitivity in

the narrative of the TWS.

Table 8 The Transitivity of TAM-marked Predicates in TWS Narration

TAM low (%) medium (%)

ki 11 (58%) 8 (42%)

nu 9 (82%) 2 (18%)

tu 0 (0%) 16 (100%)

ari 10 (63%) 6 (38%)

tari 1 (20%) 4 (80%)

At first glance, auxiliary ki appears to be the least constrained among the TAM markers with

regard to transitivity. The stark difference between tari’s percentages vis-à-vis transitivity

78

may be due to its lower sentence-final total count. We see that both ki and tari share a

difference of exactly three instances between low and mid transitivity sentence-final

predicates, although ki is more likely to govern the former and tari the later.

Below are representative examples of ki heading a low-transitivity predicate and a

medium-transitivity one.

(27) 時に諸の聲聞、及菩薩衆歡喜して佛を禮（し）て、復、本座に還しき。 (Nakada

1980: 215)

toki ni moro no syaumon oyobi bosatu syu

time LOC all GEN disciple83 and bodhisattva group

kwangi.si-te potoke wo rei.si-te mata ponsyo ni

be.overjoyed-GER buddha ACC bow-GER also original.place LOC

kwan.si-ki

return-KI

‘Then the group of all the disciples and bodhisattvas were overjoyed, bowed to the

Buddha, and returned to their original place.

(28) 復、有る八十四百千那庾多の菩薩は、隨順の法忍を證得しき。 (Nakada 1980:

221)

mata aru patijipusi pyakusen nayuta no

also some 84 hundred.thousand nayuta84 GEN

bosatu pa zuizyun no papunin wo

bodhisattva TOP follow.faithfully GEN dharma.patience85 ACC

syautoku.si-ki

attain.truth-KI

‘Also, some uncountable number of bodhisattvas attained the patience by faithfully

following the dharma.’

83 Skt. śrāvaka (see Buswell & Lopez 2014: 850).

84 This is a Skt. term indicating a large number such as ‘ten million’ (Soothill & Hodous 2014: 247).

85 This means ‘patience attained through dharma to overcome illusions’ (Soothill & Hodous 2014: 269).

79

In the above examples we find auxiliary ki governing an intransitive verb of motion,

kwan.su ‘return’ and the transitive verb syautoku.su ‘attain truth,’ which takes the truth

attained as a direct object in the accusative case. Because the ‘dharma patience’ that was

attained did not change on account of the bodhisattvas attaining it, this was coded as a

medium-transitivity predicate. We see in section 4.3.3 below that ki-marking is occurring

in these sentences due to their discourse function. Next, we turn to the only instance of

sentence-final tari in a low-transitivity predicate in the TWS narration.

(29) 如來の所說菩薩の所傳、已來未來、一朝に備に集（り）たり。 (Nakada 1958: 2)

nyorai no syosetu bosatu no syoden irai

tathāgata GEN explanation bodhisattva GEN transmission past

mirai ititeu ni tubusa.ni atumari-tari

future at.once LOC in.detail gather-TARI

‘The explanations of the Buddhas and the transmissions of the bodhisattvas, the

past and the future, have gathered in detail in one moment.’

The following example is a case of tari in a medium-transitivity predicate in the TWS

Narration.

(30) （以）今翻セ（る）所をば諸の舊本に比するに、舊き本に已に有（る）をば、今更

（に）詳明、舊き本に無き所を、斯の文に具に載セたり。 (Nakada 1958: 3)

ima pirugapese-ru tokoro wo ba moro no puruki pon ni

now translate.ARI place ACC TOP all GEN old text LOC

pi.suru ni puruki pon ni sude ni aru wo ba

compare LOC old text LOC already COP have ACC TOP

ima sara ni akiraka.ni akame puruki pon ni naki

now further COP clearly make.clear old text LOC lack

tokoro wo so no mon ni tubusa.ni nose-tari

place ACC that GEN text LOC in.detail add.TARI

80

‘When we compare what has now been translated with all the old texts, that which

was already in the old texts has been made even clearer and that which was lacking

in the old texts has been added to its rhetoric in detail.’

Both tari examples above are part of the preface to the sutra written by the translator

Xuanzang (玄奘 J. Genjō). In the low-transitivity example, he is commenting on the

importance of the text’s message. In the medium-transitivity example, he is explaining how

he has improved upon earlier Chinese renditions of the TWS by having ‘added’ (nose-tari)

‘that which was lacking in the old texts,’ to which the Japanese renderer added the

accusative marker wo.

In the above subsection I have examined a case of a medium-transitivity sentence

headed by the endoactive perfective auxiliary nu and argued it was because of the auxiliary

verb woparu read from the sinograph 已, which was strongly associated with both woparu

and nu, that led to nu instead of tu. The other medium-transitivity nu in the outermost

narrative is the following.

(31) 故ニ此の經に能（く）臭き身を濯ぎ盲目を開キ、坏器を陶シ石田を沃（し）たまひ

（ぬ）。 (Nakada 1958: 1)

yuwe ni ko no kyau ni yoku kusaki mi wo

thus COP this GEN sutra LOC successfully fetid body ACC

susugi maumoku wo piraki paipin wo dau.si

cleansing blind.eye ACC opening raw.clay86 ACC firing87

sekiden wo oku.si-tamapi-nu

rocky.field ACC make.fertile-HON-NU

86 I.e., ‘unfired pottery.’

87 I.e., ‘making porcelain.’

81

‘Thus, in this sutra, [the Buddha] succeeds in cleansing fetid bodies, opening blind

eyes, firing raw clay, and making rocky fields fertile.’

Although we find a string of transitive participles with accusative markers, Nakada (1958)

assumes the endoactive perfective nu governs the final participial form oku.si-tamapi

(‘making fertile’).88 I suspect this is due to the quality of unexpected attainment indicated

with the adverbial yoku (能く) when applied to acts not normally doable, which can be

interpreted as ‘successfully,’ at the beginning of the string of predicates. This adverbial

shifts the focus from the acts themselves to the fact that the acts occurred, which negates

the volitionality of the verb phrase, thus lowering its transitivity (see Hopper & Thompson

1980: 252). This use of nu here rather than tu shows the translator kept the semantics of

the sentence itself—perfective and non-controlled, expressing the unlikeliness of the

successful act—in mind when rendering it in Japanese.

The exoactive perfective auxiliary tu, on the other hand, heads clearly transitive

predicates as we have come to expect, with 100% of predicates it heads displaying medium

transitivity. The following example is representative.

88 As mentioned above, we know this is Nakada’s deduction due to the parentheses around the nu, which we

see as “(ぬ)” in his kakikudashi (‘Japanese transcription based on the glossed Sinitic text’, introduced in

section 1.5).

82

(32) 初に无上正等覺の心を發し、不退轉を得ツ。 (Nakada 1958: 58)

pazime ni muzyau syautougaku no kokoro

first COP unsurpassed complete.perfect.enlightenment89 GEN mind

wo hatu.si putaiten wo e-tu

ACC resolving nonretrogression90 ACC attain-TU

‘For the first time, they resolved their minds to unsurpassed, complete, perfect

enlightenment and attained nonretrogression.’

As discussed above, the verb u (‘attain (it)’) is often governed by perfective tu to index the

realization and completion of the action expressed as its “direct object.” Next, we examine

the relationship between lexical aspect and TAM marking in the TWS.

4.3.2 Lexical Aspect in TWS Narration

The table below depicts the relation between TAM marker and lexical aspect scores

in the outermost narrative of the TWS.

Table 9 The Lexical Aspect of TAM-marked Predicates in TWS Narration

TAM State (%) Activity (%) Accomplishment (%) Achievement (%)

ki 3 (16%) 4 (21%) 2 (11%) 10 (53%)

nu 0 (0%) 0 (0%) 1 (9%) 10 (91%)

tu 0 (0%) 0 (0%) 1 (6%) 14 (94%)

ari 16 (100%) 0 (0%) 0 (0%) 0 (0%)

tari 5 (100%) 0 (0%) 0 (0%) 0 (0%)

89 These two words are a common kundokubun rendition of Skt. anuttarasamyaksaṃbodhi, broken down as

“unsurpassed (anuttara), complete (samyak), and perfect enlightenment (saṃbodhi)” (Buswell & Lopez

2014: 55).

90 Skt. avaivartika (Buswell & Lopez 2014: 82).

83

Auxiliary ki is the only TAM marker used in the TWS narration that is unrestricted among

the four categories of lexical aspect, which suggests it is not a limiting factor in ki’s

employment. We first turn to one example of its use in each category.

(33) 屬て三藏玄奘法師（といふ）者有しき。 (Nakada 1958: 1)

sitagapi-te sanzau genzyau papusi to ipu pito

follow-GER canon91 Xuanzang master COM call person

imasi-ki

be.HON-KI

‘Thus, there was a person called Xuanzang, master of the Buddhist canon.’

The above example is of ki in a sentence-final predicate that is a State, imasiki (有しき)

‘there was.’ This use of ki is common when introducing characters in early Heian

kundokubun.

(34) 佉羅帝耶山、普ク皆震動し、俱胝の天の樂（不）鼓ラ（さ）ズ（ある）に自（ら）鳴

（り）き。 (Nakada 1958: 27)

gyaratiyasen amaneku mina sindau.si gutei no ten

Mt. Kharādīya92 everywhere all shaking crore93 GEN heaven

no gaku narasa-zu aru ni onozukara

GEN instrument sound(TR)-NEG be LOC naturally

nari-ki

sound(NTR)-KI

‘Mt. Kharādīya shook all over and, while no one sounded the ten million

instruments of heaven, they sounded on their own.’

91 The Buddhist canon, Skt. tripiṭaka ‘three baskets’ (see Buswell & Lopez 2014: 924).

92 The abode of Kṣitigarbha where the Buddha is claimed to have uttered the TWS. This is generally called

Karadasen (伽羅陀山) in Japanese today.

93 A very large number often rendered in as ten million or one hundred million, Skt. kauṭi.

84

The sounding of instruments is depicted here as an Activity accompanied by the shaking

of Mt. Kharādīya in response to Kṣitigarbha’s words to the Buddha, and all of this is

situated in the past with the use of the ki.

(35) 皆大に歡喜して同（じ）ク善哉なりと唱（へ）き。 (Nakada 1958: 58)

mina dai ni kwangi.si-te onaziku zen’ya nari

all greatly COP be.overjoyed-GER uniformly excellent COP

to tonape-ki

COM cheer-KI

‘Being greatly overjoyed, [they] all uniformly cheered “That is excellent!”’

Without a separately specified endpoint, the act of cheering is an Activity. However, in this

context of cheering a defined phrase, “That is excellent!” in response to the Buddha’s

words in this case, it gains an endpoint and becomes an Accomplishment, or a durative and

telic event. The auxiliary ki is also found in predicates without duration in the TWS, such

as the following example.

(36) 若し（は）、時來ル（こと）數（に）在リ、藥性、違（ふ）こと勿キことを識リキ。

(Nakada 1958: 1)

mosi pa toki kitaru koto kazu ni ari

or TOP time come NML number LOC be/have

yakusyau tagapu koto naki koto wo siri-ki

medicinal.properties different NML be.NEG NML ACC learn-KI

‘Or [with] the opportunity frequently presenting itself, [one] learned that there was

no difference in the medicine’s properties.’

The verb siru ‘learn, realize’ is telic and not durative, i.e., it is a change-of-state verb. We

find it here describing one becoming aware of a fact about medicinal properties.

85

We find the auxiliaries nu and tu most often heading Achievements as well, at 91%

and 94%, respectively. This makes sense given they themselves most basically mark a

change of state. In the examples below, each is used to head an Accomplishment, or change

of state after some duration. We saw an example of nu with an Accomplishment in (24)

above, which I reproduce here.

(37) 世尊（い）是（の）地藏菩薩の諸の功德を說き已（り）たまひヌ。 (Nakada 1958: 8)

seson i ko no dizau bosatu no moro no

lord NOM this GEN Kṣitigarbha bodhisattva GEN all GEN

kudoku wo toki-wopari-tamapi-nu

good.deeds ACC expound-finish-HON-NU

‘The lord finished expounding on all the good deeds of this bodhisattva

Kṣitigarbha.’

The one instance of sentence-final tu with an Accomplishment is when the assembly

witnesses flowers falling from the heavens and music filling the air after the words of the

Buddha.

(38) 諸の妙なる樂具其の中に充滿（すと）いふ（ことを）見ツ。 (Nakada 1958: 5)

moro no tape.naru rakugu so no naka ni

all GEN wonderful instrument that GEN within LOC

ziuman .su to ipu koto wo mi-tu

fill COM say NML ACC see-TU

‘They saw all the wonderful musical instruments fill within it.’

The scene viewed itself is durative, but the viewing of the scene comes to a conclusion,

making this particular event of seeing an Achievement rather than an Activity.

86

The stative auxiliaries ari and tari both are, understandably due to their semantics,

only found sentence-finally in clauses depicting States, such as the following.

(39) 復、有（る）ガ說きて言ヘ（り）。 (Nakada 1980: 212)

mata aru ga toki-te ipe-ri

also some GEN expound-GER say-ARI

‘Also, some had said [the following], expounding.’

(40) 達學電のゴトクに謝して、（以）肩を息メ、眞人長ク往（き）て（而）慮を寂メたり。

(Nakada 1958: 1)

datugaku inaduma no gotoku ni sya.si-te

successful.study lightning GEN like LOC disappear-GER

kata wo pisome sinnin nagaku yuki-te

shoulder ACC resting true.person long go-GER

mune wo pisome-tari

thoughts ACC quiet-TARI

‘[. . .] successful study having disappeared like lightning, resting their shoulders,

those who embody truth have long ago quieted their thoughts.’

The example of auxiliary ari in (39) above is interesting in that we find an unusual use of

genitive ga, without a clear noun to modify. In the Sinitic source text, (39) is found in a

long set of parallel structures in which different groups say something in turn introduced

with the same phrase, 復有說言, before each quote. Most of these are rendered 復、有

（る）ガ說きて言はく (mata aru ga tokite ipaku ‘Also, what some said, expounding’), using

the deictic V-aku ‘what [they] V’ nominalization that is often found before quotes in

kundokubun. In this case, the verb phrase tokite ipu ‘say, expounding’ is being nominalized

and attributed to the noun phrase aru ‘some who are present’ with the genitive particle ga.

87

The genitive ga seen in this aku noun phrase remains after aku was replaced with auxiliary

ari, even though it is in a finite form.94

The example for tari, (40), was introduced earlier in the latter half of (20). It depicts

a resulting state from a change, a perfect. The tensing difference in the English renditions,

with past tense in the ari example (39) and present tense in the tari example (40), is due to

the surrounding discourse. The ari example (39) comes within a passage that is grounded

in the past with the auxiliary ki, whereas the tari example (40) is in its story’s Abstract, at

the beginning of the sutra, and thus presented, as is typical with Abstracts, without explicit

modal or tense grounding.

In sum, while the perfective auxiliaries nu and tu are found in telic predicates

(Achievements and Accomplishments) and the stative auxiliaries ari and tari are found in

States, the established fact auxiliary ki displays no such restriction.95 The relation between

these and the other TAM markers and the sutra’s narrative discourse is discussed in the

following subsection.

94 The continued use of genitive ga when nominalized predicates became reinterpreted as finite predicates

in Late Middle Japanese led to the nominative ga we find in Japanese today (Frellesvig 2010: 366–368).

95 Auxiliary ki is less semantically “relevant” (Bybee 1985: 15) to what it governs than the perfective and

stative auxiliaries in that it indexes the speaker’s stance toward the information conveyed in the clause rather

than the lexical content of the verb stem. This is discussed further in section 5.3 below.

88

4.3.3 Discourse Function in TWS Narration

Table 10 below depicts the relation between each TAM marker and the narrative

discourse function of the sentences that it heads. As mentioned above, the Tōdaiji temple

TWS begins the first scroll with a Preface to the Chinese translation. I interpret this as a

long Abstract due to its details regarding some of the events depicted in the body of the

sutra, in addition to the translator Xuanzang’s reflections on the translation itself.96 In this

section I focus on the Abstract and compare the data of the other discourse functions with

those from the Golden Light Sutra, discussed in section 4.1.3 above.

Table 10 The Discourse Function of TAM-marked Predicates in TWS Narration

TAM Abstract (%) Orientation (%) Complication (%) Evaluation (%) Result (%)

ki 3 (16%) 3 (16%) 3 (16%) 0 (0%) 10 (53%)

nu 2 (18%) 0 (0%) 7 (64%) 0 (0%) 2 (18%)

tu 1 (6%) 0 (0%) 13 (81%) 0 (0%) 2 (2%)

ari 9 (56%) 0 (0%) 0 (%) 7 (44%) 0 (0%)

tari 4 (80%) 0 (0%) 0 (%) 1 (20%) 0 (0%)

To summarize the above results, we find ki primarily in Results at the end of narrative

passages, nu and tu in Complications in which the story progresses, and ari and tari in

sentences that comprise an Abstract. However, every TAM-marked sentence that is

functionally part of an Abstract is found in the translator’s Preface. Although not part of

the original Buddhist Hybrid Sanskrit text, in the Sinitic rendition the Preface is titled as

part of the sutra (大乘大集地藏十輪經序 daizyau daisipu dizau zipurinkyau zyo ‘Great

96 While Labov (1972: 363) defines the Abstract as a summary of the story, I follow Nara (2011: 282) in

classifying the Abstract more broadly, as ‘what the story is about.’

89

vehicle great gathering earth store ten wheels sutra preface’) and presented

orthographically in the same way as subsequent chapters of the text. Of all the sections of

the TWS’s outermost narration, the Preface contains the richest variety of TAM marking in

its Japanese rendition. It includes tokens of nu (ex. (31)), tari (ex. (40)), and ki (ex. (33)

and (36)), presented above. Below, as (41), is the first line of the Preface, which also is

headed by ki. Following it, as (42) and (43), are two lines near the Preface’s conclusion,

which are headed by tu and ari, respectively.

(41) 昔者、旭は高山を照（らししとき）に、天の宮（に）一乘の（之）駕を御メ、暉を原隰

に流（し）て、鹿苑（に）四諦の（之）輪を轉（じ）たまひき。 (Nakada 1958: 1)

mukasi pa asapi pa takayama wo terasi-si

long.ago TOP MORNING.SUN TOP tall.mountain ACC shine.upon-KI

toki ni ten no miya ni itizyau no ka

time LOC heaven GEN palace LOC one.vehicle97 GEN carriage

wo wosame ki wo genzipu ni

ACC handle ray.of.light ACC highland.wetland LOC

nagasi-te rokuwon ni sitai no

make.flow-GER Deer.Park98 LOC four.noble.truths99 GEN

rin wo ten.zi-tamapi-ki

wheel ACC turn-HON-KI

‘Long ago when the morning sun shone upon the highest mountains he turned the

wheels of the four noble truths at Deer Park, handling the single-wheeled cart and

making rays of light flow over the highlands and wetlands.’

97 Skt. ekayāna. The ‘single vehicle’ refers to the teaching that leads to Buddhahood (Buswell & Lopez

2014: 281–282).

98 Where the historical Buddha preached his first sermon (Soothill & Hodous 2014: 367).

99 More accurately understood as ‘four truths known by noble ones.’ See Buswell & Lopez (2014: 304–305).

90

(42) （於）是に翰を染メ紙を操リテ外慮を杜ギツ (Nakada 1958: 2)

koko ni pude wo some kami wo tori-te

here LOC brush ACC ink paper ACC take-GER

gwairyo wo pusagi-tu

outside.thoughts ACC shut-TU

‘Here, having inked my brush and taken my paper, [I] block out outside thoughts.’

(43) 喜足謙懷の（之）侶（は）高き節を（於）清き風に騰レリ（矣）。 (Nakada 1958: 3)

kisoku kenwe no tomogara pa takaki setu

satisfied humble GEN comrade TOP high principles100

wo kiyoki kaze ni agare-ri

ACC pure wind LOC raise-ARI

‘The satisfied and humble comrades have risen, [taking] their lofty principles onto

the pure winds.’

The narrator of the Preface (Xuanzang) first begins by explaining the reason for this sutra

to be called the Ten Wheels Sutra (in addition to it having ten scrolls) in (41). The glosser

rendered this complex sentence’s matrix predicate with auxiliary ki due to its depiction of

a past action of the Buddha, which is taken as religious truth. The Preface continues

decrying the current state of ‘final dharma’ (J. mappō 末法)101 and the importance of

understanding the Ten Wheels Sutra, leading to (42), in which the Sanskrit-to-Chinese

translator Xuanzang comments on his process of translating the text, stating that he blocks

out external thoughts in order to properly focus on the task of writing at hand. This line is

headed with auxiliary tu in the EMJ rendition because Xuanzang describes his completed,

actualized, purposeful action in blocking out these thoughts. Finally, in (43) he comments

100 Cf. Kōjien entry for 高節 (kōsetsu), defined as ‘refined principles’ (kedakai sessō けだかい節操).

101 See “mofa” in Buswell & Lopez (2014: 545).

91

on the praise his translation has received, using the auxiliary tari to describe the state of

his Buddhist colleagues. I offer a hypothesis as to why the preface contains such rich

morphological markings in the conclusion of this subsection (4.3.4).

Moving on to Orientations, we only find three tokens of auxiliary ki heading

sentences that fulfill this discourse function. All three are in the first lines of the first

chapter, in which the location of the Buddha’s preaching and his audience are introduced.

Unlike the Golden Light Sutra, this first chapter of the TWS begins with a very brief

Orientation before moving on to the didactic conversation without Orientations in the

subsequent chapters, which leads to the smaller token count we find here.

The data on Complications, Evaluations, and Results are similar to those we found

in the Golden Light Sutra. Auxiliaries ki, nu, and tu are used in Complications and Results

and ari and tari, in Evaluations. As for perfective tu in the Result function, although we

did not find it heading Result sentences in the GLS data, there are two tokens of it doing so

in the TWS narrative, such as the following example.

(44) 復有（る）无量无數の有情は、逮て果證を得ツ。 (Nakada 1958: 58)

mata aru muryau musu no iuzyau pa oyobite

also some immeasurable infinite GEN sentient.being TOP also

kwasyau wo e-tu

result.in.realization AC attain-TU

‘Also, some among the immeasurable infinite sentient beings also attained the

result of realization.’

This is the final line of the second chapter of the TWS, which is why it is coded as a Result.

Furthermore, while the third chapter and, thus, the subsequent line after (44), is missing

from the Tōdaiji rendition, it does begin with 爾時 in the Sinitic text. This is usually

92

rendered in J. as So no toki ni (‘at that time’) and is how new scenes generally begin, with

the preceding line being a Result. I now turn to the concluding remarks regarding the

outermost narration of the TWS.

4.3.4 Conclusions Regarding TWS Narration

In section 4.3, I have presented data on the relationship between the TAM auxiliaries

and the transitivity, lexical aspect, and narrative discourse function of the sentences they

head throughout the outermost narrative of Ten Wheel Sutra. Table 11 below summarizes

this section’s findings. Note that auxiliary keri is not found sentence-finally throughout this

narrative frame and therefore was not included in these analyses.

Table 11 Summary of Findings Regarding TWS Narration

TAM Total Transitivity Lexical Aspect Discourse Function

ki 19 low (58%) Achievement (53%) Result (53%)

nu 11 low (82%) Achievement (91%) Complication (74%)

tu 16 medium (100%) Achievement (94%) Complication (81%)

ari 16 low (63%) State (100%) Abstract (56%)

tari 5 medium (80%) State (100%) Abstract (80%)

The established fact auxiliary ki shows the lowest propensity to govern particular sentence-

final predicates based on all three examined parameters, with none gaining a preference

rating of over 60%. The perfective auxiliaries nu and tu differ primarily in the transitivity

of the predicate, the former governing low-transitivity predicates and the latter, those of

medium-transitivity. The stative auxiliaries ari and tari present a similar difference in their

transitivity preferences, but as there are only five tokens of tari throughout the outermost

93

narration, I hesitate to draw any conclusions regarding its leaning toward higher-transitivity

predicates than ari.

The Abstract is the most grammatically rich and complex of the narrative’s

functional parts. It contains the most animated language of the sutra outside of the

embedded narration, with not only a higher proclivity for TAM marking but also for

rhetorical devices such as hypothetical questions and sentences of self-reflection. We could

thus interpret the Abstract as a type of embedded narration itself in the voice of the

translator Xuanzang rather than the assumed outermost narrator Ānanda. One cannot help

but wonder whether the Tōdaiji monk who glossed the text felt an affiliation with this

commentary by a translator who, like him, had to carefully render important religious texts

into the vernacular from a linguistic source morphosyntacticly so far removed for the sake

of fellow practitioners, who lacked the linguistic training to access them in their “original”

forms.

4.4 An Embedded Parable in the Tōdaiji Temple Ten Wheels Sutra

We now turn to a short, but representative, embedded parable told by the Buddha

to Kṣitigarbha in the second part of the third chapter of the TWS. In this chapter the Buddha

gives numerous examples of the power of kesa, or ‘Buddhist robes,’ even when worn by

those who are not practitioners of Buddhism. Below is the entire kundokubun rendition of

the parable (from Nakada 1958: 74).

(45a) 又、善男子、昔、國王有（り）き。

mata zennansi mukasi kokuwau ari-ki

also gentlemen long.ago king is-KI

94

‘Also, gentlemen, long ago there was a king.’

(45b) 名を超福德といふ。

na wo Teupukutoku to ipu

name ACC Supreme.Virtue COM call

‘He was named Supreme Virtue.’

(45c) 有る人過（を）犯セリ。

aru pito ayamati wo wokase-ri

be person crime ACC commit-ARI

‘A certain person had committed a crime.’

(45d) 罪死に合するに應ゼ（り）。

zaisi ni gapu.suru ni ou.ze-ri

capital.offence LOC meet LOC correspond.to-ARI

‘It had met the level of a capital offense.’

(45e) 王の性仁慈にして命を斷（たむ）ことを欲（は）不。

wau no syau ninzi ni.site inoti wo tata-mu

king GEN nature benevolent COP-GER life ACC end-FUT

koto wo negapa-zu

NML ACC want-NEG

‘The king, having a benevolent nature, did not wish that [the criminal’s] life should

be ended.’

(45f) 一の大臣有（り）。

iti no daizin ari

one gen minister be

‘There was a minister [present].’

(45g) 諸の智策多し。

moro no tisaku oposi

all GEN clever.idea many

‘He had numerous clever ideas.’

95

(45h) 前みて、王に白（し）て曰ひし（く）『（為）憂へたまひ（ぬる）こと勿。終に（不令）王

に殺生の罪をば得しめジ。魁膾に付ケ不して（令）此の人を殺（さ）しめ（む）。』

（と）いふ。

susumi-te wau ni mawosi-te ipi-siku

advance-GER king LOC humbly.say-GER say-KI.NML

“urepe-tamapi-nuru koto mana tupi.ni wau ni setusyau

“fret-HON-NU NML NEG in.the.end king LOC death.penalty

no tumi wo ba e-sime-zi ke'e ni

GEN sin ACC TOP gain-cause-NEG executioner LOC

tuke-zu.site ko no pito wo korosa-sime-mu”

attach-NEG-GER this GEN person acc kill-CAUS-FUT”

to ipu

COM say

‘Advancing, he humbly said to the king, “Do not fret. In the end the sin of [ordering]

the death penalty will not be incurred by the king. We will see to it that something

kills this person without involving an executioner.” he said.’

(45i) 時に彼の大臣己ガ智力を以て、犯罪の人を將て惡（し）ク醉ヘル象に付ケツ。

toki ni ka no daizin ono ga tiriki wo

time LOC that GEN minister own GEN intellect ACC

moti-te panzai no pito wo wi-te asiku

hold-GER102 crime GEN person ACC bring.along-GER badly

wepe-ru zau ni tuke-tu

drunk-ARI elephant LOC attach-TU

102 This is a periphrastic instrumental case construction. From a phrase literally meaning ‘holding N’ (N ACC

hold-GER), by EMJ it had become grammaticalized as INST. Proof of this is in interrogative sentences where

we can see wo motite constructions marked with interrogative quasi-copula ka, as in the following example

from a Tōdaiji GLS discussed in Chapter 6 below.

何（を）以（てか）知（ること）得たまふへき。 (Tabuchi 1987: 85)

nani wo.moti-te ka siru koto e-tamapu-beki

what INST Q learn NMZ gain-HON-MODAL

‘How (=by what) are you able to learn (this)?’

96

‘Then that minister, using his own intellect, took the criminal and attached him to

a drunken elephant.’

(45j) 時に惡し（く）醉へ（る）象、鼻を以て罪人の兩ノ脛を卷キ取リて、空の中に舉上

テ、其の勢力を盡（し）て（欲）（於）地に撲カムトスルニ忽に此の人の裳に赤き色

有（る）を見、是レ袈裟なりと謂ひては心に淨信を生（じ）て便（ち）徐ク地に置

（き）て懺謝し悲號して（於）前に跪（い）テ伏セツ。

toki ni asiku wepe-ru zau pana wo moti-te

time LOC badly drunk-ARI elephant nose ACC hold-GER

zainin no putatu no piza wo maki-tori-te sora no

criminal GEN both GEN knees ACC wrap-take-GER sky GEN

naka ni age-te so no seiriki wo tukusi-te

in LOC raise-GER that GEN power ACC extinguish-GER

ti ni kudaka-mu to suru ni tatimati ni ko

earth LOC break-CONJ COM do LOC suddenly COP this

no pito no mo ni akaki iro aru wo mi

GEN person GEN lower.robe LOC red color has ACC see

kore kesa nari to omopi-te pa kokoro ni

this Buddhist.robe103 COP COM think-GER TOP heart LOC

zyausin wo syau.zi-te sunapati yauyaku ti ni

pure.faith ACC bear-GER then gradually earth LOC

oki-te zansya.si pigau.si-te mape ni pizamatui-te

place-GER repenting cry.out-GER front LOC fall.to.knees-GER

puse-tu

prostrate.oneself-TU

‘Then the badly drunken elephant wrapped up the criminal’s knees using its trunk,

raised him into the sky, and just as it was about to slam him into the ground with

all its might, it suddenly saw that this person had red color in his lower robe.

Thinking this is a Buddhist robe, its heart bore pure faith and it then gradually

placed him on the ground, cried out in repentance, and falling to its knees,

prostrated itself.’

103 J. Kesa (袈裟 ‘Buddhist robes’) is a phonetic loan from kāsāya in Sanskrit.

97

(45k) 鼻を以て足を抆ヒ心に敬重を生（じ）て、彼の人を瞻仰す。

pana wo moti-te asi wo nogopi kokoro ni kyaudiu

nose ACC hold-GER feet ACC wiping heart LOC respect

wo syau.zi-te ka no pito wo sengau.su

ACC bear-GER that GEN person ACC revere

‘Wiping his feet with its trunk, it bore respect in its heart and revered that person.’

(45l) 大臣見已（り）て馳リ還（り）て王に白す。

daizin mi-wopari-te pasiri kaperi-te wau ni

minister see-finish-GER running return-GER king LOC

mawosu

say.HUM

‘The minister finished watching this, sped back, and told the king.’

(45m) 王聞（こし）メ（し）て喜ビ愕リて未曾有なりと歎ず。

wau kikosimesi-te yorokobi ozori-te mizou nari to

king hear.HON-GER being.joyful respect-GER unprecedented COP COM

tan.zu

be.moved

‘The king, hearing this, was full of joy and awe and was deeply moved, [thinking]

“This is without precedent!”’

(45n) 便（ち）國人に勅して加三寶を敬（は）しム。

sunapati kokunin ni tyoku.si-te masumasu

thus nation LOC command-GER increasingly

sanbou wo uyamapa-simu

three.treasures104 ACC respect-CAUS

‘Thus, he commanded the nation to increasingly respect the three Buddhist

treasures.’

(45o) 斯に因（り）て殺を斷ずること、王も贍部洲にもしき。

104 The three treasures, or three jewels are the Buddha, the dharma (‘law’), and the monastic order.

98

sore ni yori-te satu wo dan.zuru koto wau mo

that LOC base-GER killing ACC cease NML king also

senbusyau ni mo si-ki

Jambudvīpa105 LOC also do-KI

‘Based on that the king also brought about a cessation of killing in Jambudvīpa as

well.

(45p) 善男子、當に是（の）如き過去の醉（へ）ル象を觀べし。

zennansi masa ni ko no gotoki kwako no wepe-ru

gentlemen surely COP this GEN like past GEN drunk-ARI

zau wo miru-besi

elephant ACC see-MODAL

‘Gentlemen, surely you should look to the likes of this drunken elephant of the

past.’

(45q) 无睱の傍生趣の身を受（け）たりと雖ども、而（も）袈裟を敬（ひ）て（不）惡業を造

ラズアルナリケリ。

muka no bausyauzyu no mi wo uke-tari

no.leisure106 GEN rebirth.as.an.animal GEN body ACC receive-TARI

to ipe-domo sikamo kesa wo uyamapi-te

COM say-although however buddhist.robes ACC respect-GER

akugyau wo tukura-zu aru nari-keri

evil.deeds ACC make-NEG have COP-KERI

‘Although they say selves reborn as animals have no leisure to follow the Buddhist

path, the fact is that they do, however, respect the Buddhist robes and will not

wreak evil on them!’

Although lacking an Abstract, this parable contains all the other narrative discourse

105 “The southern of the four great continents” Soothill & Hodous 2014: 481. See Buswell & Lopez (2014:

378–379) for details.

106 An existence in which one has no leisure to follow the Buddhist teachings (see “八無暇” in Soothill &

Hodous 2014: 38).

99

functions and concludes with a Coda headed by keri. In fact, this is the one and only time

in the entire early Heian kundokubun corpus in which the Buddha uses keri to

grammatically mark modality (i.e., factuality) without marking mirativity (i.e., new, just

discovered information). The narrative discourse functions and sentence-final predicate

morphology intersect in this embedded tale as follows. Each instance of any sentence

type is listed separately. As a reminder, ari is a stative auxiliary, ki is a past established

fact auxiliary, and tu is a perfective auxiliary.

TAM-marked and -unmarked Sentences by Narrative Function in the TWS Parable

Orientation: ki, (TAM-unmarked), ari, ari

Evaluation: (TAM-unmarked)

Orientation: (TAM-unmarked), (TAM-unmarked)

Complication: (TAM-unmarked), tu, tu

Evaluation: (TAM-unmarked)

Complication: (TAM-unmarked), (TAM-unmarked), (TAM-unmarked)

Result: ki

Coda: (TAM-unmarked), keri

This parable is representative of how the TAM auxiliaries are used in creating narrative

structure in kundokubun texts. The narrative is framed with ki at the beginning (45a) and

conclusion (45p) of the tale. There is an additional, nominalized token of ki (siku) before

the quotation in the first line serving the Complication function (45h), a reference to ‘what

he said’ that grounds the narrative in the past just before the action begins. The two lines

in the Orientation ending with the stative auxiliary ari present the background information

necessary for the action to begin, namely that someone had committed a crime (45c) and

that crime was a capital offense (45d). The two lines headed by perfective tu, (45i) and

(45j), are where we find the focused action of the story, with the rampaging elephant

100

believing itself to be before a monk rather than a criminal and thus deciding to bow in

respect. Finally, the Buddha uses keri in the final line of the Coda (45q) to present the fact

that animals will not harm those who wear Buddhist robes as something that can be inferred

from the tale, which, having been framed at its outset by the established past fact auxiliary

ki, has been presented as witnessed (historical) truth.

Although this parable is only 17 sentences long and lacks the auxiliaries nu and

tari, in this section I have presented how the stative auxiliary ari, perfective tu, past

established fact ki, and externally established fact keri are used in embedded Buddhist

kundokubun narration. I now turn to the shortest sutra examined in this study, the

Immeasurable Meanings Sutra.

4.5 The Narration of the Immeasurable Meanings Sutra

The Immeasurable Meanings Sutra (無量義経 Muryōgi kyō) (IMS) is considered

by some to be the prequel to the Lotus Sutra (J. 妙法蓮華華経 Myōhō renge kyō, Skt.

Saddharmapuṇḍarīkasūtra107) and there are records that attest to it being copied along with

the Lotus Sutra during the Nara period (710–794 CE) (Kabutogi & Nakada 1979: 130).

Although it is claimed to have been translated into Chinese by an Indian translator named

Dharmāgatayaśas (J. 曇摩伽陀耶舎 Donmakadayasha), today it is considered by some

scholars to be an apocryphon originally composed in China (Kabutogi & Nakada 1979:

127–130; Buswell & Lopez 2014: 1001–1002).

107 See Buswell & Lopez (2014: 729–730) for an overview of this sutra’s history and significance.

101

The IMS is much shorter than the Golden Light Sutra and Ten Wheel Sutra

examined above, consisting of only a preface and three chapters. The sutra is structured as

a conversation between bodhisattva Mahāvyūha (‘Great Adornment’ J. 大 荘 厳

Daisaugon) and the Buddha, in which the Buddha explains the benefits of hearing the sutra

itself and how bodhisattvas can quickly reach enlightenment.

The text examined here was in the personal collection of the acclaimed scholar of

the Lotus Sutra, Kabutogi Shōkō (兜木正亨), who acquired it in pursuit of his religious

studies research. Both he and Nakada Norio (中田祝夫) agree that it was glossed in the

first half of the Heian period, but the text itself lacks a colophon so the exact circumstances

of its glossing are unknown.

As noted, the IMS is the shortest kundokubun text examined in this dissertation,

with 141 lines of outermost narrative out of 575 total lines of the sutra.108 Another contrast

with the above texts is the lack of embedded narrative in the sutra, so this chapter concludes

with just a study of its outermost (and only level of) narration.

To begin, Table 12 below compares the total number of TAM auxiliaries in sentence-

final position throughout the sutra as opposed to the outermost narration of the IMS.

108 As with my total line count for the GLS and TWS, I include pre-quotative adverbial phrases that end in

verbs of speaking nominalized by aku, such as ipaku (曰く), in the total line count.

102

Table 12 Sentence-Final TAM auxiliaries in IMS Narration

TAM Total In Dialogue (%) In Narrative (%)

ki 35 15 (43%) 20 (57%)

keri 1 1 (100%) 0 (0%)

nu 8 3 (37%) 5 (63%)

tu 7 6 (86%) 1 (14%)

ari 32 17 (53%) 15 (47%)

tari 5 5 (100%) 0 (0%)

Among the TAM auxiliaries, ari is used sentence-finally the most evenly between dialogue

and narration, followed by ki. Because the use of TAM auxiliaries in the overarching

narrative is discussed in the subsections that follow below, let us first present, examples of

each used inside Quotations, beginning with ari:

(46) 千輻を具（へたま）へり。 (Kabutogi & Nakada 1979: 86)109

senpuku wo sonape-tamape-ri

thousand.spokes110 ACC gather-HON-ARI

‘[He] has gathered the one-thousand-spoked [wheels on his feet].’

This Quotation occurs in a scene when Mahāvyūha is telling the Buddha how wonderful

he is by describing his most excellent qualities. The use of auxiliary ari in narrative that

we have noted thus far—that of providing background information—stems from its use in

expressing conditions (i.e., States), such as the resulting condition that ari helps express in

(46).

109 See Kabutogi & Nakada (1979: 69–71) for orthographical conventions.

110 This refers to the thousand spokes on the wheel symbol on the feet of the Buddha (Soothill & Hodous

2014: 81).

103

Next, auxiliary ki is used mostly by the Buddha in Quotations, as he to relates things

that occurred in the past. After explaining that the four noble truths were originally for

humans seeking their own liberation from suffering,111 he says the following occurred.

(47) 而れども八億の諸天、来下して法を聴、菩提心を発しき。 (Kabutogi & Nakada

1979: 100)

saredomo pyakuoku no syoten raige.si-te

however 800.million GEN many.gods112 come.down-GER

papu wo kiki bodaisin wo hatu.si-ki

dharma ACC hear aspiration.to.enlightenment113 ACC resolve-KI

‘However, eight hundred million numerous gods game down and, hearing the

dharma, resolved to become enlightened.’

Here the Buddha is praising Buddhist doctrine by presenting an event—that even though

the dharma was expounded for people, it inspired even deities—grammatically marked

with ki as witnessed past fact.

Both keri and tari are found only in Quotations in the IMS. The following is the

only example of keri in the entire kundokubun rendition. The Buddha is praising

Mahāvyūha and the rest of the gathered bodhisattvas for seeking his advice:

111 J. 聲聞 syaumon; Skt. śrāvaka (Buswell & Lopez 2014: 850); see also footnote 83.

112 J. 天 ten is a translation of Skt. deva ‘divinity.’ See Buswell & Lopez (2014: 230–233) for a more details

regarding these beings.

113 Skt. bodhicitta (Buswell & Lopez 2014: 130–131).

104

(48) 善哉、善哉。善男子。善く是の時を知りけり。 (Kabutogi & Nakada 1979: 92)

zen’ya zen’ya zennansi yoku ko no toki wo

excellent excellent gentlemen well this GEN time ACC

siri-keri

learn-KERI

‘How very excellent, excellent, gentlemen, (the fact) that you will learn well of

this time.’

He goes on to say it does not take long to reach full enlightenment.114 This use of keri

strongly supports the argument that it primarily indexes epistemic modality rather than

tense in Buddhist genres. I elaborate on the use of keri in Quotations in Buddhist

kundokubun texts in section 5.3 below.

Unlike in the GLS and TWS examined above, the auxiliary tari is not found in

narrating passages of the IMS. In the genre of early Buddhist kundokubun, it is not used

with nearly the same frequency as its older semantic counterpart ari and, this sutra being

more than ten times shorter than the GLS (three chapters as opposed to thirty-one), there

may simply have not been an opportunity to use the still-emerging stative, which was

initially restricted to newer verb classes. This discrepancy is discussed further in the

following chapter, so here we turn to a representative example of tari’s use in the dialogue

of the IMS.

114 The term used in this section of the IMS is 如来 nyorai (Skt. tathāgata). While one of the epithets of the

Buddha, it specifically refers to his state of ‘having thus gone,’ i.e., reaching enlightenment. It can also be

understood as ‘having thus come,’ which is why the Sinitic character 来 (‘come’) is often used in Chinese

translations of the term. These differing interpretations of directionality are due to Sanskrit sandhi, or

assimilation, rules. Skt. tathāgata clearly begins with tathā ‘thus,’ but the remainder may represent either the

past passive participle of √gam- (‘go’) or, alternatively, an underlying assimilated preverb may be involved,

producing ā√gam- (‘come’). See Buswell & Lopez (2014: 897) for more on tathāgata and see footnote 52.

105

(49) 端坐すること六年にして、阿耨多羅三藐三菩提（を）成（ずる）こと得たり。

(Kabutogi & Nakada 1979: 98)

tanza.suru koto rokunen ni.si-te

sit.erect NML six.years COP-GER

anokutarasanmyakusanbodai wo zyau.zuru koto e-tari

supreme.perfect.enlightenment115 ACC attain NML gain-TARI

‘Having sat erect for six years, I have been able to attain perfect, supreme

enlightenment.’

In kundokubun, a verb followed by koto u (こと得) is a periphrastic expression meaning

‘manage to achieve (verb).’ Here the Buddha is explaining how he was able to attain

enlightenment after a long period of meditation, but goes on to say there are various paths

to enlightenment and he can guide his disciples to attain it more quickly. Auxiliary tari is

clearly being used here as a perfect—the Buddha is indexing the present relevance of his

change of state into an enlightened being who can now guide others on the path.

Finally, although the perfective auxiliaries are close to each other in number in the

sutra, with eight tokens of nu and seven of tu, there is a large discrepancy in their use, with

the former primarily found in the overarching narrative and the latter more common in

dialogue. Below are the first two lines of Mahāvyūha’s last turn speaking in the sutra after

hearing the Buddha command those in attendance to protect and propagate the sutra. His

use of the two perfective auxiliaries in these lines is representative of their use in

Quotations in the midst of narration.

(50) 世尊、我等、快く世尊の慈愍して我等が為に是の甚深微妙の無上大乗無量義

経を説（き）たまふことを蒙（ふり）ぬ。 (Kabutogi & Nakada 1979: 123)

115 Skt. anuttarasamyaksaṃbodhi (see Buswell & Lopez 2014: 55).

106

seson ware-ra kokoroyoku seson no zimin.si-te

Lord I-PL happily lord GEN have.compassion-GER

ware-ra ga tame ni ko no zintan bimeu no

I-PL GEN sake COP this GEN deep profound GEN

muzyau daizyau muryaugi kyau

unsurpassed great.vehicle116 immeasurable.meanings sutra

wo toki-tamapu koto wo kagapuri-nu

ACC expound-HON NML ACC receive-NU

‘Lord, we happily receive your compassionate honorable expounding of this deep

and profound unsurpassed great vehicle Immeasurable Meanings Sutra for our

sake.’

(51) 敬（み）て仏勅を受（けたてまつり）つ。 (Kabutogi & Nakada 1979: 123)

tutusimi-te bututyoku wo uke-tatematuri-tu

respect-GER budda’s.words ACC receive-HUM-TU

‘[We] respectfully have received your words.’

In both examples above, Mahāvyūha is referring to receiving the words of the Buddha.

Unlike most translations of the perfective auxiliaries in narrative passages—which are

generally rendered in the past tense due to the context of the surrounding discourse—I

render these both in the present tense in English because Mahāvyūha is commenting on an

event occurring in the moment. The reason the two different perfectives are used in these

lines is due to differences in volition involved in the two acts of “receiving.” In (50) we

find the endoactive perfective nu governing kagapuru ‘be bestowed with’ because this

action refers to hearing (a less transitive act) the words that the Buddha expounded. In the

subsequent line (51), on the other hand, the exoactive perfective tu is employed to index—

116 Skt. mahāyāna; the dominant form of Buddhism in Japan that presents a shorter path to Buddhahood

compared to other ‘vehicles,’ or schools (Buswell & Lopez 2014: 513–514), discussed in section 1.1 above.

107

and foreground—the intentionality to act on those words. The Sinograph 勅 in the word

仏勅 (bututyoku ‘Buddha’s words’) also has the nuance of ‘command,’ referring to the

Buddha’ previous speech act. I consider the quantitative differences in their employment

in the outermost narrative—that nu is employed more than tu—later in this section.

For the remainder of section 4.5, I use the phrase “IMS narration” to specifically

refer to the outermost, matrix narration, i.e., all lines outside of any dialogue or embedded

parable. Because the auxiliaries tari and keri are not found in the outermost narration, they

are excluded from the analyses in the following subsections on transitivity, lexical aspect,

and discourse function.

4.5.1 Transitivity in IMS Narration

The table below depicts the relation between TAM marker and transitivity scores

in the outermost narrative of the IMS. Just as in the sutras analyzed above, there were no

TAM-marked predicates of high transitivity in the outermost narration of the sutra.

Because the stative tari and externally established fact keri are not found sentence-finally

outside of dialogue, they are not included.

Table 13 The Transitivity of TAM-marked Predicates in IMS Narration

TAM low (%) medium (%)

ki 18 (90%) 2 (10%)

nu 3 (60%) 2 (40%)

tu 1 (100%) 0 (0%)

ari 12 (80%) 3 (20%)

108

At first glance, it is clear that TAM markers overwhelmingly govern low-transitivity

predicates throughout the narration of the IMS. As mentioned throughout this

dissertation, the “outermost narration” depicts the progress of a conversation, with most

actions those of movement in space and of speaking. We therefore find a general lack of

high-transitivity actions outside of embedded narration. That being said, although in the

above sutra analyses the exoactive perfective auxiliary tu has been strongly associated

with medium-transitivity,117 there is a single case of tu-marking in the IMS narration, in

the following low-transitivity sentence.

(52) 色を見、香を聞（き）て自然に飽足しつ。 (Kabutogi & Nakada 1979: 121)

siki wo mi kau wo kiki-te sizen ni

form118 ACC see incense119 ACC smell-GER120 natural COP

pausoku.si-tu

become.satisfied-TU

‘Having seen the form and smelled the incense, they naturally became satisfied.’

The context here is important. Naturally ‘becoming satisfied’ is a situation very much

lacking in transitivity—the subject lacks agency and there is no direct object on which to

act. As we discussed in section 2.1 above, the perfective auxiliary tu has semantic

117 As noted in its introduction in section 2.1, tu is an exoactive perfective auxiliary—it is generally found

in higher transitivity predicates. As I discuss further below, tu’s greater association with medium-transitivity

sentences rather than high-transitivity ones in kundokubun narration is simply due to a general lack of high-

transitivity tokens. When there is a high-transitivity predicate in need of a perfective in kundokubun, tu is the

one to use.

118 Skt. rūpa (see Buswell & Lopez 2014: 721–722).

119 Skt. gandha (see Buswell & Lopez 2014: 310).

120 The Sinitic character 聞, which generally signifies ‘hear/listen,’ is also used in Buddhist Hybrid Chinese

to mean ‘smell’ (Soothill & Hodous 2014: 427). It can also be found in post-Heian wabun texts in reference

to appreciating incense (Naomi Fukumori, p.c.).

109

extension as a recent past marker. In this scene flowers and incense have just fallen from

the heavens after the words of the Buddha, and (52) confirms the satisfaction the

assembly felt during that very recent experience.

As we have seen in the above analyses of the GLS and TWS, the endoactive

perfective auxiliary nu is more likely to govern lower-transitivity predicates. We find it

heading a medium-transitivity sentence-final predicate twice in the IMS narration, as in

the following example.

(53) 仏、是の経を説き已（りたまひ）ぬ。 (Kabutogi & Nakada 1979: 102)

potoke ko no kyau wo toki-wopari-tamapi-nu

buddha this GEN sutra ACC expound-finish-HON-NU

‘The Buddha finished expounding this sutra.’

The other case of medium-transitivity nu also is also in a predicate where it follows the

auxiliary verb woparu (已る ‘finish’). Kasuga (1985b) discusses the connection between

the sinograph已 and both woparu and nu but does not address the shared semantics of

the auxiliary verb woparu and the auxiliary nu that likely led to this pattern of co-

occurrence in the translation. Both auxiliaries highlight the coming to an end itself over

any agent’s intent to conclude—if indeed an agent is involved at all. They far more often

mean ‘come to an end’ rather than ‘bring to a conclusion.’ Auxiliary nu thus has a

proclivity to govern woparu predicates even when woparu governs a transitive verb, such

as toku (説く ‘expound’) in (53). This tendency was discussed in more detail in section

4.3 above.

We turn next to established past fact auxiliary ki and the stative auxiliary ari.

Below are examples of each, governing predicates of medium-transitivity.

110

(54) 無量の衆生は、阿耨多羅三藐三菩提の心を發しき。 (Kabutogi & Nakada

1979: 104)

muryau no syuzyau pa anokutarasanmyakusanbodai

immeasurable GEN sentient.beings TOP supreme.perfect.enlightenment121

no kokoro wo hatu.si-ki

GEN mind ACC resolve-KI

‘The immeasurable sentient beings resolved their minds toward supreme, perfect

enlightenment.’

In the above example auxiliary ki is governing the verb hatu.su (發す), which, when it

takes noun phrases that express thoughts or ideas as its primary complement as we see

here, means ‘resolve (oneself to [such thoughts]).’ Although high in volition, the act of

resolving causes no change in the ideas themselves, so that sentences such as this are

coded as having medium transitivity. Next is an example of the stative auxiliary ari.

(55) 諸の結漏を盡せり。 (Kabutogi & Nakada 1979: 82)

moro no keturo wo tukuse-ri

all GEN bondage.reincarnation 122 ACC extinguish-ARI

‘They have extinguished all bondage and reincarnation stemming from the

passions.’

At first glance, the verb in (55), tukusu (盡す ‘extinguish’, ‘bring to an end’), makes this

seem to be a high-transitivity situation. However, when we consider what is being

‘extinguished’ in this particular context, we find it is more properly understood as a

medium-transitivity predicate. This line is information given as part of an introduction of

12,000 monks gathered to hear the sutra. The ‘bondage and reincarnation’ are a state of

121 Skt. anuttarasamyaksaṃbodhi (see Buswell & Lopez 2014: 55).

122 ‘Bondage and reincarnation because of the passions’ (Soothill & Hodous 2014: 386).

111

being that those monks have overcome as saints (J. 阿羅漢 arakan; Skt. arhan/arhat).

Therefore, ‘extinguish’ here is more metaphorical, meaning ‘overcome’ or ‘put behind

them,’ rather than literally meaning ‘cause to cease to exist.’ This differentiation is why I

analyze this line as having medium transitivity. We now turn to the relationship between

TAM marking and lexical aspect in the outermost narration of the IMS.

4.5.2 Lexical Aspect in IMS Narration

In the following table I present the total number of sentences coded as Stative,

Activity, Accomplishment, and Achievement (see section 3.2.2 above) that end with a

TAM auxiliary in the outermost narrative of the Immeasurable Meanings Sutra.

Table 14 The Lexical Aspect of TAM-marked Predicates in IMS Narration

TAM State (%) Activity (%) Accomplishment (%) Achievement (%)

ki 10 (50%) 0 (0%) 8 (40%) 2 (10%)

nu 0 (0%) 0 (0%) 2 (40%) 3 (60%)

tu 0 (0%) 0 (0%) 0 (0%) 1 (100%)

ari 9 (60%) 0 (0%) 6 (40%) 0 (0%)

To begin, there are no sentences coded as Activities headed by a TAM auxiliary

throughout the sutra’s outermost narration. That is not to say that the narrative includes

no Activity predicates. The following is a narrative sentence without TAM-marking, which

depicts an Activity in the IMS.

(56) 是（に）於（て）三千大千世界、六種に震動す。 (Kabutogi & Nakada 1979: 102)

112

kore ni oki-te sanzen.daisen.sekai rokusyu ni

this LOC place-GER trichiliocosm123 six.types COP124

sindau.su

shake125

‘At this, the universe shook in the six ways.’

The verb sindau.su (震動す ‘shake’) is in its bare conclusive form, but I render it in

English in the past tense because it is in a narrative sequence grounded in the past due to

being framed by the past established fact auxiliary ki. In an interesting contrast with the

Golden Light Sutra narration, in which ki concludes more sentences coded as Activities

than any other lexical aspect category, in the IMS ki predominantly heads States, which

make up 50% of its tokens, such as the following.

(57) 菩薩摩訶薩ありキ。 (Kabutogi & Nakada 1979: 78)

bosatu makasatu ari-ki

bodhisattva mahāsattva be-KI

‘There were great bodhisattvas.’

The verb ari depicts a state (much like the auxiliary ari), and auxiliary ki’s semantics

cause no change to the sentence’s lexical semantics when it governs this (or any)

predicate. After States, ki most often is found in Accomplishments, 40% of the time. In

fact, although their totals differ, nu and ari are found in Accomplishment predicates 40%

123 Skt. trisāhsramahāsāhasralokadhātu (‘three-thousandfold great-thousandfold world system’). This

refers to the all-encompassing universe (Buswell & Lopez 2014: 924).

124 I interpret this ni to be the adverbial use of the participial form (‘being x, V’) of the copula nari rather

than a locative case particle because rokusyu expresses the manner in which the trichiliocosm shakes in this

scene.

125 This refers to the six great earthquakes that occurred in response to the Buddha’s advancement toward

enlightenment (Soothill & Hodous 2014: 137).

113

of the time as well. The occurrence of perfective auxiliary nu in Accomplishments, and

both it and its exoactive counterpart perfective tu in Achievements, has come to be what

we expect of these due to their perfective semantics indicating changes of state. The

stative auxiliary ari, however, is not what we would expect heading telic predicates.

There is a dichotomy of TAM marking on speech act predicates in the IMS

depending on the speaker. Actions of humble speaking, using the verb mausu, are most

often governed by ki. In other words, of the eight ki-marked sentences coded as

Accomplishments in the IMS, seven occur following Quotations spoken by Mahāvyūha

(occasionally with other bodhisattvas) in the following phrase.

(58) とまうしき。 (Kabutogi & Nakada 1979: 91, 92, 93, 106, 107, 123, 124)

to mausi-ki

COM say.HUM-KI

‘(He) humbly said.’

On the other hand, in the IMS, when the Buddha responds we find a pattern in the

grammatical marking of the dialogue’s progression not seen in other early Heian

kundokubun texts. His Quotations are followed with the following verb of speaking

governed by auxiliary ari:

(59) とのたまへり。 (Kabutogi & Nakada 1979: 92, 93, 93, 97, 107, 123)

to notamape-ri

COM say.HON-ARI

‘He honorably said.’

The stative auxiliary ari is most often defined by its atelicity, but here we find it clearly

heading telic actions that end before the next quotation, generally something said by the

Buddha’s interlocutor Mahāvyūha. One could say this repeated use of the stative

114

auxiliary ari when describing the actions of the Buddha puts more emphasis on the

content of his words rather than the fact that they were said.126

A more fine-grained analysis on this use, rare to the genre of kundokubun sutras,

of ari when describing narrative events is done in the conclusion of this section on the

IMS below. But first we examine the relationship between the TAM auxiliaries and

discourse function in the narration of the IMS.

4.5.3 Discourse Function in IMS Narration

The following table displays the relation between sentence-final TAM morphemes

in the IMS and the narrative discourse functions of those sentences.

Table 15 The Discourse Function of TAM-marked Predicates in IMS Narration

TAM Abstract (%) Orientation (%) Complication (%) Evaluation (%) Result (%)

ki 0 (0%) 11 (55%) 5 (25%) 0 (0%) 4 (20%)

nu 0 (0%) 0 (0%) 5 (100%) 0 (0%) 0 (0%)

tu 0 (0%) 0 (0%) 1 (100%) 0 (0%) 0 (0%)

ari 0 (0%) 5 (33%) 6 (40%) 4 (27%) 0 (0%)

To begin, there are no Abstracts in the IMS narration. In Orientations, we find 11 tokens

of ki and 5 of ari, which are 55% and 33% of their totals, respectively. All four TAM

auxiliaries found in the outermost narration of the IMS are used sentence-finally in

Complications, or narrative clauses that depict the progressing events in the story.

126 The choice to use a perfect rather than a (modal) past on speech-act predicates is made between itta

(‘said’) and itteiru (‘has said’) when quoting others in Japanese today. Rather than individuating the speech

act with itta (‘said’), much like ipi-ki in EMJ, one can signify more concern with the content of what was said

with itteiru (‘has said’), which seems to parallel what we find here in the IMS.

115

Our discussion in the previous section alludes to the reason we find 43% of the

ari tokens of the IMS in Complications, which is contrary to how we would imagine

stative verbs to work. This is due to the way it is used to govern “speech-act” predicates

of the Buddha speaking, discussed in the previous subsection. Auxiliary ari’s more

typical uses in Orientation and Evaluation sentences, such as the following two examples,

are much more in line with what we have seen in the GLS and TWS.

(60) 香（を）焼き、華（を）散じ、種種に供養したてまつり已（り）て、退（き）て一面に坐

せり。 (Kabutogi & Nakada 1979: 78–79)

kau wo yaki pana wo san.zi syusyu

incense ACC burning flower ACC scattering various

ni kuyau.si-tatematuri-wopari-te sirizoki-te itimen

COP make.offerings-HUM-finish-GER withdraw-GER one.side

ni za.se-ri

LOC sit-ARI

‘Having finished burning incense, scattering flowers, and making various

offerings, they withdrew and were sitting to the side.’

The above example, from the sutra’s Orientation, presents to the scene those listening to

the Buddha using the stative auxiliary. The example below is used to describe a scene

after the Buddha and his interlocutor Mahāvyūha have completed a part of their didactic

conversation.

(61) 天の妙楽の具、處處に安置せり。 (Kabutogi & Nakada 1979: 83, 103, 121)

ten no meugaku no gu syosyo

heaven GEN sublime.music GEN instrument various.places

ni anti.se-ri

LOC put.in.place-ARI

‘Heaven’s instruments of sublime music have been put in various places.’

116

This line repeats three times throughout the IMS in a parallel structure. Each time

describing the scene.

The perfective auxiliaries nu and tu are limited to Complications, such as (52)

above, discussing auxiliary tu, and the following example of nu.

(62) 仏の所に来詣しぬ。 (Kabutogi & Nakada 1979: 78, 82)

potoke no tokoro ni raikei.si-nu

buddha GEN place LOC approach-NU

‘[Each] approached the place of the Buddha.’

This line is spoken by the narrator twice in the sutra when describing assembly members

approaching the Buddha before paying him obeisance.

Although we find ki most often in Orientations, at 55%, it is also used in

Complications and Results; these are most often “speech-act” predicates of the kind that

follow Mahāvyūha’s lines, as discussed above. We now turn to section 4.5’s conclusion.

4.5.4 Conclusions Regarding IMS Narration

In this subsection I have presented data on the transitivity, lexical aspect, and

narrative discourse function of sentences headed by ki, nu, tu, and ari (but not keri and tari,

which are not attested sentence-finally) in the outermost narrative of the Immeasurable

Meanings Sutra. Table 16 summarizes these results.

117

Table 16 Summary of Findings Regarding IMS Narration

TAM Total Transitivity Lexical Aspect Discourse Function

ki 20 low (90%) unrestricted Orientation (55%)

nu 5 low (60%) Achiev. (60%) Complication (100%)

tu 1 low (100%) Achiev. (100%) Complication (100%)

ari 15 low (80%) State (60%) unrestricted

In the IMS, the established past fact auxiliary ki is found governing mostly low-transitivity

predicates, but this is more likely due to a preference for low-transitivity predicates in the

sutra in general than to ki’s semantics. Sentence-final ki governs predicates without

preference for lexical aspect and the majority of its tokens are found in Orientations. The

perfective auxiliaries nu and tu are not distinct from each other with regard to the

transitivity, lexical aspect, nor discourse function of the sentences they head.

As mentioned above, there is a discrepancy between the total number of cases of

perfectives nu and tu in narration, with five of the former and only one of the latter.

Throughout this study I have remarked on the tendency favoring lower transitivity

predicates in the outermost narration—which are generally the purview of the endoactive

nu. While it is beyond the scope of this subsection’s analysis, we do find more cases of tu

in the quoted dialogue itself. A large part of the conversation consists of the Buddha

telling Mahāvyūha what will be ‘gained’ from listening to, protecting, and propagating

the IMS. In other words, there are numerous cases of e-tu (得つ gain/attain-TU).127

127 There are even more cases of e-te-mu (得てむ attain-TU-FUT ‘you will attain’), but cases of auxiliary tu

governed by future/conjectural mu in embedded dialogue are beyond the scope of this study.

118

Even though there is only one token of sentence-final tu in the outermost narration,

it is worth discussing its contrast with nu and the bare, unmarked form in the following

three sentences, which parallel one another in both content and structure.

(63) 色を見、香を聞くときに、自然に飽足（し）ぬ。 (Kabutogi & Nakada 1979: 83)

siki wo mi kau wo kiku toki ni

form ACC seeing incense ACC smell time LOC

sizen ni pausoku.si-nu

natural COP become.satisfied-NU

‘When he saw the form and smelled the incense, he naturally became satisfied.’

The first example, (63), describes a change in Mahāvyūha’s state after he had prepared

the incense he offered to the Buddha before praising him in verse. It is an intransitive

description of a change in his internal psychological state, hence we find the endoactive

perfective auxiliary nu.

(64) 色を見、香を聞（き）て自然に飽足しつ。 (Kabutogi & Nakada 1979: 121)

siki wo mi kau wo kiki-te sizen ni

form ACC seeing incense ACC smell-GER natural COP

pausoku.si-tu

become.satisfied-TU

‘Having seen the form and smelled the incense, they naturally became satisfied.’

The sentence in (64) (also presented in (52) above) contains the only token of sentence-

final tu in the overarching narrative of the IMS. This describes a recent change of state of

the assembly after witnessing the Buddha’s sermon.

(65) 色を見、香を聞（く）に自然に飽足す。 (Kabutogi & Nakada 1979: 121)

119

siki wo mi kau wo kiku ni sizen ni

form ACC seeing incense ACC smell LOC natural COP

pausoku.su

become.satisfied

‘When they see the form and smell the incense, they naturally become satisfied.’

Finally, (65), which occurs only six sentences after (64), presents a final rain of flowers

and incense in the countless worlds of all the Buddhas. These worlds are far removed

from the scene being narrated and, subsequently, the changes that occur there are not

presented as immediate to the discourse context. Therefore, the translator elected not to

use one of the perfective auxiliaries to present the event as completed and, thus, realized.

In sum, the above examples are three different kundokubun renditions of virtually the

same, repeated, sentence in the Sinitic source text (見色聞香自然飽足). This

demonstrates that the translator was paying close attention to the particulars of the

discourse context in which each line was spoken by the sutra’s narrator when rendering

them into Japanese. In other words, he was clearly reading between the lines.

Finally, the IMS contains a unique use of the stative auxiliary ari not found in the

other sutras of this study’s dataset. It governs the “speech-act” predicate notamapu

(say.HON) that follows the quotations attributed to the Buddha throughout the narrative.

This use of ari-marking contrasts with the “speech-act” predicate that follows quotations

by the Buddha’s interlocutors, on which we find the established past fact auxiliary ki. I

therefore propose that the translator of the IMS chose to grammatically emphasize the

words of the Buddha themselves rather than the fact that they were said. In other words,

he presented them as states without a deictic anchor to the past in order to highlight their

120

relevance in the present moment. However, as these examples are limited to six sentences

in a single text, more kundokubun data are needed to draw more concrete conclusions on

this rare use of the stative auxiliary ari.

Now that we have examined the use of the TAM auxiliaries in early Heian

kundokubun renditions of the Golden Light Sutra, Ten Wheels Sutra, and Immeasurable

Meanings Sutra, in the following chapter I look at the auxiliaries in the same pairs in

which they were introduced in chapter 2, to discuss their pragmatics in narrative contexts

in the kundokubun register of Early Middle Japanese.

121

Chapter 5. The Relation Between TAM Marking and Narrative Structure in Kundokubun

While chapter 4 focused on individual layers of narrative128 in analyzing the tense,

aspect, and modality (TAM) auxiliaries in early Heian kundokubun narration, in this chapter

I present a quantitative overview of every sentence-final TAM marker in the three data

sources examined above, the Saidaiji Temple’s Golden Light Sutra (GLS), the Tōdaiji’s

Ten Wheels Sutra (TWS), and the Immeasurable Meanings Sutra (IMS) of the Kabutogi

Shōkō collection.

I now examine the TAM auxiliaries in the same pairs in which they were introduced

them in chapter 2. It should be noted that the seven narrative discourse functions—

Abstract, Orientation, Quotation, Complication, Evaluation, Result, and Coda—are coded

as such regardless of their status of being in the overarching narrative or an embedded one.

In other words, if the overarching narrator presents a dialogue that does not evolve into a

parable (i.e., embedded narration), then those lines are coded as Quotations. If a character

begins to tell a story—a sequence of events—then I code the narrative lines accordingly.

Finally, in this chapter I do not include data from the translator’s preface of the Ten

Wheel Sutra, discussed in section 4.3.4 above, due to it being outside of the contents of the

sutra itself.

128 Introduced in Figure 2 above.

122

5.1 Auxiliaries nu and tu in Heian-period Kundokubun Buddhist Texts129

Scholars such as Kasuga (1985b), Tsukishima (1967), Otsubo (1992), and

Kobayashi (2012a) have investigated the semantics of these perfective auxiliaries in Heian

period kundokubun texts. They primarily find nu governing intransitive and passive

predicates and tu governing transitive and causative predicates. As we saw in section 2.1

above, these auxiliaries’ proclivities regarding transitivity and valency are seen in

contemporary secular texts as well. Verbs commonly governed by the endoactive

perfective nu in kundokubun are naru (なる ‘become’), siru (知る ‘learn’), and itaru (至

る ‘reach’). Examples of verbs that the exoactive perfective tu often governs are u (得

‘gain, acquire’), kiku (聞く ‘hear, listen’), and usinapu (失ふ ‘lose (something)’).

Among the three early Heian kundokubun data sources, a total of 140 instances of

the endoactive perfective nu and 120 instances of the exoactive perfective tu were found

in sentence-final position. The following subsections include quantitative and qualitative

analyses of the transitivity, lexical aspect, and narrative discourse function of sentences

ending in these perfective auxiliaries.

129 The findings in this section are to be published in Bundschuh (2021c).

123

5.1.1 Transitivity of Sentence-final Perfectives

As previous studies cited above have established, the choice between these

auxiliaries is primarily one of transitivity, which bears out in their use in the kundokubun

texts. Table 5.1 overviews the transitivity of sentences headed by nu and tu.

Table 17 Transitivity of Sentence-final Perfectives

Perfective low (%) medium (%) high (%)

nu 114 (81%) 25 (18%) 1 (1%)

tu 18 (15%) 70 (58%) 32 (27%)

As expected, we find nu governing primarily low-transitivity predicates, such as the

following prototypical example, which was introduced in (62) above.

(66) 仏の所に来詣しぬ。 (Kabutogi & Nakada 1979: 78)

potoke no tokoro ni raikei.si-nu

buddha GEN place LOC approach-NU

‘[Each] approached the place of the Buddha.’

Verbs of motion in kundokubun, such as raikei.su (来詣 ‘approach’), are semantically

intransitive changes of state, making them common predicates to be governed by the

endoactive perfective nu when used in discourse. Auxiliary tu, in contrast, is found

primarily governing predicates of medium transitivity, as in the following examples.

(67) 一切の法に於て勇健の想を得つ。 (Kabutogi & Nakada 1979: 110)

issai no papu ni oki-te yūgon no sau

all GEN dharma LOC regard-GER robust GEN thought

wo e-tu

ACC gain-TU

‘They attain robust thoughts regarding all of the dharma.’

124

(68) 佛の此の甚深の空性を說きたまふを聞きつ。 (Tabuchi 1987: 58)

potoke no ko no zintan no kūsyau wo

buddha GEN this GEN deep GEN emptiness ACC

toki-tamapu wo kiki-tu

explain-HON ACC hear-TU

‘They heard the Buddha explaining this deep emptiness.’

Although what is ‘gained’ in (67) is intangible, the verb u (得 ‘gain, acquire’) is coded as

medium-transitivity due to the lack of change upon the thoughts that are attained. This

verb is almost uniformly governed by tu (in its participial form e, hence e-tu above) when

used in a narrative sequence of events. We find tu primarily governing predicates of

medium transitivity in kundokubun, such as in (68).

Verbs of listening and speaking, such as kiku (聞く ‘hear’) in (68), take that which

is spoken as an argument with the accusative marker wo, as we find here, making the

sentence medium-transitivity. The sutras investigated in this study are full of embedded

conversations, and verbs of speaking and listening are both plentiful and often governed

by tu in narratives. Thus, we find a higher percentage of medium transitivity in tu’s use

among these texts.

The 25 instances of medium transitivity nu occur predominantly when it governs

siru (知る ‘learn’), as in the following example.

(69) 法を護る人は、諸法の一味を知りぬ。 (Nakada 1980: 187)

papu wo mamoru pito pa syopapu no itimi

dharma ACC protect person TOP all.dharma GEN nature

wo siri-nu

ACC learn-NU

‘One who protects the dharma learns the nature of all of the dharma.’

125

Although siru ‘learn’ takes an argument in the accusative case, the change is both internal

to the subject and not associated with volition, thus endoactive, and so we find it

governed by nu.

The nature of kundokubun being based on translation also leads to some

predicates governed by nu with uncharacteristically high transitivity, given the auxiliary’

general association with endoactivity. Kasuga (1985b: 230) notes that, due to a strong

association between the Sinitic character 已 (‘complete, finish’) and auxiliary nu, there

are cases of nu being used for a kundokubun rendition of 已 when, as Japanese, tu would

be a more likely candidate for perfectivity marking given the context. Below are three

such examples.

(70) 親近すること得已。 (Kasuga 1985a: 28)

sinkin.suru koto e-nu

become.close NMLZ gain-NU

‘They are able to become close.’

(71) 是の語を說き已。 (Kasuga 1985a: 122)

ko no go wo toki-nu

this GEN word ACC expound-NU

‘They expound these words.’

(72) 仏、是の経を説き已りたまひぬ。 (Kabutogi & Nakada 1979: 102)

potoke ko no kyau wo toki-wopari-tamapi-nu

buddha this GEN sutra ACC expound-finish-HON-NU

‘The Buddha finished expounding this sutra.’

126

The verb u (得 ‘gain, acquire’) is most often governed by tu when marked for

perfectivity, as we see in (67) above. However, due to both the use of the Sinitic

character 已 and the semantics of /non-past clause + koto u/, which is a periphrastic

potential expression, in (70) the perfective auxiliary nu was used in the kundokubun

translation. Verbs of speaking and listening, such as toku (説く ‘expound’) found in (71)

and (72) and kiku (‘hear’) in (68) above, are generally governed by the perfective aspect

marker tu due to their taking an accusative argument and bringing a change to the scene

external to the subject of these verbs. The strong association with the sinograph 已, even

when it is explicitly read as woparu (‘finish’) in (72), presumably causes the predicate to

be marked for perfective aspect with nu (see discussion in section 4.5.1 above). We now

turn to the relation the two perfective auxiliaries have with lexical aspect.

5.1.2 Lexical Aspect of Sentence-final Perfectives

The following table summarizes the lexical aspect of sentence-final predicates

headed by one of the perfective auxiliaries.

Table 18 Lexical Aspect of Sentence-final Perfectives

Perfective State (%) Activity (%) Accomplishment (%) Achievement (%)

nu 0 (0%) 0 (0%) 14 (10%) 126 (90%)

tu 0 (0%) 3 (3%) 20 (17%) 97 (81%)

The preponderance of telic sentences, i.e., those expressing situations with a given

endpoint such as Accomplishments and Achievements, is due to the semantics of

127

perfective aspect. This temporal aspect requires an endpoint and thus the lexical aspect of

sentences headed by these auxiliaries mostly follows this restriction. At 90% and 81%,

respectively, both nu and tu primarily head Achievements, or changes of state with no

duration. Examples (66) and (67) above, raikei.si-nu (approach-NU) and e-tu (attain-TU),

respectively, are representative of these changes of state. As mentioned above, verbs of

motion such as raikei.su are not durative in kundokubun nor other varieties of registers of

Japanese, since they signify the moment of arrival.

We find the perfective auxiliaries performing a similar role in wabun texts, with

Quinn (1987: 283) finding “in classical Japanese, the Perfectives -tsu and -nu are two

suffixes which function often in marking situations as Achievements and

Accomplishments, since they express, respectively, Exoactive realization and Endoactive

realization.”

Returning to the kundokubun data, about a quarter of the tokens of each perfective

govern durative events with endpoints, or Accomplishments. This is because verbs that

indicate Activities when in their bare form generally indicate Accomplishments when

governed by a perfective, such as the case of kiku (‘listen’) in (68) above. The activity

verb kiku gains an endpoint and that endpoint is marked grammatically with tu, resulting

in the durative but telic Accomplishment kiki-tu.

The data include three cases of auxiliary tu marking an Activity, or a durative,

atelic event with no clear end point. Below is one such example.

128

(73) 貌端嚴にして、父母に偏に愛念せられつ。 (Kasuga 1985a: 192)

meutangon.ni.si-te bumo ni pitoe ni

be.beautiful-GER father.mother ABL sincere COP

ainen.se-rare-tu

deeply.love-PASS-TU

‘He was beautiful and sincerely deeply loved by our parents.’

The speaker is describing his younger brother, who had recently sacrificed himself to

feed a mother tiger and her cubs. Here, ainen.se-raru ‘be deeply loved’ is an activity

from the recent past. Auxiliary tu is also known as a marker of recent past (Quinn 1987,

Suzuki 1999, Watase 2013, among others), a use we find, albeit rarely, in early Heian

Buddhist texts. The recency here is from the speaker’s perspective. He is reporting a

recent event that involves him personally, so the evidence is high, resulting in this recent

past being indexed with tu rather than ki. In the following section we see how these two

perfectives, with their affinity with changes of state, function in the discourse.

5.1.3 Discourse Function of Sentence-final Perfectives

The table below outlines the discourse functions of the auxiliaries nu and tu in

early Heian kundokubun. Because there were no Abstracts, Orientations, Evaluations, or

Codas with a sentence-final perfective, they are not included in the table.

Table 19 Discourse Function of Sentence-final Perfectives

Perfective Quotation (%) Complication (%) Result (%)

nu 16 (11%) 115 (82%) 9 (6%)

tu 31 (26%) 87 (73%) 2 (2%)

129

Both perfectives are used a small percentage of the time, 11% for nu and 26% for tu, in

non-narrative Quotations. In narrative passages they are predominantly found in

Complications, which are events that move the story forward, such as the following

example.

(74) 審に射て象王の心を中りつ。 (Nakada 1958: 68)

tumapiraka ni i-te zauwau no kokoro wo

clear COP shoot-GER elephant.king GEN heart ACC

yaburi-tu

pierce-TU

‘Shooting clearly, they pierced the heart of the elephant king.’

Here, tu governs yaburu (‘pierce’) to depict the slaying of the elephant king, one of the

primary events of the narrative passage from which it is drawn. Although auxiliary nu is

found in Complications 82% of the time, it also heads nine Results at the end of narrative

passages among the three sutras, such as the following.

(75) 右に三帀遶りて、退きて一面に坐しぬ。 (Kasuga 1985a: 1)

migi ni sansapu mapari-te sirizoki-te itimen

right LOC three.times circumambulate-GER withdraw-GER one.side

ni za.si-nu

LOC sit-NU

‘Having circumambulated to the right three times, they withdrew and sat down to

one side.’

(76) 彼の池の邊に至りぬ。 (Tabuchi 1987: 82)

ka no ike no potori ni itari-nu

that GEN pond GEN side LOC arrive-NU

‘They arrived at that lake.’

130

The line in (75) is repeated throughout the Golden Light Sutra as characters are

introduced and sit around the Buddha before another group appears upon the stage. Verbs

of motion, such as itaru (至る ‘arrive’) found in (76), are often used at the end of scenes.

The sentence following (76) begins with ko no toki ni (是の時に ‘at this time’), which is

a common way a new scene is introduced in sutra narration. Their use sentence-finally in

quotations is often when speakers use change-of-state verbs, such as nari (‘become’),

which is most often governed by nu, and u (‘gain, attain’), which, as mentioned above, is

often governed by tu.

5.1.4 Conclusions Regarding nu and tu in Kundokubun Narration

In this subsection I have presented data regarding how the endoactive perfective

auxiliary nu and the exoactive perfective auxiliary tu relate to transitivity, lexical aspect,

and discourse function of sentences in early Heian kundokubun narration. The following

table summarizes the above findings.

Table 20 Overview of Sentence-final Perfectives

Perfective Transitivity Lexical aspect Discourse function

nu low (81%) Achievement (90%) Complication (82%)

tu medium (58%) Achievement (81%) Complication (73%)

We find that the transitivity of sentences headed by nu and tu are as expected given their

respective endo- and exoactivity, particularly when the general lack of high transitivity

predicates in sutra narration is considered. As perfectives that, beyond the semantic

extension of tu in which it is used to mark recent past tense, indicate changes of state,

131

most sentences they head are Achievements, or instantaneous telic events. Regarding

discourse function, both most often are used in Complications, or dynamic events that

advance the narrative.

Although we find perfective nu in nine Results, i.e., at the end of narrative

passages, as we saw in (75) and (76) above, we only find tu performing that same role

twice. This is because scene changes in Buddhist narratives most often include

intransitive verbs of motion. In other words, it is the difference in transitivity that leads

nu to being the more common perfective used in Results. However, a reason nu is still

generally limited to Complications is that Results in kundokubun are most often headed

by the established fact, or past, auxiliary ki, such as in the following example.

(77) 座ヨリして（而）起（ち）て、其の本處に還（り）にキ。 (Kasuga 1985a: 19)

za yori.si-te tati-te so no ponsyo ni

seat be.from-GER stand-GER that GEN original.place LOC

kaperi-ni-ki

return-NU-KI

‘He stood up from his seat and returned to his original place.’

The sentence concluding in (77) begins with Meudau bosatu pa potoke no asi o rei.si-

tatematuri-wopari-te (妙幢菩薩は佛の足を礼したてまつり已りて ‘Ruciraketu, having

finished paying obeisance to the Buddha’s feet’). Although the verb ‘return’ (還る

kaperu) is governed by the perfective nu to show its completion, because it is at the

conclusion of a scene, auxiliary ki is employed by the narrator to perform its usual role in

kundokubun narration, which is framing the discourse. This role is discussed further in

section 5.3 and chapter 6 below.

132

Kundokubun is limited in its lexical selection because it is the result of translating

Sinitic texts essentially word-for-word into Japanese. However, this study shows that the

Early Middle Japanese perfective auxiliaries nu and tu were used in translating these

religious texts in ways parallel to how we find them in more secular early Japanese

narratives. Although occasionally limited by translation conventions, such as the

correlation between the Sinitic character 已 and nu displayed in section 5.3.1 above, the

monks who rendered these Sinitic sutras into Japanese were clearly aware of their

overarching narrative structure as they used the two perfective auxiliaries at their disposal

to convey the dynamic stories embedded within the texts to a Japanese audience at the

onset of the Heian period.

5.2 Auxiliaries ari and tari in Heian-period Kundokubun Buddhist Texts

Among the three early Heian kundokubun data sources, a total of 309 instances of

the stative auxiliary ari and 101 instances of the stative auxiliary tari were found in

sentence-final position. We find over three times as many tokens of auxiliary ari because

tari is emergent in the early Heian period.

The following subsections include quantitative and qualitative analyses of the

transitivity, lexical aspect, and discourse function of sentences ending in these stative

auxiliaries.

133

5.2.1 Transitivity of Sentence-final Statives

The table below presents the number of low-, medium-, and high-transitivity

sentence-final predicates governed by the kundokubun stative auxiliaries ari and tari.

Table 21 Transitivity of Sentence-final Statives

Stative low (%) medium (%) high (%)

ari 154 (50%) 154 (50%) 1 (0%)

tari 55 (54%) 40 (40%) 6 (6%)

Both auxiliaries ari and tari end sentences of predominantly low- and medium-

transitivity. As mentioned above, Buddhist kundokubun texts contain few high-

transitivity predicates, so the data suggest transitivity is not a factor at play when

selecting one of the stative auxiliaries. The following example presents the single case of

a high-transitivity sentence headed by the stative auxiliary ari among the data set.

(78) 人天供養福を得ルこと、無邊にアラシメタマヘル。 (Kasuga 1985a: 14)

ninten kuyaupuku wo eru koto mupen ni

human.devas make.offering.merit ACC attain NML immense COP

ara-sime-tamape-ru

be-CAUS-HON-ARI

‘Has the attainment of merit by humans and devas making offerings been made

immense (by the Buddhas)?’

This example is from a Quotation by the bodhisattva Ruciraketu (J. 妙幢 Meudau)

asking the Buddha a question.130 Auxiliary ari is heading a paraphrastic causative

130 EMJ kundokubun had a syntactic rule that sentence-final predicates in information-seeking questions be

referentialized in their adnominal form, which is how we find ari (as aru) here.

134

construction, with the verb ari following the participial form of the copula nari. Below is

an example of auxiliary tari in a high-transitivity sentence-final predicate.

(79) 能ク汝等に現に勝（れ）たる報を受ケ令（め）たり。 (Kasuga 1985a: 100)

yoku nanzi-ra ni gen ni masare-taru pou wo

successfully you-PL to really COP surpass-TARI reward ACC

uke-sime-tari

accept-CAUS-TARI

‘I have successfully caused you all to receive the reward of actually having

surpassed.’

This sentence is from a Quotation of the Buddha (J. 世尊 seson) praising the four divine

kings (J. 四天王 sitennau) after they say they will respect the Golden Light Sutra and

protect those who chant it. Auxiliary tari is used to indicate that the predicate ‘cause to

accept’ is a state relevant to the conversation (i.e., a perfect). Its governing of a causative

predicate led (79) to be coded as a high-transitivity sentence. We now turn to the lexical

aspect of the statives in the data set.

5.2.2 Lexical Aspect of Sentence-final Statives

The following table summarizes the lexical aspect of sentence-final predicates

headed by one of the stative auxiliaries.

Table 22 Lexical Aspect of Sentence-final Statives

Stative State (%) Activity (%) Accomplishment (%) Achievement (%)

ari 301 (97%) 0 (0%) 8 (3%) 0 (0%)

tari 101 (100%) 0 (0%) 0 (0%) 0 (0%)

135

First, there are no sentence-final Activity or Achievement predicates headed by the

stative auxiliaries. As a reminder, the lexical aspect coding was done with the full

predicate morphology and context taken into account, which has led to the admittedly

self-evident and circular conclusion that stative auxiliaries primary head sentences that

depict States in early Heian kundokubun texts, such as the following.

(80) 梵王譬如幻の師あり、及幻の弟子ありて善ク幻術を解レリ。 (Kasuga 1985a: 93)

bonwau tatope-ba maborosi no si ari oyobi

Brahmā131 compare-COND illusion GEN master be and

maborosi no desi ari-te yoku genzyutu wo

illusion GEN apprentice be-GER well magic ACC

wakare-ri

understand-ARI

‘Brahmā, for example there is an illusion master and an illusion apprentice, and

they understand magic well.’

(81) 良福田とするに堪へたり。 (Nakada 1958: 72)

ryaupukuden to suru ni tape-tari

field.of.blessedness132 COM do LOC endure-TARI

‘You have endured to be of the field of blessedness.’

In the first example, the shining celestial nymph of the wish-fulfilling gem (J. 如意寶光

耀女天 nyoipou kwau’eu tennyo; introduced in (7) above) is presenting an example to

Brahmā of two illusionists, or magicians. The verb wakaru ‘understand’ is governed by

131 More often rendered 梵天 (J. bonten) in Sinitic (Soothill & Hodous 2014: 354). See Buswell & Lopez

(2014: 141) for greater detail in how this Hindu deity was interpreted as a protector of Buddhist teachings.

132 Soothill & Hodous (2014: 243). See puṇyakṣetra (Skt. for ‘field of merit’) in Buswell & Lopez (2014:

682).

136

auxiliary ari to denote their state of understanding magic. In (81) we find tari governing

the lower bigrade verb tapu (堪ふ ‘endure’) to index the state resulting from having

endured and refrained from impurities to be worthy of the field of merit. These words are

spoken by the Buddha (J. 佛 potoke) to Kṣitigarbha (J. 地藏 zizau) quoting Rākṣasa (J.

羅剎 rasetu) to criminals dressed as monks in a verse. Both of these examples of States

are found in dialogues, which, as I discuss in the following subsection, are the most

common place to find the stative auxiliaries in Buddhist kundokubun texts.

The eight examples of auxiliary ari heading Accomplishments are the eight cases

in the Immeasurable Meanings Sutra where it governs notamapu (say.HON) after lines

spoken by the Buddha, which I have discussed in section 4.5 above.

5.2.3 Discourse Function of Sentence-final Statives

The table below outlines the discourse functions of the stative auxiliaries ari and

tari in early Heian kundokubun. Because there are no Abstracts or Codas with a sentence-

final stative auxiliary in the data, they are not included in the table. Furthermore, two of

the examples of auxiliary ari governing notamapu (say.HON) in the Immeasurable

Meanings Sutra occur before a new paragraph begins with so no toki ni (爾（の）時に ‘at

that time’), leading to them being coded as Results. However, when rounded to the

nearest percentage point, I calculate that stative ari heads Results only 1% of the time.

Due to this and there being no cases of tari in such sentences, Results are also excluded

from the analysis below.

137

Table 23 Discourse Function of Sentence-final Statives

Stative Orientation (%) Quotation (%) Complication (%) Evaluation (%)

ari 26 (8%) 241 (78%) 6 (2%) 34 (11%)

tari 5 (5%) 83 (82%) 0 (0%) 11 (11%)

Both stative auxiliaries are overwhelmingly found in Quotations in kundokubun

narration, at 78% for auxiliary ari and 82% for tari. The examples provided in the

previous section are representative, in that the stative auxiliaries are often used in

Quotations to describe people (or personified beings).

The most common discourse function of the stative auxiliaries in narrative

passages is to pause the narration and provide background information to the scene in

Evaluation sentences, such as the following.

(82) 天の厨、天の鉢器には、天の百味、充満し盈溢せり。 (Kabutogi & Nakada

1979: 121)

ten no dyu ten no patipin ni pa

heaven GEN kitchen heaven GEN vessels LOC TOP

ten no pyakumi diuman.si yauman.se-ri

heaven GEN hundred.flavors133 being.filled be.filled-ARI

‘Heaven’s kitchens and heaven’s vessels have been completely filled with all the

good flavors of heaven.’

In this example, the narrator is describing the abundance of heaven’s flavors that have

been prepared by the Buddha and bodhisattvas for their disciples. After this description,

the narrator describes the world shaking with the words of the Buddha.

133 See Soothill & Hodous (2014: 217).

138

Along with Evaluations, the stative auxiliaries are also found in Orientations. For

example, when describing the monks who are gathered to listen to the Buddha expound

the Immeasurable Meanings Sutra, the narrator first describes them as arhats without

bonds and then says the following.

(83) 真正解脱せり。 (Kabutogi & Nakada 1979: 82)

sinsyau getatu.se-ri

truly become.enlightened-ARI

‘They truly have become enlightened.’

The use of stative auxiliaries to describe people and other beings after they are introduced

in the narrative is another one of their important functions in kundokubun narration.

5.2.4 Conclusions Regarding ari and tari in Kundokubun Narration

In this subsection, I have given an overview of how the stative auxiliaries ari and

tari relate to transitivity, lexical aspect, and discourse function of sentences in early

Heian kundokubun texts. The following table summarizes the above findings.

Table 24 Overview of Sentence-final Statives

Stative Transitivity Lexical aspect Discourse function

ari unrestricted State (97%) Quotation (78%)

tari unrestricted State (100%) Quotation (82%)

Both stative auxiliaries are unrestricted with regards to transitivity. Auxiliary ari evenly

heads low- and medium-transitivity sentences, with auxiliary tari at 54% low and 40%

medium. The general lack of high-transitivity sentence marking is only due to the fact

139

that there are relatively few actions depicted in the sutras that cause change onto a direct

object. We find more causatives in Quotation, but they are more often unmarked for TAM.

The lexical aspect of sentences that end in the stative auxiliaries, as mentioned

above, are almost all States. It is worth noting here that, although these auxiliaries are

often rendered as teiru in the Japanese of today, they lack its semantic underspecification,

i.e., whereas the auxiliary teiru today may index habitual, continuative, or stative aspect,

its EMJ predecessors were generally limited to the latter (stative aspect) (Oda 2015: 139).

Although the primary discourse function of the stative auxiliaries in early Heian

kundokubun is Quotations, this stems from their pragmatics of providing background

information, whether describing people or scenes in Quotations, at the beginning of

narrative passages (Orientations) or in the middle of narratives (Evaluations).

5.3 Auxiliaries ki and keri in Heian-period Kundokubun Buddhist Texts

As we see throughout this study, the established fact modal auxiliaries ki and keri

display a striking imbalance in their use in early Heian kundokubun Buddhist texts. After

auxiliary ari, the established past fact auxiliary ki has the most tokens throughout the three

examined sutras, with a total count of 229 sentences headed by the past established fact

auxiliary. The externally established fact auxiliary keri, on the other hand, only has 30

sentence-final tokens throughout the entirety of the three texts. For the remaining of chapter

5, I discuss the transitivity, lexical aspect, and discourse function of sentences headed by

the established fact auxiliaries ki and keri and conclude discussing their semantics and

pragmatics in the linguistic register of kundokubun.

140

5.3.1 Transitivity of Sentence-final Established Fact Modals

The table below contains the transitivity scores for sentences among the three

sutras headed by one of the established fact modal auxiliaries.

Table 25 Transitivity of Sentence-final Established Fact Modals

Modal low (%) medium (%) high (%)

ki 180 (79%) 42 (18%) 7 (3%)

keri 29 (97%) 1 (3%) 0 (0%)

We find both established fact modals in predominantly low-transitivity sentence-final

predicates, with their numbers decreasing as transitivity raises. However, this is likely the

result of there being more low-transitivity predicates than medium- or high- in

kundokubun narration rather than a semantic feature of these auxiliaries.

Below is the only case of keri governing a medium-transitivity predicate

throughout the texts investigated in this study.

(84) 善哉、善哉。善男子。善く是の時を知りけり。 (Kabutogi & Nakada 1979: 92)

zen’ya zen’ya zennansi yoku ko no toki wo

excellent excellent gentlemen well this GEN time ACC

siri-keri

learn-KERI

‘How very excellent, excellent, gentlemen, (the fact) that you will learn well of

this time.’

In this example, presented in (48) above, the Buddha is praising the bodhisattva

Mahāvyūha for asking about enlightenment, and in the subsequent lines the Buddha tells

his interlocutor that it does not take long to reach. This sentence is coded as medium-

141

transitivity because the verb siru ‘learn’ takes the direct object, ko no toki ‘this time.’

Auxiliary keri is also here due to the syntactic structure. Lines that begin with zen’ya (善

哉 ‘excellent’) most often end with expressive morphology in the subsequent predicate

emphasizing what is excellent. The pattern most often found here is aku nominalization

(the same morpheme we see in pre-speech-act adverbial phrases, e.g., ipaku, speak-NML,

‘speaking’). In zen’ya (善哉 ‘excellent’) constructions, the aku nominalization can be

best understood as ‘that (VP),’ which is similar to the English rendition I often adopt for

keri in kundokubun, ‘the fact that (VP).’

Furthermore, (84) presents evidence that a tense-based analysis of keri is

insufficient, as in the previous lines Mahāvyūha was expressing his confusion, saying 審

（らか）に（あら）不 (tumabiraka ni ara-zu ‘It is unclear’). In other words, the Buddha is

not expressing pleasure of a past fact, ‘that Mahāvyūha learned,’ but a fact regarding

what is to come, ‘that Mahāvyūha will learn,’ from the Buddha’s teaching.

There are also cases of accusative wo in verb phrases governed by keri that I

coded as low-transitivity, such as the following example.

(85) 世尊、上に說（き）つル所の如ク、菩提の正行をばすベクありケリ。 (Kasuga

1985a: 91)

seson upe ni toki-turu tokoro no gotoku bodai

Lord above LOC expound-TU place GEN like enlightenment134

no syaugyau wo ba su-beku ari-keri

GEN right.deeds135 ACC TOP do-MODAL be-KERI

134 Skt. bodhi (Buswell & Lopez 2014: 129).

135 Soothill & Hodous (2014: 193).

142

‘Lord, (I now realize the fact that) I must do the right deeds of enlightenment as

you have previously expounded.’

The above example is from a Quotation spoken by the shining nymph (J. 如意寶光耀天

女 nyoipou kwau’eu tennyo; introduced in (7) above) in response to the Buddha.

Although there is an accusative particle wo in the verb phrase, the verb su (‘do’) is

governed by the modal besi, here in its periphrastic participle beku ari being governed by

keri. Because the sentence is conveying a fact regarding what should be done, the

transitivity is essentially too far embedded for the sentence’s transitivity to be coded as

higher rather than low. She is using auxiliary keri to index the fact regarding what she

must do as the Buddha just preached. We find the adnominal form of auxiliary tu as a

recent past in this line as well.

We conclude this subsection with a representative high-transitivity sentence

example of ki.

(86) 此（の）陀羅尼は能（く）（令）一切の智慧を猛利にあり、煩惱の賊を破（せ）しめ

き。 (Nakada 1958: 23)

ko no darani pa yoku itisai no

this GEN mnemonic.device136 TOP successfully all GEN

tiwe wo myauri ni ari bonnau no zoku wo

knowledge ACC fierce COP be affliction137 GEN thief ACC

pa.se-sime-ki

be.defeated-CAUS-KI

136 Skt. dhāraṇī; a mnemonic device that often summarizes Buddhist principles but includes mantra

(‘spells’) (Buswell & Lopez 2014: 438).

137 Skt. kleśa (Buswell & Lopez 2014: 438).

143

‘This mantra, being fierce with all knowledge, successfully caused the thief of

afflictions to be defeated.’

Here Kṣitigarbha (J. 地藏 Zizau) is speaking to the Buddha (J. 佛 Potoke) and is about

to expound a mantra (J. 呪 zyu ‘spell’). This sentence displays high transitivity due to

the causative auxiliary simu being governed by auxiliary ki. Past causative constructions

in dialogue are where we find most high-transitivity sentence ki-marking. We now turn to

the lexical aspect of sentences headed by the established fact auxiliaries.

5.3.2 Lexical Aspect of Sentence-final Established Fact Modals

The following table summarizes the lexical aspect of sentence-final predicates

governed by the past established fact auxiliary ki and the externally established fact

auxiliary keri in early Heian kundokubun Buddhist texts.

Table 26 Lexical Aspect of Sentence-final Established Fact Modals

Modal State (%) Activity (%) Accomplishment (%) Achievement (%)

ki 87 (38%) 29 (13%) 59 (26%) 54 (24%)

keri 30 (100%) 0 (0%) 0 (0%) 0 (0%)

As discussed in chapter 4, auxiliary ki does not display any particular affiliation to a

specific lexical aspect. Its highest percent of use is in States, at 38%. This is likely due to

the kind of sentences we find in Orientations, which I leave for discussion in the

following subsection.

The externally established fact auxiliary keri, on the other hand, is exclusively

employed in sentences coded as States, such as (84) and (85) above. Every token of keri

144

in kundokubun is indexing the speaker’s commitment to the factuality of what he is

saying, a state of understanding, so to speak. We also find more emotionally charged

examples, such as the following.

(87) 王子は死（に）たまひケリ。 (Kasuga 1985a: 198)

wausi pa sini-tamapi-keri

prince TOP die-HON-KERI

‘The prince has died.’

This example is of a character announcing to a populace that a missing prince has been

found dead in a parable depicted as being told by the Buddha to Ānanda (J. 阿難陁

Ananda). The speaker, who is crying as he makes the announcement, is relaying new

information in the highest emotional state. In both kundokubun and varieties of Japanese

today, grammatically marking a clause as factual (generally by referential nominalization

today using the nominalizer no) has the pragmatic extension of indexing one’s emotional

response to that fact, whether positive, such as the zen’ya (‘excellent’) keri example in

(84) and the marking of a tragic fact in (87) above. We now turn to the discourse

functions of these auxiliaries of established fact.

5.3.3 Discourse Function of Sentence-final Established Fact Modals

The table below outlines the discourse functions of the auxiliaries ki and keri in

early Heian kundokubun. There are no tokens of either occurring sentence-finally in an

Abstract, so this discourse function is not included in the table.

145

Table 27 Discourse Function of Sentence-final Established Fact Modals

Modal Orient. (%) Quot. (%) Comp. (%) Eval. (%) Res. (%) Coda (%)

ki 75 (33%) 49 (21%) 59 (26%) 6 (3%) 39 (17%) 0 (0%)

keri 0 (0%) 29 (97%) 0 (0%) 0 (0%) 0 (0%) 1 (3%)

Although auxiliary ki is found most often in Orientations at 33% of its total, it is not

significantly more than the amounts we see in Quotations and Complications, at 21% and

26%, respectively. The auxiliary’s high number of tokens in Orientations do, however,

help explain why it heads States at a higher rate than other lexical aspects. In other words,

scenes established as States are often employed to introduce the new setting before the

action begins, such as the Orientation of the tale of the prince who sacrificed himself to

starving tigers out of compassion (whose death was announced in (87) above) in the GLS,

repeated by the Buddha in a verse.

(88a) 昔の時に大國有（り）キ。 (Kasuga 1985a: 196; same for all (88) examples)

mukasi no toki ni daikoku ari-ki

long.ago GEN time LOC large.kingdom be-KI

‘In a time long ago, there was a large kingdom.’

(88b) 國主をば大車と名（づけ）キ。

kokusu wo ba daisya to naduke-ki

ruler ACC TOP Mahāratha138 COM be.named-KI

‘The ruler was named Mahāratha.’

(88c) 王子をば勇猛と名（づけ）キ。

138 Skt. for ‘great chariot.’ The names of these characters are from Emmerick’s English rendition of the GLS

(Emmerick 1970: 87).

146

Wausi wo ba yuumyau to naduke-ki

prince ACC TOP Mahāsattva139 COM be.named-KI

‘The prince was named Mahāsattva.’

(88d) 常に心に施するに悋ルこと無（かり）キ。

tune ni kokoro ni se.suru ni yabusakaru koto

always COP mind LOC practice.compassion LOC lack NML

nakari-ki

have.NEG-KI

‘He never had a lack of compassion in his heart.’

(88f) 王子い二の兄有（り）キ。

wausi i ni no ani ari-ki

prince NOM two GEN older.brother have-KI

‘The prince had two older brothers.’

(88g) 大渠と大天と号（ばれ）キ。

daigo to daiten to yoba-re-ki

Mahāpraṇāda140 and Mahādeva141 COM call-PASS-KI

‘They were called Mahāpraṇāda and Mahādeva.’

All six lines of the Orientation to the verse are Statives that end in auxiliary ki. The

sentence that follows depicts the three brothers setting off into the woods and is headed

by the perfective auxiliary nu as the action begins.

After Orientations, auxiliary ki is used the most in Complications. These are mostly

speech-act verbs such as notamapu (speak.HON), discussed in section 4.5 above. I examine

139 The Sinitic rendition means ‘valor.’ The name in Skt. means ‘great being’ (See Buswell & Lopez 508).

140 The Sinitic rendition means ‘great leader’ (or ‘great ditch,’ but I assume the former). The name in Skt.

means ‘great sound.’

141 Both the Sinitic rendition and Skt. mean ‘great deity.’

147

the reason the past established fact auxiliary is often used after quotations in more detail in

section 6.2 below and argue it is to reestablish the past factuality of the discourse after a

quote. In Quotations, ki often is used to comment on a past situation, which we see in the

following line by the Buddha from the IMS.

(89) 諸の衆生は性欲不同なりといふ（ことを）知（り）て、種種に法を説（き）き。

(Kabutogi & Nakada 1979: 98).

moro no syuzyau pa syauyoku pudou nari to

all GEN living.beings TOP nature.desires not.same COP COM

ipu koto wo siri-te syusyu ni papu wo

say NML ACC learn-GER various COP dharma ACC

toki-ki

expound-KI

‘Having learned that not all living beings share the same natural desires, I

expounded the dharma in various ways.’

The Buddha goes on to tell his interlocutor Mahāvyūha how he used expedient means to

more effectively teach the dharma, or Buddhist law. The use of auxiliary ki in the Results

of extended narration has been discussed in sections 4.1 and 4.3 above. We often find it

before a new paragraph, or focus of the narrative, begins. These shifts are often marked by

so no toki ni ‘at that time’ in both the outermost and embedded narration and vocatives in

the latter.

Turning to auxiliary keri, we find it 97% of the time in quotations. In other words,

it is essentially not employed in narrative passages. The one case of it being used as a Coda

was introduced in section 4.4 on the TWS above, which I reproduce here.

(90) 无睱の傍生趣の身を受（け）たりと雖ども、而（も）袈裟を敬（ひ）て（不）惡業を造

ラズアルナリケリ。 (Nakada 1958: 74)

148

muka no bausyauzyu no mi wo uke-tari

no.leisure142 GEN rebirth.as.an.animal GEN body ACC receive-TARI

to ipe-domo sikamo kesa wo uyamapi-te

COM say-although however buddhist.robes ACC respect-GER

akugyau wo tukura-zu aru nari-keri

evil.deeds ACC make-NEG have COP-KERI

‘Although they say bodies reborn as animals have no leisure to follow the Buddhist

path, the fact is that they do, however, respect the Buddhist robes and will not

cause harm to them!’

The Buddha (or, more accurately, the monk who glossed the sutra) uses this Coda to return

the discourse to the present moment of the conversation between the Buddha and

Kṣitigarbha after telling the tale of how even a drunken elephant will not harm one who

wears Buddhist robes. We can say, however, that Codas of embedded narratives, as points

of return to the dialogue, share the same embedded discourse layer (introduced in Figure 2

above) as Quotes.

5.3.4 Conclusions Regarding ki and keri in Kundokubun Narration

Throughout this subsection I have presented data and analyses regarding how the

modal auxiliaries ki and keri relate to the transitivity, lexical aspect, and discourse

function of sentences in early Heian kundokubun narration. The following table

summarizes the above findings.

142 An existence in which one has no leisure to follow the Buddhist teachings (see “八無暇” in Soothill &

Hodous 2014: 38).

149

Table 28 Overview of Sentence-final Established Fact Modals

Modal Transitivity Lexical aspect Discourse function

ki low (79%) unrestricted Orientation (33%)

keri low (97%) State (100%) Quote (97%)

We find ki and keri predominantly heading low-transitivity sentences, at 79% and 97%,

respectively. However, as discussed in section 5.3.1 above, within the sutras’

preponderance of low-transitivity predicates, ki and keri occur in different circumstances.

Auxiliary ki is employed mostly on the peripheries of a narrative sequence, where lower-

transitivity scene establishment (such as the Orientations in (9) and (88) above) and scene

shifting (in (3) above) occur. In contrast, keri is almost exclusively employed by

characters commenting on a state of affairs they have come to realize is true (such as in

(85) above), such states of affairs generally being of low-transitivity.

The established past fact auxiliary ki heads sentences regardless of their lexical

aspect, and the discourse function of the sentences it heads most, Orientations, only take

33% of its marking. This suggests ki does not show a strong preference for a particular

discourse function among the Nara (2011)/Labov (1972) characterizations. On the other

hand, the externally established fact auxiliary keri is limited to heading States in

Quotations (at 97%) because it governs predicates describing states of affairs in order to

index the speaker’s commitment to their factuality. Thus, the translators who rendered

Sinitic sutras into kundokubun most often used auxiliary keri to index the grammatical

category of mirativity in their renditions—they used it to mark sentences presenting new

information (DeLancey 1997) or a character’s information update process (Lau &

Rooryck 2017).

150

Returning to the narrative function of the established past fact auxiliary ki, in

section 4.5.4, we briefly examined grammatical differences found in parallel structures—

i.e., variation in kundokubun renditions of the same repeated Sinitic sentence.

Although beyond the scope of this study, it is worth noting here that there are numerous

parallel structures in Buddhist texts, regardless of language. Du Bois’s (1986) study on

cross-linguistic structures in ritual speech makes the following keen observation between

parallelism and established factuality.

Parallelism promotes a perception of the utterance as an artifact—what we may

call a ‘speech tool’ rather than a ‘speech act’—and hence makes possible the

manticist equation of ritual speech with analytic truth. The basis of the authority

of ritual speech is in the end quite simple. The manticist listener, through the

structure of ritual speech and of the ritual event, is put directly in touch with a

sourceless message whose authority he can observe in its very form: it is self-

evident. (Du Bois 1986: 333)

In the Sinitic source texts, the introductory line ‘Thus I have heard,’ quotations of the

Buddha, and parallel structures bring authority to the narratives within. However,

speakers of early Heian Japanese had the auxiliary ki to index the established factuality of

their utterances and used it to frame these religious narratives in translation.143 The

following case of ki heading a sentence that presents an Evaluation, i.e., a pause in the

narrative for descriptive commentary, lends support to this argument.

(91) 聞ク者皆傷ミ悼キ悲ビ歎キ苦むこと裁フること難（かり）キ。 (Kasuga 1985a: 198)

143 Matsumoto (2011b: 67) argues that we can see auxiliary ki use making the past tense clear when setting

new scenes because it “supports the base of narrative development.” In other words, Matsumoto maintains

auxiliary ki is used to (re-)ground the narrative in the past tense. Although he does not touch upon factuality,

his assessment on ki’s use in grounding the narrative on the peripheries supports the findings in this study.

151

Kiku pito mina itami wononoki kanasibi nageki

listen person all hurting shaking.with.fear grieving lamenting

kurusimu koto osapuru koto nakari-ki

feel.pain NML restrain NML be.NEG-KI

‘All the people who heard this could not restrain their feelings of pain as they shook

with fear, grieved, and lamented.’

This sentence is coded as an Evaluation because it is describing the feelings of characters

in an embedded tale. As we see from the data, auxiliary ki only heads Evaluations 3% of

the time. However, the following line begins with 尒時に (so no toki ni ‘At that time’), a

common way to begin a new paragraph in kundokubun narrative. One type of Results I

coded were actions that occur before such so no toki ni lines. This use of paragraph-final

ki provides additional evidence that auxiliary ki acts to ground the discourse as an

established past fact at the conclusions of scenes, which is one of the focuses in the

following chapter on kundokubun and linguistic variation.

152

Chapter 6. Kundokubun and Linguistic Variation

The first part of this chapter consists of an expansion of the discussion in 5.3 with

an analysis regarding how narrative passages in kundokubun and wabun texts differ in

their uses of the auxiliaries ki and keri. The second part, in which I compare the Saidaiji

GLS analyzed above with another early Heian kundokubun partial translation of the same

sutra, as rendered at a different temple, Tōdaiji, focuses on linguistic variation within the

genre of kundokubun sutras themselves.

6.1 Framing in Kundokubun and Wabun Texts144

In this section I examine differences and similarities found in kundokubun and

wabun narrative framing strategies by comparing the Saidaiji GLS and Tōdaiji TWS

discussed above with two of the most well-known works of vernacular narrative fiction

written in the Heian period, the progenitor of the monogatari (物語 ‘tale’) genre,

Taketori monogatari (竹取物語 ‘The Tale of the Bamboo Cutter’) and the pinnacle of

the genre, Genji monogatari (源氏物語 ‘The Tale of Genji’).145 We begin with their use

in the outermost narration outside of embedded quotations (including quoted narratives).

144 An earlier version of this section was published in Bundschuh (2021).

145 Taketori monogatari is a near-contemporary of the kundokubun texts examined in this study—the earliest

estimates of its origin place it in the latter half of the 9th century CE. Genji monogatari, on the other hand,

153

6.1.1 Outermost Narration

I first turn to the earliest piece of Heian-period wabun fiction as a point of

comparison, Taketori monogatari. Its narration is known for features common to both

kundokubun and vernacular literary wabun Japanese (Sakakura 1956, Shirane 1987: 85).

Its Orientation begins as follows.

(92) 今は昔、竹取の翁といふものありけり。野山にまじりて竹を取りつゝ、萬づの事に

使ひけり。名をばさかきの造となむいひける。その竹の中に、本光る竹なむ一筋

ありける。 (Inamura 1977: 16)

Ima pa mukasi. Taketori.no.okina to ipu mono

now TOP past old.bamboo.cutter COM called person

ari-keri. Noyama ni maziri-te take wo

be-KERI field.mountain LOC be.among-GER bamboo ACC

tori-tutu yorozu no koto ni tukapi-keri. Na wo

pick-repeatedly many GEN thing ABL use-KERI name ACC

ba Sakaki.no.Miyatuko to namu ipi-keru. So no

TOP Sakaki.no.Miyatsuko(name) COM FOCUS call-KERI that GEN

take no naka ni moto pikaru take namu

bamboo GEN inside LOC base shine bamboo focus

pito.suzi ari-keru.

one.stalk be-KERI

‘(At a time) now past, there was a person called the old bamboo cutter. Going

among the fields and mountains, he repeatedly collected bamboo and used it to

make all kinds of things. Now, his name, you see, was Sakaki no Miyatsuko.

Now, it was the case that, among the bamboo, there was a stalk of bamboo whose

base was shining.’

was composed at the beginning of the 11th century CE. Furthermore, structure-wise, whereas Taketori

originated from a folktale, Genji more closely resembles a modern novel with its greater length, nuanced

character development, and detailed narration.

154

Although later in the tale we find the narrative functions of the resultative and stative

auxiliaries reflecting those in early-Heian kundokubun, the narrator of Taketori monogatari

clearly uses the externally established fact keri to frame the tale, not the past established

fact ki found in the sutras. Below are the first three lines of the Orientation of the Saidaiji

GLS’s tale of Jalavāhana, introduced in section 4.2, for comparison.

(93) 過去の無量不可思議阿僧企耶の劫に、尒時に佛有して（於）世に出現（し）たま

へリキ。名をば（曰）寶髻とまをしキ。如來應正遍知明行足善逝世間解無上士調

御丈夫天人師佛世尊とまをしキ。 (Kasuga 1985b 174)

kako no muryau pukasigi asaugiya no kopu ni

past GEN immeasurable unimaginable incalculable GEN aeon LOC

so no toki ni potoke imasi-te yo ni

that GEN time LOC Buddha be.HON-GER world LOC

syutugen.si-tamape-ri-ki. Na wo ba poukei to

appear-HON-ARI-KI name ACC TOP jewel.topknot COM

mawosi-ki. Nyorai ou syaupenti

say.HUM-KI thus.come146 worthy.one147 fully.enlightened148

myaugyausoku zenzei sekenge

perfect.knowledge.conduct149 well.departed150 knower.of.all.worlds151

muzyausi deugodyaubu tenninsi

146 如來 J. nyorai; Skt. tathāgata (see footnote 52 above). This and the following nine terms are the ten

epithets of the Buddha (J. 十號 zipugau) (Soothill & Hodous 2014: 52).

147 應 J. ou (more often rendered 應供 ougu in Sinitic script); Skt. arhat (see the entry “buddhānusmṛti”

in Buswell & Lopez 2014: 154).

148 正遍知 J. syaupenti; Skt. samyaksaṃbuddha (see the entry “buddhānusmṛti” in Buswell & Lopez 2014:

154).

149 明行足 J. myaugyausoku; Skt. vidyācaraṇasampanna (see the entry “buddhānusmṛti” in Buswell &

Lopez 2014: 154).

150 善逝 J. zenzei; Skt. sugata (see the entry “buddhānusmṛti” in Buswell & Lopez 2014: 154).

151 世間解 J. sekenge; Skt. lokavid (see the entry “buddhānusmṛti” in Buswell & Lopez 2014: 154).

155

unsurpassed152 tamer.of.people153 teacher.of.gods.and.humans154

butu seson to mawosi-ki.

buddha world.honored.one155 COM say.HUM-KI

‘Immeasurable, unimaginable, incalculable eons ago a Buddha had appeared in

the world. His name was Jeweled Topknot. He was called Thus-come, worthy

one, perfect in enlightenment, perfect in knowledge and behavior, well-departed,

knower of the world, the surpassed, supreme trainer, teacher of gods and men,

Buddha, and lord.’

Although ki is used to frame early-Heian Buddhist kundokubun narratives, the use of keri

framing, first seen in Taketori monogatari, is found throughout Heian-period wabun

narratives.

A key difference between the narratives in early Heian kundokubun sutras and

Heian-period wabun tales is epistemic grounding. Whereas the sutras are presented as

historical texts, the belles-lettres of the period make no such claim. The rhetorical role of

ki is to ground the discourse in a real or experienced past. Auxiliary keri, on the other

hand, is used when there is no epistemic ground upon which to stand. When a keri-

headed predicate grounds the discourse, it presents its predicate as “given fact,” without

relying on direct evidence (Quinn 1983), whereas ki marks the discourse as historical

truth.

152 無上士 J. muzyausi; Skt. anuttara (see the entry “anuttarasamyaksaṃbodhi” in Buswell & Lopez 2014:

55).

153 調御丈夫 J. deugodyaubu; Skt. puruṣadamyasārathi (Soothill & Hodous 2014: 52).

154 天人師 J. tenninsi; Skt. śāsṭṛ devamanuṣyānāṃ (see the entry “buddhānusmṛti” in Buswell & Lopez

2014: 154).

155 世尊 J. seson; Skt. bhagavat (Buswell & Lopez 2014: 108).

156

Beyond the use of auxiliaries, at first glance, the starkest difference between

Heian period kundokubun and wabun narration is clause length. Conventions of literary

Sinitic (a variety of Middle Chinese) composition generally restricted sentence length and

complexity to a degree not seen in EMJ wabun literature.156 There are noticeably fewer

multi-clause sentences in kudokubun. Not surprisingly, there are fewer subordinate

clauses that function as narrative Evaluations in kundokubun narration, that is, there are

fewer suspensions of action for the purpose of different Evaluative devices, such as

describing cooccurring events (Correlatives) and fewer embedded qualifying clauses

(Explicatives), compared to EMJ wabun texts.157 Another difference regarding Labovian

Evaluative clauses (see section 3.2.3) is the sparseness of focus particles, which often are

employed as Intensifiers158 in Heian period wabun narrative fiction, in kundokubun

discourse. Finally, there are clear differences in framing strategies employed by the two

registers.

To reiterate, kundokubun is a linguistic register most often used when reading

literary Sinitic texts aloud as Japanese. Many of the Sinitic source texts follow

composition conventions that limit the number of morphemes per clause, often to four or

five. When bound morphemes are in the source text, this further restricts the number of

156 See Steininger (2017) for more a detailed study on secular literary Sinitic texts in Heian Japan.

157 See Labov (1972, Chapter 9) for details on Correlatives, Explicatives, and other types of Evaluations, i.e.,

information that invites a particular valuation (attitude toward, assessment, etc.) of a narrative event or events.

158 Expressive phonology and “intensive” words. These also fall into the category of Evaluations (see Labov

1972). Some EMJ examples would be the koso, namu, and zo—only the latter is found occasionally in the

Buddhist kundokubun texts examined in this study.

157

lexemes in the resulting kundokubun rendition. To illustrate this, we turn to the first four

lines of the Saidaiji GLS’s Orientation, introduced in section 4.1, as an example.

(94a) 是（の）如キことを我レ聞きたまへキ。 (Kasuga 1985a: 1; the following as well)

ko no gotoki koto wo ware kiki-tamape-ki

this GEN like NML ACC I hear-HUM-KI

‘Thus I have humbly heard.’

(94b) 一時薄伽梵、王舍城鷲峰山の頂に、（於）最も清淨にして甚深なる法界の諸佛

（の）（之）境たる、如來の所居に在（し）キ。

itizi bagabon wausyazyau zyubusen no itadaki ni

once bhagavat Rajgir159 Vulture.Peak GEN peak LOC

motomo syauzyau ni.si-te zintan.naru popukai no

most profound COP-GER deep dharma.realm GEN

syobutu no sakai taru nyorai no syokyo ni

all.Buddhas gen region COP tathāgata160 GEN abode LOC

zai.si-ki

dwell-KI

‘Once on the top of Vulture Peak of Rajgir the Lord dwelled in the abode of the

Buddhas, the region of all Buddhas of the most profoundly deep dharma realm.’

(94c) 與には大苾芻の衆九万八千人ありキ。

tomo ni pa dai pitisu no syu kuman patisen

together LOC TOP large monk GEN gathering 90,000 8,000

nin ari-ki

people be-KI

‘Alongside him was a gathering of 98,000 supreme bodhisattvas.’

(94d) 皆是レ阿羅漢なり。

159 Skt. Rājagṛha. An ancient capital city in India near Vulture Peak (see Buswell & Lopez 2014: 694).

160 This is a common epithet for the Buddha from a past passive participial form of Skt. tathā-gam, meaning

‘one who has thus come/gone.’ See Buswell & Lopez 2014: 897 and footnote 52.

158

mina kore arakan nari

everyone these arhat161COP

‘All of these were arhats.’

If we take particles to be words and auxiliaries to be affixes, as rendered in the

morphological breakdown in the examples, then the first four sentences of the text have

49 lexemes. In contrast, the following sequence of examples presents the first four

sentences of the wabun Heian literary text, Genji monogatari. The syntactic complexity

of each sentence is immediately apparent.

(95a) いづれの御時にか、女御、更衣あまたさぶらひたまひけるなかに、いとやむごと

なき際にはあらぬが、すぐれて時めきたまふありけり。 (NKBZ: 1)

idure no opon-toki ni ka nyogo kaui amata

when GEN HON-time LOC Q consort intimate many

saburapi-tamapi-keru naka ni ito yamu.goto.naki kipa ni

be.HUM-HON-KERI among LOC very high.rank lineage LOC

pa ara-nu ga sugure-te tokimeki-tamapu ari-keri

TOP be-NEG GEN excel-GER thrive-HON be-KERI

‘In a certain reign (whose can it have been?) someone of no very great rank,

among all His Majesty’s Consorts and Intimates, enjoyed exceptional favor.’

(English translations of Genji monogatari passages by Tyler 2001: 1)

(95b) はじめより我はと思ひ上がりたまへる御方がた、めざましきものにおとしめ嫉みた

まふ。

pazime yori ware pa to omopi-agari-tamape-ru

beginning from I TOP COM think-HON-HON-ARI

opon-kata-gata mezamasiki mono ni otosimesonemi-tamapu

HON-people-PL dreadful person LOC despise-envy-HON

‘Those others who had always assumed that pride of place was properly theirs

despised her as a dreadful woman’

(95c) 同じほど、それより下臈の更衣たちは、ましてやすからず。

161 This term refers to Buddhist saints. See Buswell & Lopez (2014: 62) and footnote 64 above.

159

onazi hodo sore yori gerapu no kaui-tati pa masi-te

same extent that from low GEN intimate-PL TOP increase-GER

yasukara-zu

calm-NEG

‘while the lesser Intimates were unhappier still’

(95d) 朝夕の宮仕へにつけても、人の心をのみ動かし、恨みを負ふ積りにやありけむ、

いとあつしくなりゆき、もの心細げに里がちなるを、いよいよあかずあはれなるも

のに思ほして、人の そしりをもえ憚らせたまはず、世のためしにもなりぬべき御も

てなしなり。

asa-yupu no miya-dukape ni tuke-te mo pito no

morning-night GEN court.service LOC serve-GER also person GEN

kokoro wo nomi ugokasi urami wo opu tumori ni ya

mind ACC only move resentment ACC bear grow LOC focus

ari-kemu ito atusiku nari-yuki mono kokoro-boso-ge ni

was.perhaps very extreme become-go very worry COP

sato-gati naru wo iyo-iyo aka-zu apare.naru

return.home become ACC increasingly cease-NEG feel.regret

mono ni omoposi-te pito no sosiri wo mo

NML LOC think.HON-GER person GEN criticism ACC also

e-pabakara-se-tamapa-zu yo no tamesi ni mo

able-refrain-HON-HON-NEG society GEN topic.of.conversation LOC also

nari-nu-beki opon-motenasi nari

become-NU-MODAL HON-treatment COP

‘The way she waited on him day after day only stirred up feeling against her, and

perhaps this growing burden of resentment was what affected her health and

obliged her often to withdraw in misery to her home; but His Majesty, who could

less and less do without her, ignored his critics until his behavior seemed bound to

be the talk of all.’

As opposed to the nearly 50 lexemes found in the first four lines of the Saidaiji GLS, we

find 90 in the first four sentences, or lines, of Genji monogatari. This contrast is

representative of the differences in sentence length between Heian period kundokubun

160

and wabun.162 The above examples also demonstrate differences in framing strategies

found in the two linguistic registers. In sentence-final predicates in the Orientation

passages of the kundokubun-rendered sutras examined in this dissertation, ki is dominant,

whereas in the Orientation of wabun texts, such as Taketori monogatari and Genji

monogatari, we find a preponderance of keri. This is due to their differing narrative use:

keri indicates external evidence or legitimacy, an “externally established fact”, whereas ki

is simply an “established fact” (Quinn 1990).

Both Okada (1991) and Stinchecum (1985) argue that keri is used in narrative to

assert narrative control, essentially to remind the audience of the narrator’s presence, and

“[represent] a legitimizing element of affirmation for [literary Japanese] discourse

through which the discourse grounds itself” (Okada 1991: 42). The contrast between the

use of keri in framing wabun and that of ki framing kundokubun narratives in EMJ is clear

when we consider the use of ki in the conclusions of sutras as well, such as that of the

GLS, which was introduced in section 4.1 above.

(96) 爾時に無量無邊恒沙の大衆い佛の説を聞（きたま）へ已（り）て、皆大に歡喜して

信受し奉行しキ。 (Kasuga 1985a: 206)

so no toki ni muryau mupen gauzya no

that GEN time LOC immeasurable immense innumerable GEN

daisyu i potoke no setu wo kiki-tamape-wopari-te

group NOM Buddha GEN explanation ACC listen-HUM-finish-GER

mina dai ni kwangi.si-te sinzyu.si bugau.si-ki

all great COP be.overjoyed-GER accepting.truth enact-KI

162 We saw much less syntactic complexity and, therefore, fewer lexemes, in the opening lines of Taketori

monogatari in (6.1). Although it is clearly in the wabun register (Sakakura 1956), it may have been rendered

from a Sinitic text (Tsukishima 1955), which could be why we find this discrepancy in sentence length with

later wabun tales.

161

‘At that time the immeasurable, immense, innumerable group, having heard the

Buddha’s explanation, all in great delight, accepted and carried it out.’

There is no need for external epistemic grounding through keri because the authority of the

religious text has been accepted as witnessed fact. The TWS follows a similar framing

strategy, but the introduction of the assembly surrounding the Buddha during the sermon

is much shorter than the GLS, leading to a shorter Orientation with much fewer uses of ki.

The narrator of both these (and most other) sutras is understood to be the Buddha’s

disciple Ānanda (J. 阿難(陀) anan(da)), who is considered to have witnessed the sermons

described therein and invariably begins the first chapter with the phrase ‘Thus I have

humbly heard’ (是の如きことを我れ聞きたまへき ko no gotoki koto wo ware kikitapeki)

(Katō 1998: 13). Stinchecum (1985), in her analysis of narration in the Uji chapters of the

wabun tale Genji monogatari, comes to the following conclusion regarding ki.

As opposed to -keri, which implies psychological or temporal distance between

the speaker and the event related, and is often used in relation to hearsay, -ki

refers to certain recollection of a fact that existed in the past, usually something

the speaker has personally experienced. In the case of events not personally

experienced by the speaker, -ki refers to those which, even if the knowledge is

acquired through hearsay, are clearly and firmly engraved upon his memory.

(Stinchecum 1985: 28)

Her analysis of these auxiliaries can be extended to their use in kundokubun as well,

where keri also indexes newly noticed, until now psychologically distant, facts (keri’s

mirative use) and ki indexing witnessed facts, showing that both auxiliaries maintain their

core semantics in EMJ. Although we find stark contrasts in their employment in the

outermost narratives of wabun and kundokubun texts, I now turn to their shared

pragmatics in EMJ regardless of linguistic register.

162

6.1.2 Embedded Narration

Kundokubun and wabun share more conventions in embedded spoken narrative

compared to narration presented by the matrix narrator, a phenomenon noted by

Matsumoto (2011a). In both registers, in quotations we find both ki in Orientations and

keri in Evaluations. This shared narrative functionality reflects these auxiliaries’

pragmatics—ki is used to index past established, often witnessed, facts and keri to sum up

externally established facts, i.e., those for which we lack firsthand evidence or those we

just realized. We can see the choice of ki to highlight personally experienced facts by the

character Sama no Kami in the second chapter of Genji monogatari. As the young man

begins to share his experience of love lost, he specifies the time in his Orientation with

auxiliary ki to emphasize his personal experience of the events he describes and to draw

in his audience.

(97) はやう、まだいと下臈にはべりし時、あはれと思ふ人はべりき。 (NKBZ: 147)

hayau mada ito gerapu ni-paberi-si toki apare to

long.ago still very low.rank COP-HUM-KI time fondly COM

omopu pito paberi-ki

think person be.HUM-KI

‘Long ago [. . .] when I was still very young, there was someone who meant a

great deal to me.’ (Tyler 2001: 27)

As his story unfolds, most sentences that relate matters within his experience end

similarly with the established past fact auxiliary ki, but one is headed by keri.

(98) なほ家路と思はむ方はまたなかりけり。 (NKBZ: 151)

163

napo ipedi to omopa-mu kata pa mata

really road.home COM think-CONJ place TOP still

nakari-keri

be.NEG-KERI

‘I realized I had no other home to go to than hers.’ (Tyler 2001: 29)

Here keri is being used by Sama no Kami to convey an existing fact that had just come

into his perception, as the clause is preceded by omopimeguraseba ‘now that I think

back.’ Examples such as this are commonly referred to as keri’s ‘discovery’ or

‘exclamatory’ mirative use, but their epistemics are also the basis from which Takeoka

(1963) derives his anata naru ba ‘removed ground’ analysis of keri—that it indicates

facts that are already established but until their presentation with keri have been outside

of the perception of the speaker or the audience. We discussed this mirative use of keri in

kundokubun section 5.3.4 above.

Sama no Kami concludes his first-hand tale of regret by describing the tragic end

of his love due to his hardheadedness, and ki heads both the realis conditional clause and

its main clause consequence.

(99) いといたく思ひ嘆きてはかなくなりはべりにしかば、戯れにくくなむおぼえはべり

し。 (NKBZ: 152)

ito itaku omopi-nageki-te pakanaku

very painfully feel-lament-GER short.lived

nari-paberi-ni-sika-ba tapabure-nikuku namu

become-HUM-NU-KI-COND joke-difficult you.know

oboe-paberi-si

learn-HUM-KI

‘She was so hurt that she died. That taught me that these things are no joke.’

(Tyler 2001: 29)

164

Although Tyler’s literary translation breaks this line into two sentences, one more closely

heeding the original syntax would keep the lover’s death in a subordinate realis

conditional ki clause with only the final ki-marked predicate as the matrix clause—‘Given

that, very painfully lamenting, she became short-lived, what I learned was, you know,

it’s hard to joke about these things.’ He then goes on to compare her practical skills to

those of ideal mythical women, ending his line with another ki. Unlike in the outermost,

keri-framed narration of wabun tales, embedded narratives depicting events witnessed by

the narrator are, like those in kundokubun parables, framed with ki. Just like Sama no

Kami, the Buddha begins his narratives, such as those depicted in sections 4.2 (the GLS)

and 4.4 (the TWS) above, with ki to index his personal knowledge of the events he

describes.

6.1.3 Conclusions regarding ki and keri in Wabun and Kundokubun Narration

Throughout chapters 4 and 5 of this dissertation, we have seen that ki is used in

kundokubun discourse to epistemically ground narratives as past established facts and

keri is limited to quotations. In wabun, conversely, keri is used to discursively ground the

discourse and ki is primarily found in quotes. This section has shown that, in embedded

narration, they share the potential for ki-marked narration with keri performing an

evaluative role. Regarding these two auxiliaries in embedded tales in Genji monogatari,

Kern writes:

The auxiliary ki is an evidentiary marker, used to represent something in the past

that is within the speaker’s personal experience. The use of this auxiliary in the

[embedded Genji monogatari] tales has an important effect on the mood of the

165

storytelling [. . .] ki becomes a means by which the material is simultaneously

personalized by the author and distanced from the listener, in contrast to the more

immediate feel of those clauses not marked with ki. [. . .] keri [. . .] indicates

something from outside the speaker’s experience that is being brought into the

speaker’s knowledge. (Kern 2007: 4–5)

We can see this dichotomy between the two modal auxiliaries in the following selection

of embedded narration in Genji monogatari by the character Tō no Chūjō, who responds

to Sama no Kami’s tale, discussed above, with one of his own. He likewise begins with

ki-marking, but as he approaches his conclusion, he switches to primarily keri marking.

(100) つれなくて、つらしと思ひけるも知らで、あはれ絶えざりしも、益なき片思ひなりけ

り。今、やうやう忘れゆく際に、かれ、はた、えしも思ひ離れず、をりをり人やりなら

ぬ胸こがるる夕もあらむとおぼえはべり。これなむ、えたもつまじく頼もしげなき方

なりける。 (NKBZ 83–84)

turenaku-te turasi to omopi-keru mo sira-de

serene-GER pained COM think-KERI also learn-NEG.GER

apare tape-zari-si mo yakunaki kata-omopi

feelings end-NEG-KI also useless one.sided.feelings

nari-keri. Ima yauyau wasure-yuku kipa ni kare pata

COP-KERI now finally forget-go time LOC she in.contrast

e simo omopi-panare-zu woriwori

able not.necessarily think-separate-NEG sometimes

pito-yari-nara-nu mune kogaruru yupube mo ara-mu

forced.by.someone-NEG chest burn night also have-CONJ

to oboe-paberi. Kore namu e-tamotu-maziku

COM feel-HUM. This you.know able-maintain-NEG.CONJ

tanomosige naki kata nari-keru.

dependable be.NEG person COP-KERI

‘She seemed so serene that I never knew she was hurt, and my lasting feeling for

her went completely to waste. Even now, when I am beginning to forget her, she

probably still thinks of me and has evenings when she burns with regret, although

166

she has no one but herself to blame. She is a perfect example of the woman you

cannot keep long and cannot actually depend on.’ (Tyler 2001: 32–33)

The first keri is Tō no Chūjō evaluating the emotional state of a woman he knew. He then

shifts to ki to discuss his internal evaluation of his own emotional situation. The final two

uses of keri cover both the Evaluation and the Coda of his tale, as he evaluates the

situation and sums up his point, giving an answer to the goal of storytelling posited by

Labov (1972): how to avoid being asked the dreaded question “So what?” As a discourse

strategy used when framing wabun narratives, keri-marking indexes that the facts being

depicted are new to the listener, and therefore worth paying attention to. As we have seen

throughout this study, Buddhist texts rendered in kundokubun, in contrast, are framed as

religious truth and, therefore, the translators use the auxiliary ki to mark them as both

communally witnessed and self-evident. Furthermore, rather than concluding one’s own

narrative with a keri, the narrator often has a character who heard the words of the

Buddha give their reaction using the auxiliary, as we saw in 5.3 above. We now turn to

variation among kundokubun texts.

6.2 Variation within Early Heian Kundokubun

In this section I compare two early-Heian renditions of the Golden Light Sutra,

the first, described extensively above, is the Saidaiji temple rendition glossed ca. 830 CE.

The second is a commentary on the GLS glossed at Tōdaiji ca. 850 CE, which contains

kundokubun renditions of the 5th and 9th scrolls.163 Data from this source were collected

163 I defer to Ishizuka (2001: 31) regarding the dating of these renditions.

167

from Tabuchi (1987) who, like Kasuga (1985a), produced a kakikudashi164 from the

original kunten glosses.165 The focus of analysis in this section is the tale of Jalavāhana,

which I introduced in section 4.2 above.

To begin, Kobayashi (2012a) investigates variation between early Heian Buddhist

kundokubun texts, but primarily focuses on the lexical level—how specific Sinographs

were rendered, phonological variation of Japanese lexemes, and what morphemes were

added that are not found in the original Sinitic texts (J. 讀添語 dokutengo). The TAM

auxiliaries investigated in this dissertation, as morphemes generally not found in the Sinitic

source texts, are included in his discussions of added morphemes. However, unlike his rich

analysis of diachronic variation discussed in section 1.5 above, Kobayashi (2012a)

addresses synchronic variation by simply presenting the differences on a lexical level

without positing their pragmatic differences. Below I make an attempt to clarify the

variation between two near-contemporary renditions of the Golden Light Sutra

Figure 2 from section 1.5 above is worth reproducing for the following

discussion.

164 A transliteration of the glossed Sinitic script into Japanese script.

165 Although Tabuchi (1987) generally follows Kasuga’s methodology in producing his kakikudashi, he does

not include consonant voicing glosses (dakuten 濁点) and writes “▭” when the reading is unclear.

168

Figure 2: Framing in Buddhist Narratives

(Adapted from Matsumoto 2011a: 215)

Here I analyze sentence-final predicates based on whether they are used in the Outermost

Narration (“ON” = layer 2), the Outermost Dialogue, the conversation among those

assembled (“OD” = layer 3), the Embedded Narration told by the Buddha (“EN” = layer

4), or the Embedded Dialogue in that embedded narrative (“ED” = layer 5).

For the remainder of this subsection, “Saidaiji” and “Tōdaiji” refer specifically to

the 24th and 25th chapters of the GLS, which depict the tale of Jalavāhana, found in

Kasuga (1985a: 174–185) and Tabuchi (1987: 76–86). The two tables below depict the

heads of sentence-final forms (predicates and otherwise) throughout the parable in each

rendition.

169

Table 29 Overview of Saidaiji Sentence-final Elements

ON OD EN ED

bare verb 3 6 43 68

adjective ∅ ∅ 1 5

ki 1 ∅ 40 1

nu ∅ ∅ 10 1

tu ∅ ∅ 3 4

ari ∅ ∅ 1 3

tari ∅ ∅ 1 1

other auxiliary ∅ 5 1 38

copula ∅ 7 ∅ 3

particle ∅ 2 ∅ 1

Totals 4 20 100 125

The “other auxiliary” line refers to those that do not govern the participle, which include

two tokens of negative zu, 18 of future conjectural mu, three of negative future

conjectural zi, 21 of modal besi, and one of present conjectural ramu. Other than one

instance of negative zu, all are found in dialogue. All three particles—quasi-copula zo,

rhetorical-question marker ya, and negative imperative marker na—are found in dialogue

as well.

170

Table 30 Overview of Tōdaiji Sentence-final Elements

ON OD EN ED

bare verb 3 4 29 69

adjective ∅ ∅ 2 5

ki ∅ ∅ 58 2

nu ∅ ∅ 8 2

tu ∅ 1 2 2

ari ∅ ∅ 2 5

tari ∅ ∅ 1 ∅

other auxiliary ∅ 4 1 38

copula ∅ 6 ∅ 4

particle ∅ ∅ ∅ 2

Totals 3 15 104 129

Just as in the Saidaiji data, other than one instance of negative zu, all particles and

auxiliaries that do not govern a verb’s participle are found in dialogue. The auxiliaries

include four tokens of negative zu, 17 of future conjectural mu, one of negative

conjectural mazi, 20 of modal besi, and one of present conjectural ramu. The only

sentence-final particles in the section were one example each of quasi-copula zo and

rhetorical-question marker ya, both found in ED.

We find a total of 249 sentence-final predicates in the Saidaiji data, with only four

lines of outermost narration in these chapters (24 and 25), which only account for 1.6%

of the total. The majority are embedded in the Buddha’s telling of the tale of Jalavāhana.

The three sentence-final predicates of the outermost narration in the Tōdaiji data make up

1.3% of its total sentence-final predicate count of 231. In both, the majority of sentence-

final predicates are found in dialogue. As Matsumoto (2011a) notes of a mid-Heian

glossing of the Sudāna Sutra, embedded dialogue contains the most morphological

complexity in sentence-final predicates.

171

In comparing the Saidaiji and Tōdaiji data, we find that bare verbs outnumber

those governed by ki in sentence-final position in embedded narration of the Saidaiji

rendition. However, in the Tōdaiji version of the tale, the reverse is true. There do not

appear to be any significant contrasts among the other TAM auxiliaries. Therefore, for the

remainder of section 6.2, I compare the use of auxiliary ki in the Buddha’s narration of

the tale of Jalavāhana between the two temples’ renditions.

When we compare the lines depicted as spoken by the Buddha to the goddess of

the bodhi tree within Embedded Narration (EN) data (Saidaiji Table 29, and Tōdaiji Table

30 above), we find a clear difference in the use of auxiliary ki. In the Saidaiji EN, there are

40 cases of ki out of 100 sentence-final predicates of narration. In other words, auxiliary

ki heads 40% of the embedded narrative’s sentences. On the other hand, in the Tōdaiji

data 58 of the EN’s 108 sentence-final predicates are headed by the past established fact

auxiliary, i.e., auxiliary ki governs 56% of matrix clauses in in the Tōdaiji EN. In order to

explore this difference, we compare ki-marked sentences in the two sutras with reference

to the three parameters of transitivity, lexical aspect, and discourse function, which

formed the framework of analyses in chapters four and five above.

172

6.2.1 A Comparison of Transitivity and Sentence-Final ki

The two tables below compare the transitivity of all sentence-final elements with

predicates marked with ki in the Saidaiji and Tōdaiji renditions of the embedded narration

of the parable.166

Table 31 Correspondences between Saidaiji ki and Transitivity

Predicate transitivity: low medium high

S-F167 predicate totals 81 15 4

S-F ki totals 36 1 3

S-F ki percentages 44% 7% 75%

Table 32 Correspondences between Tōdaiji ki and Transitivity

Predicate transitivity: low medium high

S-F predicate totals 84 14 6

S-F ki totals 42 4 4

S-F ki percentages 62% 29% 67%

In both renditions of the parable, a supermajority of sentence-final predicates are of low-

transitivity, a trend seen throughout kundokubun narration. Although high-transitivity

predicates, i.e., those with causatives and transitive verbs that cause a change in their

object, are few in number in these two renditions, the percentages of those that are

166 Although the total counts include sentence-final particles, their numbers are low to the extent that I have

elected to use the term ‘predicate’ rather than ‘element’ out of utility.

167 The abbreviation ‘S-F’ stands for ‘Sentence-final’ here and in the following tables.

173

governed by ki in the Saidaiji and Tōdaiji embedded narration are 75% and 67%,

respectively. Below is an example of a high-transitivity predicate in the Tōdaiji EN.

(101) 諸（の）魚（は）並（に）死（ぬ）ことなり（と）見しめき。 (Tabuchi 1987：85)

moro no uo pa narabi.ni sinu koto nari to

all GEN fish top together die NML COP COM

mi-sime-ki

see-CAUS-KI

‘It caused him to see that all the fish would die together.’

Also, the reason that over half of the low-transitivity predicates in the Tōdaiji data are

governed by ki is that the auxiliary is often applied to “speech act” predicates at the end

of quotations. The impact of ki-marking in “speech act” predicates in the Tōdaiji GLS is

clarified in the following subsection as we compare ki-marking and lexical aspect across

the two renditions.

6.2.2 A Comparison of Lexical Aspect and Sentence-Final ki

The two tables below summarize how ki-marking and lexical aspect types

intersect in the Saidaiji and Tōdaiji renditions of the GLS.

Table 33 Correspondences between Saidaiji ki and Lexical Aspect

Lexical aspect: State Activity Accomplishment Achievement

S-F predicate totals 28 14 44 30

S-F ki totals 15 5 10 10

S-F ki percentages 54% 36% 23% 33%

174

Table 34 Correspondences between Tōdaiji ki and Lexical Aspect

Lexical aspect: State Activity Accomplishment Achievement

S-F predicate totals 28 8 38 30

S-F ki totals 20 4 26 10

S-F ki percentages 71% 50% 68% 33%

Table 33 shows that in the Saidaiji version of the tale of Jalavāhana, there is no strong

relationship between any one lexical aspect type and ki marking, although ki appears

more frequently with stative predicates than the other types. By comparison, the Tōdaiji

data show higher percentages of ki in the State, Activity, and Accomplishment

parameters. We see that in both temples’ versions of this embedded narrative, the

majority of States are in ki-marked sentences. This is likely because of the discourse

function of ki in Orientations to ground the story in an established past, which we discuss

further in the following subsection. Basically, States predominate in depicting scenes in

which the story will develop, and scene-setting is what Orientations do.

The clearest difference between the two temples’ renditions is the use of ki in

Accomplishments. Whereas in the Saidaiji data (Table 33) we find 23% of sentence-final

Accomplishment predicates governed by ki, in the Tōdaiji data (Table 34), 68% of

Accomplishments contain a sentence-final ki. As mentioned in the above subsection, the

majority of predicates coded as Accomplishments are “speech act” verbs that follow a

quote and the complementizer to. Examples contrasting this difference in the ki-marking

of predicates depicting the action of speaking are presented in the conclusion to 6.2

below. But first we examine the relationship between discourse function and the use of

auxiliary ki in the two texts.

175

6.2.3 A Comparison of Discourse Function and Sentence-Final ki

The following tables depict the total number of sentence-final predicates that

serve each discourse function, the number of those that are governed by ki, and the

percentage of the total for each category in which we find ki sentence-finally. Because I

am focusing on the embedded narration, there are no (a) Abstracts or (b) Codas, as these

discourse functions are, by definition, found in the Outermost Dialogue, due to their

functions of (a) informing the audience that a story will occur and (b) bringing the focus

of the conversation back to the time of speaking, respectively. Likewise, Quotations are

found in the layer of Embedded Dialogue rather than of Embedded Narrative. The EN

contains the remaining discourse functions—Orientations, Complications, Evaluations,

and Results—which are the focus of this section and are analyzed below.

Table 35 Correspondences between Saidaiji ki and Discourse Function

Discourse function: Orientation Complication Evaluation Result

S-F predicate totals 23 70 4 3

S-F ki totals 19 18 0 3

S-F ki percentages 83% 26% 0% 100%

Table 36 Correspondences between Tōdaiji ki and Discourse Function

Discourse function: Orientation Complication Evaluation Result

S-F predicate totals 27 68 6 3

S-F ki totals 21 35 1 3

S-F ki percentages 78% 51% 17% 100%

176

When we examine the above tables, it is clear that, in each temple’s rendition, ki is not

generally used in Evaluations and it has a proclivity to mark sentences in Orientations

and Results. Furthermore, all Orientations not headed by ki are those resetting the scene

mid-story, such as the following.

(102) 即便隨（ひ）て去ク。大池有リ (Kasuga 1985a: 179)

sunapati sitagapi-te yuku. opoike168 ari

thus follow-GER go large.lake be

‘Then [Jalavāhana] followed [the animals] and went. There was a large lake.

The first sentence in (102) is a Complication depicting the movement of the hero of the

story, Jalavāhana. The second sentence is an Orientation depicting the scene at the new

location. Because the only Orientations that are unmarked by ki occur mid-story, we can

say that 100% of Orientations at the beginning of each version of the narrative are

sentences headed by the past established fact auxiliary ki. We can also see from (102) that

the shift and reestablishment of the scene mid-story does not require ki-marking to

introduce a new scene in the continuing story.

Regarding Complications, when we compare the two texts, we find the Saidaiji

version depicted in Table 35 with 70 total sentences functioning as Complications and,

among those, 18 sentence-final predicates, or only 26% of the total, are governed by ki.

On the other hand, in the Tōdaiji text, as Table 36 shows, among the 68 total

Complications, 35 sentences, or 51% of the total, have ki in final position. In other words,

the percentage of ki-marking in Complications in the Tōdaiji rendition of Jalavāhana’s

168 Perhaps the two sinographs (大池) are read daiti. The meaning remains the same.

177

tale is roughly double that found in the Saidaiji version. The likely reason for this stark

difference is that sentence-final predicates of speaking function mostly as Complications

and the early Heian translator of the GLS at Tōdaiji was more inclined than his Saidaiji

contemporary to add the TAM marker ki to these quoting “speech act” predicates.

Finally, the tables below catalog every sentence-final predicate in the Buddha’s

telling of the tale of Jalavāhana in the two renditions of the Golden Light Sutra for its

narrative discourse function. Because the Buddha’s speech act includes the Overarching

Dialogue in addition to the Embedded Narration, the catalog includes the Abstract and

Coda of the tale as well. All sentence-final predicates governed by auxiliary ki are

underlined.

Table 37 All sentence-final predicates and their narrative discourse functions of the

Saidaiji tale of Jalavāhana narration

(1) Abstract: kike (聽ケ ‘Listen!’); si.seyo (思（せ）ヨ ‘Think!’); nen.seyo (念（せ）ヨ

‘Take care!’); toka-mu (說かむ ‘[I] shall expound’)

(2) Orientation: syutugen.si-tamepe-ri-ki (出現（し）たまへリキ ‘[he] had honorably

appeared); mawosi-ki (まをしキ ‘[he] was humbly called’); mawosi-ki (まをしキ

‘[he] was humbly called’); nari-ki (なりキ ‘[he] was’); ipi-ki (いひキ ‘[he] was

called’); si-ki (しキ ‘[he] did’); ari-ki (有（り）キ ‘[there] was’); ipi-ki (いひキ

‘[he] was called’); kiureu.si-ki (救療しキ ‘[he] healed [them]’); ari-ki (有（り）キ

‘[he] had’); ipi-ki (いひキ ‘[he] was called’); nari-ki (なりキ ‘[he] was’);

tanosima-re-ki (樂（ま）所キ ‘[they] were made happy’); sakasikari-ki (敏シカリキ

‘[he] was intelligent’); nakari-ki (無（か）リキ ‘[there] was not’); nakari-ki (無

（か）リキ ‘[there] was not’)

178

(3) Complication: omopu (オモフ ‘[they] think’); dyu.si-nu (住（し）ヌ ‘[they] stop

[there]’); mawosu (まをす ‘[they] say’); ipi-ki (いひキ ‘[he] said’); reuti.si-nu

(了知しヌ ‘he realizes’); omopi-nu (オモヒヌ ‘[he] comes to think’); ipi-ki (いひ

キ ‘[he] said’); u (得 [they] attain [it]’); nari-ni-ki (なりにキ ‘[they] became’);

syau.si-ki (請しキ ‘[they] requested’); kagapuri-ni-ki (蒙リにキ ‘[they] received’)

(4) Result: e-sime-ki (得（し）メキ ‘[he] made [them] attain [it]’)

(This is the end of chapter 24)

(5) Complication: uke-sime-ki (受ケ令メキ ‘[he] made [them] receive [it]’); su (す

‘[they] do [it]’); kango.si-ki (歡娛しキ ‘[they] delighted’); ipi-ki (いひキ ‘[they]

said’)

(6) Result: syupen.se-ri-ki (周遍せりキ ‘[it] had spread’)

(7) Orientation: ari-ki (ありキ ‘[there] was’); ipi-ki (いひキ ‘[she] was called’); ari-ki

(有リキ ‘[they] had’); naduke-ki (名（づけ）キ ‘[they] were named’)

(8) Complication: yogiri-nu (過リヌ ‘[they] pass [through]’); yuku (去ク ‘[they] are

going’); omopu (オモフ ‘[he] thinks’); yuku (去ク ‘[he] goes’)

(9) Orientation: ipu (いふ ‘[it] is called’)

(10) Complication: mi-tu (見つ ‘[he] sees’); syau.su (生す ‘[he] gives rise to [it]’);

zigen.su (示現す ‘[she] shows’); ipu (いふ ‘[she] says’); ipu (いふ ‘[he] says’);

ipu (いふ ‘[she] says’); masu (益す ‘[he] increases [it]’); marobi-kaperu (婉ビ轉

ル ‘[they] flip over’

(11) Evaluation: toma-zu (捨マず（未） ‘[they] do not stop’)

(12) Complication: nari-nu (なりヌ ‘[they] become’)

(13) Orientation: ari (有リ ‘[there] is’)

(14) Complication: tukuru (作ル ‘[he] makes’); mi-tu (見つ ‘[he] sees’)

(15) Evaluation: katasi (難し ‘[it] is difficult’)

179

(16) Complication: omopu (オモフ ‘[he] thinks’); ipu (いふ ‘[he] says’); notamapu (の

たまふ ‘[he] honorably says’); ipu (いふ ‘[he] says’); yuku (ユク ‘[they] go’);

karu (借ル ‘[they] borrow’); nari-ni-ki (なりにキ ‘[it] became’); miru (視ル

‘[he] sees’); yuku (行ク ‘[they] go’); omopu (オモフ ‘[he] thinks’); ipu (いふ

‘[he] says’); toku (說ク ‘[he] expounds’); itari-nu (至（り）ヌ ‘[they] arrive’);

kiyaku.su (喜躍す ‘[he] dances with joy’); tirasu (散す ‘[he] spreads [it]’);

paosoku.si-ni-ki (飽足しにキ ‘[they] became satisfied’); omopi-ki (オモヒキ ‘[he]

thought’); omopi-ki (オモヒキ ‘[he] thought’); mawosi-ki (まをしキ ‘[he] humbly

said’); notamapu (のたまふ ‘[he] honorably says’); ipi-ki (いひキ ‘[he] said’);

kaperi-ni-ki (還（り）にキ ‘[they] returned’); pusi-tu (臥（し）つ ‘[he] falls

asleep’); umare-nu (生レヌ ‘[they] are born’); ipi-ki (いひキ ‘[they] said); itari-

nu (至（り）ヌ ‘[they] arrive’)

(17) Evaluation: nemuri-tari (睡（り）たり ‘[he] has slept’); itase-ri (至せり ‘[they] have

brought’)

(18) Complication: kakugo.se-simu (覺悟セ令む ‘[they] make [everyone] wake up and

realize’); same-nu (寤（め）ヌ ‘[they] awaken’); yuki-ni-ki (去（き）にキ ‘[they]

went’); amepuru (雨ル ‘[they] rain [them]’); amepuru (雨ル ‘[they] rain

[them]’); uke-ki (受（け）キ ‘[they] received’); notamapu (のたまふ ‘[he]

honorably says’); mawosu (まをす ‘[they] humbly say’); kwan.su (喚す ‘[he]

summons’); itari-nu (至（り）ヌ ‘[he] arrives’); notamapu (のたまふ ‘[he]

honorably says’); ipu (いふ ‘[he] says’); notamapu (のたまふ ‘[he] honorably

says’); ipu (いふ ‘[he] says’)

(19) Evaluation: ari (有り ‘[there] is’)

(19) Result: tan.si-ki (歎しキ ‘[they] were moved [emotionally]’)

(20) Coda: siru-besi (知ル當し ‘[you] should know’); nari (なり ‘[he] is’); nari (なり

‘[he] is’); nari (なり ‘[he] is’); nari (なり ‘[he] is’); nari (なり ‘[he] is’); nari (なり

‘[he] is’); nari (なり ‘[he] is’); kiku (聽ク ‘[they] listen’); toku (說ク ‘[I]

expound’); motomu-besi (求む應し ‘[you] should seek’); hauitu.suru koto na (放

逸することナ（勿） ‘do not be self-indulgent’)

180

Table 38 All sentence-final predicates and their narrative discourse functions of the

Tōdaiji tale of Jalavāhana narration

(1) Abstract: kike (聽（け） ‘Listen!’); si.seyo (思（せ）ヨ ‘Think!’); nen.seyo (念（せ）ヨ

‘Take care!’); toka-mu (說（かむ） ‘[I] shall expound’)

(2) Orientation: notamapi-ki (のたまひき ‘[he] was honorably called’ notamapi-ki (のた

まひき ‘[he] was honorably called’); ipi-ki (（いひき） ‘[he] was called’); si-ki

(（しき） ‘[he] did’); ari-ki (有（りき） ‘[there] was’); ipi-ki (いひ（き） ‘[he] was

called’); sukuki-iyasi-ki (救ヒ療（し）き ‘[he] healed [them]’); ari-ki (有リき ‘[he]

had’); ipi-ki (いひ（き） ‘[he] was called’); nari-ki (（なり）き ‘[he] was’); tanosibi-

mi-rare-ki (樂ヒ觀所れき ‘[he] were seen with joy’); sakasiku ari-ki (敏（しくあ）リ

き ‘[he] was intelligent’); ari-ki (アリ（き） ‘[there] was’); ari-ki (アリキ ‘[there]

was’)

(3) Complication: dyu.si-nu (住（し）ヌ ‘[they] stop [there]’); ipu (いふ ‘[he] says’);

pakari-nu (忖リヌ ‘he considers’); itari-tu (至（り）ツ ‘[he] arrives’); ipi-ki (イヒキ

‘[he] said’); e-sime-ki (得しめき [he] makes them attain [it]’); nari-ni-ki (（なり）に

き ‘[they] became’); negapu (請（ふ） ‘[they] request’); kagapuri-ni-ki (蒙（り）に

き ‘[they] received’)

(4) Result: e-sime-ki (得しめき ‘[he] made [them] attain [it]’)

(This is the end of chapter 24)

(5) Complication: uke-sime-ki (受（け）しめ（き） ‘[he] made [them] receive [it]’);

yorokobi-tanosimu (歡ヒ娛シム ‘[they] joyfully celebrate’); ipu (いふ ‘[they]

say’)

(6) Result: syupen.se-ri-ki (周遍セリき ‘[it] had spread’)

(7) Orientation: ari-ki (アリき ‘[there] was’); ipi-ki (いひき ‘[she] was called’); ari-ki

(有（り）き ‘[they] had’); naduke-ki (名（つけ）き ‘[he] was named’); naduke-ki (名

（つけ）き ‘[he] was named’)

181

(8) Complication: yogiri-nu (過リヌ ‘[they] pass [through]’); pasiri-tobi-tu (奔（り）飛

（び）ツ ‘[they] run and fly’); yuku (去く ‘[they] are going’); yuku (去く ‘[he]

goes’)

(9) Orientation: ari (有リ ‘[there] is’); ipu (曰（ふ） ‘[it] is called’); ari-ki (有（り）き

‘[there] were’)

(10) Complication: syau.su (生（す） ‘[he] gives rise to [it]’); ipi-ki (いひき ‘[she]

said’); ipi-ki (いひき ‘[he] said); ipi-ki (いひき ‘[she] said’); masi-ki (益シき

‘[he] increased [it]’)

(11) Evaluation: nasi (無シ ‘[there] is none’)

(12) Complication: marobi-kaperu (婉ビ轉ル ‘[they] flip over’)

(13) Evaluation: toma-zu (捨マ不 ‘[they] do not stop’)

(14) Orientation: ari (有リ ‘[there] is’)

(15) Complication: tukuri-ki (作リき ‘[he] made’); mi-tu (見つ ‘[he] sees’)

(16) Orientation: ari (アリ ‘[there] is’); ipu (いふ ‘[it] is called’)

(17) Evaluation: ari-ki (アリき ‘[it] was’)

(18) Complication: omopi-ki (オモヒキ ‘[he] thought’); ipi-ki (いひき ‘[he] said’); ipi-ki

(いひき ‘[he] said’); ipi-ki (イヒキ ‘[he] said’); karu (借る ‘[they] borrow’); nari-

ni-ki (なりにき ‘[it] became’); mi-ki (視き ‘[he] saw’); ipi-ki (イヒキ ‘[he] said’);

ipi-ki (いひき ‘[he] said’); toku (說く ‘[he] expounds’); itari-nu (至（り）ヌ ‘[they]

arrive’); maku (散ク ‘[he] spreads [it]’); paosoku.si-ni-ki (飽足（し）ヌ ‘[they]

become satisfied’); omopi-ki (オモヒキ ‘[he] thought’); omopi-ki (オモヒキ ‘[he]

thought’); notamapi-ki (のたまひき ‘[he] honorably said’); ipi-ki (いひき ‘[he]

said’); kaperi-ni-ki (還リにき ‘[they] returned’)

(19) Evaluation: puse-ri (臥セリ ‘[he] has fallen asleep’)

(20) Complication: umare-ki (生れき ‘[they] were born’); ipi-ki (いひき ‘[they] said);

itari-nu (至（り）ヌ ‘[they] arrive’)

(21) Evaluation: nemuri-tari (睡（り）たり ‘[he] has slept’)

(22) Complication: oku (置く ‘[they] place’); oku (置く ‘[they] place’); oku (置（く）

‘[they] place’); oku (置（く） ‘[they] place’)

182

(23) Evaluation: itase-ri (至セリ ‘[they] have brought’)

(24) Complication: terase-simu (照（ら）せしむ ‘[they] make it shine’); satori-satora-

simu (覺リ悟（ら）しむ ‘[they] make [everyone] wake up and realize’); same-nu

(寤（め）ヌ ‘[they] awaken’); sari-ni-ki (去リにキ ‘[they] went’); amepuri-ki (雨

（り）き ‘[they] rained [them]’); amepuri-ki (雨（り）き ‘[they] rained [them]’); uke-

ki (受（け）き ‘[they] received’); ipi-ki (イヒキ ‘[he] said’); ipi-ki (いひき ‘[they]

said’); ipu (いふ ‘[he] says’); yobu (喚（ふ） ‘[he] summons’); itari-nu (至（り）ヌ

‘[he] arrives’); ipi-ki (イヒキ ‘[he] said’); ipi-ki (いひき ‘[he] said’); ipi-ki (イヒキ

‘[he] said’); mi-sime-ki (見しめき ‘[they] made [him] see’); toku (說く ‘[they]

expound’)

(25) Result: tan.si-ki (歎シき ‘[they] were moved [emotionally]’)

(26) Coda: siru-besi (知（る）當（し） ‘[you] should know’); nari (（な）リ ‘[he] is’); nari

(（な）リ ‘[he] is’); nari (（な）リ ‘[he] is’); nari (（な）リ ‘[he] is’); nari (（な）リ

‘[he] is’); nari (（な）リ ‘[he] is’); kiku (聽く ‘[they] listen’); toki-tu (說（き）ツ ‘[I]

expound’); motomu-besi (求む應し ‘[you] should seek’); hauitu.suru koto naku

aru-besi (放逸すること勿（く）ア（る）（應）（し） ‘[you] should not be self-

indulgent’)

When we examine “speech-act” predicates in the above tables, it is clear that we find

them primarily unmarked with ki in the Saidaiji rendition and more often ki-marked in the

Tōdaiji version. In other words, where we find unmarked ipu (‘say’) and notamapu

(‘honorably say’) in the Saidaiji rendition of the parable from Chinese, in the Tōdaiji

rendition ipi-ki (‘said’) and notamapi-ki (‘honorably said’) are used. I propose a

hypothesis regarding this difference in the conclusion of this section.

6.2.4 Conclusions Regarding Differences Between the Sutras

The clearest differences we find in the transitivity, lexical aspect, and discourse

functions of sentence-final predicates governed by the past established fact auxiliary ki

183

are all tied to the two translators’ choices regarding whether “speech-act” predicates after

quotations are generally preferred candidates for ki-marking. Verbs such as ipu (‘say’)

and notamapu (‘honorably say’) following quotations (their only complements) are low-

transitivity predicates that are aspectually Accomplishments and in their narrative

function, Complications. Whereas the translator of the Sinitic Golden Light Sutra at

Saidaiji chose to end quotes simply with ipu (‘say’) in the historical present and leave the

past tense deictic anchoring to the context (in a discourse already framed by the past

established fact auxiliary ki), the translator at Tōdaiji elected to be more explicit in tense

marking after quotations. Looking at this difference quantitatively, out of the 26 quote-

delivering predicates in the Saidaiji rendition of the parable, auxiliary ki is employed in a

total of 8, or just 31% of them. On the other hand, out of the 24 quote-delivering

predicates in the Tōdaiji rendition, we find 20 examples of sentence-final ki, or a full

83% of “speech-act” predicates throughout the Buddha’s depiction of the tale. The

proclivity of the Tōdaiji translator to use ki to this extent on quote-delivering predicates

resulted in the higher percentages of low-transitivity predicates (Accomplishments and

Complications) in the Tōdaiji rendition, compared to those of the Saidaiji version of the

tale analyzed above.

This contrast in ki-marking between the two translations cannot simply be

explained by different morphological or syntactic interpretations of the Sinitic source

text. This is because in both renditions we find cases of sentence-final ipu (‘say’) and ipi-

ki (‘said’) after quotations even though they all are spoken by characters in a tale that is

already clearly framed as having occurred in the past. In other words, there is no

184

morphosyntactic rule in early Heian kundokubun that auxiliary ki becomes necessary

when there is a shift from Embedded Quotation to Embedded Narrative. That being the

case, why is there such a clear difference in grammatical marking between these two

near-contemporary renditions of the same text?

I hypothesize that the perceived need to re-engage in epistemic grounding after a

quotation depends on the translator’s preferences. In other words, for the translator at

Saidaiji, quotations generally remain part of the Embedded Narrative and thus once they

are in place there was no particular need to reestablish that the narration is relating past

established fact. The translator at Tōdaiji, in contrast, recognized Embedded Quotations

and Embedded Narration as different layers of discourse, and thus felt there was an

evidentially motivated, rhetorical need to reframe the narrative as a witnessed past after a

quotation. This hypothesis is buttressed by the fact that we often find ki-marking before

vocatives in kundokubun narration, which was likely to draw attention to the shift in

narrative frame from Embedded Narrative to Overarching Quotation, as we see in the

following example.

(103) 是（の）如ク稱歎すること城邑に周遍せりキ。善女天、時に (Kasuga 1985a: 178)

ko no gotoku syautan.suru koto zyauopu ni

this GEN like praise NML city LOC

syupen.se-ri-ki. Zennyoten toki ni

spread-ARI-KI good.goddess time LOC

‘Thus [his] praise had spread throughout the city. Good Goddess, at that time

[. . .]’

The vocative zennyoten (善女天 ‘good goddess’) is the Buddha speaking directly to his

interlocutor, the goddess of the bodhi tree (菩提樹神 bodai zyusin), in the middle of

185

depicting the tale. Although this vocative is part of the layer of Overarching Quotation,

the surrounding discourse is the Embedded Narration. Shiina (2007) has demonstrated

that vocatives calling out to one’s interlocutor in narrative discourse re-establish the

presence of the speaker. We have discussed the externally established fact auxiliary keri’s

similar discourse function in fictional narrative above, and here I propose a parallel

function for ki in witnessed narrative such as those depicted in Buddhist kundokubun

texts.

186

Chapter 7. Conclusions

This dissertation has examined the use of Early Middle Japanese tense, aspect, and

modality auxiliaries nu, tu, ari, tari, ki, and keri in early Heian (ca. 9th century CE)

renderings of narratives in Sinitic Buddhist sutras. In chapter 1, I introduced and described

the EMJ register used for reading aloud and quoting Sinitic texts, kundokubun. Chapter 2

presented an overview of literature on the TAM auxiliaries in Heian period wabun, the more

vernacular register of Japanese. In chapter 3 I introduced the frameworks that informed my

analyses in this thesis regarding the categories of transitivity, lexical aspect, and narrative

discourse function. Chapter 4 then presented close analyses of selected narratives from

early Heian kundokubun renditions of the Golden Light Sutra, the Ten Wheels Sutra, and

the Immeasurable Meanings Sutra. Chapter 5 shifted the focus to the TAM auxiliaries

themselves with paired analyses based on their shared semantics—the perfectives nu and

tu, the statives ari and tari, and the established fact modals ki and keri. In chapter 6 I

examined differences between kundokubun and wabun and linguistic variation within

kundokubun itself.

While research into kundokubun has been an important subfield of Japanese

linguistics for the past century, there is still much work to do to explain how the translators

187

of the Sinitic source texts interpreted them for a society in which the use of Literary Sinitic

was on the decline among the educated elite (Steininger 2017).

Lurie (2011) argues that the essential nature of literacy in the Nara period and early

Heian period was that of reading Sinitic texts as Japanese. However, Steininger highlights

the following quote from the Tōzan Ōrai (東山往来):

Sutras in on reading contain many meanings. [...] Sutras in kundoku take one of

those many meanings and put it into Yamato words. Because of this, the merit of

sutras in kundoku is lessened. [...] Kundoku can be used to get the gist of the

meaning, but for the perpetual attainment of merit use the original pronunciation.

(Steininger 2017: 151, citing Jōshin 1968)

The phrase “original pronunciation” refers to ondoku (音読), which means ‘reading by

sound,’ or reading using a (phonetically localized) Chinese pronunciation of the Sinitic

text. In other words, while most Sinitic texts in early Japan were read aloud in kundokubun

to better convey their contents, the sutras examined in this study were read aloud in their

“original” Buddhist Hybrid Chinese (rather than Buddhist Hybrid Sanskrit) for liturgical

purposes. However, the emergence of kunten ‘vernacular gloss’ marking in the early Heian

period, to assist in the reading of these sutras in Japanese, suggests that some of the monks

who studied them could no longer bear the linguistic burden of translation without glossing

(Fraleigh 2019).

Although there are narrative passages in pre-Heian texts, such as the Kojiki (古事

記 ‘Record of Ancient Matters,’ as written they are unglossed Sinitic text. While there is

evidence that this Sinitic text followed Japanese syntax in certain clauses (Aldridge 2001)

and was likely read aloud as Japanese using kanbun kundoku (Lurie 2011), the way the

188

narratives were structured, i.e., how their various components were differentiated via

grammatical marking, remains unclear. Sinitic Buddhist texts that were fully marked with

kunten glosses, however, provide the reader with the precise linguistic structures used when

translating them into Japanese. While Taketori monogatari (‘The Tale of the Bamboo

Cutter,’ introduced in section 6.1) may rightfully enjoy the status of the earliest tale written

suing vernacular writing (i.e., kana), kundokubun renditions of Buddhist texts preserved

by gloss provide the earliest examples of fully structured extended narratives in Japanese.

The following table overviews the quantitative results of this study on sentence-

final TAM marking in the Golden Light Sutra, the Ten Wheels Sutra, and the Immeasurable

Meanings Sutra (introduced in Section 3.3 and examined in Chapter 4). It shows how the

monks who preserved their kundokubun renditions via gloss used the auxiliaries of tense,

aspect, and modality in relation to the transitivity, lexical aspect, and narrative discourse

function of the sentences they headed. That is, it reveals how these major resources of the

Japanese language came together to shape and communicate the stories in these foreign

texts.

Table 39 EMJ TAM Auxiliaries in Buddhist Kundokubun Narration

S-F TAM Total Count Transitivity Lexical Aspect Discourse Function

nu 140 low (81%) Achievement (90%) Complication (82%)

tu 120 medium (58%) Achievement (81%) Complication (73%)

ari 309 unrestricted State (97%) Quotation (78%)

tari 101 low (54%) State (100%) Quotation (82%)

ki 229 low (79%) State (38%) Orientation (33%)

keri 30 low (97%) State (100%) Quotation (97%)

189

Looking at the total counts, we see the most prevalent TAM auxiliary in the sutras is

stative ari, with 309 tokens. It is most often found in sentences that serve Quotation

discourse functions, expressing states that follow from events (such as perfects or

resultatives), with their transitivity evenly distributed between low- and medium-

transitivity predicates (50% each). It is no coincidence that the auxiliary most often

deployed by the translator is found in States and Quotations, because all three texts

examined in this study predominantly consist of dialogue between the Buddha and his

interlocutors. While the Buddha occasionally presents a parable (i.e., an embedded

narrative) to provide context for the discussion, he most often expounds on the dharma,

or Buddhist doctrine. This structuring leads to most sentences having the Quotation

discourse function.

Following ari, the past established fact auxiliary ki has the most sentence-final

tokens in these sutras, with a total of 229. Although the data suggest ki has a tendency to

be applied to low-transitivity States, this is due to its use as a narrative device to

evidentially ground scenes in Orientations from which the narrative progresses as a

witnessed past. Such introductions are full of States depicting the initial setting. Shifts in

scenes and narrative layers (i.e., from embedded narration to overarching dialogue) are

also prime targets for ki-marking, to re-ground the narrative as witnessed past, after a

lengthy dialogue or an extended passage of historical-present narration. Ikegami writes

that, in Japanese, once narration has been deictically anchored in the past using past tense

morphemes at the beginning of a narrative passage, a shift to the so-called historical

present actually indicates that the passage has been sufficiently framed as past tense and,

190

thus, narration that appears to be in the present is best interpreted as unmarked for tense

and understood as being in the past (Ikegami 1986: 66).169 The use of ki in concluding

extended passages suggests the translator-monks felt the witnessed past previously

established with ki for the narrative needed grammatical maintenance after the shifts

described above. Auxiliary ki’s use in grounding the narrative also clarifies why it is

found in such a high percentage (79%) of low-transitivity sentences. At the peripheries of

a narrated passage, where we often find ki, we find not just low-transitivity States, but

also low-transitivity verbs of motion and quote-delivering predicates. It is, therefore,

better to understand ki itself as being unrestricted semantically for transitivity, with these

discourse contexts leading to the apparent preference to its skewing so strongly toward

Stative predicates.

The perfective auxiliaries follow in their total token counts, with nu at 140 and tu

at 120 across the texts. We find them differing in their transitivity as we have come to

expect, given their semantics. The endoactive perfective auxiliary nu is most often found

in low-transitivity predicates and the exoactive perfective auxiliary tu is found in

medium-transitivity predicates—an unsurprising fact, since there are few high-transitivity

predicates in these sutras. Where medium-transitivity predicates do appear in

Complications, however, they are prime targets for tu-marking. We also find lower ratios

169 This is a well-documented cross-linguistic phenomenon, as Fleischman (1991) writes:

“In narrative discourse, time reference is normally established at the outset of the text, and since it tends to

be a property of large stretches of discourse, or even of entire texts, it need not in principle be reiterated in

each successive clause. [. . .] One result is that in the narrative grammars of many languages tense is in large

measure freed from its primary referential function of locating events in time, and the available morphology

is pressed into service for other, notably pragmatic, purposes.” (Fleischman 1991: 28)

191

for tu compared with nu in their use with Achievements and Complications. This contrast

arises because tu heads a greater ratio of Accomplishments (durative-cum-telic

predicates) than its endoactive counterpart nu. As our data are from narratives that

include many quoting or “speech-act” predicates, and these are Accomplishments, they

are routinely deictically grounded by the speaker relating them as recent past with tu.

With regards to discourse function, we find 15 more cases of tu inside Quotations than nu

not because tu predicates are inherently more quotable, but rather because medium-

transitivity predicates such as etu ‘gain(ed), attain(ed)’ account for so much of what is

said when discussing what followers of Buddhism attain from correct practice.

Next, we have the still-emergent stative tari, with 101 tokens across the three sutra

renditions. In comparison to the other TAM auxiliaries, it shares most features with its older

counterpart ari. Although tari displays a slightly greater ratio than ari when it comes to

being applied to low-transitivity predicates, the difference is only four percentage points.

In addition, it is found mainly in Quotations describing Buddhist deities and practices, for

example in predicates such as masare-tari ‘having surpassed (i.e., being superior).’ When

used within narrative passages, tari, like ari, is found primarily in Evaluations, at 11%.

Finally, the externally established fact auxiliary keri only has 30 tokens

throughout the data set. As discussed in Section 6.1, the Japanese genre of sutra narration

is not about epistemically distant, non-witnessed facts. The narrator begins the outermost

narrative by saying he has heard the conversation he then presents, and the embedded

parables told by the Buddha are often presented as past lives he himself lived and

experienced (both situations grammatically marked epistemically throughout the sutras

192

with the established past fact auxiliary ki). Auxiliary keri, on the other hand, is the most

embedded (i.e., removed from the discourse’s leading edge) of the TAM auxiliaries.

Furthermore, it is also most removed from the plot-driving action, being primarily used in

low-transitivity (97%) States (100%) in Quotations (97%). This is consonant with the fact

that it is most often used by characters in asserting what truths they have realized upon

hearing the words of the Buddha.

Although kundokubun came to be a highly codified variety of Japanese associated

with rendering Sinitic texts sentence-by-sentence without, presumably, concern for larger

discourse structures, this dissertation has demonstrated how, in its earliest stages, utilizers

of this register used diverse linguistic styles and employed complex narrative discourse

strategies when translating Sinitic Buddhist texts. In other words, Buddhist monks used

effective story-crafting strategies in their renditions in order to captivate their audiences

with engaging narratives, both matrix and embedded, within the sutras, like those of

Jalavāhana in the Golden Light Sutra and that of the drunken elephant in the Ten Wheels

Sutra. To do so, the practitioners who engaged in sutra translation in early Heian Japan

employed the tense, aspect, and modality auxiliaries at their disposal with maximum

rhetorical force when rendering texts in Japanese. The perfective aspect auxiliaries nu

and tu were used in narration to foreground moments of transformation and movement,

grammatically specifying changes and realizations (and implying their consequences) in

the story’s progression. The narrative function of predicates headed with the stative

aspect markers ari and tari, on the other hand, was to lay out background information in

193

order to establish a scene.170 The externally established fact auxiliary keri occurs

predominantly in quotes by characters in these stories to express their admiration,

acceptance, and subsequent internalization of the words of the Buddha. Finally, the past

established fact auxiliary ki was employed to ground the discourse in a witnessed past.

The narratives examined in this study, although translations of Sinitic religious

texts, fill a significant gap in our knowledge of how stories were structured

grammatically in Japanese in the first century of the Heian period, a time when written

genres outside of verse were coming into their own. The uses- of Early Middle Japanese

tense, aspect, and modality auxiliaries to foreground, background, and create frames in

kundokubun narratives show us that the Buddhist practitioners who rendered these texts

had full knowledge of both their overarching and embedded structures and took care to

differentiate them as they read and wrote between the lines.

170 We see this when the verb (rather than the auxiliary) ari ‘be/have’ heads a sentence as well.

194

References

Abe Akio (阿部秋生), Ken Akiyama (秋山虔) & Genʼe Imai (今井源衛). 1970.

Nihon koten bungaku zenshū (日本古典文学全集) [‘Complete works of

Japanese classical literature’] 12: Genji monogatari (源氏物語) [‘The tale of

Genji’]. Tokyo: Shōgakkan (小学館).

Aikhenvald, Alexandra Y. 2012. “The Essence of Mirativity.” Linguistic Typology

16: 435–485.

Alberizzi, Valerio Luigi. 2014. “An Introduction to kunten Glossed Texts and Their

Study in Japan.” Les dossiers de HEL: Société d'Histoire et d'Epistémologie des

Sciences du Langage 7: 1–9.

――――. 2015. “The Role of kunten Materials in the Process of Sino-Japanese

Hybridization.” Quaderni di Linguistica e Studi Orientali / Working Papers in

Linguistics and Oriental Studies 1: 233–258.

Aldridge, Edith. 2001. “Hentai Kambun Perspective on Short Scrambling.” Journal

of East Asian Linguistics 10(2): 169–200.

Biber, Douglas & Susan Conrad. 2019. Registers, Genres, and Styles: Fundamental

Varieties of Language. Cambridge: Cambridge University Press.

Binnick, Robert. 1991. Time and the Verb: A Guide to Tense and Aspect.

Oxford/New York: Oxford University Press.

Bundschuh, John. 2020. “Rethinking Japan’s Earliest Written Narratives: Early-

Heian Kundokugo Translations of Chinese Buddhist Texts.” In Jake Odagiri &

Nicholas Lambrecht, eds. The Anthology of Transborder Cultural Studies (越境

文化研究イニシアティヴ論集) 3: 43–63.

195

――――. 2021. “Evidence in Heian Buddhist Kundokugo Narration.” In Matthew

Mewhinney, ed. Evidence, Transmission, and Inheritance in Japanese

Literature and Media (Proceedings of the Association for Japanese Literary

Studies Vol. 20, 2019): 143–156.

――――. Forthcoming (2021)a. “Predicate Morphology and Narrative Structure in

Early Heian Japanese Buddhist Texts: A Case Study of the Saidaiji Golden

Light Sutra.” In Mineharu Nakayama, ed. DEALL 50th Anniversary Celebration

Issue.

――――. Forthcoming (2021)b. “Statives -(a)ri and -tari in Early Heian Japanese

Kundokubun.” In Hae-Sung Jeon, ed. Japanese/Korean Linguistics 28.

――――. Forthcoming (2021)c. “The Narrative Functions of Perfective Auxiliaries

in Early Heian Kundokubun Texts.” In Mineharu Nakayama & Wei Zhou, eds.

Buckeye East Asian Linguistics 5.

Buswell Jr., Robert E. & Donald S. Lopez Jr. 2014. The Princeton Dictionary of

Buddhism. Princeton, NJ: Princeton University Press.

Bybee, Joan. 1985. Morphology: A Study of the Relation Between Meaning and

Form. Philadelphia: John Benjamins Publishing Company.

Bybee, Joan & Paul Hopper, eds. 2001. Frequency and the Emergence of Linguistic

Structure. Philadelphia: John Benjamins Publishing Company.

Bybee, Joan, Revere Perkins & William Pagliuca. 1993. The Evolution of Grammar:

Tense, Aspect, and Modality in the Languages of the World. Chicago:

University of Chicago Press.

Chafe, Wallace. 1999. “Some Things That Narratives Tell Us About the Mind.”

Narrative Language: 79–98. Hillsdale, NJ: Lawrence Erlbaum Associates.

Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.

Dahl, Osten. 1985. Tense and Aspect Systems. New York: Blackwell Publishers.

DeLancey, Scott. 1997. “Mirativity: The Grammatical Marking of Unexpected

Information.” Linguistic Typology l: 33–52.

196

Du Bois, Jack W. 1986. “Self-Evidence and Ritual Speech.” In Wallace Chafe &

Johanna Nichols, eds. Evidentiality: The Linguistic Coding of Epistemology:

313–336. Norwood, NJ: Ablex Publishing Corporation.

Edgerton, Franklin. 1936. “The Prakrit Underlying Buddhistic Hybrid Sanskrit.”

Bulletin of the School of Oriental Studies 8(2/3): 501–516.

Emmerick, R. E. 1970. The Sūtra of Golden Light: Being a Translation of the

Suvarṇaprabhāsasūtra. London: Luzac & Company.

Fleischman, Suzanne. 1985. “Discourse Functions of Tense-aspect Oppositions in

Narrative: Toward a Theory of Grounding.” Linguistics 23: 851–882.

――――. 1991. “Verb Tense and Point of View in Narrative.” In Fleischman, Suzanne

& Linda R. Waugh, eds. Discourse-pragmatics and the Verb: The Evidence from

Romance: 26–54. Routledge.

Frellesvig, Bjarke. 2010. A History of the Japanese Language. Cambridge: Cambridge

University Press.

Fraleigh, Matthew. 2019. “Rearranging the Figures on the Tapestry: What Japanese

Direct Translation of European Texts Can Tell Us About Kanbun Kundoku.”

Japan Forum 31(1): 4–32.

Fujii Toshihiro (藤井俊博). 2013a. “Kohon setsuwashū no keri no tekusuto kinō:

nikeri, kakarimusubi no shūketsu kinō (古本説話集の「けり」のテクスト機能―

「にけり」「係り結び」の終結機能―) [‘The textual function of keri in the

‘Collection of old tales’: The conclusive functions of ni-keri and kakari-

musubi’].” Dōshisha kokubungaku (同志社国文学) [‘Doshisha studies in

Japanese literature’] 78: 130–144.

――――. 2013b. “Konjaku monogatari shū no nikeri: Tekusuto kinō no shosō (今

昔物語集の「にけり」―テキスト機能の諸相―) [‘ni-keri in Konjaku monogatari

shū: Various aspects of its textual functions’].” Hyōgen kenkyū (表現研究)

[‘Expression-formation studies’] 98: 51–60.

――――. 2016. “Yashiro-bon Heike monogatari no ‘ki’ ‘keri’ no tekusuto kinō. (屋代

本平家物語の「き」「けり」のテクスト機能: 覚一本との比較) [The textual functions

of ki and keri of the Yashiro Heike Monogatari: A comparison with the Kaku’ichi

197

text].” Dōshisha kokubungaku (同志社国文学) [‘Doshisha Japanese Literature’] 84:

247–259.

Fukumori, Naomi. 2001. “Chinese Learning as Performative Power in Makura no

sōshi and Murasaki Shikibu nikki.” Proceedings of the Association for Japanese

Literary Studies: Acts of Writing 2: 101–119.

Guest, Jennifer. 2013. Primers, Commentaries, and Kanbun Literacy in Japanese

Literary Culture, 950-1250 C.E. Dissertation, Columbia University. Ann Arbor:

ProQuest/UMI.

Hopper, Paul & Sandra A. Thompson. 1980. “Transitivity in Grammar and

Discourse.” Language 56-2: 251–299.

Hopper, Paul. 1987. “Emergent Grammar.” The Annual Proceedings of the Berkeley

Linguistics Society 13: 139–157. University of California, Berkeley.

――――. 1991. “On Some Principles of Grammaticalization.” In Elizabeth Closs

Traugott and Bernd Heine, eds. Approaches to Grammaticalization, Vol. 1: 17–35.

Amsterdam: John Benjamins.

Hosoe Ikki (細江逸記). 1932. Dōshi jisei no kenkyū (動詞時制の研究) [‘Research

on verb tense’]. Tokyo: Taibuntō (泰文堂).

Ijima Masahiro (井島正博). 2011. Chūkogo kako, kanryō hyōgen no kenkyū (中古語

過去・完了表現の研究) [‘Research on Early Middle Japanese past and

perfective expressions’]. Tokyo: Hitsuji Shobō (ひつじ書房).

Ikegami Yoshihiko (池上嘉彦). 1986. “Nihongo no katari no tekusuto ni okeru jisei

no tenkan ni tsuite (日本語の語りのテクストにおける時制の転換について)

[‘Regarding the change of tense in Japanese narrative texts’].” Kigōgaku kenkyū

(記号学研究) [‘Studia semiotica’] 6: 61–74.

Inamura Tadashi (稲村徳). 1977. Kotenbun no sōgō kenkyū (古典文の総合研究)

[‘General research on classical texts’]. Tokyo: Yūseido (有精堂).

Ishizuka Harumichi (石塚晴通). 2001. “Heian jidai shoki (平安時代初期) [‘The

early Heian period’].” In Yoshida Kanehiko (吉田金彦), Ishizuka Harumichi

(石塚晴通), Tsukishima Hiroshi (築島裕), Tsukimoto Masayuki (月本雅幸),

198

eds. Kuntengo jiten (訓点語辞典) [‘A dictionary of glossing language’]: 28–

32. Tokyo: Tōkyōdō shuppan (東京堂出版).

Janda, Richard & Brian Joseph. 1999. “The Modern Greek Negator mi(n)(-) as a

Morphological Constellation.” Greek Linguistics: Proceedings of the 3rd

International Conference on Greek Linguistics: 341–351. Athens: Elinika

Gramata.

Jōshin (定深). 1968. “Tōzan Orai (東山往来) [‘Tōzan correspondence’].” In

Ishikawa Ken (石川謙), ed. Nihon Kyōkasho Taikei: Ōrai-hen (日本教科書大

系往来編) [‘A comprehensive collection of Japanese textbooks:

Correspondence edition’]: 367–422. Kōdansha (講談社).

Kabutogi Shōkō (兜木正享) & Nakada Norio (中田祝夫). 1979. Kotenbon shiryō

sōkan 1: Muryōgikyō koten (古点本資料叢刊 1: 無量義経古点) [‘Old

glossing materials series 1: The old glosses of the Immeasurable Meanings

Sutra’]. Tokyo: Benseisha (勉誠社).

Kasuga Masaji (春日政治). 1985(1942)a. Saidaiji-bon Konkōmyō saishōōkyō koten no

kokugogakuteki kenkyū: Honbun-hen (西大寺本金光明最勝王経古点の国語学的

研究 本文編) [‘A linguistic study of the old glosses of the Saidaiji temple Golden

Light Sutra: Translation volume’]. Tokyo: Benseisha (勉誠社).

――――. 1985(1942)b. Saidaiji-bon Konkōmyō saishōōkyō koten no kokugogakuteki

kenkyū: Kenkyū-hen (西大寺本金光明最勝王経古点の国語学的研究 研究編)

[‘A linguistic study of the old glosses of the Saidaiji temple Golden Light Sutra:

Research volume’]. Tokyo: Benseisha (勉誠社).

Katō Kōji (加藤浩司). 1998. Ki, keri no kenkyū (キ・ケリの研究) [‘Research on ki and

keri’]. Tokyo: Izumi shoin (和泉書院).

Kenny, Anthony. 1963. Action, Emotion, and Will. London: Routledge and Kegan Paul.

Kern, John Christopher. 2007. The Narrative Structure of the Embedded Tales in the

“Hahakigi” Chapter of the Genji Monogatari. MA thesis: The Ohio State University.

Kobayashi Yoshinori (小林芳規). 2011. Heian jidai no bussho ni motozuku kanbun

kundokushi no kenkyū 1: Jojutsu no hōhō (平安時代の仏書に基づく漢文訓読

199

史の研究(1)叙述の方法) [‘Research on the history of kanbun kundoku based

on Heian period Buddhist texts 1: Narrative methodology’]. Tokyo: Kyūko

Shoin (汲古書院).

――――. 2012a. Heian jidai no bussho ni motozuku kanbun kundokushi no kenkyū

3: Shoki kundokugo taikei (平安時代の佛書に基づく漢文訓讀史の研(3)初期

訓讀語體系) [‘Research on the history of kanbun kundoku based on Heian

period Buddhist texts 3: The Early-period kundokugo system’]. Tokyo: Kyūko

Shoin (汲古書院).

――――. 2012b. Heian jidai no bussho ni motozuku kanbun kundokushi no kenkyū

4: Chūki kundokugo taikei (平安時代の佛書に基づく漢文訓讀史の研(4)中期

訓讀語體系) [‘Research on the history of kanbun kundoku based on Heian

period Buddhist texts 4: The Middle-period kundokugo system’]. Tokyo: Kyūko

Shoin (汲古書院).

――――. 2013. Heian jidai no bussho ni motozuku kanbun kundokushi no kenkyū

5: Kōki kundokugo taikei (平安時代の佛書に基づく漢文訓讀史の研(5)後期

訓讀語體系) [‘Research on the history of kanbun kundoku based on Heian

period Buddhist texts 5: The Late-period kundokugo system’]. Tokyo: Kyūko

Shoin (汲古書院).

――――. 2017. Heian jidai no bussho ni motozuku kanbun kundokushi no kenkyū

7: Hensen no genri (平安時代の佛書に基づく漢文訓讀史の研(7)変遷の原

理) [‘Research on the history of kanbun kundoku based on Heian period

Buddhist texts 7: Principles of change’]. Tokyo: Kyūko Shoin (汲古書院).

Kornicki, Peter Francis. 2018. Languages, Scripts, and Chinese Texts in East Asia.

Oxford: Oxford University Press.

Kumakura, Chiyuki. 1980. The Narrative Time of Genji monogatari. Dissertation,

University of California, Berkeley. Ann Arbor: ProQuest/UMI.

Labov, William. 1972. “The Transformation of Experience in Narrative Syntax.”

Language in the Inner City: 354–396. Philadelphia: University of Pennsylvania

Press.

200

Lau, Monica Laura & Johan Rooryck. 2017. “Aspect, Evidentiality, and Mirativity.”

Lingua 186-187: 110–119.

Lowe, Bryan D. 2017. Ritualized Writing: Buddhist Practice and Scriptural Cultures

in Ancient Japan. Honolulu: University of Hawai’i Press.

Lurie, David. 2011. Realms of Literacy: Early Japan and the History of Writing.

Cambridge: Harvard University Press.

Mair, Victor H. 1994. “Buddhism and the Rise of the Written Vernacular in East

Asia: The Making of National Languages.” The Journal of Asian Studies 53(3):

707–751.

Martin, Samuel E. 1987. The Japanese Language Through Time. New Haven: Yale

University Press.

Matsumoto Mitsutaka (松本光隆). 2011a. “Ishiyama-dera-zō bussetsu Taishi sudadakyō

Heian chūkiten ni okeru kundokugo no buntai (石山寺蔵仏説太子須陀拏経平安中

期点における訓読語の文体). [‘The style of the kundokugo in the mid-Heian

glossings of the Ishiyama temple Sudāna Sutra’]” Kuntengo to kunten shiryō (訓点

語と訓点資料) [‘Diacritical language and diacritical materials’] 127: 210–222.

――――. 2011b. “Chūin Sōjō Meizan no kundokugo (shita): Myōhō Renge Kyō Meizan

ten no bunmatsu hyōgen (中院僧正明算の訓読語(下): 妙法蓮華経明算点の文末

表現) [‘The kundokugo of Chu-in High Priest Meizan (part 2): The sentence-final

expressions of the Lotus Sutra Meizan glossings’].” Hiroshima daigaku daigakuin

bungaku kenkyūka ronshū (広島大学大学院文学研究科論集) [‘The Hiroshima

University Studies, Graduate School of Letters’] 71: 53–74.

――――. 2017. Heian Kamakura jidai kanbun kundokugo kaiseki ron (平安鎌倉時

代漢文訓読語解析論) [‘An analytical theory of Heian and Kamakura period

kanbun kundokugo’]. Tokyo: Kyūko Shoin (汲古書院).

Murasaki Shikibu. 2001. The Tale of Genji. Royall Tyler, trans. NY: Penguin

Classics.

201

Nakada Norio (中田祝夫). 1958. Kotenbon no kokugogakuteki kenkyū yakubun hen (古

点本の国語学的研究訳文篇) [‘A linguistic study of old glossed texts: Translation

volume’]. Tokyo: Kōdansha (講談社).

――――. 1979 (1954). Kokunbon no kokugogakuteki kenkyū sōron-hen (古点本の国語

学的研究総論篇) [‘A linguistic study of old glossed texts: General theory

volume’]. Tokyo: Benseisha (勉誠社).

――――. 1980. Kotenbon shiryō sōkan 2: Shōsōinbon Jizō jūrinkyō kango, shichi

Gangyō ten (古点本資料叢刊 2: 正倉院本地蔵十輪経(巻五・七)元慶点)

[‘Old glossing materials series 2: Ca. 880 CE glossings of the 5th and 7th scrolls

of the Shōsōin Ten Wheels Sutra’]. Tokyo: Benseisha (勉誠社).

Nara, Hiroshi. 2011. “Aspect and Discourse in Tense-Switching: A Case Study of

Natsume Sōseki’s Botchan.” Japanese Language and Literature 45: 273–305.

Oda Masaru (小田勝). 2015. Jitsurei shōkai koten bunpō sōran (実例詳解古典文法総覧

) [‘A Reference Grammar of Classical Japanese’]. Tokyo: Izumi Shoin (和泉書院).

Ogawa, Nobuo. 1983. The meaning and function of the suffixes -ki, -keri, -tu, -nu, -

tari, and -ri in Genzi Monogatari. Dissertation: University of Pennsylvania.

Okada, Richard. 1991. Figures of Resistance: Language, Poetry and Narrating in

The Tale of Genji and other Mid-Heian Texts. Durham: Duke University Press.

Okimori Takuya (沖森卓也). 2001. “Konkōmyō saishō ōkyō (金光明最勝王経)

[‘The Golden Light Sutra’]. In Kanehiko Yoshida (吉田金彦), Harumichi

Ishizuka (石塚晴通), Hiroshi Tsukishima (築島裕), Masayuki Tsukimoto (月

本雅幸), eds. Kuntengo jiten (訓点語辞典) [‘A dictionary of vernacular

glossing language’]. 134–135. Tokyo: Tōkyōdō shuppan (東京堂出版).

Omodaka Hisataka (澤瀉久孝) et al. (eds). 1967. Jidaibetsu kokugo daijiten: jōdai

hen (時代別国語大辞典: 上代編) [‘A periodized dictionary of Japanese: Old

Japanese edition’]. Tokyo: Sanseidō (三省堂).

Otsubo Heiji (大坪併治). 1961. Kuntengo no kenkyū (訓点語の研究) [‘Research on

kuntengo’]. Tokyo: Kazama shobō (風間書房).

202

――――. 1968. Kuntenshiryō no kenkyū (訓点資料の研究) [‘Research on glossed

materials’]. Tokyo: Kazama shobō (風間書房).

――――. 1977. “Setsuwa no jojutsu keishiki to shite mita jodōshi ki, keri: Kunten

shiryō o chūshin ni. (説話の叙述形式として見た助動詞キ・ケリ―訓点資料を中心

に―) [‘Auxiliaries ki and kari viewed from their narrative forms: Focusing on

glossed materials’]” Kokugogaku (国語学) [‘Japanese linguistics’] 111: 25–38.

――――. 1981. Heian jidai ni okeru kuntengo no bunpō (平安時代における訓点語の

文法) [‘The grammar of Heian period kuntengo’]. Tokyo: Kazama shobō (風間書

房).

――――. 1992. Ishiyamadera bon Daihōkōbutsu kegonkyō koten no kokugogakuteki

kenkyū (石山寺本大方廣佛華厳經古點の國語學的研究) [‘Linguistic research of

the old glossings of the Ishiyama temple Flower Garland Sutra’]. Tokyo: Kazama

shobō (風間書房).

Oya Tōru (大矢透). 1909. Kana oyobi kana jitai senkaku shiryō (仮名遣及仮名字体沿

革史料) [‘Materials on the history of the syllabaries and their written styles’].

Tokyo: Kokutei kyōkasho kyōdō hanbaisho (国定教科書共同販売所).

Quinn, Charles. 1983. “Ochō sanbun no gyōshūsei: jisei to asupekuto o chūshin ni.

(王朝散文の凝集性―時制とアスペクトを中心に―) [‘Cohesion in Heian

vernacular prose: The role of mood, tense, and aspect’].” Proceedings of the 7th

International Conference on Japanese Literature in Japan: 9–29.

――――. 1987. A Functional Grammar of Predication in Classical Japanese.

Dissertation, University of Michigan.

――――. 1990. “Suffixal Concatenation in the Classical Japanese Predicate:

Erstwhile Serial Verbs?” In Joseph, Brian & Arnold Zwicky, eds. When Verbs

Collide: Papers from the Ohio State Mini-Conference on Serial Verbs

(Columbus, Ohio, May 26–27, 1990). Working Papers in Linguistics 39: 247–

264.

――――. 2018. The Epistemic Dimension in Heian Japanese [Conference Paper

Presentation]. 27th Annual Meeting of the Association of Japanese Literary

Studies, U.C. Berkeley.

203

――――. 2021. “Classical Japanese Portal.” https://cjp.asc.ohio-state.edu/.

Sakakura Atsuyoshi (阪倉篤義). 1956. “Taketori monogatari ni okeru ‘buntai’ no

mondai (竹取物語における「文体」の問題) [‘The issue with ‘register’ in

Taketori monogatari’].” Kokugo kokubun (国語国文) [‘Japanese language and

literature’] 25(11): 683–693.

Sandness, Karen E. 1999. The Evolution of the Japanese Past and Perfective

Suffixes. Ann Arbor, MI: University of Michigan Press.

Sato, Katsuyuki. 2017. “Chapter 7. Kundoku-bun: A Hybrid Genre in Japanese

Literature.” In Elizabeth A. Thomson, Motoki Sano, and Helen de Silva Joyce.

Mapping Genres, Mapping Culture: Japanese Texts in Context: 169–189.

Philadelphia: John Benjamins Publishing Company.

Semizu, Yukino. 2006. “Invisible Translation: Reading Chinese Texts in Ancient

Japan.” In Theo Hermans, ed. Translating Others: 283–295. Manchester, UK:

St. Jerome Publishing.

Shi, Xiangdong. 2014. “The Influence of Buddhist Sanskrit on Chinese.” In Wang,

William S-Y & Chaofen Sun. The Oxford Handbook of Chinese Linguistics:

236–247. Oxford: Oxford University Press.

Shiina, Michi. 2007. “Positioning and Functioning of Vocatives: Casework in Historical

Pragmatics 1.” Hōsei daigaku bungakubu kiyō （法政大学文学部紀要） [‘Bulletin

of Faculty of Letters, Hosei University’] 55: 17–32.

Shinzato, Rumiko. 1991. “Where Do Temporality, Evidentiality and Epistemicity

Meet? A Comparison of Old Japanese -ki and -keri with Turkish -di and -miş.”

Gengo kenkyū 3: 25–57.

Shirane, Haruo. 1987. The Bridge of Dreams: A Poetics of the Tale of Genji.

Stanford: Stanford University Press.

Soothill, William E. & Lewis Hodous. 2014 (1937). A Dictionary of Chinese

Buddhist Terms: With Sanskrit and English Equivalents and a Sanskrit-Pali

Index. New York: Routledge.

Steininger, Brian. 2017. Chinese Literary Forms in Heian Japan: Poetic and

Practice. Cambridge, MA: Harvard University Asia Center.

204

Stinchecum, Amanda M. 1985. Narrative Voice in The Tale of Genji. Urbana:

University of Illinois Center for East Asian and Pacific Studies.

Suzuki Tai (鈴木泰). 1999 (1992). Kodai Nihongo dōshi no tensu, asupekuto - Genji

monogatari no bunseki (古代日本語時間表現の形態論的研究) [‘Tense and

aspect of older Japanese verbs: An analysis of Genji monogatari’]. Tokyo:

Hitsuji Shōbō (ひつじ書房).

――――. 2009. Kodai Nihongo jikan hyōgen no keitai ronteki kenkyū (古代日本語

時間表現の形態論的研究) [‘A morphological study of Classical Japanese

expressions of time’]. Tokyo: Hitsuji Shobō (ひつじ書房).

Tabuchi Masao (田渕雅生). 1987. “Tōdaiji toshokanbon Konkōmyō saishōōkyō

chūshaku no kunten (東大寺図書館蔵本金光明最勝王経註釈の訓点) [‘The

glossings of the Tōdaiji library Golden Light Sutra commentary’].” Kuntengo to

kunten shiryō (訓点語と訓点資料) [‘Diacritical language and diacritical

materials’] 76: 53–90.

Takeoka Masao (竹岡正夫). 1979 (1963). “Jodōshi ‘keri’ no hongi to kinō. (助動詞

“けり”の本義と機能) [The meaning and function of auxiliary keri’]” Umehara

Yasunori (梅原恭則), ed. Jodōshi (Ronshū Nihongo Kenkyū 7) (助動詞（論集

日本語研究 7）) [‘Auxiliaries (Collected papers in Japanese linguistics 7)’]:

178–192. Tokyo: Yūseidō (有精堂).

Takeuchi, Lone. 1987. A Study of Classical Japanese Tense and Aspect.

Copenhagen: Akademisk Forlag.

Tsukishima Hiroshi (築島裕). 1955. “Kokinshū kanajo to kanbun kundoku (古今集

仮名序と漢文訓読) [‘The kana preface of the Kokinshū and kanbun

kundoku’].” Tōkyō daigaku kyōyō gakubu jinbunka kiyō (東京大学教養学部人

文科学科紀要) [‘Tokyo University faculty of liberal arts graduate school of the

humanities bulletin’] 7: 183–206.

――――. 1963. Heian jidai no kanbun kundokugo ni tsukite no kenkyū (平安時代

の漢文訓讀語につきての硏究) [‘Research regarding Heian period kanbun

kundoku language’]. Tokyo: Tōkyō daigaku shuppan kai (東京大学出版会).

205

――――. 1967. Kōfukiji-bon Daijionji Sanzōhōshiden koten no kokugakuteki kenkyū:

Kenkyū hen (興福寺本大慈恩寺三蔵法師伝古点の国語学的研究 研究篇) [‘A

linguistic study of the old diacritics of the Kofukuji Daci'en Temple biography of

Xuanzang: Research volume’]. Tokyo: Tōkyō Daigaku Shuppankai (東京大学出版

会).

――――. 1982 (1969). Heian jidaigo shinron (平安時代語新論) [‘A new treatise

on the language of the Heian period’]. Tokyo: Tōkyō daigaku shuppan kai. (東

京大学出版会)

――――. 2001. “Kunten gogaku kenkyū-shi (訓点語学研究史) [‘A history of

glossing language research’].” In Yoshida Kanehiko (吉田金彦), Ishizuka

Harumichi (石塚晴通), Tsukishima Hiroshi (築島裕), Tsukimoto Masayuki (月

本雅幸), eds. Kuntengo jiten (訓点語辞典) [‘A dictionary of glossing

language’]: 4–21. Tokyo: Tōkyōdō shuppan (東京堂出版).

Ueda, Atsuko. 2008. “Sound, Scripts, and Styles: Kanbun kundokutai and the

National Language Reforms of 1880s Japan.” Review of Japanese Culture and

Society 20: 133–156.

Vendler, Zeno. 1957. “Verbs and Times.” The Philosophical Review 66(2): 143–160.

Watase Shigeru (渡瀬茂). 2013. Ōchō jodōshi kinō ron: Anatanaru ba, waku kōzō,

enkinhō (王朝助動詞機能論：あなたなる場・枠構造・遠近法) [‘The functions

of Heian period auxiliaries: Removed ground, framing construction,

distancing’]. Tokyo: Izumi Shoin (和泉書院).

Whitman, John, Miyoung Oh, Jinho Park, Valerio Luigi Alberizzi, Masayuki

Tsukimoto, Teiji Kosukegawa & Tomokazu Takada. 2010. “Toward an

International Vocabulary for Research on Vernacular Readings of Chinese

Texts.” Scripta 2: 1–22.

Yamaguchi Yoshinori (山口佳紀). 1986. Kodai Nihongo bunpō no seiritsu no

kenkyū (古代日本語文法の成立の研究) [‘Research on the establishment of

Old Japanese grammar’]. Tokyo: Yūseidō (有精堂).

Yoshida Kanehiko (吉田金彦), Ishizuka Harumichi (石塚晴通), Tsukishima Hiroshi

(築島裕), Tsukimoto Masayuki (月本雅幸), eds. Kuntengo jiten (訓点語辞典)

[‘A dictionary of glossing language’]. Tokyo: Tōkyōdō (東京堂).

206

Yoshizawa Yoshinori (吉澤義則). 1919. Iwasaki bunkosho-zō shōsho oyobi Nihon

shoki koshōbon ni kuwaeraretaru okototen ni tsukite (岩崎文庫所蔵尚書及び

日本書紀古鈔本に加へられたる乎古止点に就きて) [‘Regarding the

wokototen added to the old copies of the Book of Documents and Chronicles of

Japan in the Iwasaki collection’]. Tōkyō: Iwasaki bunko (岩崎文庫).

Zhu, Qingzhi. 2017. “Buddhist Chinese (Buddhist Hybrid Chinese).” In Rint

Sybesma (ed.), Encyclopedia of Chinese Language and Linguistics Volume 1

A–Dăi. 318–330. Brill: Boston.

207

Appendix A. Overview of the Tense, Aspect, and Modality Auxiliaries in Early Heian

Kundokubun

This appendix serves to provide an overview of the tense, aspect, and modality

(TAM) auxiliaries in early Heian kundokubun texts examined in this dissertation. As

discussed in section 1.4, the TAM auxiliaries I examine are specifically those that govern

the participle (J. ren’yōkei 連用形)—nu, tu, ari, tari, ki, and keriError! Reference

source not found.Error! Reference source not found.Error! Reference source not

found.Error! Reference source not found.Error! Reference source not found.. Below

I provide examples and describe the semantics and primary functions of each in

kundokubun narrative. The example numbers below correspond to those in the

dissertation-proper above, where discussion on the context of each depicted scene and its

linguistic expressions can be referenced.

Endoactive Perfective Auxiliary Nu

It expresses a realized change of state in endoactive predicates, which express

situations that generally do not involve volition or transitivity. This basic meaning in

context serves a variety of purposes, such as explicit confirmation of a situation’s

realization. Its primary narrative function, in early Heian Buddhist kundokubun texts, is

208

to foreground events such as completed movements made by the characters, as we see in

(6), (62), and (76), and sudden changes in both their internal (63) and external (18) states.

(6) 悉ク皆雲のゴトク集（り）ヌ。 (Kasuga 1985a: 4)

kotogotoku mina kumo no gotoku atumari-nu

all everyone cloud GEN like gather-NU

‘They all gathered like clouds.’

(62) 仏の所に来詣しぬ。 (Kabutogi & Nakada 1979: 78, 82)

potoke no tokoro ni raikei.si-nu

buddha GEN place LOC approach-NU

‘[Each] approached the place of the Buddha.’

(76) 彼の池の邊に至りぬ。 (Tabuchi 1987: 82)

ka no ike no potori ni itari-nu

that GEN pond GEN side LOC arrive-NU

‘They arrived at that lake.’

(63) 色を見、香を聞くときに、自然に飽足（し）ぬ。 (Kabutogi & Nakada 1979: 83)

siki wo mi kau wo kiku toki ni

form ACC seeing incense ACC smell time LOC

sizen ni pausoku.si-nu

natural COP be.satisfied-NU

‘When he saw the form and smelled the incense, he naturally became satisfied.’

(18) 時に十千の魚同時に命過して三十三天に生レヌ。 (Kasuga 1985a: 183)

toki ni zipusen no uo dauzi ni inoti

time LOC 10,000 GEN fish same.time LOC life

sugosi-te sanzipusanten ni umare-nu

pass-GER heaven LOC be.born-NU

‘Then the ten thousand fish passed away at the same time and were [re]born in

heaven.’

209

Exoactive Perfective Auxiliary Tu

In early Heian kundokubun, auxiliary tu, like its endoactive counterpart nu, is a

perfective auxiliary. In contrast, it is exoactive (Quinn 1987), most basically expressing

changes of state in more transitive and volitional predicates. Furthermore, it frequently

serves as a recent, or “high-evidence” past tense, a common extended use. In early Heian

Buddhist kundokubun texts its primary narrative function is to foreground events such as

characters’ instantaneous attainments, as we see in (32) and (51), and to explicitly mark

recently realized (i.e., completed) actions as such, as we see in (7) and (52).

(32) 初に无上正等覺の心を發し、不退轉を得ツ。 (Nakada 1958: 58)

pazime ni muzyau syautougaku no kokoro

first COP unsurpassed complete.perfect.enlightenment GEN mind

wo hatu.si putaiten wo e-tu

ACC resolving nonretrogression ACC attain-TU

‘For the first time, they resolved their minds to unsurpassed, complete, perfect

enlightenment and attained nonretrogression.’

(51) 敬（み）て仏勅を受（けたてまつり）つ。 (Kabutogi & Nakada 1979: 123)

tutusimi-te bututyoku wo uke-tatematuri-tu

respect-GER budda’s.words ACC receive-HUM-TU

‘[We] respectfully have received your words.’

(7) 尒時如意寶光耀天女、（於）大衆の中にあり、深法を說（き）たまふを聞（き）ツ。

(Kasuga 1985a: 89)

so no toki nyoipou kwau’eu tennyo

that GEN time wish-fulfilling.gem shining celestial.nymph

daisyu no naka ni ari zinbapu wo

gathering GEN in LOC be profound.truth ACC

toki-tamapu wo kiki-tu

expound-HON ACC hear-TU

210

‘Then the shining celestial nymph of the wish-fulfilling gem, being among the

crowd, heard the Buddha’s preaching of the profound truth.’

(52) 色を見、香を聞（き）て自然に飽足しつ。 (Kabutogi & Nakada 1979: 121)

siki wo mi kau wo kiki-te sizen ni

form ACC see incense ACC smell-GER natural COP

pausoku.si-tu

be.satisfied-TU

‘Having seen the form and smelled the incense, they naturally became satisfied.’

Stative Auxiliary Ari

Applied to non-stative verb participles, the stative auxiliary ari created

progressives, perfects, and resultatives. In early Heian kundokubun narrative texts, its

predicates, as perfects, provide background information (9e) that often sets the stage upon

which the action occurs in (15), (45c).

(9e) 能ク善ク調伏せり。

yoku yoku zyaubuku.se-ri

successfully well discipline-ARI

‘They have successfully properly disciplined themselves.’

(15) 曼陁羅花摩訶曼陁羅華を雨（り）て、積レルこと、（于）膝マでに至せり。 (Kasuga

1985a: 183)

mandarage makamandarage wo puri-te tumore-ru

white.lotus large.white.lotus ACC rain-GER accumulate-ARI

koto piza made ni itase-ri

NML knees to LOC bring-ARI

‘They have brought the rained accumulation of various white lotuses to reach his

knees.’

(45c) 有る人過（を）犯セリ。

211

aru pito ayamati wo wokase-ri

be person crime ACC commit-ARI

‘A certain person had committed a crime.’

Stative Auxiliary Tari

Like its older stative counterpart, auxiliary ari, stative tari is also used primarily

to express perfects that describe situations, as in (9i) and (81), and present background

information in narrative contexts (14).

(9i) 所作已に畢へたり。 (Kasuga 1985a: 1)

syosa sude.ni wope-tari

action already end-TARI

‘Actions have already ceased.’

(81) 良福田とするに堪へたり。 (Nakada 1958: 72)

ryaupukuden to suru ni tape-tari

field.of.blessedness COM do LOC endure-TARI

‘You have endured to be of the field of blessedness.’

(14) 時に長者子は高樓の上に在（り）て安隱にして（而）睡（り）たり。 (Kasuga 1985a:

183)

toki ni tyauzyasi pa kauru no upe ni ari-te

time LOC merchant.son TOP tower GEN above LOC be-GER

annon.ni.si-te nemuri-tari

be.peaceful-GER fall.asleep-TARI

‘Then the merchant’s son [Jalavāhana], being on top of a tower, has peacefully

fallen asleep.’

Externally Established Fact Modal Auxiliary Keri

212

In the kundokubun texts examined, the externally established fact modal auxiliary

keri primarily indexes mirativity, a grammatical category that marks information as new

or unexpected. While it occurs much more frequently in wabun (as discussed in section

6.1), in kundokubun it is generally limited to heading clauses that state epiphanies—

primarily facts just understood by those assembled to hear the Buddha’s sermon after

some part of it is completed, as in (85), or it caps lines spoken by the Buddha conveying

facts to a hitherto unaware interlocutor in dialogue, as in (45q) and (48).

(85) 世尊、上に說（き）つル所の如ク、菩提の正行をばすベクありケリ。 (Kasuga

1985a: 91)

seson upe ni toki-turu tokoro no gotoku bodai

Lord above LOC expound-TU place GEN like enlightenment

no syaugyau wo ba su-beku ari-keri

GEN right.deeds ACC TOP do-MODAL be-KERI

‘Lord, (I now realize the fact that) I must do the right deeds of enlightenment as

you have previously expounded.’

(45q) 无睱の傍生趣の身を受（け）たりと雖ども、而（も）袈裟を敬（ひ）て（不）惡業を造

ラズアルナリケリ。

muka no bausyauzyu no mi wo uke-tari

no.leisure GEN rebirth.as.an.animal GEN body ACC receive-TARI

to ipe-domo sikamo kesa wo uyamapi-te

COM say-although however buddhist.robes ACC respect-GER

akugyau wo tukura-zu aru nari-keri

evil.deeds ACC make-NEG have COP-KERI

‘Although they say bodies reborn as animals have no leisure to follow the Buddhist

path, the fact is that they do, however, respect the Buddhist robes and will not

cause harm to them!’

(48) 善哉、善哉。善男子。善く是の時を知りけり。 (Kabutogi & Nakada 1979: 92)

213

zen’ya zen’ya zennansi yoku ko no toki wo

excellent excellent gentlemen well this GEN time ACC

siri-keri

learn-KERI

‘How very excellent, excellent, gentlemen, (the fact) that you will learn well of

this time.’

Established Past Fact Modal Auxiliary Ki

The established past fact modal auxiliary ki is the primary framing and grounding

device of Buddhist kundokubun narration. It presents the predicate it heads as a fact

established in a past that the speaker treats as witnessed. In quotations it is used to present

a scene as a witnessed past, as in (47) and (89); and in narration it is also used to frame

scenes in their beginning, as in (33) and (45a); and at their end, as in (27) and (45o).

(47) 而れども八億の諸天、来下して法を聴、菩提心を発しき。 (Kabutogi & Nakada

1979: 100)

saredomo pyakuoku no syoten raige.si-te

however 800.million GEN many.gods come.down-GER

papu wo kiki bodaisin wo hatu.si-ki

dharma ACC hear aspiration.to.enlightenment ACC resolve-KI

‘However, eight hundred million numerous gods game down and, hearing the

dharma, resolved to become enlightened.’

(89) 諸の衆生は性欲不同なりといふ（ことを）知（り）て、種種に法を説（き）き。

(Kabutogi & Nakada 1979: 98).

moro no syuzyau pa syauyoku pudou nari to

all GEN living.beings TOP nature.desires not.same COP COM

ipu koto wo siri-te syusyu ni papu wo

say NML ACC learn-GER various COP dharma ACC

214

toki-ki

expound-KI

‘Having learned that not all living beings share the same natural desires, I

expounded the dharma in various ways.’

(33) 屬て三藏玄奘法師（といふ）者有しき。 (Nakada 1958: 1)

sitagapi-te sanzau genzyau papusi to ipu pito

follow-GER canon Xuanzang master COM call person

imasi-ki

be.HON-KI

‘Thus, there was a person called Xuanzang, master of the Buddhist canon.’

(45a) 又、善男子、昔、國王有（り）き。

mata zennansi mukasi kokuwau ari-ki

also gentlemen long.ago king is-KI

‘Also, gentlemen, long ago there was a king.’

(27) 時に諸の聲聞、及菩薩衆歡喜して佛を禮（し）て、復、本座に還しき。 (Nakada

1980: 215)

toki ni moro no syaumon oyobi bosatu syu

time LOC all GEN disciple and bodhisattva group

kwangi.si-te potoke wo rei.si-te mata ponsyo ni

be.overjoyed-GER buddha ACC bow-GER also original.place LOC

kwan.si-ki

return-KI

‘Then the group of all the disciples and bodhisattvas were overjoyed, bowed to the

Buddha, and returned to their original place.

(45o) 斯に因（り）て殺を斷ずること、王も贍部洲にもしき。

sore ni yori-te satu wo dan.zuru koto wau mo

that LOC base-GER killing ACC cease NML king also

senbusyau ni mo si-ki

Jambudvīpa LOC also do-KI

‘Based on that the king also did a cessation of killing in Jambudvīpa as well.

215

Weaving Stories with the TAM Auxiliaries

As this dissertation explains, the above tense, aspect, and modality auxiliaries

were used to weave complex and compelling stories in early Heian renditions of Sinitic

sutras. Sections 4.2, 4.3, and 6.2.2 contain visual overviews of the interweaving

perspectives that the TAM auxiliaries provide in early sutra narration. As noted in chapter

7 the perfective aspect auxiliaries nu and tu were used in narration to foreground scenes

of transformation and movement, grammatically expressing changes and realizations in

the story’s progression. The narrative function of predicates headed with the stative

aspect markers ari and tari, on the other hand, was to lay out background information in

order to establish the scene. The externally established fact auxiliary keri occurs

predominantly in quotes by characters in these stories to express their admiration,

acceptance, and subsequent internalization of the words of the Buddha. Finally, the past

established fact auxiliary ki was employed to ground the discourse in a witnessed past.

216

Appendix B. The Kundokubun Texts

In this appendix I provide an overview of extant Sinitic texts glossed in the early

Heian period, with particular detail paid to the manuscripts from which I draw data in this

study. Much of the information below is a summary of information found in Kuntengo

jiten (訓点語辞典 ‘A dictionary of glossing language’) (Yoshida et al. 2001).

To begin, of the over 4000 extant texts glossed during the Heian period (794–

1185 CE), only approximately 100 are from the early part of the period, i.e., through 900

CE. Among these, many contain only glossings that mark clause boundaries, or

punctuation, (J. kutō-ten 句読点), the ordering of lexemes (J. gojo-ten 語序点, the

predecessor to kaeri-ten 返点) and how to read and interpret individual Sinographs (J.

ongi 音義). Early Heian texts with morphosyntactic glossings (J. wokoto-ten ヲコト点) in

addition to phonological ones (J. kana-ten 仮名点) that are best suited to studying

kundokubun as a linguistic register are thus limited.

Furthermore, among early Heian Buddhist texts whose kundokubun renditions are

clear from the glossings, many are monastic regulations (J. ritsu 律, Skt. vinaya)171 and

treatises (J. ron 論, Skt. śāstra)172 that lack an outermost narrative and only occasionally

171 See Buswell & Lopez (2014: 974–975).

172 See Buswell & Lopez (2014: 784–785).

217

contain parables. Only sutras (J. kyō 経), or narratives depicting sermons by the Buddha,

are investigated in this study of the relation, in these early translations, between

grammatical marking and narrative structure. Moreover, although there were ca. 100

early Heian glossed texts generally known of at the turn of the 21st century (when the

Kuntengo jiten was published), most of these remain undocumented and understudied.

Only about a quarter of these had been sufficiently investigated to have their own entry in

the jiten (‘dictionary’), with 28 total entries.

Among these 28 texts that have been sufficiently researched, only nine are sutras

that contain narratives outside of quotations. These nine include a copy of the Aṅgulimāla

Sutra (J. Ōkutsumara kyō 央掘魔羅経) with only phonological glossings (J. ji’on字音);

a Golden Light Sutra (J. Konkōmyō saishō ōkyō 金光明最勝王経) originally stored in

the Īmuro shelter of the Mt. Hiei temple complex known as the Īmurogire (飯室切) with

fragments held at over seven art museums, which has yet to be investigated as a whole; a

Bodhisattva Maitreya Ascension to Tusita Heaven Sutra (J. Kanmiroku bosatsu jōshō

tosotsuten kyō 観弥勒菩薩上生兜率天経) that is not available to the public; a single

chapter of the Lotus Sutra (J. Myōhō renge kyō 妙法蓮華経); and two separate

fragments of the Diamond Sutra (J. Kongō hannya kyō 金剛般若経).

The remaining three sutras were investigated in this dissertation: the early-9th-

century CE complete rendering of the Golden Light Sutra glossed at Saidaiji temple; the

mid-9th-century CE rendering of the 5th and 9th scrolls of the Golden Light Sutra glossed at

Tōdaiji temple; and the late-9th-century CE rendering of eight scrolls of the Ten Wheels

218

Sutra (J. Jizō jūrin kyō 地蔵十輪経). The final text considered in this dissertation is a

complete rendering of the Immeasurable Meanings Sutra (J. Muryōgi kyō 無量義経),

which lacks an entry in the Kuntengo jiten. Although lacking a colophon, it appears to be

from the first half of the Heian period (Kabutogi & Nakada 1979). The following

subsections contain details of these texts, primarily drawn from sections above.

The Saidaiji Golden Light Sutra173

The Golden Light Sutra was likely completed in Sanskrit around the turn of the 5th

century CE. The Indian monk Dharmakṣema (J. Donmusen 曇無讖) first translated it into

Chinese for the kingdom of Northern Liang dynasty. However, his translation did not

include all the chapters and was thus incomplete. In 703 CE, the Chinese Buddhist monk

Yijing (J. Gijō 義淨), who had spent many years studying and copying sutras in India,

produced a complete Chinese translation, which is the source text for the kundokubun

rendition we examine below. This translation was quickly imported into Japan and recited

in temples throughout the kingdom by imperial decree in 741 CE to project centralized

authority and protect the state (Kasuga 1985b: 6–7; Buswell & Lopez 2014: 247–284, 877,

1028).

Saidaiji temple (西大寺) is located in Nara, Japan. Its copy of the Golden Light

Sutra is registered as a Japanese national treasure and is one of the best-preserved glossed

texts of the early Heian period. The scroll itself is a sutra copy produced in 762 CE by

173 This text is primarily discussed in sections 4.1 and 6.2.

219

Kudara no Toyomushi (百済豊虫) and glossed twice, using white ink in ca. 830 CE and

with vermilion ink in 1097 CE. The kundokubun of the earlier rendition is examined in this

dissertation.

The Tōdaiji Golden Light Sutra174

Tōdaiji temple (東大寺) is also located in Nara, Japan. Its copy of the Golden

Light Sutra lacks a colophon, and thus the exact dating is unclear, but the pattern of its

mophosyntactic glossings and the style of its phonological glossings suggest it was

glossed ca. 850 CE (Ishizuka 2001: 31).

The Ten Wheels Sutra175

The Kṣitigarbha Ten Wheels Sutra (地蔵十輪経 Jizōjūrin kyō), which I refer to as

the Ten Wheels Sutra throughout the dissertation, consists of a dialogue between the

Buddha and the bodhisattva Kṣitigarbha, Skt. ‘Earth Store’ (known as Jizō in Japan), who

rescues those reborn in one of the Buddhist hells. According to the preface of the Sinitic

rendition, it was translated into Chinese by the famous monk Xuanzang (玄奘 J. Genjō),

who was “one of the two most influential and prolific translators of Indian Buddhist texts

into Chinese” (Buswell & Lopez 2014: 1015).

174 This text is primarily discussed in section 6.2.

175 This text is primarily discussed in section 4.3.

220

The text examined in this dissertation is a kundokubun rendition glossed with white

ink at Tōdaiji temple in 883 CE on an 8th-century-CE copy of Xuanzang’s Sinitic translation,

which includes the preface detailing the importance of the sutra and the circumstances that

led to him (re-)translating it. We know the year it was glossed due to its colophon. Although

originally ten scrolls, only eight remain today. Scrolls 1, 2, 4, 8, 9, and 10 are housed in

the Tōdaiji temple library and scrolls 5 and 7 are housed in Shōsō’in (the Tōdaiji temple

repository, or treasure house), while scrolls 3 and 6 are lost.

The Immeasurable Meanings Sutra176

The Immeasurable Meanings Sutra (J. Muryōgi kyō 無量義経) is considered by

some to be the prequel to the Lotus Sutra (J. Myōhō renge kyō 妙法蓮華華経) and there

are records that attest to it being copied along with the Lotus Sutra during the Nara period

(710–794 CE) (Kabutogi & Nakada 1979: 130). Although it is claimed to have been

translated into Chinese by an Indian translator named Dharmāgatayaśas (J.

Donmakadayasha 曇摩伽陀耶舎), today it is considered by some scholars to be an

apocryphon originally composed in China (Kabutogi & Nakada 1979: 127–130; Buswell

& Lopez 2014: 1001–1002).

The Immeasurable Meanings Sutra is much shorter than the Golden Light Sutra

and Ten Wheel Sutra examined above, consisting of only a preface and three chapters. The

sutra is structured as a conversation between bodhisattva Mahāvyūha (‘Great Adornment’

176 This text is primarily discussed in section 4.5.

221

J. Daisōgon 大荘厳) and the Buddha, in which the Buddha explains how bodhisattvas can

quickly reach enlightenment and the benefits of hearing the sutra itself.

The text examined in this study was in the personal collection of the acclaimed

scholar of the Lotus Sutra Kabutogi Shōkō (兜木正亨), who acquired it in pursuit of his

religious studies research. Both he and Nakada Norio (中田祝夫) agree that it was

glossed in the first half of the Heian period, but the text itself lacks a colophon so the

exact circumstances of its glossing are unknown.