iHEmB

i

 

 

 

 

 

 

\

SITY LIBRARIES

\\ li‘iiiiiiiiiiii Hi“

iiii‘ii ii i

31293016

\\

This is to certify that the

dissertation entitled
DISCOURSE MARKERS IN JAPANESE:

CONNECTIVES, FILLERS, AND
INTERACTIONAL PARTICLES

presented by

Mieko Kimura Philips

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Linguistics

 

We M /c¢~’t‘\

Major professor

Jun L 4/036“

MSUi: an Affirmative Action/liq and Opportunity Institution 0- 12771

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIBRARY

Michigan State
Unlverslty

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOlD FINES return on or before date due.

DATE DUE

 

DATE DUE DATE NE

3: 391! '2 8 1mm.

In“ I U LUUL

W/‘f”

 

 

 

 

 

 

 

 

 

 

 

 

 

me Wm“

 

 

DISCOURSE MARKERS IN JAPANESE: CONNECT IVES, FILLERS, AND
INTERACTIONAL PARTICLES

By

Mieko Kimura Philips

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHKOSOPHY
Department of Linguistics and
Germanic, Slavic, Asian and African Languages

1998

Copyright by
Mieko Kimura Philips
1998

ABSTRACT

DISCOURSE MARKERS IN JAPANESE: CONNECTIVES, FILLERS, AND
INTERACTIONAL PARTICLES

By
Mieko Kimura Philips

This dissertation analyzes the functions of Japanese connectives, fillers, and
interactional particles as discourse markers used by native and normative speakers of
Japanese, using Schiffrin's (1987) framework. Quantitative analysis of these markers in
relation to sociolinguistic factors is also presented.

Schiffrin analyzes 11 English discourse markers using the discourse model she
proposes, consisting of five planes of talk: the ideational, exchange and action structures, the
participation framework, and the information state. She argues that with their indexical
functions, English discourse markers provide textual and participant coordinates for an
utterance, elucidating its relationship to prior and/or upcoming utterances, or to the
interlocutors.

Based upon the naturally occurring speech data, I argue that Japanese connectives,
fillers, and interactional particles operate in much the same way as English discourse
markers. Discourse markers are indispensable part of Japanese conversation with their
functions as utterance-initial, utterance-medial and utterance-final brackets of units of talk.

The connective de 'and', for example, functions as a textual coordinate, connecting
utterances which are in a sequential or cause-effect relationship. It also functions as a
marker of speaker continuation. Fillers not only function as pause-filling devices but also

display the speaker's thoughts and feelings, such as uncertainty and hesitation. Interactional

particles also reveal the speaker's feelings or attitude towards a proposition and/or the
hearer.

My data suggest that the speaker's gender and speech genre are the most significant
factor in the use of connectives. In the use of fillers, formality is the most significant factor.
Formality and speech genre are most significantly related to the use of interactional
particles.

As regards the use of Japanese discourse markers by non-native speakers, it was
found that: (1) their proficiency level and the length of formal study are the two most
significant factors in their discourse marker usage, and (2) when a marker has both
referential and interactional functions, the former are learned before the latter.

Finally, I propose that, given their importance in natural conversation, the roles and

use of discourse markers be consciously taught in Japanese language classrooms.

To all the people who supported me

ACKNOWLEDGEMENTS

There are many people both in the US. and Japan who helped and supported me at
different stages of this dissertation to whom I wish to express my appreciation.

First and foremost, I would like to express my deepest gratitude to Professor
Mutsuko Endo Hudson, the chair of my committee. From the beginning of my graduate
study, she guided me through the course work and comprehensive examinations. I am
indebted to her not only for her insightful and invaluable comments and advice on this
work but also for her encouragement, patience, and the time she spent reading earlier
manuscripts. This dissertation would not have been completed without her support. Of
course, I am solely responsible for any errors contained herein.

Next, I wish to express my gratitude to other committee members, Professors
Barbara Abbott, Susan Gass, and Dennis Preston for their valuable suggestions and
comments on this work with their respective expertise as well as their patience with me,
and to Professor Michael Lewis who served as a representative from the College of Arts
and Letters for my oral defense.

My special thanks go to Professor Steve Raudenbusch of the University of
Michigan, whose advice with statistics was indispensable.

I also wish to express my gratitude to all the people who allowed me to record and
use their speech data for this dissertation.

Finally, I wish to thank my family and friends who supported me, especially my
husband, Craig, who not only gave me moral support during the long period of time which
took me to complete this work but also spent countless hours reading earlier manuscripts

and helped me with editing.

vi

 

 

 

TABLE OF CONTENTS

List of Tables
List of Figures

CHAPTER 1
Introduction
1.1. General remarks and goals of the present study
1.2. Hypotheses
1.3. Framework of this study
1.3.1. Schiffrin's (1987) discourse model
1.3.2. Definitions of discourse markers by Schiffrin (1987)
1.3.3. Reasons for adopting Schiffrin's framework
1.4. Background to the analysis of Japanese discourse markers
1.4.1. Characteristics of spoken Japanese
1.4.2. Characteristics of the Japanese communication style
1.5. Outline of remaining chapters

CHAPTER 2
Related literature
2.1. Studies on English discourse markers
2.1.1 . Schiffrin's work on discourse markers
2.1.2. Schourup's (1982) study of English fillers as evincives
2.1.3. Discourse marker research and second language acquisition
2.2. Studies on Japanese discourse markers
2.2.1. Connectives
2.2.2. Fillers
2.2.3. Interactional particles
2.2.4. Studies on non-native use of discourse markers
2.3. Summary

CHAPTER 3
Japanese connectives as discourse markers
3.1. Preliminaries
3.2. Method
3.2.1. Data
3.2.2. Participants
3.2.3. Procedures
3.2.4. Units of talk
3.3. Quantitative results
3.4. Sociolinguistic factors and the use of connectives
3.4.1. Forrnality
3.4.2. Speakers‘ age
3.4.3. Gender
3.4.4. Speech genre
3.4.5. Summary
3.5. Qualitative analysis
3.5.1. The coordinative marker (sore)de 'and'
3.5.2. The adversative marker demo 'but'
3.5.3. The causal marker dakara 'so'
3.6. Summary of Japanese connectives

vii

CHAPTER 4
Japanese fillers as discourse markers

4.1. Preliminaries

4.2. Quantitative results

4.3. Sociolinguistic factors
4.3.1. Formality
4.3.2. Age
4.3.3. Gender
4.3.4. Speech genre

4.4. Summary

4.5. Qualitative analysis
4.5.1. Ano(o) 'well, uh': a hesitation marker
4.5.2. Nanka 'like': a marker of hesitation and uncertainty
4.5 3. M00 'really': a marker of excess
4 5.4. Yappari 'as expected': a marker of expected results
4.5.5. Ma(a) 'kind of: a marker of downplaying
4.5.6. K00 'like this': a marker of information recalling
4.5 7. Uun 'uh, well': a marker of hesitation
4.5.8. Ee(t0) ',well uh': a marker of information recalling effort
4. 5.9. Hora ',there y'know': a marker of addressee involvement
4.5.10. A ',oh' aa 'oh' and e? 0'11, huh?‘ for information management

4.6. The role of fillers 1n conversation

4.7. Summary

CHAPTER 5
Japanese interactional particle as discourse markers
5.1. Preliminaries
5.2. Quantitative analysis
5.3. Sociolinguistic factors
5.3.1. Fonnality
5.3.2. Age
5.3.3. Gender
5.3.4. Speech genre
5.3.5. Summary
5.4. Qualitative Analysis
5.4.1. Ne: a marker of agreement and cooperation
5.4.2. Yo: a marker of the speaker's emphasis
5.4.3. Y one: a marker of emphasis and cooperation
5.5. Summary

CHAPTER 6
Japanese discourse marker usage by non-native speakers
6.1. Preliminaries
6.2. Method
6.2.1. Data
6.2.2. Participants
6.2.3. Procedures
6.3. Connectives
6.4. Fillers
6.5. Interactional particles
6.6. Pedagogical implications of the present study

viii

CHAPTER 7
Conclusion
7.1. Findings of the present study
7.2. The roles of discourse markers in Japanese conversation
7.3. Discourse markers and expected norms of Japanese society
7.4. Limitations of the present study and suggestions for future study

Appendix 1. The ACTFL Oral proficiency guidelines
Appendix 2. Questionnaire for native Japanese
Appendix 3. Questionnaire for non-native subjects
Appendix 4. List of native English subjects

Bibliography

265
265
268
27 l
273

277
278
279
280

282

Table 2.1
Table 2.2
Table 2.3
Table 3.1
Table 3.2
Table 3.3
Table 3.4
Table 3.5
Table 3.6
Table 3.7
Table 3.8
Table 3.9
Table 3.10

Table 3.11

Table 3.12
Table 3.13
Table 3.14
Table 3.15
Table 4.1
Table 4.2
Table 4.3
Table 4.4
Table 4.5

LIST OF TABLES

Planes of talk on which markers function

Kamio's territory of information

Maynard's (1993) conclusion on the use of ne and yo

List of native Japanese participants

Connectives in the native Japanese data

Percentage of connectives in various studies

The tokens and rates of connectives by formality

The tokens and rates of connectives by age

The tokens and rates of connectives by the speaker's gender
The rates of connectives per 1000 words by the addressee's gender
The tokens and rates of connectives by speech genre
Functions of (sore )de on various planes of discourse

The number of tokens, percentages, rates of de and sorede
by gender

The number of tokens, percentages, and rates/1000 words
of de and sorede by speech genre

Functions of (sore )de and the discourse planes on which they operate
Fillers and the particle ne following (sore )de, demo and dakara
Functions of demo and the discourse planes on which they operate
Functions of dakara on the five planes of talk

Summary of fillers in the native Japanese data

The tokens and rates of fillers per 1,000 words by formality

The tokens and rates of fillch per 1,000 words by age

The tokens and rates of fillers per 1,000 words by speakers' gender

18
3o
31
43
47
48
50
51
53
55
57
6O

76

77
82
95
96
111
117
118
120
121

The tokens and rates of fillers per 1,000 words by addressees' gender 122

Table 4.6
Table 4.7
Table 4.8
Table 4.9
Table 5.1
Table 5.2
Table 5.3

Table 5.4
Table 5.5
Table 5.6
Table 5.7
Table 5.8
Table 5.9

Table 5.10
Table 5.11

Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 6.7
Table 6.8
Table 6.9

Table 6.10
Table 6.11

The tokens and rates of fillers per 1,000 words by speech genre
Number of tokens and % of various functions of ano(o)

Positions where ano(o) occurred in utterances

Fillers as contextual coordinates

List of interactional particles by formality and the speaker's gender
Summary of Japanese interactional particles in the collected data

The tokens and rates/1000 words of interactional particles
by formality

The tokens and rates/1000 words of interactional particles by age
The tokens and rates of interactional particles by speakers' gender
The rates of interactional particles by the addressee's gender

The rates of interactional particles by speech genre

Previous studies on ne

Summary of previous studies on yo

Summary of previous studies on yone

Functions of yo and ne in yone

List of non-native participants

The tokens and percentages of connectives by proficiency level
The rates of connectives per 1,000 words by proficiency level
The P-values of three factors influencing the use of connectives
Distribution of de and sorede by proficiency level

Distribution of functions of sorede by proficiency level
Distribution of functions of de by proficiency level

Use of demo by proficiency level

The tokens and percentages of fillers by proficiency level

The rates of fillers per 1,000 words by proficiency level

The P-values of three factors influencing the use of fillers

xi

124
138
139
177
179
181

183
185
186
187
190
192
203
217
223
228
23 1
232
234
236
238
239
242
245
246
247

Table 6.12a
Table 6.12b
Table 6.13a
Table 6.13b
Table 6.14

Table 6.15
Table 6.16
Table 7.1

The tokens and % of interactional particles by proficiency level
The breakdown of ne by position and proficiency level

The rates of interactional pa1ticles/1000 words by proficiency level
The breakdown of the rate of ne by position and proficiency level

The P—values of the factors which influenced the use of interactional
particles

The results of the questionnaires (fillers)
The results of the questionnaires (interactional pa1ticles)

Positions of connectives, fillers and interactional particles
and the units they coordinate

xii

254
255
255
256

257
261
262

270

Figure 1.1
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 5.1
Figure 5.2
Figure 7.1

LIST OF FIGURES

Positions of connectives, fillers and final particles
Uyeno's perfonnative analysis of interactional particles
Ochs' model of indexicality

Indexical relations of yo (Cook 1988: 129)
Indexical relations of ne (Cook 1988: 155)
Summary of the functions of (sore )de

Functions of but (Schiffrin 1987)

Functions of demo 'but' (Onodera 1993)
Functions of dakara 'so' (Maynard 1993)
General and specific functions of 0110(0)

General and specific functions of nanka

General and specific functions of ma(a)

General and specific functions of ee( to )
Functions of ne

Functions of yo

Positions of connectives, fillers, and interactional particles,
and the units they coordinate

xiii

 

28
32
33
34
82
83
84
97
141
149
158
169
202
216

270

 

 

TRANSCRIPTION METHODS AND CONVENTIONS

1. Japanese transliteration is given, using the Hepburn style, except that fit is spelled hu.
For double consonants such as cho, chu, and cho, t is added instead of using two cs.
e.g. hutatsu 'two', suki n natchatte 'came to love (Audrey Hepburn)’,

2. Data, examples, tables, and figures are numbered for each chapter.

3. When proper nouns appear in the data, pseudonyms are used except when it is crucial
for discussion.

4. Paralinguistic and other interactional symbols are as follows:

/

?

i1

recognizable pause

falling intonation with noticeable pause

continuing ("list") intonation, with a slight rise or fall, followed by a
short pause

rising intonation

omission

linked or continuing utterances without overlap

speech overlap

inaudible; transcription doubt

( )
((LAUGH)) indicates various types of laughs

CAPITAL
!

emphasis
animated tone

5. When transcribing, not all morphemes are separated. Only those which are important to

the present study are separated with glosses given.

6. Following abbreviations are used:

BE: the copula, be

CAUS: causative morpheme
DO. direct object particle
EMPH: emphatic particle

GEN: genitive particle

IP: interactional particle
10: indirect object

NEG: negation

NOM: nominalizer

PASS: passive morpheme
POSS: possessive marker

Q: question particle

QT: quotative marker

SUB: subject particle

TAG: tag question morpheme
TOP: topic particle

VOL: volitional form of verbs

xiv

 

 

 

 

CHAPTER 1

Introduction

1.1. General remarks and goals of the present study

The present study analyzes the elements in Japanese conversation commonly
known as 'discourse markers.‘ Discourse markers are defined by Schiffrin (1987 :326) as
"contextual coordinates, which index an utterance to the local contexts in which utterances
are produced and in which they are to be interpreted." In other words, discourse markers
reveal utterance-utterance relationships, speaker-utterance relationships, and speaker-hearer
relationships in a given discourse context.

The English and, for example, signals that the utterances before and after it are in a
coordinate and/or sequential relationship, and also signals the speaker's continued turn.
The English but signals that the prior and upcoming portions of an utterance are in a
contrastive relationship. Likewise, the Japanese dakara 'so', for example, signals that the
prior and upcoming utterances are in a causal relationship. The Japanese filler ano(o) 'uh,
well' can signal the speaker's hesitation as well as serve as an attention-getter. As such,
some discourse markers contribute to discourse coherence and others function to reveal
speaker-bearer relationships.

The need for further research on discourse markers is apparent upon examining any
transcript of ordinary conversation. Preferred discourse markers and their frequency may
vary from speaker to speaker, but there is almost no naturally-occurring conversation—
whether Japanese or English, formal or informal— in which no discourse markers are used.
Although there have been a number of studies on Japanese discourse markers in recent
years, they have been predominantly isolated works on a limited number of connectives
and final particles, such as yo and ne. Moreover, research on the use of discourse markers
by non-native speakers is scarce. Thus, an extensive study of Japanese discourse markers

is yet to be undertaken. It is hoped that the present study will make a contribution, albeit a

small one, both in the fields of Japanese discourse analysis and Japanese language
pedagogy.

The present study deals with three categories discourse markers in Japanese: (1)
logical connectives, such as sorede 'and' and demo 'but'; (2) fillers and interjections, such
as ee( to ) 'uh', nanka 'like' and 0 'oh'; and (3) final particles, such as ne 'right?’ and yo 'I
tell you'. Connectives are positioned primarily at the beginning of an utterance, and thus
function as "initial brackets," connecting the prior utterance with what follows. Fillers
primarily occur utterance-medially, and function as the "utterance-medial brackets,"
separating the prior and upcoming portions within an utterance. Lastly, final particles
primarily occur utterance-finally, and thus function as "utterance-final brackets,"
expressing the speaker's feelings and orientation to the addressee and/or the utterance to
which they are attached. Thus, schematically, the positions of connectives, fillers, and

final particles can be described as follows:

 

 

[utteranceh [connective-portion of utterance filler portion of utterance—particleh

 

 

Figure 1.1 Positions of connectives. fillers and final particles

Example (1) from my data illustrates the occurrences of conneetives, fillers, and final

particles.

(1) M]; a. Eigo benkyoo shiteru bokura da tte/ tsukawanakute sumimasu kara nee.
English study do-PROG we BE QT use-NEG get by because [P

b. Demo sono bun puresshaa ga/ koo tsuyoku natte kimasu kedo ne.
but that much pressure-SUB in this way strong become IP

WA: c. Be, 500 desu nee.
yes so BE IP

MI: a. Even we, who are studying English, can get by without using it, y'know?
b. But, for that much, the pressure (on us as English teachers), like,

increases, y'know?
WA: c. Yes, you're right, I agree.

The connective demo 'but' in (b) functions as a bracket utterance-initially, signalling that
(a) and (b) are in a contrastive relationship. The filler koo 'like this' in (b) separates the
portion before koo and after it—the former is what is being talked about, and the latter
provides the description of it, often based on the speaker's personal knowledge or
experience. The final particle ne( e) in (a), (b) and (c) signals that the speaker and addressee
know each other well and/or they are relatively close in their social status.

Connectives' functions as discourse markers seem relatively well-established. In
contrast, research on so-called fillers in Japanese has been scarce. Fillers are, in
Maynard's (1989az30-31) terms, a broad range of utterances that do not carry identifiable
or relevant propositional meanings, and are used to indicate the speaker's hesitancy or
uncertainty about his or her message content. English expressions which are referred to as
fillers include well, like, uh, you know and if you see what I mean (Brown and Yule
1983: 17). Fillers are characteristic of unplanned face-to-face interaction, and generally do
not occur in writing (Brown and Yule pp. 17-18). They are often described as devices
used to "fill in pauses, when a speaker feels threatened by his interlocutor wanting to take a
turn, using the first word that comes to mind rather than hunting for the motjuste" (Brown
and Yule p. 18).

Although some fillers are undoubtedly used as floor-holding or pause-filling
devices while searching for expressions, many fillers have important roles in conversation.
Some researchers on English fillers have considered them as an important part of
discourse. Labov and Fanshel (1977:156), for example, state that well as a discourse
marker refers backwards to some shared topic or knowledge. Stubbs (1983:70) also
recognizes the importance of well, now, and anyway as boundary markers, and Schourup
(1982) argues for the importance of expressions such as well, like, and you know as

devices to reveal interlocutors' thought processes.

Likewise, Japanese expressions which are typically considered to be fillers have
specific functions. For example, utterance-initial ano(o) 'uh, well' in Japanese as is in
Anoo chotto ohanashi ga 'uh, I would like to speak to you,’ is used to show the speaker's
hesitation before what he or she wants to say is actually expressed. Ano in Ano ne, chotto
kiite 'Hey, listen,’ is used as an attention- getting device rather than a floor—holding device.

Furthermore, each filler is restricted in terms of the situations where they can occur.

If all fillers are used only as floor-holding or pause-filling devices while searching for the

 

"right expression," any filler should suffice in any situation. However, that is not the case.
For example, ano in ano chotto ohanasi go is not replaceable by nanka 'like' or uun 'uh,
well'. This means that the expressions which are categorized as fillers have more

meaningful roles than merely filling in pauses or holding the floor.

 

Researchers, except for Labov and Fanshel (1977) and Schiffrin (1987), do not
explicitly refer to fillers as discourse markers. Many fillers, however, display typical
discourse marker characteristics proposed by Schiffrin (1987): syntactic detachability,
bracketing of units of talk, and occupying the utterance-initial position. They thus satisfy
Schiffrin's operational definition of discourse markers (see Section 1.3.3 below).
Moreover, in many instances fillers are used to show a speaker's orientation towards a
proposition, the addressee, or the situation in which the conversation is taking place.
Fillers thus can serve as contextual coordinates, satisfying the theoretical definition of
discourse markers provided by Schiffrin (1987). Therefore, in this study I treat fillers as
discourse markers.

Likewise, final particles have not been treated as discourse markers by many
researchers, except for Onodera (1993). Based on the theoretical definition given by
Schiffrin (1987:327) of discourse markers as textual and participant coordinates, final
particles are also included in this study as discourse markers. Final particles are used in

conversation to enhance "involvement" and "rapport" among interlocutors (Maynard

1989az30). The use of final particles reveals the speaker’s feelings and orientation towards

 

a proposition and/or the addressee. Example (2), taken from Makino and Tsutsui

(1986:45-46), illustrates this point.

(2) a. Yamada-san wa sensei desu.
Yamada Mr. TOP teacher BE

'Mr. Yamada is a teacher.’

b. Yamada-san wa sensei desu ne.
Yamada Mr. TOP teacher BE FP

'Mr. Yamada is a teacher, isn't he?‘

c. Watashi wa shirimasen yo.
I TOP know-NEG FP

'I don't know, I tell you.‘

d. Atashi ureshii wa.
I happy FP

'I'm happy!‘

e. Ore wa makenai 20.
I TOP lose-NEG FP

'I won't lose!‘

(2a) is a declarative sentence, which simply states a fact. Ne in (b) changes (a) into a tag
question, requesting confirmation. Yo in (c) is used to assert something emphatically to an
addressee whose social status is approximately equal to or lower than the speaker. Thus,
the use of yo reveals the speaker's judgement about the speaker—addressee relationship.
Wain (d) is used when the speaker wishes to express "femininity" in her utterance, and
thus reveals the speaker's desire to present certain image of herself in the conversational
situation. Finally, 20 in (e) signals that the speaker is a male and is asserting something
emphatically to an addressee who is of equal or lower status. Thus, final particles in (b)-
(6) above all reveal important information regarding the speaker-proposition relationship

(e.g. strong feelings expressed with yo and 20), cognitive information (e. g. the

 

confirmation particle ne), and the speaker-addressee relationship (6. g. yo and 20, which are
used with an addressee of equal or lower status). Since the main functions of final particles
concern interactional aspects of conversation, and because the particles ne( e) and so can
occur utterance-medially, they will be referred to as "interactional particles" hereafter,
following Maynard (1993). I will analyze these three categories of items, connectives,
fillers, and interactional particles, identify their main functions, and describe their
contributions to discourse coherence in Japanese conversation.

This dissertation has three goals: (1) to analyze the functions of Japanese discourse
markers as used by native speakers, and the variations in their use in relation to
sociolinguistic factors, such as gender and speech genre (e. g. conversation, interview, and
narratives); (2) to analyze the use of discourse markers by non-native speakers of Japanese
at the four proficiency levels defined by the ACTFL (the American Council on Teaching of
Foreign Languages): novice, intermediate, advanced and superior levels (see Appendix 1
for ACTFL Oral Proficiency Guidelines); and (3) to consider some pedagogical
implications of the findings of the present study in Japanese instructions with respect to

Japanese discourse markers.

1.2. Hypotheses

Based on the three goals stated above and prior studies on Japanese and English
discourse markers, the following hypotheses were formed. Hypothesis 1 relates to
connectives, Hypothesis 2 to fillers, Hypothesis 3 to interactional particles, and

Hypotheses 4 and 5 to non—native speakers' use of Japanese discourse markers.

Hypothesis 1

There will be stylistic and genre—related variations in the use of connectives. They will be
used more often in narratives than in conversations, and in formal speech than in informal
speech.

 

 

The reason for this hypothesis is that in narratives the speaker holds the floor longer than in
conversation, connecting events, explanations, and the like, which is likely to increase the
use of connectives. Also, the more formal the speech style is, more care may be taken to
speak in a coherent manner, connecting utterances, which also may increase the use of

connectives.

Hypothesis 2

The use of fillers will also be influenced by several sociolinguistic factors:
Fillers will be used more often by women than by men, by older speakers than by younger
speakers, and in formal—style speech than in informal-style speech since speakers are more

concerned with producing "appropriate speech" when speaking formally than when
speaking informally.

Hypothesis 2 is based on the following: (1) Maynard (1989:31) states that fillers
and interactional particles contribute to the effect of "social packaging," which refers to a
socially motivated act of constructing the content of the utterance in such a way as to
achieve "maximum agreeableness" to the recipient. Given the characterization of "social
packaging" by Maynard, I speculate that more fillers will be used by women and older
speakers, since they are more concerned with maintaining a harmonious relationship with
their interlocutors than younger speakers speaking informally. Furthermore, the
relationship between people conversing in formal-style Japanese tends to be less familiar
than that between persons conversing in infonnal-style Japanese, and, therefore, fillers will
be used more often in formal-style conversations than those in informal-style Japanese, so
as not to make utterances overly assertive.
Ide (1982:377) states that women and the older generation tend to use more
honorifics, which make their speech more polite, than men and the younger generation. If
fillers contribute to "social packaging," as Maynard states, they, like honorifics, may be

used more often by women and older participants than men and younger participants.

 

Hypothesis 3

Interactional particles will be used more often by women than by men; and more frequently
in informal-style speech than in fonnal-style speech.

Hypothesis 3 is based on the following: (1) Takahara (cited in Shibamoto 1985263)
states that interactional particles are used more often by female speakers than by male
speakers; and (2) Uyeno (197 1 :50) states that the closer and less formal the relationship a
speaker and addressee maintain, the more sentence particles are allowed in their
conversation. Maynard (1989az28) also states that Japanese interactional particles
contribute to encouraging rapport between interlocutors and achieving a closer monitoring
of each other's feelings.

Furthermore, since the speaker will express his/her feelings more freely when
conversing in informal-style than in formal-style, and since many interactional particles are
used to express one's feelings towards an addressee or proposition, more interactional

particles are expected be used in informal-style conversations than in formal—style

conversations.

Hypothesis 4

Non-native speakers at higher proficiency level will use more connectives, fillers and
interactional particles both in number and variety than those at lower proficiency levels.

Hypothesis 5

When a marker has both textual and interactional functions, the textual functions of
discourse markers are learned before interpersonal functions.

Hypothesis 4 is based on my experience teaching Japanese as a foreign language, and also
on the oral proficiency guidelines by the ACTFL for novice, intermediate, advanced and

superior levels. Hypothesis 5 is based on my teaching experience and my earlier pilot

Philips 1995).
in what follows, I review Schiffrin's discourse model, operational and theoretical

ons of discourse markers, and the reasons why I adopt her framework in the present

Framework of this study
Schiffrin's (1987) discourse model

Schiffrin (1987:24-29) proposes a discourse model consisting of five components
nes of talk": (1) the ideational structure; (2) the action structure; (3) the exchange
re; (4) the participation framework; and (5) the information state (1987125). The
nal structure deals with propositions themselves: their cohesive relations, topic
1s, and functional relations. The exchange structure deals with turn—taking, and the
structure deals with speech acts, such as requesting and warning. The participation
york deals with speaker-utterance relationships as well as speaker-hearer

1ships. The information state deals with the organization and management of

:dge (what the speaker and the hearer know) and meta—knowledge (what speakers
arers know or assume about their own and each other's knowledge).

Schiffrin (1987:316-317) states that many discourse markers function on more than
me of talk simultaneously, though with one primary function. For example, the

y function of the English connective and is in the ideational structure, connecting
itional units. Its secondary function is in the exchange structure, signalling the

r's continued turn. The primary function of the interjection oh is in the information
s it relates to the speaker's information management. At the same time, it may

in in the action structure if it is used to request clarification. By positing the

ms of discourse markers on different planes of talk simultaneously, Schiffrin's

can account for the multi-functionality of discourse markers.

 

 

1.3.2. Definitions of discourse markers by Schiffrin (1987 )

Connectives, fillers, and interactional particles in the present study will be analyzed
following Schiffrin's discourse model and her framework, in which discourse markers are
defined as "contextual coordinates" (p. 326). More precisely, discourse markers have two
indexical functions. The first is their function as "textual coordinates of talk," indexing an
utterance to prior and/or upcoming text, in other words, markers reveal which text an
utterance is focused on. Discourse markers index the utterance which contains them to
whatever text precedes them (proximal) and/or whatever text is to follow (distal) (p. 323).
For example, the causal connective so indexes an utterance both to prior and upcoming
texts, since it relates prior text (=causes) with upcoming text (=results). I mean focuses on
prior text, since what follows it is a restatement of what has already been mentioned prior
to I mean.

The second indexical function of discourse markers is to provide "participant
coordinates"; a marker shows whether an utterance is focused on the speaker (proximal) or
the hearer (distal) or both. For example, and focuses on the speaker, since it signals the
speaker's continued turn. Y’know focuses both on the speaker and the heater, since it is
used for speaker-hearer interaction, such as soliciting confirmation. Through these
indexical functions, Schiffrin (1987:315) states, discourse markers clarify the relationships
between utterances, the speaker and an utterance, and the speaker and the addressee, and
thus contribute to overall discourse coherence.

In sum, Schiffrin (p. 318) proposes that discourse markers "select a meaning
relation from whatever potential meanings are provided through the content of talk, and
they display that relation." In other words, when there are several possible choices in the
relations between utterances and/or an utterance and the interlocutors, markers make it clear

which meaning was selected by the speaker. (3) below is Schiffrin's example.

10

 

(3) a. Sue dislikes all linguists.
b. I like her.

Schiffrin (p. 318) states that the choice of the marker between (3a) and (3b) is determined
by several factors, such as the identity of the speaker, and the hearer's background beliefs.
For example, she states that a linguist (or one who likes linguists) would choose a
contrastive relation between (3a) and (3b) with the connective but: Sue dislikes all the
linguists, but] like her. However, a person who dislikes linguists might choose a
resultative relation with the causal marker so: Sue dislikes all linguists, so I like her. Thus,
Schiffrin (p. 318) contends that "although a marker may be able theoretically to select any
number of implicit and potential relationships, in actuality, that relationship is already fairly

constrained, such that the marker acts more to display the relationship (than to create it)."

1.3.3. Reasons for adopting Schiffrin's framework
Schiffrin's framework is a synthesis of various approaches previously employed by
other researchers for the analyses of discourse markers (e. g. Halliday & Hasan's (1976)
analysis of markers in terms of text cohesion; Sacks, Schegloff, and Jefferson's (1978)
analysis in relation to turn taking; and Owen's (1981) analysis of conversational units).
What is unique in Schiffrin's approach is her discourse model, in which discourse consists
of the aforementioned five planes of talk. It has advantages over other analyses of
discourse markers for the following reasons: (1) by employing a multi-layered discourse
model in which various elements of talk, such as speech acts, text cohesion, and turn
taking, are treated as components of discourse, it can offer a uniform account of
syntactically diverse discourse markers; 6. g. connectives, adverbs, and interjections; (2) by
regarding discourse markers as functioning on various planes of talk, it can account for the
plurifunctionality of a given discourse marker in various contexts; e. g. the use of and for

text cohesion and as a marker of a speaker’s continued turn at the same time; and (3) by

11

differentiating the primary plane from the secondary planes, Schiffrin’s analysis can
account for the relationship between the a market's primary function from its secondary
functions. Lastly, if we include a broad range of items, such as connectives, interjections,
and adverbs, as discourse markers and try to provide a uniform analysis applicable to all
these items, rather than analyzing various markers from one discrete perspective, such as
text cohesion, turn taking or speech acts, we need a theory which encompasses all such
approaches to discourse marker analysis. Schiffrin's (1987) analysis of discourse markers
does exactly that

Another contribution made by Schiffrin to the understanding of discourse markers
is her proposal of the specific conditions which qualify expressions as discourse markers

(19871328). She lists the following conditions:

(1) syntactic detachability,
(2) being commonly used in the utterance-initial position,

(3) having a range of prosodic contours, e. g tonic stress and phonological
reduction, and

(4) operating on multiple planes of discourse.

By delineating these specific conditions, we may be able to identify expressions which
function in one component of talk becoming discourse markers in another, and also be able
to analyze discourse markers in typologically varied languages.
Given these advantages which Schiffrin's approach offers, I adopt it as the
framework for the present study of Japanese discourse markers. I also employ both
qualitative and quantitative methods in my analysis, since they make complimentary

contributions to the analysis of Japanese discourse markers.

12

1.4. Background to the analysis of Japanese discourse markers
1.4.1. Characteristics of spoken Japanese

Several prominent features characterize spoken Japanese. Clancy (1982:61-76)
states that spoken Japanese is characterized by the use of the plain form of predicates,
frequent employment of interactional particles, failing to exhibit canonical verb-final order,
and considerable fragmentation and involvement. She states that a syntactic clause is
frequently broken down into a number of smaller units, which are preceded by a pause
and/or hesitations, have distinct intonation contours, and often end with heavy stress and
higher pitch on the final syllable of the last word, or with a particle such as ne or so.
According to Clancy’s study (1982:73), 67.4% of all intonation groups are shorter than a
syntactically complete clause.

Maynard (1989az23-40) lists the following as major characteristics of spoken
Japanese: fragmentation, frequent use of interactional particles, use of fillers, the
occurrence of ellipsis and postposing, and use of the plain form of verbs and adjectives.

Thus, fillers and interactional particles are both characteristic of spoken Japanese.

1.4.2. Characteristics of the Japanese communication style

One of the characteristics the Japanese style of communication which has often been
discussed by researchers is its "indirectness, ambiguity, vagueness, and impreciseness"
(e.g. Singer 1973, Christopher 1984, and Bamlund 1989). Christopher (1984:38) states
that the Japanese shun explicit, carefully reasoned statements in favor of indirect and
ambiguous ones basically designed not to communicate ideas but to feel out the other
person's mood and attitude, and that the Japanese dread provoking direct confrontation.
Bamlund (1989:39) states that homogeneity and harmony in Japanese society cultivate skill
in the use of ambiguity, circumlocution, euphemism and silence to blunt incipient disputes,
which reflects the societal norm of harmony, group-orientedness, conformity, and

cooperation.

l3

 

 

Clancy (1986) presents a similar view of Japanese communication. She (19861214)
characterizes Japanese conversation as unexplicit, indirect, and potentially ambiguous. She
states (19861216) that an important goal of socialization in Japan is to promote the
unanimity of feeling that will support the norms of verbal agreement and empathy, and that
the Japanese style of communication can work only in a rather homogeneous society in
which people actually can anticipate each other’s needs, wants, and reactions. She (p. 217)
further states that the burden of successful communication rests on the listener’s
willingness to cooperate, empathize, and intuit what the speaker has in mind.

Clancy (1986:235-240) also states that Japanese children go through what she calls
"empathy and conformity training," in which Japanese mothers actively engage themselves
in teaching their children Japanese norms for speech and behavior, both through subtle
pressure and explicit instruction. The actual items Japanese mothers teach their children
include polite formulas, socially appropriate speech, and context-appropriate tum-taking.
Clancy (p. 240) states that one of the results of this kind of empathy and conformity
training is the wish to avoid conflict, which makes it difficult for Japanese to risk angering

or hurting the feelings of addressees.

Though there is some truth to it, this kind of view seems too monolithic. The
Japanese do not use "vague" and "indirect" expressions in every situation in life. Rather, I
contend, Japanese use indirectness and vagueness as part of communication strategies
where it is socially expected, e. g. to one's superior, but they express their feelings more
straightforwardly where it is allowed, e. g. between family members and close friends.
Shibatani (19901389) states that it is Japanese cultural patterns that are responsible for what
may appear to outsiders to be illogical or vague usage.

Ide (19821382) states that women use more honorifics, and the main reason for
doing so is because women tend to be concerned with how they appear rather than with
what they actually are. In other words, women use more honorifics than men as part of

their communication strategy. Like honorifics, fillers and final particles are used as part of

14

a communication strategy in situations where indirectness and vagueness are expected. The
interactional particle ne often softens the tone of an utterance. Fillers like ma(a) 'kind of
and nanka’ 'like' give an utterance a sense of vagueness, contributing to what Maynard
(1989a131) calls "social packaging." Thus, fillers and final particles contribute to creating a
cooperative and harmonious conversational environment.

Another characteristic of the Japanese communication style is frequent use of back
channelling expressions. Maynard (198931150-177) discusses the difference in back
channelling between American and Japanese college students. Back-channel utterances,
such as 'mmhm' and 'uh huh' are used by a heater to ratify the continuing speaker's right
to hold the floor rather than attempting to take the floor. Based on casual conversation by
40 pairs of friends (20 American pairs and 20 Japanese pairs), Maynard (pp. 176-177)
reports that back-channel responses in the Japanese conversations occurred twice as often
as with the American pairs. She also states that head movement as a back-channel
expression was used three times as often among Japanese conversants in her study than
among American conversants. She (p. 169) states further that head movements and back-
channel expressions function as a signal of attention and understanding, and convey active
auditor participation, enhancing the speaker's and hearer's involvement in the conversation.
Finally, she declares that frequent and continuous back channelling is the norm within the
Japanese speech community.

One communication strategy which elicits frequent back-channel expressions is the
use of interactional particles. These particles also display important information regarding
the conversants' relative social status, the "territory of information," and the attitudes of the
speaker towards the heater and/or proposition. Maynard (1989a: 174) states that
interactional particles such as ne, yo and so provide a convenient opportunity for back-
channel expressions and head movement for the listener. Martin (19751914) characterizes
interactional particles as devices to "impart some additional hint of the speaker’s attitude

toward what he is saying—doubt, conviction, caution, inquiry, confirmation or request for

15

confirmation, and recollection." Thus, in Japanese conversations, fillers and final particles
play an important role as contextual and participant coordinates, softening the tone and

enhancing cooperative atmosphere.

1.5. Outline of remaining chapters

The organization of remaining chapters is as follows: relevant literature is reviewed
in Chapter 2. Chapters 3, 4, and 5 discuss Japanese discourse marker use by native
Japanese speakers: connectives (Chapter 3), fillers (Chapter 4) and interactional particles
(Chapter 5). Chapter 6 discusses the use of Japanese discourse markers by non-native

speakers. Finally, conclusions are presented in Chapter 7.

16

 

 

CHAPTER 2

Related Literature

2.1. Studies on English discourse markers

This section reviews two analyses of English discourse markers which are relevant
to my analysis of Japanese discourse markers: (1) Schiffrin's analysis of 11 discourse
markers in Section 2.1.1; and (2) Schourup' analysis of fillers in Section 2.1.2. Discourse

marker research in the field of second language acquisition is reviewed in Section 2.1.3..

2.1.1. Schiffrin's work on discourse markers

Schiffrin (1987) analyzes 11 English discourse markers as contextual coordinates
which operate on different planes of talk. Since an outline of her work was presented in
Chapter 1, in what follows we will review the functions of two discourse markers, and and
but, in order to see how Schiffrin analyzes English discourse markers.

According to Schiffrin, the connective and is a marker of coordination and
continuation (p.152). Its main function is on the ideational structure plane. It coordinates
idea units both at local and global levels (in the ideational structures); and signals the
speaker's continuation (in the exchange structures).

But is characterized by Schiffrin as a marker of contrast. She lists three kinds of
contrast: (1) referential contrast; (2) functional contrast; and (3) contrastive actions.
Referential contrast refer to contrasting ideas, and thus but Operates in the ideational
structure when used as a marker of referential contrast. Functional contrast refers to
contrasting functionally-different portions of discourse, and operates in the exchange
structure. Contrasting the main portion of an utterance and side-sequencing with but is an

example of functionally different contrast But as a marker of contrastive actions operates
in the action structures. Examples are: (1) making one's points after a speaker has been

interrupted, challenged or misunderstood; and (2) claiming the floor in the middle of

17

 

 

someone else's turn. Thus, Schiffrin presents an integrated view of discourse markers
from ideational and pragmatic perspectives. Table 2.1 shows the planes of talk described
above on which each discourse marker operates, as presented by Schiffrin (p. 316). The

primary plane for each discourse marker is indicated with an asterisk.

Table 2.1 Planes of talk on which markers function

 

 

fl =
Information Participation Ideational Action Exchange
state framework structure structure structure
*oh oh oh
well *well well well well
*and and and
*but but but
*or or
so so * so so so
because *because because
now *now
then *then then
I mean *I mean I mean
*y'know y'know y'know y’know

 

2.1.2. Schourup's (1982) study of English fillers as evincives
We now turn to Schourup's analysis on fillers. The review of his work is included
here for two reasons: (1) his analysis of English discourse markers as "evincives"
motivated me to examine Japanese fillers; and (2) his theory of discourse markers as
"evincives" led me to form a hypothesis about the functions of Japanese fillers and test my
hypothesis in the form of a questionnaire filled out by participants after data collection.
Schourup (1982) analyzes the English discourse particles like, well, and y'know,
and proposes a theoretical framework to account for their use. They are very much like
Japanese fillers such as ano(o) 'uh; well' and nanka 'like'. He (p. 14) states that fillers
reflect the speaker's present internal state, and argues that they are "evincives." Evincives
are, according to him (p. 15), items which enable speakers to express what they have in

mind at a particular point in a conversation without fully disclosing their thoughts, and thus

18

 

function as an interface between the speaker's thoughts and his or her actual utterances.

His definition of evincives is as follows (p. 14):

Evincive: a linguistic item that indicates that at the moment at which it is said the
speaker is engaged in, or has just then been engaged in, thinking: the evincive item
indicates that this thinking is now occurring or has just now occurred but does not
completely specify its content.

In order to analyze discourse markers as evincives, he (p. 5) proposes a tripartite
model of conversation: (l) the speaker’s private world (i.e. the speaker’s thoughts); (2) the
shared world (i.e. talk and other behavior on display and shared both by the speaker and
the addressee), and (3) the other world (i.e. the other interlocutors' thoughts, invisible to
the speaker).

Schourup's interpretation of discourse markers as reflections of the speaker's
internal state is an attractive one. When a speaker, for various reasons, does not wish to
express his or her entire thoughts explicitly, discourse markers play an important role in
bridging between the speaker’s thoughts and his or her actual utterance, revealing the
speaker's feelings and orientation toward the proposition or to the addressee.

Schourup (198217) states that many of these discourse markers become closely
associated with particular discourse situations through routine usage. The more an item is
used routinely, the more it is apt to lose contact with its literal meaning, though that
meaning may never be completely obliterated. This process is referred to by Schourup as
"routinization."

In routinized expressions, the literal meaning may be overshadowed or obscured by
the conventionalized use of the item. Schourup (p. 7) lists well, good-bye, take care and
see you as examples of routinized expressions. He states that the fact that discourse
markers maintain both the routinized meaning and the somewhat obscured original meaning

might account for the multiple functions of discourse markers.

19

Let us look at Schourup's analysis of like. He starts with the usages of the word
like in Webster’s Third New International Dictionary. Like is used as a transitive verb (to
like something), a noun (e. g. likes and dislikes), an adjective (=likely), a preposition
(=similar to), an adverb (=nearly, rather) and as a conjunction (=as if). Then he lists
examples of like from his conversation data which do not fall in any categories listed
above. He (p. 30) states that the meanings of like in these cases are reminiscent of, though
not exactly equivalent to, the meaning listed in a dictionary, and he argues that the original
meanings of like are accessible to a varying degree but not prominent (p. 42).

One of these examples is like used preceding descriptions involving numbers, e. g.
like one more week and like five till eight, in which case like is used in the sense of
"approximately; nearly." Another example he gives is like used immediately preceding a
direct quotation as "Both sides of the street can hear her yelling at us and she’s like 'come
in here and have a beer' y'know?" In this case also, like is used in the sense of "(This is)
approximately (what she said or what I heard her say)."

Schourup's next step is to determine the extent of literalness of the various
routinized uses of like, compared to the usage presented in Webster’s dictionary. He had
22 subjects (students in an introductory linguistics course) write down the meaning of like
in six instances of like in naturally occurring conversation to estimate the differences
between actual conversation and the "dictionary meaning." He (p. 44) states that all his
subjects agreed that like is used to indicate "a minor nonequivalence between overt
expression and intended meaning," which is rather close to the adverbial meaning 'nearly.‘
He (p. 45) concludes that like is used as an indicator of divergence between thought and
talk which might be caused by various circumstances of conversation, such as turn taking
requirements, unexpected turns of topic, distractions, and on-the-spot negotiations.

He states that discourse markers are used to relate what is covert to what is overt in
Ongoing conversational behavior (pp. 112-113). Each discourse marker has a core

meaning; it also has acquired various secondary functions through routine usage.

20

 

Therefore, by examining the core meaning, we are able to figure out why a particular
discourse marker is used in a specific context.

Schourup's analysis suggests that each discourse marker is associated with a "core
meaning" and came to assume other functions through its routinized use over time. His
insightful theory may partially explain why Japanese fillers are not interchangeable with
each other.

However, in the present study, since connectives, fillers and interactional particles
are treated as one category of items, discourse markers, and since Schourup's theory is not
applicable to connectives, though it will be drawn on for my analysis of fillers, Schiffrin's

approach is better suited for my analysis of Japanese discourse markers generally.

2.1.3. Discourse marker research and second language acquisition

Although McCarthy (1991:13-14) points out the importance of discourse markers,
such as right, okay, so, now and then for "framing moves" and "transactions," research on
non-native speakers' acquisition of English discourse markers is scarce. Firth (1988, cited
in McCarthy 1991149) states that, whereas his native subjects used varied expressions,
such as because, 'cos, like and see, even his advanced non-native subjects exclusively used
because as a causal connective, which made their speech sound unnatural.

As regards the roles which English discourse markers play in comprehension, there
are a number of studies. Chaudron and Richards (1986) tested comprehensibility of
spoken English with 71 ESL students at two proficiency levels, using four versions of a
recorded lecture: ( 1) no discourse markers included (the baseline version); (2) with "micro-
markers," such as and, because, well, and of course, included; (3) with "macro-markers,"
such as what I'm going to talk about today and let's go back to the beginning; and (4) with
both micro- and macro-markers. They claim that, while macro-markers significantly
increased comprehensibility for the non-native subjects, micro-markers did not aid

Comprehension and, furthermore, may have made comprehension more difficult by

21

 

detracting from the overall coherence of the lecture.

This study, however, is based on a scripted lecture as the baseline data, which is
unnatural to begin with. Furthermore, the micro-markers they added for the second
version of the tape were the ones which did not "add semantic information to the lecture"
(p. 117), which defeats the whole purpose of discourse marker presence, since typically
discourse markers add both semantic and pragmatic information to a discourse.

Chaudron and Richards' study is contradicted by later studies such as Taylor and
Bro (1992) and Flowerdew and Tauroza (1995). Taylor (1992) and Taylor and Bro (1992)
tested comprehensibility of lectures by international teaching assistants, using 15 native
speakers of English as their subjects. They found that the misuse or lack of connectives,
such as and, but, and therefore, by non-native teaching assistants significantly reduced the
comprehensibility of the lectures for native speakers.

Using an authentic video-taped lecture and its modified version with discourse
markers deleted, Flowerdew and Tauroza (1995) found that the presence of discourse
markers, both connectives and fillers, significantly increased the comprehensibility of a
native English lecturer for the 62 non-native subjects in his study.

All these studies, except that by Chaudron and Richards, suggest that the presence
of discourse markers plays a significant role in making one's speech more natural and

comprehensible.

2.2. Studies on Japanese discourse markers

This section presents a review of relevant literature on Japanese connectives, fillers,
and interactional particles. Japanese connectives, fillers, and interpersonal particles have
been analyzed by several researchers: (1) the connective dakara 'so': Maynard (1989b and
1993) and Karatsu (1993); datte 'but/because': Maynard (1993) and Mori (1994), (sore)de
'and; so': Ito (1995); (2) the filler nanka 'like': Saito (1992) and filler ano(o) 'uh, well':
Cook (1993); and (3) interactional particles: Uyeno (1972), Oishi (1985), Ishikawa (1988),

22

Cook (1988 and 1990), Maynard (1993), and Kamio (1994). In the following section, I

will review studies which are relevant to my analysis of Japanese discourse markers.

2.2.1. Connectives

Maynard analyzes the Japanese conjunctions dakara 'so' (1989b; 1993) and datte
(1993), following Schiffrin’s (1987) analysis of English discourse markers. Dakara has
traditionally been described as a 'consequential conjunction' or 'cause-and-result'
conjunction. Maynard argues that, in addition to this widely recognized function on the
textual semantic level, dakara has three additional functions as a discourse marker: (1)
additional explanation, (2) (reluctant) repetition, (3) and tum-end signal. As all three
additional functions of dakara take place at the interactional level, Maynard concludes
(1989b1411) that the multiple functions of dakara can be explained only by recognizing
multiple levels of discourse, as proposed by Schiffrin.

[to (1995) analyzes the function of sorede and de 'and' using Schiffrin's
framework. She (pp. 61-62) concludes that both sorede and de have functions similar to
the English and as a structural coordinator of units of talk (in the ideational structure), and
also as a marker of speaker continuation (in the exchange structure.) Furthermore, (sore)de
has the additional functions of requesting the other interlocutor's continuation (in the action
structure) and seeking further information (in the action structure and the participation
framework). Thus, Ito also acknowledges the need for recognizing multiple planes of
discourse in order to account for the plurifunctionality of (sore)de as a marker of
coordination at the textual semantic level and as a marker of speaker continuation at the
interactional level.

Onodera (1993), as a prelude to her diachronic analysis of demo 'but' as a
discourse marker, analyzes it using Schiffrin's framework. She states that, besides its
function as a referential contrast marker, demo connects the following 3 types of contrast:

(1) pragmatically inferable contrast in the ideational structure; (2) functional contrast (in the

23

exchange structure); and (3) contrasting actions (in the action structure). Contrasting
actions include the following: (1) point-making; (2) claiming the floor; (3) opening the
conversation; and (4) changing the topic.

Thus, Onodera also acknowledges the significance of Schiffrin's framework in the
analysis of Japanese discourse markers in order to fully account for the three types of
contrast demo exhibits. The three researchers whose work was reviewed above all agree
that Schiffrin's discourse model and her framework for analysis of discourse markers are
best suited for the analysis of Japanese discourse markers, which have multiple functions
on different levels of discourse; e.g. at the textual semantic and the interactional levels.
Furthermore, they all find similarities between Japanese connectives and their English

counterparts, and and (sore)de, but and demo, and so and dakara.

2.2.2. Fillers

Compared to the studies of connectives, research on Japanese fillers has been
scarce. In what follows 1 will review major studies on Japanese fillers.

Maynard (1989a) discusses fillers briefly, stating that the frequent use of fillers is
one of the characteristics of spoken Japanese. Although she does not treat fillers as
discourse markers, she argues for their importance and provides an interesting insight: that
they should be categorized into two types: (1) language-production-based fillers, such as ee
'well' and uuunto 'uh', which are used to fill a potential silence when smooth speech is
either cognitively or productively hindered, and (2) socially motivated fillers, such as nanka
'like', which are used to achieve "maximum agreeableness" with the recipient by
maximizing the similarities and minimizing the differences, ensuring that the recipient's
interpersonal emotion and rapport are protected (p. 32). Maynard (p. 31) further states that
"frequent use of final particles and fillers helps hide the message, delaying and softening its
delivery until the speaker is certain that the interpersonal feelings are intact when the

Semantic content is conveyed to the other interactant." She (p. 31) also states that "fillers

24

help create a casual friendly discourse with a pleasant emotional appeal to one's partner,"
even when the content of messages is not necessarily preferred.

Distinguishing two types of fillers is an important contribution Maynard made as
regards fillers, though she does not provide actual data to support her claim or refer to the
frequency and distributional characteristics of fillers. Thus, more detailed examination of
fillers is necessary to understand their contributions in conversation.

Cook (1993) discusses the use of ano(o) 'well' in discourse from the perspective
of politeness. Ano(o) originally is a demonstrative adjective used to point to an object
located at some distance from both the speaker and the listener. Cook (p. 23) discusses the
following uses of ano(o) as a filler in her data: (1) a turn-initiator, (2) an attention-getter,
(3) a proposition highlighter, (4) a new topic introducer, and (5) a mitigation device
prefacing a face-threatening act. Based on her data and the function of ano(o) as a
demonstrative adjective, Cook (p. 33) concludes that the filler ano(o) is used to align the
speaker and the listener on the same side with respect to the subsequent utterance in face-to-
face interaction, functioning as a "positive politeness (Brown and Levinson 19871101-
128)" marker to obtain the addressee's cooperation and create interpersonal rapport.

Cook's observation about ano(o) is insightful. However, I believe that the use of
ano(o) as an alignment marker between interlocutors is also related to its anaphoric function
as well as its role as a demonstrative adjective. In the anaphoric usage, ano is employed
when the speaker knows or assumes that the heater, as well as the speaker him/herself,
knows the referent of the anaphoric demonstrative (Kuno 19731283). Martin (1975:1067)
states that the anaphoric use of ano is to point out "obviousness," reminding the heater that
the reference is well known to both the speaker and the heater as a result of having just
been explained, or is widely known to one and all. The speaker treats a piece of
information as something shared or obvious; thus the anaphoric use of ano, by pointing out
obviousness and sharedness of information, aligns the speaker and the heater and enhances

the cooperative atmosphere of the conversation.

25

Saito (1992) analyzes the discourse marker nanka 'like' in Japanese conversation.

She (1992153) concludes that nanka is used as a softener, a tum-initiator, t0pic changer,
and a pause-filler. Saito's other findings include: (1) nanka occurs more often in a relaxed
conversation than in formal one; (2) nanka is less often used in the mixed-sex groups than
in the same-sex groups; and (3) that men use nanka more often than women. This is an
interesting point, since, given Japanese society's expectation for women to be more polite
and less assertive than men, we might presume that women use softening devices more
often. However, the number of Saito's subjects was small (n=8), and thus, the

relationship between gender and discourse markers deserves further research.

2.2.3. Interactional particles

Compared to connectives and fillers, more extensive research has been done on
interactional particles. Interactional particles do not contribute to referential meaning, but
display contextual information and a variety of social meanings. They thus play important
roles in face-to-face interaction, providing information about various aspects of
conversation such as: (1) the speaker; e.g. their gender; (2) the speaker-proposition
relationship, e.g. "territory of information" (see 2.2.4.3) and degree of the speaker's
certainty; and (3) the speaker-addressee relationship, 6. g. degree of intimacy and their
relative social status. In casual everyday conversation, we can hardly find utterances which
do not contain interactional particles. Maynard (19931101) states that, in her data, 31.8%
of pause—bounded phrasal units (PPUs) are accompanied by a particle. Thus, studying
interactional particles will provide us with important clues to understanding Japanese
conversational discourse.

In what follows, a review of the following studies is presented: (1) Uyeno’s (1971)
performative analysis of the particle ne; (2) Kamio's (1979 and 1994) analysis of ne from
the perspective of "territory of information"; (3) Maynard's (1993) analysis of ne and yo in

relation to the amount of information possessed by the speaker and the heater, and (4)

26

Cook's (1988; 1992) indexical approach to ne and ya.

2.2.3.1. Uyeno's (1971) performative analysis of interactional particles
Uyeno (1971) undertook one of the earliest analyses of Japanese interactional
particles. She examines several interactional particles in the perfomrative framework
developed in the late 1960's through early 1970's by generative semanticists. In this view,
the illocutionaty force of a sentence, such as warning, informing, and ordering, is
represented in the logical form by a perfomative verb, which may or may not appear overtly
in the surface form of the sentence. Uyeno states that Japanese interactional particles can
be analyzed in terms of a trifold representation: a surface structure, a logical form and a set
of presuppositions, and that interactional particles are derived from the underlying
performative verb in the logical form (p. 139). Uyeno (p. 59) considers the following in

analyzing each interactional particle:

Sex of the speaker:

. Sentence styles and relative social status;

Sentence types;

. Co-occurrence with other particles;

Occurrence in reported speech events;

Implications and effects; and

. Logical structure, presuppositions and surface structure.

oar-honour»

Uyeno (p. 132) presents the following as the presuppositions of the final particle ne and its

logical structure.

i. W:

X: male/ female
Y: 2nd person
X {>, =, < } Y, except the cases where the relative social status of X greatly

exceeds or its extremely lower than that of Y.
V: STATE/ ORDER/ ASK/ SUGGEST

27

 

ii. Logical structure

A
NP NP NP \ll
1 I l
X Y S 1 STATE
/\ ORDER
NP V1 ASK

SUGGEST

Figure 2.1 Uyeno's performative analysis of interactional particles

As a surface structure, the following are derived:
V=STATE1 iku ne; ikimasu ne ' you'll go, won't you?; I'll go, O.K.?' etc.
V=ORDER1 ikinasai ne 'go (polite); iku yo ne 'you'll go, won't you?, etc.
V=ASK1 iku ka ne; ikimasu ka ne 'I wonder if he'll go', etc.
V=SUGGEST1 ikoo ne 'let's go, OK?', etc.

Uyeno (p. 130- 132) claims that the particle ne implies "the speaker's intention that the
option of judgement on the given information is left to the addressee, and thus gives a
softening effect on the basic nature of each sentence type." Therefore, the addressee feels
the speaker's consideration toward him or her, thus enhancing the addressee's participation
in the conversation. Hence, ne is called the "particle of rapport" by Uyeno.

Uyeno's analysis is largely based on constructed isolated sentences, not on actual
conversation data. Furthermore, although some contextual information is provided based
on her native intuition, interactional aspects of conversation, which are crucial to the
analysis of discourse markers, are totally lacking. Thus, her analysis seems to be severely
limited. As Cook (1988185) argues, at any particular moment in a conversation, a speaker
constantly adjusts his or her presuppositions and subsequent choice of interactional

particles, depending on the hearer's reaction, to achieve the speaker's goal in the

28

conversation. Social meanings are inferred from the context in which an utterance is
produced. The interpretation of interactional particles and their social meanings are a joint
product of the speaker and the heater, requiring ongoing contextual information and
constant negotiation between the conversants. In the view that regards conversation not
only as a linguistic but a social and interpersonal activity, the functions of interactional

particles cannot be captured by a purely structural approach like Uyeno's.

2.2.3.2. Kamio's (1994) analysis of ne by the " territory of information"

Kamio (1979, 1990, and 1994) analyzes the final particle ne from a cognitive
perspective: "the territory of information." The "territory of information" refers to the
location of the source of information or knowledge. For example, if information is derived
from the speaker's experience or expertise, the information belongs to his or her territory,
whether or not it is shared with the heater. On the other hand, if the information is derived
from the hearer's experience or expertise, it belongs to the hearer's territory.

Table 2.2 is Kamio's representation of the territory of information and the use of ne
in the four territories of information (1990178). He argues that ne marks the information
which belongs to the hearer's territory (Territory C) or to the shared territory (Territory B)
as opposed to the speaker's territory (Territory A). He states (1990165) that in Territories
A and D, the use of ne is optional, and when it is used, it functions as a marker of
solidarity and politeness.

Kamio's analysis is somewhat similar to Maynard's (1993) analysis summarized in
Section 2.2.3.3. The use of ne, however, cannot be accounted for solely by the
"information territory." Ishikawa (1991122) states that 40% of ne usage in her data (letters
between close friends) occurred when information was only in the writer's territory,
whereas 60% of ne occurred when information was shared both by the writer and the
reader. Hence, she argues that both a cognitive perspective and an affective perspective

must be taken into consideration in the analysis of ne.

29

Table 2.2 Kamio's territory of information

 

 

 

 

 

 

 

Speaker's territory
In Out
A Direct form with D Non-direct form with
Out optional yo or ne. optional yo or ne.
Atsui desu (yo)(ne) Atsui soo desu (ne).
Hearer's hot is (yo )(ne) hot Ihear (ne)
territory B Direct form + mandatory C Non—direct form +
In ne mandatory ne
Atsui desu ne. Atsui soo desu ne.
hot is ne hot I hear ne

 

 

 

Although there may be differences in the use of ne in written Japanese vs. spoken
Japanese,l given Ishikawa's research, it is unwise to adopt Kamio's approach as the sole

means of analysis of ne.

2.2.3.3. Maynard's (1993) analysis of ne and ya

Maynard (1993) presents an analysis of yo and ne in a framework similar to
Kamio's. Prior to this work, Maynard (1989a) briefly discussed interactional particles in
relation to "self-contexualization," or "the ongoing process of defining oneself in relation to
one's interactional environment" (p. 4). She states that the frequent use of interactional
particles by conversants helps intensify their degree of involvement in the conversation,
and, further, that their sensitivity to varying levels of involvement is an important part of
the self-contextualization process.

In the discussion of characteristics of spoken Japanese, Maynard (1993) analyzes

sentence-final ne and yo in relation to information sharing. She (p. 106) states that the use

 

 

1 Correspondence between close friends lies somewhere between "spoken" language and "written" language,
'as can be seen from the use of interpersonal particles, such as ne 'confinnation', yo 'assertion', kana
uncertarnty' and a tag-question desho. which do not occur in formal written Japanese.

30

of ne and ya is regulated by the amount of information the speaker and/or the heater have.

The following is her conclusion.

Table 2.3 Maynard's (1993) conclusion on the use of ne and ya.

 

 

 

 

 

Amount of information a speaker or a heater possesses Particle used
1. the speaker's information > the hearer's information yo
2. the speaker's information S the hearer's information ne

 

Maynard's observations account for the the sentence-final ne and ya to some extent.
Ne and ya, however, are not used solely in relation to the amount of information the
interlocutors possess. For example, as we shall see in Chapter 5, the use of yo and ne and
the choice between yo and ne depends on various factors; e.g. how much emphasis the
speaker wishes to express or the relationship between the interlocutors, such as relative
social status and familiarity with one another other. Moreover, yo is often optional.

Maynard (p. 107) supplements her cognitive-oriented explanation of the choice of
ya and ne with an interactional perspective to determine the use of yo and ne: yo is likely to
be used when a piece of information is presented as "information-centered," whereas ne is
likely to be used when information is presented as "addressee-centered." Though
incorporation of the interactional approach makes Maynard's analysis more attractive, as I
noted regarding Kamio's analysis, it is better to analyze interactional particles with a multi-

faceted approach like the one proposed by Schiffrin (1987 ).

2.2.3.4. Cook's (1988) indexical approach
We now review Cook’s analysis of yo and ne, which emphasizes the function of
interactional particles in the affective domain. In contrast to Kamio's and Maynard's

interpretations, Cook's (1988) analysis of the interactional particles ne and ya is

31

 

predominantly from an affective perspective: interactional particles are regarded as "social
indexicals," following Ochs' model of indexicality. Ochs (cited in Cook 19881111)
proposes that, in any language, "affective and epistemological dispositions are directly
indexed, and that they serve as contextual building blocks of other socio—cultutal
dimensions, such as the speaker, the listener, their social identities, their relationship, and
speech acts." Ochs defines affective dispositions as feelings, moods, and attitudes of
participants toward a proposition, and epistemological dispositions as pertaining to the
participants' beliefs or knowledge vis-a-vis a proposition. Ochs' model of indexicality is

illustrated as follows:

 

 

Contextual dimension 2
(direct index) (indirect index)

Linguistic resource Contextual dimension 1

 

 

 

Figure 2.2 Ochs' model of indexicality

Direct indexicality refers to "a feature of the communicative event which is indicated
directly and unmediated" (Cook 1988:] 12), and indirect indexicality refers to "a feature of
the communicative event, indicated indirectly through the indexing of some other feature of
the communicative event." The direct indexical meanings help constitute indirect indexical
meanings; e. g. the direct indexing of the epistemological stance of uncertainty helps to
constitute the speech act of question.

Interactional particles, in this view, provide information on the social roles and
feelings of the speaker towards the addressee and a proposition. Cook (pp. 115-118)
notes, for example, that the final particle 20, which has traditionally been characterized as a
particle of male speakers, is used by female speakers in her data when they wish to express
fierceness or threats to a heater. She (p. 118) argues that the direct indexing of 20 is not

"male" but "coarse intensity," and this direct meaning of coarse intensity evokes indirect

32

H H

meanings, such as "male gender," "self-determination, threatening," or "excitement,"
regardless of the gender of the speaker.

Cook (p. 120) further states that the link between direct and indirect meanings
depends on the participants' knowledge, including socio-cultural associations, and that
members of a speech community share certain contextual assumptions which they bring to
a communicative situation. Cook (p. 120) also argues that conversants negotiate and
interpret a range of direct and indirect meanings indexed by a particular indexical sign in a
given speech context, and if a participant does not share many assumptions about the
context, indexing is likely to fail.

Cook (p. 118) argues that Ochs' model of indexicality can account for the dynamic
nature of social meanings and the complex indexical relations which Japanese interactional
particles exhibit Cook (pp. 154-155) characterizes the indexical relations of yo and ne as

follows.

Linguistic resource Direct Meaning Indirect meaning
.1. assertive attitude

2. speech acts
a. warning

b. advice
/ c. instructions
Pointing to g d. announcement
yo speaker's __ e. explanation

utterance \\\ f. report
\ g. request/command

h. insistence
i. contradiction

 

3. social relations
a. higher status
b. knowing party

Figure 2.3 Indexical relations of yo (Cook 1988: 129)

 

 

Linguistic resource Direct meaning Indirect meaning

1. speech functions/acts
a. initiating interaction
Interlocutors' b. introducing a new topic
ne general attitude of 4 c. displaying agreement/
mutual agreement \ confirmation/cooperation
d. seeking agreement/
confinnation/cooperation
e. mitigating face threatening acts

 

Figure 2.4 Indexical relations of ne (Cook 19881155).

Cook (p. 130) states that all the uses of ya contain the direct meaning of "pointing
to the speaker's utterance," and the various indirect meanings listed above derive from a
specific context for each conversation. For example, in order for an utterance to count as a
warning, the speaker must assume that the addressee is unaware of the future occurrence of
an adverse event, so the speaker must draw the addressee's attention and warn him or her.
Yo functions to draw the addressee's attention by pointing to the speaker's utterance.

In accounting for the meanings of ne, Cook proceeds in the same manner. All the
uses of ne assume an "interlocutor's general attitude of mutual agreement." In order to
introduce a new topic in conversation and maintain it, the speaker needs the addressee's
cooperation, and ne seeks this by eliciting his or her involvement. Hence, the indirect
meaning of "introducing a new topic" derives from the direct meaning of ne as an indexical
of an "interlocutor's general attitude of mutual agreement."

Cook's (198 8) analysis of interactional particles from an affective perspective as
social indexicals is a good one. The major reason for using interactional particles is for
affective reasons, and that is why they are not used in writing except for personal

correspondence, which is written as if one is speaking to the specific addressee. Her list of

34

 

the meanings of each particle is also very thorough, and it might also be pertinent to the
analysis of connectives and fillers.

There are, however, some weaknesses in Cook's analysis:

(1) First, not all ya or ne can be accounted for by the affective approach. As we
observed in reviewing Maynard's approach to yo and ne, the speaker's choice between
these two particles involve more than a cognitive or affective perspective alone.

(2) Secondly, Cook does not explain why the direct meaning of a particle leads to
many indirect meanings and functions in a variety of domains, such as cognitive, affective,
and tum-taking domains. If we employ Schiffrin's view that discourse is a synthesis of
multi-dimensional planes of talk, such as information state and participation and exchange
structures, and that expressions and functions have not one-to-one but many-to-many
relationships, we can offer a more straightforward account as to why the indirect meanings
of interactional particles derive from their direct meaning. By viewing interactional
particles as contextual coordinates which function on various planes of discourse, we will
be able to explain why yo is sometimes used for speech acts, and other times for social
relations or both.

Despite the weakness of Cook's analysis, her approach accounts for the use of ya
and ne in many situations. Therefore, I will be adopting part of her analysis in my analysis

of interactional particles in Chapter 5.

2.2.4. Studies on non-native use of discourse markers

Despite a vast number of studies on second language acquisition, research on the
use of discourse markers by non-native speakers has been scarce except for some recent
work on the interactional particles ne and ya: e. g. Sawyer (1992), Ohta (1994), and Mine
(1995). One reason may be that, as in Schiffrin’s definition, many discourse markers are
syntactically detachable, and many do not carry referential meanings. This means that

anyone, native or non—native speakers, can communicate without using discourse markers,

35

 

 

 

perhaps except for connectives and a few interactional particles, if the sole purpose of
communication is simply getting a message across. However, as incorrect use of yo and
ne can be detrimental to conversation (e.g. use of yo to one's superior), appropriate use of
discourse markers is mandatory for successful communication. Thus, teaching how to use
discourse markers should be an integral part of language instruction. In what follows, I
will review studies on the use of discourse markers by non-native speakers.

Sawyer (1992) provides a longitudinal study on the acquisition of the frnal particle
ne over a period of one year. His data were obtained by interviewing eleven students of
Japanese from a variety of language backgrounds. He found that: (1) acquisition of ne by
his subjects was slow compared to the acquisition of grammatical and lexical items; (2) ne
was first used in memorized chunks such as soo desu ne ‘yes, indeed’; and (3) individual
learners showed a considerable variation in manner and speed of acquiring ne (pp. 97-99).

Sawyer offers two reasons for the slow appearance of ne in the students’ speech, in
comparison with other items: (1) students' focus on other items due to predominantly
grarnmar-oriented instruction and/or (2) their concern with referential functions rather than
interpersonal functions. Based on his study, Sawyer (p. 106) suggests that less formal
affectively-oriented Japanese instruction is necessary, since "creating and maintaining
affective unity" (Cook 1990) is a very important part of Japanese conversation.

Sawyer’s study makes two contributions: (1) it provides data on the actual
longitudinal development of the acquisition of ne by non-native speakers, which can be
used as a basis for future studies, and (2) it offers helpful suggestions as to how Japanese
instruction should be conducted to produce speakers of Japanese who are concerned with
affectively-oriented communication.

Ohta (1994) discusses the acquisition of interactional particles by frrst-year college
students of Japanese over a period of one year in relation to their teachers' use of
interactional particles. She concludes that: ( l) a larger number and variety of interactional

particles are used in ordinary conversation in Japanese than in the classroom; (2) ne was the

36

 

 

 

particle most frequently used in the classroom by the three teachers she studied,
constituting 85.4% of all the particles; (3) ne was the only particle used by her subjects
during the entire year; and (4) the three teachers' use of interactional particles was
influenced by their stance towards their classroom roles (e. g. the teachers who focused
their instruction on communication used more particles than the one whose focus was on
grammar instruction).

Ohta's study may shed light on why ne is usually the first particle acquired by non-
native speakers, namely, ne is the one students of Japanese are exposed most frequently in
the classroom.

Mine's (1995) longitudinal study deals with the acquisition of sentence-final
expressions, including interactional particles, by non-native speakers of Japanese over 8
months. She states (p. 68) that ne and ya are the first sentence-final expressions acquired
by her subjects, followed by yone. Her study is important because it suggests an

acquisition order of Japanese interactional particles by non-native speakers.

2.3. Summary

The studies reviewed above shed light on various aspects of discourse markers.
Schourup's analysis of discourse markers as evincives, Maynard's analysis of several
discourse connectives and interactional particles in the framework of self-contextualization,
and Cook's analysis of interactional particles as social indexicals are particularly relevant to
the present study for the following reasons: (1) Schourup's analysis of English discourse
markers as evincives may be applicable to the analysis of Japanese discourse markers, and
may help us understand why a particular discourse marker is used in a particular situation;
(2) Maynard's analysis is important, since self-contextualization is culture specific, and the
use of discourse markers is related to expected communication patterns in the society in
which a conversation is taking place; and (3) Cook's analysis of interactional particles as

social indexical and affect markets is important because the very purpose of using

37

discourse markers is to index the orientation of the conversants toward each other and also
toward propositions.

While some of the studies of Japanese connectives use Schiffrin's framework, no
studies of other discourse markers, such as adverbs, interjections and interactional
particles, have been conducted using this framework. Furthermore, most studies of
interactional particles focus on one or another of Schiffrin's five discourse planes. By
applying Schiffrin's framework to connectives, fillers, and interactional particles, it is

hoped that the present study will lead to a better understanding of Japanese discourse

markers.

38

 

CHAPTER 3

Japanese Connectives as Discourse Markers

3.1. Preliminaries

Japanese connectives, like English connectives, play important roles in
conversation, providing contextual coordinates, bracketing various units of talk (e.g.
propositions, speech acts, and turns) and indexing an utterance to the local context in which
it is produced and interpreted. They coordinate propositional units in the ideational
structure as well as functioning on other planes of talk, displaying the speaker's orientation
to the proposition and the heater, and contributing to discourse coherence. For example,
de 'and' coordinates ideas and signal the speaker's continuation. Demo 'but' signals an
upcoming utterance as contrasting. Dakara 'so' expresses cause and effect relationships,
and may also express the speaker's irritation toward the addressee for having to provide
repeated or self-evident information (Maynard 1989a).

In the present section, the functions of Japanese connectives are examined using
both quantitative and qualitative approaches, as they make complementary contributions to
my analysis. The quantitative analysis section presents speakers' general preferences for
particular discourse markers, listing number of discourse marker occurrences, their
frequencies per 1000 words, and what percentage of total discourse marker use each
represents. The use of connectives is also discussed in relation to four sociolinguistic
variables: formality of speech (formal vs. informal), age (older vs. younger), gender (men
vs. women) and the genre (interviews, conversations and narratives). In the qualitative
analysis section, the three most frequently used connectives in my data, (sore )de 'and,’
demo 'but,‘ and dakara 'so', are examined to investigate their functions as contextual
coordinates in Japanese conversation.

My data show that Japanese connectives are similar to English connectives in their

functions and frequencies. Among the connectives which appeared in my data,

39

coordinative connectives such as de 'and' were used most often, constituting 46% of all
connectives used. Coordinative connectives coordinate propositional units in the ideational
structure and signal speaker continuation in the exchange structure. Adversative
connectives such as demo 'but' constituted 28% of connectives, and were used to indicate
the upcoming contrast in the ideational, action and exchange structures. The causal
connective dakara 'so' was the third most common connective (18% of all the connectives),
and was used to express cause-effect relationships. These three categories of connectives
constituted 92% of the connectives used in my data, indicating their importance in spoken
Japanese.

The use of connectives in my data, however, did not show significant a relationship
to speech styles or the age of speakers. This fact may indicate that connectives have
relatively well-defined roles and perform similar functions as contextual coordinates
regardless of the the formality of speech or the age of speaker. The use of connectives,
however, was significantly related to gender; women used connectives at much higher rate
than men. More particularly, the use of de 'and' and demo 'but' was significantly
influenced by gender difference. In what follows, I examine the frequency and functions

of Japanese connectives as discourse markers.

3.2. Method
3.2.1. Data

The main aim of the present study is to examine: (1) the use of discourse markers
by native Japanese speakers in spoken discourse; (2) the relationship between
sociolinguistic factors such as the interlocutors' gender and age and the use of discourse
markers; and (3) the differences between native and non-native speakers of Japanese in the
use of discourse markers.

In order to analyze the use of Japanese discourse markers, four types of data were

40

obtained: (1) casual dyadic conversations by college-age native speakers of Japanese (4
same-sex pairs and 2 cross-sex pairs; ages 18-21); (2) 12 interviews by the college-age
participants from whom the casual conversation data were collected; (3) conversations by
older participants (4 same-sex and 2 cross-sex pairs; ages 36-66); and (4) 6 narratives (ages
24-45).

The conversation data were obtained by asking the participants to talk about any
topic they chose for 30 minutes with a tape recorder in front of them. For the collection of
these data, unless the researcher was one of the conversants, the participants were left alone
to minimize what Labov (1972) calls the "observer's paradox": the more aware the
respondents are that their speech is being observed, the less natural their performances will
be, and thus the produced speech will be less natural.

The interview data were collected by interviewing the same college students from
whom the conversation data had previously been obtained. The interviews were conducted
in fonnal-style Japanese2 due to the following factors: (1) relationship between the
researcher and the subjects was not close enough to use informal-style Japanese;3 and (2)
to be used as the baseline data for the study of interview data of non-native subjects, most
of whom used forrnal-style Japanese. Each interview took 15 to 25 minutes. Although
the main job of the interviewer was to elicit responses from the interviewees, care was
taken to make the interaction "conversation-like" to the extent possible. For example, the
interviewer not only asked questions, but expressed her views and opinions on many of the
topics during the interviews.

Separate narrative data were obtained by asking three male and three female native

speakers to describe their happiest experience or most troubling experience in the US. The

 

2 Formal-style Japanese refers to the speech style in which the polite form of the copula (e.g. desu) and the
plain verb forms (e.g. -masu) are used.

3 Informal-style Japanese refers to the speech style in which the plain verb endings (e.g. -ru) and the plain
form of the copula (e. g. do) are used.

41

length of the narratives varied between two to ten minutes.4

Data collection took place in a dorm lounge and coffee shops to ensure a casual
atmosphere. The participants were told that their conversations and interviews would be
used for the researcher's conversation analysis, but were not told what aspect of their
speech was of particular interest. At the end of their interviews and conversations, the
participants were asked to fill out a questionnaire regarding their backgrounds (see

Appendix 2 for the questionnaire format.)

3.2.2. Participants

Two types of participants were recruited for the data collection. The 'younger'
group consists of six male and six female Japanese college students, ages 18-21
(participants #7-#14 and #19-#24 in Table 3.1). They came from three universities in
Japan to attend a summer English program in the US. They were classmates in their
respective Japanese universities and knew each other well enough to converse in the
informal speech style.

The 'older' group consists of twelve native speakers, ages 36 to 66 (Participants
#1-6 and 13-18). Though they knew each other well as friends or through their work, their
relationship to each other was not as close as that of the younger participants. Moreover,
due to the age difference in each pair, they all used fonnal-style Japanese.

The participants from whom the narrative data were collected were graduate

students and Japanese teachers at a local university. Table 3.1 is the list of participants.

‘

4 Data from 4 triadic conversations were also collected, which will be used throughout the present thesis
Whenever pertinent. Besides the Japanese data, 6 dyadic conversations and 4 narratives were collected from ,,
native English speakers speaking English, to be compared with the Japanese data, using the same procedure
as the Japanese data. The results of the analysis are presented throughout this thesis where the information
ts pertinent. See Appendix 4 for a list of the participants in English conversation.

42

 

Table 3.1 List of native Japanese participants

 

 

 

 

 

 

 

name sex age occupation hometown type of data topic

1 M] M 0 teacher Chiba C USA

2 WA M 0 teacher Tokyo C USA

3 JM M 0 teacher Chiba C travelling

4 WK M 0 teacher Tokyo C travelling

5 TH M 0 teacher Tokyo C travelling/ food
6 MA M 0 teacher Tokyo C daily life

7 FY M Y student Tokyo C/ I college life

8 SH M Y student Kanagawa C/ I college life

9 K M Y student Shizuoka C/ I USA/ classes
10 KM M Y student Shizuoka C/ I USA/ classes
11 T1" M Y student Kanagawa C/ I college life

12 Y] M Y student Chiba C/ I college life

13 TA F 0 teacher Tokyo C daily life

14 YO F 0 student Tokyo C daily life

15 YK F 0 teacher Tokyo C daily life

16 IM F 0 teacher Tokyo C USA

17 KH F 0 teacher Tokyo C travelling/ food
18 Mieko F 0 student Tokyo C daily life

19 IY F Y student Tokyo C/ I movies/ pastime
20 NM F Y student Tokyo C/ I movies/ pastime
21 UM F Y student Kanagawa C/ I travelling

22 KY F Y student Tokyo C/ I travelling

23 AR F Y student Tokyo C/ I USA

24 KB F Y student Kanagawa C/ I USA

25 K1 M 0 teacher Saitama N happy moment
26 UK M 0 student Tokyo N troubling time
27 WY M 0 teacher Niigata N troubling time
28 YY F Y student Tokyo N happy moment
29 YU F Y student Tokyo N happy moment
30 WM F Y student Tochigi N embarrassing time

 

O=older speaker (ages 33-66); Y=younger speaker (ages 18-21); C=conversation1 I=interview; N=narrative

3.2.3. Procedures

After collecting the data, the first two minutes of the recordings were transcribed

but categorically excluded from my quantitative analysis, since in the conversation data

some pairs were not yet fully engaged in natural conversation during the first minute or so.

Having been left alone with a tape recorder, some pairs could not agree upon what they

should talk about, while others started to talk immediately but in formal Japanese. After

one or two minutes, however, all the pairs became actively engaged in conversation. The

43

 

ten-minute segments following the initial two minutes of the recordings were fully
analyzed. The first two minuets of the interview data were also excluded from quantitative
analysis to maintain consistency. Regarding the narratives, only a portion of the
participants' narratives, namely, what Labov (1972:363-175) calls the "abstract" through
the "coda," were analyzed.

After transcribing the data, the markers were classified into three categories: (1)
connectives, (2) fillers, and (3) interactional particles. The number of occurrences of each
discourse marker was tallied, and the frequency of each discourse marker per 1,000 words
("the rate" hereafter) was calculated by dividing the number of occurrences of each
discourse marker by the number of the words uttered, and then multiplying the result by
1,000. After the rates were calculated, in order to see the relationship between
sociolinguistic factors and the use of discourse markers, the two-tailed t statistical
significance test was run for factors which have two variables, such as gender (male vs.
female) and age (younger vs. older). For factors which have three or more variables, e.g.
speech genre (interviews, conversations and narratives), the one-way ANOVA (analysis of
variance) test was performed. In total, 120 minutes of the interviews, 120 minutes of the

conversations, and approximately 25 minutes of the narratives were analyzed.

3.2.4. Units of talk

Defining 'units of talk' is an important task in discourse market analysis, since
Schiffrin (1987:31) operationally defines discourse markers as "sequentially dependent
elements which bracket units of talk." However, she deliberately left the definition of
'units of talk' vague in her analysis, stating that utilizing a precise unit would place a limit
on her analysis by restricting her attention to just that unit (p. 31). Therefore, her 'units of
talk' can be sentences, tone groups, actions, and so on.

Other researchers propose more concrete 'units of talk' based on prosodic cues

such as intonation or pauses (Halliday and Hasan 1976; Chafe 1982; and Iwasaki 1993,

44

 

inter alia). Halliday and Hasan (1976:325-326) state that text is organized into units of
information, expressed by intonation patterns in English, and that these information units
take the form of an unbroken succession of intonation units or tone groups. Chafe's 'idea
unit' is another 'unit of talk' based on intonation and pauses (1982137 and 1987122). He
states that idea units are 'a sequence of words combined under a single, coherent intonation
contour, are typically bounded by pauses, and usually exhibit one of a small set of syntactic
structures' (1987122). He adds that 'idea units' often coincide syntactically with a single
clause, and further states that 'they are a striking, probably universal property of spoken
language' (p. 37).

Regarding Japanese, however, Maynard (1989:23-24) argues that it is necessary to
establish a smaller unit than Chafe's idea units, since many of the units of talk defined by
phonological features, such as clear utterance-final pauses or intonation contour, are
phrases rather than clauses, i.e. morphological units of independent lexical items plus
junction words such as particles and do not contain verbals. Therefore, Maynard proposes
the use of pause-bounded phrasal units (PPU) as the unit of analysis (p. 23). Maynard
also lists the following characteristics for PPU boundaries: (1) a pause-predicting tone; (2)
a pause-waming speed; (3) skipped beats; or (4) a distinct intonation contour (p. 23).

The characteristics Maynard assigns to PPUs, however, seem to be similar to
Chafe’s idea units or Halliday's "information units" in that they are pronounced in one
continuous flow, often do not contain verbs, and may best be described as phrases rather
than clauses. Furthermore, in my data, most of these PPUs are clause-sized, both in
English (1,346 clauses in 1,487 intonation units=90.5%) and in Japanese (4,996 clauses in
5,667 intonation units=88.2% of the total corpus.)5 Matsumoto (1997) also reports that

the majority of Japanese intonation units consist of clauses rather than phrases (81% of

 

5 Short expressions such as 'O.K.' or 'sure' as a response to 'Will you call me tonight?‘ and one-word
answers to a question such as 'Tokyo' as a response to 'Where did you go yesterday?’ are treated as a clause
rather than a phrase in the present study.

45

1,121 intonation units in Matsumoto's study are clauses).

In the present study, I will consider the syntactic clause as the basic unit of analysis
for the following reasons: (1) most intonation units in my data were clause-sized; and (2) to
employ a consistent unit for formal and informal Japanese and non-native speakers'
Japanese (at least for the novice and intermediate level speakers of Japanese, intonation
contours and pauses were not a reliable indicator of intonation units, and therefore the use

of strictly intonation-based units was not practical).

3.3. Quantitative analysis

This section presents a quantitative analysis of the connectives in my data. The
number of occurrences, the percentage of total discourse marker use each connective
represents, and the rate per 1,000 words are presented in the present section. In Section
3.4, the use of connectives is discussed in relation to four sociolinguistic factors, formality
(formal vs. informal speech styles), gender (men vs. women), age (older vs. younger) and

speech genres (interviews, conversations and narratives).

3.3.1. The tokens, rates, and percentages of connectives

In this section, I discuss the number of occurrences of connectives, the rate of
connectives per 1,000 words, and what percentage of total discourse marker use each
marker represents. There were individual variations in the use of connectives. The
quantitative analysis presented below, therefore, should be regarded as a general tendency
in the use of connectives in my data.

Following Martin's categorization (19751818), I classify the connectives which
occurred in my data in the following categories: coordinative (e. g. de 'and' and soshite
'and'), adversative (e.g. demo 'but' and dakedo 'but'), causal (e.g. dakara 'so'),
consequential (e.g. ja(a) 'then'), disjunctive (e.g. aruiwa 'or'), and explanatory (e.g.

nazenaraba 'because').

46

 

Coordinative connectives were the most commonly used connectives, constituting
46% of the total connectives used. De 'and' was the most frequently used coordinative
connective (24%), followed by sorede 'and' (11%), and ato 'and/in addition' (6.6%).

Demo 'but’ was the most frequently used adversative connective, constituting 23%
of all the connectives in my data. The causal connective dakara 'so' constituted 18% of all
connectives used. The three most frequently used connectives, (sore )de 'and', demo 'but',
and dakara 'so', constituted 76% of all the connectives in my data. Table 3.2 summarizes
the number of occurrences, what percentage of total discourse marker use each one

represents, and the rate of each connective in my data.

Table 3.2 Connectives in the native Japanese data (n=36)

 

 

 

 

 

 

 

 

 

 

 

marker type marker (meaning) tokens (%) rate/1000 words
coordinative de 'and' 262 (24%) 7.32
sorede6 119 (11%) 3.33
ato 73 (6.6%) 2.04
sorekara7 45 (4.1%) 1.26
sos hi teg 7 (0.6%) 0.2
total (coord) 506 (46%) 14.1
adversative demo9 'but' 255 (23%) 7. 1 3
dattelo 35 (3.2%) 0.98
tada 15 (1 .4%) 0.42
tokoroga 1 (0.1%) 0.03
total (adv) 306 (28%) 8.55
causal dakara 'so' 1 98 ( 1 8%) 5.53
total (consq) (La 'then', etc.) 80 (7.3%) 2.24
total (disj) (aruiwa 'or', etc.) 11 (1%) 0.31
total (expl) nazenaraba 'because' 1 (0.09) 0.03
total 1102 (100%) 30.8

 

 

 

coord=coordinative connectives; adv=adversative1consq=consequentia11 disj=disjunctive; expl=explanatory

 

6 Variants of sorede, such as sonde and soide 'and' are also included in this category.
7 This category includes: sorekara 'and', soreto 'and', soreni 'and', and shikamo 'and'.
8 Of the seven occurrences of soshite 'and', three were used in its colloquial variant site 'and'.
9 This category includes: demo 'but', dakedo 'but' and shikashi 'but'.

10 Datte can be categorized either as an adversative or an explanatory connective. It is classified as an

adversative connective throughout this thesis.

47

 

The number of tokens of consequential, disjunctive, and explanatory connectives
was small, constituting 8.3% of the total connectives use. The consequential category
consists of ja( a ) 'then' (45 tokens), soshitara 'then' (21 tokens), soosuruto 'then' (8
tokens), dattara 'then' (4 tokens) and sorenara 'then' (2 tokens), constituting 7.3% of the
connectives. Disjunctive category consists of toyuuka 'or rather' (4 tokens), aruiwa 'or' (4
tokens) and soretomo 'or' (3 tokens). Only one token of explanatory connective occurred.
The low percentage of the explanatory connective is perhaps attributable to the agglutinative
nature of Japanese. Explanatory clauses in Japanese generally end with the bound
explanatory conjunctive particles such as -kara 'because' and -node 'because', rather than
beginning with independent connectives such as nazenaraba 'because'. Hence the
infrequent occurrence of explanatory connectives.

The overall percentages of connectives in my data were similar to those obtained by
other researchers, as shown in Table 3.3, which compares the number of tokens and the
percentages of connectives in four studies: the National Language Research Institute's 1955
study (cited in Onodera 199319), Onodera (1993111), Ito (1995124), and the present study.
In- all of these studies, the percentage of coordinative connectives, such as de and sorede,
was the highest, followed by the adversative connective demo or the causal connective

dakara, indicating the general preference for these connectives in Japanese conversation.

Table 3.3 Percentages of connectives in various studies

 

 

 

 

NLRI Onodera Ito (1995) present

(1955) (1993) 4 varieties study
connectives meaning 85 varieties 160 varieties 28 varieties
sorede, de, etc. 'and' 417 (27%) 539 (29.6%) 179 (43.3%) 353 (31.9%)
demo, etc. 'but' 196 (12.6%) 304 (16.7%) 96 (23.2%) 252 (22.9%)
dakara, etc. 'so' 188 (12.1%) 299 (16.4%) 118 (28.6%) 198 (18.0%)
sorekara, etc 'then' 138 (8.9%) 76 (4.1%) N/A 45 (4.1%)
jaa 'well then' 99 (6.4%) 311 (17.1%) N/A 45 (4.1%)
datte 'but/because' N/A 82 (4.5%) 20 (4.8%) 35 (3.2%)

% of connectives 67% of 88.4% of 100% of 84.2% of
number of tokens 1558 tokens 1819 tokens 423 tokens 1102 tokens

 

 

 

 

48

 

 

 

3.4. Sociolinguistic factors and the use of connectives

In this section, I examine the use of connectives in relation to four sociolinguistic
factors: (1) formality of speech in Section 3.4.1; (2) age in Section 3.4.2; (3) gender in
Section 3.4.3; and (4) speech genres in Section 3.4.4.

3.4.1. Formality

"Formal speech style" is defined in the present study as the speech style in which
the polite verb endings (e.g. -masu and -masen) and the polite form of the copula (e.g.
desu and deshita) are used. "Informal style speech" is defined as the speech style in which
the plain form of verb endings (e.g. -ru and -ta) and the plain form of the copula (e.g. da
and datta) are used.

In order to examine the differences in the use of connectives by formality, an
unmatched two-tailed t-test was run on the rates of connectives per 1,000 words at a 95%
significance level. A P-value of 0.05 or lower indicates a significant relationship between
formality and the use of connectives, and a P-value between 0.05 and 1.0 indicates a strong
tendency towards significance.ll

My results show that the difference in formality did not significantly affect the total

rate of connectives, contrary to a part of Hypothesis 1 presented in Ch 1:

Hypothesis 1

There will be a stylistic or genre-related variation in the use of connectives. They will be
used more often in narratives than in conversations, in formal speech than in informal
speech.

 

‘1 The statistical analysis does not show a clear relationship for connectives with a small number of

tokens such as the explanatory connective in my data. Therefore, the result is indicated as "indeterrninable"
in such a case.

49

The total rate of connectives was higher for informal speech than in formal speech
(33.8/1000 words and 29.6/1000 words, respectively), but the difference was not
statistically significant. By category, the rates of coordinative, adversative, disjunctive and
explanatory connectives were higher in informal speech than in formal speech, but the
differences were not significant or indetenninable due to low occurrences. Table 3.4
summarizes the number of tokens and the rate for each connective, with numbers in bold

indicating the higher rate between formal and informal speech.

Table 3.4 The tokens and rates of connectives by formality (two-tailed t-test; n=36)

 

 

 

 

 

 

 

 

 

 

 

 

marker meaning formal (n=24) informal (n=12) P-Value
tokens rate/1000 wds tokens rate/1000 wds

de 'and' 163 6.4 99 9.5 ns
sorede 'and' 87 3.4 32 3.1 ns

ato 'and' 55 2.2 18 1.7 ns
sorekara 'and' 34 l .3 11 1.1 ns
soshite 'and' 6 0.2 1 0.1 ns
total (coord) 345 13.6 161 l 5 . 4 ns
demo 'but' 168 6.6 87 8 .3 ns
datte 'but' 13 0.5 22 2 . 1 0.055
tada 'but' 15 0.6 ---------- 0.068
tokoroga 'but' 1 0.04 ---------- ID
total (adv) 197 7.8 109 10 4 ns
dakara 'so' 146 5.8 52 50 ns
total (consq) 5 1 2 . 0 29 1 .5 ns
total (disj) 9 0.36 2 2.8 ns
total (expl) 1 0.04 0 0. 2 ID
total 749 29.6 353 3 3 . 8 ns

 

 

 

 

 

 

 

degree of freedom: 34 ; confidence level: 95%; ----1 non-occurrence; P5005: significant correlation;
0.05<PSO.1: strong tendency; ID: indetenninable due to low occurrence; nsmot significant

As for individual connectives, the rates of the informal coordinative connective de
'and' and the informal adversative connective datte 'but' were higher in informal
conversations than in formal conversations, as well as another adversative connective demo
'but'. However, none of the individual connectives showed a significant relationship to

formality, though the rates of datte 'but' and tada 'but' were on the margin of significance

50

(P=0.055 and 0.068, respectively). I therefore conclude that formality does not
significantly influence the use of connectives, and that the functions of connectives as

contextual coordinates are similar in formal and informal speech.

3.4.2. Speakers' age

Speakers' age was not significantly related to the use of connectives. The rates
were higher for older speakers in every category except for coordinative connectives de
'and' (6.7/1000 words for older speakers vs. 7.6/1000 words for younger speakers) and
ato 'and; in addition' (0.77/1000 words and 2.77/1000 words, respectively), as well as the
total rate (34/1000 words vs. 29.3/1000 words), but the differences were not statistically
significant. Table 3.5 is a summary of the number of tokens and the rates of connectives
by the speakers' age, with the numbers in bold indicating the higher rate between the older
and the younger speakers.

Table 3.5 The tokens and rates of connectives by age (two-tailed t-test; n=36)

 

 

 

 

 

 

 

 

 

 

 

 

marker meaning older (n=12) younger (n=24) P-value
tokens rate/1000 wds tokens rate/1000 wds

de and 79 6.7 183 7 .6 ns
sorede and 42 3.6 77 3 2 ns
ato and 9 0.77 64 2 7 0.020
sorekara and 29 2.48 16 0.7 ns
soshite and 5 0.43 2 0.1 ns
total (coord) 164 14_ 342 1 4 . 2 ns
demo but 87 7.4 168 6.97 ns
datte but 12 1.0 23 0.95 ns
tam but 7 0.6 8 0.33 ns
tokoroga but 1 0.09 ---------- 1])
total (adv) 107 9.2 199 8.3 ns
dakara so 79 6.8 119 4.9 ns
total (consq) 42 3 .6 38 1 .6 0.045
total (disj) 5 0.4 6 0.25 as
total (expl) ---------- 1 0.04 11)
total 397 34 705 29.3 ns

 

 

 

 

 

 

 

--: non-occurrence; P5005: significant conelation; 0.05<PSO.11 strong tendency; ID: indetenninable due
to low occurrence; ns=not significant

51

The only items which showed statistically significant differences were ato 'and; in
addition' and the consequential connectives (P=0.020 and 0.045, respectively). Therefore,

I conclude that connectives have similar functions across speakers' age.

3.4.3. Gender

Women used more connectives than men overall and the difference was statistically
significant (P=0.000). They also used more coordinative, adversative, and causal
connectives than men, and the differences for coordinative and adversative connectives
were statistically significant.

Ide (1997160) cites a study on essays written by Japanese 4th to 9th graders and
states that male students used "positive conjunctions" such as soshite 'and' and dakara
'so', whereas female students preferred negative conjunctions such as demo 'but'. She
speculates that the reason for this phenomenon is that female students are less willing to
show confidence in what they are saying, and thus modify their opinions with less decisive
forms such as demo. However, probably due to the difference between written Japanese
and spoken Japanese, my data did not support this observation, since coordinative,
adversative, and causal connectives were all used by women at a higher rate than by men.
The consequential connectives also were used by women more often than men, and the
difference showed a strong tendency towards significance.

As regards individual connectives, the rates of de 'and', ato 'and, in addition', and
demo 'but' were significantly higher for women than for men. Sorekara 'and, then' was
the only connective which was used more often by men than women, but the difference
was not significant. Thus, according to my data, women use more connectives than men,
and the speaker's gender influences the use of connectives. Table 3.6 summarizes the
number of tokens and rates per 1000 words, with the numbers in bold representing the

higher rate for each connective between male and female speakers.

52

Table 3.6 The tokens and rates of connectives by the speaker's gender (two-tailed t-test; n=36)

 

 

 

 

 

 

 

 

 

 

 

 

marker meaning men (n=18) women (n=18) P-value
tokens rate/1000 wds tokens rate/1000 wds

de 'and' 61 3.9 201 10 0.001
sorede 'and' 44 2.8 75 3 . 7 ns

ato 'and' 18 1.1 55 2.7 0.048
sorekara 'and' 27 l . 7 18 0.9 ns
soshite 'and' 4 0. 3 3 0.2 ns
total (coord) 154 9.8 352 l 7 . 5 0.001
demo 'but' 74 4.7 181 9.0 0.009
datte 'but' 9 0.6 26 1.3 ns
tada 'but' 10 0.6 5 0.3 ns
tokorgga 'but' 1 0.06 ---------- ID
total (adv) 94 6.0 212 l 0 . 6 0.008
dakara 'so' 72 4.6 126 6.3 ns
total (consq) 21 1.3 59 2 . 9 0.087
total (disj) 6 0.4 5 0.3 ns
total (expl) l 0 . 0 6 ----- --_--- ID
total 348 22.1 754 3 7 . 6 0.000

 

 

 

 

 

 

 

---: non-occurrence; P5005: significant correlation; 0.05<P50.11 strong tendency; ID: indetenninable due
to low occurrence; ns=not significant

The younger men in my data talked about a wide variety of topics, each one briefly,
whereas women and older men discussed a few topics at great length. For example, FY
and SH, a young male pair, changed topics 6 times in 10 minutes: the State of Michigan,
their university in Japan, other universities in Japan, American culture, discrimination, and
their girlfriends. IY and and NM, a young women's pair, had only 2 topics: movies and
their families. TA and I, an older women's pair, had just one topic, TH's house. MI and
WA, an older men's pair, had one topic: comparing the US. and Japan. Since young men
changed their topics frequently, they did not have as many occasions to use connectives as

the other three categories of speaker, hence the low rates of connectives usage.

53

Tannen (1994199) states that, at all age levels in her study”, female speakers' talk
was tightly focused; they talked about a small number of topics at length, whereas male
speakers' talk was diffused, and they changed their topics often. The older men in my data
did not conform to the pattern described by Tannen, but the women and younger men
conformed to her characterization. Although Tannen did not comment on the use of
connectives in her study, perhaps tightly cohesive speech on a small number of topics by
women and older men in my data contributed to the larger rates of connectives compared to
those of younger men.

I now examine the rates of connectives in relation to the addressee's gender.
According to the t-test results, both men's and women's use of connectives were somewhat
influenced by the addressee's gender, though not significantly.

The male speakers' total rate was higher when speaking to a woman than to a man.
Particularly, the rates of coordinative connectives were high with female addressees. The
rate of sorede 'and' was significantly higher with a female addressee (P=0.001), and the
rates of sorekara 'and', dakara 'so', the consequential total and the total rate of connectives
were also higher when speaking to a woman than to a man, though not statistically
significant. Sorede 'and', sorekara 'and/then' and dakara 'so' are all markers of speaker
continuation. Therefore, we can say that male speakers used markers of continuation more
often when speaking to a woman than to a man.

The older male speakers in my data held the floor for a long time when they were
speaking to women, stating their opinions or describing their experience, and their female
partners had much shorter speech segments than the male speakers. On the other hand,
when a man was talking to another man, especially among the young participants, the floor
was turned over much more quickly between the conversants. This might partially account

for the high rates of coordinative and causal connectives in men's speech when speaking to

 

12 Tannen analyzed video-taped English conversations, using 8 same-sex pairs at four age levels: second-
graders, six-grmiers, tenth-graders, and 25-year-olds.

54

women than to a man (14.2/1000 words with a female addressee vs. 7.2/1000 words with
a male addressee for coordinative connectives and 7.8/1000 words with a female addressee
vs. 4.8/1000 words with a male addressee for causal connectives). The rates of
adversative connectives were higher with male addressees than with female addressees, but
the difference was not significant.

As regards the female speakers, the total rate of connectives use was higher when

speaking to another woman than to a man, as well as the rates for coordinative and causal

connectives. This is due to the fact that women had much longer speech segments with

female addressees than with male addressees. The rates of adversative connectives were

higher when speaking to a man than to a woman, but the difference was not statistically

significant. Table 3.7 summarizes the rates of connectives by the addressee's gender.

Table 3.7 The rates of connectives per 1000 words by the addressee's gender (two-tailed t-test; n=24)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Male speakers Female speakers

Mv.M Mv.F P- Fv.M Fv.F P-

(n=8) (n=4) value (n=4) (n=8) value
marker meaning token (rate) token (rate) token (rate) token (rate)
de 'and' 27 (4.3) 15( 4.5) ns 32 (8.0) 104 (12.2) ns
sorede 'and' 3 (0.5) 11 ( 3.3) 0.010 10 (2.5) 50( 5.9) ns
ato 'and' 2 (0.3) 4( 1.2) ns 8 (2.0) 13( 1.5) ns
sorekara 'and' 11 (1.8) 16( 4.8) ns 7 (1.7) 6( 0.7) ns
soshite 'and' 2(03) 1( 0.3) 1D ----- 3( 0.4) ID
total (coord) 45 (7.2; 47 (14.2) ns 57 (14._2) 176 (20.6) 0.083
demo 'but' 42 (6.7) 13( 3.9) ns 39 (9.7) 80( 9.4) ns
datte 'but' 6 (1.0) 2( 0.6) ns 15 (3.7) 11 ( 1.3) 0.078
tada 'but' 2 (0.3) 2( 0.6) ID 2 (0.5) 1 ( 0.1) ID
tokoroga 'but' ----- 1( 0.3) ID ---------- N/A
total (adv) 50 (8.0) 18 ( 5.4) ns 56 (13.9) 92 (10.8) ns
dakara 'so' 30 (4.8) 26( 7.8) ns 16 (4.0) 59( 6.9) ns
total (consq) 8 (1.3) 10 ( 3.0) ns 19 (4.7) 34 ( 4.0) ns
total (disj) ----- 3( 0.9) ID 1(0.3) 3( 0.4) ID
total (expl) ---------- N/A ---------- N/A
total 133 (21.2) 104 (31.4) us 149 (37.1) 364 (42.6) ns

 

 

 

--: non-occurrence; P5005: significant correlation; 0.05<P50.1: strong tendency; ID: indetenninable due

to low occurrence; ns=not significant; N/A: no P-value (=no occurrence)

55

As we saw above, the speaker's gender is significantly related to the use of
connectives, whereas the addressee's gender showed only a marginal relationship (except
the rate of sorede 'and' for male speakers with a female addressee). One note of caution is
that in my study, the speech data were obtained not from the same individual speaking to a
same-sex partner and an opposite-sex partner, but from different individuals conversing
with same-sex or opposite-sex partners. Moreover, the number of subjects was relatively
small (n=24 in total). Therefore, these findings may indicate a general tendency, but
cannot be said to confirm how a particular individual's speech is likely to vary depending

on the sex of his/her addressee. Thus, we will have to await further research on this issue.

3.4.4. Speech genre

Speech genre influenced the use of the connectives, which supported part of my
Hypothesis 1. The rate of de 'and' was the highest in narratives, and the difference was
statistically significant (P=0.013). The percentage of coordinative and causal connectives
together constituted 86% of the connectives used in narratives, indicating their importance
in narratives, where the speaker tells a story, connecting utterances in a coherent manner
and holding the floor for a long time. It is also possible that when the speaker does not
have to worry about fighting for the floor, coordinative and causal connectives are more
likely to be used than in conversations.

The rates of the contrast markers demo 'but' and datte 'but, because' were highest
in conversation. As Schiffrin (19871152) points out, one of the functions of adversative
connectives is marking an upcoming contrasting action. Therefore, they are less likely to
be used in interviews and narratives, where the speaker "monopolizes" the floor, than in
conversations, where the interlocutors can present opposing views to each other. Table 3.8
summarizes the number of tokens and the rates of connectives by speech genre with the

numbers in bold indicating the highest rate of the three speech genres.

56

Table 3.8 The tokens and rates of connectives by speech genre (one-way ANOVA test; n=42)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

marker meaning interviews (n=12) conversations (n=24) narratives (n=6) P-
token (%) rate token (%) rate token (%) rate Value
de 'and' 84 (23.9) 6.2 178 (23.7) 8.0 49 (54.4) 15.2 0.013
sorede 'and' 45 (12.8) 3.3 74 ( 9.9) 3.3 12 (13.3) 3.7 ns
ato 'and' 46 (13.1) 3.3 27( 3.6) 1.2 4 ( 4.4) 1.2 0.002
sorekara 'and' 5 ( 1.4) 0.3 40 ( 5.3) 1.8 ---------- ns
soshite 'and' 1 ( 0.3) 0.1 6 ( 0.8) 0.3 1 ( 1.1) 0.3 ns
total (coord) 181 (51.4) 13.3 325 (43.3) 14.7 66 (73.3) 20.5 ns
demo 'but' 81 (23) 5.9 174 (23.2) 7.9 4 ( 4.4) 1.2 0.020
datte 'but' 1 ( 0.3) 0.1 34 ( 4.5) 1.5 1 ( 1.1) 0.3 0.024
tada 'but' 8 ( 2.3) 0.6 7 ( 0.9) 0.3 ---------- ns
tokoroga 'but' ---------- 1 ( 0.13) 0.1 3 ( 3.3) 0.9 0.003
total (adv) 90 (25.6) 6.6 216 (28.8) 9.8 8 ( 8.9) 2.5 0.027
dakara 'so' 67 (19) 4.9 131 (17.5) 5.9 12 (13.3) 3.7 ns
total (consq) 9 ( 2.6) 0.7 71 ( 9.5) 3.2 4 ( 4.4) 1.2 0.008
total (disj) 4 ( 1.1) 0.3 7 ( 0.9) 0.3 ---------- ns
total (erqu) 1 ( 0.3) 0.1 -------------------- ID
total 352 (100) 25.8 750 (100) 33.9 90 (100) 28 ns

 

 

 

 

 

 

----1 non-occurrence; P5005: significant correlation; 0.05<P50. 11 strong tendency; ID: indetenninable due
to low occurrence; ns=not significant

3.4.5.

Summary

most often used, followed by adversative and causal connectives. This order was

We have seen that, in my data on spoken Japanese, coordinative connectives were

consistent with other research in Japanese and also Schiffrin's (1987) findings on English

connectives.

The use of Japanese connectives was examined in relation to four sociolinguistic

factors: formality, age, gender, and speech genre. Gender of the speaker was the most

significant factor influencing the use of connectives in my data, followed by speech genre.

It was shown that women used more connectives than men, and that the speaker's

gender is more significant than the addressee's gender. As in Tannen's study (1994), the

female participants in my data talked at length on a small number of topics, and the young

male participants discussed briefly a large number of topics, which may have contributed to

the high rates of connectives in women's speech.

57

 

Speech genre influenced the use of coordinative, causal, and adversative
connectives. Coordinative and causal connectives were used most frequently in the
narratives, suggesting that Japanese coordinative and causal connectives function as
markers of coordination and continuation, like the English and and so (see Schiffrin 1987).
Adversative connectives were used most often in conversations, where opposing views are
most likely to be expressed by the interlocutors.

Age and formality were not significant factors in the use of connectives in my data,
suggesting that connectives have similar functions across age and style. Thus, the gender-
related part of my hypothesis was supported by my data, but not the part related to
formality.

3.5. Qualitative analysis
In this section, a qualitative analysis of the three most frequently used connectives

in my data, (sore )de 'and', demo 'but', and dakara 'so', is presented.

3.5.1. The coordinative marker (sore)de 'and'

Sorede and its shortened variant de 'and' were the two most common coordinative
connectives used by the native participants, together constituting 35% of the total
connectives appearing in my data. In this section de and sorede are treated as one
connective (sore )de, except when they need to be distinguished (see, for example, 3.5.1.3
for a discussion of de vs. sorede).

(Sore )de is often treated as a causal connective. For example, A Dictionary of
Basic Japanese Grammar (Makino and Tsutsui 19861413) defines sorede as: "a conjunction
to indicate that what is stated in the preceding sentence is the reason or cause for what is
stated in the following sentence." Kojien Dictionary gives the following definitions for
sorede: (1) for that reason; and (2) thus [translation mine]. Martin (19751818) lists (sore )de

as a consequential and sequential conjunction, but not as a coordinative conjunction.

58

Several studies, however, indicate that the main function of sorede is to express a
coordinative relationship rather than a causal one (see Morita 1980; Ito 1995, and Philips
1995; Hudson 1998, inter alia).

Sorede consists of two components: the demonstrative pronoun sore 'that' and the
gerundive form of the copula do 'to be'. According to The Kojien Dictionary, gerundives
have the following functions: (1) conjoin prior and upcoming structures, events, etc.; (2)
express causal relationships; (3) express conditions; and (4) express adversative
relationships. The multifunctionality of (sore )de seems to derive from the
multifunctionality of the gerundive -de.

Ito (1995) lists the following as its three functions: (1) market of coordination of
units (PPUs in her analysis); (2) marker of speaker continuation; and (3) eliciting the other
interlocutor's continuation and requesting confirmation. Hudson (1998) states that sorede
is used in the following situations: (1) coordination of idea units; (2) expressing a cause—
effect relationship; (3) expressing contrast, similar to English but; and (4) in topic change.
Philips (1995) lists the following four functions for (sorede): (1) (sequential) coordination;
(2) speaker continuation; (3) requesting the previous speaker's continued turn; and (4)
connecting units with casual relationships.

Table 3.9 presents my summary of various functions of (sore )de as previously
reported. (*) indicates the planes of talk defined in Schiffrin (1987) on which each
function takes place. As shown in this table, (sore )de functions on every discourse plane
except in the information state, and it is very much like English and, which Schiffrin
(1987) characterizes as a marker of idea coordination and speaker continuation (see Ch. 2
of this thesis for the summary of various discourse markers analyzed by Schiffrin (1987).

Since a detailed analysis of (sore )de is presented in Ito (1995), the current study
focuses its attention on a review of the canonical functions of (sore )de, as it occurred in my
data. First, we consider the function of (sore )de in the ideational structure: (1) as a

coordination marker of idea units and (2) as a marker of causal relationship in Section

59

3.5.1.1. Secondly, I discuss the functions of (sore )de as a marker of continuation in the
exchange and action structures in Section 3.5.1.2. Finally, the functions of de and sorede

will be compared in Section 3.5.1.3.

Table 3.9 Functions of (sore )a'e on various planes of discourse

 

 

 

 

Planes of talk
Info. Ideational Participation Action Exchange
functions state structure framework structure structure
1. coordination *
2. causal marker *
3. contrast marker *
4. speaker continuation * *
5. elicit other's turn * * *
6. topic change *

 

 

 

3.5.1.1. (S ore)de in the ideational structure
3.5.1.1.1. Connecting idea units

As Schiffrin states regarding the English and, (sore )de is often used for more than
one function simultaneously. When (sore )de coordinates two idea units, for example, it
also functions as a marker of speaker continuation (see Hudson 1998). When (sore )de is
used for co—construction by several speakers, it simultaneously functions as a marker of
idea coordination and as an entry-device. In this section, however, we consider the roles
of (sore)de separately for ease of discussion.

As described in previous studies, both de and sorede functioned as a marker of idea
units or event coordination both at a local- and global-level in my data. (Sore )de as an
event connector occurred most frequently in narratives. (1) illustrates the function of de as

a coordinator of local-level units. Here YY is describing her experience in a library.

(1) YY: a. De atashi/ de okane koo dashite
and I and money like this took out

60

 

b. de tsukue no ue ni oita n desu yo.
and desk GEN top on put assert IP

c. De sono otoko no ko ga te o dashiteta n desu yone.
and that boy SUB hand DO extend assert IP

(1. De atashi kizukanakatta n desu yo.
and I notice assert IP

: a. And I took money out like this,

b. and put it on the counter.

c. And that boy was putting his hand out (towards me).
(1. And I did not notice that.’

De in (b), (c) and (d) is used to describe YY's experience choronologically, connecting

local-level idea units in the order of occurrence.

(2) is an example of de connecting global-level units. UM is telling her friend

where various high schools take their students for school trips.

a. Sukii iku tokoro mo areba nan da kke/
ski go places also exist what is it

b. Okinawa made itchau tokoro mo areba/
Okinawa to go places also exist

0. Hokkaidoo ittari/
Hokkaido go—and

(I. de uchi no gakkoo wa Kanazawa/ Hida Takayarna/ kana?
and my school TOP Kanazawa Hida Takayarna maybe

a. 'There are schools which go skiing

b. Some schools go as far as Okinawa,

c. some go to Hokkaido,

d. and my school goes to Kanazawa and Hida Takayama, I believe.‘

In (2), places other schools go for school trips are connected by other means than de or

sorede. In (a) and (b), UM uses mo areba 'there are cases such as', and in (c) she uses

-tari 'and so forth'. Finally, she talks about her own school, beginning with de. This de is

61

used to separate her own school from other schools in the area, connecting two global

units.

In (3), de connects both local- and global—level units. NM is recounting how her

parents, who knew each other in their home town, came to Tokyo, dated there again, and

eventually got married.

(3)

NM a. Saisho okaasan ga Tookyoo kita no kana?

first mother SUB Tokyo came Iwonder

b. De uchi no otoosan ga Tookyoo ni kite/ shigoto de/ otagai ne/
and my GEN father SUB Tokyo to go-and work for together [P

c. De okaasan wa koo motto nanka senmon gakkoo ni mata kayotte
and mother TOP this more kind of vocational school again attend-and

(1. de otoosan shigoto de/
and father work for

e. Mitaka toka ni sundete/

Mitaka or something at live-and

f. okaasan Setagaya toka yuu kara chikakatta rashii n da kedo/
mother Setagaya or something said so close perhaps assert but

g. nde soko de mo nando ka deeto shitari toka shitete/
and there also several times date did-and etc. do-and

h. de otagai mattete asondeta n da tte.
and each other wait for spend time-past assert QT

: a. 'I guess my mother came to Tokyo first,

b. and my father came later to Tokyo to work, both of them.

c. And my mother went to a vocational school there again,

(1. and my father worked.

e. My father lived in Mitaka.

f. My mother told me she lived in Setagaya, so they lived close to each other.
g. And they dated there several times, waited for each other,

h. and spent time together.‘

In (3), events are connected with de or its variant nde. De in (b) connects two events

sequentially: (1) NM's mother coming to Tokyo first, and (2) her father coming to Tokyo

next, thus functioning as a connector of local-level units. De in (d) connects two local-level

62

units: the two reasons why NM's parents came to Tokyo, and de in (h) also connects two
local sequential units: (1) their dates (g) and their having time together (h).

De in (c), in contrast, connects larger, more global units: (1) NM's parents' coming
to Tokyo, and (2) the reasons for doing so. Thus, de is used for connecting both local-
level and global-level units. As a marker of coordination of units, whether local or global,
de functions on two planes of talk proposed by Schiffrin (1987): (l) in the ideational
structures, connecting two ideas units, and (2) in the exchange structure, signalling speaker
continuation.

Sorede also functions as a connector of local- and global-level units. In Example

(4), sorede connects two local-level units.

(4) IY: a. Anoo/ are nan da kke Dizunii no Pokohantasu mita no ne
uh that what was it Disney GEN Pocahontas saw IP IP

NM: b. Un Pokohantasu?
Oh Pocahontas

IY: c. Un demo ne Mika gaa/ kinooo/
yeah but IP Mika SUB yesterday

(I. nanka wakannai Eego no/ nan da kke eega mite

what I don't know English GEN what movies saw

§

[In
uh huh

E?

f. sorede/ anoo un wakannai tte yappari.
and uh oh yeah not understand as you know

a. 'y'know? that/ what was it? I saw Pocahontas by Disney, y'know?
b. Huh? Pocahontas?

c. Yeah, but yesterday, Mika

(1. she saw an English/ well/ she saw a movie she couldn't understand
e. uh huh?

f. and/ uh/ oh, she said she didn't understand it as you can guess.‘

2; age

1n (0, sorede is used to connect two local-level units: (1) her friend Mika saw an English

movie, and (2) the fact that she did not understand it.

63

Both de and sorede function as markers of coordination of local and global idea
units. Interactionally, they signal the speaker's intention to continue his or her turn,

indicating that they have more to say.

3.5.1.1.2. (S ore)de as a causal marker

Both de and sorede can be used as connectors of cause and effect in the ideational
structure. However, the functional distinctions between (sore )de as a coordinative marker
and a causal marker are not always clear. In my data, the instances of (sore )de which are
replaceable with dakara 'so' were counted as causal markers, and the rest were counted as
coordinative markers. According to this classification, 76% of sorede was used as a
coordinative marker, and 24% was used as a causal marker. The percentage of de used as
a causal marker was 8% in my data. Therefore, although de is a shortened variant of
sorede, it seems to have lost much of its function as a causal marker, and is mostly used as

a coordinative marker. (5) is an example of sorede used as a causal marker.

(5) FY: a. Amerika-jin toka wa "omae wa sore de ii no ka" tte iwaretari suru toka
Americans TOP you TOP that with OK NOM Q QT say-PASS etc.

b. sorede nanka doki tte suru tte yuu ka/
so kind of astonished QT say Q

FY: a. 'Americans asked me, "Is it OK. with you?"
b. that's why I was kind of astonished.‘

In his illustration about how Americans value individual's wishes in everyday life, FY
describes his astonishment when his host father unexpectedly asked for FY's permission
so that the family's children can play with FY. In (b), sorede is used to express the cause

of his astonishment.

De was also used as a causal market, as shown in (6). TA is talking about house
hunting, and de is used to explain the reason why she began to look at houses in another

area.

(6) TA: a. Sorede moo zutto mitete/
and really long time looking-and

b. zenzen sukina no ga mitukara nakute
not at all like one SUB could not find-and

c. de chotto kokora atari mo mite miyoo kana nante/ maaketto o ne/
so a little here area also look try maybe market-DO IP

TA: a.'I kept looking at houses for a long time,
b. (but) I couldn't find anything I liked,
c. so, I thought of looking at houses in this area also.‘

In (5) and (6) above, we have seen that both sorede and de are used as a cause-effect
marker. Given the fact that sorede started out as a causal marker and that that still is the
primary function listed in dictionaries today, the percentages of sorede and de as causal
markers seem to be low. It is possible that (sore )de as a coordinating market will assume
more prevalent function in the future, but I leave that issue to future research. In the next

section, we examine the functions of (sore )de in the exchange and action structures.

3.5.1.2. (Sore)de in the exchange and action structures: A marker of
speaker-continuation

(Sore )de as a marker of speaker continuation may operate on three planes of talk
defined by Schiffrin (1987): (1) in the ideational structure, connecting idea units; (2) in the
exchange structure, signalling that the speaker has more to say and is not yet ready to
relinquish his or her turn; and (3) in the action structure, requesting the previous
interlocutor to continue his/her turn.

(Sore )de occurs in the following situations as a marker of speaker continuation: (1)

as a floor-holding device when the speaker is not yet ready to relinquish his or her turn; (2)

65

when presenting contrasting ideas within the speaker's turn; (3) when the speaker wishes
to present functionally different information, e.g. digression and supplementary
information; (4) as a tum-entry device for co-construction of a text; and (5) when
requesting another speaker's continued turn. I discuss each of these functions of (sore) de

below.

3.5.1.2.1. (Sore)de as a floor-holding device

The frrst example in which (sore )de is used in the exchange structure is as a floor-
holding device. As previous studies show (Ito 1995), (sore )de is used as a marker of
speaker continuation to signal the speaker's continued turn. (Sore )de is void of semantic
meaning in this case, and is often used with other connectives and/or fillers. (7) is such an

example.

(7) IM: a. Tada garnushara ni totteta dake de/ tan'i dake 0.
only randomly took only and credit only DO

b. Sorede sorede maa anoo dakara ne/
and and kind of uh so [P

c. sorede] atashi ga hutarime ka nanka datta n desu
and I SUB second or something was assert

d. chiisai daigaku datta kara.

small college was because

e. Dakara sonoo tan'i o kanzan suru nante koto wa kangaenakatta.
so uh credit DO transfer do such thing TOP did not think

IM: a. 'I just took courses randomly.
b. and and well uh, so, and
c. I was the second person sent as an exchange student

d. because my college was a small one
e. so, I didn't think of transferring the credits.’

Sorede in (b-c) is used with 2 fillers (maa 'kind of' and anoo 'uh') simply to keep the floor
while searching for words. The use of dakara in (b) indicates that IM is trying to explain

66

the reason why she was just taking courses rather than working toward a degree. (8) is an

example of de used as a floor-holding device.

(8) IY 1 a. Soide nakunatte kara otoosan sugoi suki n natchatte.
and passed away since father very like became

b. Demo otoosan sooyuu gara ja nai kara/ Tora-san-kee da kara.
but father such taste not because Tora-san-type BE because

c. Sonde saa uun de kekkyoku nee/
and IP uh and after all [P

d. natsu ni/ oneechan mo/
summer in older sister also

. onee-chan horaa eega sukina n da yo.
older sister horror movies like assert IP

0

. 'My father began to love (Audrey Hepbum's) movies since she passed away.
. But it's not like my father, 'cause he is more of a Tora-san type person,

. And y'know, uh and after all

. in the summer, my older sister, too

. she likes horror movies, y'know.‘

(DD-00"”

In this example, after holding the floor with sonde saa uun de kekkyoku nee 'and, uh and
after all', YI begins to talk about a totally different subject, her older sister's favorite
movies. Thus, examples (7) and (8) show that (sore )de is sometimes used simply to
maintain the speaker's turn and to show his/her intention to continue while searching for an

appropriate expression.

3.5.1.2.2. Connecting contrasting ideas

The second situation where (sore )de was used as a marker of speaker continuation
in my data was for connecting two contrasting units. (Sore )de is generally understood as a
coordinating marker like the English and. And in my English data was used not only to
coordinate two units of "like kind," but also in an environment shared with but or so, or

even with because. (9) illustrates a use of and connecting two contrasting units in my data.

67

(9) PE: a. What upsets me is that we talked about having the pressure on one instructor
b. to make the final decision,
c. and we try to move away from that.
(1. And I don't think we ever did, y'know.
e. And I, and I, that upsets me.

In ((1), and connects two contrasting idea units: (1) what the faculty tried to do in the
speaker's institution, and (2) the results of that effort, a failure. Thus, and can be used as a
contrast marker rather than as a coordinating marker.

(Sore )de also occurred in an environment shared with demo 'but' and dakara 'so' in
my data. Sorede as a contrast marker is cited in Hudson (1998) and de as a contrast marker

is noted in Ito (1995128-29). (10) is an example of de connecting contrasting units from my

data.

(10) WA: a. Demo wareware daitai shimetemasu yone hutsuu daigaku de wa.
but we usually wear IP ordinarily college at TOP

M]: b. Maa atsukattara boku wa shirnemasen kedo.
well hot if I-TOP wear-NEG but

c. De jugyoo no am hi wa daitai shimeru to.
and class GEN have day-TOP usually wear

WA: a. 'But we usually wear ties on a college campus, don't we?

MI: b. Well, I don't wear one when it's hot, but
c. But when I have classes, I wear one.’

In (0), MI states that he generally does not wear a tie during a hot season, and he continues
on with de in (c), saying that he does wear one when he has classes. In this example, since
the contrast is already expressed with -kedo at the end of (b), de in (c) connects two
contrasting units, and functions as a marker of speaker continuation simultaneously. In
total, 15/262 tokens (5.7%) of de and 1/119 token (0.8%) of sorede were used for

connecting contrasting ideas in my data.

68

All the cases of (sore )de as a contrast marker in my data occurred within the same
speaker's turn. This is another indication that (sore )de in this situation was used primarily
to keep the speaker's turn, connecting two contrasting idea units. According to my data, it
seems that if the upcoming contrast is to be expressed within the same speaker's turn,
either (sore )de or adversative markers such as demo 'but' and dakedo 'but' can be used. If
the upcoming contrast is across speakers, however, a "true" adversative marker must be

used rather than de. (11) illustrates this point.

(11) WA: a. J isseekatsu no naka de wa nanka koo/ nanka onnaji-yoona hito ga
real life GEN within TOP kind of this well same-kind person-SUB

b. takusan dete kuru yoona/ kanji ga shimasu nee/ un.
many appear like feeling-SUB do IP yeah

MI: c. Demo osoraku min-/ wakai hito-tachi nanka jibun no shitai yooni
but perhaps young people etc. self GEN want to do as

d. shitai to yuu kimochi tte no wa daibu am to omou n desu kedo ne.
want to do desire TOP fair amount have QT think assert but IP

WA: a. 'I feel that in reality, the same kind of people,
b. many people will be produced who are like each other.

M]: c. But perhaps young people of today have a fair amount of feelings
d. for what they really want to do, I think.’

In (d), it would be odd to use de instead of demo 'but'. So, if contrast is to be expressed

across speakers, a true adversative connective is in order.

3.5.1.2.3. Connecting functionally different units

(Sore )de as a marker of speaker continuation was also used to connect functionally
different units within a speaker's turn; e.g. the main topic of conversation and background
information or supplementary information. (12) is an example of de used to differentiate a

repair from the main information.

69

 

(12) AR: a. Suupaa dc ne/ nanka nee/ nanka nee/
supermarket at IP like IP like [P

b. hajimete kinoo kaado o tsukatta no ne?
for the first time yesterday card-DO used IP IP
SK: c. Un un un.
yeah yeah yeah

AR: (1. Anoo/ hajimete ja nai ka.
well the first time NEG Q

e. De nanka jibun de koo yatte yaru yatsu na no ne?
and like self by this do do one IP IP

AR: a. 'At a supermarket, well, uh, y'know
b. I used a charge card for the first time yesterday.
c. Yeah yeah yeah.
AR: d. Well, it wasn't the first time, maybe.
e. And it was a kind of machine one has to put a card through oneself

In ((1), AR corrects herself saying that it was not the first time, and in (e), she returns to her
description about using a charge card. Thus, de marks functionally different portions of an
utterance, and signals speaker continuation simultaneously.

When (sore )de was used to connect functionally different portions of an utterance,
the units connected with it were generally on the current topic of conversation, though a

sub-topic shift occurred occasionally. (13) is an example of a sub—topic shift with de.

(13) KB: a. Ano sa/ koko no saa/ ja ne/ koko no sa/ Eego no jugyoo tte saa/
Well IP here IP then [P here IP English class QT [P

b. konna taihen da to omotta?
this much demanding QT think-PAST

YI: c. Zenzen.

not at all

(I. Shukudai ga taihen da yo.
homework—SUB demanding BE IP

e. Mecha taihen datta.
very hard BE-PAST

70

KE: f. Shukudai ga taihen da yone.
homework-SUB hard BE [P

g. De atashi sa/ 5-ko mo Shukudai ga am no/ Bunkyoo no repooto ga.
and I IP item EMPH. homework-SUB exist IP placenarne GEN report-SUB

h. De koko de yaroo to omotte ne
and here do-VOL QT think IP

KE: a. 'Hey, here, then, English classes here y'know
b. did you expect them to be this demanding?
c. Not at all.
(1. There's a lot of homework.
e. Very hard, it was.
KE: f. Homework is really demanding, isn't it.
g. And I have five assigned papers for Bunkyo.
h. And I planned to do that while I am here, y'know?‘

§

In (g), KE shifts the sub-topic of conversation from the homework assignments for her
summer English program in the US. to her other homework for her university in Japan.
Thus, KE uses de in this situation to keep her turn while changing the topic of
conversation, and at the same time de differentiates the main topic of conversation from a

sub-topic of conversation.

3.5.1.2.4. (S ore)de as a turn-entry device for co-construction

Five tokens of de were used as a tum-entry device in my data. In this situation,
though the floor is taken over by the new speaker, what follows de was limited to a
cooperative supplementation of the prior speaker's utterance, not an expression of
opposition. Thus, de in this case was used to jointly construct a single discourse between a
prior and a current speaker, and to enhance the interactional relationship between the two

interlocutors. 13

 

13 Ono and Yoshida (199611 18-120) state that co-constniction in Japanese is extremely rare at the syntactic
or discourse level. Co-construction at the syntactic level was observed once in my data in a young men's
pair. Co-construction at the discourse level, however, occurred with male and female and older and younger
Speakers; 5 such cases began with de.

71

(14) KB:

KH:

a. Kooyuu kenshuu tte yuu no gal
this kind of program SUB

b. sugooku taisetsuna koto ja nai no kashira to omou n desu yonee.
very important thing must be QT think assert IP

c. Hai/ de yahari anoo/ chishiki naki keeken to
Yes and as expected uh knowledge-without experience and

d. keeken naki chishiki o kurabemasu to ne/
experience-without knowledge-DO compare when [P

e. watashi wa keeken no hoo ga mono o yuu to omou n desu yone.
I TOP experience SUB powerful QT think assert [P

a. 'This kind of study abroad program

b. is a very important experience (for the students), I think.

c. Yes, and as you know, comparing knowledge without experience
(1. and experience without knowledge,

e. I think experience is much more important.’

In this example, as soon as KH has finished her turn, TH agrees with her with hat 'yes',

and immediately starts his turn with de, supplementing what KH has stated.

Such joint-construction of a single text was often observed in triadic conversations,

in which two of the conversants were female speakers who shared experience, and the

other conversant was not as close as the two. In (15), SM and IK, both female speakers,

are telling NW, a male speaker, about their experience.

(15) SM:

NW:

SM:

a. Asia resutoran toka aru jan/ asoko no Sebun-irebun no mae.
Asia restaurant etc. exist TAG there Seven Eleven in front of

b. Un am am.
Yeah exist exist

c. Asoko ni haittanda yone?
there in enter assert IP

(1. De haitta n.
and entered [P

e. De atashi ga koohii ga nomitai.
and I SUB coffee SUB want to drink

72

SM: a. ’There is an Asia restaurant, etc. in front of Seven Eleven, y'know?
NW1 b. Yeah yeah.
SM: c. We went in there, right?
IK: (1. And we entered,
e. and I wanted to drink coffee.‘

In (c), SM gives background information to NW about the restaurant, and then invites IK
to enter the conversation by saying yone 'right?'. In ((1), IK uses de as an entry device to
begin her narrative. As with (14), the unit prefaced with de has to be a "cooperative" type
which supports a prior statement.

A similar usage of the English and was observed in my data. In (16), AP and CP
are talking about psychoanalysis.

(16) AP: a. It's difficult to measure um the amount of good it's doing for society
b. because we don't probably have enough records about]

c. the exact urn nature of ( )
(1. how many people are benefitting from it.
CP: e. Right.
AP: f. And it' 3 true that there have been la lot of ( )
CP: g. [And how much therapy do you need,

h. I mean if you are going by the medical industry and health 1nsurance.

In (g), CP starts his turn with and in the middle of AP's utterance. And in CP's utterance
here connects his utterance in (g-h) with AP's utterance in (f), supporting AP's argument in
(a-d). Thus, CP takes the turn from AP with and, but he support AP's argument.

Thus, both the Japanese de and the English and are used across speakers as a tum-

entry device for co—construction of a single text.

3.5.1.2.5. Requesting other speaker's continuation
Another type of (sore)de in my data was in requesting more information, in other
words, requesting the previous speaker's continued turn. In most cases, (sore )de was

used as a preface to a question, requesting clarification, or elaboration on what has

73

previously has been stated. There were two tokens of de (both cases were by myself) used

by itself to request the previous speaker's continuation. In (17), KM is describing his

experience in Detroit.

(17) KM:

Mieko:

KM:

Mieko:

Mieko:

Mieko:

Mieko:

KM:

Mick :

KM'

Mieko:
Mieko:

a. Detoroito itta n desu kedo
Detroit went assert but

b. Aa honto ni?
oh really

0. Uun/ nanka/ nanka/ nanka/ un nanka ayashii machi da na toka
well like like like yeah kind of questionable town BE IP something

(1. nanka koo/ sakaete
like like this flourish

e. E? dauntaun itta no?
oh downtown went Q

f. Ano ano/
uh uh

g. Doo deshita?
how was

h. Kanari abunakatta.
fairly dangerous-PAST

i. Ee/ de?

yes and

j. De yoru 9:00 sugi toka chotto aruitetara/
and night 9100 after about a little walking-when

1. nanka moo zen'in mirareta n desu yo.
somehow really everyone see-PASS assert 1P

a. 'We went to Detroit.

b. Oh, really?

0. Uh, like, like, like, kind of a questionable town,
d. somehow, well, lively

e. Oh? you went to Downtown Detroit?

f. Uh/ uh/

g. How was it?

h. It was pretty dangerous.

1. I see, and?

j. And when we were taking a little walk after 9100 at night,
1. we were all stared at for some reason, y'know?’

74

 

In (1), I used de with rising intonation to request more information from KM. In this
function, (sore )de primarily operates in the action structure, urging another interlocutor's
continued turn.

I have examined the functions of (sore )de as a marker of idea coordination and
speaker continuation in 5 situations above. (Sore )de coordinates local- or global-level idea
units, and also connects units which have cause-effect relationships. As a marker of
speaker continuation, (sore )de signals the present speaker's continued turn and is also used
for requesting the prior speaker's continued turn. In its textual function, (sore )de indexes
an utterance both to prior and upcoming units and indicates their relationship to each other,
e.g. coordination or causal relationships. In its interpersonal function, (sore )de primarily
signals the speaker's continued turn. When (sore )de is used for cooperative entry for
jointly constructing discourse or to urge the current speaker's continued turn for more
information, it indexes the utterance both to the speaker and the hearer. Thus, (sore )de
contributes to displaying discourse coherence through its textual function, coordinating idea

units, and by its interpersonal functions, signalling speaker continuation.

3.5.1.3 S orede and d e

Ito (1995153) observed the following differences between sorede and de in her
spoken Japanese data: (1) de generally connects units at local levels, while sorede is used to
connect units at a more global level; and (2) de shares some of its functions with demo
'but' or dakedo 'but', whereas sorede does not; and (3) there is a distributional difference
by gender between de and sorede: sorede is used more often by male speakers than female
speakers, whereas no such difference exists for de.

As regards the difference between sorede and de, Hudson (1998) states that it is
difficult to clearly differentiate the two since they appear as several variants such sorede:

soide, sonde, hoide, honde, nde, and de on a continuum. In what follows, I argue that: (1)

75

there is no clear difference in the use of de and sorede by gender, (2) de and sorede share
the same functions as a marker of idea coordination and speaker continuation; (3) the
difference between de and sorede in part derives from a speaker's preference for a shorter
variant in speaking, i.e. it is easier and takes less time to pronounce de than sorede, and de
is more colloquial than sorede, since de is used primarily in speaking; and (4) de and
sorede often alternate as a connector of either local or global units to break textual
regularity, rather than based on their intrinsic qualitative differences.

First, we will look at use of de and sorede in relation to gender. My data did not
conform with Ito's finding. As regards the percentage difference between de and sorede in
relation to gender, both men and women used de more than sorede, though women's
percentage of de was higher than men's (men: de: 58.1% vs. sorede: 41.9% ; women: de:
72.8% vs. sorede: 27.2%).

Concerning the rate per 1,000 words, de usage was much higher for women than
men and the difference was significant (10/1000 words for women vs. 3.88/1000 words
for men; P=0.001). In contrast, the gender difference was small for the rate of sorede, and
was not statistically significant since the P—value was more than 0.05 (3.74/1000 words for
women vs. 2.8/1000 words for men; P=0249.) Thus, my finding was the opposite of
Ito's and, therefore, as regards distributional asymmetry between de and sorede based on
gender, we must leave the issue for future research. Table 3.10 summarizes the use of de

and sorede in my data.

Table 3.10 The number of tokens, percentages, rates of de and sorede by gender

 

 

 

 

 

 

 

men (n=18) women (n=18) P-value
marker tokens (%) rate tokens (%) rate
de 61 (58.1%) 3.88 201 (72.8%) 10 0.001
sorede 44 (41.9%) 2.8 75 (27.2%) 3.74 0.249 (=ns)
total 105 (100%) 6.68 276 (100%) 13.74

 

 

 

 

 

ns=not significant

76

Concerning the use of de and sorede in relation to speech genre, a clear difference
was observed in my data. The percentage of de was the highest in the narratives (80.3%),
followed by the conversations (65.3%) and the interviews (65.1%). The rate of de per
1,000 words was also the highest in the narratives (15.2/1000 words), followed by the
conversations (8.04/1000 words) and the interviews (6.15/1000 words), and the difference
was statistically significant (P=0.0137).

In contrast, there was almost no difference in the rates of sorede in the three speech
genres (3.3/1000 words in the interviews, 3.34/1000 words in the conversations, and
3.73/1000 words in the narratives), and the difference was not statistically significant
(P=09962). It is evident that in narratives de was preferred to sorede.

Thus, my data show that both gender and speech genre are significantly related to
the use of de, but not to that of sorede. Table 3.11 summarizes the number of tokens,
percentages, and the rates/1000 words of de and sorede in three speech genres: interviews,

conversations, and narratives.

Table 3.11 The number of tokens, rates/1000 words, and percentages of de and sorede by speech genre

 

 

 

interviews conversations narratives P-value
token (%) rate rates rate token (%) rates
de 84 (65.1%) 6.15 79 (65.3%) 8.04 49 (80.3%) 15.2 0.0137

sorede 45 (34.9%) 3.3 42 (34.7%) 3.34 12 (19.6%) 3.73 0.9962 (=ns)

 

 

 

 

 

 

 

 

 

 

total 129 (100%) 9.45 121 (100%) 11.38 61 (100%) 18.93

 

ns=not significant

Next, we consider the functional differences between de and sorede. At least in my
data, there was no clear difference between de and sorede. All the functions discussed thus
far were shared by both de and sorede, though there was some difference in terms of which

of the two was preferred based on the situation.

77

First, the percentages of the two connectives used as a causal marker were
different. As mentioned in Section 3.5.1.1.2, in my data, 24% of sorede was used as a
causal marker, whereas only 8% of de was used in that function. It seems that de is
preferred as a marker of coordination, whereas sorede is preferred as a causal marker (also
see Hudson 1998).

Secondly, sorede does sound more formal than de, though the distinction was not
clear in my data. Six native speakers of Japanese to whom I addressed the question of the
distinction between de and sorede all stated that sorede is more formal sounding than de.

In my data also the rate of sorede was higher in conversations in formal style Japanese than
in informal style, though the difference was not statistically significant.14 Therefore, in a
very formal situations, such as political speech and formal lectures, sorede is more likely to
be used than de.

As for the units which de and sorede connect, there was not a clear distinction in
my data. Both de and sorede were used to connect units at both local- and global-levels.

(18) is an example of sorede being used as a connector of local units.

(18) SH: a. Detroit de chotto toraburu ga okichatta n desu yo.
Detroit in alittle trouble SUB happened assert [P

b. Basu ga yotee doori ni tsukanakute
bus-SUB as was scheduled arrive-NEG

c. sorede noriokurechatte.
and get on late

(1. De chanto tsuku no ka doo ka tte huu ni basu-gaisya ni ittara
and as expected arrive whether or not QT like bus-company 10 said when

e. tsuku kara daijoobu da tte itteta n desu kedo
arrive so OK BE QT said assert but

f. kekkyoku okurechatta 11 de
after all was late

 

 

14 .The rates of sorede was 3.43/1000 words in formal speech and 3.06/1000 words in informal speech,
Which was not statistically significant (P=0989). The rate of de in formal speech was 6.43/1000 words and
9-47/1000 words in informal speech, which also was not statistically significant (P=0189).

78

SH:

a. '(We had to spend a night in Detroit because) we had a little trouble there.
b. Our bus (from Canada) did not arrive on time,

c. and we couldn't get on the bus,

(1. and we had asked the bus company if the bus would arrive on time,

e. they said that it would, and we would be all right, but

f. it didn't arrive on time after all'

The structure of (18) is as follows:

(1 8)’ a. Topic 1: SUMMARY trouble in Detroit
b. Topic 2: EVENT in Detroit in Detroit the bus from Canada was late
c. sorede EVENT in Detroit and/so they couldn't get on the connecting bus
d. De Topic 31 EVENT in Canada And inquiry at the bus company in Canada
e. EVENT in Canada what they were told at the company in Canada
f. SUMMARY of (d) and (e) the bus from Canada was late after all

In (c), sorede connects two local-level units: (1) their bus from Canada was late, and (2)

they could not get on the bus which was leaving from Detroit to come back to their

dormitory. On the other hand, de in (e) is used to connect more global units: (1) the event

in Detroit on the previous day and (2) event in Canada before they left for Detroit.

(19) is another example of de connecting global units, and sorede connecting local

units.

(19) IY:

a. Ano Kanagawa no tomodachi n chi ni asobi ni itta no ne.
uh Kanagawa GEN friend GEN house to visited IP IP

b. De asoko ni nanka ojisan/ sono tomodachi no otoosan ga
and there well a man that friend GEN father-SUB

c. sugoi ii ojisan na 11 da kedo/
very nice man assert but

(I. onna 3-nin otoko 3-nin de asobi ni itte/
woman people man people by visited-and

e. de nanka otoko 3-nin wa nanka/ yoku wakannai geemu o shitete
and like man people-TOP like well understand-NEG garne-DO did and

79

f. soide atashi-tachi wa zutto sono reezaa disuku o misaserareta no ne.
and we -TOP the whole time that laser disk -DO watch-PASS IP IP

IY: a. 'Well, we went to visit a friend in Kanagawa.
b. and there was uh a man there/ my friend's father.
c. A very nice man.
d. Three girls and three guys went there, and

e. and the three guys were playing with a game I didn't know well,
f. and we, the girls, were made to watch that laser disk the whole time.

The structure of (19) is as follows:

(19)' a. Topic 1: EVENT visiting IY's friend's house
b-c. de Topic 2: EVENT and description of her friend's father
d. Topic 3: EVENT people who visited the friends
de Topic 41 EVENT and what the boys did that night at the friend's house
f. soide Event and what the girls did there

Sorede in (f) connects two local units: (1) what the three girls did at a friend's house; and
(2) what the three boys did. In contrast, de in (b) and (e) connects larger units. De in (b)
connects: (1) the fact that IY and her friends went to visit a friend in Kanagawa; and (2) the
description of her friend’s father. De in (e) separates the background information—that fact
that three girls and three boys visited their friend's house—from a event in the narrative—
what they did there. Thus, de in (e) marks the beginning of IY's long narrative about that
night.

Examples like these make the roles of sorede and de fuzzy. Furthermore, it is
difficult to clearly differentiate their functions when there are so many variants of sorede.
Therefore, while I agree with Ito that de is used more often in narratives to connect local-
level units (Ito 1995155), I doubt that there are clear-cut functional differences between the
two. The high occurrence of de in narratives is perhaps due to a phonological factor: de is
shorter than sorede and thus easier to say—it saves time to use de when many events must

be connected. In my data, when sorede was used in narratives, it often appeared as a one-

80

syllable variant rule or two-syllable variants, soide or honde, rather than in its full three—
syllable form, sorede, which may support my argument that the shorter form is preferred
when sorede is used over and over again. Another possible reason for the frequency of de
is that, because sorede is used as a causal market more frequently than de, when one wants
to coordinate units without implying causal relationships, de is preferred.

Another factor influencing the alternation between de and sorede parallels
Schiffrin's discussion of the alternation between and and 'zero' (i.e. asyndetic) connection
within a narrative. She (19871130) states that both and and 'zero' can be used to connect
events, but there is a tendency that one or the other is used to separate a general conclusion
from a list of specific events. That is, if a list of specific events has been connected using
and within a narrative, 'zero' is used to mark a segment which is a general conclusion, or
vice versa, to break away from textual regularity (p. 131). The Japanese sorede and de
often seem to function in the same way. That is, if a series of events have been connected
with de, sorede is used to separate a general conclusion from the series of events. If sorede
has been used to list events in a narrative, de is likely to be used with the conclusion.

As I have stated above, there are several reasons to prefer de in speaking: (I) it is
shorter than sorede and takes less time to pronounce, particularly when a series of events
are to be connected, as in narratives; (2) it lacks strong causal implication, and thus is
preferred in situations where causal relationships do not exist; and (3) it lacks the formal
tone sorede carries with it, and thus more common in casual conversations.

My data illustrate that both de and sorede are used to connect either local or global

units, and that functional differences between sorede and de are not clear-cut.

3.5.1.4. Summary of the functions of ( sore )de
In this section, I have examined the functions of (sore )de as a marker of
coordination and speaker continuation. Figure 3.1 summarizes its functions discussed in

this section.

81

General Functions Specific Functions

l I. ideational structure l (1) to connect idea units, events, etc.
(a marker of coordination)

l
(sore )de 1 H. exchange structure l
(speaker continuation) l (2) to connect functionally different units

 

. III. action structure to request the continuation of prior speaker

Figure 3.1 Summary of the functions of (sore )de

Table 3.12 summarizes the function of (sore )de in relation to Schiffrin's

discourse planes. The primary plane for each function is indicated by (*), and the

(2) to coordinate units in cause-effect relations
(1) to hold the floor while searching for words

(3) for tum-entry for co-construction of text

secondary planes are indicated by (+). No mark indicates that a particular plane is not

relevant to the specific function listed on the left or lacks that function.

Table 3.12 Functions of ( sore )de and the discourse planes on which they operate

 

Planes of talk

 

Info. Ideational Participation Action Exchange

functions state structure framework structure structure

 

 

1. sequential coordination *

'X-

2. causal coordination

 

 

3. floor-holding device + +
4. expressing contrast +
5. connect functionally +
differentiated units
6. tum-entry for co-construction + +
7. request other's continuation + + *

 

Note that many of the functions listed above are not discrete ones, but overlap each

other. For example, when (sore )de is used for sequential coordination of idea structures or

as a causal marker, interactionally, it also signals speaker continuation. What makes this

82

overlap possible is perhaps, as Schiffrin contends, because the functions listed above
simultaneously are placed on different planes of talk, such as the ideational structure, the

exchange structure and the action structure.

3.5.2. The adversative marker demo 'but'

In this section, we analyze the second most common connective demo 'but,‘ which
constituted 20% of all connectives used in my data. Demo 'but' is an adversative
connective which functions like the English but in many ways, connecting contrasting
units. We first consider the main functions of demo.

Schiffrin (19871152) characterizes but as a marker of upcoming contrast, and lists

its functions as follows:

’1. referential contrast: ex. My father is tall, but my mother is short.
but:
a contrast 11. functional contrast: l e.g. (1) giving more information than requested:
marker t contrast between A: Do you speak French?

functionally B: No, but I speak Spanish.

different units (2) side-sequencing

 

marker of speaker interrupted, misunderstood, or challenged

i
l
l
1111. contrasting actions: <l(1)point-making1 present points which have been
return 1(2) claiming the floor

Figure 3.2 Functions of but (Schiffrin 1987)

Onodera (1993) analyzes demo 'but', following Schiffrin's classification of but,
though adding two more functions. She states that referential contrast is the contrast
between contraries that are overtly stated in the discourse and "pragmatically inferable
contrast" is the contrast between contraries that are inferred rather than explicitly stated (p.
99). She argues that it is necessary to have "pragmatically inferable contrast" as a separate
category, since Japanese speakers do not generally say exactly what they mean, and

therefore, contrast must be inferred. (20) is Onodera's example of pragmatically inferable

83

COUII’BSI.

(20) Miki: a. Dee anoo/ watashi wa narna de mo ii keredomo
and uh I-TOP fresh even if OK but

b. honto wa anoo/ sukoshi abura de itameta hoo ga ii ka mo shirenai.
actually uh a little oil in saute had better may

c. Demo/ shinnari shitara anmari itamesuginakutte ii kara/ de sore irete
but soft become much overcook-NEG OK so and that put in

Miki: a. And well, though I don't mind fresh (onion),
b. it might be better to saute (the onion) in a little oil.
c. when (the onion) becomes soft, don't overcook it, and put it in'

She states that demo in (c) contrasts inferred "advantage of cooking" vs. "potential

disadvantage of cooking". Figure 3.3 summarizes Onodera's findings about demo.

ll. referential contrast

4 I]. pragmatically inferable contrast

demo:
a contrast III. functional contrast leg. (1) giving more information than requested
marker 1 (2) side-sequencing

 

LIV. contrasting actions 1(1) point-making
i (2) claiming the floor

(3) opening conversation
l(4) topic/subtopic change

Figure 3.3 Functions of demo 'but' (Onodera 1993)

My data on demo largely agree with Onodera's findings. In what follows, I

discuss each function of demo which appeared in my data.

3.5.2.1. Referential contrast
Referential contrast refers to the contrast between two propositions. Of all the uses

of demo in my data, 42.4% had this function. In (21) TA is talking about her garden.

84

(21) TA: a. Anoo natsu wa, natsu wa nee/ kyonen wa sonoo yoku teire shita node
well summer-TOP summer-TOP IP last year-TOP uh often care did because

b. zutto bara ga saitete/
for a long time rose-SUB bloom

c. yuri toka ne/ saiteta n desu ne?
lily etc. IP bloomed BE IP

d. Uun demo kotoshi wa dakara nannimo shite nai nde/
well but this year -TOP so nothing do NEG because

e. nannimo saite nai n desu yo.
nothing bloom-NEG BE IP

TA: a. 'Well in summer, in summer, last year, because I did that
b. roses were in bloom all through the summer and
c. lilies, too, were in bloom.

(1. Well, but this year, because I haven't done anything, so
e. nothing is in bloom, y'know.’

In ((1), demo 'but' is used to contrast TA's garden last summer vs. this summer: (1) last
year, as she tended the garden often, she had many flowers, but (2) this year, she has not
done anything, so she does not have any flowers. Thus, demo 'but' is used to connect

contrasting units in the ideational structures defined by Schiffrin (1987).

3.5.2.2. Functional contrast

Functional contrast is the contrast between functionally different portions of
discourse, and is interactionally motivated. Schiffrin (1987: 159-60) cites an example from
a question-and-answer pair where the respondent gives more information than was actually
requested in order to satisfy interactional demands such as mitigating a face-threatening
situation. In such cases, the contrast is between what was requested and what was not,
and Schiffrin speculates that the reason for giving more information than requested is the
speaker's desire to be socially and informationally cooperative (p. 161). She also states

that her respondents often provided the requested information first, and added a disclaimer

85

utterance prefaced with but (p. 162).

The use of demo as a marker of functional contrast constituted 21.9% in my data.

In (22), I asked NM what kind of work she wanted to do after graduating from college.

(22) Mieko:

NM:

Mieko:

a. Keezai gakubu de/ sotsugyoo shite donna shigoto shitai to omoirnasu ka.
economics department at graduate do and what kind work want to do QT think Q

b. Ee yukuyuku wa/ moshi kanoo de areba
uh future-TOP if possible

c. jibun de jigyoo ga shitai n desu kedo ne.
self by business-SUB want to assert but [P

(1. Demo ima wa ne/ kekkoo onna no hito wa tsubusareru jidai da kara
but now-TOP IP fairly women person-TOP crushed time BE because

e. uun ato okane mo nai to dekinai kara.
uh and money also NEG if cannot do because

a. 'You're an economics major, and what kind of work do you want to do
after graduation?

b. Uh in the future, if possible,

c. I want to have my own business, y'know.

(1. But presently, it is a time when women get crushed (in business), so

e. uh besides, I cannot have my own business if] don't have money.’

Demo in ((1) serves two functions: (1) functional contrast, contrasting the requested portion

of NM's answer (b-c) and additional information (d-e); and (2) referential contrast between

what she wants to do and a possibility of not being able to do it. Demo in (d) operates in:

(1) the ideational structure, connecting two functionally different idea units; and (2) the

exchange structure, signalling the speaker's continuation. There were many cases of demo

serving dual functions, marking referential and functional contrast simultaneously, as in

this example.

3.5.2.3.

Contrasting actions

Finally, we look at demo used for contrasting actions. Schiffrin (19871173) also

86

characterizes but as a marker of speaker-retum. She states that but is used to present points
which have been interrupted, misunderstood and/or challenged (p. 174). She further

states that but is used as a floor-clairning device at a non-transition place, due to the
speaker's strong desire to present his or her points in the middle of other interlocutor's tum
(p. 174). In what follows, we look at five functions of demo as a marker of contrasting
actions, which constituted 35.7% of demo in my data: (1) in point-making in Section
3.5.2.3.1; (2) for claiming the floor in Section 3.5.2.3.2; (3) for topic change in Section
3.5.2.33; (4) for returning to a prior topic of conversation in section 3.5.2.3.4; and (5)

for requesting further information in Section 3.5.2.3.5.15

3.5.2.3.1. Point-making

Schiffrin (1987: 171-174) states that points presented with but tend to be repeated
arguments from prior discourse segments. Onodera agrees with Schiffrin, and presents
examples of demo being used for making "repeated points." While I found many cases of
repeated points made with demo in my data, it is evident that totally new points are also
often made with it. (23) is an example of a "repeated point" being prefaced with demo, and
(24) is an example of a "new point" being prefaced with demo, both from my data. In
(23), IM is describing the trouble she had in the US.

(23) IM: a. Kore gurai no heya nee/ semai heya o hutari de shea suru/
this about GEN room 1P small room-DO two peOple with share do

b. shiteta n desu kara nee
did assert because [P

c. ima kangaetara shinjirarenai desu kedo.
now think when believe-NEG BE but

d. Uun dakara sono toki ni iroiro arimashita nee.
well so that time at various had IP

 

15 In many of these cases, demo was also used to express referential contrast simultaneously.

87

Mieko:

Mieko:
1M:

e. E? konna ni semai heya desu ka?
what like this small room BE Q

f. Moo chotto hiroi karno shinnai.
more a little big may

g. Moo chotto hiroi kanaa/
more a little big perhaps

h. Demo nee/ kekkoo semai mono ga arimashita yo.
but IP pretty small thing-SUB was [P

a. 'A room about the size of this room/ the two of us share the small room
b. I mean we shared, so

c. now, I can't believe we did that, but

(1. Well, so, we had various problems then.

e. What? such a small room as this?

f. Maybe it was a little bigger.

g. Perhaps a little bigger.

h. But it was pretty small, y'know.

In (f-g), in response to my comment of disbelief in (e), IM makes a concession. Then,

using demo in (h), she goes back to the point she had made earlier: the room she shared

with her roommate was small. IM seems to feel strongly that it is important to emphasize

how small her dorm room was, which serves as background information for her later

description of various troubles she had with her roommate.

We now look at an example of a new point being prefaced with demo. I have asked

NM how she feels about the trade problems between the US. and Japan.

(24) Mieko:

NM:

a. Booeki masatsu ni tsuite wa doo omoirnasu ka?
trade friction about -TOP what think Q

b. 800 desu nee/ tte ka atashi mo N ihon-jin da kara na 11 desu kedo/
well QT Q I also Japanese BE because assert but

c. ma Amerika ga kawatte hoshii na to omoirnasu.
well the U.S.-SUB change want IP QT think

(1. Ano tatoeba kuruma no mondai ni shite mo
well for example automobile problem concerning also

e. yappari ano nanka ippootekini nanka Amerika wa Nihon ga
as you know well like one-sidedly like the U.S.-TOP Japan-SUB

88

warui mitaina koto itteru kedo/
bad like thing say-FROG but

f. demo Amerika mo sore narini un doo kana/ koo ganbatte
but the US. also that way yeah how I wonder this try hard

g. jibun-tachi no kuni de tsukutta kuruma o uru doryoku o shite hoshii.
self pl. GEN country in made automobile-DO sell effort-DO do want

Mieko: a. 'How do you feel about the trade friction?
' b. Well, what shall I say, because I'm Japanese,
c. I hope the US. will change.
(1. Well, for example, concerning the automobile problems we have,
e. as you know well, Americans say one-sidedly that Japanese are bad,

f. but I want Americans, what shall I say, to try harder,
g. to make more effort to sell the cars they make.’

In (f), NM introduces the point she wants to make with demo, after a long introductory
remark about her feelings about the trade friction between the US. and Japan (be). This
was the first time the trade issue was discussed between NM and me, so NM was not
returning to a point which was interrupted or misunderstood. In my data, making a new
point with demo was more common than returning to an old point.

Thus, demo can be used to either reiterate an old point after interruption or
challenges, or present a new point one wants to make. Demo as a marker of point-making
operates in the: (1) action structure, asserting the point the speaker wishes to make; and (2)
participation framework, indicating the speaker's strong belief or commitment to the point

presented.

3.5.2.3.2. Claiming the floor

Both Schiffrin (19871170) and Onodera (1993: 128-131) state in their respective
studies that but and demo are used in the middle of another speaker's turn to claim the
floor. In my data also, though the number was small (there were only 5 such examples),

demo was used to claim the floor to defend one's position, challenge the other speaker, or

89

assert one's opinions. (25) illustrates a use of demo to claim the floor to defend one's

position.

(25) YK:

Mieko:

Mieko:

Mieko:

Mieko:

a. Uun nanka saikin anoo ata/ watito atarashii ne? shiritsu no gakkoo nanka
well like lately well new pretty new IP private school etc.

kaikaku ga susumerareteru dcsho?
improvement-SUB undertake TAG

b. Unto hora/ sensee no ebaryueeshon nanka suru yoo ni natte ne?
well y'know teacher GEN evaluation etc. do like become IP

c. E? gakusee ga?
oh student-SUB

(1. S00.
right

e. Watashi shinai hoo ga ii to omou.
I do-NEG better QT think

f. Gakusee no ebaryueeshon/ datte kawaisoo ja nai desu ka sensee wa.
student evaluation because pitiable TAG teacher-TOP

g. Uc kara cheaman no eba
top from chairman eva

h. Demo nce/ iya demoo/ ano sore dc hora ano
but IP no but well that with y'know uh

i. gakusee no niizu? ni atta kyooiku ga sarete kuru n dattara
student GEN needs to match education-SUB do-PASS come if

j. dc maji/ yappari sooyuu/ ano majime ni yatteru gakusee tte yuu no wa
and scri as expected such uh serious do-PROG student QT say one-TOP

k. sonna jibun ga raku ni naru yoona ebaryueeshon kaitari shinai to omoo.
such self-TOP easy become like evaluation write do-NEG QT think

3. 'Well, like, lately, new/ at pretty new private schools and so forth,
there have been some changes taking place, y'know?

b. Well, y'know, they began to do teachers' evaluation, y'know?

c. What? students do that?

(1. Right.

c. I think it's better if they don't.

f. Student evaluation] because, poor teachers!

g. From the top, they have the chairperson's eva

h. But y'know, but with evaluation, if

1. they can improve education to match the needs of their students,

j. and seri/ y'know those/ uh students who study seriously

k. they won't write evaluations so that they will have easy classes, I think.

90

In (h), after I stated that student evaluation of teachers is unnecessary, YK begins her turn
in the middle of my explanation to present an opposing view, using demo twice. In this
example, demo prefaccs the argument YK wishes to present, and simultaneously operates
as a tum-initiator for her lengthy explanations why she thinks student evaluation is
necessary. Demo also functions as a marker of what Onodera calls pragmatically inferable
contrast, contrasting the merits and demerits of doing students' evaluation of teachers.
This kind of demo was seen only between two conversants who are about equal in age and

status.

3.5.2.3.3. Topic change

When demo was used at the beginning of a next speaker's turn, it was often used
for a topic change, and was the most common connective to be used for a topic change in
my data. Demo as a marker of t0pic change was used regardless of the relationship, age,
or status difference between conversants. In (26), WY had been talking about the trouble
he had in the US. As soon as he finished a lengthy description of the trouble he had at the
immigration office when he arrived in the US, he immediately began talking about

another problem he encountered with demo.

(26) WY: a. Aa korc wa antan taru mon dc atu naa to yuu ishiki o motta to omoirnasu nee.
oh this-TOP dismal thing BE FP QT thinking-DO had QT think IP

b. Demo motto nce/ ima saisho no sooyuu shitsumon o sarcta toki ni nc/
but more IP now first GEN that kind question-DO do-PASS when IP

c. nani ga komatta ka to omotte sugu ichiban saisho ni atama ni ukan da no wa
what-SUB troubled Q QT think soon first head appeared NOM-TOP

konna koto ja nai n desu nee.
this thing BE-NEG assert IP

WY: a. 'Oh this is a dismal beginning, I thought, y'know?

b. But something else/ when I was asked that question,
c. what came to mind first was nothing like what I have been telling you.‘

91

Demo in (b) is used for a sub—topic change, though the overall topic is still WY's
troubles in the US. In this function, demo operates in the exchange structure, continuing

his turn and providing more information than requested.

3.5.2.3.4. Returning to a prior topic of conversation
Another function of demo in my data was to mark returning to a prior topic. In this
function, demo was used to preface a question. In (27), YK and I, both J apanesc teachers,

are talking about what we did in class on the previous day.

(27) Mieko: a. Kinoo no kurasu wa doo deshita?
yesterday GEN class-TOP how was

YK: b. Kinoo wa hutsuu dcshita ne. Ippun-kan supiichi yatte, kanji yattc/
yesterday-TOP ordinary was IP one-minute speech did Chinese characters did

Mieko: c. Ja misuterii wa shinakute.
then mystery-TOP do-NEG

YK: d. Shinakute/ sorede paatikuru to yomirnon yatta n desu yo.
did-NEG and particles and reading did assert [P

Mieko: e. Aa yomimono/ are sugoi jikan kakaru dcsho.
oh reading that very time consuming TAG

YK: f. Ee yominono are anoo tango wakan nai desu yone gakusee/ nanka nc
yes reading that uh vocabulary understand-NEG TAG student kind of [P

g. dc shotchuu ura mite/ kotac-awase shite tte kanji dc.
and often back see answer-match do QT feel

Mieko: h. Atashi tobashichatta n desu yone.
I skipped assert IP

YK: i. Atashi mo soo surcba yokatta na.
I also so do if was good FP

j. Demo misuterii wa yatta 11 desu ka?
but mystery -TOP did assert Q

Mieko: a. 'How was yesterday's class?

YK: b. As usual. We did a one-minute speech, and Chinese characters.
Mieko: c. Then you didn't do the "Mystery" activity.

92

 

YK:
Mieko:
YK:

Mieko:
YK:

d. No, we didn't. And we did reading and particles.

e. Oh reading! That's very time-consuming, isn't it?

f. Yes, reading. There are many words the students don't know.

g. So, they often looked at the back side of the sheet when we went over the

answers,

h. I skipped it, y'know?
i. I should have done that, too.
j. But did you do the "Mystery" activity?

In (j), after talking about other class activities, YK asks a question about the "Mystery"

activity which was introduced in the conversation in (c-d). Demo in (j) signals YK's wish

to return to a prior topic of conversation.

3.5.2.3.5.

Requesting clarification or further information

Another function of demo is prefacing a request for clarification or further

information. In this function, demo is used to preface a question. In (28), FY and SH are

talking about SH's friend, who is required to live in a dormitory for the first two years in

college.

(28) SH:

SH:

SH:

SH:

a. Kyoosce na no ichi, ni-nen wa.
requirement IP one two year-TOP

b. De orc/ no tomodachi no tomodachi ga T. Daigaku ittcru toka ittc/
and I GEN friend GEN friend -SUB placcnamc university go etc. say

c. nde sono hanashi o iroiro kiite/ kiitc n da kedo/ un
and that story -DO various hear hear assert but yeah

(1. Ryoo desho? sono hito.
dorm TAG that person

e. Ryoo/ nanka.

dorm well

f. Demo ryoo dc mo koshitsu desho? tabun.
but dorm even private room TAB probably

a. 'Freshmen and sophomores are required (to live in a dorm.)
b. And my friend's fiiend goes to T. University,

c. and I've heard many stories, but yeah.

(1. He lives in the dorm, right?

e. In the dorm, well.

93

FY: f. But even though it's a dorm, it's probably a private room, right?

In (1), demo prefaccs FY's request for clarification from SH about the living conditions of
SH's friend's friend. It seems that FY is testing his hypothesis about the living
arrangements at T. University against the information presented by SH. In this function,

demo operates in the exchange structure (tum-initiator).

3.5.2.4. Demo and co-occurring discourse markers

We have seen 7 functions of demo in this section: demo for referential contrast,
functional contrast, and 5 functions of demo connecting contrasting actions. Demo is
fundamentally a contrast marker, and its main function is in the ideational structure.
However, 35.7% of demo was used for contrasting actions, as an entry-device to claim the
floor and prefacing various conversational moves such as making one's points, both
repeated and new, and challenging the other interlocutor.

As Onodera (19931100) observes, it is difficult for many Japanese speakers to
present contrasting opinions or actions in a straightforward way. This view can be
supported by the fact that many instances of demo in my data were immediately followed
by other discourse markers, in particular with frllcrs and the interactional particle ne, which
have a mitigating or softening effect.

Demo was followed by fillers and the particle ne much more often than (sore )de
'and' or dakara 'so'. In particular, the percentages of hedging or softening expressions,
such as nanka 'like', ma( a ) 'kind of', yappari 'as cxpcctcd', nante yuu no 'what shall I
say,‘ and ne occurring with demo were higher than with (sore )de or dakara. This suggests
that when people use demo, though they may feel a need to express their contrasting
opinions, they also feel it is necessary to somehow justify their insistence or contrasting
behavior, or to mitigate the "rudeness" of having to do so; and one of the means that

speakers employ to mitigate their action is to use softening expressions such as fillers.

94

Table 3.13 summarizes the number of fillers and the interactional particle ne which

immediately followed (sore )de 'and', demo 'but' and dakara 'so'.

Table 3.13 Fillers and the particle ne following (sore )de, demo and dakara

 

 

 

 

expressions following (sore )de 'and' demo 'but' dakara 'so'
the connectives

nanka 'like' 21 (5.5%) 25 (9.8%) 5 (2.5%)
ano(o) 'well' 19 (5.0%) 18 (7.1%) 15 (7.6%)
ma(a) 'kind of 13 (3.4%) 19 (7.5%) ' 5 (2.5%)
yappari 'as expected' 7 (1.8%) 14 (5.5%) 4 (2.0%)
ee(to) 'uh' 6 (1.6%) 1 (0.4%) 2 (0.1%)
sono(o) 'well' 3 (0.8%) 2 (0.4%) 4 (2.0%)
mm 'well' 2 (0.5%) 6 (2.4%) 3 (1.5%)
nante yuu no 1 (0.3%) 8 (3.1%) 2 (0.1%)

'what shall I say?‘

hora 'y'know' --------- 1 (0.4%) 1 (0.5%)
ne 8 (2.1%) 23 (9.0%) 5 (2.0%)
without fillers 301 (79.0%) 138 (54.1%) 152 (76.8%)
total 381 (100%) 255 (100%) 198 (100%)

 

 

3.5.2.5 Summary

In this section, we looked at the functions of the adversative marker demo. It
functions much like the English but, connecting units in contrasting relationships, such as
referential and functional contrasts and contrasting actions. Hedging expressions, such as
nanka 'like' and ma( a ) 'kind of are used with demo to soften the tone and mitigate face
threatening acts, such as presentation of opposing views and challenges. Table 3.14
summarizes the functions of demo and the planes of talk on which each function operates.
The primary plane for each function is indicated by (*), and the secondary planes are
indicated by (+).

95

Figure 3.14 Functions of demo and the discourse planes on which they operate

 

 

 

 

 

Planes of talk
Info. Ideational Participation Action Exchange
functions state structure framework structure structure
1. referential contrast * +
2. functional contrast + *
3. point-making * + + +
4. claiming the floor + + + *
5. topic change + + + *
6. returning to a prior topic * + +
7. requesting further info. * + + +

 

3.5.3. The Causal marker dakara 'so'

In this section, we will look at the functions of the causal marker dakara 'so',
which was the third most frequently used connective in my data. Dakara consists of the
copula do and a conjunctive particle kara 'because' (Morita 19801237). The portion prior to
kara 'because' expresses reasons or causes, and the portion following kara expresses the
result.

Dakara is classified as a consequential conjunction (Martin 19751818) and
sometimes appears in its more complete forms: soredesukara (sore 'that' desu 'is' kara
'because'); soredakara; desukara or ndakara. K enkyusha ’s New Japanese-English
Dictionary lists the following as the translation for dakara: 'so, accordingly, therefore,
consequently, for that reason, on that ground, that is why, so that'.

As a discourse marker, Maynard (1993:95-97) summarizes the functions of dakara
as shown in Figure 3.4. According to Maynard (1993:70-98), as a cause-result marker,
dakara 'so' is used to connect discourse segments [X] and [Y], primarily indicating the
semantic cause-and-rcsult relationship between them. Interactionally, as a marker of

additional explanation, dakara is used to signal a point in discourse where an additional

96

 

explanation about [X] begins. Dakara may also express the speaker's irritation, and signal
that [X] has been already mentioned or is assumed to be understood or self-evident in the
current discourse, so the speaker adds (reluctantly) an explanatory statement [Y] (Maynard
1993188). As an end-of-the-tum marker, dakara signals the speaker's willingness to yield
the turn. As for the distribution of dakara in the above mentioned functions, Maynard
(1993195) states that 57.75% of dakara in her conversation data was used as a cause-result

marker; 36.62% as an additional explanation marker; and 5.63% as an cnd-of-tum marker.

 

 

semantic source: context-bound meaning Modal contextualization
Dakara connects discourse r 1. semantic a cause-result marker
segments [X] and [Y] in
that [X] provides a semantic
and/or interactional reason l a) add explanation
for [Y] to naturally follow. 2. interactional -------------- i b) signal end of turn

c) express irritation for

repeated explanation

Figure 3.4 Functions of dakara 'so' (Maynard 1993)

The data collected for the present study largely agree with Maynard's analysis of
dakara. In the following section, we examine the functions of dakara as: (1) a cause and
effect marker in Section 3.5.3. 1; (2) a marker of supplementary information in Section
3.5.3.2; (3) a tum-yielding device in Section 3.5.3.3; and (4) a marker of repeated

information in Section 3.5.3.4.

3.5.3.1 Dakara as a marker of semantic cause and effect
In my data, 70.8% of dakara was used to connect both local and global units with
cause-effect relationships. (29) is an example of dakara connecting two local-level units in

cause-effect relations. YK is talking about a robbery in her office.

97

(29) YK: a. Nanka kanojo ga kekkyoku hora sotogawa desho?
well she SUB after all see outside TAG

b. ano hito no suwaru/ nee/ suwaru toko ga.
that person-GEN sit IP sit place-SUB

c. Dakara are mitai desu yo/
so that like I?

d. kanojo no mono ga hotondo torareta n ja nai no kashira.
she GEN things SUB mostly stolen TAG I wonder

YK: a. 'Well she sits close to the door in the office, y'know?
b. the place where she sits.

c. So, it looks like
(1. that mostly her things were stolen, I think.

In this example, YK is speculating about why her co-worker’s things were stolen by a
burglar. She says the reason is that her co-worker sits closest to the door and, therefore,
her items were easily accessible to the robber. Dakara indicates that two local units, [YK's
co-worker sits nearest to the entrance (a-b)] and [mostly her belongings were stolen(c-d)],
are in a direct cause-effect relationship.

Dakara also is used to connect global units in cause—effect relationships. This use
of dakara was typically found in narratives. In (30), I asked WY to talk about his most
troubling moment in the U.S. WY starts describing the trouble he had concerning his room
and board fee at the cashier's office, which in the end turned out to be one of his happy
moments in the U.S. The receptionist at the office insisted that he had not paid his fee

when, in fact, he had.

(30) WY: a. Kono hito to hanashite mo dame da kara ne
this person with talk even no good because IP

b. bosu ni hanashi o shiyoo tte itte
boss to talk DO let's QT said-and

c. boku wa hetana Eego nagaramo koo setsumee shita n desu ne.
I TOP bad English although this explanation did assert IP

98

d. Kaikoo ichiban sono josee ga nani o itta ka ttsuttara nee
open mouth first that lady-SUB what DO said Q 1P

e. "Mr. W, I'm very sorry" to osshatta n desu yo.
QT said assert [P

f. "I'm sorry" tte kotoba wa Amerika-jin wa mettani iwanai mon to
word TOP American TOP seldom say QT

omottemashita kara ne.
thought because [P

g. "watashi wa mooshiwakenai to omotteru" tte yuu iikata de ayamararete/
1 TOP sorry QT think QT say expression by apologize-and

h. zenbu kochira de tehazu o shimasu kara tte yuu koto ni narimashite ne.
everything this by arrange DO do because QT say situation became 1P

i. Sore ga kekkatekini wa dakara watashi wa sugoku ureshii koto de.
that SUB as a result TOP so I TOP very happy thing

a. 'Because it's no use talking to this person,

b. (my friend) suggested that we talk to her boss,

c. with my bad English, I explained my situation.

d. What she said as soon as she opened her mouth was

e. she said, "Mr. W, I'm sorry."

f. I had thought Americans seldom say "I'm sorry,"

g. "I'm deeply sorry" she apologized, and

h. she said that she would make all the necessary arrangements.
i. 80, in the end, that turned out to be a very happy moment.‘

In (30), dakara prefaces the coda of the narrative: that was a very happy moment. Dakara

does not simply connect only one event to the conclusion, but connects three events (c)—(h)

as the reasons which made him feel happy, functioning globally over a wide range of the

narrative. The structure of (30) is as follows:

(30)'

dakara (i)

EVENT: WY's trouble at the cashier's office (prior to the narrative)

(a-b) EVENT: he decided to talk to the receptionist's boss

EVENT: the boss listened to his story,

(d-g) EVENT: the boss apologized to him about their mistake,

EVENT: the boss said she would make all the necessary arrangements,
CONCLUSION: he felt happy.

99

Dakara as a cause—effect marker occurs not only within one's turn, but may occur

across speakers. In (31), CY and I are talking about our college classmates.

(31) Mieko: a. 800 yonee/ moo 20-nen da mon nee datte.
right IP already year BE IP but

b. Kao datte saa/ minna hagetari shiteru n ja nai?
face even 1P everyone bald etc. do TAG

c. Onna no ko no hoo ga kawannai yonee.
girl SUB change-NEG IP

OY: d. Soo soo dakara kookoo no doosookai no toki mo sa/
right so high school reunion time also [P

e. onna no ko wa hotondo kawaranai no yone.

girl TOP almost change-NEG IP IP
f. Demo otoko no ko wa muzan ni mo hagetari hutotchattari shite saa
but boy TOP shockingly EMPH. bald fat do [P
g. zenzen dare ga dare da ka wakannai no yo.

at all who-SUB who BEQ understand-NEG IP IP
Mieko: a. 'You're right. It's been 20 years.
b. Their faces, everyone has changed, gotten bald, and the like.
c. Girls don't change much, do they?
OY: (1. Yeah yeah, so, when I went to my high school reunion, y'know
e. most girls haven't changed.

f. But, to my shock, some boys became bald or fat,
g. I couldn't tell who is who.’

After I said that girls have not changed in their appearance in 20-years after graduation, OY
eXpresses agreement in (d), and begins a long description of her high school class reunion.
The units connected by dakara in (h) are: (1) girls do not change much over the years, and
(2) OY could still recognize her female classmates from high school after 25 years, but not
male classmates.
In this example, the reason is given by myself, which was confirmed by OY, and
the result is given by OY. This type of discourse is similar in format to the co-construction

with de discussed in Section 3.5.1.2.4, except that when the other interlocutor begins

100

 

supplementation with dakara, a cause-effect relationship must be present between the

previous and the next speakers' utterances. The structure of (31) is as follows:

(31)' (a-c) REASON: girls do not change much over time
(d) AGREEMENT/ dakara STATEMENT: in her high school reunion
(e) STATEMENT: most girls have not changed
(f) demo STATEMENT: but boys have changed
(g) RESULT: OY could not recognize her male classmates

Thus, dakara together with the expression of agreement, soo soo 'yeah yeah' in this case,
functions as a device to begin one's utterance which is in a cause-effect relationship with
the prior speaker's last utterance. This type of dakara not only operates in the ideational
structures as a connector of propositional units, but also in the exchange structures as a

tum-entry device.

3.5.3.2. Marker of supplementary information

Thus far, we have seen dakara function as a connector of units in causal relations:
{[X: cause/reason] Dakara [Y: result/consequence]}. As has been pointed out by Maynard
(1989b; 1993b), however, dakara also connects units which do not show an apparent
cause-effect relationship: {[X: utterance] Dakara [Y: supplementary information“. In this
function, dakara signals the beginning of a supplementary explanation, reason, or
description, expressing the meanings: 'to add', 'to explain further' or even 'in other
words,’ rather than 'so', that's why', or 'consequently'.

Maynard (1993:77) argues that dakara as a marker of supplementary information is
used to point out that a relevant conversational move [X] has already been made, and that
the speaker wishes to provide further information. She further states that what follows

dakara is given as evidence supporting the previous point (p. 77).

101

3.5.3.2.1. As a marker of "self-initiated" supplementary information
First, we will look at the case of dakara used as a marker of self-initiated

supplementary information, as in (32).

(32) Mieko: a. Dooshite Amerika koyoo to omotta n desu ka?
why the U.S. come QT thought assert Q

IY: b. E? atashi Eebunka na 11 desu kedo/
Oh I English literature BE but

c. uun yappari mukoo de mo Eego wa benkyoo dekimasu kedo

well as you know over there in also English-TOP study can but

(I. uun nanka hanareta toko kara Nihon mitari toka
uh well far away place from Japan see-want and

e. Amerika-jin to iroirona kuni no hito no seekatsu o tooshite
American and various country GEN people GEN life -DO through

f. a Nihon-jin o mitari toka
oh Japanese -DO see and

g. dakara honmono no Eego no ne? hatsuon toka mo kikitakatta shi
so real English GEN IP pronunciation etc. also hear and

h. nitijookaiwateki na koto mo kiso o shi? nante yuu n deshoo/
everyday conversation-like thing also basics-DO what shall I say

i. bunpoo ga doo toka ja nakute/ tomodachi dooshi no kaiwa toka
grammar-SUB how etc. NEG-and friends each other-GEN conversation, etc.

Mieko: a. 'What made you decide to come to the U.S.?
IY: b. Oh? I'm majoring in English literature, and
c. as you know, I can study English in Japan too, but
(1. I wanted to see Japan from a far away place, and
e. through Americans and people from various countries,
f. I wanted to see Japan, and
g. so, real English, I wanted to hear real pronunciation, and
h. everyday conversation also, what shall I say,
i. not just English grammar, but conversations between friends, etc.’

In (1)), IY begins to explain why she decided to participate in the summer English study
Program (d-f). Then, she continues her explanation in (g-i) beginning with dakara. In (g),

dakara is used to connect IY's utterance (d-f) and another reason for coming to the U.S. (g-

102

i), elaborating her statement in (d-f), and responding to my question in (a). Perhaps the
explanation in (g-i) was brought about as a result of IY's realization that additional

explanation was in order. The structure for (32) is as follows:

(32)' (a) QUESTION: request for IY's reason for coming to the U.S.
(b-c) ANSWER: REASON 1: IY's major is English literature;
and she can study English in Japan, but
(d-f) REASON 2: IY wanted to see Japan and Japanese from outside

(g-i) dakara ADDITIONAL INFORMATION: she wanted to study real English

In this function, dakara corresponds to 'to explain further' or 'to add' in English, and

operates in the ideational structure, connecting two idea units.

3.5.3.2.2. As a marker of "other-initiated" supplementary information
Let us now consider the second use of dakara as a marker of supplementary
information, where a supplementary explanation is given in response to the other

interlocutor's utterance. (33) is Maynard's example (1993:77).

(33) B: a. Yoku Chiba iku n da yo ano hito.
often go NOM be IP that person

A: b. Aa ima wa Chiba ni sunderu n ja nai no ka.
oh now TOP Chiba in live NOM BE—NEG NOM Q

B. c. Iya dakara/ uchi to ato Chiba ni N arashino tte tokoro ga atte
no so/therefore home and Chiba in Narashino QT place SUB there is

d. soko ni itoko ga sunderu no ne.
there in cousin SUB live IP IP

e. De nanka Chiba ni geetobooru tomodachi ga iru kara/
and like in croquet friend SUB there is because

f. soo tenki no ii hi wa taitee dakara itoko n chi itte asondete
yes weather SUB good day TOP mostly so/therefore cousin LK home go—and play

103

 

a. 'She goes to Chiba often.

b. Oh, doesn't she live in Chiba now?

c. No, so she's at our house and there's a place called Narashino in Chiba,

d. and that's where my cousin lives.

e. And she has some friends to play Japanese croquet with in Chiba, so

f. yeah, when the weather's good, most of the time, so she goes to my
cousin's place and plays there.‘

57"?!”

In (c) and (f), what follow dakara are not a result of the prior units. In fact, the segments
after dakara in (c) and (f) both explain why B's grandmother goes to Chiba often and what
she does there. Therefore, the only function dakara plays in (32) is to indicate where a
supplementary explanation begins. Maynard (1993:79) provides the following structure
for (33):

(33)' Maynard's interpretation of the structure of (33):
[X]: B: c. No, (my grandmother does not live in Chiba).

dakara
[Y]: B: Grandmother lives at home.
B's cousin lives in Narashino, Chiba.
Grandmother has friends there with whom she plays Japanese croquet.
On days when the weather is good, grandmother goes to the cousin's place
plays there.

n99?

According to (33)', Maynard seems to find both [X] (cause) and [Y] (effect) in B's own
utterance. Unlike (32), in which the supplementary information was provided with the
speaker's own initiative, in (33), B's supplementary explanation seems to have been
triggered by A's utterance about the current residence of B's grandmother. In (c), B first
responds to A's question with iya 'no', and begins the supplementary explanation with
dakara, pointing out that relevant information has already been supplied in (a). Therefore,

if A had not asked about B's grandmother's current residence, B's explanation probably

104

would not have occurred.16 Thus, dakara has two functions in (33): (l) to mark the
beginning of the supplementary explanation; and (2) to point out that relevant information
has already be given.

In (33), we find the prior conversational movement [X] in earlier segments of the
conversation. However, there were cases in my data in which a relevant piece of
information could not be found prior to dakara. In (34), KY has been talking about her

area of study.

(34) Mieko: a. Senkoo wa?
major TOP

KY: b. A, keezai desu.

oh economics BE

Mieko: c. De tokuni keezai no donna koto ni tsuite benkyoo shiteru n desu ka.
and particularly economics GEN what kind things about study do-PROG BE Q

KY: (1. Yooroppa no keezai no ugoki o ima chotto benkyoo shitemasu.
Europe GEN economy GEN movement now a little study do—PROG

Mieko: e. Hai/ uun dooshite keezai o benkyoo shiyoo to omotta n desu ka?
yes uh why economics TOP study do VOL QT thought assert Q

KY: f. A dakara saisho dakara shakaitekina koto ni tsuite kyoomi ga atte/
oh so first so social thing in about interest SUB have-and

g. de hoogaku ni shiyoo ka keezai ni shiyoo ka mayotte/
and law do or economics do or wondered-and

h. keezai o tooshite sekai to tsunagatteru kara to omotte/
economics DO through world with connect because QT thought-and

i. nanka un omoshirosoo da na to omotte/ un.
kind of yeah interesting seem FP QT thought yeah

Mieko: a. 'Your major is?
KY: b. Oh, economics.
Mieko: c. And what area in economics in particular are you studying?
KY: (1. 'I'm studying the economic trend in Europe, etc.
Mieko: e. O.K. Uh why did you decide to study economics?
KY: f. To explain, I was interested in social issues, and
g. I wondered whether I should study law or economics.

 

16 Hudson (1998) states that the English expression Like I said is similar to the Japanese dakara used for
(reluctant) repetition.

105

h. I thought I'd be better connected with the world by studying economics,
i. I thought it's kind of interesting.‘

In (34), as with (33), what follows dakara in (f) is not the result of some prior segments in
the conversation. Rather, the role of each unit is completely reversed: the unit before
dakara is the result, namely that KY is studying economics, and the unit after dakara refers
to the reasons for studying economics, contrary to the canonical use of dakara, in which the
reason or cause precedes the result. The structure in (34) is: {[result/consequence].
Dakara [reason/cause] }.

In (34), it is difficult to find a relevant conversational move [X] in the portion prior
to dakara, since it was the first time KY was asked about the reasons for studying
economics, and there was nothing which indicated her motives prior to this segment.
Therefore, it seems that my question in (e) led to KY's use of dakara in (f). She first
responded to my question with a 'oh' to mark a receipt of information, and began her
explanation about the reason for studying economics with dakara. In this case, the
translation for dakara is more like 'to explain', rather than 'like I said'.

Dakara which marks the beginning of supplementary information operates in the:
(1) action structure, providing additional information; and (2) exchange structure, when

dakara is used as a tum-initiator.

3.5.3.3. Signalling the end of one's turn

In this section, we look at two instances of dakara which are used to signal the end
of the speaker's turn. Maynard (1989az410) states that dakara at the end of the turn or a
narrative signals the end of the current turn, the speaker's wish to yield the turn, or to end
the conversational narrative framework. I first discuss dakara which marks the beginning
of a summary statement, which often occurs in the coda segment of a narrative. In (35),

YU has been describing her happiest experience in the U.S.

106

(35) YU: a. Ichiban ureshikatta koto to ieba/ saisho ni kita toki ni/
happiest event QT say first at came when

b. kuukoo ni borantia yatteru hito ga kitete
airport to volunteer do person-SUB come-and

c. de "Anata Y desu ka" toka itte
and you name BE Q etc. said

(1. de nanka moo kuukoo kara doomu made moo okutte kurete
and like EMPH airport from dorm to EMPH took kindly

e. de sonoato zutto moo zutto nanka ima mo desu kedo/
and after that long time EMPH long time like now also BE but

f. nanka koto aru goto ni/ tatoeba sankusugibingu toka ne/
like event have every for example Thanksgiving etc. IP

g. nanka ima de mo koo koosai o tsuzukete kurete iru to yuu
like now also like relation-DO continue kindly BE QT say

h. nanka komatta koto ga attara itsu de mo irasshai mitaina koto o itte
something trouble -SUB have if whenever come like thing-DO say

kureru n de
kindly and

i. dakara sore ga sugoku yokatta to omoirnasu.
so that-SUB very good-PAST QT think

Mieko: j. Ii desu nee.
good BE IP

YU: a. 'Speaking about the happiest experience,
b. a volunteer person met me at the airport, and
c. she said, "Are you Y?" and
d. and she took me from the airport to my dorm, and
e. since then, even now,
f. at every special occasion, such as Thanksgiving,
g. she has kept in touch with me until now, and
h. she tells me to come to her when I am in trouble, and
i. so, I think that's my happiest experience.
Mieko: j. That's nice!’

In (35), following 21 intonation units, YU finished her narrative with one clause prefaced

with dakara. Dakara in (i) connects the cause (events in (a-h)) with the result (concluding

107

statement in (i)), and prefaces the summary statement which follows and signals the end of
YU's turn.

The second type of dakara occurred not only in narratives, but in conversations and
interviews. After giving reasons, explanations or descriptions, speakers often ended their
turn with an utterance prefaced with dakara or simply with dakara alone as in (36). YK and

I have been talking about student evaluations of teachers.

(36) YK: a. Nihon no gakusee tte/ warito majime ni torikumu n ja nai desu ka ne?
Japanese student QT pretty seriously deal with TAG Q [P

b. Anmari chakashitari suru hito wa inai n ja/
not many make fun of person-TOP exist-NEG

c. sukunai to omoo yo demo/ dakara.
few QT think IP but so

Mieko: d. Atashitachi mo majime ni kakimasu mon ne.
we also seriously write IP

YK: a. 'Japanese students/ they treat (evaluation) pretty seriously, don't they?
b. I don't think there will be many/
c. few students who will treat them lightly, though. That's why (I think

student evaluation of teachers is good.)
Mieko: d. We write out teachers' evaluation seriously, too, don't we?

In (c), there was a definite pause between demo 'but' and dakara, and dakara was
pronounced with the utterance-fmal falling intonation. In both (35) and (36), immediately
after dakara or the utterance prefaced with dakara, I took the turn, commenting on the
narrative and asking follow-up questions in (35), and commenting on how we, as students,
fill out evaluations seriously in (36).

Dakara in this function seems to work just like the English so. Schiffrin

(1987:223) characterizes the function of so as a result marker:

...so indicates that a speaker has reached a point in the presentation of his/her idea
at which a bearer can infer what would come next even if it is not explicitly stated.

108

The joint accessibility of unstated information creates a point of potential closure for
a discourse topic...it is possible for either speaker or hearer to move on to a new
topic of talk. Hence, a potential participation transition...

...so does convey a 'result' meaning even if no result follows, because so instructs
the hearer to recover a conclusion (an inference, a claim) which has already been

presented, or which is otherwise mutually known because of e.g. just-presented
reasons and/or support.

Examples (35) and (36) illustrate that dakara has a function similar to so; it is used at a
point where both a change in speakers and topics is possible, and it signals the speaker's

willingness to turn over the floor to the other interlocutor.

3.5.3.4. Marker of repeated information

Finally, we will consider the function of dakara as a marker of (reluctantly) repeated
information. Maynard (1993:92) states that when dakara is used for this function, the
speaker may convey his or her reluctance or irritation about having to repeat a piece of
information which has already been given earlier or self-evident. There were two tokens of
this type of dakara in my data and, in both cases, dakara was pronounced with an emphatic
tone.l7 Both were in informal conversation. In (37), AR and SK are talking about the

lunch they had had.

(37) AR: 3. Nanka ne/ sutiimu raisu tte no wa sa/ hutsuu no raisu jan taita gohan no/
well IP steam rice QT TOP IP ordinary rice TAG boiled rice GEN

b. sore na noni atashi sore ja nakutte motto nanka
that but I that BE NEG more like

c. raisu/ nanka chaahan mitaina no ga aru jan
rice well fried rice like SUB there is TAG

d. are ga tabetakutte kooyatte yubisashi-nagara demo yatta no ne?
that SUB eat-want this way point while but did IP IP

e. de sutiimud raisu tte yutte/
and steamed rice QT said

\

17 There were other examples of dakara used for repeated information or supplementary information, but
the tone of irritation were not as obvious as with these two.

109

f. a sutiimud raisu tte hakumai datta no ne.
oh steam rice QT white rice was IP IP

SK: g. Hutsuu raisu tte yuttara rice kureta yo.
usually rice QT say rice give 1P

h. Gohan ga tabetakatta no?
rice -SUB eat-want Q

AR: i. Iya chigau/ dakara sono chaahan ga tabetakatta n da kedo.
no wrong so that fried rice-SUB eat-want IP BE but
AR: a. 'Well, "steamed rice" means ordinary rice, boiled rice, right?

b. but I/ not boiled rice, but something else,
c. but there is something like fried rice, isn't there?
(1. I wanted to eat fried rice, and so I ordered, pointing to the rice,
e. and I said "steamed rice,"
f. oh, now I know steamed rice means white boiled rice,
SK: g. I usually got rice when I said "rice."
h. Did you want to eat boiled rice?
AR: i. No, wrong. Like I said/ I wanted to eat fried rice.’

In (d), AR has already mentioned that she wanted to eat fried rice. In (i), AR repeats that
she wanted to eat fried rice, responding to SK's question in (h), with a tone of minor
irritation and annoyance towards SK's slowness understanding her point. In this function,
dakara Operates in the participation framework and action structure, conveying the
speaker's irritation and pointing out that something has already been mentioned or should

be evident from the context.

3.5.3.5. Summary

In this section, we examined the functions of dakara 'so'. Dakara primarily is a
cause-effect marker which connects idea units, indicating that the unit preceding dakara is
the cause of the unit following it. Dakara has additional functions at the interactional level,
marking the beginning of self- or other-initiated supplementary information. Utterance-
final dakara (following a pause and pronounced with utterance-final intonation), or an
utterance prefaced with dakara, marks the speaker's willingness to yield his or her turn and

is often used in the concluding statement of a narrative. It also prefaces a self-evident or

110

repeated piece of information and may convey the speaker's irritation or annoyance for
having to state it again.

Dakara indexes an utterance both to prior and upcoming units, since it connects
units in a cause—effect relationship. Interactionally, dakara indexes an utterance both to the
speaker and the bearer, sometimes expressing the speaker's irritation or annoyance towards
the addressee, prefacing additional information, and signalling the end of the speaker's
turn.

Table 3.15 summarizes the function of dakara presented in this section. The
primary plane is indicated by (*), and the secondary planes are indicated by (+). No mark

indicates that a particular plane is not relevant to the functions listed on the left.

Table 3.15 Functions of dakara on the five planes of talk

 

 

 

 

Planes of talk
Info. Ideational Participation Action Exchange
functions state structure framework structure structure
1. cause-effect marker * +
2. supplementary information + * +
3. concluding statement * + +
4. repetition marker + * + +

 

 

3.6. Summary of Japanese connectives

In this chapter, a quantitative and a qualitative analyses of Japanese connectives
were presented. In the collected data, it was found that connectives were used more often
by older speakers than younger speakers, but the difference was not statistically significant.
Connectives were used significantly more frequently by women than men, which was
probably due to the small rate of connectives use by the younger male participants, who
had quick floor changes compared to older males and female participants. Coordinative

connectives were used more frequently in narratives than in interviews or conversations,

111

 

supporting their characteristics as markers of idea coordination and speaker continuation.
Adversative connectives, demo 'but' and datte 'but, because' were used more frequently in
conversations, probably due to their use for contrastive actions, such as presentation of
opposing views, challenges and defending one's opinions.

It was argued that (sore )de functions as a marker of idea coordination and speaker
continuation, and there were no noticeable functional differences between de and sorede in
my data. The adversative demo was presented as a contrast marker in the ideational, action
and exchange structures. Demo is used for: (1) referential contrast; (2) functional contrast
(contrasting functionally different units); and (3) contrastive actions (point-making,
claiming the floor, topic change, and requesting clarification or further information). Demo
also functions as a marker of speaker's continuation or his/her return to a prior point.
Demo is frequently followed by fillers and the interactional particle ne as hedging
expressions, which may be attributable to Japanese speakers' tendency to hesitate in
presenting contrasting views.

Dakara connects units which have semantic cause-effect relationships, connecting
units in which the unit prior to dakara expresses the cause, and the unit following dakara
expresses result. Dakara is also used to signal the beginning of supplementary
information, to summarize a prior utterance, or to signal the end of one's turn.

All the connectives presented in this section not only operate as connectors of idea
units, but function interactionally in the action and exchange structures and participation
framework, e.g. as a tum-entry or tum-exit device, or as a device to express the speaker's
orientation toward propositions or to other participants in the conversation. Thus, they all
index an utterance to the local context and to the interlocutors, contributing to discourse

coherence.

112

CHAPTER 4

Japanese fillers as discourse markers

4.1. Preliminaries

In this section, I analyze fillers in Japanese. Fillers are a broad range of utterances,
including interjections and adverbs, that typically occur in casual face—to-face interactions or
telephone conversations. They are considered not to carry identifiable propositional
meanings. However, just as connectives function as contextual coordinates with their
indexical functions in ideational, action and exchange structures, coordinating idea units,
speech acts and tum-taking, as described in Chapter 3, fillers have important functions in
the participation framework and the information state, displaying the speaker's belief about
the information possessed by the bearer and the speaker's feelings and orientation toward a
proposition and/or the addressee.

As textual and participant coordinates, fillers make a significant contribution to
discourse coherence through their indexical functions regarding proposition-proposition,
speaker-proposition and speaker-bearer relationships. For example, when ano(o) 'well,
uh' functions as an attention-getter, it indexes an utterance both to the speaker and the
hearer, since it expresses the speaker's wish to get the hearer's attention. It also functions
as a marker of speaker continuation, and thus it is not used at the utterance-fmal position,
unlike some other fillers such as nanka 'like' and ma(a) 'kind of. As a textual coordinate,
ano(o) indexes an utterance both to prior and upcoming texts, separating the two and
highlighting the latter in repairs, disagreement and (sub)topic shifts.

Ee( to) 'uh' marks the speaker's search for an appropriate expression. As a textual
coordinate, it indicates that the upcoming utterance is the result of the search. Although
both ano(o) and ee( to ) can be used as a place-holder while the speaker is searching for
appropriate expressions, ano(o) is more likely to be used in formal settings and to one's

superiors than ee( to ), though there certainly exits individual preferences for one or the

113

other. Thus, the choice between the two depends on the speech situation as well as
personal preference, which is another indication of the functions of ano(o) and ee( to ) as
participant coordinates. A 'oh', aa 'oh' and e? 'oh, huh?‘ index an utterance to the speaker
and to prior text, since they mark the speaker's receipt and management of information.
Thus, although fillers can be and are often used as floor-holding or pause-filling devices
while one searches for appropriate expressions, they also play important roles in
conversation cognitively, socially, and interactionally.

Despite the important roles fillers play in conversation, research on Japanese fillers
has been scarce. Maynard (1989a:30) categorizes fillers into two groups: (1) language-
production—based fillers, which are employed to avoid potential silence or to hold one's
turn during a search for the right expression, and (2) socially motivated fillers, which are
used to make the utterance softer or less impactful.

Takubo (1995, 1996) analyzes several fillers from the perspective of information
management, in particular, the "mental space theory." He states that the "mental space"
serves as the data base for actual utterances, and that fillers express the speaker's
information management processes. Takubo's characterization of Japanese fillers is similar
to Schourup's analysis of English fillers as evincives, which reflect speakers' unexpressed
thought processes or their current mental state (1982: 14).

Koide (cited in Maynard 1989az32) states that "hesitation fillers" such as ee( to) 'uh'
occur most frequently in the following three contexts: (1) when a speaker approaches the
most important point of the message; (2) when a speaker attempts to summarize the content;
and (3) when the speaker designs his or her talk to be situationally appropriate. Koide also
states that hesitation fillers express politeness and soften a statement in conversation.

Thus, Maynard, Takubo and Koide agree that fillers are used not only as a floor-holding
device when smooth speech production is hindered, but also for various other purposes to
produce situationally appropriate utterances.

One of the reasons that fillers should be regarded as more than simple floor-holding

114

devices is that most fillers are not interchangeable with each other. If all fillers are used
simply to keep the speaker's turn or to avoid silence, any filler should suffice in any such
situation. The truth is that many fillers are associated with specific functions. Some of the
functions derive from a filler's original meaning as an adverb, as in the case of nanka 'like'
and ma(a) 'kind of; other fillers may lack propositional meanings but become associated
with particular functions through their routinized use, as in the case of a 'oh', and ee( to )
'uh '.

Thus, without hearing the upcoming utterance, hearers can often infer what kind of
message will follow a particular filler. For example, nanka 'like' is associated with the
speaker's non-assertiveness and uncertainty about a prior or upcoming proposition. Ma(a)
'kind of indicates the downplaying of a proposition by the speaker. Moo 'really' indicates
the speaker's emphatic feelings about the upcoming proposition. These effects are
conveyed by the fillers without speakers overt mention of their feelings. If fillers are
simply a floor-holding device, why must a particular filler be used in a particular situation?
I will show in what follows that fillers are used to indicate the speaker's thoughts or
evaluation processes, and that they should be treated as discourse markers.

In Sections 4.2-4.4, I present quantitative descriptions of all fillers in my native
Japanese speakers' data, and, in Section 4.5, a qualitative analysis of selected fillers. As
with connectives, fillers have several functions simultaneously, primarily in the information
state and the participation framework defined by Schiffrin (1987); therefore, I adopt her

approach in analyzing their roles in conversation.

4.2. Quantitative results
Following Maynard (1989a), I classify fillers' functions into two categories: (1)
language-production-based fillers, which are used as pause-filling devices when a smooth

speech production is either cognitively or productively hindered; and (2) socially motivated

115

fillers, which are used to express the speaker's feelings, emotions, and orientation, such as
hesitancy and uncertainty.

I further classify language-production-based fillers into two categories: (1)
production-related fillers; e.g. ee( to) 'uh', soo desu nee 'let me see', and are 'that thing';
and (2) information management-type fillers, 0 'oh', aa 'oh (I see)‘, an 'yeah', and e? 'oh,
huh?'. Production-related fillers are used when the speaker is trying to recall an expression
or having trouble formulating an appropriate utterance. Information-management fillers are
used to signal a receipt of information, and whether or not the information was anticipated.
Socially motivated fillers are used as part of communication strategies termed "social
packaging" by Maynard (1989b:31)," reflecting the speaker's desire to construct an
utterance in a way agreeable to the recipient in Japanese conversation. Fillers which are
used for interpersonal purposes such as the speaker's hesitation, uncertainty, or intense
feelings are included in the category of social fillers in the present study.

This classification, however, is not a clear—cut characterization of each filler, since
many of the fillers serve more than one function simultaneously. For example, nanka 'like'
is used when the speaker wishes to express uncertainty about the proposition itself and/or
when he or she feels the need to sound less certain because of the speaker-addressee
relationship (e.g. the addressee is the speaker's superior and thus it is desirable to sound
less imposing and definite). Ano(o) 'well' and mm 'well, uh' can also serve either as
production-type or social-type fillers depending on the context in which they occur, and it
is often difficult to distinguish their functions.

The total rate of fillers per 1,000 words was much higher than that of connectives
(30.8/1000 words for connectives vs. 71.4/1000 words for fillers). As regards individual
fillers, ano(o) 'well, uh' was the one most frequently used, followed by nanka 'like' and
moo 'really'. These three constituted 49.7% of all fillers in my data. Table 4.1

summarizes the number of tokens, what percentage of total filler use each one represents,

116

and the rate per 1,000 words of each filler used by the native Japanese participants in order

of decreasing frequency.

Table 4.1 Summary of fillers in the native Japanese data (n=36)

 

 

 

 

marker meaning tokens (%) rate/lOOOwds
ano(o) uh, well 527 (20.7%) 14.7
nanka like 467 (18.4%) 13.0
moo really 270 (10.6%) 7.54
yappari asexpected 201 ( 7.9%) 5.62
ma(a) kind of 189 ( 7.4%) 5.28
0 oh 147( 5.8%) 4.10
koo like this 135 ( 5.3%) 3.77
«1 oh (I see) 105 ( 4.1%) 2.93
mm uh, well 97 ( 3.8%) 2.71
ee(to) uh 94( 3.7%) 2.6
sono(o) uh, well 89 ( 3.5%) 2.49
nante yuu no what shall I say? 86 ( 3.4%) 2.40
an yeah 47( 1.8%) 1.31
e? oh, huh? 35 ( 1.4%) 0.98
soo (desu ) ne( e ) well, let me see 34 ( 1.3%) 0.95
hora there! l8( 0.7%) 0.5
me that 16( 0.6%) 0.45
total 2,541 (100%) 71.4

 

 

 

 

4.3. Sociolinguistic factors

We now look at the rates of fillers per 1,000 words in relation to four
sociolinguistic factors: (1) formality (formal vs. informal); (2) age (older vs. younger); (3)
the speaker's and the addressee's gender (men vs. women); and (4) speech genre
(conversation, interview, and narrative). As with connectives, after the rates were
calculated, in order to see whether or not these factors significantly influence the use of
discourse markers, the unmatched two-tailed t-test was run for formality, gender, and age.

For speech genre, the one-way AN OVA test was performed.

4.3.1. Formality

Speech styles influenced the use of fillers. The total rate of filler use was much

117

higher in formal speech than informal speech; the difference was statistically significant
(P=0.003). Not only the rate but the types of fillers used were influenced by speech styles:
colloquial sounding fillers such as nanka 'like' were used more often in informal speech
than in formal speech, though the differences were not statistically significant. In contrast,
fonnal-sounding ones such as ano(o) 'well', sono(o) 'well', and soo desu nee 'let me see'
were used more often in formal speech than in informal speech; the P-values were 0.0000,
0.019, and 0.020, respectively, indicating a strong relationship between speech formality
and the use of certain fillers. Table 4.2 summarizes the number of tokens, rates/1000
words and the P-value of each filler. A P-value of 0.05 or lower indicates a significant
relationship between formality and the use of fillers and a P—value between 0.05 and 0.1

indicates a strong tendency towards significance.

Table 4.2 The tokens and rates of fillers per 1,000 words by formality (two-tailed t-test; n=36)

 

 

 

 

 

 

marker formal (n=24) informal (n=12) P-value
number of words 25,334 10,454 df=34
tokens rates/1000 wds tokens rates/1000 wds cl=95%

ano(o) 478 1 8 . 9 49 4.69 0.000
nanka 310 12.2 157 15.0 ns
moo 188 7.42 82 7 . 84 ns
yappari 188 7.42 13 1.24 0.000
ma(a) 167 6.5 9 22 2.1 0.044
koo 115 4.54 44 0.57 ns
um: 88 3 .4 7 69 0.86 0.030
ee(to) 87 3 .43 7 0.67 0.046
sono(o) 83 3 .28 49 0.57 0.019
a 79 3.12 68 6 . 5 0.025
m 73 2.88 32 3 .06 ns
nante yuu no 56 2.21 30 2.87 ns

un 39 l .5 4 8 0.77 0.053
soo (desu) ne(e) 33 1 .30 l 0.1 0.020
e? 21 0.83 14 l .34 ns
hora 12 0.47 6 0 . 5 7 us

are 5 0.2 1 1 l .05 0.036
total 2022 7 9 . 8 535 51.2 0.003

 

 

 

 

 

 

df: degree of freedom; cl: confidence level; P5005: significant relationship; 0.05<PSO.1: strong tendency;

ns: not significant

 

The rate of the social-type filler ma(a) 'kind of per 1,000 words was significantly
higher in formal speech than in informal speech (6.59 in formal speech vs. 2.1 in informal
speech; P=0.044), which may suggest that "social packaging" is more prevalent in formal

speech situations than when the participants are speaking informally.

4.3.2. Age

The age of speakers influenced the use of fillers. Older speakers used fillers more
often than younger speakers (85.8 vs. 64.5/1000 words), and the difference was
statistically significant (P=0.05). The most notable difference was in the use of ano(o)
'well, uh' and sono(o) 'well, uh',l which were much higher for older speakers than
younger speakers (P=0.000 for both ano(o) and sono(o) ). The rate of the social filler
ma(a) 'kind of' was also higher for the older participants than the younger participants
(P=0.007). This may be partially due to the fact that the older conversants used fonnal-
style Japanese, in which social packaging is common, while the younger participants' data
include both formal and informal speech. This area must await future research which
includes older speakers' informal Japanese data.

As regards production-related fillers, such as ee( to ) 'uh', soo desu ne( e ) 'let me
see', and nante yuu no 'what shall I say', there was no statistically significant difference
between the older and younger conversants. As for information management fillers,
younger speakers' rates were higher than those of older speakers, and the differences were
statistically significant, except the rates of aa 'oh' (the P—values for a 'oh', un 'yeah', and
e? 'oh, huh?’ were 0.033, 0.007, and 0.028, respectively). Table 4.3 summarizes the

number of tokens, the rate/1000 words, and the P—value for each filler by participants' age.

 

1 Sono(o), like ano(o), is a demonstrative adjective originally.

119

Table 4.3 The tokens and rates of fillers per 1,000 words by age (two-tailed t-test; n=36)

 

 

 

 

 

 

marker older (n=12) younger (n=24) P-value
number of words 1 1,686 24,102 df=34
tokens rates/1000 wds tokens rates/1000 wds cl=95%

ano(o) 306 26.2 221 9.2 0.000
nanka 121 10.4 346 14.4 ns
ma(a) 107 9.2 82 3.4 0.007
yappari 90 7 .7 1 l 1 4.6 0.083
moo 82 7.0 188 7 .8 ns
koo 76 6 .5 59 2.5 0.024
sono(o) 62 5 . 3 27 1.1 0.000
ee( to) 37 3 . 2 57 2.4 ns
mm 29 2.5 68 2 .8 ns

a2 28 2.4 77 3 . 2 ns

a 26 2.2 121 5 .0 0.033
nante yuu no 17 1.5 69 2.9 ns
hora 1 l 0 .9 7 0.3 ns
soo (desu) ne( e ) 5 0.4 29 l . 2 us

an 3 0.3 44 l .8 0.007
e? 3 0.3 32 1 . 3 0.028
are ---------- l6 0. 7 0.092
total 1003 8 5 . 8 1554 64.5 0.050

 

 

 

 

 

 

 

---: no occurrence; P5005: significant relationship; 0.05<PSO. 1: strong tendency; ns: not significant

4.3.3. Gender

My results show that the speaker's gender was not a significant factor in the use of

fillers, contrary to my Hypothesis 2:

Hypothesis 2

The use of fillers will also be influenced by several sociolinguistic factors:
Fillers will be used more often by women than by men, by older speakers than by younger
speakers, and in formal-style speech than in informal-style speech since speakers are more

concerned with producing "appropriate speech" when speaking formally than when
speaking informally.

As shown in Table 4.4, there was not much difference between men and women in the total

rate (67.8/1000 words for men vs. 74.3/1000 for women). The only category which

120

showed a significant difference was the information management fillers a 'oh', un 'yeah',

and e? 'oh, huh?', whose P-values were 0.001, 0.004, and 0.009, respectively. Of other

fillers, the rate of the social filler ma( a) 'kind of was significantly related to gender

(P=0.027), and the rates of nante yuu no 'what shall I say', hora 'therel', and koo 'like

this' showed some relationship to the speakers' gender (P=0.084, 0.084, and 0.099,

respectively). This result shows that Hypothesis 2 is incorrect or inconclusive at best.

Table 4.4 The tokens and rates of fillers per 1,000 words by speakers' gender (two-tailed t—test; n=36)

 

 

 

 

 

 

marker men (n=18) women (n=18) P—value
number of words 15,729 20,059 df=34
tokens rates/1000 wds tokens rates/1000 wds cl=95%

ano(o) 245 l 5 . 6 282 14.1 ns
nanka 156 9.9 31 1 l 5 . 5 ns

moo 123 7 . 8 147 7.3 ns
ma(a) 121 7. 7 68 3.4 0.027
koo 91 5 . 8 44 2.2 0.099
yappari 88 5.6 1 13 5.6 ns

m 45 2.9 60 3 .0 ns
sono(o) 40 2 . 5 4 69 2.4 ns
ee(to) 39 2.48 55 2 . 7 ns

mm 28 1.8 49 3 .4 ns

a 27 1.7 120 6.0 0.001
nante yuu no 22 1.4 64 3. 2 0.084
soo (desu) ne(e) 17 l . l 17 0.9 ns

are 7 0.5 9 0.5 ns

e? 4 0.3 31 l .6 0.009
hora 3 0.2 15 0 .8 0.084
total 1067 67.8 1490 74.3 ns

 

 

 

 

 

 

P3005: significant relationship; 0.05<PSO.1: strong tendency; ns: not significant

 

We now examine the rate of filler usage in relation to the addressee's gender. As
shown in Table 4.5, address's gender was not significantly related to the use of fillers,
except for ee( to) 'uh,‘ soo desu ne(e) 'let me see,’ and nanka 'like' (P=0.041, 0.036, and
0.016, respectively). The rate of koo 'like this' showed some relationship to the

addressee's gender (P=0.08).

121

The total rate of fillers per 1,000 words was higher for both men and women when
they were speaking to a woman than when speaking to a man. As regards the female
speakers, the rates of most fillers, except nanka 'like', moo 'really', and sono(o) 'well'
were higher with a female addressee than with a male addressee, which was contrary to my
expectation that women would use more fillers, especially softeners, when speaking to
men. Male speakers used more fillers when speaking to a female addressee, but the
difference was small (66.3/1000 words with a male addressee vs. 77.5/1000 words with a

female addressee). Hypothesis 2 was incorrect in this regard as well.

Table 4.5 The rates of fillers per 1,000 words by addressees' gender (one-way ANOVA test)

 

 

 

 

 

 

marker M v. M M v. F F v. M F v. F P—value
n=8 n: n=4 n=8
number of words 6260 3316 4018 8546 df=34
rates! rates rates rates cl=95%

nanka 13.7 2.1 14.9 14.6 0.016
ano(o) 13.3 28.6 5.0 18.4 ns
koo 1 0 .4 0.6 1.7 2.6 0.080
ma(a) 8.0 8.4 1.7 5.2 ns
yappari 5.6 4.8 1.7 5.3 ns
moo 4.3 9. 1 l 1 .0 7.4 ns

or 2.24 2.1 3.0 3.2 ns
sono(o) 2.2 5.7 3.0 2.7 ns
nante yuu no 2. 1 1.2 1.5 2.8 us

a 2. l 0.9 5.5 6.6 ns
mm 1.1 2.1 0.3 2.7 ns

an 0.5 0.3 0.7 0.7 ns
hora 0.5 ----- 0.3 l .5 ns

ee( to) 0.2 8 . 8 0.5 1 . 4 0.041
e? 0.2 0.9 0.5 l .3 ns
soo (desu) ne(e) ----- 0.6 0.3 0.4 0.036
are ----- 1 .2 0.5 0.6 ns
total 66.3 77.5 51. 77.1 ns

 

 

 

 

 

 

--: no occurrence; P3005: significant relationship; 0.05<PSO.1: strong tendency; ns=not significant

The use of fillers by the speaker's gender and the addressee's gender indicate that

the use of fillers is not significantly related to gender. This is perhaps due to the fact that

the older conversants spoke in formal Japanese in which the gender difference is generally

122

 

not prominent, and that among the younger speakers, especially among students, little
gender difference is present not only in the use of fillers but in their speech in general, even

when they speak in informal Japanese. This area needs further research.

4.3.4. Speech genre

We now compare the use of fillers in the interviews, casual conversations and
narratives. Speech genres were not significantly related to the total rate of fillers, though
the use of several fillers was significantly related to speech genres.

First, the rate of the production filler ee( to ) 'uh' was the highest in narratives,
where speakers needed to keep the floor for a long time, and the difference was significant
(P=0.007). The the rate of soo desu nee 'let me see,‘ a floor holder, was the highest in the
interviews (P=0.001).

Secondly, the rates of information management fillers were low in narratives,
where speakers did not need to respond to others' utterances. The highest rate for
information management fillers, however, was in the interviews, contrary to my
assumption that information management markers are generally used in response to what
the other interlocutor has said and thus are likely to appear most often in casual
conversations, where two or more interlocutors share similar responsibility in speaking, in
contrast with interviews, where interviewees hold the floor for a longer time than the
interviewer.

Social-type fillers such as ma(a) ’kind of and nanka 'like' did not show a
significant relationship to speech genres. I had anticipated that the rates for social-type
fillers would be highest in casual conversations, where speakers need more social
packaging in responding to others' utterances and presenting their points tactfully, but this
was also proven to be incorrect.

The P-values of the total rate and those of most individual fillers suggest that speech

genre does not significantly influence the use of fillers. Table 4.6 is a summary of the

123

number of tokens, the percentage and the rate/1000 words for each filler in the three speech
genres, with bold numbers indicating the highest rate of the three speech genres for each

marker.

Table 4.6 The tokens and rates of fillers per 1,000 words by speech genre (one-way ANOVA test; n=42)

 

 

 

 

 

 

marker conversation (n=24) interviews (n=24) narratives (n=6) P-
22140 wds 13648 wds 3217 wds Value
token (%) rate token (%) rate token (%) rate

ano(o) 355 (23.1) 16 172 (16.9) 12.6 43 (18.6) 13.4 ns

nanka 278 (18.1) 12.6 189 (18.5) 13.8 49 (21.2) 15.2 ns

moo 164 (10.7) 7.4 106 (10.4) 7.8 25 (10.8) 7.8 ns
ma(a) 129( 8.4) 5.8 60( 5.9) 4.4 21( 9.1) 6.5 ns
yappari 103( 6.7) 4.7 98( 9.6) 7.2 12( 5.2) 3.7 ns
koo 96( 6.2) 4.3 39( 3.8) 2.9 21( 9.1) 6.5 ns

a 94( 6.1) 4.3 53( 5.2) 3.9 6( 2.6) 1.9 ns

sono(o) 68( 4.4) 3.1 21( 2.1) 1.5 13( 5.6) 4.0 ns

01 60( 3.9) 2.7 45( 4.4) 3.3 ---------- 0.076

nante yuu no 47 ( 3.1) 2.1 39 ( 3.8) 2.9 6 ( 2.6) 1.9 ns

ee(to) 44( 2.9) 2.0 50( 4.9) 3.7 24 (10.4) 7.5 0.007

mm 38( 2.5) 1.7 59( 5.8) 4.3 8( 3.5) 2.5 0.008

hora 17( 1.1) 0.8 1( 0.1) 0.07 ---------- as

e? 17( 1.1) 0.8 18( 1.8) 1.3 ---------- ns

un 11( 0.7) 0.5 36( 3.5) 2.6 1( 0.4) 0.3 ns

are 11 ( 0.7) 0.5 5( 0.5) 0.37 ---------- ns

soo (desu) ne(e) 6( 0.4) 0.3 28( 2.8) 2.1 2( 0.9) 0.6 0.001

total 1538 (100) 74 7 1019 (100) 69.5 231 (100) 71.8 ns

 

 

 

 

 

 

 

 

 

--: non-occurrence; P5005: significant conelation; 0.05<PSO.1: strong tendency; ID: indetenninable due
to low occurrence; ns=not significant

4.4. Summary

In this section, I presented a quantitative analysis of fillers. My findings are:

1. Formality was significantly related to the use of fillers: they were used more
often in formal speech than in informal speech (P=0.003).

2. Age also was a significant factor in the use of fillers: older speakers used more
fillers than younger speakers (P=0.05). Ano(o) 'well', sono(o) 'well', ma(a) 'kind of',
and koo 'like this' were used more often by older speakers than by younger speakers,

whereas information management fillers were used more often by younger speakers than

124

by older speakers.

3. Neither the speakers' or addressees' gender was significantly related to the use
of fillers, except for the information management fillers a 'oh', un 'yeah', and e? 'oh,
huh?', which were used more often by women than by men (P=0.001, 0.004 and 0.009,
respectively).

4. Speech genre was not a significant factor to any fillers except the production
markers ee( to) 'uh', soo desu nee 'let me see' and mm 'uh, well'. Ee( to) was used most

often in narratives, whereas soo desu nee and mm were used most often in interviews.

4.5. Qualitative analysis

In this section, in order to show that frllers function as textual and participation
coordinates which indicate the speaker's thoughts or internal evaluation processes, I
analyze eight non-infonnation-management—type fillers first in descending order of
frequency: ano(o) 'well', nanka 'like', moo 'really', yappari 'as expected', ma(a) 'kind
of, koo 'like this', ee(to) 'uh, well' and mm 'well' in Sections 4.5.1—4.9. In Section

4.5.10 functions of information-management fillers is discussed.

4.5.1. Ano(o) 'well, uh': a hesitation marker

Ano(o) was the most frequently used filler in my data with 527 tokens, constituting
21% of all the fillers used. In its original use as a demonstrative adjective, ano 'that' refers
to objects which are away from both the speaker and the addressee. As an anaphoric
demonstrative, it refers to shared information or experience between the speaker and the
bearer and enhances the closeness between the two (cf. Kuno 1973; Martin 1975). Morita
(1989:69) states that ano is also used by a writer when s/he "wishes to make the readers
feel as if they share the same perspective as the writer" [translation mine].

According to Cook (1993:23), as a discourse marker, ano(o) is used in the

following situations: (1) to get the the addressee's attention; (2) when the speaker starts a

125

conversation or a new turn; (3) to highlight a proposition that immediately follows ano(o),
(4) to start a new topic, and (5) when the speaker disagrees with others. She (1993:33)
concludes that ano(o), like its function as a demonstrative adjective, is used to align the
speaker and the addressee on the same side and serves to obtain the addressee's
cooperation in face-to-face interactions.

Takubo (1995:1025) states "ano(o) is fundamentally used when the speaker is
searching for or constructing appropriate expressions directed to an addressee, and that
"ano(o) appears only in face-to-face interaction, not in soliloquy " [translation mine].

My data confirm Cook's and Takubo's observations. Ano(o) is used when the
speaker needs the addressee's cooperation before introducing a new conversational move.
In other words, ano(o) is used when the speaker is about to make a new conversational
move such as presenting an important point or shifting a topic, as if to warn the addressee
about the upcoming conversational move and mitigate its impact.

Ano(o) shares many functions with the English well and uh. Well prefaces a
disagreement (Pomerantz 1984:72) or a dispreferred response (W ootton 1981:67), starts a
new turn (Sacks, Schegloff and Jefferson 1974:32), and prefaces a topic shift or self- or
other-repair (Schourup 1982:53-60). Uh is often used tum-initially, or when the speaker is
searching for an appropriate expression. Schourup (1982:110) states that uh indicates that
the speaker is hesitating to try to drink of the best or most accurate thing to say next, or to
remember something, or is reluctant to say what is to follow.

In what follows, we examine the functions of ano(o) and show that the general
function of ano(o) is to express hesitation before new conversational moves. First, we
review the functions of ano(o) proposed in Cook (1993) in Sections 4.5.1.1-4.5.1.5, and

then consider it additional functions found in my data in Sections 4.5.1.6-4.5.1.7.

4.5.1.1. Getting the addressee's attention

Probably the most common function which the Japanese speakers associate with

126

ano(o) is as an attention-getter when talking to a stranger. For example, in the
questionnaire2 filled out by the participants in the present study, 6/22 (27.2%) of them
associated ano(o) with talking to strangers, getting their attention, while 5/22 (22.7%)
associated it with searching for an appropriate expression. One participant stated that she
uses it when speaking to a person older than herself, suggesting the use of ano(o) in polite
situations. Though native speakers often use fillers without consciously considering their
functions, the result of the questionnaire shed some light on the use of fillers.

Ano(o) also is used before requesting a favor. For example, when students visit a
professor's office asking for a letter of recommendation, they will typically start their
request with ano(o), as ano(o) suisenjoo onegai shitai n desu ga....'ano, may I ask you to
write a letter of recommendation for me.’ In this case, ano(o) serves two functions: (1) to
get the addressee's attention; and (2) to express hesitancy, and, consequently, deference
towards the addressee by delaying the request for a moment. Without ano(o) in this case,
the speaker sounds too forward, as if s/he is requesting a naturally expected service. Ano
in this situation operates in the participation framework, showing the speaker's hesitation
toward the addressee.

In my data, since the recordings started after speakers sat down for interviews,
casual conversations or narratives, perhaps there was no need to call other's attention, no
example of ano(o) as an attention-getter was observed. (1) is an example of ano(o) which
is closest to its use as an attention-getter in my data. Only the tokens of ano(o) which are

pertinent to the specific function depicted in this section are in bold.

(1) KB: a. Nanika hanasanakya ikenai n desho?
something speak-must TAG

b. Shitara/ demo sa/ nanka sugoi wazatorashii to omowanai?
then but IP somehow very artificial QT think-NEG

 

2 See Section 6.6.1 for the results of the questionnaire.

127

c. Ano sa/ koko no sa/ jaa nee/ koko no sa/ Eego no jugyoo tte sa
well IP here-LOC [P then IP here-LOC IP English class JP

KE: a. 'We have to talk about something, right?

b. Then, but, don't you think it's very artificial?
c. Well, here, then, speaking of the English classes here'

KB and Y1 have been asked to converse for the data collection for this thesis. In (a) and
(b), KE is commenting on how artificial it is to talk about something for the purpose of data
collection with a tape recorder in front of them. Then in (c), she changes the topic
completely and begins talking about their summer English courses. Ano seems to be used
for three functions here: (1) as a floor-holder; (2) to signal KE's hesitation before the
introduction of a new topic of conversation; and (3) to get the addressee's attention and

cooperation for the new topic, for which ano nee or ano saa is often used.

4.5.1.2. Prefacing a new speaker's turn
Ano(o) as a marker of hesitation was used at the tum-initial position. In my data,

21 tokens of ano(o) occurred in this position. In (2), I asked IM why she came to the U.S.

(2) Mieko: a. Mawari ni itandesu ka?
around in was BE Q

b. sono nante Amerika toka itte benkyoo shiteru hito toka ga.
uh what the U.S. etc. go study do-PROG person etc.-SUB

IM: c. Ano ne/ amari inakatta desu ne/ mada ne.
well 1P very was-NEG BE IP yet [P

(1. Demo ano kookoo no toki ni anoo Oosutoraria to kookan toka de
but well high school time in well Australia with exchange etc. by

Mieko: a. 'Were there people around you
b. that / what shall I say/ people who went to the U.S., etc. and studied?
IM: c. Well, there weren't many yet.
(1. But uh in high school, well by an exchange (program) with
Australia'

128

In (c), ano occurs at the beginning of IM's turn; she starts her answer with ano ne 'well
y'know'. In this example, since IM is giving requested information, there is no need for
her to be hesitant. Furthermore, since her turn is requested and expected by the
interviewer, she has no need to use ano to hold the floor. Therefore, ano in (2) must be
analyzed as being used for other purposes, rather than as a hedging expression or floor-
holding device. It seems to be used to get the addressee's attention and indicate the
importance of the upcoming utterance. Ano in this function operates in the participation
framework, signalling the speaker's orientation to the upcoming proposition as well as in

the action structure, soliciting the addressee's cooperation.

4.5.1.3. Highlighting a proposition

According to Cook (1993), ano(o) is frequently used to highlight upcoming
information. In my data, ano(o) as a proposition highlighter or point-making was its most
frequently used function (343/527 tokens=65.l%). (3) is an example from my data to

illustrate this function of ano(o).

(3) KY: a. De uun de un motomoto sono daigaku]
and well and yeah originally that university

b. a kookoo toka tsukutta mokuteki ga
oh high school etc. founded purpose-SUB

c. ano gakkai-in no tame no daigaku kookoo toka ja nakutte
well group member for GEN university high school etc. BE-NEG

(1. arm shakai ni yuuigina jinzai o sodateru tte yuu hooshin ga atta kara
well society for useful people-DO raise QT principle-SUB had because

KY: a. 'And, well, and, yeah, originally, uh our university
b. oh the purpose for building the high school and so forth
c. well was not to build high schools and colleges for the (religious) group
members, but
d. well was based on the principle of producing graduates who will be
useful for society'

129

KY is making a point that her university was not built only to benefit the religious group
she belongs to, but to produce people useful to the society in general. In (c) and (d),
before introducing the point, she uses ano to make the addressee ready to listen to her
proposition.

As a marker of point-making, ano(o) was used in one of the following four
positions: (1) intonation-unit-initial position; (2) immediately after the topic marker wa; (3)
after the subject marker ga; and (4) after adverbial phrases or clauses. In my data, 17.9%
of ano(o) was used after an adverbial phrase or clause, such as tokuni 'particularly',
sugoku 'very' or hijooni 'extremely', and 10.1% occurred following the topic marker wa.
It seems that by inserting ano(o) between the introductory portion of an utterance and the
following main information, the speaker tries to focus addressees' attention on upcoming

information.

4.5.1.4. (Sub)topic shifting
The third function of ano(o) discussed in Cook (1993) is the use of ano(o) before a
topic change. In my data also, 29/527 tokens (5 .5%) of ano(o) was used to preface a new

topic or subtopic introduction. (4) illustrates the use of ano(o) prefacing a subtopic change.

(4) KY: a. J a kondo obaachan chi asobi ni oide yo.
then next time grandma house visit come IP

UM: b. Ikitaai.
go—want
KY: c. Tada yo.
free IP

UM: d. Ano koosoku basu de ike n no?
well express bus by go-can Q

KY: e. Un soo soo soo.
yeah right right right

KY: a. 'Then come and visit (me at) my grandma's house.
UM: b. I want to do that.

130

KY:

UM:

KY:

c. It's free.
(1. Well, can I come by an express bus?
e. Yeah, right right.’

Prior to the segment above, KY has described the beauty of the city where her grandmother

lives. Then she invites UM to come and visit her grandmother's house. In ((1), UM asks a

question concerning how to get to KY's grandmother's house, with ano(o) prefacing a

subtopic change.

Example (5) illustrates a major topic change with ano(o).

(5) SH:

Mieko:

SH:

Mieko:

SH:

Mieko:

SH'

Mieko:

a. Amami Ooshima/ obaachan ga mada sunderu 11 de
Amami Oshima grandmother-SUB still live-PROG

b. ichioo maitoshi kaerimasu.
generally every year return

c. Ii desu nee.
good BE [P

d. Hai, sugoku kiree de/
yes very pretty-and

e. Kiree da tte iimasu yone.
pretty BE QT say [P

f. Anoo/ kochira ni kite Eego no kurasu totte
uh here to come English class take-and

g. Nihon no Eego no kurasu to kurabete doo desu ka?
Japanese English class with compare how BE Q

a. 'My grandmother still lives in Amami Ooshima, and so

b. we generally go back there every year.

c. That's nice!

d. Yes, it's very pretty, and

e. People say it's beautiful.

f. Well, you've been taking English classes since you came here,

g. what do you think comparing them with the English classes in Japan?‘

Up to (f), SH and I were talking about his father's hometown, Amami Ooshima, and in (f),

after I made a comment on what SH said in (d), the topic was completely changed with

anoo. Anoo was used here both as a floor holder while thinking up the next question to

131

ask and also to avoid the abruptness of asking a completely different question.

Schourup (1982:49-53) states that well in English often prefaces a topic shift. He
argues that well closes previous discourse and focuses on the following discourse. He
further states that "without well the speaker would be seen as peremptorily changing the
topic without taking leave of the other participant to do so." He further argues that well as
a discourse marker is used to indicate that the present speaker is now examining the
contents of a "private world", and "by evincing covert consultation at the point just before
an abrupt topic shift, the speaker can imply that the shift is a considered one and invites the
interlocutor to fill in some reasonable motive for the shift (p.53)."

Ano(o) in (4) and (5) functions in a similar way to the English well characterized by
Schourup, serving as a warning device for the addressee of an upcoming topic change and

enlisting the addressee's cooperation in the introduction of the new topic.

4.5.1.5. Prefacing a disagreement

The last function of ano(o) discussed in Cook (1993) is for prefacing disagreement.
In my data, 17/527 tokens (3.2%) of ano(o) was used to preface disagreement. Ano(o) in
such a function often followed iya(a) 'no', uun 'well', and ma(a) 'kind of'. There were
cases, however, where ano(o) prefaced a disagreement without other expressions of

disagreement, as in (6).

(6) Mieko: a. Demo hanasu no sukoshi wa narete jishin ga tsuita desho?
but speak NOM a little-TOP accustomed confidence-SUB build TAG

b. Amerika ni kuru to.

the U.S. to come when

KK: c. Ano/ kairnono toka sooyuu no wa dekiru kedo/
well shopping etc. such one-TOP can do but

(1. tomodachi toka to atte hukaku hanasu/
friend etc. with see seriously speak

132

e. sooyuu no wa yappari Nihongo ja nai to mukuchi ni natte'
such one-TOP as you know Japanese BE-NEG tacitum become

Mieko: a. 'But you must have become used to speaking and become confident,
b. now that you have spent some time in the U.S.
KK: c. Well, I can do things like shopping, but

(1. things like meeting with friends and have serious discussions,
e. if I cannot use Japanese, I tend not to say much....'

In answering my question in (c-e), KK does not entirely agree with my assumption that he
has gotten confident about speaking English. He shows his hesitation with ano in. (c). Ano
functions as a mitigator of an upcoming face-threatening act of disagreement, Operating in
the participation framework.

Thus far, we have seen the functions of ano(o) which appeared both in Cook's and
my data. In what follows, we will look at two new functions of ano(o): prefacing repairs

in Section 4.5.1.6 and prefacing quoted segments in 4.5.1.7.

4.5.1.6. Prefacing repairs

Ano(o) was used to preface repair segments. I follow Schiffrin's classification of
repairs (1987:300-303): background and replacement repairs. According to Schiffiin,
background repairs are "subordinate asides which provide information to modify and/or
supplement hearers' understanding of surrounding material"; e.g. background information,
and repetition and paraphrases of what has previously been stated. Replacement repairs are
modification of the propositional information or speaker intentions concerning prior
utterances. Schiffrin states that well is used to preface background repairs, whereas I mean
or rather is used for replacement repairs in English.

In my Japanese data, ano(o) was used for both background and replacement
repairs. (7) is an example of the former, and (8) the latter. In (7), arm is used to preface

additional background information.

133

(7) SH: a. Motomoto uchi no daigaku wa yappari gakkai-in-san ga ooi n de
originally my college-TOP as you know member-SUB abundant and

b. sorede boku mo soo na n de
and I also so assert and

c. ano soo uchi no daigaku wa sooritsusha ga
well uh my college-TOP founder-SUB

d. ano shitteru to omou n desu kedo/ Ikeda Daisaku-san ga taterarete
well know QT think BE but SUB founded

e. ma sono hito no kengaku no seeshin tte yuu ka/
well that person GEN founding school spirit or

f. sooyuu no ni hikarete ikoo tte yuu huu ni
such attracted go QT like

SH: a. 'Originally, there are many religious group members in my college,
b. and I am a member, too, so
c. well uh the founder of my college,
d. well, I think you know, Mr. Daisaku Ikeda founded it, and

e. well the founding spirit, or something,
f. I was attracted to that, and decided to attend'

After giving two reasons, SH makes two incidental comments as regards the founder of his
college prefaced with ano before going on to the third reason for choosing his college. Ano
in this function seems to be used to signal to the bearer the speaker's shift in orientation
between the main information and background or tangential information.

(8) illustrates the use of ano(o) prefacing a replacement repair. KY attended a high

school away from home, and has been describing how her high school life was.

(8) KY: a. Ryoo de hachi-nin gurashi
dorm in eight-people living

b. hutari-beya de/ hachi-nin ga sundete
two-person-room in eight-people-SUB lived

c. de sono hachi/ ano san-nen-kan issho ni kurashita tte kanji de
and that eight well three-year-for together lived like

KY: a. 'Eight people lived in the dorm
b. in two-person rooms, eight students lived,

134

c. and those eight, I mean, for three years we lived together.’

In (c), KY starts saying hachi 'eight', but rephrased her utterance with ano and begins to
talk about how long they lived together. Ano in this example is used to preface a self-
initiated correction.

Not all self-repair involves correction. (9) is an example of ano(o) used for

clarification of the segment prior to anoo.

(9) Mieko: a. Dooshite Amerika ni kita n desu ka.

why the U.S. to came BE Q
YI: b. Soo desu nee/ Eego o manabu tame ni.
well English-DO learn to

c. Wareware wa/ anoo kokusai gakubu wa ichioo
we-TOP well international division-TOP generally

(1. Eego o kiso to shite sono ue ni iroirona keezai/ seeji o yatteku n de
English-DO bases as that top on various economics politics-DO do

Mieko: a. 'Why did you come to the U.S.?
YI: b. Well/ to learn English.
c. We, well/I mean at the International Division, generally
(1. with English as our basis, we study various fields like economics or
politics'

In (c), YI starts his utterance with wareware wa 'we'. Then he restates it with kokusai
gakubu wa 'the International Division' in order to clarify what he means by wareware. The
segment following ano(o) represents more specific and accurate information in the mind of

the speaker. Thus, in (7)-(9), ano(o) is used to alert the addressee of upcoming repairs.
4.5.1.7. Prefacing direct quotation

Another use of ano(o) found in my data is before quoted segments, as shown in

(10) and (11).

135

(10) TA: a. Atashi mo kyone/ ano kotoshi kotoshi kaette kitara
I also last year well this year this year return came when

b. M ga yuu ni/
M-SUB say

c. ano "Bad news!" toka yuu no ne.
etc. say IP IP

TA: a. 'I too, last/ well, this year, when I returned home this year,
b. M said,
c. well "Bad news!" or something, y'know.'

In (10), TA is talking about the problem she had with her house while she was away. She
reports what her house sitter, an English speaker, told her when she returned to her house.
Ano in (c), since the quoted segment is in English, not in Japanese, it is clear that ano

belongs to TA, not to the quoted segment. Ano is used to separate TA's utterance from the

direct quotation. (l l) is another example of ano prefacing a quoted segment.

(1 1) KB: a. Demo sono koojitsu ga dandan tsuujinaku natte kite shimatta n de
but that excuse-SUB gradually work NEG came and

b. sorede ano "Mata iku no?" toka minna kazoku ni iwarete shirnau n de
and well again go Q etc. everyone family by say-PASS so

KE: a. 'But that excuse gradually became unworkable,
b. and everyone in my family asks me, well, "Are you going again?" etc'

In (b) also, it is unnatural to interpret ano as part of the quoted utterance. Therefore, in

both of these cases, arm is used to separate the speakers' own words from quotations.

4.5.1.8. Ano(o) as a floor-holding device
Finally, we will treat the use of ano(o) as a floor-holding device. Consider (l 2)

and (13).

(12) MI: a. N ihon mo Amerika mo kihonteki ni wa anmari kawaranai tokoro ga
Japan also the U.S. also fundamentally-TOP very different-NEG aspect-SUB

136

(13)

WA:

aru naa to omoirnasu ne.
exist FP QT think [P

b. 800 desu nee.
right BE [P

c. Tatoeba boku/ nanka Meridian Mooru aruite te mo
for example I well Meridian Mall walk even

d. ano/ ano/ nante iimasu ka nee/ koo/ anoo Eego baajon no naka o ne/
well well what shall I say IP like this well English version of in-DO IP

e. aruiteru tte iimasu ka ne/
walk-FROG QT say Q IP

f. Nihon no machi no ano Eego baajon no naka o aruiteru tte yuu
Japan GEN town GEN well English version of in -DO walk-PROG QT say

a. 'I feel that many fundamental aspects are similar between the U.S.
and Japan.

. Yes, that's right.

. For example, when i walk around at the Meridian Mall,

. well, well, what shall I say, like, well, in the English version/

. as if we are walking, so to speak,

. I feel as if we are walking in the English version of a Japanese mall'

WOQOO‘

a. De moo uchi wa zutto kyooshi na n desu kazoku ga.
and EMPH. house-TOP for a long time teacher assert family-SUB

b. Dakara anoo nanka maa kyooshi kana toka omotteta n desu yo.
so well like kind of teacher I wonder etc. thought assert IP

a.'There has been a teacher rn my family for generations.
b. So, uh, like, well, (I) should become a teacher, Iwas thinking.

In both (12) and (13), ano(o) occurs as part of a string of floor-holding devices: in (12), as

the first one of the five (ano, ano, nante iimasu ka ne, koo, and anoo 'well, well, what

shall I say, like, well') , and, in (13), as the first of three (anoo, nanka, and maa 'well,

like, well'). A typical use of ano(o) as a floor-holding device appeared before meta-

communication expressions such as nan re yuu no? 'what do you call that' or nan da kke

'what was it?‘ Ano(o) functions as a floor-holding device so that the speaker can keep his

or her turn while searching for an appropriate expression. In my data, ano(o) was often

137

used for more than one function simultaneously; e.g. prefacing a proposition to be

highlighted and as a floor-holder simultaneously.

4.5.1.9. Ano(o): a marker of hesitation for a new conversational move

In what follows, I illustrate the use of ano(o) to mark a change in conversational
moves in face-to-face interactions as well as to function as a floor-holder. First, we
examine the distribution of the various functions of ano(o) discussed above, summarized in
Table 4.7. Of 527 tokens of ano(o), 7% were used exclusively as floor-holding device
while the speaker was searching for the next expression,3 and 0.6% was undeterminable
since another speaker started to talk before the cunent speaker's turn was over. The
remaining 92.4% of ano(o) was used to preface certain conversational moves, such as
introducing of the main portion of an utterance, disagreement, or topic change, though

many tokens of ano(o) were simultaneously used as a floor holder.

Table 4.7 Number of tokens and % of various functions of ano(o)

 

 

 

 

various functions of ano(o) (total=527 tokens)
highlight a point 343 (65.1%) tum-initial 21 (4.0%)
repair: (1) background 39 (7.4%) disagreement 17 (3.2%)
(2) replacement 32 (6.1%) direct quote 6 (1.2%)
floor—holding device 37 (7.0%) undeterminable 3 (0.6%)
topic change 29 (5.5%)

 

 

Ano(o) was used most frequently before introducing the main point of an utterance,
followed by its use before repairs. In most cases, ano(o) occurred after other fillers or by
itself, so that once ano(o) was uttered, the speaker continued his or her utterance without
further insertion of fillers. Only 37/527 tokens (7.0%) of ano(o) were followed by other
fillers.

 

3 I am using the term "floor-holding device" here exclusively for situations where ano(o) was used with
other fillers and/or connectives and the speaker was obviously trying to recall something or having trouble
formulating the next utterance.

138

This is evident when we look at the positional distribution of ano(o) as it occurred
in my data. 43.8% of ano(o) occurred clause-initially, most of which were used to mark
point-making, repairs, topic change, or disagreement. Ano following a connective or
particles was usually used for point making. The positions where ano(o) occurred are

summarized in Table 4.8.

Table 4.8 Positions where ano(o) occurred in utterances

 

 

 

 

position token (%) position token (%)
tum-initial 21 (4.0%) after no (GEN/LDC) 15 (2.8%)
clause-initial 210 (39.8%) after ga (SUB) 12 (2.3%)
after connective 63 (12.0%) after de (LOC) 5 (0.9%)
after adverb 94 (17.8%) after 0 (DO) 5 (0.9%)
after wa (TOP) 54 (10.2%) other“ 11 (2.1%)
before other fillers 37 (7.0%)

total 527 (100%)

 

 

 

What is common to all the functions of ano(o) discussed in this section is that it
marks the speaker's shift in orientation and the introduction of a new conversational move,
e.g. from less to a more important point in an utterance, from one topic of conversation to
another, from the speaker's own speech to a directly quoted segment, or from the main
topic of conversation to tangential utterances such as background and replacement repairs.
Ano(o) separates the prior segment of an utterance from an upcoming segment which is
different in orientation. It signals to the bearer that the speaker is about to present a change
in discourse.

Furthermore, ano(o) serves as a marker of speaker continuation. As such, it
occuned at the utterance-initial and medial positions, but never at the utterance-final
position. Ano(o) is used to indicate that the speaker is searching for an appropriate
expression, perhaps while taking various factors of the conversational situation into

account, such as the relationship between the speaker and the addressee, the content of the

 

4 This category includes adjectives, adjectival phrases, and prepositional phrases.

139

proposition, or the knowledge and meta—knowledge of both participants. Ano(o) followed
by several other fillers, such as anoo nanka nante yuu ka 'well, like, what shall I say'
obviously signals the speaker's wish to continue and his/her effort to search for an
appropriate expression. Even when ano(o) was used for other purposes in my data, such
as highlighting a proposition, the speaker was generally not interrupted while ano(o) was
being used, suggesting the simultaneous function of ano(o) as a floor-holder.5

Ano then serves four purposes: (1) holds the floor in the conversation while the
speaker is searching for an appropriate expression; (2) separates the prior and upcoming
portions of an utterance when the speaker wishes to present a change in the utterance (e.g.
highlight a proposition, topic shift, or disagreement); (3) calls the attention of the addressee
to an upcoming proposition, as well as soliciting the addressee's cooperation; and (4)
shows the speaker's hesitation or reluctance (e.g. in topic change and disagreement),
mitigating the possible face threatening acts brought about by the new conversational move.

Again, in many cases, ano(o) may serve more than one function simultaneously.

4.5.1.10. Summary

In this section, I examined the functions of ano(o). Most uses of ano(o) should not
be simply treated as a floor—holding device, but as a discourse marker which indicates the
speaker's orientation shift and a new conversational move, such as point-making and topic
change, though ano(o) can also function as a floor-holder simultaneously in any of these
situations. As such, ano(o) operates primarily in the participation framework and the
exchange structure proposed by Schiffrin (1987).

I also contended that ano(o) may signal to addressees the speaker's hesitation when
presenting a new conversational move and enlist the addressee's cooperation, perhaps

mitigates a face-threatening act brought about by the speaker's orientation shift in the new

 

5 Ano(o) as a floor-holder in these situations was suggested to me by Professor M. E. Hudson.

140

conversational move. Figure 4.1 summarizes the general and specific functions of ano(o)

presented in this section.

general functions specific functions

l I. place-holder l. to hold the floor while searching for an
appropriate expression

 

ano(o): II. hesitation marker 2. to preface (sub)topic shift

a hesitation 3. to preface a request

marker before

a new i 111. to call addressee's 4. to preface a repair

conversational attention to the 5. to preface a quoted speech

move upcoming proposition 6. to highlight an upcoming proposition

IV. to enlist addressee's 7. to preface a new speaker's turn

cooperation and/or 8. to preface disagreement
mitigate FTAs

Figure 4.1 General and specific functions of ano(o)

4.5.2. N anka 'like': a marker of hesitation and uncertainty

Nanka was the second most common filler in my data, constituting 18% of all the
fillers in my data with 467 tokens. Nanka derives from nani 'what' and the question
particle ka, and originally means 'something, somehow.‘ Saito (1992:53) characterizes the
discourse marker nanka as a softener, a tum-initiator, and a floor-holder, which is used to
express vagueness or the speakers' uncertainty.

Nanka 'like' in my data primarily filled the following functions: (1) as a floor-
holder while the speaker was searching for an appropriate expression; (2) as a tum-initiator,
(3) to preface the main points of an utterance; and (4) to preface a (sub)topic shift. Nanka,
like ano(o), has two general functions: (1)as a floor-holder while searching for an
appropriate expression, a production-related function, and (2) as a hesitation marker, a
socially motivated function. Nanka as a socially motivated filler is used based on the

following two factors: (1) uncertainty about the proposition itself, that is the speaker is not

141

certain about the proposition due to lack of knowledge or memory lapse; and (2) the
speaker-addressee relationship, as when the speaker expresses hesitation towards certain
addressees, such as one's superior, creating an agreeable environment for the interaction.
I discuss the use of nanka: (l) prefacing a new speaker's turn in Section 4.5.2.1;
(2) prefacing a topic-shift in Section 4.5.2.2; and (3) as a floor-holder in Section 4.5.2.3.

In Section 4.5.2.4, a summary of the functions nanka is presented.

4.5.2.1. Nanka prefacing a new speaker's turn

32 tokens (6.9%) of nanka were used prefacing a new speaker's turn.

(14) SH: a. Amerika-jin wa yoku sedai o koete hanasu ne.
American-people-TOP often generation-DO cross speak IP

FY: b. Nanka! chitchai kodomo to saa/
like small child and [P

c. kookoosee no niichan toka hanashiteru mon.
high school student young man etc. speak

SH: (1. Un tanoshisoo ni ne.
yeah happy seem IP

SH: a. 'Americans often communicate with each other, across generations, don't
they. ,
FY: b. Like, (we often see) young children and
c. high school students talking to each other.
SH: d. Yeah, looking very happy, right?‘

(15) Mieko: a. Tokyo to Osaka kurabete doo desu ka?
Tokyo and Osaka compare how BE Q

b. kookoo wa Osaka de daigaku wa Tokyo desu yone.
high school-TOP Osaka in college-TOP Tokyo BE IP

KY: c. Nanka kookoo no toki wa iroiro kisoku ga kibishikatta kara/
like high school time-TOP various regulations-SUB strict-PAST because

(1. ano anmari soto e derarenakute
uh often outside go-NEG

142

Mieko: a. 'How do you feel, comparing Tokyo and Osaka?
b. Your high school was in Osaka, and college is in Tokyo, right?
KY: c. Like, in high school, there were various strict regulations, so
(1. uh we couldn't go out much.‘

In (14), immediately after SH finishes his utterance with ne, FY begins his turn with
nanka, and gives a specific example which supports SH's prior statement. In (15), KY
begins her answer with nanka. In these examples, nanka cannot be regarding as a softener
or hedging expressions towards the addressee. Particularly in (14), since SH and FY are
good friends and FY 's utterance is a support for SH's prior statement, FY has no need to
soften his statement or try to be vague in relation to SH. Nanka therefore is used to
express uncertainty about what the speaker is about to say to convey his/her observation
about Americans communicating across generations: 'I don 't know exactly why this is
but. . ..'

Nanka in (14) and (15) also seems to be used to avoid an abrupt beginning of the new
speaker's turn. Two of the participants who filled out the questionnaire (see Section 6.6.1
for the entire results) stated that they use nanka at the beginning of an utterance without
attaching any meaning, and stated that they use it when they do not want to emphasize
something, in other words, to make a statement vague. Nanka in (14) and (15) is used
perhaps to avoid sounding too abrupt or definite.

Not only is nanka used to preface a new speaker's expected turn as in (14) and

(15), but it can also be used as a tum-entry device, as shown in (16).

(16) KO: a. N ioi kagu to ne/ Muhi no nioi ga suru n da yo.
smell smell when [P Muhi smell-SUB do BE IP

SM: b. Tabete miru?

eat try
YA: c. E? ii yo.

what OK IP
KO: (1. Nanka sa/

like IP

143

SM: e. Irna made tabeta arne no naka de ichiban oishiku nai to omou.
now until ate candy among most tasty-NEG QT think

f. Kuchi ni irete mireba?
mouth in put in try

KO: g. Nanka ne/
like IP

SM: h. tabemono da kara.
food BE because

KO: i. Irna nioi dake de mo kekkoo kiteru.

now smell only fairly come
KO: a. 'When I smell the candy, it smells like the insect cream Muhi.
SM: b. Do you want to try some?
YA: c. What? Oh that's OK.
KO: d. Like?
SM: e. This is the worst tasting candy of all the ones I've tried so far.
f. Why don't you put one in your mouth?
KO: g. Like?
SM: h. It's food, so.
KO: i. Now, I can smell the candy pretty strongly.

After KO has told YA how bad the candy smells, SM urges YA to try one in (b). YA
declines the offer in (c). KO tries to start her turn with nanka sa 'like, y'know', but SM
continues her turn and further urges YA to try the candy. In (g), KO tries again to enter the
conversation with nanka nee 'like, y'know', and again is ignored by SM, who continues
her turn. Finally, KO gains a chance to begin her utterance in (i). Since the first two tries
by KO to enter the conversation were ignored by SM, we do not know what she was going
to say after nanka. When she finally succeeds, it is by commenting on the candy's smell.
Although both attempts failed, nanka in this example was used as an entry device in the
conversation. Nanka lacks the forcefulness which ano(o) 'uh, well', demo 'but' and the
final particle nee have as a tum-initiator since it expresses uncertainty and vagueness. This

may have contributed to the failure of KO's attempts to enter the conversation.

144

As a tum-initiator, nanka operates in the participation framework, expressing the
speaker's feelings about the proposition or the addressee, in the exchange structure, and in

the action structure indicating the speaker's willingness to enter the conversation.

4.5.2.2. (Sub)topic shifting
Nanka was also used before topic and subtopic shift. (17) is an example of a
subtopic shift. WA and MI have been talking about casual attire worn by some Americans

on Fridays.

(17) WA: a. Shichoo-san toko no B-san nanka mo
mayor Mr. place GEN Mr. also

MI: b. Tii-shatsu mitaina no kitemashita kara ne.
T-shirt like one wear-PAST because IP

WA: c. Bikkuri suru yoona ne/
surprise do like IP

d. dokka no ojisan ga arawareta yoona kanji deshita yone.
somewhere middle-aged man-SUB appeared like seem was 1P

e. Aayuu tokoro wa nanka omoshiroi tte yuu ka/
such aspect-TOP like interesting QT say Q

MI: f. Nanka kigyoo demo nanka sooyuu no ga toriirerareteru mitai desu ne
like business even like such that-SUB incorporate-PASS like BE IP

g. kinyoobi toka wa moo kanzenni jibun no sukina huku de kite ii toka in.
Friday etc. TOP really completely self POSS like clothes in come may etc. QT

WA: a. 'People like Mr. B from the Mayor's office

MI: b. He was dressed in something like a T-shirt.

WA: c. A kind which shocks us, right?
(1. He looked like a strange middle-aged man out of nowhere, didn't he?
e. That kind of custom is kind of interesting, or what shall we say?

MI: f. Well, that kind of custom seems to be adopted in business also, y'know?
g. It seems that they have rules which say that they can come (to the office)

dressed as they please.

After talking about the clothing a mayor's staff member was wearing on a Friday, MI
begins to talk about the business world adopting the same kind of custom in (f). With

145

nanka in (f), the subtopic shifted from what they actually experienced at the mayor's office
to what MI has heard or read about private companies. Nanka in (f) expresses the
speaker's non-commitment due to uncertainty about the information he has. As such,
nanka was often accompanied by expressions such as ki/kanji ga suru 'seems like, feels
like' and mitai or yoo 'seems, appears', which also indicate the speaker's uncertainty or
speculation. Nanka thus operates in the participation framework, signalling the speaker's

uncertainty or non-commitment to the proposition.

4.5.2.3. As a floor-holder
Nanka occurred as a floor-holder as well. In (18), TA is talking about her college

classmates after graduation.

(18) Mieko: a. Sotsugyoo shite sugu kekkon shita hito hotondo imasen nee.
graduation do soon marry did person almost exist-NEG IP

TA: b. J. wa warini hora nanka/
J.university-TOP fairly y'know like

c. J. wa nanka/
J .Univ-TOP like

(1. joshikoo kara suisen tte yuu no ga ooi rr desu yo/ un sugoku.
girls' school from recommended one-SUB many BE IP yeah very

Mieko: a. 'There is almost no one who got married right after graduation.
TA: b. My university is, y'know, uh
c. there are, well,

(1. there are many female students who come from girls' high schools with
the school's recommendation'

In (b-c), TA uses nanka twice and hora 'you know what I mean' once while she searches
for an appropriate expression. Nanka serves two functions here: (1) to hold the floor while
TA searches for an appropriate expression; and (2) to express uncertainty as to how exactly
to characterize her university, operating in the exchange structure and the participation

framework. (19) is another example of nanka as a floor-holder.

146

(19) KM: a. Detoroito itta n desu kedo/

Detroit went explain but

b. uun nanka/ nanka/ nanka un nanka ayashii machi da na toka
well uh uh uh yeah like strange town BE 1P etc.

c. nanka koo sakaete/ sakaete nakute/ kikenna machi tte kanji ga suru.
like this flourish flourish NEG dangerous town QT feel-SUB do

KM: a. 'We went to Detroit, but
b. well, like, like, like, yeah like, kind of a strange town,
c. well uh it's not flourish, flourishing, and looks like a dangerous town.‘

In (b), KM uses four nanka before he finally comes up with his impression of Detroit. The
instances of nanka in (18) and (19) all have the dual functions of expressing the speaker's
uncertainty as to how exactly to formulate the utterance, and of maintaining one's turn
while he or she searches for an appropriate expression. Sometimes nanka in this function
was followed by expressions such as nante yuu no 'what shall I say' as nanka nante yuu
no? 'well, what shall I say?’ or koo 'like this', as nanka koo 'like this', which intensifies
the feeling of uncertainty even more.

Nanka was used with other fillers as in (18) and (19)—directly preceded by hora in
(18b) and mm in (19b) and followed by koo in (19c)—but was also used by itself as a place-

holder and as an expression of uncertainty, as shown in (20).

(20) YU: a. Sonoato zutto moo zutto/
after that long time really long time

b. nanka ima mo desu kedo
uh now also BE but

c. nanka koto aru goto ni/ tatoeba sankusugibingu toka ne/
uh event have every for example Thanksgiving etc. IP

d. nanka Iisutaa toka aru goto ni
like Easter etc. have every time

147

e. nanka ima de mo koo koosai o tsuzukete kurete iru to yuu
uh now also like relation-DO continue kindly QT say

YU: a. 'For a long time since then,
b. like, even now
c. like, at every special occasion, such as Thanksgiving,

(1. like, and Easter, on every such occasion,
e. like, even now she has kindly been in touch with me'

YU uses nanka at the beginning of each intonation unit in (b—e). Nanka is used to maintain
her turn, and at the same time it expresses her uncertainty about why things are happening
the way they are. Nanka operates in the participation framework, indicating the speaker's
uncertainty about a proposition and in the exchange structure, signalling her intention to

continue speaking.

4.5.2.4. The general functions of nanka

In Section 4.5.2, we examined three situations where nanka 'like' is used as a
hesitation and uncertainty marker. Nanka is used: (1) as a tum-entry device or prefacing a
new speaker's turn; (2) to preface a (sub)topic shift; and (3) as a floor-holder while
searching for an appropriate expression. Nanka expresses the speaker's feelings of
hesitation and uncertainty in a given situation, as if to say: I don’t know exactly, but...‘

I propose that nanka is a marker of the speaker's evaluation of his or her knowledge
about the proposition and/or the speaker's assessment of the relationship between the
addressee and him/herself. Three participants I interviewed stated that they use nanka to
soften the tone, even when they are certain about a proposition when speaking to certain
addressees (e.g. to a person who is older than the speaker). In other words, nanka is used
to signal their reluctance to express what is in their mind straightforwardly in certain
situations. In other cases, nanka is used as a production-based filler, expressing the
speaker's inability to immediately formulate exactly what s/he wants to express due to

reasons such as inaccurate information or memory lapse.

148

The use of nanka thus signals to the addressee that the speaker has gone through an
evaluation process regarding his or her knowledge about a proposition and the speaker-
addressee relationships, and, as a result, underscores the fact that what is stated with nanka
is an approximation of what is in his or her mind, which results in the tone of uncertainty,
vagueness and hesitation. The use of nanka is particularly suited to conversation where the
speaker has to express him/herself without being able to spend enough time to carefully

formulate what to say.

4.5.2.5. Summary

In this section, we examined the functions of nanka '1 ike', which is summarized in

Figure 4.2.
general functions specific functions
l I. floor-holder l. to hold the floor while searching for an
appropriate expression
nanka: H. hesitation 2. to preface a (sub)topic shift
a marker of marker 3. to preface a new speaker's turn
the speaker's i 4. to initiate a turn
uncertainty &
non-commit- III. express 5. to indicate lack of knowledge or memory lapse
ment uncertainty or 6. to indicate the speaker's unwillingness to
vagueness appear too definite and confident

 

Figure 4.2 General and specific functions of nanka

4.5.3. M 00 'really': a marker of excess

Moo was the third most common frller in my data, constituting 11% of all the fillers
with 270 tokens. It originally is an adverb, meaning 'really'. As a discourse marker, it is
used when the speaker thinks something is beyond his/her expectations and expresses the
speaker's feelings of "excess" about something in comparison with the expected amount

s/he assigns it. In (21), WA has been talking about how Americans value their informality

149

and individualism.

(21) WA: a. Kuruma mo soo desu ne, Nihon wa koo pikapika ni shiteru n desu yonee.
car also so BE IP Japan-TOP this shiny do assert [P

b. Amerika wa koo nante yuu ka moo/ moo hashirereba ii tte yuu kanji de ne/
the U.S. -TOP this what shall I say really really run if OK QT impression IP

c. ano chotto ya sotto de wa moo/ nanka]
uh a little by -top really like

(1. daibu sutanto-man ga tsukatta yoona kuruma de mo ne/
fairly stunt man-SUB used like car even if IP

e. amari ki ni shinai tte yuu ka
very mind do-NEG QT say Q

WA: a. 'We can say the same thing about cars also. Japanese keep their cars shiny.
b. In the U.S., what shall I say, really it really looks like if a car runs, it's
OK.
c. well, with a little damage really, or rather,
(1. even a car which looks like it was used by a stunt man,
e. they don't seem to mind driving it.

In (a) and (b), WA is contrasting the situation in Japan and the U.S. He uses moo three
times to express his feelings regarding how different the situation in the U.S. is from his
expectation, and that Americans drive cars whose appearance is unimaginable in Japan.

(22) is another example of moo.

(22) TT: a. Bokura wa bebii buumu no gakusee ga ichiban ookatta tokina mon de
I—pl.-TOP baby boom student-SUB most many time BE

b. ano hito kurasu nan nanka iroiro riyuu o tsukete huyashita n desu yo.
uh one class somehow various reason-DO attach increase-PAST BE [P

c. Soshitara moo gakuryoku no reberu ga zudoon to sagatte shimatte
then really academic level-SUB onom. lower do-PAST

TT: a. 'We belong to the time when there was the largest number of students
even during the baby-boom age, and
b. well, our high school decided to add another class for various reasons.
c. As a result, really, our academic level went down drarnatically'

150

TT attended a very prestigious high school which prided itself for academic excellence. In
(c), T'T expresses his dismay and disappointment with moo about how different the level of
his school was from his expectation.

In both (21) and (22), moo is used to emphasize the speakers' feelings about
something being excessive, much beyond their expectations, or something beyond a
"normal" range for better or worse. Moo operates in the participation framework, showing
the speaker's orientation or feelings about a situation or proposition, and, as a result, plays
a role in drawing addressees' attention to an upcoming proposition and involves them in the

conversation.

4.5.4. Y appari 'as expected': a marker of expected results

Yappari 'as expected' was the fourth most common filler in my data, constituting
7.9% of all the fillers used with 201 tokens. Yappari is an informal variant of an adverb
yahari (K ojien Dictionary). It has an even more informal form, yappa. Kenkyuusha's
New Japanese-English Dictionary lists the following translations for yahari: (1) also, as
well, likewise; (2) still, all the same; (3) after all; and (4) nevertheless. Morita (1980:453)
states that yahari is used when "the present situation does not differ from the speaker's
expectations" [translation mine].

In my data, yappari occurred at the utterance-initial-, medial- and utterance-final
position. As a discourse marker, yappari is used for the following functions: (1) to
confirm that something matches the speaker's expectations; (2) to point out to the addressee
that they have shared knowledge of something; and (3) to indicate that something matches
widely accepted expectations; e. g. by society in general. In order to use yappari in
discourse, speakers go through an evaluation of the situation or proposition, based on the
expected norms set by the speaker, addressee, and society, or based on past experiences.

(23) illustrates the use of yappari based on a widely accepted expectation.

151

(23) Mieko:

YI:

Mieko:
YI:

a. D00 desu ka? Eego no kurasu wa. Nani ga ichiban nobimashita?
how BE Q English GEN class-TOP what-SUB most improve

b. Soo desu ne/ yappa supiikingu ga ichiban/
let me see as expected speaking-SUB most

c. supiikingu to risuningu ga kanari/
speaking and listening-SUB considerably

d. ano Doitsu-jin toka Kankoku-jin to issho no kurasu na n de
well Germans and Koreans with together class

a. 'How are your English classes? What improved the most?
b. Well, as expected, speaking (has improved) the most/
c. speaking and listening (improved) considerably,

well we (study) together with Germans, Koreans, etc.’

In (b), YI states that his speaking and listening abilities improved the most since he came to

the U.S. He uses yappa, an informal variant of yappari, indicating that these two areas are

generally expected to improve the most by living in a country where the language is

spoken.

In (24), SH has been talking about how he has gotten used to expressing himself in

English since he came to the U.S.

(24) SH:

SH:

a. Boku saisho ni kotchi kita toki wa/
I first at here came time-TOP

b. toriaezu "hello" to shika iwanakatta n desu kedo/
at first hello QT only say-NEG-PAST assert but

c. iroiro yappa aru ja nai desu ka/ ai/ nan te yuu n desu ka/
various as expected exist BE-NEG BE Q gree/ what QT say BE Q

d. koo aisatsu kotoba mitaina no ga/ sooyuu no 0 sukoshi zutsu oboete
this greeting expression like one-SUB such ONE-DO a little by little learn

a. 'When I came here first,

b. "Hello" was the only thing I could say, but

c. there are various, y'know what I mean, gree/ what are they called
(1. well, greeting expressions or something/ I learned those a little by little'

In (c), SH seems to have trouble formulating exactly what he wants to say, and says,

152

yappa aru ja nai desu ka 'there are, you know what I mean.‘ Yappari in this context is used
to enlist addressees' participation in the conversation. In other words, yappari makes
addressees feel as if the speaker is talking about what everyone should know, and thus,
addressees feel obliged to try to guess what the speaker is trying to say, rather than
passively listening to the speaker. Yappari enhances addressees' involvement and
contributes to improved rapport between speakers and addressees as well as serving as a
pause-filling device while the speaker searches for an appropriate expression.

Thus, yappari operates on three planes of talk defined by Schiffrin (1987): (1) in
the participation framework, expressing the speaker's judgement or feelings about a
situation or a proposition; (2) in the information state as regards knowledge and meta-
knowledge of the interlocutors; and (3) in the action structure, enlisting addressee's

cooperation.

4.5.5. Ma( a ) 'kind of' : a marker of downplaying the effect

We now turn to ma( a) 'kind of'. Ma(a) was the fifth most common filler in my
data, constituting 7.4% of all the fillers used with 189 tokens. Kenkyusha's New
Japanese-English dictionary gives the following translation of ma( a ): 'well, I think, I
should say/think, it would seem, and something like that.‘ Kojien Dictionary lists ma(a) as
an adverb which indicates that something is more or less acceptable. Morita (1989:1037)
states that ma( a ) is used to express the speaker's acceptance and resignation that something
is the best possible choice or situation at a given moment, and thus may convey a sense of
uncertainty and dissatisfaction [translation mine]. It suggests that the speaker knows the
best situation for something, judges that the present situation is good enough, and accepts
it. As such, it conveys a sense of the speaker's compromise.

In what follows, I propose the function of ma(a) 'kind of also expresses the
speaker's "downplaying" effect when expressing personal views or disagreement. It is

used to express a sense of humility when talking about him/herself or his in-group

153

members (e. g. family members, and co-workers in some cases) to outsiders. In Section
4.5.5.], I discuss the function of ma( a ) to express the speaker's views non-assertively, its
use to preface disagreement in Section 4.5.5.2, and its use to express humility as regards

personal situations in Section 4.5.5.3.

4.5.5.1. When expressing opinions

The first situation where ma(a) 'kind of was used for the downplaying effect is
when expressing opinions, as shown in (25). 158/198 tokens (83.6%) of ma(a) was used
to lessen the impact of the speaker's statement, of which 12.2% occurred at the tum—initial

position of a new speaker.

(25) WA: a. Amerika wa irui wa yappa zuibun yasui naa.
the U.S-TOP clothing-TOP as expected very cheap IP

MI: b. Maa soo yuu seekatsu ni maa mitchaku shita mono toka
kind of such life for kind of closely related item and

c. hitsujuhin mitai no wa yasui desu nee.
necessity like one-TOP cheap BE IP

WA: a. 'Clothes are, as you know, very cheap in the U.S.

M]: b. Well things which are kind of closely related to everyday life, and
c. daily necessities are inexpensive, aren't they.’

In (b), MI starts his turn with maa, which serves two purposes: (1) make a less abrupt
transition between speakers like the English well; and (2) to avoid sounding too confident
and assertive about the proposition the speaker is putting forward. (26) is another example

of ma( a) being used for the downplaying effect.

(26) SH: a. Daigaku ni haitte/ anmari koo okkina mon etenakatta n de
college at enter very this bit thing get-NEG assert

b. sorede ma/ kotchi ni kite

so kind of here to come

154

c. sooyuu jibun mo hukumete koo/ kaetai na to omotte
such self also including this change IP QT think

d. ma kotchi no hoo ni kite
kind of here way to come

e. ma ima wa manzoku shiteru n desu kedo/ kotchi no seekatsu.
kind of now-TOP satisfy do assert but here-GEN life

SH: . 'After entering my college, I wasn't getting much out of it,
. so, well, by coming here

. I wanted to change things, including myself, and

. kind of, since I came here (the U.S.), and

. now I kind of am satisfied with my life here.‘

0900‘”

In (b), (d) and (e), ma is used to somewhat downplay his statement about his present
situation since coming to the U.S. In each of these cases, ma has the effect of indicating to

the hearer that SH's situation may not be the best one possible, but is acceptable.
4.5.5.2. Prefacing disagreement
Another situation where ma( a ) was used is to preface minor disagreement (22/189

tokens=11.1%), as in (27).

(27) WA: a. Nihon de wa dooshitemo kanpai toka hitokoto shichoo-san toka ga aisatsu

Japan in-TOP without fail toast etc. speech mayor etc. SUB greeting
suru toka nanka arimasu ne.
do etc. something like have IP

b. Amerika no baai wa nanka moo hajirne kara nantonaku hajimatte/
the U.S. in case -TOP somehow really beginning from in someway begin-and

c. nante yuu/ nantonaku owatte/
what shall I say/ somehow end-and

(1. ma kojinkojin no koo deai o daiji ni suru tte yuu ka
kind of individual like this encounter-DO value do or

e. sonna ki ga chotto shimasu kedo ne.
such feeling-SUB a little do but IP

MI: f. Ma boku wa nantonaku sonoo tatoeba ma tashoo keeshikibatta tokoro de wa
kind of I-TOP somehow uh for example kind of a little formal place in-TOP

155

g. aisatsu toka mo anoo Amerika mo aru yoona kanji desu keredo.
greeting etc. also well the U.S. also exist like feeling BE but

h. Sore ga maa sonnani katakurushii kanji ja nai/ tte yuu ki wa shimasu ne.
that-SUB that much formal feel BE-NEG like feeling-TOP do IP

WA: a. 'In Japan (at the beginning of a festival), we always have speeches by
someone, like the mayor,
b. but in the U.S., from the beginning, events begin without us knowing
when,
c. and, what shall I say, end before we know it,
(1. and we may be able to say that they value individual encounters
c. more (than formality), I feel that's what they value, y'know.
MI: f. Well, I somehow feel that at kind of formal occasions, for example,
g. Americans also do formal greetings or something, but
h. perhaps those greetings are not as stiff-sounding (like Japanese), I think.‘

In (t), MI begins his turn with maa, expressing a minor disagreement with WA, who had
stated that Americans value an informal atmosphere when meeting with people, rather than
valuing formality. MI expresses his view that Americans do have formality, but it is not as
formal as Japanese. Maa is followed by other expressions of uncertainty nantonaku
'somehow', sonoo 'uh' as well as another ma 'kind of. Ma 'kind of here seems to
indicate that MI has heard what WA has said, is not in complete agreement with WA, but
wishes to present his disagreement less irnpactfully.

Unlike the connective demo 'but', a marker of upcoming contrast, ma( a ) 'kind of'
makes an utterance sound softer, downplaying the impact of disagreement. Whereas demo
expresses disagreement straightforwardly, ma( a ) marks conciliatory agreement, indicating
the result of the speaker's evaluation process of the situation and partial acceptance of what
has been stated. As such, ma( a ) is not used often by young children, unlike demo which is
used even by every young children.

The participants who filled out the questionnaire after recording their speech (see
Section 6.6.1 for the results) stated the following: (1) they use ma(a) 'kind of when they
do not agree with something completely (7/ 17:41 .2%); and (2) to soften the impact of a

counter argument (3/17=23.5%). It is possible that the speaker has his or her own view

156

about something, but may not wish to express it to a certain addressee. In other words,
ma(a) is a sign of the speaker's calculation as to how best to present his/her own ideas

towards the addressee in a given situation and the resulting compromise.

4.5.5.3. To express humility

The core meaning of ma(a) is that of downplaying, and expresses the speaker's
feelings that the situation or the prior statement is not exactly what s/he thinks, but
acceptable to the extent that they can compromise. Ma(a) expresses the concession the
speaker is making, partially accepting the situation or other interlocutor's view, but
implying that they do not consider them the best choice possible. As such, ma(a) can also
be used to express the speaker's humility about a personal situation, as is illustrated in
(28).

(28) MA: a. Eeto Ema wa anoo ee nante yuu rr desu ka ne/ ee geemu sofuto o
well Ema-TOP well uh what is it called IP uh game software-DO

b. ma ichiban ookii no wa Nintendoo toka arimasu kedo ne/
kind of most big one-TOP Nintendo etc. exist but [P

c. Nintendoo/ Sega toka ma sooyuu anoo tokoro to kuraberu to
Nintendo Sega etc. kind of such well place with compare when

d. chiisai n desu keredomo/
small explain but

e. sooyuu geemu sofuto o tsukuru/ ma kaisha ni tsutometemashite
such game software—DO make kind of company at employed

MA: a. 'Well, Ema, well uh, what shall I say?/ uh game software/
b. well the biggest ones are like Nintendo and others/
c. Nintendo and Sega. Comparing with those companies
(1. it's small, but
e. that kind of video game software/ kind of company, she works for'

In (b)-(c), MA talks about the two biggest video-game software manufactures. Then in (d-

e), he mentions that his daughter also works for a video-game manufacture, though much

157

smaller than the other two. It seems that MA is trying to characterize his daughter's work
as less fancy than it sounds. This kind of downplaying about one's family member is

socially expected in Japan, and ma(a) fulfills that function in (e).

4.5.5.4. Summary

In this section, we examined the functions of ma(a) 'kind of. The general function
of ma(a) is to downplay the effect of the the speaker's utterance. Specifically, ma(a) is
used in the following situations: (1) to express the speaker's feelings that something is not
the absolutely the best one, but acceptable, such as in expressing conciliatory agreement;
(2) to soften the speaker's utterance, and thus is used to preface disagreement; and (3) to
show the speaker's humility about him/herself and/or his in-group members. Figure 4.3

summarizes the functions of ma(a) discussed in this section.

general functions specific functions
ma(a): I to express the speaker's I 1. to express conciliatory agreement
a marker of wish to present his/her 2. to express mild disagreement
downplaying l utterance less irnpactfully I 3. to state one's opinions less impactfully
the effect I 4. to express humility when talking about

personal situations

Figure 4.3 General and specific functions of ma( a )

4.5.6. K 00 'like this': A marker of information recalling

Koo originally is an adverb meaning 'like this' and was the seventh most frequently
used filler in my data (135 tokens=5.3%). Its occurrence was limited to the utterance-
medial or at the intonation-unit—final position, and it was generally followed by a short
pause before the next word was uttered. As a discourse marker koo indicates that the
speaker is in search of appropriate expressions based on his/her personal experience and

functions as a floor-holder while the search is going on. Consider (29).

158

(29) WA: a. Ma watashi mo kondo kuru toki wa soo shitara ii kana to omoirnasu ne.
well I also next time come when-TOP so do-if good FP QT think [P

b. Demo nee/ ano koo/ 10-nen gurai watashi koo/
but 1P well like this ten-years about I like this

c. mae kara koo/ miteru n desu keredomo
ago from like this watch explain but

(1. chotto Amerika no naka de mo kawatte kita yoona ki ga suru n desu.
a little the U.S. of inside in also change come-PAST like feeling-SUB do assert

WA: a. 'Well, I think I also will do that (=wearing casual shoes) when I come
here next time.
b. But, y'know, well it's like, from about 10 years ago, it's like

c. I've been, it's like, observing, but
d. I feel that the U.S. also has been changing a little bit.

In (b)-(c), with the use of koo 'like this', WA sounds as if he is trying to recall his personal
experience and currently visualizing the changes he has observed.

Koo seems to be used when the speaker is trying to bring up what is stored in his
or her mental storage space as accurately as possible. As we can infer from the original
meaning of koo, when speakers use this expression, it seems that they are trying to come
up with expressions which best describe the ways or manners in which things have
happened, are happening, or will happen in the future, based on the speaker's personal
experience. Thus, koo functions as an interface between the speaker's cognitive operation

and the upcoming utterance.

4.5.7. Uun 'uh, well': a marker of hesitation

Uun 'uh, well', a marker of hesitation and place-holder, was the ninth most common
filler in my data (the eighth most common filler was a 'oh', which is discussed in Section
4.5.10). It is a grunting sound and is used when the speaker is searching for words, or
when he or she does not want to express him/herself immediately. In (30), KY has been
talking about her impression of the United States.

159

(30) KY:

KY:

a. Uwasa ni wa kiitemashita kedo/ hiroi n desu yone.
rumor in-TOP hear-PAST but big BE IP

b. Michi mo hiroi shi/ kuruma mo ookii shi/
road also wide and car also big and

c. uun soo desu ne/ inshoo/
well let me see impression

d. senshuu Shikago ni itta n desu yo
last week Chicago to go-PAST BE IP

a. 'I had heard what people talk about (the U.S.), but it's big, y'know?
b. Roads are wide, and cars are big, and

c. we:ll/ let me see, my impression,

d. last week, we went to Chicago'

In (c), while searching what to say next, KY uses two expressions which signal that she is

in the process of thinking: mm and soo desu nee 'let me see' as well as repeating the word

inshoo 'my impression (about the U.S.). Finally, she starts talking about her trip to

Chicago.

(31) is an example of arm prefacing disagreement.

(31) Mieko:

Mieko:

a. Nee/ yappa chigau to omou n desu yone/
[P as you know different QT think assert [P

b. nanka koo sensee ni taishite mo/ reegi tadashisa ga dete kuru to
like like this teacher to about also politeness-SUB come out QT think

omoirnasen?
think

c. Uun/ demo ma sore wa keego toka tsukawaseteru kara da to omoirnasu
well but kind of that-TOP honorific etc. use-CAUSE-CAUS because BE QT think

kedo ne.
but IP

a. 'Y'know, as expected, they are different, I think.

b. somehow, like, towards teachers, they show politeness, don't you think?
c. We::ll, but that's because we make them use honorifics, etc., I think.‘

160

In (b), I stated that American students who take a Japanese course are different from other
Americans. In (c), [M expresses disagreement with demo 'but'. By uttering uun before
and maa 'kind of ' after demo, IM avoids the abruptness of disagreement.

Uun is sometimes used for background repairs, as is illustrated in (32).

(32) KB: a. Atashi ga omotta no wa Indo to Nihon no kankee toka/
I-SUB think-PAST NOM-TOP India and Japan GEN relationship etc.

b. sono Indo ga uun/ yoku aru topikku na 11 desu kedo/
uh India-SUB often exist topic assert but

c. ano shokuminchi jidai ni uketa Indo no
well colony period in received India GEN

KE: a. 'What I was thinking is the relationship between India and Japan or
b. uh India, well it's a common topic, but
c. well, the influence India received during its colonial days'

In (b), by saying uun, KE indicates to the addressee that she has to stop and think what to
say next before she goes on.

Uun is similar to ee( to ) (discussed in Section 4.5.8) in its floor-holding function.
Uun, like ee( to ), indicates that the speaker's smooth speech production is momentarily
halted. Moreover, both rum and ee( to) are used in relatively informal situations to an
addressee who is not one's superior. However, while ee( to) predominantly signals that the
speaker is simply searching for an appropriate expression, uun indicates two kinds of
hesitation: (l) production-related hesitation (the speaker has trouble formulating an
appropriate expression); and (2) hesitation based on the speaker-addressee and speaker-
proposition relationship, e.g. to an addressee to whom the speaker may not be in the
position to express overt disagreement, or the speaker is uncertain about the proposition, or
the proposition is a sensitive issue. In other words, while ee( to) carries no sense of

reluctance, hedging or disagreement, uun often indicates the speaker's hesitation or

161

reluctance to express exactly what is in his or her mind. As such, uun may also be used to

express disagreement or refusal instead of explicit iie or iya 'no'.

4.5.8. Ee(to) 'well, uh': a marker of information recalling effort

Ee( to ) is an interjection which is translated as 'well', 'let me see' and 'er' in the
Kenkyusha's New Japanese-English dictionary. It was the 10th most common filler in my
data. Ee( to ) was used in several variant forms such as ee, and uunto,6 all of which will be
represented as ee( to) throughout this study. The degree of formality is as follows: ee>
eeto> ettoo> uunto. Maynard (1989a:3 l) characterizes ee( to ) as a "language-production-
based" filler. Takubo (1995:1024-1025) characterizes it as a marker of "inward-directed
cognitive operation," and states that "it is used when the speaker is in consultation with
his/her knowledge data-base, or when he or she wants to temporarily suspend the
interaction with the addressee before continuing" [translation mine].

In my data, 71/94 tokens (75.5%) of ee( to ) were used at the clause-initial position.
Of the 71 tokens, 59.1% (42 tokens) were used to preface a new speaker's turn, which
was the highest percentage among all the fillers in my data used in this position. In what
follows, I discuss four functions of ee( to ) as a marker of the speaker's recalling effort: (1)
prefacing an answer to a question in Section 4.5.8.1; (2) as a floor-holding device in
Section 4.5.8.2; (3) prefacing repair segments in Section 4.5.8.3; and (4) prefacing a

(sub)topic shift in Section 4.5.8.4.

4.5.8.1. Ee(to) prefacing answers
Ee( to) was frequently used to preface answers to questions. 29/94 tokens (30.1%)
of all ee( to ) was used before answering WH questions, and 7 tokens (7.4%) were used

before answering yes-or-no questions. Not only answers to questions which require some

 

6 Based on its functions, uunto is treated as a variant of eeto in this thesis rather than as a variant of mm
'well'.

162

amount of thinking, but answers to very simple questions were prefaced with ee( to).

Sometimes, pre-answer ee( to ) was preceded by a 'oh' as in a ee( to) 'oh, well'. (33)-(35)

illustrate the use of ee( to ) prefacing answers to WH questions.

(33)

(34)

(35)

Mieko:

SH:

Mieko:

SH:

Mieko:

Mieko:

Mieko:

Mieko:

Mieko:

IY.

Mieko:

IY:

a. Amami-Ooshima tte doko ni aru n desu ka?
Amami-Oshima QT where at exist Q

b. Eeto anoo hanbun Okinawa haittete
uh well half Okinawa enter

c. hanbun/ sugoi dokutoku no bunka ga am It desu yo.
half very unique culture—SUB have assert IP

a. 'Where is Amami-Oshima?
b. Well, one half of it belongs to Okinawa, and
c. the other half/ it has a very unique culture, y'know.'

doko de shiteru n desu ka?
BE: (2

a. Baito
part-time work where at do

b. Eeto juuroku-goo zoi no sebun irebun de yatteru n desu kedo'
uh route along LOC 7-11 at do-PROG assert but

a. 'Where do you work part-time?
b. Uh, I work at 7-11 on Route 16.'

a. Daigaku wa doko desu ka?
college-TOP where BE Q

b. Eeto Sooka Daigaku desu.
uh Sooka University BE

c. Nan-nensee desu ka?
what-grade BE Q

d. Eeto san—nen desu.
uh third-year BE

a. Which university do you go to?
b. Uh, Sooka University.

c. What year are you in?
d. Uh, junior.‘

163

In all these examples, answers to questions are prefaced with ee( to). As stated at the
beginning of this section, ee( to) has typically been characterized as a language-production-
based filler (Maynard l989a:31), which is supposed to be used when smooth speech
production is temporarily hindered or to fill in a pause while searching for an appropriate
expression; SH is trying to give the best way to describe where Amami-Oshima is in
(33b), and AR is thinking for a second how to give information about her work place in
(34b).

Characterization of ee( to) as a language-production-based filler does not make much
sense in Example (35), where IY is only asked the name of her university and her academic
level. Giving the requested answers in (35) did not require any degree of thinking, since
the questions are simple and straightforward and about IY herself. The answers should be
readily available to IY, and there is no need to be hesitant or spend time searching for
appropriate expressions. In fact, three of the participants in the interviews responded with
ee( to) when they were asked their names, two responded with ee( to ) when they were asked
about their majors or hometowns, and four prefaced the names of their university and their
academic level with ee(to). This indicates that ee( to) can sometimes be used in situations
where there is no need to search for appropriate expressions.

Ee( to ) in these cases is perhaps used as a momentary floor-holder to signal to the
addressee that the speakers understood the questions and are ready to give their answers,
rather than being used to fill in potential silence while searching for an appropriate

expression.

4.5.8.2. As a floor-holder
There were cases of ee(to) clearly being used to signal that the speaker was having
trouble coming up with appropriate expressions, as in (36)-(37). (36) is an example where

ee( to ) is used while searching for the right word.

164

(36) UM a. Koobe wa itta yo.
Kobe~TOP go-PAST 1P

KY: b. 1i yone.
good IP

UM: c. Kyonen to ototoshi.
last year and the year before last year

(1. Eeto chotto are suru mac.
uh a little that do before

KY: e. Kuzureru mae?
destroy before

UM: f. Un kuzureru mae ni.
yeah destroy before

UM: a. 'I visited Kobe.
KY: b. That's nice.
UM: c. Last year and two years ago.
(1. Uh, a little before that (happened).

KY: e. Before it was destroyed by the earthquake?
UM: f. Yeah, before the earthquake.

In ((1), UM says that she "visited Kobe eeto a little before that (happened)," unable to come
up with the word for destruction or earthquake. KY supplies the expression kuzureru
'destroyed', and UM acknowledges the information with an 'yeah', indicating that
kuzureru was the word she was looking for.

(37) is an example where ee( to) was used to come up with an appropriate answer,

rather than a specific word.

(37) Mieko: a. Ichiban tanoshikatta koto wa nan desu ka?/ Amerika ni kite.
most enjoyable thing—TOP what BE Q the U.S. to come

AR: b. Eeto soo desu ne nanka/ a doyoobi ni Shiidaa Pointo ni itta n desu kedo

uh let's see well oh Saturday on Cedar Point to went explain but

Mieko: a. 'What was the most enjoyable experience you've had in the U.S.?
AR b. Uh, let's see, uh, oh, we went to Cedar Point on Saturday'

In (37), AR uses eeto to search through her memory to find an answer to the question—the

165

most enjoyable experience. She uses three fillers, eeto, soo desu nee, and nanka before
she begins her answer with a 'oh', which marks AR's having thought of something to say.
If ee( to) is simply used as a filler of potential silence, any kind of sound such as aa or an
will suffice, which is not the case here.7 Therefore, ee( to) has more meaning than just
filling in the pause. I propose that, in these cases, the primary function of ee( to) is to
signal to addressees that speakers are recalling information. The secondary functions of

ee( to) are to express the speaker's intention to continue and fill the potential silence.

4.5.8.3. Prefacing a repair segment
Ee( to) was also used to preface repair segments. (38) is an example of a

replacement repair.

(38) Mieko: a. Hitori-gurashi? Geshuku?

one-person live boarding house

IY: b. Geshuku/ ee apaato de
boarding house uh apartment at

Mieko: a. 'You live alone? Is it a rooming house?
IY: b. Rooming house, uh it's an apartment.’

In (b), IY began to say geshuku 'a boarding house' first, and corrects herself with ee, a
shorter and more formal variant of eeto.

(39) is an example of a background repair.

(39) NM: a. Uchi wa ne/ ano ne/
my family-TOPIP well IP

b. uunto ryooshin wa Iwate shusshin na n da kedo/
parents-TOP Iwate Prefecture hometown BE but

c. de soko de nanka kookoo ni densha de kayotteta no ne?
and there well high school to train by commute-PROG-PAST IP IP

 

7 Ano(o) 'uh' or arm 'uh' are possible alternatives to eeto in this situation.

166

NM: a. 'In my parents' case/ well y'know?
b. well, they are originally from Iwate Prefecture,
c. and they were commuting to their high schools by train, y'know?‘

In (b), NM inserts background information about her parents' home prefecture with uunto,
an even more informal variant of eeto. In these examples, ee( to ) was used to show a

momentary hesitation before introducing actual repair segments.

4.5.8.4. (Sub)topic shifting

Ee( to ) was also used to search for an appropriate expression before the speaker
introduced a new topic. It was frequently used by myself when I was interviewing the
participants, but there were other cases as well. (40) is an example of ee( to) for a subtopic

change.

(40) IY: a. Atashi ano byuutii ando biisuto mite/ ano jirnaku nashi datta kara
l uh Beauty and Beast see well subtitle without BE-PAST because

b. yoku wakan nakatta.
well understand NEG-PAST

c. Uunto ne/ sukina ne/ are nani?/ eega.
uh IP favorite [P that what movie

(1. Ichiban bideo mite/ aa atashi wa kore ga suki daa tte yuu eega.
best video see oh I-TOP this-SUB like BE QT say movie

NM: e. ((laugh)) Kakko ii to omotta no ga/ Kebin Kosunaa no
look good QT think one-SUB Kevin Costner GEN

IY: a. 'I saw the Beauty and the Beast and/ well there was no subtitle, so
b. I couldn't understand it well.
c. Uh, your favorite/ what is your favorite movie?

(1. Watching videos, the one you feel you love the most?
NM: e. ((laugh)) The one I thought the coolest was the one by Kevin Costrrer'

In (c), IY changes the subtopic with uunto from the movies she has seen to N M's favorite

movies.

167

(41) is an example of a major topic change during an interview. Prior to this
segment, KY has been describing her happiest experience since coming to the U.S., talking

about going bowling and making her personal best score.

(41) KY: a. Hyaku-juuroku-ten desu.
1 16 point BE

Mieko: b. A soo desu ka.
oh so BE Q

c. Eeto ichiban komatta koto toka iya datta koto wa nan desu ka?
well most troubled thing and bad BE thing-TOP what BE Q

KY: a. 'It was 116 points.

Mieko: b. on really?
c. Well, what was your most troubling or worst experience?‘

In (c), the topic was shifted from happy experience to difficult experience.8 The use of
ee( to ) for a topic change during interviews occurred frequently at the beginning of the
interviews, when I was getting the interviewees' background information such as their
names, hometowns, majors, and the reasons for coming to the U.S. Ee(to) was used as a
place-holder while I was thinking what I should ask next. I needed more time to think at
the beginning because the participants gave short answers while I was still drinking about
the next questions to ask.

Thus, ee( to) before a (sub)topic change functions to signal to the hearer that the

speaker is in the process of searching for an appropriate expression and to hold the floor.

4.5.8.5 Other functions of ee( to )
Besides the functions discussed thus far, ee( to ) was used at the beginning of a

narrative. Of the six narratives collected for the present study, three began with ee( to ), one

 

8 In this situation, the connective jaa 'then' rather than the filler eeto 'uh' would have been a possible
alternative, and obviously there are personal preferences as to which expression a speaker chooses at a given
circumstance.

168

began with ano, and the other two began with eeto anoo and ano eeto. Ee(to) and ano(o) in
narratives were used between the initial chat and the subsequent narratives. Ee( to ), like
ano(o), seems to have a function of marking a change in speaker orientation, such as from
an initial chat to the narrative.

We have seen that ee( to ), like other fillers discussed in this chapter, is used in
various situations where the speaker needs time while searching for the right words or an
appropriate answer, signalling the speaker's intention to continue his or her tum. As such,
it is used to preface a (sub)topic shift, repairs, and when the speaker is trying to formulate
an answer to a question. It seems that the primary function of ee( to) is to indicate the
speaker's intentions to continue their turns and to request that the addressee wait until they

come up with what to say. Figure 4.4 summarizes the functions of ee( to ) discussed above.

general functions specific functions
ee( to): l I. express that the speaker l. to preface answers to questions
a marker of is searching an 2. to preface a repair

the speaker's i appropriate expression 3. to preface (sub)topic change
information and needs time

recalling
effort r 11. floor-holder 4. to hold the floor

 

Figure 4.4 General and specific functions of ee(to)

Both ee( to) and ano(o) are used when speakers search for an appropriate
expression. Ee( to ), however, is different from ano(o) in several respects; (1) ano(o)
sounds more formal than ee( to ) and can consequently be used in formal situations and
towards one's superior; (2) the primary function of ano(o) is a hesitation marker based on
the speaker's perception of the speaker-addressee relationship and is used to solicit the
addressee's cooperation by enhancing or creating the sense of togetherness and sharedness,
while ee( to) is used to signal the speaker's consultation with his/her own mental data base

to come up with an appropriate expression. Therefore, the main functions of ano(o) are in

169

the participation framework and exchange structure proposed by Schiffrin, whereas the
main function of ee( to ) is in the information state and the exchange structure.

Ee( to) also contrasts with nanka or ma( a ), which entail creating an agreeable
conversation atmosphere. Ee( to ) does not have such a function. Thus, while ano(o),
ee( to ), mm, and nanka all can function as pause-filling devices and are translated as uh,
well, and er in dictionaries and Japanese textbooks, their functions must be clearly

distinguished and their nuances explained in the Japanese language classroom.

4.5.9. Hora 'there, y'know': A marker of addressee involvement

Hora 'look, there' is an interjection which is used to attract someone's attention to a
visible object. Kenkyusha’s New Japanese-English Dictionary lists y'know and you
remember as its translation.

Hora was the least frequent of the fifteen fillers in my data with 18 tokens. Of them,
none appeared in the narratives, one in an interview, and seventeen in casual conversations.
In what follows, I argue that hora not only functions as a language—production-based filler
as suggested by Maynard (l989a), but also as a device to enhance or create a sense of
sharedness between interlocutors.

Hora, I propose, has functions similar to y'know in English, and has the following
effects: (1) points to shared knowledge about something, and enhances the participation of
the addressee in conversation; (2) creates a sense of "sharedness" and rapport among
interlocutors; and (3) functions as a floor-holder while the speaker is trying to recall an
expression. (42) illustrates the use of hora referring to a shared piece of information

between the speaker and the addressee.

(42) YK: a. Nanka kanojo ga kekkyoku hora sotogawa desho?
like she-SUB after all y'know outside TAG

b. ano hito no suwaru/ nee suwaru toko ga.
that person-SUB sit IP sit place-SUB

170

c. Dakara are mitai desu yo/ kanojo no mono ga hotondo torareta n ja
so that like BE IP she-GEN things-SUB mostly take-PASS BE

YK: a. 'Well she is, as you know, outside (=close to the door), right?

b. the place where her desk is.
c. 80, it looks as if mostly her things were taken'

Both YK and I are familiar with the office arrangements where a burglary took place. By
using hora 'y'know' and desho 'right?’ in (a), YK tries to facilitate my understanding
regarding why mostly things which belong to one of the office workers were stolen, based
on our shared knowledge that her desk is located nearest to the entrance.

Hora is also used to create the feeling of "sharedness" between interlocutors when

they do not actually have shared knowledge about something, as illustrated in (43).

(43) Mieko: a. Sonnani takusan o-uchi miru no taihen deshita ne.
such many HON-house look NOM hard BE-PAST IP

TA: b. Iya/ sore ga hora/ hima datta desho?
no that-SUB y'know free BE TAG

Mieko: c. Demo ii kairnono desu nee.
but nice purchase BE IP

TA: (1. Demo/ demo koko wa hora gakkoo-ku ga anmari yoku nai n desho?
but but here-TOP y'know school-district-SUB very good-NEG TAG

Mieko: a. 'It must have taken you a long time to look at so many houses!
TA: b. Oh no, the fact is that I had nothing else to do, as you know?

Mieko: c. But you've purchased a nice house!
TA: (1. But, but, the school district here is not that great, y'know?’

TA uses hora in (b) and (d) as if I knew what TA's schedule was like when she was house-
hunting or about the school district in the area where TA's house is. Hora in this case
contributes to creating "pseudo-shared world" between TA and myself, eliciting the
addressee's involvement in the conversation. Furthermore, since the use of hora is limited
to casual conversations among conversants who know each other well, the speaker can use

hora in order to emphasize their close relationship, enlist addressees' cooperation, and

171

enhance the addressees' participation in the conversation.
The third function of hora is to indicate to the addressee that the speaker is in the

process of recalling something.

(44) Mieko: a. Naze asoko de oshie-hajimeta n deshita kke?
why there at teach-begin-PAST BE-PAST Q

YO: b. Saisho wa dakara hora ano sensee/
first-TOP so y'know that teacher

c. sensee tte yuu ka/ eeto nan da kke/ A-san/ kanojo no shookai dc
teacher QT say Q uh what BE Q A-Ms. she-GEN introducing by

Mieko: a. 'Why did you begin teaching there?
YO: b. At first, to explain, y'know, that teacher
c. teacher?, or what should I say? Ms. A. I was brought in by her‘

In (b), YO uses hora when she is trying to recall the name of the person who got her
involved in teaching Japanese. YO has trouble remembering the person's name and uses
three fillers: hora, eeto and nan da kke 'what do you call'. Furthermore, she says ano
sensee 'that teacher', indicating that she is referring to a teacher who is known to both Y0
and myself. In this example, hora, together with ano 'that person/thing we both have
shared knowledge about' and nan da kke 'what do you call that', contributes to increasing
the addressee's participation in the conversation.

As a discourse marker, hora creates a sense of "sharedness" among the
interlocutors and enhances their participation in the conversation. It also relates to the
knowledge and meta-knowledge among them, and thus Operates in the participation

framework and the information state.
4.5.10. A 'oh', aa 'oh', and e? 'oh, huh?’ for information management

Finally, I discuss three interjections, a, aa and e?, all of which are translated as oh

in English (K enkyusha 's New Japanese-English Dictionary). A 'oh' indicates that the

172

speaker has just become aware of something and wishes to express that realization to the

addressee and/or to him/herself in the case of soliloquy. In face-to—face interactions, it is

used in repairs or prefacing an answer to a question, where the speaker's re-orientation is
necessary. (45) illustrates a case of a prefacing a background repair segment. KY is

describing her high school.

(45) KY: a. Nanka Oosaka na n desu keredomo/ sono yama no ushiro ga/
like Osaka explain but that mountain GEN back—SUB

b. a/ koosha no ushiro ni yama ga atte
oh school building GEN behind at mountains-SUB exist

c. sono ushiro ga Nara-ken to yuu tokoro.
that behind-SUB Nara Prefecture QT say place

KY: a. '(my high school) is in Osaka, but behind the mountain is,

b. oh, behind the school building is a mountain, and
c. behind that is Nara Prefecture, a place like that.

In (b), KY starts saying something about behind the mountain, and she repairs it with a
'oh', adding that the mountain is behind the school building and continues on to say that
behind the mountain is Nara Prefecture. Thus, in (45), a signals to the addressee that the
speaker has just became aware of something, a mistake.

(46) is a case of a before answering questions. KE says (1 every time before

answering questions, as if she is acknowledging the receipt of my questions.

(46) Mieko: a. Eeto ja namae o onegai shimasu.
uh then name-DO please

KE: b. A, KE desu.
oh KE BE

Mieko: c. Hai, de Ajia keezai o benkyoo shiteru n desu ka?
OK and Asia economics-DO study do-PROG BE Q

KE: (1. A, hai.
oh yes

173

Mieko: e. Tokuni Ajia no donna koto ni tsuite benkyoo shiteru n desu ka?
particularly Asia-GEN what thing about study do-PROG BE Q

KE: f. A mada soko made haitte nai n desu kedo.
oh yet that as far as enter-NEG BE but

Mieko: 3. 'Your name, please.

KE: b. Oh, I'm KE.

Mieko: c. O.K., and you're studying Asian economics?

KB: (1. Oh, yes.

Mieko: e. Particularly about what in Asia are you studying?

KE: f. Oh, we haven't gotten into the details yet, but well now the world'

The second interjection aa 'oh' indicates a speaker's receipt of information. Aa can
occur alone or as aa soo desu ka 'oh, is that right'. In (47), YK and I are talking about

inexpensive radios, and in (b), YK signals that she received the information with aa 'oh'.

(47) Mieko: a. Kore ne/ chooshi ga warukute tokidoki ugokanai n desu yone.
this IP condition-SUB bad sometimes move-NEG BE IP IP

YK: b. Aa soo na no. Uchi ni mo am It desu kedo ne/ kooyuu no zenzen ne.
so IP house at also exist BE but IP this kind one not at all IP

Mieko: a. 'This one, y'know, its condition is bad, and sometimes it doesn't work.
YK: b. Oh, I see. I have one like this at home too, but this kind of (radio) is not
at all good.‘

E? 'oh, huh?’ is used to indicate receipt of unanticipated information, or indicates

that the speaker did not hear or understand what has been said.

(48) Mieko: a. Amerika ni kitai na to omotta no wa donna riyuu na n desu ka?
the U.S. to come-want QT think-PAST NOM-TOP what kind reason BE Q

NM: b. E? sonoo benkyoo shiteru koto mo arimasu.
oh uh study do-PROG fact also exist

c. Ano benkyoo shiteru no wa nichi/ nichiboo/ nichibee booeki?
well study do-PROG-GEN-TOP Japan! Japan tra Japan-U.S. trade

Mieko: a. 'What made you want to come to the U.S.?
' b. Oh? well, what I'm studying is part of the reason/
c. Well, what I'm studying is J apan/ Japan trade/ Japan-U.S. trade'

174

The frllers in examples (45)-(48) display the speaker's cognitive and information
management processes. A, aa, and e? are all response markers concerning the speaker's
receipt of information. They signal whether or not the speaker understood the received
information and indicate whether the information was anticipated or not. They also indicate
the speaker's cognitive re-orientation in the case of a repair. As such, a aa and e? all

operate in the information state.

4.6. The role of fillers in conversation

We have seen how fillers function in Japanese conversation. Unlike the
connectives discussed in Chapter 3 which mostly function as utterance-initial brackets,
fillers primarily bracket utterance-intemal units, separating prior and upcoming portions
within an utterance. Fillers such as ee( to) 'uh' and all the information management fillers
index an utterance to the speaker, displaying the speaker's cognitive processes and
orientation to the information, whereas socially motivated fillers such as nanka 'like',
ano(o) 'uh, well', ma( a ) 'kind of, and hora 'there, y'know' all index an utterance both to
the speaker and the addressee, displaying the speaker's feelings to both the addressee and
to the proposition.

Except by a few researchers such as Maynard (1989a), Cook (1993) and Takubo
(1992 and 1995) and Takubo and Kinsui (1996), fillers have been treated predominantly as
pause-filling or floor-holding devices in face-to-face interactions, and consequently have
not been given much importance. We have seen, however, that even the items which have
been characterized solely as production-related fillers such as ee( to ) 'uh' and hora 'there'
have important functions unique to each one. Many times, fillers have simultaneous roles
as a place-holder and for other social and interactional purposes.

There are many reasons why speakers stall momentarily: they may not be able to

recall a word instantly; they may be weighing various options, wondering which one best

175

suits the purpose of the utterance in a given situation; they may feel that they should
withhold some information from certain addressees; or they may want to avoid abrupt topic
changes in face-tO-face interactions. Each filler, in its unique ways, represents the
speaker's thoughts and/or evaluation or judgement regarding the speech situations,
propositions or speaker-bearer relationships, operating primarily in the information state or
participation framework. That is why most fillers are not interchangeable with each other,
and interlocutors, in many cases, can guess what kind of utterance or conversational move
is to follow simply by hearing certain fillers.

In some instances what makes this relationship between a particular filler and its
implied meaning is the original meaning of the filler (e.g. nanka 'somehow', yappari 'as
expected' and ma( a ) 'kind of); other fillers come to be associated with particular situations
and assume specific functions in a given speech community through their routinized usage
(e.g. ee( to )). In still other cases the conventionalized usages may be far removed from the
expression's literal meanings, as in the case of ano 'uh; well' and hora 'look'.

The routinized expressions assume shared functions within a speech community.
For example, when native Japanese speakers hear ma(a) 'kind of, they can sense that the
speaker is trying to downplay the effect of his or her utterance; when the speaker utters
nanka ”like', addressees can sense the speaker's uncertainty. When the Japanese hear
ee( to ) 'uh', koo 'like this' or hora 'there; y'know', they can feel that the speaker is going
through his/her mental storage space, trying to figure out appropriate expressions or to
recall something.

Through these routinized functions, fillers help the speaker display his/her feelings
and orientation towards a proposition and/or addressee without overtly mentioning them.
They may also enhance the involvement of the interlocutors in conversation and contribute
to discourse coherence. Thus, fillers are essential for successful face-to-face interaction.
Table 4.9 shows the indexical functions of the fillers discussed in this chapter as textual

and participant coordinates.

176

Table 4.9 Fillers as contextual coordinates

 

 

 

marker original position within as participation as textual
meaning an intonation unit coordinate coordinate

ano that initial/medial speaker/hearer prior/upcoming

nanka something initial/medial/final/alone speaker/bearer prior/upcoming

ma( a ) kind of initial/medial/fmal/alone speaker/bearer prior/upcoming

ee( to ) well; uh initial/medial speaker upcoming

moo really initial/medial/final speaker/bearer upcoming

yappari as expected initial/medial/final/alone speaker/hearer prior/upcoming

uun well; uh initial/medial/final/alone speaker/heater prior/upcoming

koo in this way initial/medial speaker upcoming

hora look; there initial/medial/final speaker/hearer upcoming

a.’,aa, e? oh initial/medial speaker prior

 

 

4.7.

using Schiffrin's (1987) framework. Based on the fact that fillers are not interchangeable
with each other, and that specific fillers are associated with specific agreed-upon functions

Summary

In this chapter, I examined the functions of Japanese fillers as discourse markers

within a speech community, I have argued that fillers have important roles, just as

connectives do, disclosing information such as the speaker's feelings, orientation, and/or

changes in conversational moves, and therefore are an indispensable part of face-to-face

interactions.

177

 

CHAPTER 5

Japanese interactional particles as discourse markers

5.1. Preliminaries

We now turn to the last group of discourse markers used by native Japanese
participants, interactional particles. Interactional particles, like connectives and fillers, play
important roles as contextual coordinates in face-to-face interactions in Japanese,
expressing the speaker's feelings and orientation towards the addressee or proposition
(Makino and Tsutsui 1986:45-49). Martin (1975:914) states that interactional particles are
used to "impart some additional hint of the speaker's attitude toward what he is saying—
doubt, conviction, caution, inquiry, confirmation or request for confinnation, recollection,
etc." In ordinary conversation, especially in the informal type, it is virtually impossible to
talk without interactional particles.

In my data, the frequency of interactional particles was 55.7 tokens per 1,000
words, whereas the frequency of fillers was 77.1/1000 words and that of connectives was
30.8/1000 words (the rate of connectives in the English data was 52.9/1000 words, and
that of fillers was 28.3/1000 words). These numbers alone indicate how prevalent the use
of fillers and interactional particles is in Japanese conversation. Interactional particles were
used for various purposes. Ne, for example, was used to request confirmation or
agreement, to soften the tone when expressing disagreement, as an attention—checking
device, and as an attention-getter. Yo was used in warning, advice, and invitation. Kana
and kashira were used to express the speaker's uncertainty about a proposition.

One of the differences between interactional particles and other discourse markers is
the position where they occur in discourse; unlike connectives and fillers, which function
as utterance-initial- and utterance-medial—brackets, interactional particles function as
utterance- or phrase-final brackets. The units they bracket are also different. Connectives

mainly bracket propositions, speech acts and turns, indexing an utterance to prior and/or

178

upcoming utterances and fillers bracket prior and upcoming units within an utterance. In
contrast, interactional particles generally index an utterance to prior units of discourse.

The principal aim of this chapter is not to dispute the validity of previous research
on Japanese interactional particles, but to argue for the incorporation of a multi-faceted
approach like Schiffrin's in their analyses. Adapting such a discourse model is particularly
important in analyzing them since they coordinate interactional units rather than idea units.
In addition, they reveal cognitive information; e.g. whether the information belongs to the
speaker, the hearer, or is shared by both. They also reveal the speaker-hearer relationship;
e.g. their relative social status difference and how close they are to each other. Yo, for
example, is not generally used toward one's superior. Some particles are used only in
informal settings (e.g. no, so, and yona) and others can be used both in formal and
informal settings (e. g. ne and yone), thus revealing the speaker's judgement about whether
the situation is formal or informal. Furthermore, many particles reveal the speaker's
gender; e.g. wa and kashira are mostly used by female speakers, and the addressee-directed
na, yona, and 20 are generally used by male speakers. Table 5.1 is a classification of

interactional particles by formality and the speaker's gender.

Table 5.1 List of interactional particles by formality and the speaker's gender

 

 

 

 

 

particles formal informal

predominantly feminine wa kashira, no
redominantly masculine yona, 20

neutral ne, yo, yonel kana, sa

 

 

 

 

Particles such as yo 'I assert', no 'I assert', 20 'I assert' and kana 'I wonder' also
express the speaker's feelings about a proposition; e. g. how confident the speaker is about

his or her utterance. Furthermore, the use of these particles is not rigidly fixed but varies

 

1 Most Of these particles have a short and long variants (e.g. ne and nee, so and saa). They will be
represented by their short variants throughout the present study unless the distinction is important for the
discussion.

179

according to circumstances. For example, yo is generally characterized as a particle of
assertion when a piece of information is assumed to be known only to the speaker,
operating in the information state defined by Schiffiin (1987). However, as Maynard
(1993:107—108) states, when the speaker wishes to avoid sounding rude, or to enhance the
atmosphere of cooperation and rapport, ne may be used instead of yo. On the other hand,
as we shall see in Section 5.4.2.3, yo may be used in place of ne to show agreement, if the
speaker feels strongly about a proposition in a situation where normally ne is expected.
The choice of particles is not independently determined but related to the context in which
they occur. Thus, it is preferable to employ an approach in which we can analyze the
functions of interactional particles from multiple perspectives. Therefore, Schiffrin's
(1987) framework is adopted in the present study for the analysis of interactional particles.
In what follows, I analyze the functions of Japanese interactional particles as
discourse markers. Quantitative analysis is presented in Section 5.2, followed by a
discussion of interactional particle use in relation to formality, age, gender and speech

genre in Section 5.3. In Section 5.4, the functions of ne, yo and yone, are examined.

5.2. Quantitative analysis

A total of 11 interactional particles were examined, and the frequency per 1,000
words and the percentage of each particle in total particle use were calculated. Ne was by
far the most frequently used of all, constituting 52% of all the interactional particles used,
followed by yo (13%)and yone.2 These three constituted 76% of all the interactional
particles in my data. These percentages were similar to those discussed in Maynard (1993);
she reports that ne constituted 41.18% and yo 14.83% in her data. Of all the ne used, the
percentage for the utterance-final ne was the highest (35% of all interactional particles),

followed by the utterance-medial ne (16%). Since ne and yone express a cooperative

 

2 Yone is a combination of ya and ne, but it was counted as a separate particle. Likewise, yona (the
combination of ya and na) was counted as a separate particle.

180

 

attitude, such as confirmation and agreement, the high percentage of of ne and yone reflects
the importance of cooperative attitude in face-to-face interactions.

Table 5.2 shows the numbers of tokens, the rate per 1,000 words, and the
percentage of each interactional particle in my data. The numbers in parentheses show the

breakdown Of ne and na according to their positions within an utterance.3

Table 5.2 Summary of Japanese interactional particles in the collected data (n=36)

 

 

particles functions4 tokens / 1000 wds %

ne total confirmation/rapport 1039 29 52
(initial) ( 10) (0.28) ( 0.5)
(medial) (314) (8.77) (16)
(frnal) (689) (19.3) (35)
(alone) ( 26) (0.73) ( 1.2)

yo speaker's emphasis 251 7.0 13

na total confirmation/rapport/emotion 1 5 1 4 .2 7.6
(media15) ( 5) (0.01) ( 0.3)
(final 1) ( 29) (0.08) ( 1.5)
(final 26) (117) (0.33) ( 5.9)

no explanation/emotion/question 133 0.37 6.7

kana uncertainty (male/female) 115 0.32 5.8

kashira uncertainty (female) 4 0.01 0.2

sa informality/assertion 77 0.22 3.9

wa assertion with femininity 1 0.003 0.1

20 assertion with masculinity 1 0.003 0.1

yone assertion+confirmation 216 0.6 11
ona assertion+confirmation 4 0.01 0.2

total 1992 55.7 100

 

 

 

 

 

 

 

Okamoto (1995:304) reports that the use of strongly feminine particles was

infrequent in her study of informal Japanese conversations. In my data also, interactional

 

3 All particles other than ne and na, and sa, only occur at the utterance-final position. All occurrences of
sa in my data was utterance-medially.

4 The characterization of the functions of the interactional particles are based on the following: Uyeno
(1971 ); Makino and Tsutsui (1986), and Cook (1988).

5 All utterance-medial and final nal are considered as addressee-directed in the present study; e.g. K:

K inoo naa Morita-san to rural dansu shita yo ore and K inoo no eega omoshirokatta na 'Yesterday's

movie was interesting, wasn't it?’ It is generally used by male speakers in casual conversation with male
addressees.

6 This category refers to the self-directed na, as in Dokka ikitai na '1 want to go somewhere.’ This type of
no is used both by male and female speakers to male and female addressees.

181

particles which are associated with gender appeared infrequently. For example, the particle
kashira (feminine) and yona (masculine) were used only four times each, and wa7
(feminine) and 20 (masculine) were used just once each. This may be partially due to the
limited types of subjects used for the present study, all of whom were teachers and college
students. With data from more varied conversation situations, such as conversations
between married couples and from a wide variety of occupations where expressions of
femininity or masculinity are common, the results may have been different. This is an area

which awaits further research.

5.3. Sociolinguistic factors

We now consider the use of interactional particles in relation to four sociolinguistic
factors: (1) formality; (2) age; (3) gender and (4) speech genre. In order to examine the
significance of these four sociolinguistic factors on the usage of interactional particles,
based on their rates per 1,000 words, the unmatched two-tailed-test was performed for the
factors with two variables, such as formality, and one-way ANOVA test was run for those
with three or more variables; e.g. speech genre. At a 95% confidence level, a P-value of
0.05 or lower indicates a significant difference; and 0.1 indicates a strong tendency towards

significance.

5.3.1. Formality

The result of the t-test showed a statistically significant influence of formality on the
use of interactional particles. The total rate of interactional particles was much higher in
informal speech than formal speech (80.5/1000 words vs. 45 .4/1000 words), and the
difference was significant (P=0.003). The high use of interactional particles in informal

speech indicates that Japanese speakers express their feelings and attitudes toward a

 

7 Wa with falling intonation can be used both by male and female speakers, but this type of wa did not
appear in my data. The wa in my data was pronounced with rising intonation.

182

proposition or the addressee more readily in informal speech than in formal speech. Table
5.3 shows the rates of interactional particles per 1,000 words and their P-values by speech

style.

Table 5.3 The tokens and rates/1000 words of interactional particles by formality (two-tailed t-test; n=36)

 

 

 

 

 

particles formal (n=24) informal (n=12) P—value
tokens rates/1000 wds tokens rates/1000 wds cl=95%

ne total 696 27.5 448 32.8 ns
(initial) ---------- ( 10) ( 0.96) 0.001
(medial) (178) ( 7.03) (136) (13) ns
(final) (503) (19.9) (186) (17.8) us
(alone) ( 15) (0.59) ( 11) (1.05) ns

yo 144 5.68 107 1 0 . 2 0.035
na total 72 2.84 79 7.56 0.002
(medial) ( ----- ) ( ----- ) ( 5) (0.48) 0.027
(final 1) ( ----- ) ( ----- ) ( 29) (2.8) 0.011
(final 2) ( 72) (2.8) ( 45) (4.3) ns

no 1 1 0.43 122 1 1 . 7 0.000
kana 72 2.84 43 4 . 1 1 ns
kashira 4 0 . l 6 ---------- 1D

sa 1 0.04 76 7 .27 0.000
wa l 0.04 ---------- ID

20 ---------- l 0. 1 ID

yone 149 5.88 67 6.4 ns

yona ---------- 4 0.3_8 11)
total 1 150 45 .4 842 8 0 . 5 0.003

 

 

 

 

 

 

 

P5005: significant correlation; 0.05<PSO.1: strong tendency; ID=indeterminable due to low occurrence;
ns=not significant

The use of certain particles was associated with particular speech styles. Kashira and wa
were used only in formal speech, but their statistical significance was indetenninable due to
their infrequent occunences. All other particles, except for the utterance-final ne, were
used more frequently in informal speech than in formal style speech. More particularly, the
rates of utterance-medial and the no, sa and yona and the addressee-directed na were much
higher in informal style speech than in formal speech, confirming Uyeno's (1971)

observation that these particles are characteristic of informal speech. The differences in the

183

rates of no and sa, both particles of assertion, also proved to be significantly associated
with informal style speech.

As regards ne, while the total rate and the rate of the utterance-final ne did not
show a statistically significant difference between formal and informal speech (19.9/1000
words in formal and 17.8/1000 words in informal), the utterance-initial ne showed a strong

association with informal speech (P=0.001).

5.3.2. Age

Speakers' ages also influenced the use of some interactional particles in my data,
though not the rate of particles as a whole. The total rate was higher for older speakers
than younger speakers (62.1 and 52.5 per 1,000 words, respectively), but the difference
was not statistically significant.

Seven items (ne as a whole, the final ne, na as a whole, the medial na, the
addressee-directed na, sa, and yone) were significantly related to age (P5005), and the
particle no showed a strong tendency (0.05<PS01). The rates of the utterance-final ne and
yone were significantly higher for older speakers than younger speakers (P=0.004 and
0.003, respectively). The rates for na, no and sa were higher among the younger speakers
than the older ones, particularly the use of the utterance-medial and addressee—directed no.
This is probably due to the fact that the older speakers all used formal-style Japanese, in
which no, the addressee-directed na, and sa are not used often, whereas the younger
speakers used both formal and informal style Japanese, in which these particles are used
frequently. Therefore, we must await for future study to see the independent effect of age
and formality on interactional particle use.

The use of kana, kashira, wa and 20 were indetenninable due to their low rates.
The rates of the initial and the single-utterance ne, yo, and the self-directed na were not
significantly related to the speaker's age. Table 5.4 summarizes the rates of interactional

particles used by participants' age.

184

Table 5.4 The tokens and rates/1000 words of interactional particles by age (two-tailed t-test; n=36)

 

 

 

 

 

particles older (n=12) younger (n=24) P-value
tokens rates/1000 wds tokens rates/1000 wds cl=95%

ne total 457 39.1 582 24.1 0.005
(initial) ---------- ( 10) ( 0.4) ns
(medial) (144) (12.3) (170) ( 7.1) ns
(final) (301) (25.8) (388) (16.1) 0.004
(alone) ( 12) ( 1.0) ( 14) ( 0.6) ns
yo 84 7.2 167 6.9 ns
na total 16 1.4 135 5.6 0.011
(medial) ( ----- ) ( ----- ) ( 5) (0.21) 0.005
(final 1) ( ----- ( ----- ( 29) (1.2) 0.008
(final 2) ( 16) (1.4) (101) (4.19) ns

no 9 3.1 124 5.1 0.061
kana 36 0.3 79 3.3 ID
kashira 4 0.34 ---------- ID
sa 1 0.09 76 3.2 0.047
wa l 0.09 ----- ----- ID
20 ---------- l 0.04 ID
yone 118 10.1 98 4.1 0.003
yona ----- ----- 4 0.2 ID
total 726 6 2 . l 1266 52.5 ns

 

 

 

 

 

 

P5005: significant correlation; 0.05<PS01: strong tendency; ID: indetenninable; ns=not significant

5.3.3. Gender

In this section, the use of interactional particles is examined in terms of three

gender-related issues: (1) the use of interactional particles in relation to the speaker's

gender; (2) the use of interactional particles in relation to the gender of the addressee; and

(3) some changes occuning in the use of the gender-associated particles.

Some previous research has suggested that women use more interactional particles

than men (Peng 1981 and Hori 1979, reported in Shibamoto 1985). My data did not

support this observation. Only the initial and medial ne showed a statistically significant
relationship to gender (P=0.022 and 0.021, respectively). Though women's rates, except

no and 20, were higher than those of men, the differences were not significant. I conclude

that, at least in my data, the speakers' gender was not related to the frequency of

185

 

interactional particles. Table 5.5 summarizes the use of interactional particles by speakers'

gender.

Table 5.5 The tokens and rates of interactional particles by speakers' gender (two-tailed t-test; n=36)

 

 

 

 

 

particles men (n=18) women (n=18) P-value
tokens rates/1000 wds tokens rates/1000 wds cl=95%

ne total 448 26.8 591 29.5 ns
(initial) ( ----- ) ( ----- ) ( 10) ( 0.5) 0.022
(medial) ( 79) ( 5.0) (235) (11.7) 0.021
(final) (360) (22.9) (329) (16.4) us
(alone) ( 9) ( 0.6) ( 17) ( 0.9) ns
yo 79 5.0 172 8.6 ns

na total 89 5.7 63 3.1 0.086
(medial) (5) (0.3) ( ----- ) ( ----- ) 118
(final 1) (29) (1.8) (~----) ----- ) 0.081
(final 2) (55) (3.5) ( 63) (3.1) 0.058
no 16 1.0 117 5.8 ns
kana 44 2.8 71 3.5 ns
kashira ---------- 4 0.2 ID

sa 13 0.9 64 3.2 ns
wa ---------- 1 0.05 ID

zo 1 0.06 ---------- ID
yone 79 5.0 137 6.8 ns
yona 2 0.1 2 0.1 ID
total 726 47.4 1266 60.9 ns

 

 

 

 

 

 

P5005: significant correlation; 0.05<PSO 1: strong tendency; ID: indetenninable; nsmot significant

 

We now consider the use of interactional particles in relation to the addressee's
gender. For both men and women, the total rate of interactional particles per 1,000 words
was higher in speaking to a man than to a woman, though the difference was of marginal
significance for women (P=0.071) and not significant for men. The only particle which
showed a significant difference was the utterance-final ne and yone by female speakers to
female addressees (P=0.022 and 0.024, respectively). The addressee's gender did not
affect the use of other interactional particles in any significant way.

As regards the use of strongly feminine and masculine particles such as wa

(feminine) and 20 (masculine) it has been reported that both men and women have a

186

tendency to use them when talking to a same-sex partner (Peng 1981, cited in McGloin
1991 :25). My data showed mixed results. While the feminine particle wa (l token), the
addressee-directed na (34 tokens) and yona (4 tokens), were used with same-sex partners,
the feminine particle kashira (4 tokens) and the masculine particle 20 were used to opposite-
sex partners. However, since the number of tokens of these particles in my data, except
for na(a), was extremely small, we await future research concerning this point. Table 5.6

summarizes the use of interactional particles in relation to the addressee's gender.

Table 5.6 The rates of interactional particles by the addressee's gender (one-way ANOVA test; n=24)

 

 

 

 

particles M v. M M v. F P-value F v. M F v. F P-value
(n=8) (n=4) (n=4) (n=8)

ne total 39 23.2 ns 42.1 36.3 ns
(initial) ( ----- ) (----) N/A ( 1.5) ( 0.5) ns
(medial) ( 7.0) ( 4.8) ns (13.7) (19.3) ns
(final) (30.5) (18.4) ns (24.9) (15.8) 0.022
(alone) ( 1.4) ( ----- ) ns ( 2.0) ( 0.7) ns

yo 4.2 7 .8 ns 8.5 12.3 ns

na total 8.3 1.5 ns 4.5 2.3 ns
(medial) (0-8) ( ----- ) HS ( ----- ) ( ----- ) N/A
(final 1) (4.6) ----- ) ns ----- ) ( ----- ) N/A
(final 2) (2.9) (1.5) ns (4.5) (2.3) ns

no 1.9 1.2 ns 17.2 5.4 ns

kana 2.6 3.3 ns 2.7 4.8 ns

kashira ----- ----- N/A 0.75 0.1 ID

sa 1 .4 1.2 ns 1 1 2.3 ns

wa ---------- N/A ----- 0. 1 ID

20 ----- 0.3 ID ---------- N/A

yone 9 . 7 2.1 0.093 1 5 . 4 6.4 0.024
ona 0.3 ----- 1D 0.5 ----- ID

total 6 7 . 4 40.7 ns 1 03 70.1 0.071

 

 

 

 

 

 

 

 

P5005: significant correlation; 0.05<PSO.1: strong tendency; ID: indetenninable; N/A: no P-value due to
no occurrence; ns=not significant

Finally, we consider the changes taking place in the use of gender-associated
interactional particles. Recently, it has been noted that the distinction between Japanese
women's and men's speech is becoming less prominent (e.g. Reynolds 1993 and Okamoto

1995). My data on interactional particles provide some support for their conclusions.

187

While there still are strongly gender-associated interactional particles, some of them are
coming to be used by both sexes. For example, the particle no in declarative sentences,
which has been associated with female speech (e.g. Shibamoto 1985), was also used by
men in my data. Among men, 4/16 tokens (25%) of no were used in declarative sentences
and 12/16 tokens (75%) in questions, whereas among female speakers, 77/117 tokens
(65.8%) were in declarative sentences, and 40/117 tokens (34.2%) were in questions.
This supports McGloin's (1991) observation that the use of no in declarative sentences by
male speakers is increasing. Thus, while no in declarative sentences may still be
characterized as feminine, it is certainly undergoing a change.

Another particle which showed a change in its use is the utterance-final yona, which
has traditionally been associated with male speech (e.g. Uchida 1993, cited in Maynard
1997:73). It was used twice by AR, one of the young female speakers, when speaking to a
male addressee. Among the male speakers, yona was used twice in total (once each by two
speakers). Even though yona was used by only one female speaker and was "self-
directed," the use of yona by AR suggests that its use is also undergoing some change.

The use of yona by a female speaker, the use of no in declarative sentences by men,
the low occurrence of strongly masculine and feminine interactional particles such as 20,
wa and kashira, and the high occurrence of da (the plain form of the copula desu
considered to be masculine) among young female speakers all indicate that the
neutralization of masculine and feminine speech is indeed taking place in Japanese, at least
among young speakers of Japanese in casual conversations. It is interesting to consider
these changes in the light of Eckert's (1990) and Labov's (1991) observations on gender
and linguistic changes.

Labov (1991:205) argues that, as regards phonological changes (e.g. the Northern
Cities Chain Shift in vowels), women lead men at the initial stages of linguistic changes,
though women become more conservative than men in accepting changes as women are

recognized by the speech community. Eckert (1990:264-265) argues that not only the

188

speaker's sex but also his or her social status must be considered in the discussion of
gender-related linguistic phenomena. Both Labov and Eckert agree that women's
phonological behavior in part relates to their relative "powerlessness" in relation to men and
to their greater need to use social symbols, such as speech, for self-presentation (Eckert
1989:264; Labov 1991:144).

The strongly-masculine particle yona was used by a young female college student,
AR, and the plain form of copula, do, was used predominantly by younger female
speakers, though some of the Older female speakers also used it. Up to the time of college
graduation, very little power difference exists between men and women in Japanese
society. It may be that only in this "power-neutral" situation is women's speech and men's
speech less distinct, and, as women become recognized in society, their speech becomes

more conservative, as in the case of the older participants in the present study.

5.3.4. Speech genre

We now consider the use of interactional particles in the three types of speech
genres: interviews, casual conversations and narratives. In order to examine the
relationship between speech genre and the use of interactional particles, the one-way
AN OVA test was run at 95% confidence level on the rates of interactional particles per
1,000 words.

The total rate as well as the rates of all individual particles except for ya and na were
highest in conversations (the total rates: 708/1000 words in conversations, 41/1000 words
in narratives and 31 . 1/1000 words in interviews). The P-value for the total rate is 0.0000,
indicating a significant relationship between speech genre and the usage rates of
interactional particles. The result indicates that interactional particles, as their name
indicates, are truly important in two-way conversations, compared to one-way interactions

like narratives.

189

The speech genre particularly affected the use of ne (utterance-medially as well as
the ne as a whole), no, and sa. The particles sa, wa, and yona were used only in
conversations. Conversation was the only genre which included informal Japanese, which
probably explain the high rates of the informal particles no and sa. In narratives, ne, yo
and yone were frequent, constituting 91.9% of all particles used, compared to 77.9% in
interviews and 75% in conversations. Table 5.7 is a summary of the use of interactional

particles in interviews, conversations and in narratives.

Table 5.7 The rates of interactional particles by speech genre (one-way ANOVA test; n=42)

 

 

 

 

 

 

particles interviews (n=12) conversations (n=24) narratives (n=6) P-value
rates % rates % rates %

ne: total 17.5 (56.4) 3 6 . 1 (51) 29.5 (70.4) 0.0043
(initial) ---------- (0.45) (0.64) ( ----- ) ----- ns
(medial) ( 2.5) (8.0) (12.6) (17.9) (6.8) (16.3) 0.0193
(final) (30.5 ) (47.6) (18.4) (31.1) (24.9) (54.1) 0.1152
(alone) ( 0.22) (0.7) (1.04) (1.5) ( ----- ) ----- ns

yo 4.4 (14.2) 4.3 (12.2) 4.7 (11.1) ns

na: total 4.1 (13.2) 1.5 (6.1) 1.9 (4.4) ns
(medial) ( ----- ) (:----) (0-23) (0-32) ( ----- ) ( ----- ) 118
(final 1) ( ----- ) ----- ) (1.3) (1.9) ----- ) ----- ns
(final 2) (4.1) (13.2) (2.8) (13.2) (1.9) (4.4) ns

no 0.2 (0.5) 5 .9 (8.4) 0.9 (2.2) 0.0516
kana 2.6 (8.5) 3 .6 (5.0) 0.6 (1.5) 0.0694
kashira ----- ( ----- ) 0. 2 (0.3) ----- ( ----- ) 1D

sa ----- ( ----- ) 3 .5 (4.9) ----- ( ----- ) 0.0445
wa ----- ( ----- ) 0.05 (0.06) ----- ( ----- ) ID

20 ---------- ) 0 .05 (0.06) ----- ( ----- ID

yone 2.3 (7.3) 8 .4 (l 1.8) 4.6 (10.4) 0.0340

yona ----- ( ----- ) j 0. 2 (0.3) ----- ( ----- ) ID

total 31.1 (100) 70.8 (100) 42 (100) 0.0000

 

 

 

 

 

P3005: significant correlation; 005<P

5.3.5.

Summary

3

.1: strong tendency; ID: indetenninable; ns=not significant

In this section, a quantitative analysis of Japanese interactional particles was

presented. My data showed the following:

190

 

l. The rates of interactional particles were higher in informal speech than in formal
speech, a statistically significant result.

2. The relationship between age and the use of interactional particles was
inconclusive.

3. Neither the speaker's nor the addressee's gender were significantly related to the
rates of interactional particles.

4. Some of the strongly gender-related particles are beginning to be used by both
sexes; e.g. yona (a masculine particle) and no (a feminine particle) were used by both men
and women.

5. The rates of interactional particles are significantly related to speech genre: the
highest was in the conversation, and the lowest in the narratives, indicating the importance

of interactional particles particularly in informal conversations.

5.4. Qualitative analysis

In this section, the three most commonly used interactional particles, ne, yo and
yone are analyzed as contextual coordinates which show a speaker's feelings and
orientation toward propositions and/or the addressee. First, I present an analysis of ne in
Section 5.4.1, followed by an analysis Of yo in Section 5.4.2, and finally an analysis of
yone in Section 5.4.3.

All Japanese interactional particles occur in the utterance-fmal position, except for
ne, na and sa, which can occur both in the phrase-final and utterance-fmal positions. Ne
also occurs utterance-initially or as a whole utterance by itself. Most research thus far has
focused on the utterance-final ne. In the present study, a unified account of ne in all

positions is proposed.

5.4.1. Ne: a marker of agreement and cooperation

Ne has been characterized as a particle of confirmation and agreement (Makino and

191

Tsutsui 1986, and Takubo and Kinsui 1996) ), shared information (Kamio 1994), and
cooperative attitude and interpersonal involvement (Cook 1990). Table 5.8 summarizes the

previous analyses of ne by several researchers.

Table 5.8 Previous studies on ne

 

 

 

 

 

studies perspective characterization of ne

Uyeno (1971:113) speech acts speaker requests/provides agreement or rapport

Makino & Tsutsui functional speaker requests confirmation or agreement

(1986:286) from hearer about some shared knowledge

Cook (1988:155) indexical] interlocutors' general attitude of mutual
affective agreement

Ishikawa (1988:23; cognitive (1) information is in hearer's territory

1990:31) interactional (2) displays/elicits speaker's friendly attitude

 

Maynard (1993: 106) cognitive (1) only the hearer has the information
(2) speaker (S)'s infoe hearer (H)'s info.
(3) 8'8 info.> H's info. (to present information
with a sense of detachment)

 

 

Kamio (1994:86) cognitive info. in hearer's territory 2 speaker's territory
Takubo & Kinsui (1996) cognitive indication that speaker is calculating the validity
of the proposition

 

As shown in Table 5.8, ne has generally been analyzed from two perspectives: (l)
the cognitive perspective, e.g. Kamio's "territory of information" (1990), Maynard's
"amount of knowledge and meta-knowledge" (1993), Takubo's (1992) and Takubo and
Kinsui's (1996) "mental space" theory; and (2) the affective/interpersonal perspective, e.g.
Cook (1988 and 1990). Only Ishikawa (1988 and 1990) analyzes ne from both cognitive
and interpersonal perspectives.

I agree with Ishikawa's approach and propose that Schiffrin's framework be
adopted for the analysis of ne for the following two reasons. First, by incorporating the
five planes of talk proposed in her discourse model, Schiffrin's approach can account for
the various functions of ne which operate on more than one plane of talk. Secondly, since
all instances of ne, regardless of their positions in an utterance, are derived from the same

origin (see Onodera 1993), all the functions of ne must be regarded as related to each other

192

in some way, and Schiffrin's framework can capture the commonality in all uses of ne.

It is evident that not all functions of ne can be explained solely from a cognitive
perspective such as the "territory of information" or "mental space" theories, since ne
expresses the speaker's orientation towards the addressee as well. Takubo and Kinsui's
(1996) mental space theory may be able to account for most uses of the utterance-final ne.
Neither their theory nor Kamio's theory, however, can fully account for the function of the
utterance-medial ne as an attention—checking device, which thus must be accounted for by
interactional as well as cognitive terms. Thus, since it is impossible to analyze the
functions of ne and other Japanese interactional particles from just one perspective, it is
necessary to adopt a multi-faceted approach like Schiffrin's.

In what follows, I argue that ne is a marker of speaker-bearer cooperation, and also
that it displays the speaker's assessment of the addressee, as ne is generally used with an
addressee whom the speaker knows well. As such, ne is used in situations where the
speaker wishes to display his or her cooperative attitude to the addressee or to enlist the
addressee's cooperation, including mitigating upcoming face threatening acts. My
approach is close to Cook's (1990) analysis ne, which is from an affective perspective. In
addition, I incorporate a cognitive approach in my analysis, since the use of ne is related to
the interlocutors' knowledge and meta-knowledge as well, and thus cannot be analyzed
solely from an affective perspective.

Six specific functions of ne were found in my data, all relating to cooperation: (I) to
request or express agreement; (2) to request or provide confirmation; (3) to soften the
preceding expression; (4) to preface a repair; (5) as an attention-checking device; and (6) as
an attention-getter. The fnst three functions above are examples of ne used utterance-

finally, (4) and (5) utterance-medially, and (6) utterance-initially.

5.4.1.1. Requesting or expressing agreement

First, I discuss the function of ne used to request or express agreement. The most

193

typical use of ne as an agreement marker was in expressions such as soo desu ne 'yes,
indeed' or as a single-utterance Ne, which can be interpreted as an abbreviation of a repeat
of an immediately preceding utterance. In my data, of 689 tokens of the utterance-final ne,
148 tokens (21.5%) were used as soo desu ne 'yes, indeed.’ There were 26 tokens (3.8%)
of ne as single-utterances. Ne in both of these cases was used to express agreement with a
prior utterance. Thus, over 25% of all the instances of ne in my data were used to express
agreement.

Example (1) illustrates the use of ne in this function. FY and SH are talking about a
university in Japan which is considered to have excellent facilities, which they have not yet

visited.

(1) FY: a. 'T Daigaku ne/ sugoi ne.

name university IP awesome IP

SH: b. Kondo itte mitai ne.
this time go try [P

c. Kengaku shite mitai ne.
observe do try IP

FY: (1. Ne. Chotto ryoo haitte mitai ne.
a little dorrn enter try IP

SH: e. Haitte mitai ne nanka. So kkaa. Naruhodo ne.
enter try kind of so Q Isee IP

FY: a. 'T University is awesome, isn't it?
SH: b. I want to visit it some day, don't you?
c. I want to see what it is like, don't you?

FY: d. I agree. I want to enter the dormitory, don't you?
SH: e. Yeah, I want to try entering, somehow. I see, I see.‘

In (a), after hearing about the university, FY says "It's awesome" to which SH responds,
"I want to visit it one day" and "I want to see it, don't you?" SH ends his utterances with
ne, which elicits agreement from FY. In ((1), FY provides agreement with a single-

utterance "Ne" and continues with "I want to go inside their dormitory, don't you?,"

194

ending his utterance with ne also. Immediately after FY's ne, SH starts his turn, saying "I
want to go inside the dormitory too, don't you?" again ending with ne. Thus, in this
example, we can see ne being used repeatedly and for both requesting and expressing
agreement by the two speakers, displaying their cooperative attitude toward each other. Ne
in this function is generally pronounced with a middle sustained pitch with a glottal stop
following it or the vowel elongated as nee, rather than with definite utterance-final falling
intonation.

Ne as a marker requesting or providing agreement operates in the following planes
of talk: (1) action structures, eliciting and providing agreement; (2) participation
fiarnework, displaying the speaker's desire to involve other interlocutors in the

conversation; (3) exchange structures, when a speaker-change occurs.

5.4.1.2. Requesting or providing confirmation
The second function of ne is to request confirmation from the bearer. In (2), MA,

an avid soccer fan, and I are talking about the World Cup soccer matches.

(2) Mieko: a. Sakkaa sugokatta desu ne/ kotoshi no natsu wa/ anoo Amerika kappu.
soccer exciting-PAST BE IP this year's summer-TOP uh America cup

MA: b. Aa waarudo kappu ne/ kotoshi no waarudo kappu wa sugokatta.
oh world cup IP this year's world cup-TOP exciting-PAST

Mieko: c. A waarudo kappu desu ka.
oh world cup BE Q

d. Atashi Nihon ga kuru ka to omotte tanoshirni ni shitetara
I Japan-SUB come Q QT think look forward to

e. yosen de makechatta n desu ne?
elimination match in lose-PAST assert IP

MA: f. Ee zannen nagara hikiwakete ne/ anoo hikiwakete/
yes regrettable tie IP uh tie-and

g. sore de saigo no/ uun 30—byoo gurai de makete shimairnashita ne.
that with last uh 30 seconds about in lose-PAST IP

195

Mieko: a. 'Soccer matches were exciting, this summer, well the Americas Cup.
MA: b. Oh you mean the World Cup. This year's World Cup WAS exciting.
Mieko: c. Oh, it was the World Cup, was it?
(1. I thought Japan might come to the U.S., and was looking forward to it,
e. but they lost during the elimination round, right?
MA: f. Yes, regrettably, they tied y'know/ well they tied and
g. in the last, well, 30 seconds, they ended up losing, y'know.'

In (e), "Japan was defeated during the elimination round ne" was pronounced with rising
intonation, which was interpreted as a tag question. In response, MA says ee 'yes' and
provides an account of how Japan lost the match. Ne in (e) also serves as a device to yield
the turn to the other interlocutor, since the question elicits the addressee's response. Ne in
this function is used when the speakers assume that the addressee has an equal amount of
or more information. In (g), MA finishes providing information with ne, pronounced with
falling intonation.

Thus, ne is used both to request and provide confirmation. In these functions, ne
operates in the information state (displaying the speaker's assumption about the hearer's
knowledge) as well as in action structures (requesting confirmation and cooperation), the
participation framework (showing interest in what other interlocutors have to say), and the

exchange structure (initiating speaker—change).

5.4.1.3. In disagreement
Ne is also used to soften the tone, following a negative response. Here, too, the
speaker solicits the addressee's cooperation. One such example is the use of ne following

an expression of disagreement, as shown in (3).

(3) Mieko: a. Amerika-jin igai no hito ga hanasu Eego no hoo ga
American other persons-SUB speak English-SUB

b. wakan-yasui to omoirnasen? nantonaku.
understand-easy QT think-NEG somehow

196

YI: c. 800 de wa nai desu ne.
so BE NEG BE IP

(1. Doitsu-jin wa yoku shaberu n desu kedo/
German-TOP often speak BE but

e. dokutoku no namari mitaina no ga atte/ are wa chotto kikitorinikui shi
unique accent like one-SUB exist that-TOP a little hear-difficult and

Mieko: a. 'English spoken by people other than Americans,
b. don't you think it's easier to understand it, somehow?
Y1: c. That's not the case, y'know.

d. Germans speak a lot, but
e. they have unique accent-like thing, and it's hard to understand them, and'

In (c), YI expresses disagreement soo de wa nai desu 'that's not so,‘ adding ne at the end.
Then he starts explaining specific reasons why he thinks it is not the case. In this situation,
since the information belongs exclusively to the speaker, soo de wa nai desu 'that's not so'
alone or with the final particle of assertion yo, as soo de wa nai desu yo would be expected
if we simply base our analysis on the "territory of information." However, the speaker's
choice was ne, and without the ne in this case, his answer would have sounded too
assertive and blunt. Additionally, ne may add a sense of detachment as if the speaker is
expressing his/her opinion objectively.8

Therefore, ne in this example can be analyzed as expressing the speaker's effort to
appear cooperative towards the addressee by softening the tone and mitigating the face
threatening act which has just occurred. At the same time, the speaker requests a
cooperative acceptance from the addressee of the negative response he makes. Ne has the
effect of softening the impact of disagreement as well as inject a sense of objectiveness.
This is an example which shows that ne cannot be analyzed by a cognitive approach alone.

Another example of ne for requesting cooperation was observed in its frequent use
after the adversative marker demo. In my data, 23/255 tokens (9.0%) of demo were

immediately followed by ne as demo nee 'but y'know'. This percentage is much higher

 

8 See Maynard ( 1993a: 106) where she states that ne may be used when the speaker wishes to present
his/her remarks with a sense of detachment or distance.

197

than other connectives, e.g. de ne 'and y'know' (8/38122.1%) and dakara ne 'so y'know'
(5/198=2.5%). It is quite plausible that speakers should try to mitigate, by the use of ne,
the face threatening acts which they are about to introduce. Ne in this function, then,
operates in the participation framework, displaying the speaker's concem about the

speaker-bearer relationship.

5.4.1.4. Prefacing repairs
Another function of ne is to preface repairs, which also necessitates the hearer's

cooperation. In (4), TI‘ has been talking about his impressions of Americans.

(4) TT: a. 'Minna ga sorezore ano "Hi" tte itta toki ni miseru egao tte yuu no ga
everyone-SUB each well hi QT say when show smile QT say one-SUB

b. hijooni nanka shin/

very something

c. nee/ inshoobukai tte yuu ka.
impressive QT say Q

Mieko: d. Are wa ii desu nee.
that-TOP nice BE IP

TT: a. 'The smile everyone shows when they say "Hi"
b. very/ well, shin/
c. y'know/ it impressed me or what shall I say.
Mieko: (1. That's really nice, isn't it.

In (b), TT says shin (perhaps the initial part of shinjirarenai 'incredible'). Then he pauses
and says nee 'y'know' and rephrases his remark, saying inshoobukai 'impressive'. There
were only four cases of ne used immediately before a repair in my data. It is still an
impOrtant use of ne, since it helps create rapport and a cooperative atmosphere between
interlocutors. It also expresses the speaker's desire to appeal to the addressee in making a
repair, so that a successful correction can be made with the cooperation from the addressee.

Ne prefacing repairs operates in the participation framework (displaying the speaker's

198

desire for cooperation from the addressee) and action structures (enlisting cooperation).

5.4.1.5.

As an attention-checking device

Ne was frequently used utterance-medially as an attention-checking device, another

situation where the addressee's cooperation is needed. In (5), NR is presenting her ideas

about the use of fillers in Japanese conversation.

(5)

NR:

Mieko:

NR:

Mieko:

NR:

Mieko:

NR:

Mieko:

NR:

Mieko:

a. Dakedo nee/ tatoeba nee/ purezenteeshon no umai hito tte no wa sa/
but IP for example IP presentation-SUB good person QT one—TOP IP

Un.
yeah

b. sooyuu no ga sukunai no yo.
such one-SUB few IP [P

c. Nde nee/ atashi omou n da kedo nee/
and IP I think BE but IP

Un.
yeah

(1. uchi no ofisu nanka de mo sa/ koo kaigi o yaru ja nai.
my-GEN office something even IP this meeting-DO do TAG

e. Sooruruto nee/ ano heta ni nee/
then IP well badly IP

f. Nihontekina hassoo o motteru no wa ne
Japanese way of thinking have one-TOP [P

g. otoko no hito ni kagitte nee/ sooyuu no ga ookute nee/
men limit [P such one—SUB many IP

Urr un.
yeah yeah

h. ooi tte yuu ka
many or what

a. 'But y'know, for example y'know, those who are good at
presentations
Yeah.
b. they don't use many fillers, I tell you.
c. And y'know, what I think is y'know
Yeah.
d. in my office y'know, we have meetings, as you imagine.

199

e. Then y'know, awkwardly y'know/

f. those who stick to a Japanese way of thinking y'know

g. typically among men y'know/ they are plenty of them y'know/
Mieko: Yeah, yeah.
NR: h. or what?‘

NR uses ne nine times in this segment, almost after each phrase while she presents her
ideas. When she does not use ne( e), another attention-checking particle sa is used as in
((1). By the fiequent use of ne and sa, NR can elicit the hearer's interest and cooperation,
and also check whether or not the addressee is paying attention to her utterance, entitling
her to continue her turn. The effect of ne as an attention-checking device is evident in the
frequent back—channelling expressions an (un) 'yeah (yeah),' which indicate the
addressee's interest and willingness to hear the story. Ne as an attention-checking device
operates in the participation framework, displaying the speaker's intention to continue and
his/her concem about the addressee's attention to his/her utterance and in the action

structure, eliciting the addressee's cooperation.

5.4.1.6. As an attention-getter

The utterance-initial ne is used to get the addressee's attention when, for example,
initiating a turn or introducing a new topic into the conversation. This is another situation
where the speaker needs the addressee's cooperation. In (6), ne is used to get attention

before initiating a turn and topic change.

(6) KB: a. Dakara ato l-kko de Matsudo da yone.
so more one with Matsudo BE 1P

YI: b. Un Tookyoo to wa na bakari no——
yeah Tokyo QT-TOP name only

KE: c. —Nee sakki nee/
IP a while ago [P

d. sakki nee/ Masumi-chan ni yutta n da kedo/ kyoo no asa sugoku nee/
a while ago IP Masumi-Miss to say BE but today GEN morning very IP

200

e. koohii nomi ni ittara nee
coffee drink to go-PAST when IP

a. '80, it'll be Matsudo if we go one more station.

b. Yeah, it's Tokyo only by name.

c. Y'know a little while ago, y'know

d. a little while ago y'know, I told this to Masumi too, but this morning, very
e. when I went to have coffee, y'know?

5.3%.

In (c), KE starts her turn when YI has barely finished speaking, using nee. By the use of
nee, KE mitigates the face threatening act of taking a tum from YI. Nee in this example
operates in the: (l) participation framework, displaying the speaker's eagerness to tell
something to the addressee and to mitigate the face threatening act which is about to come
up; (2) exchange structure, initiating a turn; and (3) action structure, enlisting the
addressee's cooperation. The second nee in (c) and nee in (d) are used to enlist the
addressee's cooperation in introducing a new topic and checking his/her attention. It thus
functions in the action structure and participation framework. In all these cases, ne( e)
serves to create a cooperative atmosphere so that the speaker can proceed with his or her
intended remark.

We have thus far examined various functions of ne and have seen that ne operates
on three planes of talk proposed by Schiffrin. Cognitively, ne is used when the speaker
assumes that the hearer has as much or more knowledge about a proposition as the speaker
(Kamio 1990). Interactionally, ne is used when the speaker is requesting or expressing a
cooperative attitude. From the examples examined thus far, ne seems to be more
significant as an interactional marker of cooperation than as a cognitive marker.

As we have seen in several examples above, as a textual coordinate, ne indexes an
utterance both to prior and upcoming segments. As a participant coordinate, it indexes an
utterance both to the speaker and the hearer, enhancing agreeableness between the
interlocutors and displaying the speaker's judgement about the speaker-addressee

relationship as "familiar."

201

Thus, ne in several situations cannot be explained solely from cognitive
perspectives. In (1b) and (1c), for example, since SH is expressing his own thoughts, he
either should use no particle or use yo, if we base our analysis on the cognitive approach.
However, SH uses ne instead. The most plausible explanation for this is that he is
requesting agreement from FY, which means an interactional need and motive can override
the cognitive function of particles (the use of ne rather than no particle or the use of yo in
this case). Ne as an attention-getter or as an attention-checking device are other examples
where ne cannot be accounted for solely from a cognitive perspective. Hence it is
important to incorporate an affective perspective in an analysis of ne.

As stated in the preliminary section of this chapter, the validity of previous research
on ne itself is not questioned. Rather, I am proposing that, given various functions of ne,

we should employ a more integrated approach to account for all the roles of ne.

5.4.1.4. Summary

In this section, I examined various functions of ne as a marker of cooperation when
requesting (or providing) agreement or confirmation, in repairs, and as an attention-
checking device. Figure 5.] summarizes the general and specific functions of ne as

discussed in this section.

 

 

General function Specific function
i I. interactional [(1) to request or provide agreement
marker of i (2) to soften disagreement
ne 1 cooperation (3) to preface repairs

(4) to check the addressee's attention
(5) to get attention
i(6) as a tum-initiator

L 11. cognitive marker of

 

territory of information---- (7) to request or provide confirmation

Figure 5.1 Functions of ne

202

We have also seen ne used to mitigate face threatening acts such as expressing
disagreement or taking another speaker's turn. Ne should not be analyzed from one point
of view alone, such as cognitive or interpersonal, but should be analyzed from various

perspectives.

5.4.2. Y o: a marker of the speaker's emphasis

Yo was the second most common final particle in my data, constituting 13% of all
the particles used. Yo has been characterized as a particle of assertion or speaker's
insistence (e.g. Uyeno 1971 and Makino 1986) and is often translated as "I tell you," "I

warn you," or "I'm sure." Table 5.9 summarizes previous research on yo.

Table 5.9 Summary of previous studies on yo

 

 

 

 

 

studies perspective characterization of yo

Uyeno (1971:109) speech acts speaker's insistence

Makino & Tsutsui (1986:543) functional assertion or (fairly) strong conviction

Cook (1988:129) indexical/ point to speaker's utterance (direct meaning)
affective assertive attitude (indirect meaning)

Maynard (1993:106) cognitive speaker's info. > hearer's info.

 

Takubo & Kinsui (1996:72) cognitive speaker tells hearer to register something in
the "indirect-experience" domain

 

As Table 5.9 shows, cognitively, yo is used when the speaker assumes that s/he has more
knowledge than the addressee. Functionally, yo conveys the speaker's assertive attitude,
strong conviction or insistence.

In what follows, I argue that yo is a marker which conveys the speaker's emphasis
on a proposition and simultaneously displays his/her orientation towards the addressee, i.e.
either the speaker is higher in status than the addressee, or they are equal and/or in a
friendly relationship. The term "emphasis" here is defined as the "speaker's attempt to

strongly convey to the addressee the importance of the message." A speaker might wish to

203

i ___l i,
L /_ -

convey emphasis for various reasons: (1) to provide emphatically information that the
addressee does not have; (2) to show his/her strong conviction about the proposition; or (3)
to show his/her strong feeling about the proposition. In the rest of this section, the terms
"emphasis," "intensity" and "speaker's conviction" are all used interchangeably.

In what follows, five specific functions of yo as an emphatic marker are examined.
I first discuss the use of yo when presenting information unknown to the bearer in Section
5.4.2.1 , then the use of yo in disagreement in Section 5.4.2.2, in providing confirmation

or agreement in Section 5.4.2.3, in advising and warning in Section 5.4.2.4, and finally in

request or invitation in Section 5.4.2.5.

5.4.2.1. Presenting information unknown to the hearer

Many cases of yo in my data were used to present information unknown to the
hearer. In Kamio's (1990) term, the information belongs to the speaker's territory.
However, yo in this situation is mostly optional, except in advice or in warnings. In other
words, the speaker can accomplish the task of conveying his/her exclusive information
without yo. The reason the speaker chooses to use yo instead of the bare form, I argue, is
to emphasize the information and/or to convey the sense of closeness in familiar
conversations. As such, yo is not usually used to one's social superiors.

In my data, 234/251 tokens (93.2%) of all instances of yo were used for this
purpose, whereas 12 tokens (4.8%) were used for confirming prior utterances, 3 tokens
(1 .2%) for warning, and 2 tokens (0.8%) were for invitation.

I now discuss two situations where yo is used to convey the speaker's emphasis in
presenting information unknown to the addressee: (1) presenting the speaker's views; and

(2) describing the speaker's own experience.

5.4.2.1.1. Presenting the speaker's views

The first situation where ya is used to emphasize a proposition unknown to the

204

addressee is when the speaker presents his/her views. In (7), SH and FY are talking about

American society.

(7) SH: a. Ma demo nee/ de Amerika wa tonikaku jitsuryoku shugi da ne.
well but IP and the U.S.-TOP anyhow ability principle BE IP

FY: b. 800 da ne/ Amerika wa.
so BE [P the U.S.-TOP

c. Sore wa ii koto da to omou yo boku wa.
that-TOP good thing BE QT think I-TOP

SH: a. 'We]! but, y'know, the U.S. is really an ability-based society, isn't it?

FY: b. You are right, the U.S. is.
c. I think that's good, y'know?’

In (b), FY first expresses his agreement with ne to what SH has just said. Then in (c), he
adds his own view, ending his utterance with yo. In my data, 51/251 tokens (20.3%) of
yo were used to express the speaker's views, of which 16 tokens (6.4%) were
accompanied by omou 'I tlrink', as in this example.

It is widely agreed among researchers (e.g. Maynard 1993; Kamio 1990; and
Takubo and Kinsui 1996) that cognitively yo is used when the speaker assumes what s/he
has just said is unknown to the addressee. FY does exactly that in (7c) above.

However, (c) would have been acceptable without the yo as an expression of FY's
view. Therefore, the use of yo must convey more than the fact that the information belongs
to the speaker. I argue that the reason FY uses yo in (c) is to emphasize his opinion in
order to convey to SH how strongly he feels about the proposition; yo also signals to SH
that they have friendly and equal (or high—to-low) relationship .

When a speaker presents personal views or experiences which belong to his or her
territory, and if s/he feels strongly, yo is used. However, even if the information belongs
to the speaker's tenitory, if s/he does not wish to convey how strongly s/he feels about the

matter, s/he can always choose not to employ yo, or use ne instead. Therefore, the use of

205

interactional particles such as yo and ne is determined by the interaction of three factors: (1)
the speaker's knowledge and meta-knowledge; and (2) the speaker's feelings and sense of
certainty; (3) the speaker-hearer relationship, which determines in part whether or not it is
appropriate for the speaker to express feelings emphatically and/or convey the sense of
friendliness.

Yo operates in the information state, signalling the speaker's assumptions about the
addressee's knowledge about an upcoming utterance. At the same time, yo operates in the
participation framework, displaying the speaker's strong feelings about the proposition
presented and his or her orientation towards the addressee, pointing out their friendly and

equal or high-to-low relationship.

5.4.2.1.2. Describing the speakers' experiences
Another function of yo was to emphatically present information unknown to the
addressee when describing the speaker's own experiences. In (8), IM is recounting the

trouble she had with her first roommate in the U.S.

(8) IM: a. Anoo nanka karada ni warusoona mono bakkari tabeteta n desu yo/
well something body for bad-seem thin only eat-PAST assert 1P

b. anoo daietto shiteta n desu yonee/ kanojo ga.
well diet do—PROG-PAST BE IP she-SUB

c. Sorede maa yasetai bakkari ni/ demo arnai mon wa tabetai n desu yo.
and kind of slim down-want but sweet thing-TOP eat-want assert 1P

IM: a. '(she) only ate things which looked bad for her body, I tell you.
b. well she was on diet, y'know. '
c. And well because she really wanted to lose weight, but she wanted to eat
sweets, I tell you.‘

In (a) and (c), [M uses yo. In this example, as in (7) above, the speaker could have
presented her description without yo if she did not wish to convey how strongly she felt

about the situation or to show her willingness to be friendly. Therefore, yo seems to

206

display IM's desire to convey her emphatic feelings about the matter and her friendliness
towards the addressee at the same time.
Yo was also used to report what the speaker had heard which s/he assumed was

unknown to the addressee, as illustrated in (9).

(9) Mieko: a. Uchi no shujin mo are ga ichiban suki datta mitai.
my husband also that-SUB most like BE-PAST seem

AR: b. A nanka ittemashita yo ima.
oh something say-PROG-PAST IP now

c. Uchi no okusan wa kirai na n da yo tte.
my wife-TOP dislike assert IP QT

Mieko: a. 'My husband also liked that ride most, it seems.
AR: b. Oh, he was telling me just now, y'know,
c. "My wife doesn't like that."

In (b), AR reported to me what she heard prior to this segment. When a speaker quotes
what someone said that is unknown to the addressee, s/he often uses yo, as in itteta yo
's/he told me y'know'.

In (9), the use of yo is natural cognitively, since AR is reporting what she has
heard, but assumes it is unknown to the addressee. At the same time yo signals AR's
emphasis on the presented information and friendly posture. The same utterance presented I
without yo will lack emphatic tone and familiarity.

Thus, yo in this function Operates in the information state, signalling the speaker's
assumptions that s/he has exclusive knowledge, and in the participation framework,

displaying his/her friendly feelings towards the addressee.

5.4.2.2. In disagreement

Another function of yo was to express disagreement emphatically, as in (10).

207

(10) Mieko: a. Majimena ii gakusee datta n desu ne.
serious good student BE-PAST BE IP

IM: b. Ee? sonna koto nai desu yo/ tada nanka gamushara ni totteta dake de.
what such thing NEG BE IP only something randomly take-PROG-PAST only

Mieko: a. 'You were a serious and good student, weren't you?

IM: b. Oh no, that's not true, I tell you. I was simply taking (courses)
randomly.’

After hearing that [M finished taking all the necessary credits for graduation by the end of
her junior year in college, I inferred that IM was a hard-working student and made the
comment in (a). IM expresses her disagreement with yo in (b), perhaps out of humility.
To decline compliments is a social norm in Japanese society. Cognitively, the use of yo in
(b) is quite natural, since IM knows herself best. As in the case of (7)-(9), IM's utterance
would have been acceptable without yo, which suggests that yo is used to indicate the
source of information as well as to emphasize her denial without spoiling the friendly
relationship between the speaker and the addressee. (11) is another example of yo in

disagreement.

(11) KY: a. M ga kita no 2-nen datta kke.
M-SUB come-PAST-NOM sophomore-year BE-PAST Q

UM: b. 2-nen da yo. 2-nen no
second-year BE IP second-year GEN

KY: c. 2-nen no haruyasumi da kara
sophomore-year GEN spring break BE so

UM: d. 800 da yone. A chigau. 1-nen no haru-yasumi.
so BE IP oh wrong freshman-year GEN spring break

e. Huyu da/ huyu huyu huyu.

winter BE winter winter winter

KY: f. E chigau yo. Haru-yasumi da yo. Atashi san-gatsu oboeteru mon.
what wrong IP spring-break BE IP I March remember assert

UM: g. A so kkaa.
oh so Q

208

KY: a. 'Wasn't it our sophomore year when you came to my house?
UM: b. Our sophomore year, I'm sure. Our sophomore year' 5
KY: c. Our sophomore year' 5 spring break.
UM: d. It was, wasn't it? Oh, wrong! The spring break of our freshman year.
e. In the winter it was. Winter, winter, winter.
KY: f. What? That's wrong, I'm sure. During the spring break, I tell you.
I remember it was in March.
UM: g. Oh I see.’

In (1 1), not remembering exactly when UM visited KY's house, they both assert that it
was their sophomore year. KY's use of yo in e chigau yo and haruyasumi da yo in (f)
expresses her certainty that it was during a spring break, not a winter break, which she
substantiates with her remark atashi 3 -gatsu oboeteru mon T remember it was in March.‘
KY uses yo to emphasize her certainty as well as point out their friendly relationship.

In this example, the information about the time of the visit belongs to both
interlocutors' territories since they shared the experience and furthermore, they both believe
that they are correct. The use of yo seems to be related to various factors here; e.g. the
territory of information, the speaker's certainty about the information, how much of that
certainty they wish to display to the addressee, and the familiarity which exist between the
interlocutors. Thus, yo operates in the information state, displaying the speaker's
assumption about the territory and amount of information, and in the participation
framework, indicating the speaker's wish to emphasize what has just been said and his/her

desire to reassure the friendly relationship among the interlocutors.

5.4.2.3. Providing confirmation or agreement

Yo was also used to provide confirmation or agreement, as illustrated in (12).

( 12) SH: a. Okinawa no chikaku na n de/
Okinawa-LOC near assert—and

b. dokutoku no ano jamisen tte yuu no ga aru n desu yo.
unique well jamisen QT say one-SUB exist assert IP

209

Mieko: c. E? ano hebi no kawa tsukau?
oh that snake—GEN skin use

SH: (1. S00 na n desu yo.

so assert IP

SH: a. 'Because (Amami-Oshima is) near Okinawa,

b. they have a unique musical instrument called "jamisen."
Mieko: c. Oh? that is the instrument made with snake skin?
SH: (1. That's right!‘

In ((1), SH confirms with yo my assumption that jamisen is made with snake skin. Since
SH has more knowledge about the instrument, the use of yo here is natural.
(13) is an example of yo used when both interlocutors have information. AR and I

are talking about amusement parks.

(13) AR: a. Shiidaa into wa asui desu one.
P0 Y y
place name-TOP inexpensive BE IP

Mieko: b. 22-doru gurai desho? 1-nichi de.
22 dollar about TAG one-day for

AR: c. Hurii pasu de/ un.
free pass with yeah

Mieko: d. Datte Dizuniirando nante Tookyoo no sugoku takai desu mon nee.
because Disneyland -TOP Tokyo one very expensive BE IP

AR: e. Takai desu yo.
expensive BE IP

AR: a. 'Cedar Point is inexpensive, isn't it?
Mieko: b. It's about 22 dollars a day, isn't it?
AR: 0. With unlimited number of rides, yeah.

Mieko: d. Disneyland, the one in Tokyo, is very expensive, isn't it?
AR: e. It certainly is!’

In (13), both interlocutors have first-hand knowledge about the amusement parks because
they both have been there. In (e), AR could have used ne instead of yo, as in takai desu ne

since she is expressing agreement with what I said. The use of ne here, however, makes

AR's statement mere agreement without displaying her own feelings. In contrast, the use

210

of yo expresses AR's strong feelings about how expensive Tokyo Disneyland is.

5.4.2.4. In advising and warning

Yo is also used in friendly advising or warning, another situation where the
speaker's emphasis is called for. In (14), KO and YA are talking about the terrible-
smelling candy KO has purchased. After tasting one piece of candy, YA warns KO about

the danger of trying a new product in a teasing tone.

(14) KO: a. Nanka onnaji no kau no wa noo ga nai to omotte
well same one buy-NOM-TOP brain-SUB exist-NEG QT think-and

b. chotto choosen/ booken o shite mita no/ atashi ni shite wa mezurashiku.
a little challenge adventure-DO do try-PAST [PI for TOP rare

YA: c. Ame de booken suru no wa kiken da yo.
candy with adventure do-NOM-TOP dangerous BE IP

KO: a. 'I though it is kind of silly of me to keep buying the same candy, and

b. I challenged a little/ tried some adventure/ unusual for me.
YA: (1. It's dangerous to be adventurous with candy, I tell you!’

In (15), IM is reporting what she told her roommate about her eating habits.
(15) IM: a. Karada ni warusoona mono bakkari tabeteru kara
body for bad-looking thing only eat-PROG because

lo. "karada ni warui n ja nai no? chanto tabeta hoo ga ii yo" ttsuttara
body for bad TAG Q properly eat had better say-PAST when

IM: a. '(She) was eating only unhealthy things, so

b. I said to her, "Aren't they bad for your health? You should eat more
properly, I tell you'”

In (b), yo is used to emphasize the importance of her advice to her roommate. Unlike
Examples (1) - (13) above, where yo was optional, the yo in (14) and (15) is mandatory, if
the advice is personally directed to the addressee. Y 0 indicates the importance of the

message as well as making it as personal and directly relevant to the addressee. Without

211

the yo, the advice in (14) and (15) would sound like mere factual statements.

5.4.2.5. In request or invitation
Yo was also used for requesting and invitation in my data. In (16), AR and SK
have been conversing, but SK had not said much before this segment. So, AR tells SK to

contribute more to the conversation.

(16) AR: a. S-kun anmari shabette nai jan. Shabette nai yo chotto.
name Mr. much speak-NEG TAG speak-NEG IP a little

b. Tanomimasu yo.
request IP

SK: c. Dekinai/ dekinai. Nanka wadai o kaete kudasai.

cannot cannot something topic-DO change please

AR: a. '8, you haven't been saying much. You haven't, hey.
b. I ask you (to speak more) yo.
SK: c. I can't. I can't. Please change the topic.’

(17) is an example of yo in invitation, in which KY invites UM with yo to come to visit

her at her grandmother's house.

(17) UM: a. Aa kireedaroo ne.
oh pretty BE-AUX IP

KY: 1). J a kondo obaachan chi asobi ni oide yo.
then next time grandma house visit come IP

UM: a. 'Oh that (KY's grandmother's hometown) must be pretty!
KY: b. Then, come and visit me there next time yo.‘

Yo in both (16) and ( 17) may be translated into "Really" in English, expressing the

speakers' strong feelings about their remarks. Yo also makes the utterance sound much

more personal. Rather than having to do with the territory of information, yo in these

212

functions operates in the participation framework, expressing the speaker's strong feelings
and friendliness.

We have seen that yo primarily functions in the information state and the
participation framework, indicating the source of knowledge and expressing the speaker's
emphasis when presenting opinions, experiences, assumptions, warnings or invitations.
In all situations discussed above, except for warning and advice, the task could have been
achieved without yo, or with the use of ne in case of expressing agreement.

The speaker's choice of yo is determined by several factors: (1) the location and
amount of knowledge and meta-knowledge among the interlocutors; (2) how much
emphasis the speaker wishes to convey; (3) the speaker's judgement as regards the
relationship among the interlocutors (the addressee's status must be lower or equal to the
speaker) and his/her desire to make the utterance personal and friendly. Yo cannot be used
when a piece of information belongs solely to the hearer. Likewise, even if the speaker has
exclusive knowledge, yo need not be used if s/he does not wish to convey a sense of
emphasis and friendliness.

As a textual coordinate, yo indexes an utterance to prior text, emphasizing the
proposition. As a participant coordinate, yo indexes an utterance both to the speaker and
the bearer, expressing the speaker's assumptions about knowledge and meta-knowledge,

and about the speaker-hearer relationship.

5.4.2.6. Y0 and n e

Thus far, we have seen several cases where both ne and yo are possible as
interactional particles; e.g. following an expression of disagreement in Section 5.4.2.2, or
providing confnmation in Section 5.4.2.3. In some situations yo is preferred where ne
would typically be more appropriate, as in Example (1 3). In other situations, to soften the
tone or present an utterance with a sense of detachment, ne is used where yo or no particle

would be expected, given the interlocutors' knowledge and meta-knowledge.

213

Of the utterance-fmal ne, 61 /689 tokens (9.0%) occurred where yo would normally
be expected. The percentage was particularly high in the interview data, constituting
82.3% of all such instance of ne (51/61 tokens were in the interviews). 11/61 tokens were
in formal conversation, and there was no instances of this type of ne in informal
conversation. In contrast, the cases of yo used where ne would be expected were few
(only 5 tokens=2.0%).

(18) is an example, in which ne is used when providing unknown information to

the addressee.

(18) Mieko: a. Nani ga/ donna koto ni kyoomi ga arimasu ka?
what-SUB what kind thing in interest-SUB have Q

IY: b. Nihon desu ka? Keezai desu ne.
Japan BE Q economy BE IP

Mieko: a. 'What/ what kind of things are you interested in?
IY: b. About Japan? Its economy, I'd say.’

IY provides the requested information in (b) with ne. Since the information is known
exclusively to the speaker IY, yo or the bare form would have been expected. This type of
ne was also common in narratives in their evaluation and coda portions. Thus, the choice
of ne and yo is not always clear-cut.

Let us examine how this type of ne is accounted for by previous studies: Maynard
(1993), Kamio (1994) and Takubo and Kinsui (1996). First, according to Maynard's
(1993) characterization of ne and yo, yo should be used in this situation since the
information exclusively belongs to IY. She (19933:106-108) states, however, that ne may
be used when the speaker wishes to present his or her views with a sense of detachment or
distance or to avoid sounding rude. She also accounts for the noncanonical uses of yo and
ne based on whether the speaker's utterance is "information-centered" or "addressee-

centered" (e.g. 1993: 109 and 1997 :88). She states that when an utterance is made with the

214

information as its main focus, yo is likely to be used, and when it is "addressee-centered"
ne is more likely to be chosen (p. 106). However, according to my data, yo is used not
only when an utterance is "information-centered" but also when the speaker wishes to
present a message directed to the addressee as personal and friendly, which is "addressee-
centered" as well.

Next, Kamio (1994:96) refers to this type of ne as an "optional ne," and states that
the "optional ne may appear when (i) a given piece of information does not fall into the
hearer's territory, but (ii) it is close to the speaker, or equally far from both." In (18), the
piece of information is closer to the speaker than to the bearer. Thus, Kamio's explanation
is applicable. However, he simply states when this type of ne may occur, but provides no
explanation of its effect on discourse. Furthermore, Kamio does not address cases where
yo is preferred over ne when both of them are possible.

Finally, Takubo and Kinsui (1996:71) argue that the utterance-final ne indicates that
"the speaker has gone through an internal search to reach the conclusion just stated"
(translation mine). However, they, like Kamio, do not address the alternation of yo and ne
when they are both possible, nor do they discuss the effect of such alternations on
discourse.

These three views base their analyses on knowledge and meta-knowledge among
interlocutors, and their analyses account for some instances of ne being used in place of yo.
However, only Maynard addresses the cases where ya is preferred over ne when both are
possible choices. Furthermore, the cognitive approach cannot account for the following:
(1) the non-utterance-final uses of ne; (2) the choice between ne and yo when both are
possible; and (3) the effect of ne or yo on the discourse. In other words, it is impossible to
explain all the different functions of ne or ya from a cognitive perspective or
indexical/affective perspective alone.

Interactional particles, as well as connectives and fillers, are used in an ever

developing discourse. The speaker must take various factors into account to determine

215

which particle is appropriate in a given situation. In order to do so, not only cognitive but
also interactional factors such as the speaker-bearer relationship, societal norms, and
speech situations (e.g. friendly vs. argumentative) play important roles. As discourse
markers are always used in context, it is imperative to analyze them from multiple

perspectives.

5.4.2.7. Summary

In this section, the use of the final particle yo as a marker of the speaker's emphatic
appeal was examined. Yo is used when the speaker wishes to emphasize his or her
propositions when presenting unknown information to addressees, e.g. expressing the
speaker's views. It also reveals the speaker's judgement about the speaker-addressee
relationship as regards their relative status and familiarity. In addition, yo conveys the
speaker's desire to present a message as friendly and personal, as in warning or advice

directed to the addressee. Figure 5.2 summarizes the functions of yo discussed above.

 

General functions Specific functions of yo
i I. cognitive marker of ------- l (l) to present information unknown to addressee
territory of information I e. g. the speaker's views and experience
11. interactional marker of ------ r (2) disagreement
yo i speaker's emphasis r (3) in expressing continuation or agreement
III. to express the speaker's--- ' (4) advice and warning
judgement on speaker- . (5) request and invitation
L addressee relationship

 

Figure 5.2 Functions of yo

5.4.3. Yone: a marker of emphasis and cooperation
Now we will turn to the last particle in this chapter, yone, the collocated form of yo

and ne. Yone has been analyzed by several researchers from different perspectives. Table

216

5.10 summarizes past research.

Table 5.10 Summary of previous studies on yone

 

 

 

 

 

studies perspective characterization of yone
Makino & Tsutsui functional yone means 'I assert the following and don't
(1986:545) you agree?’
Cook (1988:253-254) indexical/ additive meaning of yo and ne, i.e. assertive
affective attitude+attitude of general agreement/harmony
Nazikian (1994) cognitive/ soften speaker's statement;
interactional confirm info. the speaker is uncertain about
Takubo & Kinsui (1996) cognitive confrrrn uncertain information

 

Table 5.10 shows that Makino and Tsutsui, Cook, and Nazikian interpret yone as
an interactional particle which carries the combined meanings of yo and ne. All of these
researchers agree that the function of yone cannot be analyzed solely based on the
knowledge and meta-knowledge of the interlocutors. If we analyze yone from only the
territory of information, there will be a conflict: (1) yo is used when the speaker has more
information; and (2) ne is used when the hearer has more information or when the
information is shared between the interlocutors.

I agree that functions of yone must be analyzed from both cognitive and
interpersonal perspectives. I further propose that, by using Schiffrin's discourse model,
we can better account for the functions of yone. In what follows, I examine the functions
of yone as a marker of emphasis and cooperation.

Yone was used for three functions in my data: (1) to present information, opinions
or experiences which the speaker assumes are unknown to the addressee; (2) to request or
express agreement on shared information; and (3) to request confirmation. In each
function, yone conveys the combined meaning of yo and ne. In the first function, yo is
used to indicate that the information is in the speaker's tenitory, and that the speaker
wishes to present it emphatically, and ne is used to enlist the addressee's cooperative

listenership. In the second function, yo is used to mark the speaker's emphasis and ne is

217

used to request or express agreement. In the third function, yo is used to express the
speaker's strong conviction about his or her assumptions and ne is used to request

confirmation. We examine each of these situations below.

5.4.3.1. Presenting speakers' opinions and experiences
One of the situations where the speaker solicits the addressee's cooperative attitude
is when emphatically presenting his or her own views or experiences. Yone was used in

such cases. In (19), K1 is describing a happy experience in the U.S.

(l9) KI: a. Boku ga Amerika ni kita toki wa 20-nen gurai mae da kara/
I-SUB the U.S. to come-PAST when-TOP 20 year about ago BE because

b. ano i/ anoo boku wa/ nan da/ sentakuki mo mita koto nakatta n da yone.
well well I-TOP what BE washing machine even see-PAST-NEG BE IP

KI: a. 'I came to the U.S. about 20-years ago, so
b. well, i/ well, what? I hadn't even seen a washing machine, y'know.'

In (b), emphasizing the fact that he had not seen a washing machine before coming to the
U.S., K1 was able to effectively present his narrative, which followed (b) above, on how
happy he felt using a washing machine in the U.S. instead of washing his clothes by hand.

In (20), MI is talking about his experience of driving a beat-up car in Japan.

(20) WA: a. sutantoman ga tsukatta yoona kuruma de mo nee/ amari ki ni shinai tte yuu ka
stunt man-SUB use-PAST like car even IP much mind-NEG QT say Q

MI: b. Demo boku mo soo datta n desu yone.
but I too so BE-PAST assert IP

0. Anoo saisho kuruma katta toki ni/ anoo moo ichiban saisho kara jiko okoshite
well first car but-PAST when well EMPH. very first from accident happen

(1. are kaikaetara doo toka mawari ga nanka yuu n desu yone.
that buy-change how etc. other-SUB something say assert IP

WA: a. 'Even cars which look like as if they were used by stunt men/ they don't
seem to mind.

218

MI: b. But, I was just like that, y'know. .
c. Well, when I bought my first car, I had an accident at the begirrnrng, and
(1. people around us suggested we buy a new car, y'know.'

In both (19) and (20), the speakers are presenting their own experiences which are
unknown to the addressees. Therefore, ne in yone is not used to request or give
confumation or agreement, since there is nothing to be agreed upon or confirmed. The ne
in yone is pronounced with a slightly rising intonation in these cases. All instances of yone
of this type are replaceable with yo alone or with the bare form. Yo in yone in (19) and
(20) operates in the information state, indicating that the information only belongs to the
speaker. It also operates in the participation framework, expressing the speaker's desire to
express emphatic feelings about the proposition, and his/her judgement that the speaker-
hearer relationship is a close one. Ne in these examples functions to request cooperation in
listening or to elicit the hearer's attention so that the speaker can continue with the story,
operating in the participation framework. Such ne also has the effect of softening the
speaker's tone. In sum, yo and ne in yone operate on different planes of talk.

Since there is no conflict as regards yo and ne in the information state in this
analysis (the speaker is the only source of knowledge), and the functions of yo and ne in
the participation framework and the action structure are not mutually exclusive, we can
account for the non-conflict of the co-occurring yo and ne.

This use of yone occurred frequently in narratives. Both in conversations and
narratives, addressees often provided back-channel expressions such as ee 'yes' or an un
'yeah yeah', following yone. This type of yone did not result in a speaker—change, which
is another indication that ne is used simply to soften the tone of the speaker's utterance and
to request cooperative attitude from the addressee in listening to the story, not agreement or
confirmation.

In what follows, yone which resulted in a speaker-change is discussed.

219

5.4.3.2. Requesting or expressing agreement about shared experience
The second situation where yone was used in my data is when requesting or

providing agreement about shared experiences, as shown in (21) .

(21) TH: a. Taki no mizu no tsumetasa ga dono kurai ka tte yuu no O ne/ te de sawareba
water fall-GEN water-GEN coolness-SUB how about Q QT say-NOM-DO hand with touch

KH: b. Te de sawaritakatta desu yonee.
hand with touch—want BE IP

TH: c. Sono gurai chikaku made ikitakatta desu yone.
that about closeness to go-want-PAST BE IP

KH: d. Ikitakatta desu yonee.
go-want-PAST BE IP

TH: a. 'How cool the water of the water fall was, if we could touch it with our
hands,

KH: b. I wanted to touch it with my hands, and you too, right?

TH: c. I wanted to go as close to the fall as possible, don't you agree?

KH: d. I wanted to go, you're right.

Yo nee in (b) and (d) are pronounced with falling intonation with an elongated ne. Yone in
(c) is pronounced with a slightly rising intonation. Yone in (b), (c) and (d) are all
replaceable by either yo (if only emphatically presenting the speaker's desire to the
addressee) or ne (if merely requesting or expressing agreement). Both KH and TH express
their strong desire and request or express agreement in this example with the use of yone.
In (b), KH provides support to TH's remark with yonee. In (c), TH gives support to what
KH has said in (b) and in turn requests agreement with yone. In ((1), KH expresses
agreement to TH's remark in (c) with yonee.

Thus, yo in (21) is used to express the speaker's emphasis (how strongly he or she
feels about the proposition), operating in the participation fiarnework. It also indicates the
close speaker-addressee relationship. Ne is used to request or provide agreement,

operating in the action structure.

220

5.4.3.3. Requesting confirmation
The last function of yone was to request confnrnation on matters about which the
speaker assumes that the bearer has more information. In (22), KM and I are talking about

an automobile factory.

(22) Mieko: a. Ranshingu ni mo Ooruzumoobiiu no koojoo ga aru n desu yone.
Lansing in also Oldsmobile-GEN factory-SUB exist assert IP

KM: b. Aa.
oh
Mieko: c. Ikirnashita?
go-PAST
KM: d. Ittenai desu kedo/ basu no ano l-ban no tokoro desu yone.

go-NEG BE but bus-GEN well numberl place BE IP

Mieko: e. Un soo desu yone.
yeah so BE IP

Mick : a. 'There is an Oldsmobile factory in Lansing, too, y'know.
KM: 11). Oh, I see.

Mieko: c. Have you been there?

KM: d. I haven't, but it's along Number 1 bus route, right?
Mieko: e. Yeah, that's right.

In this example, I, as a long-time resident of the area, know where the factory is. KM has
been in this area only for three weeks, and thus is not familiar with its location. In ((1), KM
uses yone with a slight rising intonation to confirm that his assumption is correct. In this
case, yone is replaceable with ne to request confirmation. Assuming that yo expresses the
speaker's conviction about the proposition, yone means something like the following: 'I
believe my assumption about this matter is correct (yo), and I request that you confirm my
assumption (ne).' This is the interpretation commonly given to yone (e.g. Makino and
Tsutsui 1986 and Takubo and Kinsui 1996). Yo operates in the participation framework,
displaying speakers' conviction about the proposition and the speaker's judgement

regarding the speaker-addressee relationship; ne operates in the information state

221

expressing the speaker's belief that the addressee has more knowledge than the speaker,

and also in action structure, requesting confirmation.

5.4.3.4. Summary of yone

I have examined the functions of yone as a marker which has the combined
meanings of yo and ne in three situations: (1) when the speaker had exclusive information;
(2) when information is considered to be shared between interlocutors; and (3) when the
addressee is believed to have more information. In the first function, yo indicates the
territory of information and/or the speaker's certainty and strong feelings about the
proposition, and to express friendliness to the addressee. Ne is used to request the hearer's
cooperation in listening to the speaker. In the second function, yo expresses the speaker's
strong feelings and certainty about the proposition, and ne is used to request or express
agreement. In the third function, yo expresses the speaker's conviction about the
assumption s/he is putting forward, and ne is used to request confirmation of the validity of
the assumption.

In the combination yone, yo marks the following: (1) information is in the
speaker's territory; (2) the speaker wishes to present his or her utterance emphatically;
and/or (3) the speaker and the addressee have a close relationship. Ne indicates: (1) the
speaker's assumption that the bearer has more knowledge about the proposition; and (2) the
speaker's wishes to elicit the hearer's agreement, confirmation, or cooperative attitude
when listening. Pragmatically, yone is used when the speaker wishes to emphasize a
proposition in a friendly way and, request confinnation and/or agreement, and check the
addressee's cooperation at the same time.

By applying the multi-faceted discourse model to an analysis of Japanese
interactional particles, we can solve the seemingly conflicting juxtaposition of yo and ne.
Y0 and ne in yone do not represent two conflicting cognitive situations since they work on

different planes of discourse, e.g. when yo is operating in the information state, ne

222

operates in the action structure, when yo operates in the participation framework, ne
operates in the information state, and/or action structure, and so forth. Since both yo and
ne operate on several planes of talk simultaneously, yone as a combination of two
discourse makers can be understood as working at least on two planes. Table 5.11

summarizes the functions of yone discussed above.

Table 5.11 Functions of yo and ne in yone

 

 

 

y a ne

(information state) (information state)

to indicate the information is in the to indicate that the information is in the hearer's
speaker's territory territory

(participation framework) (participation framework)

1. to convey emphatic feeling 1. to express the speaker's judgement of the

2. to express the speaker's judgement speaker-addressee relationship as familiar
of the speaker-addressee relationship
(close relationship; the speaker is
srmerior or equal to the addressee)

 

 

(action structure) (action structure)
1. to assert 1. to elicit the addressee's attention and
cooperative listening attitucb

2. to request confirmation or agreement
3. to point out the friendliness between the

speaker and the addressee
4. to check the addressee's attention

 

 

As a textual coordinate, yone indexes an utterance to a prior portion of the
utterance, marking the speaker's conviction and information territory in the utterance to
which they are attached. As a participant coordinate, yone indexes an utterance both to the
speaker and the hearer, marking information management by the speaker and pointing out

the familiar friendly speaker-addressee relationship.

5.5. Summary

In this chapter, I have presented an analysis of Japanese interactional particles, ne,

223

 

yo, and yone as contextual coordinates which index the speaker-proposition and speaker-
hearer relationship, using Schiffrin's discourse mode. Interactional particles as discourse
markers are an indispensable part of face-to—face interactions in Japanese. They display
information about the speaker's feelings or orientation towards a proposition, speaker-
hearer relations, and knowledge and meta-knowledge. I have proposed that: (1) ne is a
marker of cooperative attitude, reveals that the information is in the hearer's territory,
requests or expresses agreement or confirmation, and checks the addressee's attention; (2)
yo is a marker of the speaker's conviction about a proposition, identifies the information
territory, and expresses the speaker's judgement about the speaker-hearer relationship (e. g.
the speaker is superior to the addressee or they are equals); and (3) yone marks the
speaker's conviction about a proposition, reveals his/her assumption about knowledge and
meta-knowledge, and elicits (or expresses) a cooperative attitude, confirmation, or
agreement.

I have argued that a multi-faceted discourse model better accounts for Japanese
interactional particle use, including the occurrence of ne in various positions of an
utterance, and better explains the seemingly contradictory juxtaposition of yo and ne as

yone.

224

CHAPTER 6

Japanese discourse marker usage by non-native speakers

6.1. Preliminaries

The importance English discourse markers play in comprehension has been argued
by several researchers. Tyler (1992) and Tyler and Bro (1992) claim that when
connectives are misused by non-native speakers, or not used where normally expected
comprehension difficulty occurs among native listeners. Flowedew and Tauroze (1995)
argue that both connectives and fillers used by native speakers during a lecture significantly
increase comprehensibility for non-native speakers either by giving the listeners more time
to process information with filled pauses or by their semantic or pragmatic framing
functions.

The purpose of this chapter is two-fold. The first is to compare the differences in
the use of discourse markers between the native and non-native subjects. The second is to
examine the differences in discourse marker use among non-native participants at four
proficiency levels defined by the ACTFL (see Appendix 1). While the ACT'FL Oral
Proficiency Guidelines for the advanced level, for example, refer specifically to the use of
communication strategies, such as pause fillers and stalling devices, research on the use of
Japanese discourse markers by non—native speakers is scarce to date. This chapter aims to

test the following hypotheses:

Hypothesis 4

Non-native speakers at higher proficiency level will use more connectives, fillers and
interactional particles both in number and variety than those at lower proficiency levels.
Hypothesis 5

When a marker has both textual and interactional functions, the textual functions of
discourse markers are learned before interpersonal functions.

225

In order to investigate the acquisition of Japanese discourse markers by non-native
speakers, a longitudinal study, in which a certain linguistic performance of the same
subjects is evaluated at several points in time, would be preferable. This thesis, however,
employes a cross-sectional method, in which a linguistic performance of a number of
subjects is evaluated at a certain point in time. A longitudinal study was not a realistic
choice for the present study, since it takes years of research, following the language
acquisition process of the same individuals. Furthermore, realistically speaking, few
learners reach the superior level in Japanese. Liskin—Gasparro (cited in Hadley, 1993:28)
reports that 2400-2760 hours of instruction is necessary for adult English speakers with
high language learning aptitude to reach the superior level in Japanese in contrast with 480
hours for Romance languages and 480-720 hours for other Indo-European languages.
Japanese is thus rated as one of the most time-consuming languages to leam. I adopted a
cross-sectional research method for the present study for the reasons given above.

In what follows I discuss the research method employed in the present study in
Section 6.2, findings on non-native speakers' use of connectives in Section 6.3, their filler
usage in Section 6.4, and their usage of interactional particles in Section 6.5. In Section

6.6, I discuss pedagogical implications of the present study.

6.2. Method
6.2.1. Data

The data from non-native subjects (NNSs hereafter) come from interviews in
formal-style Japanese. The topics of the interviews for speakers at lower proficiency levels
were limited to their immediate surroundings such as family, school, and daily routines.
The questions for speakers at higher proficiency levels were similar to those asked of the
native speakers of Japanese, including U.S.-Japan relations and economic, social and
educational issues. Care was taken to ask questions on similar topics and at a similar

degree of difficulty within each proficiency level. In order to maintain consistency between

226

the natives' and non-natives' data, only the interview data from the native Japanese

participants are included in the comparative analysis in this chapter.

6.2.2. Participants

Thirty-two non-native speakers participated in this study. Of the thirty-two
recordings, data marred by poor recording quality or those of individuals whose
proficiency level was unclear were excluded from my analysis. Three male and three
female participants were included in each of the four proficiency levels on the ACTFL
Proficiency Scale: (1) novice; (2) intermediate ; (3) advanced; and (4) superior levels. All
of the novice level subjects were at novice-high level, and the intermediate level subjects
were at intermediate-mid to intermediate-high level. The advanced level participants range
from advanced to advanced-high level. The number of native Japanese participants used in
this chapter is 12.

Since the objective of this study was to compare the use of discourse markers at the
four proficiency levels and with that of native Japanese speakers, regardless of the non-
native speakers' personal backgrounds, the participants' native language, age, the length of
Japanese studies or the length of stay in Japan were not controlled, though the t-test and
one-way AN OVA test were performed to investigate the relationship between the use of
discourse markers and these factors. At the end of their interviews, the participants were
asked to fill out a questionnaire regarding their backgrounds and their use of some fillers
and interactional particles (See Appendix 3 for the questionnaire format). Table 6.1 is a list

of the non-native participants.

227

 

 

 

 

 

 

Table 6.1 List of non-native participants
subject level sex age occupation native lang. time in length of #of words
Japan formal study in 10 min.
1 BH NH M O businessman English none 6 mos. 238
2 KK NH M 0 teacher English none 1 year 245
3 AG NH M Y student English none 6 mos. 342
4 TC NH F 0 teacher English 7 mos. 6 mos. 231
5 JL NH F 0 teacher English none 9 mos. 324
6 AP NH F 0 student English 3 years 1 year 342
7 CA 1M M Y student English 9 mos. 15 mos. 572
8 PC IH M O editor English 3 years 1 year 551
9 DD 1M M Y student English 9 mos. 1 year 306
10 NH IM F 0 teacher English 6 mos. 1 year 386
11 KO IH F Y teacher English 2 years 2 years 522
12 WJ IM F Y student English 9 mos. 1 year 534
13 DM A M 0 teacher English 3 years 3 years 791
14 TD A M Y student English 3 years 3 years 764
15 FM A+ M 0 teacher English 3 mos. 3 years 759
16 MB A F 0 teacher English 3 years 4 years 571
17 EV A F Y teacher English 3 years 4 years 674
18 SH A+ F 0 teacher English 3 years 3 years 769
19 JY S M 0 teacher English 3 years 6 years 1194
20 TK S M O teacher Korean 1 years 6 years 1202
21 SM S M O teacher Korean 4 years 6 years 1087
22 BS S F Y student English 3 years 4 years 1107
23 KA S F 0 teacher Chinese 2 mos. 8 years 1214
24 YK S F 0 teacher Korean 1 year 5 years 1132

 

 

NH=novice-high Level; M=intermediate-mid Level, IH=intermediate-high Level; A=advanced level;
A+=advanced-high Level; S=superior level; O=older (37-61 years old); Y=younger (19-23 years old)

6.2.3.

Procedures

I adopted the ACTFL's Oral Proficiency Interview structure as my interview

format, namely, beginning with a wann-up, level checks, probes, and ending with a wind-

down. All the interviews consisted of questions and answers and one or two role-plays,

though the role-play portions are not included in my analysis. As with the native Japanese

interview data, the first two minutes of the recordings was categorically excluded from

analysis and the following ten-minute segment of each interview was fully transcribed and

analyzed. A total of 240 minutes of recordings, 60 minutes for each proficiency level,

were analyzed.

228

After transcribing the data, the frequencies of discourse markers per 1,000 words
("the rates") and their percentages were calculated. Based on the rates of markers per
1,000 words, the t-test or the one-way ANOVA test was run on the following variables at a
95% confidence level: (1) age; (2) gender, (3) proficiency level; (4) the length of formal

Japanese studies; and (5) the time NNSs had spent in Japan prior to the interviews.

6.3. Connectives
6.3.1. Tokens, rates per 1,000 words, and percentages

In this section, quantitative findings on connective marker use by non-native
subjects are presented. According to my data, Hypothesis 4 was supported. Usage rates
per 1,000 words were higher and the variety of connectives was more diversified among
the superior level subjects and advanced level subjects than those at the lower levels. For
both native and non-native subjects, preferences for connectives was in exactly the same
order: de 'and' was the most commonly used, followed by sorede 'and', demo 'but' and
dakara 'so'.

My data show that novice and intermediate level subjects expressed coordination or
contrast, using a limited variety of connectives, most of which are introduced relatively
early in Japanese instruction, e.g. soshite 'and', sorekara 'and/then', and demo 'but'. The
novice level subjects used only one kind of connective, demo 'but', which was used by
AG and TC one time each. The reason demo was the only connective used by the novice
level subjects is perhaps: (1) all the novice level participants used the same textbook, in
which demo was introduced as the fnst connective; and (2) the function of demo as a
connector of contrasting idea units is relatively straightforward and thus easy to learn at an
early stage.

The intermediate level subjects used 5 connectives: ato 'and, in addition', sorekara
'and, then' and soshite 'and', demo 'but,‘ and dakara 'so'. (Sore)de 'and', the most

common coordinative connective among the native and the superior level subjects, was not

229

used by the intermediate level subjects. Nor were consequential (e.g. jaa 'then, if so' and
dattara 'then; if so'), disjunctive (e.g. aruiwa 'or' and soretomo 'or') or explanatory
connectives (e.g. to iu no wa 'that is to say' and nazenara 'because') used by the subjects at
this level. Among the novice and intermediate level subjects, adversative connectives
constituted high percentages (100% and 48.6%, respectively).

The advanced and superior level subjects connected utterances using a variety of
connectives to express various types of coordination and contrast. The advanced level
subjects used 7 connectives, including 3 varieties of coordinative connectives. Among the
coordinative connectives, sorede 'and' was most commonly used, followed by sorekara
'and' and de 'and'. The reasons for the high occurrence of sorede compared to de at this
level seem to be: (1) when sorede is introduced in the classroom as a causal or coordinative
conjunction, it is always introduced in the form of sorede, not de; (2) de is a colloquial
variant; thus students of Japanese are not likely to be exposed to it in the classroom—the
usual way for them to become familiar with de is through casual interactions with native
Japanese; (3) de is not very salient in everyday Japanese speech (i.e. it is usually
pronounced quickly and does not carry stress), and even the advanced level speakers'
listening skill may not be high enough to recognize it in spoken Japanese. Thus, learners at
the advanced level can use sorede in the form of sorede, but not yet as de. As regards
other connectives, they used one kind of adversative connective, demo 'but', and there was
one token of the disjunctive connective aruiwa 'or'. Consequential or explanatory
connectives were not used by the subjects at this level.

The superior level subjects used 10 connectives, and there was much more
diversification of coordinative and adversative connectives. Their percentages of
connectives usage are similar to those of the native participants. The percentages of
coordinative connectives were high among the advanced and superior level subjects (53.6%
and 56.7%, respectively), while the percentages of adversative connectives were high

among the novice and intermediate level subjects.

230

 

Table 6.2 shows the number of tokens and percentages of connectives used by the

non-native subjects at four proficiency levels and the native participants. The first set of

numbers are the number of tokens; those in parentheses represent the percentages

represented by each connective .

Table 6.2 The tokens and percentages of connectives by proficiency levels (n=36)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

marker native superior advanced intermediate novice
# of subjects n=1 2 n=6 n=6 n: n=6
total time analyzed 120 min. 60 min. 60 min. 60 mm 60 min.
# of total words 11,686 6,936 4,328 2,871 1,665
token (%) token (%) token (%) token (%) token (%)
de 'and' 84 (23.9) 51 (29.8) 5 ( 7.3) ----------
sorede 'and' 45 (12.8) 27 (14) 17 (10.1) ----------
ato 'and 46 (13.1) ---------- 6 (16.2) .....
sorekara 'and' 10 (1.4) 14 ( 0.6) 15 (17.4) 3 ( 8.1) -----
soshite 'and' 1 (0.3) 5 ( 7.6) 0 (2.9) 2 ( 5.4) -----
total (coord) 321 (46.6) 97 (56.7) 37 (53.6) 11 (29.7) -----
demo 'but' 81 (21.9) 22 (12.9) 23 (33.3) 18 (48.6) 2 (100)
datte 'but' 1 (2.27) ....................
tada 'but' 8 (1.14) 1 (0.6) ---------- -----
tokoroga 'but' ------ 3 ( 1.8) ...............
total (adv) 90 (25.6) 26 (15.2) 23 (33.3) 18 (48.6) 2 (100)
dakara 'so' 67 (19) 40 (23.4) 8 (11.6) 8 (21.6) -----
total (consq) 9 (2.6) 2 (1.2) ...............
total (disj) 4( 1.14) ----- 1 ( 1,5) ..........
total (expl) 1 (0.28) ....................
total 352 (100) 163 (100) 69 (100) 37 (100) 2 (100)

 

 

 

 

 

 

consequential connectives: e.g. ja(a) 'then'; dattara 'then'; sorenara 'then'; disjunctive connectives: e.g. aruiwa

'or'; soretomo 'or'; explanatory connectives: nazenara 'because'

We now turn to the rates of connective marker use per 1,000 words. The total rate

of connectives among the NNSs was highest for the superior level subjects (24/1000
words), followed by the advanced (15.9/1000 words), the intermediate (12.9/1000

words), and the novice level subjects (1 .2/1000 words). As shown in Table 6.3, the rate

for the native subjects was 25.8/1000 words, which is similar to the superior level

speakers' rate. The largest rate difference was between the novice level and the

231

 

intermediate level subjects (1.2 and 12.9/1000 words, respectively). In contrast, there was

a small difference between the intermediate and the advanced level subjects.

Table 6.3 The rates of connectives per 1,000 words by proficiency level (n=36)

 

 

 

 

 

 

 

 

 

 

 

marker meaning native superior advanced interrned. novice
de 'and' 6.2 7 .4 1.2 ----------
sorede 'and' 3.3 3 .9 3 .9 ----------
ato 'and' 3.4 ---------- 2.1 .....
sorekara 'and' 0.4 2 3 . 5 1 .0 -----
soshite 'and' 0.07 0.7 ----- 0.7 -----
total (coord) 13.3 1 4 8.55 3.8 -----
demo 'but' 5.9 3.2 5.3 6 .3 1.2
date 'but' 0.07 ....................
tada 'but' 0.6 0 . l ...............
tokoroga 'but' ----- 0 . 4 ...............
total (advers.) 6.6 3.7 5.3 6.3 1 2
dakara 'so' 4.9 5.8 1.9 2.8 -----
total (consq) 0.7 0 . 3 ...............
total (disj) 0.3 ----- 0.2 ..........
total (expl) 0.07 ....................
total 25.8 24 15.9 12.9 1 2

 

 

 

 

 

 

 

6.3.2. Sociolinguistic factors and the use of connectives by NNSs

In order to see what kinds of factors influence the use of connectives, the
unmatched two-tailed t-test was run for NNSs' age and gender, and the one-way AN OVA
test was performed on the following factors: (1) proficiency level (novice, intermediate,
advanced and superior), (2) the length of the participants' formal Japanese studies (less
than 1 year; 1.0-1.9 years, 2.0-2.9 years, and over 3 years); and (3) the time NNSs had
spent in Japan prior to the interviews (0, less than one year, 1.0-1.9 years, 2.0-2.9 years,
and over 3 years). A P-value of 0.05 or less indicates a significant relationship, and
0.053P30.1 indicates a strong tendency towards significance at a 95% confidence level.
6.3.2.1. Age and gender

Speakers' age was not a significant factor in the use of connectives among the

232

NNSs, as was also the case with native subjects. The only connective which showed a
marginal relationship to the speakers' age was the adversative connective demo 'but'
(P=0.090). Each connective perhaps has relatively well-established functions, so there is
little variation based on the age of the NNSs.

As regards gender, we first examine the differences in the native speakers'
interview data to use as a basis for comparing native (n=12) and non-native subjects
(n=24). The native subjects' use of connectives in interviews showed a statistically
significant relationship to gender (the total: 32.2/1000 words for women and 18/1000
words for men; P=0.006). The use of individual connectives was also influenced by
gender. For example, women's preference for the following connectives was statistically
significant or showed a strong tendency: de 'and' (the rates: 8.67 vs. 3.09, respectively;
P=0.071), demo 'but' (the rates: 8.54 vs. 4.23, respectively;P=0.006), and dakara 'so'
(the rates: 6.8 vs. 2.6 and P: 0.032).

For the NNSs, gender was not a significant factor when proficiency levels were not
taken into consideration. Only the use of de 'and' showed marginal significance (women's
rate was higher than men's; P=0.071). However, among the superior level NNSs, the
gender difference was clearly exhibited and parallelled the native subjects' tendencies. For
example, the rates of de 'and' and demo 'but' were much higher among females than males
(7.24 vs. 0.57/1000 words for de, and 23.7 vs. 4.3/1000 words for demo). Thus,
although gender was not a significant factor in the use of connectives at novice,
intermediate, and advanced levels, it was among superior level and native subjects. The
fact that the use of connectives among the superior level subjects parallelled that of the
native participants in terms of the number, variety, and their preference for particular
discourse markers according to their gender indicates that superior level speakers truly do
have "native-like" discourse strategy in the use of connectives, as defined in the ACTFL

proficiency guidelines.

233

6.3.2.2. Other factors

Of the other three factors, proficiency level and the length of formal Japanese
studies, and the time spent in Japan, the fnst two showed much more significant
relationships to the use of connectives than the time participants spent in Japan, as shown

in Table 6.4.

Table 6.4 The P-values of the three factors influencing the use of connectives (n=24)

 

 

 

 

 

 

 

 

 

 

 

marker meaning The P-values for the 3 factors
Prof. level formal Japanese study time in Japan

de 'and' 0.0396 0.0376 ns

sorede 'and' 0.0097 0.0725 ns

ato 'and' ns ns ns

sorekara 'and' ns ns ns

soshite 'and' 0.0295 ns ns

total (coord) 0.0363 0.0467 ns

demo 'but' 0.0295 0.0363 ns

datte 'but' N/A N/A N/A

tada 'but' ns ns N/A

tokoroga 'but' ns ns ns

subtotal 0.0342 0.0358 ns

dakara s 0' 0.0534 0.0518 0.0213

total (consq) ns ns ns

total (disj) ns 0.0040 ns

total (expl) ID ID ID

total 0.0018 0.0014 0.0513

 

 

 

 

 

 

P3005: significant correlation; 005<PS01: strong tendency; ns: not significant; ID: indetenninable due to
low occurrence, N/A: no P-value (=no occurrence)

The total rate of connectives use was strongly related to the proficiency level (P:
0.0018). As regards individual connectives, de 'and' (P=0.0396), sorede 'and'
(P=0.0097) soshite 'and' (P=0.0295), demo 'but' (P=0.0295), and dakara 'so'
(P=0.0534) showed significant relationships. This result supports Hypothesis 4: NNSs
proficiency level is a significant factor which influences the use of connectives.

The length of formal Japanese studies was also strongly related to the use of

connectives: the longer formal study, the more frequent the use of connectives. The total

234

rate of connectives, those of (sore )de 'and' (P=0.0376 for de and 0.0725 for sorede),
demo 'but' (P=0.0363), dakara 'so' (P=0.0518), and coordinative, adversative,
disjunctive category totals are all significantly related to the length of formal study
(P=0.0467, 0.0358, 0.0040, respectively).

The length of time the NNSs spent in Japan was not as significant as the other two
factors. Only the rate of dakara 'so' and the total rate were related to connectives use (P:
0.0213 and 0.0513, respectively). From these results, it is observed that the length of
formal study is important in attaining high proficiency levels, which in turn influences the

use of discourse markers.

6.3.3. Discussion

In this section, I first discuss my findings regarding the use of connectives by
NNSs and, secondly, how the two most common connectives, (sore )de 'and' and demo
'but' were used by the NNSs.

My major findings are as follows: (1) the NNSs at novice, intermediate, and
advanced levels used a limited number of connectives, and the superior level subjects' use
of connectives was similar to that of the native subjects; (2) the NNSs at the novice and
intermediate levels used connectives mainly for connecting idea units, while those at the
advanced and the superior levels use connectives for ideational and interpersonal functions;
(3) usage errors occurred only among the novice level subjects (some novice level subjects
tried to use connectives, but often erred in form and placement);1 and (4) the ideational
functions of a connective are learned before its interpersonal functions; i.e. the order of
learning is from ideational to interpersonal. As the NNSs used connectives for

interpersonal functions, their ideational functions decreased in percentage.

 

1 The following connectives were used inconectly and thus were not included in the count: BH: 1 error in
sorekara 'and, then'; KK: 5 errors in demo 'but'; TC: 1 error in jaa 'then'; AG: 3 errors in to 'and'; and 2
errors in dakara 'so' and IL: 1 error in sorekara 'and/then'.

235

6.3.3.1. The use of (sore)de by non-native subjects

(Sore )de 'and' was the most commonly used connective by the non-native subjects,
though it was used only by the advanced and superior level subjects. In this section, I
discuss the use of de and sorede 'and' by the non-native subjects. First, their distributional
characteristics are discussed in Section 6.3.3.1.1, and their use by non-native subjects is

discussed in Section 6.3.3.1 .2.

6.3.3.1.]. Distribution of do and sorede 'and' in non-native subjects

I begin my discussion with distributional characteristics of de and sorede among the
NNSs and compare them with those among the native subjects. The most striking fact
about de and sorede among the NNSs was that they were not used by novice or
intermediate level subjects, though the intermediate level subjects used three other
coordinative connectives, ato 'and, in addition', sorekara 'and, then', and soshite 'and'.

At the advanced and the superior levels, both de and sorede were used. The
following differences in usage between the NNSs at the advanced and superior levels were
observed: (1) the advanced level subjects used sorede more Often than de (sorede: 17/22
tokens=77.3%; de: 5/22 tokens=22.7%), whereas the superior level subjects used de more
often than sorede (sorede: 27/78 tokens=34.6% ; and de: 51fl8 tokens=65.4%). The
superior level subjects' percentages of de and sorede were similar to those of native
subjects. Table 6.5 summarizes the tokens and percentages of de and sorede as used by

advanced and superior level subjects and native subjects.

Table 6.5 Distribution of de and sorede by proficiency level

 

 

 

 

 

native superior advanced
sorede 'and' 45 (34.9%) 17 (34.6%) 17 (77.3%)
de 'and' 84 (65.1%) 51 (65.4%) 5 (22.7%)
total 129 (100%) 78 (100%) 22 (100%)

 

236

These distributional differences may in part be related to limited instruction about de
and sorede, and partially due to the complexity of their functions:

- (l) Sorede is introduced relatively late in Japanese instruction;2

(2) The coordinative de is not introduced in most introductory textbooks, which
suggests that many NNSs do not receive explicit instruction on the use of de as a
coordinative or causal marker;

(3) Generally speaking, de is pronounced quickly by native speakers of Japanese,
and is often inaudible even to native Japanese (for example, at times, I had to listen to the
recordings of my data several times to hear it), and, therefore, even NNSs who have lived
in Japan may have difficulty hearing it; and

(4) Both de and sorede have many functions (e. g. sequential idea coordination,
causal coordination, and speaker continuation), and NNSs may have difficulty figuring out
their appropriate usages.

All the above suggests that, if language teachers hope to teach natural Japanese in

classroom, explicit instruction on the use of de and sorede is important.

6.3.3.1.2. Functions of ( sore )de among the NNSs

We now turn to the differences in the usage of de and sorede between the advanced
and superior level NNSs on the one hand, and the native subjects on the other. In terms of
the functions of (sore )de as a marker of coordination and speaker-continuation discussed in
Section 3.5.1, the following were observed in the native subjects' interview data: (1)
( sore )de 'and' as a marker of (sequential) idea units coordination at both local- and global-

levels; (2) as a causal marker; (3) as a connector of contrasting units within a speaker's

 

2 In the Introduction to Modern Japanese by Mizutarri and Mizutarri, for example, sorede is introduced in
Chapter 22 out of 30 chapters, which is usually covered in the second half of the first year or in the first
half of the second year at college level. In Learn Japanese by Young and Nakajirna, sorede is introduced in
Chapter 4 out of 15 chapters as a causal marker, but not as a coordinative connective.

237

turn; and (4) as a connector of functionally different units, such as the main and tangential
portions of an utterance. Interactionally, (sore )de was used as a floor-holding device for
speaker continuation.

Among the NNSs the dominant functions of de and sorede differed by proficiency
level. The advanced level subjects used sorede mostly as a causal marker (58.8%) and as a
sequential connector of local-level idea units (35.3%). These two functions constitute
94.1% of all the uses of sorede for the advanced level subjects. The superior level subjects
used sorede in more diversified functions, such as sequential local- and global-level unit
connector (40.7% and 11.1%, respectively), as a causal marker (22.2%), and floor-
holding device (13.3%). For superior level subjects, as sorede assumes various functions,
its importance as a causal marker decreases. Table 6.6 summarizes the usages of sorede by

NNSs and native Japanese.

Table 6.6 Distributions of functions of sorede by proficiency level

 

 

 

 

 

functions of sorede native superior advanced
causal marker 17 (37.8%) 6 (22.2%) 10 (58.8%)
local-level sequential coordination 12 (26.7%) 11 (40.7%) 6 (35.3%)
floor-holding device 6 (13.3%) 4 (14.8%) 1 (5.9%)
global—level sequential coordination 4 (8.9%) 3 (11.1%)

connect functionally different unit 4 (8.9%) 3 (11.1%)

total 45 (100%) 27 (100%) 17 (100%)

 

De was used for various functions by native and NNSs. The advanced level NNSs
used de for 3 functions: (1) local-level sequential coordination (3/5 tokens=60%); (2)
global-level sequential coordination (1/5 tokens=20%); and (3) floor-holding device (1/5
tokens=20%). The superior level subjects used de for 4 functions: (1) local-level
sequential coordination (29/51 tokens=56.9%); (2) global-level sequential coordination
(10/51 tokens=19.6%); and (3) as a floor-holding device (10/51 tokens=19.6%, each); and
(4) as a causal marker (1/51 tokens=3.9%). These results suggest that for both de and

238

sorede, ideational functions are learned first, followed by their interactional functions.
One point of interest about the use of de is that its percentage as a causal marker
was much lower among NNSs than native speakers. This is probably due to the fact that
de is usually not explicitly taught in the classroom and it is difficult for NNSs to leam this
function. Even when they hear de used by native Japanese, they may not know exactly
how and when it is used. Explicit instruction on the use of de is therefore indispensable.
Table 6.7 summarizes the use of de by the NNSs and the native participants. Notice the
striking similarities in percentages between the superior level subjects and the advanced

level subjects, as well as the differences between them and the native subjects.

Table 6.7 Distributions of functions of de by proficiency level

 

 

 

 

 

functions of de native superior advanced
local-level sequential coordination 27 (32.1%) 29 (56.9%) 3 (60%)
floor-holding device 28 (33.3%) 10 ( 19.6%) 1 (20%)
global-level sequential coordination 12 (15.5%) 10 (19.6%) 1 (20%)
causal marker 9 (10.7%) 2 (3.9%)

connect functionally differentiated units 6 (7.1%)

connect contrasting units 1 (1.2%)

total 84 (100%) 51 (100%) 5 (100%)

 

6.3.3.2. The use of demo 'but' by non-native speakers

Demo 'but' was characterized as a marker of upcoming contrast in Section 3.5.2 of
this thesis. It was the second most commonly used connective both by native and non-
native subjects in my data, and was used conectly by the speakers at all four proficiency
levels. It was the only connective used by the novice level subjects, and was also the
connective most frequently used by the intermediate level subjects. In what follows, I
compare the use of demo by NNSs at the four proficiency levels and also with that of the
native subjects.

During the native participants' interviews, demo 'but' was used in three functions:

(1) to connect contrasting idea units (referential contrast); (2) to connect functionally

239

different units (functional contrast), e.g. main ideas and side-sequencing; and (3) to mark
an upcoming contrasting action, e.g. point-making, disagreement, or challenge.

As was the case with (sore )de, demo was used in its ideational function among the
novice level speakers but not in its interactional functions. Demo was used twice by novice
level speakers,3 and both were used to connect contrasting idea units; e.g. Uchi wa chiisai
desu. Demo ii desu. 'My house is small. But it is nice' (Subject TC). Again, this perhaps
indicates that the functions of a connective as a connector of ideational units are learned
before its interactional functions.

The percentage of demo for referential contrast was higher for the NNSs at lower
proficiency levels (100% for novice level, 66.7% for intermediate level) than at higher
proficiency levels (56.5 for advanced level and 45.5 for superior level). The percentages
of functional contrast and contrasting actions were higher among the NNSs at the advanced
and superior levels than the intermediate or the novice level subjects (see Table 6.8). (l) is

an example of functional contrast with demo used by KA, a superior level speaker.

(l) Mieko: a. Senmon ga Nihongo na n desu ne?
major-SUB Japanese assert Q

KA: b. Ee/ daigaku/ Pekin daigaku de 5-nen-kan senmon to shite narattee/

yes university Beijin university at 5-years-for major as learned-and

0. demo jugyoo wa/ ano Pekin daigaku wa soogoo daigaku desu kara/
but classes-TOP well Beijin university-TOP multiversity BE because

(1. tada Nihongo no senmon to yuu/ itte mo/
only Japanese GEN major QT say say though

e. N ihongo no jugyoo dake ja nakute/

Japanese class only not

f. ano nihongo to kankee nai/ sooyuu karnoku o torimashita.
well Japanese with relation NEG such subject-DP took

Mieko: a. 'Your major is Japanese, right?
KA: b. Yes. University, at Beijin University, I studied Japanese for 5 years as

 

3 Novice-level speakers are characterized by their word-level utterances rather than sentence-level.
Therefore, infrequent use of connectives are quite natural even for novice-high level speakers

240

my major, and
c. but classes, well, Beijin University is a multiversity, so
(1. even though I was a Japanese major,
e. not only Japanese,
f. but well, I also took courses unrelated to Japanese.’

KA used demo in (c) to differentiate her direct answer to my question from additional

information in (c-f).

(2) is an example of demo used by YK, another superior level speaker, connecting

contrasting actions; i.e. challenging.

(2) YK:

Mieko:

Mieko:

a. Sooshi kaimee wa kekkyoku Nihon-jin kara miru to
changing names-TOP after all Japanese from look when

b. taishita mondai ja nai n desu yone.
serious problem BE NEG assert 1P

c. Un demo yappari ne/ namae kaetaku nai
yeah but as expected [P name change-want NEG

d. Iya demo rekishi kara mite miru to
no but history from look look when

e. Nihon no baai wa onna no hito/ kekkon shitara namae kaemasu yone?
Japanese case-TOP female person marry do-when name change IP

a. 'From the Japanese point of view, changing (Koreans') names was

b. not a big thing, y'know?

c. Yeah, but, as you know, we don't want [to change our names

d. [But from historical viewpoint,

e. In the case of Japan, women change their names when they get married,
don't they?‘

In (c), I began to say that no one likes to change his or her name, which was interrupted by

YK's statement in (d), arguing that because it was nothing serious for Japanese to change

their names, they forced Koreans to change their names, which they resent to this day.

Thus, demo in (c) is used to present a counter-argument and claim the floor. The

interactional use of demo by KA and YK is similar to that of native subjects. This is an

indication that superior level speakers possess native-like proficiency not only in syntax

and vocabulary, but in the use of connectives as discourse markers, referentially and

241

interactionally. Table 6.8 is a summary of the various uses of demo by native and non-

native subjects.

Table 6.8 Use of demo by proficiency level

 

functions native superior advanced interrned. novice

 

referential contrast 36 (44.4%) 10 (45.5%) 13 (56.5%) 12 (66.7%) 2 (100%)
functional contrast 36 (44.4%) 8 (36.4%) 5 (21.3%) 3 (16.7%)
contrasting actions 9 (11.2%) 4 (18.2%) 5 (21.3%) 3 (16.7%)

 

 

 

total 81 (100%) 22 (100%) 23 (100%) 18 (100%) 2 (100%)

 

Notice that the use of demo in its referential function decreases with proficiency,
but that the reverse is true of the functional use. In this cross-sectional study of the
functional distribution of (sore )de 'and' and demo 'but' we find the following: when a
connective has more than one function, its most proto-typical function is learned first, i.e.
the referential function is learned before the interactional functions. For example, (sore )de
is introduced as a causal marker first, and thus, NNSs learn its function as a causal marker.
As we have seen in Table 6.6, the advanced level subjects in the present study used sorede
as a causal marker 58.8% of the time. Among the superior level subjects, the percentage of
its function as a marker of sequential coordination was higher than the percentage as a
causal marker (33.3% vs. 22.2%). The percentage of sorede as a floor-holding device, an
interactional function, was 5.9% among the advanced level speakers, while it was 14.8%
among the superior level subjects. As sorede acquired more functions, its percentage as a
causal marker decreased.

As regards demo, its proto-typical function is to express a referential contrast, and
this was the only usage found among the novice level subjects. Its interactional function as
a contrasting action marker occuned at the intermediate level and above, supporting my
Hypothesis 5. It will be interesting to test this hypothesis in a longitudinal study in the

future.

242

6.3.4. Summary

In this section, I have presented a quantitative analysis of connectives use by the
NNSs and discussed the two connectives most often used by the NNSs, (sore )de 'and' and
demo 'but'. My findings are:

1. The NNSs at the advanced and superior levels used considerably more
connectives both in variety and number than those at intermediate and novice levels.

2. The most striking difference in the variety and the rate of connectives was
between novice level and intermediate level: (1) one variety of connective at novice level vs.
5 varieties at intermediate; and (2) the rate of 1.2/1000 words at novice level vs. 12.9/1000
at intermediate level.

3. The proficiency level and the length of formal Japanese study most significantly
affected the discourse marker use by the non-native subjects, followed by the length of the
time they lived in Japan. Age and gender were not significant factors.

4. There were distributional differences in the use of de and sorede between the
advanced and the superior level subjects. Possible reasons are: (1) classroom instruction
mainly introduces sorede as a causal marker; and (2) de as a coordinative connective is
seldom taught.

5. Finally, my findings suggest that a proto-typical function for each connective is
learned first, followed by peripheral functions; e.g. as a causal marker fnst in the case of
sorede 'and', followed by its function as a marker of sequential coordination, the referential
function of demo 'but' before its function as a marker of functional contrast or contrasting

actions.

243

6.4. Fillers
6.4.1. Tokens, rates per 1000 words, and percentages

First, I discuss the variety and percentages of fillers. The NNSs at higher
proficiency level used more varied fillers than those at lower proficiency levels, supporting
Hypothesis 4. Among the fillers, ano(o) 'uh, well' was used most frequently by all the
NNSs.

In Section 4.2, fillers were classified into two categories: (1) language-production-
based fillers, which are used to fill a potential silence when a smooth speech production is
hindered; and (2) socially motivated fillers, which contribute to enhancing agreeable
conversation atmosphere. The language-production-based fillers were further classified
into: (1) production-related fillers, e.g. ee( to ) 'uh' and soo desu ne( e ) 'let me see' and (2)
information management fillers, a 'oh', aa 'Oh (I see)‘, an 'yeah', and e? 'oh, huh?'.

The novice level speakers used only a production-related filler, ee( to ) 'uh'. The
intermediate level speakers used six fillers, two production-type fillers ee( to ) and soo desu
ne( e ) 'let me see', two information-management fillers, a 'oh' and aa 'Oh (I see)’, and
ano(o) 'uh, well' and arm 'well, uh', which can function both as production-related and
socially motivated fillers. Ano(o) constituted 77.1% of all fillers used by the intermediate
level subjects. Ano(o) use was particularly notable among those who spent some time in
Japan (6 to 9 months).

The advanced level speakers used 13 fillers. They also used ano(o) frequently
(71.2% of all the fillers they used). Hora 'therel', are 'that thing', e? 'oh, huh?', and koo
'like this' were not yet used by the advanced level subjects.

The superior level subjects' use of filler was more diversified, and was similar to
that of native speakers in their number and variety. They used 15 fillers. Their percentage
of ano(o) was consequently much lower than those of other proficiency levels. Are 'that
thing' and e? 'oh, huh?’ were not used by them. Table 6.9 summarizes the use of fillers.

The numbers in parentheses represent the percentage of each filler.

244

Table 6.9 The tokens and percentages of fillers by proficiency level

 

 

 

 

 

 

 

 

marker original native superior advanced intermediate novice
meaning tokens (%) tokens (%) tokens (%) tokens (%) tokens (%)
nanka 'like' 189 (18.5) 52 (12.5)4 5 ( 1.5) ----- 5 (100)
ano(o) 'uh. well' 172 (16.9) 145 (34.8) 242 (71.2) 54 (77.1) .....
moo 'really' 106 (10.4) 23 ( 5.5) 8 ( 2.4) ----------
yappari 'as expected' 98( 9.6) 18( 4.3) 11 ( 3.2) ----------
ma(a) 'kind of 60( 5.9) 81 (19.4) 12( 3.5) ..........
mm 'uh.well' 59( 5.9) 10( 2.4) 13( 3.8) 2(2.9) -----
a 'oh' 53( 5.2) 3( 0.7) 1 ( 0.3) 5 (7.1) -----
ee(to} 'uh' 50( 4.9) 18( 4.3) 8( 2.4) 3 (4.3) -----
00 'oh (I see)’ 45( 4.4) 9( 2.2) 3( 0.9) 2 (2.9) .....
nante yuu no 'what is it?’ 39 ( 3.8) 29 ( 5.3) 13 ( 3.8) ----------
koo 'like this' 39( 3.8) 15( 3.6) ...............
an 'yeah' 36( 3.5) 1( 0.2) 1( 0.3) ..........
soo desu ne( e) 'let's see' 28 ( 2.8) 9( 2.2) 15 ( 4.4) 4 (5,7) .....
sono(o) 'uh, well' 21 ( 2.1) 8( 1.9) 8 ( 2.4) ----------
6? 'oh, huh?’ 18( 1.8) ....................
are 'thatthing' 5( 05) ....................
hora 'there!‘ 1 ( 0.1) 3 (0,7) ...............
total 1019(100 417 (100) 340(100) 70(100) 5(100)

 

 

 

As regards the total rate of fillers per 1,000 words, the advanced level subjects' rate was

the highest (69.1/1000 words) due to the high rate of ano(o) (55.9/1000 words). Ano(o)

was used mostly as a pause-filling device by the intermediate level subjects, and was not

used for a socially motivated purposes at this level.

Information management fillers were rarely used by the NNSs. The total rates of a

'oh', aa 'oh (I see)‘, an 'yeah', and e? 'oh, huh?’ ranged between 1.16 to 2.44/1000

words, compared to the native subjects' rate, 11.1/1000 words. Table 6.10 summarizes

the rates of fillers per 1,000 words by proficiency level, with numbers in bold indicating

the highest rates among the NNSs.

 

4 44 of 52 tokens (84.6%) of nanka were produced by as, a young female speaker.

245

 

Table 6.10 The rates of fillers per 1,000 words by proficiency level

 

 

 

 

marker meaning native spperior advanced intermediate novice
nanka 'like' 13.8 7.5 1.2 ----------
ano(o) 'uh, well' 12.6 21.0 55.9 18.8 -----
moo 'really' 7.8 3.3 1.9 ----------
yappari 'as expected' 7.2 2.6 2.5 ----------
ma(a) 'kind of' 4.4 12.0 2.8 ----------
uun 'uh, well' 4.3 1.4 3 0.7 -----
0 'oh' 3.8 0.4 0.23 1.7 -----
ee(to) 'uh' 3.7 2.6 1.9 1.0 3.0
m 'Oh (I see)‘ 3.3 1.3 0.7 0.7 -----
nante yuu no 'what is it?’ 2.9 3.2 3 ----------
koo 'like this' 2.9 2.2 ...............
un 'yeah' 2.6 0.1 0.2 ----------
soo desu ne( e) 'let's see' 2.1 1.3 3.5 1.4 -----
sono(o) 'uh, well' 1.5 1.2 1.9 ----------
e? 'oh, huh?‘ 1.3 --------------------
are 'that thing' 0.4 um um ..........
hora 'there' 0.07 0 . 4 ...............
total total 74.7 60.4 78.6 24.4 3.0

 

 

 

 

 

 

 

 

6.4.2. Sociolinguistic factors and the use of fillers
6.4.2.1. Age and gender

As with connectives, speakers' age or gender did not significantly affect the use of
fillers among the NNSs. The only filler which showed a strong relationship to speakers'
age was nanka 'like' (P=0.040), which was used predominantly by the younger speakers.5

As regards gender, I discuss the use of fillers by the native Japanese subjects first.
The women's rates were higher for the total rate of fillers, and the difference was
significant. For example, women used the following frllers more often than men: ee( to )
'uh' (5.47 vs. 1.46/1000 words, respectively; P=0.052) and a 'oh' (5.6 vs. 1.8/1000
words, respectively; P=0.015). A socially motivated filler, ma(a) 'kind of, showed a
significant difference (6.99 for men vs. 2.27 for women per 1000 words; P=0.001).

Among the NNSs, the only filler which showed a significant relationship to gender

was aa 'oh' (P=0.022), which was used more often by women than by men. As regards

 

5 Without the outlier BS who used nanka 44 times out of 52, the P-value is 0.487, which means age was
not a factor influencing the use of nanka.

246

other fillers, there was no clear difference between female and male speakers at any of the
four proficiency levels.
6.4.2.2. Other factors

As with connectives, the result of the one-way ANOVA test shows that the
subjects' proficiency levels and the length of formal study were the most significant factors
promoting the total use of fillers (P=0.0000 for both). Table 6.11 summarizes the P-values

in relation to NNSs' proficiency level, length of formal study, and time spent in Japan.

Table 6.11 The P—values of the three factors influencing the use of fillers

 

 

 

 

 

marker original The P-values for the 3 factors
meaning Prof. level formal Japanese study time in Japan

ano(o) 'uh; well' 0.0003 0.0020 0.0986

nanka 'like' ns ns ns

moo 'really' 0.0542 0.0245 ns

yappari 'as expected' ns ns ns

ma(a) 'kind of' 0.0000 0.0035 ns

koo 'like this' 0.0507 ns 0.0040

mm 'uh; well' 0.0998 0.0002 ns

sono(o) 'uh; well' ns 0.0781 ns

ee(to) 'uh' ns ns ns

nante yuu no 'what is it?‘ ns 0.0413 ns

soo desu nee 'let me see' ns ns ns

hora 'there' ns ns ns

are 'that thing' N/A N/A N/A

a.’ 'oh' 0.0722 0.0639 ns

a2 'oh (I see)‘ ns ns us

an 'oh yeah' ns ns ns

e? 'oh?; huh?‘ N/A N/A N/A

total 0.0000 0.0000 0.0409

 

 

 

 

P3005: significant correlation; 0.05<P301: strong tendency; ns: not significant; N/A: no P-value (=no

occurrence)

The rates of two fillers, ano(o) 'uh, well' and ma(a) 'kind of showed a strong relationship

to proficiency levels (P=0.0003 and 0.0000, respectively). and the rates of four fillers,

moo 'really', koo 'like this', uun 'uh, well', and a 'oh' were also related to proficiency

247

 

levels (0.05<P30.1).

As regards the length of formal Japanese studies, the rates of five fillers ano(o) 'uh,
well', moo 'really', ma(a) 'kind of, mm 'uh, well', and nante yuu no 'what shall I say?‘
all showed a significant relation (0.05<P). The rates of sono(o) 'uh, well' and a 'oh' also
showed some relationship to the length of formal Japanese studies (0.05<P30.1).

The amount of time the NNSs spent in Japan also showed some relationship to the
use of fillers, but not as strong as the other two factors (the P-value of the total rate was
0.0409). The rate of koo 'like this' showed a significant relationship and that of ano(o)
showed some relationship to the time the NNSs spent in Japan (P=0.004 and 0.0986,

respectively).

6.4.3. Discussion

The following are the major findings as regards the use of fillers by the NNSs:

1. Like connectives, the advanced and the superior level subjects used more fillers
in number and variety than the intermediate and novice level subjects (see Table 6.10).

2. The NNSs at novice and intermediate levels used fillers mostly for production-
related purposes while searching for an appropriate expression, whereas those at advanced
and superior levels used various fillers for both production-related and social purposes.
For example, at the novice level, a pause-filling filler ee(to) 'uh' was the only one used. At
the intermediate level, three more variety of fillers, soo desu ne( e ) 'let me see', uun 'uh,
well' and ano(o) 'uh, well', occurred as pause-filling devices, as well as a 'oh' and aa
'oh', information management fillers. Socially motivated fillers (e.g. ma( a ) 'kind of) or
interaction-related ones (e.g. yappari 'as expected', moo 'really' and hora 'there, as you
know') were not used at these levels.

3. The rate of ano(o) per 1,000 words among the advanced level subjects was the
highest (4.4 times more than the native subjects and 2.6 times more than the superior level

subjects).

248

4. Serious errors in usage were not made by any of the subjects.

Findings 1-3 above suggest that: (l) language-production fillers such as ee( to) 'uh' and soo
desu ne( e ) 'let me see' are learned before socially or interactionally-oriented ones such as
ma( a ) 'kind of for down-playing and yappari 'as expected'; and (2) if the same filler can
be used either for language-production purposes, e.g. when searching for an appropriate
expression, and social/interactional reasons, e.g. to enhance agreeable atmosphere and/or
the addressee's involvement in conversation, its use as a language-production function is
learned before social/interactional functions.

Let us turn to findings (2) and (3). First, as regards the use of language production
fillers, as shown in Table 6.9, ee( to ) 'uh' was the only one used by the novice level
subjects, specifically K and JL. In the case of IL, 2 out of the 3 times she used ee( to),
there was approximately a 2—second-pause before she uttered it, suggesting that she was
consciously thinking what she must use to fill the pause. K and JL used English fillers
as well as Japanese fillers.

The intermediate level subjects used three additional production-related fillers, soo
desu nee 'let me see', ano(o) 'uh, well' and mm 'uh, well', and none of the subjects used
English fillers. Eeto, ano(o) and soo desu ne( e ) are usually the first fillers introduced in
introductory Japanese textbooks.6 Ano(o) in particular is introduced in most Japanese
textbooks as a polite hesitation marker (Situational Functional Japanese) or as the most
common hesitation marker (Japanese: The Spoken Language). NH, an intermediate level
subject, commented in the questionnaire (see Section 6.6.1) that she uses it whenever she
needs to stall because she heard Japanese people use it when she was living in Japan.

Another intermediate level speaker, PK, stated that he uses ano(o) in place of um when

 

6 For example, eeto 'uh' is introduced in the first chapter out of 7 chapters in Yookoso. Ano(o) 'uh, well'
is introduced in Ch. 1 out of 8 chapters in Situational Functional Japanese and Ch 4 out of 12 chapters in

Japanese: The spoken language. 500 desu nee 'let me see' is introduced in Ch. 8 out of 30 chapters in the

Introduction to Modern Japanese.

249

speaking in Japanese.

This leads to my finding (3): the high rate of ano(o) among the advanced level
subjects. As we saw in Table 6.9, ano(o) constituted 71.2% of all the fillers used by the
advanced level subjects and their rate of ano(o) per 1,000 words was higher than the NNSs
at any level or the native subjects. A characteristic of the usage of ano(o) by the
intermediate and advanced level subjects was that, unlike the native subjects, the use of
ano(o) did not show any consistent pattern. Most instances of ano(o) used by the native
subjects were analyzed as a hesitation marker for specific purposes, such as to highlight a
proposition, and to preface topic shift, disagreement, and side-sequencing. The superior
level speakers' use of ano(o) was similar to the that of native speakers.

Most instances of ano(o) by the intermediate and advanced level subjects, in
contrast, were analyzed as a floor-holding device. Of 242 tokens of ano(o) used by the
advanced level speakers, 6 preceded expressions of disagreement, and 5 prefaced repairs,
and others were used while searching for appropriate expressions. (3) illustrates the use of

ano(o) for language-production-related purposes by an advanced level subject.

(3) DM: a. Irna minna anoo N ihon no gakkoo de benkyoo shiteru ne/
now everyone uh Japanese school at study do-PROG [P

b. doo yuu no kanaa/ anoo/ kochira no improvement shitai de ne?
how say NOM I wonder uh here improvement do-want IP

Mieko: c. Huun/ improvement ttee no wa nan desu ka?
uh huh improvement QT NOM-TOP what is Q

DM (1. Ano/ doo yuu no kana/ ano/ doo/ huu/ doo yuu huu/ anoo/
uh how say NOM wonder uh how way how say way uh

nante yuu no kana/ systenr/ nante yuu no kana/ doo/ nante
what say-NOM wonder system what say-NOM wonder how become

e. kochira no gakkoo ga motto ii/ aa shitai tte ne.
here school-SUB better that way do—want QT IP

f. Sore benkyoo ano N ihon no kyooiku o benkyoo shiteru.
that study uh Japanese education-DO study do-PROG

250

DM: a. Everyone now uh, is studying in Japanese school now, y'know?
b. what shall I say, uh something we want to improve here, y'know?
Mieko: c. Uh huh? What does the word "improvement" mean?
DM: d. Well, what shall I say? uh/ how?/ how?/ uh/
I wonder how to say it/ system/ how shall I say it? how/ what?
e. our schools here/ better/ to change them in a certain way, y'know?
f. They are studying that/ uh/ they're studying Japanese education.’

In (a)—(f), DM uses ano(o) 6 times. Ano(o) in (d) can be interpreted either as a tum—initial
use of ano or as a pause-filling device. However, given that the whole utterance in (d)
consists of a string of meta-communication fillers, such as doo yuu no kanaa 'how shall I
say it?‘ and nan te yuu no kanaa 'what shall I say it,‘ it is more natural to interpret ano(o) in
(d) as a pause-filling device while searching for words, rather than as a tum-initial use.
Compare (3) with Example (4), which illustrates the use of other fillers by SM, a
superior level subject, talking about the Korean and American educational systems.

(4) Mieko: a. Kankoku to Amerika no kyooiku seedo o tootte kita gakusee o kurabete
Korea and America education system-DO go through student-DO compare

b. koo jinzai-teki ni wa dochira no hoo ga yuushuuna hito ga
uh human resource-DO which one -SUB excellent person-SUB

dete kuru n deshoo nee.
come out BE Q

SM: c. Sonoo/ nante yuu ka/ anoo hikaku tte yuu no wa
well what shall I say uh comparison TOP

muzukashii desu ne/ yappari.
difficult BE IP as you know

(1. Demo aru imi de wa desu ne/ anoo/ Amerika no shisutemu no hoo ga koo/
but in a sense IP well American system -SUB like this

e. yuushuuna / ano eriito no koto ittemasu kedo ne/
excellent well elite about say-PROG but 1P

f. maa toppu reberu no gakusee tachi o tsukuru ni wa
kind of top level student pl. -DO produce for-TOP

g. Amerika no hoo ga muite iru n ja naika to omou n desu kedo ne/ ee.
America -SUB suited TAG QT think assert but [P yes

251

Mieko: a. 'When you compare students who went through Korean educational system
with those who went through American educational system,
b. which one do you drink produces better human resources?
SM: c. Well, what shall I say, uh, it's hard to compare them, as you know.
(1. But in a sense, uh, American system, like
e. excellent, well I'm talking about the elite,
f. for producing kind of the top-level students,
g. American educational system may be better, I drink, yes.‘

Compared to DM, who used only ano(o) and nante (doo) yuu no 'what shall I say' as
pause-filling fillers, SM used 6 different kinds of fillers: sonoo 'uh, well', nante yuu ka
'what shall I say', ano(o) 'uh, well', yappari 'as expected', koo 'like this', and maa 'kind
of. Ano in (c) and (d) is probably used as a pause-filling device, but ano in (e) is used for
background repair. SM uses sono and koo in lines (c) and (e) where ano could have been
used instead. Thus, this example shows diversification of the use of ano(o) by a superior
level speaker and the use of odrer fillers in place of ano(o).

The frequent use of ano(o) by the intermediate and advanced level subjects in my
data is perhaps attributable to: (1) NNSs' familiarity widr the ano(o) drrough their
classroom instruction; (2) dreir exposure to ano(o) used by native Japanese speakers; (3)
their lack of knowledge of other frllers; and (4) transfer of an English communication
strategy into speaking Japanese, e.g. translating the English um as ano(o) in Japanese.

It seems that drey often overgeneralize the function of ano(o) to situations where its
use is not appropriate, and use it for any situation where they wish to express hesitation,
including situations where odrer frllers are better suited. As NNSs become familiar widr
other stalling devices, such as nanka 'like' and hora 'there; you know what I mean', and
also widr socially motivated fillers, dreir use of ano(o) is expected to decrease, as was the

case widr the superior level subjects in the present study.

6.4.4. Summary

In this section, the findings regarding filler use by NNSs in comparison widr that

252

by native speakers were presented. As was the case widr connectives, the NNSs'
proficiency level and the length of formal Japanese study were the most significant factors
affecting the use of fillers, followed by the length of the time spent in Japan. The results
show that fillers are used for language-production-related purposes before dreir use for
social or interactional functions, supporting Hypothesis 5. In the case of a filler with bodr
production-related and social functions, e.g. ano(o) 'uh, well' and arm 'uh, well', its use
as a production filler is learned first. Finally, as non-native speakers reach higher
proficiency levels, their use of frllers become diversified and dreir pause-filling function

becomes less important.

6.5. Interactional particles
6.5.1. Quantitative analysis

The findings of interactional particles also supported Hypodresis 4. As widr
connectives and fillers, the NNSs at superior and advanced levels used a larger variety of
interactional particles and used them more frequently than drose at the intermediate or
novice levels. None of the novice level subjects used any interactional particles during the
interviews. However, during the role-plays (not included in the quantitative analysis), two
subjects used interactional particles: KK used the confirmation particle ne and the assertion
particle yo, each one once, and AP used ne once and yo twice. This shows that drey knew
how to use ne and yo correctly, given appropriate circumstances.

The intermediate subjects used three particles, ne (84.5%), yo (14.1%) and yone
(1.4%). Thus, judging from the novice and intermediate subjects, ne and yo are the most
likely to be the first particles learned by NNSs, followed by yone.7 Mine (1995), based on
her 8—month longitudinal study on the acquisition of sentence-final expressions by non-

native subjects of Japanese, suggests a general order of acquisition of interactional

 

7 Sakurai (1997) obtained similar results with her studies of yo and ne by first drrough fourth year college
students of Japanese in her cross-sectional study.

253

particles in Japanese: ne and yo are learned first, and followed by yone (p. 68). My cross-
sectional data confirmed her findings. Ohta (1994), based on her one-year longitudinal
study, also found that ne was the only interactional particle used by her first-year Japanese
subjects, suggesting drat ne is the first particle learned in Japanese class.

The advanced level subjects used 6 interactional particles and their total number of
particles used was much higher than drat of the intermediate level subjects. The percentage
of ne was the highest (78.8%), and kana, a particle of uncertainty, was the second highest
in percentage. This concurs with Mine's finding drat the self—directed no and kana are
learned relatively early by non-native speakers, following ne and yo. Thus, Mine's claim
about a general order of acquisition is again supported by my cross-sectional study.

The superior level subjects' use of interactional particles was similar to drat by the
native subjects in terms of the variety of particles. However, dreir percentages of ne and
yone were much higher than that of the native subjects, constituting 86.9% of all particles
in contrast to the native subjects' 63.7%. Table 6.12a summarizes the the tokens and

percentages of interactional particles used by the NNSs.

Table 6.12a The tokens and percentages of interactional particles by proficiency level

 

 

 

 

 

native superior advanced intermed. novice
marker functions tokens (%) tokens (%) tokens (%) tokens (%) tokens (%)
ne cooperation 239 (56.4) 246 (68.7) 152 (78.8) 60 (84.5)9 .....
yo assertion 60 (14.2) 35 (9.8) 7 (3.6) 10 (14,1) .....
na emotion 56 (13.2) 6 (1.7) 2 (1.0) ----------
no question 2 (0.5) 1 (0.3) ...............
kana uncertainty 36 (8.5) 5 (1.4) 23 (11.9) ----------
kashira uncertainty (F)8 ---------- 1 (0,5) ..........
yone confurnation 31 (7.3) 65 (18.2) 8 (4.2) 1 (1.4) -----
total 424 (100) 358 (100) 193 (100) 70 (100) -----

 

 

 

 

 

 

 

Widr regard to the position of ne, NNSs at intermediate and advanced levels usually

 

8 Kashira is exclusively used by female speakers.
9 30 tokens of ne out of 60 were produced by a single subject, PC.

254

employed it sentence-fmally (98.3% and 80.3%, respectively). The superior level subjects
used it utterance-medially (35.4%), sentence-fmally (64.2%) and by itself (0.4%). This
suggests that ne in the sentence-final position is learned before other positions. Thus, not
only the frequency of ne, but its position is related to proficiency level.

It is evident that as the learner becomes more proficient, ne (and ya) constitute a
decreasing percentage of total particle use: intermediate 84.5% (98.6%); advanced: 78.8%
(82.4%); superior: 68.7% (78.5%); native subjects: 56.4% (70.6%). Table 6.12b

summarizes the breakdown of the use of ne by its position.

Table 6.12b The breakdown of ne by position and proficiency level (tokens and %)

 

 

 

 

 

native superior advanced interrned. novice
marker tokens (%) tokens (%) tokens (%) tokens (%) tokens (%)
ne medial 34 (14.2) 87 (35.4) 30 (19.7) 1 (1.7) -----
final 202 (84.5) 158 (64.2) 122 (80.3) 59 (98.3) -----
alone 3 (1.5) 1 (0.4) ...............
total 239 (100) 246 (100) 152 (100) 60 (100) -----

 

 

 

 

 

 

 

As regards the rates of particles per 1,000 words, the total rates were higher for the NNSs
at advanced and superior levels than for drose at novice and intermediate levels. Of all the
interactional particles used by the NNSs, the most frequently occuning one was ne by the
NNSs at all levels, and the rates were higher than that of the native participants. Table

6.13a summarizes the rates of interactional particles by proficiency level.

Table 6.13a The rates of interactional particles/1000 words by proficiency level

 

 

 

 

marker functions native superior advanced interrned. novice
ne cooperation 17.5 35 3 5 . 1 20.9 -----
yo assertion 4.4 5 1.6 3.5 .....
na emotion 4.1 ' 0.9 0.5 ----------
no question 0.2 0 . 1 ...............
kana uncertainty 2.6 0.7 5 .3 ----------
kashira uncertainty (F) ---------- 0 . 2 ..........
yone confnmation 2.3 9 . 4 l .9 0.4 -----
total 31.1 52 44.6 24.7 -----

 

 

 

 

 

 

 

255

Table 6.13b summarizes the breakdown of the use of ne by position.

Table 6.13b The breakdown of ne by position and proficiency level (rates/1000 words)

 

 

 

marker functions native superior advanced interrned. novice

ne medial 2.49 13 6.93 0.35 -----
final 14.8 23 28.2 20.6 -----
alone 0.22 0. 1 ...............

total 17.5 35 35.1 20.9 -----

 

 

 

 

 

 

 

 

6.5.2. Sociolinguistic factors and the use of interactional particles
6.5.2.1. Age and gender

We now examine the use of interactional particles in relation to age and gender.
While age was not a significant factor, gender showed some relationship to the use of
interactional particles in the NNSs' data. In particular, ne was used more often by men
than women (82.6 vs. 54.4/1000 words), and the difference is statistically significant (P:
0.035). The rate of no and the total rate of interactional particles also showed a strong
tendency drat their use is influenced by gender. Gender was not a significant factor in the
use of interactional particles in native Japanese interviews, however.
6.5.2.2. Other factors

Widr regards to the other three factors, proficiency level and formal Japanese study
were more significantly related to the use of interactional particles than was time spent in
Japan, parallelling the result obtained widr fillers and connectives as discussed in Sections
6.3.2.2 and 6.4.2.2.

The proficiency level and formal Japanese study significantly affected the rates of
ne (P=0.0157 and 0.0116, respectively) kana (P=0.0086 and 0.0000, respectively), and

yone (P=0.0009 and 0.0158, respectively) as well as the total rates of particles (P=0.0018

256

 

and 0.0017, respectively). Thus, proficiency level and the length of formal study
significantly promoted the use of interactional particles. The time the NNSs spent in Japan
was related to the rate of ne (P=0.0520) and yone (P=0.0399) as well as the total rate of
particle usage (P=0.0476). Table 6.14 summarizes the P-values of the rates of interactional
particles to gender, proficiency level, the lengdr of formal study and the time spent in

Japan.

Table 6.14 The P—values of the factors which influenced the use of interactional particles

 

 

 

 

 

The P-values for the 4 factors

marker functions gender Prof. level formal Japanese study time in Japan
ne cooperation 0.035 0.0157 0.0116 0.0520
yo assertion ns ns ns ns
na emotion 0.064 N/A N/A N/A
no question ID ID ID ID
kana uncertainty ns 0.0086 0.0000 ns
kashira uncertainty (F) ID ID ID ID

one confirmation ns 0.0009 0.0158 0.0399
total 0.065 0.0018 0.0017 0.0476

 

 

 

 

 

 

P3005: significant correlation; 0.05<P30. 1: strong tendency; ns: not significant; ID: indetenninable due to
low occurrence, N/A: no P-value (no occurrence)

6.5.3. Discussion

Let us turn to the use by the NNSs of the most common final particle, ne. My
findings are as follows: (1) the frequency of ne as a percentage of all particles used was
higher at lower proficiency levels than higher levels (84.5% at the intermediate level;
78.8% at the advanced level; and 68.7% at the superior level) (2) no errors in interactional
particles usage occurred except for overuse of ne by one intermediate level subject; (3) the
occurrence rates of ne by the NNSs were higher than that of the native subjects.

The high rates of ne by the NNSs may be attributed to the following facts: (1) ne is
the most commonly used final particle among native speakers (Maynard 1993:102); (2) it is

257

introduced early in Japanese instruction; '0 and (3) it is the most frequently used particle by
Japanese language teachers in the classroom according to Ohta (1994:310), constituting
86% of all the particles used by the three teachers in her data, followed by deshoo (5%), yo
(5%), kana (4%), etc.

Ne, as a marker of cooperative attitude, as discussed in Chapter 5, creates an
agreeable atmosphere in face-to-face interaction. Its overuse, however, makes one's
speech unnatural. The frequent use of ne by the NNSs may derive from their
misperception of the functions of ne as well as the complexity of its functions. In the
questionnaire filled out after the conversations and interviews, 4 native subjects wrote drat
they use ne with close friends, and one wrote drat she does not use ne to an older person.
An intermediate level subject wrote drat she uses ne whenever she uses ano(o). Anodrer
intermediate level subject wrote drat he uses ne for emphasis. It is quite plausible drat
NNSs at the intermediate level do not yet know exactly when they should or should not use
ne. As Martin (1975:916) states, overuse of ne can be irritating to the listener.

The overuse of ne is likely to be related to the lack of familiarity of odrer particles.
Mine (1995:70) states that during the first month of her longitudinal study, ne was often
misused, and drat by the eighth month, the usage error decreased as her subjects learned
other particles appropriate to various situations. (5) is an example of overuse of ne by one

of my intermediate subjects.

(5) Mieko: a. Gokazoku wa naka ga ii desu ka?
family-TOP relationship-SUB good BE Q

b. Ironna koto o issho ni shimasu ka?
various dring-DO together do Q

PC: c. Aa soo desu ne.
oh so BE IP

 

'0 For example, ne is introduced in Ch. 1 of Y ookoso and Japanese: The spoken language (out of 7 and 12
chapters, respectively), in Ch. 2 (out of 12 chapters) of Situational Functional Japanese, and in Ch. 3 (out
of 30 chapters) of the Introduction to Modern Japanese.

258

(1. A3] daitai hataraku dake desu ne.
oh usually work only BE IP

Mieko: e. Hataraku dake?

work only

PC: f. 800 omoirnasu ne.
so think IP

Mieko: a. 'Are your family members close to each other?
1). Do they do various drings together?

PC: c. Oh yes.
(1. Oh we just work.

Mieko: e. Just work?

PC: f. I think so.’

In (c), (d), and (f), PC ended his utterance with ne pronounced with a slightly rising
intonation. Since the information in these utterances is about his own family, which I
know nothing about, he should have used the bare form or yo, rather than ne. If ne was
pronounced widr a falling or a middle sustained intonation, it would not have sounded as
odd, since native speakers sometimes use ne when presenting information widr a sense of
detachment (see Section 5.4.3). PC's intonation and frequent use of ne, however, made
his utterances sound unnatural. Though he used bodl ne and yo during his interview,
perhaps he did not have clear understanding of the use of yo, ne, and the bare form of
verbs.

Overusing particles in conversation, as well as lack thereof, can create problems,
such as sounding too insistent and too familiar in the case of yo or sounding too familiar in
the case of ne. Therefore, it is advisable to include explicit instruction about how to use
interactional particles, including when to use or not use them, in various contexts and at

various stages of instruction.

6.5.4. Summary
To summarize the use of interactional particles by non-native subjects, ne was the

most commonly used particle. The most significant factors affecting the use of interactional

259

particles are proficiency levels and the length of formal study, followed by the time spent in
Japan and gender. My cross-sectional data and Mine's longitudinal study both suggest drat

ne and ya are learned first by NNSs.

6.6. Pedagogical implications of the present study
In this section, perceptions of some fillers and interactional particles by native and

non-native subjects are discussed first, followed by pedagogical suggestions.

6.6.1. Natives' and NNSs' perceptions of fillers and particles

The results of my questionnaire regarding the fillers indicate drat native
speakers attach various roles to a given filler. Native and non-native participants have
largely similar perceptions about the roles of the frllers surveyed, though drere were some
discrepancies, particularly among the intermediate level participants (see Appendix 2 and 3
for the format of the questionnaires).

It should be noted that for each of the four fillers, ee( to) 'uh', ano(o) 'well', nanka
'like', and ma(a) 'kind of', there were several native participants who stated drat drey try
not to use them (TH, a 66-year old male teacher, stated he avoids using ano(o) and ee( to),
TA, a 40-year old female teacher, and KH, a 56 year old female teacher, stated they do not
use nanka and ma( a )), which indicates drat the use of fillers is not always considered to be
appropriate by native speakers.

Table 6.15 summarizes the results of the questionnaire as regards fillers.11

 

1' The format of the questionnaire was revised after interviewing the NNSs. Therefore, they did not
comment on nanka 'like' and maa 'kind of. Not every participant answered all the questions, and some of
them gave more than one answer for some questions. Therefore, the totals are sometimes different from the
number of participants.

260

Table 6.15 The results of the questionnaires (fillers)

 

 

 

 

 

 

 

 

 

 

situations where the fillers are used native (n=24) NNSs (n=24)
ee(to) when searching for appropriate expression 11 (40.7%) 8 (36.4%)
'uh' while thinking 10 (30%) 6 (27.3%)
while recalling something 2 (7.4%) 3 (13.6%)
when not sure 1 (3.7%) -----
to stall or pause ----- 3 (13.6%)
at the beginning of utterance ----- 2 (9.1%)
do not use widr friends 2 (7.4%) -----
do not use it often 1 (3.7%) -----
total 27 (100%) 22 (100%)
ano(o) to a stranger 6 (27.2%)
'well; searching for appropriate expression 5 (22.7%) ' 10 (50%)
uh' while thinking 3 (13.6%) 6 (30%)
to say things which are hard to say 3 (13.6%)
to get by a situation 3 (13.6%)
when not sure 2 (9.1%) 1 (5%)
use instead of um I (5%)
at the beginning (of a long explanation) 2 (10%)
total 22 (100%) 20 (100%)
nanka when words do not come out 5 (22.7%)
'like' when I don't want to emphasize something 4 (18.2%)
to express uncertainty 4 (18.2%)
at the beginning of an utterance (no meaning) 2 (9.1%)
to people who are familiar 1 (4.5%)
to complain 1 (4.5%)
use it without thinking 1 (4.5%)
use it like a conjunction 1 (4.5%)
try not to use it often 3 (13.6%)
total 22 (100%)
ma(a) when do not agree completely 7 (41.2%)
'kind to present a counter argument softly 4 (23.5%)
of' to fill a pause (no meaning) 3 (17.6%)
to express doubt 2 (11.8%)
o express minor conclusions 1 (5.9%)
total12 17 (100%)

 

 

 

As regards the roles of the interactional particles ne and yo, native and non-native

responses were again similar. Among the native subjects, SH (a male college student)

stated he does not use ne often, and FY, KM, NM, and KY (all college students, FY a

male, the odrer drree are females) stated drat they try not to use yo often.

 

‘2 5 native speakers confused maa 'kind of with a female interjection maa 'oh' and answered accordingly,

which is not reflected in Table 6.15.

261

 

The result of the questionnaire indicate that native and non-native subjects assign
similar functions to the fillers and interactional particles surveyed, though the perception
differences were greater among the novice and intermediate level subjects than among the
advanced and superior level subjects. Table 6.16 summarizes the results of the

questionnaire.

Table 6.16 The results of the questionnaires (interactional particles)

 

 

 

 

 

 

 

 

 

particle situations where the particles are used native non-native
ne ask for agreement 10 (31.3%) 10 (40%)
to express agreement 6 (18.8%) 9 (36%)
to confirm 6 (18.8%) 2 (8%)
to familiar people 4 ((12.5%) 1 (4%)
for invitation 1 (3.1%) -----
to be polite l (3.1%) -----
when speaking to women 1 (3.1%) -----
for emphasis ----- 1 (4%)
for shared information ----- l (4%)
always use with ano ----- 1 (4%)
don't use much 1 (3.1%) -----
do not use to an older person 2 (6.3%) -----
total 32 (100%) 25 (100%)
ya to emphasize something 9 (37.5%) 5 (33.3%)
to give information 4 (16%) 4 (26.7%)
to assert something 4 (16%) 4 (26.7%)
when advising 1 (4.2%) -----
with people who are familiar 2 (8.3%) -----
to express strong feelings ----- 2 (13.3%)
don't use much 4 (16.7%) -----
masculine sounding l (4.2%) -----
total 24 (100%) 15 (100%)
6.6.2. Classroom implications

 

The results of the present study suggests pedagogical areas which merit attention:

1. Explicit instructions on discourse markers should be included in Japanese
language instruction from the beginning and expanded at every level for the following
reasons: (1) discourse markers are an integral part of Japanese conversation, connecting

utterances, expressing the speaker's orientation to the proposition and to the addressee, and

262

widrout them, one's speech will sound unnatural; (2) all discourse markers analyzed in the
present study have more than one function and are used in a variety of situations, and drus
dreir functions and the contexts in which they occur may not be easily understood widrout
explicit explanations; and (3) many students will not have enough opportunities to hear
native speaker's natural conversation.

Aldrough it is impossible to teach all discourse markers at once, the instruction
should consciously include the use of discourse markers in a wide variety of contexts at
various stages of instruction, such as the use of (sore )de as a floor-holding device or the
use of filler, nanka 'like' to express uncertainty.

As specific teaching strategies, I suggest the following be included as part of
explicit classroom instruction of discourse markers:

(1) Increase the teachers' use of discourse markers and make dreir use salient.
Ohta (1994:314) states drat the frequency of interactional particles used in the classroom is
much less than in normal conversation. Larsen-Freeman (1976, cited in Larsen-Freeman
and Long 1991 :91) claims drat input frequency is significantly correlated with morpheme
acquisition order in her ESL classroom. Kasper and Blum-Kulka (1993:35) argue drat
"attention to input is a necessary condition for any learning at all." Many of the
intermediate and advanced level subjects in the present study were not aware of fillers such
as nanka 'like' and and ma(a) 'kind of'.

By increasing the use of discourse markers in the classroom and by pointing out
their usage, teachers can make students aware of the functions of discourse markers. Ohta
(1994:318) points out the importance of the use of "ne-marked assessments (i.e. teacher's
personal reaction to students' utterance, such as ii desu nee! 'how nice!‘) and eye-gaze as
strategies to increase the use of ne and drawing the students' attention to it. It is also
important drat teachers emphasize communicative activities using natural language as well

as instruction in grammar and vocabulary, including the use of discourse markers.

263

(2) Use recordings of natural, authentic conversations in class at various stages of
instruction. It is important not simply to listen to the recordings but practice how to use
particular discourse markers which the teacher wishes to introduce. By analyzing the use
of discourse markers in naturally-occurring conversations and by drilling on them, students
will have better understanding of the functions of discourse markers.

(3) Provide opportunities for students to use discourse markers whenever possible.
Hadley (1993:79-87) argues that contexts must be provided in language learning so
learners can apply dreir knowledge to coping widr real-life situations. The importance of
context is illustrated by my two novice level subjects, K and AP, who used interactional
particles yo and ne in their role plays but not during interviews. This suggests that drey
knew how to use these particles when provided appropriate contexts. Therefore, learners
should practice the use of discourse markers in environments in which discourse marker
use is imperative, such as in role-plays and guided conversations.

(4) Include the use of discourse markers as part of the evaluation of students'
progress. By being tested on the use of discourse markers, students are more likely to use
discourse markers consciously.

Okumura (1995:34-36), based on her study of teachers' and nonteachers'
perception of elementary leamers' spoken Japanese, states drat one of the most important
criteria for judging non-native speakers' proficiency is fluency and appropriateness as well
as correct grammar or vocabulary. The use of discourse markers certainty is part of
appropriateness. Kasper and Blum-Kulka (1993:35) state drat "explicit teacher-provided
information about the pragrnatics of the second language can play a role in learning,
provided that it is accurate and not based solely on fallible native speaker intuition." An
analysis of discourse markers such as the present study will serve as an important source of

such information.

264

CHAPTER 7

Conclusion

This chapter summarizes the findings regarding the roles of discourse markers in

Japanese conversation and discusses limitations of the present study.

 

7.1. Findings of the present study

This study focused its attention on Japanese connectives, fillers and interactional

 

particles as discourse markers as used by native and non-native speakers of Japanese,
using Schiffrin's (1987) framework. 1 have shown drat connectives, fillers, and
interactional particles all function as discourse markers in Japanese, and drat drey play
important roles in Japanese conversation, conjoining idea units and indicating the speaker's
feelings and orientation toward a proposition and/or the addressee. The importance of
connectives, fillers, and interactional particles in conversation was in part supported by
dreir frequent occurrence in my data, and partly by their various functions regarding the
proposition-proposition, speaker-proposition and speaker-addressee relationships.

To my knowledge, the present study is the fnst attempt to analyze connectives,
fillers, and interactional particles as items all belonging to one category, discourse markers,
and in one framework, though some connectives and interactional particles have previously
been analyzed separately (e.g. Maynard 1989a, Onodera 1993, [to 1995, Oishi 1985,
Ishikawa 1988 and Cook 1988, inter alia). The filler ano(o) has been analyzed by Cook
(1993). I also presented an analysis of discourse marker usage by non-native subjects at
four proficiency levels defined by the American Council on the Teaching of Foreign
Languages (ACTFL), and compared it with that of native speakers.

The primary function of connectives is to conjoin idea units, but drey also operate at

the interactional level. For example, (sore )de 'and' sequentially coordinates idea units,

 

expresses cause-effect relationships, and also operates as a floor-holding device or as an

265

entry—device for co-construction. Demo 'but' connects contrasting idea units (referential
contrast), as well as connecting functionally different units (functional contrast) and

contrasting actions (e.g. claiming the floor during the odrer speaker's turn). Dakara 'so'
connects units in cause-effect relationships, and can also signal the speaker's irritation at
having to repeat what has previously been said as well as to signal the end of one's turn.

My data show drat the most significant factors which influence the occunence rate
of connectives by native speakers are dreir gender and the speech genre. Connectives were
used more frequently by female speakers than by male speakers. De 'and' was used most
often in narratives where the speaker holds the floor for a long time. Formality and the
speakers' age were not significant factors in the use of connectives.

Fillers' primary function as discourse markers is in participation framework and
exchange structures, revealing the cognitive process, such as information management, and
the speaker's feelings and orientation towards a proposition and/or the addressee. For
example, a 'oh' and aa 'oh' express the speaker's receipt of information. Some fillers can
also be used in the exchange structure as tum-entry devices (e.g. ano(o) 'uh, well' and
nanka 'like'). The main function of ano(o) is to express the speaker's hesitation before
introducing a new conversational move, such as topic change, disagreement or highlighting
a proposition. Nanka 'like' indicates the speaker's uncertainty and hesitation, revealing his
or her attitude towards the proposition and/or the addressee. Many fillers, including
ano(o), uun 'uh, well' and nanka 'like' can function both as floor-holding devices, while
the speaker searches for an appropriate expression, and as socially motivated fillers to
create an agreeable atmosphere for conversation. Ma( a ) 'kind of expresses the speaker's
compromise and downplaying. Thus, it is a useful device to express a conciliatory
agreement without overtly stating how one really feels.

The most significant factors which influence the fillers usage are formality and the
speaker's age. Fillers are used more frequently in formal-style speech, where "social
packaging" is more prevalent than in informal-style speech. The speakers' gender and

266

 

 

speech genre were not significant factors in the use of fillers.

Interactional particles, except ne, na, and sa, occur exclusively utterance-fmally,
indicating the speaker's feelings and orientation towards a proposition and/or the
addressee. Ne is used to indicate cooperative attitudes or to align the interlocutors on the
same side, as Cook (1993) observes, and drus is used to seek and express agreement and
confirmation. It is also used to soften the tone in disagreement, as an attention-checking
device, and as an entry device when used utterance—initially. Yo is used to express the
speaker's emphasis and reveals the speaker's judgement as to the speaker-addressee
relationship as a familiar and personal one. Yone indicates the speaker's conviction and
cooperative attitude simultaneously, and also works as a device to request agreement or
confirmation and to check the addressee's attention.

The use of interactional particles was most significantly affected by formality,
speakers' age, and speech genre. They were used more often in informal-style speech,
confirming Uyeno's observation that "the closer and less formal relation a speaker and
addressee maintain, the more sentence particles are allowed in their conversation (1971:50).
Older participants used more interactional particles than younger ones.

As regards non-native speakers' use of Japanese discourse markers, my data show
the following:

1. The NNSs at higher proficiency levels, such as advanced and superior levels,
used more discourse markers both in number and variety; the superior level speakers' use
of connectives was similar to drat by native speakers, indicating drat superior level speakers
truly have native-like interactive and discourse strategies as defined in the ACTFL Oral
Proficiency Guidelines (the ACTFL 1988:11).

2. The NNSs at lower proficiency levels used connectives for ideational functions
primarily, while those at higher proficiency levels used drem both for ideational and
interactional functions, as so native speakers. This confirmed my hypodresis drat the

ideational functions of connectives are learned before interactional ones.

267

 

 

 

3. The intermediate and advanced level subjects used ano(o) at a high rate and
primarily for production-related purposes. This may indicate that the functions of ano(o)
for social purposes have not yet been learned at these proficiency levels, or drey do not
know any other fillers drey can use as pause-filling devices.

4. The rates of interactional particles were higher among advanced and superior
level subjects than drose at the novice and intermediate levels, and native speakers.

5. Of the sociolinguistic factors examined for the present study, proficiency level
and lengdr of formal Japanese study most significantly influence the use of discourse
markers by non-native subjects. Age, gender and the length of the time spent in Japan
were not as significantly related as the odrer two factors.

Given my findings about NNSs' discourse marker usage and frequency, I
proposed that explicit instruction on the specific functions of discourse markers be
incorporated at various stages of instruction in order to develop non-native speakers'

overall competence.

7 .2. The roles of discourse markers in Japanese conversation

My data showed how frequently connectives, fillers and interactional particles are
used in Japanese conversation by native speakers. Many connectives are optional. For
example, hearers may understand whether two consecutive idea units are in coordinate or
contrasting relationship from the meanings of the units dremselves. The use of
connectives, however, facilitates the understanding of discourse structures, makes the
relationships between utterances clearer, and leads to a better discourse coherence. ‘

Many fillers are optional as well. They, however, are useful as floor-holders while
the speaker searches for an appropriate expression. They also contribute to creating a
socially and culturally expected interactional atmosphere by showing the speaker's
emotions and orientation, such as hesitation and uncertainty (e.g. nanka 'like' and maa

'kind of), since appearing uncertain is somewhat expected in many situations in Japanese

268

 

 

conversation, even if the speaker is certain about somedring. Fillers may also display the
speaker's emotions (e.g. moo 'really'), and help enhance the addressee's involvement in
conversation (e.g. hora 'you know what I mean' and yappari 'as expected').

Interactional particles also contribute to creating an agreeable atmosphere; e.g. by
softening the speaker's tone and expressing cooperative attitudes. Additionally, drey
display the speaker's feelings and emotions, such as emphasis and excitement by the use of
yo, or the expression of closeness by ne and yo. The speaker can use drem as attention-
getting or attention-checking devices as well (e.g. the utterance-initial ne and utterance-
medial ne and sa). Interactional particles, as Uyeno (1971 :49) and Maynard (1997:87)
state, are used by the speaker in order to communicate widr the addressee in an emotional
and empathy-creating way.

As regards the positions and the speech units they bracket, connectives primarily
occur utterance-initially as boundary-markers, connecting ideational (e.g. de 'and'), tum-
taking (e.g. dakara 'so' at the end of one's turn) and speech act units (e.g. the use of demo
'but' for point-making). Fillers primarily occur utterance-medially, indexing utterance-
internal units, such as phrases, to prior and/or upcoming units and to the interlocutors. For
example, ma( a ) 'kind of expresses downplaying of the effect of the upcoming utterance
and nanka 'like' shows the speaker's uncertainty or desire to appear uncertain. Fillers may
also bracket turns (e.g. nanka 'like' as a tum-initiator) and speech acts (ano(o) 'uh' as an
attention-getter, as well as a tum-initiator). Interactional particles occur primarily utterance-
fmally, with the exception of sa, which occurs bodr phrase and utterance-fmally, and ne,
which occurs utterance-initially and phrase-fmally, as well as utterance-initially or by itself.

The most important function discourse markers have is as textual and participant
coordinates. In other words, they, as Schiffrin (1987:321) states, select and display, rather
than create, relationships between propositions, a proposition and the interlocutors, and the

speaker and the addressee, by indexing an utterance to prior and/or upcoming units and to

269

the interlocutors. All Japanese connectives, fillers and interactional particles do exactly
that.

Table 7.1 and Figure 7.1 summarize the functions of connectives, fillers and
interactional particles discussed in this thesis. Bold indicates the primary position for each

category of discourse marker.

Table 7.1 Positions of connectives, fillers, and interactional particles and the units they coordinate

 

 

 

 

 

type of marker positions main functions secondary functions
connectives utterance-initial coordinate idea units, express the speaker's
utterance-final turns, and speech acts orientation toward a
prgrosition or the hearer
fillers utterance-medial express the speaker's tum-initiator; attention-
utterance-initial cognitive process and getter, etc.
utterance-final feelings/ attitudes toward
the proposition/ hearer
interactional utterance-final express the speaker's tum-initiator
particles utterance-initial knowledge and meta- attention-getter
utterance-medial knowledge and feelings/ attention-checker
attitudes

 

l. connectives: primarily utterance-initial brackets (also utterance-final for dakara 'so')

[idea unit I
connective I turn
Lspeech act .I

2. fillers: primarily utterance-medial brackets (secondary: utterance-initial (e.g. ano(o) 'uh,
well' and maa 'kind of and eeto 'uh') or final brackets (e.g. nanka 'like' )

phrase filler phrase
3. interactional particles: utterance-final (secondary: utterance medial and final for sa and
no; and utterance-initial, medial, and final and by itself for ne)

I idea unit I
I turn I interactional particle
[speech act

Figure 7.] Positions of connectives, fillers, and interactional particles and the units they coordinate

270

 

7.3. Discourse markers and expected norms of Japanese society

Many researchers have stated drat wa 'harrnony, peace, and unity' is one of the
most important social values of Japanese society (e.g. Reischauer 1977; Barrrlund 1989;
Wierzbicka 1991 inter alia). Wierzbicka (1991:358) characterizes wa as follows: "it is a
desire for unity, closeness, concord, and peace and a refusal to credit or blame individuals
within a group, and made possible by conscious and common effort by the participants in a
society." Other characteristics attributed to Japanese society is modesty, self-effacement,
humility, and avoiding open conflict and confrontation widr odrers (e.g. Reischauer 1977
and Smith 1983).

In order to achieve harmony widrin the group and avoid open confrontation,
Japanese children receive what Clancy (1986:232-240) calls "empadry and conformity
training" in which, at quite an early stage of dreir life, mothers actively teach their children
Japanese norms for speech and behavior. Clancy states drat in giving directives, Japanese
mothers tended to appeal to social norms, unlike American mothers, who tended to express
dreir own feelings (p. 218). She further states that this early training in empathy and
conformity leads Japanese children to understand the feelings and expectations of others,
and that helps set the stage for the successful functioning of the Japanese indirect, intuitive
mode of communication (p. 240).

As shown in Chapters 3, 4, and 5 of this dresis, discourse markers play important
roles in Japanese communication bodr at the textual and the interactional levels. Fillers and
interactional particles particularly contribute to achieving wa in Japanese society. For
example, ma(a) 'kind of expresses downplaying and humility when the speaker is talking
to an addressee who is superior to the speaker, and/or when the speaker is talking about
him/herself or his/her family members. Clancy (1986:240) states drat one kind of
conformity training Japanese children is given by dreir mothers is how to avoid saying no

directly, because any overt expression of conflict could jeopardize the harmony between the

271‘

speaker and the listener. She argues that the Japanese try various indirect strategies for
saying no, the most extreme of which is simply saying yes. Ma( a ) functions precisely for
this purpose, giving conciliatory agreement when the speaker does not truly agree with
others.

Ano(o) also contributes to achieving harmony, fnst by aligning the speaker and the
listener on the same side, appealing to the sense of sharedness between drem, and secondly
by expressing the speaker's hesitancy for imposing somedring on the addressee or
reluctance to express the upcoming proposition. Nanka 'like' is another filler which is
used to accomplish in-group harmony. It can be used when the speaker desires not to
appear too confident or definite in the eyes of a certain addressee, such as one's superior.
Uun 'uh; well' and soo desu ne( e ) 'let me see' are still other fillers which can be used when
the speaker does not wish to say no overtly.

Interactional particle such as ne and yone also contribute to achieving harmony in
various situations. Both ne and yone are used to request or express agreement. Ne is also
used to check the addressee's attention in the middle of the speaker's turn.

There are fillers in English which express uncertainty, hesitation, or downplaying.
Owen (1983) states drat well is used to preface a face-drreatening response such as non-
compliance with a request, or rejection of an offer. Schiffrin (1987:280-282) states that
y’know is used to appeal to shared knowledge between the speaker and the hearer, and to
draw the hearer's attention to material which is important for the speaker. Kind of and sort
of are other fillers which can be used for uncertainty or downplaying.

As Maynard (1997:20) states, some languages have many language-explicit devices
that respond to relationality-based cues, whereas others have fewer and must use odrer
strategies in order to accomplish the same objective. She lists Japanese as an example of
the former, and English as that of the latter. Furthermore, in any society, drere may be
situations where harmonious and cooperative relations are valued. And, perhaps, drere are

situations when one feels the need to downplay one's utterance or not telling the whole

272

truth. What is crucial is the extent to which individuals in a particular society try to achieve
the goals of promoting harmonious and cooperative relationship with others, avoiding open
confrontation.

The Japanese are much more concerned than Americans, for example, widr
presenting dremselves as cooperative and harmonious beings in social interactions, or, in
Maynard's terms (1997: 19) "are more likely to be trapped in the self-imposed concept of
society." Since reluctance to disagree with another's opinions or refuse a request is so
strong in Japanese society, Uchida (1974, cited in Clancy 1986:215) states drat her subjects
rarely used no except within their families. As discussed by many researchers (Reischauer
1977, Smith 1983, Christopher 1984, and Bamlund 1989), Japanese society seems to
dictate the individual's effort to achieve harmonious relationships more strongly than
American society. The variety and frequency of discourse markers, fillers and interactional
particles in particular, provide another piece of evidence to support the characterization of
Japanese society presented by these researchers. And, needless to say, this facet of
Japanese society helps account for the very significant functions of discourse markers

described in this thesis.

7 .4. Limitations of the present study and suggestions for future study

There are several factors which limited the scope of this study. First, the number of
the subjects was relatively small, especially drat of non-native subjects (36 native and 24
non-native subjects). It is hoped that a similar study widr a greater number of subjects will
be carried out in the future to test the hypodreses presented here.

Second, the type of data was limited. The present study did not include informal
speech by older (over 35 years old) native Japanese. Lack of male speakers' data in
informal-style conversation compelled me to exclude the informal speech data collected

from older women in order to maintain consistency. All of the native subjects were

273

students and teachers. Studies that include data from various social groups need to be
conducted, particularly in investigating gender differences in the use of discourse markers.

Third, in the present study, except for the college-age native subjects, conversation,
interview and narrative data were collected from different individuals. Though my data
showed tendencies in the use of discourse markers in the three speech genres, it would also
be interesting to see how the same individuals' use of discourse markers varies according
to speech genre. This issue needs to be addressed in future studies.

Fourth, in the present study, I was able to get feedback about the use of discourse
markers only from a handful of the participants, since most of them left the country
immediately after recording. The subjects' feedback provided important information about
dreir use of discourse markers. A more systematic method of collecting subjects' feedback
is desirable.

Fifth, the present study on the use of discourse markers by non-native speakers
was limited to cross-sectional data. Future studies need to address the issue in longitudinal
situations in order to see the acquisition order of Japanese discourse markers by non-native
speakers.

Lastly, in the present study, only interview data were analyzed for non-native
speakers for the following reason: it was difficult to find conversation data from novice and
superior level speakers. Novice level speakers are not yet able to converse widr anodrer
novice level speaker, and superior level non-native speakers are difficult to find in the area
where the research was carried out. It is desirable to include non-native conversation data
in future studies.

Despite all drese limitations, 1 hope I have been able to shed light on the use of
Japanese discourse markers by native and non-native speakers. Discourse markers,
particularly fillers, may be used without the speaker's awareness and sometimes in spite of
his or her conscious effort not to use drem (e. g. TH who stated in the questionnaire drat he

tries not to use ano(o) and eeto used drem more often than any odrer older male speakers

274

and TA and KH who stated that they try to avoid using ma( a ) and nanka also used them,
though less often than other older female speakers). Without drem, the speaker may sound
too confident, too arrogant, and too self—assuring or too inconsiderate of the addressee. I
have shown that these discourse markers are indispensable and integral elements in face-to-
face interactions in Japanese, and that the Japanese society dictates their appropriate use.
Therefore, if learners of Japanese aim at achieving native-like proficiency, explicit
instructions on the appropriate use of discourse markers must be incorporated in classroom

instruction.

275

APPENDICES

276

Novice

Nov ice-High

Intermediate

Advanced

Superior

APPENDIX 1

The ACTFL Proficiency Guidelines
(cited in Hadley 1993:502-504)

The Novice level is characterized by the ability to communicate minimally
with leamed material.

Able to satisfy partially the requirements of basic communicative exchanges
by relying heavily on learned utterances but occasionally expanding these
drrough simple recombinations of dreir elements. Can ask questions or
make statements involving leamed material. Shows signs of spontaneity
although this falls short of real autonomy of expression. Speech continues
to consist of learned utterances radrer than of personalized, situationally
adapted ones.

The Intermediate level is characterized by the speaker's ability to:

-create with the language by combining and recombining learned elements,
though primarily in a reactive mode;

-initiate, minimally sustain, and close in a simple way basic communicative
tasks; and

-ask and answer questions.

The Advanced level is characterized by the speaker's ability to:

-converse in a clearly participatory fashion;

~initiate, sustain, and bring to closure a wide variety of communicative
tasks, including those that require an increased ability to convey meaning
with diverse language strategies due to a complication or an unforeseen turn
of events;

-satisfy the requirements of school and work situations; and

-narrate and describe with paragraph-length connected discourse.
-shortcomings can often be smoothed over by communicative strategies,
such as pause fillers, stalling devices, and different rates of speech

The Superior level is characterized by the speaker's ability to:
-participate effectively in most formal and informal conversations on
practical, social, professional, and abstract topics; and

-support opinions and hypothesize using native-like interactive and
discourse strategies.

277

APPENDIX 2

THE QUESTIONNAIRE FOR NATIVE JAPANESE SPEAKERS

INFORMATION SHEET Pair # Date

 

PLEASE ANSWER THE FOLLOWING QUESTIONS. (Fiamfiflfl 1C 15%;: 1‘ a v). E! 21233-5 is
$33? were. )

1. Name: 2. Age (35%):

 

3. Gender (1432”): Male (%) Female (it)

4. Classification ($35): Fr (#51?) So (:51?) Jr (3513) Sr (129513) Grad (kéfi) £15

 

 

Other (%UDflE) Occupation (fifi):

5. Major field of study (@137)

 

6. Hometown (dig): State (1%): City (111):

 

7. How close are you to your conversation partner? (%3EIH$&®§§5‘§I1T®B‘NC¥1K 9 ifi’o )
a.. very close friend (fifi) b. good friend (26%) c. acquaintance (931841))
(1. don't know the person well (.1: < in 57211)) e. other (360%)

 

8. For what reasons do you use the following expressions when you are speaking?

(£3341. B‘Afsflfilcifltwfifiélfihiim )
a. ano(o) b0)( 5)
b. ee(to) iiflk)
C-ma(a) 1(5))
d.nanka tit/oi»
e. ne(e) 23(1)

f. yo J:

Thank you very much for your cOOperation. fillflfl E 5 ‘15?) '9 133‘ k 5 I." 31035 LTC.

278

APPENDIX 3
THE QUESUONNARE FOR NON-NATIVE SPEAKERS OF JAPANESE

INFORMATION SHEET
PLEASE ANSWER THE FOLLOWING QUESTIONS.

 

1. Name: 2. Age:
3. Gender: Male Female

4. Classification: Fr. So. Jr. Sr. Grad. Other:

 

 

 

 

5. Major field of study:
6. Hometown: City: State:
7. How long have you studied Japanese? years months

 

 

8. How long have you lived in Japan, if at all?

 

9. What were your main reasons for living in Japan? study, work, travelling,

other:

 

10. About how many Japanese friends do you have?

 

11. For what reasons do you personally use the following expressions?
a. ano(o) &) U)( 5 ):

b. ee(to) it in 3: ):
c. nee 213(31):
(1. yo .1::

12. What are the things you pay particular attention when speaking Japanese?
a. accuracy b. naturalness of expressions 0. pronunciation/intonation
d. express what you want to express
e. other (Please specify)

 

Thank you very much for your cooperation. E 5 ‘1) $ 0 735‘ 2‘: 5 C‘é'lfldi L713.

279

APPENDIX 4

List of native English speakers for the English data

 

 

 

 

 

 

 

subject sex age occupation home state data topic words
PE M O teacher New York C foreign language 720
CP M O teacher Oregon C philosophy 1157
SS M O businessman Michigan C Americans 791
HE F O housewife Virginia C foreign language 1326
AP F O teacher Michigan C philosophy 838
SH F O counsellor Michigan C everyday life 1047
J F M Y student Michigan C Americans 785
MC M Y student Michigan C college life 655
BH M Y student Michigan C college life 1037
J H F Y student Michigan C USA & the world 1031
HS F Y student Michigan C USA & the world 1169
EP F Y student Michigan C everyday life 303
CB M Y student Michigan N happy moment 523
TD M Y student Michigan N happy moment 599
TR F Y student Michigan N troubling time 264
LV F Y student Michigan N troubling time 525

 

 

 

 

 

 

 

 

O=older; Y=younger; C=conversation; N=narratives

280

 

BIBLIOGRAPHY

281

BIBLIOGRAPHY

ACTFL. 1988. Oral Proficiency Interview: Tester Training Manual. New York: ACTFL,
Inc.

Anderson, Elaine S. 1986. The acquisition of register variation by Anglo-American
children. Language Socialization across cultures, ed. by Banbi B. Schieffelin
and Elinor Ochs, 153-161. Cambridge: Cambridge University Press.

Banrlund, Dean C. 1989.Communicative styles of Japanese and Americans: Images and
realities. Belmont: Wadsworth Publishing Co.

Brown, Gillian and Yule, George. 1983. Discourse analysis. Cambridge: Cambridge
University Press.

Brown, Penelope. 1980. How and why are women more polite: Some evidence from a
Mayan community. Women and language in literature and society, ed. by Sally
McConnell-Ginet, Ruth Borker and Nelly Furrnan, 111-136. New York: Praeger

Publishers.

Chafe, Wallace L. 1982. Integration and involvement in speaking, writing, and oral
literature. Spoken and written language: Exploring orality and literacy, ed. by
Deborah Tannen, 35-54. Norwood: Ablex Publishing Co.

------- 1987. Cognitive constraints on information flow. Coherence and grounding in
discourse, ed. by Russell Tomlin, 21-51. Amsterdam: John Benjamins.

Chaudron, Craig and Jack C. Richards. 1986. The effect of discourse markers on the
comprehension of lectures. Applied Linguistics 7. 113-127.

Christopher, Robert C. 1984. The Japanese mind. London: Pan Publishing Co.

Clancy, Patricia M. 1982. Written and spoken style in Japanese narratives. Spoken and
written Language: exploring orality and literacy, ed. by Deborah Tannen, 55-76.

Norwood, NJ: Ablex Publishing.

------- 1985. The acquisition of Japanese. The cross-linguistic study of language
acquisition, ed. by D. I. Slobin, 373-524. New Jersey: Erlbaum

------- 1986. The acquisition of communicative style in Japanese. Language socialization
across cultures, ed. by Banbi B. Schieffelin and Elinor Ochs, 213-250.

Cambridge: Cambridge University Press.

Cook, Haruko Minegishi. 1988. Sentential particles in Japanese conversation: A study of
indexicality. Ph.D. dissertation, University of Southern California.

-------- 1990. The sentence-final particle ne as a tool for cooperation in Japanese

conversation. J apanese/ Korean Linguistics, ed. by Hajime Hoji, 2944.
Stanford: The Center for the Study of Language and Information.

282

-------- 1993. Function of the filler ano in Japanese. J apanese/ Korean Linguistics, ed.
by S. Choi, 19-38. Stanford: CSLI.

Doi, Takeo. 1971. The anatomy of dependence: The key analysis of Japanese behavior.
Tokyo: Kodansha International.

Du Bois, John, Schuetze-Cobum, Stephan, Paolini, Danae, and Cumming, Susanna.
1992. Discourse transcription. Santa Barbara Papers in Linguistics 4. Santa
Barbara: University of California Press.

Eckert, Penelope. 1989. The whole woman: Sex and gender differences in variation.
Language Variation and Change 1. 245-267.

Gass, Susan M. and Larry Selinker. 1994. Second language acquisition: An introductory
course. New Jersey: Lawrence Erlbaum.

Firth, A. 1988. Communicative strategies of Danish learners and English native-speakers.
Ms. English Language Research, University of Birmingham.

Flowerdew, John and Steve Tauroza. 1995. The effect of discourse markers on second
language lecture comprehension. Studies in Second Language Acquisition 17.
435-458.

GivOn, Talmy. 1982. Logic vs. pragmatics, with human languages as the referee: Toward
an empirically viable epistemology. Joumal of Pragmatics 6. 81-133.

Gradall, D. and Swann, J. 1989. Gender voices. Cambridge: Blackwell.
Gumperz, John J. 1982. Discourse strategies. Cambridge: Cambridge University Press.

Hadley, Alice Omaggio. 1993. Teaching language in context. Boston, MA: Heinle &
Heinle Publishers.

Halliday, M.A.K. and Hasan, Ruquaiya. 1976. Cohesion in English. London: Longman.

Hatch, Evelyn. 1992. Discourse and language education. Cambridge: Cambridge
University Press.

Heritage, John. 1994. Constituting and maintaining activities across sequences: And-
prefacing as a feature of question design. Language in Society 23. 1-29.

Hudson, Mutsuko Endo. 1998. So?: On Japanese connectives sorede, dakara, and ja.
J apanese/ Korean linguistics Volume 7, ed. by Noriko Akatsuka, et. al., 59—75.
Stanford: The Center for the Study of Language and Information.

------- 1997. On Being 'PC': 'Pragmatic Correctrress' in elementary Japanese textbooks.
Progress in Japanese linguistics and pedagogy: A collection in honor of professor
Akira Miura's 70th birthday, ed. by Hubbard, M, T. Sakamoto and J. Davis.
Tokyo: ALC Publishing. 247-262

Ide, Sachiko. 1982. Japanese sociolinguistics: Politeness and women’s language.
Lingua 57. 357-385.

283

Ide, Sachiko. 1997. Women's language, men's language. Broken silence: Voices of
Japanese feminism, ed. by Sandra Buckley, 32-65. Berkley: University of
California Press. '

Ishikawa, Minako. 1991. Hutatsu no apuroochi ni yoru bunmatsu-joshi ne no danwa-
bunseki: Soogo sayooron-teki shakaigengogaku to ihen bunseki no hokanteki
hoohooron [An analysis of sentence-final particle ne by two approaches:
Complimentary methodology with interactional sociolinguistics and variation
analysis]. Southern Review 6. 17-26.

Ito, Rika. 1995. The Functions of the Japanese Conjunction (Sore )de as a discourse
marker. M.A. thesis, Michigan State University.

Iwasaki, Schoichi. 1993. The structure of intonation unit in Japanese. Japanese/ Korean
Linguistics Volume 3, ed. by Soonja Choi, 39-53. Stanford: Center for the Study
of Language and Information.

Kamio, Akio. 1979. On the notion speaker's territory of information: A functional
analysis of certain sentence-final forms in Japanese. Explorations in linguistics:
Papers in honor of Kazuko Inoue, ed. by G. Bedell, E. Kobayashi and M. Muraki,
213-231. Tokyo: Kenkyuusha.

-------- 1990. Joohoo no nawabari-riron [Territory of Information]. Tokyo:
Taishuukan.

-------- 1994. The theory of territory of information: The case of Japanese. Journal of
Pragmatics 21. 67-100.

Kuno, Susumu. 1973. The structure of Japanese language. Cambridge, MA: MIT Press.

Labov, William. 1972. The transformation of experience in narrative syntax. Language in
the inner city. Philadelphia, PA: University of Pennsylvania Press.

------- 1990. The intersection of sex and social class in the course of linguistic change.
Language Variation and Change 2. 205-254.

Lakoff, Robin. 1975. Language and woman’s Place. New York: Harper & Row,
Publishers.

Larsen-Freeman, Diane and Michael H. Long. 1991. An introduction to second
language acquisition research. London: Longman.

Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.

Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge
University Press.

Makino, Seiichi and Tsutsui Michio. 1986. A dictionary of basic Japanese grammar.
Tokyo: The Japan Times.

Martin, Samuel E. 1975. A reference grammar of Japanese. Rudand: Tuttle.
Matsumoto, Kazuko. 1997. Japanese intonation units and syntactic structure. Paper

presented at 1997 LSA meeting.

284

Maynard, Senko Kumiya. 1989a. Japanese conversation: Self-contextualization through
structure and interactional management. Norwood: Ablex Publishing Corporation.

------- l989b. Functions of the discourse marker dakara in Japanese conversation.
Text 9 (4). 389-414.

------- 1993a. Kaiwa bunseki [Conversation Analysis]. Tokyo: Kuroshio
Shuppan.

------- 1993b. Discourse modality: subjectivity, emotion and voice in the Japanese
language. Philadelphia: John Benjamins.

------- 1997. Japanese communication: Language and thought in context. Honolulu:
University of Hawaii Press

McCarthy, Michael. 1991. Discourse analysis for language teachers. Cambridge:
Cambridge University Press.

McGloin, Naomi H. 1991. Sex difference and sentence-interactional particles. Aspects of
Japanese women's language, ed. by S. Ide and N. H. McGloin, 23-41. Tokyo:
Kuroshio Shuppan.

Mine, Huyuki. 1995. Nihongo gakushuu sha no kaiwa ni okeru bunmatsu-hyoogen no
shuutoku katei ni kansuru kenkyuu [Research on the acquisition process of
sentence-final expressions in conversations by students of Japanese]. Nihongo
kyooiku 86, 65-80.

Morita, Yoshiyuki. 1980. Kiso Nihongo l and 2 [Basic Japanese 1 and 2]. Tokyo:
Kadokawa Publishing.

Ochs, Elinor. 1986. Introduction. Language socialization across cultures, ed. by Bambi
B. Schieffelin and E. Ochs, 1-13. Cambridge: Cambridge University Press.

Ohta, Amy Snyder. 1994. Socializing the expression of affect: An overview of affective
particle use in the Japanese as a foreign language classroom. Issues in Applied
Linguistics 5. 303-325.

Oishi, Toshio. 1985. A Description of Japanese interactional particles in context. Ph.D.
dissertation, University of Michigan.

Okamoto, Shigeko. 1992. "Tasteless" Japanese: Less "feminine" speech among young
Japanese women. Gender Articulated: Language and the socially constructed self,
ed. by K. Hall and M. Bocholtz, 297-325. New York: Roudedge.

Okumura, Akiko. 1995. Teachers' and nonteachers' Perception of elementary leamers'
spoken Japanese. The Modern Language Journal 79. 29-39.

Ono, Tsuyoshi and Yoshida, Eri. 1996. A study of co—construction in Japanese: We
don't finish each other's sentences. Japanese/Korean Linguistics, ed. by Noriko
Akatsuka, Shoichi Iwasaki, and Susan Strauss. Stanford: CSLI.

Onodera, Noriko Okada. 1993. Pragmatic change in Japanese: Conjunctions and
interjections as discourse markers. Ph.D. dissertation, Georgetown University.

285

Owen, Marion. 1981. Conversational units and the use of 'Well...'. Conversation and
discourse, ed. by Paul Werth, 99-116. London: Croom Helm.

Philips, Mieko Kimura. 1995. An analysis of Japanese discourse markers. Paper
presented 1995 Association of Asian Studies annual meeting.

Philips, Susan U., Steele, Susan, Tanz, Christine, (eds.) 1987. Language, gender and
sex in comparative perspective. Cambridge: Cambridge University Press.

Pomerantz, Anita. 1984. Agreeing and disagreeing with assessments: some features of
preferred/dispreferred turn shapes. Structures of Social Action, ed. by Maxwell
Atrkinson and John Heritage, 57—101.

Reischauer, Edwin O. 1977. The Japanese. Cambridge, MA: Harvard University Press.

Reynolds-Akiba, Katsue (ed.) 1993. Onna to Nihongo [Women and Language in
Japanese]. Tokyo: Yuushindo.

Sacks, Harvey, Schegloff, Emanuel A., and Jefferson, Gail. 1978. A Simplest
systematics for the organization of turn taking for Conversation, ed. by Jim
Schenkein, 7-55. New York: Academic Press.

Saito, Makiko. 1992. A study of discourse marker Nanka in Japanese. M.A. thesis,
Michigan State University.

Sakurai, Shizuka. 1997. Interactional particles ne and yo in L2 Japanese. A paper
presented at 1997 SLRF annual conference.

Sawyer, Mark. 1992. The development of pragrnatics in Japanese as a second language:
The sentence-fmal particle ne. Pragmatics of Japanese as native and target
language, ed. by G. Kasper, 82-113. Honolulu: University of Hawaii Press.

Schiffrin, Deborah. 1987. Discourse markers. Cambridge: University Press.

-------- 1994. Approaches to discourse. Oxford: Blackwell.

Schourup, Lawrence Clifford. 1982. Common discourse particles in English
conversation. Ph.D. dissertation, Ohio State University.

Shibamoto, Janet S. 1985. Japanese women's language. New York: Academic Press.

-------- 1987. The Womanly woman: Manipulation of stereotypical and nonstereotypical
features of Japanese female speech. Language, gender & sex in comparative
perspective, ed. by Susan U. Philips, Susan Steele & Christine Tanz. Cambridge:
Cambridge University Press.

Shinmura, Izuru. 1969. Kojien dictionary. Tokyo: Iwanarni Shoten.

Singer, Kurt. 1973. Mirror, sword and jewel: A study of Japanese characteristics. New
York: Braziller.

Smith, Janet 8. Women in charge: Politeness and directives in the speech of Japanese
women. Language in Society 21 . 59-82.

286

Smith, Robert J. 1983. Japanese society. Cambridge: Cambridge University Press.
Stubbs, Michael. 1983. Discourse analysis. Chicago: The University of Chicago Press.

-------- 1984. Applied discourse analysis and educational linguistics. Applied
sociolinguistics, ed. by P. Trudgill, 203-244. London: Academic Press.

Takubo, Yukinori. 1995. Onseigengo no gengogakuteki modern 0 mezashite:
onseitaiwakanrihyooshiki o chuushin ni [Toward a linguistic model of speech
performance: Conversation management markers. Joohoo shori 'Inforrnation
Processing], 1020-1026.

Takubo, Yukinori, Kinsui, Satoshi. 1996. Hukusuu no shinteki-ryooiki ni yoru
danwakanri [Discourse management in terms of mental domains]. Ninchi Kagaku
[Cognitive Science]. 59-74.

Tannen, Deborah. 1982. The Oral/literate continuum in discourse. Spoken and written
language, ed. by D. Tannen, 1-16. Norwood: Ablex Publishing Co.

-------- 1984. Conversational style: Analyzing talk among friends. Norwood, NJ: Ablex.

-------- 1989. Talking voices: repetition, dialogue, and imagery in conversational
discourse. Cambridge: Cambridge University Press.

-------- 1994. Gender and discourse. Oxford: Oxford University Press.
-------- (ed.) 1993a. Framing in discourse. Oxford: Oxford University Press.

-------- (ed.) 1993b. Gender and conversational interaction. Oxford: Oxford University
Press.

Traugott, Elizabeth Closs. 1982. From propositional to textual and expressive meanings:
some semantic-pragmatic aspects of grarnmaticalization. Perspectives on historical
linguistics, ed. by Winfred P. Lehmann and Y. Malkiel, 245-171. Philadelphia:

John Benjamins.

Tyler, Andrea. 1992. Discourse structure and the perception of incoherence in
international teaching assistants' spoken discourse. TESOL Quarterly 26. 713-
729.

Tyler, Andrea and John Bro. 1992. Discourse structure in nonnative English discourse:
The effect of ordering and interpretive cues on perceptions of comprehensibility.
Studies in Second Language Acquisition 14. 71-86.

Uyeno, Tazuko Yamanaka. 1971. A study of Japanese modality—A performative
analysis of sentence particles. Ph.D. dissertation, University of Michigan.

Watanabe, Suwako. 1993. Cultural differences in framing: American and Japanese
group discussions. Framing in discourse, ed. by Deborah Tannen. Oxford:
Oxford University Press.

Wetzel, Patricia J. 1988. Are "powerless" communication strategies the Japanese norm?
Language in Society 17. 555-564.

287

Whorf, Benjamin Lee. 1956. Language, Thought, and reality: Selected writings of
Benjamin Lee Whorf. Cambridge, MA: MIT Press.

Wierzbicka, Anna. 1991. Japanese key words and core cultural values. Language in
Society 20. 333-385.

Wootton, A. J. 1981. The management of grantings and rejections by parents in request
sequences. Semiotica 37. 59-89.

288

 

 

 

r 1 !

 

"‘11111111111111“