Intonation Units in Japanese Conversation: Syntactic, Informational, and Functional Structures (Studies in Language Companion Series)

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 For Evolutionary Anthropology, Leipzig

University of Texas, Arlington

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 For Evolutionary Anthropology, Leipzig

Russell Tomlin

Brian MacWhinney

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

8

TM

The paper used in this publication meets the minimum requirements 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, 1955Intonation 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 isbn 90 272 3075 7 (Eur.) / 1 58811 364 7 (US) (Hb; alk. paper)

2002033023

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

1

3

Chapter 2 Information flow in spoken discourse 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

7

 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 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

43

Chapter 4 Syntactic structure of the intonation unit in conversational Japanese 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

51

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 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

141





Table of contents

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

169

References Index 191

179

159

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 University 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 pertinent 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. 3.2. 3.3. 3.4. 3.5.

Data for this study Major topics of 16 conversations Number of IUs produced in 16 conversations Topics of 16 conversational segments Number of four types of IUs in 16 conversational segments

4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7. 4.8. 4.9. 4.10. 4.11. 4.12. 4.13. 4.14.

Distribution of IU syntactic structure types Distribution of eight clausal IU types Distribution of 12 phrasal IU types Distribution of post-predicate phrases among IU syntactic types Proportion of IUs involving postposing by IU syntactic type Proportion of full clauses vs. semi-clauses Distribution of five types of NP IUs Distribution of five types of independent phrasal NP IUs Frequency and features of four types of stray NP IUs Distribution of adjectival, adverbial, and mixed phrasal IU types Distribution of six types of independent adverbial IUs Average number of IUs per clause Frequency of the number of IUs per clause Summary of results of IU syntactic structure analysis

5.1. 5.2. 5.3. 5.4. 5.5. 5.6. 5.7. 5.8. 5.9. 5.10.

Overt NPs, IUs, and clauses Average number of overt NPs per IU by IU syntactic type Distribution of grammatical roles among IU syntactic types Grammatical role and information status Grammatical role and NP syntactic form Information status and NP syntactic form Summary of results of analysis of overt NPs Distribution of IU information structure types Two types of IUs with zero NPs Types of IUs with two new NPs

 Tables and Figures

5.11. 5.12. 5.13. 5.14. 5.15.

Frequency of clause types by the number of overt arguments Frequency of clauses with overt S, A, and O arguments Frequency of clauses with zero, one, and two overt arguments Proportion of overt arguments by clause type Proportion of IUs/clauses with more than one new NP

6.1. 6.2.

Distribution of IU functional structure types Frequency of the number of functional components per IU

Figures 4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7.

Distribution of four major IU syntactic types Distribution of six IU syntactic types Distribution of 20 IU syntactic types Proportion of full clauses vs. semi-clauses Distribution of five types of NP IUs Distribution of five types of independent phrasal NP IUs Frequency of the number of IUs per clause

5.1. 5.2. 5.3. 5.4. 5.5. 5.6. 5.7. 5.8. 5.9.

Distribution of six grammatical roles Proportion of three information statuses within grammatical roles Distribution of eight NP syntactic forms Distribution of IU information structure types Frequency of five clause types Proportion of clauses with zero, one, and two arguments Frequency of clauses with zero, overt S, A, and O arguments Proportion of overt vs. null arguments 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 greaterthan 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 selfinterruption; 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 indicate 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 ACC ALL CAU COM COMP CON COP DAT DES EMP FP GEN HON IMP INS LOC

ablative accusative allative causative comitative comparative conditional copula dative desiderative emphatic final particle genitive honorific imperative instrumental locative

NEG NML NOM PASS PAST PF PL POL POT PROG PRS Q QT STA SOF TAG TOP

negative nominalizer nominative passive past pause filler plural polite form potential progressive presumptive question marker quotative stative softener tag 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 speakers 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 previous 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 discourse. 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 information 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 structure of IUs in English and Japanese, have shown, however, that while English 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 warranted 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 structure, 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 operating 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 object 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 constructions 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 interacts 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 others 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 functional 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 patterns in their production of IUs. In its analysis of the syntactic and information structures of the Japanese IU, the present study focuses on nominal references 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 functional structure of the IU, the study focuses on the linear order of functional components within an IU as well as interrelationships among those components. Overall, in exploring the syntactic, informational, and functional structures 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 information 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, information 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”. Referential 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 complexity” (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 ‘cognitive 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 textfrequency 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 “activation 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 almighty, 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 information, that is, to integrate or “address” the new information onto the appropriate 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 asserted, or foregrounded at point n in the discourse will become a shared, backgrounded 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 particular rhetorical purpose and its associated theme” (Tomlin 1985: 119). The continuum thus places propositions describing events which are more crucial to the developing theme higher on that continuum. Although intended to be a genre-independent definition, according to Tomlin, his characterization of the foreground-background information seems to be useful primarily (or even only) in analyses of highly structured narrative discourse, not in analyses of broader conversational interaction. Clearly, more research should be





Chapter 2

directed toward the development of precise, operational definitions of the foreground 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 discourse, is presupposed; and new events tend to be introduced in the predicate (cf. Section 2.4.5). In backgrounded clauses, by contrast, new information 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 decision 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 formulated 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 recoverability 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 predictable/recoverable to the listener anaphorically or situationally represents “given/old” information; an element that is presented by the speaker as being 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, givenness/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 interlocutors. 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 “brandnew” 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 discoursemodel” (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 classified, 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 distinction 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 “unshared” vs. “shared”. However, he emphasizes that sharedness, which is a component 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 question from whether or not it is assumed to be already part of the listener’s knowledge. . . . 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 “activation 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 proposes 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)

speakeroriented

t1

t2

t3

pause onset

word1 word2 word3 ..... IU onset

IU completion

active semi-active inactive

listeneroriented

given

active

accessible new

active semi-active inactive

given

active

accessible new

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 activation 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 information 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 emphasize 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 contexts 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 discourse, by the environment in which a conversation takes place, or by the social 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 require the use of demonstrative pronouns, proper names (e.g. Larry in (2.3b)), common nouns with definite or demonstrative articles, or the addition of modifiers (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 production 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 example, 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 unshared, whereas the idea of Larry in (2.3b) was assumed to be shared by the listener. 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 basic 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; accordingly, relevant dimensions for the segmentation of discourse need to be specified. Prosodic features reflect various interactional, cognitive, and linguistic 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 research has naturally focused on the prosodic structure of discourse and segmented spoken interaction into prosodic units. As previous studies have noted, a prosodic unit is simultaneously the domain of the syntactic as well as information 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 perception (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 prosodybased 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 identification 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, hesitational, 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 consciousness, whose properties apparently belong to our built-in informationprocessing capabilities”. In other words, the IU as a whole is a verbalization of the particular information on which the speaker is focusing his/her consciousness 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”, “regulatory”, and “fragmentary”. Substantive IUs convey substantive ideas of events, states, or referents, participating in the communication of propositional content. 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 socalled “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: “textual”, “interactional”, “cognitive”, and “validational”. The textual subtype functions 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 between 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, validational subtype expresses the speaker’s judgment of the validity of the information being conveyed (e.g. maybe). As Chafe (1993) emphasizes, however, the boundaries between these subtypes of regulatory IUs are not clear-cut. (For example, 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. b. c. d. e. f. g. h. i. j. k.

(R) (S) (R) (F) (F) (S) (S) (S) (S) (S) (S)

A: ++ Well, isn’t she healthy? B: + Mhm, A: ++ I mean she I know she has C: More or less. A: + She has [something with her] gallbladder, B: [ gallbladder and, ] ++ heart trouble and, [back problems.] 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 substantive 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 presumably 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 canonical pattern of information flow—from given to new—within units of English spoken discourse. The new element carrying “tonic prominence” and thus “information focus” is typically the last functional element within the “information 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 processing 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” connectives 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 characteristics 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 validity of the clause centrality proposal, examined the syntactic structure of substantive IUs produced by native speakers of English and Japanese in interactive conversations. Their major findings are summarized in (2.13).15 (2.13) English Japanese

Clausal 54% 45%

Non-clausal 46% 55%

English Japanese

Full-clause 83% 24%

Semi-clause 17% 76%

English Japanese

NP IUs 12% 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 verbal predicates, whereas the clausal IUs in Japanese are mostly subjectless semiclauses that consist of verbal predicates only. (Clearly this can be attributed to the grammatical difference between the two languages that Japanese allows 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.

b.

c.

d.

e.

f.

g.

h.

A: Yamato san ga ne:, Yamato hon nom fp ‘Mr. Yamato’ kekkyoku ne:, in short fp ‘in short’ wareware o ne:, us acc fp ‘us’ sofuto no ne:, software gen fp ‘on software’ shigoto nitaisuru. job toward ‘of (our) job’ hyooka ga ano hito ne:, evaluation nom that person fp ‘evaluation, he’ shitenai. do-prog-neg ‘doesn’t do’ 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 consisting 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 interactional 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 collectively 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 researchers. Maynard (1989), for example, refers to such smaller phrasal units as “pause-bounded phrasal units” (PPUs), which, in my interpretation, are essentially 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 fragmentation is emphasized and made more prominent (Maynard 1989: 21–22). Likewise, 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, underscoring 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 linguistics by Tokieda (1950) and his followers. They identify three metafunctions of language: “descriptive”, “expressive”, and “interpersonal”. The descriptive function, which is performed by the act of jojutsu ‘description’, concerns the communication of propositions. The expressive function, which is performed by the act of chinjutsu ‘expression’, reveals the speaker’s internal state, or subjectivity. The interpersonal function concerns itself with speech acts as Halliday claims, or the speaker’s sensitivity toward the addressee. Unlike Halliday, however, 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 Tokiedaschool Japanese linguists, concludes that language must incorporate four metafunctions: “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 interactional. 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 proposition 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 elements 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-toface 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 ID CO ID IT LD ID ID CO IT LD ID CO IT

34% 22% 17% 2% 12% 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 important 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 permits 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 constraint” (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 multifunctionality of Japanese IUs discussed above. He argues that a strong motivational 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 information in addition to propositional, ideational information. That is, the frequent 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 addition to textual cohesive information and propositional/ideational information 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 information, 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 between 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 opportunity 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 incorporation 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 information? 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 communication 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. Apparently, 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 previously 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 narrative 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) a. b. c. e. f. g. h. i. j. k. l.

GIVEN he

ACCESSIBLE cláss

he

his nótes the róom

he

NEW would come into at three or f precísely one minute after the hóur, would immédiately open up in the front of st

évery lécture

this Wesleyan school

áfter the fírst stárted the same wáy was at Wésleyan, was still a mén’s

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 newargument 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 consisting 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 applicability of the quantity and role constraints of PAS. Among the most recent studies, Ashby and Bentivoglio (1993), Smith (1996), and Kärkkäinen (1996), for example, have shown that PAS holds in spoken French and Spanish, modern 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 argument 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 45minute 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 of conversation

F1 F2 F3 F4 F5 F6 F7 F8 Total

45 min 45 min 45 min 45 min 45 min 45 min 45 min 45 min 360 min

Participants (Age) Saori (27) Yukiko (21) Mari (22) Akiko (24) Masayo (22) Eri (26) Yukari (24) Kumiko (21) N=16 (Average=23.8)

Yumi Misako Orie Mika Yasue Noriko Tae Haruko

(27) (23) (23) (26) (22) (27) (25) (21)

Method of the study

Table 3.1. (continued) Dyad

Length of conversation

Participants (Age)

M1 M2 M3 M4 M5 M6 M7 M8 Total

45 min 45 min 45 min 45 min 45 min 45 min 45 min 45 min 360 min

Masao (25) Takeshi (21) Shigeo (25) Takuya (26) Osamu (24) Ken (21) Ichiro (22) Hiroshi (22) N=16 (Average=23.5)

TOTAL

720 min

N=32 (AVERAGE=23.7)

Shinji Naoto Jun Hideki Koichi Isao Seiji Yutaka

(26) (22) (26) (26) (25) (22) (22) (21)

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 F2 F3 F4 F5 F6 F7 F8

roommates, danger in America, riots in Los Angeles oriental food, Easter, roommates, politics in Japan sports, returnees, university majors, literature courses driving, M’s trip to the Middle East, cooking, marriage Japanese friends, summer vacation, Japanese restaurants conferences, research projects, parties, pierced earrings funerals, sociology, anthropology, UCLA departments movies, K’s visit to New York, English classes, assignments

M1 M2 M3 M4 M5 M6 M7 M8

Asians, success in life, American food, traveling in Japan men vs. women, American girls, T’s American friends computers, TV programs, Japanese students, dormitories laws, honorifics, Japanese universities, northern part of Japan cooking, UCLA courses, registration, course units job-hunting, apartment life, Bruce Lee, playing tennis abnormal weather, classes, graduation, future plans 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. b. c. d. e.

falling intonation (transcribed as .). continuing intonation (transcribed as ,).5 rising intonation (transcribed as ?). rise-fall intonation (transcribed as ^). rise-fall-rise intonation (transcribed as ∼).

Spoken Japanese, like spoken English, exhibits a relatively high degree of alignment 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 incompleteness. 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 contour 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 (45 min)

Average number of IUs per min

F1 F2 F3 F4 F5 F6 F7 F8 Total (Average)

1,526 1,598 1,441 1,643 1,285 1,345 1,268 1,302 11,415 (1,426.9)

M1 M2 M3 M4 M5 M6 M7 M8 Total (Average)

1,512 1,410 1,256 1,203 1,228 1,432 1,357 1,396 10,794 (1,349.3)

(29.9)

TOTAL (AVERAGE)

22,209 (1,388.1)

(30.8)

33.9 35.5 32.0 36.5 28.7 29.9 28.2 28.9 (31.7) 33.6 31.3 27.9 26.7 27.3 31.8 30.2 31.0





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 segment 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 taking place, and thus the co-participants’ verbal contributions to discourse production are fairly balanced between the two.6 In addition, the segments were carefully selected in order for different topics and different types of conversational 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 F2 F3 F4 F5 F6 F7 F8

danger in America Easter university majors marriage summer vacation research projects sociology movies

M1 M2 M3 M4 M5 M6 M7 M8

success in life American girls TV programs honorifics UCLA courses apartment life graduation 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 conversational 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 segments, 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 conversational 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

F1 F2 F3 F4 F5 F6 F7 F8 Total

100 100 100 100 100 100 100 100 800 (81) 100

20 13 23 31 15 32 19 20 173 (17) 21.6

1 1 3 1 1 2 1 0 10 (1.0) 1.3

1 1 1 1 3 1 0 1 9 (0.9) 1.1

121 116 127 133 119 135 120 121 992 (100%) 124.0

100 100 100 100 100 100 100 100 800 (81) 100

14 20 26 16 15 22 28 24 165 (17) 20.6

1 0 3 1 1 1 1 1 9 (0.9) 1.1

1 3 4 1 0 2 0 2 13 (1.3) 1.6

116 123 133 118 116 125 129 127 987 (100%) 123.4

1,600 (81) 100

338 (17) 21.1

19 (0.9) 1.2

22 (1.1) 1.4

1,979 (100%) 123.7

Average M1 M2 M3 M4 M5 M6 M7 M8 Total Average TOTAL AVERAGE

Total

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 semiclause 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] [PVP] [PNP] [PAP]

= = = =

overt subject + verbal/nominal/adjectival predicate Ø-subject + verbal predicate Ø-subject + nominal predicate Ø-subject + adjectival predicate

Type 2: Clausal IU as part of multi-IU clause: a clausal IU that can communicate 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 involving 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] [CVP] [CNP] [CAP]

= = = =

overt subject + verbal/nominal/adjectival predicate Ø-subject + verbal predicate Ø-subject + nominal predicate 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] [CNPa] [CNPp] [CAPa] [CAPp] [CAvP] [CXP]

= = = = = = =

verb phrase argument NP predicate NP attributive AP predicative AP adverbial phrase 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] [NP] [AP] [AvP] [XP]

= = = = =

verb phrase noun phrase adjective phrase adverbial phrase 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] [VP] [NP] [AP] [AvP] [XP]

= = = = = =

[FC] + [CFC] [PVP] + [CVP] + [CVPp] + [VP] [PNP] + [CNP] + [CNPa] + [CNPp] + [NP] [PAP] + [CAP] + [CAPa] + [CAPp] + [AP] [CAvP] + [AvP] [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 “basegenerated”, “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 multiIU 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 semiclause 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 independent 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 nominal predicate (i.e. [PNP]). The IU (4.3a), uttered by speaker E but conveying 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 integrated 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)

yame na:, [CVP] O: kocchi no daigaku taihen da kara 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)

(4.7) a.

(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 multiIU 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 complete propositions either singly or collectively, are centrally concerned with the communication of propositional content. These IUs constitute propositionally “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 “redundant” part of the conversational language (cf. Dik 1989). That is, substantive 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 latter are peripheral, to the communication of ideas. In brief, Japanese conversational 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 Independent Part of multiIU IU clause

Phrasal IU Part of multiIndependent IU clause IU

Total

[FC]

263

[CFC]

40

[PVP]

392

[CVP]

106

[CVPp]

34

[VP]

7

[VP] [NP]

[CNP]

17

94 23

112

174

[CNPa] [CNPp]

[NP]

[PNP]

[AP]

[CAP]

7

18 11

14

81

[CAPa] [CAPp]

[AP]

[PAP]

131 (8%)

[CAvP]

90

[AvP]

82

[AvP]

[CXP]

27

[XP]

8

[XP]

172 (11%) 35 (2%)

Total

910 (57%)

170 (11%)

[FC]

297 (18%)

223 (14%)

303 (19%) 539 (34%) 420 (26%)

1,600 (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 semiclausal 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

Frequency

700 600 500 400 300 200 100 Independent phrasal IU

Phrasal IU as part of multi-IU clause

Clausal IU as part of multi-IU clause

Independent clausal IU

0

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

%

[FC] [PVP] [PNP] [PAP]

263 392 174 81

29 43 19 9

Total

910

100

Type 2 IU

N

%

[CFC] [CVP] [CNP] [CAP]

40 106 17 7

24 62 10 4

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 syntactic types, independent clausal IUs which communicate complete propositions 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

Frequency

250 200 150 100 50

NP AP AvP XP

CAvP CXP VP

CAPa CAPp

CFC CVP CNP CAP CVPp CNPa CNPp

FC PVP PNP PAP

0

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 consisting of verbal predicates or verb phrases (i.e. [VP] IUs) were most frequently produced (34%). In sum, the results indicate that the Japanese conTable 4.3. Distribution of 12 phrasal IU types Type 3 IU [CVPp] [CNPa] [CNPp] [CAPa] [CAPp] [CAvP] [CXP] Total

N

%

Type 4 IU

N

%

34 94 23 18 11 90 27

11 32 8 6 4 30 9

[VP] [NP]

7 112

3 50

[AP]

14

6

[AvP] [XP]

82 8

37 4

297

100

Total

223

100





Chapter 4

versational co-participants produced preferentially clausal IUs, typically independent 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 “unexpectedly” 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 comparable 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 conversational 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 proposal” 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 postposing. 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] [PVP] [CVP] Type 1-Total

23 7 1 31

25

[CNPa] [CAPa] [CAvP] [CXP] Type 2-Total

26 6 32 2 66

54

[NP] [AP] [AvP] Type 3-Total

20 1 5 26

21

123

100

TOTAL





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.

b.

c.

J:

deteta. [CVPp] appear-past-sta ‘was (there on the magazine)’ kiji ga. [CNPa] article nom ‘an article’ 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 immediately 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 foregoundbackground alternation in connected discourse (cf. Section 2.2). The clauseexternal postposed phrases, on the other hand, can be seen to be added for purposes 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 multiIU clauses; they directly participate in the construction of propositions. Type 3 post-predicate phrases, in contarst, only repeat or restate “redundantly” the information 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 comparison, 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 examination. 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 IU syntactic type

Total number of IUs N

N

IUs involving postposing %

[FC] [PVP] [CVP] [CNPa] [CAPa] [CAvP] [CXP]

263 392 106 94 18 90 16

23 7 1 26 6 32 2

8.7 1.8 0.9 27.7 33.3 35.6 12.5

[NP] [AP] [AvP]

112 9 82

20 1 5

17.9 11.1 6.1

1,600

123

7.7

Total

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

Clause type

Independent clausal IU N %

Clausal IU as part of multi-IU clause N %

Clausal IU Total N %

Full clause Semi-clause Total

263 647 910

40 130 170

303 777 1,080

29 71 100

23 77 100

28 72 100

. 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 constituting 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

Frequency

600 500

Full clause

400 Semi-clause 300 200 100

Clausal IU Total

Clausal IU as part of multi-IU clause

Independent clausal IU

0

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 nominal/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 NP IU type

CNPp

NP

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] [CNP] [CNPa] [CNPp] [NP]

174 17 94 23 112

41 4 22 6 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.

b.

c. d.

e.

f.

g.

h.

H: nihon de isoona kakkoo.= [PNP] Japan loc exist-likely appearance ‘(It’s) an appearance (which is) likely to be found in Japan.’ K: =ikeike. [PNP] ikeike ‘(It’s) ikeike.’ H: +ikeike. ((laugh)) [NP] ‘ikeike’ K: ((laugh)) ano ko ga, [NP] that girl nom ‘she’ [ dakara, ] so ‘so’ H: [ano hito] shakaijin desho?= [FC] that person working person tag ‘She is working, isn’t she?’ K: =janai no? tag fp ‘I guess so.’ 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 risefall-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 complete 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 selfproduced 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 Type 1 Type 2 Type 3 Type 4 Type 5 Total

N Stray Lead Topic/LDed Postposed/RDed Repeated Others (Vocatives)

%

24 19 30 18 19 2

21 17 27 16 17 2

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 referentially/informationally relatable to the immediately following self-produced IU clauses. Repeated NPs can be interactionally relatable to the immediately preceding other-produced IU clauses (see detailed discussion below). In brief, independent 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 production 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

Frequency

20

15

10

5

Others

Repeated

Postposed/ RDed

Topic/LDed

Lead

Stray

0

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 otherinterruptions 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 otherinterruptions or self-repairs. Type 4: “argument” NP IUs that are “stray” in the sense that predicate elements that would be formally required to express a full proposition are neither self-supplied nor other-supplied, but rather assumed by the conversational coparticipants 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 functions 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 frequency

Elements yet to be uttered in order to constitute a full proposition

Reconstructability of elements not uttered by the original speaker

Distance between the original NP and supplied elements

Type 1 N=13 Type 2 N= 3 Type 3 N= 3 Type 4 N= 5

Not supplied Other-supplied Self-supplied Not supplied

No Yes/No — Yes

— Adjacent Non-adjacent —

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 different 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 typical conversational acts which demonstrate the co-participants’ interactional involvement. The NP IU (4.12d) was begun by K, but because of H’s interrupted 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 constitute 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 collaborative finishes as in (4.15b), as well as Type 1 stray NPs involving interruptions as in (4.12d), are the “products of active interaction” among the coparticipants 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 iand 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 proposition, 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 communication 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 independent 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 completions, 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 communication of propositions is concerned, seems yet to have important implications for such matters as the process of proposition construction and information structure 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 production 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 information because the referent was already activated in the hearer’s consciousness 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.

c.

d.

e.

f.

g.

h.

i.

j.

(4.20) a.

tatoeba, [AvP] for example ‘for example’ +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,’ 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.’ ++desho? tag ‘wouldn’t it?’ de, and ‘and’ manten toreru tte no muzukashii kara, [FC] full mark take-pot qt nml difficult because ‘Because getting a full mark is difficult,’ ja tatoeba nanahyaku to shitara, [PVP] so for example 700 qt do-con ‘So, suppose (it’s) 700, for example’ baabaru yonhyaku da mon.@ [FC] verbal 400 cop nml ‘the verbal (section’s score will be) 400.’ +dakara taihen da na: to omotte, [PVP] so hard cop fp qt think-and ‘So (I) think that (it’s) hard.’ (F3)

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.

e.

→f.

g.

h.

ato sono hoomuwaaku no,@ [AP] and the homework gen ‘and of the Homework (=TV drama title)’ nan da kke. what cop q ‘What shall I say?’ ano dooseishiteru onna no hito mo, [NP] that cohabit-prog woman gen person also ‘the woman who is cohabiting also’ are kashu mitaidesu ne. [CFC] that singer appear-pol fp ‘she seems to be a singer’ +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 multiIU 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.

→b.

c.

d.

e.

f.

g.

h.

i.

T: honde ato kinniku shinkoo mo tsuyoi ne. [FC] and rest muscle faith also strong fp ‘And “muscle faith” is also strong.’ +kinniku ni taisuru shinkoo ga.@@ [NP] muscle toward faith nom ‘faith in muscle’ maa nihonjin wa kekko ne, [CXP] pf Japanese top quite fp ‘well Japanese quite’ dochira ka to surimu desho?@ [CNPp] which q qt slim tag ‘(are) relatively slim, aren’t they?’ minna ningen hosoi desho? [FC] all people slim tag ‘People are all slim, aren’t they?’ 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.’ T: ne. fp ‘You are right.’ N: ++ano macchoman. [PNP] pf Macchoman ‘Uh (there’s) the Macchoman (=comic character).’ 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. social security number ‘social security number’

[NP]

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 (literally)’ 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 contained 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 participant 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, reinforcement, 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 patterns of IU production involving the lead, postposed, and repeated NP IUs. Most typically, lead NPs were prefixed to the immediately following singleIU clause, in which the repeated/restated NP functions as the subject of that clausal IU. The robust pattern of clause-external postposing involved the repeating/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 immediately 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 selfproduced repetition, the repeated NPs concern other-produced repetition. We can conclude from these findings that one of the recurrent patterns of discourse production in the form of IUs in conversational Japanese is to “prefix” or “suffix” a single-IU clause that expresses a complete proposition with clauseexternal 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 extraclausal 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 discoursepragmatic 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 multiIU 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 independent 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 [PAP] [CAP] [CAPa] [CAPp] [AP] Total

N

%

81 7 18 11 14

62 5 14 8 11

131

100

AvP IU type [CAvP]

[AvP]

N

XP IU type

N

52

[CXP]

27

82

48

[XP]

172

100

90

%

% 77

8

23

35

100





Chapter 4

Table 4.11. Distribution of six types of independent adverbial IUs AvP type Type 1 Type 2 Type 3 Type 4 Type 5

N Stray Lead Topic/LDed Postposed/RDed Repeated Sentential adverb

Total

%

40 7 1 5 6 23

49 9 1 6 7 28

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.= company gen mainstream cop neg because fp ‘Because (it’s) not the company’s mainstream.’

[PNP]





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 of IUs

Number of clauses

Average number of IUs per clause

Single-IU full clause Multi-IU full clause Full clause-Total

263 233 496

263 105 368

1.00 2.22 1.35

Single-IU semi-clause Multi-IU semi-clause Semi-clause-Total

647 234 881

647 106 753

1.00 2.21 1.17

910 467 1,377

910 211 1,121

1.00 2.21 1.23

Single-IU clause Multi-IU clause Clause-Total

Syntactic structure of the intonation unit in conversational Japanese

Table 4.13. Frequency of the number of IUs per clause Full clause Number of IUs per clause

N

One Two Three Four Five+

263 83 21 1 0

Total

368

Semi-clause

%

N

71 23 6 0.3 0

%

647 86 18 2 0

100

753

86 11 2 0.3 0 100

Total N

%

910 169 39 3 0 1,121

81 15 4 0.3 0 100

1000 900 800

Frequency

700 Full clause

600 500

Semi-clause

400

Total

300 200 100

Five+

Four

Three

Two

One

0

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 fragmentation” 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 Unit type Single-IU clause Multi-IU clause Independent phrase Total

Number of clauses/phrases N %

Number of IUs involved N %

910 211 223

68 16 16

910 467 223

57 29 14

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 total 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 constituents. 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 substantive IUs produced by the Japanese conversational co-participants. That is, what this suggests is that the clause is the syntactic exponent of the Japanese substantive 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 applicability 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.

f.

samaa sesshon totta no yo.$ [PVP] summer session take-past fp fp ‘(I) took summer session (courses).’ 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 relationships 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 production 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 information statuses, six categories of NP grammatical roles, and eight categories of NP syntactic forms. In coding the IUs for the preferred information structure, 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/nonnew, 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 coparticipants) (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 nominals, 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 adjectives, 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

Single-IU clause Multi-IU clause Independent phrase TOTAL

Number of clauses

Average Average number of NPs number of NPs per clause per IU

Number of NPs

Number of IUs

884

910

910

0.97

0.97

358

467

211

0.77

1.70

175

233

____

0.75

____

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 of NPs

Number of IUs

[FC] [PVP] [PNP] [PAP] Total

390 297 185 12 884

263 392 174 81 910

1.48 0.76 1.06 0.15 0.97

[CFC] [CVP] [CNP] [CAP] Total

47 55 18 1 121

40 106 17 7 170

1.18 0.52 1.09 0.14 0.71

[CVPp] [CNPa] [CNPp] [CAPa] [CAPp] [CAvP] [CXP] Total

11 95 23 15 1 60 32 237

34 94 23 18 11 90 27 297

0.32 1.01 1.00 0.83 0.09 0.67 1.19 0.80

[VP] [NP] [AP] [AvP] [XP] Total

3 112 10 39 11 175

7 112 14 82 8 223

0.43 1.00 0.71 0.48 1.38 0.78

1,417

1,600

0.89

TOTAL

c.

buronkusu tte yuu tokoro ga aru no? 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?’

Average number of NPs per IU

[CFC]

 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 iand 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 clauseexternal 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 nonarguments. 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 A O

IU syntactic type

S

[FC] [PVP] [PNP] [PAP] Total

238 16 1

27 5

255

32

[CFC] [CVP] [CNP] [CAP] Total

36 4

5

40

5

25

52

15

4 28

[CVPp] [CNPa] [CNPp] [CAPa] [CAPp] [CAvP] [CXP] Total

14 146 1 5 166 25

Oblique

Non-argument NPs Predicate Independent nominal NP

59 115 5 6 185

52 15 178 1 246

–– –– –– ––

390 297 185 12 884

3 24 1 1 29

3 2 17

–– –– –– ––

47 55 18 1 121

6

1

–– –– –– –– –– –– ––

11 95 23 15 1 60 32 237 3 112 10 39 11 175

22

23

11 63

3 18

8 40

Total

15 1 60 10 92

24

[VP] [NP] [AP] [AvP] [XP] Total

–– –– –– –– –– ––

–– –– –– –– –– ––

–– –– –– –– –– ––

–– –– –– –– –– ––

–– –– –– –– –– ––

3 112 10 39 11 175

TOTAL

358 (25%)

55 (4%)

231 (16%)

306 (22%)

292 (21%)

175 (12%)

1,417 (100%)

 Chapter 5 400 350

Frequency

300 250 200 150 100 50

Independent NP

Predicate nominal

Oblique

O

A

S

0

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 Information S status given accessible new Total

189 (53) 28 (8) 141 (39)

Argument NPs A O 39 (71) 5 (9) 11 (20)

99 (43) 18 (8) 114 (49)

358 55 231 (100%) (100%) (100%)

Non-argument NPs Oblique Predicate Independent Total nominal NP 100 (33) 28 (9) 178 (58)

97 (33) 15 (5) 180 (62)

88 (50) 23 (13) 64 (37)

612 (43) 117 (8) 688 (49)

306 (100%)

292 (100%)

175 (100%)

1,417 (100%)





Chapter 5

Total

Independent NP Predicate nominal

new

Oblique

accessible given

O

A

S

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 nominal

Independent NP

Total

p-pro

15 (4) 47 (13) 152 (43) 34 (9) 35 (10) 31 (9) 32 (9) 12 (3)

15 (28) 4 (7) 21 (38) 5 (9) 5 (9) 1 (2) 4 (7) 0 (0)

0 (0) 22 (9) 122 (53) 22 (9) 18 (8) 28 (12) 15 (7) 4 (2)

10 (3) 11 (4) 182 (59) 34 (11) 21 (7) 43 (14) 5 (2) 0 (0)

2 (1) 7 (2) 144 (49) 19 (7) 36 (12) 52 (18) 30 (10) 2 (1)

2 (1) 12 (7) 91 (52) 18 (10) 16 (9) 17 (10) 14 (8) 5 (3)

44 (3) 103 (7) 710 (50) 132 (10) 131 (9) 172 (12) 102 (7) 23 (2)

358 (100%)

55 (100%)

231 (100%)

306 (100%)

292 (100%)

175 (100%)

1,417 (100%)

d-pro bareN dem+N N-no+N adj+N compNP nomVP Total



Chapter 5 800 700 600

Frequency

500 400 300 200 100

nomVP

compNP

adj+N

N-no+N

dem+N

bareN

pro-d

0

pro-p



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 expression 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 propensity 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) adjective + NPs tend to appear in predicate nominals (30%) or obliques (25%); and (g) complex NPs tend to appear in S roles (31%) or predicate nominals (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 N %

new N

Total

NP form

N

%

N

%

p-pro d-pro bareN dem+N N-no+N adj+N compNP nomVP

44 102 260 110 29 37 20 10

7 17 42 18 5 6 3 2

0 1 93 5 5 7 5 1

0 1 80 4 4 6 4 1

0 0 357 17 97 128 77 12

0 0 52 2 14 19 11 2

44 103 710 132 131 172 102 23

3 8 50 9 9 12 7 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 “Nonlexical 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 Nonlexical 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 Arole 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 produced the NPs as clause-internal arguments (45%) or non-arguments (43%), and as bare nouns (50%). Second, the speakers produced a roughly comparable 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 preferred 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 N= 729 (51%)

S OBL O

N=217 N=128 N=117

(30%) (18%) (16%)

bareN dem+N d-pro

N=353 N=115 N=103

(48%) (16%) (14%)

New N= 688 (49%)

PN OBL S

N=180 N=178 N=141

(26%) (26%) (21%)

bareN adj+N N-no+N

N=357 N=128 N= 97

(52%) (19%) (14%)

Total N=1,417 (100%)

Arguments S N=358 A N= 55 O N=231 Total N=644 Non-arguments OBL N=306 PN N=292 Total N=598 Independent NPs N=175

bareN adj+N others

N=710 N=172 N=535

(50%) (12%) (38%)

(25%) ( 4%) (16%) (45%) (22%) (21%) (43%) (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 G

547 483

34.3 30.3

86 50 20 12

5.4 3.1 1.2 0.7

7 5 3 2

0.4 0.3 0.2 0.1

G+N G+G N+G N+N G+G+N G+G+G G+N+G N+G+G 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

Frequency

400

300

200

100

N+G+G

G+N+G

G+G+G

G+G+N

N+N

N+G

G+G

G+N

G

N

Zero NPs

0

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 Ø-A Ø-A+O Ø-O Total

164 62 34 4 264

43 16 9 1 69

No associated NPs Adverbial phrase Verb phrase Adjectival phrase Mixed phrase Total

75 25 17 4 121

20 6 4 1 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 Oblique (Adv) Argument Oblique (Topic) Argument Total

+ + + +

Argument Oblique (Adv) Argument Argument

N

%

7 3 2 0

58 25 17 0

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 conversational interaction. The coded NPs include all the NPs contained in the IUs, viz. arguments (S, A, and O), non-arguments (obliques and predicate nominals), and clause-external independent NPs. Those NPs contained in the intraIU 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. nonreferential 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 nominal 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 type

Transitive high

Transitive low

Intransitive Adjectival predicate

Single-IU Ø clause A O A+O Total

22 6 61 4 93

Ø A O A+O Total

21 Ø 4 S 129 14 168 Total

Multi-IU Ø clause A O A+O Total

4 2 19 6 31

Ø A O A+O Total

1 Ø 4 S 52 8 65 Total

Clause Total

Ø A O A+O Total

26 8 80 10 124 (11%)

Ø A O A+O Total

159 Ø 107 S

266 Total

Nominal predicate

81 Ø 76 S

157 Total

Number of overt arguments

174 Zero 52 One Two

457 435 18

226 Total

910

13 Ø 47 S

6 Ø 14 S

11 Zero 24 One Two

35 162 14

60 Total

20 Total

35 Total

211

22 Ø 172 Ø 87 Ø 185 8 S 154 S 90 S 76 181 22 233 Total 326 Total 177 Total 261 (21%) (29%) (16%) (23%)

Zero One Two

492 597 32

Total

1,121 (100%)

 Chapter 5 350 300

Frequency

250 200 150 100 50

Nominal predicate

Adjectival predicate

Intransitive

Transitive low

Transitive high

0

Clause type

Figure 5.5. Frequency of five clause types

clauses. Of all the two-participant clauses which are the minority, low transitives 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 overwhelmingly consists of clauses at the lower extreme of the transitivity continuum— clauses unrelated to events or actions which typically express speakers’ feelings, 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 nonevent-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 transitivity. That is, conversational Japanese, and possibly conversational discourse in general, is preferentially composed of non-event- or non-action-related materials such as descriptions of circumstances, expressions of mental states and subjective opinions and evaluations. Thus, viewed in terms of transitivity, conversation 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 spontaneous conversation abounds in clauses of very low transitivity. This suggests that more attention should be directed to studies of the grammar of nontransitive, 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 high

26 (21)

––

8 (6)

80 (65)

10 (8)

124 (100%)

Transitive low

22 (9)

––

8 (4)

181 (78)

22 (9)

233 (100%)

Intransitive

172 (53)

154 (47)

––

––

––

326 (100%)

Adjectival predicate

87 (49)

90 (51)

––

––

––

177 (100%)

Nominal predicate

185 (71)

76 (29)

––

––

––

261 (100%)

Total

492 (44)

320 (29)

16 (1)

261 (23)

32 (3)

1,121 (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 majority (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 importantly, clauses with two overt arguments were produced remarkably infrequently by the speakers. That is, 91% of the transitives and 97% of the clauses contained zero or one overt argument. This suggests Japanese speakers’ notable 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 O

Adjectival 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 nominal. (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 A O One argument-Total

320 16 261 597

29 1 23 53

Two arguments (A+O) TOTAL

32

3

1,121

100

 Chapter 5 500 450 400

Frequency

350 300 250 200 150 100 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 proposed 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 reminiscent of Du Bois’ (1987) “One Lexical Argument Constraint”, which prohibits more than one lexical argument per clause. When we compare the two constraints 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 arguments. 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 indicates 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 Clause type Transitive high Transitive low Intransitive Adjectival predicate Nominal predicate Total

N

A %

O

N

%

____

18

14

____

30

13

N

Total %

N

%

90

73

108

44

203

87

233

50

154

47

____

____

154

47

90

51

____

____

90

51

76

29

____

____

76

29

320

42

661

45

48

13

293

82

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, Arole 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 nonnew (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 contain 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 Unit type

N

IUs/clauses with more than one new NP N %

IU Single-IU clause Multi-IU clause Clause

1,600 910 211 1,221

12 12 68 80

0.7 1.3 32.2 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%

Frequency

70% 60%

More than one new NP

50% One new NP 40% 30% 20% 10%

Clause

Multi-IU clause

Single-IU clause

IU

0%

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? everybody celebrate-prog nml tag ‘everybody is celebrating (Easter)’ c. +chanto. [CAvP] properly ‘properly’ (F2)

[CFC/A]

(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 Arole 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 contained two or three new NPs, in violation of the one new NP per IU constraint. In other words, if the speaker had not opted to express the proposition 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 produced the multi-IU clause, that is, divided or fragmented the clause into separate 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, considering 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 responsible 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 clausecentered 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 multiIU 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 nontransitive clauses without overt S (40%). The results have shown that Japanese conversational co-participants conform 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 communicating 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 multiIU 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 introducing 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 conversational 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 conversational 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 component 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 reexamine 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] [TX] [CG] [IT]

= = = =

ideational textual cognitive interactional

The ideational [ID] component conveys a full proposition or part of a proposition, participating in the content of the conversation. The ideational component may be comprised of a clause or a phrase or a word, each of which contributes 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 processing, 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.) Typically, 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 relationship involved in the co-production of conversational discourse. This component functions to show the speaker’s sensitivity towards the addressee, or concern for the hearer’s understanding. It functions to solicit the hearer’s involvement 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 ID TX CG ID TX ID ID CG

434 126 66 58 9

27.1 7.9 4.1 3.6 0.6

47 45 40 13 12 4 1

2.9 2.8 2.5 0.8 0.7 0.3 0.1

7 5 1 1

0.4 0.3 0.1 0.1

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

1,600

100

component types, and 4 four-component types. The 17 types are listed in descending 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 component 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 interactional components, 318 textual components, and 139 cognitive components. 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

Frequency

500 400 300 200 100

TX ID CG IT

CG ID CG IT

CG ID TX IT

TX ID TX IT

ID CG IT

TX ID CG

CG ID TX

TX ID TX

TX ID IT

CG ID IT

ID TX IT

TX ID

ID CG

CG ID

ID IT

ID TX

ID

0

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 immediately 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 tendency to use the ideational component to produce the IU type ID or ID-IT. Second, of the 581 interactional [IT] components, 89% appeared immediately 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 component 63% (vs. before the ideational component 37%) of the time. This indicates 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 clausefinal position.) More specifically, when the textual component was co-used with the ideational component in the two-component IUs, the textual component 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) I:

[ ID ] [ TX ] 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 production. Given that the cognitive component has to do with the speaker’s cognitive processing, and most typically, cognitive planning of ideational content 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 propositionconveying 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 components). Those IUs composed fully of the four components were rarely produced, 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 One Two Three Four Total

N

%

731 693 162 14

46 43 10 1

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 component. The textual and cognitive components appeared either immediately before or immediately after the ideational component. Given that the textual component concerns linkages as well as specific relationships between ideas or propositions expressed in the ideational component, and the cognitive component concerns the planning of ideas or propositions expressed in the ideational component, it follows that both of them need to occur adjacent 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, unlike 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 IUinitially 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 preideational (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, comprised fully of the four components. The components are placed in the following 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 twocomponent 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 composed 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, overwhelmingly, Japanese speakers produce IUs such that they perform “one function at a time”. This is reminiscent of previously formulated constraints on the IU informational composition such as Chafe’s (1987, 1994) “one new concept at a time”/“one new idea” constraint and Du Bois’ (1987) “one new argument” constraint, 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 followed 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.

b.

c.

d.

e.

f.

g.

h.

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).’ 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)’ +fuku toka. clothes and [ ID ] ‘clothes and so on’ 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.’ Y: kono mae ne, this before fp [ ID ] [IT] ‘the other day’ ->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’ ++hajimete.= first time [ ID ] ‘for the first time’ M: =animaru paaku∼ animal park [ ID ] ‘(Is it) the Animal Park?’



 Chapter 6

i.

j.

Y: waarudo animaru paaku tte shitteru? world animal park top know [ ID ] ‘(Do you) know the World Animal Park?’ +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 independent 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 direct 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 comprised 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 simplest functional structure type, namely, ID, without the other optional components serving as idea-linking devices. In Iwasaki’s (1993) study, if we collapse the ideational [ID] component and the language-specific cohesive [CO] component, 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 independently communicates a full proposition. The “marked” case of phrasality 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 information. 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 interactional information more frequently than he/she would be able to do so by producing 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 produced, 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.

i.

hyooka o. evaluation acc ‘(high) evaluation’ 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 simultaneously 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 governed 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 followed by the interactional component. That is, the substantive IU in conversational Japanese has been found to be predominantly unifunctional, or minimally 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 consists solely of the ideational component, without the other components serving a coherence-creating, regulatory, or interpersonal function, or as an idealinking 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 conversational 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 preferred 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 substantive 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 subjects 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 centrality 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 directly 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 conversation 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 propositional content, is important discourse functionally, and that it can be explained in terms of discourse-communicative, interactional, and information flow-related motivations (Matsumoto 1997b, 1998a). Another important finding discussed in Chapter 4 along with the robustness 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 conversational 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 speakers 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 conversational Japanese. Importantly, the coding revealed an approximately equal

Conclusion

number of overtly expressed Given and New NPs. This suggests that conversational 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 communicating ideational content. It was additionally found that one new NP was introduced 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 nontransitive, one-participant clauses with very low transitivity (typically intransitive or nominal predicate clauses). It was also shown that the speakers preferentially produced clauses with no overt arguments or one overt argument, 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 conversational Japanese and the “one overt argument per clause constraint”, I argued 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 nominal 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 conversational 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 occurrence 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 motivating 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 incorporated into one substantive IU, the speakers preferentially used only one component (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 component followed by the interactional component ([ID] [IT]). Given the notable prevalence of [ID] in the data, I have further suggested that speakers communicate ideational information most typically by using the simplest functional 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 component 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 semiclausal, 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 Japanese 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 conversational Japanese are limited by five major constraints. The syntactic structure 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 production 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 general, interactively co-construct conversational discourse, with strict cognitive constraints operating on their processing capacities, relying on the immediate 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 coparticipants 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 minimal 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 information 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 attention 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 smallest idea-conveying discourse units will cluster into larger thematically coherent 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 multiparty 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 production of phrasal IUs. Additionally, I have suggested in Chapter 6 (Section 6.4) that another factor facilitating the production of phrasal IUs in Japanese conversation is the interactional motivation on the part of the speaker. By dividing 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 exploration 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 precision 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 equipments 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 scrambled 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 “segmentative” 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 speakerhearer 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 expressions (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 “basegenerated”, “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 independent 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 postpredicate 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 crosslinguistic 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: This study Du Bois

S 39 22

A 20 3

O 49 25

Oblique 58 39

Total 49 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 functional 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 complex 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 rightdislocations 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 crosslinguistic 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 SocialSemiotic 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 selfcorrection 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 Multidisciplinary 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 A See A role A role 41–42, 103–105, 108–117, 124–125, 127–134, 136–137, 161, 175–176 See also Transitive subject Abelson, Robert P. 170 Accelerated syllable 21 See also Anacrusis Accent 17, 23–24, 171, 173 See also Primary; Secondary accent Accent unit (AU) 23–25, 171, 173 defined 23–24 substantive vs. regulatory 171 Accessibility 9 Accessible information/NP 15–17, 19, 24–25, 40, 81, 100, 102–103, 111–112, 115 Acknowledgment 71–72, 87, 91 Acoustic analysis 167 Acoustic unit 2, 21 See also Declination unit Action 10–12, 105, 126–127, 170 Activation cost 9, 14–19, 41, 115, 171 and definiteness 18 and identifiability 18–19 defined 16 expression of 17–18, 115 perspectives on 16–17 See also Given-accessible-new distinction Activation states 2, 15, 24, 39, 115 See also Active; Inactive; Semi-active state Active consciousness 23, 26, 102 Active state 2, 15–17, 21–22, 39, 81, 138, 170, 176 focally vs. peripherally 15, 176

Added information 25 See also Starting point Addressee 13, 34, 35, 143 Adjectival IU 31, 53, 83, 89, 91 Adjectival/adjective phrase (AP) 31, 52–53, 55, 58, 61, 63, 65–66, 76, 83, 89–91, 103, 107, 109, 121, 175–176 See also Attributive; Predicate AP Adjectival predicate 52, 57, 62, 67, 99, 103, 105, 112, 121–122, 124–125, 127, 132, 177 Adjectival predicate clause 62, 99, 105, 121, 125–129, 132–133, 149, 177 Adjective 18, 38–39, 53, 63, 65, 91, 104–105, 108, 115, 171, 174–175 See also Attributive; Predicate adjective Adjective + NP 104, 108, 113–115, 117 Adjunct 54, 57, 177 Adverb/adverbial 38, 41, 54, 63, 89–92, 102, 105, 110–111, 173–175 See also Sentential adverb Adverbial/adverbial phrasal IU 31, 53, 60, 89–90, 96, 100, 118, 121, 136, 147, 154 Adverbial phrase 52–53, 56–58, 61, 63, 65–66, 82, 85–86, 89–92, 97, 99–100, 102–103, 107, 109, 121, 123, 149, 173 Affect 34, 37, 90 Affirmation 70, 72, 87, 170 Agent 103, 134, 170 Agreement 46, 71–72, 75, 87 Anacrusis 21 Anaphora 9 See also Zero anaphora Anaphoric zero, vs. non-anaphoric 124 Andersen, Roger W. 46 AP See Adjectival phrase

 Index

Argument 5, 8–9, 33, 41–42, 52–54, 56, 60, 69–70, 79, 89, 96, 101, 103, 105, 108–111, 116–118, 120–125, 127–134, 138–139, 147, 155–156, 160–161, 163–165, 171–173, 176 defined 103 overt vs. null 132–134 types and definitions 103 vs. non-argument 103–104, 110, 116, 118, 138–139, 160, 176 vs. predicate 53, 56, 69–70 See also A role; Core argument; Core-oblique distinction; O role; Preferred Argument Structure; S role Argument slot 118, 120, 132, 161 filled vs. unfilled with overt NP 132–133 Argument-oblique combination 124 A-role argument See A role Article 11, 18–19 See also Definite; Indefinite article Ashby, William J. 3, 42, 71 Aspect 12, 105, 170 Assessments 127 Assumed familiarity 14 Atkinson, J. Maxwell 46 Attention focus 27–28, 170 Attitudes 126–127 Attributive adjective/AP 18, 53, 55, 65, 89, 103, 173, 176 vs. predicate adjective/AP 53, 89 AU See Accent unit Auditory unit 2 B Backchannel xvi, 32, 87, 172–174 defined 172 in narrative 173 in spoken Japanese 32 types xvi See also Continuer; Reactive token; Turn-internal listener backchannel

Background 10–12, 15, 21, 25, 39, 127, 166, 170 defined 10–11 vs. foreground 10–12, 170 Backgrounding 11–12, 65, 165, 170 Bantu languages 39 Bare noun 104, 108, 113–117, 139, 160 Base-generated topic See Genuine topic Baseline pitch level 21 Basic word order 3–4, 63, 169 vs. marked word order 4, 63 See also Canonical; Unmarked word order Beckman, Mary E. 28–29, 46 Bentivoglio, Paola 3, 42 Birner, Betty J. 1, 14, 169 Boden, Deirdre 166 Bolinger, Dwight 46 Boundary, IU 2, 20–21, 46, 143, 150, 171 marker/signal of 20, 46, 143, 150 Brand-new information 14 See also Evoked; Unused information Brazil, David 20 Brown, Gillian 17–18, 30, 172 Bybee, Joan L. 161 C Canonical word order 4, 164 Case marker 104, 169 See also Case-marking particle Case-marker drop phenomenon 169 Case-marking particle 33, 147, 169, 173 Casual conversation/spoken discourse 51, 169–170, 173, 176 CD See Communicative dynamism Chafe, Wallace L. 1–3, 7, 9, 13–29, 36, 39–42, 46, 48–49, 54, 62, 83, 97, 99, 102, 113, 120, 122, 124, 137, 142–143, 151–152,159–160, 164, 166, 170–172, 175–176 Chesterman, Andrew 18 Chinese See Mandarin Choice, speaker’s 1, 65, 165 Choice question 72

Index 

Clancy, Patricia M. 3–4, 28, 32, 63, 65, 74, 87, 97, 132, 159–160, 164, 174 Clarification request 71–72 Clark, Herbert H. 13, 18 Clausal IU 3, 26, 29, 32–33, 37, 51–53, 55, 57–62, 64–69, 71, 80–81, 85, 87, 89, 98–100, 106, 121–122, 137, 147, 154–157, 159–160, 171–172, 175, 177 defined 51 robustness of 160 types and distribution 52, 59 vs. phrasal IU 3, 36–37, 57–60, 62, 155–157, 172 See also Full clausal; Independent clausal; Semi-clausal IU; Single-IU clause Clausal IU as part of multi-IU clause 52, 56, 58–60, 64–66, 69, 89 definition and subtypes 52 Clausal IU strategy, vs. phrasal 33 Clausal object 52, 57, 104–105, 124, 174, 177 Clausal structure 53–54, 57–58, 64, 71–72, 98, 159, 174 Clausal subject 104–105, 174 Clause 2–3, 8–9, 12, 20, 25–27, 29–33, 37–38, 41–42, 51, 53–54, 62–63, 71–72, 74, 79, 94–95, 97–99, 101, 105–106, 117–118, 121–132, 134, 137–139, 141–142, 144, 152, 156–163, 165–167, 172, 176 and IU in English 2–3, 25–27 and IU in Japanese 3, 29–32, 94–98 and NP 105–106 as basic unit of discourse production/ information flow/processing 2–3, 27 as exponent of IU 26, 51, 63, 99, 159, 163 breakup/division of 32, 37, 101, 139, 141, 156–158 defined 53–54, 159 English vs. Japanese 33

tight vs. loose internal structure of 33 types and definitions 105 vs. phrase 29–33, 51–53, 57–60, 98, 172 See also Adjectival predicate; Full; Intransitive; Multi-IU; Nominal predicate; One-participant; Semi-; Single-IU; Transitive; Two-participant clause Clause centrality proposal 3, 26, 29, 33, 62, 99, 152, 159 defined 25–26, 152 Clause chaining 27 Clause core 41, 122–123 Clause vs. phrase centrality 30, 51, 99 Clause-external NP 103–104 See also Independent NP Clause-external postposing 64, 84–85, 87–88, 165 See also Postposed NP Clause-internal NP 81, 103, 106, 108, 110 See also Argument; Oblique Clause-internal postposing 64–65, 85, 88, 165 Code 35, 44, 48, 51–54, 104, 142–144, 147, 155 Coding 31, 46, 54, 62, 101–102, 121, 123, 141, 143, 155, 160 Coding categories 48, 101 Coding scheme 32–33, 142, 144, 175 Cognitive activation, vs. search 8 Cognitive component 142–143, 145–146, 148–152, 162 Cognitive connective 143, 150 Cognitive constraint/limitation 23, 36, 163–164 Cognitive cost 16, 138 See also Activation cost Cognitive function 143–144, 151 Cognitive planning See Planning Cognitive process/processing 15, 143, 148 Cognitive subtype 22, 49, 91, 143, 173 Coherence 7, 10, 33, 38, 143, 158, 162 Cohesion 34–35, 142

 Index

Cohesive component 34–37, 141–142, 151, 154–155 Cohesive function 142 Cohesive information 3, 36–37 Collaborative completion/finish 71, 76–77, 79, 88, 91, 93, 174–175 Comment 83 See also Genuine topic Common noun 18 Communication 22, 32, 34, 36–37, 57–58, 64, 71, 79, 81, 87–88, 156, 158–160, 166, 170–171 See also Conversational; Spoken communication Communication breakdown 174 Communicative dynamism (CD) 25, 57, 171 defined 171 Communicative goal 11 Competitive referent 171 Complement 176–177 Complete proposition 31, 52, 54–55, 57, 60, 66, 71, 74, 79, 81, 88, 92–94, 99, 106, 122, 154, 175 Complete propositional strategy 37 Complex clause 174 See also Clausal object; subject; Embedded; Subordinate clause Complex NP 70, 83, 93, 104, 108, 113–115, 131, 174, 177 See also Nominal complement; Relative clause Complexity, syntactic 177 Comprehension 32, 49 Comrie, Bernard 103 Concept 2, 15, 18, 25, 39–41, 102–103, 122–123, 137–139, 152, 162, 164–165, 170, 176 Conceptual unitariness 40–41, 102 Conjunction 27–28, 49, 52, 141–144, 148–149, 151, 173, 177 See also Coordinating; Subordinating conjunction Conjunctive See Conjunction Connected discourse 2, 5, 7, 27, 38–39, 65

Connective 27–28, 144, 148, 150–151 See also Cognitive; Textual connective Consciousness 2, 13–15, 17, 21–26, 39, 81, 86, 100, 102, 164 and IUs 21–22 focal vs. peripheral 13, 15 focus of 2, 15, 21–22, 25, 39, 164 See also Active consciousness Constituent 4, 24, 31, 33, 98, 103, 137, 169, 173–174 Constraints 2–3, 22, 24–25, 38–39, 41–42, 102, 119, 121–123, 131, 137, 139, 152, 160, 163–164, 177 on information flow 38–42 on the flow rate of new information/ new information quantity 2, 39–42, 137 PAS 41–42, 123, 161 See also No more than two functional components per IU; No more than two IUs per clause; One new NP per IU; One overt argument per clause; One overt NP per IU constraint Constructed example 169 Content See Ideational; Propositional content Content, of thought 166 Content words 41 Context 8, 13, 18, 22, 25, 28, 71–74, 76, 79, 87, 92, 102, 151, 163–165, 169, 171, 173 anaphoric vs. cataphoric 8 Contextual salience 18–19 Continuer 87, 172 Continuing intonation 20, 27–28, 46, 173 Contrastive referent 9, 171 Conversation See Casual; Dialogic, Dyadic; English; Everyday; Face-to-face; Female; Informal; Interactive; Japanese; Male; Multi-party; Naturally occurring; Ordinary; Spontaneous; Telephone conversation

Index 

Conversation analysis 173 Conversational acts 76, 88 Conversational communication 159, 161 Conversational co-participants See Co-participants Conversational discourse 48, 51, 77, 123, 126, 131, 139, 143–144, 151, 158, 161, 163, 166–167 low transitivity/non-event/nonaction-oriented nature of 126–127, 161 Conversational English 27, 62, 113, 172 Conversational interaction 5, 11, 30, 49–50, 124, 127, 155, 161, 164, 167 Conversational Japanese 5, 32, 43, 51, 57, 62–63, 65, 88, 94, 96–97, 101, 106, 110–111, 113, 115–117, 119, 122–124, 126, 128, 131, 133–134, 138, 141, 152, 154, 158–163, 166 Conversational language 49, 57, 87, 127, 172 Conversational monologue 27 Conversational narrative 40 Conversational space 10, 118 Cooperative principle 13 Coordinate sentence 177 Coordinating conjunction 28, 141 Co-participants 18, 22, 30, 32, 48–49, 54, 61, 69, 73–76, 79, 88, 92, 99, 102, 106, 117, 137, 139, 143, 152, 158, 163, 166, 169, 173 Copied postposed NP 55, 72, 85, 88 Copula 41, 62, 105, 173 See also Adjectival predicate; Linking verb; Nominal predicate Copula + PP/NP construction 41 Core argument 41, 52–54, 103, 120–121, 171, 176 Core grammatical role 103, 113 See also A; O; S role Core-oblique distinction 113, 134 Corpus 41, 67, 88, 105, 134 Couper-Kuhlen, Elizabeth 3, 19 Courtroom discourse 166

Croft, William 3, 19, 72, 88, 158, 167, 175, 177 Cross-linguistic applicability/validity 29, 42, 88, 99, 159 Cross-linguistic comparison 62, 144, 172, 176 Cruttenden, Alan 2, 20–21, 46 Crystal, David 2, 20

D Data analysis 48 Data base 44–45 Data coding 51, 101, 141 Data transcription 46 Declarative clause 46 Declination unit 2 Definite article 11, 18–19 Definite NP 7, 9, 17 vs. indefinite NP 7, 17 Definiteness 1, 9, 18 See also Identifiability Demonstrative adjective + NP 104, 113–115, 117 Demonstrative pronoun 18, 104, 113–115, 117, 176 Derived topic, vs. base-generated 174 Descriptive function 34 Detached NP IU 56, 69, 71, 88, 90, 159–160, 175 See also Independent phrasal NP IU Di Cristo, Albert 29, 46 Dialogic conversation/interaction 5, 157 Dik, Simon C. 57 Direct object 9, 31, 55, 103, 105, 123, 126, 147–148, 154, 169, 176 vs. indirect object 9 Direct object marker 169 Disagreement 71–72 Discourse See Conversational; Spoken discourse Discourse analysis 167 Discourse context See Context Discourse functions 65, 160, 175 Discourse marker 22

 Index

Discourse production 2, 5, 19, 32, 39, 48, 81, 88, 138, 148, 164 Discourse unit 2, 31, 42, 54, 138, 159, 163, 166 See also Clause; Intonation unit Discourse-functional studies 3, 169 Discourse-new, hearer-new information 14 See also Brand-new information Discourse-new, hearer-old information 14 See also Unused information Discourse-old, hearer-old information 14 See also Evoked information Discourse-pragmatic factors 72, 88 Displaced mode 49 Distance 8, 74, 177 See also Referential distance Ditransitive clause/verb 176 Dixon, Robert M. W. 54, 103 Doctor-patient interaction 166 Dorgeloh, Heidrun 169 Downing, Pamela 169 Drew, Paul 166 Du Bois, John W. 1–3, 18–21, 25, 41–42, 46, 48, 54, 102–103, 111–113, 116–118, 122–124, 130–131, 133–134, 137–138, 152, 160–161, 171, 175–176 Duranti, Alessandro 169 Durie, Mark 71, 88 Dyad 44–45, 47–50, 174 Dyadic conversation 44, 46, 166 E Edwards, Jane A. 46 Efficiency 167 Elaborating NP 175 Elicited narrative 41, 172 Embedded clause 20, 121, 124, 177 Emotions 171 Emphasis 65, 85, 88 Encoding 27, 166 English 3, 12, 18, 20, 23, 29–33, 37, 42, 46, 62, 112, 116–117, 120, 137, 148, 171–173 See also Conversational; Spoken English

English clause 33, 177 English conversation 10, 29, 32, 113, 172, 176 English IU 3, 29–30 English speakers 25, 28–29, 33, 37, 157, 160 English spoken discourse 25 Entity 14, 17–18, 33, 102, 170, 174–175 Episodic memory 10, 38 Epistemic stance See Stance Evaluation 49, 126 Evaluative subtype See Validational subtype Event 3, 11–12, 22, 26–27, 41, 49, 54, 102, 126–127, 143, 152, 159, 171, 175 Event idea 26, 175 Everyday conversation 127, 165, 172 Evoked information 14, 102 See also Given information Experiential function 33 Explicit NP/subject See Overt NP/ subject Expressive function 34 Extra-clausal NP See Clause-external NP

F Face-to-face conversation 35, 44 Falling intonation 20, 27, 46 False start 49, 78, 80 Feedback 32, 157 Feelings 126–127 Female conversation, vs. male 44–45, 47–48, 166, 172 Figure-ground distinction 170 Filled pause See Pause filler Final/IU-final particle 49, 70, 150–151 Finality, vs. non-finality 46, 171 Finnish 97 Firbas, Jan 25, 57, 171 First person 170, 176 Floor, of conversation 34, 143, 174 Flow, of conversation 34, 65, 143 Ford, Cecilia E. 19, 161

Index 

Foreground 10–12, 170 defined 10–11 Foreground-background alternation 11, 26, 65 Foreground-background distinction/ continuum 10–12, 170 Foregrounded clause, vs. backgrounded 12 Foregrounding, vs. backgrounding 12, 170 Foregrounding device, vs. backgrounding 11–12 Fox, Barbara A. 1, 10, 13, 33, 65, 169 Fragmentary IU 22, 24, 49–50, 78 defined 22, 49 Fragmentation 29–30, 32, 96, 160, 172 in English vs. Japanese 29–30, 32–33, 172 in Japanese and Mandarin 171–172 in spoken Japanese 32 Fragmentedness 30 Fragments, discourse 88, 159 Free clause 11 French 42, 170 Frequency 4, 8, 37, 158 Fujii, Noriko 169 Full clausal IU 53, 58, 80–81, 85, 99–100, 110, 122, 159 independent 58, 110 vs. semi-clausal IU 66–67, 94 Full clause 29, 31, 52, 54–56, 59, 62, 66–67, 69–70, 78, 85, 89, 91, 94–97, 99–100, 106, 108, 122, 136, 171, 175 defined 52, 54 vs. semi-clause 62, 66–67, 94–95 Full NP, vs. pronoun 9, 17, 117, 171 Full proposition 31, 35, 56–57, 74, 77, 142, 147–148, 155, 174–175 Function words 41 Functional components 5, 34–36, 141–142, 144–145, 147, 149–152, 155, 158, 162–164 defined 34 linear order of 5, 35, 141, 150–152

obligatory vs. optional 150–151, 162 preferred number of 141, 149–150, 155, 162 types and definitions 142–143 See also Cognitive; Ideational; Interactional; Textual component Functional linguistics 127 Functional sentence perspective 171 Functional structure, IU 3, 5–6, 35, 43, 48, 141–142, 151, 158–159, 163, 177 orderly nature of 3, 35, 151, 162 Functional structure types, IU 36, 144–149, 151–155, 158, 162 distribution of 144–146

G Gee, James P. 166 Geluykens, Ronald 71 Gender differences 166 Generic referent 120, 171 Genitive case 104, 173 Genre 11–12, 30 Genuine topic 54, 57, 72, 81, 83, 100, 174–175 defined 174 Gesture 28, 163–164 Given A Constraint 41, 113, 116, 133, 161 Given followed by new 12, 25 See also Unmarked information structure Given information/NP 10, 13–17, 19, 24–25, 38–40, 43, 79, 81, 87, 100–103, 111–113, 115–120, 123–124, 131, 133–134, 139, 160–163, 165, 171, 177 defined 12–14, 102 constraint on the flow rate of 38–39 Given-accessible-new distinction 9, 15–16, 41, 171 See also Activation cost Givenness 9, 12–15, 170

 Index

ways of establishing 170 See also Accessibility; Newness Given-new contract 13 Given-new distinction 12–14, 144 See also Predictability; Recoverability; Saliency; Shared knowledge Givón, Talmy 1–2, 7–11, 13, 18, 27, 38–39, 41–42, 54, 117, 122, 137, 165, 170, 175 Grammar 1, 8, 11, 127, 161 Grammatical categories 53, 58, 61, 98, 175 Grammatical relation 103 Grammatical role 101, 103–104, 109–117, 138–139, 164, 176 and information status 111–113 and NP syntactic form 113–115 types and definitions 103–104 See also Argument; Independent NP; Oblique; Predicate nominal Green, Georgia M. 169 Grice, H. Paul 13 Grimes, Joseph E. 2 Grounding 1, 10–12, 38 and foreground-background distinction 11–12 and given-new information 10–11 and transitivity 12 defined 10–11 See also Background; Foreground H Haiman, John 27 Halliday, Michael A. K. 1–2, 12–13, 17–19, 24–27, 33–34, 48, 120, 141–142, 165, 172 Hasan, Ruqaiya 13, 34 Haviland, Susan E. 13 Hawkins, John A. 1, 18 Hayashi, Makoto 77 Hearer, and/vs. speaker 2, 10–15, 22, 46, 73, 79, 86, 88, 93, 102, 143, 154, 165, 170, 174–175 See also Listener Heavy added information constraint 25 Hebrew 42

Helasvuo, Marja-Liisa 3, 88, 97 Heritage, John 46, 166 Hesitation 20 High focus, vs. low 170 High transitive clause 105, 125–129, 132–133 Hinds, John 4, 54, 63, 165–166, 169 Hirschberg, Julia 19, 47 Hirst, Daniel 29, 46 Hoji, Hajime 3 Hopper, Paul J. 1, 3, 10–12, 42, 105, 125–127, 161, 170, 176 Horie, Kaoru 113 Human agent/subject 134, 170 Hypotaxis 33 Hypothesis 38, 43, 51, 101, 141, 162–163 I Iconicity 170 Idea 16–17, 19, 22, 24–29, 41–42, 49, 57, 122, 138, 142–143, 150, 158, 164, 170, 175 subcategorization of 41 See also Event; Referent; State Idea unit 171 Idea-conveying IU 36, 146, 163, 166 See also Substantive IU Ideation 36, 155 Ideational component 34–36, 43, 141–143, 145–152, 154–155, 158, 162–163, 177 Ideational content 31, 50, 57, 148, 154, 156, 158, 161, 163, 172 Ideational function 33, 141–142, 144 Ideational information 3, 22, 30, 32, 36–37, 151–152, 155–157, 162 Identifiability 9, 14, 18–19, 102, 171 and activation cost 19 defined 18 properties of 18–19 See also Contextual salience; Definiteness; Sharedness; Sufficiently identifying verbalization

Index 

Identifiable referent 18–19, 81, 100, 102 vs. non-identifiable referent 19 Immediate mode 49 Inactive state 2, 15–17, 39, 42, 138 Indefinite article 11 Indefinite NP 7–8, 17 Independent adverbial IU 89–92, 100 types and distribution 90 Independent clausal IU 51–52, 55, 58–60, 62, 66, 89, 98–99, 106, 110–111, 137, 154, 174–175 definition and subtypes 52 vs. clausal IU as part of multi-IU clause 58, 66, 89 vs. independent phrasal IU 54–55, 58 Independent NP 54, 71–72, 79, 81, 83, 103–104, 108–114, 117, 124, 175 See also Clause-external NP; Non-argument Independent phrasal IU 51, 53, 55, 58–60, 65, 71, 89–90, 93, 98, 103–104, 113, 173 definition and subtypes 53 vs. phrasal IU as part of multi-IU clause 52–53, 58, 65–66, 89 Independent phrasal NP IU 54, 70–88, 111, 159 functions of 88 types, definitions and distribution 71–72 See also Detached NP IU; Lead; Left-dislocated; Postposed; Repeated; Right-dislocated; Stray; Topic NP Independent semi-clausal IU 58, 60–61, 67, 98, 106, 110–111, 123, 162 Indirect object 9, 176 Inferrable information 14–15, 18 Informal conversation 26, 79, 88, 109, 160, 176 Informal form 54, 70, 177 Information block 2

Information flow 1–3, 5, 7, 19–20, 22, 25, 38–39, 49, 88, 120, 160, 164–165 basic/canonical/unmarked pattern of 25, 120 constraints on 38–42 defined 1 units of 1–3, 5, 19–20, 38, 164 Information focus 25 Information pressure 117–118 Information pressure quotient (IPQ) 117 Information processing 2, 27, 39, 164 Information status 1, 12–18, 81, 88, 100–102, 104, 111–112, 115–117, 139, 160–161, 164–165, 175 and grammatical role 111–113 and NP syntactic form 115 as property of nominal reference 102 See also Accessible; Given; New information Information/informational structure, IU 3, 5–6, 23–25, 43, 48, 81, 97, 101, 118–119, 139, 144, 152, 159–160, 163 in English 23–25 in Japanese 118–120 See also Unmarked information structure Information structure types, IU 118–120, 139 distribution of 118–121 Information unit 2, 25–26, 172 Informational factor, and multi-IU clause 158 See also One new NP per IU constraint Informational hybrid 38 Information-flow factors 1, 72, 88 Infrastructure, of thought 166 Initial pause 15, 20, 75 Insertion/inserted sequence 74, 78 Institutional talk 166 Intensifier 31, 118

 Index

Intention, speaker’s 21, 34, 65 Interactants 11, 18, 37, 47, 67, 83, 86, 90, 161, 166 Interaction 19, 22, 32, 37, 77, 79, 85, 87, 127, 143 See also Conversational interaction Interactional component 34–36, 43, 142–143, 145–148, 150–152, 154–158, 162–163 Interactional factors 72, 88 Interactional function 143–144, 162 Interactional information 3, 32, 36–37, 151–152, 155–158 Interactional involvement, co-participants’ 76, 88 See also Collaborative completion; Interruption; Repair Interactional motivation, and multi-IU clause 158, 166 Interactional NP 175 Interactional particle 31, 33, 35, 37, 75, 143, 147, 150, 155, 157, 166, 177 as marker of IU boundary 150 See also Final particle Interactional subtype 22, 49, 173 Interactive conversation 29, 48, 79, 117, 157, 161 Interlocutor 1, 10, 14, 49, 73, 87, 91, 157–158, 167, 172 Interpersonal function 33–34, 147, 158 Interrogativity 46 Interruption 49, 77, 88 See also Other-; Self-interruption Intonation 4, 27–28, 46, 165, 167, 171 functions of 46, 165, 171 See also Continuing; Falling; Rise-fall; Rise-fall-rise; Rising intonation Intonation contour xvi, 2, 20–21, 28, 46, 75, 169, 171, 173 acoustic measurement of 169, 173 as signal of IU boundary 20, 46 coherent/unified 2, 20–21, 46 types xvi, 46 Intonation group 2, 20

Intonation unit (IU) xv, 1–3, 5, 15, 20–23, 25–28, 33, 37, 43, 47–49, 51, 57, 60, 62, 66, 68, 71, 89, 94, 101–102, 105–106, 117–118, 124, 134, 137–138, 141, 144, 146, 152, 156, 159–161, 163–164, 167, 171, 173 and AU 23–24 and clause 2–3, 25–27, 29–32, 62, 94–98 and consciousness/language production 21–22 and NP 105–106 as basic unit of discourse production/ information flow 1, 5, 19, 164 defined 20 identification criteria 20–21 linkage 25–28, 164 prosodic features of 21 singleness/unitariness of 144, 164 types and size of 22–23 See also Functional; Informational; Syntactic structure Intonational phrase 20 Intra-IU clause 57, 104–105, 124, 174, 177 Intransitive clause 105, 121, 124–129, 131–133, 161, 176 Intransitive subject 41, 103, 132 See also S role Intransitive verb 103, 124 Inversion 169 IPQ See Information pressure quotient Irrealis 171 IU See Intonation unit IUs with zero NPs 118–121 types of 118, 120–121 See also No associated NPs; Zero anaphora Iwasaki, Shoichi 3, 28–32, 34–37, 42, 46, 48, 62, 67, 97, 99, 116, 127, 130, 141–144, 150–151, 154–155, 159–160, 162, 171–172

Index 

J Japanese 3–4, 9, 17, 29–32, 34, 37, 46, 62, 100, 116–117, 120, 130, 134, 137, 142–144, 148, 152, 157, 169–173 as SOV/verb-final language 130 Japanese clause 33, 105, 161, 166 Japanese conversation 5, 29, 44, 48, 69, 96, 130, 134, 144, 152, 158–159, 161, 166, 169, 176 Japanese intonation unit (IU) 3–5, 28, 30, 33–37, 42, 48, 51, 60, 99–100, 118, 142, 144, 146, 149, 151, 153, 155, 158–160, 163 exponent of 51, 63, 99, 159, 162–163 multifunctionality of 33–38 simple/non-elaborate nature of 160, 162–163 unifunctional nature of 152–155, 158 Japanese speakers 4–5, 32–33, 37, 51, 81, 97–98, 109, 115, 118, 124, 128, 139, 149, 152, 155, 157, 159–163 Japanese spoken discourse 3–4, 19, 35, 42, 102, 129 Jones, L. B. 11 Jones, L. K. 11 Judgment 22, 34, 49, 173–174 K Kärkkäinen, Elise 3, 42, 123 Kinesis 105, 170 Kreckel, Marga 20 Kuno, Susumu 3–4, 13, 65, 130, 165, 169 L Labov, William 11 Ladd, D. Robert 19, 46 Lambrecht, Knud 1, 161, 175 Lampert, Martin D. 46 Language 1, 17, 21, 33–34, 127, 164 See also Conversational; Spoken language

Language production 21, 42 Language universal 113 Latching xvi LDed NP See Left-dislocated NP Lead component 34–35, 142–143, 151 Lead NP 71–72, 78, 80–81, 87–88, 92, 159, 175, 177 definition and types 71 Left-dislocated (LDed) NP 7, 9, 31, 71–72, 81–83, 159, 175 Lengthening 21, 173 Lerner, Gene H. 74 Lexical argument 42, 130–131, 161 Lexical NP 114–116, 131, 134, 176 vs. non-lexical/pronominal NP/ pronoun 114–116, 131, 134 Lexicalized phrase 41 Light starting point constraint 25 Light subject constraint 25, 113 Lightness 25 Linkage, IU 2, 22, 25–29, 49, 143, 150, 164 as clause linkage 2, 26, 29 connective vs. non-connective type 27–28 non-elaborate nature of 28, 164 Linking, IU See Linkage Linking device, IU/idea 28, 92, 155, 158, 162 elaborate vs. non-elaborate 28 Linking verb 62 See also Copula Listener, and/vs. speaker 10, 13–19, 21–22, 32, 35, 42, 46, 86, 157, 172, 174 See also Hearer Listener backchannel See Backchannel Listener response 71–72, 87, 91 Listener-oriented perspective, on activation cost 16–17 Listener-orientedness 14 Logical function 33 Longacre, Robert E. 11 Loose internal structure, of Japanese clause 33, 166 Loud talk xv Loudness 173

 Index

Low transitive clause 105, 125–129, 131–133 M Main clause 10, 52, 57, 121, 170, 177 vs. subordinate clause 52, 57, 121 Makino, Seiichi 177 Male conversation, vs. female 44–45, 47–48, 166, 172 Mandarin 9, 17, 81, 88, 105, 171–172 Marked word order 4, 54, 63, 169 vs. basic/unmarked word order 4, 54, 63 See also Postposing; Scrambling Marshall, Catherine R. 18 Matsumoto, Kazuko xi, 4, 51, 63, 101–102, 141, 159–163, 165–167, 169–170, 172 Maynard, Senko K. 3, 28, 30, 32, 65, 97, 127, 155, 159–160, 172 Metafunction 33–34 See also Cohesive; Descriptive; Experiential; Expressive; Ideational; Interactional; Interpersonal; Logical; Subjective; Textual function Mithun, Marianne 169, 176 Mixed phrasal IU 56, 89, 93 Mixed phrase 53, 63, 174 Modifier 18, 89, 91, 114 Mori, Junko 19 Multifunctionality 36, 144, 151, 155 Multi-IU clause 51–58, 60, 64–66, 69, 71, 73, 85, 89, 94, 96–98, 101, 108, 124–125, 128, 134–139, 141, 147, 149, 154–158, 160–161, 164–166, 174–175 and coding of interactional information 155–158 and Japanese vs. English syntax 33, 157–158 and multifunctionality 155–158 and one new NP per IU constraint 134–138, 157, 161–162, 166 and post-predicate phrase 65, 164–165, 175

and syntactic factors 158, 177 full vs. semi- 56–57, 94–97 marked production of 139, 158, 160–161, 164, 166 vs. single-IU clause 94–98, 124–125, 128, 134–135, 138, 160 Multi-IU/unit turn 83, 172 Multi-party conversation 166 Multiple IUs, one clause strategy 97

N Narrative 3, 11–12, 30, 32, 35, 40–41, 48, 88, 111–113, 117–118, 127, 172–173, 175–176 vs. conversation 30, 173–174 See also Conversational narrative Narrative skeleton 11 Native speakers 1, 4, 29, 44 Natural vs. elicited discourse 30 Naturally occurring conversation/ discourse 4–5, 17, 130, 169 Nespor, Marina 20 New information/NP 2, 10, 12–17, 19, 24–25, 38–43, 88, 100–103, 111–113, 115–121, 123–124, 131, 133–139, 152, 157–158, 160, 165, 170–171, 174–177 and cognitive cost 16, 138 and IU/clause 134–138 constraint on the flow rate/quantity of 39–42, 137 defined 12–14, 103 integration of 38–39 See also Brand-new; Unused information Newness 9, 12–15 News interview 166 No associated NPs 120–121, 136 No more than two functional components per IU constraint 149, 152, 163 No more than/up to two functions per IU constraint 36, 155 No more than two IUs per clause constraint 95, 163

Index 

Nominal 5, 39, 63, 102–105, 111, 116, 118, 123, 144, 161–162, 164–165, 174 See also Noun; Noun phrase Nominal complement clause 104, 174 Nominal information 43, 100, 117–119, 152, 157, 161, 163 Nominal phrase See Noun phrase Nominal predicate 39, 52, 55, 58, 67, 69–70, 99–100, 103, 105, 112, 124–125, 127, 131–132, 147, 151, 161, 173, 177 Nominal predicate clause 62, 99, 105, 121, 124–129, 131–133, 147, 151, 161 Nominal reference 5, 25, 102, 137, 160 Nominal referent See Referent Nominalized VP 104, 113–115 Nominalizer 35, 104, 143, 154–155 Nominative case 147 Non-action 126 Non-anaphoric zero 9, 120, 124 defined 124 vs. anaphoric zero 124 Non-argument 5, 103, 108–111, 116–118, 123–124, 138–139, 160, 176 defined 103 types and definitions 103–104 See also Independent NP; Oblique; Predicate nominal Non-clausal IU 3, 29–31, 62, 161, 167, 171 See also Phrasal IU Non-event 126 Non-finite clause 174 Non-finite predicate/-te form 35, 143, 154 Non-ideational information 37, 155–156, 158 See also Cohesive; Interactional information Non-identifiable referent 19, 113 Non-lexical A Constraint 42, 116, 133–134, 161 Non-lexical NP 116, 134 See also Pronoun Non-new information 102, 134 See also Accessible; Given information

Non-newness 25 Non-postposed phrase See Pre-predicate phrase Non-referential NP 104, 171 Non-referential zero, vs. referential 124 Non-specific referent 171 Non-substantive IU 22, 48, 154, 165 See also Fragmentary; Regulatory IU Non-tracking NP 113 Non-transitive clause 125–128, 130, 139, 161 See also Adjectival predicate; Intransitive; Nominal predicate; One-participant clause Non-verbal predicate 103 See also Adjectival; Nominal predicate Noonan, Michael 169 Notational conventions 46 Noun 9–10, 17–20, 38, 41, 53, 104, 108, 114–116, 139, 160, 171 See also Bare; Common noun; Proper name Noun phrase (NP) 7, 10, 17–20, 31, 52–53, 55–56, 58, 65, 69–72, 81, 91, 100, 103–121, 123–124, 139, 159–161, 163–164, 169, 171, 173, 175–176 and IU/clause 105–106 argument vs. predicate 53, 56, 69–70 clause-internal vs. clause-external 103 definite vs. indefinite 7 lexical vs. pronominal 114–116, 131, 134 See also Argument; Grammatical role; Information status; Non-argument; Preferred NP types; Syntactic form NP See Noun phrase NP IU 29–31, 53–56, 58, 60, 62, 68–76, 78–81, 83, 85–91, 96–98, 104, 111, 147, 154, 159, 171, 174–175 types and distribution 52–53, 68–69 See also Independent phrasal NP IU NP-no + NP 104, 108, 113–115, 117

 Index

Null argument 9, 118, 124, 132–133 See also Overt; Zero argument Null argument language 9 Null argument strategy 118 O O See O role O role 4, 30, 41–42, 56, 70, 75–76, 82, 85–87, 103–105, 108–114, 117, 122–125, 127–134, 136, 139, 156–157, 161, 163, 165, 169–170, 176 See also Transitive object Object 4, 9, 38, 41, 53–54, 69, 97, 110, 169–170 See also Direct; Indirect; Transitive object Oblique 103, 108–117, 122–124, 136–139, 160, 176 See also Argument; Core-oblique distinction Observed example 54, 104, 147 Ochs, Elinor 3, 164, 169 O’Dowd, Elizabeth 42 OF See Overall frequency Old information 13–14, 38–39 See also Given information One clause/function/new NP/overt argument at a time 155, 158 One clause at a time constraint 2, 26–27, 36, 39 One IU, one clause strategy 97, 160 See also Multiple-IUs, one clause strategy; Single-IU clause One Lexical Argument Constraint 42, 131, 161 One New Argument Constraint 2, 41, 122–124, 138, 152, 160–161 One new concept at a time constraint 2, 36, 39–40, 122, 124, 152, 160 One new idea constraint 39, 122, 124, 152, 160 One new NP per IU constraint 81, 88, 121–124, 134, 136–139, 157, 160–161, 163, 165–166, 177 and IU/clause 138 and multi-IU clause 134–138, 157, 161–162, 166

and post-predicate new NP 165 See also Two new NPs per IU One overt argument per clause constraint 128, 131, 139, 161, 163 One overt NP per IU constraint 120, 123, 127, 139, 160, 163 One unit per proposition constraint 2, 39, 122 One-chunk-per-clause processing principle 39, 122 One-participant clause 125–127, 130, 161 vs. two-participant clause 125–126 See also Non-transitive clause Ono, Tsuyoshi 3–4, 30–31, 64, 72, 74, 169 Operational definition 12, 101–102 Opinions 46, 49, 126–127 Oral discourse/communication See Spoken discourse/communication Ordinary conversation 127, 166 and subjectivity 127 vs. institutional talk 166 O-role argument See O role Other-interruption 71, 73–74, 76, 79, 90 Overall frequency (OF) 8, 165 Overlapping xvi, 76 Overt argument 124–125, 127–133, 139, 152, 155, 161, 163–165 vs. null argument 132–134 Overt NP 104–107, 109, 111, 116–117, 119–121, 123, 127, 131–132, 134, 137, 139, 160–161, 163, 165 Overt subject 29, 52, 55, 58, 60, 67, 98–99, 105–106, 122, 159 P Parallelism 177 Parataxis 33 Partial proposition 55, 73–74 Partial propositional strategy 36–37 Particle 31–35, 37, 49, 70, 75, 143–144, 147, 150–151, 155, 157, 166, 169, 173, 176–177 See also Case-marking; Interactional particle

Index 

PAS See Preferred Argument Structure Passive clause/verb 176 Patient 103, 105, 170 Pattern 1, 5, 25, 33, 36, 41–42, 63, 87–88, 96, 115, 118, 130, 132, 147, 159, 161, 164–165 Pause xvi, 2, 16, 20–21, 34–35, 46, 49, 75, 86, 142–143, 149 types xvi See also Initial pause Pause filler 34–35, 49, 143, 149 Pause-bounded phrasal unit (PPU) 32 Pawley, Andrew 2, 26–27, 36, 39, 41 Payne, Doris L. 169 Pear Film/Story 32, 41, 172, 175–176 See also Elicited narrative Perception 19, 127, 173 Performance error 80–81, 88, 159 Personal pronoun 104, 108, 113–115, 176 Phonological prominence 17 Phrasal IU 3, 29–33, 36–37, 51, 53–56, 58–62, 64–65, 68–69, 71, 89, 93, 96, 98, 101, 103–104, 113, 121–122, 141, 147, 154–158, 161, 164, 166–167, 172–175, 177 as element/part of multi-IU clause 52–53, 64–65, 69, 89, 164 defined 51 marked use of 161 types and distribution 52–53, 61, 89 vs. clausal IU 57–60, 155–157, 172 See also Independent phrasal IU Phrasal IU as part of multi-IU clause 52–53, 56, 58–60, 64–65, 69, 89, 164 definition and subtypes 52 Phrasal IU option, vs. clausal 33, 158 Phrasality, marked case of 155 Phrase 10, 17–20, 30–31, 51–58, 61, 63–65, 74, 97–100, 103, 106, 121, 123, 142, 149, 159, 164, 172, 174–176 types 52–53

vs. clause 29–33, 51–53, 57–60, 98, 172 Phrase-orientedness 33 PI See Potential interference Pierrehumbert, Janet B. 19, 28, 46 Pitch 21, 32, 46, 171, 173 Planning 32, 34, 91, 142, 148, 150 Point of view 127 See also Subjectivity Polysynthetic language 23 Possessor 18 Postposed NP, clause-external 71–72, 84–88, 159, 165, 174 See also Copied; Restated postposed NP Postposed phrase, clause-internal/external See Post-predicate phrase Postposing 4, 54, 63–66, 87, 102, 164–165, 169–170, 175, 177 and backgrounding/defocusing 65, 85, 165 and intonation 4 and IU 63–66 and multi-IU clause 65, 164–165, 175 defined 4 functions of 65, 85, 88, 165 types of 64–65 See also Clause-external; Clause-internal postposing Postposition 176 See also Case-marking particle Post-predicate element/phrase 4, 31, 53–54, 56–57, 63–65, 70, 97, 129, 136, 154, 164–165, 169–170, 174–175, 177 vs. pre-predicate element/phrase 4, 54, 63, 164–165, 169 See also Marked word order Postverbal element See Post-predicate element Potential interference (PI) 8, 165 PP See Prepositional phrase PPU See Pause-bounded phrasal unit Predicate 12, 25–26, 31–33, 37, 39, 41, 51–54, 69, 74–80, 89, 92, 99,

 Index

104, 106, 125, 128–130, 159, 161, 173 types of 52 See also Adjectival; Nominal; Verbal predicate Predicate/predicative adjective/AP 53, 89, 105 Predicate nominal 70, 103–105, 108–117, 124, 131, 138–139, 160, 171 See also Nominal predicate Predictability 12–13 Preferred Argument Structure (PAS) 42, 123, 161 defined 41–42 quantity and role constraints of 42 See also Given A; Non-lexical A; One Lexical Argument; One New Argument Constraint Preferred argument structure 124 Preferred clause structure 101, 124, 128–130, 161 Preferred clause types 101, 124–125, 128, 139 Preferred functional structure, IU 5, 43, 141, 146 Preferred information structure, IU 5, 42–43, 101, 118, 123–124, 139 Preferred nominal structure 5, 161 Preferred NP types 101, 109, 116, 139, 160, 164 See also Grammatical role; Information status; Syntactic form Preferred structures, IU 5, 163–164 Preferred syntactic structure, IU 5, 43, 51, 60–61 Preposing 169 Preposition 25, 41 Prepositional phrase (PP) 30, 41 Pre-predicate element/phrase 4, 54, 63, 164–165, 169 See also Basic word order Presentative NP 175 Presupposition 11 Preverbal element See Pre-predicate element Primary accent 18, 23–25, 171

vs. secondary accent 18 Primary function, of linguistic element 144 Prince, Ellen F. 1, 12, 14–15, 170 Production, IU 5, 15, 33, 47–48, 81, 87, 97, 102, 117, 139, 155, 163, 166 See also Discourse; Language; Speech production Pronominal NP See Pronoun Pronoun 7, 9, 17–18, 65, 76, 83, 86, 91, 104, 108, 112, 114–117, 131, 133–134, 165, 171, 175–176 stressed vs. unstressed 7, 9 vs. full NP 9, 17, 117 vs. lexical NP 114–116, 131, 134 See also Demonstrative; Personal pronoun Proper name 18, 72, 170 Proposition 3, 11, 27, 33–34, 36–39, 57–58, 64–65, 71, 73–74, 79, 81, 85, 87–88, 137, 141–142, 144, 147, 150, 152, 154, 158–159, 164, 172, 174, 177 See also Complete; Full; Partial proposition Propositional content 22, 32, 35, 53, 57, 87, 117, 139, 141–142, 154, 160, 167 Propositional strategy 36–37 See also Complete; Partial propositional strategy Prosodic unit 3, 19–20, 171 Prosody 19, 28, 158, 164 Purpose, of communication/interaction 11, 30, 166 Q Quantitative analysis 48 Question 70, 72, 76, 79 See also Choice; Wh-; Yes/no question Question words 171 R RD See Referential distance RDed NP/phrase See Right-dislocated NP/phrase

Index 

Reactive token 87, 174 Realis 12 Recipient 173 See also Hearer; Listener Recoverability 12–13, 79 Redundancy 117 Reference 5, 8, 10–11, 18, 25, 102, 137, 160, 170 See also Nominal; Verbal reference Referent 7–11, 13–14, 18–19, 22, 24–26, 31, 41, 49, 81, 83, 85–86, 102–104, 124, 131, 133, 138–139, 143, 158, 163, 170–171, 175 and mental representation of 175 defined 175 given vs. accessible vs. new 10, 13, 25, 102–103 identifiable vs. non-identifiable 18–19 vs. event/state 26, 41 Referent-activating NP 81 Referent-anchoring NP 175 Referential accessibility 8 aspects of 8 See also Referential complexity; continuity Referential complexity 8, 165 measures and definitions 8 See also PI; SR Referential continuity 8, 165 measure and definition 8 See also RD Referential distance (RD) 8–9, 165, 175 Referential interference 8 Referential NP 175 Referential zero, vs. non-referential 124 Referent-listing NP 175 Referring expression, definite vs. indefinite 17 See also Noun phrase Regulatory function 22, 49, 158, 162, 173 Regulatory IU 22–24, 35, 49–50, 70, 76, 83, 91, 142–143, 165–166, 173, 175

defined 22, 49 subtypes and definitions 22, 49 vs. substantive IU 22–23, 49–50, 165–166, 173, 175 See also Cognitive; Interactional; Textual; Validational subtype Reid, Wallis 170 Reinforcement 65, 85 Reinhart, Tanya 170 Relative clause 10, 18, 20, 70, 93, 104, 174 Repair 65, 71, 73–75, 79, 85, 88 See also Self-repair Repeated NP 71–72, 86–88, 91, 159, 175 Repetition 86, 88, 174–175 Research question (RQ) 5, 43, 51, 101, 141 Reset 21 See also Baseline pitch level Restated postposed NP 72, 85 Resumptive pronoun 65, 83, 85, 175 See also Left-; Right-dislocated NP Right-dislocated (RDed) NP 7, 71–72, 84–86, 159 Right-dislocated (RDed) phrase 53–54, 65 Right-dislocation 65 Rise-fall intonation 46, 75 Rise-fall-rise intonation 46, 70 Rising intonation 35, 46, 70 Roberts, Celia 166 RQ See Research question S S See S role S role 41, 103–105, 108–117, 121–122, 124–125, 127–134, 136, 138–139, 160–161, 163, 165, 175–176 See also Intransitive subject Sacks, Harvey 173 Saeki, Tetsuo 65 Saito, Mamoru 3 Sakapulteko 41, 112–113, 118, 130, 175 Saliency 12–13 S-A-O categories 176 Sarangi, Srikant 166

 Index

Schank, Roger C. 170 Schegloff, Emanuel A. 19, 65, 87, 172–173 Scheibman, Joanne 3, 127 Schema 102, 170, 176 defined 170 Schiffrin, Deborah 22 Schuetze-Coburn, Stephan 2–3, 19, 21, 176 Scrambling 4, 169, 174 Second person 170, 176 Secondary accent 18 Segmentation, of spoken discourse 2, 19, 27, 171–172 See also Prosodic unit Segmentative/delimitative function, of intonation 171 Self-interruption xv Self-repair 71, 73–75, 79, 90 Selkirk, Elizabeth O. 20 Selting, Margaret 3, 19 Semi-active state 2, 15–17, 39, 81, 100, 102, 170, 176 Semi-clausal IU 43, 57, 59–60, 67, 70, 85, 88, 96–98, 100, 112, 122, 137, 159, 162, 171 vs. full clausal IU 66–67, 94 See also Independent semi-clausal IU Semi-clause 29, 31, 35, 37, 52, 54–55, 57, 59, 62, 66–67, 69, 89, 92–97, 99, 112, 122, 136, 159, 163, 171–172 defined 52, 54 vs. full clause 62, 94–95 Seneca 23 Sensitivity, toward addressee 34–35, 143 Sentence 4, 13, 19, 25, 39, 130, 177 Sentential adverb 89–92, 175 Sequence 2, 11, 20–21, 26–27, 40, 46, 50, 55–57, 70, 74, 78, 83, 91, 93, 102, 106, 153, 157, 177 Shared information 18–19, 118 vs. unshared information 14, 19 Shared knowledge 12–13, 18, 102

Sharedness 14, 18–19 See also Identifiability Shibatani, Masayoshi 3–4, 54, 64, 130, 174 Silva-Corvalán, Carmen 169 Simultaneous talk See Overlapping Single-IU clause 54–55, 64–66, 71, 87–88, 94, 96–99, 106, 108, 111, 122, 124–125, 128, 135, 137–138, 160, 174 full vs. semi- 94–95 vs. multi-IU clause 94–98, 124–125, 128, 134–135, 138, 160 See also Clausal IU; Independent clausal IU Singleness 164 Smith, Wendy 3, 42 Softening device 177 Sound stretching xv See also Lengthening SOV language/order See Subject-object-verb Spanish 42 Speaker-hearer interactional information See Interactional information Speaker-hearer relationship 30, 143, 171 Speaker-oriented perspective, on activation cost 16–17 Speakers of English See English speakers Speakers of Japanese See Japanese speakers Speaker-selectedness 14 Speakership 154 Specification, further 65, 85, 88 Speech 1, 19–20, 35, 143, 155, 173, 175 Speech acts 33–34 Speech production 19–20 Spoken communication 1, 28, 32, 39, 144, 158, 163 Spoken discourse 1–2, 7, 9, 12, 17, 19–20, 25, 27, 30, 35, 39, 42, 88, 130, 137, 166, 169–171, 175 spontaneous/casual vs. planned 170

Index 

Spoken English 2–3, 17, 26, 28–30, 37, 46, 63, 97, 151, 172 Spoken Japanese 3, 29, 32, 34, 36–37, 46, 116, 155, 157–158, 177 Spoken language 19, 164 non-elaboration of IU structures and characteristics of 163–164 See also Conversational language Spontaneous spoken discourse/ conversation 1–2, 4, 20, 27, 127, 144, 164, 170, 175 SR See Switch reference S-role argument See S role Stance 34, 126 Starting point 19, 21, 25 See also Added information State 3, 22, 26–27, 41, 49, 54, 102, 126–127, 143, 152, 159 State idea 26, 175 Stein, Dieter 127 Storage locus 38 Story preface 173 Storytelling 172 Strategy, speaker’s 29, 39, 65, 81, 88, 118–119, 177 Stray NP 71–79, 88, 93, 159, 173–175 defined 71 types and definitions 73–74 Stress 32 Structures, IU 3, 5, 48, 144, 163–164, 166 non-elaborate nature of 164 Subject 9, 12, 19, 25, 37, 52, 54–55, 57, 67, 69–70, 96, 100, 103, 110, 112–113, 147–148, 159, 169–171, 176 vs. object 9, 110, 169 vs. predicate 12, 25, 37 See also Intransitive; Transitive subject Subject marker 169 Subjectivity 34, 127 See also Point of view Subject-object relation 169

Subject-object-verb (SOV) 3–4, 130, 161, 169 as basic word order in Japanese 3–4, 169 Subject-predicate structure 25 Subordinate clause 52, 57, 104–105, 121, 124, 174 vs. main clause 52, 57, 121 Subordinating conjunction 148 Subordination 27 Substantive IU 22–26, 29, 35–36, 48–51, 57, 62, 83, 99, 123, 139, 142, 146–147, 149, 151–155, 158–160, 162–166, 171, 173, 175 defined 22, 49 vs. non-substantive IU 22, 48, 154, 165 See also Fragmentary; Regulatory IU Subtopics, of conversation 172 Sufficiently identifying verbalization 18–19 Sugito, Miyoko 29, 47 Suzuki, Ryoko 4, 64 Svartvik, Jan 19 Switch reference (SR) 8, 165 Syder, Frances H. 2, 26–27, 36, 39, 41 Syntactic factors, and multi-IU clause 158, 177 Syntactic form, NP 101, 104, 111, 113–117, 139, 164 and grammatical role 113–115 and information status 115 Syntactic fragmentation See Fragmentation Syntactic structure, IU 3, 5–6, 29, 32, 43, 48, 51, 65, 98, 144, 152, 159, 162–163 in English vs. Japanese 3, 29–33, 62 Syntactic structure types, IU 51–53, 57–58, 60, 63, 66, 98–99, 106–107, 109, 159, 164 and NP 106–107 and postposing 63–66 core/essential vs. peripheral/ redundant 57

 Index

defined 52–53 distribution of 57–60 See also Independent clausal; phrasal IU; Clausal; Phrasal IU as part of multi-IU clause Syntax/syntactic organization 33, 40, 157 English vs. Japanese, and multi-IU clause 33, 157–158 T Tag-like expression 143 See also Interactional particle Takami, Kenichi 65, 165 Talk-in-interaction 19, 88, 127 Tanaka, Hiroko 19 Tannen, Deborah 164, 170 Tao, Hongyin 3, 9, 19, 29–32, 37, 48, 62, 67, 72, 81, 87–88, 97, 99, 105, 124, 126, 159, 171, 175–176 TCU See Turn constructional unit Telephone conversation 35 Temporal juncture 11 Tendencies 122, 131, 155, 161 See also Pattern Tense 12, 170 Teramura, Hideo 65 Text 8–9, 11 Textual component 143, 145–146, 148–150, 152, 162 Textual connective 143, 148, 150 Textual function 33, 144, 151 Textual information 151 Textual subtype 22, 49, 173 Thematic importance 8, 165 measures and definitions 8 See also OF; TP Theme 11 Third person 176 Thompson, Sandra A. 1, 3, 10, 12–13, 19, 27, 30, 32, 42, 72, 74, 105, 113, 125–127, 170, 176 Thought 21, 42, 166 aspects of 166 See also Content; Infrastructure

Thought organization 164 Tightness, of clause 33 Tokieda, Motoki 34, 144 Tomlin, Russell S. 1, 10–11 Tone group 19, 25–26 Tone unit 2, 19 Tonic foot/syllable 24 Tonic prominence 25 Topic 1, 7–10, 12, 45, 48–49, 54, 57, 71–72, 76, 81, 83, 90, 100, 122, 134, 136, 159, 170, 172, 174–176 accessible vs. inaccessible 7–8 continuous vs. discontinuous 7–8 important vs. unimportant 8 See also Genuine topic; Referent Topic continuity 1, 7, 9–10, 134, 170 defined 7 vs. action/location/temporal continuity 10 vs. topic discontinuity 7 See also Referential distance; Topic persistence Topic marker 54 Topic NP, clause-external 71–72, 81–83, 159, 175 See also Genuine topic Topic persistence (TP) 8–9, 165 Topic shift 172 Topicality 7–9 aspects and measures of 8–9 See also Referential accessibility; Thematic importance Topics, of conversational data 45, 48–49 TP See Topic persistence Transcriber xvi, 21 Transcript 46, 50, 54 Transcription xvi, 43, 46–47 Transition relevance 173 Transitive clause 105, 110, 125, 127–128, 130–131, 139, 148, 161, 176 vs. non-transitive clause 124–126 See also High transitive; Low transitive clause Transitive object 41, 103, 132 See also O role

Index

Transitive subject 41, 103, 132 See also A role Transitive verb 103, 122–124 Transitive-intransitive bivalency 67 Transitivity 12, 105, 124–127, 130, 161, 170 and grounding 12 clauses of high vs. low 125–127 continuum 126 defined 105, 170 features of 12, 170 Truncated word/unit xv, 22 See also False start; Fragmentary IU; Interruption Tsutsui, Michio 177 Turn 19, 46, 83, 172–173, 175 Turn constructional unit (TCU) 173 Turn organization 19 Turn-internal listener backchannel 172 Turn-taking 48 Two new NPs per IU 122–124, 160 See also Argument-oblique combination Two-argument clause/structure 128, 130, 161 Two-participant clause 125–126, 130 vs. one-participant clause 125–126 See also Transitive clause Two-party conversation See Dyadic conversation

V Validational subtype 22, 49, 173 Value 37, 156 van Dijk, Teun A. 166 Vandepitte, Sonia 19, 47, 171 Venditti, Jennifer J. 29, 47 Verb 4, 20, 31, 38–39, 41, 52–53, 57–58, 61–62, 96, 102–103, 105, 121–124, 126, 130, 154, 170–171, 176–177 See also Ditransitive; Intransitive; Transitive verb Verb phrase (VP) 31, 39, 52–53, 58, 61, 67, 93, 97, 104, 107, 109, 115, 121 Verb-adverb combination 41, 102 Verbal predicate 25, 29, 31, 52, 55, 58–62, 67, 96–100, 105–106, 110, 112, 159, 171, 174 Verbal reference 102 Verb-final language, vs. verb-initial 130 Verb-final requirement 4 Verb-object combination 41 Verb-particle combination 41 Vocalization 15, 20 Vocatives 72 Vogel, Irene 20 Voice 12 VP See Verb phrase VP IU 53, 58, 61, 93, 96–97, 121

U Uncodable IU 49–50 Unifunctionality 153, 155 Unit See Discourse unit; Intonation unit; Prosodic unit Unitariness 164 Unmarked information structure 12, 25, 120 Unmarked word order 54, 129 Unshared information 14, 19 Unused information 14 Utterance 2, 13, 15, 17, 19, 22, 25, 33–34, 77, 81, 86–87, 92, 100, 127, 173, 175

W Wallace, Stephen 11, 170 Ward, Gregory L. 1, 14, 47, 169 Wh-question 79 Word 2, 20, 23–25, 27, 32, 34, 38–40, 46, 50, 142, 171 Word order 3–4, 12, 54, 63, 129, 164, 169 in Japanese spoken discourse 3–4 marked vs. unmarked/basic/ canonical 3–4, 54, 63, 164, 169 SOV 3–4, 169 See also Basic; Marked word order



 Index

Wright, Susan 127 Writing 27 Written discourse 170

Y Yes/no question 70, 76 Yule, George 17–18, 30, 172

Z Zero anaphora 7–9, 11, 29, 120–121, 172 Zero argument 124, 128 See also Anaphoric; Non-anaphoric zero Zero form 55, 112, 116, 124, 132–134, 161 Zero-form NP 9, 116–117, 133, 161 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. Workshop 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. Amsterdam, 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 Argumentation, 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. Amsterdam/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/Philadelphia, 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, incorporation 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 linguistics. 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 semanticcommunicative 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, informational 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.