Professional Documents
Culture Documents
The SLCS series has been established as a companion series to Studies in Language,
International Journal, sponsored by the Foundation “Foundations of Language”.
Series Editors
Werner Abraham Michael Noonan
University of Vienna University of Wisconsin, Milwaukee
Editorial Board
Joan Bybee Ekkehard König
University of New Mexico Free University of Berlin
Ulrike Claudi Christian Lehmann
University of Cologne University of Erfurt
Bernard Comrie Robert Longacre
Max Planck Institute University of Texas, Arlington
For Evolutionary Anthropology, Leipzig
Brian MacWhinney
William Croft Carnegie-Mellon University
University of Manchester
Marianne Mithun
Östen Dahl University of California, Santa Barbara
University of Stockholm
Edith Moravcsik
Gerrit Dimmendaal University of Wisconsin, Milwaukee
University of Leiden
Masayoshi Shibatani
Martin Haspelmath Rice University and Kobe University
Max Planck Institute
Russell Tomlin
For Evolutionary Anthropology, Leipzig
University of Oregon
Volume 65
Intonation Units in Japanese Conversation: Syntactic, Informational, and
Functional structures
by Kazuko Matsumoto
Intonation Units in
Japanese Conversation
Syntactic, informational,
and functional structures
Kazuko Matsumoto
Aichi University of Education, Japan
Preface xi
Tables and Figures xiii
Transcription conventions xv
Abbreviations xvii
Chapter 1
Introduction 1
1.1 Information flow 1
1.2 Intonation units 1
1.3 Word order in Japanese spoken discourse 3
1.4 Purpose of the study 5
1.5 General research questions 5
1.6 Organization of the book 5
Chapter 2
Information flow in spoken discourse 7
2.1 Topicality 7
2.1.1 Topic continuity 7
2.1.2 Referential accessibility and thematic importance 8
2.1.3 Referential distance and topic persistence 9
2.2 Grounding 10
2.2.1 Grounding and given-new information 10
2.2.2 Foreground vs. background 11
2.3 Information status 12
2.3.1 Approaches to the notion of givenness/newness 12
2.3.2 Activation cost: Given-accessible-new distinction 15
2.3.3 The expression of activation cost 17
2.3.4 Identifiability, definiteness, and activation cost 18
Table of contents
Chapter 3
Method of the study 43
3.1 Research questions and hypotheses 43
3.2 Data base 44
3.3 Data transcription 46
3.4 Intonation units produced in 16 conversations 47
3.5 Analysis of intonation units 48
3.5.1 Data for quantitative analyses of Japanese intonation units 48
3.5.2 Topics of 16 conversational segments 48
3.5.3 Substantive, regulatory, and fragmentary intonation units 49
Chapter 4
Syntactic structure of the intonation unit
in conversational Japanese 51
4.1 Data coding 51
4.2 Distribution of IU syntactic structure types 57
4.2.1 Clausal vs. phrasal intonation units 57
4.2.2 Preferred syntactic structure of the Japanese intonation
unit 60
4.3 Distribution of post-predicate phrases among IU syntactic types 63
4.4 Clausal intonation units: Full clauses vs. semi-clauses 66
Table of contents
Chapter 5
Information structure of the intonation unit
in conversational Japanese 101
5.1 Data coding 101
5.2 NPs, intonation units, and clauses 105
5.3 Preferred NP types: Grammatical roles, information statuses,
and syntactic forms 109
5.4 Preferred information structure of the Japanese intonation unit 118
5.4.1 Distribution of IU information structure types 118
5.4.2 The one new NP per IU constraint 121
5.5 Preferred clause structure in conversational Japanese 124
5.5.1 Preferred clause types and preferred argument structure 124
5.5.2 Overt vs. null arguments 132
5.6 Multi-IU clauses and the one new NP per IU constraint 134
5.7 Summary 139
Chapter 6
Functional structure of the intonation unit
in conversational Japanese 141
6.1 Data coding 141
6.2 Distribution of IU functional structure types 144
6.3 Preferred functional structure of the Japanese intonation unit 146
6.3.1 Preferred number of functional components per IU 149
6.3.2 Linear order of functional components within an IU 150
6.4 Multi-IU clauses and multifunctionality 155
6.5 Summary 158
Table of contents
Chapter 7
Conclusion 159
7.1 Japanese intonation units: Syntactic, informational,
and functional structures 159
7.2 Prospects 164
Notes 169
References 179
Index 191
Preface
Tables
3.1. Data for this study
3.2. Major topics of 16 conversations
3.3. Number of IUs produced in 16 conversations
3.4. Topics of 16 conversational segments
3.5. Number of four types of IUs in 16 conversational segments
4.1. Distribution of IU syntactic structure types
4.2. Distribution of eight clausal IU types
4.3. Distribution of 12 phrasal IU types
4.4. Distribution of post-predicate phrases among IU syntactic types
4.5. Proportion of IUs involving postposing by IU syntactic type
4.6. Proportion of full clauses vs. semi-clauses
4.7. Distribution of five types of NP IUs
4.8. Distribution of five types of independent phrasal NP IUs
4.9. Frequency and features of four types of stray NP IUs
4.10. Distribution of adjectival, adverbial, and mixed phrasal IU types
4.11. Distribution of six types of independent adverbial IUs
4.12. Average number of IUs per clause
4.13. Frequency of the number of IUs per clause
4.14. Summary of results of IU syntactic structure analysis
5.1. Overt NPs, IUs, and clauses
5.2. Average number of overt NPs per IU by IU syntactic type
5.3. Distribution of grammatical roles among IU syntactic types
5.4. Grammatical role and information status
5.5. Grammatical role and NP syntactic form
5.6. Information status and NP syntactic form
5.7. Summary of results of analysis of overt NPs
5.8. Distribution of IU information structure types
5.9. Two types of IUs with zero NPs
5.10. Types of IUs with two new NPs
Tables and Figures
Figures
4.1. Distribution of four major IU syntactic types
4.2. Distribution of six IU syntactic types
4.3. Distribution of 20 IU syntactic types
4.4. Proportion of full clauses vs. semi-clauses
4.5. Distribution of five types of NP IUs
4.6. Distribution of five types of independent phrasal NP IUs
4.7. Frequency of the number of IUs per clause
5.1. Distribution of six grammatical roles
5.2. Proportion of three information statuses within grammatical roles
5.3. Distribution of eight NP syntactic forms
5.4. Distribution of IU information structure types
5.5. Frequency of five clause types
5.6. Proportion of clauses with zero, one, and two arguments
5.7. Frequency of clauses with zero, overt S, A, and O arguments
5.8. Proportion of overt vs. null arguments
5.9. Proportion of IUs/clauses with more than one new NP
6.1. Distribution of IU functional structure types
Transcription conventions
Sequential Relationships
= Equal signs come in pairs and mark latching; the talk connected by
equal signs is continuous with no discernible pause between them;
used only for an inter-speaker (not intra-speaker) transition.
[ ] Brackets mark overlapping or simultaneous talk; left brackets in-
dicate the point of onset; right brackets indicate the point of
resolution.
Pauses
+ very short micropause which is barely noticeable
(0.1–0.2 seconds of silence)
++ medium-length pause which is noticeable
(0.3–0.6 seconds of silence)
+++ long pause
(0.7–0.9 seconds of silence)
(2.0) Numbers in parentheses indicate elapsed extra-long silence in tenths
of a second (silence of longer than 1.0 second).
Intonation Contours
. falling
, continuing
? rising
^ rise-fall
∼ rise-fall-rise
Listener Backchannels
@ listener backchannels which show affirmative response to the
speaker’s utterance
$ listener backchannels which show wonder, awareness, or surprise
# laughter as listener backchannels; the more symbols, the more
laughters.
Other Symbols
(word) Parentheses surrounding a word or words indicate uncertainty
about the transcription; empty parentheses mark that no hearing
could be achieved.
(( )) Double parentheses mark the transcriber’s descriptions of events or
comments.
Abbreviations
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.
Past research has noted the intermittent nature of spontaneous spoken dis-
course. That is, spontaneous speech has the property of being produced in a
series of brief spurts. These spurts of language, or the coherent chunks into
which speakers segment talk, have been considered the basic units of informa-
tion flow. These units have been given different names by different discourse
Chapter 1
the clause corresponding to “the mental proposition that stands for some state
or event” as the basic unit of information storage in coherent discourse.
Iwasaki and Tao (1993), in their comparative study of the syntactic struc-
ture of IUs in English and Japanese, have shown, however, that while En-
glish IUs are mostly clausal, Japanese IUs tend to be non-clausal, or phrasal.
This suggests that Chafe’s (1987) “clause centrality proposal” may not be war-
ranted cross-linguistically. On the other hand, their findings are consistent
with Clancy’s (1982) and Maynard’s (1989) observation that spoken Japanese
is “highly fragmented”, that is, a syntactic clause is frequently broken up into
a number of units smaller than those found in spoken English. Moreover,
Japanese IUs, according to Iwasaki (1993), have very orderly functional struc-
ture, and a built-in mechanism which allows speakers to additionally convey
cohesive and interactional information while at the same time accomplishing
the primary task of communicating ideational information.
Previous discourse-functional research on IUs as basic prosodic units of
information flow has mostly dealt with spoken narratives (e.g. Chafe 1980a;
Clancy 1980, 1982; Croft 1995; Du Bois 1987). However, an apparent
shift from narrative-focused to conversation-centered research can be seen
in more recent studies (e.g. Clancy et al. 1996; Couper-Kuhlen & Selting
1996; Ochs et al. 1996; Ono & Thompson 1995, 1996). Importantly, many
of the recent IU-based functional studies are cross-linguistic in nature,
searching for language-independent or language-specific constraints operat-
ing on the syntactic, informational, or functional structure of the IU (e.g.
Ashby & Bentivoglio 1993; Chafe 1994; Du Bois 1987, 2003; Du Bois et al.
2003; Helasvuo 2001; Iwasaki 1993; Iwasaki & Tao 1993; Kärkkäinen 1996;
Scheibman 2002; Schuetze-Coburn 1994; Smith 1996; Tao 1996; Thompson
1997; Thompson & Hopper 2001). Past research in Japanese IUs has already
dealt with their syntactic and functional structures in comparison to their
English counterparts. However, their information structure and the possible
relation between the structures—syntactic, functional, and informational—
have not been investigated to date.
The purpose of this book is to explore the syntactic, informational, and func-
tional structures of IUs as basic units of discourse production and information
flow in naturally occurring Japanese conversations. Put differently, this study
attempts to examine, in terms of information flow, not only types of IUs that
speakers of Japanese use in dialogic conversational interactions, but also pat-
terns in their production of IUs. In its analysis of the syntactic and informa-
tion structures of the Japanese IU, the present study focuses on nominal ref-
erences and patterns in the speakers’ production of those nominals. The study
tries to elucidate, above all, the preferred nominal structure, where nominals
include both arguments and non-arguments. In its investigation of the func-
tional structure of the IU, the study focuses on the linear order of functional
components within an IU as well as interrelationships among those compo-
nents. Overall, in exploring the syntactic, informational, and functional struc-
tures of the IU in conversational Japanese, this study is interested not only in
the preferred IU structures which are typical of the way Japanese speakers talk
in connected discourse, but in possible relationships between the structures
and their implications as well.
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?
analyzing it. Chapters 4, 5, and 6 present and discuss the results of the study.
That is, Chapters 4, 5, and 6, responding to RQ1, RQ2, and RQ3, respectively,
present and discuss the results of the analysis of the syntactic structure, in-
formation structure, and functional structure of the Japanese IU, respectively.
Finally, Chapter 7 concludes the study with a summary of the results presented
in Chapters 4 through 6 and discusses implications of this study.
Chapter 2
. Topicality
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”.
Givón’s RD, Chafe (1994) claims, is a rough reflection of what he calls “ac-
tivation cost”—the given-accessible-new distinction (see Section 2.3.2). That
is, the “topicality” measured by RD is largely equatable with activation cost:
(a) zero anaphora and unstressed pronouns unambiguously express givenness;
(b) stressed pronouns express contrastive given or accessible referents; and
(c) left-dislocated NPs may verbalize accessible or new referents. Chafe further
remarks that Givón’s RD, which limits its “look-back” to 20 clauses, does not
distinguish accessibility from newness in many cases. For Chafe, the finding of
Givón (1990) that definite nouns showed a scattered RD distribution supports
his own view that definiteness, or identifiability is independent of activation
cost (cf. Section 2.3.4).
Givón (1990) makes two important observations in his quantitative studies
of topicality. One is the lower RD and higher TP of the subject over the object,
and the lower RD and higher TP of the direct object over the indirect object,
which has led him to state that “the subject is consistently more topical than
the direct object, and the direct object more topical than the indirect object”
(Givón 1990: 901). Chafe (1994: 184) interprets this as saying that “subjects are
most often given and of primary importance, and direct objects rank somewhat
lower on the scales of activation cost and importance, and all other roles rank
lower still”. Another important observation is the predominance of anaphoric
pronouns over full NPs in oral discourse. This reflects Givón’s (1990: 918) view
that continued activation of the current active referent is the norm in coherent
multi-propositional discourse.
Givón’s topic continuity-based account discussed above is far from al-
mighty, in my view. First and foremost, we should question whether the theory
has the capacity to explain differnt types of anaphora that can occur in natural
spoken discourse. This is especially important when we work with so-called
“null argument” languages such as Japanese and Mandarin, which allow use of
abundant zero-form NPs. If part of the zero-marking argument NPs in these
languages are actually independent of anaphoric continuity processes, as Tao
(1996) has convincingly shown, this means that the topic continuity theory
cannot be applied to these “non-anaphoric zeros” (see Sections 5.4–5.5). In
other words, the theory seems not entirely effective in accounting for all types
of anaphora occurring in spoken discourse. We should also note that, as Givón
Chapter 2
(1990) himself points out, topic continuity, or the recurrence of the same
referent is not the only required element of discoursal coherence. In future
discourse studies, more attention should be paid to other types of continuity,
which include temporal continuity and continuity of location and action.
. Grounding
The notion of grounding has been used in two different, although not totally
unrelated, ways by discourse researchers. One is represented by Fox and
Thompson’s (1990) and Givón’s (1990) view of grounding which has to do
with the distinction and interaction between given and new information. The
other is represented by Hopper’s (1979) and Tomlin’s (1985) view of grounding
which is concerned with the foreground-background distinction/continuum.
More detailed discussion will follow.
Grounding is the principal way in which speakers make referents carrying new
information “relevant” for listeners. That is, “to ground a noun phrase is to
locate its referent in conversational space, to make its referent relevant for
the hearer by relating it to a given referent already established in the prior
discourse. Grounding a noun phrase is a way of warranting its introduction
at the point where it is mentioned” (Fox & Thompson 1990: 303). Fox and
Thompson (1990), with reference to relative clause constructions in English
conversations, discuss three types of grounding. These include what they call
“main-clause grounding”, where the main clause situates a newly introduced
head NP by relating it to a given referent in its own clause. For Fox and
Thompson (1990: 301), then, “grounding is essentially a background task,
as opposed to asserting . . . That is, a grounding [relative] clause does not
assert . . . , but merely locates the referent in conversational space”.
A similar view of grounding is taken by Givón (1990). He states that
the function of given, redundant, or topical information in the clause is to
“ground” the new, asserted information to the already stored given informa-
tion, that is, to integrate or “address” the new information onto the appro-
priate location within the storage space in the episodic memory. Therefore,
in communicating with their interlocutors, speakers include given, shared, or
presupposed information in their clauses as a “background” for asserted, new
information, which, for Givón (1987), constitutes “foreground”. The differen-
Information flow in spoken discourse
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
two important properties: their overt concern with both the speaker and
the listener (i.e. “speaker-selectedness” and “listener-orientedness”), and their
two-way division of information statuses into given/old and new.
The binary given-new distinction, however, has proved inadequate in
more recent discourse studies (cf. Birner & Ward 1998). Apparently, given-
ness/newness should be treated as a matter of degree, not as an all-or-nothing
phenomenon (Chafe 1976). Prince (1981) proposes a taxonomy of what she
terms “assumed familiarity”, which reflects the fact that language users must
actually operate on the basis of what they assume to be familiar to their in-
terlocutors. Her taxonomy of information status includes under the heading
“assumed familiarity” three categories: (a) new, (b) inferrable, and (c) evoked.
Prince defines an entity which the speaker first introduces into the discourse as
“new”. A “new” discourse entity falls into one of two categories: it is “brand-
new” when “the hearer may have had to create a new entity”; it is “unused”
when “the hearer may be assumed to have a corresponding entity in his/her
own model and simply has to replace it in (or copy it into) the discourse-
model” (Prince 1981: 235). “If some NP is uttered whose entity is already in
the discourse-model, it represents an ‘evoked’ entity” (where “evoked” is used
as an equivalent for “given”). “A discourse entity is ‘inferrable’ if the speaker
assumes the hearer can infer it, via logical—or, more commonly, plausible—
reasoning, from discourse entities already evoked or from other inferrables”
(Prince 1981: 236).
In Prince (1992: 309), information statuses of discourse entities are classi-
fied, in terms of two distinct divisions, discourse-old/-new and hearer-old/-new,
into four types. They are (a) discourse-new, hearer-new (i.e. brand-new);
(b) discourse-new, hearer-old, (i.e. unused); (c) discourse-old, hearer-old
(i.e. evoked); and (d) discourse-old, hearer-new (this type presumably does
not occur in natural discourse). Clearly, one of the advantages of this distinc-
tion is that it captures the fact that while what is old/given in the discourse will
be familiar to the hearer as well, what is assumed by the speaker to be new to
the discourse may not be new to the hearer. Chafe (1994) notes that Prince’s
(1981) use of “brand-new” vs. “unused” is nearly identical to his use of “un-
shared” vs. “shared”. However, he emphasizes that sharedness, which is a com-
ponent of identifiability, is independent of activation cost (see Section 2.3.2).
I agree with Chafe’s (1994: 175) statement that “whether or not a referent is
assumed to be newly activated in the listener’s consciousness is a different ques-
tion from whether or not it is assumed to be already part of the listener’s knowl-
edge. . . . As a term, therefore, ‘unused’ has the disadvantage of conflating the
separate domains of sharedness and activation cost”.
Information flow in spoken discourse
Chafe (1987), assuming that givenness and newness are partial manifestations
of basic cognitive processes, introduces the notion of “accessible” information
as an intermediate type that exists between “given” and “new”. He thus pro-
poses a three-way breakdown into given, accessible, and new information as
opposed to the simple binary distinction of given vs. new.
Chafe (1987) hypothesizes that a particular concept, at a particular time,
may be in any one of three different states of activation. These he terms “active”,
“semi-active”, and “inactive”. “An active concept is one that is currently lit up,
a concept in a person’s focus of consciousness. A semi-active concept is one
that is in a person’s peripheral consciousness, a concept of which a person
has background awareness, but which is not being directly focused on. An
inactive concept is one which is currently in a person’s long-term memory,
neither focally nor peripherally active” (Chafe 1987: 25). It is assumed in
his model that the speaker’s utterance of an IU functions to activate all the
concepts it contains for the hearer while deactivating others, and to bring about
changes in the activation states of information in the hearer’s mind. That is,
the production of an IU, it is assumed, involves changes in activation states,
which take place first in the speaker’s mind during the initially occurring pause
and then in the hearer’s mind during the following period of vocalization
(cf. Section 2.4.2). According to Chafe (1987), then, “given” concepts are
those that were “already active” for the speaker prior to uttering an IU, and
which the speaker assumed to be already active in the mind of the hearer
as well.5 “Accessible” or “previously semi-active”—which could be equated
with Prince’s (1981) “inferrable”—concepts are those that the speaker, before
uttering an IU, transferred from the semi-active to the active state in his/her
own mind.6 “New” or “previously inactive” concepts are those that the speaker,
before uttering an IU, transferred from the inactive to the active state.
The temporal aspects of givenness, accessibility, and newness in relation to
the speaker’s and the listener’s cognitive processes can be visualized as shown
in (2.2), which is from Chafe (1994: 74).
Chapter 2
(2.2) t1 t2 t3
word1 word2 word3 .....
pause IU IU
onset onset completion
given
speaker- active active
oriented accessible
semi-active
new
inactive
given
listener- active active
oriented accessible
semi-active
new
inactive
At t1, the onset of the pause, a particular idea is active, semi-active, or
inactive; and at t2, the onset of the IU, this idea is now active. If it was
already active at t1, it is “given” information at t2. If it was semi-active at
t1, it is “accessible” information at t2. If it was inactive at t1, it is “new”
information at t2. Concerning what Chafe (1994: Chapter 6) terms “activation
cost” (i.e. cognitive cost involved in the three processes, which thereby refers
to the given-accessible-new distinction), given information is presumably least
costly in the transition from t1 to t2 because it was already active at t1.
Accessible information is somewhat more costly; and new information is the
most costly of all, because converting an idea from the inactive to the active
state supposedly requires more mental effort on the part of the speaker.
The figure given in (2.2) presents two perspectives on activation cost,
i.e. speaker-oriented and listener-oriented. The speaker-oriented perspective,
shown in the top left of (2.2), is concerned only with what is happening in the
the speaker’s mind. At t2, the end of the pause or the onset of the IU, to repeat,
all the ideas to be verbalized in the following IU would be active for the speaker.
From the speaker’s point of view, then, to repeat, an idea that was already
active at t1, the beginning of the pause, would constitute “given” information;
one that was semi-active, “accessible” information; and one that was inactive,
“new” information. The listener-oriented perspective, shown in the bottom
right of (2.2), on the other hand, incorporates the speaker’s understanding of
what is happening in the mind of the listener. Here the speaker assumes that a
particular idea is active, semi-active, or inactive in the listener’s mind at t2, the
onset of the IU. Information status is determined by the speaker’s assumption
Information flow in spoken discourse
of what will take place in the listener’s mind between t2 and t3, the completion
of the IU. The speaker assumes that hearing the IU will either (a) continue an
idea that is already active for the listener, (b) activate an idea that was previously
semi-active for the listener, or (c) activate an idea that was previously inactive
for the listener. In the cases of (a), (b), and (c), the speaker will verbalize the
idea as “given”, “accessible”, and “new” information, respectively. That is, we
can say that it is the speaker’s anticipation of the activation process in the
listener’s mind that determines the information status of the verbalized idea.
Chafe (1994) emphasizes that although language works better when activa-
tion cost is listener-oriented, the two perspectives—the first dependent solely
on the speaker’s consciousness and the second on the speaker’s assessment of
the listener’s consciousness—may not need to be chosen categorically between
them. Rather, Chafe (1994: 75) argues, “typically a speaker may assume that the
processes in the listener’s mind are in harmony with those in the speaker’s own
mind, allowing for the time lag occupied by the utterance of the IU. In other
words, the events pictured in the section labeled speaker-oriented [in (2.2)]
are likely to be mirrored in the assumptions represented in the section labeled
listener-oriented”.
Chafe claims that of his three identification criteria (i.e. intonational, hesita-
tional, and syntactic), intonation contour is the most consistent signal of an IU
boundary in English, whereas the presence of a pause or the clausal syntactic
structure is a less important criterion for the identification of such units. In
English, Chafe shows, IUs are not uniformly separated by hesitations (pauses
often occur within such units); and some IUs are less than clauses (e.g. phrases)
or more than clauses (e.g. IUs that contain an embedded relative clause).
Du Bois et al. (1992: 100) present the following as the major prosodic cues
that contribute to identifying the boundaries of IUs:
Information flow in spoken discourse
Regarding the insights IUs can yield into the nature of language production
and thought processes, Chafe (1993: 38–39) states the following:
As an intuitive starting point, I can observe that I am able to focus my
consciousness on only a very small amount of information at one time and
that this focus of consciousness changes quite rapidly as thinking proceeds.
Furthermore, it is plausible to suppose that during the production of language
a speaker will focus on the information he or she is verbalizing at that
moment. Against this background, an intonation unit is plausibly viewed as
the verbal representation of just the information that is in the speaker’s focus
of consciousness at the moment it is uttered. A speaker’s intention in uttering
an intonation unit must then be to introduce something resembling that
particular focus of consciousness into the attentive listener’s consciousness.
If each intonation unit, indeed, corresponds to a focus of consciousness,
intonation units can give us important insights into how much and what kinds
of information can be active at one time in a speaker’s mind.
Chapter 2
In sum, for Chafe (1980b: 48), IUs are “linguistic expressions of focuses of con-
sciousness, whose properties apparently belong to our built-in information-
processing capabilities”. In other words, the IU as a whole is a verbalization of
the particular information on which the speaker is focusing his/her conscious-
ness at a particular moment. The IU is also viewed as the linguistic expression
of information that is, at first, active in the consciousness of the speaker, and
then, by the utterance of the unit, active in the consciousness of the listener
as well (cf. Section 2.3.2). The question of “how much” and “what kinds” of
information can be activated within each IU will be dealt with in terms of
constraints on information flow in Section 2.6.
Chafe (1993, 1994) categorizes IUs into three major types: “substantive”, “regu-
latory”, and “fragmentary”. Substantive IUs convey substantive ideas of events,
states, or referents, participating in the communication of propositional con-
tent. The two other non-substantive types are not directly concerned with the
transmission of substantive ideational information. Regulatory IUs function
to regulate interaction or information flow in discourse (IUs consisting of so-
called “discourse markers” belong to this category; see Schiffrin 1987). Both
substantive and regulatory IUs are successfully completed units. Fragmentary
IUs, on the other hand, are truncated units that the speaker breaks off before
the completion of their projected contour.
The regulatory type can be further divided into at least four subtypes: “tex-
tual”, “interactional”, “cognitive”, and “validational”. The textual subtype func-
tions to regulate the linkage, or to show a particular kind of linkage, between
IUs (e.g. and then, but, so). The interactional subtype involves interaction be-
tween the conversational co-participants, indicating the speaker’s attentiveness
to the hearer (e.g. mhm, you know). The cognitive subtype signals ongoing
mental activities on the part of the speaker (e.g. let’s see). The fourth, vali-
dational subtype expresses the speaker’s judgment of the validity of the infor-
mation being conveyed (e.g. maybe). As Chafe (1993) emphasizes, however, the
boundaries between these subtypes of regulatory IUs are not clear-cut. (For ex-
ample, the function of the regulatory IU well may be textual, interactional, or
cognitive, depending on the conversational context where it is uttered.) What
is evident is that IUs that serve several regulatory functions contrast sharply
with IUs that convey substantive ideational information (cf. Section 3.5.3).
The taxonomy of IUs is shown in (2.5).
Information flow in spoken discourse
The distinction between substantive (S), regulatory (R), and fragmentary (F)
IUs is illustrated in (2.6), an excerpt from a conversation between three
participants, A, B, and C (Chafe 1994: 63–64).
(2.6) a. (R) A: ++ Well,
b. (S) isn’t she healthy?
c. (R) B: + Mhm,
d. (F) A: ++ I mean she -
e. (F) I know she has -
f. (S) C: More or less.
g. (S) A: + She has [something with her] gallbladder,
h. (S) B: [ gallbladder and, ]
i. (S) ++ heart trouble and,
j. (S) [back problems.]
k. (S) A: [ She has heart ] trouble,
With regard to the size of IUs in English, Chafe (1994: 64–65) observes that
while regulatory IUs tend to be one word in length, substantive IUs are fairly
constrained to a modal length of four words.12 He indicates that these figures
are applicable to English only. In languages that pack more information into a
word, IUs generally contain fewer words; for example, IUs in Seneca, a highly
polysynthetic language, tend to be two words in length (Chafe 1994: 148). Thus,
for Chafe, the fact that the number of words per IU remains within a narrow
range for any given language reflects a cognitive constraint on the capacity of
active consciousness, i.e. how much information can be active in the speaker’s
mind at one time.
Chafe (1993) introduces the term “accent units” (AUs) as subunits of IUs.
An AU, according to Chafe, contains only one primary accent; it consists
of the word containing that primary accent—which roughly corresponds to
Chapter 2
what Halliday (1985) calls “tonic foot” that contains a “tonic syllable”—
plus whatever other words belong to the same constituent as that word.
Chafe hypothesizes that the AU is the domain of activation of information
in consciousness, that is, information is activated and deactivated within
each AU. In other words, AUs are the loci of three types of information,
given, accessible, and new. An IU, being composed of one or more AUs,
verbalizes a cluster of ideas; and this cluster may include more than one given
or accessible information, but only one new information, according to his
proposed constraint (as will be shown in Section 2.6).
In the following conversational excerpts, (2.7) and (2.8), which are from
Chafe (1993: 38), AUs are separated by the “pipe” symbol (|).
(2.7) a. A: (1) ++ and thé:n | (2) the má:n,
b. ->(1) ++ ùh: hèr bóyfriend whatever | (2) was gonna móve
- ->ìn | (3) wíth them.
(2.8) a. A: ++ And so in betwéen the -
b. ->(1) + okay the fírst two ròoms | (2) are at the t + at the +
- ->frónt pàrt of the hàll.
c. B: ++ Mhm,
d. A: ->(1) + and so + betwéen those ++ the èntrances to those
- ->ròoms and the bàthroom | (2) there’s a lò:ng stretch of hállway.
Example (2.7) consists of two substantive IUs, (2.7a, b). The IU (2.7a) consists
of two AUs, (2.7a-1) and then and (2.7a-2) the man, each of which carries a
primary accent. The first AU and then is regulatory, whereas the second AU
the man is substantive.13 The IU (2.7b), on the other hand, is composed of
three substantive AUs, (2.7b-1), (2.7b-2), and (2.7b-3), where primary accents
are placed on boyfriend, move, and with, respectively. Likewise, (2.8) consists of
one fragmentary IU, (2.8a), one regulatory IU, (2.8c), and two substantive IUs,
(2.8b) and (2.8d). Both of these substantive IUs consist of two substantive AUs.
These examples indicate that regulatory AUs can be subunits of both regulatory
and substantive IUs.
Turning to the activation states of information verbalized in the four sub-
stantive AUs in (2.7), we can observe the following. First, the AU (2.7a-2) the
man verbalizes given information. Second, the AU (2.7b-1) uh her boyfriend
whetever can be taken to express accessible information, because it has the
same referent as the man activated in (2.7a), and the added characterization
of that referent (i.e. the idea of him being the woman’s boyfriend) is presum-
ably accessible. The AU (2.7b-2) was gonna move in also expresses accessible
infomation because the idea of moving-in was already activated earlier in the
Information flow in spoken discourse
discourse. The only AU that verbalizes new information is the last AU, (2.7b-3)
with them, where the very utterance of the primary accented preposition with
introduces the man’s participation in the activity as a totally new, previously
unmentioned concept (Chafe 1993). Similarly, in the IUs (2.8b) and (2.8d),
the second AUs, (2.8b-2) and (2.8d-2), respectively, express new information,
with the preceding AUs providing a context/background for the following one
new idea. In sum, the AUs (2.7b-3), (2.8b-2), and (2.8d-2) are the domains
in which only one idea is newly activated within the IUs (2.7b), (2.8b), and
(2.8d), respectively. Examples (2.7)–(2.8) appear to illustrate the basic canon-
ical pattern of information flow—from given to new—within units of English
spoken discourse. The new element carrying “tonic prominence” and thus “in-
formation focus” is typically the last functional element within the “informa-
tion unit” or “tone group” (which can be regarded as essentially the same as
the IU) (Halliday 1985).
Chafe (1987) argues that the information structure of an IU usually
consists of a “starting point” and an “added information”, which are manifested
linguistically in the subject-predicate structure. He proposes two constraints
relevant to the information structure of IUs in discourse. First, the starting
point is usually a given—or occasionally an accessible, but rarely a new—
referent (if it is new, it is normally of trivial importance); hence, subjects are
governed by the “light starting point constraint” (the “light subject constraint”
in Chafe 1994), where the lightness could be taken to refer to non-newness.
Second, the added information typically contains one new concept, though it
may also contain some accessible or given concepts; hence, the “heavy added
information constraint”. These two constraints appear to be in accord with
Halliday’s (1967b, 1985) view of the “unmarked” structure of information
within the “information unit” as being “given followed by new”, Firbas’ (1992)
claim of the correspondence of the order of words in a sentence and an increase
in “communicative dynamism” (CD),14 and Du Bois’ (1987) remark that given
status is the norm for nominal references, whereas new status is the norm for
verbal predicates.
Chafe (1987, 1988, 1994) observes that the majority of substantive IUs in
English spoken discourse take the form of complete single clauses. This means
that an English speaker’s typical way of verbalizing a focus of consciousness is
through the format of a clause. More specifically, Chafe (1987: 38) states that
“the clause appears to be the prototypical IU type, from which most other types
Chapter 2
are derived, or are deviations”. Likewise, Halliday (1967b: 242), concerning the
relationship between the realization of his “information unit” phonologically,
in the “tone group”, and syntactically, in the clause, states that “in the unmarked
case (in informal conversation) the information unit will be mapped on to
the clause”. That is, Chafe’s as well as Halliday’s claim is that the clause is the
syntactic exponent of the IU, and thus clause linkage is the predominant type of
IU linkage in spoken English. I refer to this as the “clause centrality proposal”.
It has been shown that the mean proportion of single-clause substantive
IUs in spoken English is about 60–70% (Chafe 1988, 1994). Clausal IUs express
ideas of states or events, and usually each IU verbalizes a different state or event
from that which precedes it. That is, state and event ideas are highly transient
in active consciousness, constantly being replaced by other state and event
ideas. This dynamic process of successive and transient activations seems to be
analogous to Givon’s (1987) view of the foreground-background alternation
discussed in Section 2.2.1. The continual replacement of state and event ideas
also reflects that our consciousness is in constant change, restlessly moving
from one idea to another. The sequence in (2.9) below illustrates such rapid
progression from one state/event idea to the next (Chafe 1994: 66).
(2.9) a. A: ++ Cause I had a ++ a thick patch of barley there,
b. B: ++ mhm,
c. A: + about the size of the + kitchen and living room,
d. ++ and I went over it.
e. + and then,
f. ++ when I got done,
g. I had a little bit left,
h. + so I turned around,
i. and I went and sprayed it twice.
j. + and it’s just as yellow as ++ can be.
Example (2.9) contains eight substantive clausal IUs. Of these, (2.9a), (2.9c),
(2.9g), and (2.9j) verbalize ideas of states, whereas (2.9d), (2.9f), and (2.9h, i)
verbalize ideas of events. This excerpt shows how one newly activated state/
event idea is rapidly replaced by another newly activated state/event idea
as clausal IUs are successively produced. This means that ideas of states
or events expressed in predicates are normally new, in contrast to ideas of
referents, which are more persistent (for example, the referent verbalized as
I is repeatedly used in (2.9)) (cf. Section 2.4.5).
Related to the clause centrality proposal discussed above is Pawley and
Syder’s (1983: 564–565) “one clause at a time constraint”. This constraint ba-
Information flow in spoken discourse
sically suggests that humans can encode or formulate only the contents of “one
clause at a time”. The clause is thus treated as the basic unit for information pro-
cessing and segmentation in human spoken discourse. They argue that there is
a fundamental limit in humans’ verbal processing such that in a single span of
attention focus, it is only possible to plan the content of a novel clause of up to
about ten words; therefore, to encode the full lexical content of a longer novel
sequence requires two or more separate encoding operations. The one clause at
a time constraint, according to Pawley and Syder, allows speakers to maintain
maximal fluency within the limits of their encoding capacity, and underlies the
characteristic “clause-chaining” style of spontaneous connected discourse, i.e.
the preponderance of conjoined and adjoined clauses with much less use of
subordination than in formal writing (cf. Haiman & Thompson 1988). Givón
has also identified the mental proposition which codes some cognized state or
event and which often surfaces as a clause in natural connected discourse as the
basic unit of information storage and discourse processing. That is, for Givón
(1990: 896), just as for Chafe, Halliday, and Pawley and Syder, “the basic unit of
stored information in coherent discourse is the mental proposition that stands
for some state or event”.
Chafe (1988) discusses two major kinds of linkages that exist between IUs
in conversational English. First, there are the linkages signaled by intonation
alone, where falling intonation signals closure of an idea or idea sequence, and
continuing intonation signals continuation from one idea to another. Second,
there are the linkages signaled by explicit connective words such as and, but,
and so. He reports that the first “non-connective type” accounted for about
44% of the cases of IU linkages in his study, whereas the remaining 56%
of the cases involved more specific linkage markers. In addition, half of the
“connective-type” IU linkages involved the conjunction and.
The IU sequences (2.10)–(2.12) show how IUs are linked to one another in
English conversational monologues.
(2.10) a. A: ++ I came home,
b. I was really exhausted,
c. I was eating a popsicle,
d. ++ I was sitting there in my chair,
e. ++ just eating my popsicle,
(2.11) a. A: and then another day,
b. ++ it was really hot,
c. it was in the summer and,
d. + my room was small.
Chapter 2
Beckman (1988), Sugito (1989, 1990), Hirst and Di Cristo (1998), Iwasaki
(2002), or Venditti (2003) (cf. Section 3.3).
i. zettai.
never
‘never’
Of the nine IUs contained in (2.14), only one, (2.14g), is a semi-clause consist-
ing of a verb with a missing subject, according to Iwasaki and Tao (1993). The
remaining eight IUs are non-clausal, phrasal units; (2.14a), (2.14c), (2.14f),
and (2.14h) are nominal, (2.14b) and (2.14i) are adverbial, and (2.14d, e) are
adjectival IUs. We can observe that five of the eight phrasal IUs contain the in-
teractional particle ne in the IU-final position. It should be noted that speaker
A could have conveyed the same ideational content as the nine short IUs col-
lectively do by means of one full-clausal IU, which could roughly correspond
in English to ‘Mr. Yamato never values our software job’.
Iwasaki and Tao (1993) showed that the proportion of NP IUs in Japanese
is higher than in English, but failed to show where those NPs appeared, or what
forms those NP IUs took syntactically—for example, whether they occurred
as clause-internal constituents or as clause-external independent IUs. Clearly
their study lacks such detailed information. Moreover, they coded (2.14g) as a
semi-clause; but given that the nine IUs collectively constitute a full proposition
meaning that ‘Mr. Yamato never values our software job’, I would code (2.14g)
as a VP, which is clearly propositionally dependent on the other phrasal IUs.
That is, in my view, the nine phrasal IUs, (2.14a–i), collectively constitute a
full clause, with (2.14a) functioning as the subject, and (2.14g) functioning as
the verbal predicate of that multi-IU full clause. (Alternatively, (2.14a) could
be coded as a left-dislocated NP topicalizing the referent ‘Mr. Yamato’, and
(2.14b), as a clause-external adverbial IU.) The two post-predicate phrasal
IUs, (2.14h) and (2.14i), could be taken to function as the direct object
and an adverbial intensifier, respectively, of the negated verb ‘do’ in (2.14g)
(see Section 4.3). (Note that the o-marked NP ‘us’ in (2.14c) cannot be the
direct object of the verb ‘do’, and the ga-marked NP ‘evaluation’ in (2.14f),
which is modified by the preceding adjectival phrasal IUs, (2.14d, e), lacks its
predicate element.)
In short, Iwasaki and Tao’s (1993) coding is too simple in that it treats
each IU as a syntactically complete clausal or phrasal discourse unit; it does
not consider the possibility that multiple IUs will constitute a clause. Another
drawback in their coding is that they looked at only the surface grammatical
forms of IUs in distinguishing “clauses” from “phrases”, thereby eliminating
the possibility that some of what we formally identify as “phrases” may
communicate a complete proposition as “clauses” normally do (cf. Ono &
Chapter 2
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.
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
Given in (2.16) is one of his substantive IU examples that contain all of the four
functional components in the linear order specified in (2.15).
(2.16) ano tabi nante hitoride shita koto nakatta no ne?
pf trip sof alone do-past nml exist-neg-past nml fp
[LD][ ID ][CO][IT]
‘uh (I) had never taken a trip alone.’
The IU (2.16) begins with ano ‘uh’, a pause filler which constitutes the lead
component. This is followed by the ideational component, where a full propo-
sition is conveyed by the subjectless semi-clause ‘(I) had never taken a trip
alone’. The nominalizer no, a device which signals a cohesive tie between ele-
ments in discourse, comprises the cohesive component, according to Iwasaki.
The interactional particle ne, placed IU-finally with rising intonation, attempts
to involve the addressee in the current speech situation; this constitutes the
interactional component.
Iwasaki (1993), using the four codes, analyzed IUs produced in three
types of Japanese spoken discourse (i.e. narratives, telephone and face-to-
face conversations). The IUs that he analyzed include both substantive and
regulatory IUs (cf. Section 2.4.4). The main results are presented in (2.17).
(2.17) ID 34%
ID CO 22%
ID IT 17%
LD ID 2%
ID CO IT 12%
LD ID CO IT 0.3%
Chapter 2
It was found in his study that the IU in spoken Japanese typically consists
of no more than two functional components. More specifically, IUs with one
or two components accounted for 85%, whereas IUs with three components
accounted for 13%, and IUs with four components occupied only 0.3%. In
particular, as shown in (2.17), the one- and two-component IUs which include
the ideational [ID] component (i.e. ID, ID-CO, ID-IT, and LD-ID) occupy
75% of all the IUs analyzed (87% if the three-component type ID-CO-IT is
added). Among them, the most popular functional structure type is the one
which consists of the ideational component only (i.e. ID), occupying 34%. On
the other hand, the types ID-CO and ID-IT jointly occupy 39%; this means
that those types consisting of the ideational component with the cohesive
or interactional component are similarly the preferred pattern. In addition,
of all the IUs examined, those which include the interactional component
accounted for 33%.
Iwasaki (1993) interprets these findings as follows. First, the most impor-
tant task of IUs is to convey ideational information. (This is also reflected by
the large proportion of idea-conveying substantive IUs that constitute a given
discourse; cf. Section 3.5.3.) Second, while the IU in spoken Japanese tends
to communicate ideational information alone, it more often than not conveys
additional information, cohesive or interactional, at the same time. That is,
“Japanese IUs have a built-in mechanism which allows the speaker to attend
to different concerns of communication other than ideation itself ” (Iwasaki
1993: 50). Third, there is a constraint operating in spoken Japanese that per-
mits speakers to incorporate up to two functions within one IU. Iwasaki relates
the “no more than two functions per IU” constraint formulated in his study to
the cognitive limitation on how much information speakers can handle within
an IU—particularly to Pawley and Syder’s (1983) “one clause at a time con-
straint” (discussed in Section 2.4.6), which limits an IU to the expression of
no more than one proposition, and to Chafe’s (1987) “one new concept at a
time constraint” (to be discussed in Section 2.6.2), which limits an IU to the
expression of no more than one piece of newly activated information.
Most importantly, Iwasaki (1993) explains the occurrence of phrasal IUs,
which he claims to be pervasive in Japanese discourse, in terms of the multi-
functionality of Japanese IUs discussed above. He argues that a strong moti-
vational factor responsible for the phrasal IU structure in spoken Japanese is
“the multi-faceted task which the speaker must carry out in one IU” (Iwasaki
1993: 50)—the coding of non-referential, speaker-hearer interactional infor-
mation in addition to propositional, ideational information. That is, the fre-
quent use of the “partial propositional strategy”, i.e. use of phrasal IUs (as
Information flow in spoken discourse
but not in A. Du Bois (1987) also proposes what he calls the “One Lexical
Argument Constraint” (i.e. avoid more than one lexical argument per clause)
and the “Non-lexical A Constraint” (i.e. avoid lexical A’s). These grammatical
constraints, together with the pragmatic constraints given above, constitute
what he terms “Preferred Argument Structure” (PAS).
Many studies have provided evidence for the cross-linguistic applicabil-
ity of the quantity and role constraints of PAS. Among the most recent stud-
ies, Ashby and Bentivoglio (1993), Smith (1996), and Kärkkäinen (1996), for
example, have shown that PAS holds in spoken French and Spanish, mod-
ern Hebrew, and American English discourse, respectively (see also Du Bois
2003; Du Bois et al. 2003; O’Dowd 1990; Thompson 1997; Thompson &
Hopper 2001). Japanese spoken discourse, however, has not been investigated
extensively in terms of PAS or the preferred information structure of the IU
(cf. Iwasaki 1985; Section 2.4.5). It therefore remains to be explored whether
Japanese IUs follow the same pattern as those of the other languages in the
arrangement of given and new information and in the type and number of ar-
gument NPs to be expressed within an IU as well as a clause. We will address
this issue in Chapter 5.
Given the constraints on the flow rate of new information per discourse
unit proposed by Givón, Chafe, and Du Bois, it seems that we can safely say
that a segment of spoken discourse, or an IU is subject to a strong limitation on
how much new information it can express. This would mean that thought, or at
least language production, proceeds in terms of one activation of a previously
inactive idea at a time; and the speaker, and presumably the listener as well, is
able to handle no more than one new idea at a time (Chafe 1994: 109).
Chapter 3
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).
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
The data that I collected and analyzed are face-to-face, two-party conversations
between native speakers of Tokyo Japanese. The participants are 16 male and 16
female university undergraduate or graduate students, all of whom are in their
twenties and different individuals. The average age of the 32 participants is
23.7. The data were provided by 8 male-male dyads and 8 female-female dyads.
The two participants in each of the 16 dyads are friends who have known each
other very well and had conversed many times before the recording was done
for this study. In order to obtain naturalistic conversational data, the researcher
did not set up the conversations; instead, the participants self-administered
the recordings. In other words, the data were collected in such a way that
the participants took the initiative in recording their own conversations; most
of the data were recorded in places such as restaurants, coffee shops, the
participants’ apartments, or dormitory rooms. Each of the 16 dyadic pairs of
participants audio-recorded their own conversations for about 60 minutes as
they conversed continuously in a casual and spontaneous manner. The last 45-
minute segment was then selected from each conversation and was transcribed
for analysis. Thus the data base for the present study consists of 720-minute
(i.e. 12-hour) dyadic Japanese conversations between same-sex friends.
A summary of the Japanese conversational data used for this study is given
in Table 3.1, where the participants’ names are pseudonyms. The boldfaced
codes F1 . . . . . F8 indicate the eight 45-minute female dyadic conversations,
whereas the boldfaced codes M1 . . . . . M8 indicate the eight 45-minute male
dyadic conversations.1
Table 3.2 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
Dyad Topics
Spoken Japanese, like spoken English, exhibits a relatively high degree of align-
ment between intonation contours and the projection of turn completion.
The falling intonation in Japanese thus indicates finality or completeness or
idea closure, often used with declarative clauses. The continuing intonation
and non-question rising intonation, in contrast, express non-finality or in-
completeness. The rising pitch also indicates interrogativity, as in English.
The remaining two contours serve more or less “non-grammatical” functions
which appear to be peculiar to Japanese. The rise-fall pitch contour expresses
a “discoursal” meaning which involves the speaker’s appeal to the listener, or
the speaker’s immediate communicative expectations of the hearer; this con-
tour type functions to seek agreement or to impose the speaker’s opinion
on the hearer. The rise-fall-rise pitch contour, which expresses an “emotive”
or “attitudinal” meaning, shows the speaker’s doubt or dissatisfaction (see
Beckman & Pierrehumbert 1986; Bolinger 1989; Chafe 1988; Hirst & Di Cristo
1998; Iwasaki 1992, 2002; Pierrehumbert & Beckman 1988; Pierrehumbert &
Method of the study
Hirschberg 1990; Sugito 1989, 1990; Vandepitte 1989; Venditti 2003; Ward &
Hirschberg 1985).
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.
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.
The topics of the 16 conversational segments are listed in Table 3.4. The codes
F1, F2, . . . . . , and F8 indicate the segments containing 100 substantive IUs
selected from the conversations F1, F2, . . . . . , and F8, respectively. Likewise,
the codes M1, M2, . . . . . , and M8 indicate the 100-IU segments selected from
the conversations M1, M2, . . . . . , and M8, respectively.
The selected segments are interactive conversational portions instead of
narrative portions—portions where active conversational turn-taking is tak-
ing place, and thus the co-participants’ verbal contributions to discourse pro-
duction are fairly balanced between the two.6 In addition, the segments were
carefully selected in order for different topics and different types of conver-
sational discourse involving different activities to be represented in the data
(e.g. talking about the immediate environment in which the co-participants are
Method of the study
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.
In order to obtain 100 substantive IUs, I coded the IUs in each conversation
for four categories: substantive, regulatory, fragmentary, and uncodable. As
explained in Section 2.4.4, substantive IUs convey substantive ideas of events,
states, or referents. Fragmentary IUs are units which were begun but were not
completed for various reasons (e.g. because of the speaker’s “false starts” or
interruptions by the other interlocutor). Regulatory IUs function to regulate
conversational interaction or information flow. This category consists of the
following five subtypes: (a) textual subtype, which functions to indicate a
particular kind of linkage between IUs; (b) interactional subtype, which signals
the speaker’s attentiveness to what the interlocutor is saying and his/her
comprehension of it; (c) cognitive subtype, which indicates ongoing mental
processes on the part of the speaker; (d) validational (or evaluative) subtype,
which concerns itself with the speaker’s judgment of the validity, or the
speaker’s evaluation, of the information being conveyed; and (e) combination
type, which incorporates multiple regulatory functions within an IU. For
example, the regulatory IU ano dakara sa has three features: cognitive (marked
by the pause filler ano ‘uh’), textual (marked by the conjunction dakara ‘so’),
and interactional (marked by the final particle sa) (cf. Section 6.1). Finally,
uncodable IUs involve uncertainty or inaudibility of words, which is marked
by parentheses in the transcripts.
Chapter 3
Table 3.5 lists the number of the four types of IUs contained in the 16 con-
versational segments (where [S]=substantive, [R]=regulatory, [F]=fragmentary,
and [X]=uncodable).
Table 3.5 indicates that the selected segments involving a sequence of 100
substantive IUs included, on average, 21.1 regulatory IUs, 1.2 fragmentary IUs,
and 1.4 uncodable IUs. This means that of all the IUs contained in the 16 seg-
ments, 81% are substantive, 17% are regulatory, 0.9% are fragmentary, and
1.1% are uncodable. The largest number of regulatory IUs were produced in F6,
where 24% of all the IUs in the segment are regulatory.7 Importantly, the results
indicate that the overwhelming majority of IUs that speakers produce in con-
versational interaction are substantive units that transmit ideational content.
Dyad S R F X Total
F1 100 20 1 1 121
F2 100 13 1 1 116
F3 100 23 3 1 127
F4 100 31 1 1 133
F5 100 15 1 3 119
F6 100 32 2 1 135
F7 100 19 1 0 120
F8 100 20 0 1 121
Total 800 173 10 9 992
(81) (17) (1.0) (0.9) (100%)
Average 100 21.6 1.3 1.1 124.0
M1 100 14 1 1 116
M2 100 20 0 3 123
M3 100 26 3 4 133
M4 100 16 1 1 118
M5 100 15 1 0 116
M6 100 22 1 2 125
M7 100 28 1 0 129
M8 100 24 1 2 127
Total 800 165 9 13 987
(81) (17) (0.9) (1.3) (100%)
Average 100 20.6 1.1 1.6 123.4
TOTAL 1,600 338 19 22 1,979
(81) (17) (0.9) (1.1) (100%)
AVERAGE 100 21.1 1.2 1.4 123.7
Chapter 4
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?
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
into two types: argument and predicate; similarly, the AP was classified into
attributive and predicative.2 Type 3 IUs are phrasal IU elements of a larger
multi-IU clause; they consist of the following seven subtypes:
[CVPp] = verb phrase
[CNPa] = argument NP
[CNPp] = predicate NP
[CAPa] = attributive AP
[CAPp] = predicative AP
[CAvP] = adverbial phrase
[CXP] = mixed phrase (e.g. [CNPa + CAvP])
Type 4: Independent phrasal IU: a phrase that is not integrated into the
clausal structure and that as a clause-external element, mainly functions to
perform “peripheral” work rather than the “core” task of communicating
propositional content, unlike Types 1–3. For example, left/right-dislocated
phrases and copied/restated postposed phrases belong to this group (see
Section 4.5.2 for detailed discussion). Type 4 “detached” phrasal IUs consist
of the following five subtypes:
[VP] = verb phrase
[NP] = noun phrase
[AP] = adjective phrase
[AvP] = adverbial phrase
[XP] = mixed phrase (e.g. [NP + AvP])
In addition, I used six codes as superordinate codes, namely, [FC], [VP],
[NP], [AP], [AvP], and [XP]. These boldfaced codes represent IU syntactic
types based on grammatical categories. For example, as shown below, [FC]
(full clausal IUs) consists of two types, i.e. [FC] and [CFC]; [VP] (VP IUs)
consists of four types, i.e. [PVP], [CVP], [CVPp], and [VP]; [NP] (NP IUs)
consists of five types, i.e. [PNP], [CNP], [CNPa], [CNPp], and [NP].
[FC] = [FC] + [CFC]
[VP] = [PVP] + [CVP] + [CVPp] + [VP]
[NP] = [PNP] + [CNP] + [CNPa] + [CNPp] + [NP]
[AP] = [PAP] + [CAP] + [CAPa] + [CAPp] + [AP]
[AvP] = [CAvP] + [AvP]
[XP] = [CXP] + [XP]
In this study I defined a clause as “a unit of discourse which consists of
a predicate and its associated core arguments (i.e. a subject and an object,
Chapter 4
which may or may not be overtly expressed) and adjuncts (i.e. locative,
temporal, and manner adverbials performing circumstantial functions, which
are optionally present)” (Chafe 1980b; Dixon 1979; Du Bois 1987). The
arguments and adjuncts may be in the pre-predicate position (unmarked
word order) or in the post-predicate position (marked word order involving
postposing) (cf. Section 1.3). I also defined a clause as “a propositionally
complete discourse unit which codes some state or event” (Givón 1990). That
is, the basic assumption in my coding is that a clause consisting of at least a
predicate conveys a complete proposition.3 Thus, when a core argument or an
adjunct appeared independently as a “topic” IU, marked by the so-called “topic
marker” wa (or its informal version tte) (cf. Hinds et al. 1987), I considered it
to be within the clausal structure, coding it as a Type 3 phrasal IU (see (4.6b)
below). On the other hand, when a wa-marked NP IU functioned as a “base-
generated”, “genuine” topic4 (Shibatani 1990) which would constitute neither a
core argument nor an adjunct, I considered it to be outside the clausal domain,
coding it as an independent NP IU, namely, Type 4 (see (4.19a) below).
As noted above, a “full clause” is defined to consist of an overtly expressed
subject plus a predicate, whereas a “semi-clause” consists of a predicate only
without an explicit subject. This means that while they differ in whether the
subject is overtly expressed or assumed by the co-participants, both clause
types may contain an overtly expressed object. I coded phrasal IUs as part of a
multi-IU clause only when they were produced by the same speaker and only
when they were adjacently produced (i.e. when they were not interspersed with
other IUs in the transcripts). This means that phrasal IUs which have adjacently
been produced by the co-participants such that they constitute a collaboratively
completed clause were not counted as constituting a multi-IU clause; rather
they were considered independent (see (4.3)). This also means that phrasal IUs
with intervening IUs/propositions which would otherwise constitute a multi-
IU clause were coded as independent IUs (see (4.16) below). Further, I treated
phrasal IUs which have been produced adjacently to a propositionally complete
single-IU or multi-IU clause such that an element of that clause is copied or
restated as being outside the clausal domain (see (4.4)). Accordingly, as will
be shown in Section 4.5.2, Type 4 IUs typically includes, besides “genuine”
topics, copied/restated postposed phrases, left/right-dislocated phrases, and
repeated phrases.
Observed IU examples are provided in (4.1)–(4.8). Examples (4.1)–(4.4)
illustrate Type 1 or Type 4 independent IUs. (In the examples below, assigned
codes are given in square brackets.)
Syntactic structure of the intonation unit in conversational Japanese
Examples (4.1) and (4.2) illustrate Type 1 independent clausal IUs. The IU
(4.1) is a full clause which consists of the overt subject NP ano hito ‘he’ and
the nominal predicate insei da ‘(is) a graduate student’. The IU (4.2) is a semi-
clause which consists of the verbal predicate atta ‘met’, where the subject ‘I’ and
the direct object ‘her’ are assumed, taking zero forms. Both (4.1) and (4.2), as
single-IU clauses, independently convey a complete proposition.
Examples (4.3a) and (4.4b), on the other hand, illustrate Type 4 indepen-
dent phrasal IUs. The sequence in (4.3) is an example of the collaboratively
completed clause. Given that the two speakers, E and R, are involved, I did not
code (4.3) as a multi-IU clause; rather I coded (4.3a) sangatsu no ‘of March’
as an independent attributive AP IU (i.e. [AP]), and (4.3b) juukyuu? ‘(Is it)
nineteenth?’ as an independent semi-clause which solely consists of the nom-
inal predicate (i.e. [PNP]). The IU (4.3a), uttered by speaker E but convey-
ing only a partial proposition, is made propositionally complete by speaker
R’s immediate suppliance of (4.3b). (If (4.3a) and (4.3b) had been uttered by
the same speaker, they would be coded as [CAPa] and [CNPp], respectively.)
The IU (4.4b) seikaku ga ‘(their) characteristics’ is an example of the copied
post-predicate nominal phrase. Given that the NP IU (4.4b) cannot be inte-
grated into the immediately preceding full clause (4.4a) ‘(their) characteristics
differ’, but simply repeats the subject NP already uttered in (4.4a), (4.4b) is a
Chapter 4
The IUs (4.6a) and (4.7a, b) exemplify Type 2 IUs—clausal IUs as part of multi-
IU clauses, whereas (4.5a, b), (4.6b), and (4.8a, b) exemplify Type 3 IUs—
phrasal IUs as part of multi-IU clauses. The sequence in (4.5) is a multi-IU full
clause which is composed of the subject argument NP (4.5a) ‘this Sunday’ and
the predicate NP (4.5b) ‘Easter’; these two nominal phrasal IUs jointly convey
a full proposition. Example (4.8) is another multi-IU full clause; this sequence
is comprised of the predicate NP IU (4.8a) and the post-predicate mixed
phrasal IU (4.8b) (which consists of the adverbial phrase hotondo ‘almost’
Syntactic structure of the intonation unit in conversational Japanese
Table 4.1 displays the distribution of the 20 IU syntactic structure types as well
as the four major syntactic types.
Chapter 4
Clausal IU Phrasal IU
Independent Part of multi- Part of multi- Independent Total
IU IU clause IU clause IU
[FC] 263 [CFC] 40 [FC] 303
(19%)
[PVP] 392 [CVP] 106 [CVPp] 34 [VP] 7 [VP] 539
(34%)
[CNPa] 94 [NP] 112 [NP] 420
[PNP] 174 [CNP] 17 [CNPp] 23 (26%)
[CAPa] 18 [AP] 14 [AP] 131
[PAP] 81 [CAP] 7 [CAPp] 11 (8%)
[CAvP] 90 [AvP] 82 [AvP] 172
(11%)
[CXP] 27 [XP] 8 [XP] 35
(2%)
Total 910 170 297 223 1,600
(57%) (11%) (18%) (14%) (100%)
Table 4.1 shows that of the 1,600 substantive IUs produced in the 16
conversational segments, (a) 68% are clausal IUs, whereas 32% are phrasal
IUs; (b) 71% are independent clausal or phrasal IUs, whereas 29% are clausal
or phrasal IUs constituting part of multi-IU clauses; and (c) 86% are “core”
IUs which are centrally concerned with the communication of propositions
(Type 1–3 IUs), whereas 14% are “peripheral”, “redundant” phrasal IUs placed
outside the clausal structure (Type 4 IUs). It also indicates that of the four
major IU syntactic types, Type 1 independent clausal IUs were most frequently
produced, accounting for 57% (see Figure 4.1). In addition, of the six IU
syntactic types based on grammatical categories, IUs consisting of verbal
predicates or verb phrases (i.e. [VP]) and IUs consisting of nominal predicates
or nominal phrases (i.e. [NP]) were most frequently produced, occupying 34%
and 26%, respectively. Full clausal IUs as independent IUs or as elements of
multi-IU clauses (i.e. [FC]) were third most frequently produced, accounting
for 19% (see Figure 4.2). We can also see from Table 4.1 that of the 20
IU syntactic types, the speakers produced [PVP] IUs (independent semi-
clausal IUs consisting of verbal predicates only) most frequently (25% of
the 1,600 IUs).7 This is followed in frequency by [FC] IUs (independent full
clausal IUs consisting of overt subjects plus predicates) (16%) and [PNP] IUs
(independent semi-clausal IUs consisting of nominal predicates only) (11%)
Syntactic structure of the intonation unit in conversational Japanese
1000
900
800
700
600
Frequency
500
400
300
200
100
0
Independent
Independent
part of multi-IU
part of multi-IU
phrasal IU
clausal IU
Phrasal IU as
Clausal IU as
clause
clause
IU syntactic type
(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
Type 1 IU N % Type 2 IU N %
[FC] 263 29 [CFC] 40 24
[PVP] 392 43 [CVP] 106 62
[PNP] 174 19 [CNP] 17 10
[PAP] 81 9 [CAP] 7 4
Total 910 100 Total 170 100
Chapter 4
600
500
400
Frequency
300
200
100
0
FC VP NP AP AvP XP
IU syntactic type
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%.
The main results regarding the preferred IU syntactic structure types are
the following: (a) of the 1,600 substantive IUs produced by the 32 speakers,
68% are clausal, as opposed to phrasal, IUs; (b) of the four major IU syn-
tactic types, independent clausal IUs which communicate complete proposi-
tions were most frequently produced (57%); (c) of the 20 IU syntactic types,
independent semi-clausal IUs consisting of verbal predicates without overt
Syntactic structure of the intonation unit in conversational Japanese
400
350
300
250
Frequency
200
150
100
50
0
FC
PVP
PNP
PAP
CFC
CVP
CNP
CAP
CVPp
CNPa
CNPp
CAPa
CAPp
CAvP
CXP
VP
NP
AP
AvP
XP
IU syntactic type
subjects (i.e. [PVP] IUs) were found to be the most preferred IU type (25%);
and (d) of the six syntactic types based on grammatical categories, IUs con-
sisting of verbal predicates or verb phrases (i.e. [VP] IUs) were most fre-
quently produced (34%). In sum, the results indicate that the Japanese con-
Type 3 IU N % Type 4 IU N %
[CVPp] 34 11 [VP] 7 3
[CNPa] 94 32 [NP] 112 50
[CNPp] 23 8
[CAPa] 18 6 [AP] 14 6
[CAPp] 11 4
[CAvP] 90 30 [AvP] 82 37
[CXP] 27 9 [XP] 8 4
Total 297 100 Total 223 100
Chapter 4
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.
The 1,600 substantive IUs contained not only pre-predicate elements (which
represent basic word order) but post-predicate elements (which represent
marked word order) as well (cf. Section 1.3; Note 3 in Chapter 1). I have
coded the frequency of post-predicate, or postposed phrases which occurred
in the 16 selected segments, in order to see what proportion of the IUs
involved postposing and to find out patterns in the speakers’ production of
those marked word order constructions. Table 4.4 presents the distribution of
post-predicate phrases among 10 IU syntactic types (the remaining 10 types
involved no postposings). The coded postposed phrases include nominals,
adjectives, adverbials, and mixed phrases.8
Table 4.4 indicates that of the 1,600 IUs, 7.7% (N=123) involved postpos-
ing. The postposed phrases listed in Table 4.4 can be grouped into three types,
depending on the relationship between the postposed phrase and the IU/clause
IU syntactic type N %
[FC] 23
[PVP] 7
[CVP] 1
Type 1-Total 31 25
[CNPa] 26
[CAPa] 6
[CAvP] 32
[CXP] 2
Type 2-Total 66 54
[NP] 20
[AP] 1
[AvP] 5
Type 3-Total 26 21
TOTAL 123 100
Chapter 4
in which it occurs (cf. Clancy 1982; Hinds 1976; Matsumoto 1995a, 1997c;
Ono & Suzuki 1992; Shibatani 1990). Examples are given in (4.9)–(4.11), where
postposed NPs are boldfaced.
Type 1: postposed phrases constituting the final part of clausal IUs, which may
be independent single-IU clauses, as in (4.9), or clausal IU elements
of multi-IU clauses. (Those constituting the final part of phrasal
predicate IUs such as [CVPp] were not found in the data.)
Type 2: postposed phrasal IUs constituting elements of multi-IU clauses, as in
(4.10b) and (4.10c).
Type 3: independent postposed phrasal IUs external to the clausal structure,
i.e. copied/restated postposed phrases, as in (4.11c).
Postposings are thus of two types: clause-internal (Type 1 and Type 2) and
clause-external (Type 3). Clause-internal postposed phrases, integrated into
the clausal structure, constitute part of single-IU clauses (Type 1), or part of
clausal IU elements (Type 1) or phrasal IU elements (Type 2) of multi-IU
clauses. As clause-internal elements, they participate in the communication of
propositions. Clause-external postposed phrases (Type 3), placed outside the
Syntactic structure of the intonation unit in conversational Japanese
clausal domain, only repeat or restate phrases already uttered in the immedi-
ately preceding clause. They take the form of copied/restated post-predicate
phrases or right-dislocated phrases, with resumptive pronouns remaining in
the preceding clause in the case of right-dislocation (see detailed discussion in
Section 4.5.2). The two types of postposings differ not only in terms of the IU
syntactic structure but also functionally, that is, they seem to serve different
discourse functions in conversational Japanese. The clause-internal postposing
can be seen as linked to the speaker’s focused attention on the information that
is most urgently sought within the flow of conversation, which in turn triggers
the backgrounding of the postposed information. Apparently, we can relate the
production of clause-internal postposed phrases to the dynamic foregound-
background alternation in connected discourse (cf. Section 2.2). The clause-
external postposed phrases, on the other hand, can be seen to be added for pur-
poses such as emphasis, repair, and further specification of already expressed
elements; this suggests that Type 3 postposing entails the speaker’s conscious
intentional strategic choices (cf. Clancy 1982; Fox et al. 1996; Kuno 1978a;
Maynard 1989; Saeki 1975; Schegloff 1979; Schegloff et al. 1977; Takami 1995;
Teramura 1984).
In sum, Type 1 and Type 2 post-predicate phrases, although they may be
backgrounded or defocused, constitute “core” elements of single-IU or multi-
IU clauses; they directly participate in the construction of propositions. Type 3
post-predicate phrases, in contarst, only repeat or restate “redundantly” the in-
formation already given in the immediately preceding clause, thus functioning
principally as a device of emphasis or reinforcement. Table 4.4 indicates that
79% of the post-predicate phrases involved clause-internal postposing, and
that 54% appeared as phrasal IU elements of multi-IU clauses (Type 2).
Table 4.5 presents the proportion of IUs which involved postposing by IU
syntactic type.
Table 4.5 indicates that the adverbial phrases, attributive adjective phrases,
and nominal phrases as part of multi-IU clauses exhibited the highest rate
of postposings ([CAvP]=36%, [CAPa]=33%, [CNPa]=28%). In accord with
this, the phrasal IUs as constituents of multi-IU clauses were most highly
postposed—22% of them constituted post-predicate elements. (By compar-
ison, only 3% of the clausal IUs and 12% of the independent phrasal IUs
involved postposings.) The speakers’ preferential production of postposed
phrases as phrasal IU elements of multi-IU clauses deserves further ex-
amination.
Detailed discussion of the clausal IUs, NP IUs, AP IUs, AvP IUs, and XP
IUs follows, focusing on the types and functions of NP IUs.
Chapter 4
Table 4.6 displays the proportions of full clauses vs. semi-clauses within Type 1
independent clausal IUs and Type 2 clausal IUs as part of multi-IU clauses (see
also Figure 4.4).
Table 4.6 shows that of the total number of clausal IUs (N=1,080), 72%
are semi-clauses, whereas 28% are full clauses. It indicates that in both Type 1
and Type 2 clausal IUs, the majority are subjectless semi-clauses as opposed
to full clauses, and they account for about 70–80%. We can also see from
Table 4.6 that of the 1,080 clausal IUs, 84% are independent single-IU clauses
communicating complete propositions, whereas 16% are clausal IUs constitut-
ing part of multi-IU clauses. The results additionally indicate that of the 1,600
IUs, (a) 49% are semi-clauses ([PVP] + [PNP] + [PAP] + [CVP] + [CNP] +
[CAP]); (b) 19% are full clauses ([FC] + [CFC]); and (c) 40% are independent
single-IU semi-clauses ([PVP] + [PNP] + [PAP]) (see Table 4.1).
Syntactic structure of the intonation unit in conversational Japanese
800
700
600
500
Frequency
Full clause
400
Semi-clause
300
200
100
0
Independent
part of multi-IU
Clausal IU
clausal IU
Clausal IU as
Total
clause
Clausal IU type
Further coding of the corpus has shown the following regarding the
structure of the full clauses as well as semi-clauses: (a) of the 263 [FC] IUs,
51% (N=135) consist of S(subject)+VP; (b) of the 40 [CFC] IUs, 75%
(N=30) consist of S(subject)+VP; (c) of the 647 independent semi-clausal IUs
([PVP] + [PNP] + [PAP]), 61% (N=392) consist of VP; and (d) of the 130
semi-clausal IUs as elements of multi-IU-clauses ([CVP] + [CNP] + [CAP]),
82% (N=106) consist of VP. This means that 55% of the full clauses, 64%
of the semi-clauses, and 61% of the clausal IUs contained verbal predicates.
This demonstrates the speakers’ notable preference for verbal over nomi-
nal/adjectival predicates. (The clausal IUs will also be discussed in terms of
transitive-intransitive bivalency in Section 5.5.)
In sum, the results presented above reveal that the Japanese interactants
prefer to produce IUs which are syntactically semi-clausal lacking explicit
subjects, especially propositionally complete independent semi-clausal IUs
which consist of verbal predicates. This finding is in accordance with Iwasaki
and Tao (1993), as discussed in Section 4.2.2.
Chapter 4
180
160
140
120
Frequency
100
80
60
40
20
0
PNP CNP CNPa CNPp NP
NP IU type
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).
NP IU type N %
[PNP] 174 41
[CNP] 17 4
[CNPa] 94 22
[CNPp] 23 6
[NP] 112 27
Total 420 100
Syntactic structure of the intonation unit in conversational Japanese
Table 4.7 shows that of the 420 NP IUs, (a) 55% are phrasal IUs, whereas
45% are clausal IUs; (b) 51% are NPs functioning as predicates, whereas 49%
are argument NPs or extra-clausal NPs; and (c) 68% are independent clausal or
phrasal NP IUs, whereas 32% are elements of multi-IU clauses. It also indicates
that of the five NP IU types, [PNP] (independent single-IU semi-clauses with
nominal predicates only) is the most frequently produced type, accounting for
41%. This is followed in frequency by [NP] (clause-external “detached” NP
IUs) (27%) and [CNPa] (argument NP IUs as constituents of multi-IU clauses)
(22%). (Of the 94 [CNPa] IUs, further coding has shown, 78% occurred within
the multi-IU full clauses, functioning as subject or object NPs, whereas 22%
constituted object NPs of the multi-IU semi-clauses.)
The occurrence of NP IUs in Japanese conversation is illustrated in (4.12).
In this excerpt, the female co-participants K and H are talking about one of
their common friends whom H claims to be too gaily dressed.
(4.12) a. H: nihon de isoona kakkoo.= [PNP]
Japan loc exist-likely appearance
‘(It’s) an appearance (which is) likely to be found in Japan.’
b. K: =ikeike. [PNP]
ikeike
‘(It’s) ikeike.’
c. H: +ikeike. ((laugh)) [NP]
‘ikeike’
d. K: ((laugh)) ano ko ga, [NP]
that girl nom
‘she’
e. [ dakara, ]
so
‘so’
f. H: [ano hito] shakaijin desho?= [FC]
that person working person tag
‘She is working, isn’t she?’
g. K: =janai no?
tag fp
‘I guess so.’
h. H: gakusei?= [PNP]
student
‘(Is she) a student?’
Chapter 4
i. K: =uu:n∼
‘I don’t think so.’
j. H: shakaijin da yo ne.@ [CNPp]
working person cop fp fp
‘(is) a working woman’
k. +are ne. [CNPa]
that fp
‘she’
l. K: ->ano: karaa kontakuto o kaitai tte
pf colored contact lenses acc buy-des qt
-->itteru hito. [PNP]
say-prog person
‘Well, (she is) the person who says that (she) wants to buy colored
contact lenses.’ (F8)
The sequence in (4.12) is composed of eight NP IUs. The IUs (4.12a, b),
(4.12h), and (4.12l) are [PNP] IUs, i.e. subjectless semi-clausal IUs with
nominal predicates. The IU (4.12j) is a predicate NP ([CNPp]), and the
IU (4.12k) is a post-predicate subject argument NP ([CNPa]); the two IUs
collectively form a multi-IU full clause. The IUs (4.12c) and (4.12d) are [NP]
IUs, i.e. clause-external independent phrasal NP IUs. The [NP] IU (4.12c) is
a repeated-as-affirmation NP, which functions to affirm that the term ikeike
(a slang for flamboyant appearance) supplied by K is exactly what H meant.
The [NP] IU (4.12d) was begun by K, but because of H’s interrupted yes/no
questions, seems to have failed to express the intended proposition. The
sequence contains one [FC] IU, (4.12f), which consists of two NPs, the subject
NP ano hito and the predicate NP shakaijin. (Note that the two regulatory
IUs—(4.12g) janai no? (consisting of two final particles with rising intonation)
and (4.12i) uu:n∼ (consisting of informal ‘yes’ with doubt-indicating rise-
fall-rise contour)—function as somewhat “softened” affirmative and negative
response, respectively, to the preceding questions.) The sequence contains 12
NPs, two of which, (4.12a) and (4.12l), are complex NPs with relative clauses
(cf. Note 7). (For the syntactic forms, grammatical roles, and information
statuses of NPs contained in the 420 NP IUs, see Section 5.3.)
In what follows I will explore the types and functions of [NP] IUs.
Syntactic structure of the intonation unit in conversational Japanese
My analysis has identified five types of independent phrasal NP IUs. These are
listed and defined below (the same classification can be applied to the other
syntactic categories of independent phrasal IUs):
Type 1: “Stray” NPs—NPs that are “detached” from the immediate discourse
context, where an IU which is yet to be uttered to produce a com-
plete proposition is not supplied by the speaker who has produced
that NP. “Stray” NPs result when other-interruptions, self-repairs, or
collaborative completions occur.
Type 2: “Lead” NPs—NPs that function as “leads” to the following clausal IU
which expresses a complete proposition (cf. Durie 1994).
Type 3: Topic NPs or left-dislocated (LDed) NPs that are outside the clausal
structure (cf. Ashby 1988; Geluykens 1992).
Type 4: Postposed NPs or right-dislocated (RDed) NPs produced immediately
following a single-IU/multi-IU clause.
Type 5: NPs repeated as listener responses—tokens of acknowledgment,
agreement/disagreement, or clarification request—to the IU that the
speaker has just uttered.
CLAUSES
NPs in single-IU and multi-IU clauses
COMMUNICATION OF PROPOSITIONS
Postposed/RDed NPs
Repeated NPs
The independent, detached NP IUs can be classified into two groups, which
seem to differ in the degree of relatability to the surrounding IUs/clauses.
“Stray” NP IUs cannot be syntactically attached/related to any of the self-
produced IUs (i.e. IUs produced by the speaker who uttered the NPs) in
Chapter 4
NP type N %
Type 1 Stray 24 21
Type 2 Lead 19 17
Type 3 Topic/LDed 30 27
Type 4 Postposed/RDed 18 16
Type 5 Repeated 19 17
Others (Vocatives) 2 2
Total 112 100
the immediate context. Topic NPs and postposed NPs, on the other hand,
can be syntactically related, although not integratable, to the immediately
following/preceding self-produced clause(s). “Lead” NPs can be referen-
tially/informationally relatable to the immediately following self-produced IU
clauses. Repeated NPs can be interactionally relatable to the immediately pre-
ceding other-produced IU clauses (see detailed discussion below). In brief, in-
dependent NP IUs appear to differ in the degree to which they are “detached”
from the clausal structure in the immediate discourse context (cf. Croft 1995;
Ono & Thompson 1994; Tao 1996). Importantly, as I will show shortly, the pro-
duction of such clause-external independent NP IUs is motivated by various
discourse-pragmatic, interactional, and information-flow factors.
Table 4.8 presents the frequency of the five types of independent phrasal
NP IUs produced in the 16 conversational segments (see also Figure 4.6).
Table 4.8 shows that of the 112 independent NP IUs, topic/LDed NPs
(27%) and stray NPs (21%) accounted for 48%, with the remaining three types
occupying comparable percentages (16–17%). (The category “others” includes
vocatives, i.e. proper names used in address.) Further analysis has shown the
following: (a) of the 30 topic/LDed NP IUs, 63% are “genuine” topics, and
10% are LDed NPs (the remaining 27% are NPs preposed for comparison
or as alternatives in choice questions); (b) of the 18 postposed/RDed NPs,
44% are restated postposed NPs, 33% are copied postposed NPs, and 23% are
RDed NPs; and (c) the 19 repeated-as-listener-response NPs consist of tokens
of acknowledgment (47%), affirmation/agreement (21%), and disagreement
(5%) (the remaining 27% are concerned with clarification request).
I will elaborate on each of the five types of independent NP IUs in order
below with illustrative examples.
Syntactic structure of the intonation unit in conversational Japanese
30
25
20
Frequency
15
10
0 RDed
Postposed/
Lead
Topic/LDed
Repeated
Others
Stray
NP type
Stray NPs
The first type, “stray” NP IUs, can be further divided into the following four
subtypes:
Type 1: NP IUs that were uttered but totally “left alone” or “detached”
from the surrounding context, because neither the speaker himself/herself nor
the hearer subsequently supplied an IU which would complete the intended
proposition. Type 1 stray NPs involve discoursal operations such as other-
interruptions and self-repairs, and in most cases, the remaining-to-be-uttered
elements are non-recoverable.
Thus the NP IU uttered by the first speaker, together with the immediately
following IU supplied by the other, jointly constitute a full proposition. The
resulting proposition/clause, however, may or may not be equal to the one that
the original speaker intended to produce (cf. Clancy et al. 1996; Lerner 1991;
Ono & Thompson 1996).
Type 3: NP IUs that are “detached” only from the immediate environment;
the speaker who uttered the NP subsequently supplies another IU to make a
partial proposition complete, but supplies it non-adjacently to the “original”
NP IU, that is, other IUs/propositions intervene. Type 3 stray NP IUs invariably
involve insertion sequences, which vary in length from fairly short to rather
long (e.g. 10-IU sequences). Like Type 1, Type 3 stray NPs may involve other-
interruptions or self-repairs.
Type 4: “argument” NP IUs that are “stray” in the sense that predicate ele-
ments that would be formally required to express a full proposition are neither
self-supplied nor other-supplied, but rather assumed by the conversational co-
participants because those elements are recoverable from the context. Type 4
stray NP IU differs from the other types in that its intended proposition can be
inferred from the context shared by the co-participants; thus it virtually func-
tions as a “clause” expressing a full proposition where some elements can be
taken to be ellipted. It may be said that Type 4 stray NPs are “formally phrases
but functionally clauses”. They are “stray” NPs uttered in the “right” context,
where elements yet to be uttered to produce a complete proposition can be
recovered.9
Table 4.9 summarizes the main features of the four types of stray NP IUs
and presents the frequency of each type. This shows that Type 1 is the most
frequently used stray NP type, accounting for 54%.
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
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
Topic/Left-dislocated NPs
The third type, which involves NPs functioning as “genuine” topics or LDed
clause-external elements, is illustrated in (4.19) and (4.20).
(4.19) →a. M: datte senhyaku tte koto wa, [NP]
because 1,100 qt thing top
‘because speaking of the requirement that (the SAT score must be)
1,100,’
Chapter 4
b. tatoeba, [AvP]
for example
‘for example’
c. +suugaku o manten totta to shitemo, [PVP]
math acc full mark take-past qt do-con
‘Even if (you) get a full mark in math,’
d. suugaku ga manten de yatto sanbyaku.@ [FC]
math nom full mark cop barely 300
‘The full math score would barely leave (the other section’s score)
300.’
e. ++desho?
tag
‘wouldn’t it?’
f. de,
and
‘and’
g. manten toreru tte no muzukashii kara, [FC]
full mark take-pot qt nml difficult because
‘Because getting a full mark is difficult,’
h. ja tatoeba nanahyaku to shitara, [PVP]
so for example 700 qt do-con
‘So, suppose (it’s) 700, for example’
i. baabaru yonhyaku da mon.@ [FC]
verbal 400 cop nml
‘the verbal (section’s score will be) 400.’
j. +dakara taihen da na: to omotte, [PVP]
so hard cop fp qt think-and
‘So (I) think that (it’s) hard.’ (F3)
(4.20) a. S: suzuki mari to ka yuu. [PVP]
Suzuki Mari qt q say
‘(She is) named Mari Suzuki.’
b. nandaka bairingyaru na n da soodesu yo. [PNP]
sof bilingual gal cop mnl cop hear-pol fp
‘(I) hear (that she is) a bilingual gal.’
c. J: ->aa nanka shuukanshi ni
yeah something weekly magazine loc
-->deteta na.@ [FC]
appear-past-sta fp
‘Yeah, (there) was some article (about her) on a weekly magazine.’
Syntactic structure of the intonation unit in conversational Japanese
Postposed/Right-dislocated NPs
Repeated NPs
nal NP (or the only NP in the case of semi-clausal IUs) contained in the im-
mediately preceding other-produced single-IU clause. Of the three patterns,
the lead and postposed NPs exemplify intra-speaker “redundant” repetitious
use of NPs, whereas the repeated NPs exemplify inter-speaker repetition of
NPs. That is, while the copied leads and copied postposed NPs concern self-
produced repetition, the repeated NPs concern other-produced repetition. We
can conclude from these findings that one of the recurrent patterns of dis-
course production in the form of IUs in conversational Japanese is to “prefix”
or “suffix” a single-IU clause that expresses a complete proposition with clause-
external detached NP IUs which are redundant propositionally but important
discourse-functionally.
In conclusion, the “independent phrasal” NP IUs or “detached” NPs,
though placed outside the clausal domain and thus not directly concerned
with the communication of propositions, cannot simply be deemed fragments
of discourse or performance errors. It has been argued in this section that
Japanese conversational co-participants produce five types of detached NPs,
each of which is an important phenomenon from a discourse-functional
point of view, and that their occurrence in informal conversation is motivated
by various discourse-pragmatic, interactional, and information-flow factors.
Concerning the cross-linguistic applicability of the discourse phenomena dealt
with in the present section, Croft (1995) and Tao (1996)12 observe similar
independent lone NPs in a corpus of English oral narratives and Mandarin
Chinese conversation, respectively. It may be that the detached NP as an extra-
clausal independent IU occurs cross-linguistically in spoken discourse (see also
Durie 1994; Helasvuo 2001).
Stray NPs occur as a result of conversational acts by which co-participants
exhibit strong interactional involvement such as collaborative completions
and interruptions. Lead NPs, which usually carry new information, serve the
important function of regulating information flow in discourse specifically
by “converting” the information status of NPs to be copied or restated in
the following clause. This can be seen as one of the discourse strategies
speakers employ in order for the conversation to conform to the “one new
NP per IU constraint”. Further, clause-external postposed NPs which appear
as separate IUs apparently perform more interactionally oriented discourse-
pragmatic functions than their clause-internal counterparts—for example,
their functions such as emphasis, repair, and further specification can be seen
to be more directly related to the speaker’s concern about the hearer. Finally,
use of detached NPs as repetitions is another common type of interactional
strategies speakers use in talk-in-interaction.
Syntactic structure of the intonation unit in conversational Japanese
We have seen in Section 4.2 that the adjectival IUs (i.e. [AP]) accounted for
8% of the 1,600 substantive IUs. The adverbial IUs (i.e. [AvP]) and mixed
phrasal IUs (i.e. [XP]) accounted for 11% and 2%, respectively (see Table 4.1).
Table 4.10 shows the distribution of the adjectival, adverbial, and mixed
phrasal IUs.
We can see the following from Table 4.10 concerning the distribution of
the five types of AP IUs: (a) 67% are clausal IUs, whereas 33% are phrasal
IUs; (b) 76% are predicate APs, whereas 24% are attributive APs; and (c) 73%
are independent clausal or phrasal IUs, whereas 27% are elements of multi-
IU clauses. In short, the results indicate the speakers’ preferential production
of AP IUs as independent and clausal [PAP] IUs. The adverbial IUs and
mixed phrasal IUs occurred either as constituents of multi-IU clauses ([CAvP],
[CXP]) or as independent IUs ([AvP], [XP]). Regarding the distribution of
these IU types, Table 4.10 indicates the following. The AvP IUs were used as
elements of multi-IU clauses and as independent adverbials in comparable
proportions. The XP IUs, on the other hand, were used preferentially as
elements of multi-IU clauses. Further coding has shown that both [CAPa] and
[CAvP] IUs tend to occur within multi-IU semi-clauses as opposed to multi-IU
full clauses. This indicates the speakers’ propensities to use attributive AP IUs as
modifiers of non-subject NPs and adverbial IUs as circumstantial modifiers of
subjectless predicates. In contrast, the mixed phrasal [CXP] IUs tend to occur
within multi-IU full clauses, as in the case of the [CNPa] argument NP IUs
(cf. Section 4.5.1).
Table 4.11 presents the distribution of six types of clause-external indepen-
dent adverbial IUs.
Table 4.11 shows that of the 82 independent [AvP] IUs, about 50% are
“stray” adverbials, and about 30% are sentential adverbs13 which regulatorily
AvP type N %
Type 1 Stray 40 49
Type 2 Lead 7 9
Type 3 Topic/LDed 1 1
Type 4 Postposed/RDed 5 6
Type 5 Repeated 6 7
Sentential adverb 23 28
Total 82 100
function to link the preceding and following IUs (e.g. kekkyoku ‘after all’,
tatoeba ‘for example’, tsumari ‘in other words’). This indicates that unlike
the detached [NP] IUs, which showed a fairly balanced distribution among
the five types (see Table 4.8), the [AvP] IUs exhibit a skewing toward the
stray type. Of the 40 stray [AvP] IUs, 40% belong to Type 1 which involves
self-repair or other-interruption (Type 2=2%, Type 3=25%, Type 4=33%) (cf.
Section 4.5.2).
Examples of independent phrasal [AP], [AvP], and [XP] IUs are given
below as illustrations.
The excerpt (4.24) contains two [AvP] IUs and two [AP] IUs. Here the
interactants S and I are conversing about potential factors that will affect job
placement in Japanese companies.
(4.24)→a. S: gyakuni ne,@ [AvP]
conversely fp
‘conversely’
b. hu::n.
pf
mhm
→c. yappari, [AvP]
after all
‘after all’
d. are ga aru yo. [FC]
that nom exist fp
‘That exists.’
→e. +daigaku no, [AP]
university gen
‘university’s’
Syntactic structure of the intonation unit in conversational Japanese
The IUs (4.24a) and (4.24c) are [AvP] IUs; (4.24e, f) are [AP] IUs. In this
sequence, speaker S first utters ‘conversely’, a sentential adverb, to link the
preceding IUs with the following IUs. Then, inserting the cognitive type
regulatory IU ‘mhm’ (which suggests S’s mental planning for the coming IUs),
S utters ‘after all’, an additional sentential abverb. The first AP IU (4.24e)
‘university’s’ is a RDed and stray (Type 1) adjective, which was intended to
be a modifier of some head NP (which was not supplied) and thus to be
part of a larger NP restating the resumptive subject pronoun ‘that’ in the
full clause (4.24d). The second AP IU (4.24f) ‘university’s’ is a stray (Type 2)
adjective. Speaker S, while searching for an appropriate head NP, repeats the
AP, but before coming up with a “target” NP, the other interlocutor supplies
the NP ‘name value’ in (4.24g). The result is collaborative completion. Speaker
S then negates the supplied term ‘name value’ and finally supplies the target
NP ‘cliquism’ in (4.24i). The repeated-as-listener-response NP IU (4.24j) batsu
functions as a token of acknowledgment.
Example (4.25), where speaker Y and H are talking about one of the major
Japanese newspapers, contains three detached [AvP] IUs.
(4.25)→a. Y: yappa sono:, [AvP]
after all pf
‘after all uh’
b. kaisha no shuryuu de nai kara ne.= [PNP]
company gen mainstream cop neg because fp
‘Because (it’s) not the company’s mainstream.’
Chapter 4
. 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.
1000
900
800
700
Full clause
600
Frequency
500 Semi-clause
400 Total
300
200
100
0
One
Two
Three
Four
Five+
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
. Summary
c. full clausal IUs consisting of overt subjects plus predicates ([FC] = [FC] +
[CFC]) (19%).
Further, of the 20 IU syntactic types, the three most pervasive types have
been found to be the following:
a. single-IU semi-clauses with verbal predicates only ([PVP]) (25% of the
1,600 IUs).
b. single-IU full clauses composed of overt subjects plus predicates ([FC])
(16%).
c. single-IU semi-clauses with nominal predicates only ([PNP]) (11%).
In sum, the results indicate that clausal IUs, and especially independent
clausal IUs conveying complete propositions are representative of the substan-
tive IUs produced by the Japanese conversational co-participants. That is, what
this suggests is that the clause is the syntactic exponent of the Japanese substan-
tive IU, as I argued in Section 4.2.2. This finding supports the “clause centrality
proposal” advocated by Chafe and others, suggesting a cross-linguistic applica-
bility of the proposal. This, on the other hand, casts doubt on Iwasaki and Tao’s
(1993) argument for the phrase-oriented nature of Japanese IUs, although, as I
suggested, the treatment of nominal/adjectival predicate clausal IUs is crucially
involved in the clause vs. phrase dichotomy issue in question. Apparently, the
clause centrality that this chapter has found robust in Japanese substantive IUs
merits further investigation.
Example (4.31) consists of [PVP] and [FC] IUs, the two most preferred
syntactic types of the Japanese substantive IU.
(4.31) a. Y: ->mae atashi ga ano yuupen de
before I nom pf UPenn loc
-->benkyooshiteta toki ni ne? [FC]
study-past-prog when loc fp
‘When I was studying at uh UPenn before’
b. S: itsu yuupen nanka itteta no? [PVP]
when UPenn sof go-past-sta fp
‘When were (you) at UPenn?’
c. Y: u:n kocchi ni, [AvP]
pf here all
‘uhm to this place’
d. ++yuusiieruei ga hajimaru mae ni, [FC]
ucla nom begin before loc
‘Before UCLA (classes) began,’
Chapter 4
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
This chapter, testing Hypothesis 2.1 and expanding the original research
question RQ2, addresses the following four related questions (Matsumoto
1997a, 1998b, 2000a, 2000b, 2002, 2003):
a. What are the preferred NP types in conversational Japanese in terms of
grammatical role, information status, and syntactic form? What relation-
ships exist among these properties of NPs?
b. What is the preferred information structure of the intonation unit (IU) in
conversational Japanese? What is the preferred arrangement of given and
new information within an IU?
c. What are the preferred clause types in conversational Japanese? What is the
preferred clause structure in terms of the number and type of arguments
contained per clause?
d. Is the information structure of the IU related to the speaker’s produc-
tion of multi-IU clauses, or the breakup of a clause into phrasal IUs in
conversational Japanese? If so, in what way?
In coding the IUs for the preferred NP types, I used three categories of infor-
mation statuses, six categories of NP grammatical roles, and eight categories
of NP syntactic forms. In coding the IUs for the preferred information struc-
ture, I used two categories of information statuses. In coding the clauses for
the preferred clause types, I used five categories. Further, in coding the clauses
for the preferred clause structure, I used three categories of grammatical roles.
The coding categories and their operational definitions follow.
Chapter 5
In analyzing the 1,600 substantive IUs for the three information statuses,
given, accessible, and new, I coded only nominal references. As Du Bois
(1987: 816) emphasizes, “practically, information status for nominals is more
amenable to reliable operational definition and quantification”. That is, it is
much easier to determine whether a nominal reference to an entity is new/non-
new, or whether there are one or two new/non-new entities within a single
IU, than it is to determine whether a sequence of verb + adverb, for example,
constitutes two pieces or one complex piece of new/non-new information. I
myself experienced difficulties previously in coding the information statuses of
verbal and adverbial elements in relation to postposing in Japanese discourse
(Matsumoto 1995a). Another justification for not coding verbal mentions
has to do with the fact that events and states are highly transient in active
consciousness, and thus verbs do not usually refer back repeatedly to a single
event or state during the successive production of IUs. This means that new
status is clearly the “default” for verbal references (Chafe 1994; Du Bois
1987; cf. Sections 2.4.5–2.4.6). Moreover, given that verbs normally carry new
information, a more precise characterization of conceptual “unitariness”—
what, for example, verb + adverb combination, should be counted as a unitary
concept—must first be established in relation to the proposed constraints on
new information quantity per IU, if we are to code verbal tokens as well
(see discussion in Section 2.6.2). While we lack such precise definitions of
conceptual unitariness involving verbal elements, it seems safe to confine the
given, accessible, and new statuses, together with identifiabilty, to the discourse
properties of nominal references.
The definitions of the three categories of information statuses are provided
below (cf. Du Bois 1987: 816):
a. given: a referent1 which was mentioned within 25 IUs2 previously in the
discourse (i.e. the transcribed 45-minute conversation), or a referent which
is given from the conversational context itself (e.g. the conversational co-
participants) (Chafe 1976).
b. accessible: a referent which was mentioned more than 25 IUs previously
in the discourse, or a referent which was previously unmentioned in the
discourse but is part of a previously evoked schema (Chafe 1987; Du Bois
1980), or a referent which the speaker assumes to be identifiable to the
hearer (i.e. to be previously semi-active in the hearer’s consciousness) by
situation or prior knowledge already shared by the participants.
Information structure of the intonation unit in conversational Japanese
c. new: a referent which is neither (a) nor (b), that is, a referent which
was introduced into the discourse as a previously unmentioned, totally
new concept.
Given that previous research has found that given and accessible concepts
virtually exhibit the same patterning (Du Bois 1987), and accessible mentions
appeared rarely in the data, as will be shown below, I will also make a
binary distinction between “Given” and “New” (with capitals G and N), where
“given” and “accessible” concepts are subsumed under the category “Given”
(the categories “New” and “new” refer to identical concepts).
NP syntactic forms
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
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
Of the 10 NPs contained in (5.1), two NPs, those in (5.1a, b), are clause-
external independent NPs. The remaining eight NPs are clause-internal NPs.
Of these, five are contained in the single-IU clauses (5.1e, f) and (5.1h, i),
and three are contained in the multi-IU clauses comprised of (5.1c, d) and
(5.1k, l). The eight clause-internal NPs consist of four arguments and four non-
arguments. The four clause-internal argument NPs consist of the following:
two S roles, which occur in (5.1c) and (5.1h), one A role, and one O role, both
of which occur in the full clause (5.1i). One of the S-role NPs is a complex NP;
the other consists of adjective + NP. The A-role NP is a personal pronoun,
and the O-role NP is composed of NP-no + NP. The four non-argument
NPs, on the other hand, consist of two oblique NPs, those in (5.1d) and
(5.1k), and two predicate nominals, those in (5.1e, f). Four of the 10 NPs are
bare nouns.
Information structure of the intonation unit in conversational Japanese
Does this example illustrate the typical way Japanese speakers use overt
NPs in informal conversation? What types of NPs do they use preferentially?
This issue will be addressed in the next section.
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
400
350
300
250
Frequency
200
150
100
50
0
Oblique
Predicate
nominal
Independent
NP
S
Grammatical role
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
Total
Independent
NP
Predicate
nominal
new
Oblique accessible
given
O
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).
800
700
600
500
Frequency
400
300
200
100
0
dem+N
pro-p
pro-d
bareN
N-no+N
adj+N
compNP
nomVP
NP syntactic form
Table 5.5 indicates that of the 1,417 NPs, (a) 90% (N=1,270) are lexical,
whereas 10% (N=147) are pronominal;7 and (b) 50% (N=710) are bare nouns
without any modifiers, with the other syntactic forms being in low percentages
(see also Figure 5.3). Table 5.5 also shows the following concerning the expres-
sion of the grammatical roles: (a) A roles tend to take the form of bare nouns
(38%) or personal pronouns (28%); and (b) S roles (43%), O roles (53%),
obliques (59%), predicate nominals (49%), and independent NPs (52%) tend
to take the form of bare nouns. Overall, the speakers exhibited a strong propen-
sity to express the grammatical roles with bare nouns, which accounted for
about 40–60% across the role types. Further, the A roles, in comparison to
the other roles, are less likely to be expressed with lexical NPs; while the
overwhelming majority of the S roles (83%), O roles (91%), obliques (93%),
predicate nominals (97%), and independnet NPs (92%) are lexical, 65% of
the A roles are lexical.
Further, Table 5.5 indicates the following regarding the grammatical-role
positions in which the eight syntactic forms typically occur: (a) personal pro-
Information structure of the intonation unit in conversational Japanese
p-pro 44 7 0 0 0 0 44 3
d-pro 102 17 1 1 0 0 103 8
bareN 260 42 93 80 357 52 710 50
dem+N 110 18 5 4 17 2 132 9
N-no+N 29 5 5 4 97 14 131 9
adj+N 37 6 7 6 128 19 172 12
compNP 20 3 5 4 77 11 102 7
nomVP 10 2 1 1 12 2 23 2
Total 612 100 117 100 688 100 1,417 100
Chapter 5
In sum, the major finding displayed in Tables 5.4–5.6 is that bare nouns,
thus lexical NPs were used preferentially across the grammatical roles and
information statuses. Pertinent to this finding is the infrequent use of pronouns
in the data (only 10% of all overt NPs), and this is most likely to be linked to
the fact that in spoken Japanese many Given NPs are not expressed, or take
zero forms. Given that 65% of the A roles are lexical, it appears that the “Non-
lexical A Constraint” (Du Bois 1987) does not hold in conversational Japanese.
On the other hand, the finding that the A roles contained a relatively higher
percentage of New information, i.e. 20% suggests the weak applicability of the
Given A Constraint in conversational Japanese (cf. Iwasaki 1985). However,
we need to be careful in interpreting these findings in relation to the Non-
lexical A and Given A Constraints. That is, these constraints do not seem to
hold or hold less strongly in Japanese, to the extent that we look at overt NPs
exclusively. As stated above, many Given NPs (including A-role NPs), which
would normally take pronominal forms in other languages such as English,
take zero forms in Japanese. Therefore, it may be that the constraints actually
hold (even though not as strongly as in Du Bois 1987), if we code zero-form
A-role NPs as pronominal (as the most attenuated, special type of pronouns)
and as Given. A roles may be more likely to be non-lexical and Given, if A-
role nominals taking zero forms are equally coded. I will address this issue in
Section 5.5.
Table 5.7 summarizes the results of analysis of the 1,417 overt NPs in terms
of the three features of grammatical role, information status, and syntactic
form (where OBL=oblique, and PN=predicate nominal).
Table 5.7 shows the following. First, overwhelmingly, the speakers pro-
duced the NPs as clause-internal arguments (45%) or non-arguments (43%),
and as bare nouns (50%). Second, the speakers produced a roughly compara-
ble proportion of Given (51%) and New NPs (49%), with Given NPs slightly
outnumbering New NPs. Third, the speakers placed Given information most
frequently in S roles (30%), and they expressed it typically with bare nouns
(48%). By contrast, the speakers placed New information mostly in predicate
nominals (26%) or oblique NPs (26%) (i.e. clause-internal non-arguments),
and they expressed it typically with bare nouns (52%). Thus, the most pre-
ferred type of overt NPs are bare nouns with S role and Given information
status, and bare nouns with non-argument roles and New information status.
As discussed above, partly because maximally attenuated, zero-form Given
nominals were uncoded, the analysis revealed that the percentage of overtly
expressed Given NPs is nearly equal to that of New NPs, exhibiting, as a
result, the higher proportion of New NPs in all grammatical role types. This
Information structure of the intonation unit in conversational Japanese
Table 5.8 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
only 7% (N=117) contain both Given and New NPs. The three most preferred
IU information structure types are thus the following: (a) IUs with one New NP
(34%), (b) IUs with one Given NP (30%), and (c) IUs with zero NPs (24%).
It also shows that (a) of the IUs containing two NPs, the type G + N is most
prevalent (51%); and (b) of the IUs containing three NPs, the type G + G + N
is most prevalent (41%).
The results given above suggest a number of constraints on the quantity of
explicit NPs containable within one IU in conversational Japanese. The finding
that none of the IUs in the database contained more than three NPs suggests
a constraint on the maximum number of NPs that may be overtly expressed
within an IU. That is, one IU may contain no more than three overt NPs. The
results show that while the overall production of NPs strictly conforms to this
constraint, which confines the upper limit of NP quantity per IU to three, the
maximum number of Given NPs per IU is three, whereas the maximal quantity
of New NPs per IU is two. This suggests that one IU may contain no more than
three Given NPs and no more than two New NPs.
Another important finding is that among the IUs with zero, one, two, and
three NPs that are permitted by this constraint, IUs with only one NP are most
prevalent, whereas IUs with zero NPs are less common, and IUs with two or
three NPs are quite rare. This indicates that the speakers prefer to use NPs
singly rather than multiply each time they produce an IU, such that each IU
contains one piece of Given or New nominal information. More specifically,
the speakers used the strategy of placing “one NP per IU” 65% of the time,
and the strategy of placing “zero or one NP per IU” 89% of the time. This
Chapter 5
600
500
400
Frequency
300
200
100
G+G+G
G+N+G
N+G+G
Zero NPs
G+N
G+G
N+G
N+N
G+G+N
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
IU type N %
Zero anaphora
Ø-S 164 43
Ø-A 62 16
Ø-A+O 34 9
Ø-O 4 1
Total 264 69
No associated NPs
Adverbial phrase 75 20
Verb phrase 25 6
Adjectival phrase 17 4
Mixed phrase 4 1
Total 121 31
TOTAL 385 100
slots or because core arguments are already present in adjacent IUs. (In coding
the IUs for existence/non-existence of core argument NPs, when both main
and subordinate/embedded clauses are present in an IU, only the main clause
was coded.)
Table 5.9 indicates that the cases of zero anaphora accounted for about
70% of the IUs with zero NPs. It also shows that 43% of such IUs have their
S-role slots unfilled with overt NPs. Of the remaining cases of no associated
NPs, the majority are adverbial phrasal IUs which originally lack NP slots.
This is followed by verb phrasal IUs, in which no NPs need to be present
because associated core arguments are contained in the immediately following
or preceding IUs. In sum, the majority of the IUs with no explicit NPs are
clauses involving zero anaphora, typically intransitive or nominal/adjectival
predicate clausal IUs whose S-role slots are unfilled (see further discussion in
Section 5.5).
The 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
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).
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)
The previous section dealt with the preferred ways Japanese speakers arrange
Given and New NPs as they successively produce IUs and clauses in conver-
sational interaction. The coded NPs include all the NPs contained in the IUs,
viz. arguments (S, A, and O), non-arguments (obliques and predicate nomi-
nals), and clause-external independent NPs. Those NPs contained in the intra-
IU subordinate/embedded clauses were also coded. This section examines the
preferred ways Japanese speakers structure different types of clauses in terms of
the number and type of arguments contained per clause. Thus, only argument
NPs are coded in this section.
Table 5.11 presents the distribution of the 1,121 clauses (which do not
include intra-IU clauses) among the five clause types with different degrees
of transitivity and with different numbers of overtly expressed arguments
(which include clausal arguments, mostly clausal objects). In Table 5.11, Ø
indicates zero arguments, i.e. arguments that are subcategorized by the verb
(A and O for transitive verbs, S for intransitive verbs and adjectival/nominal
predicates) but are not overtly realized. As in Table 5.9, zero arguments
include both “anaphoric zeros” (i.e. referential zeros which can be linked with
specific referents previously mentioned) and “non-anaphoric zeros” (i.e. non-
referential zeros not derived from prior mentions and thus independent of
anaphoric continuity processes; inherently null arguments which always take
zero forms) (cf. Tao 1996; Section 5.4.1).
Table 5.11 indicates that (a) of the 910 single-IU clauses, intransitive (29%)
and nominal predicate clauses (25%) are most common; and (b) of the 211
Information structure of the intonation unit in conversational Japanese
multi-IU clauses, low transitive (31%) and intransitive clauses (28%) are most
prevalent. Of the 1,121 clauses, intransitive clauses are most pervasive (29%),
followed by nominal predicate (23%), low transitive (21%), and adjectival
predicate clauses (16%), whereas high transitive clauses are least common
(11%) (see also Figure 5.5). Table 5.11 also shows that transitive clauses (32%)
are about as frequent as intransitive clauses (29%). In sum, the three most
preferred clause types are, with descending frequency, intransitive, nominal
predicate, and low transitive clauses, with high transitives being the most
dispreferred clause type.
Of the total number of clauses, one-participant clauses, or clauses with
predicates which take only one argument (S) accounted for 68%, whereas
two-participant clauses, or clauses with predicates which take two arguments
(A and O) accounted for 32%. This means that the majority of the clauses
are non-transitive, or one-participant clauses which are low in transitivity
(cf. Hopper & Thompson 1980; Section 2.2.2). Of all the one-participant
clauses which constitute the majority, 43% are intransitive, 34% are nomi-
nal predicate, and 23% are adjectival predicate clauses. This indicates that the
perferred one-participant clause types are intransitive or nominal predicate
350
300
250
Frequency
200
150
100
50
0
Intransitive
high
low
Adjectival
Nominal
Transitive
Transitive
predicate
predicate
Clause type
clauses. Of all the two-participant clauses which are the minority, low transi-
tives accounted for 70%. This means that even the two-participant clauses tend
to be low in transitivity, not involving action verbs which exert an effect on the
direct object NP. The findings suggest that conversational Japanese overwhelm-
ingly consists of clauses at the lower extreme of the transitivity continuum—
clauses unrelated to events or actions which typically express speakers’ feel-
ings, attitudes, and opinions, or subjective stance. In short, the results of this
study are in support of the claim for the low transitivity-centered and non-
event-oriented nature of conversational discourse (cf. Tao 1996; Thompson &
Hopper 2001).
The analysis has revealed that overwhelmingly, conversational Japanese
consists of non-transitive, one-participant clauses which are very low in tran-
sitivity. That is, conversational Japanese, and possibly conversational discourse
in general, is preferentially composed of non-event- or non-action-related ma-
terials such as descriptions of circumstances, expressions of mental states and
subjective opinions and evaluations. Thus, viewed in terms of transitivity, con-
versation is reminiscent of what Hopper and Thompson (1980) identified as
Information structure of the intonation unit in conversational Japanese
Transitive 26 –– 8 80 10 124
high (21) (6) (65) (8) (100%)
Transitive 22 –– 8 181 22 233
low (9) (4) (78) (9) (100%)
Intransitive 172 154 –– –– –– 326
(53) (47) (100%)
Adjectival 87 90 –– –– –– 177
predicate (49) (51) (100%)
Nominal 185 76 –– –– –– 261
predicate (71) (29) (100%)
Total 492 320 16 261 32 1,121
(44) (29) (1) (23) (3) (100%)
We can see from Table 5.11 that (a) about 50% of the single-IU clauses
contained zero arguments, and about 50%, one overt argument, (b) the ma-
jority (about 80%) of the multi-IU clauses contained one overt argument, and
(c) the multi-IU clauses contained a higher proportion of two overt arguments
than the single-IU clauses (7% vs. 2%). As shown in Table 5.13, of the 1,121
clauses, 53% contained one overt argument, 44% contained zero arguments,
and only 3% contained two overt arguments (see also Figure 5.7). Most im-
portantly, clauses with two overt arguments were produced remarkably infre-
quently by the speakers. That is, 91% of the transitives and 97% of the clauses
contained zero or one overt argument. This suggests Japanese speakers’ no-
table preference for use of zero-argument or one-argument clauses over use of
two-argument clauses. The preferred clause types that the present study found
are thus the following: clauses with zero arguments or one overt argument,
typically, non-transitive clauses with zero arguments or overt S, and transitive
clauses with overt O only.
The finding that clauses with two overt arguments are extremely rare
has led me to propose the “one overt argument per clause constraint”. What
this constraint says is that speakers avoid more than one overt argument per
clause, such that one clause contains zero arguments or one overt argument
(typically S or O). The preferred surface syntactic structure of the clause in
conversational Japanese can hence be represented as follows (N=argument NP,
and P=predicate):
Information structure of the intonation unit in conversational Japanese
Total
Nominal
predicate A+O
Adjectival O
predicate
A
Intransitive S
Zero
Transitive
low
Transitive
high
Figure 5.6. Proportion of clauses with zero, one, and two arguments
(5.5) N(S/O) P
The configuration (5.5) indicates that in the unmarked word order situation
which does not involve post-predicate NPs, the maximal surface structure
which is consistently preferred in Japanese discourse is a single overt argument
in the S or O role followed by a predicate, either verbal, adjectival, or nom-
inal. (Note that the argument NP may not be overt; the surface clause form,
Table 5.13. Frequency of clauses with zero, one, and two overt arguments
Clause type N %
Zero arguments 492 44
S 320 29
A 16 1
O 261 23
One argument-Total 597 53
Two arguments (A+O) 32 3
TOTAL 1,121 100
Chapter 5
500
450
400
350
300
Frequency
250
200
150
100
50
0
Ø S A O A+O
Argument type
P with no overt argument NP is also one variant of the preferred pattern.) This
is perfectly compatible with the preferred clause structure in Sakapulteko pro-
posed in Du Bois (1987: 823), except that the order of argument and predicate
(verb) is reversed (Japanese is a verb-final language, whereas Sakapulteko is a
verb-initial language).9
The argument structure in (5.5) also indicates that although linguists
generally define Japanese as an SOV (Subject-Object-Verb) language, and have
treated sentences with two overt lexical arguments such as John ga Mary o
butta. ‘John hit Mary.’ (Kuno 1973: 3) and Taroo ga hon o katta. ‘Taro bought
a book.’ (Shibatani 1990: 258) as if they were representative clausal forms,
this treatment can never be justified, given that in naturally occurring spoken
discourse such highly transitive two-argument structures are rarely produced
(see also Iwasaki 2002: Chapter 6). In summary, the vast majority of clauses in
Japanese conversation are non-transitive, one-participant clauses with zero or
overt S arguments, and the minority transitive, two-participant clauses, which
themselves tend to be low in transitivity, overwhelmingly contain overt O
Information structure of the intonation unit in conversational Japanese
arguments only; highly transitive clauses with two overtly expressed arguments
are atypical in conversational discourse.
The clause types most preferentially produced in conversational Japanese
are illustrated in (5.6)–(5.8). Example (5.6) is an intransitive clause with overt S
‘tigers’; (5.7) is a nominal predicate clause without overt S, which solely consists
of the complex predicate nominal ‘a person (you) don’t know’; and (5.8) is a
low transitive clause with overt O ‘grade’.
(5.6) M: tora toka mo iru no? [FC]
tiger sof also exist q
‘Are there tigers (there), too?’ (F5)
(5.7) Y: shiranai hito? [PNP]
know-neg person
‘(Is she) a person (you) don’t know?’ (F2)
(5.8) K: MAda ((laugh)) gureido shiranai no. [PVP]
yet grade know-neg q
‘(You) don’t know (your) grade yet.’ (M5)
The one overt argument per clause constraint formulated above is reminis-
cent of Du Bois’ (1987) “One Lexical Argument Constraint”, which prohibits
more than one lexical argument per clause. When we compare the two con-
straints in terms of their applicability in conversational Japanese, we would
need to note that arguments in the one overt argument per clause constraint
include both lexical and pronominal NPs (pronouns accounted for only about
10% of all overt NPs), and the lexical arguments are a subset of the overt argu-
ments. Thus the fact that the one overt argument per clause constraint holds
in the Japanese data suggests that the One Lexical Argument Constraint will
hold as well. In sum, there exists a notable tendency in conversational Japanese
to limit the quantity of overtly expressed arguments within a clause to a
maximum of one. That is, speakers either overtly express only one argument
(preferentially S or O) or do not express arguments at all within one clause,
with the simultaneous expression of A and O arguments per clause being
strongly dispreferred. (The skewed selection of O over A as a single argument
that is allowed to be overtly expressed by the one overt argument per clause
constraint is clearly pertinent to the finding discussed in Section 5.3 that A is
overwhelmingly given and O tends to carry more newly introduced referents.)
Chapter 5
Table 5.14 presents the proportion of three argument slots that the speakers
filled with overt NPs (see also Figure 5.8).
Table 5.14 indicates, first of all, that of the 1,478 argument slots contained
in the 1,121 clauses, 45% (N=661) were filled with either overt S, A, or O,
whereas 55% (N=817) received no overt coding, remaining zero. It also indi-
cates the following regarding the proportion of the three argument slots filled
vs. unfilled with overt NPs: (a) of the 764 S-role slots, 42% (N=320) were filled
with overt NPs (58% were left unfilled); (b) of the 357 A-role slots, only 13%
(N=48) were filled (87% were left unfilled); and (c) of the 357 O-role slots,
82% (N=293) were filled with overt NPs (18% were left unfilled). We can
also see from Table 5.13 that (a) the percentages of unfilled A-role slots are
comparable between the high and low transitives; (b) the percentage of filled
O-role slots is higher in the low transitives; and (c) as we have already seen
above, while nearly 50% of the S-role slots of the intransitive and adjectival
predicate clauses were filled, those of the nominal predicate clauses tend to
remain zero.
The results indicate an interesting pattern in the overt expressibility of the
three roles. That is, O-role slots tend to receive overt coding, whereas S-role
and A-role slots tend to remain zero forms. In particular, the finding that the
speakers did not fill about 90% of the available A-role slots with overt NPs
should be noted. What this suggests is that the transitive subject slot (A),
by comparison to the transitive object (O) and intransitive subject (S) slots,
is strongly dispreferred as a site for overt NPs (cf. Clancy 2003). Assuming
S A O Total
Clause type N % N % N % N %
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
that new information must always be overt, this means that A-role slots are
disfavored for the expression of new nominal referents. In other words, A-
role slots constitute preferentially the sites for presupposed, given information.
What pertains to this observation is the speculation I have made above in
Section 5.3—that if we code zero-form A role NPs as Given and as the most
attenuated, special type of pronouns, both the Given A and Non-lexical A
Constraints (Du Bois 1987) will hold in conversational Japanese. Given that
the overwhelming majority of the A-role NPs have been found to take zero
forms, thus Given and pronominal, this leads us to the conclusion that the
Given A Constraint, which was found to hold only weakly, and the Non-lexical
Chapter 5
I now take up the question that I raised in Section 4.7: what factors are
responsible for the production of multi-IU clauses? Discussion of this issue
seems relevant at this point because this can be related to the “one new NP per
IU constraint” formulated above. We should remember that this constraint is
on the amount of new information to be contained within an IU, not within a
clause. What this suggests is that a clause composed of multiple IUs may con-
tain freely more than one new NP. Given the finding that the majority of the
multi-IU clauses consist of two or three IUs (see Section 4.7), it is conceivable
that one multi-IU clause may contain maximally two or three new NPs. My
analysis has shown that this is certainly the case with the multi-IU clauses
contained in the Japanese data. As displayed in Table 5.15, of the 211 multi-
Information structure of the intonation unit in conversational Japanese
IU clauses, 32% (N=68) contained more than one newly introduced NP. By
contrast, only 1.3% (N=12) of the 910 single-IU clauses contained more than
one new NP (see also Figure 5.9).
100%
90%
80%
70%
60%
Frequency
30%
20%
10%
0%
IU
clause
clause
Single-IU
Multi-IU
Clause
Unit type
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
We can see from Table 5.15 that of the 1,121 clauses, 6.5% (N=80)
contained more than one new NP. This means that about 94% of the clauses
contained no more than one new NP. I take this to mean that the “one new
NP per IU” constraint can also be interpreted as the “one new NP per clause”
constraint. That is, the constraint is applicable to the clause as well, although
less strongly (about 99% of the IUs contained no more than one new NP).
Bringing the above-presented results together, we can state, in conclusion,
that the “one new NP constraint” applies most strongly to the IU (only 0.7%
of the 1,600 IUs contained more than one new NP), less strongly to the
single-IU clause (only 1.3% of the single-IU clauses contained more than
one new NP), and only weakly to the multi-IU clause (32% of the multi-IU
clauses contained more than one new NP). However, given the relatively small
percentage of the multi-IU clauses (about 19% of all clauses) and the clause-
centered characteristics of the IU evidenced in Chapter 4, the constraint could,
on the whole, be taken to apply not only to the IU but also to the clause. That
is, the “one new NP constraint” limits the quantity of new nominal referents
that can be introduced within a discourse unit in conversational Japanese,
such that one IU/clause may contain no more than one new NP.
An additional relevant finding that should be presented here concerns
the grammatical role types of multiple new NPs contained in the 68 multi-
IU clauses. These new NPs were found to be argument + non-argument (e.g.
S role + oblique) or non-argument + non-argument (e.g. oblique + predicate
nominal), but not argument + argument, combinations. That is, none of
the multi-IU clauses contained two new argument NPs. This indicates that
Du Bois’ (1987) One New Argument Constraint discussed in Section 5.4.2 can
apply not only to the IU but also to the clause in conversational Japanese.
That is, speakers avoid introducing more than one new argument NP within
one IU and within one clause as well. The speaker’s avoidance of more
than one new nominal referent per IU/clause, it appears, is related to the
cognitive cost involved in the process of introducing a new concept. That
is, activating a previously inactive concept, or converting an idea from the
inactive to the active state is supposedly most costly (cf. Section 2.3.2). It
is conceivable that the discourse production activity of introducing a new
referent engages the speaker’s whole verbalization capacities, and therefore
the simultaneous introduction of two new referents within one IU/clause is
excessively burdensome.
Information structure of the intonation unit in conversational Japanese
. Summary
The results of the present IU information stucture analysis have provided the
following answers to the research questions that I posed at the beginning of
this chapter:
a. preferred NP types: (a) grammatical role: S-role NPs (25%), oblique NPs
(22%), and predicate nominals (21%); (b) information status: Given NPs
(51%) and New NPs (49%); (c) syntactic form: bare nouns (50%); and
(d) most typical NP types: bare nouns with S role and Given information
status, and bare nouns with non-argument roles (obliques or predicate
nominals) and New information status.
b. preferred IU information structure types: IUs containing only one New NP
(34%) and IUs containing only one overt Given NP (30%).
c. preferred clause types: clauses with one overt argument (53%), typically
non-transitive clauses with overt S (29%) and transitive clauses with overt
O only (23%); and clauses with no overt arguments (44%), typically non-
transitive clauses without overt S (40%).
This chapter, responding to RQ3 and testing Hypothesis 3.1, explores the
functional structure of the intonation unit (IU) in conversational Japanese.
The following four questions are addressed (Matsumoto 1999a):
a. What is the preferred functional structure of the IU in conversational
Japanese?
b. What is the preferred number of functional components per IU in conver-
sational Japanese?
c. What linear order do the functional components follow within an IU in
conversational Japanese?
d. Is the functional structure of the IU related to the speaker’s production of
multi-IU clauses, or the division of a clause into phrasal IUs in conversa-
tional Japanese? If so, in what way?
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
of events, states, or referents (see Section 3.5); this means that all of the IUs
analyzed in this study contain [ID].
The textual [TX] component also participates in the content, but does
textual referential work by connecting one IU to another. That is, the textual
component has to do with the creation of coherence in discourse while
regulating the linkage between successively produced IUs. Those elements
coded as [TX] in this study include such textual connectives as de/sorede ‘and’,
demo ‘but’, kedo ‘though’,1 and kara ‘as’.
The cognitive [CG] component has to do with the regulation of the flow
of conversation. It indicates that the speaker is involved in some cognitive pro-
cessing, or is trying to keep the floor of the conversation, with the implication
to the hearer that more IUs are coming. (I use the term “cognitive” following
Chafe’s 1994 “cognitive” subtype of regulatory IUs; see Section 2.4.4.) Typi-
cally, in Japanese, cognitive connectives, or pause fillers such as ano, nanka,
and maa serve this cognitive function. Both the textual [TX] and cognitive
[CG] components are concerned with the linkage of ideas expressed in the
ideational [ID] components, functioning as IU-boundary signals. However,
the main difference between the two is that [TX] participates in the content,
whereas [CG] does not.
The interactional [IT] component has to do with interaction between the
conversational co-participants, or the speaker-hearer interpersonal relation-
ship involved in the co-production of conversational discourse. This compo-
nent functions to show the speaker’s sensitivity towards the addressee, or con-
cern for the hearer’s understanding. It functions to solicit the hearer’s involve-
ment into the on-going speech event. The interactional function is served by
such linguistic elements as interactional particles (e.g. ne, sa, yo) and tag-like
expressions (e.g. janai, desho).
In sum, four major changes have been made of Iwasaki’s (1993) coding
system: (a) elimination of [CO] (cohesive), (b) elimination of [LD] (lead),
(c) creation of [TX] partly as a replacement for [CO], and (d) creation of [CG]
as a replacement for [LD]. Thus, of the elements coded as [CO] in Iwasaki’s
(1993) study, IU-final overt conjunctives such as kedo ‘though’ were coded as
[TX], while -te forms and nominalizers like no and wake comprised part of
[ID] in this study. IU-initial conjunctives such as sorede ‘and’, which constitute
part of [ID] in Iwasaki (1993), were assigned the code [TX] in this study. Pause
fillers like ano ‘uh’, which constitute the IU-initial [LD] component in Iwasaki
(1993), were coded in the present study as [CG], which presumably occur not
only IU-initially but also elsewhere. One of the major advantages of this study
is that [CO], the component supposedly specific to the Japanese language, has
Chapter 6
been eliminated, so that the coding scheme can be applied to languages other
than Japanese, and therefore commensurable cross-linguistic comparisons are
made possible.
In this study, when two connectives or particles with the same function
occurred adjacently in the IU-initial or IU-final position (although such cases
are rare), they were taken as belonging to one functional component and
assigned only one functional code, namely, [TX], [CG], or [IT] (see (6.4)
below). In addition, this study, as in Tokieda-school linguistics, allows the
idea of multifunctionality of certain linguistic elements—that one linguistic
element may have more than one function (e.g. cognitive and interactional)
(see Section 2.5.2). That is, certain linguistic elements should be analyzed as
simultaneously having more than one function, having a certain degree of
membership in one functional category. For example, it can be considered
that conjunctions like sorede ‘and’ have primarily a textual function, but they
equally serve some degree of ideational function (cf. Iwasaki’s 1993 treatment).
This study assumes that a given linguistic element may have multiple functions
but must have only one primary function, and the assigned code indicates such
primary function. Thus, all conjunctions were coded as [TX], not [ID].
The approach that the present chapter takes is advantageous and important
in that it allows us to direct attention to units that would otherwise be
disregarded if we look at only clauses as units of communicating propositions
and only nominals as units of expressing given/new informational distinction.
The basic assumption underlying this approach is that the overall IU structures
will be elucidated more precisely if we examine IUs not merely from the
perspective of syntactic and informational composition but also from the
perspective of functions of syntactic components that comprise IUs. That
is, it is the belief of the present study that the breakdown of Japanese IUs
into functional components will lead us to a more complete understanding
of the structural characteristics of IUs that constitute Japanese conversational
discourse. Specifically, the goal of this chapter lies in providing an additional
support for the claimed “single” nature of IUs as basic linguistic units of
spontaneous spoken communication, as argued for in Chapters 4 and 5.
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
ID 731 45.7
ID IT 434 27.1
ID TX 126 7.9
CG ID 66 4.1
TX ID 58 3.6
ID CG 9 0.6
ID TX IT 47 2.9
TX ID IT 45 2.8
CG ID IT 40 2.5
TX ID TX 13 0.8
CG ID TX 12 0.7
TX ID CG 4 0.3
ID CG IT 1 0.1
TX ID TX IT 7 0.4
CG ID TX IT 5 0.3
TX ID CG IT 1 0.1
CG ID CG IT 1 0.1
Total 1,600 100
component types, and 4 four-component types. The 17 types are listed in de-
scending order of frequency within each category. The order of the components
in the two-, three-, and four-component types corresponds to the temporal
order in which these components were actually uttered within each IU. For
example, the two-component type ID-TX indicates that the ideational compo-
nent is temporally followed by the textual component. Of the 17 IU functional
types, ID represents the minimal use of the functional components, being the
simplest type. The IUs consisting of two components such as ID-IT and those
with three components such as ID-TX-IT are functionally more complex. The
types such as TX-ID-TX-IT represent the maximal use of the four components,
being the most elaborate of all the functional structure types.
Table 6.1 shows that the 1,600 substantive IUs contained a total of 2,638
functional components, which consist of 1,600 ideational components, 581 in-
teractional components, 318 textual components, and 139 cognitive compo-
nents. This means the following. First, of all the components contained in the
IUs, the ideational components account for 61% (the interactional, textual, and
Chapter 6
800
700
600
500
Frequency
400
300
200
100
0
TX ID CG
ID CG
ID
CG ID TX IT
TX ID CG IT
TX ID
CG ID
CG ID IT
ID CG IT
ID IT
TX ID TX
CG ID TX
ID TX
ID TX IT
TX ID IT
TX ID TX IT
CG ID CG IT
IU functional structure type
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).
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
The IUs (6.2) and (6.3) are examples of the simplest functional structure
type, composed solely of the obligatory ideational component. The clausal IU
(6.2) expresses a full proposition, with the subject ‘I’ and the direct object
‘pictures’ being assumed. The IU (6.3) ‘with (my) friends’ is an adverbial
phrasal IU, which, as part of a multi-IU clause, contributes to building up
a proposition. Examples (6.4) and (6.5) are two-component IUs comprised
of the ideational and interactional components. The IU (6.4) consists of the
subjectless nominal predicate clause ‘(it is) an investment’ which constitutes
the ideational component plus the particles yo and ne which collectively
constitute the interactional component. (The particle ne was most frequently
used in this study to serve the interpersonal function.) The NP IU (6.5), an
argument of a multi-IU clause, is composed of the ideational component (the
NP ‘(my) mother’ with the nominative case-marking particle ga) followed by
the interactional component (the interactional particle sa).
The results also reveal the following about the discoursal patterns of the
four functional components. First, of the 1,600 ideational [ID] components,
46% appeared singly, forming the functional type ID, and 32% were im-
mediately followed by the interactional component, forming the type ID-IT,
Chapter 6
The IU (6.6) is composed of [ID], the subjectless transitive clause ‘(I) have
(one)’ plus [TX], the conjunctive kedo ‘though’ (the textual connective most
frequently used in the present data). The IU expresses a full proposition in
which the subject ‘I’ and the direct object ‘one’ are assumed.
Finally, the cognitive [CG] component appeared before the ideational
component 89% (vs. after the ideational component 11%) of the time. This
finding can be explained by the role of this component in discourse produc-
tion. Given that the cognitive component has to do with the speaker’s cog-
nitive processing, and most typically, cognitive planning of ideational con-
tent to be expressed in about-to-be-uttered IUs, it naturally follows that
this component occurs overwhelmingly before the ideational component.
Of the two-component IU types involving the cognitive component, the
type CG-ID (N=66), as exemplified by (6.7), was preferred over the type
ID-CG (N=9).
Functional structure of the intonation unit in conversational Japanese
(6.7) [CG] [ ID ]
H: ano daigaku jidai yo nen kan de,
pf university days four years during loc
‘uh in (my) four-year university days’ (M4)
The IU (6.7) begins with [CG], the pause filler ano ‘uh’, which is followed
by [ID], the adverbial phrase ‘in (my) four-year university days’. The two
components together constitute a substantive IU element of a proposition-
conveying multi-IU clause.
Table 6.2 lists the frequency of the number of the functional components used
per IU.
Table 6.2 shows that 89% (N=1,424) of the 1,600 substantive IUs consist of
one functional component (i.e. the ideational component) or two functional
components (i.e. the ideational component plus one of the optional compo-
nents). Those IUs composed fully of the four components were rarely pro-
duced, occupying only 1% of the total number of the IUs. This suggests that
Japanese IUs tend to be unifunctional, or at least minimally multifunctional.
In other words, Japanese speakers prefer to incorporate no more than two
functional components—preferentially only one—within an IU. This can be
termed the “no more than two functional components per IU” constraint.
While the one-component functional type ID is predominant in the data,
of the multi-component types, some were more preferentially produced. As
shown in Table 6.1, the most preferred two-component type is ID-IT, the most
preferred three-component type is ID-TX-IT, and the most frequently used
four-component type is TX-ID-TX-IT. The IU (6.8) given below is an example
of the most favored three-component functional type, ID-TX-IT. The IU
consists of the ideational component (the subjectless adjectival predicate clause
‘(it is) difficult’), the textual component (the conjunctive kedo ‘though’), and
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)
The results show that the three optional functional components occurred in
specific linear positions in an IU relative to the obligatory ideational com-
ponent. The textual and cognitive components appeared either immediately
before or immediately after the ideational component. Given that the tex-
tual component concerns linkages as well as specific relationships between
ideas or propositions expressed in the ideational component, and the cogni-
tive component concerns the planning of ideas or propositions expressed in
the ideational component, it follows that both of them need to occur adja-
cent to the ideational component. The interactional component, on the other
hand, occurred only in the IU-final position. This means that the interactional
component does not always appear adjacent to the ideational component, un-
like the textual and cognitive components. This also suggests that interactional
particles such as ne can function as non-prosodic markers of IU boundaries
(cf. Section 2.4.2).
The linear order of the four functional components within an IU can thus
be represented as follows:
(6.9) [TX/CG] [ID] [TX/CG] [IT]
What (6.9) indicates is that when an IU contains all of the four functional
components, the textual/cognitive component is followed by the ideational
component, which is followed by the textual/cognitive component, which is
in turn followed by the interactional component. To the extent that (6.9)
allows the textual and cognitive components to occur in the same linear
positions, this means that no fixed linear order can be observed among the four
functional components, in contrast to Iwasaki’s (1993) treatment (see (2.15)
in Section 2.5.2). That is, the pre-ideational and post-ideational positions are
not unifunctional; they may be occupied by either the textual or cognitive
component. The fact that only these components are allowed to be substituted
for one another could be attributed to the following: A given connective can
Functional structure of the intonation unit in conversational Japanese
have either the textual or cognitive function in a given context. For example,
and can function as a textual connective; it can equally function as a cognitive
connective as well (cf. Section 2.4.4). The fact that the two components
take the same linear positions can be explained by the multifunctionality of
connective words which would appear in these positions.
To make another comparison, unlike in Iwasaki (1993), the cognitive
component, which corresponds to his IU-initial “lead”, occurred not only IU-
initially but also in other positions including IU-finally (although in much
lower proportions) in this study. In addition, given that the language-specific
cohesive component has been eliminated, the proposed linear order in (6.9)
could be applied cross-linguistically. For example, in spoken English, the
conjunction and, as a textual or cognitive connective, can occur in both pre-
ideational (IU-initial) and post-ideational (IU-final) positions (Chafe 1988).
On the other hand, the results of the present study are compatible with
Iwasaki’s (1993) in that the interactional component appeared only IU-finally.
The configuration (6.9), nevertheless, demonstrates that the functional
structure of the Japanese substantive IU is very orderly, such that the most
important and obligatory ideational component is preceded and followed by
either the textual or cognitive component, and the interactional component is
placed IU-finally. (Note that (6.9) represents a maximal functional structure,
given that [TX], [CG], and [IT] are optional components.) Given in (6.10)
is an example of substantive IUs that contain all of the four functional
components in the linear order specified in (6.9).
(6.10) [ TX ] [ ID ] [ TX ] [IT]
Y: soreni natsu da kara sa,
and summer cop because fp
‘And because (it is) summer’ (F5)
The IU (6.10) exemplifies the most complex functional structure type, com-
prised fully of the four components. The components are placed in the fol-
lowing order: textual (the conjunction soreni ‘and’), ideational (the nominal
predicate clause ‘(it is) summer’), textual (the conjunction kara ‘because’), and
interactional (the IU-final particle sa). Example (6.10) shows that the most
important ideational information is surrounded by two types of information,
textual and interactional. Note that although TX-ID-TX-IT is the type most
frequently produced in the four-component category, such four-component
IUs rarely occur in actual conversational discourse.
To summarize, the main findings of the present IU functional structure
analysis are the following:
Chapter 6
Of the 10 IUs, five IUs, (6.11a), (6.11d), and (6.11h)–(6.11j), are indepen-
dent clausal IUs each of which expresses a complete proposition. All of these
IUs constitute the simplest functional structure type, ID, the sole function of
which is to convey ideational content. The IUs (6.11b, c) collectively constitute
a multi-IU clause, in which the post-predicate NP IU (6.11c) serves as the di-
rect object of the verb ‘take’ uttered in the clausal IU (6.11b). Both of these are
unifunctional, contributing solely to the completion of a proposition. The IUs
(6.11e)-(6.11g) also constitute a multi-IU clause, in which (6.11e) and (6.11g)
are adverbial phrasal IUs. Of these, (6.11e, f) are two-component IUs com-
prised of the ideational component followed by the interactional component.
By uttering these IUs, speaker Y performs two functions—communicating
propositional content and soliciting hearer M’s involvement into her own
speakership. Overall, (6.11) strongly evidences the unifunctional nature of the
substantive IU in conversational Japanese.
The results of the present study are mostly compatible with those of
Iwasaki (1993) discussed in Section 2.5.2. First, concerning the percentage
of the functional structure type ID, his analysis resulted in 34%, compared
to 46% in this study. This is undoubtedly attributable to the fact that while
my data consist of substantive IUs only, his IUs include both substantive and
non-substantive IUs. The difference can also be explained by the fact that part
of his ID-CO is interpretable as my ID, given that part of his cohesive [CO]
components consist of such non-conjunctives as nominalizers and -te forms,
which were coded as part of the ideational [ID] component in my analysis.
As to the frequency of the functional type ID-IT, his analysis revealed 17%,
compared to 27% in my analysis. This is similarly attributable to the fact
that part of his ID-CO-IT is interpretable as my ID-IT for the same reason
given above.
Second, as to the percentage of IUs that include the interactional [IT]
component, his study showed 32%, which is quite comparable to 36% in my
study. The difference can partly be explained by the fact that IU-final no, which
Functional structure of the intonation unit in conversational Japanese
h. hyooka o.
evaluation acc
‘(high) evaluation’
i. zettai.
never
‘never’
speakers. However, the results of the present study show that this phrasal IU
option is exercised not all the time, but only in “special” occasions. These
include when the speaker introduces more than one new nominal referent,
and when the speaker is strongly motivated to communicate interactional
information to the interlocutor. Besides the informational and interactional
factors, other factors such as syntactic ones that will govern the production of
multi-IU clauses in spoken communication need to explored in future research
(see Croft 1995).3
. Summary
The present chapter has concerned itself with the analysis of the functional
composition of IUs as basic prosody-based units of Japanese conversation. It
has been shown in this chapter that overwhelmingly, the Japanese substantive
IU consists of the ideational component only, although it is sometimes fol-
lowed by the interactional component. That is, the substantive IU in conver-
sational Japanese has been found to be predominantly unifunctional, or min-
imally multifunctional. This means that typically, IUs are produced such that
they serve “one function at a time”. In other words, speakers opt not to perform
multiple functions but rather concentrate on one function—communication
of ideational content—within one substantive IU. The results suggest that the
preferred way Japanese conversational co-participants communicate ideas or
propositions is by means of the simplest functional structure type which con-
sists solely of the ideational component, without the other components serv-
ing a coherence-creating, regulatory, or interpersonal function, or as an idea-
linking device. Importantly, it has also been argued that one additional factor
responsible for the marked production of multi-IU clauses in spoken Japanese
is the interactional motivation on the part of the speaker. The breakup of a
clause into phrasal IUs will undoubtedly increase the frequency with which
non-ideational, interactional information can be communicated in conversa-
tional discourse.
Chapter 7
Conclusion
number of overtly expressed Given and New NPs. This suggests that conver-
sational Japanese, while making abundant use of zero-form Given NPs (55%
of the argument slots in the data were not filled with overt NPs), utilizes overt
NPs with both information statuses in comparable proportions in communi-
cating ideational content. It was additionally found that one new NP was intro-
duced every 1.6 clauses and every 2.3 IUs on average; this suggests the relatively
high-rate introduction of new nominal information by Japanese interactants in
conversational communication (Matsumoto 1997a).
Furthermore, it was found that the speakers preferentially produced non-
transitive, one-participant clauses with very low transitivity (typically intran-
sitive or nominal predicate clauses). It was also shown that the speakers
preferentially produced clauses with no overt arguments or one overt argu-
ment, and that the one-argument clauses typically contained S or O, not A.
That is, a single overt argument in the S or O role followed by a predicate
represents the maximal preferred clause structure in conversational Japanese;
this has been formulated as N(S/O) P. By contrast, clauses with overt A and
O arguments are strongly dispreferred, that is, two-argument clauses rarely
occur in actual interactive conversational discourse; this is consistent with
Lambrecht (1987) and Du Bois (1987). Given the low transitivity of conver-
sational Japanese and the “one overt argument per clause constraint”, I ar-
gued as a corollary that the traditional treatment of the highly transitive SOV
clause, which typically involves two overt lexical arguments, as canonical or
representative of Japanese clauses is not justified; but rather, more attention
needs to be directed to studies of grammar of one-participant clauses. Also
relevant to this is the finding that the overwhelming majority of A-role NPs
in Japanese conversation take zero forms. The results of the preferred nom-
inal structure analysis of the present study indicate that Du Bois’ (1987) PAS
constraints—the “One Lexical Argument”, “One New Argument”, “Non-lexical
A”, and “Given A” Constraints—hold in conversational Japanese (Matsumoto
2000a, 2002, 2003). Importantly, such recurrent patterns or tendencies in con-
versational interaction can be understood as having the power to shape the
foundations of grammar (Bybee & Hopper 2001; Du Bois 2003; Ford et al.
2003).
I further proposed in Chapter 5 that given the higher proportional oc-
currence of multiple new NPs per multi-IU clause, the marked production of
multi-IU clauses (i.e. the marked use of non-clausal, phrasal IUs) in Japanese
conversation can be related to the “one new NP per IU constraint”. That is,
my argument is that the one new NP per IU constraint is one of the motivat-
ing factors that will lead Japanese speakers to produce multi-IU clauses, or to
Chapter 7
divide a clause into separate IU elements, placing one new nominal concept
within one IU (Matsumoto 2000b; cf. Section 7.2).
Third, we found in Chapter 6 that the Japanese substantive IU tends
to be functionally non-elaborate, which supports the findings of Iwasaki’s
(1993) analysis. Among the four functional components—ideational [ID],
textual [TX], cognitive [CG], and interactional [IT]—that could be incorpo-
rated into one substantive IU, the speakers preferentially used only one compo-
nent (the obligatory ideational component) or two components (the obligatory
ideational component + one of the three other optional components) per IU.
That is, functionally, the substantive IU in conversational Japanese consists of
one or two components; the preferred number of functional components per
IU is no more than two. More specifically, the substantive IU preferentially
consists of the ideational component only ([ID]) or the ideational compo-
nent followed by the interactional component ([ID] [IT]). Given the notable
prevalence of [ID] in the data, I have further suggested that speakers com-
municate ideational information most typically by using the simplest func-
tional structure type which consists solely of the ideational component, with
none of the other components serving an interactional, coherence-creating, or
regulatory function, or as an idea-linking device. The analysis also revealed
that the maximal structure of the substantive IU in conversational Japanese
is very orderly, such that the most important and obligatory ideational com-
ponent is preceded and followed by the textual/cognitive component, and
the interactional component is placed IU-finally. This has been formulated as
[TX/CG] [ID] [TX/CG] [IT] (Matsumoto 1999a).
The above-presented major findings of the present study can be further
summarized as follows:
a. Japanese speakers prefer to produce IUs which are syntactically semi-
clausal, especially propositionally complete independent semi-clausal IUs.
b. Japanese speakers prefer to produce IUs which consist of a single nominal
carrying new or given information.
c. Japanese speakers prefer to produce IUs which functionally consist of the
ideational component only or the ideational plus interactional components.
Thus, the following hypotheses this study attempted to test have all been
confirmed (see Section 3.1).
HYPOTHESIS 1.1: The syntactic structure of the IU in conversational Japanese
tends to be semi-clausal.
Conclusion
rate IUs, as I have argued, can be seen as being motivated by the characteristics
of spoken language itself (Chafe 1982, 1988, 1994: Chapter 4; Clancy 1982;
Ochs 1979; Tannen 1982). The presence of a directly shared context and the
availability of prosodic and gestural resources for supplementary use negate the
necessity for elaborating the IU structures and linkages, on the one hand; and
the speakers’ focus on verbalizing ideas on the run—their most urgent task—
discourages such elaboration, on the other. Besides various extrinsic factors, we
also need to take into account intrinsic factors such as cognitive constraints on
humans’ information processing capacity. The “unitary” nature of substantive
IUs that has surfaced in this study suggests that speakers can handle only one—
proposition, new concept, overt argument, or functional component—within
one IU, a linguistic expression of the speaker’s focus of consciousness, a mini-
mal unit of thought organization (Chafe 2000). That is, “the magical number
one appears to be fundamental to the way the mind handles the flow of infor-
mation through consciousness and language” (Chafe 1994: 119). In conclusion,
the findings of the present study supports the “unitariness” or “singleness” of
the IU as the basic linguistic unit of discourse production and information flow
in spontaneous conversational interaction.
. Prospects
We have seen in Section 4.3 that about 8% of the substantive IUs contained
post-predicate phrases. This means that the overwhelming majority of the IUs
in the database consist of pre-predicate elements. We can therefore interpret
the results given above as reflecting the preferred structures of IUs which
involve non-postposed elements, or the canonical word order. Accordingly, an
exploration of postposing in relation to the preferred IU structures elucidated
in this study, an examination of how the structures of IUs involving postposing
deviate from the preferred IU patterns seems worth conducting in future
research. In addition, the way postposed nominals differ from the preferred
NP types in terms of grammatical role, information status, and syntactic form
is worth investigating.
We found in Section 4.3 that the post-predicate elements typically occured
as phrasal IU elements of multi-IU clauses. That is, the multi-IU clauses
exhibited a higher rate of postposing than the other IU syntactic types. It thus
seems evident that the production of postposed elements is related to that
of multi-IU clauses. In what way, then, is postposing related to the marked
production of multi-IU clauses? One answer would be that their relationship
Conclusion
be explored in detail. Further, how these two types of IUs, substantive and
regulatory—which form different aspects of thought, i.e. its “content” and
“infrastructure” respectively (Chafe 1998, 2000)—interact with each other as
they are successively produced by conversational co-participants needs further
study. What is also worthy of examination is how substantive IUs as the small-
est idea-conveying discourse units will cluster into larger thematically coher-
ent units in the collaborative and interactive construction of conversational
discourse (cf. Chafe 1980b, 1987, 1994; Gee 1986; Hinds 1980).
Second, given that this is a study of two-party conversations between
same-sex friends where a fairly large amount of information is supposed to be
shared or presupposed, other types of interactions merit further investigation.
For example, it would be interesting to explore interactions such as multi-
party conversations involving three or more participants, dyadic male-female
conversations, and conversations among participants who have met for the
first time and thus share very little background information. Another related
and interesting area worth exploring in subsequent research concerns gender
differences, that is, how males and females significantly differ in the production
of various types of IUs (cf. Matsumoto 1996; Section 3.4). Moreover, it is
worthwhile to conduct comparative studies of the structures of IUs produced
in different types of spoken discourse—not only ordinary conversations but
also other types of spoken discourse occurring in professional, institutional
encounters, for example, doctor-patient interaction, news interviews, and
courtroom discourse (cf. Boden & Zimmerman 1991; Drew & Heritage 1992;
Sarangi & Roberts 1999; van Dijk 1997). A comparative analysis of ordinary
conversation and institutional talk in particular would help us understand
the relationships between the IU structures, the roles of interactants, and the
purpose of communication (cf. Heritage 1997; Matsumoto 1999b).
Third, I proposed in this study that several factors are responsible for
the marked production of multi-IU clauses in conversational Japanese. I have
claimed in Chapter 5 (Section 5.6) that the “one new NP per IU constraint” is
one of the important motivating forces that will lead speakers into the produc-
tion of phrasal IUs. Additionally, I have suggested in Chapter 6 (Section 6.4)
that another factor facilitating the production of phrasal IUs in Japanese con-
versation is the interactional motivation on the part of the speaker. By divid-
ing a clause into phrasal IUs and terminating them with interactional particles
such as ne, the speaker can perform interactional encoding more frequently.
Importantly, it has also been pointed out in Chapter 2 (Section 2.5.1) that the
“loose” internal structure of Japanese clauses underlies speakers’ use of phrasal
IUs in conversational Japanese. The relation of the speaker’s marked produc-
Conclusion
tion of non-clausal IUs to other factors needs further study, both within and
across languages. Future research needs to be directed toward further explo-
ration of what speakers and/or their interlocutors can gain in conversational
interaction by breaking up a clause into phrasal IUs at the cost of efficiency
in communicating propositional content (see Croft 1995; Matsumoto 1998b,
1999a, 2000b, 2001).
Finally, it is expected that acoustically based analysis of intonation, which
is beyond the scope of the present research, will be effectively incorporated
into, and will interplay with, the discourse analysis of auditorily defined IUs.
Future studies are expected to have access to acoustic properties of sounds and
to supplement our perceptual observations with physical observations. This
approach will undoubtedly yield a greater depth of understanding and preci-
sion to our investigation of how speakers organize their messages into coherent
units as they jointly and interactively construct conversational discourse.
Notes
Chapter 1
. Hinds (1983: 53), for example, claims that “of 567 clauses, 8 (1.4%) evidenced a scram-
bled word order [i.e. OSV]”. Another piece of evidence for the basicness of SOV order comes
from the so-called “case-marker drop” phenomenon. When two NPs, whose subject-object
relation is unclear from the context, appear preverbally without case-marking particles as in
(i), the first NP (John) is invariably interpreted as the subject, and the second NP (Mary), as
the object, but not vice versa. This suggests that the SOV order is the unmarked basic word
order in Japanese (Matsumoto 1995b).
(i) John Mary butta yo.
hit-past fp
‘John hit Mary.’
Note that Kuno’s (1973) examples are constructed examples. In naturally occurring casual
Japanese conversation, case markers such as ga (the subject marker) and o (the direct object
marker) are often not used (cf. Fujii & Ono 2000; Ono et al. 2000).
. In this paper I will use the terms “postposed” and “post-predicate” interchangeably; I
will also use the terms “non-postposed” and “pre-predicate” interchangeably. Thus the term
“postposing”, as I use it here, refers to the placement of elements in a post-predicate position;
it does not imply movement of constituents from the “canonical” preverbal position.
. Past discourse-functional studies typically dealt with functions of marked word order
constructions used in naturally occurring spoken discourse, providing explanations for
pragmatic reordering in various syntactically based languages (cf. Downing & Noonan 1995;
Mithun 1987; Payne 1992). The marked word order constructions that have been discussed
in the literature are of three types: preposing, postposing, and inversion (e.g. Birner & Ward
1998; Dorgeloh 1997; Duranti & Ochs 1979; Fox 1985; Green 1980; Silva-Corvalán 1983;
Ward 1988).
Notes
. Matsumoto (1995a) has shown that of 1,526 IUs, 84 (5.5%) involved postposing. Note
that post-predicate elements in Japanese are found only in spontaneous casual spoken
discourse. They do not normally occur in written discourse or planned spoken discourse.
Chapter 2
. According to Givón (1988: 67), the scale accords with the iconicity principle for topic
continuity (i.e. “the more continuous/predictable is the topic/referent NP, the less overt
expression it needs to receive”), which, in turn, is an instance of the general iconicity
principle in communication (i.e. “the more predictable the information, the less coding it
receives”).
. Reid (1977) argues that the two past tense forms in French, passé simple and imparfait,
call for “high focus” and “low focus” of attention, respectively. These can be taken to be the
foregrounding and backgrounding forms of the verb, respectively. His data show that the
passé simple tends to be used with actions, affirmative verbs, human, singular, first-person,
proper-name subjects, main characters of discourse, and main clauses. These seem to accord
with Hopper and Thompson’s (1980) high-transitivity features.
. Transitivity is a global property of an entire clause such that an activity is carried over
or transferred from an agent to a patient. It has ten component parts: participants, kinesis,
aspect, punctuality, volitionality, affirmation, mode, agency, affectedness of O (object), and
individuation of O. These features collectively allow clauses to be coded as more or less
transitive—the more high transitive features a clause has, the more transitive it is, and the
closer it is to cardinal transitivity (Hopper & Thompson 1980: 251–253).
. Reinhart (1984) and Wallace (1982) argue that the foreground-background distinction in
discourse rests on the innate, universal perceptual “figure-ground” distinction. They suggest
that there is a striking correlation between the perceptual criteria determining the figure and
those determining the discourse foreground, that is, foreground in discourse is “perceptually
more salient” than background.
. Givenness of an idea may be established in a “textually evoked” way; that is, an idea is first
introduced into a discourse as new information and then remains given for a certain period
of time. It may also be established in a “situationally evoked” way; for example, because of
the active status of the ideas of the speaker and the hearer, references to first and second
persons (i.e. I and you) are typically given (Chafe 1994: 78–79; Prince 1981: 236).
. Chafe (1987: 29–30) discusses two ways in which concepts become semi-active: (a)
through deactivation from an earlier active state, and (b) through evocation of a “schema”,
which is generally regarded as a cluster of interrelated expectations (Schank & Abelson 1977;
Tannen 1979). When a schema is evoked in a discourse, some of the expectations or concepts
associated with it are assumed to change into the semi-active state. For example, a “class”
schema includes such concepts as “students”, “classroom”, and “lecture” as accessible entities.
Notes
. Exceptions to this pattern include the occurrence of “competitive” referents (where given
information is expressed with a full noun), and that of “contrastive” referents (where given
information receives a primary accent) (Chafe 1994: 76–77).
. Noun phrases that are outside the domain of identifiability are of two types: those that
express generic referents, and those that are non-referential (e.g. non-referential it, negative
and universal pronouns, question words, predicate nouns, and non-specific referents in an
irrealis context) (Chafe 1994: 102–105).
. The segmentation of spoken discourse into prosodic units, it should be noted, is most
closely concerned with one of the grammatical functions of intonation, i.e. its “segmenta-
tive” or “delimitative” function, while simultaneously involving its other functions as well
(e.g. to indicate finality/non-finality, show the speaker’s emotions, or regulate the speaker-
hearer relationship) (Vandepitte 1989; cf. Section 3.2).
. Chafe (1980b) uses the term “idea units”, which are essentially the same as IUs.
. It is important to realize that the IU analysis is a unit summary system, where the
intonation contour is indicated only once per unit. In other words, a symbol such as a
period indicating a fall in pitch placed at the end of an IU does not show an intonational
event taking place just at that point; rather it represents a set of intonational movements that
occur over the course of the unit in which it appears (Du Bois et al. 1992: 113).
. Chafe (1994: 65) notes that a more careful identification of IU boundaries in his more
recent work has led him to revise the modal length of substantive IUs in English, which was
previously identified as five words (Chafe 1980b, 1987, 1993).
. AUs, like IUs, are either substantive or regulatory. A substantive AU contains at least
one open-class word such as a noun, verb, or adjective; whereas a regulatory AU verbalizes
a linkage, or it may be an interpersonal or cognitive formula (e.g. and then, you know, let’s
see). The distinction between given, accessible, and new information thus applies only to
substantive AUs; regulatory AUs are outside the domain of activation cost (Chafe 1993: 40).
. Linguists working in the tradition of “functional sentence perspective” claim that
linguistic elements vary in their degree of CD (communicative dynamism), which is
characterized as “the relative extent to which a linguistic element contributes towards the
further development of the communication” (Firbas 1992: 8).
. Iwasaki and Tao (1993) defined the terms as follows: “An IU is ‘clausal’ if it contains a
verbal predicate; any IUs lacking a verbal predicate are ‘non-clausal’. The clausal unit which
contains a verbal predicate and its associated core arguments (at least the subject) is a ‘full
clause’; the clausal unit with a verbal predicate but without its associated core arguments is a
‘semi-clause’. ‘Nominal’ (NP) IUs refer to any non-clausal unit which consists of a nominal
element, whether or not it is modified”.
. Iwasaki and Tao (1993) examined Mandarin Chinese IUs as well. They found the
proportions of non-clausal IUs (61%), semi-clausal IUs (63%), and NP IUs (23%) in
Mandarin comparable to those in Japanese. That is, the Mandarin and Japanese data were
Notes
found to be similar in terms of the amount of “syntactic fragmentation” they exhibited. The
preferential use of semi-clauses in Japanese and Mandarin can be attributed to the fact that
both languages, unlike English, allow for abundant use of zero anaphora.
. Brown and Yule (1983: 159–160), examining Halliday’s conversational English data, note
that the “information unit” is more likely to take the form of the phrase rather than the
clause, and argue for abandoning the clause as the unmarked syntactic domain of the unit.
Maynard (1989) likewise reports that her English conversation data displayed a similar rate
of segmentation into phrasal units as was found in comparable Japanese data. This would
suggest that spoken English is also “fragmented”.
. The Pear Film narratives were collected as part of a research project investigating the
verbalization of remembered experience (Chafe 1980a). For the project, a brief film without
dialog—which portrays various adventures of a boy with a basket of pears on his bicycle—
was shown to speakers of several different languages; they were then asked to tell what
had happened in it to an interviewer of the same age and sex. An obvious advantage of
this approach is that we can focus on the cross-linguistic comparison of linguistic forms
while rigidly controlling the content. Its major disadvantage, on the other hand, would
be that these narratives, elicited under controlled conditions, do not constitute natural,
representative samples of our everyday interactive use of conversational language.
. Concerning this issue, Iwasaki (1993: 41) states as follows: “For the speaker, who attends
to more than one type of information, it is easier to deal with a smaller amount of
the ideational content of information. This can be achieved by producing part of the
proposition, or clause.” What evidence can there be for the argument that the speaker
can handle smaller rather than larger ideational content more easily when faced with the
need to process multiple types of information? It might be easier (and definitely more
efficient) for the speaker to deal with clausal information rather than phrasal information.
Moreover, it is not always the case that phrasal IUs contain smaller amount of information
than clausal IUs.
Chapter 3
. There were fairly frequent topic shifts, and subtopics appeared and recurred during the
45-minute conversations. It seems that the female conversations generally involved more
topic shifts than the male conversations.
. The transcribed IUs do not include what Maynard (1986) calls “turn-internal listener
backchannels”, or what Schegloff (1982) calls “continuers”—brief backchannelling expres-
sions (e.g. un ‘uh-huh’) that the interlocutor who assumes primarily a listener’s role sends
during the other interlocutor’s speaking turn, especially in a long multi-unit turn (e.g.
storytelling).
Notes
. 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.
. Of the 32 regulatory IUs produced in F6, the number and percentage of the subtypes
are as follows: interactional (N=14, 44%), cognitive (N=7, 22%), textual (N=6, 19%), and
validational (N=5, 15%). My analysis has shown that one linguistic form can serve different
regulatory functions in different contexts, or simultaneously serve multiple functions in a
given context (cf. Section 2.4.4). Although this study focuses on substantive IUs, detailed
research on regulatory IUs will also be needed in the future.
Chapter 4
. The nominal predicates occurred either with or without the so-called “copula” (see (4.1)
and (4.3b) for examples). Given that the copula in Japanese can be attached to almost all
types of constituents, even to adverbs and conjunctions, I do not consider it to be equivalent
to the “be-verb” in English.
. I coded the phrasal IUs for the five syntactic categories based on the functions they
perform within the clause they belong to. For example, besides ga- and o-marked NPs, I
coded NP-ni as in John ni au ‘meet John’ as argument NP, whereas NP-ni as in Tokyo ni
sumu ‘live in Tokyo’ was coded as AvP. In addition, I coded NP-no (NP plus the genitive
case-marking particle no) as attributive AP.
. This means that the study assumes that independent phrasal IUs without formally
expressed predicates (which may be assumed as in the case of Type 4 stray NPs to be
Notes
discussed in Section 4.5.2) are propositionally incomplete and thus cannot constitute a
clause conveying a full proposition.
. Shibatani (1990: 275–278) distinguishes two types of topics in Japanese. One is a “base-
generated”, “genuine” topic that is not integratable into the clausal structure and that
expresses an entity about which a judgment is made, as in (i). The other is a “derived” topic
which he considers is a stylistic device similar to scrambling, as in (ii). In this study I coded
only “base-generated”, “genuine” wa-marked NPs as in (i) as topics.
(i) Tookyoo wa daremo konakatta.
Tokyo top no one come-neg-past
‘Tokyo is such that no one came (from there).’
(ii) Tookyoo kara wa daremo konakatta.
Tokyo abl top no one come-neg-past
‘From Tokyo, no one came.’
. This indicates that the IUs in (4.1) have been taken from male dyad M5. Hereafter,
sources of examples will be shown likewise.
. In this study, an intra-IU clause or a non-finite clause was not counted as propositionally
complete; it was coded simply as part of the single-IU/multi-IU clause in which it occurs.
. Of the 1,600 IUs, 14% (N=231) contained a total of 244 intra-IU clauses, constituting
complex clauses or complex NPs. Of the 244 intra-IU clauses, relative clauses (36%), clausal
objects (34%), and subordinate clauses (20%) accounted for 90% (nominal complement
clauses and clausal subjects accounted for the remaining 10%). It was also found that like
the clausal IUs, the intra-IU clauses tend to consist of subjectless verbal predicates only, and
they were produced typically within the independent single-IU clauses, especially within
[PVP] IUs.
. Further analysis has shown that of the 123 postposed phrases, 55% are nominals, and
36% are adverbials (the remaining 9% are adjectives and mixed phrases). It was also shown
that of the postposed NPs, 32% are Type 1, 38% are Type 2, and 30% are Type 3, whereas
73% of the postposed adverbials belong to Type 2. This suggests that postposed adverbials,
unlike postposed NPs, have a marked tendency to appear as phrasal IU constituents of
multi-IU clauses.
. I classified the “proposition-constituting” NP IUs as one type of stray NPs, given that
occasionally the speaker’s intended meaning cannot be understood by the listener, and thus
communication breakdown results. This seems to justify that this type of NPs may still be
termed “stray”.
. Clancy et al. (1996) classify collaborative finishes as one type of what they call
“reactive tokens”, which include non-lexical backchannels, lexical reactive expressions, and
repetitions. Given that most collaborative finishes are contentful, and moreover, provide
new information, I am not in favor of treating them simply as the non-primary speaker’s
non-floor-taking reactive tokens. Of the two adjacently uttered IUs which constitute a
collaboratively constructed proposition, Clancy et al. (1996) focus on the second IU uttered
Notes
by the hearer as in (4.15c), whereas this study focuses on the first IU uttered by the primary
speaker as in (4.15b). In my coding scheme, the second IU is normally treated as an inde-
pendent clausal IU conveying a full proposition which assumes the NP just uttered by the
primary speaker in the immediately preceding turn.
. Some of the lead NPs were expanded into other grammatical categories such as adjectives
and adverbials in the following clause, as in kyoo ‘today’ (NP lead) → kyoo no ‘today’s’
(AP target).
. Comparing the independent NPs identified and discussed in Croft (1995), Tao (1996),
and this study, Croft’s “presentative” NPs would correspond to the “genuine topic” NPs
in my classification of detached NP IUs; his “topic” NPs would correspond to my “LDed”
NPs involving resumptive pronouns. I coded two subtypes of Tao’s “referential” NP IUs—
“referent-listing” NPs and “referent-anchoring” NPs (i.e. NPs describing the same referent
from different angles, which Croft calls “elaborating” NPs)—as “independent clausal” NPs,
given that each of them constitutes a complete proposition and provides a piece of newly
introduced information. Further, Tao’s “interactional” NP IUs consist of two subtypes: NPs
as repetitions and NPs as collaborative finishes. The former correspond to the “repeated”
NPs, whereas the latter are clearly related to “Type 2 stray” NPs in the taxonomy provided
in the present study. It may be concluded that the grammatically independent, detached
NP with a specific discourse function is a cross-linguistically occurring phenomenon in
spontaneous spoken discourse.
. Although I coded sentential adverbs as clause-external substantive IUs, they may be
classified as regulatory IUs, given that their function is not so much to express ideas as to
show particular kind of relationships between the preceding and following IUs.
. A conspicuous feature of the multi-IU clauses is that 30% of them included post-
predicate phrasal IU elements, as illustrated in (4.29). In particular, full clauses composed
of three IUs exhibited the highest rate of postposing; 52% of them involved post-predicate
elements. The relationship between the production of multi-IU clauses and post-predicate
phrases is worth examining in future research (cf. Section 4.3).
Chapter 5
. Following Chafe (1996: 41), “referents” are defined here more broadly as “ideas of
people, objects, or abstractions” that could function as participants in event or state
ideas. By definition, then, all NPs have “referents” and thus can be coded for information
statuses. As Lambrecht (1994: 37) notes, the information structure analysis is concerned
with referents (i.e. “entities and states of affairs designated by linguistic expressions in
particular utterances”) and the abstract mental representations of these discourse referents
in the mind of speech participants.
. Du Bois (1987) uses 20 IUs for this measure in his analysis of the Pear Story Sakapulteko
narratives, following Givón’s (1983a) measure of RD, and given the one-to-one correspon-
Notes
dence between the clause and the IU in his data (cf. Sections 2.1, 2.6.2). I used 25 IUs (=20
IUs multiplied by 1.23) instead based on the finding that in my Japanese data one clause
contained 1.23 IUs on average (see Section 4.7). This means that I have chosen 25 IUs as
an arbitrary measure during which an active concept will be deactivated into a semi-active
state, or a concept changes from a focally active to a peripherally active state (Chafe 1987).
. There are some problems with the use of the S-A-O categories as a heuristic as well as
the notion of argument structure (Mithun & Chafe 1999; Thompson & Hopper 2001). This
study has nevertheless chosen to use the S-A-O schema primarily because important cross-
linguistic comparisons with previous information-flow studies such as Du Bois (1987), Tao
(1996), and Schuetze-Coburn (1994), all of which employed the schema, are made possible.
. Thus, for example, in a passive clause such as Mary ga John ni butareta. ‘Mary was hit by
John’, I coded ‘Mary’ as S (the single argument of the intransitive passive verb butareta), and
‘John’ as oblique. Typically the roles S and A are marked by the particle ga/wa, whereas the
role O is marked by the particle o/wa. In casual Japanese conversation, however, many NPs
have no postpositions with them (cf. Note 2 in Chapter 1).
. The rest of the oblique NPs consist of the following: NPs occurring in attributive
adjectival phrases (N=19, 6%), topics (N=11, 3%), and complements (N=2, 1%).
. The following provides a comparison between the present study and Du Bois (1987) in
the percentage of New NPs in the four grammatical roles:
S A O Oblique Total
This study 39 20 49 58 49
Du Bois 22 3 25 39 20
The higher proportion of New NPs in the Japanese data may also be attributed to the
fact that in the Pear Story narratives the number of NPs that could be introduced are
more restricted than in informal conversations, in which a wider range of NPs could be
introduced.
. About 70% of the pronouns are demonstrative, as opposed to personal, pronouns. Of the
103 demonstrative pronouns, 61% (N=63) are sore ‘it’, which includes soo ‘so’ (N=7) (23%
are are ‘that’, and 16% are kore ‘this’). Of the 44 personal pronouns, 84% (N=37) are 1st
person ‘I’ atashi/boku/ore, which includes ‘we’ (N=3) (9% are 2nd person ‘you’ anata, and
7% are 3rd person ‘he/she’ kare/kanojo).
. In Du Bois (1987), (a) a new argument NP was introduced every 4.32 clauses (IPQ=0.23)
on average; (b) a new non-argument NP was introduced every 6.45 clauses (IPQ=0.16) on
average; and (c) a new NP was introduced every 2.59 clauses (IPQ=0.39) on average.
. Du Bois (2003) argues that this argument pattern applies not only to intransitive and
transitive clauses but also to ditransitive clauses involving verbs such as give, tell, and show.
Among the three core arguments that a ditransitive verb takes, i.e. subject (A), direct object
(O), and indirect object (I), he shows, only the O role freely allows lexical NPs and new
information in English conversation. This means that in both transitive and ditransitive
clauses, the O role constitutes the only lexical/new argument allowable within a clause.
Notes
. As I argued in Section 4.5.2, the production of “lead” NPs seems also related to the one
new NP per IU constraint. In most cases, the speakers first introduced new information as
a lead phrasal IU and then “converted it into given” in the following clausal IU, in order
to prevent the clause from containing two new NPs. This can be considered as one of the
speakers’ strategies for conforming to this constraint.
Chapter 6
. The conjunction kedo ‘though’, an informal form of keredomo ‘although’, often occurs
independently without main clauses in spoken Japanese, presumably as a softening device
which makes an asserted proposition less definite (Makino & Tsutsui 1989). In this study
every occurrence of kedo, with or without main clauses, was coded as [TX].
. In this study the functional structure of an intra-IU clause was not coded separately
but was taken simply as constituting part of the ideational [ID] component (cf. Note 7 in
Chapter 4). This means that some of the IUs actually exhibit more complex functional
structures. For example, (i) was coded as TX-ID, but actually involves more complex func-
tional composition, i.e. TX-ID-IT-ID, where the sequence ID-IT (the adjectival predicate
semi-clause muzukashii ‘(it is) difficult’ + the interactional particle na) constitutes the
intra-IU embedded clausal object of the verb ‘think’.
(i) M: dakara muzukashii na: to omotte.
so difficult fp qt think-and
[ TX ] [ ID ]
{ TX } { ID } {IT} { ID }
‘So (I) think that (it is) difficult.’ (F3)
Similarly, some of the IUs involving postposing exhibit structures more functionally com-
plex than those coded in the present study. For example, (ii), although coded as ID, actually
involves more complex structure, ID-IT-ID, with the interactional particle na separating the
nominal predicate hantai da ‘is converse (to yours)’ and the post-predicate element atashi
‘my (case)’.
(ii) M: hantai da na: atashi. ((laugh))
converse cop fp I
[ ID ]
{ ID } {IT} { ID }
‘is converse (to yours), my (case).’ (F4)
. Croft (1995) proposes three syntactic factors that will cause an English clause to be
broken up into multiple IUs. They are “parallelism” (i.e. coordinate sentences are almost
always broken), “complexity” (i.e. complex subject NPs tend to be broken), and “distance”
(i.e. adjuncts as opposed to complements tend to be found on separate IUs). Whether these
constraints apply cross-linguistically awaits future research.
References
Andersen, Roger W. (1991). A Poor Man’s Guide to Interlanguage Analysis. Los Angeles:
Department of TESL and Applied Linguistics, University of California, Los Angeles.
Ashby, William J. (1988). The syntax, pragmatics, and sociolinguistics of left- and right-
dislocations in French. Lingua, 75, 203–229.
Ashby, William J. & Paola Bentivoglio (1993). Preferred argument structure in spoken
French and Spanish. Language Variation and Change, 5, 61–76.
Atkinson, J. Maxwell & John Heritage (Eds.). (1984). Structures of Social Action: Studies of
Conversation Analysis. Cambridge: Cambridge University Press.
Beckman, Mary E. & Janet B. Pierrehumbert (1986). Intonational structure in Japanese and
English. Phonology Yearbook, 3, 255–309.
Birner, Betty J. & Gregory L. Ward (1998). Information Status and Noncanonical Word Order
in English. Amsterdam: John Benjamins.
Boden, Deirdre & Don H. Zimmerman (Eds.). (1991). Talk and Social Structure: Studies in
Ethnomethodology and Conversation Analysis. Cambridge: Polity Press.
Bolinger, Dwight (1989). Intonation and its Uses. Stanford, CA: Stanford University Press.
Brazil, David (1985). The Communicative Value of Intonation in English. Birmingham:
English Language Research, University of Birmingham.
Brown, Gillian & George Yule (1983). Discourse Analysis. Cambridge: Cambridge University
Press.
Bybee, Joan L. & Paul J. Hopper (Eds.). (2001). Frequency and the Emergence of Linguistic
Structure. Amsterdam: John Benjamins.
Chafe, Wallace L. (1972). Discourse structure and human knowledge. In R. O. Freedle &
J. B. Carroll (Eds.), Language Comprehension and the Acquisition of Knowledge (pp. 41–
69). Washington, DC: V. H. Winston.
Chafe, Wallace L. (1974). Language and consciousness. Language, 50, 111–133.
Chafe, Wallace L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point
of view. In C. Li & S. A. Thompson (Eds.), Subject and Topic (pp. 27–55). New York:
Academic Press.
Chafe, Wallace L. (1979). The flow of thought and the flow of language. In T. Givón
(Ed.), Discourse and Syntax: Syntax and Semantics, Vol. 12 (pp. 159–181). New York:
Academic Press.
Chafe, Wallace L. (Ed.). (1980a). The Pear Stories: Cognitive, Cultural, and Linguistic Aspects
of Narrative Production. Norwood, NJ: Ablex.
Chafe, Wallace L. (1980b). The deployment of consciousness in the production of a
narrative. In W. L. Chafe (Ed.), The Pear Stories: Cognitive, Cultural, and Linguistic
Aspects of Narrative Production (pp. 9–50). Norwood, NJ: Ablex.
References
Chafe, Wallace L. (1982). Integration and involvement in speaking, writing, and oral
literature. In D. Tannen (Ed.), Spoken and Written Language: Exploring Orality and
Literacy (pp. 35–53). Norwood, NJ: Ablex.
Chafe, Wallace L. (1987). Cognitive constraints on information flow. In R. Tomlin (Ed.),
Coherence and Grounding in Discourse (pp. 21–51). Amsterdam: John Benjamins.
Chafe, Wallace L. (1988). Linking intonation units in spoken English. In J. Haiman &
S. A. Thompson (Eds.), Clause Combining in Grammar and Discourse (pp. 1–27).
Amsterdam: John Benjamins.
Chafe, Wallace L. (1993). Prosodic and functional units of language. In J. A. Edwards &
M. D. Lampert (Eds.), Talking Data: Transcription and Coding in Discourse Research
(pp. 33–43). Hillsdale, NJ: Lawrence Erlbaum.
Chafe, Wallace L. (1994). Discourse, Consciousness, and Time: The Flow and Displacement of
Conscious Experience in Speaking and Writing. Chicago: University of Chicago Press.
Chafe, Wallace L. (1996). How consciousness shapes language. Pragmatics and Cognition, 4,
35–54.
Chafe, Wallace L. (1998). Language and the flow of thought. In M. Tamasello (Ed.), The
New Psychology of Language: Cognitive and Functional Approaches to Language Structure
(pp. 93–111). Mahwah, NJ: Lawrence Erlbaum.
Chafe, Wallace L. (2000). The interplay of prosodic and segmental sounds in the expression
of thoughts. In M. L. Juge & J. L. Moxley (Eds.), Proceedings of the 23rd Annual Meeting
of the Berkeley Linguistic Society 1997 (pp. 389–401). Berkeley, CA: Berkeley Linguistic
Society.
Chesterman, Andrew (1991). On Definiteness: A Study with Special Reference to English and
Finnish. Cambridge: Cambridge University Press.
Clancy, Patricia M. (1980). Referential choice in English and Japanese narrative discourse.
In W. L. Chafe (Ed.), The Pear Stories: Cognitive, Cultural, and Linguistic Aspects of
Narrative Production (pp. 127–202). Norwood, NJ: Ablex.
Clancy, Patricia M. (1982). Written and spoken style in Japanese narratives. In D. Tannen
(Ed.), Spoken and Written Language: Exploring Orality and Literacy (pp. 55–76).
Norwood, NJ: Ablex.
Clancy, Patricia M. (2003). The lexicon in interaction: Developmental origins of Preferred
Argument Structure in Korean. In J. W. Du Bois, L. E. Kumpf, & W. J. Ashby (Eds.),
Preferred Argument Structure: Grammar as Architecture for Function (pp. 81–108).
Amsterdam: John Benjamins.
Clancy, Patricia M., Sandra A. Thompson, Ryoko Suzuki, & Hongyin Tao (1996). The
conversational use of reactive tokens in English, Japanese, and Mandarin. Journal of
Pragmatics, 26, 355–387.
Clark, Herbert H. & Susan E. Haviland (1977). Comprehension and the given-new contract.
In R. O. Freedle (Ed.), Discourse Production and Comprehension (pp. 1–40). Norwood,
NJ: Ablex.
Clark, Herbert H. & Catherine R. Marshall (1981). Definite reference and mutual
knowledge. In A. K. Joshi, B. L. Webber, & I. A. Sag (Eds.), Elements of Discourse
Understanding (pp. 10–63). Cambridge: Cambridge University Press.
Comrie, Bernard (1978). Ergativity. In W. P. Lehmann (Ed.), Syntactic Typology (pp. 329–
394). Austin: University of Texas Press.
References
Ford, Cecilia E., Barbara A. Fox, & Sandra A. Thompson (Eds.). (2002). The Language of
Turn and Sequence. Oxford: Oxford University Press.
Ford, Cecilia E., Barbara A. Fox, & Sandra A. Thompson (2003). Social interaction and
grammar. In M. Tamasello (Ed.), The New Psychology of Language: Cognitive and
Functional Approaches to Language Structure, Vol. 2 (pp. 119–143). Mahwah, NJ:
Lawrence Erlbaum.
Fox, Barbara A. (1985). Word-order inversion and discourse continuity in Tagalog. Text, 5,
39–54.
Fox, Barbara A. & Sandra A. Thompson (1990). A discourse explanation of the grammar of
relative clauses in English conversation. Language, 66, 297–316.
Fox, Barbara A., Makoto Hayashi, & Robert Jasperson (1996). Resources and repair: A cross-
linguistic study of syntax and repair. In E. Ochs, E. A. Schegloff, & S. A. Thompson
(Eds.), Interaction and Grammar (pp. 185–237). Cambridge: Cambridge University
Press.
Fujii, Noriko & Tsuyoshi Ono (2000). The occurrence and non-occurrence of the Japanese
direct object marker o in conversation. Studies in Language, 24, 1–39.
Gee, James P. (1986). Units in the production of narrative discourse. Discourse Processes, 9,
391–422.
Geluykens, Ronald (1992). From Discourse Process to Grammatical Construction: On Left
Dislocation in English. Amsterdam: John Benjamins.
Givón, Talmy (1975). Focus and the scope of assertion: Some Bantu evidence. Studies in
African Linguistics, 6, 185–205.
Givón, Talmy (Ed.). (1983a). Topic Continuity in Discourse: A Quantitative Cross-Language
Study. Amsterdam: John Benjamins.
Givón, Talmy (1983b). Topic continuity in discourse: An introduction. In T. Givón
(Ed.), Topic Continuity in Discourse: A Quantitative Cross-Language Study (pp. 1–41).
Amsterdam: John Benjamins.
Givón, Talmy (1984). Syntax: A Functional-Typological Introduction, Vol. I. Amsterdam:
John Benjamins.
Givón, Talmy (1987). Beyond foreground and background. In R. S. Tomlin (Ed.), Coherence
and Grounding in Discourse (pp. 175–188). Amsterdam: John Benjamins.
Givón, Talmy (1988). Topic continuity in discourse: The functional domain of switch
reference. In J. Haiman & S. A. Thompson (Eds.), Clause Combining in Grammar and
Discourse (pp. 51–82). Amsterdam: John Benjamins.
Givón, Talmy (1990). Syntax: A Functional-Typological Introduction, Vol. II. Amsterdam:
John Benjamins.
Green, Georgia M. (1980). Some wherefores of English inversions. Language, 56, 582–601.
Grice, H. Paul (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Speech Acts:
Syntax and Semantics, Vol. 3 (pp. 41–58). New York: Academic Press.
Grimes, Joseph E. (1975). The Thread of Discourse. The Hague: Mouton.
Haiman, John & Sandra A. Thompson (Eds.). (1988). Clause Combining in Grammar and
Discourse. Amsterdam: John Benjamins.
Halliday, Michael A. K. (1967a). Intonation and Grammar in British English. The Hague:
Mouton.
References
Halliday, Michael A. K. (1967b). Notes on transitivity and theme in English: Part 2. Journal
of Linguistics, 3, 199–244.
Halliday, Michael A. K. (1973). Explorations in the Functions of Language. London: Edward
Arnold.
Halliday, Michael A. K. (1985). An Introduction to Functional Grammar. London: Edward
Arnold.
Halliday, Michael A. K. (1989). Language, Context, and Text: Aspects of Language in a Social-
Semiotic Perspective. London: Oxford University Press.
Halliday, Michael A. K. (1994). An Introduction to Functional Grammar (2nd ed.). London:
Edward Arnold.
Halliday, Michael A. K. & Ruqaiya Hasan (1976). Cohesion in English. London: Longman.
Hawkins, John A. (1978). Definiteness and Indefiniteness: A Study in Reference and
Grammaticality Prediction. London: Croom Helm.
Hayashi, Makoto (2003). Joint Utterance Construction in Japanese Conversation. Amsterdam:
John Benjamins.
Helasvuo, Marja-Liisa (2001). Syntax in the Making: The Emergence of Syntactic Units in
Finnish Conversation. Amsterdam: John Benjamins.
Heritage, John (1997). Conversation analysis and institutional talk: Analyzing data. In
D. Silverman (Ed.), Qualitative Research: Theory, Method and Practice (pp. 161–182).
London: Sage.
Hinds, John (1976). Postposing in Japanese. Eoneo: The Journal of the Linguistic Society of
Korea, 1, 113–125.
Hinds, John (1980). Japanese conversation, discourse structure, and ellipsis. Discourse
Processes, 3, 263–286.
Hinds, John (1982). Ellipsis in Japanese Discourse. Alberta, Canada: Linguistic Research, Inc.
Hinds, John (1983). Topic continuity in Japanese. In T. Givón (Ed.), Topic Continuity
in Discourse: A Quantitative Cross-Language Study (pp. 47–93). Amsterdam: John
Benjamins.
Hinds, John, Senko K. Maynard, & Shoichi Iwasaki (Eds.). (1987). Perspectives on
Topicalization: The Case of Japanese WA. Amsterdam: John Benjamins.
Hirst, Daniel & Albert Di Cristo (Eds.). (1998). Intonation Systems: A Survey of Twenty
Languages. Cambridge: Cambridge University Press.
Hoji, Hajime (1985). Logical Form Constraints and Configurational Structures in Japanese.
Ph.D. dissertation, University of Washington, Seattle.
Hopper, Paul J. (1979). Aspect and grounding in discourse. In T. Givón (Ed.), Discourse and
Syntax: Syntax and Semantics, Vol. 12 (pp. 213–241). New York: Academic Press.
Hopper, Paul J. & Sandra A. Thompson (1980). Transitivity in grammar and discourse.
Language, 56, 251–299.
Horie, Kaoru (2000). Core-oblique distinction and nominalizer choice in Japanese and
Korean. Studies in Language, 24, 77–102.
Iwasaki, Shoichi (1985). The ‘Given A Constraint’ and the Japanese particle ga. In
S. DeLancey & R. S. Tomlin (Eds.), Proceedings of the First Annual Pacific Linguistics
Conference (pp. 152–167). Eugene: Department of Linguistics, University of Oregon.
Iwasaki, Shoichi (1992). Subjectivity in Grammar and Discourse: Theoretical Considerations
and a Case Study of Japanese Spoken Discourse. Amsterdam: John Benjamins.
References
Iwasaki, Shoichi (1993). The structure of the intonation unit in Japanese. In S. Choi (Ed.),
Japanese/Korean Linguistics, Vol. 3 (pp. 39–53). Stanford, CA: Center for the Study of
Language and Information.
Iwasaki, Shoichi (2002). Japanese. Amsterdam: John Benjamins.
Iwasaki, Shoichi & Hongyin Tao (1993). A comparative study of the structure of the
intonation unit in English, Japanese, and Mandarin Chinese. Paper presented at the
67th Annual Meeting of the Linguistic Society of America, Los Angeles, CA.
Jones, L. B. & Jones, L. K. (1979). Multiple levels of information in discourse. In L. K. Jones &
R. E. Longacre (Eds.), Discourse Studies in Mesoamerican Languages (pp. 3–28).
Arlington: University of Texas, Arlington.
Kärkkäinen, Elise (1996). Preferred argument structure and subject role in American
English conversational discourse. Journal of Pragmatics, 25, 675–701.
Kreckel, Marga (1981). Tone units as message blocks in natural discourse: Segmentation of
face-to-face interaction by naive, native speakers. Journal of Pragmatics, 5, 459–476.
Kuno, Susumu (1973). The Structure of the Japanese Language. Cambridge, MA: MIT Press.
Kuno, Susumu (1978a). Danwa no Bumpoo [Grammar of Discourse]. Tokyo: Taishuukan.
Kuno, Susumu (1978b). Generative discourse analysis in America. In W. Dressler (Ed.),
Current Trends in Textlinguistics (pp. 275–294). Berlin: Mouton de Gruyter.
Kuno, Susumu (1978c). Japanese: A characteristic OV language. In W. P. Lehmann (Ed.),
Syntactic Typology: Studies of the Phenomenology of Language (pp. 57–138). Austin:
University of Texas Press.
Labov, William (1972). The transformation of experience in narrative syntax. In Language
in the Inner City: Studies in the Black English Vernacular (pp. 354–396). Philadelphia:
University of Pennsylvania Press.
Ladd, D. Robert (1996). The Intonational Phonology. Cambridge: Cambridge University
Press.
Lambrecht, Knud (1987). On the status of SVO sentences in French discourse. In
R. S. Tomlin (Ed.), Coherence and Grounding in Discourse (pp. 217–261). Amsterdam:
John Benjamins.
Lambrecht, Knud (1994). Information Structure and Sentence Form: Topic, Focus, and the
Mental Representations of Discourse Referents. Cambridge: Cambridge University Press.
Lerner, Gene H. (1991). On the syntax of sentences-in-progress. Language in Society, 20,
441–458.
Longacre, Robert E. (1976). The discourse structure of the flood narrative. In G. MacRae
(Ed.), Society of Biblical Literature 1976 Seminar Papers (pp. 235–262). Missoula, MT:
Scholars Press.
Makino, Seiichi & Michio Tsutsui (1989). A Dictionary of Basic Japanese Grammar. Tokyo:
The Japan Times.
Matsumoto, Kazuko (1995a). Fragmentation in conversational Japanese. Japan Association
for Language Teaching Journal, 17, 238–253.
Matsumoto, Kazuko (1995b). Wh-preposing, scrambling, and topicalization in Japanese.
The Bulletin of Aichi University of Education, 44, 45–62.
Matsumoto, Kazuko (1996). Intonation Units in Conversational Japanese: Structure and
Linkage. Ph.D. dissertation, University of California, Los Angeles.
Matsumoto, Kazuko (1997a). NPs in Japanese conversation. Pragmatics, 7, 163–181.
References
Ochs, Elinor, Emanuel A. Schegloff, & Sandra A. Thompson (Eds.). (1996). Interaction and
Grammar. Cambridge: Cambridge University Press.
O’Dowd, Elizabeth (1990). Discourse pressure, genre and grammatical alignment—after
Du Bois. Studies in Language, 14, 365–403.
Ono, Tsuyoshi & Ryoko Suzuki (1992). Word order variability in Japanese conversation:
Motivations and grammaticization. Text, 12, 429–445.
Ono, Tsuyoshi & Sandra A. Thompson (1994). Unattached NPs in English conversation. In
S. Gahl, A. Dolbey, & C. Johnson (Eds.), Proceedings of the 20th Annual Meeting of the
Berkeley Linguistic Society (pp. 402–419). Berkeley, CA: Berkeley Linguistic Society.
Ono, Tsuyoshi & Sandra A. Thompson (1995). What can conversation tell us about syntax?
In P. W. Davis (Ed.), Alternative Linguistics: Descriptive and Theoretical Modes (pp. 213–
271). Amsterdam: John Benjamins.
Ono, Tsuyoshi & Sandra A. Thompson (1996). Interaction and syntax in the structure of
conversational discourse. In E. H. Hovy & D. R. Scott (Eds.), Discourse Processing: An
Interdisciplinary Perspective (pp. 67–96). Heidelberg: Springer-Verlag.
Ono, Tsuyoshi, Sandra A. Thompson, & Ryoko Suzuki (2000). The pragmatic nature of the
so-called subject marker ga in Japanese: Evidence from conversation. Discourse Studies,
2, 55–84.
Pawley, Andrew & Frances H. Syder (1983). Natural selection in syntax: Notes on adaptive
variation and change in vernacular and literary grammar. Journal of Pragmatics, 7,
551–579.
Pawley, Andrew & Frances H. Syder (2000). The one-clause-at-a-time hypothesis. In
H. Riggenbach (Ed.), Perspectives on Fluency (pp. 163–199). Ann Arbor: University of
Michigan Press.
Payne, Doris L. (Ed.). (1992). Pragmatics of Word Order Flexibility. Amsterdam: John
Benjamins.
Pierrehumbert, Janet B. & Mary E. Beckman (1988). Japanese Tone Structure. Cambridge,
MA: MIT Press.
Pierrehumbert, Janet B. & Julia Hirschberg (1990). The meaning of intonational contours
in the interpretation of discourse. In P. R. Cohen, J. Morgan, & M. E. Pollack (Eds.),
Intentions in Communication (pp. 271–311). Cambridge, MA: MIT Press.
Prince, Ellen F. (1981). Toward a typology of given-new information. In P. Cole (Ed.),
Radical Pragmatics (pp. 223–255). New York: Academic Press.
Prince, Ellen F. (1992). The ZPG letter: Subjects, definiteness, and information-status. In
W. C. Mann & S. A. Thompson (Eds.), Discourse Description: Diverse Linguistic Analyses
of a Fund Raising Text (pp. 295–325). Amsterdam: John Benjamins.
Reid, Wallis (1977). The quantitative validation of a grammatical hypothesis: The passé
simple and the imparfait. In J. A. Kegl et al. (Eds.), Proceedings of the 7th Annual
Meeting of the North East Linguistic Society (pp. 315–333). Cambridge, MA: North East
Linguistic Society.
Reinhart, Tanya (1984). Principles of gestalt perception in the temporal organization of
narrative texts. Linguistics, 22, 779–809.
Sacks, Harvey, Emanuel A. Schegloff, & Gail Jefferson (1974). A simplest systematics for the
organization of turn-taking for conversation. Language, 50, 696–735.
References
Saeki, Tetsuo (1975). Gendai Nihongo no Gojun [Word Order of Modern Japanese]. Tokyo:
Kasama Shoin.
Saito, Mamoru (1985). Some Asymmetries in Japanese and their Theoretical Implications.
Ph.D. dissertation, MIT, Cambridge, MA.
Sarangi, Srikant & Celia Roberts (Eds.). (1999). Talk, Work and Institutional Order: Discourse
in Medical, Mediation and Management Settings. Berlin: Mouton de Gruyter.
Schank, Roger C. & Robert P. Abelson (1977). Scripts, Plans, Goals, and Understanding: An
Inquiry into Human Knowledge Structures. Hillsdale, NJ: Lawrence Erlbaum.
Schegloff, Emanuel A. (1979). The relevance of repair to syntax-for-conversation. In
T. Givón (Ed.), Discourse and Syntax: Syntax and Semantics, Vol. 12 (pp. 261–286).
New York: Academic Press.
Schegloff, Emanuel A. (1982). Discourse as an interactional achievement: Some uses
of ‘uh huh’ and other things that come between sentences. In D. Tannen (Ed.),
Analyzing Discourse: Text and Talk, Georgetown University Round Table on Languages
and Linguistics 1981 (pp. 71–93). Washington, DC: Georgetown University Press.
Schegloff, Emanuel A. (1996). Turn organization: One direction for inquiry into grammar
and interaction. In E. Ochs, E. A. Schegloff, & S. A. Thompson (Eds.), Interaction and
Grammar (pp. 52–133). Cambridge: Cambridge University Press.
Schegloff, Emanuel A. (1998). Reflections on studying prosody in talk-in-interaction.
Language and Speech, 41, 235–263.
Schegloff, Emanuel A., Gail Jefferson, & Harvey Sacks (1977). The preference for self-
correction in the organization of repair in conversation. Language, 53, 361–381.
Scheibman, Joanne (2002). Point of View and Grammar: Structural Patterns of Subjectivity in
American English Conversation. Amsterdam: John Benjamins.
Schiffrin, Deborah (1987). Discourse Markers. Cambridge: Cambridge University Press.
Schuetze-Coburn, Stephan (1994). Prosody, Syntax, and Discourse Pragmatics: Assessing
Information Flow in German Conversation. Ph.D. dissertation, University of
California, Los Angeles.
Schuetze-Coburn, Stephan, Marian Shapley, & Elizabeth Weber (1991). Units of intonation
in discourse: A comparison of acoustic and auditory analyses. Language and Speech, 34,
207–234.
Selkirk, Elizabeth O. (1984). Phonology and Syntax: The Relation between Sound and
Structure. Cambridge, MA: MIT Press.
Selting, Margaret (1996). On the interplay of syntax and prosody in the constitution of
turn-constructional units and turns in conversation. Pragmatics, 6, 371–388.
Selting, Margaret (2000). The construction of units in conversational talk. Language in
Society, 29, 477–517.
Selting, Margaret & Elizabeth Couper-Kuhlen (Eds.). (2001). Studies in Interactional
Linguistics. Amsterdam: John Benjamins.
Shibatani, Masayoshi (1990). The Languages of Japan. Cambridge: Cambridge University
Press.
Silva-Corvalán, Carmen (1983). On the interaction of word order and intonation: Some
OV constructions in Spanish. In F. Klein-Andreu (Ed.), Discourse Perspectives on Syntax
(pp. 117–140). New York: Academic Press.
References
Smith, Wendy (1996). Spoken narratives and preferred clause structure: Evidence from
modern Hebrew discourse. Studies in Language, 20, 163–189.
Stein, Dieter & Susan Wright (Eds.). (1995). Subjectivity and Subjectivization. Cambridge:
Cambridge University Press.
Sugito, Miyoko (Ed.). (1989). Nihongo to Nihongo Kyooiku [Japanese and Japanese Language
Teaching], Vol. 2. Tokyo: Meiji Shoin.
Sugito, Miyoko (Ed.). (1990). Nihongo to Nihongo Kyooiku [Japanese and Japanese Language
Teaching], Vol. 3. Tokyo: Meiji Shoin.
Svartvik, Jan (1982). The segmentation of impromptu speech. In N. E. Enkvist (Ed.),
Impromptu Speech: A Symposium (pp. 131–145). Åbo, Finland: Åbo Akademi.
Takami, Kenichi (Ed.). (1995). Nichieigo no Uhoo Idoo Koobun [Rightward Movement
Constructions in English and Japanese]. Tokyo: Hitsuji Shoboo.
Tanaka, Hiroko (1999). Turn-Taking in Japanese Conversation: A Study in Grammar and
Interaction. Amsterdam: John Benjamins.
Tannen, Deborah (1979). What’s in a frame? Surface evidence for underlying expectations.
In R. O. Freedle (Ed.), New Directions in Discourse Processing (pp. 137–181). Norwood,
NJ: Ablex.
Tannen, Deborah (Ed.). (1982). Spoken and Written Language: Exploring Orality and
Literacy. Norwood, NJ: Ablex.
Tao, Hongyin (1996). Units in Mandarin Conversation: Prosody, Discourse, and Grammar.
Amsterdam: John Benjamins.
Teramura, Hideo (1984). Nihongo no Sintakkusu to Imi [Japanese Syntax and Semantics],
2 vols. Tokyo: Kuroshio Shuppan.
Thompson, Sandra A. (1997). Discourse motivations for the core-oblique distinction as a
language universal. In A. Kamio (Ed.), Directions in Functional Linguistics (pp. 59–82).
Amsterdam: John Benjamins.
Thompson, Sandra A. & Paul J. Hopper. (2001). Transitivity, clause structure, and argument
structure: Evidence from conversation. In J. L. Bybee & P. J. Hopper (Eds.), Frequency
and the Emergence of Linguistic Structure (pp. 27–60). Amsterdam: John Benjamins.
Tokieda, Motoki (1950). Nohon Bumpoo [Japanese Grammar]. Tokyo: Iwanami.
Tomlin, Russell S. (1985). Foreground-background information and the syntax of
subordination. Text, 5, 85–122.
Tomlin, Russell S. (Ed.). (1987). Coherence and Grounding in Discourse. Amsterdam: John
Benjamins.
Vandepitte, Sonia (1989). Pragmatic function of intonation: Tone and cognitive
environment. Lingua, 79, 265–297.
van Dijk, Teun A. (Ed.). (1997). Discourse as Social Interaction, Discourse Studies: A Multi-
disciplinary Introduction, Vol. 2. London: Sage.
Venditti, Jennifer J. (2003). The J_ToBI model of Japanese intonation. In S.-A. Jun (Ed.),
Prosodic Transcription and Typology: A Unified Approach. Oxford: Oxford University
Press.
Wallace, Stephen (1982). Figure and ground: The interrelationships of linguistic categories.
In P. J. Hopper (Ed.), Tense-Aspect: Between Semantics and Pragmatics (pp. 201–223).
Amsterdam: John Benjamins.
References
Ward, Gregory L. (1988). The Semantics and Pragmatics of Preposing. New York: Garland.
Ward, Gregory L. & Julia Hirschberg (1985). Implicating uncertainty: The pragmatics of
fall-rise intonation. Language, 61, 747–776.
Index
Argument 5, 8–9, 33, 41–42, 52–54, 56, Background 10–12, 15, 21, 25, 39, 127,
60, 69–70, 79, 89, 96, 101, 103, 166, 170
105, 108–111, 116–118, defined 10–11
120–125, 127–134, 138–139, vs. foreground 10–12, 170
147, 155–156, 160–161, Backgrounding 11–12, 65, 165, 170
163–165, 171–173, 176 Bantu languages 39
defined 103 Bare noun 104, 108, 113–117, 139, 160
overt vs. null 132–134 Base-generated topic See Genuine topic
types and definitions 103 Baseline pitch level 21
vs. non-argument 103–104, 110, Basic word order 3–4, 63, 169
116, 118, 138–139, 160, 176 vs. marked word order 4, 63
vs. predicate 53, 56, 69–70 See also Canonical; Unmarked word
See also A role; Core argument; order
Core-oblique distinction; O Beckman, Mary E. 28–29, 46
role; Preferred Argument Bentivoglio, Paola 3, 42
Structure; S role Birner, Betty J. 1, 14, 169
Argument slot 118, 120, 132, 161 Boden, Deirdre 166
filled vs. unfilled with overt NP Bolinger, Dwight 46
132–133 Boundary, IU 2, 20–21, 46, 143, 150,
Argument-oblique combination 124 171
A-role argument See A role marker/signal of 20, 46, 143, 150
Article 11, 18–19 See also Definite; Brand-new information 14 See also
Indefinite article Evoked; Unused information
Ashby, William J. 3, 42, 71 Brazil, David 20
Aspect 12, 105, 170 Brown, Gillian 17–18, 30, 172
Assessments 127 Bybee, Joan L. 161
Assumed familiarity 14
Atkinson, J. Maxwell 46
C
Attention focus 27–28, 170
Canonical word order 4, 164
Attitudes 126–127
Case marker 104, 169 See also
Attributive adjective/AP 18, 53, 55, 65,
Case-marking particle
89, 103, 173, 176
Case-marker drop phenomenon 169
vs. predicate adjective/AP 53, 89
Case-marking particle 33, 147, 169, 173
AU See Accent unit
Casual conversation/spoken discourse
Auditory unit 2
51, 169–170, 173, 176
CD See Communicative dynamism
B Chafe, Wallace L. 1–3, 7, 9, 13–29, 36,
Backchannel xvi, 32, 87, 172–174 39–42, 46, 48–49, 54, 62, 83, 97, 99,
defined 172 102, 113, 120, 122, 124, 137,
in narrative 173 142–143, 151–152,159–160, 164, 166,
in spoken Japanese 32 170–172, 175–176
types xvi Chesterman, Andrew 18
See also Continuer; Reactive token; Chinese See Mandarin
Turn-internal listener Choice, speaker’s 1, 65, 165
backchannel Choice question 72
Index
Clancy, Patricia M. 3–4, 28, 32, 63, 65, tight vs. loose internal structure of
74, 87, 97, 132, 159–160, 164, 174 33
Clarification request 71–72 types and definitions 105
Clark, Herbert H. 13, 18 vs. phrase 29–33, 51–53, 57–60, 98,
Clausal IU 3, 26, 29, 32–33, 37, 51–53, 172
55, 57–62, 64–69, 71, 80–81, 85, See also Adjectival predicate; Full;
87, 89, 98–100, 106, 121–122, Intransitive; Multi-IU; Nominal
137, 147, 154–157, 159–160, predicate; One-participant;
171–172, 175, 177 Semi-; Single-IU; Transitive;
defined 51 Two-participant clause
robustness of 160 Clause centrality proposal 3, 26, 29, 33,
types and distribution 52, 59 62, 99, 152, 159
vs. phrasal IU 3, 36–37, 57–60, 62, defined 25–26, 152
155–157, 172 Clause chaining 27
See also Full clausal; Independent Clause core 41, 122–123
clausal; Semi-clausal IU; Clause vs. phrase centrality 30, 51, 99
Single-IU clause Clause-external NP 103–104 See also
Clausal IU as part of multi-IU clause 52, Independent NP
56, 58–60, 64–66, 69, 89 Clause-external postposing 64, 84–85,
definition and subtypes 52 87–88, 165 See also Postposed NP
Clausal IU strategy, vs. phrasal 33 Clause-internal NP 81, 103, 106, 108,
Clausal object 52, 57, 104–105, 124, 110 See also Argument; Oblique
174, 177 Clause-internal postposing 64–65, 85,
Clausal structure 53–54, 57–58, 64, 88, 165
71–72, 98, 159, 174 Code 35, 44, 48, 51–54, 104, 142–144,
Clausal subject 104–105, 174 147, 155
Clause 2–3, 8–9, 12, 20, 25–27, 29–33, Coding 31, 46, 54, 62, 101–102, 121,
37–38, 41–42, 51, 53–54, 62–63, 123, 141, 143, 155, 160
71–72, 74, 79, 94–95, 97–99, Coding categories 48, 101
101, 105–106, 117–118, Coding scheme 32–33, 142, 144, 175
121–132, 134, 137–139, Cognitive activation, vs. search 8
141–142, 144, 152, 156–163, Cognitive component 142–143,
165–167, 172, 176 145–146, 148–152, 162
and IU in English 2–3, 25–27 Cognitive connective 143, 150
and IU in Japanese 3, 29–32, 94–98 Cognitive constraint/limitation 23, 36,
and NP 105–106 163–164
as basic unit of discourse production/ Cognitive cost 16, 138 See also
information flow/processing Activation cost
2–3, 27 Cognitive function 143–144, 151
as exponent of IU 26, 51, 63, 99, Cognitive planning See Planning
159, 163 Cognitive process/processing 15, 143,
breakup/division of 32, 37, 101, 148
139, 141, 156–158 Cognitive subtype 22, 49, 91, 143, 173
defined 53–54, 159 Coherence 7, 10, 33, 38, 143, 158, 162
English vs. Japanese 33 Cohesion 34–35, 142
Index
Conversation analysis 173 Croft, William 3, 19, 72, 88, 158, 167,
Conversational acts 76, 88 175, 177
Conversational communication 159, Cross-linguistic applicability/validity
161 29, 42, 88, 99, 159
Conversational co-participants See Cross-linguistic comparison 62, 144,
Co-participants 172, 176
Conversational discourse 48, 51, 77, Cruttenden, Alan 2, 20–21, 46
123, 126, 131, 139, 143–144, Crystal, David 2, 20
151, 158, 161, 163, 166–167
low transitivity/non-event/non-
action-oriented nature of D
126–127, 161 Data analysis 48
Conversational English 27, 62, 113, 172 Data base 44–45
Conversational interaction 5, 11, 30, Data coding 51, 101, 141
49–50, 124, 127, 155, 161, 164, 167 Data transcription 46
Conversational Japanese 5, 32, 43, 51, Declarative clause 46
57, 62–63, 65, 88, 94, 96–97, 101, Declination unit 2
106, 110–111, 113, 115–117, 119, Definite article 11, 18–19
122–124, 126, 128, 131, 133–134, Definite NP 7, 9, 17
138, 141, 152, 154, 158–163, 166 vs. indefinite NP 7, 17
Conversational language 49, 57, 87, 127, Definiteness 1, 9, 18 See also
172 Identifiability
Conversational monologue 27 Demonstrative adjective + NP 104,
Conversational narrative 40 113–115, 117
Conversational space 10, 118 Demonstrative pronoun 18, 104,
Cooperative principle 13 113–115, 117, 176
Coordinate sentence 177 Derived topic, vs. base-generated 174
Coordinating conjunction 28, 141 Descriptive function 34
Co-participants 18, 22, 30, 32, 48–49, Detached NP IU 56, 69, 71, 88, 90,
54, 61, 69, 73–76, 79, 88, 92, 99, 102, 159–160, 175 See also Independent
106, 117, 137, 139, 143, 152, 158, phrasal NP IU
163, 166, 169, 173 Di Cristo, Albert 29, 46
Copied postposed NP 55, 72, 85, 88 Dialogic conversation/interaction 5, 157
Copula 41, 62, 105, 173 See also Dik, Simon C. 57
Adjectival predicate; Linking verb; Direct object 9, 31, 55, 103, 105, 123,
Nominal predicate 126, 147–148, 154, 169, 176
Copula + PP/NP construction 41 vs. indirect object 9
Core argument 41, 52–54, 103, Direct object marker 169
120–121, 171, 176 Disagreement 71–72
Core grammatical role 103, 113 See Discourse See Conversational; Spoken
also A; O; S role discourse
Core-oblique distinction 113, 134 Discourse analysis 167
Corpus 41, 67, 88, 105, 134 Discourse context See Context
Couper-Kuhlen, Elizabeth 3, 19 Discourse functions 65, 160, 175
Courtroom discourse 166 Discourse marker 22
Index
Identifiable referent 18–19, 81, 100, 102 Information flow 1–3, 5, 7, 19–20, 22,
vs. non-identifiable referent 19 25, 38–39, 49, 88, 120, 160,
Immediate mode 49 164–165
Inactive state 2, 15–17, 39, 42, 138 basic/canonical/unmarked pattern of
25, 120
Indefinite article 11
constraints on 38–42
Indefinite NP 7–8, 17
defined 1
Independent adverbial IU 89–92, 100
units of 1–3, 5, 19–20, 38, 164
types and distribution 90 Information focus 25
Independent clausal IU 51–52, 55, Information pressure 117–118
58–60, 62, 66, 89, 98–99, 106, Information pressure quotient (IPQ)
110–111, 137, 154, 174–175 117
definition and subtypes 52 Information processing 2, 27, 39, 164
vs. clausal IU as part of multi-IU Information status 1, 12–18, 81, 88,
clause 58, 66, 89 100–102, 104, 111–112,
vs. independent phrasal IU 54–55, 115–117, 139, 160–161,
58 164–165, 175
Independent NP 54, 71–72, 79, 81, 83, and grammatical role 111–113
103–104, 108–114, 117, 124, 175 and NP syntactic form 115
See also Clause-external NP; as property of nominal reference
Non-argument 102
Independent phrasal IU 51, 53, 55, See also Accessible; Given; New
58–60, 65, 71, 89–90, 93, 98, information
103–104, 113, 173 Information/informational structure, IU
definition and subtypes 53 3, 5–6, 23–25, 43, 48, 81, 97,
vs. phrasal IU as part of multi-IU 101, 118–119, 139, 144, 152,
clause 52–53, 58, 65–66, 89 159–160, 163
Independent phrasal NP IU 54, 70–88, in English 23–25
111, 159 in Japanese 118–120
See also Unmarked information
functions of 88
structure
types, definitions and distribution
Information structure types, IU
71–72
118–120, 139
See also Detached NP IU; Lead;
distribution of 118–121
Left-dislocated; Postposed;
Information unit 2, 25–26, 172
Repeated; Right-dislocated;
Informational factor, and multi-IU clause
Stray; Topic NP
158 See also One new NP per IU
Independent semi-clausal IU 58, 60–61, constraint
67, 98, 106, 110–111, 123, 162 Informational hybrid 38
Indirect object 9, 176 Information-flow factors 1, 72, 88
Inferrable information 14–15, 18 Infrastructure, of thought 166
Informal conversation 26, 79, 88, 109, Initial pause 15, 20, 75
160, 176 Insertion/inserted sequence 74, 78
Informal form 54, 70, 177 Institutional talk 166
Information block 2 Intensifier 31, 118
Index
Intention, speaker’s 21, 34, 65 Intonation unit (IU) xv, 1–3, 5, 15,
Interactants 11, 18, 37, 47, 67, 83, 86, 20–23, 25–28, 33, 37, 43, 47–49,
90, 161, 166 51, 57, 60, 62, 66, 68, 71, 89, 94,
Interaction 19, 22, 32, 37, 77, 79, 85, 87, 101–102, 105–106, 117–118,
127, 143 See also Conversational 124, 134, 137–138, 141, 144,
interaction 146, 152, 156, 159–161,
Interactional component 34–36, 43, 163–164, 167, 171, 173
142–143, 145–148, 150–152, and AU 23–24
154–158, 162–163 and clause 2–3, 25–27, 29–32, 62,
Interactional factors 72, 88 94–98
Interactional function 143–144, 162 and consciousness/language
Interactional information 3, 32, 36–37, production 21–22
151–152, 155–158 and NP 105–106
Interactional involvement, as basic unit of discourse production/
co-participants’ 76, 88 See also information flow 1, 5, 19, 164
Collaborative completion; defined 20
Interruption; Repair identification criteria 20–21
Interactional motivation, and multi-IU linkage 25–28, 164
clause 158, 166
prosodic features of 21
Interactional NP 175
singleness/unitariness of 144, 164
Interactional particle 31, 33, 35, 37, 75,
types and size of 22–23
143, 147, 150, 155, 157, 166, 177
See also Functional; Informational;
as marker of IU boundary 150
Syntactic structure
See also Final particle
Intonational phrase 20
Interactional subtype 22, 49, 173
Interactive conversation 29, 48, 79, 117, Intra-IU clause 57, 104–105, 124, 174,
157, 161 177
Interlocutor 1, 10, 14, 49, 73, 87, 91, Intransitive clause 105, 121, 124–129,
157–158, 167, 172 131–133, 161, 176
Interpersonal function 33–34, 147, 158 Intransitive subject 41, 103, 132 See
Interrogativity 46 also S role
Interruption 49, 77, 88 See also Intransitive verb 103, 124
Other-; Self-interruption Inversion 169
Intonation 4, 27–28, 46, 165, 167, 171 IPQ See Information pressure quotient
functions of 46, 165, 171 Irrealis 171
See also Continuing; Falling; Rise-fall; IU See Intonation unit
Rise-fall-rise; Rising intonation IUs with zero NPs 118–121
Intonation contour xvi, 2, 20–21, 28, 46, types of 118, 120–121
75, 169, 171, 173 See also No associated NPs; Zero
acoustic measurement of 169, 173 anaphora
as signal of IU boundary 20, 46 Iwasaki, Shoichi 3, 28–32, 34–37, 42, 46,
coherent/unified 2, 20–21, 46 48, 62, 67, 97, 99, 116, 127, 130,
types xvi, 46 141–144, 150–151, 154–155,
Intonation group 2, 20 159–160, 162, 171–172
Index
Low transitive clause 105, 125–129, and syntactic factors 158, 177
131–133 full vs. semi- 56–57, 94–97
marked production of 139, 158,
160–161, 164, 166
M
vs. single-IU clause 94–98,
Main clause 10, 52, 57, 121, 170, 177
124–125, 128, 134–135, 138, 160
vs. subordinate clause 52, 57, 121
Multi-IU/unit turn 83, 172
Makino, Seiichi 177
Multi-party conversation 166
Male conversation, vs. female 44–45,
Multiple IUs, one clause strategy 97
47–48, 166, 172
Mandarin 9, 17, 81, 88, 105, 171–172
Marked word order 4, 54, 63, 169 N
vs. basic/unmarked word order 4,
Narrative 3, 11–12, 30, 32, 35, 40–41,
54, 63
48, 88, 111–113, 117–118, 127,
See also Postposing; Scrambling
172–173, 175–176
Marshall, Catherine R. 18
vs. conversation 30, 173–174
Matsumoto, Kazuko xi, 4, 51, 63,
See also Conversational narrative
101–102, 141, 159–163, 165–167,
Narrative skeleton 11
169–170, 172
Native speakers 1, 4, 29, 44
Maynard, Senko K. 3, 28, 30, 32, 65, 97,
Natural vs. elicited discourse 30
127, 155, 159–160, 172
Naturally occurring conversation/
Metafunction 33–34 See also
discourse 4–5, 17, 130, 169
Cohesive; Descriptive; Experiential;
Expressive; Ideational; Interactional; Nespor, Marina 20
Interpersonal; Logical; Subjective; New information/NP 2, 10, 12–17, 19,
Textual function 24–25, 38–43, 88, 100–103,
Mithun, Marianne 169, 176 111–113, 115–121, 123–124,
Mixed phrasal IU 56, 89, 93 131, 133–139, 152, 157–158,
Mixed phrase 53, 63, 174 160, 165, 170–171, 174–177
Modifier 18, 89, 91, 114 and cognitive cost 16, 138
Mori, Junko 19 and IU/clause 134–138
Multifunctionality 36, 144, 151, 155 constraint on the flow rate/quantity
Multi-IU clause 51–58, 60, 64–66, 69, of 39–42, 137
71, 73, 85, 89, 94, 96–98, 101, defined 12–14, 103
108, 124–125, 128, 134–139, integration of 38–39
141, 147, 149, 154–158, See also Brand-new; Unused
160–161, 164–166, 174–175 information
and coding of interactional Newness 9, 12–15
information 155–158 News interview 166
and Japanese vs. English syntax 33, No associated NPs 120–121, 136
157–158 No more than two functional components
and multifunctionality 155–158 per IU constraint 149, 152, 163
and one new NP per IU constraint No more than/up to two functions per IU
134–138, 157, 161–162, 166 constraint 36, 155
and post-predicate phrase 65, No more than two IUs per clause
164–165, 175 constraint 95, 163
Index
PAS See Preferred Argument Structure vs. clause 29–33, 51–53, 57–60, 98,
Passive clause/verb 176 172
Patient 103, 105, 170 Phrase-orientedness 33
Pattern 1, 5, 25, 33, 36, 41–42, 63, PI See Potential interference
87–88, 96, 115, 118, 130, 132, 147, Pierrehumbert, Janet B. 19, 28, 46
159, 161, 164–165 Pitch 21, 32, 46, 171, 173
Pause xvi, 2, 16, 20–21, 34–35, 46, 49, Planning 32, 34, 91, 142, 148, 150
75, 86, 142–143, 149 Point of view 127 See also Subjectivity
types xvi Polysynthetic language 23
See also Initial pause Possessor 18
Pause filler 34–35, 49, 143, 149 Postposed NP, clause-external 71–72,
Pause-bounded phrasal unit (PPU) 32 84–88, 159, 165, 174 See also
Pawley, Andrew 2, 26–27, 36, 39, 41 Copied; Restated postposed NP
Payne, Doris L. 169 Postposed phrase, clause-internal/external
Pear Film/Story 32, 41, 172, 175–176 See Post-predicate phrase
See also Elicited narrative
Postposing 4, 54, 63–66, 87, 102,
Perception 19, 127, 173 164–165, 169–170, 175, 177
Performance error 80–81, 88, 159
and backgrounding/defocusing 65,
Personal pronoun 104, 108, 113–115, 85, 165
176
and intonation 4
Phonological prominence 17
and IU 63–66
Phrasal IU 3, 29–33, 36–37, 51, 53–56,
and multi-IU clause 65, 164–165,
58–62, 64–65, 68–69, 71, 89, 93,
175
96, 98, 101, 103–104, 113,
defined 4
121–122, 141, 147, 154–158,
161, 164, 166–167, 172–175, 177 functions of 65, 85, 88, 165
as element/part of multi-IU clause types of 64–65
52–53, 64–65, 69, 89, 164 See also Clause-external;
defined 51 Clause-internal postposing
marked use of 161 Postposition 176 See also
types and distribution 52–53, 61, Case-marking particle
89 Post-predicate element/phrase 4, 31,
vs. clausal IU 57–60, 155–157, 172 53–54, 56–57, 63–65, 70, 97,
See also Independent phrasal IU 129, 136, 154, 164–165,
Phrasal IU as part of multi-IU clause 169–170, 174–175, 177
52–53, 56, 58–60, 64–65, 69, 89, vs. pre-predicate element/phrase 4,
164 54, 63, 164–165, 169
definition and subtypes 52 See also Marked word order
Phrasal IU option, vs. clausal 33, 158 Postverbal element See Post-predicate
Phrasality, marked case of 155 element
Phrase 10, 17–20, 30–31, 51–58, 61, Potential interference (PI) 8, 165
63–65, 74, 97–100, 103, 106, PP See Prepositional phrase
121, 123, 142, 149, 159, 164, PPU See Pause-bounded phrasal unit
172, 174–176 Predicate 12, 25–26, 31–33, 37, 39, 41,
types 52–53 51–54, 69, 74–80, 89, 92, 99,
Index
Spoken English 2–3, 17, 26, 28–30, 37, Subject-object-verb (SOV) 3–4, 130,
46, 63, 97, 151, 172 161, 169
Spoken Japanese 3, 29, 32, 34, 36–37, as basic word order in Japanese
46, 116, 155, 157–158, 177 3–4, 169
Spoken language 19, 164 Subject-predicate structure 25
non-elaboration of IU structures and Subordinate clause 52, 57, 104–105,
characteristics of 163–164 121, 124, 174
See also Conversational language vs. main clause 52, 57, 121
Spontaneous spoken discourse/ Subordinating conjunction 148
conversation 1–2, 4, 20, 27, 127, Subordination 27
144, 164, 170, 175 Substantive IU 22–26, 29, 35–36, 48–51,
SR See Switch reference 57, 62, 83, 99, 123, 139, 142,
S-role argument See S role 146–147, 149, 151–155,
Stance 34, 126 158–160, 162–166, 171, 173, 175
Starting point 19, 21, 25 See also defined 22, 49
Added information vs. non-substantive IU 22, 48, 154,
State 3, 22, 26–27, 41, 49, 54, 102, 165
126–127, 143, 152, 159 See also Fragmentary; Regulatory IU
State idea 26, 175 Subtopics, of conversation 172
Sufficiently identifying verbalization
Stein, Dieter 127
18–19
Storage locus 38
Sugito, Miyoko 29, 47
Story preface 173
Suzuki, Ryoko 4, 64
Storytelling 172
Svartvik, Jan 19
Strategy, speaker’s 29, 39, 65, 81, 88,
Switch reference (SR) 8, 165
118–119, 177
Syder, Frances H. 2, 26–27, 36, 39, 41
Stray NP 71–79, 88, 93, 159, 173–175
Syntactic factors, and multi-IU clause
defined 71 158, 177
types and definitions 73–74 Syntactic form, NP 101, 104, 111,
Stress 32 113–117, 139, 164
Structures, IU 3, 5, 48, 144, 163–164, and grammatical role 113–115
166 and information status 115
non-elaborate nature of 164 Syntactic fragmentation See
Subject 9, 12, 19, 25, 37, 52, 54–55, 57, Fragmentation
67, 69–70, 96, 100, 103, 110, Syntactic structure, IU 3, 5–6, 29, 32,
112–113, 147–148, 159, 43, 48, 51, 65, 98, 144, 152, 159,
169–171, 176 162–163
vs. object 9, 110, 169 in English vs. Japanese 3, 29–33, 62
vs. predicate 12, 25, 37 Syntactic structure types, IU 51–53,
See also Intransitive; Transitive 57–58, 60, 63, 66, 98–99,
subject 106–107, 109, 159, 164
Subject marker 169 and NP 106–107
Subjectivity 34, 127 See also Point of and postposing 63–66
view core/essential vs. peripheral/
Subject-object relation 169 redundant 57
Index
U
W
Uncodable IU 49–50
Wallace, Stephen 11, 170
Unifunctionality 153, 155
Ward, Gregory L. 1, 14, 47, 169
Unit See Discourse unit; Intonation
unit; Prosodic unit Wh-question 79
Unitariness 164 Word 2, 20, 23–25, 27, 32, 34, 38–40, 46,
Unmarked information structure 12, 50, 142, 171
25, 120 Word order 3–4, 12, 54, 63, 129, 164,
Unmarked word order 54, 129 169
Unshared information 14, 19 in Japanese spoken discourse 3–4
Unused information 14 marked vs. unmarked/basic/
Utterance 2, 13, 15, 17, 19, 22, 25, canonical 3–4, 54, 63, 164, 169
33–34, 77, 81, 86–87, 92, 100, 127, SOV 3–4, 169
173, 175 See also Basic; Marked word order
Index