You are on page 1of 234

Intonation Units in Japanese Conversation

Studies in Language Companion Series (SLCS)

The SLCS series has been established as a companion series to Studies in Language,
International Journal, sponsored by the Foundation “Foundations of Language”.

Series Editors
Werner Abraham Michael Noonan
University of Vienna University of Wisconsin, Milwaukee

Editorial Board
Joan Bybee Ekkehard König
University of New Mexico Free University of Berlin
Ulrike Claudi Christian Lehmann
University of Cologne University of Erfurt
Bernard Comrie Robert Longacre
Max Planck Institute University of Texas, Arlington
For Evolutionary Anthropology, Leipzig
Brian MacWhinney
William Croft Carnegie-Mellon University
University of Manchester
Marianne Mithun
Östen Dahl University of California, Santa Barbara
University of Stockholm
Edith Moravcsik
Gerrit Dimmendaal University of Wisconsin, Milwaukee
University of Leiden
Masayoshi Shibatani
Martin Haspelmath Rice University and Kobe University
Max Planck Institute
Russell Tomlin
For Evolutionary Anthropology, Leipzig
University of Oregon

Volume 65
Intonation Units in Japanese Conversation: Syntactic, Informational, and
Functional structures
by Kazuko Matsumoto
Intonation Units in
Japanese Conversation
Syntactic, informational,
and functional structures

Kazuko Matsumoto
Aichi University of Education, Japan

John Benjamins Publishing Company


Amsterdam/Philadelphia
TM The paper used in this publication meets the minimum requirements
8
of American National Standard for Information Sciences – Permanence
of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data

Matsumoto, Kazuko, 1955-


Intonation units in Japanese conversation : syntactic, informational and
functional structures / Kazuko Matsumoto.
p. cm. (Studies in Language Companion Series, issn 0165–7763 ; v.
65)
Includes bibliographical references and indexes.
1. Japanese language--Intonation. 2. Japanese language--Discourse
analysis. I. Title. II. Series.

Pl544.5 M37 2003


495.6’16-dc21 2002033023
isbn 90 272 3075 7 (Eur.) / 1 58811 364 7 (US) (Hb; alk. paper)

© 2003 – John Benjamins B.V.


No part of this book may be reproduced in any form, by print, photoprint, microfilm, or
any other means, without written permission from the publisher.
John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands
John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
For my teachers, who taught me the beauty of scholarly pursuit
Table of contents

Preface xi
Tables and Figures xiii
Transcription conventions xv
Abbreviations xvii

Chapter 1
Introduction 1
1.1 Information flow 1
1.2 Intonation units 1
1.3 Word order in Japanese spoken discourse 3
1.4 Purpose of the study 5
1.5 General research questions 5
1.6 Organization of the book 5

Chapter 2
Information flow in spoken discourse 7
2.1 Topicality 7
2.1.1 Topic continuity 7
2.1.2 Referential accessibility and thematic importance 8
2.1.3 Referential distance and topic persistence 9
2.2 Grounding 10
2.2.1 Grounding and given-new information 10
2.2.2 Foreground vs. background 11
2.3 Information status 12
2.3.1 Approaches to the notion of givenness/newness 12
2.3.2 Activation cost: Given-accessible-new distinction 15
2.3.3 The expression of activation cost 17
2.3.4 Identifiability, definiteness, and activation cost 18
 Table of contents

2.4 Units of information flow 19


2.4.1 Segmentation of discourse into prosodic units 19
2.4.2 Intonation units: Definition and identification criteria 20
2.4.3 Intonation units, consciousness, and language production 21
2.4.4 Types and size of intonation units 22
2.4.5 Intonation units, accent units, and information structure 23
2.4.6 Intonation units, clauses, and IU linkage 25
2.5 Japanese intonation units 28
2.5.1 Syntactic fragmentation in spoken Japanese 29
2.5.2 Multi-functional nature of Japanese intonation units 33
2.6 Constraints on information flow 38
2.6.1 The function and flow rate of given information 38
2.6.2 Constraints on the flow rate of new information 39

Chapter 3
Method of the study 43
3.1 Research questions and hypotheses 43
3.2 Data base 44
3.3 Data transcription 46
3.4 Intonation units produced in 16 conversations 47
3.5 Analysis of intonation units 48
3.5.1 Data for quantitative analyses of Japanese intonation units 48
3.5.2 Topics of 16 conversational segments 48
3.5.3 Substantive, regulatory, and fragmentary intonation units 49

Chapter 4
Syntactic structure of the intonation unit
in conversational Japanese 51
4.1 Data coding 51
4.2 Distribution of IU syntactic structure types 57
4.2.1 Clausal vs. phrasal intonation units 57
4.2.2 Preferred syntactic structure of the Japanese intonation
unit 60
4.3 Distribution of post-predicate phrases among IU syntactic types 63
4.4 Clausal intonation units: Full clauses vs. semi-clauses 66
Table of contents 

4.5 NP intonation units 68


4.5.1 Distribution of five types of NP intonation units 68
4.5.2 Independent phrasal NP intonation units: Types and
functions 71
4.6 Adjectival, adverbial, and mixed phrasal intonation units 89
4.7 Intonation units and clauses: Single-IU clauses vs. multi-IU
clauses 94
4.8 Summary 98

Chapter 5
Information structure of the intonation unit
in conversational Japanese 101
5.1 Data coding 101
5.2 NPs, intonation units, and clauses 105
5.3 Preferred NP types: Grammatical roles, information statuses,
and syntactic forms 109
5.4 Preferred information structure of the Japanese intonation unit 118
5.4.1 Distribution of IU information structure types 118
5.4.2 The one new NP per IU constraint 121
5.5 Preferred clause structure in conversational Japanese 124
5.5.1 Preferred clause types and preferred argument structure 124
5.5.2 Overt vs. null arguments 132
5.6 Multi-IU clauses and the one new NP per IU constraint 134
5.7 Summary 139

Chapter 6
Functional structure of the intonation unit
in conversational Japanese 141
6.1 Data coding 141
6.2 Distribution of IU functional structure types 144
6.3 Preferred functional structure of the Japanese intonation unit 146
6.3.1 Preferred number of functional components per IU 149
6.3.2 Linear order of functional components within an IU 150
6.4 Multi-IU clauses and multifunctionality 155
6.5 Summary 158
 Table of contents

Chapter 7
Conclusion 159
7.1 Japanese intonation units: Syntactic, informational,
and functional structures 159
7.2 Prospects 164

Notes 169
References 179
Index 191
Preface

This book is based on my doctoral dissertation which was submitted to the


University of California, Los Angeles for partial satisfaction of the requirements
for the degree Doctor of Philosophy in Applied Linguistics (Matsumoto 1996).
The dissertation originally contained a chapter that dealt with linkage between
intonation units. However, its content turned out to exhibit an overlap with
that of the functional structure chapter; therefore, I have deleted the linkage
chapter from this book. Thus the book deals specifically with the structures
of the intonation unit, namely, syntactic, informational, and functional. As a
result, this book no longer shows redundancy, being more coherent than the
dissertation. In addition, the literature review chapter has been shortened so
that it contains only those ideas, concepts, past studies, and so on that are
directly relevant to the theme of the book. Further, the references have been
updated. While these revisions were in progress, other versions of parts of the
present book appeared as journal articles (Matsumoto 1997a, 1998a, 1999a,
2000a, 2000b).
This book grew out of graduate courses and seminars that I took at UCLA
as a doctoral student and interactions I had with professors and students
there. I would like to thank especially Roger Andersen and Shoichi Iwasaki,
co-chairs of my doctoral committee, for providing me with detailed, valuable
comments and suggestions, from which I benefited considerably. I also wish
to thank Marianne Celce-Murcia, Elinor Ochs, John Schumann, Asif Agha,
Robert Kirsner, John Heritage, Emanuel Schegloff, Dominique Sportiche, and
Timothy Stowell for their guidance and encouragement. It is precisely this
interdisciplinary nature of the UCLA graduate program in Applied Linguistics
that contributed to the birth of this book.
Another academic environment out of which this book grew is the Univer-
sity of California, Santa Barbara where I conducted my post-graduate research
as a visiting Fulbright scholar. The Linguistics Department’s friendly and yet
rigidly scholarly atmosphere contributed significantly to the creation of new
ideas. These I have added and discussed in the present book. I am grateful to
Wallace Chafe for his detailed, insightful comments on various earlier versions
 Preface

of the present book, suggestions, and encouragement. This book consists in an


elaboration of Chafe’s theory; thus undoubtedly, he is the person most perti-
nent and most important to the topics discussed here. Special thanks are also
due to Sandra Thompson, Patricia Clancy, and John Du Bois for their helpful
discussions of the issues addressed in this book. Finally, for help and advice
during the publication process of the present book, I thank Michael Noonan,
the co-editor of this series, and Cornelis Vaes of the John Benjamins Publishing
Company in Amsterdam.
Tables and Figures

Tables
3.1. Data for this study
3.2. Major topics of 16 conversations
3.3. Number of IUs produced in 16 conversations
3.4. Topics of 16 conversational segments
3.5. Number of four types of IUs in 16 conversational segments
4.1. Distribution of IU syntactic structure types
4.2. Distribution of eight clausal IU types
4.3. Distribution of 12 phrasal IU types
4.4. Distribution of post-predicate phrases among IU syntactic types
4.5. Proportion of IUs involving postposing by IU syntactic type
4.6. Proportion of full clauses vs. semi-clauses
4.7. Distribution of five types of NP IUs
4.8. Distribution of five types of independent phrasal NP IUs
4.9. Frequency and features of four types of stray NP IUs
4.10. Distribution of adjectival, adverbial, and mixed phrasal IU types
4.11. Distribution of six types of independent adverbial IUs
4.12. Average number of IUs per clause
4.13. Frequency of the number of IUs per clause
4.14. Summary of results of IU syntactic structure analysis
5.1. Overt NPs, IUs, and clauses
5.2. Average number of overt NPs per IU by IU syntactic type
5.3. Distribution of grammatical roles among IU syntactic types
5.4. Grammatical role and information status
5.5. Grammatical role and NP syntactic form
5.6. Information status and NP syntactic form
5.7. Summary of results of analysis of overt NPs
5.8. Distribution of IU information structure types
5.9. Two types of IUs with zero NPs
5.10. Types of IUs with two new NPs
 Tables and Figures

5.11. Frequency of clause types by the number of overt arguments


5.12. Frequency of clauses with overt S, A, and O arguments
5.13. Frequency of clauses with zero, one, and two overt arguments
5.14. Proportion of overt arguments by clause type
5.15. Proportion of IUs/clauses with more than one new NP
6.1. Distribution of IU functional structure types
6.2. Frequency of the number of functional components per IU

Figures
4.1. Distribution of four major IU syntactic types
4.2. Distribution of six IU syntactic types
4.3. Distribution of 20 IU syntactic types
4.4. Proportion of full clauses vs. semi-clauses
4.5. Distribution of five types of NP IUs
4.6. Distribution of five types of independent phrasal NP IUs
4.7. Frequency of the number of IUs per clause
5.1. Distribution of six grammatical roles
5.2. Proportion of three information statuses within grammatical roles
5.3. Distribution of eight NP syntactic forms
5.4. Distribution of IU information structure types
5.5. Frequency of five clause types
5.6. Proportion of clauses with zero, one, and two arguments
5.7. Frequency of clauses with zero, overt S, A, and O arguments
5.8. Proportion of overt vs. null arguments
5.9. Proportion of IUs/clauses with more than one new NP
6.1. Distribution of IU functional structure types
Transcription conventions

Segmentation of Discourse into Intonation Units (IUs)


A carriage return is used to indicate the end of an IU. This means that each
IU appears on a separate line.
-> The two symbols come in pairs to indicate that the IU continues
--> across two lines; the hyphen plus greater-than symbol (->) marks
the beginning of the first line, and a double hyphen plus greater-
than sign (-->) signals the beginning of the second line.
Characteristics of Speech Delivery
wor- A hyphen after a word or part of a word signals a cut-off or self-
interruption; it signals a truncated or uncompleted word.
WOrd Upper case indicates loud talk raised in pitch or volume; the louder,
the more upper case.
wo::d Colons mark the prolongation of the preceding sound; the more
colons, the greater the sound stretching.
˚ ˚ The degree signs indicate that the talk between them was markedly
softer or quieter than the surrounding talk.
< The less-than sign marks that the immediately following talk started
with a rush.
> < The combination of more-than and less-than symbols indicates that
the talk between them is compressed or rushed.
< > The use of less-than and more-than symbols in the reverse order
indicates that the talk between them is markedly slowed or drawn
out.
 Transcription conventions

Sequential Relationships
= Equal signs come in pairs and mark latching; the talk connected by
equal signs is continuous with no discernible pause between them;
used only for an inter-speaker (not intra-speaker) transition.
[ ] Brackets mark overlapping or simultaneous talk; left brackets in-
dicate the point of onset; right brackets indicate the point of
resolution.
Pauses
+ very short micropause which is barely noticeable
(0.1–0.2 seconds of silence)
++ medium-length pause which is noticeable
(0.3–0.6 seconds of silence)
+++ long pause
(0.7–0.9 seconds of silence)
(2.0) Numbers in parentheses indicate elapsed extra-long silence in tenths
of a second (silence of longer than 1.0 second).
Intonation Contours
. falling
, continuing
? rising
^ rise-fall
∼ rise-fall-rise
Listener Backchannels
@ listener backchannels which show affirmative response to the
speaker’s utterance
$ listener backchannels which show wonder, awareness, or surprise
# laughter as listener backchannels; the more symbols, the more
laughters.
Other Symbols
(word) Parentheses surrounding a word or words indicate uncertainty
about the transcription; empty parentheses mark that no hearing
could be achieved.
(( )) Double parentheses mark the transcriber’s descriptions of events or
comments.
Abbreviations

ABL ablative NEG negative


ACC accusative NML nominalizer
ALL allative NOM nominative
CAU causative PASS passive
COM comitative PAST past
COMP comparative PF pause filler
CON conditional PL plural
COP copula POL polite form
DAT dative POT potential
DES desiderative PROG progressive
EMP emphatic PRS presumptive
FP final particle Q question marker
GEN genitive QT quotative
HON honorific STA stative
IMP imperative SOF softener
INS instrumental TAG tag
LOC locative TOP topic marker
Chapter 1

Introduction

This chapter provides a broad overview of this book. It first presents a brief
overview of prior research on information flow, intonation units, and word
order in Japanese, upon which the present study is based. It then states the
purpose of this research and the general research questions addressed in this
study. Finally, it describes the organization of this book.

. Information flow

Information flow refers to “the interactionally determined choices that speak-


ers make which determine intonational, grammatical, and lexical choices”
(Fox & Thompson 1990: 297) in the process of communicating with their
interlocutors. Aspects of information flow which have been discussed in previ-
ous studies include information status (e.g. Birner & Ward 1998; Chafe 1976,
1987, 1994; Du Bois 1987; Halliday 1985, 1994; Lambrecht 1994; Prince 1981,
1992), topic continuity (e.g. Givón 1983a, 1984, 1990), grounding (e.g. Givón
1990; Hopper 1979; Tomlin 1985, 1987), and definiteness (e.g. Chafe 1972,
1976, 1994; Du Bois 1980; Hawkins 1978). The discourse-level explanations
that these studies provided have shown that the information-flow factors, both
cognitive and interactional, play a significant role in determining the grammar
of syntactic constructions that native speakers use in spoken communication
and the distributional patterns of those constructions.

. Intonation units

Past research has noted the intermittent nature of spontaneous spoken dis-
course. That is, spontaneous speech has the property of being produced in a
series of brief spurts. These spurts of language, or the coherent chunks into
which speakers segment talk, have been considered the basic units of informa-
tion flow. These units have been given different names by different discourse
 Chapter 1

researchers: for example, “tone units” (Crystal 1975), “information blocks”


(Grimes 1975), “information units” (Halliday 1985), and “intonation groups”
(Cruttenden 1986).
More recently, Chafe (1987) has given them the name “intonation units”
(IUs). An IU, roughly speaking, is a sequence of words uttered under a
single coherent intonation contour, usually preceded by a brief pause. For
Chafe (1980b: 48), IUs are “linguistic expressions of focuses of consciousness,
whose properties apparently belong to our built-in information-processing
capabilities”. IUs are not acoustically measured units but perceptual, auditory
units which are comparable to Crystal’s (1969) “tone units” and Cruttenden’s
(1986) “intonation groups”. Concerning the relationship between auditorily
based IUs and acoustic declination units, previous research has evidenced an
almost perfect correspondence between the boundaries of these two types of
discourse units (see Schuetze-Coburn et al. 1991). Thus, in the present study,
I consider the IU as a valid, well-established auditory unit.1
Chafe (1987) hypothesizes that a concept may be in any one of three
different states of activation at a particular time: “active”, “semi-active”, or
“inactive”. His model assumes that the effect of the speaker’s utterance of an
IU on the hearer is to activate all the concepts it contains while deactivating
others, and to bring about changes in the activation states of information in
the hearer’s mind. Chafe (1987) proposes that only one “previously inactive”, or
“new” concept can be changed to the “active” state within a single IU, which he
calls the “one new concept at a time constraint”. This means that a single IU can
express no more than one new concept. This constraint on new information
quantity per unit, Chafe (1987: 32) argues, results naturally from “the cognitive
basis of an IU: the expression of a single focus of consciousness”. Other
discourse researchers have also proposed similar constraints on the amount
of new information that can be transferred within a unit of information flow:
for example, Givon’s (1975) “one [new] unit per proposition” constraint and
Du Bois’ (1987) “One New Argument Constraint”.
Chafe (1987) further notes that in spoken English the majority of IUs
take the form of clauses, that is, the clause is the prototypical IU type, and
thus clause linkage is the predominant type of IU linkage. Consistent with
his observation, other discourse researchers have also treated the clause as the
basic unit of discourse production and information flow in human spoken
discourse. Pawley and Syder (1983), for example, have proposed that humans
can encode or formulate only the contents of “one clause at a time”, arguing
that the clause is the basic unit for information processing and segmentation
in spontaneous connected discourse. Likewise, Givón (1990: 896) has identified
Introduction 

the clause corresponding to “the mental proposition that stands for some state
or event” as the basic unit of information storage in coherent discourse.
Iwasaki and Tao (1993), in their comparative study of the syntactic struc-
ture of IUs in English and Japanese, have shown, however, that while En-
glish IUs are mostly clausal, Japanese IUs tend to be non-clausal, or phrasal.
This suggests that Chafe’s (1987) “clause centrality proposal” may not be war-
ranted cross-linguistically. On the other hand, their findings are consistent
with Clancy’s (1982) and Maynard’s (1989) observation that spoken Japanese
is “highly fragmented”, that is, a syntactic clause is frequently broken up into
a number of units smaller than those found in spoken English. Moreover,
Japanese IUs, according to Iwasaki (1993), have very orderly functional struc-
ture, and a built-in mechanism which allows speakers to additionally convey
cohesive and interactional information while at the same time accomplishing
the primary task of communicating ideational information.
Previous discourse-functional research on IUs as basic prosodic units of
information flow has mostly dealt with spoken narratives (e.g. Chafe 1980a;
Clancy 1980, 1982; Croft 1995; Du Bois 1987). However, an apparent
shift from narrative-focused to conversation-centered research can be seen
in more recent studies (e.g. Clancy et al. 1996; Couper-Kuhlen & Selting
1996; Ochs et al. 1996; Ono & Thompson 1995, 1996). Importantly, many
of the recent IU-based functional studies are cross-linguistic in nature,
searching for language-independent or language-specific constraints operat-
ing on the syntactic, informational, or functional structure of the IU (e.g.
Ashby & Bentivoglio 1993; Chafe 1994; Du Bois 1987, 2003; Du Bois et al.
2003; Helasvuo 2001; Iwasaki 1993; Iwasaki & Tao 1993; Kärkkäinen 1996;
Scheibman 2002; Schuetze-Coburn 1994; Smith 1996; Tao 1996; Thompson
1997; Thompson & Hopper 2001). Past research in Japanese IUs has already
dealt with their syntactic and functional structures in comparison to their
English counterparts. However, their information structure and the possible
relation between the structures—syntactic, functional, and informational—
have not been investigated to date.

. Word order in Japanese spoken discourse

It is generally agreed among linguists that typologically, Japanese is an


SOV language (see Hoji 1985; Iwasaki 2002; Kuno 1973, 1978c; Saito 1985;
Shibatani 1990). That is, the basic, syntactically defined word order in Japanese
is SOV (Subject-Object-Verb), as exemplified in (1.1). Although the SOV order
 Chapter 1

is undoubtedly basic, given its predominant frequency in naturally occurring


discourse and native speakers’ intuitions,2 Japanese allows fronting of an ob-
ject NP to a pre-subject position. This phenomenon, known as “scrambling”,
is shown in (1.2). Examples (1.1) and (1.2) are from Kuno (1973: 3 and 4),
respectively.
(1.1) John ga Mary o butta.
nom acc hit-past
‘John hit Mary.’
(1.2) Mary o John ga butta.
acc nom hit-past
‘Mary, John hit.’

Although scrambling allows reordering of preverbal elements, it is generally


accepted that the strict verb-final requirement must be met in Japanese.
That is, constituents may not be scrambled into a postverbal position. Thus,
sentences like John ga butta Mary o. ‘John hit, Mary.’ (SVO order) and butta
John ga Mary o. ‘hit, John Mary.’ (VSO order) (Kuno 1973: 4) are considered
unacceptable in Japanese.
Japanese spontaneous spoken discourse, however, does exhibit construc-
tions which apparently violate the verb-final requirement. That is, speakers do
utter constructions in which an element or elements appear after the verb. This
is exemplified in (1.3), which is from Matsumoto (1995a: 239).
(1.3) nihon de KAngaerarenai ne sonna koto:.
Japan loc think-pot-pass-neg fp such thing
‘(is) unthinkable in Japan, such a thing.’

The placement of constituents in the post-predicate position, the phenomenon


often referred to as “postposing”,3 constitutes, together with scrambling,
marked word order constructions4 in Japanese. The postposing phenomenon
is particularly relevant to the present study of Japanese IUs in that it inter-
acts significantly with intonation. More specifically, it has been observed that
some of the postposing constructions involve an intonational break between
the verb and the postverbal element (thus constituting two IUs), while oth-
ers do not (thus constituting one IU), as in (1.3) (Clancy 1982; Hinds 1976;
Matsumoto 1995a, 1997c, 2000c; Ono & Suzuki 1992; Shibatani 1990). In sum,
Japanese spoken discourse consists not only of basic canonical SOV word order
constructions involving pre-predicate elements, but also of marked word order
constructions involving post-predicate elements.5
Introduction 

. Purpose of the study

The purpose of this book is to explore the syntactic, informational, and func-
tional structures of IUs as basic units of discourse production and information
flow in naturally occurring Japanese conversations. Put differently, this study
attempts to examine, in terms of information flow, not only types of IUs that
speakers of Japanese use in dialogic conversational interactions, but also pat-
terns in their production of IUs. In its analysis of the syntactic and informa-
tion structures of the Japanese IU, the present study focuses on nominal ref-
erences and patterns in the speakers’ production of those nominals. The study
tries to elucidate, above all, the preferred nominal structure, where nominals
include both arguments and non-arguments. In its investigation of the func-
tional structure of the IU, the study focuses on the linear order of functional
components within an IU as well as interrelationships among those compo-
nents. Overall, in exploring the syntactic, informational, and functional struc-
tures of the IU in conversational Japanese, this study is interested not only in
the preferred IU structures which are typical of the way Japanese speakers talk
in connected discourse, but in possible relationships between the structures
and their implications as well.

. General research questions

The general research questions (RQs) this study addresses are the following:
RQ 1: What is the preferred syntactic structure of the IU in conversational
Japanese?
RQ 2: What is the preferred information structure of the IU in conversational
Japanese?
RQ 3: What is the preferred functional structure of the IU in conversational
Japanese?

. Organization of the book

This book consists of seven chapters. Chapter 2 presents background infor-


mation relevant to the present study by reviewing past functional linguistic
research on information flow in spoken discourse. Chapter 3 describes the data
used in this study and the methodological procedures that were followed in
 Chapter 1

analyzing it. Chapters 4, 5, and 6 present and discuss the results of the study.
That is, Chapters 4, 5, and 6, responding to RQ1, RQ2, and RQ3, respectively,
present and discuss the results of the analysis of the syntactic structure, in-
formation structure, and functional structure of the Japanese IU, respectively.
Finally, Chapter 7 concludes the study with a summary of the results presented
in Chapters 4 through 6 and discusses implications of this study.
Chapter 2

Information flow in spoken discourse

This chapter reviews previous research on aspects of information flow in


spoken discourse. These include topicality, grounding, information status,
units of information flow, and constraints on information flow. The discussion
will focus on Chafe’s information-flow research on intonation units, and in so
doing, discourse-pragmatic concepts and terms relevant to the present study
are presented and defined.

. Topicality

.. Topic continuity

Topic continuity refers to the recurrence of topics/referents that contributes to


thematic coherence across a multi-clause chain in human connected discourse.
“Coherent discourse is thus characterized by equi-topic clause-chains” (Givón
1990: 902). This means that in cohesive units of discourse, topic continuity is
expected as the unmarked case. Topic change or discontinuity, by contrast,
is disruptive, surprising, or hard to process; therefore, more coding material
must be assigned to new/discontinuous topics in order to clearly identify
their referents (Givón 1983b: 18). Givón’s (1983b: 17) cross-linguistic scale for
coding topic accessibility is shown in (2.1).
(2.1) most continuous/accessible topic
zero anaphora
unstressed/bound pronouns
stressed/independent pronouns
right-dislocated definite NPs
neutral-ordered definite NPs
left-dislocated definite NPs
Y-moved NPs
cleft/focus constructions
referential indefinite NPs
most discontinuous/inaccessible topic
 Chapter 2

This scale identifies zero anaphora as the most continuous/accessible topic and
referential indefinite NPs as the most discontinuous/inaccessible topic.1 The
criteria involved here concern how accessible or predictable the topic is to the
hearer, given distance from prior mention in discourse, referential interference
from other referents, and so on. According to Givón (1990: 897), then, “more
coherent discourse, with continuous or recurrent sub-elements, is organized in
a way that makes the information mentally more accessible to the hearer”.

.. Referential accessibility and thematic importance

Givón (1983a, 1990) discusses two discourse-pragmatic aspects of topicality,


which he calls “referential accessibility” and “thematic importance”. Referen-
tial accessibility of nominal topics, which primarily derives from the textually
shared anaphoric, or preceding discourse context, is a composite product of
such factors as “referential continuity” (which concerns itself with referential
gap between the current and last occurrence of a referent) and “referential com-
plexity” (which has to do with referential competition between semantically
similar referents in the immediately preceding context). Thematic importance,
by contrast, is concerned with the degree of participation of a referent within
the cataphoric, or subsequent discourse context. From the hearer’s perspective,
then, “referential accessibility pertains to the ‘cognitive search’ for the nominal
referent in mental storage space; thematic importance pertains to the ‘cogni-
tive activation’ of important topics, and non-activation of unimportant ones”
(Givón 1990: 903). This reflects Givón’s general view of grammar as mental
processing instructions to the hearer (see Givón 1990: Chapter 20).
Givón’s (1990) quantitative measures of referential accessibility are of
three kinds: (a) referential distance (RD) (i.e. the number of clauses from
the last occurrence in the preceding discourse), (b) switch reference (SR)
(i.e. whether the preceding clause has the referent as an argument or not),
and (c) potential interference (PI) (i.e. the number of semantically compatible
referents within the preceding 1–2 clauses). Of these, RD is the measure
of referential continuity, while SR and PI are the measures of referential
complexity. Givón’s quantitative measures of thematic importance are of two
kinds: (a) topic persistence (TP) (i.e. the number of times the referent persists
as argument in the subsequent 10 clauses following the current clause), and
(b) overall frequency (OF) (i.e. the total number of times the same referent
appears as clausal argument in the discourse). TP and OF are the text-
frequency measures of local and global thematic importance, respectively.
Information flow in spoken discourse 

Among the five text-based measures of topicality, it is RD and TP that have


most often been employed in previous discourse studies.

.. Referential distance and topic persistence

Givón’s RD, Chafe (1994) claims, is a rough reflection of what he calls “ac-
tivation cost”—the given-accessible-new distinction (see Section 2.3.2). That
is, the “topicality” measured by RD is largely equatable with activation cost:
(a) zero anaphora and unstressed pronouns unambiguously express givenness;
(b) stressed pronouns express contrastive given or accessible referents; and
(c) left-dislocated NPs may verbalize accessible or new referents. Chafe further
remarks that Givón’s RD, which limits its “look-back” to 20 clauses, does not
distinguish accessibility from newness in many cases. For Chafe, the finding of
Givón (1990) that definite nouns showed a scattered RD distribution supports
his own view that definiteness, or identifiability is independent of activation
cost (cf. Section 2.3.4).
Givón (1990) makes two important observations in his quantitative studies
of topicality. One is the lower RD and higher TP of the subject over the object,
and the lower RD and higher TP of the direct object over the indirect object,
which has led him to state that “the subject is consistently more topical than
the direct object, and the direct object more topical than the indirect object”
(Givón 1990: 901). Chafe (1994: 184) interprets this as saying that “subjects are
most often given and of primary importance, and direct objects rank somewhat
lower on the scales of activation cost and importance, and all other roles rank
lower still”. Another important observation is the predominance of anaphoric
pronouns over full NPs in oral discourse. This reflects Givón’s (1990: 918) view
that continued activation of the current active referent is the norm in coherent
multi-propositional discourse.
Givón’s topic continuity-based account discussed above is far from al-
mighty, in my view. First and foremost, we should question whether the theory
has the capacity to explain differnt types of anaphora that can occur in natural
spoken discourse. This is especially important when we work with so-called
“null argument” languages such as Japanese and Mandarin, which allow use of
abundant zero-form NPs. If part of the zero-marking argument NPs in these
languages are actually independent of anaphoric continuity processes, as Tao
(1996) has convincingly shown, this means that the topic continuity theory
cannot be applied to these “non-anaphoric zeros” (see Sections 5.4–5.5). In
other words, the theory seems not entirely effective in accounting for all types
of anaphora occurring in spoken discourse. We should also note that, as Givón
 Chapter 2

(1990) himself points out, topic continuity, or the recurrence of the same
referent is not the only required element of discoursal coherence. In future
discourse studies, more attention should be paid to other types of continuity,
which include temporal continuity and continuity of location and action.

. Grounding

The notion of grounding has been used in two different, although not totally
unrelated, ways by discourse researchers. One is represented by Fox and
Thompson’s (1990) and Givón’s (1990) view of grounding which has to do
with the distinction and interaction between given and new information. The
other is represented by Hopper’s (1979) and Tomlin’s (1985) view of grounding
which is concerned with the foreground-background distinction/continuum.
More detailed discussion will follow.

.. Grounding and given-new information

Grounding is the principal way in which speakers make referents carrying new
information “relevant” for listeners. That is, “to ground a noun phrase is to
locate its referent in conversational space, to make its referent relevant for
the hearer by relating it to a given referent already established in the prior
discourse. Grounding a noun phrase is a way of warranting its introduction
at the point where it is mentioned” (Fox & Thompson 1990: 303). Fox and
Thompson (1990), with reference to relative clause constructions in English
conversations, discuss three types of grounding. These include what they call
“main-clause grounding”, where the main clause situates a newly introduced
head NP by relating it to a given referent in its own clause. For Fox and
Thompson (1990: 301), then, “grounding is essentially a background task,
as opposed to asserting . . . That is, a grounding [relative] clause does not
assert . . . , but merely locates the referent in conversational space”.
A similar view of grounding is taken by Givón (1990). He states that
the function of given, redundant, or topical information in the clause is to
“ground” the new, asserted information to the already stored given informa-
tion, that is, to integrate or “address” the new information onto the appro-
priate location within the storage space in the episodic memory. Therefore,
in communicating with their interlocutors, speakers include given, shared, or
presupposed information in their clauses as a “background” for asserted, new
information, which, for Givón (1987), constitutes “foreground”. The differen-
Information flow in spoken discourse 

tiation manifests itself in the grammar of nominal referents, as in the use of


definite articles and zero anaphora as primarily backgrounding devices, and
indefinite articles as primarily foregrounding devices.
Importantly, discourse grounding is dynamic, not static. While a text is
being co-produced by conversational interactants, a proposition that is as-
serted, or foregrounded at point n in the discourse will become a shared, back-
grounded presupposition at point n plus 1 (Givón 1987). A discourse can thus
be seen as consisting of such dynamic foreground-background alternations.
It should also be noted that the determination of what to foreground and
what to background at a particular point in the discourse consists solely in the
speaker’s communicative goal (i.e. what the speaker wishes to communicate to
the hearer), not dictated by the textual environment within the discourse.

.. Foreground vs. background

The dual notion of “foreground” and “background” has been developed by


Longacre (1976), Hopper (1979), Jones and Jones (1979), Tomlin (1985),
and others. Briefly, foreground information is information which is more
important, central, or crucial to the development of the overall discourse
theme. Background information is information which serves to elaborate,
explicate, or enrich foreground information. Given that the foreground refers
to propositions which the speaker has chosen to make more relevant to the
hearer, or has lent more importance to than other propositions, a sequence
of temporally ordered event clauses in a narrarive—Labov’s (1972) “narrative
skeleton”—have been treated as the foreground. These portions are considered
to crucially contribute to the speaker’s communicative goal. The background,
in contrast, has been argued to consist of less central material—“free clauses”
which are not constrained by any “temporal juncture” (Labov 1972) and thus
are outside the sequence of action continuity.
Tomlin (1985) rejects the binary notion of grounding and proposes instead
the foreground-background continuum. Grounding is seen here as “a process
which ranks propositions according to importance with respect to some par-
ticular rhetorical purpose and its associated theme” (Tomlin 1985: 119). The
continuum thus places propositions describing events which are more cru-
cial to the developing theme higher on that continuum. Although intended
to be a genre-independent definition, according to Tomlin, his characteriza-
tion of the foreground-background information seems to be useful primarily
(or even only) in analyses of highly structured narrative discourse, not in anal-
yses of broader conversational interaction. Clearly, more research should be
 Chapter 2

directed toward the development of precise, operational definitions of the fore-


ground and background which will apply more broadly to any genre of spoken
discourse.
It has been shown that speakers of the world’s languages use various
linguistic devices for signaling the foreground-background distinction. Such
linguistic markers include tense-aspect morphology,2 word order, and voice
(Hopper 1979). Moreover, in any language, foregrounded clauses commonly
tend to involve the “unmarked” information structure, namely, “given followed
by new”. That is, the subject, usually being the central character in the dis-
course, is presupposed; and new events tend to be introduced in the predi-
cate (cf. Section 2.4.5). In backgrounded clauses, by contrast, new informa-
tion often resides in parts other than the predicate, for example, in the subject
(Hopper 1979: 220).
Hopper and Thompson (1980), in their discussion of transitivity,3 argue
that foregrounding in English is determined by a cluster of properties, not a
single morphosyntactic feature, and that this cluster of properties is precisely
that which characterizes high transitivity—which includes “action”, “telic”,
“punctual”, “volitional”, “affirmative”, and “realis”. That is, “the likelihood that
a clause will receive a foreground interpretation is proportional to the height
of that clause on the scale of transitivity. From the performer’s [i.e. speaker’s]
viewpoint, the decision to foreground a clause will be reflected in the deci-
sion to encode more transitivity features in the clause” (Hopper & Thompson
1980: 284). In short, the proposed hypothesis is that high transitivity correlates
with foregrounding, and low transitivity, with backgrounding.4 It is important
to note that their view of transitivity accords with the foreground-background
continuum discussed above. To the extent that the proposal has been formu-
lated based solely on narrative data, more research is also needed to determine
whether or not the hypothesis will apply to conversational data as well.

. Information status

.. Approaches to the notion of givenness/newness

The notion of given vs. new information has been given different definitions by
different linguists. Typically, the given-new distinction has been characterized
in terms of predictability/recoverability, knowledge shared by the speaker and
the hearer, and saliency (Prince 1981). First, givenness/newness in the sense
of predictability/recoverability is represented by Halliday’s (1967b, 1985) and
Information flow in spoken discourse 

Halliday and Hasan’s (1976) “given-new” information, and Kuno’s (1978b)


“old-new” information. This notion has to do with the predictability or re-
coverability from the context of a particular linguistic item in a sentence.
That is, under this rubric, an element that is treated by the speaker as pre-
dictable/recoverable to the listener anaphorically or situationally represents
“given/old” information; an element that is presented by the speaker as be-
ing not predictable/recoverable to the listener, “new” information. Givón’s
(1990: 897) characterization of the “old-new” distinction likewise falls under
this category: “by ‘old’ one means ‘assumed by the speaker to be accessible to
the hearer’, and by ‘new’, ‘assumed by the speaker inaccessible to the hearer’ ”. It
should be noted that this approach recognizes that the notion of given-new is
listener-oriented.
Second, the notion of givenness/newness in the sense of shared knowledge
is represented by Clark and Haviland’s (1977: 4) definition of “given-new”
information: “given” is “information the speaker believes the listener already
knows and accepts as true”, and “new” is “information the speaker believes
the listener does not yet know”. Underlying their notion of the given-new
distinction is the “given-new contract” agreed to by the speaker and the listener,
which was seen as one aspect of Grice’s (1975) “cooperative principle”. Their
listener-oriented view is best represented by their recognition that “the speaker
tries, to the best of his ability, to make the structure of his utterance congruent
with his knowledge of the listener’s mental world” (Clark & Haviland 1977: 4).
Third, givenness/newness in the sense of saliency is represented by Chafe’s
(1974) notion of “given-new” information, in which the binary partition
of information statuses into “given” and “new” is related to the notion of
consciousness. More specifically, Chafe (1976: 30) defines “given” information
as “that knowledge which the speaker assumes to be in the consciousness of
the addressee at the time of the utterance”, and “new” information as “what
the speaker assumes he[/she] is introducing into the addressee’s consciousness
by what he[/she] says.” Put differently, as Fox and Thompson (1990: 299–
300) redefine them, “given” is a status given to “a referent presumed to
be in the hearer’s focal consciousness”, and “new” is a status given to “a
referent introduced into the discourse, presumed not to be in the hearer’s focal
consciousness”. Thus, for Chafe, givenness is a status established by the speaker,
and this notion is fundamentally a matter of the speaker’s belief that the
concept is saliently present in the hearer’s consciousness. We should note that
the three approaches to givenness/newness discussed above are not mutually
exclusive, to the extent that the concepts of predictability/recoverability, shared
knowledge, and saliency are not mutually independent. Moreover, they share
 Chapter 2

two important properties: their overt concern with both the speaker and
the listener (i.e. “speaker-selectedness” and “listener-orientedness”), and their
two-way division of information statuses into given/old and new.
The binary given-new distinction, however, has proved inadequate in
more recent discourse studies (cf. Birner & Ward 1998). Apparently, given-
ness/newness should be treated as a matter of degree, not as an all-or-nothing
phenomenon (Chafe 1976). Prince (1981) proposes a taxonomy of what she
terms “assumed familiarity”, which reflects the fact that language users must
actually operate on the basis of what they assume to be familiar to their in-
terlocutors. Her taxonomy of information status includes under the heading
“assumed familiarity” three categories: (a) new, (b) inferrable, and (c) evoked.
Prince defines an entity which the speaker first introduces into the discourse as
“new”. A “new” discourse entity falls into one of two categories: it is “brand-
new” when “the hearer may have had to create a new entity”; it is “unused”
when “the hearer may be assumed to have a corresponding entity in his/her
own model and simply has to replace it in (or copy it into) the discourse-
model” (Prince 1981: 235). “If some NP is uttered whose entity is already in
the discourse-model, it represents an ‘evoked’ entity” (where “evoked” is used
as an equivalent for “given”). “A discourse entity is ‘inferrable’ if the speaker
assumes the hearer can infer it, via logical—or, more commonly, plausible—
reasoning, from discourse entities already evoked or from other inferrables”
(Prince 1981: 236).
In Prince (1992: 309), information statuses of discourse entities are classi-
fied, in terms of two distinct divisions, discourse-old/-new and hearer-old/-new,
into four types. They are (a) discourse-new, hearer-new (i.e. brand-new);
(b) discourse-new, hearer-old, (i.e. unused); (c) discourse-old, hearer-old
(i.e. evoked); and (d) discourse-old, hearer-new (this type presumably does
not occur in natural discourse). Clearly, one of the advantages of this distinc-
tion is that it captures the fact that while what is old/given in the discourse will
be familiar to the hearer as well, what is assumed by the speaker to be new to
the discourse may not be new to the hearer. Chafe (1994) notes that Prince’s
(1981) use of “brand-new” vs. “unused” is nearly identical to his use of “un-
shared” vs. “shared”. However, he emphasizes that sharedness, which is a com-
ponent of identifiability, is independent of activation cost (see Section 2.3.2).
I agree with Chafe’s (1994: 175) statement that “whether or not a referent is
assumed to be newly activated in the listener’s consciousness is a different ques-
tion from whether or not it is assumed to be already part of the listener’s knowl-
edge. . . . As a term, therefore, ‘unused’ has the disadvantage of conflating the
separate domains of sharedness and activation cost”.
Information flow in spoken discourse 

Having summarized the major approaches to givenness/newness, I will


focus, in what follows, on Chafe’s consciousness-based approach to the notion
of information status. The discussion will center on what Chafe labels “activa-
tion states” and “activation cost” in terms of speakers’ production of intonation
units (IUs).

.. Activation cost: Given-accessible-new distinction

Chafe (1987), assuming that givenness and newness are partial manifestations
of basic cognitive processes, introduces the notion of “accessible” information
as an intermediate type that exists between “given” and “new”. He thus pro-
poses a three-way breakdown into given, accessible, and new information as
opposed to the simple binary distinction of given vs. new.
Chafe (1987) hypothesizes that a particular concept, at a particular time,
may be in any one of three different states of activation. These he terms “active”,
“semi-active”, and “inactive”. “An active concept is one that is currently lit up,
a concept in a person’s focus of consciousness. A semi-active concept is one
that is in a person’s peripheral consciousness, a concept of which a person
has background awareness, but which is not being directly focused on. An
inactive concept is one which is currently in a person’s long-term memory,
neither focally nor peripherally active” (Chafe 1987: 25). It is assumed in
his model that the speaker’s utterance of an IU functions to activate all the
concepts it contains for the hearer while deactivating others, and to bring about
changes in the activation states of information in the hearer’s mind. That is,
the production of an IU, it is assumed, involves changes in activation states,
which take place first in the speaker’s mind during the initially occurring pause
and then in the hearer’s mind during the following period of vocalization
(cf. Section 2.4.2). According to Chafe (1987), then, “given” concepts are
those that were “already active” for the speaker prior to uttering an IU, and
which the speaker assumed to be already active in the mind of the hearer
as well.5 “Accessible” or “previously semi-active”—which could be equated
with Prince’s (1981) “inferrable”—concepts are those that the speaker, before
uttering an IU, transferred from the semi-active to the active state in his/her
own mind.6 “New” or “previously inactive” concepts are those that the speaker,
before uttering an IU, transferred from the inactive to the active state.
The temporal aspects of givenness, accessibility, and newness in relation to
the speaker’s and the listener’s cognitive processes can be visualized as shown
in (2.2), which is from Chafe (1994: 74).
 Chapter 2

(2.2) t1 t2 t3
word1 word2 word3 .....
pause IU IU
onset onset completion

given
speaker- active active
oriented accessible
semi-active
new
inactive

given
listener- active active
oriented accessible
semi-active
new
inactive
At t1, the onset of the pause, a particular idea is active, semi-active, or
inactive; and at t2, the onset of the IU, this idea is now active. If it was
already active at t1, it is “given” information at t2. If it was semi-active at
t1, it is “accessible” information at t2. If it was inactive at t1, it is “new”
information at t2. Concerning what Chafe (1994: Chapter 6) terms “activation
cost” (i.e. cognitive cost involved in the three processes, which thereby refers
to the given-accessible-new distinction), given information is presumably least
costly in the transition from t1 to t2 because it was already active at t1.
Accessible information is somewhat more costly; and new information is the
most costly of all, because converting an idea from the inactive to the active
state supposedly requires more mental effort on the part of the speaker.
The figure given in (2.2) presents two perspectives on activation cost,
i.e. speaker-oriented and listener-oriented. The speaker-oriented perspective,
shown in the top left of (2.2), is concerned only with what is happening in the
the speaker’s mind. At t2, the end of the pause or the onset of the IU, to repeat,
all the ideas to be verbalized in the following IU would be active for the speaker.
From the speaker’s point of view, then, to repeat, an idea that was already
active at t1, the beginning of the pause, would constitute “given” information;
one that was semi-active, “accessible” information; and one that was inactive,
“new” information. The listener-oriented perspective, shown in the bottom
right of (2.2), on the other hand, incorporates the speaker’s understanding of
what is happening in the mind of the listener. Here the speaker assumes that a
particular idea is active, semi-active, or inactive in the listener’s mind at t2, the
onset of the IU. Information status is determined by the speaker’s assumption
Information flow in spoken discourse 

of what will take place in the listener’s mind between t2 and t3, the completion
of the IU. The speaker assumes that hearing the IU will either (a) continue an
idea that is already active for the listener, (b) activate an idea that was previously
semi-active for the listener, or (c) activate an idea that was previously inactive
for the listener. In the cases of (a), (b), and (c), the speaker will verbalize the
idea as “given”, “accessible”, and “new” information, respectively. That is, we
can say that it is the speaker’s anticipation of the activation process in the
listener’s mind that determines the information status of the verbalized idea.
Chafe (1994) emphasizes that although language works better when activa-
tion cost is listener-oriented, the two perspectives—the first dependent solely
on the speaker’s consciousness and the second on the speaker’s assessment of
the listener’s consciousness—may not need to be chosen categorically between
them. Rather, Chafe (1994: 75) argues, “typically a speaker may assume that the
processes in the listener’s mind are in harmony with those in the speaker’s own
mind, allowing for the time lag occupied by the utterance of the IU. In other
words, the events pictured in the section labeled speaker-oriented [in (2.2)]
are likely to be mirrored in the assumptions represented in the section labeled
listener-oriented”.

.. The expression of activation cost

Activation cost is manifested in such linguistic phenomena as the use of a full


noun phrase or a pronoun, as well as in the use or non-use of phonological
prominence, namely, accent. In spoken English, new and accessible informa-
tion are usually expressed with accented full noun phrases. Given information,
on the other hand, is expressed in a more attenuated manner, typically with
a weakly accented pronoun.7 In languages like Japanese and Mandarin, given
information is normally expressed in the most attenuated way, that is, ellipted
(Chafe 1994). Moreover, given and accessible entities are usually referred to by
definite referring expressions, whereas new entities are usually introduced by
indefinite referring expressions (Brown & Yule 1983). It is important to em-
phasize here, however, that in naturally occurring spoken discourse we should
not expect one-to-one mapping between information status and its syntactic
as well as phonological manifestation. For example, phonological prominence
is not always associated with the use of indefinite referring expressions, and
speakers do not always employ them to refer to newly introduced entities. As
Halliday (1967b) stresses, the information structure is speaker-selected; it is
the speaker, not the structure of discourse, that determines the information
statuses of discourse entities. “These are options on the part of the speaker, not
 Chapter 2

determined by the textual or situational environment; what is new is in the last


resort what the speaker chooses to present as new, and predictions from the
discourse have only a high probability of being fulfilled” (Halliday 1967b: 211).
As Brown and Yule (1983: 189) note, no rules but rather only regularities exist
for the syntactic and phonological forms of expressions of information statuses
used by the speaker; hence, “it is the exploitation of these regularities in con-
texts of discourse which allows us to assess the information status attributed to
an entity by speakers”.

.. Identifiability, definiteness, and activation cost

One of the important discourse properties of referents that interact with


activation cost is identifiability. In English, this is often, though by no means
always, associated with the use of definite articles, i.e. definiteness (Chesterman
1991; Chafe 1976; Clark & Marshall 1981; Givón 1984; Hawkins 1978). As
Du Bois (1980: 218) defines it, “identifiability is a property of the relation
between reference and referent. . . . If a noun phrase is said to be identifiable,
this means that the hearer can establish a link between the noun phrase and the
concept it refers to”. Chafe (1994: Chapter 8) discusses three properties that a
referent must have in order to be identifiable: (a) it must be shared; (b) it must
be contextually salient; and (c) it must be verbalized in a sufficiently identifying
way. First, the sharing of knowledge of a referent between the speaker and the
listener can be direct (as when the referent itself is already known to both the
speaker and the listener), or indirect (as when the referent is inferrable from
other shared knowledge). For example, in (2.3b) below, the referent of Larry
is apparently already known to both interactants—it is thus shared. (In (2.3),
which is from Chafe (1994: 71–72), and in examples given below, the symbols
(´) and (`) indicate primary and secondary accents, respectively.)
(2.3) a. I tàlked to a láwyer last nìght,
b. I tàlked to Lárry last nìght,

Second, the contextual salience of a referent may be established by the dis-


course, by the environment in which a conversation takes place, or by the so-
cial group the conversational co-participants belong to—for example, Larry in
(2.3b) must refer to one person named Larry who was more salient than any
others to both interactants. Third, the sufficiently identifying language may re-
quire the use of demonstrative pronouns, proper names (e.g. Larry in (2.3b)),
common nouns with definite or demonstrative articles, or the addition of mod-
ifiers (e.g. possessors, attributive adjectives, relative clauses). In English, the use
Information flow in spoken discourse 

of definite articles typically functions to signal that noun phrases do express


identifiable referents which fulfill the three criteria of sharedness, salience, and
sufficiently identifying verbalization.8
It is important to note, within the framework of Chafe’s discourse pro-
duction model, which I will follow in the present study of Japanese spoken
discourse, that identifiability and activation cost are independent in the sense
that identifiable referents may be either given, accessible, or new. For exam-
ple, the referent of a lawyer in (2.3a) and the referent of Larry in (2.3b) are
both “new” in that both referents were newly activated at the time of utterance;
however, the idea of a lawyer in (2.3a) was assumed by the speaker to be un-
shared, whereas the idea of Larry in (2.3b) was assumed to be shared by the lis-
tener. That is, a lawyer is “new and non-identifiable”, whereas Larry is “new but
identifiable”. Non-identifiable referents, according to Chafe (1994), are nearly
always new; and subjects, or “starting points” are nearly always identifiable.

. Units of information flow

.. Segmentation of discourse into prosodic units

Previous research in discourse processing has identified prosodic units as ba-


sic units of information flow in natural spoken interaction. Discourse data
are not presegmented in contrast to data used in sentence-oriented studies.
Moreover, discourse data can actually be segmented in numerous ways; ac-
cordingly, relevant dimensions for the segmentation of discourse need to be
specified. Prosodic features reflect various interactional, cognitive, and lin-
guistic processes that are at work during the production of spoken language
(Couper-Kuhlen & Selting 1996; Ladd 1996; Pierrehumbert & Hirschberg
1990; Selting & Couper-Kuhlen 2001; Vandepitte 1989). Therefore, past re-
search has naturally focused on the prosodic structure of discourse and seg-
mented spoken interaction into prosodic units. As previous studies have noted,
a prosodic unit is simultaneously the domain of the syntactic as well as in-
formation structure of discourse (e.g. Chafe 1987, 1994; Croft 1995; Du Bois
1987; Halliday 1985; Tao 1996). It is a unit of speech production and percep-
tion (e.g. Schuetze-Coburn 1994; Svartvik 1982). It is also a major component
in turn organization (e.g. Ford & Thompson 1996; Ford et al. 1996, 2002; Mori
1999; Schegloff 1996, 1998; Selting 1996, 2000; Tanaka 1999). The prosody-
based units of talk-in-interaction have been given different terms by different
researchers—for example, “tone groups” (Halliday 1967a, 1985), “tone units”
 Chapter 2

(Brazil 1985; Crystal 1975; Kreckel 1981), “intonation groups” (Cruttenden


1986), “intonational phrases” (Nespor & Vogel 1986; Selkirk 1984), and most
recently, “intonation units” (IUs) (Chafe 1987, 1993, 1994).9
In what follows, I will discuss recent developments in discourse research on
IUs as basic prosodic units of information flow in natural spontaneous spoken
discourse.

.. Intonation units: Definition and identification criteria

The IU is generally defined as “a sequence of words combined under a single,


coherent intonation contour” (Chafe 1987: 22), or “a stretch of speech uttered
under a single coherent intonation contour” (Du Bois et al. 1992: 17). That is,
the IU is defined basically in terms of a unified intonation contour. According
to Chafe (1987: 24), an IU roughly consists of an “initial pause” followed by
a period of vocalization. Thus, in his model, a discourse can be viewed as
a succession of IUs, which in turn consist of alternating pauses (. . . ) and
vocalizations (xxx), as schematically shown in (2.4). (The lengths of pauses
and vocalizations vary in actual speech production.)
(2.4) . . . xxx . . . xxx . . . xxx . . . xxx . . . xxx . . . xxx
[ IU ] [ IU ] [ IU ] [ IU ] [ IU ] [ IU ]
Chafe (1980b: 14) provides the following three criteria for the identifica-
tion of IUs:10
a. IUs tend to end with a continuing or falling intonation contour.
b. IUs are typically separated by at least a brief pause.
c. IUs tend to consist of a single clause, which contains one verb plus
accompanying noun phrases that are associated with it.

Chafe claims that of his three identification criteria (i.e. intonational, hesita-
tional, and syntactic), intonation contour is the most consistent signal of an IU
boundary in English, whereas the presence of a pause or the clausal syntactic
structure is a less important criterion for the identification of such units. In
English, Chafe shows, IUs are not uniformly separated by hesitations (pauses
often occur within such units); and some IUs are less than clauses (e.g. phrases)
or more than clauses (e.g. IUs that contain an embedded relative clause).
Du Bois et al. (1992: 100) present the following as the major prosodic cues
that contribute to identifying the boundaries of IUs:
Information flow in spoken discourse 

a. coherent contour: a unified intonation contour, i.e. one displaying overall


gestalt unity.11
b. reset: a resetting of the baseline pitch level at the beginning of the unit.
c. pause: a pause at the beginning of the unit (in effect, between two units).
d. anacrusis: a sequence of accelerated syllables at the beginning of the unit.
e. lengthening: a prosodic lengthening of syllables at the end of the unit (e.g.
of the last syllable in the unit).

Du Bois et al. note that a prototypical IU may well be characterized by all


of these five prosodic features; however, their presence is neither a necessary
nor a sufficient criterion of IU status, because some of the cues (e.g. pause,
lengthening) can occur for reasons other than an IU boundary. Moreover,
some of the cues (e.g. resetting of the baseline pitch level) may be hard to
identify unequivocally under certain conditions. They therefore suggest that
in especially difficult cases, the transcriber should weigh all of these factors
together in order to arrive at a reliable determination of IU status (see Chafe
1994: 56–61; Cruttenden 1986: 35–45; Du Bois et al. 1992: Chapter 22). Clearly,
more research will need to be directed toward refinements of the specification
of the relevant prosodic properties of IUs and the relation between auditory
IUs and acoustic units (cf. Schuetze-Coburn et al. 1991).

.. Intonation units, consciousness, and language production

Regarding the insights IUs can yield into the nature of language production
and thought processes, Chafe (1993: 38–39) states the following:
As an intuitive starting point, I can observe that I am able to focus my
consciousness on only a very small amount of information at one time and
that this focus of consciousness changes quite rapidly as thinking proceeds.
Furthermore, it is plausible to suppose that during the production of language
a speaker will focus on the information he or she is verbalizing at that
moment. Against this background, an intonation unit is plausibly viewed as
the verbal representation of just the information that is in the speaker’s focus
of consciousness at the moment it is uttered. A speaker’s intention in uttering
an intonation unit must then be to introduce something resembling that
particular focus of consciousness into the attentive listener’s consciousness.
If each intonation unit, indeed, corresponds to a focus of consciousness,
intonation units can give us important insights into how much and what kinds
of information can be active at one time in a speaker’s mind.
 Chapter 2

In sum, for Chafe (1980b: 48), IUs are “linguistic expressions of focuses of con-
sciousness, whose properties apparently belong to our built-in information-
processing capabilities”. In other words, the IU as a whole is a verbalization of
the particular information on which the speaker is focusing his/her conscious-
ness at a particular moment. The IU is also viewed as the linguistic expression
of information that is, at first, active in the consciousness of the speaker, and
then, by the utterance of the unit, active in the consciousness of the listener
as well (cf. Section 2.3.2). The question of “how much” and “what kinds” of
information can be activated within each IU will be dealt with in terms of
constraints on information flow in Section 2.6.

.. Types and size of intonation units

Chafe (1993, 1994) categorizes IUs into three major types: “substantive”, “regu-
latory”, and “fragmentary”. Substantive IUs convey substantive ideas of events,
states, or referents, participating in the communication of propositional con-
tent. The two other non-substantive types are not directly concerned with the
transmission of substantive ideational information. Regulatory IUs function
to regulate interaction or information flow in discourse (IUs consisting of so-
called “discourse markers” belong to this category; see Schiffrin 1987). Both
substantive and regulatory IUs are successfully completed units. Fragmentary
IUs, on the other hand, are truncated units that the speaker breaks off before
the completion of their projected contour.
The regulatory type can be further divided into at least four subtypes: “tex-
tual”, “interactional”, “cognitive”, and “validational”. The textual subtype func-
tions to regulate the linkage, or to show a particular kind of linkage, between
IUs (e.g. and then, but, so). The interactional subtype involves interaction be-
tween the conversational co-participants, indicating the speaker’s attentiveness
to the hearer (e.g. mhm, you know). The cognitive subtype signals ongoing
mental activities on the part of the speaker (e.g. let’s see). The fourth, vali-
dational subtype expresses the speaker’s judgment of the validity of the infor-
mation being conveyed (e.g. maybe). As Chafe (1993) emphasizes, however, the
boundaries between these subtypes of regulatory IUs are not clear-cut. (For ex-
ample, the function of the regulatory IU well may be textual, interactional, or
cognitive, depending on the conversational context where it is uttered.) What
is evident is that IUs that serve several regulatory functions contrast sharply
with IUs that convey substantive ideational information (cf. Section 3.5.3).
The taxonomy of IUs is shown in (2.5).
Information flow in spoken discourse 

(2.5) Intonation Units


Completed Non-completed
Substantive Fragmentary
Regulatory
Textual
Interactional
Cognitive
Validational

The distinction between substantive (S), regulatory (R), and fragmentary (F)
IUs is illustrated in (2.6), an excerpt from a conversation between three
participants, A, B, and C (Chafe 1994: 63–64).
(2.6) a. (R) A: ++ Well,
b. (S) isn’t she healthy?
c. (R) B: + Mhm,
d. (F) A: ++ I mean she -
e. (F) I know she has -
f. (S) C: More or less.
g. (S) A: + She has [something with her] gallbladder,
h. (S) B: [ gallbladder and, ]
i. (S) ++ heart trouble and,
j. (S) [back problems.]
k. (S) A: [ She has heart ] trouble,
With regard to the size of IUs in English, Chafe (1994: 64–65) observes that
while regulatory IUs tend to be one word in length, substantive IUs are fairly
constrained to a modal length of four words.12 He indicates that these figures
are applicable to English only. In languages that pack more information into a
word, IUs generally contain fewer words; for example, IUs in Seneca, a highly
polysynthetic language, tend to be two words in length (Chafe 1994: 148). Thus,
for Chafe, the fact that the number of words per IU remains within a narrow
range for any given language reflects a cognitive constraint on the capacity of
active consciousness, i.e. how much information can be active in the speaker’s
mind at one time.

.. Intonation units, accent units, and information structure

Chafe (1993) introduces the term “accent units” (AUs) as subunits of IUs.
An AU, according to Chafe, contains only one primary accent; it consists
of the word containing that primary accent—which roughly corresponds to
 Chapter 2

what Halliday (1985) calls “tonic foot” that contains a “tonic syllable”—
plus whatever other words belong to the same constituent as that word.
Chafe hypothesizes that the AU is the domain of activation of information
in consciousness, that is, information is activated and deactivated within
each AU. In other words, AUs are the loci of three types of information,
given, accessible, and new. An IU, being composed of one or more AUs,
verbalizes a cluster of ideas; and this cluster may include more than one given
or accessible information, but only one new information, according to his
proposed constraint (as will be shown in Section 2.6).
In the following conversational excerpts, (2.7) and (2.8), which are from
Chafe (1993: 38), AUs are separated by the “pipe” symbol (|).
(2.7) a. A: (1) ++ and thé:n | (2) the má:n,
b. ->(1) ++ ùh: hèr bóyfriend whatever | (2) was gonna móve
- ->ìn | (3) wíth them.
(2.8) a. A: ++ And so in betwéen the -
b. ->(1) + okay the fírst two ròoms | (2) are at the t + at the +
- ->frónt pàrt of the hàll.
c. B: ++ Mhm,
d. A: ->(1) + and so + betwéen those ++ the èntrances to those
- ->ròoms and the bàthroom | (2) there’s a lò:ng stretch of hállway.
Example (2.7) consists of two substantive IUs, (2.7a, b). The IU (2.7a) consists
of two AUs, (2.7a-1) and then and (2.7a-2) the man, each of which carries a
primary accent. The first AU and then is regulatory, whereas the second AU
the man is substantive.13 The IU (2.7b), on the other hand, is composed of
three substantive AUs, (2.7b-1), (2.7b-2), and (2.7b-3), where primary accents
are placed on boyfriend, move, and with, respectively. Likewise, (2.8) consists of
one fragmentary IU, (2.8a), one regulatory IU, (2.8c), and two substantive IUs,
(2.8b) and (2.8d). Both of these substantive IUs consist of two substantive AUs.
These examples indicate that regulatory AUs can be subunits of both regulatory
and substantive IUs.
Turning to the activation states of information verbalized in the four sub-
stantive AUs in (2.7), we can observe the following. First, the AU (2.7a-2) the
man verbalizes given information. Second, the AU (2.7b-1) uh her boyfriend
whetever can be taken to express accessible information, because it has the
same referent as the man activated in (2.7a), and the added characterization
of that referent (i.e. the idea of him being the woman’s boyfriend) is presum-
ably accessible. The AU (2.7b-2) was gonna move in also expresses accessible
infomation because the idea of moving-in was already activated earlier in the
Information flow in spoken discourse 

discourse. The only AU that verbalizes new information is the last AU, (2.7b-3)
with them, where the very utterance of the primary accented preposition with
introduces the man’s participation in the activity as a totally new, previously
unmentioned concept (Chafe 1993). Similarly, in the IUs (2.8b) and (2.8d),
the second AUs, (2.8b-2) and (2.8d-2), respectively, express new information,
with the preceding AUs providing a context/background for the following one
new idea. In sum, the AUs (2.7b-3), (2.8b-2), and (2.8d-2) are the domains
in which only one idea is newly activated within the IUs (2.7b), (2.8b), and
(2.8d), respectively. Examples (2.7)–(2.8) appear to illustrate the basic canon-
ical pattern of information flow—from given to new—within units of English
spoken discourse. The new element carrying “tonic prominence” and thus “in-
formation focus” is typically the last functional element within the “informa-
tion unit” or “tone group” (which can be regarded as essentially the same as
the IU) (Halliday 1985).
Chafe (1987) argues that the information structure of an IU usually
consists of a “starting point” and an “added information”, which are manifested
linguistically in the subject-predicate structure. He proposes two constraints
relevant to the information structure of IUs in discourse. First, the starting
point is usually a given—or occasionally an accessible, but rarely a new—
referent (if it is new, it is normally of trivial importance); hence, subjects are
governed by the “light starting point constraint” (the “light subject constraint”
in Chafe 1994), where the lightness could be taken to refer to non-newness.
Second, the added information typically contains one new concept, though it
may also contain some accessible or given concepts; hence, the “heavy added
information constraint”. These two constraints appear to be in accord with
Halliday’s (1967b, 1985) view of the “unmarked” structure of information
within the “information unit” as being “given followed by new”, Firbas’ (1992)
claim of the correspondence of the order of words in a sentence and an increase
in “communicative dynamism” (CD),14 and Du Bois’ (1987) remark that given
status is the norm for nominal references, whereas new status is the norm for
verbal predicates.

.. Intonation units, clauses, and IU linkage

Chafe (1987, 1988, 1994) observes that the majority of substantive IUs in
English spoken discourse take the form of complete single clauses. This means
that an English speaker’s typical way of verbalizing a focus of consciousness is
through the format of a clause. More specifically, Chafe (1987: 38) states that
“the clause appears to be the prototypical IU type, from which most other types
 Chapter 2

are derived, or are deviations”. Likewise, Halliday (1967b: 242), concerning the
relationship between the realization of his “information unit” phonologically,
in the “tone group”, and syntactically, in the clause, states that “in the unmarked
case (in informal conversation) the information unit will be mapped on to
the clause”. That is, Chafe’s as well as Halliday’s claim is that the clause is the
syntactic exponent of the IU, and thus clause linkage is the predominant type of
IU linkage in spoken English. I refer to this as the “clause centrality proposal”.
It has been shown that the mean proportion of single-clause substantive
IUs in spoken English is about 60–70% (Chafe 1988, 1994). Clausal IUs express
ideas of states or events, and usually each IU verbalizes a different state or event
from that which precedes it. That is, state and event ideas are highly transient
in active consciousness, constantly being replaced by other state and event
ideas. This dynamic process of successive and transient activations seems to be
analogous to Givon’s (1987) view of the foreground-background alternation
discussed in Section 2.2.1. The continual replacement of state and event ideas
also reflects that our consciousness is in constant change, restlessly moving
from one idea to another. The sequence in (2.9) below illustrates such rapid
progression from one state/event idea to the next (Chafe 1994: 66).
(2.9) a. A: ++ Cause I had a ++ a thick patch of barley there,
b. B: ++ mhm,
c. A: + about the size of the + kitchen and living room,
d. ++ and I went over it.
e. + and then,
f. ++ when I got done,
g. I had a little bit left,
h. + so I turned around,
i. and I went and sprayed it twice.
j. + and it’s just as yellow as ++ can be.
Example (2.9) contains eight substantive clausal IUs. Of these, (2.9a), (2.9c),
(2.9g), and (2.9j) verbalize ideas of states, whereas (2.9d), (2.9f), and (2.9h, i)
verbalize ideas of events. This excerpt shows how one newly activated state/
event idea is rapidly replaced by another newly activated state/event idea
as clausal IUs are successively produced. This means that ideas of states
or events expressed in predicates are normally new, in contrast to ideas of
referents, which are more persistent (for example, the referent verbalized as
I is repeatedly used in (2.9)) (cf. Section 2.4.5).
Related to the clause centrality proposal discussed above is Pawley and
Syder’s (1983: 564–565) “one clause at a time constraint”. This constraint ba-
Information flow in spoken discourse 

sically suggests that humans can encode or formulate only the contents of “one
clause at a time”. The clause is thus treated as the basic unit for information pro-
cessing and segmentation in human spoken discourse. They argue that there is
a fundamental limit in humans’ verbal processing such that in a single span of
attention focus, it is only possible to plan the content of a novel clause of up to
about ten words; therefore, to encode the full lexical content of a longer novel
sequence requires two or more separate encoding operations. The one clause at
a time constraint, according to Pawley and Syder, allows speakers to maintain
maximal fluency within the limits of their encoding capacity, and underlies the
characteristic “clause-chaining” style of spontaneous connected discourse, i.e.
the preponderance of conjoined and adjoined clauses with much less use of
subordination than in formal writing (cf. Haiman & Thompson 1988). Givón
has also identified the mental proposition which codes some cognized state or
event and which often surfaces as a clause in natural connected discourse as the
basic unit of information storage and discourse processing. That is, for Givón
(1990: 896), just as for Chafe, Halliday, and Pawley and Syder, “the basic unit of
stored information in coherent discourse is the mental proposition that stands
for some state or event”.
Chafe (1988) discusses two major kinds of linkages that exist between IUs
in conversational English. First, there are the linkages signaled by intonation
alone, where falling intonation signals closure of an idea or idea sequence, and
continuing intonation signals continuation from one idea to another. Second,
there are the linkages signaled by explicit connective words such as and, but,
and so. He reports that the first “non-connective type” accounted for about
44% of the cases of IU linkages in his study, whereas the remaining 56%
of the cases involved more specific linkage markers. In addition, half of the
“connective-type” IU linkages involved the conjunction and.
The IU sequences (2.10)–(2.12) show how IUs are linked to one another in
English conversational monologues.
(2.10) a. A: ++ I came home,
b. I was really exhausted,
c. I was eating a popsicle,
d. ++ I was sitting there in my chair,
e. ++ just eating my popsicle,
(2.11) a. A: and then another day,
b. ++ it was really hot,
c. it was in the summer and,
d. + my room was small.
 Chapter 2

(2.12) a. B: ++ it’s just about going away now,


b. + I had it about two weeks.
c. ++ anyway,
d. + so I was sort of nervous about that,
e. + for a day or two.
f. + and then I forgot about it.
Example (2.10) illustrates IU linkages of the non-connective type: the IUs
are linked with continuing intonation contours alone, which are marked with
commas (Chafe 1988: 10). Example (2.11) illustrates, on the other hand, IU
linkages of the connective type: the IUs are linked with the coordinating
conjunction and (Chafe 1988: 11). Note that and is used IU-initially in (2.11a),
but IU-finally in (2.11c). In (2.12), which is from Chafe (1988: 17), both types
of IU linkages can be seen. In (2.12d) and (2.12f), so and and then are explicitly
used as IU-linking connectives, respectively. In (2.12c), the conjunctive anyway
occupies the complete IU by itself; this indicates the speaker’s momentary
focus entirely on the linkage. In (2.12a, b) and (2.12e), falling or continuing
intonation contours alone are used to link the IUs.
Chafe (1988: 22–23) observes that unit linkages in spoken English are of a
relatively simple nature. That is, English speakers tend to favor non-elaborate
linking devices—using falling/continuing intonation alone or “handy” con-
nectives such as and—to link one IU to another. He notes that the minimal
use of elaborate linking devices in spoken English can be linked to the charac-
teristics of spoken communication. Specifically, the non-elaborate nature of IU
linking has to do not only with speakers’ attention focus on the verbalization
of ideas, which leaves them little time for making the unit linkages explicit, but
also with the presence of a directly shared context and the supplementary use of
prosody and gestures, which help to make the connections between ideas more
apparent and therefore make the elaborate use of linking devices unnecessay.

. Japanese intonation units

This section discusses the syntactic and functional characteristics of IUs in


spoken Japanese, based on the results of prior research conducted by such
discourse researchers as Clancy, Iwasaki, and Maynard. For detailed discussion
of characteristic features of Japanese intonation and prosodic properties of
Japanese IUs, see Beckman and Pierrehumbert (1986), Pierrehumbert and
Information flow in spoken discourse 

Beckman (1988), Sugito (1989, 1990), Hirst and Di Cristo (1998), Iwasaki
(2002), or Venditti (2003) (cf. Section 3.3).

.. Syntactic fragmentation in spoken Japanese

As discussed in Section 2.4.6, if the clause is the prototypical English IU type,


and thus clause linkage is the robust form of IU linkage in spoken English
(Chafe 1987, 1988, 1994), the question to be raised is whether or not this is
the universal strategy speakers of different languages employ to link one idea
to another. Recent research has shown that the answer seems to be negative.
Iwasaki and Tao (1993), in their attempt to test the cross-linguistic valid-
ity of the clause centrality proposal, examined the syntactic structure of sub-
stantive IUs produced by native speakers of English and Japanese in interactive
conversations. Their major findings are summarized in (2.13).15
(2.13) Clausal Non-clausal
English 54% 46%
Japanese 45% 55%
Full-clause Semi-clause
English 83% 17%
Japanese 24% 76%
NP IUs
English 12%
Japanese 22%
The percentages given in (2.13) indicate the following. First, while there is a
tendency for English IUs to be more clausal, Japanese shows a preference for
non-clausal, or phrasal IUs that lack verbal predicates. Second, the clausal IUs
in English are overwhelmingly full clauses composed of overt subjects and ver-
bal predicates, whereas the clausal IUs in Japanese are mostly subjectless semi-
clauses that consist of verbal predicates only. (Clearly this can be attributed
to the grammatical difference between the two languages that Japanese al-
lows for abundant use of zero anaphora, whereas English does not.) Finally,
in Japanese conversation, use of NP IUs is about twice as frequent as in English
conversation.16
Based on the higher frequency of non-clausal IUs (including NP IUs) in
the Japanese data, they conclude that speakers of Japanese are more likely to
“fragment” the clause than English speakers. They suggest that this is true even
when the speaker could have produced a single clause that would communicate
 Chapter 2

the same ideational information as a collection of the pieces would. However,


this should by no means be interpreted to suggest that spoken English does
not involve “syntactic fragmentation”. First and foremost, the proportions of
non-clausal, phrasal IUs in Japanese (55%) vs. English (46%) do not seem to
differ significantly. Spoken English, although with fewer NP IUs, might contain
more other types of phrasal IUs such as PP IUs. Moreover, the clause vs. phrase
centrality of English IUs might depend on the genre or type of spoken discourse
examined (e.g. narrative vs. conversation, “natural” vs. “elicited”). It might
also depend on such factors as the relationship between the co-participants
and the purpose of conversational interaction. The degree of “fragmentedness”
in spoken English in comparison to other languages such as Japanese should
therefore be examined more closely in future research (cf. Brown & Yule 1983;
Maynard 1989; Ono & Thompson 1994).17
Consider the IUs in (2.14), which Iwasaki and Tao (1993: 7) provide as one
of the best illustrations of the fragmentedness of Japanese IUs.
(2.14) a. A: Yamato san ga ne:,
Yamato hon nom fp
‘Mr. Yamato’
b. kekkyoku ne:,
in short fp
‘in short’
c. wareware o ne:,
us acc fp
‘us’
d. sofuto no ne:,
software gen fp
‘on software’
e. shigoto nitaisuru.
job toward
‘of (our) job’
f. hyooka ga ano hito ne:,
evaluation nom that person fp
‘evaluation, he’
g. shitenai.
do-prog-neg
‘doesn’t do’
h. hyooka o.
evaluation acc
‘(high) evaluation’
Information flow in spoken discourse 

i. zettai.
never
‘never’

Of the nine IUs contained in (2.14), only one, (2.14g), is a semi-clause consist-
ing of a verb with a missing subject, according to Iwasaki and Tao (1993). The
remaining eight IUs are non-clausal, phrasal units; (2.14a), (2.14c), (2.14f),
and (2.14h) are nominal, (2.14b) and (2.14i) are adverbial, and (2.14d, e) are
adjectival IUs. We can observe that five of the eight phrasal IUs contain the in-
teractional particle ne in the IU-final position. It should be noted that speaker
A could have conveyed the same ideational content as the nine short IUs col-
lectively do by means of one full-clausal IU, which could roughly correspond
in English to ‘Mr. Yamato never values our software job’.
Iwasaki and Tao (1993) showed that the proportion of NP IUs in Japanese
is higher than in English, but failed to show where those NPs appeared, or what
forms those NP IUs took syntactically—for example, whether they occurred
as clause-internal constituents or as clause-external independent IUs. Clearly
their study lacks such detailed information. Moreover, they coded (2.14g) as a
semi-clause; but given that the nine IUs collectively constitute a full proposition
meaning that ‘Mr. Yamato never values our software job’, I would code (2.14g)
as a VP, which is clearly propositionally dependent on the other phrasal IUs.
That is, in my view, the nine phrasal IUs, (2.14a–i), collectively constitute a
full clause, with (2.14a) functioning as the subject, and (2.14g) functioning as
the verbal predicate of that multi-IU full clause. (Alternatively, (2.14a) could
be coded as a left-dislocated NP topicalizing the referent ‘Mr. Yamato’, and
(2.14b), as a clause-external adverbial IU.) The two post-predicate phrasal
IUs, (2.14h) and (2.14i), could be taken to function as the direct object
and an adverbial intensifier, respectively, of the negated verb ‘do’ in (2.14g)
(see Section 4.3). (Note that the o-marked NP ‘us’ in (2.14c) cannot be the
direct object of the verb ‘do’, and the ga-marked NP ‘evaluation’ in (2.14f),
which is modified by the preceding adjectival phrasal IUs, (2.14d, e), lacks its
predicate element.)
In short, Iwasaki and Tao’s (1993) coding is too simple in that it treats
each IU as a syntactically complete clausal or phrasal discourse unit; it does
not consider the possibility that multiple IUs will constitute a clause. Another
drawback in their coding is that they looked at only the surface grammatical
forms of IUs in distinguishing “clauses” from “phrases”, thereby eliminating
the possibility that some of what we formally identify as “phrases” may
communicate a complete proposition as “clauses” normally do (cf. Ono &
 Chapter 2

Thompson 1994). It is evident that a more sophisticated coding scheme will


be needed for a closer examination of the syntactic structure of the IU.
The “highly fragmented” nature of spoken Japanese, as argued for in
Iwasaki and Tao (1993), has also been pointed out by other discourse re-
searchers. Maynard (1989), for example, refers to such smaller phrasal units as
“pause-bounded phrasal units” (PPUs), which, in my interpretation, are essen-
tially the same as the phrasal IUs found in (2.14). She notes that the tendency
towards fragmentation in Japanese is expressed to an even greater degree than
in comparable American English situations—not so much through the process
of the fragmentation itself as through the use of particles and other devices at
the unit-final position. For example, the use of heavy stress and higher pitch
on the final syllable of the last word is employed in such a way that fragmen-
tation is emphasized and made more prominent (Maynard 1989: 21–22). Like-
wise, Clancy (1982) reports that in her study of the “Pear Film”18 Japanese oral
narratives, 67% of the IUs that she examined were short phrasal units which
did not include the predicate of the case frame being verbalized, underscor-
ing that in spoken Japanese a single clause is often communicated in several
distinct IUs.
Clancy (1982: 74) suggests that such “syntactic fragmentation” may serve
social, interactive functions in that it invites the listener’s more frequent
participation in the interactive oral communication by providing some kind
of verbal feedback, including backchannels, at the end of most IUs. This jointly
creates a more cooperative interaction than is typical of English conversation.
She also notes that it may also simplify the planning and comprehension tasks
of the speaker and the listener, respectively, thus allowing cognitive advantages
for the conversational co-participants. Similarly, Iwasaki (1993) argues that a
strong motivational factor responsible for the claimed phrasal IU structure in
spoken Japanese is the coding of non-referential, speaker-hearer interactional
information in addition to propositional, ideational information (see detailed
discussion in Section 2.5.2). On the other hand, the fragmented pattern of
discourse production is clearly less efficient as far as the communication of
propositonal content is concerned; it is presumably more time-consuming
to convey a piece of information by means of several separate IUs than by
means of a single unit. In sum, the main argument presented in the previous
studies is that the fragmentation, or the breakup of a clause into smaller units
in conversational Japanese is greatly influenced by social, and speaker-hearer
interactional considerations.
What the earlier studies have shown, in effect, is options available to
Japanese speakers when producing IUs. That is, they can produce a clausal IU,
Information flow in spoken discourse 

or alternatively they can break up a clause and produce phrasal IUs. English
speakers, by contrast, usually have only the clausal IU strategy available to
them. Importantly, this difference in the speakers’ IU-production patterns can
also be explained in terms of the fundamental structural difference between
English and Japanese—the “tightness” of the clause as a unit. English clauses
are tight-knit, coherent units with a predicate and its associated arguments
(especially a subject) clearly united and overtly expressed. Japanese clauses,
on the other hand, are a more loosely organized entity, in which arguments
can be freely unexpressed, and constituents are generally more independent
thus more separable from one another than the counterparts in English clauses
(see Fox et al. 1996). It is therefore reasonable to state that due to the nature
of Japanese syntax, Japanese speakers are allowed to fragment IUs into phrasal
units “more readily” than English speakers. (It seems that the use of particles—
not only interactional particles like ne and sa but also case-marking particles
like ga and o—play a role in making elements in an utterance more distinct
from one another.) In short, previous research has evidenced that Japanese
speakers have two different options, clausal and phrasal IUs, whereas English
speakers usually have only one option, clausal IUs. This difference is significant
because it is related to the “tight” vs. “loose” internal structure of clauses in the
two languages.
Given the need for a more sophisticated coding scheme that I suggested
above, and the commonly made claims for the “phrase-orientedness” of
Japanese IUs, what remains to be done is to retest the validity of the clause
centrality proposal. Let us explore this issue in Chapter 4.

.. Multi-functional nature of Japanese intonation units

Halliday (1973, 1989), opposing the tradition that shows partiality to clauses
as a means of conveying propositions and as the major domain of inquiry,
proposes that language is composed of three metafunctions: “ideational”,
“interpersonal”, and “textual”. The ideational function further consists of two
functions: “experiential” and “logical”. The experiential function represents
the real world as we apprehend it in our experience; the logical function
represents relations among propositions such as hypotaxis and parataxis.
The interpersonal function is concerned with the conduct of different types
of speech acts. The textual function concerns the creation of coherence in
discourse. Halliday (1989) claims that these metafunctions cannot be isolated
in utterances, but the clause serves these functions simultaneously.
 Chapter 2

Halliday’s multifunctional view of language is shared in Japanese linguis-


tics by Tokieda (1950) and his followers. They identify three metafunctions of
language: “descriptive”, “expressive”, and “interpersonal”. The descriptive func-
tion, which is performed by the act of jojutsu ‘description’, concerns the com-
munication of propositions. The expressive function, which is performed by
the act of chinjutsu ‘expression’, reveals the speaker’s internal state, or subjec-
tivity. The interpersonal function concerns itself with speech acts as Halliday
claims, or the speaker’s sensitivity toward the addressee. Unlike Halliday, how-
ever, the Tokieda-school linguists assume that the functions can be isolated in
utterances, so that it is possible to point out a linguistic element and name its
function, although there are cases where one linguistic element serves multiple
functions.
Iwasaki (1993), reviewing the views proposed by Halliday and the Tokieda-
school Japanese linguists, concludes that language must incorporate four meta-
functions: “ideational”, “interactional”, “cohesive”, and “subjective”. To refer to
linguistic elements that represent the metafunctions in a particular utterance,
he introduces the term “functional components”, which is to be distinguished
from the abstract notion of “metafunctions” of language. He thus argues that
the Japanese IU maximally consists of four functional components, which are
arranged in the following linear order: ideational, subjective, cohesive, and in-
teractional. According to his definition, the ideational component participates
in the construction of a proposition. The subjective component expresses the
speaker’s internal state such as affect and epistemic stance, or the speaker’s
subjective judgment. (The subjective function is claimed to be accomplished
in Japanese by such lexical words as rashii ‘seem’ and highlighting particles
like dake ‘only’.) The cohesive component serves what Halliday terms “textual”
function, i.e. to create cohesion among elements in discourse (cf. Halliday &
Hasan 1976). Finally, the interactional component functions to exhibit the
speaker’s sensitivity toward the addressee.
Iwasaki further adds two revisions to his four-component analysis of
IUs so that it can handle real speech data more efficiently and completely:
fusing the subjective component with the ideational component, and adding a
component which he terms “lead”. He therefore proposes that the IU in spoken
Japanese can be analyzed in terms of the following four functional components,
namely, the lead [LD], ideational [ID], cohesive [CO], and interactional [IT]
components. The lead, which consists of pause fillers such as ano ‘uh’ that
signal the speaker’s cognitive planning, functions to regulate the flow of
conversation, signaling more material to follow and the speaker’s intention
of keeping the floor of conversation. The ideational component transmits
Information flow in spoken discourse 

propositional content. The cohesive component, which does textual referential


work, consists of markers of cohesion such as nominalizers like no and wake,
conjunctives such as kara ‘because’ and kedo ‘though’, and non-finite predicate
forms such as -tara ‘if ’ and -te ‘and’, according to Iwasaki. The interactional
component, composed of interactional particles and expressions like ne, sa, yo,
janai, and desho, serves the speaker’s need to show sociolinguistic sensitivity
toward the addressee by soliciting the listener’s involvement in the current
speech event.
Iwasaki (1993: 46) argues that Japanese IUs have very orderly functional
structure, such that in a maximal structure the four components invariably
occur in a fixed linear order within an IU, as shown in (2.15).
(2.15) [LD] [ID] [CO] [IT]

Given in (2.16) is one of his substantive IU examples that contain all of the four
functional components in the linear order specified in (2.15).
(2.16) ano tabi nante hitoride shita koto nakatta no ne?
pf trip sof alone do-past nml exist-neg-past nml fp
[LD][ ID ][CO][IT]
‘uh (I) had never taken a trip alone.’

The IU (2.16) begins with ano ‘uh’, a pause filler which constitutes the lead
component. This is followed by the ideational component, where a full propo-
sition is conveyed by the subjectless semi-clause ‘(I) had never taken a trip
alone’. The nominalizer no, a device which signals a cohesive tie between ele-
ments in discourse, comprises the cohesive component, according to Iwasaki.
The interactional particle ne, placed IU-finally with rising intonation, attempts
to involve the addressee in the current speech situation; this constitutes the
interactional component.
Iwasaki (1993), using the four codes, analyzed IUs produced in three
types of Japanese spoken discourse (i.e. narratives, telephone and face-to-
face conversations). The IUs that he analyzed include both substantive and
regulatory IUs (cf. Section 2.4.4). The main results are presented in (2.17).
(2.17) ID 34%
ID CO 22%
ID IT 17%
LD ID 2%
ID CO IT 12%
LD ID CO IT 0.3%
 Chapter 2

It was found in his study that the IU in spoken Japanese typically consists
of no more than two functional components. More specifically, IUs with one
or two components accounted for 85%, whereas IUs with three components
accounted for 13%, and IUs with four components occupied only 0.3%. In
particular, as shown in (2.17), the one- and two-component IUs which include
the ideational [ID] component (i.e. ID, ID-CO, ID-IT, and LD-ID) occupy
75% of all the IUs analyzed (87% if the three-component type ID-CO-IT is
added). Among them, the most popular functional structure type is the one
which consists of the ideational component only (i.e. ID), occupying 34%. On
the other hand, the types ID-CO and ID-IT jointly occupy 39%; this means
that those types consisting of the ideational component with the cohesive
or interactional component are similarly the preferred pattern. In addition,
of all the IUs examined, those which include the interactional component
accounted for 33%.
Iwasaki (1993) interprets these findings as follows. First, the most impor-
tant task of IUs is to convey ideational information. (This is also reflected by
the large proportion of idea-conveying substantive IUs that constitute a given
discourse; cf. Section 3.5.3.) Second, while the IU in spoken Japanese tends
to communicate ideational information alone, it more often than not conveys
additional information, cohesive or interactional, at the same time. That is,
“Japanese IUs have a built-in mechanism which allows the speaker to attend
to different concerns of communication other than ideation itself ” (Iwasaki
1993: 50). Third, there is a constraint operating in spoken Japanese that per-
mits speakers to incorporate up to two functions within one IU. Iwasaki relates
the “no more than two functions per IU” constraint formulated in his study to
the cognitive limitation on how much information speakers can handle within
an IU—particularly to Pawley and Syder’s (1983) “one clause at a time con-
straint” (discussed in Section 2.4.6), which limits an IU to the expression of
no more than one proposition, and to Chafe’s (1987) “one new concept at a
time constraint” (to be discussed in Section 2.6.2), which limits an IU to the
expression of no more than one piece of newly activated information.
Most importantly, Iwasaki (1993) explains the occurrence of phrasal IUs,
which he claims to be pervasive in Japanese discourse, in terms of the multi-
functionality of Japanese IUs discussed above. He argues that a strong moti-
vational factor responsible for the phrasal IU structure in spoken Japanese is
“the multi-faceted task which the speaker must carry out in one IU” (Iwasaki
1993: 50)—the coding of non-referential, speaker-hearer interactional infor-
mation in addition to propositional, ideational information. That is, the fre-
quent use of the “partial propositional strategy”, i.e. use of phrasal IUs (as
Information flow in spoken discourse 

opposed to the “complete propositional strategy”, i.e. use of clausal IUs) in


spoken Japanese is a consequence of the multi-faceted task imposed on the
speaker by the discourse structure of the Japanese language (Iwasaki 1993: 41).
As Iwasaki and Tao (1993: 6–7) state:

Japanese has grammaticized speaker-hearer interactional information in ad-


dition to textual cohesive information and propositional/ideational informa-
tion in the construction of discourse. In other words, Japanese speakers are
given the opportunity to code all these kinds of information in an intonation
unit. Thus it is typical for speakers of Japanese to divide ideational informa-
tion, generally realized in clausal forms, into pieces. This explains, in Japanese,
the frequent use of semi-clauses and NP intonation units accompanied by
interactional particles.

This would alternatively explain, according to Iwasaki, why IUs in spoken


English tend to be clausal. He argues that English speakers, unlike Japanese
speakers, are free of the task of incorporating the non-ideational, cohesive and
interactional information in an IU. This thus enables them to concentrate
on the formation and communication of a proposition, which is normally
accomplished in English by the clausal format, typically with a subject and
a predicate expressed within one IU. That is, for Iwasaki, the clausal vs.
phrasal syntactic structural differences between English and Japanese IUs can
be attributed to their functional structural differences.
While this interpretation may sound plausible, the claimed relationship be-
tween the multi-faceted task that the Japanese speaker must carry out within
one IU and the frequent use of the partial propositional strategy in Japanese
spoken interaction is far from clear.19 Specifically, how can we relate the oppor-
tunity given to Japanese speakers to code in an IU cohesive and interactional
information in addition to ideational information to their tendency to divide
a clause into shorter phrasal IUs? Unlike interactional information, cohesive
information invariably co-occurs with predicates. In other words, the incorpo-
ration of a cohesive component into an IU necessitates the presence of a clausal
IU. Therefore, the question to be answered can now be restated as the following:
why do Japanese speakers divide a proposition-conveying clause into pieces
when faced with the need to communicate additional interactional informa-
tion? In my view, the key to this question seems to lie in the “value” placed
by Japanese interactants on non-ideational, interactional coding in discourse.
Presumably, the breakup of a clause into phrasal IUs will affect the “frequency”
with which such non-ideational, interactional information can be coded in in-
 Chapter 2

teractive discourse. (See (2.14) in Section 2.5.1 as an illustration.) This issue


will be explored in Chapter 6.

. Constraints on information flow

.. The function and flow rate of given information

Propositions or clauses transacted in coherent discourse tend to carry both


given and new information, that is, they tend to be “informational hybrids”
(Givón 1990: 898). This is because clauses with only given information are
redundant, providing the hearer with no motivation for attending; whereas
clauses with only new information are incoherent, providing no grounding
point for the information to cohere. Thus, a plausible hypothesis that can be
formulated in discourse-structure terms of coherence and grounding is that
the chunks of given, or old information in the clause “serve to ‘ground’ the new
information to the already stored old information. Cognitively they furnish the
‘address’ or ‘label’ for the ‘storage locus’ (‘file’) in the episodic memory” (Givón
1990: 899) (cf. Section 2.2.1). If given information is to fulfill this function,
then, how many chunks are required within a unit of information flow in
connected discourse?
Givón (1990: 898) provides the following constraint on the flow rate of
given information: “A clause in connected discourse tends to contain at least
one/more than one chunk of old information.” (Here a “chunk” corresponds
to the lexical word that codes either the subject or object noun, the verb,
an adjective, or an adverb.) Givón (1984: 263–264) further proposes two
seemingly contradictory hypotheses. The first hypothesis states that “the more
chunks of old information the proposition has, the ‘faster’ will be the process
of integrating it [i.e. new information] into the coherence structure of the
discourse, because the search through the pre-existing network would proceed
on the basis of a larger number of ‘clues’, and will thus have less ‘indeterminacy’
in it”. Conversely, the second hypothesis states that “the more old information
chunks there are in a proposition, the more complex—thus ‘slower’—will
be the task of integrating it [i.e. new information] into the pre-existing
coherence network, since a larger number of addresses will have to be checked
systematically”. It seems that both of the hypotheses—the first on the speed,
and the second on the cost, of integrating new information—hold, depending
on the quantity of given information that has to be processed per proposition.
The two hypotheses would suggest that there is an optimal number of given
Information flow in spoken discourse 

concepts which would facilitate the integration of a chunk of new information


within a unit of spoken discourse, and which is presumably more than one, but
obviously not infinite.

.. Constraints on the flow rate of new information

Prior functional research on information-flow constraints in spoken commu-


nication has highlighted the limitations on the amount of new information that
can be transferred within a unit of discourse production. Specifically, some of
the key discourse researchers commonly proposed that the maximum amount
of new information within each unit in spoken discourse is only one. Appar-
ently, to activate a previously inactive concept exacts more cognitive effort than
any other kind of change in the activation states (i.e. the activation of a previ-
ously active or semi-active concept, or the deactivation of an active concept);
therefore, this is most likely to submit to constraint (see Section 2.3.2). The
rate of given information flow, on the other hand, has not been shown to be as
constrained as that of new information flow to date.
Givón (1975), for example, noted that speakers of several Bantu languages
employ a strategy of placing only one piece of new information per proposition
or VP. That is, he suggested, based on Bantu speakers’ discourse production,
that “there exists a strategy of information processing in language such that
the amount of new information per a certain unit of message-transaction is
restricted in a fashion—say ‘one unit per proposition’ ” (Givón 1975: 202–204).
This constraint has been further reformulated as the “one-chunk-per-clause
processing principle” in Givón (1984: 258), which states that “the majority
of sentences/clauses in connected discourse will have only one chunk—be it
a nominal, predicate (verb, adjective) or adverbial word/phrase—under the
scope of asserted new information. All other elements in the clause will tend to
be topical, background or presupposed old information”.
Chafe (1987) proposes that only one previously inactive or new concept
can be changed to the active state within a single IU. He calls this the “one
new concept at a time constraint”, by analogy to Pawley and Syder’s (1983)
“one clause at a time constraint” discussed in Section 2.4.6. This means that a
single IU can express no more than one new concept or idea. Chafe (1987: 32)
argues that this constraint on new information quantity per unit results
naturally from “the cognitive basis of an IU: the expression of a single focus of
consciousness. Such a focus can evidently contain no more than one previously
inactive concept”. This constraint is called the “one new idea constraint” in
Chafe (1994).
 Chapter 2

The effect of this constraint is illustrated in (2.18), a conversational narra-


tive excerpt composed of 12 IUs (Chafe 1987: 23), and (2.19), where the three
types of concepts, given, accessible, and new, verbalized in each of the IUs in
(2.18) are listed (Chafe 1987: 33).
(2.18) a. A: ++ And + he would come into cláss,
b. ++ a:t + uh: you know three or f
c. + precísely one minute after the hóur,
d. or something like thát,
e. ->++ a:nd he: + would + immédiately open his
- ->++ nótes up,
f. ++ in the front of the róom,
g. + and he st
h. and évery ++ évery lécture,
i. ++ áfter the fírst,
j. + stárted the same wáy.
k. This was + u:m at Wésleyan,
l. when Wesleyan was still ++ a mén’s school.
(2.19) GIVEN ACCESSIBLE NEW
a. he cláss would come into
b. at three or f
c. precísely one minute after the hóur,
e. he his nótes would immédiately open up
f. the róom in the front of
g. he st
h. évery lécture
i. áfter the fírst
j. stárted the same wáy
k. this was at Wésleyan,
l. Wesleyan was still a mén’s
school
According to Chafe (1987: 34), “the one new concept at a time constraint
dictates that each of the expressions in the righthand column [in (2.19) above]
must express a unitary concept”. He emphasizes that what he means by this is
“conceptual unitariness”, that is, whether the concept expressed by a sequence
of words is unitary or not; it is, therefore, important to realize that “there
is no consistent relation between the status of being a unitary concept and
the length or syntax of the verbalization of the concept” (Chafe 1987: 34).
Types of word sequences expressing a single new concept in this example
Information flow in spoken discourse 

include the following: copula + PP/NP constructions (e.g. was at Wesleyan,


was still a men’s (school)), verb-particle combinations (e.g. would immediately
open up), and adverbials (e.g. precisely one minute after the hour). Although
Chafe concludes that none of the IUs in his narrative contradicts his proposed
constraint, his interpretation of the “unitariness” seems to lack consistency.
For example, Chafe (1987: 35) treats verb-object combinations such as makes
a difference and give a lecture—which, according to Chafe, are lexicalized
phrases—as expressing a whole unitary new concept, instead of treating the
verb and the noun as verbalizing two new concepts (cf. Chafe 1994: Chapter 9;
Pawley & Syder 2000). He also treats verb-adverb combinations like started
the same way in (2.18j) as an expression of a unitary new concept. Moreover,
treating prepositions as expressing a new concept contradicts his notion that
it is only content words (which express ideas), not function words (which
express non-idea information), that are associated with activation cost, or
the given-accessible-new distinction (Chafe 1994: 80). (Note that in (2.7) in
Section 2.4.5, the preposition with is interpreted to express new information.)
It follows that a more precise characterization of the “unitariness”, as well as
a more useful subcategorization of “ideas”—not merely into states/events and
referents as discussed in Section 2.4.6—is needed, if the claim that an IU does
not ordinarily express more than one new concept is to be advocated (see
further discussion in Section 5.1).
One more constraint on new information quantity per unit, similar to
Givón’s (1975, 1984) and Chafe’s (1987), has been proposed by Du Bois
(1987) in his analysis of the Pear Film elicited narratives in Sakapulteko (cf.
Note 19 in this chapter). One of the most important discourse findings of
his study is that new nominal mentions follow a coherent and significant
pattern. His IU-based analysis has shown that substantial numbers of clauses
contained zero or one new-argument mention, with the former predominating
(occupying about 70%), but not even a single clause contained two new-
argument mentions. That is, speakers “avoid more than one new argument
per clause”; this Du Bois terms the “One New Argument Constraint”. (Here
the “clause”, which he calls “clause core”, is defined as a subset of the IU con-
sisting of a predicate and its core argument NPs, i.e. S (intransitive subject),
A (transitive subject), and O (transitive object); in his corpus, most IUs were
found to be simple clauses.) His analysis further revealed that speakers avoid
introducing a new referent in the A-role argument position; hence, the “Given
A Constraint” (i.e. avoid new A’s). The two pragmatic constraints, combined
together, mean that there is a maximum of one new referent per clause,
and this single new-argument mention typically appears in S or O roles,
 Chapter 2

but not in A. Du Bois (1987) also proposes what he calls the “One Lexical
Argument Constraint” (i.e. avoid more than one lexical argument per clause)
and the “Non-lexical A Constraint” (i.e. avoid lexical A’s). These grammatical
constraints, together with the pragmatic constraints given above, constitute
what he terms “Preferred Argument Structure” (PAS).
Many studies have provided evidence for the cross-linguistic applicabil-
ity of the quantity and role constraints of PAS. Among the most recent stud-
ies, Ashby and Bentivoglio (1993), Smith (1996), and Kärkkäinen (1996), for
example, have shown that PAS holds in spoken French and Spanish, mod-
ern Hebrew, and American English discourse, respectively (see also Du Bois
2003; Du Bois et al. 2003; O’Dowd 1990; Thompson 1997; Thompson &
Hopper 2001). Japanese spoken discourse, however, has not been investigated
extensively in terms of PAS or the preferred information structure of the IU
(cf. Iwasaki 1985; Section 2.4.5). It therefore remains to be explored whether
Japanese IUs follow the same pattern as those of the other languages in the
arrangement of given and new information and in the type and number of ar-
gument NPs to be expressed within an IU as well as a clause. We will address
this issue in Chapter 5.
Given the constraints on the flow rate of new information per discourse
unit proposed by Givón, Chafe, and Du Bois, it seems that we can safely say
that a segment of spoken discourse, or an IU is subject to a strong limitation on
how much new information it can express. This would mean that thought, or at
least language production, proceeds in terms of one activation of a previously
inactive idea at a time; and the speaker, and presumably the listener as well, is
able to handle no more than one new idea at a time (Chafe 1994: 109).
Chapter 3

Method of the study

This chapter first states the research questions and hypotheses of the present
study. It then describes the data base and the procedures of data transcription
and data analysis used in the study to test those hypotheses. This chapter
also presents the results of a preliminary quantitative analysis of intonation
units (IUs).

. Research questions and hypotheses

This study addresses the following three research questions (RQs). Predictions
are stated in terms of hypotheses:
RQ 1: What is the preferred syntactic structure of the IU in conversational
Japanese?
HYPOTHESIS 1.1: The syntactic structure of the IU in conversational Japanese
tends to be semi-clausal.
RQ 2: What is the preferred information structure of the IU in conversational
Japanese?
HYPOTHESIS 2.1: The information structure of the IU in conversational
Japanese tends to consist of one piece of new or given
nominal information.
RQ 3: What is the preferred functional structure of the IU in conversational
Japanese?
HYPOTHESIS 3.1: The functional structure of the IU in conversational
Japanese tends to consist of the ideational component only
or the ideational plus interactional components.
 Chapter 3

. Data base

The data that I collected and analyzed are face-to-face, two-party conversations
between native speakers of Tokyo Japanese. The participants are 16 male and 16
female university undergraduate or graduate students, all of whom are in their
twenties and different individuals. The average age of the 32 participants is
23.7. The data were provided by 8 male-male dyads and 8 female-female dyads.
The two participants in each of the 16 dyads are friends who have known each
other very well and had conversed many times before the recording was done
for this study. In order to obtain naturalistic conversational data, the researcher
did not set up the conversations; instead, the participants self-administered
the recordings. In other words, the data were collected in such a way that
the participants took the initiative in recording their own conversations; most
of the data were recorded in places such as restaurants, coffee shops, the
participants’ apartments, or dormitory rooms. Each of the 16 dyadic pairs of
participants audio-recorded their own conversations for about 60 minutes as
they conversed continuously in a casual and spontaneous manner. The last 45-
minute segment was then selected from each conversation and was transcribed
for analysis. Thus the data base for the present study consists of 720-minute
(i.e. 12-hour) dyadic Japanese conversations between same-sex friends.
A summary of the Japanese conversational data used for this study is given
in Table 3.1, where the participants’ names are pseudonyms. The boldfaced
codes F1 . . . . . F8 indicate the eight 45-minute female dyadic conversations,
whereas the boldfaced codes M1 . . . . . M8 indicate the eight 45-minute male
dyadic conversations.1

Table 3.1. Data for this study

Dyad Length Participants (Age)


of conversation
F1 45 min Saori (27) Yumi (27)
F2 45 min Yukiko (21) Misako (23)
F3 45 min Mari (22) Orie (23)
F4 45 min Akiko (24) Mika (26)
F5 45 min Masayo (22) Yasue (22)
F6 45 min Eri (26) Noriko (27)
F7 45 min Yukari (24) Tae (25)
F8 45 min Kumiko (21) Haruko (21)
Total 360 min N=16 (Average=23.8)
Method of the study 

Table 3.1. (continued)

Dyad Length Participants (Age)


of conversation
M1 45 min Masao (25) Shinji (26)
M2 45 min Takeshi (21) Naoto (22)
M3 45 min Shigeo (25) Jun (26)
M4 45 min Takuya (26) Hideki (26)
M5 45 min Osamu (24) Koichi (25)
M6 45 min Ken (21) Isao (22)
M7 45 min Ichiro (22) Seiji (22)
M8 45 min Hiroshi (22) Yutaka (21)
Total 360 min N=16 (Average=23.5)
TOTAL 720 min N=32 (AVERAGE=23.7)

Table 3.2 lists the major topics of the sixteen 45-minute conversations.
The topics are listed in the order that they were talked about in each of the
conversations.2

Table 3.2. Major topics of 16 conversations

Dyad Topics

F1 roommates, danger in America, riots in Los Angeles


F2 oriental food, Easter, roommates, politics in Japan
F3 sports, returnees, university majors, literature courses
F4 driving, M’s trip to the Middle East, cooking, marriage
F5 Japanese friends, summer vacation, Japanese restaurants
F6 conferences, research projects, parties, pierced earrings
F7 funerals, sociology, anthropology, UCLA departments
F8 movies, K’s visit to New York, English classes, assignments
M1 Asians, success in life, American food, traveling in Japan
M2 men vs. women, American girls, T’s American friends
M3 computers, TV programs, Japanese students, dormitories
M4 laws, honorifics, Japanese universities, northern part of Japan
M5 cooking, UCLA courses, registration, course units
M6 job-hunting, apartment life, Bruce Lee, playing tennis
M7 abnormal weather, classes, graduation, future plans
M8 Thailand, newspapers, cosmetic surgery, boxing, dating
 Chapter 3

. Data transcription

I transcribed the conversations using the notational conventions developted


by Andersen (1991) and Atkinson and Heritage (1984). In transcribing the
data, I put each IU on a separate line in the transcripts for coding purposes.3 I
also sequentially numbered the IUs in each of the 16 dyadic conversations. As
explained in Chapter 2, an IU, according to Chafe (1987: 22), is “a sequence of
words combined under a single, coherent intonation contour, usually preceded
by a pause”. Among Chafe’s (1980b) three criteria—intonational, hesitational,
and syntactic—for identifying IUs, I used the intonational criterion (i.e. the
presence of a unified intonation contour) as the single most reliable indicator
of an IU boundary in this study. This means that neither the presence of a
pause nor the clausal syntactic structure was counted as a necessary criterion
for determining the IU status (cf. Cruttenden 1986; Du Bois et al. 1992, 1993;
Edwards & Lampert 1993; Ladd 1996).
In this study I distinguished five intonation contours as signals of IU
boundaries.4 Thus the end of each IU in the transcripts was marked by one
of the following five symbols:
a. falling intonation (transcribed as .).
b. continuing intonation (transcribed as ,).5
c. rising intonation (transcribed as ?).
d. rise-fall intonation (transcribed as ^).
e. rise-fall-rise intonation (transcribed as ∼).

Spoken Japanese, like spoken English, exhibits a relatively high degree of align-
ment between intonation contours and the projection of turn completion.
The falling intonation in Japanese thus indicates finality or completeness or
idea closure, often used with declarative clauses. The continuing intonation
and non-question rising intonation, in contrast, express non-finality or in-
completeness. The rising pitch also indicates interrogativity, as in English.
The remaining two contours serve more or less “non-grammatical” functions
which appear to be peculiar to Japanese. The rise-fall pitch contour expresses
a “discoursal” meaning which involves the speaker’s appeal to the listener, or
the speaker’s immediate communicative expectations of the hearer; this con-
tour type functions to seek agreement or to impose the speaker’s opinion
on the hearer. The rise-fall-rise pitch contour, which expresses an “emotive”
or “attitudinal” meaning, shows the speaker’s doubt or dissatisfaction (see
Beckman & Pierrehumbert 1986; Bolinger 1989; Chafe 1988; Hirst & Di Cristo
1998; Iwasaki 1992, 2002; Pierrehumbert & Beckman 1988; Pierrehumbert &
Method of the study 

Hirschberg 1990; Sugito 1989, 1990; Vandepitte 1989; Venditti 2003; Ward &
Hirschberg 1985).

. Intonation units produced in 16 conversations

The total number of IUs collaboratively produced by the two interactants in


each of the 16 conversations and the average number of IUs produced per
minute are given in Table 3.3.
Table 3.3 shows that the transcription of the 12-hour conversation has
yielded a total of 22,209 IUs, and that, on average, the 32 participants produced
1,388.1 IUs per 45-minute conversation and 30.8 IUs per minute. It also
indicates that the female group produced more IUs than the male group. The
highest-rate production of IUs is found in F4, where 36.5 IUs were produced
per minute, whereas the lowest-rate production is found in M5, where 26.7 IUs

Table 3.3. Number of IUs produced in 16 conversations

Dyad Number of IUs Average number


(45 min) of IUs per min
F1 1,526 33.9
F2 1,598 35.5
F3 1,441 32.0
F4 1,643 36.5
F5 1,285 28.7
F6 1,345 29.9
F7 1,268 28.2
F8 1,302 28.9
Total 11,415
(Average) (1,426.9) (31.7)
M1 1,512 33.6
M2 1,410 31.3
M3 1,256 27.9
M4 1,203 26.7
M5 1,228 27.3
M6 1,432 31.8
M7 1,357 30.2
M8 1,396 31.0
Total 10,794
(Average) (1,349.3) (29.9)
TOTAL 22,209
(AVERAGE) (1,388.1) (30.8)
 Chapter 3

were produced per minute. The results indicate that the female dyads tend to
be associated with faster, more active production of IUs, and therefore, more
“dense” conversational discourse than the male dyads.

. Analysis of intonation units

.. Data for quantitative analyses of Japanese intonation units

From each of the 16 transcribed Japanese conversations, I selected a seg-


ment containing 100 substantive IUs with accompanying non-substantive IUs
(cf. Section 2.4.4; see Table 3.5). To test the three hypotheses, I then analyzed
the 1,600 substantive IUs in terms of the following:
a. syntactic structure (e.g. clausal vs. phrasal, full clausal vs. semi-clausal)
(cf. Chafe 1987, 1994; Iwasaki & Tao 1993).
b. information structure (e.g. new only, given plus new) (cf. Du Bois 1987,
2003; Du Bois et al. 2003).
c. functional structure (e.g. ideational only, ideational plus interactional)
(cf. Halliday 1985, 1994; Iwasaki 1993).

That is, the 1,600 substantive IUs taken from the 16 conversations constitute
the data base for the quantitative analyses of the structures of the Japanese IU
which will be dealt with in the following three chapters. The coding categories
used in the analyses are given in Sections 4.1, 5.1, and 6.1.

.. Topics of 16 conversational segments

The topics of the 16 conversational segments are listed in Table 3.4. The codes
F1, F2, . . . . . , and F8 indicate the segments containing 100 substantive IUs
selected from the conversations F1, F2, . . . . . , and F8, respectively. Likewise,
the codes M1, M2, . . . . . , and M8 indicate the 100-IU segments selected from
the conversations M1, M2, . . . . . , and M8, respectively.
The selected segments are interactive conversational portions instead of
narrative portions—portions where active conversational turn-taking is tak-
ing place, and thus the co-participants’ verbal contributions to discourse pro-
duction are fairly balanced between the two.6 In addition, the segments were
carefully selected in order for different topics and different types of conver-
sational discourse involving different activities to be represented in the data
(e.g. talking about the immediate environment in which the co-participants are
Method of the study 

Table 3.4. Topics of 16 conversational segments

Dyad Topic Dyad Topic

F1 danger in America M1 success in life


F2 Easter M2 American girls
F3 university majors M3 TV programs
F4 marriage M4 honorifics
F5 summer vacation M5 UCLA courses
F6 research projects M6 apartment life
F7 sociology M7 graduation
F8 movies M8 newspapers

situated, talking about what happened in the past, describing people or things,
exchanging opinions, and talking about future plans). That is, the data include
both the “immediate and displaced modes” of conversational language (Chafe
1994: Chapter 15), although most of the IUs, according to my observation, are
concerned with the latter rather than the former.

.. Substantive, regulatory, and fragmentary intonation units

In order to obtain 100 substantive IUs, I coded the IUs in each conversation
for four categories: substantive, regulatory, fragmentary, and uncodable. As
explained in Section 2.4.4, substantive IUs convey substantive ideas of events,
states, or referents. Fragmentary IUs are units which were begun but were not
completed for various reasons (e.g. because of the speaker’s “false starts” or
interruptions by the other interlocutor). Regulatory IUs function to regulate
conversational interaction or information flow. This category consists of the
following five subtypes: (a) textual subtype, which functions to indicate a
particular kind of linkage between IUs; (b) interactional subtype, which signals
the speaker’s attentiveness to what the interlocutor is saying and his/her
comprehension of it; (c) cognitive subtype, which indicates ongoing mental
processes on the part of the speaker; (d) validational (or evaluative) subtype,
which concerns itself with the speaker’s judgment of the validity, or the
speaker’s evaluation, of the information being conveyed; and (e) combination
type, which incorporates multiple regulatory functions within an IU. For
example, the regulatory IU ano dakara sa has three features: cognitive (marked
by the pause filler ano ‘uh’), textual (marked by the conjunction dakara ‘so’),
and interactional (marked by the final particle sa) (cf. Section 6.1). Finally,
uncodable IUs involve uncertainty or inaudibility of words, which is marked
by parentheses in the transcripts.
 Chapter 3

Table 3.5 lists the number of the four types of IUs contained in the 16 con-
versational segments (where [S]=substantive, [R]=regulatory, [F]=fragmentary,
and [X]=uncodable).
Table 3.5 indicates that the selected segments involving a sequence of 100
substantive IUs included, on average, 21.1 regulatory IUs, 1.2 fragmentary IUs,
and 1.4 uncodable IUs. This means that of all the IUs contained in the 16 seg-
ments, 81% are substantive, 17% are regulatory, 0.9% are fragmentary, and
1.1% are uncodable. The largest number of regulatory IUs were produced in F6,
where 24% of all the IUs in the segment are regulatory.7 Importantly, the results
indicate that the overwhelming majority of IUs that speakers produce in con-
versational interaction are substantive units that transmit ideational content.

Table 3.5. Number of four types of IUs in 16 conversational segments

Dyad S R F X Total
F1 100 20 1 1 121
F2 100 13 1 1 116
F3 100 23 3 1 127
F4 100 31 1 1 133
F5 100 15 1 3 119
F6 100 32 2 1 135
F7 100 19 1 0 120
F8 100 20 0 1 121
Total 800 173 10 9 992
(81) (17) (1.0) (0.9) (100%)
Average 100 21.6 1.3 1.1 124.0
M1 100 14 1 1 116
M2 100 20 0 3 123
M3 100 26 3 4 133
M4 100 16 1 1 118
M5 100 15 1 0 116
M6 100 22 1 2 125
M7 100 28 1 0 129
M8 100 24 1 2 127
Total 800 165 9 13 987
(81) (17) (0.9) (1.3) (100%)
Average 100 20.6 1.1 1.6 123.4
TOTAL 1,600 338 19 22 1,979
(81) (17) (0.9) (1.1) (100%)
AVERAGE 100 21.1 1.2 1.4 123.7
Chapter 4

Syntactic structure of the intonation unit


in conversational Japanese

This chapter, in answering RQ1 and testing Hypothesis 1.1, will attempt to
elucidate what syntactic types of intonation units (IUs) Japanese speakers
preferentially use in casual conversational discourse. It additionally responds
to the question that I raised in Section 2.5.1 concerning the clause vs. phrase
centrality debate. Thus the following three related questions are addressed in
this chapter (Matsumoto 1997b, 1998a, 2000b, 2001):
a. What is the preferred syntactic structure of the IU in conversational
Japanese?
b. Is the clause the exponent of the Japanese IU, as discussed in Section 2.4.6?
Or is the phrase the exponent of the Japanese IU, as discussed in
Section 2.5.1?
c. Is conversational Japanese highly fragmented, as illustrated in (2.14)
in Section 2.5.1? How often do Japanese speakers divide a clause into
phrasal IUs?

. Data coding

I first classified the substantive IUs into two general syntactic categories: clausal
and phrasal. An IU was coded as “clausal” if it contained a predicate; an IU
without a predicate was coded as “phrasal”. The clausal category was further
classified into two syntactic types: independent clausal IUs (Type 1) and clausal
IUs as part of a larger multi-IU clause (Type 2). Likewise, the phrasal category
was classified into two syntactic types: phrasal IUs as part of a larger multi-IU
clause (Type 3) and independent phrasal IUs (Type 4). These four syntactic
types were further subdivided into a total of 20 IU syntactic types. Listed and
defined below are Type 1–4 IUs and the 20 syntactic types that belong to them.
(Codes are given in square brackets; these will be used in subsequent sections.
 Chapter 4

The code-initial P as in [PVP] indicates “Predicate”, whereas the code-initial C


as in [CVP] indicates “part of a larger Clause”.)

Type 1: Independent clausal IU: a clausal IU that communicates a complete


proposition by itself. It can be either a “full clause” which consists of an overt
subject plus a predicate or a “semi-clause” which consists of a predicate only
without an overt subject. The predicate is of three types: verbal, nominal,1
and adjectival. Thus a full clause may consist of a verbal/nominal/adjectival
predicate and a subject NP as its associated core argument ([FC]); a semi-
clause may consist of a verbal predicate ([PVP]), a nominal predicate ([PNP]),
or an adjectival predicate ([PAP]) only. This study classified not only main
clauses but also subordinate clauses into this category. Type 1 IUs consist of
the following four subtypes:
[FC] = overt subject + verbal/nominal/adjectival predicate
[PVP] = Ø-subject + verbal predicate
[PNP] = Ø-subject + nominal predicate
[PAP] = Ø-subject + adjectival predicate

Type 2: Clausal IU as part of multi-IU clause: a clausal IU that can com-


municate a complete proposition only in conjunction with the preceding
or following clausal/phrasal IU(s). As in Type 1, it can be either a full
clause consisting of an overt subject plus a verbal/nominal/adjectival predicate
([CFC]) or a semi-clause consisting of a verbal/nominal/adjectival predicate
only ([CVP]/[CNP]/[CAP]). Typically, this group comprises clauses involv-
ing such verbs as yuu ‘say’ and omou ‘think’ and their clausal objects. Type 2
IUs, which are clausal IU elements of a larger multi-IU clause, consist of the
following four subtypes:
[CFC] = overt subject + verbal/nominal/adjectival predicate
[CVP] = Ø-subject + verbal predicate
[CNP] = Ø-subject + nominal predicate
[CAP] = Ø-subject + adjectival predicate

Type 3: Phrasal IU as part of multi-IU clause: a phrasal IU that, together


with the preceding or following clausal/phrasal IU(s), constitutes a multi-IU
clause in which multiple IUs collectively convey a complete proposition. The
phrase is of five types: VP (verb phrase), NP (nominal phrase), AP (adjectival
phrase), AvP (adverbial phrase), and XP (complex phrase consisting of phrases
of different syntactic categories, e.g. NP + AvP). The NP was further divided
Syntactic structure of the intonation unit in conversational Japanese 

into two types: argument and predicate; similarly, the AP was classified into
attributive and predicative.2 Type 3 IUs are phrasal IU elements of a larger
multi-IU clause; they consist of the following seven subtypes:
[CVPp] = verb phrase
[CNPa] = argument NP
[CNPp] = predicate NP
[CAPa] = attributive AP
[CAPp] = predicative AP
[CAvP] = adverbial phrase
[CXP] = mixed phrase (e.g. [CNPa + CAvP])

Type 4: Independent phrasal IU: a phrase that is not integrated into the
clausal structure and that as a clause-external element, mainly functions to
perform “peripheral” work rather than the “core” task of communicating
propositional content, unlike Types 1–3. For example, left/right-dislocated
phrases and copied/restated postposed phrases belong to this group (see
Section 4.5.2 for detailed discussion). Type 4 “detached” phrasal IUs consist
of the following five subtypes:
[VP] = verb phrase
[NP] = noun phrase
[AP] = adjective phrase
[AvP] = adverbial phrase
[XP] = mixed phrase (e.g. [NP + AvP])
In addition, I used six codes as superordinate codes, namely, [FC], [VP],
[NP], [AP], [AvP], and [XP]. These boldfaced codes represent IU syntactic
types based on grammatical categories. For example, as shown below, [FC]
(full clausal IUs) consists of two types, i.e. [FC] and [CFC]; [VP] (VP IUs)
consists of four types, i.e. [PVP], [CVP], [CVPp], and [VP]; [NP] (NP IUs)
consists of five types, i.e. [PNP], [CNP], [CNPa], [CNPp], and [NP].
[FC] = [FC] + [CFC]
[VP] = [PVP] + [CVP] + [CVPp] + [VP]
[NP] = [PNP] + [CNP] + [CNPa] + [CNPp] + [NP]
[AP] = [PAP] + [CAP] + [CAPa] + [CAPp] + [AP]
[AvP] = [CAvP] + [AvP]
[XP] = [CXP] + [XP]
In this study I defined a clause as “a unit of discourse which consists of
a predicate and its associated core arguments (i.e. a subject and an object,
 Chapter 4

which may or may not be overtly expressed) and adjuncts (i.e. locative,
temporal, and manner adverbials performing circumstantial functions, which
are optionally present)” (Chafe 1980b; Dixon 1979; Du Bois 1987). The
arguments and adjuncts may be in the pre-predicate position (unmarked
word order) or in the post-predicate position (marked word order involving
postposing) (cf. Section 1.3). I also defined a clause as “a propositionally
complete discourse unit which codes some state or event” (Givón 1990). That
is, the basic assumption in my coding is that a clause consisting of at least a
predicate conveys a complete proposition.3 Thus, when a core argument or an
adjunct appeared independently as a “topic” IU, marked by the so-called “topic
marker” wa (or its informal version tte) (cf. Hinds et al. 1987), I considered it
to be within the clausal structure, coding it as a Type 3 phrasal IU (see (4.6b)
below). On the other hand, when a wa-marked NP IU functioned as a “base-
generated”, “genuine” topic4 (Shibatani 1990) which would constitute neither a
core argument nor an adjunct, I considered it to be outside the clausal domain,
coding it as an independent NP IU, namely, Type 4 (see (4.19a) below).
As noted above, a “full clause” is defined to consist of an overtly expressed
subject plus a predicate, whereas a “semi-clause” consists of a predicate only
without an explicit subject. This means that while they differ in whether the
subject is overtly expressed or assumed by the co-participants, both clause
types may contain an overtly expressed object. I coded phrasal IUs as part of a
multi-IU clause only when they were produced by the same speaker and only
when they were adjacently produced (i.e. when they were not interspersed with
other IUs in the transcripts). This means that phrasal IUs which have adjacently
been produced by the co-participants such that they constitute a collaboratively
completed clause were not counted as constituting a multi-IU clause; rather
they were considered independent (see (4.3)). This also means that phrasal IUs
with intervening IUs/propositions which would otherwise constitute a multi-
IU clause were coded as independent IUs (see (4.16) below). Further, I treated
phrasal IUs which have been produced adjacently to a propositionally complete
single-IU or multi-IU clause such that an element of that clause is copied or
restated as being outside the clausal domain (see (4.4)). Accordingly, as will
be shown in Section 4.5.2, Type 4 IUs typically includes, besides “genuine”
topics, copied/restated postposed phrases, left/right-dislocated phrases, and
repeated phrases.
Observed IU examples are provided in (4.1)–(4.8). Examples (4.1)–(4.4)
illustrate Type 1 or Type 4 independent IUs. (In the examples below, assigned
codes are given in square brackets.)
Syntactic structure of the intonation unit in conversational Japanese 

(4.1) S: ano hito insei da yo ne.@ [FC]


that person graduate student cop fp fp
‘He (is) a graduate student, isn’t he?’ (M7)5
(4.2) H: ni kai gurai atta yo. [PVP]
two times about meet-past fp
‘(I) met (her) about two times.’ (F8)
(4.3) a. E: sangatsu no, [AP]
March gen
‘of March’
b. N: ++juukyuu? [PNP]
nineteen
‘(Is it) nineteenth?’ (F6)
(4.4) a. T: sore yori mo seikaku ga chigau. [FC]
it comp emp characteristics nom differ
‘Rather than that, (their) characteristics differ.’
b. +seikaku ga. [NP]
characteristics nom
‘(their) characteristics’ (M2)

Examples (4.1) and (4.2) illustrate Type 1 independent clausal IUs. The IU
(4.1) is a full clause which consists of the overt subject NP ano hito ‘he’ and
the nominal predicate insei da ‘(is) a graduate student’. The IU (4.2) is a semi-
clause which consists of the verbal predicate atta ‘met’, where the subject ‘I’ and
the direct object ‘her’ are assumed, taking zero forms. Both (4.1) and (4.2), as
single-IU clauses, independently convey a complete proposition.
Examples (4.3a) and (4.4b), on the other hand, illustrate Type 4 indepen-
dent phrasal IUs. The sequence in (4.3) is an example of the collaboratively
completed clause. Given that the two speakers, E and R, are involved, I did not
code (4.3) as a multi-IU clause; rather I coded (4.3a) sangatsu no ‘of March’
as an independent attributive AP IU (i.e. [AP]), and (4.3b) juukyuu? ‘(Is it)
nineteenth?’ as an independent semi-clause which solely consists of the nom-
inal predicate (i.e. [PNP]). The IU (4.3a), uttered by speaker E but convey-
ing only a partial proposition, is made propositionally complete by speaker
R’s immediate suppliance of (4.3b). (If (4.3a) and (4.3b) had been uttered by
the same speaker, they would be coded as [CAPa] and [CNPp], respectively.)
The IU (4.4b) seikaku ga ‘(their) characteristics’ is an example of the copied
post-predicate nominal phrase. Given that the NP IU (4.4b) cannot be inte-
grated into the immediately preceding full clause (4.4a) ‘(their) characteristics
differ’, but simply repeats the subject NP already uttered in (4.4a), (4.4b) is a
 Chapter 4

clause-external independent phrasal NP. (For discussion of functional subtypes


of Type 4 “detached” NP IUs, see Section 4.5.2.)
Examples (4.5)–(4.8) illustrate multi-IU clauses, which involve Type 2
and Type 3 IUs. In each case, two IUs collectively constitute a propositionally
complete clause.
(4.5) a. M: kon shuu no nichiyoobi? [CNPa]
this week gen Sunday
‘this Sunday’
b. +iisutaa da yo ne.@ [CNPp]
Easter cop fp fp
‘(is) Easter’ (F2)
(4.6) a. K: atsui kara:. [CAP]
hot because
‘because (it is) hot’
b. +ima wa. [CAvP]
now top
‘now’ (M5)
(4.7) a. O: kocchi no daigaku taihen da kara yame na:, [CVP]
here gen university tough cop because stop-imp fp
‘because universities here (are) tough, (you’d better) stop
(applying)’
b. to ka itta n da kedo, [CVP]
qt q say-past nml cop though
‘though (I) told (him) that’ (F3)
(4.8) a. S: owari desho? [CNPp]
end tag
‘(is coming to) an end, isn’t it?’
b. +hotondo are. [CXP]
almost that
‘that (drama) almost’ (M3)

The IUs (4.6a) and (4.7a, b) exemplify Type 2 IUs—clausal IUs as part of multi-
IU clauses, whereas (4.5a, b), (4.6b), and (4.8a, b) exemplify Type 3 IUs—
phrasal IUs as part of multi-IU clauses. The sequence in (4.5) is a multi-IU full
clause which is composed of the subject argument NP (4.5a) ‘this Sunday’ and
the predicate NP (4.5b) ‘Easter’; these two nominal phrasal IUs jointly convey
a full proposition. Example (4.8) is another multi-IU full clause; this sequence
is comprised of the predicate NP IU (4.8a) and the post-predicate mixed
phrasal IU (4.8b) (which consists of the adverbial phrase hotondo ‘almost’
Syntactic structure of the intonation unit in conversational Japanese 

followed by the subject NP are ‘that’). In (4.6), the adjectival clausal IU


(4.6a) ‘because (it is) hot’ and the post-predicate adverbial phrase (4.6b) ‘now’
collectively constitute a multi-IU semi-clause which communicates a complete
proposition. As explained above, I coded the wa-marked phrase (4.6b) ima
wa—which functions as an adjunct of the adjectival predicate atsui—as part of
the multi-IU clause, not as a clause-external “genuine” topic. Example (4.7)
is another multi-IU semi-clause. This sequence consists of two subjectless
semi-clausal IUs, i.e. (4.7a) which functions as the clausal object of the verb
‘say’ and (4.7b) which functions as the main clause; these two [CVP] IUs
collectively constitute a full proposition. Note that kocchi no daigaku taihen da
kara ‘because universities here (are) tough’ contained in (4.7a) is an intra-IU
subordinate clause.6 (For more observed IU examples, see sections to follow.)
To sum up, of the four major IU syntactic types used in the present study,
Type 1–3 IUs, being integrated into the clausal structure and conveying com-
plete propositions either singly or collectively, are centrally concerned with the
communication of propositional content. These IUs constitute proposition-
ally “essential” or “core” part of the conversational language; these are the IUs
that conversational Japanese is expected to consist of, if it is to be maximally
economical in conveying propositions. Type 4 IUs, being outside the clausal
domain, are not so much concerned with the communication of ideational
content as with different types of “peripheral” work, as I will demonstrate
shortly in Section 4.5.2. These IUs more or less constitute propositionally “re-
dundant” part of the conversational language (cf. Dik 1989). That is, substan-
tive IUs can be classified into two types which seem to differ in the degree
to which they contribute to the communication of propositional content and
its further development. This notion may be equated with Firbas’ (1992) CD
(cf. Note 14 in Chapter 2). We can thus say that Type 1–3 IUs are high in CD,
whereas Type 4 IUs are low in CD. The former are central, whereas the lat-
ter are peripheral, to the communication of ideas. In brief, Japanese conver-
sational language, and possibly conversational language in general, consists of
both “core/essential” and “peripheral/redundant” substantive IUs.

. Distribution of IU syntactic structure types

.. Clausal vs. phrasal intonation units

Table 4.1 displays the distribution of the 20 IU syntactic structure types as well
as the four major syntactic types.
 Chapter 4

Table 4.1. Distribution of IU syntactic structure types

Clausal IU Phrasal IU
Independent Part of multi- Part of multi- Independent Total
IU IU clause IU clause IU
[FC] 263 [CFC] 40 [FC] 303
(19%)
[PVP] 392 [CVP] 106 [CVPp] 34 [VP] 7 [VP] 539
(34%)
[CNPa] 94 [NP] 112 [NP] 420
[PNP] 174 [CNP] 17 [CNPp] 23 (26%)
[CAPa] 18 [AP] 14 [AP] 131
[PAP] 81 [CAP] 7 [CAPp] 11 (8%)
[CAvP] 90 [AvP] 82 [AvP] 172
(11%)
[CXP] 27 [XP] 8 [XP] 35
(2%)
Total 910 170 297 223 1,600
(57%) (11%) (18%) (14%) (100%)

Table 4.1 shows that of the 1,600 substantive IUs produced in the 16
conversational segments, (a) 68% are clausal IUs, whereas 32% are phrasal
IUs; (b) 71% are independent clausal or phrasal IUs, whereas 29% are clausal
or phrasal IUs constituting part of multi-IU clauses; and (c) 86% are “core”
IUs which are centrally concerned with the communication of propositions
(Type 1–3 IUs), whereas 14% are “peripheral”, “redundant” phrasal IUs placed
outside the clausal structure (Type 4 IUs). It also indicates that of the four
major IU syntactic types, Type 1 independent clausal IUs were most frequently
produced, accounting for 57% (see Figure 4.1). In addition, of the six IU
syntactic types based on grammatical categories, IUs consisting of verbal
predicates or verb phrases (i.e. [VP]) and IUs consisting of nominal predicates
or nominal phrases (i.e. [NP]) were most frequently produced, occupying 34%
and 26%, respectively. Full clausal IUs as independent IUs or as elements of
multi-IU clauses (i.e. [FC]) were third most frequently produced, accounting
for 19% (see Figure 4.2). We can also see from Table 4.1 that of the 20
IU syntactic types, the speakers produced [PVP] IUs (independent semi-
clausal IUs consisting of verbal predicates only) most frequently (25% of
the 1,600 IUs).7 This is followed in frequency by [FC] IUs (independent full
clausal IUs consisting of overt subjects plus predicates) (16%) and [PNP] IUs
(independent semi-clausal IUs consisting of nominal predicates only) (11%)
Syntactic structure of the intonation unit in conversational Japanese 

1000
900

800
700
600
Frequency

500
400

300
200
100
0
Independent

Independent
part of multi-IU

part of multi-IU

phrasal IU
clausal IU

Phrasal IU as
Clausal IU as

clause

clause

IU syntactic type

Figure 4.1. Distribution of four major IU syntactic types

(see Figure 4.3). (For further analysis of the full clauses vs. semi-clauses, see
Section 4.4.)
Table 4.2 shows the distribution of the eight clausal IU types. Table 4.3, on
the other hand, shows the distribution of the 12 phrasal IU types.
Table 4.2 indicates the following. Of Type 1 independent clausal IUs,
[PVP] IUs (subjectless semi-clausal IUs with verbal predicates) were produced
most frequently, accounting for 43%. Of Type 2 clausal IUs as part of

Table 4.2. Distribution of eight clausal IU types

Type 1 IU N % Type 2 IU N %
[FC] 263 29 [CFC] 40 24
[PVP] 392 43 [CVP] 106 62
[PNP] 174 19 [CNP] 17 10
[PAP] 81 9 [CAP] 7 4
Total 910 100 Total 170 100
 Chapter 4

600

500

400
Frequency

300

200

100

0
FC VP NP AP AvP XP
IU syntactic type

Figure 4.2. Distribution of six IU syntactic types

multi-IU clauses, similarly, the type which was most frequently produced is
[CVP] (subjectless semi-clausal IUs with verbal predicates); this accounted
for 62%. Table 4.3 likewise indicates the following. Of Type 3 phrasal IUs
as part of multi-IU clauses, the speakers most frequently produced [CNPa]
IUs (argument NP IUs) (32%) and [CAvP] IUs (adverbial IUs) (30%). Of
Type 4 independent phrasal IUs, [NP] IUs were produced most frequently,
accounting for 50%.

.. Preferred syntactic structure of the Japanese intonation unit

The main results regarding the preferred IU syntactic structure types are
the following: (a) of the 1,600 substantive IUs produced by the 32 speakers,
68% are clausal, as opposed to phrasal, IUs; (b) of the four major IU syn-
tactic types, independent clausal IUs which communicate complete proposi-
tions were most frequently produced (57%); (c) of the 20 IU syntactic types,
independent semi-clausal IUs consisting of verbal predicates without overt
Syntactic structure of the intonation unit in conversational Japanese 

400

350

300

250
Frequency

200

150

100

50

0
FC
PVP
PNP
PAP
CFC
CVP
CNP
CAP
CVPp
CNPa
CNPp
CAPa
CAPp
CAvP
CXP
VP
NP
AP
AvP
XP
IU syntactic type

Figure 4.3. Distribution of 20 IU syntactic types

subjects (i.e. [PVP] IUs) were found to be the most preferred IU type (25%);
and (d) of the six syntactic types based on grammatical categories, IUs con-
sisting of verbal predicates or verb phrases (i.e. [VP] IUs) were most fre-
quently produced (34%). In sum, the results indicate that the Japanese con-

Table 4.3. Distribution of 12 phrasal IU types

Type 3 IU N % Type 4 IU N %
[CVPp] 34 11 [VP] 7 3
[CNPa] 94 32 [NP] 112 50
[CNPp] 23 8
[CAPa] 18 6 [AP] 14 6
[CAPp] 11 4
[CAvP] 90 30 [AvP] 82 37
[CXP] 27 9 [XP] 8 4
Total 297 100 Total 223 100
 Chapter 4

versational co-participants produced preferentially clausal IUs, typically inde-


pendent semi-clausal IUs which solely consist of verbal predicates.
Let us now compare the results of the present study with those of Iwasaki
and Tao (1993) discussed in Section 2.5.1. The fact that the proportion of
clausal IUs was found to be higher in this study (68% out of the 1,600 IUs in
this study vs. 45% out of the 756 IUs in their study) appears to be attributable
to the difference in coding between the two studies. That is, I coded IUs with
verbal, nominal, and adjectival predicates as clausal, whereas they coded only
IUs with verbal predicates as “clausal”. Thus, in their study, IUs consisting of
nominal/adjectival predicates were coded as “non-clausal”, which apparently
led to the lower proportion of clausal IUs. As I will show shortly in Section 4.4,
the percentage of clausal IUs with verbal predicates obtained in this study
(N=663, 41% out of the 1,600 IUs) is nearly commensurate with that of
“clausal” IUs (45%) in Iwasaki and Tao (1993). Likewise, the percentages of full
clauses vs. semi-clauses in the two studies are comparable with each other. In
this study, as will be shown in Section 4.4, out of the 1,080 clausal IUs, 28% are
full clauses, whereas 72% are semi-clauses; in their study, out of the 343 clausal
IUs, 24% are full clauses, whereas 76% are semi-clauses. The two studies also
exhibit comparable proportions of NP IUs (26% out of the 1,600 IUs in this
study vs. 22% out of the 756 IUs in their study).
To be noted especially is the fact that my Japanese data exhibited “unex-
pectedly” such a high degree of correspondence between the IU and the clause,
i.e. 68% of the 1,600 IUs were found to be clausal. This is clearly compara-
ble to the proportion of clausal IUs (about 60–70%) that Chafe (1988, 1994)
found in his conversational English data. Some caution is needed, however, in
interpreting this result. For one thing, by “clausal” IUs Chafe seems to mean
“complete single clauses”, and these correspond to Type 1 independent clausal
IUs in the present study (which occupy 57%). We can, therefore, reinterpret the
finding by stating that the proportion of single-clause substantive IUs in con-
versational Japanese was found to be slightly lower than that in conversational
English. That is, the results suggest that the IU correlates with the clause (viz.
the propositionally complete independent clause) to a lesser degree in Japanese
than in English. Relatedly, the existence of nominal and adjectival predicate
clauses—without linking verbs but only with copulas optionally present—in
Japanese, but not in English, appears to make commensurable cross-linguistic
comparisons difficult to establish. Clearly, the question of how to code these
two types of clauses which are peculiar to Japanese would be crucial to the
issue of the proportion of clausal vs. phrasal IUs and the “clause centrality pro-
posal” discussed in Section 2.4.6. Nevertheless, if we interpret both indepen-
Syntactic structure of the intonation unit in conversational Japanese 

dent and non-independent clausal IUs as “clausal”, and given their substantial
combined proportion out of the 1,600 substantive IUs, i.e. 68%, it seems safe
to conclude that the clause is the typical IU type. In other words, the clause is
the grammatical exponent of the IU in conversational Japanese, as in the case
of spoken English.

. Distribution of post-predicate phrases among IU syntactic types

The 1,600 substantive IUs contained not only pre-predicate elements (which
represent basic word order) but post-predicate elements (which represent
marked word order) as well (cf. Section 1.3; Note 3 in Chapter 1). I have
coded the frequency of post-predicate, or postposed phrases which occurred
in the 16 selected segments, in order to see what proportion of the IUs
involved postposing and to find out patterns in the speakers’ production of
those marked word order constructions. Table 4.4 presents the distribution of
post-predicate phrases among 10 IU syntactic types (the remaining 10 types
involved no postposings). The coded postposed phrases include nominals,
adjectives, adverbials, and mixed phrases.8
Table 4.4 indicates that of the 1,600 IUs, 7.7% (N=123) involved postpos-
ing. The postposed phrases listed in Table 4.4 can be grouped into three types,
depending on the relationship between the postposed phrase and the IU/clause

Table 4.4. Distribution of post-predicate phrases among IU syntactic types

IU syntactic type N %
[FC] 23
[PVP] 7
[CVP] 1
Type 1-Total 31 25
[CNPa] 26
[CAPa] 6
[CAvP] 32
[CXP] 2
Type 2-Total 66 54
[NP] 20
[AP] 1
[AvP] 5
Type 3-Total 26 21
TOTAL 123 100
 Chapter 4

in which it occurs (cf. Clancy 1982; Hinds 1976; Matsumoto 1995a, 1997c;
Ono & Suzuki 1992; Shibatani 1990). Examples are given in (4.9)–(4.11), where
postposed NPs are boldfaced.
Type 1: postposed phrases constituting the final part of clausal IUs, which may
be independent single-IU clauses, as in (4.9), or clausal IU elements
of multi-IU clauses. (Those constituting the final part of phrasal
predicate IUs such as [CVPp] were not found in the data.)
Type 2: postposed phrasal IUs constituting elements of multi-IU clauses, as in
(4.10b) and (4.10c).
Type 3: independent postposed phrasal IUs external to the clausal structure,
i.e. copied/restated postposed phrases, as in (4.11c).

(4.9) H: chishiki nasa sugiru kara na ore. [FC]


knowledge exist-neg exceed because fp I
‘Because have too little knowledge, I.’ (M8)
(4.10) a. J: deteta. [CVPp]
appear-past-sta
‘was (there on the magazine)’
b. kiji ga. [CNPa]
article nom
‘an article’
c. juu nen ai no. [CAPa]
ten year love gen
‘on the Ten-year love (=TV drama title)’ (M3)
(4.11) a. K: ((laugh)) ano roozu tte nanka, [CNPa]
that Rose top pf
‘Rose’
b. ano atakku # sare yasusoona kanji jan. [CVPp]
pf attack-pass easy-looking feeling tag
‘uh it looks like gets easily attacked (by men)’
c. ano roozu tte. [NP]
that Rose top
‘Rose’ (F8)

Postposings are thus of two types: clause-internal (Type 1 and Type 2) and
clause-external (Type 3). Clause-internal postposed phrases, integrated into
the clausal structure, constitute part of single-IU clauses (Type 1), or part of
clausal IU elements (Type 1) or phrasal IU elements (Type 2) of multi-IU
clauses. As clause-internal elements, they participate in the communication of
propositions. Clause-external postposed phrases (Type 3), placed outside the
Syntactic structure of the intonation unit in conversational Japanese 

clausal domain, only repeat or restate phrases already uttered in the immedi-
ately preceding clause. They take the form of copied/restated post-predicate
phrases or right-dislocated phrases, with resumptive pronouns remaining in
the preceding clause in the case of right-dislocation (see detailed discussion in
Section 4.5.2). The two types of postposings differ not only in terms of the IU
syntactic structure but also functionally, that is, they seem to serve different
discourse functions in conversational Japanese. The clause-internal postposing
can be seen as linked to the speaker’s focused attention on the information that
is most urgently sought within the flow of conversation, which in turn triggers
the backgrounding of the postposed information. Apparently, we can relate the
production of clause-internal postposed phrases to the dynamic foregound-
background alternation in connected discourse (cf. Section 2.2). The clause-
external postposed phrases, on the other hand, can be seen to be added for pur-
poses such as emphasis, repair, and further specification of already expressed
elements; this suggests that Type 3 postposing entails the speaker’s conscious
intentional strategic choices (cf. Clancy 1982; Fox et al. 1996; Kuno 1978a;
Maynard 1989; Saeki 1975; Schegloff 1979; Schegloff et al. 1977; Takami 1995;
Teramura 1984).
In sum, Type 1 and Type 2 post-predicate phrases, although they may be
backgrounded or defocused, constitute “core” elements of single-IU or multi-
IU clauses; they directly participate in the construction of propositions. Type 3
post-predicate phrases, in contarst, only repeat or restate “redundantly” the in-
formation already given in the immediately preceding clause, thus functioning
principally as a device of emphasis or reinforcement. Table 4.4 indicates that
79% of the post-predicate phrases involved clause-internal postposing, and
that 54% appeared as phrasal IU elements of multi-IU clauses (Type 2).
Table 4.5 presents the proportion of IUs which involved postposing by IU
syntactic type.
Table 4.5 indicates that the adverbial phrases, attributive adjective phrases,
and nominal phrases as part of multi-IU clauses exhibited the highest rate
of postposings ([CAvP]=36%, [CAPa]=33%, [CNPa]=28%). In accord with
this, the phrasal IUs as constituents of multi-IU clauses were most highly
postposed—22% of them constituted post-predicate elements. (By compar-
ison, only 3% of the clausal IUs and 12% of the independent phrasal IUs
involved postposings.) The speakers’ preferential production of postposed
phrases as phrasal IU elements of multi-IU clauses deserves further ex-
amination.
Detailed discussion of the clausal IUs, NP IUs, AP IUs, AvP IUs, and XP
IUs follows, focusing on the types and functions of NP IUs.
 Chapter 4

Table 4.5. Proportion of IUs involving postposing by IU syntactic type

Total number of IUs IUs involving postposing


IU syntactic type N N %
[FC] 263 23 8.7
[PVP] 392 7 1.8
[CVP] 106 1 0.9
[CNPa] 94 26 27.7
[CAPa] 18 6 33.3
[CAvP] 90 32 35.6
[CXP] 16 2 12.5
[NP] 112 20 17.9
[AP] 9 1 11.1
[AvP] 82 5 6.1
Total 1,600 123 7.7

Table 4.6. Proportion of full clauses vs. semi-clauses

Independent Clausal IU as part Clausal IU


clausal IU of multi-IU clause Total
Clause type N % N % N %
Full clause 263 29 40 23 303 28
Semi-clause 647 71 130 77 777 72
Total 910 100 170 100 1,080 100

. Clausal intonation units: Full clauses vs. semi-clauses

Table 4.6 displays the proportions of full clauses vs. semi-clauses within Type 1
independent clausal IUs and Type 2 clausal IUs as part of multi-IU clauses (see
also Figure 4.4).
Table 4.6 shows that of the total number of clausal IUs (N=1,080), 72%
are semi-clauses, whereas 28% are full clauses. It indicates that in both Type 1
and Type 2 clausal IUs, the majority are subjectless semi-clauses as opposed
to full clauses, and they account for about 70–80%. We can also see from
Table 4.6 that of the 1,080 clausal IUs, 84% are independent single-IU clauses
communicating complete propositions, whereas 16% are clausal IUs constitut-
ing part of multi-IU clauses. The results additionally indicate that of the 1,600
IUs, (a) 49% are semi-clauses ([PVP] + [PNP] + [PAP] + [CVP] + [CNP] +
[CAP]); (b) 19% are full clauses ([FC] + [CFC]); and (c) 40% are independent
single-IU semi-clauses ([PVP] + [PNP] + [PAP]) (see Table 4.1).
Syntactic structure of the intonation unit in conversational Japanese 

800

700

600

500
Frequency

Full clause
400
Semi-clause
300

200

100

0
Independent

part of multi-IU

Clausal IU
clausal IU

Clausal IU as

Total
clause

Clausal IU type

Figure 4.4. Proportion of full clauses vs. semi-clauses

Further coding of the corpus has shown the following regarding the
structure of the full clauses as well as semi-clauses: (a) of the 263 [FC] IUs,
51% (N=135) consist of S(subject)+VP; (b) of the 40 [CFC] IUs, 75%
(N=30) consist of S(subject)+VP; (c) of the 647 independent semi-clausal IUs
([PVP] + [PNP] + [PAP]), 61% (N=392) consist of VP; and (d) of the 130
semi-clausal IUs as elements of multi-IU-clauses ([CVP] + [CNP] + [CAP]),
82% (N=106) consist of VP. This means that 55% of the full clauses, 64%
of the semi-clauses, and 61% of the clausal IUs contained verbal predicates.
This demonstrates the speakers’ notable preference for verbal over nomi-
nal/adjectival predicates. (The clausal IUs will also be discussed in terms of
transitive-intransitive bivalency in Section 5.5.)
In sum, the results presented above reveal that the Japanese interactants
prefer to produce IUs which are syntactically semi-clausal lacking explicit
subjects, especially propositionally complete independent semi-clausal IUs
which consist of verbal predicates. This finding is in accordance with Iwasaki
and Tao (1993), as discussed in Section 4.2.2.
 Chapter 4

180

160

140

120
Frequency

100

80

60

40

20

0
PNP CNP CNPa CNPp NP
NP IU type

Figure 4.5. Distribution of five types of NP IUs

. NP intonation units

.. Distribution of five types of NP intonation units

We have seen in Section 4.2 that the NP IUs (i.e. [NP]) accounted for 26% of
the 1,600 substantive IUs. We have also seen that the NP IUs accounted for
18% of the 1,080 clausal IUs and 44% of the 520 phrasal IUs (see Table 4.1).
Table 4.7 displays the distribution of the five types of NP IUs, i.e. [PNP],
[CNP], [CNPa], [CNPp], and [NP] (see also Figure 4.5).

Table 4.7. Distribution of five types of NP IUs

NP IU type N %
[PNP] 174 41
[CNP] 17 4
[CNPa] 94 22
[CNPp] 23 6
[NP] 112 27
Total 420 100
Syntactic structure of the intonation unit in conversational Japanese 

Table 4.7 shows that of the 420 NP IUs, (a) 55% are phrasal IUs, whereas
45% are clausal IUs; (b) 51% are NPs functioning as predicates, whereas 49%
are argument NPs or extra-clausal NPs; and (c) 68% are independent clausal or
phrasal NP IUs, whereas 32% are elements of multi-IU clauses. It also indicates
that of the five NP IU types, [PNP] (independent single-IU semi-clauses with
nominal predicates only) is the most frequently produced type, accounting for
41%. This is followed in frequency by [NP] (clause-external “detached” NP
IUs) (27%) and [CNPa] (argument NP IUs as constituents of multi-IU clauses)
(22%). (Of the 94 [CNPa] IUs, further coding has shown, 78% occurred within
the multi-IU full clauses, functioning as subject or object NPs, whereas 22%
constituted object NPs of the multi-IU semi-clauses.)
The occurrence of NP IUs in Japanese conversation is illustrated in (4.12).
In this excerpt, the female co-participants K and H are talking about one of
their common friends whom H claims to be too gaily dressed.
(4.12) a. H: nihon de isoona kakkoo.= [PNP]
Japan loc exist-likely appearance
‘(It’s) an appearance (which is) likely to be found in Japan.’
b. K: =ikeike. [PNP]
ikeike
‘(It’s) ikeike.’
c. H: +ikeike. ((laugh)) [NP]
‘ikeike’
d. K: ((laugh)) ano ko ga, [NP]
that girl nom
‘she’
e. [ dakara, ]
so
‘so’
f. H: [ano hito] shakaijin desho?= [FC]
that person working person tag
‘She is working, isn’t she?’
g. K: =janai no?
tag fp
‘I guess so.’
h. H: gakusei?= [PNP]
student
‘(Is she) a student?’
 Chapter 4

i. K: =uu:n∼
‘I don’t think so.’
j. H: shakaijin da yo ne.@ [CNPp]
working person cop fp fp
‘(is) a working woman’
k. +are ne. [CNPa]
that fp
‘she’
l. K: ->ano: karaa kontakuto o kaitai tte
pf colored contact lenses acc buy-des qt
-->itteru hito. [PNP]
say-prog person
‘Well, (she is) the person who says that (she) wants to buy colored
contact lenses.’ (F8)

The sequence in (4.12) is composed of eight NP IUs. The IUs (4.12a, b),
(4.12h), and (4.12l) are [PNP] IUs, i.e. subjectless semi-clausal IUs with
nominal predicates. The IU (4.12j) is a predicate NP ([CNPp]), and the
IU (4.12k) is a post-predicate subject argument NP ([CNPa]); the two IUs
collectively form a multi-IU full clause. The IUs (4.12c) and (4.12d) are [NP]
IUs, i.e. clause-external independent phrasal NP IUs. The [NP] IU (4.12c) is
a repeated-as-affirmation NP, which functions to affirm that the term ikeike
(a slang for flamboyant appearance) supplied by K is exactly what H meant.
The [NP] IU (4.12d) was begun by K, but because of H’s interrupted yes/no
questions, seems to have failed to express the intended proposition. The
sequence contains one [FC] IU, (4.12f), which consists of two NPs, the subject
NP ano hito and the predicate NP shakaijin. (Note that the two regulatory
IUs—(4.12g) janai no? (consisting of two final particles with rising intonation)
and (4.12i) uu:n∼ (consisting of informal ‘yes’ with doubt-indicating rise-
fall-rise contour)—function as somewhat “softened” affirmative and negative
response, respectively, to the preceding questions.) The sequence contains 12
NPs, two of which, (4.12a) and (4.12l), are complex NPs with relative clauses
(cf. Note 7). (For the syntactic forms, grammatical roles, and information
statuses of NPs contained in the 420 NP IUs, see Section 5.3.)
In what follows I will explore the types and functions of [NP] IUs.
Syntactic structure of the intonation unit in conversational Japanese 

.. Independent phrasal NP intonation units: Types and functions

My analysis has identified five types of independent phrasal NP IUs. These are
listed and defined below (the same classification can be applied to the other
syntactic categories of independent phrasal IUs):
Type 1: “Stray” NPs—NPs that are “detached” from the immediate discourse
context, where an IU which is yet to be uttered to produce a com-
plete proposition is not supplied by the speaker who has produced
that NP. “Stray” NPs result when other-interruptions, self-repairs, or
collaborative completions occur.
Type 2: “Lead” NPs—NPs that function as “leads” to the following clausal IU
which expresses a complete proposition (cf. Durie 1994).
Type 3: Topic NPs or left-dislocated (LDed) NPs that are outside the clausal
structure (cf. Ashby 1988; Geluykens 1992).
Type 4: Postposed NPs or right-dislocated (RDed) NPs produced immediately
following a single-IU/multi-IU clause.
Type 5: NPs repeated as listener responses—tokens of acknowledgment,
agreement/disagreement, or clarification request—to the IU that the
speaker has just uttered.

The five types of independent NP IUs are schematically shown in (4.13).


This indicates that these NPs, located outside the clausal domain, are only
peripherally concerned with the communication of propositions, as opposed
to NPs integrated into the clausal structure (i.e. NPs occurring in single-IU
and multi-IU clauses) which directly feed into the propositions conveyed.
(4.13) Stray NPs
Lead NPs
Topic/LDed NPs

CLAUSES
NPs in single-IU and multi-IU clauses
COMMUNICATION OF PROPOSITIONS

Postposed/RDed NPs
Repeated NPs
The independent, detached NP IUs can be classified into two groups, which
seem to differ in the degree of relatability to the surrounding IUs/clauses.
“Stray” NP IUs cannot be syntactically attached/related to any of the self-
produced IUs (i.e. IUs produced by the speaker who uttered the NPs) in
 Chapter 4

Table 4.8. Distribution of five types of independent phrasal NP IUs

NP type N %

Type 1 Stray 24 21
Type 2 Lead 19 17
Type 3 Topic/LDed 30 27
Type 4 Postposed/RDed 18 16
Type 5 Repeated 19 17
Others (Vocatives) 2 2
Total 112 100

the immediate context. Topic NPs and postposed NPs, on the other hand,
can be syntactically related, although not integratable, to the immediately
following/preceding self-produced clause(s). “Lead” NPs can be referen-
tially/informationally relatable to the immediately following self-produced IU
clauses. Repeated NPs can be interactionally relatable to the immediately pre-
ceding other-produced IU clauses (see detailed discussion below). In brief, in-
dependent NP IUs appear to differ in the degree to which they are “detached”
from the clausal structure in the immediate discourse context (cf. Croft 1995;
Ono & Thompson 1994; Tao 1996). Importantly, as I will show shortly, the pro-
duction of such clause-external independent NP IUs is motivated by various
discourse-pragmatic, interactional, and information-flow factors.
Table 4.8 presents the frequency of the five types of independent phrasal
NP IUs produced in the 16 conversational segments (see also Figure 4.6).
Table 4.8 shows that of the 112 independent NP IUs, topic/LDed NPs
(27%) and stray NPs (21%) accounted for 48%, with the remaining three types
occupying comparable percentages (16–17%). (The category “others” includes
vocatives, i.e. proper names used in address.) Further analysis has shown the
following: (a) of the 30 topic/LDed NP IUs, 63% are “genuine” topics, and
10% are LDed NPs (the remaining 27% are NPs preposed for comparison
or as alternatives in choice questions); (b) of the 18 postposed/RDed NPs,
44% are restated postposed NPs, 33% are copied postposed NPs, and 23% are
RDed NPs; and (c) the 19 repeated-as-listener-response NPs consist of tokens
of acknowledgment (47%), affirmation/agreement (21%), and disagreement
(5%) (the remaining 27% are concerned with clarification request).
I will elaborate on each of the five types of independent NP IUs in order
below with illustrative examples.
Syntactic structure of the intonation unit in conversational Japanese 

30

25

20
Frequency

15

10

0 RDed
Postposed/
Lead

Topic/LDed

Repeated

Others
Stray

NP type

Figure 4.6. Distribution of five types of independent phrasal NP IUs

Stray NPs

The first type, “stray” NP IUs, can be further divided into the following four
subtypes:

Type 1: NP IUs that were uttered but totally “left alone” or “detached”
from the surrounding context, because neither the speaker himself/herself nor
the hearer subsequently supplied an IU which would complete the intended
proposition. Type 1 stray NPs involve discoursal operations such as other-
interruptions and self-repairs, and in most cases, the remaining-to-be-uttered
elements are non-recoverable.

Type 2: NP IUs that constitute part of a multi-IU clause collaboratively


constructed by the co-participants; NP IUs that were left isolated by the
speaker and therefore express only a partial proposition, but that were made
propositionally complete by the other interlocutor’s provision of another IU.
 Chapter 4

Thus the NP IU uttered by the first speaker, together with the immediately
following IU supplied by the other, jointly constitute a full proposition. The
resulting proposition/clause, however, may or may not be equal to the one that
the original speaker intended to produce (cf. Clancy et al. 1996; Lerner 1991;
Ono & Thompson 1996).

Type 3: NP IUs that are “detached” only from the immediate environment;
the speaker who uttered the NP subsequently supplies another IU to make a
partial proposition complete, but supplies it non-adjacently to the “original”
NP IU, that is, other IUs/propositions intervene. Type 3 stray NP IUs invariably
involve insertion sequences, which vary in length from fairly short to rather
long (e.g. 10-IU sequences). Like Type 1, Type 3 stray NPs may involve other-
interruptions or self-repairs.

Type 4: “argument” NP IUs that are “stray” in the sense that predicate ele-
ments that would be formally required to express a full proposition are neither
self-supplied nor other-supplied, but rather assumed by the conversational co-
participants because those elements are recoverable from the context. Type 4
stray NP IU differs from the other types in that its intended proposition can be
inferred from the context shared by the co-participants; thus it virtually func-
tions as a “clause” expressing a full proposition where some elements can be
taken to be ellipted. It may be said that Type 4 stray NPs are “formally phrases
but functionally clauses”. They are “stray” NPs uttered in the “right” context,
where elements yet to be uttered to produce a complete proposition can be
recovered.9
Table 4.9 summarizes the main features of the four types of stray NP IUs
and presents the frequency of each type. This shows that Type 1 is the most
frequently used stray NP type, accounting for 54%.

Table 4.9. Frequency and features of four types of stray NP IUs

Type and Elements yet to be Reconstructability Distance between the


frequency uttered in order to of elements not original NP and
constitute a full uttered by the supplied elements
proposition original speaker
Type 1 N=13 Not supplied No —
Type 2 N= 3 Other-supplied Yes/No Adjacent
Type 3 N= 3 Self-supplied — Non-adjacent
Type 4 N= 5 Not supplied Yes —
Total N=24
Syntactic structure of the intonation unit in conversational Japanese 

Type 1 stray NP is illustrated in (4.14), where the female co-participants A


and M are discussing how difficult it would be to prepare meals and bring up
children in married life.
(4.14) a. A: ->maa ie ni iru n dattara
pf home loc stay nml cop-con
-->betsu da kedo, [PAP]
different cop though
‘If (you) stay at home, (it=cooking every day) (would be) a diffe-
rent matter.
b. NE::^
fp
‘wouldn’t it?’
c. M: demo ie ni itemo ne, [PVP]
but home loc stay-con fp
‘But even if (you) stay at home,’
→d. atashi, [NP]
I
‘I’
e. ++kodomo o sodateru no toka ne? [CNPa]
child acc raise nml sof fp
‘bringing up children’
f. chiisai to [sugoi taihen. [CAPp]
small con very hard
‘(is) very hard if (they are) small.
g. A: [HAAdo deshoo ne::^ [CAPp]
hard cop-pres fp
‘will (be) hard!.
h. +are ne:.@ [CNPa]
that fp
‘that’
i. M: <are Su:goi taihen. [FC]
that very hard
‘That (is) very hard.’ (F4)

The NP IU (4.14d) is a Type 1 stray NP. Speaker M, after uttering the NP


atashi ‘I’, self-repairs during the medium-length initial pause in (4.14e) and
begins to utter a new IU. This leaves the NP IU (4.14d) totally “stray”; no
“matching” predicate IU is to be supplied in subsequent discourse. (Note that
in (4.14b) and (4.14g), speaker A is using rise-fall intonation contours (^)
effectively with the interactional particle ne to seek agreement from M.)
 Chapter 4

Another example of Type 1 stray NP is (4.12d) in Section 4.5.1. The NP


ano ko ga ‘she’ was produced as a result of other-interruption, one of the typ-
ical conversational acts which demonstrate the co-participants’ interactional
involvement. The NP IU (4.12d) was begun by K, but because of H’s inter-
rupted yes/no questions—(4.12f) ‘She is working, isn’t she?’ and (4.12h) ‘(Is
she) a student?’, the first one overlapping with K’s regulatory IU (4.12e) ‘so’—
has failed to express the intended proposition. That is, speaker K has left the
NP IU (4.12d) totally “stray”, with no “matching” predicate IU following in
subsequent context.
Type 2 stray NP, which involves interactional collaborative finishes,10 is
illustrated in (4.15), where the talk is concerned with the purpose of the
author’s research project for which T and R are in the very process of recording
their conversation.
(4.15) a. T: iya kono hito no ne?@ [AP]
pf this person gen fp
‘uh this person’s’
→b. sono: kenkyuu naiyoo ga, [NP]
pf reseach content nom
‘uh research topic’
c. H: +doko ni aru ka. [PVP]
where loc exist q
‘Where does (it) consist in?’
d. T: doko ni aru ka tte yuu no o sa, [NP]
where loc exist q qt say nml acc fp
‘the (problem) of where (it) consists in’
e. H: +soo desu ne:.
so cop-pol fp
‘Well, let me see.’
f. sore ja nai to shitara, [PNP]
it cop neg qt do-con
‘If (that’s) not it (=use of pronouns),’
g. ->tatoeba keigo no tsukaikata
for example honorifics gen how to use
-->toka sa, [PNP]
sof fp
(It’s) use of honorifics, ‘for example,’ (M4)

The NP IU (4.15b) kenkyuu naiyoo ga ‘research topic’ is a Type 2 stray NP,


the “matching” predicate of which was not self-supplied by T, but other-
Syntactic structure of the intonation unit in conversational Japanese 

supplied by R in the immediately following IU (4.15c) to produce a complete


clause/proposition. In other words, speaker R finishes speaker T’s utterance
such that the adjacently produced two IUs, (4.15b) and (4.15c), jointly con-
stitute a full proposition. Given speaker T’s utterance (4.15d) (Type 1 stray
NP), which slightly modifies R’s just-supplied predicate IU (4.15c), however,
the proposition completed by R appears to be what the original speaker T
intended to produce. It seems evident that Type 2 stray NPs involving col-
laborative finishes as in (4.15b), as well as Type 1 stray NPs involving inter-
ruptions as in (4.12d), are the “products of active interaction” among the co-
participants who are jointly and collaboratively constructing a conversational
discourse (cf. Hayashi 2003). (Another example of Type 2 stray NP is (4.26c)
in Section 4.6.)
Type 3 stray NP, which involves a non-adjacently self-produced predicate
IU, is best illustrated in (4.16).
(4.16) →a. Y: kurasu no kotachi ga:, [NP]
class gen student-pl nom
‘(my) classmates’
b. buronkusu? [NP]
Bronx
‘Bronx?’
c. buronkusu tte yuu tokoro ga aru no? [CFC]
bronx qt say place nom exist q
‘is there a place called Bronx?’
d. +manhattan ni? [CAvP]
Manhatten in
‘in Manhatten?’
e. S: buronkusu∼ [PNP]
Bronx
‘(Is it really) Bronx?’
f. Y: burankusu ka na? [PNP]
branx q fp
(It may be) Branx.
g. wakannai. [PVP]
know-neg
‘(I) don’t know.’
h. nanka abunai chiiki ga aru n da tte? [FC]
pf dangerous area nom exist nml cop qt
‘(I hear) that there is a dangerous area.’
 Chapter 4

i. S: atashi mo amerika no koto wa shiranai. [FC]


I also America gen thing top know-neg
‘I don’t know about things in America, either.’
j. Y: <de soko ni i-
and there all
‘and there -’
k. ->soko no rokku konsaato ni itte
there gen rock concert all go-and
-->KAette KOnai. [CVP]
return-and come-neg
‘(they) went to a rock concert there and have never come back or
something’
l. to ka itte::, [CVP]
qt q say-and
‘(they=my classmates) say that’ (F1)

In (4.16), eight IUs intervene between the stray subject NP IU (4.16a) ‘my
classmates’ and its “matching” multi-IU predicate, (4.16k, l) ‘say that they
(=‘two girls and one boy’ to be supplied subsequently) went to a rock concert
there and never come back’. During the inserted IU sequence (4.16b-i), speaker
Y interacts with S regarding the area named Bronx, and thus “derails” from her
narration of a series of dangerous incidents that she heard actually happened
in New York. (Note that (4.16b) is a lead NP to be copied in the full clause
(4.16c), and (4.16j) is a fragmentary IU involving Y’s false start.)
Type 4 stray NP is found in (4.17), where the male undergraduate students
I and K are conversing in K’s apartment about how K bought various expensive
household items.
(4.17) a. K: kore nanka wa roon na n desu yo.@ [FC]
this sof top installment cop nml cop-pol fp
‘(I bought) this on the installment plan.’
b. I: aa sore kiita.@ [PVP]
yeah it hear-past
‘Yeah, (I’ve) heard it.’
c. K: ato ni kai gurai desu ka.@ [FC]
rest two times about cop q
‘(I think I have) about two more times (of payment) to go.
d. +ato moo ni kai. [FC]
rest more two times
‘(I have) two more times to go.’
Syntactic structure of the intonation unit in conversational Japanese 

→e. I: kuuraa wa? [NP]


air conditioner top
‘(What about) the air conditioner?’
f. K: ->+kuuraa wa obaachan ga katte
air conditioner top grandma nom buy-and
-->kuremashita. [FC]
give-past-pol
‘(My) grandma bought the air conditioner (for me).’
g. I: II NA::. [PAP]
good fp
‘I envy you!’
h. SOO yuu no ichiban ii yo:. ((laugh))# [FC]
so say nml most good fp
‘That is best!’
i. ->amaereru toki wa moo
depend-pot time top immediately
-->amaechau mon ne.# [PVP]
depend nml fp
‘When (you) can depend on (someone), (you) immediately do so,
don’t you? (though you are not supposed to do so)’ (M6)

The NP IU (4.17e) kuuraa wa? (which is given information already uttered in


the anaphoric context) is a Type 4 stray NP, which, although an “argument”
NP alone without a predicate, successfully conveys speaker I’s intended propo-
sition, i.e. ‘how (e.g. on the installment plan, by lump-sum payment) did K buy
the air conditioner?’ The intended proposition was taken as such by hearer K as
well; thus K provides the “proper” response (4.17f) ‘(my) grandma bought the
air conditioner (for me)’ to I’s intended “wh-question”. The successful commu-
nication between I and K in this excerpt naturally rests on the recoverability of
the “argument” NP’s predicate elements which are not expressed but shared
by the co-participants by prior discourse and interaction. Thus the indepen-
dent NP IU (4.17e) virtually functions as a “clause” communicating a complete
proposition.
In sum, stray NPs are found principally in aspects of conversation in which
co-participants are strongly interactionally involved such as collaborative com-
pletions, other-interruptions, and self-repairs. The occurrence of stray NPs in
informal interactive conversation can thus be related to the speaker’s concern
about, or orientation toward, the hearer.
 Chapter 4

Lead NPs

The second type of independent phrasal NP IUs, “lead” NPs, are of two types:
(a) those to be copied/repeated, and (b) those to be modified/restated, in the
following clause.11 The NP IU (4.18d) Hiroki (male first name) given below
illustrates the to-be-repeated lead NP, which is repeated in the immediately
following full clausal IU (4.18e) as the subject of the predicate ‘return’. Recall
that such NPs as (4.18d) have been treated in discourse research as performance
errors, specifically, “false starts”. The to-be-restated NP lead is illustrated in
(4.25) in Section 4.6, where the NP IU (4.25f) saki ga is to be restated as
shooraisei ga in the following clausal IU (4.25h).
(4.18) a.
M: tooka? [PNP]
ten
‘(Is it=the date of your return) (September) ten?’
b. Y: un.
‘Yeah.’
c. chotto inai n da kedo, [PVP]
little exist-neg nml cop though
‘Though (I) won’t be (here) for a little while,’
→d. hiroki toka, [NP]
Hiroki sof
‘Hiroki (=Y’s boyfriend)’
e. ->+hiroki nanka hachigatu no nijuuku ni
Hiroki sof August gen 29 loc
-->kaette kuru desho? [FC]
return-and come tag
‘Hiroki is coming back on August 29, isn’t he?
f. natsu ni kaette kitara ne, [PVP]
summer loc return-and come-con fp
‘When (I) come back this summer,’
g. daibingu no menkyo toru kara. [PVP]
diving gen licence take as
‘(I) will take a diving licence.’
h. +datte HIma ja::n. [PAP]
because free fp
‘Because (I will be) free’
i. >gakkoo hajimaru made.< [FC]
school begin until
‘until school starts’ (F5)
Syntactic structure of the intonation unit in conversational Japanese 

The lead NP IU, which is apparently “redundant” as far as the communica-


tion of propositions is concerned, seems yet to have important implications for
such matters as the process of proposition construction and information struc-
ture of the IU. In my view, the existence of NP leads is important especially
from an information-structural point of view. First, what lead NPs indicate
about discourse production in the form of IUs is that one of the IU produc-
tion strategies Japanese speakers employ is to first utter an independent NP IU
and then use that NP as a “stepping stone” to build a complete proposition by
incorporating that same previously uttered NP or a slightly formally revised
NP in the next clausal IU. Second, what lead NPs imply for the information
structure is that this pattern of IU production may be related to the strategy
of avoiding more than one new NP within an IU. (For discussion of the “one
new NP per IU constraint”, see Section 5.4.2.) That is, uttering an NP first as an
independent lead IU has the effect of converting the information status of the
following clause-internal NP—which refers to the same referent as the lead NP
and would remain “new” without the lead—into “given” or “accessible” (i.e.
“previously active” or “previously semi-active”) before the utterance of that NP
in the “target” clausal IU (cf. Section 2.3.2). For example, Hiroki in (4.18d) is a
newly activated but identifiable NP, whereas Hiroki in (4.18e) is now given in-
formation because the referent was already activated in the hearer’s conscious-
ness before the utterance of that NP (cf. “referent-activating” NPs in Mandarin
in Tao 1996). It is through the utterance of the lead NP (4.18d) that the full
clausal IU (4.18e) has resulted in containing one new NP, ‘August 29’.
In sum, lead NPs should not be treated as performance errors. They have
important implications for the speakers’ discourse production strategies as well
as for the information structure of the IU. Moreover, their referent-activating
and information flow-regulating functions are interactionally motivated.

Topic/Left-dislocated NPs

The third type, which involves NPs functioning as “genuine” topics or LDed
clause-external elements, is illustrated in (4.19) and (4.20).
(4.19) →a. M: datte senhyaku tte koto wa, [NP]
because 1,100 qt thing top
‘because speaking of the requirement that (the SAT score must be)
1,100,’
 Chapter 4

b. tatoeba, [AvP]
for example
‘for example’
c. +suugaku o manten totta to shitemo, [PVP]
math acc full mark take-past qt do-con
‘Even if (you) get a full mark in math,’
d. suugaku ga manten de yatto sanbyaku.@ [FC]
math nom full mark cop barely 300
‘The full math score would barely leave (the other section’s score)
300.’
e. ++desho?
tag
‘wouldn’t it?’
f. de,
and
‘and’
g. manten toreru tte no muzukashii kara, [FC]
full mark take-pot qt nml difficult because
‘Because getting a full mark is difficult,’
h. ja tatoeba nanahyaku to shitara, [PVP]
so for example 700 qt do-con
‘So, suppose (it’s) 700, for example’
i. baabaru yonhyaku da mon.@ [FC]
verbal 400 cop nml
‘the verbal (section’s score will be) 400.’
j. +dakara taihen da na: to omotte, [PVP]
so hard cop fp qt think-and
‘So (I) think that (it’s) hard.’ (F3)
(4.20) a. S: suzuki mari to ka yuu. [PVP]
Suzuki Mari qt q say
‘(She is) named Mari Suzuki.’
b. nandaka bairingyaru na n da soodesu yo. [PNP]
sof bilingual gal cop mnl cop hear-pol fp
‘(I) hear (that she is) a bilingual gal.’
c. J: ->aa nanka shuukanshi ni
yeah something weekly magazine loc
-->deteta na.@ [FC]
appear-past-sta fp
‘Yeah, (there) was some article (about her) on a weekly magazine.’
Syntactic structure of the intonation unit in conversational Japanese 

d. ato sono hoomuwaaku no,@ [AP]


and the homework gen
‘and of the Homework (=TV drama title)’
e. nan da kke.
what cop q
‘What shall I say?’
→f. ano dooseishiteru onna no hito mo, [NP]
that cohabit-prog woman gen person also
‘the woman who is cohabiting also’
g. are kashu mitaidesu ne. [CFC]
that singer appear-pol fp
‘she seems to be a singer’
h. +honto wa. [CAvP]
truth top
‘actually’ (M3)

The complex NP (4.19a) is a genuine topic NP, whereas (4.20f) is a LDed NP


involving the resumptive pronoun ‘that’ in the following clause. In the multi-
IU turn in (4.19), speaker M is talking about the SAT score 1,100 which she
heard is required of returnee students for admission to some of the Japanese
high schools in Tokyo. Of the 10 IUs constituting the turn, six substantive
IUs, (4.19b-d) and (4.19g-i), function to elaborate or explain the topic NP IU
(4.19a) senhyaku tte koto wa ‘the requirement that (the SAT score must be)
1,100’. That is, the sequence in (4.19) consists of a genuine topic-indicating IU,
(4.19a), followed by a multi-IU “comment” (composed of eight IUs (4.19b-i)
including regulatory IUs) and a conclusive opinion-indicating IU, (4.19j). The
topic NP IU (4.19a) can be regarded as setting up a “framework” (which is
fairly large in this case covering eight IUs) within which further referents are
identified and predicated (Chafe 1976: 50).
The IU sequence (4.20), on the other hand, is part of a longer sequence
where the male interactants J and S are conversing about actors and actresses
performing in two Japanese TV dramas, one of which is entitled ‘Homework’.
The preposed, LDed independent NP IU (4.20f) ano dooseishiteru onna no hito
mo ‘the woman who is cohabiting’—modified by the independent adjectival IU
containing the drama title (4.20d) sono hoomuwaaku no ‘of the Homework’—
appears to function to “topicalize” the referent ‘the woman who is cohabiting in
the drama Homework’, placing it high in focus among the “competing” actors
and actresses to be possibly talked about.
 Chapter 4

Postposed/Right-dislocated NPs

The fourth type, which involves clause-external post-predicate placement of


NPs as postposed or RDed NPs, is illustrated in the excerpts (4.21) and (4.22).
(4.21) a. T: honde ato kinniku shinkoo mo tsuyoi ne. [FC]
and rest muscle faith also strong fp
‘And “muscle faith” is also strong.’
→b. +kinniku ni taisuru shinkoo ga.@@ [NP]
muscle toward faith nom
‘faith in muscle’
c. maa nihonjin wa kekko ne, [CXP]
pf Japanese top quite fp
‘well Japanese quite’
d. dochira ka to surimu desho?@ [CNPp]
which q qt slim tag
‘(are) relatively slim, aren’t they?’
e. minna ningen hosoi desho? [FC]
all people slim tag
‘People are all slim, aren’t they?’
f. N: ->dakara nihon de kinniku wa gyagu
so Japan loc muscle top gag
-->ni naru mon na.@ [FC]
become thing fp
‘So muscle becomes a gag in Japan.’
g. T: ne.
fp
‘You are right.’
h. N: ++ano macchoman. [PNP]
pf Macchoman
‘Uh (there’s) the Macchoman (=comic character).’
i. mukimukiman toka ne? [PNP]
Mukimukiman and fp
‘(There’s) the Mukimukiman (=comic character) and so on.’ (M2)
(4.22) a.K: are doo shita no. [PVP]
that how do-past fp
‘What did (you) do with that?’
→b. ++soosharu sekyuritii nambaa. [NP]
social security number
‘social security number’
Syntactic structure of the intonation unit in conversational Japanese 

c. O: iya ichiyoo, [AvP]


pf provisionally
‘uhm provisionally’
d. +dakara ano:,=
so pf
‘so well’
e. K: =nai n desho? [PVP]
exist-neg nml tag
‘(You) don’t have (one), do you?’
f. O: mottemasu kedo, [PVP]
have-pol though
‘(I) have (one), but’
g. Kaitenai n desu yo. [CVP]
write-neg-sta nml cop-pol fp
‘(I) haven’t put (it)’
h. +ano: gakkoo no hoo ni wa. [CAvP]
pf school gen side loc top
‘uhm on (my) school (documents)’ (M5)

Of the two types of postposed NP IUs, copied/repeated and restated/modified,


(4.21b) kinniku ni taisusu shinkoo ga ‘faith in muscle’ illustrates the restated
postposed NP IU, which restates the NP kinniku shinkoo ‘muscle faith (liter-
ally)’ contained in the preceding full clausal IU (4.21a). (The copied/repeated
postposed NP IU is illustrated in (4.4) in Section 4.1 and (4.11) in Section 4.3,
where (4.4b) seikaku ga and (4.11c) ano Roozu te simply repeat the NPs con-
tained in the preceding single-IU and multi-IU full clause, respectively.) Both
types of postposed NP IUs seem to have the function of emphasizing the
proposition conveyed in the preceding clause or reinforcing/clarifying with
emphasis the repeated/restated referent for the hearer. That is, clause-external
postposed NPs occupying separate IUs are more explicitly related to partici-
pant interaction than clause-internal postposed NPs constituting the final part
of single-IU/multi-IU clauses, which are generally taken as backgrounded or
defocused elements. In short, the detached post-predicate NPs, while repeating
or restating “redundantly” the information given in the immediately preceding
clause, do function as a more interactionally oriented device of emphasis, re-
inforcement, repair, or further specification of already expressed elements (see
Section 4.3).
The IU (4.22b) illustrates the RDed NP IU with the resumptive pronoun
‘that’ remaining in the preceding semi-clausal IU (4.22a). The RDed NP IU
can be considered one type of restated postposed NP IUs, whose function
 Chapter 4

is not so much reinforcing as clarifying the referent which was coded by


the pronoun. Thus, in (4.22), speaker K supplies the NP IU ‘social security
number’ postverbally in order to clarify for hearer O what he means by the
pronoun ‘that’. The IU-initial medium-length pause in (4.22b) seems to have
been secured so that the intended referent, which was created by the very
utterance of ‘that’, can be activated/accessed in hearer O’s consciousness before
getting clarified.

Repeated NPs

Finally, the fifth type, which involves interactive repetition of speaker-uttered


NPs by the listener, is illustrated in (4.23), where the interactants M and O are
talking about O’s younger brother’s prospective undergraduate major at UCLA.
(4.23) a. M: ->demo yaritai mono ni mo
but do-des thing on also
-->yoru daroo shi. [PVP]
depend prs and
‘But (it=whether to go to a Japanese vs. American university) will
probably also depend on what (he=O’s younger brother) wants to
major.’
b. +ne^
fp
‘won’t it?’
c. nan da kke. [PNP]
what cop q
‘What (is it)?’
d. aakitekuchaa ka? [PNP]
architecture q
‘(Is it) architecture?’
e. O: iya moo are wa yameta no. [PVP]
no already that top give up-past fp
‘No, (he) already gave that up.’
f. M: yameta no? [PVP]
give up-past q
‘(He) gave (it) up?’
g. de kekkyoku? [AvP]
and after all
‘and after all?’
Syntactic structure of the intonation unit in conversational Japanese 

h. O: nanka enjiniaringu. [PNP]


sof engineering
‘(It’s) engineering.’
→i. M: enjiniaringu. [NP]
engineering
‘engineering’ (F3)

The final IU (4.23i) ‘engineering’ illustrates the repeated-as-listener-response


NP, which M supplies in response to O’s utterance (4.23h) ‘engineering’
(which means that ‘it (=O’s brother’s prospective major) is engineering’)
in order to acknowledge to O, seemingly with some wonder, that M has
definitely received the word ‘engineering’ which has just been uttered by
O and the proposition conveyed as well. (Another example of repeated NP
IU can be found in (4.24) in Section 4.6, where the NP IU (4.24j) batsu
functions as a token of acknowledgment as in (4.23h). One more example
is (4.12c) ikeike, which may be taken to function as a token of affirmation
of, or agreement with, the proposition conveyed by the interlocutor in the
immediately preceding clause.)
NP IUs repeated as listener response have to do more with speaker-hearer
interaction than with the dynamic process of referring, that is, the function
of repeated NP IUs is interactional rather than referential (cf. Tao 1996).
Thus NP IUs repeated, especially as tokens of acknowledgment, may even
be considered “lexical” versions of backchannels, which in some contexts
probably do nothing more than serve as “continuers” (Schegloff 1982). This
also accords with Clancy et al.’s (1996) treatment of such repeated NPs as one
form of backchannels, which they call “reactive tokens”. Given that the repeated
NPs convey given information, they contribute neither to the communication
of new propositions nor to their further development. As indicated above,
repeated NPs constitute “redundant” part of conversational language, as far as
the conveying of propositional content is concerned. However, their existence
is important interactionally.
Finally, the analysis revealed the following concerning the typical pat-
terns of IU production involving the lead, postposed, and repeated NP IUs.
Most typically, lead NPs were prefixed to the immediately following single-
IU clause, in which the repeated/restated NP functions as the subject of that
clausal IU. The robust pattern of clause-external postposing involved the re-
peating/restating of the subject NP contained in the immediately preceding
clausal IU. Most typically, repeated NP IUs took the form of repeating the fi-
 Chapter 4

nal NP (or the only NP in the case of semi-clausal IUs) contained in the im-
mediately preceding other-produced single-IU clause. Of the three patterns,
the lead and postposed NPs exemplify intra-speaker “redundant” repetitious
use of NPs, whereas the repeated NPs exemplify inter-speaker repetition of
NPs. That is, while the copied leads and copied postposed NPs concern self-
produced repetition, the repeated NPs concern other-produced repetition. We
can conclude from these findings that one of the recurrent patterns of dis-
course production in the form of IUs in conversational Japanese is to “prefix”
or “suffix” a single-IU clause that expresses a complete proposition with clause-
external detached NP IUs which are redundant propositionally but important
discourse-functionally.
In conclusion, the “independent phrasal” NP IUs or “detached” NPs,
though placed outside the clausal domain and thus not directly concerned
with the communication of propositions, cannot simply be deemed fragments
of discourse or performance errors. It has been argued in this section that
Japanese conversational co-participants produce five types of detached NPs,
each of which is an important phenomenon from a discourse-functional
point of view, and that their occurrence in informal conversation is motivated
by various discourse-pragmatic, interactional, and information-flow factors.
Concerning the cross-linguistic applicability of the discourse phenomena dealt
with in the present section, Croft (1995) and Tao (1996)12 observe similar
independent lone NPs in a corpus of English oral narratives and Mandarin
Chinese conversation, respectively. It may be that the detached NP as an extra-
clausal independent IU occurs cross-linguistically in spoken discourse (see also
Durie 1994; Helasvuo 2001).
Stray NPs occur as a result of conversational acts by which co-participants
exhibit strong interactional involvement such as collaborative completions
and interruptions. Lead NPs, which usually carry new information, serve the
important function of regulating information flow in discourse specifically
by “converting” the information status of NPs to be copied or restated in
the following clause. This can be seen as one of the discourse strategies
speakers employ in order for the conversation to conform to the “one new
NP per IU constraint”. Further, clause-external postposed NPs which appear
as separate IUs apparently perform more interactionally oriented discourse-
pragmatic functions than their clause-internal counterparts—for example,
their functions such as emphasis, repair, and further specification can be seen
to be more directly related to the speaker’s concern about the hearer. Finally,
use of detached NPs as repetitions is another common type of interactional
strategies speakers use in talk-in-interaction.
Syntactic structure of the intonation unit in conversational Japanese 

. Adjectival, adverbial, and mixed phrasal intonation units

We have seen in Section 4.2 that the adjectival IUs (i.e. [AP]) accounted for
8% of the 1,600 substantive IUs. The adverbial IUs (i.e. [AvP]) and mixed
phrasal IUs (i.e. [XP]) accounted for 11% and 2%, respectively (see Table 4.1).
Table 4.10 shows the distribution of the adjectival, adverbial, and mixed
phrasal IUs.
We can see the following from Table 4.10 concerning the distribution of
the five types of AP IUs: (a) 67% are clausal IUs, whereas 33% are phrasal
IUs; (b) 76% are predicate APs, whereas 24% are attributive APs; and (c) 73%
are independent clausal or phrasal IUs, whereas 27% are elements of multi-
IU clauses. In short, the results indicate the speakers’ preferential production
of AP IUs as independent and clausal [PAP] IUs. The adverbial IUs and
mixed phrasal IUs occurred either as constituents of multi-IU clauses ([CAvP],
[CXP]) or as independent IUs ([AvP], [XP]). Regarding the distribution of
these IU types, Table 4.10 indicates the following. The AvP IUs were used as
elements of multi-IU clauses and as independent adverbials in comparable
proportions. The XP IUs, on the other hand, were used preferentially as
elements of multi-IU clauses. Further coding has shown that both [CAPa] and
[CAvP] IUs tend to occur within multi-IU semi-clauses as opposed to multi-IU
full clauses. This indicates the speakers’ propensities to use attributive AP IUs as
modifiers of non-subject NPs and adverbial IUs as circumstantial modifiers of
subjectless predicates. In contrast, the mixed phrasal [CXP] IUs tend to occur
within multi-IU full clauses, as in the case of the [CNPa] argument NP IUs
(cf. Section 4.5.1).
Table 4.11 presents the distribution of six types of clause-external indepen-
dent adverbial IUs.
Table 4.11 shows that of the 82 independent [AvP] IUs, about 50% are
“stray” adverbials, and about 30% are sentential adverbs13 which regulatorily

Table 4.10. Distribution of adjectival, adverbial, and mixed phrasal IU types

AP IU type N % AvP IU type N % XP IU type N %


[PAP] 81 62
[CAP] 7 5 [CAvP] 90 52 [CXP] 27 77
[CAPa] 18 14
[CAPp] 11 8
[AP] 14 11 [AvP] 82 48 [XP] 8 23
Total 131 100 172 100 35 100
 Chapter 4

Table 4.11. Distribution of six types of independent adverbial IUs

AvP type N %

Type 1 Stray 40 49
Type 2 Lead 7 9
Type 3 Topic/LDed 1 1
Type 4 Postposed/RDed 5 6
Type 5 Repeated 6 7
Sentential adverb 23 28
Total 82 100

function to link the preceding and following IUs (e.g. kekkyoku ‘after all’,
tatoeba ‘for example’, tsumari ‘in other words’). This indicates that unlike
the detached [NP] IUs, which showed a fairly balanced distribution among
the five types (see Table 4.8), the [AvP] IUs exhibit a skewing toward the
stray type. Of the 40 stray [AvP] IUs, 40% belong to Type 1 which involves
self-repair or other-interruption (Type 2=2%, Type 3=25%, Type 4=33%) (cf.
Section 4.5.2).
Examples of independent phrasal [AP], [AvP], and [XP] IUs are given
below as illustrations.
The excerpt (4.24) contains two [AvP] IUs and two [AP] IUs. Here the
interactants S and I are conversing about potential factors that will affect job
placement in Japanese companies.
(4.24)→a. S: gyakuni ne,@ [AvP]
conversely fp
‘conversely’
b. hu::n.
pf
mhm
→c. yappari, [AvP]
after all
‘after all’
d. are ga aru yo. [FC]
that nom exist fp
‘That exists.’
→e. +daigaku no, [AP]
university gen
‘university’s’
Syntactic structure of the intonation unit in conversational Japanese 

→f. daigaku no:, [AP]


university gen
‘university’s’
g. I: ++neemu baryuu mitaina? [PNP]
name value sof
‘(Is it something like) name value?’
h. S: neemu baryuu ja nakute, [PNP]
name value cop neg-and
‘(It’s) not name value, but’
i. daigaku batsu mitaina. [PNP]
university cliquism sof
‘(It’s something like) university cliquism.’
j. I: batsu. [NP]
cliquism
‘cliquism’ (M7)

The IUs (4.24a) and (4.24c) are [AvP] IUs; (4.24e, f) are [AP] IUs. In this
sequence, speaker S first utters ‘conversely’, a sentential adverb, to link the
preceding IUs with the following IUs. Then, inserting the cognitive type
regulatory IU ‘mhm’ (which suggests S’s mental planning for the coming IUs),
S utters ‘after all’, an additional sentential abverb. The first AP IU (4.24e)
‘university’s’ is a RDed and stray (Type 1) adjective, which was intended to
be a modifier of some head NP (which was not supplied) and thus to be
part of a larger NP restating the resumptive subject pronoun ‘that’ in the
full clause (4.24d). The second AP IU (4.24f) ‘university’s’ is a stray (Type 2)
adjective. Speaker S, while searching for an appropriate head NP, repeats the
AP, but before coming up with a “target” NP, the other interlocutor supplies
the NP ‘name value’ in (4.24g). The result is collaborative completion. Speaker
S then negates the supplied term ‘name value’ and finally supplies the target
NP ‘cliquism’ in (4.24i). The repeated-as-listener-response NP IU (4.24j) batsu
functions as a token of acknowledgment.
Example (4.25), where speaker Y and H are talking about one of the major
Japanese newspapers, contains three detached [AvP] IUs.
(4.25)→a. Y: yappa sono:, [AvP]
after all pf
‘after all uh’
b. kaisha no shuryuu de nai kara ne.= [PNP]
company gen mainstream cop neg because fp
‘Because (it’s) not the company’s mainstream.’
 Chapter 4

→c. H: =tashikani. [AvP]


certainly
‘(That’s) certainly (true).’
d. nikkei dakara anteishiteru kara sa, [CFC]
Nikkei so stabilize-sta because fp
‘because the Nikkei is stable’
e. shinbunsha de ichiban. [CAvP]
newspaper company among most
‘most among the newspaper publishing companies’
f. saki ga:, [NP]
future nom
‘future’
g. ++ne,
fp
h. shooraisei ga aru desho. [PVP]
promise nom exist tag
‘(It) is promising.’
i. Y: soo na no? [PAP]
so cop fp
‘(Is that) so?’
j. H: ->ichiban shooraisei ga aru to omou yo
most promise nom exist qt think fp
-->nikkei wa. [PVP]
Nikkei top
‘(I) think that is most promising, the Nikkei.’
→k. Y: doo yuu imi de? [AvP]
how say sense in
‘in what sense?’ (M8)

The IU (4.25a) ‘after all’ is a sentential adverb used as a cohesive IU-linking


device; (4.25c) and (4.25k) are stray (Type 4) adverbials. The IU (4.25c)
‘certainly’—without a predicate but uttered in the “right” context, immediately
following the semi-clause (4.25b)—expresses a complete proposition (which
would be equivalent to ‘that’s certainly true’). Thus in uttering the adverb
(4.25c), speaker H affirms Y’s immediately preceding utterance. Likewise,
(4.25k) ‘in what sense?’ builds on the proposition conveyed by the immediately
preceding complex semi-clause (4.25j), virtually expanding that proposition.
In (4.25k), although without an overtly expressed predicate, the proposition
equivalent to ‘is the Nikkei newspaper most promising?’ may be taken to be
assumed by the co-participants. (The IU (4.25f) is a lead NP.)
Syntactic structure of the intonation unit in conversational Japanese 

Finally, clause-external detached [XP] and [VP] IUs functioning as “leads”


are illustrated in (4.26).
(4.26)→a. S: amerika de hontoni suugaku, [XP]
America loc earnestly mathematics
‘mathematics earnestly in America’
→b. amerika de hontoni suugaku benkyoo, [VP]
America loc earnestly mathematics study
‘study mathematics earnestly in America’
c. sukide benkyooshiteru yatsu tte sa, [NP]
like-and study-prog person top fp
‘people who like (mathematics) and so are studying (it)’
d. M: +anmari inai n janai? [PVP]
not many exist-neg nml tag
‘(Such people) are not many, aren’t they?’
e. S: i- iru kedo mo, [PVP]
exist though emp
‘There are (such people), but’
f. iru kedo mo, [PVP]
exist though emp
‘(Such a person) exists, but’
g. soide hana ga denakattara, [FC]
and flower nom come out-neg-past-con
‘And if flowers did not come out (=if he did not succeed),’
h. sore shinu n da ne. [FC]
it die nml cop fp
‘he would perish.’ (M1)

At the beginning of the sequence, speaker S, by uttering the two independent


phrasal IUs, the [XP] IU (4.26a) and the [VP] IU (4.26b), gradually builds up
a target IU, and thus approaches a target proposition he intended to convey.
Given that the target IU is a complex NP as shown in (4.26c), both of the
[XP] and [VP] IUs can be considered to be intended as elements of a relative
clause. Speaker S, however, failed to construct the target proposition because
of hearer M’s “timely” provision of the semi-clause (4.26d), which, together
with the complex NP (4.26c), conveys a complete proposition. This is another
example of collaborative completion. (The NP (4.26c) is a Type 2 stray NP.)
The only thing we can deduce from the sequence is that the proposition speaker
S originally intended has something to do with the last four IUs (4.26e-h).
 Chapter 4

. Intonation units and clauses: Single-IU clauses vs. multi-IU clauses

Table 4.12 lists the number of single-IU/multi-IU full clauses and semi-clauses
that the 32 speakers produced in the 16 conversational segments, and the
number of IUs contained in the four clause types. It also presents the average
number of IUs contained per full clause, semi-clause, and clause (full clause
plus semi-clause).
Table 4.12 indicates that on average, the multi-IU full clauses contained
2.22 IUs, the multi-IU semi-clauses contained 2.21 IUs (this means that
both types of multi-IU clauses contained a comparable number of IUs), and
the multi-IU clauses (full clauses + semi-clauses) contained 2.21 IUs. That
is, on average, 2.21 IUs collectively constituted a multi-IU clause in the 16
conversational segments. It also indicates that on average, the full clauses
contained 1.35 IUs, the semi-clauses contained 1.17 IUs, and the clauses
contained 1.23 IUs. That is, on average, 1.35 IUs collectively comprised a full
clause, 1.17 IUs collectively comprised a semi-clause, and 1.23 IUs collectively
comprised a clause in the 16 segments.
As shown in Table 4.13, of the total number of clauses (N=1,121), 81%
consist of one IU (i.e. single-IU clauses), and the remaining 19% consist of
more than one IU (i.e. multi-IU clauses). More specifically, 15% consist of two
IUs; clauses comprised of three or more IUs are extremely rare, accounting for
only 4% (see also Figure 4.7). This suggests that there is an operative constraint
in conversational Japanese that limits the number of IUs contained within
a clause to “no more than two”. In other words, the constraint confines the
production of clauses, such that one IU singly, or two IUs—uttered adjacently
and by the same speaker—collectively, communicate a complete proposition.

Table 4.12. Average number of IUs per clause

Clause type Number Number Average number


of IUs of clauses of IUs per clause
Single-IU full clause 263 263 1.00
Multi-IU full clause 233 105 2.22
Full clause-Total 496 368 1.35
Single-IU semi-clause 647 647 1.00
Multi-IU semi-clause 234 106 2.21
Semi-clause-Total 881 753 1.17
Single-IU clause 910 910 1.00
Multi-IU clause 467 211 2.21
Clause-Total 1,377 1,121 1.23
Syntactic structure of the intonation unit in conversational Japanese 

Table 4.13. Frequency of the number of IUs per clause

Full clause Semi-clause Total


Number of IUs
per clause N % N % N %
One 263 71 647 86 910 81
Two 83 23 86 11 169 15
Three 21 6 18 2 39 4
Four 1 0.3 2 0.3 3 0.3
Five+ 0 0 0 0 0 0
Total 368 100 753 100 1,121 100

1000

900

800

700
Full clause
600
Frequency

500 Semi-clause

400 Total

300

200

100

0
One

Two

Three

Four

Five+

Number of IUs per clause

Figure 4.7. Frequency of the number of IUs per clause

This constraint may generally be termed the “no more than two IUs per
clause” constraint (i.e. speakers avoid including more than two IUs per clause).
Note that this is not a categorical rule, and that the overwhelming majority
(81%) of the clauses consist of only one IU. In brief, the results show that
 Chapter 4

grammatically independent single-IU clauses are predominant in Japanese


conversation (cf. Section 4.2).
Further, my analysis revealed that the patterns of IU combination shown in
(4.27) and (4.28) are the most representative of the multi-IU clauses produced
by the speakers. That is, most typically, the multi-IU full clause comprised
[CNPa] (subject argument NP IU) and [CVPp] (verb phrasal IU), whereas
the multi-IU semi-clause comprised [CAvP] (adverbial phrasal IU) and [CVP]
(subjectless semi-clausal IU consisting of a verbal predicate).
(4.27) a. S: sono waku wa, [CNPa]
the limit top
‘the quota’
b. kakuhosareteru kara ne. [CVPp]
secure-pass-sta because fp
‘because is secured’ (M7)
(4.28) a. M: koko de, [CAvP]
here loc
‘here’
b. ichi jikan jugyoo kiite, [CVP]
one hour class listen-and
‘(you) listen to class (lecture) for one hour and’ (M1)

Examples (4.29) and (4.30) illustrate maximally possible “syntactic frag-


mentation” in conversational Japanese.
(4.29) a. A: sofu mo ne, [CNPa]
grandfather also fp
‘(my) grandfather also’
b. su:goku sukide:, [CVPp]
very much like-and
‘likes very much’
c. +nanka kaimono toka ne, [CNPa]
pf shopping and fp
‘uh shopping and’
d. +oryoori toka. [CNPa]
cooking-pol and
‘cooking and so on’ (F4)
Syntactic structure of the intonation unit in conversational Japanese 

(4.30) a. N: ja ichioo, [CAvP]


so provisionally
‘so provisionally’
b. ++ja raishuu wa,@ [CAvP]
so next week top
‘so next week’
c. gogo ni, [CAvP]
afternoon in
‘in the afternoon’
d. +ja kitemorau yooni shite,@@ [CVP]
so come-cau so that do-and
‘so (I’ll) see to it that (I’ll) have (you) visit (me)’ (F6)

In (4.29), three NP IUs, (4.29a) (subject NP IU) and (4.29c, d) (post-predicate


object NP IUs),14 and one VP IU, (4.29b), collectively constitute a multi-IU
full clause. In (4.30), one semi-clausal IU with a verbal predicate, (3.30d), and
three avderbial phrases, (4.30a-c), collectively comprise a multi-IU semi-clause.
As I have shown above, clauses composed of four IUs such as (4.29) and (4.30)
are atypical of the multi-IU clauses in Japanese discourse.
In sum, the results indicate that the full clause contained more IUs than
the semi-clause (1.35 vs. 1.17) on average, and that the average number of IUs
per clause is 1.23. This suggests that conversational Japanese, although it does
not show a one-to-one correspondence between the clause and the IU as Chafe
(1993, 1994) claims that spoken English does, is not as “highly fragmented”
as has been proposed in previous studies (Clancy 1982; Iwasaki & Tao 1993;
Maynard 1989). Given the large proportion of single-IU clauses (81%) vs. the
small proportion of multi-IU clauses (19%), and the predominant use of two
IUs per multi-IU clause as shown in Table 4.13, it is possible to conclude that,
contrary to our expectations, conversational Japanese generally conforms to
the “one IU, one clause” strategy, but probably to a lesser degree than spoken
English (see also Helasvuo’s 2001 discussion on the use of the same strategy in
Finnish conversation).
If the “one IU, one clause” strategy is the “basic”, “unmarked” rule of
IU production in conversational Japanese, then, when do speakers use the
“multiple IUs, one clause” strategy? In other words, what factors lead Japanese
speakers into the non-canonical production of multi-IU clauses? What factors
lead them to divide a clause into separate IUs? Is the information structure
of the IU related to this? If so, in what way? This issue will be explored in
Section 5.6.
 Chapter 4

Table 4.14. Summary of results of IU syntactic structure analysis

Number of clauses/phrases Number of IUs involved


Unit type N % N %
Single-IU clause 910 68 910 57
Multi-IU clause 211 16 467 29
Independent phrase 223 16 223 14
Total 1,344 100 1,600 100

. Summary

Table 4.14 provides a summary of the results of the present IU syntactic


structure analysis.
Table 4.14 indicates, first of all, that the 1,600 substantive IUs constituted
a total of 1,344 clauses or phrases. Of these, about 70% are single-IU clauses;
these are comprised of 910 clausal IUs which occupy about 60% of the to-
tal number of IUs. The percentage of multi-IU clauses is much lower, i.e.
16%, and they are comprised of 467 clausal or phrasal IUs as their constitu-
ents. The remaining 16% are independent phrases that do not belong to the
clausal structure but perform various “peripheral” functions; these phrases
are composed of 223 phrasal IUs. In short, the overwhelming majority of the
syntactic units that the Japanese speakers produced in the present study are
propositionally complete single-IU clauses, rather than multi-IU clauses (in
which a proposition is conveyed by multiple IU constituents) or clause-external
detached phrases.
We have seen in Section 4.1 that the following IU syntactic types are
prevalent in the data:
a. independent clausal IUs ([FC] + [PVP] + [PNP] + [PAP]) (57% of the
1,600 IUs).
b. subjectless semi-clausal IUs ([PVP] + [PNP] + [PAP] + [CVP] + [CNP] +
[CAP]) (49%).
c. independent semi-clausal IUs ([PVP] + [PNP] + [PAP]) (40%).

Of the IU syntactic types based on grammatical categories, the following


have been found to be pervasive:
a. semi-clausal IUs consisting of verbal predicates without overt subjects
([PVP] + [CVP]) (31% of the 1,600 IUs).
b. NP IUs ([NP] = [PNP] + [CNP] + [CNPa] + [CNPp] + [NP]) (26%).
Syntactic structure of the intonation unit in conversational Japanese 

c. full clausal IUs consisting of overt subjects plus predicates ([FC] = [FC] +
[CFC]) (19%).

Further, of the 20 IU syntactic types, the three most pervasive types have
been found to be the following:
a. single-IU semi-clauses with verbal predicates only ([PVP]) (25% of the
1,600 IUs).
b. single-IU full clauses composed of overt subjects plus predicates ([FC])
(16%).
c. single-IU semi-clauses with nominal predicates only ([PNP]) (11%).

In sum, the results indicate that clausal IUs, and especially independent
clausal IUs conveying complete propositions are representative of the substan-
tive IUs produced by the Japanese conversational co-participants. That is, what
this suggests is that the clause is the syntactic exponent of the Japanese substan-
tive IU, as I argued in Section 4.2.2. This finding supports the “clause centrality
proposal” advocated by Chafe and others, suggesting a cross-linguistic applica-
bility of the proposal. This, on the other hand, casts doubt on Iwasaki and Tao’s
(1993) argument for the phrase-oriented nature of Japanese IUs, although, as I
suggested, the treatment of nominal/adjectival predicate clausal IUs is crucially
involved in the clause vs. phrase dichotomy issue in question. Apparently, the
clause centrality that this chapter has found robust in Japanese substantive IUs
merits further investigation.
Example (4.31) consists of [PVP] and [FC] IUs, the two most preferred
syntactic types of the Japanese substantive IU.
(4.31) a. Y: ->mae atashi ga ano yuupen de
before I nom pf UPenn loc
-->benkyooshiteta toki ni ne? [FC]
study-past-prog when loc fp
‘When I was studying at uh UPenn before’
b. S: itsu yuupen nanka itteta no? [PVP]
when UPenn sof go-past-sta fp
‘When were (you) at UPenn?’
c. Y: u:n kocchi ni, [AvP]
pf here all
‘uhm to this place’
d. ++yuusiieruei ga hajimaru mae ni, [FC]
ucla nom begin before loc
‘Before UCLA (classes) began,’
 Chapter 4

e. samaa sesshon totta no yo.$ [PVP]


summer session take-past fp fp
‘(I) took summer session (courses).’
f. S: ->honto wa yuupen ga daiichi kiboo
truth top UPenn nom first choice
-->datta no atashi.$ [FC]
cop-past fp i
‘To tell the truth, UPenn was the first choice, as for me.’ (F1)

The full clause (4.31a) comprises the given subject NP atashi ‘I’ followed by
the new verbal predicate ‘study at UPenn’, where ‘UPenn’ is a newly activated
identifiable NP. The second full clause (4.31d) consists of the given subject
‘UCLA’ and the new verbal predicate ‘begin’. Importantly, the “given”, more
specifically, “accessible”, information status of the subject NP arises from
the utterance of the immediately preceding independent “stray” adverbial IU
(4.31c) ‘to this place’. This “detached” adverbial phrase has led the subject NP
of (4.31d) ‘UCLA’ to be “previously semi-active” in the hearer’s consciousness
(cf. Section 2.3.2). The last full clausal IU (4.31f) contains the given subject
NP ‘UPenn’ and the new nominal predicate ‘first choice’ followed by the given
“genuine” topic NP ‘I’. The semi-clausal IUs (4.31b) and (4.31e) consist of the
verbal predicates ‘go’ and ‘take’, respectively, each carrying new information,
with the subjects ‘you’ and ‘I’ being assumed, respectively.
If these illustrate the typical arrangement of given and new information
within an IU, we can then expect that most of the substantive IUs in Japanese
contain new verbal information only or given nominal information plus new
verbal information. I will explore the information structure of the Japanese IU
in the next chapter.
Chapter 5

Information structure of the intonation unit


in conversational Japanese

This chapter, testing Hypothesis 2.1 and expanding the original research
question RQ2, addresses the following four related questions (Matsumoto
1997a, 1998b, 2000a, 2000b, 2002, 2003):
a. What are the preferred NP types in conversational Japanese in terms of
grammatical role, information status, and syntactic form? What relation-
ships exist among these properties of NPs?
b. What is the preferred information structure of the intonation unit (IU) in
conversational Japanese? What is the preferred arrangement of given and
new information within an IU?
c. What are the preferred clause types in conversational Japanese? What is the
preferred clause structure in terms of the number and type of arguments
contained per clause?
d. Is the information structure of the IU related to the speaker’s produc-
tion of multi-IU clauses, or the breakup of a clause into phrasal IUs in
conversational Japanese? If so, in what way?

. Data coding

In coding the IUs for the preferred NP types, I used three categories of infor-
mation statuses, six categories of NP grammatical roles, and eight categories
of NP syntactic forms. In coding the IUs for the preferred information struc-
ture, I used two categories of information statuses. In coding the clauses for
the preferred clause types, I used five categories. Further, in coding the clauses
for the preferred clause structure, I used three categories of grammatical roles.
The coding categories and their operational definitions follow.
 Chapter 5

Information statuses: Given, accessible, and new

In analyzing the 1,600 substantive IUs for the three information statuses,
given, accessible, and new, I coded only nominal references. As Du Bois
(1987: 816) emphasizes, “practically, information status for nominals is more
amenable to reliable operational definition and quantification”. That is, it is
much easier to determine whether a nominal reference to an entity is new/non-
new, or whether there are one or two new/non-new entities within a single
IU, than it is to determine whether a sequence of verb + adverb, for example,
constitutes two pieces or one complex piece of new/non-new information. I
myself experienced difficulties previously in coding the information statuses of
verbal and adverbial elements in relation to postposing in Japanese discourse
(Matsumoto 1995a). Another justification for not coding verbal mentions
has to do with the fact that events and states are highly transient in active
consciousness, and thus verbs do not usually refer back repeatedly to a single
event or state during the successive production of IUs. This means that new
status is clearly the “default” for verbal references (Chafe 1994; Du Bois
1987; cf. Sections 2.4.5–2.4.6). Moreover, given that verbs normally carry new
information, a more precise characterization of conceptual “unitariness”—
what, for example, verb + adverb combination, should be counted as a unitary
concept—must first be established in relation to the proposed constraints on
new information quantity per IU, if we are to code verbal tokens as well
(see discussion in Section 2.6.2). While we lack such precise definitions of
conceptual unitariness involving verbal elements, it seems safe to confine the
given, accessible, and new statuses, together with identifiabilty, to the discourse
properties of nominal references.
The definitions of the three categories of information statuses are provided
below (cf. Du Bois 1987: 816):
a. given: a referent1 which was mentioned within 25 IUs2 previously in the
discourse (i.e. the transcribed 45-minute conversation), or a referent which
is given from the conversational context itself (e.g. the conversational co-
participants) (Chafe 1976).
b. accessible: a referent which was mentioned more than 25 IUs previously
in the discourse, or a referent which was previously unmentioned in the
discourse but is part of a previously evoked schema (Chafe 1987; Du Bois
1980), or a referent which the speaker assumes to be identifiable to the
hearer (i.e. to be previously semi-active in the hearer’s consciousness) by
situation or prior knowledge already shared by the participants.
Information structure of the intonation unit in conversational Japanese 

c. new: a referent which is neither (a) nor (b), that is, a referent which
was introduced into the discourse as a previously unmentioned, totally
new concept.

Given that previous research has found that given and accessible concepts
virtually exhibit the same patterning (Du Bois 1987), and accessible mentions
appeared rarely in the data, as will be shown below, I will also make a
binary distinction between “Given” and “New” (with capitals G and N), where
“given” and “accessible” concepts are subsumed under the category “Given”
(the categories “New” and “new” refer to identical concepts).

Grammatical roles: S, A, O, and oblique

In obtaining NP types, I first classified nominals into clause-internal NPs and


clause-external NPs. The clause-external NPs are those contained in Type 4
independent phrasal IUs (cf. Section 4.1). I then classified the clause-internal
NPs into two types: arguments and non-arguments. The argument NPs bear a
direct grammatical relation to the verb, whereas the non-argument NPs do not.
I further distinguished three categories of argument NPs and two categories of
non-argument NPs. The arguments consist of three core grammatical roles: S,
A, and O3 (cf. Comrie 1978; Dixon 1979; Du Bois 1987). The non-arguments
consist of oblique NPs and predicate nominals. That is, in my usage, the
roles S, A, and O are clause-internal core arguments which, unlike the other
roles, participate in the structuring of grammatical relations; the obliques and
predicate nominals are clause-internal non-arguments; and the independent
NPs are clause-external constituents. In this study I applied the symbols S, A,
and O to surface grammatical roles.4
The six categories of NP grammatical roles are defined below:
a. S: the single argument of an intransitive verb (i.e. “intransitive subject”),
or the subject of a non-verbal (nominal or adjectival) predicate.
b. A: the argument of a transitive verb that functions as its subject (and could
be the agent NP) (i.e. “transitive subject”).
c. O: the argument of a transitive verb that functions as its direct object (and
could be the patient NP) (i.e. “transitive object”).
d. obliques: clause-internal non-argument NPs other than predicate nomi-
nals, which include NPs contained in adverbial phrases and NPs contained
in attributive adjectival phrases.
 Chapter 5

e. predicate nominals: NPs functioning as predicates.


f. independent NPs: clause-external NPs contained in independent phrasal
IUs (e.g. [NP] IUs).

NP syntactic forms

I distinguished the following eight syntactic forms of NPs. Their simplified


codes, which will be used in Section 5.3, are given in square brackets, and
observed examples are given for each category:
a. personal pronoun [p-pro]: atashi/boku/ore ‘I’, anata ‘you’, kare ‘he’, kanojo
‘she’.
b. demonstrative pronoun [d-pro]: kore ‘this’, sore ‘it’, are ‘that’, soo ‘so’ (as in
soo suru/yuu/omou ‘do/say/think so’).
c. bare noun [bareN]: NPs which are established dictionary items; gakusei
‘student’, shinbun ‘newspaper’, bunka ‘culture’.
d. demonstrative adjective + NP [dem+N]: kono chiiki ‘this area’, ano hito
‘that person’, sonna hanashi ‘such a story’.
e. NP-no + NP [N-no+N]: nihon no daigaku ‘universities in Japan’, Keiko no
tomodachi ‘Keiko’s friend’ (where no is the genitive case marker).
f. adjective + NP [adj+N]: atarashii eiga ‘new movie’, ichi nen ‘one year’
(numeral + NP).
g. complex NP [compNP]: shiranai hito ‘a person whom (you) don’t know’
(relative clause + NP), inakamono tte yuu uwasa ‘the rumor that (he is) a
rustic’ (nominal complement clause + NP).
h. nominalized VP [nomVP]: sara arau no ‘washing dishes’, nihongo hanasu
koto ‘speaking Japanese’ (where no and koto are mominalizers).

I coded each of the 1,600 substantive IUs for overt NPs. I applied the definitions
of the information statuses to all the NPs contained in the IUs, including
“non-referential” predicate nominals which cannot be linked to any specific
referents. I also coded NPs contained in the intra-IU subordinate clauses
and clausal subjects/objects for the three information statuses and surface
grammatical role types as well (cf. Note 7 in Chapter 4). However, I coded
nominals composed of NP-no + NP, complex NPs, and nominalized VPs as a
complex whole with a unitary role and a unitary information status. Further,
I coded the independent NPs as roleless (without S/A/O), given that they are
not clause-internal elements; however, each of them was assigned one of the
information statuses.
Information structure of the intonation unit in conversational Japanese 

Clause types: Transitive, intransitive, and adjectival/nominal predicate

As discussed in Section 2.2.2, transitivity is defined by Hopper and Thompson


(1980) as the global property of a clause involving the effectiveness of an action
transferred from one participant to another. Given its scalar quality, it is useful
to classify clauses in terms of the degree of transitivity. Tao (1996) determined
the degree of transitivity of the clause primarily on the basis of the kinesis of
the verbal action in categorizing Mandarin clauses. Based on his taxonomy, I
classified Japanese clauses into five types, and coded the clauses in the database
according to the following definitions:
a. high transitive clauses: clauses with verbs which take two arguments (A
and O) and denote some action exerting an effect on the direct object NP
(patient argument).
b. low transitive clauses: clauses with verbs which take two arguments (A and
O) and denote some action having relatively little or no effect on the direct
object NP; typically, clauses containing verbs of possession (motsu ‘have’),
verbs of saying (yuu ‘say’), and verbs denoting mental activities (omou
‘think’, shiru ‘know’).
c. intransitive clauses: clauses with verbs which take only one argument (S).
d. adjectival predicate clauses: stative clauses which consist of predicate ad-
jectives, optionally with copula and/or one overt subject argument (S);
given that they can incorporate aspectual marking, modification by
adverbs, and so on, as verbal predicates do, these can be considered as
higher in transitivity than nominal predicate clauses.
e. nominal predicate clauses: descriptive clauses which consist of predicate
nominals, optionally with copula and/or one overt subject argument (S).

. NPs, intonation units, and clauses

Table 5.1 displays the relation between the NPs, IUs, and clauses in the present
corpus. (Note that the number of NPs listed includes that of NPs contained in
the intra-IU subordinate clauses and clausal subjects/objects.)
As shown in Table 5.1, the 1,600 substantive IUs contained 1,417 overt
nominals. This means that the average number of NPs contained per IU is 0.9,
and the average number of NPs contained per clause is 1.1. This suggests that
there almost exists a one-to-one correspondence between the IU/clause and the
 Chapter 5

Table 5.1. Overt NPs, IUs, and clauses


Unit type Number Number Number Average Average
of NPs of IUs of clauses number of NPs number of NPs
per IU per clause
Single-IU 884 910 910 0.97 0.97
clause
Multi-IU 358 467 211 0.77 1.70
clause
Independent 175 233 ____ 0.75 ____
phrase
TOTAL 1,417 1,600 1,121 0.89 1.11

NP in conversational Japanese, such that one IU/clause contains approximately


one overt NP.
Table 5.2 presents the average number of overt NPs per IU by IU syntactic
type.
Table 5.2 indicates that of the four IU syntactic types, the independent
clausal IUs showed the highest rate of NP inclusion, i.e. 0.97 NP per IU on
average. Particularly, the single-IU full clauses ([FC] IUs) exhibited the largest
average number of NPs, i.e. 1.48. The independent semi-clausal IUs with
verbal predicates ([PVP] IUs), although produced most frequently, showed
the average of 0.76 NP. It also indicates that the majority of the 1,417 NPs
are clause-internal NPs (N=1,242, 88%)—mostly, single-IU clause-internal
NPs (N=884, 62%), those occurring within the single-IU full clauses (N=390,
28%), in particular. In short, the results show the speakers’ notable tendency to
produce NPs as elements of the independent single-IU clauses which express a
complete proposition, especially as elements of the single-IU full clauses with
an overtly expressed subject and a predicate.
The occurrence of NPs within IUs/clauses is illustrated in (5.1) (=(4.16)),
where the female co-participants S and Y are talking about danger in New York.
The sequence contains 10 NPs, which are boldfaced.
(5.1) a. Y: kurasu no kotachi ga:, [NP]
class gen student-pl nom
‘(my) classmates’
b. buronkusu? [NP]
Bronx
‘Bronx?’
Information structure of the intonation unit in conversational Japanese 

Table 5.2. Average number of overt NPs per IU by IU syntactic type

IU syntactic type Number Number Average number


of NPs of IUs of NPs per IU
[FC] 390 263 1.48
[PVP] 297 392 0.76
[PNP] 185 174 1.06
[PAP] 12 81 0.15
Total 884 910 0.97
[CFC] 47 40 1.18
[CVP] 55 106 0.52
[CNP] 18 17 1.09
[CAP] 1 7 0.14
Total 121 170 0.71
[CVPp] 11 34 0.32
[CNPa] 95 94 1.01
[CNPp] 23 23 1.00
[CAPa] 15 18 0.83
[CAPp] 1 11 0.09
[CAvP] 60 90 0.67
[CXP] 32 27 1.19
Total 237 297 0.80
[VP] 3 7 0.43
[NP] 112 112 1.00
[AP] 10 14 0.71
[AvP] 39 82 0.48
[XP] 11 8 1.38
Total 175 223 0.78
TOTAL 1,417 1,600 0.89

c. buronkusu tte yuu tokoro ga aru no? [CFC]


Bronx qt say place nom exist q
‘is there a place called Bronx?’
d. +manhattan ni? [CAvP]
Manhatten in
‘in Manhatten?’
e. S: buronkusu∼ [PNP]
Bronx
‘(Is it really) Bronx?’
 Chapter 5

f. Y: burankusu ka na? [PNP]


Branx q fp
(It may be) Branx.’
g. wakannai. [PVP]
know-neg
‘(I) don’t know.’
h. nanka abunai chiiki ga aru n da tte? [FC]
pf dangerous area nom exist nml cop qt
‘(I hear) that there is a dangerous area.’
i. S: atashi mo amerika no koto wa shiranai. [FC]
I also America gen thing top know-neg
‘I don’t know about things in America, either.’
j. Y: <de soko ni i-
and there all
‘and there -’
k. ->soko no rokku konsaato ni itte
there gen rock concert all go-and
-->KAette KOnai. [CVP]
return-and come-neg
‘(they) went to a rock concert there and have never come back or
something’
l. to ka itte::, [CVP]
qt q say-and
‘(they=my classmates) say that’ (F1)

Of the 10 NPs contained in (5.1), two NPs, those in (5.1a, b), are clause-
external independent NPs. The remaining eight NPs are clause-internal NPs.
Of these, five are contained in the single-IU clauses (5.1e, f) and (5.1h, i),
and three are contained in the multi-IU clauses comprised of (5.1c, d) and
(5.1k, l). The eight clause-internal NPs consist of four arguments and four non-
arguments. The four clause-internal argument NPs consist of the following:
two S roles, which occur in (5.1c) and (5.1h), one A role, and one O role, both
of which occur in the full clause (5.1i). One of the S-role NPs is a complex NP;
the other consists of adjective + NP. The A-role NP is a personal pronoun,
and the O-role NP is composed of NP-no + NP. The four non-argument
NPs, on the other hand, consist of two oblique NPs, those in (5.1d) and
(5.1k), and two predicate nominals, those in (5.1e, f). Four of the 10 NPs are
bare nouns.
Information structure of the intonation unit in conversational Japanese 

Does this example illustrate the typical way Japanese speakers use overt
NPs in informal conversation? What types of NPs do they use preferentially?
This issue will be addressed in the next section.

. Preferred NP types: Grammatical roles, information statuses,


and syntactic forms

Table 5.3 displays the distribution of the six grammatical roles among the 20
IU syntactic types (see also Figure 5.1).
Table 5.3. Distribution of grammatical roles among IU syntactic types

Argument NPs Non-argument NPs


IU S A O Oblique Predicate Independent Total
syntactic nominal NP
type
[FC] 238 27 14 59 52 –– 390
[PVP] 16 5 146 115 15 –– 297
[PNP] 1 1 5 178 –– 185
[PAP] 5 6 1 –– 12
Total 255 32 166 185 246 884
[CFC] 36 5 3 3 –– 47
[CVP] 4 25 24 2 –– 55
[CNP] 1 17 –– 18
[CAP] 1 –– 1
Total 40 5 25 29 22 121
[CVPp] 4 6 1 –– 11
[CNPa] 52 15 28 –– 95
[CNPp] 23 –– 23
[CAPa] 15 –– 15
[CAPp] 1 –– 1
[CAvP] 60 –– 60
[CXP] 11 3 8 10 –– 32
Total 63 18 40 92 24 237
[VP] –– –– –– –– –– 3 3
[NP] –– –– –– –– –– 112 112
[AP] –– –– –– –– –– 10 10
[AvP] –– –– –– –– –– 39 39
[XP] –– –– –– –– –– 11 11
Total –– –– –– –– –– 175 175
TOTAL 358 55 231 306 292 175 1,417
(25%) (4%) (16%) (22%) (21%) (12%) (100%)
 Chapter 5

400

350

300

250
Frequency

200

150

100

50

0
Oblique

Predicate
nominal

Independent
NP
S

Grammatical role

Figure 5.1. Distribution of six grammatical roles

Table 5.3 indicates the following: (a) of the 1,417 NPs contained in the
1,600 substantive IUs, clause-internal argument NPs and non-argument NPs
accounted for roughly comparable proportions, i.e. 45% (N=644) and 43%
(N=598), respectively; (b) of the six grammatical role types, S roles exhibited
the highest percentage (25%), which is followed in frequency by obliques
(22%) and predicate nominals (21%); and (c) only 4% are A roles, the
frequency of which is notably low in comparison to the other grammatical
roles. Of the 644 argument NPs, 56% are S roles, 8% are A roles, and 36%
are O roles. This suggests that A is much less likely to be overtly expressed than
O in conversational Japanese, given that each transitive clause in the database
equally contains both an A-role slot and an O-role slot (see Section 5.5). Of the
598 non-argument NPs, obliques and predicate nominals each accounted for
about 50%. Further, 90% of the oblique NPs are those occurring in adverbials.5
Table 5.3 also shows that 67% of the S roles and 49% of the A roles
occurred as subjects of the independent full clausal [FC] IUs (IUs), whereas
63% of the O roles occurred as objects of the independent semi-clausal IUs with
verbal predicates ([PVP] IUs). In addition, 38% of the oblique NPs occurred
Information structure of the intonation unit in conversational Japanese 

in adverbials within [PVP] IUs, 61% of the predicate nominals occurred


in the independent semi-clausal [PNP] IUs, and 64% of the clause-external
independent NPs appeared in the independent [NP] IUs. This means that,
as stated above, the majority of overt nominals were used as clause-internal
elements of the independent single-IU clauses, mostly, [FC] or [PVP] IUs.
The relationships among the grammatical roles, information statuses, and
syntactic forms of NPs are shown in Tables 5.4–5.6.
Table 5.4 indicates that although new NPs exhibited the highest percentage
of the three information statuses, the proportions of Given (given + accessible)
vs. New (=new) NPs are roughly comparable, the percentage of Given only
slightly outnumbering that of New (51% vs. 49%). By constrast, it shows, the
use of accessible NPs is the least preferred in conversational Japanese (this is
consistent with prior research on narrative discourse; similarly, in Du Bois
1987, only 7.5% of all nominal mentions were accessible). Table 5.4 also
indicates the following as to the distribution of Given and New information
within the six grammatical role types: (a) A roles strongly tend to be Given
(80%); (b) S roles and independent NPs tend to be Given (about 60%); (c)
obliques and predicate nominals tend to be New (about 60%); and (d) O roles
are Given/New about 50% of the time (see Figure 5.2). Further, concerning
the grammatical roles in which Given and New information typically occur,
Table 5.4 shows that Given information tends to occur in the S role (30%)
or oblique (18%) positions, whereas New information tends to occur in the
predicate nominal (26%) or oblique (26%) positions.
Several things should be noted in the findings given above with respect to
the relationship between grammatical role and information status. First, the
percentage of New information is notably higher in all role types, compared

Table 5.4. Grammatical role and information status


Argument NPs Non-argument NPs
Information S A O Oblique Predicate Independent Total
status nominal NP
given 189 39 99 100 97 88 612
(53) (71) (43) (33) (33) (50) (43)
accessible 28 5 18 28 15 23 117
(8) (9) (8) (9) (5) (13) (8)
new 141 11 114 178 180 64 688
(39) (20) (49) (58) (62) (37) (49)
Total 358 55 231 306 292 175 1,417
(100%) (100%) (100%) (100%) (100%) (100%) (100%)
 Chapter 5

Total

Independent
NP

Predicate
nominal
new

Oblique accessible

given
O

0% 20% 40% 60% 80% 100%

Figure 5.2. Proportion of three information statuses within grammatical roles

to, for example, Du Bois’ (1987) Sakapulteko narrative data (in Du Bois’ study
20% of the 864 NPs are New; in this study 49% of the 1,417 NPs are New).6 The
higher proportion of New NPs in my Japanese data is clearly attributable to the
fact that Given NPs which were maximally attenuated (i.e. not overtly realized)
were not coded. This led to the lower percentage of Given NPs, which in turn
led to the higher percentage of New NPs. As demonstrated in Section 4.2,
subjectless semi-clausal IUs consisting of verbal/nominal/adjectival predicates
are most prevalent in the data, accounting for 49% (N=777) of the 1,600
substantive IUs; this means that at least 777 Given NPs (probably more if
we include O roles taking zero forms) were not realized in overt forms. We
can thus speculate that if the 777 zero subjects of these semi-clauses had been
overtly expressed as pronouns, as is normal in English, the frequency of Given
NPs would have been much higher—specifically, it would have increased to
67%. Second, the finding that the vast majority of the S and A roles are Given
Information structure of the intonation unit in conversational Japanese 

is compatible with Chafe’s (1994) “light subject constraint”. The finding that
80% of the A roles are Given further indicates that the “Given A Constraint”
(Du Bois 1987) holds in conversational Japanese as well, but not as strongly as
in Sakapulteko narratives, where 97% of the A roles were Given (see further
discussion below in this section). In addition, it is important to note that
the S and A roles—but not O roles—exhibited information-flow properties
noticeably different from obliques. That is, the core subject roles (which are
mostly Given) contrast sharply with obliques (which are mostly New), but O
roles do not. It seems, therefore, that the present study does not provide a
compelling support for Thompson’s (1997) claim for a core-oblique distinction
in discourse as a language universal (in her study of English conversation
core roles and obliques were found to pattern differently, i.e. oblique NPs, as
opposed to core NPs, tend to be new, non-identifiable, and non-tracking).
It seems that in conversational Japanese the core-oblique distinction exists,
but not so strongly as in conversational English (cf. Horie 2001). Finally,
the tendency of the independent NPs to be Given is compatible with the
“redundant” and “peripheral” nature of Type 4 clause-external independent
phrasal IUs (cf. Sections 4.1, 4.5.2).

Table 5.5. Grammatical role and NP syntactic form

NP form S A O Oblique Predicate Independent Total


nominal NP
p-pro 15 15 0 10 2 2 44
(4) (28) (0) (3) (1) (1) (3)
d-pro 47 4 22 11 7 12 103
(13) (7) (9) (4) (2) (7) (7)
bareN 152 21 122 182 144 91 710
(43) (38) (53) (59) (49) (52) (50)
dem+N 34 5 22 34 19 18 132
(9) (9) (9) (11) (7) (10) (10)
N-no+N 35 5 18 21 36 16 131
(10) (9) (8) (7) (12) (9) (9)
adj+N 31 1 28 43 52 17 172
(9) (2) (12) (14) (18) (10) (12)
compNP 32 4 15 5 30 14 102
(9) (7) (7) (2) (10) (8) (7)
nomVP 12 0 4 0 2 5 23
(3) (0) (2) (0) (1) (3) (2)
Total 358 55 231 306 292 175 1,417
(100%) (100%) (100%) (100%) (100%) (100%) (100%)
 Chapter 5

800

700

600

500
Frequency

400

300

200

100

0
dem+N
pro-p

pro-d

bareN

N-no+N

adj+N

compNP

nomVP
NP syntactic form

Figure 5.3. Distribution of eight NP syntactic forms

Table 5.5 indicates that of the 1,417 NPs, (a) 90% (N=1,270) are lexical,
whereas 10% (N=147) are pronominal;7 and (b) 50% (N=710) are bare nouns
without any modifiers, with the other syntactic forms being in low percentages
(see also Figure 5.3). Table 5.5 also shows the following concerning the expres-
sion of the grammatical roles: (a) A roles tend to take the form of bare nouns
(38%) or personal pronouns (28%); and (b) S roles (43%), O roles (53%),
obliques (59%), predicate nominals (49%), and independent NPs (52%) tend
to take the form of bare nouns. Overall, the speakers exhibited a strong propen-
sity to express the grammatical roles with bare nouns, which accounted for
about 40–60% across the role types. Further, the A roles, in comparison to
the other roles, are less likely to be expressed with lexical NPs; while the
overwhelming majority of the S roles (83%), O roles (91%), obliques (93%),
predicate nominals (97%), and independnet NPs (92%) are lexical, 65% of
the A roles are lexical.
Further, Table 5.5 indicates the following regarding the grammatical-role
positions in which the eight syntactic forms typically occur: (a) personal pro-
Information structure of the intonation unit in conversational Japanese 

nouns tend to occur in S or A roles (34% each); (b) demonstrative pronouns


and nominalized VPs appear in S-role positions about 50% of the time; (c) bare
nouns tend to appear in obliques (26%) or S roles (21%); (d) demonstrative
adjective + NPs tend to occur in S roles (26%) or obliques (26%); (e) NP-no +
NPs tend to occur in predicate nominals (27%) or S roles (27%); (f) adjec-
tive + NPs tend to appear in predicate nominals (30%) or obliques (25%);
and (g) complex NPs tend to appear in S roles (31%) or predicate nomi-
nals (29%). In addition, the pronouns occurred tyically in the S-role position
(42%), whereas most of the lexical NPs occurred within the S-role, oblique, or
predicate nominal positions (about 20% each).
Table 5.6 indicates the following pattern as to the expression of activation
cost in conversational Japanese: (a) given information tends to be expressed
with bare nouns (42%) or demonstrative adjective + NPs (18%); (b) accessible
information is expressed overwhelmingly with bare nouns (80%); and (c) new
information tends to be expressed with bare nouns (52%) or adjective +
NPs (19%). This indicates that Japanese speakers strongly tend to express
all three of the activation states of NPs with bare nouns, as is the case with
the expression of the grammatical roles. Table 5.6 also shows the following
concerning the distribution of Given vs. New information within the eight
NP syntactic forms: (a) the personal and demonstrative pronouns expressed
Given information 100% of the time; (b) the overwhelming majority of
demonstrative adjective + NPs expressed Given information (87%); (c) mostly,
NP-no + NPs (74%), adjective + NPs (74%), and complex NPs (76%) expressed
New information; and (d) the bare nouns and nominalized VPs expressed
Given/New information about 50% of the time.

Table 5.6. Information status and NP syntactic form

given accessible new Total


NP form N % N % N % N %

p-pro 44 7 0 0 0 0 44 3
d-pro 102 17 1 1 0 0 103 8
bareN 260 42 93 80 357 52 710 50
dem+N 110 18 5 4 17 2 132 9
N-no+N 29 5 5 4 97 14 131 9
adj+N 37 6 7 6 128 19 172 12
compNP 20 3 5 4 77 11 102 7
nomVP 10 2 1 1 12 2 23 2
Total 612 100 117 100 688 100 1,417 100
 Chapter 5

In sum, the major finding displayed in Tables 5.4–5.6 is that bare nouns,
thus lexical NPs were used preferentially across the grammatical roles and
information statuses. Pertinent to this finding is the infrequent use of pronouns
in the data (only 10% of all overt NPs), and this is most likely to be linked to
the fact that in spoken Japanese many Given NPs are not expressed, or take
zero forms. Given that 65% of the A roles are lexical, it appears that the “Non-
lexical A Constraint” (Du Bois 1987) does not hold in conversational Japanese.
On the other hand, the finding that the A roles contained a relatively higher
percentage of New information, i.e. 20% suggests the weak applicability of the
Given A Constraint in conversational Japanese (cf. Iwasaki 1985). However,
we need to be careful in interpreting these findings in relation to the Non-
lexical A and Given A Constraints. That is, these constraints do not seem to
hold or hold less strongly in Japanese, to the extent that we look at overt NPs
exclusively. As stated above, many Given NPs (including A-role NPs), which
would normally take pronominal forms in other languages such as English,
take zero forms in Japanese. Therefore, it may be that the constraints actually
hold (even though not as strongly as in Du Bois 1987), if we code zero-form
A-role NPs as pronominal (as the most attenuated, special type of pronouns)
and as Given. A roles may be more likely to be non-lexical and Given, if A-
role nominals taking zero forms are equally coded. I will address this issue in
Section 5.5.
Table 5.7 summarizes the results of analysis of the 1,417 overt NPs in terms
of the three features of grammatical role, information status, and syntactic
form (where OBL=oblique, and PN=predicate nominal).
Table 5.7 shows the following. First, overwhelmingly, the speakers pro-
duced the NPs as clause-internal arguments (45%) or non-arguments (43%),
and as bare nouns (50%). Second, the speakers produced a roughly compara-
ble proportion of Given (51%) and New NPs (49%), with Given NPs slightly
outnumbering New NPs. Third, the speakers placed Given information most
frequently in S roles (30%), and they expressed it typically with bare nouns
(48%). By contrast, the speakers placed New information mostly in predicate
nominals (26%) or oblique NPs (26%) (i.e. clause-internal non-arguments),
and they expressed it typically with bare nouns (52%). Thus, the most pre-
ferred type of overt NPs are bare nouns with S role and Given information
status, and bare nouns with non-argument roles and New information status.
As discussed above, partly because maximally attenuated, zero-form Given
nominals were uncoded, the analysis revealed that the percentage of overtly
expressed Given NPs is nearly equal to that of New NPs, exhibiting, as a
result, the higher proportion of New NPs in all grammatical role types. This
Information structure of the intonation unit in conversational Japanese 

Table 5.7. Summary of results of analysis of overt NPs

Information status Grammatical role Syntactic form

Given S N=217 (30%) bareN N=353 (48%)


N= 729 OBL N=128 (18%) dem+N N=115 (16%)
(51%) O N=117 (16%) d-pro N=103 (14%)
New PN N=180 (26%) bareN N=357 (52%)
N= 688 OBL N=178 (26%) adj+N N=128 (19%)
(49%) S N=141 (21%) N-no+N N= 97 (14%)
Total Arguments bareN N=710 (50%)
N=1,417 S N=358 (25%) adj+N N=172 (12%)
(100%) A N= 55 ( 4%) others N=535 (38%)
O N=231 (16%)
Total N=644 (45%)
Non-arguments
OBL N=306 (22%)
PN N=292 (21%)
Total N=598 (43%)
Independent NPs
N=175 (12%)

suggests that conversational Japanese contains both presupposed and asserted


material such that overtly expressed Given and New nominal information
balance out. Put differently, Japanese conversational co-participants produce
and manipulate an approximately equal number of overtly expressed Given
and New NPs in communicating propositional content. In languages like
English, it is normally the case that the major portion of discourse consists of
Given information (anaphoric pronouns) with only a small portion comprised
of New information (full NPs) (Givón 1990). In Japanese, however, which
allows the use of abundant zero-form NPs, a smaller amount of overtly
expressed Given information is utilized. This is presumably more economical
in the sense that the degree of redundancy caused by the repeated expression
of Given NPs is lowered, and therefore the IU production process becomes less
time-consuming.
Related to the higher proportional occurrence of new information in the
present Japanese data is the relatively high ratio of new NPs to IUs or clauses.
That is, an Information Pressure Quotient (IPQ) (Du Bois 1987) turned
out to be fairly high, somewhat contrary to Du Bois’ (1987: 835) prediction
that in interactive conversations, as opposed to narratives, especially intimate
conversations between family members and close friends, information pressure
is often low. The quotients are provided below. These indicate that the data
 Chapter 5

involved higher information pressure conditions than Du Bois’ Sakapulteko


narrative data:8
a. A new argument NP was introduced every 4.21 clauses (IPQ=0.24), and
every 6.02 IUs (IPQ=0.17) on average.
b. A new non-argument NP was introduced every 2.66 clauses (IPQ=0.38),
and 3.79 IUs (IPQ=0.26) on average.
c. A new NP was introduced every 1.63 clauses (IPQ=0.61), and 2.33 IUs
(IPQ=0.43) on average.

The high-rate introduction of new nominal information (i.e. a new NP every


1.6 clauses/2.3 IUs) by the Japanese speakers seems to be related to their
strategy of not expressing overtly already shared, given information. Assuming
that the number of nominals to be expressed in a given conversational space
is limited, the so-called “null argument” strategy is likely to increase the
probability that more new nominals will come into the space, taking the place
of otherwise overtly expressed given NPs. This will supposedly result in an
increase in the number of new nominals.
We now turn to patterns in the arrangement of the Given and New NPs
within the IUs.

. Preferred information structure of the Japanese intonation unit

.. Distribution of IU information structure types

Table 5.8 presents the distribution of the 729 Given (G) NPs and the 688 New
(N) NPs within the 1,600 substantive IUs (see also Figure 5.4). The information
structure types involving multiple NPs represent the linear order of those NPs
in an IU. For example, the type G + N indicates that Given NP is followed by
New NP within an IU.
Table 5.8 indicates that of the total number of IUs, the majority (65%)
contain one NP, 10% contain two NPs, and only 1% contain three NPs (IUs
with four or more NPs were not found). The information structure type “zero
NPs” (24%) is concerned with one of two cases: (a) NP argument slots are not
filled, remaining empty, or (b) the IU originally contains no NP slots, as in
the case of adverbial IUs consisting of intensifiers such as zenzen ‘absolutely’
(see more detailed discussion below). Table 5.8 also shows that 42% (N=676)
contain at least one New NP, 41% (N=656) contain at least one Given NP, 35%
(N=559) contain New NPs only, 34% (N=539) contain Given NPs only, and
Information structure of the intonation unit in conversational Japanese 

Table 5.8. Distribution of IU information structure types

IU information structure type N %

Zero NPs 385 24.0


N 547 34.3
G 483 30.3
G+N 86 5.4
G+G 50 3.1
N+G 20 1.2
N+N 12 0.7
G+G+N 7 0.4
G+G+G 5 0.3
G+N+G 3 0.2
N+G+G 2 0.1
Total 1,600 100

only 7% (N=117) contain both Given and New NPs. The three most preferred
IU information structure types are thus the following: (a) IUs with one New NP
(34%), (b) IUs with one Given NP (30%), and (c) IUs with zero NPs (24%).
It also shows that (a) of the IUs containing two NPs, the type G + N is most
prevalent (51%); and (b) of the IUs containing three NPs, the type G + G + N
is most prevalent (41%).
The results given above suggest a number of constraints on the quantity of
explicit NPs containable within one IU in conversational Japanese. The finding
that none of the IUs in the database contained more than three NPs suggests
a constraint on the maximum number of NPs that may be overtly expressed
within an IU. That is, one IU may contain no more than three overt NPs. The
results show that while the overall production of NPs strictly conforms to this
constraint, which confines the upper limit of NP quantity per IU to three, the
maximum number of Given NPs per IU is three, whereas the maximal quantity
of New NPs per IU is two. This suggests that one IU may contain no more than
three Given NPs and no more than two New NPs.
Another important finding is that among the IUs with zero, one, two, and
three NPs that are permitted by this constraint, IUs with only one NP are most
prevalent, whereas IUs with zero NPs are less common, and IUs with two or
three NPs are quite rare. This indicates that the speakers prefer to use NPs
singly rather than multiply each time they produce an IU, such that each IU
contains one piece of Given or New nominal information. More specifically,
the speakers used the strategy of placing “one NP per IU” 65% of the time,
and the strategy of placing “zero or one NP per IU” 89% of the time. This
 Chapter 5

600

500

400
Frequency

300

200

100

G+G+G

G+N+G

N+G+G
Zero NPs

G+N

G+G

N+G

N+N

G+G+N

IU information structure type

Figure 5.4. Distribution of IU information structure types

means that overwhelmingly, the speakers avoided using more than one overt
NP at a time, i.e. within a single IU; this may be termed the “one overt NP per
IU constraint”. We should also note that although the use of two or three NPs
per IU was found to be uncommon, among those more complex IUs, the most
preferred types are G + N and G + G + N, as noted above. This is compatible
with the information structure of the IU that has been argued to be unmarked
in English discourse (i.e. information flows from Given to New within an IU)
(Chafe 1994; Halliday 1985, 1994; cf. Section 2.4.5).
Table 5.9 displays a breakdown of the IUs with zero NPs into two types:
(a) cases of “zero anaphora” which involve unfilled NP argument slots, and (b)
cases of “no associated NPs” which originally lack NP slots. In the former case
of zero anaphora, IUs contain no NPs because core arguments are not overtly
expressed (this type includes “non-anaphoric zeros”, which always remain
empty in Japanese and which would be supplied typically by generic they or
temporal/meteorological it in English; cf. Section 5.5.1). In the latter case of
no associated NPs, IUs contain no NPs either because they originally lack NP
Information structure of the intonation unit in conversational Japanese 

Table 5.9. Two types of IUs with zero NPs

IU type N %

Zero anaphora
Ø-S 164 43
Ø-A 62 16
Ø-A+O 34 9
Ø-O 4 1
Total 264 69
No associated NPs
Adverbial phrase 75 20
Verb phrase 25 6
Adjectival phrase 17 4
Mixed phrase 4 1
Total 121 31
TOTAL 385 100

slots or because core arguments are already present in adjacent IUs. (In coding
the IUs for existence/non-existence of core argument NPs, when both main
and subordinate/embedded clauses are present in an IU, only the main clause
was coded.)
Table 5.9 indicates that the cases of zero anaphora accounted for about
70% of the IUs with zero NPs. It also shows that 43% of such IUs have their
S-role slots unfilled with overt NPs. Of the remaining cases of no associated
NPs, the majority are adverbial phrasal IUs which originally lack NP slots.
This is followed by verb phrasal IUs, in which no NPs need to be present
because associated core arguments are contained in the immediately following
or preceding IUs. In sum, the majority of the IUs with no explicit NPs are
clauses involving zero anaphora, typically intransitive or nominal/adjectival
predicate clausal IUs whose S-role slots are unfilled (see further discussion in
Section 5.5).

.. The one new NP per IU constraint

The finding that 12 of the 1,600 IUs contained two new NPs merits discussion,
especially in relation to the previously formulated constraints on the quantity
of new information per IU discussed in Section 2.6.2. To be noted first of all is
that the IUs with two new NPs are extremely limited, occupying only 0.7%
of the total number of IUs. This means that overwhelmingly, the speakers
avoided introducing more than one new NP per IU; I will refer to this as
 Chapter 5

the “one new NP per IU constraint”. This constraint is basically equivalent


to Chafe’s (1987, 1994) “one new concept at a time/one new idea constraint”,
which states that a single IU can express no more than one new concept or idea.
The one new NP per IU constraint can also be related to Givón’s (1975, 1984)
“one [new] unit per proposition” constraint/“one [new]-chunk-per-clause
processing principle”, and Du Bois’ (1987) “One New Argument Constraint”
(i.e. speakers avoid more than one new argument per clause) as well.
The operation of the one new NP per IU constraint is illustrated by (5.2)–
(5.3), where new NPs are boldfaced.
(5.2) K: shusseki ritsu takai mono. [FC/S]
attendance percentage high because
‘Cause the percentage of (my class) attendance (is) high.’ (M5)
(5.3) Y: samaa sesshon totta no yo. [PVP/O]
summer session take-past fp fp
‘(I) took summer session (courses).’ (F1)

Both (5.2) and (5.3) are single-IU clauses each of which expresses a complete
proposition. Example (5.2) is a full clausal IU, which contains only one new
NP, ‘the percentage of attendance’ (S-role argument of the adjectival predicate
‘is high’); (5.3) is a semi-clausal IU, which contains only one new NP, ‘summer
session (courses)’ (O-role argument of the transitive verb ‘take’).
The one new NP per IU constraint cannot be formulated as a categorical
rule, unlike Du Bois’ (1987) One New Argument Constraint (none of his
“clause cores” contained two new-argument mentions); but it does indicate
a strong tendency in the present conversational Japanese data. Let us now
examine the relation between these two constraints more closely.
Table 5.10 presents a breakdown of the 12 IUs containing two new NPs into
four types. Further analysis has shown that all of them are clausal, of which six
are single-IU full clauses with overt subjects, and six are subjectless single-IU
semi-clauses (none of the phrasal IUs contained two new NPs).

Table 5.10. Types of IUs with two new NPs

Type and order of two new NPs N %

Oblique (Adv) + Argument 7 58


Argument + Oblique (Adv) 3 25
Oblique (Topic) + Argument 2 17
Argument + Argument 0 0
Total 12 100
Information structure of the intonation unit in conversational Japanese 

As shown in Table 5.10, of the 12 IUs, those containing one new oblique NP
followed by one new argument NP are most prevalent (accounting for 75%);
and most importantly, none of the 12 IUs contained two new argument NPs.
An example of the most frequently found combination of two new NPs within
an IU is given in (5.4).
(5.4) T: rainen ni kootoo shiken ukeru yo. [PVP/OBL·O]
next year loc oral exam take fp
‘In the next year, (I) will take an oral exam.’ (M4)

Example (5.4), an independent subjectless semi-clausal IU, contains two newly


introduced nominals (which are boldfaced), i.e. the obique NP rainen ‘next
year’, which occurs within the adverbial phrase, and the O-role argument
NP kootoo shiken ‘an oral exam’, which functions as the direct object of the
transitive verb ‘take’. Note that such IUs with two new nominals are extremely
exceptional in actual conversational discourse.
What these results indicate is that while the speakers overwhemingly
avoided more than one new NP per IU, they absolutely avoided more than
one new argument NP per IU. This means that the one new NP per IU
constraint that I formulated above does not contradict Du Bois’ (1997) One
New Argument Constraint. On the contrary, the two are compatible with each
other. Since Du Bois (1987) excluded obliques from his “clause core” and
did not code them in formulating his PAS, his constraints cover only part of
the NPs, namely arguments; they do not say anything about NPs in oblique
positions (cf. Section 2.6.2). Pointing to the narrow scope of his constraints,
he himself suggests the possibility that an IU may contain more than one
new nominal mention (Du Bois 1987: 832–833). What I have shown in this
study by coding all types of NPs—not only arguments but also non-arguments
including obliques—is that one IU may actually express two new nominal
concepts, to the extent that one of the two new NPs is argument and the
other is oblique. That is, “two new NPs per IU” (the occurrence of two new
NPs within one IU) is possible, to the extent that “two new arguments per
IU” is avoided; this is consistent with the One New Argument Constraint (cf.
Kärkkäinen 1996).
In sum, the results given and discussed above regarding the preferred
information structure of the substantive IU in conversational Japanese can be
summarized as follows. First, the speakers included no more than three explicit
NPs per IU; they placed preferentially zero NPs or one overt NP within an
IU, conforming to the “one overt NP per IU constraint”. More specifically,
“one New NP” or “one overt Given NP” has been found to represent the most
 Chapter 5

preferred information structure of the IU in conversational Japanese. Second,


the speakers avoided introducing more than one new NP per IU (although
not absolute avoidance), conforming to the “one new NP per IU constraint”
(cf. Chafe’s 1987, 1994 one new concept at a time/one new idea constraint).
Third, the speakers’ introduction of two new NPs within one IU (“two new
NPs per IU”), which itself rarely occurs, is governed by the constraint that
the two new NPs are argument + oblique combinations. This lends support
to Du Bois’ (1987) One New Argument Constraint. (See related discussion of
the relationship between the one new NP per IU constraint and the speaker’s
production of multi-IU clauses in Section 5.6.)

. Preferred clause structure in conversational Japanese

.. Preferred clause types and preferred argument structure

The previous section dealt with the preferred ways Japanese speakers arrange
Given and New NPs as they successively produce IUs and clauses in conver-
sational interaction. The coded NPs include all the NPs contained in the IUs,
viz. arguments (S, A, and O), non-arguments (obliques and predicate nomi-
nals), and clause-external independent NPs. Those NPs contained in the intra-
IU subordinate/embedded clauses were also coded. This section examines the
preferred ways Japanese speakers structure different types of clauses in terms of
the number and type of arguments contained per clause. Thus, only argument
NPs are coded in this section.
Table 5.11 presents the distribution of the 1,121 clauses (which do not
include intra-IU clauses) among the five clause types with different degrees
of transitivity and with different numbers of overtly expressed arguments
(which include clausal arguments, mostly clausal objects). In Table 5.11, Ø
indicates zero arguments, i.e. arguments that are subcategorized by the verb
(A and O for transitive verbs, S for intransitive verbs and adjectival/nominal
predicates) but are not overtly realized. As in Table 5.9, zero arguments
include both “anaphoric zeros” (i.e. referential zeros which can be linked with
specific referents previously mentioned) and “non-anaphoric zeros” (i.e. non-
referential zeros not derived from prior mentions and thus independent of
anaphoric continuity processes; inherently null arguments which always take
zero forms) (cf. Tao 1996; Section 5.4.1).
Table 5.11 indicates that (a) of the 910 single-IU clauses, intransitive (29%)
and nominal predicate clauses (25%) are most common; and (b) of the 211
Information structure of the intonation unit in conversational Japanese 

multi-IU clauses, low transitive (31%) and intransitive clauses (28%) are most
prevalent. Of the 1,121 clauses, intransitive clauses are most pervasive (29%),
followed by nominal predicate (23%), low transitive (21%), and adjectival
predicate clauses (16%), whereas high transitive clauses are least common
(11%) (see also Figure 5.5). Table 5.11 also shows that transitive clauses (32%)
are about as frequent as intransitive clauses (29%). In sum, the three most
preferred clause types are, with descending frequency, intransitive, nominal
predicate, and low transitive clauses, with high transitives being the most
dispreferred clause type.
Of the total number of clauses, one-participant clauses, or clauses with
predicates which take only one argument (S) accounted for 68%, whereas
two-participant clauses, or clauses with predicates which take two arguments
(A and O) accounted for 32%. This means that the majority of the clauses
are non-transitive, or one-participant clauses which are low in transitivity
(cf. Hopper & Thompson 1980; Section 2.2.2). Of all the one-participant
clauses which constitute the majority, 43% are intransitive, 34% are nomi-
nal predicate, and 23% are adjectival predicate clauses. This indicates that the
perferred one-participant clause types are intransitive or nominal predicate

Table 5.11. Frequency of clause types by the number of overt arguments

Clause Transitive Transitive Intransitive Adjectival Nominal Number of


type high low predicate predicate overt
arguments
Single-IU Ø 22 Ø 21 Ø 159 Ø 81 Ø 174 Zero 457
clause A 6 A 4 S 107 S 76 S 52 One 435
O 61 O 129 Two 18
A+O 4 A+O 14
Total 93 Total 168 Total 266 Total 157 Total 226 Total 910
Multi-IU Ø 4 Ø 1 Ø 13 Ø 6 Ø 11 Zero 35
clause A 2 A 4 S 47 S 14 S 24 One 162
O 19 O 52 Two 14
A+O 6 A+O 8
Total 31 Total 65 Total 60 Total 20 Total 35 Total 211
Clause Ø 26 Ø 22 Ø 172 Ø 87 Ø 185 Zero 492
Total A 8 A 8 S 154 S 90 S 76 One 597
O 80 O 181 Two 32
A+O 10 A+O 22
Total 124 Total 233 Total 326 Total 177 Total 261 Total 1,121
(11%) (21%) (29%) (16%) (23%) (100%)
 Chapter 5

350

300

250
Frequency

200

150

100

50

0
Intransitive
high

low

Adjectival

Nominal
Transitive

Transitive

predicate

predicate
Clause type

Figure 5.5. Frequency of five clause types

clauses. Of all the two-participant clauses which are the minority, low transi-
tives accounted for 70%. This means that even the two-participant clauses tend
to be low in transitivity, not involving action verbs which exert an effect on the
direct object NP. The findings suggest that conversational Japanese overwhelm-
ingly consists of clauses at the lower extreme of the transitivity continuum—
clauses unrelated to events or actions which typically express speakers’ feel-
ings, attitudes, and opinions, or subjective stance. In short, the results of this
study are in support of the claim for the low transitivity-centered and non-
event-oriented nature of conversational discourse (cf. Tao 1996; Thompson &
Hopper 2001).
The analysis has revealed that overwhelmingly, conversational Japanese
consists of non-transitive, one-participant clauses which are very low in tran-
sitivity. That is, conversational Japanese, and possibly conversational discourse
in general, is preferentially composed of non-event- or non-action-related ma-
terials such as descriptions of circumstances, expressions of mental states and
subjective opinions and evaluations. Thus, viewed in terms of transitivity, con-
versation is reminiscent of what Hopper and Thompson (1980) identified as
Information structure of the intonation unit in conversational Japanese 

the “backgrounded” portions of narrative discourse. The findings suggest that


the role transitive clauses play in ordinary conversation is remarkably small,
in contrast to the fact that much attention has been given in recent functional
linguistics to highly transitive clauses. That is, while most current studies of
clause grammar are rigidly based on highly transitive examples, everyday spon-
taneous conversation abounds in clauses of very low transitivity. This suggests
that more attention should be directed to studies of the grammar of non-
transitive, one-participant clauses, instead of clauses of high transitivity which
are rarely produced in actual everyday conversational interaction.
The low transitivity of conversational language seems to be determined
largely by what kinds of things we do when we engage in talk-in-interaction.
In ordinary conversation, rather than talking about events or actions, we
tend to describe states, convey our attitudes, express our perceptions, feelings,
and opinions, and provide our assessments of people and situations. That is,
ordinary talk is mostly concerned with speakers’ points of view, i.e. how things
are from speakers’ perspective. Importantly, what these reflect is “subjectivity”
in our use of language in everyday conversational interactions (cf. Iwasaki
1993; Maynard 1993; Scheibman 2002; Stein & Wright 1995). In short, “the
degree of transitivity of the utterances people use in everyday interaction, and
indeed the very nature of clause grammar itself, is tightly related to what they
are doing with their talk” (Thompson & Hopper 2001: 54).
Table 5.12 presents the distribution of the five types of clauses with zero,
overt S, A, and O arguments (see also Figure 5.6).
Table 5.12 indicates the following concerning the quantity and type of
overtly expressed arguments: (a) both of the high and low transitives tend to
contain one overt argument, typically O; (b) about half of the intransitives
and adjectival predicate clauses contain overt S argument, whereas nominal
predicate clauses tend to occur without overt S argument; and (c) transitives
with overtly expressed A and O are rare, accounting for only 9% of all the
transitives, and only 3% of all the clauses. Of the total number of clauses,
those without overtly expressed arguments are most pervasive (44%), those
containing overt S (29%) or overt O (23%) are less frequent, whereas clauses
with overt A are extremely rare, accounting for only 1%. (The finding that
S roles in nominal predicate clauses tend to remain zero could be related to
the “one overt NP per IU constraint” formulated above in Section 5.4.1. The
presence of overt S in nominal predicate clauses means “two overt NPs within
one clause/IU”, in violation of the constraint. S roles in adjectival predicate
clauses, on the other hand, can be explicit more freely because the constraint
would not be affected by the presence of overt S.)
 Chapter 5

Table 5.12. Frequency of clauses with overt S, A, and O arguments

Clause type Zero S A O A+O Total

Transitive 26 –– 8 80 10 124
high (21) (6) (65) (8) (100%)
Transitive 22 –– 8 181 22 233
low (9) (4) (78) (9) (100%)
Intransitive 172 154 –– –– –– 326
(53) (47) (100%)
Adjectival 87 90 –– –– –– 177
predicate (49) (51) (100%)
Nominal 185 76 –– –– –– 261
predicate (71) (29) (100%)
Total 492 320 16 261 32 1,121
(44) (29) (1) (23) (3) (100%)

We can see from Table 5.11 that (a) about 50% of the single-IU clauses
contained zero arguments, and about 50%, one overt argument, (b) the ma-
jority (about 80%) of the multi-IU clauses contained one overt argument, and
(c) the multi-IU clauses contained a higher proportion of two overt arguments
than the single-IU clauses (7% vs. 2%). As shown in Table 5.13, of the 1,121
clauses, 53% contained one overt argument, 44% contained zero arguments,
and only 3% contained two overt arguments (see also Figure 5.7). Most im-
portantly, clauses with two overt arguments were produced remarkably infre-
quently by the speakers. That is, 91% of the transitives and 97% of the clauses
contained zero or one overt argument. This suggests Japanese speakers’ no-
table preference for use of zero-argument or one-argument clauses over use of
two-argument clauses. The preferred clause types that the present study found
are thus the following: clauses with zero arguments or one overt argument,
typically, non-transitive clauses with zero arguments or overt S, and transitive
clauses with overt O only.
The finding that clauses with two overt arguments are extremely rare
has led me to propose the “one overt argument per clause constraint”. What
this constraint says is that speakers avoid more than one overt argument per
clause, such that one clause contains zero arguments or one overt argument
(typically S or O). The preferred surface syntactic structure of the clause in
conversational Japanese can hence be represented as follows (N=argument NP,
and P=predicate):
Information structure of the intonation unit in conversational Japanese 

Total

Nominal
predicate A+O

Adjectival O
predicate
A

Intransitive S

Zero
Transitive
low

Transitive
high

0% 20% 40% 60% 80% 100%

Figure 5.6. Proportion of clauses with zero, one, and two arguments

(5.5) N(S/O) P

The configuration (5.5) indicates that in the unmarked word order situation
which does not involve post-predicate NPs, the maximal surface structure
which is consistently preferred in Japanese discourse is a single overt argument
in the S or O role followed by a predicate, either verbal, adjectival, or nom-
inal. (Note that the argument NP may not be overt; the surface clause form,

Table 5.13. Frequency of clauses with zero, one, and two overt arguments

Clause type N %
Zero arguments 492 44
S 320 29
A 16 1
O 261 23
One argument-Total 597 53
Two arguments (A+O) 32 3
TOTAL 1,121 100
 Chapter 5

500

450

400

350

300
Frequency

250

200

150

100

50

0
Ø S A O A+O
Argument type

Figure 5.7. Frequency of clauses with zero, overt S, A, and O arguments

P with no overt argument NP is also one variant of the preferred pattern.) This
is perfectly compatible with the preferred clause structure in Sakapulteko pro-
posed in Du Bois (1987: 823), except that the order of argument and predicate
(verb) is reversed (Japanese is a verb-final language, whereas Sakapulteko is a
verb-initial language).9
The argument structure in (5.5) also indicates that although linguists
generally define Japanese as an SOV (Subject-Object-Verb) language, and have
treated sentences with two overt lexical arguments such as John ga Mary o
butta. ‘John hit Mary.’ (Kuno 1973: 3) and Taroo ga hon o katta. ‘Taro bought
a book.’ (Shibatani 1990: 258) as if they were representative clausal forms,
this treatment can never be justified, given that in naturally occurring spoken
discourse such highly transitive two-argument structures are rarely produced
(see also Iwasaki 2002: Chapter 6). In summary, the vast majority of clauses in
Japanese conversation are non-transitive, one-participant clauses with zero or
overt S arguments, and the minority transitive, two-participant clauses, which
themselves tend to be low in transitivity, overwhelmingly contain overt O
Information structure of the intonation unit in conversational Japanese 

arguments only; highly transitive clauses with two overtly expressed arguments
are atypical in conversational discourse.
The clause types most preferentially produced in conversational Japanese
are illustrated in (5.6)–(5.8). Example (5.6) is an intransitive clause with overt S
‘tigers’; (5.7) is a nominal predicate clause without overt S, which solely consists
of the complex predicate nominal ‘a person (you) don’t know’; and (5.8) is a
low transitive clause with overt O ‘grade’.
(5.6) M: tora toka mo iru no? [FC]
tiger sof also exist q
‘Are there tigers (there), too?’ (F5)
(5.7) Y: shiranai hito? [PNP]
know-neg person
‘(Is she) a person (you) don’t know?’ (F2)
(5.8) K: MAda ((laugh)) gureido shiranai no. [PVP]
yet grade know-neg q
‘(You) don’t know (your) grade yet.’ (M5)

The one overt argument per clause constraint formulated above is reminis-
cent of Du Bois’ (1987) “One Lexical Argument Constraint”, which prohibits
more than one lexical argument per clause. When we compare the two con-
straints in terms of their applicability in conversational Japanese, we would
need to note that arguments in the one overt argument per clause constraint
include both lexical and pronominal NPs (pronouns accounted for only about
10% of all overt NPs), and the lexical arguments are a subset of the overt argu-
ments. Thus the fact that the one overt argument per clause constraint holds
in the Japanese data suggests that the One Lexical Argument Constraint will
hold as well. In sum, there exists a notable tendency in conversational Japanese
to limit the quantity of overtly expressed arguments within a clause to a
maximum of one. That is, speakers either overtly express only one argument
(preferentially S or O) or do not express arguments at all within one clause,
with the simultaneous expression of A and O arguments per clause being
strongly dispreferred. (The skewed selection of O over A as a single argument
that is allowed to be overtly expressed by the one overt argument per clause
constraint is clearly pertinent to the finding discussed in Section 5.3 that A is
overwhelmingly given and O tends to carry more newly introduced referents.)
 Chapter 5

.. Overt vs. null arguments

Table 5.14 presents the proportion of three argument slots that the speakers
filled with overt NPs (see also Figure 5.8).
Table 5.14 indicates, first of all, that of the 1,478 argument slots contained
in the 1,121 clauses, 45% (N=661) were filled with either overt S, A, or O,
whereas 55% (N=817) received no overt coding, remaining zero. It also indi-
cates the following regarding the proportion of the three argument slots filled
vs. unfilled with overt NPs: (a) of the 764 S-role slots, 42% (N=320) were filled
with overt NPs (58% were left unfilled); (b) of the 357 A-role slots, only 13%
(N=48) were filled (87% were left unfilled); and (c) of the 357 O-role slots,
82% (N=293) were filled with overt NPs (18% were left unfilled). We can
also see from Table 5.13 that (a) the percentages of unfilled A-role slots are
comparable between the high and low transitives; (b) the percentage of filled
O-role slots is higher in the low transitives; and (c) as we have already seen
above, while nearly 50% of the S-role slots of the intransitive and adjectival
predicate clauses were filled, those of the nominal predicate clauses tend to
remain zero.
The results indicate an interesting pattern in the overt expressibility of the
three roles. That is, O-role slots tend to receive overt coding, whereas S-role
and A-role slots tend to remain zero forms. In particular, the finding that the
speakers did not fill about 90% of the available A-role slots with overt NPs
should be noted. What this suggests is that the transitive subject slot (A),
by comparison to the transitive object (O) and intransitive subject (S) slots,
is strongly dispreferred as a site for overt NPs (cf. Clancy 2003). Assuming

Table 5.14. Proportion of overt arguments by clause type

S A O Total
Clause type N % N % N % N %

Transitive ____ 18 14 90 73 108 44


high
Transitive ____ 30 13 203 87 233 50
low
Intransitive 154 47 ____ ____ 154 47

Adjectival 90 51 ____ ____ 90 51


predicate
Nominal 76 29 ____ ____ 76 29
predicate
Total 320 42 48 13 293 82 661 45
Information structure of the intonation unit in conversational Japanese 

TOTAL

O-Total

O-Transitive
low
O-Transitive
high

A-Total
Ø
A-Transitive
low
Overt
A-Transitive
high

S-Total

S-Nominal
predicate
S-Adjectival
predicate

S-Intransitive

0% 20% 40% 60% 80% 100%

Figure 5.8. Proportion of overt vs. null arguments

that new information must always be overt, this means that A-role slots are
disfavored for the expression of new nominal referents. In other words, A-
role slots constitute preferentially the sites for presupposed, given information.
What pertains to this observation is the speculation I have made above in
Section 5.3—that if we code zero-form A role NPs as Given and as the most
attenuated, special type of pronouns, both the Given A and Non-lexical A
Constraints (Du Bois 1987) will hold in conversational Japanese. Given that
the overwhelming majority of the A-role NPs have been found to take zero
forms, thus Given and pronominal, this leads us to the conclusion that the
Given A Constraint, which was found to hold only weakly, and the Non-lexical
 Chapter 5

A Constraint, which was found to be inapplicable, when we looked at overtly


expressed NPs only, actually hold in conversational Japanese as well, as in many
other languages.
In sum, with respect to allowing new information and lexical NPs, the way
S and O pattern is distinct from the way A patterns in discourse. That is, S and
O roles are more likely than A roles to accomodate new and lexical NPs. In
Japanese conversation about 90% of A roles take zero forms, thus given and
non-lexical; and overtly expressed A roles themselves strongly tend to be given
and less lexical than the other roles. Interestingly, the present study suggests
that S and O, at least in Japanese, do not seem to pattern in the same way; it has
been found that O roles are more likely than S roles to contain new information
and lexical NPs (in Du Bois 1987, for example, S and O were found to pattern
almost in the identical way). This merits further examination. As to the degree
of allowing new information and lexical NPs, then, the present corpus suggests
a continuum: O is most likely to be new and lexical, A is most likely to be non-
new (given) and non-lexical (pronominal), and S lies somewhere between O
and A (see related discussion of core-oblique distinction in Section 5.3). (The
non-appearance of new and lexical mentions in A roles can also be linked to the
role of topic continuity in discourse—the A position is more likely to be filled
by human agent arguments which tend to persist as topic through successive
clauses and thus strongly tend to be given and pronominal; cf. Du Bois 1987,
Section 2.1.)

. Multi-IU clauses and the one new NP per IU constraint

I now take up the question that I raised in Section 4.7: what factors are
responsible for the production of multi-IU clauses? Discussion of this issue
seems relevant at this point because this can be related to the “one new NP per
IU constraint” formulated above. We should remember that this constraint is
on the amount of new information to be contained within an IU, not within a
clause. What this suggests is that a clause composed of multiple IUs may con-
tain freely more than one new NP. Given the finding that the majority of the
multi-IU clauses consist of two or three IUs (see Section 4.7), it is conceivable
that one multi-IU clause may contain maximally two or three new NPs. My
analysis has shown that this is certainly the case with the multi-IU clauses
contained in the Japanese data. As displayed in Table 5.15, of the 211 multi-
Information structure of the intonation unit in conversational Japanese 

Table 5.15. Proportion of IUs/clauses with more than one new NP

Total number of IUs/clauses IUs/clauses with


more than one new NP
Unit type N N %
IU 1,600 12 0.7
Single-IU clause 910 12 1.3
Multi-IU clause 211 68 32.2
Clause 1,221 80 6.5

IU clauses, 32% (N=68) contained more than one newly introduced NP. By
contrast, only 1.3% (N=12) of the 910 single-IU clauses contained more than
one new NP (see also Figure 5.9).

100%

90%

80%

70%

60%
Frequency

More than one new NP


50%
One new NP
40%

30%

20%

10%

0%
IU

clause

clause
Single-IU

Multi-IU

Clause

Unit type

Figure 5.9. Proportion of IUs/clauses with more than one new NP


 Chapter 5

The occurrence of multiple new NPs within a multi-IU clause is illustrated


in (5.9)–(5.11), where new NPs are boldfaced.
(5.9) a. H: =roozu ga sa, [CNPa/S]
Rose nom fp
‘Rose’
b. baito kara ne? [CAvP/OBL]
part-time job abl fp
‘from (her) part-time job’
c. ->+juu:ichi ji han gurai ni
eleven o’clock half about loc
- ->kaette kuru. [CVPp/OBL]
return-and come
‘comes back around eleven thirty’ (F8)
(5.10) a. M: demo katei de, [CAvP/OBL]
but home loc
‘but at home’
b. minna iwatteru n desho? [CFC/A]
everybody celebrate-prog nml tag
‘everybody is celebrating (Easter)’
c. +chanto. [CAvP]
properly
‘properly’ (F2)
(5.11) a. Y: shakai men no wadai nanka wa, [CNPa/O]
social page gen topic sof top
‘topics (such as those) in the local news page (of the newspaper)’
b. terebi kara, [CAvP/OBL]
tv abl
‘from TV’
c. +erareru kara, [CVP]
get-pot because
‘because (we) can get’ (M8)

The multi-IU full clause (5.9), composed of three IUs, contains three new NPs
(one S-role NP and two oblique NPs). The three IUs each contain one newly
introduced NP, thus conforms to the one new NP per IU constraint. In the
multi-IU full clause (5.10), two new NPs (one A-role NP and one oblique
NP) are introduced; the post-predicate adverbial IU (5.10c) originally lack NP
slots (this illustrates the case of “no associated NPs” discussed in Section 5.4.1).
Likewise, the multi-IU semi-clause (5.11) contains two new NPs (one O-role
Information structure of the intonation unit in conversational Japanese 

NP and one oblique NP); the semi-clausal IU (5.11c) contains no overt A-


role NP.
What is to be noted here is that in each case, if the speaker had uttered
the new NPs within a single-IU clause instead of placing them within the
multi-IU clause as they actually did, that single-IU clause would have con-
tained two or three new NPs, in violation of the one new NP per IU con-
straint. In other words, if the speaker had not opted to express the proposi-
tion by taking the form of the multi-IU clause, within which one IU contains
one new NP, that is, if the speaker had not divided the clause into separate
IU constituents, an independent clausal IU containing multiple new nominal
references would have resulted. Conversely, we can say that the speaker pro-
duced the multi-IU clause, that is, divided or fragmented the clause into sepa-
rate IU elements, in order to conform to the one new NP per IU constraint.
Given that almost one-third of the multi-IU clauses involved multiple new
NPs, it seems clear that the constraint is one of the important potential forces
that has motivated the speakers to produce multi-IU clauses. However, con-
sidering the fact that the remaining two-thirds did not involve multiple new
NPs, it appears, at the same time, that the constraint is not solely responsi-
ble for the production of multi-IU clauses. Other factors are also involved in
the speaker’s tendency to break the clause into separate IUs; this would merit
further investigation.
In sum, I propose that the speaker’s production of multi-IU clauses can be
explained, although not entirely, in terms of the one new NP per IU constraint.
The Japanese conversational co-participants are led to divide a clause into
separate IUs, placing one new nominal concept within one IU, by the operation
of the one new NP per IU constraint.10
Finally, we now turn to the IU/clause dichotomy with respect to the
constraint on new information quantity formulated above. As discussed in
Section 2.6.2, a number of constraints on new information quantity have been
proposed by some of the key discourse researchers (e.g. Chafe 1987, 1994;
Du Bois 1987; Givón 1975, 1984). However, one of the issues left unclarified is
whether the proposed constraint may be applied to the IU or the clause, given
that strictly speaking, we cannot equate IUs with clauses 100% of the time, even
in a type of spoken discourse which exhibits highly preferential alignment of
IUs with clauses such as English, let alone in Japanese, which we have found less
clausal than English (see Section 4.2). Hence, what remains to be elucidated is
the exact relationship between the proposed constraints on new information
quantity and the IU/clause dichotomy.
 Chapter 5

We can see from Table 5.15 that of the 1,121 clauses, 6.5% (N=80)
contained more than one new NP. This means that about 94% of the clauses
contained no more than one new NP. I take this to mean that the “one new
NP per IU” constraint can also be interpreted as the “one new NP per clause”
constraint. That is, the constraint is applicable to the clause as well, although
less strongly (about 99% of the IUs contained no more than one new NP).
Bringing the above-presented results together, we can state, in conclusion,
that the “one new NP constraint” applies most strongly to the IU (only 0.7%
of the 1,600 IUs contained more than one new NP), less strongly to the
single-IU clause (only 1.3% of the single-IU clauses contained more than
one new NP), and only weakly to the multi-IU clause (32% of the multi-IU
clauses contained more than one new NP). However, given the relatively small
percentage of the multi-IU clauses (about 19% of all clauses) and the clause-
centered characteristics of the IU evidenced in Chapter 4, the constraint could,
on the whole, be taken to apply not only to the IU but also to the clause. That
is, the “one new NP constraint” limits the quantity of new nominal referents
that can be introduced within a discourse unit in conversational Japanese,
such that one IU/clause may contain no more than one new NP.
An additional relevant finding that should be presented here concerns
the grammatical role types of multiple new NPs contained in the 68 multi-
IU clauses. These new NPs were found to be argument + non-argument (e.g.
S role + oblique) or non-argument + non-argument (e.g. oblique + predicate
nominal), but not argument + argument, combinations. That is, none of
the multi-IU clauses contained two new argument NPs. This indicates that
Du Bois’ (1987) One New Argument Constraint discussed in Section 5.4.2 can
apply not only to the IU but also to the clause in conversational Japanese.
That is, speakers avoid introducing more than one new argument NP within
one IU and within one clause as well. The speaker’s avoidance of more
than one new nominal referent per IU/clause, it appears, is related to the
cognitive cost involved in the process of introducing a new concept. That
is, activating a previously inactive concept, or converting an idea from the
inactive to the active state is supposedly most costly (cf. Section 2.3.2). It
is conceivable that the discourse production activity of introducing a new
referent engages the speaker’s whole verbalization capacities, and therefore
the simultaneous introduction of two new referents within one IU/clause is
excessively burdensome.
Information structure of the intonation unit in conversational Japanese 

. Summary

The results of the present IU information stucture analysis have provided the
following answers to the research questions that I posed at the beginning of
this chapter:
a. preferred NP types: (a) grammatical role: S-role NPs (25%), oblique NPs
(22%), and predicate nominals (21%); (b) information status: Given NPs
(51%) and New NPs (49%); (c) syntactic form: bare nouns (50%); and
(d) most typical NP types: bare nouns with S role and Given information
status, and bare nouns with non-argument roles (obliques or predicate
nominals) and New information status.
b. preferred IU information structure types: IUs containing only one New NP
(34%) and IUs containing only one overt Given NP (30%).
c. preferred clause types: clauses with one overt argument (53%), typically
non-transitive clauses with overt S (29%) and transitive clauses with overt
O only (23%); and clauses with no overt arguments (44%), typically non-
transitive clauses without overt S (40%).

The results have shown that Japanese conversational co-participants con-


form to the three quantity constraints, namely, the “one overt NP per IU”, “one
new NP per IU”, and “one overt argument per clause” constraints, while com-
municating propositional content and constructing conversational discourse
through successive production of substantive IUs. They preferentially express
one NP within one IU, introduce one new nominal referent within one IU, and
express one argument, typically S or O, within one clause. Finally, given the
finding of the higher proportional occurrence of multiple new NPs per multi-
IU clause, I proposed that the speaker’s “marked” use of multi-IU clauses can
be related to the “one new NP per IU constraint” (i.e. speakers avoid introduc-
ing more than one new NP per IU). That is, the one new NP per IU constraint,
I have argued, is one of the motivating factors that will lead Japanese speakers
to the production of multi-IU clauses, or the breakup of a clause into separate
IU elements, with one new nominal concept being placed within one IU.
Chapter 6

Functional structure of the intonation unit


in conversational Japanese

This chapter, responding to RQ3 and testing Hypothesis 3.1, explores the
functional structure of the intonation unit (IU) in conversational Japanese.
The following four questions are addressed (Matsumoto 1999a):
a. What is the preferred functional structure of the IU in conversational
Japanese?
b. What is the preferred number of functional components per IU in conver-
sational Japanese?
c. What linear order do the functional components follow within an IU in
conversational Japanese?
d. Is the functional structure of the IU related to the speaker’s production of
multi-IU clauses, or the division of a clause into phrasal IUs in conversa-
tional Japanese? If so, in what way?

. Data coding

There are three major problems with Iwasaki’s (1993) analysis discussed in
Section 2.5.2. The first one concerns the coding of conjunctions. He codes the
coordinating conjunction sorede ‘and’ as [ID] (ideational), whose function is
defined to convey propositional content. I would rather argue that the principal
function of conjunctives such as ‘and’ is to link propositions although they link
IUs by virtue of a particular relationship, which differs from one conjunction
to another. That is, I would code sorede ‘and’ as Iwasaki’s [CO] (cohesive),
which is defined to do textual referential work, although I do not reject the idea
that it simultaneously has specific semantic content (Note here that specifying
relations between propositions constitutes one type of Halliday’s 1973, 1989
“ideational” function.) Another related issue that should be addressed is: if
conjunctions like kedo ‘though’ are coded as [CO], why are conjunctions like
sorede ‘and’ not coded similarly as [CO]? In my view, both types of conjunctives
 Chapter 6

function to link propositions and contribute to maintaining textual cohesion


in discourse.
The second problem concerns the distinction made between the ideational
[ID] and cohesive [CO] components. As Iwasaki (1993: 51) himself notes,
the ideational component may contain some words which have inherent
cohesive function and which do not occupy the linear position specified in
(2.15), i.e. after [ID] and/or before [IT] (interactional) (see Section 2.5.2).
(Demonstrative words like kore ‘this’ and sono ‘that’ are claimed to be such
examples.) That is, [ID] may serve cohesive function at the same time. In
addition, most of the lexical items coded as [CO] in his study seem specific
to the Japanese language. In a more generalized framework, therefore, [ID]
and [CO] will need to be collapsed. Further, both [ID] and [CO] concern the
expression of propositional content, unlike [LD] (lead) and [IT]. These suggest
that it will be more advantageous to fuse the cohesive component with the
ideational component.
The third problem is concerned with the linear position the lead compo-
nent is claimed to take in an IU. It seems that filled pauses like ano ‘uh’, which
are mainly the indicators of in-progress cognitive planning on the part of the
speaker (Chafe 1979: 162), occur not just at the beginning of the IU as Iwasaki
argues, but in other parts of the IU as well. Given such potential drawbacks of
Iwasaki’s (1993) analysis and room for revision, this chapter sets out to reex-
amine the functional composition of the IU, using a revised coding scheme.
In the present study of the functional structure of the Japanese substantive
IU, I used four codes. These are given in (6.1). The use of these codes is based on
Halliday (1985) and Iwasaki (1993) as well as Chafe’s (1993, 1994) functional
analysis of regulatory IUs. The four functional components are defined below.
(6.1) [ID] = ideational
[TX] = textual
[CG] = cognitive
[IT] = interactional
The ideational [ID] component conveys a full proposition or part of a propo-
sition, participating in the content of the conversation. The ideational compo-
nent may be comprised of a clause or a phrase or a word, each of which con-
tributes to building up a proposition. This study, unlike Halliday (1973, 1989),
does not consider specifying relations between propositions as the ideational
function. Thus, unlike Iwasaki (1993), I did not code conjunctions such as
sorede ‘and’ as [ID]. The 1,600 IUs are all substantive IUs that express ideas
Functional structure of the intonation unit in conversational Japanese 

of events, states, or referents (see Section 3.5); this means that all of the IUs
analyzed in this study contain [ID].
The textual [TX] component also participates in the content, but does
textual referential work by connecting one IU to another. That is, the textual
component has to do with the creation of coherence in discourse while
regulating the linkage between successively produced IUs. Those elements
coded as [TX] in this study include such textual connectives as de/sorede ‘and’,
demo ‘but’, kedo ‘though’,1 and kara ‘as’.
The cognitive [CG] component has to do with the regulation of the flow
of conversation. It indicates that the speaker is involved in some cognitive pro-
cessing, or is trying to keep the floor of the conversation, with the implication
to the hearer that more IUs are coming. (I use the term “cognitive” following
Chafe’s 1994 “cognitive” subtype of regulatory IUs; see Section 2.4.4.) Typi-
cally, in Japanese, cognitive connectives, or pause fillers such as ano, nanka,
and maa serve this cognitive function. Both the textual [TX] and cognitive
[CG] components are concerned with the linkage of ideas expressed in the
ideational [ID] components, functioning as IU-boundary signals. However,
the main difference between the two is that [TX] participates in the content,
whereas [CG] does not.
The interactional [IT] component has to do with interaction between the
conversational co-participants, or the speaker-hearer interpersonal relation-
ship involved in the co-production of conversational discourse. This compo-
nent functions to show the speaker’s sensitivity towards the addressee, or con-
cern for the hearer’s understanding. It functions to solicit the hearer’s involve-
ment into the on-going speech event. The interactional function is served by
such linguistic elements as interactional particles (e.g. ne, sa, yo) and tag-like
expressions (e.g. janai, desho).
In sum, four major changes have been made of Iwasaki’s (1993) coding
system: (a) elimination of [CO] (cohesive), (b) elimination of [LD] (lead),
(c) creation of [TX] partly as a replacement for [CO], and (d) creation of [CG]
as a replacement for [LD]. Thus, of the elements coded as [CO] in Iwasaki’s
(1993) study, IU-final overt conjunctives such as kedo ‘though’ were coded as
[TX], while -te forms and nominalizers like no and wake comprised part of
[ID] in this study. IU-initial conjunctives such as sorede ‘and’, which constitute
part of [ID] in Iwasaki (1993), were assigned the code [TX] in this study. Pause
fillers like ano ‘uh’, which constitute the IU-initial [LD] component in Iwasaki
(1993), were coded in the present study as [CG], which presumably occur not
only IU-initially but also elsewhere. One of the major advantages of this study
is that [CO], the component supposedly specific to the Japanese language, has
 Chapter 6

been eliminated, so that the coding scheme can be applied to languages other
than Japanese, and therefore commensurable cross-linguistic comparisons are
made possible.
In this study, when two connectives or particles with the same function
occurred adjacently in the IU-initial or IU-final position (although such cases
are rare), they were taken as belonging to one functional component and
assigned only one functional code, namely, [TX], [CG], or [IT] (see (6.4)
below). In addition, this study, as in Tokieda-school linguistics, allows the
idea of multifunctionality of certain linguistic elements—that one linguistic
element may have more than one function (e.g. cognitive and interactional)
(see Section 2.5.2). That is, certain linguistic elements should be analyzed as
simultaneously having more than one function, having a certain degree of
membership in one functional category. For example, it can be considered
that conjunctions like sorede ‘and’ have primarily a textual function, but they
equally serve some degree of ideational function (cf. Iwasaki’s 1993 treatment).
This study assumes that a given linguistic element may have multiple functions
but must have only one primary function, and the assigned code indicates such
primary function. Thus, all conjunctions were coded as [TX], not [ID].
The approach that the present chapter takes is advantageous and important
in that it allows us to direct attention to units that would otherwise be
disregarded if we look at only clauses as units of communicating propositions
and only nominals as units of expressing given/new informational distinction.
The basic assumption underlying this approach is that the overall IU structures
will be elucidated more precisely if we examine IUs not merely from the
perspective of syntactic and informational composition but also from the
perspective of functions of syntactic components that comprise IUs. That
is, it is the belief of the present study that the breakdown of Japanese IUs
into functional components will lead us to a more complete understanding
of the structural characteristics of IUs that constitute Japanese conversational
discourse. Specifically, the goal of this chapter lies in providing an additional
support for the claimed “single” nature of IUs as basic linguistic units of
spontaneous spoken communication, as argued for in Chapters 4 and 5.

. Distribution of IU functional structure types

The coding yielded a total of 17 IU functional structure types. The types and
their frequency are listed in Table 6.1 (see also Figure 6.1).2 The functional
types consist of one one-component type, 5 two-component types, 7 three-
Functional structure of the intonation unit in conversational Japanese 

Table 6.1. Distribution of IU functional structure types

IU functional structure type N %

ID 731 45.7
ID IT 434 27.1
ID TX 126 7.9
CG ID 66 4.1
TX ID 58 3.6
ID CG 9 0.6
ID TX IT 47 2.9
TX ID IT 45 2.8
CG ID IT 40 2.5
TX ID TX 13 0.8
CG ID TX 12 0.7
TX ID CG 4 0.3
ID CG IT 1 0.1
TX ID TX IT 7 0.4
CG ID TX IT 5 0.3
TX ID CG IT 1 0.1
CG ID CG IT 1 0.1
Total 1,600 100

component types, and 4 four-component types. The 17 types are listed in de-
scending order of frequency within each category. The order of the components
in the two-, three-, and four-component types corresponds to the temporal
order in which these components were actually uttered within each IU. For
example, the two-component type ID-TX indicates that the ideational compo-
nent is temporally followed by the textual component. Of the 17 IU functional
types, ID represents the minimal use of the functional components, being the
simplest type. The IUs consisting of two components such as ID-IT and those
with three components such as ID-TX-IT are functionally more complex. The
types such as TX-ID-TX-IT represent the maximal use of the four components,
being the most elaborate of all the functional structure types.
Table 6.1 shows that the 1,600 substantive IUs contained a total of 2,638
functional components, which consist of 1,600 ideational components, 581 in-
teractional components, 318 textual components, and 139 cognitive compo-
nents. This means the following. First, of all the components contained in the
IUs, the ideational components account for 61% (the interactional, textual, and
 Chapter 6

800

700

600

500
Frequency

400

300

200

100

0
TX ID CG
ID CG
ID

CG ID TX IT
TX ID CG IT
TX ID
CG ID

CG ID IT

ID CG IT
ID IT

TX ID TX
CG ID TX
ID TX

ID TX IT
TX ID IT

TX ID TX IT

CG ID CG IT
IU functional structure type

Figure 6.1. Distribution of IU functional structure types

cognitive components occupy 22%, 12%, and 5%, respectively). Second, of all
the IUs analyzed, 100% contained the ideational component (given that all of
them are idea-conveying substantive IUs), and 36% contained the interactional
component (20% and 9% contained the textual and cognitive components,
respectively).

. Preferred functional structure of the Japanese intonation unit

Table 6.1 indicates that the most frequently produced IU functional structure
type is ID (46%), and the second most frequently produced functional type is
ID-IT (27%). This means that the majority of the substantive IUs consist of
the ideational component only or the ideational component followed by the
interactional component.
Functional structure of the intonation unit in conversational Japanese 

In (6.2)–(6.5), observed substantive IU examples and the codes assigned to


them are given. The IUs (6.2) and (6.3) exemplify the most preferred, and (6.4)
and (6.5) exemplify the second most preferred, arrangement of the functional
components in an IU.
(6.2) [ ID ]
Y: MOchiron totta.
of course take-past
‘Of course, (I) took (pictures).’ (F5)
(6.3) [ ID ]
K: tomodachi to:,
friends com
‘with (my) friends’ (F8)
(6.4) [ ID ] [ IT ]
M: tooshi da yo ne.
investment cop fp fp
‘(It is) an investment, isn’t it?’ (M1)
(6.5) [ ID ] [IT]
S: haha ga sa:,
mother nom fp
‘(my) mother’ (F1)

The IUs (6.2) and (6.3) are examples of the simplest functional structure
type, composed solely of the obligatory ideational component. The clausal IU
(6.2) expresses a full proposition, with the subject ‘I’ and the direct object
‘pictures’ being assumed. The IU (6.3) ‘with (my) friends’ is an adverbial
phrasal IU, which, as part of a multi-IU clause, contributes to building up
a proposition. Examples (6.4) and (6.5) are two-component IUs comprised
of the ideational and interactional components. The IU (6.4) consists of the
subjectless nominal predicate clause ‘(it is) an investment’ which constitutes
the ideational component plus the particles yo and ne which collectively
constitute the interactional component. (The particle ne was most frequently
used in this study to serve the interpersonal function.) The NP IU (6.5), an
argument of a multi-IU clause, is composed of the ideational component (the
NP ‘(my) mother’ with the nominative case-marking particle ga) followed by
the interactional component (the interactional particle sa).
The results also reveal the following about the discoursal patterns of the
four functional components. First, of the 1,600 ideational [ID] components,
46% appeared singly, forming the functional type ID, and 32% were im-
mediately followed by the interactional component, forming the type ID-IT,
 Chapter 6

TX-ID-IT, or CG-ID-IT. Of all the ideational components which appeared


non-singly (N=869), 50% co-occurred with the interactional component only,
forming the type ID or ID-IT. This demonstrates the speakers’ marked ten-
dency to use the ideational component to produce the IU type ID or ID-IT.
Second, of the 581 interactional [IT] components, 89% appeared imme-
diately after the ideational component. This means that overwhelmingly, the
interactional component was used to produce the type ID-IT, TX-ID-IT, or
CG-ID-IT.
Third, the textual [TX] component appeared after the ideational compo-
nent 63% (vs. before the ideational component 37%) of the time. This indi-
cates that the majority of the textual connectives used in the data are IU-final
“subordinating” conjunctions. (In Japanese, unlike in English, “subordinating”
conjunctions such as kedo ‘though’ and kara ‘because’ occur in the clause-
final position.) More specifically, when the textual component was co-used
with the ideational component in the two-component IUs, the textual compo-
nent appeared after the ideational component 69% (vs. before the ideational
component 31%) of the time. That is, the type ID-TX (N=126) was produced
more frequently than the type TX-ID (N=58). An example of the preferred
functional structure type ID-TX is given in (6.6).
(6.6) [ ID ] [ TX ]
I: motteru kedo::,
have though
‘though (I) have (one)’ (M7)

The IU (6.6) is composed of [ID], the subjectless transitive clause ‘(I) have
(one)’ plus [TX], the conjunctive kedo ‘though’ (the textual connective most
frequently used in the present data). The IU expresses a full proposition in
which the subject ‘I’ and the direct object ‘one’ are assumed.
Finally, the cognitive [CG] component appeared before the ideational
component 89% (vs. after the ideational component 11%) of the time. This
finding can be explained by the role of this component in discourse produc-
tion. Given that the cognitive component has to do with the speaker’s cog-
nitive processing, and most typically, cognitive planning of ideational con-
tent to be expressed in about-to-be-uttered IUs, it naturally follows that
this component occurs overwhelmingly before the ideational component.
Of the two-component IU types involving the cognitive component, the
type CG-ID (N=66), as exemplified by (6.7), was preferred over the type
ID-CG (N=9).
Functional structure of the intonation unit in conversational Japanese 

(6.7) [CG] [ ID ]
H: ano daigaku jidai yo nen kan de,
pf university days four years during loc
‘uh in (my) four-year university days’ (M4)

The IU (6.7) begins with [CG], the pause filler ano ‘uh’, which is followed
by [ID], the adverbial phrase ‘in (my) four-year university days’. The two
components together constitute a substantive IU element of a proposition-
conveying multi-IU clause.

.. Preferred number of functional components per IU

Table 6.2 lists the frequency of the number of the functional components used
per IU.
Table 6.2 shows that 89% (N=1,424) of the 1,600 substantive IUs consist of
one functional component (i.e. the ideational component) or two functional
components (i.e. the ideational component plus one of the optional compo-
nents). Those IUs composed fully of the four components were rarely pro-
duced, occupying only 1% of the total number of the IUs. This suggests that
Japanese IUs tend to be unifunctional, or at least minimally multifunctional.
In other words, Japanese speakers prefer to incorporate no more than two
functional components—preferentially only one—within an IU. This can be
termed the “no more than two functional components per IU” constraint.
While the one-component functional type ID is predominant in the data,
of the multi-component types, some were more preferentially produced. As
shown in Table 6.1, the most preferred two-component type is ID-IT, the most
preferred three-component type is ID-TX-IT, and the most frequently used
four-component type is TX-ID-TX-IT. The IU (6.8) given below is an example
of the most favored three-component functional type, ID-TX-IT. The IU
consists of the ideational component (the subjectless adjectival predicate clause
‘(it is) difficult’), the textual component (the conjunctive kedo ‘though’), and

Table 6.2. Frequency of the number of functional components per IU

Number of components per IU N %


One 731 46
Two 693 43
Three 162 10
Four 14 1
Total 1,600 100
 Chapter 6

the interactional component (the IU-final particle ne). (For examples of ID-IT
and TX-ID-TX-IT, see (6.4)–(6.5) and (6.10), respectively.)
(6.8) [ ID ] [ TX ] [IT]
T: muzukashii kedo ne::.
difficult though fp
‘though (it is) difficult’ (M2)

.. Linear order of functional components within an IU

The results show that the three optional functional components occurred in
specific linear positions in an IU relative to the obligatory ideational com-
ponent. The textual and cognitive components appeared either immediately
before or immediately after the ideational component. Given that the tex-
tual component concerns linkages as well as specific relationships between
ideas or propositions expressed in the ideational component, and the cogni-
tive component concerns the planning of ideas or propositions expressed in
the ideational component, it follows that both of them need to occur adja-
cent to the ideational component. The interactional component, on the other
hand, occurred only in the IU-final position. This means that the interactional
component does not always appear adjacent to the ideational component, un-
like the textual and cognitive components. This also suggests that interactional
particles such as ne can function as non-prosodic markers of IU boundaries
(cf. Section 2.4.2).
The linear order of the four functional components within an IU can thus
be represented as follows:
(6.9) [TX/CG] [ID] [TX/CG] [IT]

What (6.9) indicates is that when an IU contains all of the four functional
components, the textual/cognitive component is followed by the ideational
component, which is followed by the textual/cognitive component, which is
in turn followed by the interactional component. To the extent that (6.9)
allows the textual and cognitive components to occur in the same linear
positions, this means that no fixed linear order can be observed among the four
functional components, in contrast to Iwasaki’s (1993) treatment (see (2.15)
in Section 2.5.2). That is, the pre-ideational and post-ideational positions are
not unifunctional; they may be occupied by either the textual or cognitive
component. The fact that only these components are allowed to be substituted
for one another could be attributed to the following: A given connective can
Functional structure of the intonation unit in conversational Japanese 

have either the textual or cognitive function in a given context. For example,
and can function as a textual connective; it can equally function as a cognitive
connective as well (cf. Section 2.4.4). The fact that the two components
take the same linear positions can be explained by the multifunctionality of
connective words which would appear in these positions.
To make another comparison, unlike in Iwasaki (1993), the cognitive
component, which corresponds to his IU-initial “lead”, occurred not only IU-
initially but also in other positions including IU-finally (although in much
lower proportions) in this study. In addition, given that the language-specific
cohesive component has been eliminated, the proposed linear order in (6.9)
could be applied cross-linguistically. For example, in spoken English, the
conjunction and, as a textual or cognitive connective, can occur in both pre-
ideational (IU-initial) and post-ideational (IU-final) positions (Chafe 1988).
On the other hand, the results of the present study are compatible with
Iwasaki’s (1993) in that the interactional component appeared only IU-finally.
The configuration (6.9), nevertheless, demonstrates that the functional
structure of the Japanese substantive IU is very orderly, such that the most
important and obligatory ideational component is preceded and followed by
either the textual or cognitive component, and the interactional component is
placed IU-finally. (Note that (6.9) represents a maximal functional structure,
given that [TX], [CG], and [IT] are optional components.) Given in (6.10)
is an example of substantive IUs that contain all of the four functional
components in the linear order specified in (6.9).
(6.10) [ TX ] [ ID ] [ TX ] [IT]
Y: soreni natsu da kara sa,
and summer cop because fp
‘And because (it is) summer’ (F5)

The IU (6.10) exemplifies the most complex functional structure type, com-
prised fully of the four components. The components are placed in the fol-
lowing order: textual (the conjunction soreni ‘and’), ideational (the nominal
predicate clause ‘(it is) summer’), textual (the conjunction kara ‘because’), and
interactional (the IU-final particle sa). Example (6.10) shows that the most
important ideational information is surrounded by two types of information,
textual and interactional. Note that although TX-ID-TX-IT is the type most
frequently produced in the four-component category, such four-component
IUs rarely occur in actual conversational discourse.
To summarize, the main findings of the present IU functional structure
analysis are the following:
 Chapter 6

a. The most frequently produced IU functional types are ID (46%) and


ID-IT (27%).
b. About 40% of the IUs terminate with the [IT] component.
c. The [TX] component tends to take the post-[ID] position, whereas the
[CG] component tends to take the pre-[ID] position.
d. The overwhelming maority (about 90%) of the IUs belong to one- or two-
component functional types, whereas three- or four-component types are
extremely limited; this can be termed the “no more than two functional
components per IU” constraint.
e. The linear arrangement of the four functional components in an IU can be
formulated as [TX/CG] [ID] [TX/CG] [IT].

The results given above suggest that the substantive IU in conversational


Japanese tends to be unifunctional or minimally multifunctional. That is, IUs
that Japanese conversational co-participants produce are preferentially com-
posed of one or two components—specifically, the ideational component only
or the ideational component followed by the interactional component. The
finding that maximally multifunctional IUs are rarely produced in Japanese
conversation suggests that speakers opt not to perform multiple functions
in an IU, even though they can choose to do so whenever possible; rather,
they concentrate on a single task of expressing ideational information (or
ideational plus interactional information) within one IU. In sum, overwhelm-
ingly, Japanese speakers produce IUs such that they perform “one function
at a time”.
This is reminiscent of previously formulated constraints on the IU infor-
mational composition such as Chafe’s (1987, 1994) “one new concept at a
time”/“one new idea” constraint and Du Bois’ (1987) “one new argument” con-
straint, both of which allow only one newly introduced concept within an IU.
This is also reminiscent of Chafe’s (1994) clause centrality proposal which says
that the IU tends to consist of a single clause that expresses a proposition which
codes some state or event. Similarly, in Chapters 4 and 5, which explored the
syntactic and informational composition of the IU in Japanese, we found that
the majority of substantive IUs take the form of a clause which introduces one
piece of new nominal information and which contains one overtly expressed
argument NP. Given the results of the present chapter, we can add to these
findings that the substantive IU in conversational Japanese tends to consist
solely of the ideational [ID] functional component (which is sometimes fol-
lowed by the interactional [IT] component).
Functional structure of the intonation unit in conversational Japanese 

The unifunctionality of the Japanese substantive IU is exemplified by


(6.11). This 10-IU sequence, in which female speakers Y and M are conversing
casually, consists of the two most preferred functional types, namely, ID
and ID-IT.
(6.11) a. Y: SU:goi yokeina mono mo motte icchatte.
very unnecessary thing even have-and go-past-and
[ ID ]
‘(I mistakenly) took even unnecessary things (with me).’
b. M: anmari motte ikanai koto ni suru.
not many have-and go-neg thing on decide
[ ID ]
‘(I will) decide not to take many (with me this time)’
c. +fuku toka.
clothes and
[ ID ]
‘clothes and so on’
d. onaji mono kireba ii ya to omotte. @@@#
same thing wear-con OK fp qt think-and
[ ID ]
‘(I) think that (it’ll be) OK if (I) wear the same clothes.’
e. Y: kono mae ne,
this before fp
[ ID ] [IT]
‘the other day’
f. ->san diego animaru paaku ni itte kita n da
San Diego Animal Park all go-and come-past nml cop
[ ID ]
- ->yo.
fp
[IT]
‘(I) went to the San Diego Animal Park’
g. ++hajimete.=
first time
[ ID ]
‘for the first time’
h. M: =animaru paaku∼
animal park
[ ID ]
‘(Is it) the Animal Park?’
 Chapter 6

i. Y: waarudo animaru paaku tte shitteru?


world animal park top know
[ ID ]
‘(Do you) know the World Animal Park?’
j. +zuu ja nakutte:,
zoo cop neg-and
[ ID ]
‘(It is) not a zoo, and’ (F5)

Of the 10 IUs, five IUs, (6.11a), (6.11d), and (6.11h)–(6.11j), are indepen-
dent clausal IUs each of which expresses a complete proposition. All of these
IUs constitute the simplest functional structure type, ID, the sole function of
which is to convey ideational content. The IUs (6.11b, c) collectively constitute
a multi-IU clause, in which the post-predicate NP IU (6.11c) serves as the di-
rect object of the verb ‘take’ uttered in the clausal IU (6.11b). Both of these are
unifunctional, contributing solely to the completion of a proposition. The IUs
(6.11e)-(6.11g) also constitute a multi-IU clause, in which (6.11e) and (6.11g)
are adverbial phrasal IUs. Of these, (6.11e, f) are two-component IUs com-
prised of the ideational component followed by the interactional component.
By uttering these IUs, speaker Y performs two functions—communicating
propositional content and soliciting hearer M’s involvement into her own
speakership. Overall, (6.11) strongly evidences the unifunctional nature of the
substantive IU in conversational Japanese.
The results of the present study are mostly compatible with those of
Iwasaki (1993) discussed in Section 2.5.2. First, concerning the percentage
of the functional structure type ID, his analysis resulted in 34%, compared
to 46% in this study. This is undoubtedly attributable to the fact that while
my data consist of substantive IUs only, his IUs include both substantive and
non-substantive IUs. The difference can also be explained by the fact that part
of his ID-CO is interpretable as my ID, given that part of his cohesive [CO]
components consist of such non-conjunctives as nominalizers and -te forms,
which were coded as part of the ideational [ID] component in my analysis.
As to the frequency of the functional type ID-IT, his analysis revealed 17%,
compared to 27% in my analysis. This is similarly attributable to the fact
that part of his ID-CO-IT is interpretable as my ID-IT for the same reason
given above.
Second, as to the percentage of IUs that include the interactional [IT]
component, his study showed 32%, which is quite comparable to 36% in my
study. The difference can partly be explained by the fact that IU-final no, which
Functional structure of the intonation unit in conversational Japanese 

was coded as a nominalizer, thus [CO] in Iwasaki (1993), was coded as an


interactional particle, thus [IT] in my coding (cf. Maynard 1989).
Third, regarding the preferred number of functional components per IU,
the distributional percentages have been found to be very similar across the two
studies. That is, 85% of his IUs vs. 89% of my IUs contain one component or
two components (42% of his IUs vs. 46% of my IUs contain one component;
in both studies, 43% of the IUs contain two components). Those with three
components are also comparable between the two studies (13% of his IUs vs.
10% of my IUs). Moreover, the percentages of IUs with four components are
both extremely low (0.3% of his IUs vs. 0.9% of my IUs). Although the two
studies used different codes, the results of my study clearly support Iwasaki’s
(1993) “up to two functions per unit” constraint, which allows speakers of
Japanese to incorporate up to two functional components within one IU in
conversational interaction.
The results of this study suggest, in accordance with Iwasaki (1993), that
in conversational speech the most fundamental task of substantive IUs—
communicating ideational information—is largely accomplished through the
production of IUs with the ideational component only. That is, speakers most
typically accomplish the ideation-communicating task by means of the sim-
plest functional structure type, namely, ID, without the other optional compo-
nents serving as idea-linking devices. In Iwasaki’s (1993) study, if we collapse
the ideational [ID] component and the language-specific cohesive [CO] com-
ponent, we can see that 56% of his IUs consist of the ideational component
alone. This apparently strengthens my argument for the unifunctionality of
Japanese IUs. In short, Japanese speakers have strong tendencies to perform
“one function at a time”, just as they have strong tendencies to produce “one
clause at a time”, “one new NP at a time”, and “one overt argument at a time”,
as has been shown in Chapters 4 and 5.

. Multi-IU clauses and multifunctionality

I now go back to the issue of the relation between the multifunctionality of


spoken Japanese and the speaker’s use of phrasal IUs taken up in Section 2.5.2
in terms of Iwasaki’s (1993) argument. Based on the findings of Chapter 4,
I claim that the “default” case is the production of a clausal IU which inde-
pendently communicates a full proposition. The “marked” case of phrasal-
ity results when the speaker is strongly motivated to code non-ideational,
interactional information—which is generally considered important and highly
 Chapter 6

valued in Japanese communication—simultaneously with ideational informa-


tion. That is, when the speaker feels the need, presumably the stronger need
than usual, to convey interactional information, an IU, which would normally
take the form of a clause, will be divided up into shorter phrasal units. This
is because by producing multiple phrasal IUs, the speaker can encode interac-
tional information more frequently than he/she would be able to do so by pro-
ducing only one clausal IU with the same ideational content. In other words,
the best way for the speaker to code interactional information frequently is to
divide a clausal IU into pieces, such that each piece contains an interactional
component IU-finally. It could also be claimed, in line of this argument, that
the stronger the need on the part of speakers to convey such non-ideational,
interactional information, the more frequently multi-IU clauses will be pro-
duced, that is, the more frequently the breakup of a clause into phrasal IUs
will occur.
By way of illustration, let us reexamine (2.14) in Section 2.5.1, which is
reproduced below in (6.12).
(6.12) a. A: Yamato san ga ne:,
Yamato hon nom fp
‘Mr. Yamato’
b. kekkyoku ne:,
in short fp
‘in short’
c. wareware o ne:,
us acc fp
‘us’
d. sofuto no ne:,
software gen fp
‘on software’
e. shigoto ni taisuru.
job toward
‘of (our) job’
f. hyooka ga ano hito ne:,
evaluation nom that person fp
‘evaluation, he’
g. shitenai.
do-prog-neg
‘doesn’t do’
Functional structure of the intonation unit in conversational Japanese 

h. hyooka o.
evaluation acc
‘(high) evaluation’
i. zettai.
never
‘never’

The sequence in (6.12), we should note, is an excerpt from a conversation


occurring in a bar between two Japanese colleagues complaining about their
boss’ low evaluation of their work. Presumably, in that kind of situation, the
speaker would normally try to solicit more interactional feedback from the
listener than in usual conversational situations, and consequently, is more likely
to code interactional information through the use of interactional particles.
This is exactly how speaker A has resulted in “fragmenting” a clause into
phrasal IUs. By producing short phrasal IUs and terminating most of them
with the interactional particle ne, A has communicated with his interlocutor
“more interactionally” than when producing a single clausal IU which would
at best contain one interactional component. (As we can see from the examples
given above, the interactional component in spoken Japanese typically consists
of the interactional particle ne.) That is, we have to keep in mind that the IUs in
(6.12) have been uttered in a relatively “unusual”, or “special” situation. Thus
the multi-IU clause such as (6.12) could not be regarded as typical of IUs that
Japanese speakers normally produce in dialogic interactive conversation.
It is not always the case that Japanese speakers produce phrasal IUs, but
they do so only in “special” situations, one of which is when the speaker is
strongly motivated to communicate interactional information while simulta-
neously conveying ideational information. By dividing a clause into distinct
phrasal IUs, the speaker can code speaker-hearer interactional information
more frequently than when he/she produces a single clausal IU. While the
speaker’s need to code interactional information may be a strong motivational
force which will make the IU phrasal and multifunctional, the breakup of a
clause into phrasal IUs, or the production of multi-IU clauses seems to be gov-
erned by other factors as well. As I claimed in Section 5.6, another discoursal
situation in which the speaker uses phrasal IUs involves the introduction of
more than one new NP. Through the operation of the “one new NP per IU
constraint”, the speaker is led to divide a clause into separate IUs such that each
phrasal IU contains one piece of newly introduced nominal information.
As noted in Section 2.5.1, the syntactic organization of Japanese allows
Japanese speakers to produce short phrasal IUs more readily than English
 Chapter 6

speakers. However, the results of the present study show that this phrasal IU
option is exercised not all the time, but only in “special” occasions. These
include when the speaker introduces more than one new nominal referent,
and when the speaker is strongly motivated to communicate interactional
information to the interlocutor. Besides the informational and interactional
factors, other factors such as syntactic ones that will govern the production of
multi-IU clauses in spoken communication need to explored in future research
(see Croft 1995).3

. Summary

The present chapter has concerned itself with the analysis of the functional
composition of IUs as basic prosody-based units of Japanese conversation. It
has been shown in this chapter that overwhelmingly, the Japanese substantive
IU consists of the ideational component only, although it is sometimes fol-
lowed by the interactional component. That is, the substantive IU in conver-
sational Japanese has been found to be predominantly unifunctional, or min-
imally multifunctional. This means that typically, IUs are produced such that
they serve “one function at a time”. In other words, speakers opt not to perform
multiple functions but rather concentrate on one function—communication
of ideational content—within one substantive IU. The results suggest that the
preferred way Japanese conversational co-participants communicate ideas or
propositions is by means of the simplest functional structure type which con-
sists solely of the ideational component, without the other components serv-
ing a coherence-creating, regulatory, or interpersonal function, or as an idea-
linking device. Importantly, it has also been argued that one additional factor
responsible for the marked production of multi-IU clauses in spoken Japanese
is the interactional motivation on the part of the speaker. The breakup of a
clause into phrasal IUs will undoubtedly increase the frequency with which
non-ideational, interactional information can be communicated in conversa-
tional discourse.
Chapter 7

Conclusion

. Japanese intonation units: Syntactic, informational,


and functional structures

We explored in the previous three chapters recurrent patterns in the way


Japanese speakers structure their discourse in the process of conversational
communication. Specifically, we investigated the syntactic, informational, and
functional structures of the substantive intonation unit (IU) in conversational
Japanese, focusing our primary interest on the exploration of the most pre-
ferred IU type, or the exponent of the Japanese substantive IU. We found, first
of all, in Chapter 4 that three syntactic types are characteristic of Japanese sub-
stantive IUs: (a) semi-clausal IUs that consist of verbal predicates without overt
subjects, (b) NP IUs formally comprised of nominal phrases (including NP IUs
functioning as semi-clauses), and (c) full clausal IUs that consist of overt sub-
jects and predicates. Given the notable prevalence of the clausal IUs in the data,
I concluded that the clause—a unit of discourse that expresses a proposition
which codes some state or event—is the syntactic exponent of the Japanese
substantive IU (Matsumoto 2000b). This is in concert with the “clause central-
ity proposal” advocated by Chafe (1987, 1994) and against the phrase-centered
quality of Japanese IUs argued for by Clancy (1982), Maynard (1989), and
Iwasaki and Tao (1993). The apparent discrepancy between the conclusions
drawn in the present study and those in the previous studies needs to be further
examined in the future in relation to the issue of the cross-linguistic validity of
the clause centrality proposal and with a larger body of conversational data.
In our discussion of NP IUs, which pervaded the data, we have focused
on the clause-external independent phrasal NP IUs. We have shown that
these NPs, though “detached” from the clausal structure and thus not di-
rectly concerned with the communication of propositions, cannot simply be
deemed fragments of discourse or performance errors; they are indeed an
important phenomenon, worthy of study. We found that Japanese conversa-
tion manifests five types of detached NPs—“stray”, “lead”, topic/left-dislocated,
postposed/right-dislocated, and repeated NPs—and that these five types each
 Chapter 7

have important discourse functions. I have argued that the occurrence of


clause-external detached NPs in informal conversation, although not directly
concerned with the core task of conversation, i.e. communication of propo-
sitional content, is important discourse functionally, and that it can be ex-
plained in terms of discourse-communicative, interactional, and information
flow-related motivations (Matsumoto 1997b, 1998a).
Another important finding discussed in Chapter 4 along with the robust-
ness of clausal IUs is that the overwhelming majority of the clauses in the data
consisted of one IU. That is, single-IU clauses are predominant in conversa-
tional Japanese. The coding further demonstrated that one clause contained an
average of 1.23 IUs. Hence, I explicitly proposed that conversational Japanese
cannot be regarded as highly fragmented, contrary to the arguments forcefully
put forth in previous research (e.g. Clancy 1982; Iwasaki 1993; Maynard 1989).
I argued that rather than employing syntactic fragmentation, Japanese speak-
ers basically conform to the “one IU, one clause” strategy, although apparently
to a lesser degree than English speakers. This proposal reflects my view that
the production of single-IU clauses is the “default”, whereas that of multi-IU
clauses is “marked” in conversational Japanese (Matsumoto 2000b).
Second, we found in Chapter 5 that the information structure of the
Japanese substantive IU tends to be non-elaborate. Focusing on nominal
references including both arguments and non-arguments, we found that the
IUs typically contained a single overt NP with New or Given information
status. This suggests the speakers’ preferential placement of no more than one
overt NP (i.e. zero NPs or one overt NP) within one IU, which I term the “one
overt NP per IU constraint”. In addition, it was found that none of the IUs
contained more than three overt NPs. Another related and noteworthy finding
is that the speakers avoided introducing more than one new NP per IU. That is,
Japanese speakers conform to the “one new NP per IU constraint” (Matsumoto
1998b, 2000a) and previously formulated constraints such as Chafe’s (1987,
1994) “one new concept at a time/one new idea constraint” as well. The study
further showed that the extremely rare occurrence of “two new NPs per IU”
is itself governed by the constraint that the two new NPs are argument +
non-argument combinations, thus conforming to Du Bois’ (1987) “One New
Argument Constraint”.
We also found in Chapter 5 that the speakers preferentially produced two
types of NPs: (a) bare nouns with S role and Given information status, and
(b) bare nouns with non-argument roles (obliques or predicate nominals)
and New information status. These represent the preferred NP types in con-
versational Japanese. Importantly, the coding revealed an approximately equal
Conclusion 

number of overtly expressed Given and New NPs. This suggests that conver-
sational Japanese, while making abundant use of zero-form Given NPs (55%
of the argument slots in the data were not filled with overt NPs), utilizes overt
NPs with both information statuses in comparable proportions in communi-
cating ideational content. It was additionally found that one new NP was intro-
duced every 1.6 clauses and every 2.3 IUs on average; this suggests the relatively
high-rate introduction of new nominal information by Japanese interactants in
conversational communication (Matsumoto 1997a).
Furthermore, it was found that the speakers preferentially produced non-
transitive, one-participant clauses with very low transitivity (typically intran-
sitive or nominal predicate clauses). It was also shown that the speakers
preferentially produced clauses with no overt arguments or one overt argu-
ment, and that the one-argument clauses typically contained S or O, not A.
That is, a single overt argument in the S or O role followed by a predicate
represents the maximal preferred clause structure in conversational Japanese;
this has been formulated as N(S/O) P. By contrast, clauses with overt A and
O arguments are strongly dispreferred, that is, two-argument clauses rarely
occur in actual interactive conversational discourse; this is consistent with
Lambrecht (1987) and Du Bois (1987). Given the low transitivity of conver-
sational Japanese and the “one overt argument per clause constraint”, I ar-
gued as a corollary that the traditional treatment of the highly transitive SOV
clause, which typically involves two overt lexical arguments, as canonical or
representative of Japanese clauses is not justified; but rather, more attention
needs to be directed to studies of grammar of one-participant clauses. Also
relevant to this is the finding that the overwhelming majority of A-role NPs
in Japanese conversation take zero forms. The results of the preferred nom-
inal structure analysis of the present study indicate that Du Bois’ (1987) PAS
constraints—the “One Lexical Argument”, “One New Argument”, “Non-lexical
A”, and “Given A” Constraints—hold in conversational Japanese (Matsumoto
2000a, 2002, 2003). Importantly, such recurrent patterns or tendencies in con-
versational interaction can be understood as having the power to shape the
foundations of grammar (Bybee & Hopper 2001; Du Bois 2003; Ford et al.
2003).
I further proposed in Chapter 5 that given the higher proportional oc-
currence of multiple new NPs per multi-IU clause, the marked production of
multi-IU clauses (i.e. the marked use of non-clausal, phrasal IUs) in Japanese
conversation can be related to the “one new NP per IU constraint”. That is,
my argument is that the one new NP per IU constraint is one of the motivat-
ing factors that will lead Japanese speakers to produce multi-IU clauses, or to
 Chapter 7

divide a clause into separate IU elements, placing one new nominal concept
within one IU (Matsumoto 2000b; cf. Section 7.2).
Third, we found in Chapter 6 that the Japanese substantive IU tends
to be functionally non-elaborate, which supports the findings of Iwasaki’s
(1993) analysis. Among the four functional components—ideational [ID],
textual [TX], cognitive [CG], and interactional [IT]—that could be incorpo-
rated into one substantive IU, the speakers preferentially used only one compo-
nent (the obligatory ideational component) or two components (the obligatory
ideational component + one of the three other optional components) per IU.
That is, functionally, the substantive IU in conversational Japanese consists of
one or two components; the preferred number of functional components per
IU is no more than two. More specifically, the substantive IU preferentially
consists of the ideational component only ([ID]) or the ideational compo-
nent followed by the interactional component ([ID] [IT]). Given the notable
prevalence of [ID] in the data, I have further suggested that speakers com-
municate ideational information most typically by using the simplest func-
tional structure type which consists solely of the ideational component, with
none of the other components serving an interactional, coherence-creating, or
regulatory function, or as an idea-linking device. The analysis also revealed
that the maximal structure of the substantive IU in conversational Japanese
is very orderly, such that the most important and obligatory ideational com-
ponent is preceded and followed by the textual/cognitive component, and
the interactional component is placed IU-finally. This has been formulated as
[TX/CG] [ID] [TX/CG] [IT] (Matsumoto 1999a).
The above-presented major findings of the present study can be further
summarized as follows:
a. Japanese speakers prefer to produce IUs which are syntactically semi-
clausal, especially propositionally complete independent semi-clausal IUs.
b. Japanese speakers prefer to produce IUs which consist of a single nominal
carrying new or given information.
c. Japanese speakers prefer to produce IUs which functionally consist of the
ideational component only or the ideational plus interactional components.

Thus, the following hypotheses this study attempted to test have all been
confirmed (see Section 3.1).
HYPOTHESIS 1.1: The syntactic structure of the IU in conversational Japanese
tends to be semi-clausal.
Conclusion 

HYPOTHESIS 2.1: The information structure of the IU in conversational


Japanese tends to consist of one piece of new or given
nominal information.
HYPOTHESIS 3.1: The functional structure of the IU in conversational Japa-
nese tends to consist of the ideational component only or
the ideational plus interactional components.

The analysis revealed that the structures of the substantive IU in conver-


sational Japanese are limited by five major constraints. The syntactic struc-
ture of the IU is limited by the “no more than two IUs per clause” constraint.
The information structure of the IU is limited by the “one overt NP per IU”,
“one new NP per IU”, and “one overt argument per clause” constraints. The
functional structure of the IU is limited by the “no more than two functional
components per IU” constraint. Overall, these constraints confine the produc-
tion of substantive IUs in such a way that a discourse unit (i.e. IU or clause)
preferentially contains only one structurally important element (e.g. new NP,
overt argument). This is how Japanese speakers, and possibly speakers in gen-
eral, interactively co-construct conversational discourse, with strict cognitive
constraints operating on their processing capacities, relying on the immedi-
ate context, having recourse to gestural as well as contextual resources, and
concentrating, above all, on the expression of ideational content.
In sum, the results presented in Chapters 4–6 demonstrate the speakers’
marked preference for IUs which are simple or non-elaborate in terms of the
syntactic, informational, and functional structures. That is, elaborate types
of IUs that contain two or more elements (e.g. overtly expressed NPs, new
nominal referents) are not preferred for the transmission of ideational content
in spoken communication. This means, in effect, that one IU typically contains
only one structurally important element, as represented most explicitly by the
“one new NP per IU” constraint and the “one overt argument per clause”
constraint. We can conclude that the exponent of the Japanese substantive IU
that the present study has uncovered in relation to the formulated constraints
is the following: a semi-clause composed of one IU, with one overt NP,
with one piece of new nominal information, with one overt S/O argument,
and with one functional component (note that this represents a maximal
preferred IU structure). What this suggests is that Japanese conversational co-
participants interactively organize discourse into a succession of structurally
simple discourse units (Matsumoto 1999c, to appear).
The observed relatively simple nature of the idea-conveying substantive IU
in conversational Japanese, or the speakers’ minimal use of structurally elabo-
 Chapter 7

rate IUs, as I have argued, can be seen as being motivated by the characteristics
of spoken language itself (Chafe 1982, 1988, 1994: Chapter 4; Clancy 1982;
Ochs 1979; Tannen 1982). The presence of a directly shared context and the
availability of prosodic and gestural resources for supplementary use negate the
necessity for elaborating the IU structures and linkages, on the one hand; and
the speakers’ focus on verbalizing ideas on the run—their most urgent task—
discourages such elaboration, on the other. Besides various extrinsic factors, we
also need to take into account intrinsic factors such as cognitive constraints on
humans’ information processing capacity. The “unitary” nature of substantive
IUs that has surfaced in this study suggests that speakers can handle only one—
proposition, new concept, overt argument, or functional component—within
one IU, a linguistic expression of the speaker’s focus of consciousness, a mini-
mal unit of thought organization (Chafe 2000). That is, “the magical number
one appears to be fundamental to the way the mind handles the flow of infor-
mation through consciousness and language” (Chafe 1994: 119). In conclusion,
the findings of the present study supports the “unitariness” or “singleness” of
the IU as the basic linguistic unit of discourse production and information flow
in spontaneous conversational interaction.

. Prospects

We have seen in Section 4.3 that about 8% of the substantive IUs contained
post-predicate phrases. This means that the overwhelming majority of the IUs
in the database consist of pre-predicate elements. We can therefore interpret
the results given above as reflecting the preferred structures of IUs which
involve non-postposed elements, or the canonical word order. Accordingly, an
exploration of postposing in relation to the preferred IU structures elucidated
in this study, an examination of how the structures of IUs involving postposing
deviate from the preferred IU patterns seems worth conducting in future
research. In addition, the way postposed nominals differ from the preferred
NP types in terms of grammatical role, information status, and syntactic form
is worth investigating.
We found in Section 4.3 that the post-predicate elements typically occured
as phrasal IU elements of multi-IU clauses. That is, the multi-IU clauses
exhibited a higher rate of postposing than the other IU syntactic types. It thus
seems evident that the production of postposed elements is related to that
of multi-IU clauses. In what way, then, is postposing related to the marked
production of multi-IU clauses? One answer would be that their relationship
Conclusion 

can be explained in part in terms of the “one new NP per IU constraint”


formulated above. It appears that just as the production of multi-IU clauses
is motivated by the one new NP per IU constraint, as I argued in Chapter 5,
the consistent placement of postposed new NPs in an independent IU, which
was found in Matsumoto (1995a, 2000c), can be related to the one new NP per
IU constraint as well. Specifically, postposed NPs with new information status
are placed in a separate IU instead of being appended to the final portion of an
IU, in order to prevent the IU from containing more than one new NP by the
operation of this constraint. On the other hand, the coherent pattern exhibited
by the post-predicate new NPs provides evidence that speakers do not segment
discourse randomly, but sort it into IUs in such a way that information flow is
facilitated. This presumably reflects the speaker’s interactionally determined
choice to make postposed new information, although backgrounded, more
salient to the hearer who is processing it. It furthermore suggests that the
IU has a function of highlighting new information; this seems to accord with
Halliday’s (1967b) claim that one of the functions of intonation is to mark off
which information the speaker is treating as new and which as given.
Some predictions can be made regarding the properties of postposed
elements. As I discussed in Section 4.3, if clause-internal postposing has a
backgrounding function, it is expected to involve typically given information,
or if new, that which is of trivial importance, i.e. referentially less important
than the other new concepts uttered in the immediate context (cf. Kuno 1978a;
Takami 1995). If this is the case, it will generally exhibit low TP values (Givón
1983a, 1990). Similarly, given that clause-external postposing has to do with
the repeating or restating of already uttered elements, it should naturally entail
given information (cf. Section 5.3). If postposed NPs tend to be pronouns, as
was found in Matsumoto (2000c), this will need to be linked to the finding of
the present study that pronouns accounted for only 10% of the overt nominals
produced by the speakers. Further, if the majority of postposed NPs are S roles,
as was shown in Matsumoto (2000c), this will accord with the finding of this
study that the majority of the clauses contained overt S or O arguments. It will
also be useful to compare postposed NPs and non-postposed NPs in terms
of the degrees of referential continuity (RD values), referential complexity
(PI and SR values), and thematic importance (TP and OF values) (Givón
1983a; Hinds 1983; cf. Section 2.1).
Besides postposing, other issues and areas future research should pay at-
tention to in relation to the results of the present study and as its extensions
include the following. First, given that this study dealt with substantive IUs
only, non-substantive, regulatory IUs used in everyday conversations need to
 Chapter 7

be explored in detail. Further, how these two types of IUs, substantive and
regulatory—which form different aspects of thought, i.e. its “content” and
“infrastructure” respectively (Chafe 1998, 2000)—interact with each other as
they are successively produced by conversational co-participants needs further
study. What is also worthy of examination is how substantive IUs as the small-
est idea-conveying discourse units will cluster into larger thematically coher-
ent units in the collaborative and interactive construction of conversational
discourse (cf. Chafe 1980b, 1987, 1994; Gee 1986; Hinds 1980).
Second, given that this is a study of two-party conversations between
same-sex friends where a fairly large amount of information is supposed to be
shared or presupposed, other types of interactions merit further investigation.
For example, it would be interesting to explore interactions such as multi-
party conversations involving three or more participants, dyadic male-female
conversations, and conversations among participants who have met for the
first time and thus share very little background information. Another related
and interesting area worth exploring in subsequent research concerns gender
differences, that is, how males and females significantly differ in the production
of various types of IUs (cf. Matsumoto 1996; Section 3.4). Moreover, it is
worthwhile to conduct comparative studies of the structures of IUs produced
in different types of spoken discourse—not only ordinary conversations but
also other types of spoken discourse occurring in professional, institutional
encounters, for example, doctor-patient interaction, news interviews, and
courtroom discourse (cf. Boden & Zimmerman 1991; Drew & Heritage 1992;
Sarangi & Roberts 1999; van Dijk 1997). A comparative analysis of ordinary
conversation and institutional talk in particular would help us understand
the relationships between the IU structures, the roles of interactants, and the
purpose of communication (cf. Heritage 1997; Matsumoto 1999b).
Third, I proposed in this study that several factors are responsible for
the marked production of multi-IU clauses in conversational Japanese. I have
claimed in Chapter 5 (Section 5.6) that the “one new NP per IU constraint” is
one of the important motivating forces that will lead speakers into the produc-
tion of phrasal IUs. Additionally, I have suggested in Chapter 6 (Section 6.4)
that another factor facilitating the production of phrasal IUs in Japanese con-
versation is the interactional motivation on the part of the speaker. By divid-
ing a clause into phrasal IUs and terminating them with interactional particles
such as ne, the speaker can perform interactional encoding more frequently.
Importantly, it has also been pointed out in Chapter 2 (Section 2.5.1) that the
“loose” internal structure of Japanese clauses underlies speakers’ use of phrasal
IUs in conversational Japanese. The relation of the speaker’s marked produc-
Conclusion 

tion of non-clausal IUs to other factors needs further study, both within and
across languages. Future research needs to be directed toward further explo-
ration of what speakers and/or their interlocutors can gain in conversational
interaction by breaking up a clause into phrasal IUs at the cost of efficiency
in communicating propositional content (see Croft 1995; Matsumoto 1998b,
1999a, 2000b, 2001).
Finally, it is expected that acoustically based analysis of intonation, which
is beyond the scope of the present research, will be effectively incorporated
into, and will interplay with, the discourse analysis of auditorily defined IUs.
Future studies are expected to have access to acoustic properties of sounds and
to supplement our perceptual observations with physical observations. This
approach will undoubtedly yield a greater depth of understanding and preci-
sion to our investigation of how speakers organize their messages into coherent
units as they jointly and interactively construct conversational discourse.
Notes

Chapter 1

. The precise measurement of intonation contours by means of acoustic phonetic equip-


ments is beyond the scope of the present study, which is concerned with the units that
human co-participants in conversations auditorily perceive as coherent chunks of spoken
discourse.

. Hinds (1983: 53), for example, claims that “of 567 clauses, 8 (1.4%) evidenced a scram-
bled word order [i.e. OSV]”. Another piece of evidence for the basicness of SOV order comes
from the so-called “case-marker drop” phenomenon. When two NPs, whose subject-object
relation is unclear from the context, appear preverbally without case-marking particles as in
(i), the first NP (John) is invariably interpreted as the subject, and the second NP (Mary), as
the object, but not vice versa. This suggests that the SOV order is the unmarked basic word
order in Japanese (Matsumoto 1995b).
(i) John Mary butta yo.
hit-past fp
‘John hit Mary.’
Note that Kuno’s (1973) examples are constructed examples. In naturally occurring casual
Japanese conversation, case markers such as ga (the subject marker) and o (the direct object
marker) are often not used (cf. Fujii & Ono 2000; Ono et al. 2000).

. In this paper I will use the terms “postposed” and “post-predicate” interchangeably; I
will also use the terms “non-postposed” and “pre-predicate” interchangeably. Thus the term
“postposing”, as I use it here, refers to the placement of elements in a post-predicate position;
it does not imply movement of constituents from the “canonical” preverbal position.

. Past discourse-functional studies typically dealt with functions of marked word order
constructions used in naturally occurring spoken discourse, providing explanations for
pragmatic reordering in various syntactically based languages (cf. Downing & Noonan 1995;
Mithun 1987; Payne 1992). The marked word order constructions that have been discussed
in the literature are of three types: preposing, postposing, and inversion (e.g. Birner & Ward
1998; Dorgeloh 1997; Duranti & Ochs 1979; Fox 1985; Green 1980; Silva-Corvalán 1983;
Ward 1988).
 Notes

. Matsumoto (1995a) has shown that of 1,526 IUs, 84 (5.5%) involved postposing. Note
that post-predicate elements in Japanese are found only in spontaneous casual spoken
discourse. They do not normally occur in written discourse or planned spoken discourse.

Chapter 2

. According to Givón (1988: 67), the scale accords with the iconicity principle for topic
continuity (i.e. “the more continuous/predictable is the topic/referent NP, the less overt
expression it needs to receive”), which, in turn, is an instance of the general iconicity
principle in communication (i.e. “the more predictable the information, the less coding it
receives”).

. Reid (1977) argues that the two past tense forms in French, passé simple and imparfait,
call for “high focus” and “low focus” of attention, respectively. These can be taken to be the
foregrounding and backgrounding forms of the verb, respectively. His data show that the
passé simple tends to be used with actions, affirmative verbs, human, singular, first-person,
proper-name subjects, main characters of discourse, and main clauses. These seem to accord
with Hopper and Thompson’s (1980) high-transitivity features.

. Transitivity is a global property of an entire clause such that an activity is carried over
or transferred from an agent to a patient. It has ten component parts: participants, kinesis,
aspect, punctuality, volitionality, affirmation, mode, agency, affectedness of O (object), and
individuation of O. These features collectively allow clauses to be coded as more or less
transitive—the more high transitive features a clause has, the more transitive it is, and the
closer it is to cardinal transitivity (Hopper & Thompson 1980: 251–253).

. Reinhart (1984) and Wallace (1982) argue that the foreground-background distinction in
discourse rests on the innate, universal perceptual “figure-ground” distinction. They suggest
that there is a striking correlation between the perceptual criteria determining the figure and
those determining the discourse foreground, that is, foreground in discourse is “perceptually
more salient” than background.

. Givenness of an idea may be established in a “textually evoked” way; that is, an idea is first
introduced into a discourse as new information and then remains given for a certain period
of time. It may also be established in a “situationally evoked” way; for example, because of
the active status of the ideas of the speaker and the hearer, references to first and second
persons (i.e. I and you) are typically given (Chafe 1994: 78–79; Prince 1981: 236).

. Chafe (1987: 29–30) discusses two ways in which concepts become semi-active: (a)
through deactivation from an earlier active state, and (b) through evocation of a “schema”,
which is generally regarded as a cluster of interrelated expectations (Schank & Abelson 1977;
Tannen 1979). When a schema is evoked in a discourse, some of the expectations or concepts
associated with it are assumed to change into the semi-active state. For example, a “class”
schema includes such concepts as “students”, “classroom”, and “lecture” as accessible entities.
Notes 

. Exceptions to this pattern include the occurrence of “competitive” referents (where given
information is expressed with a full noun), and that of “contrastive” referents (where given
information receives a primary accent) (Chafe 1994: 76–77).

. Noun phrases that are outside the domain of identifiability are of two types: those that
express generic referents, and those that are non-referential (e.g. non-referential it, negative
and universal pronouns, question words, predicate nouns, and non-specific referents in an
irrealis context) (Chafe 1994: 102–105).

. The segmentation of spoken discourse into prosodic units, it should be noted, is most
closely concerned with one of the grammatical functions of intonation, i.e. its “segmenta-
tive” or “delimitative” function, while simultaneously involving its other functions as well
(e.g. to indicate finality/non-finality, show the speaker’s emotions, or regulate the speaker-
hearer relationship) (Vandepitte 1989; cf. Section 3.2).

. Chafe (1980b) uses the term “idea units”, which are essentially the same as IUs.

. It is important to realize that the IU analysis is a unit summary system, where the
intonation contour is indicated only once per unit. In other words, a symbol such as a
period indicating a fall in pitch placed at the end of an IU does not show an intonational
event taking place just at that point; rather it represents a set of intonational movements that
occur over the course of the unit in which it appears (Du Bois et al. 1992: 113).

. Chafe (1994: 65) notes that a more careful identification of IU boundaries in his more
recent work has led him to revise the modal length of substantive IUs in English, which was
previously identified as five words (Chafe 1980b, 1987, 1993).

. AUs, like IUs, are either substantive or regulatory. A substantive AU contains at least
one open-class word such as a noun, verb, or adjective; whereas a regulatory AU verbalizes
a linkage, or it may be an interpersonal or cognitive formula (e.g. and then, you know, let’s
see). The distinction between given, accessible, and new information thus applies only to
substantive AUs; regulatory AUs are outside the domain of activation cost (Chafe 1993: 40).

. Linguists working in the tradition of “functional sentence perspective” claim that
linguistic elements vary in their degree of CD (communicative dynamism), which is
characterized as “the relative extent to which a linguistic element contributes towards the
further development of the communication” (Firbas 1992: 8).

. Iwasaki and Tao (1993) defined the terms as follows: “An IU is ‘clausal’ if it contains a
verbal predicate; any IUs lacking a verbal predicate are ‘non-clausal’. The clausal unit which
contains a verbal predicate and its associated core arguments (at least the subject) is a ‘full
clause’; the clausal unit with a verbal predicate but without its associated core arguments is a
‘semi-clause’. ‘Nominal’ (NP) IUs refer to any non-clausal unit which consists of a nominal
element, whether or not it is modified”.

. Iwasaki and Tao (1993) examined Mandarin Chinese IUs as well. They found the
proportions of non-clausal IUs (61%), semi-clausal IUs (63%), and NP IUs (23%) in
Mandarin comparable to those in Japanese. That is, the Mandarin and Japanese data were
 Notes

found to be similar in terms of the amount of “syntactic fragmentation” they exhibited. The
preferential use of semi-clauses in Japanese and Mandarin can be attributed to the fact that
both languages, unlike English, allow for abundant use of zero anaphora.

. Brown and Yule (1983: 159–160), examining Halliday’s conversational English data, note
that the “information unit” is more likely to take the form of the phrase rather than the
clause, and argue for abandoning the clause as the unmarked syntactic domain of the unit.
Maynard (1989) likewise reports that her English conversation data displayed a similar rate
of segmentation into phrasal units as was found in comparable Japanese data. This would
suggest that spoken English is also “fragmented”.

. The Pear Film narratives were collected as part of a research project investigating the
verbalization of remembered experience (Chafe 1980a). For the project, a brief film without
dialog—which portrays various adventures of a boy with a basket of pears on his bicycle—
was shown to speakers of several different languages; they were then asked to tell what
had happened in it to an interviewer of the same age and sex. An obvious advantage of
this approach is that we can focus on the cross-linguistic comparison of linguistic forms
while rigidly controlling the content. Its major disadvantage, on the other hand, would
be that these narratives, elicited under controlled conditions, do not constitute natural,
representative samples of our everyday interactive use of conversational language.

. Concerning this issue, Iwasaki (1993: 41) states as follows: “For the speaker, who attends
to more than one type of information, it is easier to deal with a smaller amount of
the ideational content of information. This can be achieved by producing part of the
proposition, or clause.” What evidence can there be for the argument that the speaker
can handle smaller rather than larger ideational content more easily when faced with the
need to process multiple types of information? It might be easier (and definitely more
efficient) for the speaker to deal with clausal information rather than phrasal information.
Moreover, it is not always the case that phrasal IUs contain smaller amount of information
than clausal IUs.

Chapter 3

. F1 is the data used in Matsumoto’s (1995a) case study.

. There were fairly frequent topic shifts, and subtopics appeared and recurred during the
45-minute conversations. It seems that the female conversations generally involved more
topic shifts than the male conversations.

. The transcribed IUs do not include what Maynard (1986) calls “turn-internal listener
backchannels”, or what Schegloff (1982) calls “continuers”—brief backchannelling expres-
sions (e.g. un ‘uh-huh’) that the interlocutor who assumes primarily a listener’s role sends
during the other interlocutor’s speaking turn, especially in a long multi-unit turn (e.g.
storytelling).
Notes 

. In identifying the five IU-demarcating intonation contours, I basically relied on my


auditory perception. However, for the purpose of measuring the degree of accuracy of my
perception and making my perceptual judgment more reliable, I supplementally used sound
spectrographic display at the UCLA Department of Linguistics Phonetic Laboratory for
acoustic measurement of pitch contours of some IUs selected from each conversation.

. When IUs terminated with continuing intonation contours, the final syllables of some of
these IUs involved pitch accent, loudness, and lengthening. It seems that these additional
prosodic features do function to make IUs segmented more distinctly from the surrounding
discourse. Those IUs marked with such accented and lengthened final syllables are often
found in casual speech of young Japanese females, especially those in their teens or
early twenties.

. From a conversation-analytic point of view, narratives differ from conversations in that


“the co-participants will, on the whole, pass every opportunity that they will otherwise
have to produce a turn of their own. That is, narratives, unlike conversations, precisely
involve unlinking between possible completion of a ‘turn constructional unit’ (TCU) and
‘transition relevance’ [see Sacks et al. 1974]. The function of a ‘story preface’ (e.g. A funny
thing happened to me today), then, is to invite the recipient to suspend transition relevance
of possible completion of TCUs. For the duration of the story, the co-participants tend
to backchannel at possible completion points, but will not use these places to start up
independent utterances of their own” (Schegloff 1992: lecture at UCLA).

. Of the 32 regulatory IUs produced in F6, the number and percentage of the subtypes
are as follows: interactional (N=14, 44%), cognitive (N=7, 22%), textual (N=6, 19%), and
validational (N=5, 15%). My analysis has shown that one linguistic form can serve different
regulatory functions in different contexts, or simultaneously serve multiple functions in a
given context (cf. Section 2.4.4). Although this study focuses on substantive IUs, detailed
research on regulatory IUs will also be needed in the future.

Chapter 4

. The nominal predicates occurred either with or without the so-called “copula” (see (4.1)
and (4.3b) for examples). Given that the copula in Japanese can be attached to almost all
types of constituents, even to adverbs and conjunctions, I do not consider it to be equivalent
to the “be-verb” in English.

. I coded the phrasal IUs for the five syntactic categories based on the functions they
perform within the clause they belong to. For example, besides ga- and o-marked NPs, I
coded NP-ni as in John ni au ‘meet John’ as argument NP, whereas NP-ni as in Tokyo ni
sumu ‘live in Tokyo’ was coded as AvP. In addition, I coded NP-no (NP plus the genitive
case-marking particle no) as attributive AP.

. This means that the study assumes that independent phrasal IUs without formally
expressed predicates (which may be assumed as in the case of Type 4 stray NPs to be
 Notes

discussed in Section 4.5.2) are propositionally incomplete and thus cannot constitute a
clause conveying a full proposition.

. Shibatani (1990: 275–278) distinguishes two types of topics in Japanese. One is a “base-
generated”, “genuine” topic that is not integratable into the clausal structure and that
expresses an entity about which a judgment is made, as in (i). The other is a “derived” topic
which he considers is a stylistic device similar to scrambling, as in (ii). In this study I coded
only “base-generated”, “genuine” wa-marked NPs as in (i) as topics.
(i) Tookyoo wa daremo konakatta.
Tokyo top no one come-neg-past
‘Tokyo is such that no one came (from there).’
(ii) Tookyoo kara wa daremo konakatta.
Tokyo abl top no one come-neg-past
‘From Tokyo, no one came.’

. This indicates that the IUs in (4.1) have been taken from male dyad M5. Hereafter,
sources of examples will be shown likewise.

. In this study, an intra-IU clause or a non-finite clause was not counted as propositionally
complete; it was coded simply as part of the single-IU/multi-IU clause in which it occurs.

. Of the 1,600 IUs, 14% (N=231) contained a total of 244 intra-IU clauses, constituting
complex clauses or complex NPs. Of the 244 intra-IU clauses, relative clauses (36%), clausal
objects (34%), and subordinate clauses (20%) accounted for 90% (nominal complement
clauses and clausal subjects accounted for the remaining 10%). It was also found that like
the clausal IUs, the intra-IU clauses tend to consist of subjectless verbal predicates only, and
they were produced typically within the independent single-IU clauses, especially within
[PVP] IUs.

. Further analysis has shown that of the 123 postposed phrases, 55% are nominals, and
36% are adverbials (the remaining 9% are adjectives and mixed phrases). It was also shown
that of the postposed NPs, 32% are Type 1, 38% are Type 2, and 30% are Type 3, whereas
73% of the postposed adverbials belong to Type 2. This suggests that postposed adverbials,
unlike postposed NPs, have a marked tendency to appear as phrasal IU constituents of
multi-IU clauses.

. I classified the “proposition-constituting” NP IUs as one type of stray NPs, given that
occasionally the speaker’s intended meaning cannot be understood by the listener, and thus
communication breakdown results. This seems to justify that this type of NPs may still be
termed “stray”.

. Clancy et al. (1996) classify collaborative finishes as one type of what they call
“reactive tokens”, which include non-lexical backchannels, lexical reactive expressions, and
repetitions. Given that most collaborative finishes are contentful, and moreover, provide
new information, I am not in favor of treating them simply as the non-primary speaker’s
non-floor-taking reactive tokens. Of the two adjacently uttered IUs which constitute a
collaboratively constructed proposition, Clancy et al. (1996) focus on the second IU uttered
Notes 

by the hearer as in (4.15c), whereas this study focuses on the first IU uttered by the primary
speaker as in (4.15b). In my coding scheme, the second IU is normally treated as an inde-
pendent clausal IU conveying a full proposition which assumes the NP just uttered by the
primary speaker in the immediately preceding turn.

. Some of the lead NPs were expanded into other grammatical categories such as adjectives
and adverbials in the following clause, as in kyoo ‘today’ (NP lead) → kyoo no ‘today’s’
(AP target).

. Comparing the independent NPs identified and discussed in Croft (1995), Tao (1996),
and this study, Croft’s “presentative” NPs would correspond to the “genuine topic” NPs
in my classification of detached NP IUs; his “topic” NPs would correspond to my “LDed”
NPs involving resumptive pronouns. I coded two subtypes of Tao’s “referential” NP IUs—
“referent-listing” NPs and “referent-anchoring” NPs (i.e. NPs describing the same referent
from different angles, which Croft calls “elaborating” NPs)—as “independent clausal” NPs,
given that each of them constitutes a complete proposition and provides a piece of newly
introduced information. Further, Tao’s “interactional” NP IUs consist of two subtypes: NPs
as repetitions and NPs as collaborative finishes. The former correspond to the “repeated”
NPs, whereas the latter are clearly related to “Type 2 stray” NPs in the taxonomy provided
in the present study. It may be concluded that the grammatically independent, detached
NP with a specific discourse function is a cross-linguistically occurring phenomenon in
spontaneous spoken discourse.

. Although I coded sentential adverbs as clause-external substantive IUs, they may be
classified as regulatory IUs, given that their function is not so much to express ideas as to
show particular kind of relationships between the preceding and following IUs.

. A conspicuous feature of the multi-IU clauses is that 30% of them included post-
predicate phrasal IU elements, as illustrated in (4.29). In particular, full clauses composed
of three IUs exhibited the highest rate of postposing; 52% of them involved post-predicate
elements. The relationship between the production of multi-IU clauses and post-predicate
phrases is worth examining in future research (cf. Section 4.3).

Chapter 5

. Following Chafe (1996: 41), “referents” are defined here more broadly as “ideas of
people, objects, or abstractions” that could function as participants in event or state
ideas. By definition, then, all NPs have “referents” and thus can be coded for information
statuses. As Lambrecht (1994: 37) notes, the information structure analysis is concerned
with referents (i.e. “entities and states of affairs designated by linguistic expressions in
particular utterances”) and the abstract mental representations of these discourse referents
in the mind of speech participants.

. Du Bois (1987) uses 20 IUs for this measure in his analysis of the Pear Story Sakapulteko
narratives, following Givón’s (1983a) measure of RD, and given the one-to-one correspon-
 Notes

dence between the clause and the IU in his data (cf. Sections 2.1, 2.6.2). I used 25 IUs (=20
IUs multiplied by 1.23) instead based on the finding that in my Japanese data one clause
contained 1.23 IUs on average (see Section 4.7). This means that I have chosen 25 IUs as
an arbitrary measure during which an active concept will be deactivated into a semi-active
state, or a concept changes from a focally active to a peripherally active state (Chafe 1987).

. There are some problems with the use of the S-A-O categories as a heuristic as well as
the notion of argument structure (Mithun & Chafe 1999; Thompson & Hopper 2001). This
study has nevertheless chosen to use the S-A-O schema primarily because important cross-
linguistic comparisons with previous information-flow studies such as Du Bois (1987), Tao
(1996), and Schuetze-Coburn (1994), all of which employed the schema, are made possible.

. Thus, for example, in a passive clause such as Mary ga John ni butareta. ‘Mary was hit by
John’, I coded ‘Mary’ as S (the single argument of the intransitive passive verb butareta), and
‘John’ as oblique. Typically the roles S and A are marked by the particle ga/wa, whereas the
role O is marked by the particle o/wa. In casual Japanese conversation, however, many NPs
have no postpositions with them (cf. Note 2 in Chapter 1).

. The rest of the oblique NPs consist of the following: NPs occurring in attributive
adjectival phrases (N=19, 6%), topics (N=11, 3%), and complements (N=2, 1%).

. The following provides a comparison between the present study and Du Bois (1987) in
the percentage of New NPs in the four grammatical roles:
S A O Oblique Total
This study 39 20 49 58 49
Du Bois 22 3 25 39 20
The higher proportion of New NPs in the Japanese data may also be attributed to the
fact that in the Pear Story narratives the number of NPs that could be introduced are
more restricted than in informal conversations, in which a wider range of NPs could be
introduced.

. About 70% of the pronouns are demonstrative, as opposed to personal, pronouns. Of the
103 demonstrative pronouns, 61% (N=63) are sore ‘it’, which includes soo ‘so’ (N=7) (23%
are are ‘that’, and 16% are kore ‘this’). Of the 44 personal pronouns, 84% (N=37) are 1st
person ‘I’ atashi/boku/ore, which includes ‘we’ (N=3) (9% are 2nd person ‘you’ anata, and
7% are 3rd person ‘he/she’ kare/kanojo).

. In Du Bois (1987), (a) a new argument NP was introduced every 4.32 clauses (IPQ=0.23)
on average; (b) a new non-argument NP was introduced every 6.45 clauses (IPQ=0.16) on
average; and (c) a new NP was introduced every 2.59 clauses (IPQ=0.39) on average.

. Du Bois (2003) argues that this argument pattern applies not only to intransitive and
transitive clauses but also to ditransitive clauses involving verbs such as give, tell, and show.
Among the three core arguments that a ditransitive verb takes, i.e. subject (A), direct object
(O), and indirect object (I), he shows, only the O role freely allows lexical NPs and new
information in English conversation. This means that in both transitive and ditransitive
clauses, the O role constitutes the only lexical/new argument allowable within a clause.
Notes 

. As I argued in Section 4.5.2, the production of “lead” NPs seems also related to the one
new NP per IU constraint. In most cases, the speakers first introduced new information as
a lead phrasal IU and then “converted it into given” in the following clausal IU, in order
to prevent the clause from containing two new NPs. This can be considered as one of the
speakers’ strategies for conforming to this constraint.

Chapter 6

. The conjunction kedo ‘though’, an informal form of keredomo ‘although’, often occurs
independently without main clauses in spoken Japanese, presumably as a softening device
which makes an asserted proposition less definite (Makino & Tsutsui 1989). In this study
every occurrence of kedo, with or without main clauses, was coded as [TX].

. In this study the functional structure of an intra-IU clause was not coded separately
but was taken simply as constituting part of the ideational [ID] component (cf. Note 7 in
Chapter 4). This means that some of the IUs actually exhibit more complex functional
structures. For example, (i) was coded as TX-ID, but actually involves more complex func-
tional composition, i.e. TX-ID-IT-ID, where the sequence ID-IT (the adjectival predicate
semi-clause muzukashii ‘(it is) difficult’ + the interactional particle na) constitutes the
intra-IU embedded clausal object of the verb ‘think’.
(i) M: dakara muzukashii na: to omotte.
so difficult fp qt think-and
[ TX ] [ ID ]
{ TX } { ID } {IT} { ID }
‘So (I) think that (it is) difficult.’ (F3)
Similarly, some of the IUs involving postposing exhibit structures more functionally com-
plex than those coded in the present study. For example, (ii), although coded as ID, actually
involves more complex structure, ID-IT-ID, with the interactional particle na separating the
nominal predicate hantai da ‘is converse (to yours)’ and the post-predicate element atashi
‘my (case)’.
(ii) M: hantai da na: atashi. ((laugh))
converse cop fp I
[ ID ]
{ ID } {IT} { ID }
‘is converse (to yours), my (case).’ (F4)

. Croft (1995) proposes three syntactic factors that will cause an English clause to be
broken up into multiple IUs. They are “parallelism” (i.e. coordinate sentences are almost
always broken), “complexity” (i.e. complex subject NPs tend to be broken), and “distance”
(i.e. adjuncts as opposed to complements tend to be found on separate IUs). Whether these
constraints apply cross-linguistically awaits future research.
References

Andersen, Roger W. (1991). A Poor Man’s Guide to Interlanguage Analysis. Los Angeles:
Department of TESL and Applied Linguistics, University of California, Los Angeles.
Ashby, William J. (1988). The syntax, pragmatics, and sociolinguistics of left- and right-
dislocations in French. Lingua, 75, 203–229.
Ashby, William J. & Paola Bentivoglio (1993). Preferred argument structure in spoken
French and Spanish. Language Variation and Change, 5, 61–76.
Atkinson, J. Maxwell & John Heritage (Eds.). (1984). Structures of Social Action: Studies of
Conversation Analysis. Cambridge: Cambridge University Press.
Beckman, Mary E. & Janet B. Pierrehumbert (1986). Intonational structure in Japanese and
English. Phonology Yearbook, 3, 255–309.
Birner, Betty J. & Gregory L. Ward (1998). Information Status and Noncanonical Word Order
in English. Amsterdam: John Benjamins.
Boden, Deirdre & Don H. Zimmerman (Eds.). (1991). Talk and Social Structure: Studies in
Ethnomethodology and Conversation Analysis. Cambridge: Polity Press.
Bolinger, Dwight (1989). Intonation and its Uses. Stanford, CA: Stanford University Press.
Brazil, David (1985). The Communicative Value of Intonation in English. Birmingham:
English Language Research, University of Birmingham.
Brown, Gillian & George Yule (1983). Discourse Analysis. Cambridge: Cambridge University
Press.
Bybee, Joan L. & Paul J. Hopper (Eds.). (2001). Frequency and the Emergence of Linguistic
Structure. Amsterdam: John Benjamins.
Chafe, Wallace L. (1972). Discourse structure and human knowledge. In R. O. Freedle &
J. B. Carroll (Eds.), Language Comprehension and the Acquisition of Knowledge (pp. 41–
69). Washington, DC: V. H. Winston.
Chafe, Wallace L. (1974). Language and consciousness. Language, 50, 111–133.
Chafe, Wallace L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point
of view. In C. Li & S. A. Thompson (Eds.), Subject and Topic (pp. 27–55). New York:
Academic Press.
Chafe, Wallace L. (1979). The flow of thought and the flow of language. In T. Givón
(Ed.), Discourse and Syntax: Syntax and Semantics, Vol. 12 (pp. 159–181). New York:
Academic Press.
Chafe, Wallace L. (Ed.). (1980a). The Pear Stories: Cognitive, Cultural, and Linguistic Aspects
of Narrative Production. Norwood, NJ: Ablex.
Chafe, Wallace L. (1980b). The deployment of consciousness in the production of a
narrative. In W. L. Chafe (Ed.), The Pear Stories: Cognitive, Cultural, and Linguistic
Aspects of Narrative Production (pp. 9–50). Norwood, NJ: Ablex.
 References

Chafe, Wallace L. (1982). Integration and involvement in speaking, writing, and oral
literature. In D. Tannen (Ed.), Spoken and Written Language: Exploring Orality and
Literacy (pp. 35–53). Norwood, NJ: Ablex.
Chafe, Wallace L. (1987). Cognitive constraints on information flow. In R. Tomlin (Ed.),
Coherence and Grounding in Discourse (pp. 21–51). Amsterdam: John Benjamins.
Chafe, Wallace L. (1988). Linking intonation units in spoken English. In J. Haiman &
S. A. Thompson (Eds.), Clause Combining in Grammar and Discourse (pp. 1–27).
Amsterdam: John Benjamins.
Chafe, Wallace L. (1993). Prosodic and functional units of language. In J. A. Edwards &
M. D. Lampert (Eds.), Talking Data: Transcription and Coding in Discourse Research
(pp. 33–43). Hillsdale, NJ: Lawrence Erlbaum.
Chafe, Wallace L. (1994). Discourse, Consciousness, and Time: The Flow and Displacement of
Conscious Experience in Speaking and Writing. Chicago: University of Chicago Press.
Chafe, Wallace L. (1996). How consciousness shapes language. Pragmatics and Cognition, 4,
35–54.
Chafe, Wallace L. (1998). Language and the flow of thought. In M. Tamasello (Ed.), The
New Psychology of Language: Cognitive and Functional Approaches to Language Structure
(pp. 93–111). Mahwah, NJ: Lawrence Erlbaum.
Chafe, Wallace L. (2000). The interplay of prosodic and segmental sounds in the expression
of thoughts. In M. L. Juge & J. L. Moxley (Eds.), Proceedings of the 23rd Annual Meeting
of the Berkeley Linguistic Society 1997 (pp. 389–401). Berkeley, CA: Berkeley Linguistic
Society.
Chesterman, Andrew (1991). On Definiteness: A Study with Special Reference to English and
Finnish. Cambridge: Cambridge University Press.
Clancy, Patricia M. (1980). Referential choice in English and Japanese narrative discourse.
In W. L. Chafe (Ed.), The Pear Stories: Cognitive, Cultural, and Linguistic Aspects of
Narrative Production (pp. 127–202). Norwood, NJ: Ablex.
Clancy, Patricia M. (1982). Written and spoken style in Japanese narratives. In D. Tannen
(Ed.), Spoken and Written Language: Exploring Orality and Literacy (pp. 55–76).
Norwood, NJ: Ablex.
Clancy, Patricia M. (2003). The lexicon in interaction: Developmental origins of Preferred
Argument Structure in Korean. In J. W. Du Bois, L. E. Kumpf, & W. J. Ashby (Eds.),
Preferred Argument Structure: Grammar as Architecture for Function (pp. 81–108).
Amsterdam: John Benjamins.
Clancy, Patricia M., Sandra A. Thompson, Ryoko Suzuki, & Hongyin Tao (1996). The
conversational use of reactive tokens in English, Japanese, and Mandarin. Journal of
Pragmatics, 26, 355–387.
Clark, Herbert H. & Susan E. Haviland (1977). Comprehension and the given-new contract.
In R. O. Freedle (Ed.), Discourse Production and Comprehension (pp. 1–40). Norwood,
NJ: Ablex.
Clark, Herbert H. & Catherine R. Marshall (1981). Definite reference and mutual
knowledge. In A. K. Joshi, B. L. Webber, & I. A. Sag (Eds.), Elements of Discourse
Understanding (pp. 10–63). Cambridge: Cambridge University Press.
Comrie, Bernard (1978). Ergativity. In W. P. Lehmann (Ed.), Syntactic Typology (pp. 329–
394). Austin: University of Texas Press.
References 

Couper-Kuhlen, Elizabeth & Margaret Selting (Eds.). (1996). Prosody in Conversation:


Interactional Studies. Cambridge: Cambridge University Press.
Croft, William (1995). Intonation units and grammatical structure. Linguistics, 33, 839–882.
Cruttenden, Alan (1986). Intonation. Cambridge: Cambridge University Press.
Crystal, David (1975). The English Tone of Voice. London: St. Martin.
Dik, Simon C. (1989). The Theory of Functional Grammar, Part I: The Structure of the Clause.
Dordrecht, Holland: Foris.
Dixon, Robert M. W. (1979). Ergativity. Language, 55, 59–138.
Dorgeloh, Heidrun (1997). Inversion in Modern English: Form and Function. Amsterdam:
John Benjamins.
Downing, Pamela & Michael Noonan (Eds.). (1995). Word Order in Discourse. Amsterdam:
John Benjamins.
Drew, Paul & John Heritage (Eds.). (1992). Talk at Work: Interaction in Institutional Settings.
Cambridge: Cambridge University Press.
Du Bois, John W. (1980). Beyond definiteness: The trace of identity in discourse. In
W. L. Chafe (Ed.), The Pear Stories: Cognitive, Cultural, and Linguistic Aspects of
Narrative Production (pp. 203–274). Norwood, NJ: Ablex.
Du Bois, John W. (1987). The discourse basis of ergativity. Language, 63, 805–855.
Du Bois, John W. (2003). Discourse and grammar. In M. Tamasello (Ed.), The New
Psychology of Language: Cognitive and Functional Approaches to Language Structure,
Vol. 2 (pp. 47–87). Mahwah, NJ: Lawrence Erlbaum.
Du Bois, John W., Lorraine E. Kumpf, & William J. Ashby (Eds.). (2003). Preferred Argument
Structure: Grammar as Architecture for Function. Amsterdam: John Benjamins.
Du Bois, John W., Stephan Schuetze-Coburn, Susanna Cumming, & Danae Paolino (1993).
Outline of discourse transcription. In J. A. Edwards & M. D. Lampert (Eds.), Talking
Data: Transcription and Coding in Discourse Research (pp. 45–89). Hillsdale, NJ:
Lawrence Erlbaum.
Du Bois, John W., Stephan Schuetze-Coburn, Danane Paolino, & Sussana Cumming (1992).
Discourse Transcription: Santa Barbara Papers in Linguistics, Vol. 4. Santa Barbara:
Department of Linguistics, University of California, Santa Barbara.
Duranti, Alessandro & Elinor Ochs (1979). Left-dislocation in Italian conversation. In
T. Givón (Ed.), Discourse and Syntax: Syntax and Semantics, Vol. 12 (pp. 377–416).
New York: Academic Press.
Durie, Mark (1994). A case study of pragmatic linking. Text, 14, 495–529.
Edwards, Jane A. & Martin D. Lampert (Eds.). (1993). Talking Data: Transcription and
Coding in Discourse Research. Hillsdale, NJ: Lawrence Erlbaum.
Firbas, Jan (1992). Functional Sentence Perspective in Written and Spoken Communication.
Cambridge: Cambridge University Press.
Ford, Cecilia E. & Sandra A. Thompson (1996). Interactional units in conversation:
Syntactic, intonational, and pragmatic resources for the management of turns. In
E. Ochs, E. A. Schegloff, & S. A. Thompson (Eds.), Interaction and Grammar (pp. 134–
184). Cambridge: Cambridge University Press.
Ford, Cecilia E., Barbara A. Fox, & Sandra A. Thompson (1996). Practices in the
construction of turns: The ‘TCU’ revisited. Pragmatics, 6, 427–454.
 References

Ford, Cecilia E., Barbara A. Fox, & Sandra A. Thompson (Eds.). (2002). The Language of
Turn and Sequence. Oxford: Oxford University Press.
Ford, Cecilia E., Barbara A. Fox, & Sandra A. Thompson (2003). Social interaction and
grammar. In M. Tamasello (Ed.), The New Psychology of Language: Cognitive and
Functional Approaches to Language Structure, Vol. 2 (pp. 119–143). Mahwah, NJ:
Lawrence Erlbaum.
Fox, Barbara A. (1985). Word-order inversion and discourse continuity in Tagalog. Text, 5,
39–54.
Fox, Barbara A. & Sandra A. Thompson (1990). A discourse explanation of the grammar of
relative clauses in English conversation. Language, 66, 297–316.
Fox, Barbara A., Makoto Hayashi, & Robert Jasperson (1996). Resources and repair: A cross-
linguistic study of syntax and repair. In E. Ochs, E. A. Schegloff, & S. A. Thompson
(Eds.), Interaction and Grammar (pp. 185–237). Cambridge: Cambridge University
Press.
Fujii, Noriko & Tsuyoshi Ono (2000). The occurrence and non-occurrence of the Japanese
direct object marker o in conversation. Studies in Language, 24, 1–39.
Gee, James P. (1986). Units in the production of narrative discourse. Discourse Processes, 9,
391–422.
Geluykens, Ronald (1992). From Discourse Process to Grammatical Construction: On Left
Dislocation in English. Amsterdam: John Benjamins.
Givón, Talmy (1975). Focus and the scope of assertion: Some Bantu evidence. Studies in
African Linguistics, 6, 185–205.
Givón, Talmy (Ed.). (1983a). Topic Continuity in Discourse: A Quantitative Cross-Language
Study. Amsterdam: John Benjamins.
Givón, Talmy (1983b). Topic continuity in discourse: An introduction. In T. Givón
(Ed.), Topic Continuity in Discourse: A Quantitative Cross-Language Study (pp. 1–41).
Amsterdam: John Benjamins.
Givón, Talmy (1984). Syntax: A Functional-Typological Introduction, Vol. I. Amsterdam:
John Benjamins.
Givón, Talmy (1987). Beyond foreground and background. In R. S. Tomlin (Ed.), Coherence
and Grounding in Discourse (pp. 175–188). Amsterdam: John Benjamins.
Givón, Talmy (1988). Topic continuity in discourse: The functional domain of switch
reference. In J. Haiman & S. A. Thompson (Eds.), Clause Combining in Grammar and
Discourse (pp. 51–82). Amsterdam: John Benjamins.
Givón, Talmy (1990). Syntax: A Functional-Typological Introduction, Vol. II. Amsterdam:
John Benjamins.
Green, Georgia M. (1980). Some wherefores of English inversions. Language, 56, 582–601.
Grice, H. Paul (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Speech Acts:
Syntax and Semantics, Vol. 3 (pp. 41–58). New York: Academic Press.
Grimes, Joseph E. (1975). The Thread of Discourse. The Hague: Mouton.
Haiman, John & Sandra A. Thompson (Eds.). (1988). Clause Combining in Grammar and
Discourse. Amsterdam: John Benjamins.
Halliday, Michael A. K. (1967a). Intonation and Grammar in British English. The Hague:
Mouton.
References 

Halliday, Michael A. K. (1967b). Notes on transitivity and theme in English: Part 2. Journal
of Linguistics, 3, 199–244.
Halliday, Michael A. K. (1973). Explorations in the Functions of Language. London: Edward
Arnold.
Halliday, Michael A. K. (1985). An Introduction to Functional Grammar. London: Edward
Arnold.
Halliday, Michael A. K. (1989). Language, Context, and Text: Aspects of Language in a Social-
Semiotic Perspective. London: Oxford University Press.
Halliday, Michael A. K. (1994). An Introduction to Functional Grammar (2nd ed.). London:
Edward Arnold.
Halliday, Michael A. K. & Ruqaiya Hasan (1976). Cohesion in English. London: Longman.
Hawkins, John A. (1978). Definiteness and Indefiniteness: A Study in Reference and
Grammaticality Prediction. London: Croom Helm.
Hayashi, Makoto (2003). Joint Utterance Construction in Japanese Conversation. Amsterdam:
John Benjamins.
Helasvuo, Marja-Liisa (2001). Syntax in the Making: The Emergence of Syntactic Units in
Finnish Conversation. Amsterdam: John Benjamins.
Heritage, John (1997). Conversation analysis and institutional talk: Analyzing data. In
D. Silverman (Ed.), Qualitative Research: Theory, Method and Practice (pp. 161–182).
London: Sage.
Hinds, John (1976). Postposing in Japanese. Eoneo: The Journal of the Linguistic Society of
Korea, 1, 113–125.
Hinds, John (1980). Japanese conversation, discourse structure, and ellipsis. Discourse
Processes, 3, 263–286.
Hinds, John (1982). Ellipsis in Japanese Discourse. Alberta, Canada: Linguistic Research, Inc.
Hinds, John (1983). Topic continuity in Japanese. In T. Givón (Ed.), Topic Continuity
in Discourse: A Quantitative Cross-Language Study (pp. 47–93). Amsterdam: John
Benjamins.
Hinds, John, Senko K. Maynard, & Shoichi Iwasaki (Eds.). (1987). Perspectives on
Topicalization: The Case of Japanese WA. Amsterdam: John Benjamins.
Hirst, Daniel & Albert Di Cristo (Eds.). (1998). Intonation Systems: A Survey of Twenty
Languages. Cambridge: Cambridge University Press.
Hoji, Hajime (1985). Logical Form Constraints and Configurational Structures in Japanese.
Ph.D. dissertation, University of Washington, Seattle.
Hopper, Paul J. (1979). Aspect and grounding in discourse. In T. Givón (Ed.), Discourse and
Syntax: Syntax and Semantics, Vol. 12 (pp. 213–241). New York: Academic Press.
Hopper, Paul J. & Sandra A. Thompson (1980). Transitivity in grammar and discourse.
Language, 56, 251–299.
Horie, Kaoru (2000). Core-oblique distinction and nominalizer choice in Japanese and
Korean. Studies in Language, 24, 77–102.
Iwasaki, Shoichi (1985). The ‘Given A Constraint’ and the Japanese particle ga. In
S. DeLancey & R. S. Tomlin (Eds.), Proceedings of the First Annual Pacific Linguistics
Conference (pp. 152–167). Eugene: Department of Linguistics, University of Oregon.
Iwasaki, Shoichi (1992). Subjectivity in Grammar and Discourse: Theoretical Considerations
and a Case Study of Japanese Spoken Discourse. Amsterdam: John Benjamins.
 References

Iwasaki, Shoichi (1993). The structure of the intonation unit in Japanese. In S. Choi (Ed.),
Japanese/Korean Linguistics, Vol. 3 (pp. 39–53). Stanford, CA: Center for the Study of
Language and Information.
Iwasaki, Shoichi (2002). Japanese. Amsterdam: John Benjamins.
Iwasaki, Shoichi & Hongyin Tao (1993). A comparative study of the structure of the
intonation unit in English, Japanese, and Mandarin Chinese. Paper presented at the
67th Annual Meeting of the Linguistic Society of America, Los Angeles, CA.
Jones, L. B. & Jones, L. K. (1979). Multiple levels of information in discourse. In L. K. Jones &
R. E. Longacre (Eds.), Discourse Studies in Mesoamerican Languages (pp. 3–28).
Arlington: University of Texas, Arlington.
Kärkkäinen, Elise (1996). Preferred argument structure and subject role in American
English conversational discourse. Journal of Pragmatics, 25, 675–701.
Kreckel, Marga (1981). Tone units as message blocks in natural discourse: Segmentation of
face-to-face interaction by naive, native speakers. Journal of Pragmatics, 5, 459–476.
Kuno, Susumu (1973). The Structure of the Japanese Language. Cambridge, MA: MIT Press.
Kuno, Susumu (1978a). Danwa no Bumpoo [Grammar of Discourse]. Tokyo: Taishuukan.
Kuno, Susumu (1978b). Generative discourse analysis in America. In W. Dressler (Ed.),
Current Trends in Textlinguistics (pp. 275–294). Berlin: Mouton de Gruyter.
Kuno, Susumu (1978c). Japanese: A characteristic OV language. In W. P. Lehmann (Ed.),
Syntactic Typology: Studies of the Phenomenology of Language (pp. 57–138). Austin:
University of Texas Press.
Labov, William (1972). The transformation of experience in narrative syntax. In Language
in the Inner City: Studies in the Black English Vernacular (pp. 354–396). Philadelphia:
University of Pennsylvania Press.
Ladd, D. Robert (1996). The Intonational Phonology. Cambridge: Cambridge University
Press.
Lambrecht, Knud (1987). On the status of SVO sentences in French discourse. In
R. S. Tomlin (Ed.), Coherence and Grounding in Discourse (pp. 217–261). Amsterdam:
John Benjamins.
Lambrecht, Knud (1994). Information Structure and Sentence Form: Topic, Focus, and the
Mental Representations of Discourse Referents. Cambridge: Cambridge University Press.
Lerner, Gene H. (1991). On the syntax of sentences-in-progress. Language in Society, 20,
441–458.
Longacre, Robert E. (1976). The discourse structure of the flood narrative. In G. MacRae
(Ed.), Society of Biblical Literature 1976 Seminar Papers (pp. 235–262). Missoula, MT:
Scholars Press.
Makino, Seiichi & Michio Tsutsui (1989). A Dictionary of Basic Japanese Grammar. Tokyo:
The Japan Times.
Matsumoto, Kazuko (1995a). Fragmentation in conversational Japanese. Japan Association
for Language Teaching Journal, 17, 238–253.
Matsumoto, Kazuko (1995b). Wh-preposing, scrambling, and topicalization in Japanese.
The Bulletin of Aichi University of Education, 44, 45–62.
Matsumoto, Kazuko (1996). Intonation Units in Conversational Japanese: Structure and
Linkage. Ph.D. dissertation, University of California, Los Angeles.
Matsumoto, Kazuko (1997a). NPs in Japanese conversation. Pragmatics, 7, 163–181.
References 

Matsumoto, Kazuko (1997b). Independent phrasal intonation units in conversational


discourse. The Bulletin of Aichi University of Education, 46, 29–36.
Matsumoto, Kazuko (1997c). Postposing in Japanese discourse. Studies in Foreign Languages
and Literature, 33, 79–88.
Matsumoto, Kazuko (1998a). Detached NPs in Japanese conversation: Types and functions.
Text, 18, 417–444.
Matsumoto, Kazuko (1998b). On the ‘one new NP per unit’ constraint in spoken Japanese.
The Bulletin of Aichi University of Education, 47, 17–25.
Matsumoto, Kazuko (1999a). Intonation units and functional composition in Japanese
conversational discourse. Functions of Language, 6, 163–194.
Matsumoto, Kazuko (1999b). And-prefaced questions in institutional discourse. Linguistics,
37, 251–274.
Matsumoto, Kazuko (1999c). Units of conversation in Japanese. In T. Yamamoto,
M. Nakamura, & M. Kamiya (Eds.), Symphony of Words: Papers in Honor of Masahiko
Kanno (pp. 105–114). Tokyo: Eihoosha.
Matsumoto, Kazuko (2000a). Intonation units, clauses and preferred argument structure in
conversational Japanese. Language Sciences, 21, 63–86.
Matsumoto, Kazuko (2000b). Japanese intonation units and syntactic structure. Studies in
Language, 24, 515–564.
Matsumoto, Kazuko (2000c). Marked word order and information status. The Bulletin of
Aichi University of Education, 49, 11–17.
Matsumoto, Kazuko (2001). Phrasal vs. clausal units in spoken interaction. The Bulletin of
Aichi University of Education, 50, 95–101.
Matsumoto, Kazuko (2002). On the structure of the clause in spoken discourse. The Bulletin
of Aichi University of Education, 51, 99–104.
Matsumoto, Kazuko (2003). Transitivity and clauses in conversational English and Japanese.
The Bulletin of Aichi University of Education, 52, 125–131.
Matsumoto, Kazuko (to appear). Unit linking in conversational Japanese. Language Sciences.
Maynard, Senko K. (1986). On back-channel behavior in Japanese and English casual
conversation. Linguistics, 24, 1079–1108.
Maynard, Senko K. (1989). Japanese Conversation: Self-Contexualization through Structure
and Interactional Management. Norwood, NJ: Ablex.
Maynard, Senko K. (1990). An Introduction to Japanese Grammar and Communication
Strategies. Tokyo: The Japan Times.
Maynard, Senko K. (1993). Discourse Modality: Subjectivity, Emotion and Voice in the
Japanese Language. Amsterdam: John Benjamins.
Mithun, Marianne (1987). Is basic word order universal? In R. S. Tomlin (Ed.), Coherence
and Grounding in Discourse (pp. 281–328). Amsterdam: John Benjamins.
Mithun, Marianne & Wallace L. Chafe (1999). What are S, A, and O? Studies in Language,
23, 579–606.
Mori, Junko (1999). Negotiating Agreement and Disagreement in Japanese: Connective
Expressions and Turn Construction. Amsterdam: John Benjamins.
Nespor, Marina & Irene Vogel (1986). Prosodic Phonology. Dordrecht, Holland: Foris.
Ochs, Elinor (1979). Planned and unplanned discourse. In T. Givón (Ed.), Discourse and
Syntax: Syntax and Semantics, Vol. 12 (pp. 51–80). New York: Academic Press.
 References

Ochs, Elinor, Emanuel A. Schegloff, & Sandra A. Thompson (Eds.). (1996). Interaction and
Grammar. Cambridge: Cambridge University Press.
O’Dowd, Elizabeth (1990). Discourse pressure, genre and grammatical alignment—after
Du Bois. Studies in Language, 14, 365–403.
Ono, Tsuyoshi & Ryoko Suzuki (1992). Word order variability in Japanese conversation:
Motivations and grammaticization. Text, 12, 429–445.
Ono, Tsuyoshi & Sandra A. Thompson (1994). Unattached NPs in English conversation. In
S. Gahl, A. Dolbey, & C. Johnson (Eds.), Proceedings of the 20th Annual Meeting of the
Berkeley Linguistic Society (pp. 402–419). Berkeley, CA: Berkeley Linguistic Society.
Ono, Tsuyoshi & Sandra A. Thompson (1995). What can conversation tell us about syntax?
In P. W. Davis (Ed.), Alternative Linguistics: Descriptive and Theoretical Modes (pp. 213–
271). Amsterdam: John Benjamins.
Ono, Tsuyoshi & Sandra A. Thompson (1996). Interaction and syntax in the structure of
conversational discourse. In E. H. Hovy & D. R. Scott (Eds.), Discourse Processing: An
Interdisciplinary Perspective (pp. 67–96). Heidelberg: Springer-Verlag.
Ono, Tsuyoshi, Sandra A. Thompson, & Ryoko Suzuki (2000). The pragmatic nature of the
so-called subject marker ga in Japanese: Evidence from conversation. Discourse Studies,
2, 55–84.
Pawley, Andrew & Frances H. Syder (1983). Natural selection in syntax: Notes on adaptive
variation and change in vernacular and literary grammar. Journal of Pragmatics, 7,
551–579.
Pawley, Andrew & Frances H. Syder (2000). The one-clause-at-a-time hypothesis. In
H. Riggenbach (Ed.), Perspectives on Fluency (pp. 163–199). Ann Arbor: University of
Michigan Press.
Payne, Doris L. (Ed.). (1992). Pragmatics of Word Order Flexibility. Amsterdam: John
Benjamins.
Pierrehumbert, Janet B. & Mary E. Beckman (1988). Japanese Tone Structure. Cambridge,
MA: MIT Press.
Pierrehumbert, Janet B. & Julia Hirschberg (1990). The meaning of intonational contours
in the interpretation of discourse. In P. R. Cohen, J. Morgan, & M. E. Pollack (Eds.),
Intentions in Communication (pp. 271–311). Cambridge, MA: MIT Press.
Prince, Ellen F. (1981). Toward a typology of given-new information. In P. Cole (Ed.),
Radical Pragmatics (pp. 223–255). New York: Academic Press.
Prince, Ellen F. (1992). The ZPG letter: Subjects, definiteness, and information-status. In
W. C. Mann & S. A. Thompson (Eds.), Discourse Description: Diverse Linguistic Analyses
of a Fund Raising Text (pp. 295–325). Amsterdam: John Benjamins.
Reid, Wallis (1977). The quantitative validation of a grammatical hypothesis: The passé
simple and the imparfait. In J. A. Kegl et al. (Eds.), Proceedings of the 7th Annual
Meeting of the North East Linguistic Society (pp. 315–333). Cambridge, MA: North East
Linguistic Society.
Reinhart, Tanya (1984). Principles of gestalt perception in the temporal organization of
narrative texts. Linguistics, 22, 779–809.
Sacks, Harvey, Emanuel A. Schegloff, & Gail Jefferson (1974). A simplest systematics for the
organization of turn-taking for conversation. Language, 50, 696–735.
References 

Saeki, Tetsuo (1975). Gendai Nihongo no Gojun [Word Order of Modern Japanese]. Tokyo:
Kasama Shoin.
Saito, Mamoru (1985). Some Asymmetries in Japanese and their Theoretical Implications.
Ph.D. dissertation, MIT, Cambridge, MA.
Sarangi, Srikant & Celia Roberts (Eds.). (1999). Talk, Work and Institutional Order: Discourse
in Medical, Mediation and Management Settings. Berlin: Mouton de Gruyter.
Schank, Roger C. & Robert P. Abelson (1977). Scripts, Plans, Goals, and Understanding: An
Inquiry into Human Knowledge Structures. Hillsdale, NJ: Lawrence Erlbaum.
Schegloff, Emanuel A. (1979). The relevance of repair to syntax-for-conversation. In
T. Givón (Ed.), Discourse and Syntax: Syntax and Semantics, Vol. 12 (pp. 261–286).
New York: Academic Press.
Schegloff, Emanuel A. (1982). Discourse as an interactional achievement: Some uses
of ‘uh huh’ and other things that come between sentences. In D. Tannen (Ed.),
Analyzing Discourse: Text and Talk, Georgetown University Round Table on Languages
and Linguistics 1981 (pp. 71–93). Washington, DC: Georgetown University Press.
Schegloff, Emanuel A. (1996). Turn organization: One direction for inquiry into grammar
and interaction. In E. Ochs, E. A. Schegloff, & S. A. Thompson (Eds.), Interaction and
Grammar (pp. 52–133). Cambridge: Cambridge University Press.
Schegloff, Emanuel A. (1998). Reflections on studying prosody in talk-in-interaction.
Language and Speech, 41, 235–263.
Schegloff, Emanuel A., Gail Jefferson, & Harvey Sacks (1977). The preference for self-
correction in the organization of repair in conversation. Language, 53, 361–381.
Scheibman, Joanne (2002). Point of View and Grammar: Structural Patterns of Subjectivity in
American English Conversation. Amsterdam: John Benjamins.
Schiffrin, Deborah (1987). Discourse Markers. Cambridge: Cambridge University Press.
Schuetze-Coburn, Stephan (1994). Prosody, Syntax, and Discourse Pragmatics: Assessing
Information Flow in German Conversation. Ph.D. dissertation, University of
California, Los Angeles.
Schuetze-Coburn, Stephan, Marian Shapley, & Elizabeth Weber (1991). Units of intonation
in discourse: A comparison of acoustic and auditory analyses. Language and Speech, 34,
207–234.
Selkirk, Elizabeth O. (1984). Phonology and Syntax: The Relation between Sound and
Structure. Cambridge, MA: MIT Press.
Selting, Margaret (1996). On the interplay of syntax and prosody in the constitution of
turn-constructional units and turns in conversation. Pragmatics, 6, 371–388.
Selting, Margaret (2000). The construction of units in conversational talk. Language in
Society, 29, 477–517.
Selting, Margaret & Elizabeth Couper-Kuhlen (Eds.). (2001). Studies in Interactional
Linguistics. Amsterdam: John Benjamins.
Shibatani, Masayoshi (1990). The Languages of Japan. Cambridge: Cambridge University
Press.
Silva-Corvalán, Carmen (1983). On the interaction of word order and intonation: Some
OV constructions in Spanish. In F. Klein-Andreu (Ed.), Discourse Perspectives on Syntax
(pp. 117–140). New York: Academic Press.
 References

Smith, Wendy (1996). Spoken narratives and preferred clause structure: Evidence from
modern Hebrew discourse. Studies in Language, 20, 163–189.
Stein, Dieter & Susan Wright (Eds.). (1995). Subjectivity and Subjectivization. Cambridge:
Cambridge University Press.
Sugito, Miyoko (Ed.). (1989). Nihongo to Nihongo Kyooiku [Japanese and Japanese Language
Teaching], Vol. 2. Tokyo: Meiji Shoin.
Sugito, Miyoko (Ed.). (1990). Nihongo to Nihongo Kyooiku [Japanese and Japanese Language
Teaching], Vol. 3. Tokyo: Meiji Shoin.
Svartvik, Jan (1982). The segmentation of impromptu speech. In N. E. Enkvist (Ed.),
Impromptu Speech: A Symposium (pp. 131–145). Åbo, Finland: Åbo Akademi.
Takami, Kenichi (Ed.). (1995). Nichieigo no Uhoo Idoo Koobun [Rightward Movement
Constructions in English and Japanese]. Tokyo: Hitsuji Shoboo.
Tanaka, Hiroko (1999). Turn-Taking in Japanese Conversation: A Study in Grammar and
Interaction. Amsterdam: John Benjamins.
Tannen, Deborah (1979). What’s in a frame? Surface evidence for underlying expectations.
In R. O. Freedle (Ed.), New Directions in Discourse Processing (pp. 137–181). Norwood,
NJ: Ablex.
Tannen, Deborah (Ed.). (1982). Spoken and Written Language: Exploring Orality and
Literacy. Norwood, NJ: Ablex.
Tao, Hongyin (1996). Units in Mandarin Conversation: Prosody, Discourse, and Grammar.
Amsterdam: John Benjamins.
Teramura, Hideo (1984). Nihongo no Sintakkusu to Imi [Japanese Syntax and Semantics],
2 vols. Tokyo: Kuroshio Shuppan.
Thompson, Sandra A. (1997). Discourse motivations for the core-oblique distinction as a
language universal. In A. Kamio (Ed.), Directions in Functional Linguistics (pp. 59–82).
Amsterdam: John Benjamins.
Thompson, Sandra A. & Paul J. Hopper. (2001). Transitivity, clause structure, and argument
structure: Evidence from conversation. In J. L. Bybee & P. J. Hopper (Eds.), Frequency
and the Emergence of Linguistic Structure (pp. 27–60). Amsterdam: John Benjamins.
Tokieda, Motoki (1950). Nohon Bumpoo [Japanese Grammar]. Tokyo: Iwanami.
Tomlin, Russell S. (1985). Foreground-background information and the syntax of
subordination. Text, 5, 85–122.
Tomlin, Russell S. (Ed.). (1987). Coherence and Grounding in Discourse. Amsterdam: John
Benjamins.
Vandepitte, Sonia (1989). Pragmatic function of intonation: Tone and cognitive
environment. Lingua, 79, 265–297.
van Dijk, Teun A. (Ed.). (1997). Discourse as Social Interaction, Discourse Studies: A Multi-
disciplinary Introduction, Vol. 2. London: Sage.
Venditti, Jennifer J. (2003). The J_ToBI model of Japanese intonation. In S.-A. Jun (Ed.),
Prosodic Transcription and Typology: A Unified Approach. Oxford: Oxford University
Press.
Wallace, Stephen (1982). Figure and ground: The interrelationships of linguistic categories.
In P. J. Hopper (Ed.), Tense-Aspect: Between Semantics and Pragmatics (pp. 201–223).
Amsterdam: John Benjamins.
References 

Ward, Gregory L. (1988). The Semantics and Pragmatics of Preposing. New York: Garland.
Ward, Gregory L. & Julia Hirschberg (1985). Implicating uncertainty: The pragmatics of
fall-rise intonation. Language, 61, 747–776.
Index

A Added information 25 See also


A See A role Starting point
A role 41–42, 103–105, 108–117, Addressee 13, 34, 35, 143
124–125, 127–134, 136–137, 161, Adjectival IU 31, 53, 83, 89, 91
175–176 See also Transitive subject Adjectival/adjective phrase (AP) 31,
Abelson, Robert P. 170 52–53, 55, 58, 61, 63, 65–66, 76, 83,
Accelerated syllable 21 See also 89–91, 103, 107, 109, 121, 175–176
Anacrusis See also Attributive; Predicate AP
Accent 17, 23–24, 171, 173 See also Adjectival predicate 52, 57, 62, 67, 99,
Primary; Secondary accent 103, 105, 112, 121–122, 124–125,
Accent unit (AU) 23–25, 171, 173 127, 132, 177
defined 23–24 Adjectival predicate clause 62, 99, 105,
substantive vs. regulatory 171 121, 125–129, 132–133, 149, 177
Accessibility 9 Adjective 18, 38–39, 53, 63, 65, 91,
104–105, 108, 115, 171, 174–175
Accessible information/NP 15–17, 19,
See also Attributive; Predicate
24–25, 40, 81, 100, 102–103,
adjective
111–112, 115
Adjective + NP 104, 108, 113–115, 117
Acknowledgment 71–72, 87, 91
Adjunct 54, 57, 177
Acoustic analysis 167
Adverb/adverbial 38, 41, 54, 63, 89–92,
Acoustic unit 2, 21 See also
102, 105, 110–111, 173–175
Declination unit
See also Sentential adverb
Action 10–12, 105, 126–127, 170
Adverbial/adverbial phrasal IU 31, 53,
Activation cost 9, 14–19, 41, 115, 171
60, 89–90, 96, 100, 118, 121, 136,
and definiteness 18 147, 154
and identifiability 18–19 Adverbial phrase 52–53, 56–58, 61, 63,
defined 16 65–66, 82, 85–86, 89–92, 97, 99–100,
expression of 17–18, 115 102–103, 107, 109, 121, 123, 149, 173
perspectives on 16–17 Affect 34, 37, 90
See also Given-accessible-new Affirmation 70, 72, 87, 170
distinction Agent 103, 134, 170
Activation states 2, 15, 24, 39, 115 See Agreement 46, 71–72, 75, 87
also Active; Inactive; Semi-active state Anacrusis 21
Active consciousness 23, 26, 102 Anaphora 9 See also Zero anaphora
Active state 2, 15–17, 21–22, 39, 81, 138, Anaphoric zero, vs. non-anaphoric 124
170, 176 Andersen, Roger W. 46
focally vs. peripherally 15, 176 AP See Adjectival phrase
 Index

Argument 5, 8–9, 33, 41–42, 52–54, 56, Background 10–12, 15, 21, 25, 39, 127,
60, 69–70, 79, 89, 96, 101, 103, 166, 170
105, 108–111, 116–118, defined 10–11
120–125, 127–134, 138–139, vs. foreground 10–12, 170
147, 155–156, 160–161, Backgrounding 11–12, 65, 165, 170
163–165, 171–173, 176 Bantu languages 39
defined 103 Bare noun 104, 108, 113–117, 139, 160
overt vs. null 132–134 Base-generated topic See Genuine topic
types and definitions 103 Baseline pitch level 21
vs. non-argument 103–104, 110, Basic word order 3–4, 63, 169
116, 118, 138–139, 160, 176 vs. marked word order 4, 63
vs. predicate 53, 56, 69–70 See also Canonical; Unmarked word
See also A role; Core argument; order
Core-oblique distinction; O Beckman, Mary E. 28–29, 46
role; Preferred Argument Bentivoglio, Paola 3, 42
Structure; S role Birner, Betty J. 1, 14, 169
Argument slot 118, 120, 132, 161 Boden, Deirdre 166
filled vs. unfilled with overt NP Bolinger, Dwight 46
132–133 Boundary, IU 2, 20–21, 46, 143, 150,
Argument-oblique combination 124 171
A-role argument See A role marker/signal of 20, 46, 143, 150
Article 11, 18–19 See also Definite; Brand-new information 14 See also
Indefinite article Evoked; Unused information
Ashby, William J. 3, 42, 71 Brazil, David 20
Aspect 12, 105, 170 Brown, Gillian 17–18, 30, 172
Assessments 127 Bybee, Joan L. 161
Assumed familiarity 14
Atkinson, J. Maxwell 46
C
Attention focus 27–28, 170
Canonical word order 4, 164
Attitudes 126–127
Case marker 104, 169 See also
Attributive adjective/AP 18, 53, 55, 65,
Case-marking particle
89, 103, 173, 176
Case-marker drop phenomenon 169
vs. predicate adjective/AP 53, 89
Case-marking particle 33, 147, 169, 173
AU See Accent unit
Casual conversation/spoken discourse
Auditory unit 2
51, 169–170, 173, 176
CD See Communicative dynamism
B Chafe, Wallace L. 1–3, 7, 9, 13–29, 36,
Backchannel xvi, 32, 87, 172–174 39–42, 46, 48–49, 54, 62, 83, 97, 99,
defined 172 102, 113, 120, 122, 124, 137,
in narrative 173 142–143, 151–152,159–160, 164, 166,
in spoken Japanese 32 170–172, 175–176
types xvi Chesterman, Andrew 18
See also Continuer; Reactive token; Chinese See Mandarin
Turn-internal listener Choice, speaker’s 1, 65, 165
backchannel Choice question 72
Index 

Clancy, Patricia M. 3–4, 28, 32, 63, 65, tight vs. loose internal structure of
74, 87, 97, 132, 159–160, 164, 174 33
Clarification request 71–72 types and definitions 105
Clark, Herbert H. 13, 18 vs. phrase 29–33, 51–53, 57–60, 98,
Clausal IU 3, 26, 29, 32–33, 37, 51–53, 172
55, 57–62, 64–69, 71, 80–81, 85, See also Adjectival predicate; Full;
87, 89, 98–100, 106, 121–122, Intransitive; Multi-IU; Nominal
137, 147, 154–157, 159–160, predicate; One-participant;
171–172, 175, 177 Semi-; Single-IU; Transitive;
defined 51 Two-participant clause
robustness of 160 Clause centrality proposal 3, 26, 29, 33,
types and distribution 52, 59 62, 99, 152, 159
vs. phrasal IU 3, 36–37, 57–60, 62, defined 25–26, 152
155–157, 172 Clause chaining 27
See also Full clausal; Independent Clause core 41, 122–123
clausal; Semi-clausal IU; Clause vs. phrase centrality 30, 51, 99
Single-IU clause Clause-external NP 103–104 See also
Clausal IU as part of multi-IU clause 52, Independent NP
56, 58–60, 64–66, 69, 89 Clause-external postposing 64, 84–85,
definition and subtypes 52 87–88, 165 See also Postposed NP
Clausal IU strategy, vs. phrasal 33 Clause-internal NP 81, 103, 106, 108,
Clausal object 52, 57, 104–105, 124, 110 See also Argument; Oblique
174, 177 Clause-internal postposing 64–65, 85,
Clausal structure 53–54, 57–58, 64, 88, 165
71–72, 98, 159, 174 Code 35, 44, 48, 51–54, 104, 142–144,
Clausal subject 104–105, 174 147, 155
Clause 2–3, 8–9, 12, 20, 25–27, 29–33, Coding 31, 46, 54, 62, 101–102, 121,
37–38, 41–42, 51, 53–54, 62–63, 123, 141, 143, 155, 160
71–72, 74, 79, 94–95, 97–99, Coding categories 48, 101
101, 105–106, 117–118, Coding scheme 32–33, 142, 144, 175
121–132, 134, 137–139, Cognitive activation, vs. search 8
141–142, 144, 152, 156–163, Cognitive component 142–143,
165–167, 172, 176 145–146, 148–152, 162
and IU in English 2–3, 25–27 Cognitive connective 143, 150
and IU in Japanese 3, 29–32, 94–98 Cognitive constraint/limitation 23, 36,
and NP 105–106 163–164
as basic unit of discourse production/ Cognitive cost 16, 138 See also
information flow/processing Activation cost
2–3, 27 Cognitive function 143–144, 151
as exponent of IU 26, 51, 63, 99, Cognitive planning See Planning
159, 163 Cognitive process/processing 15, 143,
breakup/division of 32, 37, 101, 148
139, 141, 156–158 Cognitive subtype 22, 49, 91, 143, 173
defined 53–54, 159 Coherence 7, 10, 33, 38, 143, 158, 162
English vs. Japanese 33 Cohesion 34–35, 142
 Index

Cohesive component 34–37, 141–142, Connective 27–28, 144, 148, 150–151


151, 154–155 See also Cognitive; Textual connective
Cohesive function 142 Consciousness 2, 13–15, 17, 21–26, 39,
Cohesive information 3, 36–37 81, 86, 100, 102, 164
Collaborative completion/finish 71, and IUs 21–22
76–77, 79, 88, 91, 93, 174–175 focal vs. peripheral 13, 15
Comment 83 See also Genuine topic focus of 2, 15, 21–22, 25, 39, 164
Common noun 18 See also Active consciousness
Communication 22, 32, 34, 36–37, Constituent 4, 24, 31, 33, 98, 103, 137,
57–58, 64, 71, 79, 81, 87–88, 156, 169, 173–174
158–160, 166, 170–171 See also Constraints 2–3, 22, 24–25, 38–39,
Conversational; Spoken 41–42, 102, 119, 121–123, 131,
communication 137, 139, 152, 160, 163–164, 177
Communication breakdown 174 on information flow 38–42
Communicative dynamism (CD) 25, on the flow rate of new information/
57, 171 new information quantity 2,
defined 171 39–42, 137
Communicative goal 11 PAS 41–42, 123, 161
Competitive referent 171 See also No more than two functional
Complement 176–177 components per IU; No more
Complete proposition 31, 52, 54–55, 57, than two IUs per clause; One
60, 66, 71, 74, 79, 81, 88, 92–94, 99, new NP per IU; One overt
106, 122, 154, 175 argument per clause; One overt
Complete propositional strategy 37 NP per IU constraint
Complex clause 174 See also Clausal Constructed example 169
object; subject; Embedded; Content See Ideational; Propositional
Subordinate clause content
Complex NP 70, 83, 93, 104, 108, Content, of thought 166
113–115, 131, 174, 177 See also Content words 41
Nominal complement; Relative Context 8, 13, 18, 22, 25, 28, 71–74, 76,
clause 79, 87, 92, 102, 151, 163–165,
Complexity, syntactic 177 169, 171, 173
Comprehension 32, 49 anaphoric vs. cataphoric 8
Comrie, Bernard 103 Contextual salience 18–19
Concept 2, 15, 18, 25, 39–41, 102–103, Continuer 87, 172
122–123, 137–139, 152, 162, Continuing intonation 20, 27–28, 46,
164–165, 170, 176 173
Conceptual unitariness 40–41, 102 Contrastive referent 9, 171
Conjunction 27–28, 49, 52, 141–144, Conversation See Casual; Dialogic,
148–149, 151, 173, 177 See also Dyadic; English; Everyday;
Coordinating; Subordinating Face-to-face; Female; Informal;
conjunction Interactive; Japanese; Male;
Conjunctive See Conjunction Multi-party; Naturally occurring;
Connected discourse 2, 5, 7, 27, 38–39, Ordinary; Spontaneous; Telephone
65 conversation
Index 

Conversation analysis 173 Croft, William 3, 19, 72, 88, 158, 167,
Conversational acts 76, 88 175, 177
Conversational communication 159, Cross-linguistic applicability/validity
161 29, 42, 88, 99, 159
Conversational co-participants See Cross-linguistic comparison 62, 144,
Co-participants 172, 176
Conversational discourse 48, 51, 77, Cruttenden, Alan 2, 20–21, 46
123, 126, 131, 139, 143–144, Crystal, David 2, 20
151, 158, 161, 163, 166–167
low transitivity/non-event/non-
action-oriented nature of D
126–127, 161 Data analysis 48
Conversational English 27, 62, 113, 172 Data base 44–45
Conversational interaction 5, 11, 30, Data coding 51, 101, 141
49–50, 124, 127, 155, 161, 164, 167 Data transcription 46
Conversational Japanese 5, 32, 43, 51, Declarative clause 46
57, 62–63, 65, 88, 94, 96–97, 101, Declination unit 2
106, 110–111, 113, 115–117, 119, Definite article 11, 18–19
122–124, 126, 128, 131, 133–134, Definite NP 7, 9, 17
138, 141, 152, 154, 158–163, 166 vs. indefinite NP 7, 17
Conversational language 49, 57, 87, 127, Definiteness 1, 9, 18 See also
172 Identifiability
Conversational monologue 27 Demonstrative adjective + NP 104,
Conversational narrative 40 113–115, 117
Conversational space 10, 118 Demonstrative pronoun 18, 104,
Cooperative principle 13 113–115, 117, 176
Coordinate sentence 177 Derived topic, vs. base-generated 174
Coordinating conjunction 28, 141 Descriptive function 34
Co-participants 18, 22, 30, 32, 48–49, Detached NP IU 56, 69, 71, 88, 90,
54, 61, 69, 73–76, 79, 88, 92, 99, 102, 159–160, 175 See also Independent
106, 117, 137, 139, 143, 152, 158, phrasal NP IU
163, 166, 169, 173 Di Cristo, Albert 29, 46
Copied postposed NP 55, 72, 85, 88 Dialogic conversation/interaction 5, 157
Copula 41, 62, 105, 173 See also Dik, Simon C. 57
Adjectival predicate; Linking verb; Direct object 9, 31, 55, 103, 105, 123,
Nominal predicate 126, 147–148, 154, 169, 176
Copula + PP/NP construction 41 vs. indirect object 9
Core argument 41, 52–54, 103, Direct object marker 169
120–121, 171, 176 Disagreement 71–72
Core grammatical role 103, 113 See Discourse See Conversational; Spoken
also A; O; S role discourse
Core-oblique distinction 113, 134 Discourse analysis 167
Corpus 41, 67, 88, 105, 134 Discourse context See Context
Couper-Kuhlen, Elizabeth 3, 19 Discourse functions 65, 160, 175
Courtroom discourse 166 Discourse marker 22
 Index

Discourse production 2, 5, 19, 32, 39, English clause 33, 177


48, 81, 88, 138, 148, 164 English conversation 10, 29, 32, 113,
Discourse unit 2, 31, 42, 54, 138, 159, 172, 176
163, 166 See also Clause; English IU 3, 29–30
Intonation unit English speakers 25, 28–29, 33, 37, 157,
Discourse-functional studies 3, 169 160
Discourse-new, hearer-new information English spoken discourse 25
14 See also Brand-new information Entity 14, 17–18, 33, 102, 170, 174–175
Discourse-new, hearer-old information Episodic memory 10, 38
14 See also Unused information Epistemic stance See Stance
Discourse-old, hearer-old information Evaluation 49, 126
14 See also Evoked information Evaluative subtype See Validational
Discourse-pragmatic factors 72, 88 subtype
Displaced mode 49 Event 3, 11–12, 22, 26–27, 41, 49, 54,
Distance 8, 74, 177 See also 102, 126–127, 143, 152, 159, 171, 175
Referential distance Event idea 26, 175
Ditransitive clause/verb 176 Everyday conversation 127, 165, 172
Dixon, Robert M. W. 54, 103 Evoked information 14, 102 See also
Doctor-patient interaction 166 Given information
Dorgeloh, Heidrun 169 Experiential function 33
Downing, Pamela 169 Explicit NP/subject See Overt NP/
Drew, Paul 166 subject
Du Bois, John W. 1–3, 18–21, 25, 41–42,
Expressive function 34
46, 48, 54, 102–103, 111–113,
Extra-clausal NP See Clause-external
116–118, 122–124, 130–131,
NP
133–134, 137–138, 152, 160–161,
171, 175–176
Duranti, Alessandro 169 F
Durie, Mark 71, 88 Face-to-face conversation 35, 44
Dyad 44–45, 47–50, 174 Falling intonation 20, 27, 46
Dyadic conversation 44, 46, 166 False start 49, 78, 80
Feedback 32, 157
E Feelings 126–127
Edwards, Jane A. 46 Female conversation, vs. male 44–45,
Efficiency 167 47–48, 166, 172
Elaborating NP 175 Figure-ground distinction 170
Elicited narrative 41, 172 Filled pause See Pause filler
Embedded clause 20, 121, 124, 177 Final/IU-final particle 49, 70, 150–151
Emotions 171 Finality, vs. non-finality 46, 171
Emphasis 65, 85, 88 Finnish 97
Encoding 27, 166 Firbas, Jan 25, 57, 171
English 3, 12, 18, 20, 23, 29–33, 37, 42, First person 170, 176
46, 62, 112, 116–117, 120, 137, 148, Floor, of conversation 34, 143, 174
171–173 See also Conversational; Flow, of conversation 34, 65, 143
Spoken English Ford, Cecilia E. 19, 161
Index 

Foreground 10–12, 170 obligatory vs. optional 150–151,


defined 10–11 162
Foreground-background alternation 11, preferred number of 141, 149–150,
26, 65 155, 162
Foreground-background distinction/ types and definitions 142–143
continuum 10–12, 170 See also Cognitive; Ideational;
Foregrounded clause, vs. backgrounded Interactional; Textual
12 component
Foregrounding, vs. backgrounding 12, Functional linguistics 127
170 Functional sentence perspective 171
Foregrounding device, vs. backgrounding Functional structure, IU 3, 5–6, 35, 43,
11–12 48, 141–142, 151, 158–159, 163,
Fox, Barbara A. 1, 10, 13, 33, 65, 169 177
Fragmentary IU 22, 24, 49–50, 78 orderly nature of 3, 35, 151, 162
defined 22, 49 Functional structure types, IU 36,
Fragmentation 29–30, 32, 96, 160, 172 144–149, 151–155, 158, 162
in English vs. Japanese 29–30, distribution of 144–146
32–33, 172
in Japanese and Mandarin 171–172
in spoken Japanese 32 G
Fragmentedness 30 Gee, James P. 166
Fragments, discourse 88, 159 Geluykens, Ronald 71
Free clause 11 Gender differences 166
French 42, 170 Generic referent 120, 171
Frequency 4, 8, 37, 158 Genitive case 104, 173
Fujii, Noriko 169 Genre 11–12, 30
Full clausal IU 53, 58, 80–81, 85, Genuine topic 54, 57, 72, 81, 83, 100,
99–100, 110, 122, 159 174–175
independent 58, 110 defined 174
vs. semi-clausal IU 66–67, 94 Gesture 28, 163–164
Full clause 29, 31, 52, 54–56, 59, 62, Given A Constraint 41, 113, 116, 133,
66–67, 69–70, 78, 85, 89, 91, 161
94–97, 99–100, 106, 108, 122, Given followed by new 12, 25 See also
136, 171, 175 Unmarked information structure
defined 52, 54 Given information/NP 10, 13–17, 19,
vs. semi-clause 62, 66–67, 94–95 24–25, 38–40, 43, 79, 81, 87,
Full NP, vs. pronoun 9, 17, 117, 171 100–103, 111–113, 115–120,
Full proposition 31, 35, 56–57, 74, 77, 123–124, 131, 133–134, 139,
142, 147–148, 155, 174–175 160–163, 165, 171, 177
Function words 41 defined 12–14, 102
Functional components 5, 34–36, constraint on the flow rate of 38–39
141–142, 144–145, 147, Given-accessible-new distinction 9,
149–152, 155, 158, 162–164 15–16, 41, 171 See also Activation
defined 34 cost
linear order of 5, 35, 141, 150–152 Givenness 9, 12–15, 170
 Index

ways of establishing 170 Helasvuo, Marja-Liisa 3, 88, 97


See also Accessibility; Newness Heritage, John 46, 166
Given-new contract 13 Hesitation 20
Given-new distinction 12–14, 144 See High focus, vs. low 170
also Predictability; Recoverability; High transitive clause 105, 125–129,
Saliency; Shared knowledge 132–133
Givón, Talmy 1–2, 7–11, 13, 18, 27, Hinds, John 4, 54, 63, 165–166, 169
38–39, 41–42, 54, 117, 122, 137, 165, Hirschberg, Julia 19, 47
170, 175 Hirst, Daniel 29, 46
Grammar 1, 8, 11, 127, 161 Hoji, Hajime 3
Grammatical categories 53, 58, 61, 98, Hopper, Paul J. 1, 3, 10–12, 42, 105,
175 125–127, 161, 170, 176
Grammatical relation 103 Horie, Kaoru 113
Grammatical role 101, 103–104, Human agent/subject 134, 170
109–117, 138–139, 164, 176 Hypotaxis 33
and information status 111–113 Hypothesis 38, 43, 51, 101, 141,
and NP syntactic form 113–115 162–163
types and definitions 103–104
See also Argument; Independent NP;
I
Oblique; Predicate nominal
Iconicity 170
Green, Georgia M. 169
Idea 16–17, 19, 22, 24–29, 41–42, 49, 57,
Grice, H. Paul 13
122, 138, 142–143, 150, 158,
Grimes, Joseph E. 2
164, 170, 175
Grounding 1, 10–12, 38
subcategorization of 41
and foreground-background
See also Event; Referent; State
distinction 11–12
Idea unit 171
and given-new information 10–11
Idea-conveying IU 36, 146, 163, 166
and transitivity 12
See also Substantive IU
defined 10–11
Ideation 36, 155
See also Background; Foreground
Ideational component 34–36, 43,
141–143, 145–152, 154–155, 158,
H 162–163, 177
Haiman, John 27 Ideational content 31, 50, 57, 148, 154,
Halliday, Michael A. K. 1–2, 12–13, 156, 158, 161, 163, 172
17–19, 24–27, 33–34, 48, 120, Ideational function 33, 141–142, 144
141–142, 165, 172 Ideational information 3, 22, 30, 32,
Hasan, Ruqaiya 13, 34 36–37, 151–152, 155–157, 162
Haviland, Susan E. 13 Identifiability 9, 14, 18–19, 102, 171
Hawkins, John A. 1, 18 and activation cost 19
Hayashi, Makoto 77 defined 18
Hearer, and/vs. speaker 2, 10–15, 22, 46, properties of 18–19
73, 79, 86, 88, 93, 102, 143, 154, 165, See also Contextual salience;
170, 174–175 See also Listener Definiteness; Sharedness;
Heavy added information constraint 25 Sufficiently identifying
Hebrew 42 verbalization
Index 

Identifiable referent 18–19, 81, 100, 102 Information flow 1–3, 5, 7, 19–20, 22,
vs. non-identifiable referent 19 25, 38–39, 49, 88, 120, 160,
Immediate mode 49 164–165
Inactive state 2, 15–17, 39, 42, 138 basic/canonical/unmarked pattern of
25, 120
Indefinite article 11
constraints on 38–42
Indefinite NP 7–8, 17
defined 1
Independent adverbial IU 89–92, 100
units of 1–3, 5, 19–20, 38, 164
types and distribution 90 Information focus 25
Independent clausal IU 51–52, 55, Information pressure 117–118
58–60, 62, 66, 89, 98–99, 106, Information pressure quotient (IPQ)
110–111, 137, 154, 174–175 117
definition and subtypes 52 Information processing 2, 27, 39, 164
vs. clausal IU as part of multi-IU Information status 1, 12–18, 81, 88,
clause 58, 66, 89 100–102, 104, 111–112,
vs. independent phrasal IU 54–55, 115–117, 139, 160–161,
58 164–165, 175
Independent NP 54, 71–72, 79, 81, 83, and grammatical role 111–113
103–104, 108–114, 117, 124, 175 and NP syntactic form 115
See also Clause-external NP; as property of nominal reference
Non-argument 102
Independent phrasal IU 51, 53, 55, See also Accessible; Given; New
58–60, 65, 71, 89–90, 93, 98, information
103–104, 113, 173 Information/informational structure, IU
definition and subtypes 53 3, 5–6, 23–25, 43, 48, 81, 97,
vs. phrasal IU as part of multi-IU 101, 118–119, 139, 144, 152,
clause 52–53, 58, 65–66, 89 159–160, 163
Independent phrasal NP IU 54, 70–88, in English 23–25
111, 159 in Japanese 118–120
See also Unmarked information
functions of 88
structure
types, definitions and distribution
Information structure types, IU
71–72
118–120, 139
See also Detached NP IU; Lead;
distribution of 118–121
Left-dislocated; Postposed;
Information unit 2, 25–26, 172
Repeated; Right-dislocated;
Informational factor, and multi-IU clause
Stray; Topic NP
158 See also One new NP per IU
Independent semi-clausal IU 58, 60–61, constraint
67, 98, 106, 110–111, 123, 162 Informational hybrid 38
Indirect object 9, 176 Information-flow factors 1, 72, 88
Inferrable information 14–15, 18 Infrastructure, of thought 166
Informal conversation 26, 79, 88, 109, Initial pause 15, 20, 75
160, 176 Insertion/inserted sequence 74, 78
Informal form 54, 70, 177 Institutional talk 166
Information block 2 Intensifier 31, 118
 Index

Intention, speaker’s 21, 34, 65 Intonation unit (IU) xv, 1–3, 5, 15,
Interactants 11, 18, 37, 47, 67, 83, 86, 20–23, 25–28, 33, 37, 43, 47–49,
90, 161, 166 51, 57, 60, 62, 66, 68, 71, 89, 94,
Interaction 19, 22, 32, 37, 77, 79, 85, 87, 101–102, 105–106, 117–118,
127, 143 See also Conversational 124, 134, 137–138, 141, 144,
interaction 146, 152, 156, 159–161,
Interactional component 34–36, 43, 163–164, 167, 171, 173
142–143, 145–148, 150–152, and AU 23–24
154–158, 162–163 and clause 2–3, 25–27, 29–32, 62,
Interactional factors 72, 88 94–98
Interactional function 143–144, 162 and consciousness/language
Interactional information 3, 32, 36–37, production 21–22
151–152, 155–158 and NP 105–106
Interactional involvement, as basic unit of discourse production/
co-participants’ 76, 88 See also information flow 1, 5, 19, 164
Collaborative completion; defined 20
Interruption; Repair identification criteria 20–21
Interactional motivation, and multi-IU linkage 25–28, 164
clause 158, 166
prosodic features of 21
Interactional NP 175
singleness/unitariness of 144, 164
Interactional particle 31, 33, 35, 37, 75,
types and size of 22–23
143, 147, 150, 155, 157, 166, 177
See also Functional; Informational;
as marker of IU boundary 150
Syntactic structure
See also Final particle
Intonational phrase 20
Interactional subtype 22, 49, 173
Interactive conversation 29, 48, 79, 117, Intra-IU clause 57, 104–105, 124, 174,
157, 161 177
Interlocutor 1, 10, 14, 49, 73, 87, 91, Intransitive clause 105, 121, 124–129,
157–158, 167, 172 131–133, 161, 176
Interpersonal function 33–34, 147, 158 Intransitive subject 41, 103, 132 See
Interrogativity 46 also S role
Interruption 49, 77, 88 See also Intransitive verb 103, 124
Other-; Self-interruption Inversion 169
Intonation 4, 27–28, 46, 165, 167, 171 IPQ See Information pressure quotient
functions of 46, 165, 171 Irrealis 171
See also Continuing; Falling; Rise-fall; IU See Intonation unit
Rise-fall-rise; Rising intonation IUs with zero NPs 118–121
Intonation contour xvi, 2, 20–21, 28, 46, types of 118, 120–121
75, 169, 171, 173 See also No associated NPs; Zero
acoustic measurement of 169, 173 anaphora
as signal of IU boundary 20, 46 Iwasaki, Shoichi 3, 28–32, 34–37, 42, 46,
coherent/unified 2, 20–21, 46 48, 62, 67, 97, 99, 116, 127, 130,
types xvi, 46 141–144, 150–151, 154–155,
Intonation group 2, 20 159–160, 162, 171–172
Index 

J Language production 21, 42


Japanese 3–4, 9, 17, 29–32, 34, 37, 46, Language universal 113
62, 100, 116–117, 120, 130, 134, Latching xvi
137, 142–144, 148, 152, 157, LDed NP See Left-dislocated NP
169–173 Lead component 34–35, 142–143, 151
as SOV/verb-final language 130 Lead NP 71–72, 78, 80–81, 87–88, 92,
Japanese clause 33, 105, 161, 166 159, 175, 177
Japanese conversation 5, 29, 44, 48, 69, definition and types 71
96, 130, 134, 144, 152, 158–159, 161, Left-dislocated (LDed) NP 7, 9, 31,
166, 169, 176 71–72, 81–83, 159, 175
Japanese intonation unit (IU) 3–5, 28, Lengthening 21, 173
30, 33–37, 42, 48, 51, 60, Lerner, Gene H. 74
99–100, 118, 142, 144, 146, 149, Lexical argument 42, 130–131, 161
151, 153, 155, 158–160, 163 Lexical NP 114–116, 131, 134, 176
exponent of 51, 63, 99, 159, vs. non-lexical/pronominal NP/
162–163 pronoun 114–116, 131, 134
multifunctionality of 33–38 Lexicalized phrase 41
simple/non-elaborate nature of Light starting point constraint 25
160, 162–163 Light subject constraint 25, 113
unifunctional nature of 152–155, Lightness 25
158 Linkage, IU 2, 22, 25–29, 49, 143, 150,
Japanese speakers 4–5, 32–33, 37, 51, 164
81, 97–98, 109, 115, 118, 124, 128, as clause linkage 2, 26, 29
139, 149, 152, 155, 157, 159–163 connective vs. non-connective type
Japanese spoken discourse 3–4, 19, 35, 27–28
42, 102, 129 non-elaborate nature of 28, 164
Jones, L. B. 11 Linking, IU See Linkage
Jones, L. K. 11 Linking device, IU/idea 28, 92, 155, 158,
Judgment 22, 34, 49, 173–174 162
elaborate vs. non-elaborate 28
Linking verb 62 See also Copula
K
Listener, and/vs. speaker 10, 13–19,
Kärkkäinen, Elise 3, 42, 123
21–22, 32, 35, 42, 46, 86, 157, 172,
Kinesis 105, 170
174 See also Hearer
Kreckel, Marga 20
Listener backchannel See Backchannel
Kuno, Susumu 3–4, 13, 65, 130, 165, 169
Listener response 71–72, 87, 91
Listener-oriented perspective, on
L activation cost 16–17
Labov, William 11 Listener-orientedness 14
Ladd, D. Robert 19, 46 Logical function 33
Lambrecht, Knud 1, 161, 175 Longacre, Robert E. 11
Lampert, Martin D. 46 Loose internal structure, of Japanese
Language 1, 17, 21, 33–34, 127, 164 clause 33, 166
See also Conversational; Spoken Loud talk xv
language Loudness 173
 Index

Low transitive clause 105, 125–129, and syntactic factors 158, 177
131–133 full vs. semi- 56–57, 94–97
marked production of 139, 158,
160–161, 164, 166
M
vs. single-IU clause 94–98,
Main clause 10, 52, 57, 121, 170, 177
124–125, 128, 134–135, 138, 160
vs. subordinate clause 52, 57, 121
Multi-IU/unit turn 83, 172
Makino, Seiichi 177
Multi-party conversation 166
Male conversation, vs. female 44–45,
Multiple IUs, one clause strategy 97
47–48, 166, 172
Mandarin 9, 17, 81, 88, 105, 171–172
Marked word order 4, 54, 63, 169 N
vs. basic/unmarked word order 4,
Narrative 3, 11–12, 30, 32, 35, 40–41,
54, 63
48, 88, 111–113, 117–118, 127,
See also Postposing; Scrambling
172–173, 175–176
Marshall, Catherine R. 18
vs. conversation 30, 173–174
Matsumoto, Kazuko xi, 4, 51, 63,
See also Conversational narrative
101–102, 141, 159–163, 165–167,
Narrative skeleton 11
169–170, 172
Native speakers 1, 4, 29, 44
Maynard, Senko K. 3, 28, 30, 32, 65, 97,
Natural vs. elicited discourse 30
127, 155, 159–160, 172
Naturally occurring conversation/
Metafunction 33–34 See also
discourse 4–5, 17, 130, 169
Cohesive; Descriptive; Experiential;
Expressive; Ideational; Interactional; Nespor, Marina 20
Interpersonal; Logical; Subjective; New information/NP 2, 10, 12–17, 19,
Textual function 24–25, 38–43, 88, 100–103,
Mithun, Marianne 169, 176 111–113, 115–121, 123–124,
Mixed phrasal IU 56, 89, 93 131, 133–139, 152, 157–158,
Mixed phrase 53, 63, 174 160, 165, 170–171, 174–177
Modifier 18, 89, 91, 114 and cognitive cost 16, 138
Mori, Junko 19 and IU/clause 134–138
Multifunctionality 36, 144, 151, 155 constraint on the flow rate/quantity
Multi-IU clause 51–58, 60, 64–66, 69, of 39–42, 137
71, 73, 85, 89, 94, 96–98, 101, defined 12–14, 103
108, 124–125, 128, 134–139, integration of 38–39
141, 147, 149, 154–158, See also Brand-new; Unused
160–161, 164–166, 174–175 information
and coding of interactional Newness 9, 12–15
information 155–158 News interview 166
and Japanese vs. English syntax 33, No associated NPs 120–121, 136
157–158 No more than two functional components
and multifunctionality 155–158 per IU constraint 149, 152, 163
and one new NP per IU constraint No more than/up to two functions per IU
134–138, 157, 161–162, 166 constraint 36, 155
and post-predicate phrase 65, No more than two IUs per clause
164–165, 175 constraint 95, 163
Index 

Nominal 5, 39, 63, 102–105, 111, 116, Non-newness 25


118, 123, 144, 161–162, 164–165, 174 Non-postposed phrase See
See also Noun; Noun phrase Pre-predicate phrase
Nominal complement clause 104, 174 Non-referential NP 104, 171
Nominal information 43, 100, 117–119, Non-referential zero, vs. referential 124
152, 157, 161, 163 Non-specific referent 171
Nominal phrase See Noun phrase Non-substantive IU 22, 48, 154, 165
Nominal predicate 39, 52, 55, 58, 67, See also Fragmentary; Regulatory IU
69–70, 99–100, 103, 105, 112, Non-tracking NP 113
124–125, 127, 131–132, 147, 151, Non-transitive clause 125–128, 130,
161, 173, 177 139, 161 See also Adjectival
Nominal predicate clause 62, 99, 105, predicate; Intransitive; Nominal
121, 124–129, 131–133, 147, 151, 161 predicate; One-participant clause
Nominal reference 5, 25, 102, 137, 160 Non-verbal predicate 103 See also
Nominal referent See Referent Adjectival; Nominal predicate
Nominalized VP 104, 113–115 Noonan, Michael 169
Nominalizer 35, 104, 143, 154–155 Notational conventions 46
Nominative case 147 Noun 9–10, 17–20, 38, 41, 53, 104, 108,
Non-action 126 114–116, 139, 160, 171 See also
Non-anaphoric zero 9, 120, 124 Bare; Common noun; Proper name
defined 124 Noun phrase (NP) 7, 10, 17–20, 31,
vs. anaphoric zero 124 52–53, 55–56, 58, 65, 69–72, 81, 91,
Non-argument 5, 103, 108–111, 100, 103–121, 123–124, 139,
116–118, 123–124, 138–139, 159–161, 163–164, 169, 171, 173,
160, 176 175–176
defined 103 and IU/clause 105–106
types and definitions 103–104 argument vs. predicate 53, 56,
See also Independent NP; Oblique; 69–70
Predicate nominal clause-internal vs. clause-external
Non-clausal IU 3, 29–31, 62, 161, 167, 103
171 See also Phrasal IU definite vs. indefinite 7
Non-event 126 lexical vs. pronominal 114–116,
Non-finite clause 174 131, 134
Non-finite predicate/-te form 35, 143, See also Argument; Grammatical
154 role; Information status;
Non-ideational information 37, Non-argument; Preferred NP
155–156, 158 See also Cohesive; types; Syntactic form
Interactional information NP See Noun phrase
Non-identifiable referent 19, 113 NP IU 29–31, 53–56, 58, 60, 62, 68–76,
Non-lexical A Constraint 42, 116, 78–81, 83, 85–91, 96–98, 104,
133–134, 161 111, 147, 154, 159, 171, 174–175
Non-lexical NP 116, 134 See also types and distribution 52–53,
Pronoun 68–69
Non-new information 102, 134 See See also Independent phrasal NP IU
also Accessible; Given information NP-no + NP 104, 108, 113–115, 117
 Index

Null argument 9, 118, 124, 132–133 and post-predicate new NP 165


See also Overt; Zero argument See also Two new NPs per IU
Null argument language 9 One overt argument per clause constraint
Null argument strategy 118 128, 131, 139, 161, 163
One overt NP per IU constraint 120,
123, 127, 139, 160, 163
O
One unit per proposition constraint 2,
O See O role
39, 122
O role 4, 30, 41–42, 56, 70, 75–76, 82,
One-chunk-per-clause processing
85–87, 103–105, 108–114, 117,
principle 39, 122
122–125, 127–134, 136, 139,
One-participant clause 125–127, 130,
156–157, 161, 163, 165, 169–170, 176
161
See also Transitive object
vs. two-participant clause 125–126
Object 4, 9, 38, 41, 53–54, 69, 97, 110,
See also Non-transitive clause
169–170 See also Direct; Indirect;
Ono, Tsuyoshi 3–4, 30–31, 64, 72, 74,
Transitive object
169
Oblique 103, 108–117, 122–124,
Operational definition 12, 101–102
136–139, 160, 176 See also
Opinions 46, 49, 126–127
Argument; Core-oblique distinction
Oral discourse/communication See
Observed example 54, 104, 147
Spoken discourse/communication
Ochs, Elinor 3, 164, 169
Ordinary conversation 127, 166
O’Dowd, Elizabeth 42
and subjectivity 127
OF See Overall frequency
vs. institutional talk 166
Old information 13–14, 38–39 See
O-role argument See O role
also Given information
Other-interruption 71, 73–74, 76, 79, 90
One clause/function/new NP/overt
Overall frequency (OF) 8, 165
argument at a time 155, 158
Overlapping xvi, 76
One clause at a time constraint 2,
Overt argument 124–125, 127–133, 139,
26–27, 36, 39
152, 155, 161, 163–165
One IU, one clause strategy 97, 160
vs. null argument 132–134
See also Multiple-IUs, one clause
Overt NP 104–107, 109, 111, 116–117,
strategy; Single-IU clause
119–121, 123, 127, 131–132, 134,
One Lexical Argument Constraint 42,
137, 139, 160–161, 163, 165
131, 161
Overt subject 29, 52, 55, 58, 60, 67,
One New Argument Constraint 2, 41,
98–99, 105–106, 122, 159
122–124, 138, 152, 160–161
One new concept at a time constraint 2,
36, 39–40, 122, 124, 152, 160 P
One new idea constraint 39, 122, 124, Parallelism 177
152, 160 Parataxis 33
One new NP per IU constraint 81, 88, Partial proposition 55, 73–74
121–124, 134, 136–139, 157, Partial propositional strategy 36–37
160–161, 163, 165–166, 177 Particle 31–35, 37, 49, 70, 75, 143–144,
and IU/clause 138 147, 150–151, 155, 157, 166, 169,
and multi-IU clause 134–138, 157, 173, 176–177 See also
161–162, 166 Case-marking; Interactional particle
Index 

PAS See Preferred Argument Structure vs. clause 29–33, 51–53, 57–60, 98,
Passive clause/verb 176 172
Patient 103, 105, 170 Phrase-orientedness 33
Pattern 1, 5, 25, 33, 36, 41–42, 63, PI See Potential interference
87–88, 96, 115, 118, 130, 132, 147, Pierrehumbert, Janet B. 19, 28, 46
159, 161, 164–165 Pitch 21, 32, 46, 171, 173
Pause xvi, 2, 16, 20–21, 34–35, 46, 49, Planning 32, 34, 91, 142, 148, 150
75, 86, 142–143, 149 Point of view 127 See also Subjectivity
types xvi Polysynthetic language 23
See also Initial pause Possessor 18
Pause filler 34–35, 49, 143, 149 Postposed NP, clause-external 71–72,
Pause-bounded phrasal unit (PPU) 32 84–88, 159, 165, 174 See also
Pawley, Andrew 2, 26–27, 36, 39, 41 Copied; Restated postposed NP
Payne, Doris L. 169 Postposed phrase, clause-internal/external
Pear Film/Story 32, 41, 172, 175–176 See Post-predicate phrase
See also Elicited narrative
Postposing 4, 54, 63–66, 87, 102,
Perception 19, 127, 173 164–165, 169–170, 175, 177
Performance error 80–81, 88, 159
and backgrounding/defocusing 65,
Personal pronoun 104, 108, 113–115, 85, 165
176
and intonation 4
Phonological prominence 17
and IU 63–66
Phrasal IU 3, 29–33, 36–37, 51, 53–56,
and multi-IU clause 65, 164–165,
58–62, 64–65, 68–69, 71, 89, 93,
175
96, 98, 101, 103–104, 113,
defined 4
121–122, 141, 147, 154–158,
161, 164, 166–167, 172–175, 177 functions of 65, 85, 88, 165
as element/part of multi-IU clause types of 64–65
52–53, 64–65, 69, 89, 164 See also Clause-external;
defined 51 Clause-internal postposing
marked use of 161 Postposition 176 See also
types and distribution 52–53, 61, Case-marking particle
89 Post-predicate element/phrase 4, 31,
vs. clausal IU 57–60, 155–157, 172 53–54, 56–57, 63–65, 70, 97,
See also Independent phrasal IU 129, 136, 154, 164–165,
Phrasal IU as part of multi-IU clause 169–170, 174–175, 177
52–53, 56, 58–60, 64–65, 69, 89, vs. pre-predicate element/phrase 4,
164 54, 63, 164–165, 169
definition and subtypes 52 See also Marked word order
Phrasal IU option, vs. clausal 33, 158 Postverbal element See Post-predicate
Phrasality, marked case of 155 element
Phrase 10, 17–20, 30–31, 51–58, 61, Potential interference (PI) 8, 165
63–65, 74, 97–100, 103, 106, PP See Prepositional phrase
121, 123, 142, 149, 159, 164, PPU See Pause-bounded phrasal unit
172, 174–176 Predicate 12, 25–26, 31–33, 37, 39, 41,
types 52–53 51–54, 69, 74–80, 89, 92, 99,
 Index

104, 106, 125, 128–130, 159, vs. secondary accent 18


161, 173 Primary function, of linguistic element
types of 52 144
See also Adjectival; Nominal; Verbal Prince, Ellen F. 1, 12, 14–15, 170
predicate Production, IU 5, 15, 33, 47–48, 81, 87,
Predicate/predicative adjective/AP 53, 97, 102, 117, 139, 155, 163, 166 See
89, 105 also Discourse; Language; Speech
Predicate nominal 70, 103–105, production
108–117, 124, 131, 138–139, 160, 171 Pronominal NP See Pronoun
See also Nominal predicate Pronoun 7, 9, 17–18, 65, 76, 83, 86, 91,
Predictability 12–13 104, 108, 112, 114–117, 131,
Preferred Argument Structure (PAS) 42, 133–134, 165, 171, 175–176
123, 161 stressed vs. unstressed 7, 9
defined 41–42 vs. full NP 9, 17, 117
quantity and role constraints of 42 vs. lexical NP 114–116, 131, 134
See also Given A; Non-lexical A; One See also Demonstrative; Personal
Lexical Argument; One New pronoun
Argument Constraint Proper name 18, 72, 170
Preferred argument structure 124 Proposition 3, 11, 27, 33–34, 36–39,
Preferred clause structure 101, 124, 57–58, 64–65, 71, 73–74, 79, 81, 85,
128–130, 161 87–88, 137, 141–142, 144, 147, 150,
Preferred clause types 101, 124–125, 152, 154, 158–159, 164, 172, 174, 177
128, 139 See also Complete; Full; Partial
Preferred functional structure, IU 5, 43, proposition
141, 146 Propositional content 22, 32, 35, 53, 57,
Preferred information structure, IU 5, 87, 117, 139, 141–142, 154, 160, 167
42–43, 101, 118, 123–124, 139 Propositional strategy 36–37 See also
Preferred nominal structure 5, 161 Complete; Partial propositional
Preferred NP types 101, 109, 116, 139, strategy
160, 164 See also Grammatical role; Prosodic unit 3, 19–20, 171
Information status; Syntactic form Prosody 19, 28, 158, 164
Preferred structures, IU 5, 163–164 Purpose, of communication/interaction
Preferred syntactic structure, IU 5, 43, 11, 30, 166
51, 60–61
Preposing 169
Q
Preposition 25, 41
Quantitative analysis 48
Prepositional phrase (PP) 30, 41
Question 70, 72, 76, 79 See also
Pre-predicate element/phrase 4, 54, 63, Choice; Wh-; Yes/no question
164–165, 169 See also Basic word
Question words 171
order
Presentative NP 175
Presupposition 11 R
Preverbal element See Pre-predicate RD See Referential distance
element RDed NP/phrase See Right-dislocated
Primary accent 18, 23–25, 171 NP/phrase
Index 

Reactive token 87, 174 defined 22, 49


Realis 12 subtypes and definitions 22, 49
Recipient 173 See also Hearer; Listener vs. substantive IU 22–23, 49–50,
Recoverability 12–13, 79 165–166, 173, 175
Redundancy 117 See also Cognitive; Interactional;
Reference 5, 8, 10–11, 18, 25, 102, 137, Textual; Validational subtype
160, 170 See also Nominal; Verbal Reid, Wallis 170
reference Reinforcement 65, 85
Referent 7–11, 13–14, 18–19, 22, 24–26, Reinhart, Tanya 170
31, 41, 49, 81, 83, 85–86, Relative clause 10, 18, 20, 70, 93, 104,
102–104, 124, 131, 133, 174
138–139, 143, 158, 163, Repair 65, 71, 73–75, 79, 85, 88 See
170–171, 175 also Self-repair
and mental representation of 175 Repeated NP 71–72, 86–88, 91, 159, 175
defined 175 Repetition 86, 88, 174–175
given vs. accessible vs. new 10, 13, Research question (RQ) 5, 43, 51, 101,
25, 102–103 141
identifiable vs. non-identifiable Reset 21 See also Baseline pitch level
18–19 Restated postposed NP 72, 85
vs. event/state 26, 41 Resumptive pronoun 65, 83, 85, 175
Referent-activating NP 81 See also Left-; Right-dislocated NP
Referent-anchoring NP 175 Right-dislocated (RDed) NP 7, 71–72,
Referential accessibility 8 84–86, 159
aspects of 8 Right-dislocated (RDed) phrase 53–54,
See also Referential complexity; 65
continuity Right-dislocation 65
Referential complexity 8, 165 Rise-fall intonation 46, 75
measures and definitions 8 Rise-fall-rise intonation 46, 70
See also PI; SR Rising intonation 35, 46, 70
Referential continuity 8, 165 Roberts, Celia 166
RQ See Research question
measure and definition 8
See also RD
Referential distance (RD) 8–9, 165, 175 S
Referential interference 8 S See S role
Referential NP 175 S role 41, 103–105, 108–117, 121–122,
Referential zero, vs. non-referential 124 124–125, 127–134, 136, 138–139,
Referent-listing NP 175 160–161, 163, 165, 175–176 See
Referring expression, definite vs. also Intransitive subject
indefinite 17 See also Noun Sacks, Harvey 173
phrase Saeki, Tetsuo 65
Regulatory function 22, 49, 158, 162, Saito, Mamoru 3
173 Sakapulteko 41, 112–113, 118, 130, 175
Regulatory IU 22–24, 35, 49–50, 70, 76, Saliency 12–13
83, 91, 142–143, 165–166, 173, S-A-O categories 176
175 Sarangi, Srikant 166
 Index

Schank, Roger C. 170 Sharedness 14, 18–19 See also


Schegloff, Emanuel A. 19, 65, 87, Identifiability
172–173 Shibatani, Masayoshi 3–4, 54, 64, 130,
Scheibman, Joanne 3, 127 174
Schema 102, 170, 176 Silva-Corvalán, Carmen 169
defined 170 Simultaneous talk See Overlapping
Schiffrin, Deborah 22 Single-IU clause 54–55, 64–66, 71,
Schuetze-Coburn, Stephan 2–3, 19, 21, 87–88, 94, 96–99, 106, 108, 111,
176 122, 124–125, 128, 135,
Scrambling 4, 169, 174 137–138, 160, 174
Second person 170, 176 full vs. semi- 94–95
Secondary accent 18 vs. multi-IU clause 94–98,
Segmentation, of spoken discourse 2, 124–125, 128, 134–135, 138, 160
19, 27, 171–172 See also Prosodic See also Clausal IU; Independent
unit clausal IU
Segmentative/delimitative function, of Singleness 164
intonation 171 Smith, Wendy 3, 42
Self-interruption xv Softening device 177
Self-repair 71, 73–75, 79, 90 Sound stretching xv See also
Selkirk, Elizabeth O. 20 Lengthening
Selting, Margaret 3, 19 SOV language/order See
Semi-active state 2, 15–17, 39, 81, 100, Subject-object-verb
102, 170, 176 Spanish 42
Semi-clausal IU 43, 57, 59–60, 67, 70, Speaker-hearer interactional information
85, 88, 96–98, 100, 112, 122, See Interactional information
137, 159, 162, 171 Speaker-hearer relationship 30, 143, 171
vs. full clausal IU 66–67, 94 Speaker-oriented perspective, on
See also Independent semi-clausal IU activation cost 16–17
Semi-clause 29, 31, 35, 37, 52, 54–55, Speakers of English See English speakers
57, 59, 62, 66–67, 69, 89, 92–97, Speakers of Japanese See Japanese
99, 112, 122, 136, 159, 163, speakers
171–172 Speaker-selectedness 14
defined 52, 54 Speakership 154
vs. full clause 62, 94–95 Specification, further 65, 85, 88
Seneca 23 Speech 1, 19–20, 35, 143, 155, 173, 175
Sensitivity, toward addressee 34–35, 143 Speech acts 33–34
Sentence 4, 13, 19, 25, 39, 130, 177 Speech production 19–20
Sentential adverb 89–92, 175 Spoken communication 1, 28, 32, 39,
Sequence 2, 11, 20–21, 26–27, 40, 46, 144, 158, 163
50, 55–57, 70, 74, 78, 83, 91, 93, 102, Spoken discourse 1–2, 7, 9, 12, 17,
106, 153, 157, 177 19–20, 25, 27, 30, 35, 39, 42, 88,
Shared information 18–19, 118 130, 137, 166, 169–171, 175
vs. unshared information 14, 19 spontaneous/casual vs. planned
Shared knowledge 12–13, 18, 102 170
Index 

Spoken English 2–3, 17, 26, 28–30, 37, Subject-object-verb (SOV) 3–4, 130,
46, 63, 97, 151, 172 161, 169
Spoken Japanese 3, 29, 32, 34, 36–37, as basic word order in Japanese
46, 116, 155, 157–158, 177 3–4, 169
Spoken language 19, 164 Subject-predicate structure 25
non-elaboration of IU structures and Subordinate clause 52, 57, 104–105,
characteristics of 163–164 121, 124, 174
See also Conversational language vs. main clause 52, 57, 121
Spontaneous spoken discourse/ Subordinating conjunction 148
conversation 1–2, 4, 20, 27, 127, Subordination 27
144, 164, 170, 175 Substantive IU 22–26, 29, 35–36, 48–51,
SR See Switch reference 57, 62, 83, 99, 123, 139, 142,
S-role argument See S role 146–147, 149, 151–155,
Stance 34, 126 158–160, 162–166, 171, 173, 175
Starting point 19, 21, 25 See also defined 22, 49
Added information vs. non-substantive IU 22, 48, 154,
State 3, 22, 26–27, 41, 49, 54, 102, 165
126–127, 143, 152, 159 See also Fragmentary; Regulatory IU
State idea 26, 175 Subtopics, of conversation 172
Sufficiently identifying verbalization
Stein, Dieter 127
18–19
Storage locus 38
Sugito, Miyoko 29, 47
Story preface 173
Suzuki, Ryoko 4, 64
Storytelling 172
Svartvik, Jan 19
Strategy, speaker’s 29, 39, 65, 81, 88,
Switch reference (SR) 8, 165
118–119, 177
Syder, Frances H. 2, 26–27, 36, 39, 41
Stray NP 71–79, 88, 93, 159, 173–175
Syntactic factors, and multi-IU clause
defined 71 158, 177
types and definitions 73–74 Syntactic form, NP 101, 104, 111,
Stress 32 113–117, 139, 164
Structures, IU 3, 5, 48, 144, 163–164, and grammatical role 113–115
166 and information status 115
non-elaborate nature of 164 Syntactic fragmentation See
Subject 9, 12, 19, 25, 37, 52, 54–55, 57, Fragmentation
67, 69–70, 96, 100, 103, 110, Syntactic structure, IU 3, 5–6, 29, 32,
112–113, 147–148, 159, 43, 48, 51, 65, 98, 144, 152, 159,
169–171, 176 162–163
vs. object 9, 110, 169 in English vs. Japanese 3, 29–33, 62
vs. predicate 12, 25, 37 Syntactic structure types, IU 51–53,
See also Intransitive; Transitive 57–58, 60, 63, 66, 98–99,
subject 106–107, 109, 159, 164
Subject marker 169 and NP 106–107
Subjectivity 34, 127 See also Point of and postposing 63–66
view core/essential vs. peripheral/
Subject-object relation 169 redundant 57
 Index

defined 52–53 Thought organization 164


distribution of 57–60 Tightness, of clause 33
See also Independent clausal; phrasal Tokieda, Motoki 34, 144
IU; Clausal; Phrasal IU as part Tomlin, Russell S. 1, 10–11
of multi-IU clause Tone group 19, 25–26
Syntax/syntactic organization 33, 40, Tone unit 2, 19
157 Tonic foot/syllable 24
English vs. Japanese, and multi-IU clause Tonic prominence 25
33, 157–158 Topic 1, 7–10, 12, 45, 48–49, 54, 57,
71–72, 76, 81, 83, 90, 100, 122,
T 134, 136, 159, 170, 172, 174–176
Tag-like expression 143 See also accessible vs. inaccessible 7–8
Interactional particle continuous vs. discontinuous 7–8
Takami, Kenichi 65, 165 important vs. unimportant 8
Talk-in-interaction 19, 88, 127 See also Genuine topic; Referent
Tanaka, Hiroko 19 Topic continuity 1, 7, 9–10, 134, 170
Tannen, Deborah 164, 170 defined 7
Tao, Hongyin 3, 9, 19, 29–32, 37, 48, 62, vs. action/location/temporal
67, 72, 81, 87–88, 97, 99, 105, 124, continuity 10
126, 159, 171, 175–176 vs. topic discontinuity 7
TCU See Turn constructional unit See also Referential distance; Topic
Telephone conversation 35 persistence
Temporal juncture 11 Topic marker 54
Tendencies 122, 131, 155, 161 See also Topic NP, clause-external 71–72, 81–83,
Pattern 159, 175 See also Genuine topic
Tense 12, 170 Topic persistence (TP) 8–9, 165
Teramura, Hideo 65
Topic shift 172
Text 8–9, 11
Topicality 7–9
Textual component 143, 145–146,
aspects and measures of 8–9
148–150, 152, 162
See also Referential accessibility;
Textual connective 143, 148, 150
Thematic importance
Textual function 33, 144, 151
Topics, of conversational data 45, 48–49
Textual information 151
Textual subtype 22, 49, 173 TP See Topic persistence
Thematic importance 8, 165 Transcriber xvi, 21
measures and definitions 8 Transcript 46, 50, 54
See also OF; TP Transcription xvi, 43, 46–47
Theme 11 Transition relevance 173
Third person 176 Transitive clause 105, 110, 125, 127–128,
Thompson, Sandra A. 1, 3, 10, 12–13, 130–131, 139, 148, 161, 176
19, 27, 30, 32, 42, 72, 74, 105, 113, vs. non-transitive clause 124–126
125–127, 170, 176 See also High transitive; Low
Thought 21, 42, 166 transitive clause
aspects of 166 Transitive object 41, 103, 132 See also
See also Content; Infrastructure O role
Index 

Transitive subject 41, 103, 132 See also V


A role Validational subtype 22, 49, 173
Transitive verb 103, 122–124 Value 37, 156
Transitive-intransitive bivalency 67 van Dijk, Teun A. 166
Transitivity 12, 105, 124–127, 130, 161, Vandepitte, Sonia 19, 47, 171
170 Venditti, Jennifer J. 29, 47
and grounding 12 Verb 4, 20, 31, 38–39, 41, 52–53, 57–58,
clauses of high vs. low 125–127 61–62, 96, 102–103, 105, 121–124,
continuum 126 126, 130, 154, 170–171, 176–177
defined 105, 170 See also Ditransitive; Intransitive;
features of 12, 170 Transitive verb
Truncated word/unit xv, 22 See also
Verb phrase (VP) 31, 39, 52–53, 58, 61,
False start; Fragmentary IU;
67, 93, 97, 104, 107, 109, 115, 121
Interruption
Verb-adverb combination 41, 102
Tsutsui, Michio 177
Verbal predicate 25, 29, 31, 52, 55,
Turn 19, 46, 83, 172–173, 175
58–62, 67, 96–100, 105–106, 110,
Turn constructional unit (TCU) 173
112, 159, 171, 174
Turn organization 19
Verbal reference 102
Turn-internal listener backchannel 172
Verb-final language, vs. verb-initial 130
Turn-taking 48
Two new NPs per IU 122–124, 160 See Verb-final requirement 4
also Argument-oblique combination Verb-object combination 41
Two-argument clause/structure 128, Verb-particle combination 41
130, 161 Vocalization 15, 20
Two-participant clause 125–126, 130 Vocatives 72
vs. one-participant clause 125–126 Vogel, Irene 20
See also Transitive clause Voice 12
Two-party conversation See Dyadic VP See Verb phrase
conversation VP IU 53, 58, 61, 93, 96–97, 121

U
W
Uncodable IU 49–50
Wallace, Stephen 11, 170
Unifunctionality 153, 155
Ward, Gregory L. 1, 14, 47, 169
Unit See Discourse unit; Intonation
unit; Prosodic unit Wh-question 79
Unitariness 164 Word 2, 20, 23–25, 27, 32, 34, 38–40, 46,
Unmarked information structure 12, 50, 142, 171
25, 120 Word order 3–4, 12, 54, 63, 129, 164,
Unmarked word order 54, 129 169
Unshared information 14, 19 in Japanese spoken discourse 3–4
Unused information 14 marked vs. unmarked/basic/
Utterance 2, 13, 15, 17, 19, 22, 25, canonical 3–4, 54, 63, 164, 169
33–34, 77, 81, 86–87, 92, 100, 127, SOV 3–4, 169
173, 175 See also Basic; Marked word order
 Index

Wright, Susan 127 Z


Writing 27 Zero anaphora 7–9, 11, 29, 120–121,
Written discourse 170 172
Zero argument 124, 128 See also
Anaphoric; Non-anaphoric zero
Zero form 55, 112, 116, 124, 132–134,
Y 161
Yes/no question 70, 76 Zero-form NP 9, 116–117, 133, 161
Yule, George 17–18, 30, 172 Zimmerman, Don H. 166
In the STUDIES IN LANGUAGE COMPANION SERIES (SLCS) the following volumes
have been published thus far or are scheduled for publication:
1. ABRAHAM, Werner (ed.): Valence, Semantic Case, and Grammatical Relations. Work-
shop studies prepared for the 12th Conference of Linguistics, Vienna, August 29th to
September 3rd, 1977. Amsterdam, 1978.
2. ANWAR, Mohamed Sami: BE and Equational Sentences in Egyptian Colloquial Arabic.
Amsterdam, 1979.
3. MALKIEL, Yakov: From Particular to General Linguistics. Selected Essays 1965-1978.
With an introd. by the author + indices. Amsterdam, 1983.
4. LLOYD, Albert L.: Anatomy of the Verb: The Gothic Verb as a Model for a Unified Theory
of Aspect, Actional Types, and Verbal Velocity. Amsterdam, 1979.
5. HAIMAN, John: Hua: A Papuan Language of the Eastern Highlands of New Guinea.
Amsterdam, 1980.
6. VAGO, Robert (ed.): Issues in Vowel Harmony. Proceedings of the CUNY Linguistics
Conference on Vowel Harmony (May 14, 1977). Amsterdam, 1980.
7. PARRET, H., J. VERSCHUEREN, M. SBISÀ (eds): Possibilities and Limitations of
Pragmatics. Proceedings of the Conference on Pragmatics, Urbino, July 8-14, 1979. Am-
sterdam, 1981.
8. BARTH, E.M. & J.L. MARTENS (eds): Argumentation: Approaches to Theory Formation.
Containing the Contributions to the Groningen Conference on the Theory of Argumenta-
tion, Groningen, October 1978. Amsterdam, 1982.
9. LANG, Ewald: The Semantics of Coordination. Amsterdam, 1984.(English transl. by John
Pheby from the German orig. edition “Semantik der koordinativen Verknüpfung”, Berlin,
1977.)
10. DRESSLER, Wolfgang U., Willi MAYERTHALER, Oswald PANAGL & Wolfgang U.
WURZEL: Leitmotifs in Natural Morphology. Amsterdam, 1987.
11. PANHUIS, Dirk G.J.: The Communicative Perspective in the Sentence: A Study of Latin
Word Order. Amsterdam, 1982.
12. PINKSTER, Harm (ed.): Latin Linguistics and Linguistic Theory. Proceedings of the 1st
Intern. Coll. on Latin Linguistics, Amsterdam, April 1981. Amsterdam, 1983.
13. REESINK, G.: Structures and their Functions in Usan. Amsterdam, 1987.
14. BENSON, Morton, Evelyn BENSON & Robert ILSON: Lexicographic Description of
English. Amsterdam, 1986.
15. JUSTICE, David: The Semantics of Form in Arabic, in the mirror of European languages.
Amsterdam, 1987.
16. CONTE, M.E., J.S. PETÖFI, and E. SÖZER (eds): Text and Discourse Connectedness.
Amsterdam/Philadelphia, 1989.
17. CALBOLI, Gualtiero (ed.): Subordination and other Topics in Latin. Proceedings of the
Third Colloquium on Latin Linguistics, Bologna, 1-5 April 1985. Amsterdam/Philadelphia,
1989.
18. WIERZBICKA, Anna: The Semantics of Grammar. Amsterdam/Philadelphia, 1988.
19. BLUST, Robert A.: Austronesian Root Theory. An Essay on the Limits of Morphology.
Amsterdam/Philadelphia, 1988.
20. VERHAAR, John W.M. (ed.): Melanesian Pidgin and Tok Pisin. Proceedings of the First
International Conference on Pidgins and Creoles on Melanesia. Amsterdam/Philadelphia,
1990.
21. COLEMAN, Robert (ed.): New Studies in Latin Linguistics. Proceedings of the 4th
International Colloquium on Latin Linguistics, Cambridge, April 1987. Amsterdam/
Philadelphia, 1991.
22. McGREGOR, William: A Functional Grammar of Gooniyandi. Amsterdam/Philadelphia,
1990.
23. COMRIE, Bernard and Maria POLINSKY (eds): Causatives and Transitivity. Amster-
dam/Philadelphia, 1993.
24. BHAT, D.N.S. The Adjectival Category. Criteria for differentiation and identification.
Amsterdam/Philadelphia, 1994.
25. GODDARD, Cliff and Anna WIERZBICKA (eds): Semantics and Lexical Universals.
Theory and empirical findings. Amsterdam/Philadelphia, 1994.
26. LIMA, Susan D., Roberta L. CORRIGAN and Gregory K. IVERSON (eds): The Reality of
Linguistic Rules. Amsterdam/Philadelphia, 1994.
27. ABRAHAM, Werner, T. GIVÓN and Sandra A. THOMPSON (eds): Discourse Grammar
and Typology. Amsterdam/Philadelphia, 1995.
28. HERMAN, József: Linguistic Studies on Latin: Selected papers from the 6th international
colloquium on Latin linguistics, Budapest, 2-27 March, 1991. Amsterdam/Philadelphia,
1994.
29. ENGBERG-PEDERSEN, Elisabeth et al. (eds): Content, Expression and Structure. Studies
in Danish functional grammar. Amsterdam/Philadelphia, 1996.
30. HUFFMAN, Alan: The Categories of Grammar. French lui and le. Amsterdam/Philadel-
phia, 1997.
31. WANNER, Leo (ed.): Lexical Functions in Lexicography and Natural Language Processing.
Amsterdam/Philadelphia, 1996.
32. FRAJZYNGIER, Zygmunt: Grammaticalization of the Complex Sentence. A case study in
Chadic. Amsterdam/Philadelphia, 1996.
33. VELAZQUEZ-CASTILLO, Maura: The Grammar of Possession. Inalienability, incorpora-
tion and possessor ascension in Guaraní. Amsterdam/Philadelphia, 1996.
34. HATAV, Galia: The Semantics of Aspect and Modality. Evidence from English and Biblical
Hebrew. Amsterdam/Philadelphia, 1997.
35. MATSUMOTO, Yoshiko: Noun-Modifying Constructions in Japanese. A frame semantic
approach. Amsterdam/Philadelphia, 1997.
36. KAMIO, Akio (ed.): Directions in Functional Linguistics. Amsterdam/Philadelphia, 1997.
37. HARVEY, Mark and Nicholas REID (eds): Nominal Classification in Aboriginal Australia.
Amsterdam/Philadelphia, 1997.
38. HACKING, Jane F.: Coding the Hypothetical. A Comparative Typology of Conditionals in
Russian and Macedonian. Amsterdam/Philadelphia, 1998.
39. WANNER, Leo (ed.): Recent Trends in Meaning-Text Theory. Amsterdam/Philadelphia,
1997.
40. BIRNER, Betty and Gregory WARD: Information Status and Noncanonical Word Order in
English. Amsterdam/Philadelphia, 1998.
41. DARNELL, Michael, Edith MORAVSCIK, Michael NOONAN, Frederick NEWMEYER
and Kathleen WHEATLY (eds): Functionalism and Formalism in Linguistics. Volume I:
General papers. Amsterdam/Philadelphia, 1999.
42. DARNELL, Michael, Edith MORAVSCIK, Michael NOONAN, Frederick NEWMEYER
and Kathleen WHEATLY (eds): Functionalism and Formalism in Linguistics. Volume II:
Case studies. Amsterdam/Philadelphia, 1999.
43. OLBERTZ, Hella, Kees HENGEVELD and Jesús Sánchez GARCÍA (eds): The Structure of
the Lexicon in Functional Grammar. Amsterdam/Philadelphia, 1998.
44. HANNAY, Mike and A. Machtelt BOLKESTEIN (eds): Functional Grammar and Verbal
Interaction. 1998.
45. COLLINS, Peter and David LEE (eds): The Clause in English. In honour of Rodney
Huddleston. 1999.
46. YAMAMOTO, Mutsumi: Animacy and Reference. A cognitive approach to corpus linguis-
tics. 1999.
47. BRINTON, Laurel J. and Minoji AKIMOTO (eds): ollocational and Idiomatic Aspects of
Composite Predicates in the History of English. 1999.
48. MANNEY, Linda Joyce: Middle Voice in Modern Greek. Meaning and function of an
inflectional category. 2000.
49. BHAT, D.N.S.: The Prominence of Tense, Aspect and Mood. 1999.
50. ABRAHAM, Werner and Leonid KULIKOV (eds): Transitivity, Causativity, and TAM.
In honour of Vladimir Nedjalkov. 1999.
51. ZIEGELER, Debra: Hypothetical Modality. Grammaticalisation in an L2 dialect. 2000.
52. TORRES CACOULLOS, Rena: Grammaticization, Synchronic Variation, and Language
Contact.A study of Spanish progressive -ndo constructions. 2000.
53. FISCHER, Olga, Anette ROSENBACH and Dieter STEIN (eds.): Pathways of Change.
Grammaticalization in English. 2000.
54. DAHL, Östen and Maria KOPTJEVSKAJA-TAMM (eds.): Circum-Baltic Languages.
Volume 1: Past and Present. 2001.
55. DAHL, Östen and Maria KOPTJEVSKAJA-TAMM (eds.): Circum-Baltic Languages.
Volume 2: Grammar and Typology. 2001.
56. FAARLUND, Jan Terje (ed.): Grammatical Relations in Change. 2001.
57. MEL’C UK, Igor: Communicative Organization in Natural Language. The semantic-
communicative structure of sentences. 2001.
58. MAYLOR, Brian Roger: Lexical Template Morphology. Change of state and the verbal
prefixes in German. 2002.
59. SHI, Yuzhi: The Establishment of Modern Chinese Grammar. The formation of the
resultative construction and its effects. 2002.
60. GODDARD, Cliff and Anna WIERZBICKA (eds.): Meaning and Universal Grammar.
Theory and empirical findings. Volume 1. 2002.
61. GODDARD, Cliff and Anna WIERZBICKA (eds.): Meaning and Universal Grammar.
Theory and empirical findings. Volume 2. 2002.
62. FIELD, Fredric W.: Linguistic Borrowing in Bilingual Contexts. 2002.
63. BUTLER, Chris: Structure and Function – A Guide to Three Major Structural-Functional
Theories. Part 1: Approaches to the simplex clause. 2003.
64. BUTLER, Chris: Structure and Function – A Guide to Three Major Structural-Functional
Theories. Part 2: From clause to discourse and beyond. 2003.
65. MATSUMOTO, Kazuko: Intonation Units in Japanese Conversation. Syntactic, informa-
tional and functional structures. 2003.
66. NARIYAMA, Shigeko: Ellipsis and Reference Tracking in Japanese. 2003.
67. LURAGHI, Silvia: On the Meaning of Prepositions and Cases. The expression of semantic
roles in Ancient Greek. n.y.p.
68. MODER, Carol Lynn and Aida MARTINOVIC-ZIC (eds.): Discourse Across Languages
and Cultures. n.y.p.
69. TANAKA, Lidia: Gender, Language and Culture. A study of Japanese television interview
discourse. n.y.p.