View
216
Download
2
Category
Preview:
Citation preview
THE EFFECT OF NOUN PHRASE
INTERPRETATION ON SENTENCE
PROCESSING IN JAPANESE
a dissertation
submitted to the department of english linguistics
of kobe shoin graduate school of letters
in partial fulfillment of the requirements
for the degree of
doctor of philosophy
Adviser: GUNJI Takao
By
IGUCHI Yoko
November, 2006
c© Copyright 2006
by
IGUCHI Yoko
Dissertation Committee
Chairperson: GUNJI Takao
Member: Joseph Emonds
Member: Taisuke Nishigauchi
Acknowledgments
A great number of people have contributed to making my life a wonderful experience. Many
people gave me a lot of support and encouragement to complete the dissertation. I am very
fortunate to have been supported by many great people during my study and completing
the dissertation.
First and foremost, I am immensely indebted to Takao Gunji, my first adviser. I wish
to express my special gratitude to him for his generosity in sharing his vast of knowledge,
countless support and his encouragement. He inspires me all the time, and I deeply respect
him not only as a linguist but also as a mentor.
I would like to thank Joseph Emonds, who suggested great deal of things for me every
time, and I really enjoyed talking with him. I am impressed with his wise comments.
I also wish to thank Taisuke Nishigauchi for his constant support and valuable advice.
I have learned a great deal from him, and I really enjoyed his class and comments.
I am also deeply grateful to Edson T. Miyamoto for lots of advice for conducting exper-
iments. The collaboration with him to conduct experiments made a wonderful change in
my life, and made me think that the research is really interesting. I owe him almost all of
the technical expertise that are central to my dissertation.
My special thanks are also due to Michinao Matsui, Kenjiro Matsuda, and Phillipe
Spaelti for so much advice. My knowledge of linguistics and statistics is from them.
The experiments reported in this work could not have been completed without support
of many people. I wish to thank from the heart all the participants in Kobe Shoin Women’s
University.
I am indebted to the graduate students of Kobe Shoin for their really warm support
and encouragement.
My final and the most sincere thanks are due to my friends and my family. I could not
have done this thesis without their generous help and encouragement.
iv
Contents
Acknowledgments iv
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Human sentence processing in Japanese . . . . . . . . . . . . . . . . 1
1.1.2 Incrementality in Japanese . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 The purposes of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Organization of dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Properties of nominative ga 10
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 The properties of nominative ga . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Ga as a single subject . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Ga marks an object . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Ga ga constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Clause boundaries in Japanese processing 20
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Case marking information or “scrambling” . . . . . . . . . . . . . . . . . . . 20
3.3 Nominative case markers as clause boundaries . . . . . . . . . . . . . . . . . 25
3.4 Structural ambiguity effect of nominative case repetition . . . . . . . . . . . 32
3.5 Animacy information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4 Saturation of nouns 39
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Relational nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.3 Unsaturated and saturated nouns . . . . . . . . . . . . . . . . . . . . . . . . 50
v
4.4 The relation between Saturation and argument structure . . . . . . . . . . . 52
4.5 ‘Saturation’ in our experiments . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5 Experiment 1 61
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2.4 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.3.1 Results for the NP-ga NP-ga fragments . . . . . . . . . . . . . . . . 75
5.3.2 Results for the NP-ni NP-ga NP-ga fragments . . . . . . . . . . . . . 76
5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6 Experiment 2 82
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.2.3 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.2.4 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.5 Corpus analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7 Experiment 3 91
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2.3 Procedure and analysis . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
vi
7.3.1 Clausal structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.3.2 Roles of the nominative NP . . . . . . . . . . . . . . . . . . . . . . . 94
7.3.3 Length effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
8 Conclusion 103
8.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
A Fragments used in Experiment 1 105
B Fragments used in Experiment 2 113
C Fragments used in Experiment 3 118
D Completions provided by participants for experiment 1 123
E Completions provided by participants for experiment 2 136
F Completions provided by participants for experiment 3 145
Bibliography 154
vii
Chapter 1 Introduction
1.1 Background
Over the past few decades a considerable number of studies have been conducted on human
language from many fields of study (e.g., theoretical linguistics, psycholinguistics, com-
putational linguistics, traditional linguistics, sociolinguistics, cognitive psychology, among
others). Theoretical linguistics tries to search for a universal grammar, that is, properties
all languages have in common.
Many researchers in the fields of linguistics, psychology, cognitive science and neuro-
science believe that the universal theory could possibly assist in understanding how the
human brain works. Hence much research has been performed on sentence processing in
English and other languages.
A central concern of psycholinguistics is the issue of whether the human sentence pro-
cessing mechanism is in fact universal, which has been assumed either implicitly or explicitly
by virtually all researchers in the field.
Most of the work in Japanese sentence processing has been influenced by the work on
English sentences. However, more recently, a lot of researchers are working on Japanese
language with theoretical and experimental paradigms to examine the various properties of
Japanese.
1.1.1 Human sentence processing in Japanese
There have been many studies to investigate how people understand and produce sentences.
Many models to construct a universal sentence processing mechanism in the human brain
have been discussed for a long time.
A verb comes relatively early in English sentence as shown in (1), so that the verb can
play a crucial role in English sentence processing, and also seems to be a powerful generator
of expectancies about what kinds of phrases will be continued.
(1) I gave a book to Mary
Many studies suggest that people utilize a verb information to process sentences (e.g.,
Trueswell, Tanenhaus, & Garnsey, 1994). The information is incrementally integrated as
a sentence is processed (Marslen-Wilson, 1975; Tanenhaus, Spivey-Knowlton, Eberhard, &
1
2 Chapter 1. Introduction
Sedivy, 1995). Therefore, verbs can affect the sentence processing relatively early in English.
(2b) is a well known example of an English sentence having a garden path effect (Bever
1970: 316), and (3) is an English filler-gap sentence (Frazier et al. 1983: 203).
(2) a. The horse raced that car, and won.
b. The horse raced past the barn fell.
c. The horse (which was) raced past the barn fell.
(3) Everyone liked the woman who the little childi started [PROj to sing those stupid
French songs for [ti] last Christmas].
When people understand sentences, a parser1 processes the words from left to right,
which means the words come up in sequence. For example, for the garden path sentence in
(2b), the parser makes a hypothesis that the arguments up to the word the barn are in the
same clause, henceforth the verb raced is a main verb as in (2a). However when it processes
the verb fell, it realizes that fell is the main verb which should be read as (2c) and the guess
was wrong. This is a garden path effect and the parser is required to reanalyze the syntactic
structure of the sentence (raced is part of a reduced relative clause which was raced past
the barn).
Japanese is a subject-object-verb (SOV), head-final, left-branching language, and has
the productive usage of empty categories, while it has no relative pronouns. Because of
these characteristics, it is well known that the syntactic structure of Japanese sentences are
considerably ambiguous until the end of phrases (Inoue and Fordor 1995, Mazuka and Itoh
1995, Berwick and Fong 1995). For instance, there are many possible constructions that
could follow the three words shown in (4).
(4) Naomi-ga Ken-ni hon-o ...
Naomi-nom Ken-dat book-acc
For example, a simple clause such as (5) is possible.
(5) Naomi-ga Ken-ni hon-o ageta.
Naomi-nom Ken-dat book-acc gave.
‘Naomi gave a book to Ken.’
1as a syntactic processor of a human brain
Section 1.1. Background 3
Also there are other phrases that can follow these words such as those shown in (6). In
(5), the three words are in the same clause in a simple construction, while they are in the
different clauses in (6), such as relative clauses or sentential complement.
(6) a. Naomi-ga [ Ken-ni hon-o ageta ] hito-o mitsuketa.
Naomi-nom Ken-dat book-acc gave person-acc found
‘Naomi found the person who gave a book to Ken.’
b. Naomi-ga Ken-ni hon-o [kita ] hito-ni uraseta.
Naomi-nom Ken-dat book-acc came person-dat to let sell
‘Naomi made Ken sell the books to people who came.’
c. Naomi-ga Ken-ni [hon-o Taroo-ga katta] to itta.
Naomi-nom Ken-dat book-acc Taro-nom bought comp said
‘Naomi said to Ken that Taro bought a book.’
There are many possibilities for phrases that come after them, so that Japanese sentences
are relatively ambiguous until the end of sentences. Another property of Japanese which
makes processing difficult is word order flexibility. In general, the verb has to be at the end
of the clause in Japanese, however the word order is flexible. The sentences in (7) are all
grammatical in Japanese and the meanings are almost the same. The type of sentence such
as (7) has been considered to be derived by a transformation and is called “scrambling”
(Ross, 1967; see also general information about scrambling in Nakayama, 1999; Nemoto,
1999).
(7) a. Naomi-ga Ken-ni hon-o ageta.
Naomi-nom Ken-dat book-acc gave.
b. Naomi-ga hon-o Ken-ni ageta.
Naomi-nom book-acc Ken-dat gave.
c. Ken-ni Naomi-ga hon-o ageta.
Ken-dat Naomi-nom book-acc gave.
d. Hon-o Naomi-ga Ken-ni ageta.
book-acc Naomi-nom Ken-dat gave.
‘Naomi gave a book to Ken.’
It has been shown that the frequency of scrambled sentences such as (7) in Japanese
is low (Kuno, 1973a, Yamashita & Suzuki, 1995). In a generative grammar of a syntactic
4 Chapter 1. Introduction
theory, in the early 1980s, it was assumed that Japanese is a typical example of a non-
configurational language because of the flexible word order (Hale, 1980; Farmer 1980, 1984,
Whitman, 1979; Miyagawa, 1980), so that Japanese sentences were assumed to be derived
from a phrase structure rule like (8) (Farmer, 1980).
(8) x′ −→ x′ ∗ x
In this hypothesis, where Japanese has a flat structure, some of the syntactic structures
of the sentences in (7) are shown as (9).
(9) a.IP
NP
Naomi-ga
NP
Ken-ni
NP
hon-o
V
ageta
b.IP
NP
hon-o
NP
Naomi-ga
NP
Ken-ni
V
ageta
Since the middle 1980’s, the Japanese syntactic structure for a sentence with a canonical
word order has been argued to be as in (10) in the government and binding framework (Hoji,
1985). It was argued that Japanese has VP, which includes object and verb.
(10)IP
NP
NP-ga
VP
NP
NP-o
V
V
In this theory, scrambled word order is derived from their canonical word order by the
operation of scrambling, and the scrambled argument is adjoined to IP as shown in (11)
(c.f., Saito, 1985, among others).
(11)
Section 1.1. Background 5
IP
NP
NP-oi
IP
NP
NP-ga
VP
NP
ti
V
V
Numerous attempts have been made by researchers to analyze the property of scrambling,
but it is not clear yet. In the movement approach (Saito, 1985; Hoji, 1985), the nature of
scrambling shows both the properties of traditional A-movement exemplified by raising or
passive construction in English, which shows the “antireconstruction effect”, and traditional
A′-movement such as wh-question or topicalization in English, which reconstructs movement
in the base position with respect to binding (Bayer and Kornfilt, 1994; Dayal 1994; Deprez
1994; Frank, Lee and Rambow, 1992; Lee, 1993; Lee and Santorini 1994; Mahajan, 1990;
Miyagawa, 1995; Neeleman, 1994; Saito, 1992; Webelhuth, 1992; Wiliams, 1994), There
have been various approaches even among the constraint-based frameworks: the subcat
approach (Gunji, 1987), the slash approach (Gunji, 1987), and the linearization ap-
proach (Yatabe, 1996; Gunji, 1999). In the Optimality-Theory Syntax (OT Syntax: Prince
and Smolensky, 1993), which is free from the complicated movement typology, scrambling in
German is analyzed from the perspective of ranking syntactic constraints (e.g., grammatical
roles) in combination with discourse constraints (e.g., new information or focus) (Hye-Won
Choi, 2001). Although this character in Japanese is a very interesting phenomenon, we will
try not to commit to the theoretical status of scrambling, since it is not a main concern of
the present research.
In the psycholinguistic research, empirical works also have been conducted to investigate
this word order effect in processing of Japanese (e.g., Yamashita, 1997, Mazuka, Itoh &
Kondo, 2002; Miyamoto & Takahashi, 2002).
1.1.2 Incrementality in Japanese
Many computational models have been proposed. A number of studies have discussed the
interaction between syntactic, semantic, and pragmatic information in English sentence
processing. There are many models for sentence processing: some researchers suggested
that only syntactic information is used for making decision in the beginning of parsing
(e.g., Frazier, 1987; Mitchell, 1987, 1989; Mitchell, Corley, & Garnham, 1992; Rayner,
6 Chapter 1. Introduction
Carlson, & Frazier, 1983). Among them prople propose “two-stage serial models” which
use only syntactic information such as grammatical category and syntactic relationships in
the first stage sensitive only to syntactic information; other information is used for parsing
in the second stage (Ferreira & Clifton, 1986; Frazier, 1987; Frazier, et al., 1983).
On the other hand, other researchers suggested that other information affects even in
the beginning of the sentence (e.g., Tanenhaus, Carlson, & Trueswel, 1989; Taraban &
McClelland, 1988, 1990). For instance, “constraint-based probabilistic models”(Altman,
1999, 2002; MacDonald, 1994; McRae et al., 1998; Trueswell, Tanenhause, & Garnsey,
1994), with significant interactions between other information such as lexical and world
knowledge, are sensitive to syntactic, semantic and pragmatic constraints. There are also
some researchers who stand between these two extremes (e.g., Altman, 1988; Altman &
Steedman, 1988; Crain & Steedman, 1985).
In different perspectives, a “delay model” proposes that parsing decisions wait until
more information can be available (e.g., Weinberg 1993, Gorrell 1995). In this model,
misunderstanding a sentence or making a mistake in computing are decreased. However, the
cost of holding words are increased until the argument, which is available to parse, shows up.
“Head-driven models” are also considered as one of the delay models for sentence processing,
and in this model, a grammatical head is an ideal candidate for processing (Abney, 1989;
Pritchett, 1991, 1992).
If we assume that sentences are processed incrementally without delay, the resulting
parsing decisions would have a high probability of being incorrect. This type of model is
called an “incremental model”.
However, these models are originally proposed in English processing. When trying to
adopt these models to Japanese sentence processing, problems arise because of the syntactic
distinction.
There are also models that have been proposed for Japanese sentence processing (e.g.,
Mazuka & Lust, 1990; Inoue, 1991; Mazuka, 1991). If we assume that Japanese sentences
are processed incrementally without delay, the resulting parsing decisions would have a high
probability of being incorrect. However, the cost of holding the words is relatively high,
especialy in Japanese because of the syntactic property of being a head-final language.
A number of studies have shown that sentences are processed incrementally, and native
Japanese speakers start interpreting sentences immediately even before the clause-ending
predicate is read or heard.
There have been numerous data to support this. For instance, questionnaire data (Ue-
hara & Bradley, 2002), reading times (Miyamoto, 2002), event related potentials (ERPs)
Section 1.2. The purposes of the dissertation 7
(Garnsey et al., 2001), reading times (Mazuka, Itoh & Kondo, 2002, among others.).
Kamide, Altmann & Haywood (2003) conduct an eye-tracking experiments, and they find
that a hamburger is more likely to be expected after hearing (12a) than (12b).
(12) a. Ueetoresu-ga kyaku-ni ...
waitress-nom customer-dat
b. Ueetoresu-ga kyaku-o ...
waitress-nom customer-acc
1.2 The purposes of the dissertation
A considerable body of evidence indicates that native Japanese speakers start interpreting
sentences immediately even before the clause-ending predicate is read or heard.
One crucial source of information during this process is case marking. In particular,
because of their special syntactic properties, nominative NPs have been claimed to indicate
the beginning of a new clause (Miyamoto, 2002; Uehara & Bradley, 2002). Recent work
also suggests that animacy is an important factor in determining whether a nominative NP
is interpreted as a subject or an object (Muraoka & Sakamoto, 2004).
However, most previous studies do not explore in detail the factors that affect how
two NPs are interpreted in relation to each other. We report three questionnaire studies
and a corpus count which suggest that the possible relations between NPs are critical
in determining readers’ interpretations. Although properties of each individual NP such
as animacy are important, their interpretation is affected by the relation that may hold
between the NPs (e.g., possessor-possessee, part-whole relations). This is compatible with
the claim that thematic roles (i.e., semantic roles in a sentence) are not determined in
isolation for each NP, but rather they are assigned taking the surrounding context (e.g.,
nearby NPs) into consideration.
In order to present this relation between the nominals, we will also show the analysis of
“saturation”, i.e, relational nouns from the perspective of their argument structures.
1.3 Organization of dissertation
In Chapter 2, we will introduce some properties of the nominative marker ga which are
relevant to our present experiments. The nominative marker -ga in Japanese can mark a
subject, a possessor of a subject, an adjunct and the object of a stative predicate. We will
see the properties of descriptive ga, exhaustive-listing ga and objective ga of the nominative
marker ga.
8 Chapter 1. Introduction
We will look at the ga ga constructions with the subjective ga and the objective ga.
Some of the ga ga constructions can change to the ni ga constructions and the wa ga
constructions. We will see the instances of these constructions in this chapter. We will see
the basic phenomenon of the ga and ga ga constructions.
In Chapter 3, we will begin with showing whether the “Scrambling” effect causes an
increased load on sentence processing or other factors such as repetition of the same marker
may cause it. We will introduce the works of Yamashita (1997) which conduct experiments
on these phenomenons. Secondly, we will see whether the case markers, especially nom-
inative ga markers, are the factor causing increased load of processing. Nominative case
markers are the indicator of a left clause boundary, and the creation of the boundary may
cause a longer reading time, as has been proven by the work of Miyamoto (2002). Third,
in the construction of the nominative case marker repetition, such as NP-ga NP-ga, each
NP-ga is processed in association with a different clause, that is, NP-ga is a single clause
subject. On this topic, we will examine the works of Uehara and Bradely (2002). Fourth,
even in the construction of NP-ga NP-ga, the clause boundary is not inserted when the
second NP-ga is inanimate. We will see the work of Muraoka and Sakamoto (2004) on
animacy.
In Chapter 4, we will discuss how important it is to treat noun phrases carefully in the
sentence processing experiments. Since experiments related to sentence processing some-
times do not make mention of the noun phrase differences in the test items, and both proper
noun phrases and common noun phrases are sometimes mixed in the investigation of the
ambiguity of the sentence structure following the garden path effect or the scrambling effect
(see also discussion in Inoue, 2006).
In these perspectives, we will look into a kind of common noun phrases (saturation)
related to our present experiment. First, in order to classify the common noun phrases,
we will look into the distinction between “saturated nouns” and “unsaturated nouns” from
Nishiyama (1990)’s proposal. Second, there is also a similar type of nouns such as “relational
noun” in the term of Harada (1991), Kageyama (1993), Gunji (1995), and also Kikuchi and
Shirai (2006), among others. Then we will look into our definition of such kinds of nouns
from the perspective of sentence processing: what we call “saturation” in our experiment,
saturated nouns, and unsaturated nouns.
In Chapter 5, we will show our results from a completion task. The results confirm
previous results according to which readers prefer to interpret each nominative NP as part
of a separate clause (Miyamoto, 2002; Uehara, 1997; Uehara & Bradley, 2002).
We will see the interaction between saturation and that animacy, and animacy only has
Section 1.3. Organization of dissertation 9
an effect on clause boundary insertion when the second nominative NP is unsaturated. The
result shows that animacy is modulated by Saturation.
In Chapter 6, the second experiment will be presented. We will discuss how the model
which we suggest affects our results. There are two ways in which case marking and animacy
can be used during the understanding of a sequence of NPs. In one model, each NP is
interpreted exclusively based on its case marker and its individual properties (which we will
refer to as intra-nominal properties; e.g., animacy). Hence, a nominative-marked human
NP would be the prototypical agent, an accusative-marked object would most likely receive
the role of theme, and so forth.
In a second model, although intra-nominal properties are important, they are interpreted
within the context in which the NP occurs. In other words, the interpretation of a NP
depends on its relations to other NPs (inter-nominal relations), so that the same NP (e.g.,
an animate NP) may assume one certain role in a context but a different role in another
context. We will show the second models are in accord with our results.
In Chapter 7, we will present the third experiment. The corpus count suggests that
two nominative NPs in the same clause are rarely seen in newspaper sentences. Whatever
the reason for a difficulty with multiple nominative NPs, we will show that it should not
interact with the findings of the two experiments reported earlier. Nevertheless, a third
questionnaire is conducted by replacing the first nominative marker in the fragments of
Experiment 2 with the topic marker in order to verify if the findings in the previous two
experiments still stand.
Chapter 2 Properties of nominative ga
2.1 Introduction
In this chapter, we will see some properties of the nominative marker ga which are relevant
to our present experiments. The nominative marker -ga in Japanese can mark a subject,
a possessor of a subject, an adjunct, or the object of a stative predicate. Although the
nominative marker ga has many properties, we will see the basic properties of ga which are
relevant to our experiments. That is, the properties of descriptive ga, exhaustive-listing ga
and objective ga.
We can also make the ga ga constructions with the subjective ga and the objective ga.
Some of the ga ga constructions can change to ni ga constructions, wa ga constructions
and no ga constructions. We will see the instances of these constructions in this chapter.
However, we will not try to settle on the theoretical analysis of these constructions. We
will concentrate on setting up the preliminaries to our experiment, as we examine the basic
phenomenon of these constructions.
2.2 The properties of nominative ga
In general, a nominative case marker ga is used to show the single grammatical subject.
Naomi is the subject of sentence (13a), and Ken and Naomi are subjects of the embedded
clause and the main clause, respectively, in sentence (13b).
(13) a. Naomi-ga uttatta.
Naomi-nom sang
‘Naomi sang.’
b. Ken-ga [Naomi-ga uttatta ] to itta.
Ken-nom Naomi-nom sang -comp said
‘Ken said that Naomi sang.’
The particle ga has three different uses (Kuno, 1973). Kuno (1973) calls them descriptive
ga, exhaustive-listing ga and objective ga as shown in (14)1.
1In the following, the romanization of the sentences are changed to the Hepborn method.
10
Section 2.2. The properties of nominative ga 11
(14) a. ga for neutral descriptions of actions or temporary states:
Ame-ga futte imasu.
rain-nom falling is
‘It is raining.’
b. ga for exhaustive listing. “X (and only X). . .” “It is X that. . .”:
John-ga gakusei desu.
John-nom student is‘(Of) all the people under discussion) John (and only John) is a student.’
‘It is John who is a student.’
c. ga for object marking:
Boku-wa Mary-ga suki desu.
I-top Mary-nom fond of am
‘I like Mary.’
(Kuno, 1973: 38)
2.2.1 Ga as a single subject
As we have seen, the prototypical usage of ga is to mark the single subject, and there
are two types of usage, namely, descriptive ga and exhaustive-listing ga as in (14a) and
(14b). The examples of the descriptive ga are shown in (15) below. The descriptive ga in
(14a) is restricted to use only before action verbs, existential verbs and adjectives/nominal
adjectives that represent change of states. Action verbs are used in sentence (15), existential
verbs are used in (16), and the sentences in (17) represent change of states. They are all
instances for the usage of descriptive ga.
(15) a. John-ga asoko-ni tatte iru.
John-nom that-place-loc standing is
‘John is standing there.’
b. Tegami-ga kita.
letter-nom camae
‘Mail has come.’
c. Ame-ga futte iru.
rain-nom falling is
‘It is raining.’
12 Chapter 2. Properties of nominative ga
(16) a. Tsukue-no ue-ni hon-ga aru.
table-gen top-loc book-nom is
‘There is a book on the table.’
b. Oya, asoko-ni John-ga iru.
Oh there-loc John-nom is
‘Oh, John is there.’
(17) a. Sora-ga akai.
sky-nom red
‘The sky is red.’
b. Atama-ga itai.
head-nom aching
‘I have headache.’
(Kuno, 1973: 50)
On the other hand, there are no such restrictions as descriptive ones in the exhaustive-
listing ga (14b).
However, the sentences which represent states such as (18) are a little awkward, since
they need contexts that solicit exhaustive listings such as (19).
(18) a. John-ga gakusei desu.
John-nom student am‘(Of all the people we are talking about) John (and only John) is a student; it is
John who is a student.’
b. Saru-ga ningen no senzo desu.
monkey-nom man ’s ancestor is
‘It is the monkey that is the ancestor of man.’
c. John-ga nihongo-o sitte iru.
John-nom Japanese-acc knows is
‘John (and only John) knows Japanese.’
(19) a. Dare-ga gakusei desu ka ?
who-nom student is
‘Who is a student?’
Section 2.2. The properties of nominative ga 13
b. Nani-ga ningen-no senzo desu ka ?
what-nom man-gen ancestor is
‘What is the ancestor of mankind?’
c. Dare-ga nihongo-o sitte iru ka ?
who-nom Japanese-acc knowing is
‘Who knows Japanese?’
(Kuno, 1973: 51)
If the verbs denote an action, existence, or temporary state in the sentences, ga will be
ambiguous between the neutral descriptive ga and the exhaustive listing. When there are less
permanent states in the sentences, only the meaning of the exhaustive listing interpretation
of ga will be prominent shown in (20).
(20) a. (neutral description)
Sora-ga aoi
sky-nom blue
‘Look! The sky is blue.’
b. (exhaustive listing)
‘It is the sky that is blue.’
(21) a. (neutral description)
John-ga kita
John-nom came
‘John came’
b. (exhaustive listing)
‘It was John who came.’
2.2.2 Ga marks an object
Typically, ga marks the grammatical subject, and o marks the grammatical object in
Japanese as in (22).
(22) Naomi-ga uta-o utatte imasu.
Naomi-nom song-acc sing is.
‘Naomi is singing a song.’
14 Chapter 2. Properties of nominative ga
In certain constructions, ga can appear at the grammatical position of o as in (23).
(23) a. Naomi-wa eigo-ga hanas-e-ru
Naomi-top English-nom speak-can
Naomi can speak English.
b. Naomi-wa konpuutaa-ga hoshii.
Naomi-top computer-nom want
Naomi wants a computer.
c. Naomi-wa Ken-ga suki desu.
Naomi-top Ken-nom like is
Naomi likes Ken.
Kuno (1973) shows that ga is used not only for marking the subject but also for mark-
ing the object of all transitive adjectives and nominal adjectives and of a certain class of
transitive verbs as shown in (24) trough (29).
(24) Competence: joozu ‘good at’, nigate ‘bad at’, heta ‘bad at’, tokui ‘good at, proud of’,
umai ‘good at’.
a. Dare-ga eigo-ga joozu desu ka?
who-nom English-nom good-at is
‘Who is good at English?’
b. Boku-ga nihongo-ga nigate/heta na koto wa minna yoku shitte imasu.
I-nom Japanese-nom bad-at am fact-that everyone well know
‘Everyone knows well that I am bad at Japanese.’
(25) Adjectives and Nominal Adjectives of Feeling: suki ‘fond of’, kirai ‘hateful of’, hoshii
‘want’, kowai ‘be fearful of’
a. John-ga Mary-ga suki/kirai na koto wa yoku sitte imasu.
John-nom Mary-nom fond-of/hateful-of is fact-that well know
‘I know very well that John likes/dislikes Mary.’
b. Boku-ga Mary-ga kowai.
I-nom Marynom am-fearful-of
‘I am afraid of Mary.’
Section 2.2. The properties of nominative ga 15
(26) -Tai Derivatives: yomitai ‘want to read’, tabetai ‘want to eat’, etc. Although ga is
preferred, o can be used for marking objects.
a. Boku-wa eiga-ga (o) mitai.
I-top movie-nom see-want
‘I am anxious to see movies.’
b. Boku-ga osushi-ga (o) tabe-tai koto o nando ittara wakaru
I-nom sushi-nom eat-want fact-that how-often if-I-said understand
no desu ka?
that is
‘How many times is it that I have to tell you that I am anxious to eat sushi?’
(27) Competence: dekiru and re/rare forms
a. Dare-ga nihongo-ga dekiru ka?
who-nom Japanese-nom can
‘Who can (speak) Japanese?’
b. Dare-ga nihongo-ga (o) hanas-e-ru ka?
who-nom Japanese-nom(acc) speak-can
‘Who can speak Japanese?’
(28) Nonintentional Perception: wakaru ‘understand’, kikoeru ‘hear’, mieru ‘see’
a. Anata-wa nihongo-ga wakarimasu ka?
you-top Japanese-nom understand
‘Do you understand Japanese?’
b. John-ga nihongo-ga wakaranai kara, eigo de hanashimashoo.
John-nom Japanese-nom not-understand since English in let’s speak
‘Let’s speak in English because John does not understand Japanese.’
(29) Possession, Need: aru ‘have’, iru ‘need’
a. Anata-ga okane-ga aru koto wa minna ga shitte imasu.
you-nom money-nom have fact-that everyone knowing is
‘Everyone knows that you have money.’
16 Chapter 2. Properties of nominative ga
b. Anata-ga okane-ga nai koto wa minna ga shitte imasu.
you-nom money-nom not-have that everyone knowing is
‘Everyone knows that you don’t have any money.’
c. Watakusi-wa okane-ga iru.
I-top money-nom need
‘I need money.’
(Kuno, 1973: 81)
2.3 Ga ga constructions
In the NP-ga NP-ga constructions, first NP-ga can be changed to NP-ni. Even if NP-ga
NP-ga constructions are changed to NP-ni NP-ga constructions, they share almost the same
meaning as shown in (30), (31) and (32).
(30) a. Tom-ga eigo-ga wakaru
Tom-nom English-nom understand
‘Tom understands English.’
b. Tom-ni eigo-ga wakaru
Tom-dat English-nom understand
‘Tom understands English.’
(31) a. Tom-ga osake-ga nomeru
Tom-nom sake-nom can drink
‘Tom can drink sake.’
b. Tom-ni osake-ga nomeru
Tom-dat sake-nom can drink
‘Tom can drink sake.’
(32) a. Tom-ga okane-ga aru
Tom-nom money-nom have
‘Tom has money.’
b. Tom-ni okane-ga aru
Tom-dat money-nom have
‘Tom has money.’
Section 2.3. Ga ga constructions 17
However, some NP-ga NP-ga constructions can not be changed to NP-ni NP-ga con-
structions as the following examples in (33) and (34) show.
(33) a. Tom-ga Mary-ga sukida.
Tom-nom Mary-nom like
‘Tom likes Mary.’
b. *Tom-ni Mary-ga sukida.
Tom-dat Mary-nom like
‘(lit.) Tom likes Mary.’
(34) a. Tom-ga eiga-ga mitai.
Tom-nom movie-nom want to watch
‘Tom wants to watch movie.’
b. * Tom-ni eiga-ga mitai.
Tom-dat movie-nom want to watch
In the following NP-wa NP-ga constructions, if NP-wa can have both the exhaustive
listing and neutral description interpretations, then NP-wa can be replaced by NP-ga, which
will yield grammatical sentences as in (35) - (36). On the other hand, when the NP-wa NP-
ga sentence has only the exhaustive-listing interpretation, the corresponding NP-ga NP-ga
sentences are ungrammatical (Kuno, 1973).
(35) a. Kono class-wa dansei-ga yoku dekiru.
this class-top male-nom well are-able
‘Speaking of this class, the boys do well.’
(dansei ga: exhaustive listing and neutral description)
b. Kono class-ga dansei-ga yoku dekiru.
this class-nom male-nom well are-able
‘It is this class that the boys do well in.’
(dansei ga: neutral description)
(36) a. Nihon-wa dansei-ga tanmei desu.
Japan-top male-nom short-life-span are
‘Speaking of Japan, men have a short life-span.’
(dansei ga: exhaustive listing and neutral description)
18 Chapter 2. Properties of nominative ga
b. Nihon-ga dansei-ga tanmei desu.
Japan-nom male-nom short-life-span are
‘It is Japan, that men have a short life-span.’
(dansei ga: neutral description)
(Kuno, 1973: 67)
Kuno claims that the sources of the sentence (35b) and (36b) are the following.
(37) a. Kono class-no dansei-ga yoku dekiru.
this class-gen male-gen well are-able
‘This class’s boys (and only they) do well.’
b. Nihon-no dansei-ga tanmei desu.
Japan-nom malenom short-life-span are
‘Men of Japan (and only they) have a short life-span.’
(38) a. John-no otoosan-ga shinda.
John-gen fathernom died
(i) ‘John’s father died.’ (ii) ‘It is John’s father that has died.’
b. John-ga otoosan-ga shinda.
John-nom father-nom died
‘It is John whose father died.’
Similarly,
(39) a. Yama-no ki-ga kirei desu.
mountain-gen trees-nom pretty are(i) ‘It is trees in the mountains that are pretty.’ (ii) ‘(Look!) The trees in the
mountains that are pretty.’
b. Yama-ga ki-ga kirei desu.
mountain-nom treesnom pretty are(i) ‘It is the mountains that trees are pretty in.’ (ii) ‘(Look!) The mountains -their
trees are pretty.’
(40) a. John-no kodomo-ga sensei ni sikarareta.
John-gen child-nom teacher by was-scolded(i) ‘John’s child was scolded by the teacher..’ (ii) ‘It is John’s child that was scolded
by the teacher.’
Section 2.4. Summary 19
b. John-ga kodomo-ga sensei-ni sikarareta.
John-nom childnom teacher-dat was-scolded
‘John-his child was scolded by th teacher.’
(Kuno, 1973: 70)
2.4 Summary
We have looked at some properties of the nominative marker ga which are relevant to the
present experiments. The nominative marker -ga in Japanese marks a subject, a possessor
of a subject, an adjunct, or the object of a stative predicate . We saw the properties
of descriptive ga, exhaustive-listing ga, and objective ga, and also looked at the ga ga
constructions with the subjective ga and the objective ga. We showed how some of the ga
ga constructions can change to ni ga constructions and wa ga constructions.
In our results of the experiments, we find some types of ga examples which were pre-
sented in this chapter. The crucial properties of ga are the objective ga, exhaustive listing
ga, and also when we will classify the sentences, we refer to the observation that the ga ga
construction can be replaced with the ni ga construction, no ga construction, ga o construc-
tion.
Chapter 3 Clause boundaries in Japanese
processing
3.1 Introduction
In this chapter, we will survey what factors cause difficulty in the processing of Japanese
sentences from the recent works.
As stated before a considerable body of evidence indicates that native Japanese speakers
start interpreting sentences immediately even before the clause-ending predicate is read or
heard. One crucial source of information during this process is case marking. In particular,
because of their special syntactic properties, nominative NPs have been claimed to indicate
the beginning of a new clause, although nominative markers have many syntactic properties
(Miyamoto, 2002, Uehara & Bradley, 2002). Recent work also suggests that animacy is an
important factor in determining whether a nominative NP is interpreted as a subject or an
object (Muraoka & Sakamoto, 2004).
We will begin with Yamashita (1978)’s work which examined scrambled sentences, and
some following works summarized as below.
• A “scrambling” effect causes a load on sentence processing, although other factors
may also cause it such as repetition of the same marker, (Yamashita, 1997).
• Case markers especially nominative ga markers are the factor causing load on pro-
cessing. Nominative case markers are the indicator of a left clause boundary, and the
creation of the boundary may cause a longer reading time. (Miyamoto, 2002).
• In the construction of the nominative case marker repetition, such as NP-ga NP-
ga, each NP-ga is processed with a different clause, that it, NP-ga is a single clause
subject. (Uehara and Bradely, 2002).
• Even in the construction of NP-ga NP-ga, the clause boundary is not inserted when
the second NP-ga is an inanimate. (Muraoka and Sakamoto, 2004).
3.2 Case marking information or “scrambling”
There are many studies indicating that verb information has a crucial role for sentence
processing. Basically, the word order of English sentences is Subject-Verb-Object, so that
20
Section 3.2. Case marking information or “scrambling” 21
the verb can affect the sentence comprehension in an early stage during reading sentence
fragments. In contrast to a head-initial language such as English, Japanese is a head-final
language. Because verb information cannot be utilized until the end of phrases is reached in
Japanese sentence processing, how Japanese people understand a sentence is an interesting
topic. Do they wait for the verb information to arrive or make a certain decision before
the verb shows up? If people make a certain decision using some kind of information which
comes before the verb, for instance that coming from noun phrases, what kind of noun
phrase information is available and how is such information used?
In addition to this head-final property, Japanese sentences have another relevant char-
acter: the word order is relatively free (e.g., scrambling). Many researchers have assumed
longer reading time for scrambled sentences than that for the canonical word order sentences.
As we have seen in the previous chapter, scrambled sentences are analyzed as creating a
new IP node and the structure is more complex than the canonical ones if configurational
structures like (41) are assumed. If there is no such difference, the structure may well be
non-configurational.
(41) a.
IP
NP
NP-ga
VP
NP
NP-o
V
V
b.IP
NP
NP-oi
IP
NP
NP-ga
VP
NP
ti
V
V
There is a hypothesis that the word order affects a decision about sentence structure
and that a case marker such as the nominative marker ga or the accusative marker o also
affects the decision in Japanese sentence processing.
22 Chapter 3. Clause boundaries in Japanese processing
Yamashita (1997) employs two on-line, self-paced, moving-window reading task and a
lexical decision task , in order to investigate whether word order and case markers play a
role, and whether the parser is sensitive to the variety of case-marked arguments. She uses
sentences such as (42) for her first experiment.
(42) a. Condition A: (canonical: NP-ga NP-ni NP-o order)
Wakai jimuin-ga mukuchi-na shachoo-ni omoshiroi hon-o ageta.
young secretary-nom quiet president-dat fun bookacc gave
b. Condition B: (NP-ni scrambled: NP-ni NP-ga NP-o order)
Mukuchi-na shachoo-ni wakai jimuin-ga omoshiroi hon-o ageta.
quiet president-dat young secretary-nom fun bookacc gave
c. Condition C: (NP-o scrambled: NP-o NP-ga NP-ni order)
Omoshiroi hon-o wakai jimuin-ga mukuchi-na shachoo-ni ageta.
fun bookacc young secretary-nom quiet president-dat gave
d. Condition D: (NP-ni, NP-o scrambled: NP-o NP-ni NP-ga order)
Omoshiroi hon-o mukuchi-na shachoo-ni wakai jimuin-ga ageta.
fun bookacc quiet president-dat young secretary-nom gave
‘A young secretary gave the quiet company president a fun book.’
(Yamashita, 1978: 69)
The results show that the scrambled Conditions B, C and D were read as quickly as
the canonical order of Condition A at any position. She found no effect of the word order
variation here, because there was no increased difficulty in processing scrambled sentences.
She employs a lexical decision task to examine how the parser treats both scrambled and
canonical order arguments before the verb shows up.
The next experiment measured the response times of the lexical judgment target as
shown in (43). The fragments are all presented automatically, word by word, and as soon
as the last word of the fragments disappeared, the lexical decision target word such as
Target A or Target B in (43) or a non word is presented. Participants press the return
key if the target word is a real word, and if not, presses the tab key. She measured the
recognition response time from the participants at the target word position.
(43) a. Canonical sentence fragments
Section 3.2. Case marking information or “scrambling” 23
kawaii onnanoko-ga wakai sensei-ni oishii ocha-o
cute girl-nom young teacher-dat good tea-acc
Target A: dashita Target B: nonda
served drank
b. Noncanonical, scrambled sentence fragments
wakai sensei-ni kawaii onnanoko-ga oishii ocha-o
young teacher-dat cute girl-nom good tea-acc
Target A: dashita Target B: nonda
served drank
(43a, 43b A)‘The cute girl served the young teacher good tea.’
(43a, 43b B)‘The cute girl (told) the young teacher (that she) drank good tea.’
(Yamashita, 1997: 172)
In the examples such as (43a: A), ditransitive verbs complete the sentence as a simple
clause, and it is a preferred structure. In contrast, (43a: B), there are two transitive verbs so
that the sentences will still be continued after the verb is presented, and it is a non-preferred
structure (e.g., a relative or subordinate clause). Her results are shown below.
Table 1: Mean response times (ms) by word order and verb type, and accuracy (%)of lexical decision answers
Ditransitive verbs Transitive verbs
Canonical sentences 594(99.4) 657(97.4)
Non-canonical sentences 576(98.4) 637(99.0)
The results in Table: 1 show that there is a significant difference of recognition time
between ditransitive verbs and transitive verbs. Ditransitive verbs are recognized faster
than the transitive verbs. However she did not find any difference of the recognition time
between canonical and non-canonical word order. Following the results, she claims that the
parser utilizes the case-marking information regardless of the word order of the case-marked
arguments.
In order to investigate the effects of word order and case markers, she organized another
self-paced moving-window reading experiment. She measured the time of reading first three
words using the following sentences such as (44). Condition A is in a canonical word order
and includes the same two case markers ga in the first three arguments, Condition B is in a
24 Chapter 3. Clause boundaries in Japanese processing
scrambled order and includes different case markers in the first three arguments, Condition
C is also scrambled, is not analogous to any canonical word order, and includes different
case markers in three arguments.
(44) a. Condition A: (canonical order for the subordinate structure
[PP-NP-ga-NP-ni [NP-ga-NP-o-V] V)
Denwa-de hansamuna gakusee-ga sensee-ni [ tsumetai koibito-ga
phone-on handsome student-nom teacher-dat cold girlfriend-nom
nagai tegami-o yabutta-to itta
long letter-acc tore-comp said
b. Condition B: (scrambled subordinate object
[PP-NP-ga-NP-ni [NP-o-NP-ga-V] V)
Denwa-de hansamuna gakusei-ga sensee-ni [ nagai tegami-o tsumetai
phone-on handsome student-nom teacher-dat long letter-acc cold
koibito-ga yabutta-to itta
girlfriend-nom tore-comp said
c. Condition C: (matrix indirect object and subordinate object scrambled
[PP-NP-ni-NP-ga [NP-o-NP-ga-V] V)
Denwa-de sensee-ni hansamuna gakusei-ga [ nagai tegami-o tsumetai
phone-on teacher-dat handsome student-nom long letter-acc cold
koibito-ga yabutta-to itta
girlfriend-nom tore-comp said
“On the phone, a handsome student told the teacher that the cold-hearted girlfriend
had torn the long letter.”
(Yamashiya, 1997: 178)
The result shows that the word only at position six (the first noun in the embedded
clause) took a longer reading time for the participants to process in Condition A. This
is the second occurrence of the NP-ga phrase. Note that position six in the scrambled
B is (NP-o) and C is (NP-o). She did not find any scrambling cost during the sentence
processing. She explains its longer reading time in Condition A because the same case
marker appeared twice within a possible clause. In general, the occurrence of more than
one case marker of the same type signals the existence of another clause (Inoue 1991) (cf.
“double-o constraint” Harada, 1973; Kuroda, 1978). Moreover, multiple nominative case
Section 3.3. Nominative case markers as clause boundaries 25
markers -ga are limited to particular types of predicates such as stative verbs and potential
verbs (Kuno, 1973b).
She proposes that another reason for the delay in Condition A is the inherent property
of the case marker -ga. The case marker ga is said to have special syntactic and pragmatic
characteristics, and every appearance of -ga triggers the creation of a new clause, leaving
the previous arguments in the higher, unclosed clause.
In sum, her results do not show any scrambling cost in Japanese, and suggests that non-
configurational phrase structural representation will capture these results. She suggests that
the variety of case-marked arguments does seem to affect how the parser makes decisions.
Nakayama (1995) also employed the probe recognition experiment using scrambled sen-
tence and unscrambled sentences to investigate the accessibility of the trace. Even though
he did not discuss the processing load of scrambling, his result suggests that scrambling
does not change the grammaticality of the unscrambled sentence, therefore the trace in the
scrambled phrase may be optional and absent.
On the other hand, other studies indicate extra cost for processing a scrambled sen-
tences in other languages such as German (Clahsen and Featherston, 1998), and in Dutch
(Frazier, 1987; Frazier and Flores d’Arcais, 1989; Kaan, 1997). Mazuka et al. (2002) found
scrambling effects based on her eye-movement experiment in Japanese sentence processing.
Miyamoto and Takahashi (2002) also found a processing load in scrambled sentences. Thus,
studies in sentence processing present conflicting results for the effects of scrambling (see
Miyamoto, 2004 for overview).
From Yamashita’s results, we can assume that the repetition of the same case markers,
especially nominative case markers, may cause a difficulty to process and it may be more
difficult than the scrambled sentences which are also assumed to cause a difficultly compared
to canonical sentences in some studies. Therefore we will see empirical works about the
ambiguity or difficulty of the nominative case markers and the repetitions of the same case
markers in the next section.
3.3 Nominative case markers as clause boundaries
Case markers convey some information, and if that information is utilized to interpret the
sentences in Japanese sentence processing, then how are they used?
Recent works suggest that nominative markers are important in determining how many
clauses are expected in a sentence being read (Miyamoto, 2002; Uehara & Bradley, 2002).
The property of head-final relative clauses and embedded clauses in Japanese often make
sentence processing difficult. Complementizers and relative pronouns show up relatively
26 Chapter 3. Clause boundaries in Japanese processing
early in sentences in English as shown in (45).
(45) a. John has an interesting book.
b. Mary said that John has an interesting book.
c. Mary likes John who has an interesting book.
d. Mary likes an interesting book whcih John has.
However, there are no relative pronouns in Japanese as shown in (46). Therefore the
existence of a relative clause usually cannot be detected until reading or listening the head
noun of the relative clauses. Complementizers in Japanese also show up at the end of
the embedded clause, so that it is ambiguous in the beginning of the sentence whether a
new clause should be opened or not. These characters of Japanese sentences give rise to
processing difficulty (e.g., Inoue, 1991; Mazuka & Itoh, 1995).
(46) a. Ken-ga omoshiroi hon-o motteiru.
Ken-nom interesting book-acc has
‘Ken has an interesting book.’
b. Naomi-wa Ken-ga omoshiroi hon-o motteiru to itta.
Naomi-top Ken-nom interesting book-acc has comp said
‘Naomi said that Ken has an interesting book.’
c. Naomi-wa omoshiroi hon-o motteiru Ken-ga sukida.
Naomi-top interesting book-acc has Ken-nom likes
‘Naomi likes Ken who has an interesting book.’
d. Naomi-wa Ken-ga motteiru omoshiroi hon-ga sukida.
Naomi-top Ken-nom has interesting book-nom likes
‘Naomi likes Ken who has an interesting book.’
Miyamoto (2002) argues that when people read the first two NPs as in the following
fragment in (47), they are completed with a verb such as mita ‘saw’ which is a single clause
interpretation.
(47) Obasan-ga toshiyori-o ← gakusei-ga
woman-nom old-man-acc student-nom
(Miyamoto, 2002: 322)
Section 3.3. Nominative case markers as clause boundaries 27
However when a following NP gakusei-ga ‘student-nom’ is detected, people decide that
it is a beginning of a new clause because there is no verb that takes these three NPs as
arguments. Therefore the third NP student-nom is an indicator of a clause boundary.
He proposes a strategy to insert a clause boundary as follows.
(48) Local Assignment of Clause boundaries (LACB): Assign the left boundary of a new
clause at the point where it is first clear that this new clause is necessary for the
interpretation of the sentence.
(Miyamoto, 2002: 322)
Following this strategy, the clause boundary is assigned between old-man-acc and
student-nom in (47), because student-nom is the first word that makes clear that new
clause is necessary to interpret the sentence.
A single-clause interpretation such as (49a) is initially favored during sentence process-
ing, and the place where the noun girl-dat in (49b) appears is the clear point that saw
is a part of relative clause. (49b) is called subject reanalysis (SR for short: Hirose & A.
Inoue, 1998), because only the subject woman has to be reanalyzed from the single clause
interpretation to a relative clause (Mazuka & Itoh, 1995).
(49) a. Obasan-ga toshiyori-o koosaten-de mita.
woman-nom old-man-acc intersection-loc saw
”The woman saw the old-man at the intersection.”
b. SR (subject reanalysis)
Obasan-ga [RC toshiyori-o koosaten-de mita ]
woman-nom old-man-acc intersection-loc saw
onnanoko-ni koe-o-kaketa.
girl-dat called
“The woman called the girl who saw old-man at the intersection”
(Miyamoto, 2002: 323)
(50) is called subject and object reanalysis (SOR for short: Hirose & A. Inoue, 1998),
because subject woman and object taxi has to be reanalyzed. The reanalysis at girl is easier
than the one at the word taxi because more NPs have to be displaced in the SOR sentences
(Mazuka & Itoh, 1995).
28 Chapter 3. Clause boundaries in Japanese processing
(50) SOR (subject and object reanalysis)
Obasan-ga toshiyori-o [RC koosaten-de mita ]
woman-nom old-man-acc intersection-loc saw
takushii-ni noseta.
taxi-dat put
“The woman put the old-man in the taxi that she saw at the intersection.”
(Miyamoto, 2002: 323)
The purpose of the study is to show that slowdown at the RC head (e.g., taxi) might
be avoided if the information of a clause boundary following the LACB is provided in the
early time of the processing, and also to show that the reanalysis at the RC head is harder
for SOR than for SR sentences. The reading time should slow down at taxi in the following
SOR sentence (51b), because this point is the first place which indicates the existence of
a relative clause for the single clause interpretation. On the other hand, in (51a), a clause
boundary must be inserted at the position student because the LACB indicates it. Therefore,
it should not be accompanied with a slowdown at the RC head position taxi because the
relative clause is already there and the sentence can have a more natural interpretation
than (51b).
(51) a. Canonical control for the SOR
Obasan-ga toshiyori-o [RC gakusei-ga koosaten-de mita ]
woman-nom old-man-acc student-nom intersection-loc saw
takushii-ni noseta.
taxi-dat put
“The woman put the old-man in the taxi that the student saw at the intersection.”
b. SOR
Obasan-ga toshiyori-o [RC koosaten-de mita ]
woman-nom old-man-acc intersection-loc saw
takushii-ni noseta.
taxi-dat put
“The woman put the old-man in the taxi that the student saw at the intersection.”
(Miyamoto, 2002: 324)
For another problem for (51a), Miyamoto prepares one more control sentence for SOR
such as (52). Following the LACB, the clause boundary is inserted between old man and
Section 3.3. Nominative case markers as clause boundaries 29
student in (51a), and it is the correct position for (51a). However, old man might be
reanalyzed at the embedded verb position saw to satisfy the argument structure of the verb,
namely it might be interpreted as toshiyori-o gakusei-ga kousaten-de mita ‘the student saw
an old man at the intersection’.
(52) Scrambling control for the SOR
Toshiyori-o obasan-ga [RC gakusei-ga koosaten-de mita ]
old-man-acc woman-nom student-nom intersection-loc saw
takushii-ni noseta.
taxi-dat put
“The woman put the old-man in the taxi that she saw at the intersection.”
(Miyamoto, 2002: 325)
Miyamoto (2002) conducted a phrase-by-phrase self-paced noncumulative moving-window
reading task (Just et al., 1993) using the following sentences as in (53) and (54).
(53) a. SR (subject reanalysis)
Obasan-ga [RC yoboyobo-no toshiyori-o guuzen-ni koosaten-de mita ]
woman-nom feeble old-man-acc by chance intersection-loc saw
onnanoko-ni isoide koe-o-kaketa.
girl-dat hurry called
“The woman called in a hurry the girl who saw the feeble old-man at the intersection
by chance.”
b. Control for the SR
Obasan-ga [RC yoboyobo-no toshiyori-ga guuzen-ni koosaten-de mita ]
woman-nom feeble old-man-nom by chance intersection-loc saw
onnanoko-ni isoide koe-o-kaketa.
girl-dat hurry called
“The woman called in a hurry the girl who the feeble old-man saw at the intersection
by chance.”
A difference between (53a) and (53b) is the position of the relative clause gap. (53a)
has an object gap in the relative clause and (53b) has a subject gap in the relative clause.
In his results, it took significantly longer reading time with a gap at the third position
toshiyori-ga in (53b) than at the third position toshiyori-o in (53a). However there is no
30 Chapter 3. Clause boundaries in Japanese processing
difference at the RC head position onnanoko-ni. Though a subject gap is easier to process
than the object gap in English (Kin & Just, 1991), it is not clear that Japanese should be
the same. For one thing, the distance between the RC head and object gap is shorter than
the for a subject gap. Therefore, he states that it should be refined in future work.
(54) a. SOR (subject and object reanalysis)
Obasan-ga yoboyobo-no toshiyori-o [RC guuzen-ni koosaten-de mita ]
woman-nom feeble old-man-acc by chance intersection-loc saw
takushii-ni isoide noseta.
taxi-dat hurry put
“The woman put in a hurry the feeble old-man in the taxi that she saw at
the intersection by chance.”
b. Canonical control for the SOR
Obasan-ga yoboyobo-no toshiyori-o [RC gakusei-ga koosaten-de mita ]
woman-nom feeble old-man-acc student-nom intersection-loc saw
takushii-ni isoide noseta.
taxi-dat hurry put
“The woman put in a hurry the feeble old-man in the taxi that the student saw at
the intersection by chance.”
c. Scrambled control for the SOR
Yoboyobo-no toshiyori-o obasan-ga [RC gakusei-ga koosaten-de mita ]
feeble old-man-acc woman-nom student-nom intersection-loc saw
takushii-ni isoide noseta.
taxi-dat hurry put
“The woman put in a hurry the feeble old-man in the taxi that the student saw at
the intersection by chance.”
(Miyamoto, 2002: 328)
In his results, the two controls for the SOR are slower than any other conditions at the
fourth position gakusei-ga in (54b) and (54c) because of the insertion of a clause bound-
ary. Moreover the canonical control for the SOR is slower than its scrambled counterpart,
because there is no competition about clause boundary insertion in (54c).
Comparing the sentences control for the SR and the scrambling control for the SOR,
repeated as in (55), (55a) was significantly slower than (55b) at the third word position
which were both nominative NPs.
Section 3.3. Nominative case markers as clause boundaries 31
(55) a. Control for the SR: Woman-nom [RC feeble old-man-nom
b. Scrambled control for the SOR: Feeble old-man-acc woman-nom [RC
(Miyamoto, 2002: 332)
(55b) does not have to involve the clause boundary, while (55a) has two alternative
interpretations: one with old-man as the subject of an embedded clause and the other with
old-man as the object of an upcoming stative verb such as like. Therefore, Miyamoto argued
that the competition due to a clause boundary insertion caused the slowdown rather than
the property of nominative markers.
The reading time at the fourth word position, the canonical control for the SOR, took a
longer reading time than the scrambled control for the SOR (56), repeated here below. This
suggests that the slowdown result which Yamashita (1997) obtained for (44) is more likely
to be due to a left clause boundary insertion than some intrinsic property of the nominative
markers or adjacency of the two nominative NPs.
(56) Controls for the SOR
a. Canonical: Woman-nom feeble old-man-acc [RC student-nom
b. Scrambled: Feeble old-man-acc woman-nom [RC student-nom
(Miyamoto, 2002: 332)
There is only one possible position for the clause boundary in (56b), which is between
the word woman-nom and student-nom. However there are two possible clause boundary
insertion positions in (56a), one is the right position which is just before student-nom and
the other is between woman-nom and feeble. Miyamoto argued that the LACB favors the
clause boundary before student, but this competition may cause a slowdown at this position.
Therefore he suggests that this slowdown may reflect the competition between alternative
interpretations rather than complexity of the syntactic structure.
The result of the reading time at the fourth position is in accord with the LACB.
However, a significant difference at the seventh position (RC head) does not follow from it
because a RC head should be read quickly when the clause boundary is inserted before it.
He explains this phenomenon in terms of the difficulty of processing multiple nominative
constructions. A sentence becomes easier to process when the nominative marker ga is
replaced by the topic marker wa (Uehara, 1997).
Miyamoto conducted another experiment using the same sentences replacing the first
nominative marker with the topic marker. The results show that the topic marker is easier
32 Chapter 3. Clause boundaries in Japanese processing
to process, and the expected reading time at the seventh position is shown in the experiment,
which supports the LACB.
In the next section, we will see how the nominative case markers are analyzed if more
than one nominative case marker shows up in the beginning of the sentences.
3.4 Structural ambiguity effect of nominative case repetition
Japanese nominative ga has multiple syntactic properties, and we will see these properties
in more detail in Chapter 4. Here, I will show some experimental works first.
English is a head-initial and complementizer-initial language, so it is not so difficult
to process a sentence including an embedded clause such as (57b). However, Japanese is
head-final and complementizer-final, so there will be ambiguity in processing when the two
nominative NPs show up as the first two nouns in (58b).
(57) a. John saw a cat in the street.
b. Mary said that John saw a cat in the street.
(58) a. Ken-ga dooro-de neko-o mita.
Ken-nom street-loc cat-acc saw.
‘Ken saw a cat in the street.’
b. Naomi-ga Ken-ga dooro-de neko-o mita to itta.
Naomi-nom Ken-nom street-loc cat-acc saw comp said.
‘Naomi said that Ken saw a cat in the street.’
Uehara and Bradley (2002) claims that the parser takes NP-ga as a single subject, and
even though ga has various syntactic properties, it is not ambiguous for the parser to make
a structure. In their work, they employed the sentence completion task (Forster, 1968), to
explore the ranking among competing structural analyses of NP-ga repetition.
Ga has various properties, and so we can imagine the possible construction of ga repeti-
tion sentences. For instance, if people read the sequence of NPga as shown in (59), it could
be continued using such words as (60).
(59) Taroo-ga tomodachi-ga otooto-ga
Taroo-nom friend-nom younger brother-nom
Section 3.4. Structural ambiguity effect of nominative case repetition 33
(60) a. ... naiteiru ] to oshieta ] to itta ]
crying-is that taught that said
‘Taro said that his friend taught that his younger brother was crying.’
b. ... kirai-da ] to omotte-iru ]
dislike that thinking-is
‘Taro thinks that his friend does not like his younger brother.’
c. ... monosugoku ijiwaru-da ]
extraordinarily mean-is
‘(lit.) Taro’s friend’s younger brother is very mean.’
(Uehara & Bradley, 2002: 270)
Uehara and Bradley show that each ga in an repetition sequence is four-ways ambiguous
as shown in (61) – (64), and there are logically sixty four ways for the parser to make a
structural hypothesis.
Marking the single subject of a clause in the Single Subject Construction (SSC in short)
is the most prototypical use as shown in (61). Marking two or more parallel subjects of a
clause in the Parallel Subject Construction (PSC in short) is a special use as shown in (62).
Marking the object of a stative predicate in the Nominative Object Construction (NOC
in short) is another special use as in (63). Marking the external possessor of a ga marked
subject NP in the External Possession Construction (EPC in short) is yet another special
use as shown in (64).
(61) Single Subject Construction (SSC)
[ Taro-ga [ tomodachi-ga okotta ] to itta ]
Taro-nom friend-nom got mad that said
‘Taro said taht his friend got mad.’
(62) Parallel Subject Construction (PSC)
[ Taro-ga tomodachi-ga sorezore utatta ]
Taro-nom friend-nom each sang
‘(lit.) Taro and his friend each sang.’
(63) Nominative Object Construction (NOC)
[ Taro-ga tomodachi-ga daisuki-da ]
Taro-nom friend-nom much fond-is
‘(lit.) Taro likes his friend very much.’
34 Chapter 3. Clause boundaries in Japanese processing
(64) a. [ Taro-ga tomodachi-ga totemo shinsetsu-da ]
Taro-nom friend-nom very kind-is
‘(lit.) Taro’s friend is very kind.’
b. [ Taro-no tomodachi-ga totemo shinsetsu-da ]
Taro-gen friend-nom very kind-is
‘(lit.) Taro’s friend is very kind.’
(Uehara & Bradley, 2002: 268)
They conduct two sentence completion tasks, to determine the construction of preferred
structural hypotheses. Participants are asked to complete the sentence from the given
fragments with pencil. The experimental fragments are different in the first questionnaire
and the second questionnaire. Questionnaire 1 asked all possible orderings of { NP-ga NP-ni
NP-o}, and all distinguishable orderings of {{NP-ga NP-ga} plus one of {NP-ni NP-o}},
and all distinguishable orderings of {NP-ga NP-ga NP-ni NP-o } as shown in (Table: 2).
In this experiment, all human nouns, such as role terms, kinship terms, and proper nouns,
are prepared for their fragments in order to rule out the External Possession Construction.
Therefore the completion will be SSC, PSC or NOC.
Table 2: Order of case-markers in sentence fragments
All possible orderings of ga ni o o ga ni ni ga o{ NP-ga, NP-ni, NP-o} ga o ni o ni ga ni o ga
all distinguishable ga ga o ga ga niorderings of ga o ga ga ni ga{{ NP-ga NP-ga} plus one of o ga ga ni ga ga{ NP-ni NP-o}}
all distinguishable ga ga ni o o ga ga ni ni ga ga oorderings of ga ga o ni o ga ni ga ni ga o ga{ NP-ga NP-ga NP-ni NP-o } ga ni ga o o ni ga ga ni o ga ga
ga o ga ni ga ni o ga ga o ni ga
(Uehara & Bradley, 2002; 272)
Their results are that SSC is the most preferred structure, and almost all sentence
completions are the type of SSC except for 11% which are PSC or NOC. In the competition
between PSC and NOC in this 11%, all sentences are PSC sentences, even though NOC has
Section 3.4. Structural ambiguity effect of nominative case repetition 35
a higher frequency than the PSC (some fragments rule out NOC interpretation). Following
this result, they suggest that the nominative case-marker ga is not ambiguous in building
structure, and that it is taken as a subject marker for parsing.
In the second questionnaire, fragments contain both a human noun and inanimate noun
in order to avoid the PSC interpretations, namely they contain inanimate nouns in the right
hand pattern in Table: 3. Therefore the competing pattern will be SSC, NOC and EPC
here. The fragments pattern is shown in Table: 3.
Table 3: Ordering of case markers in sentence fragments
H-ga H-ga H-ga I-gaH-ga H-ga H-o H-ga I-ga I-oH-ga H-ga H-ni H-ga I-ga H-niH-ga H-ga H-ni H-o H-ga I-ga H-ni H-o
H=human noun, I=inanimate noun
(Uehara & Bradley, 2002; 272)
Table 4: Completion response frequency, as a function of structural interpretation,Database 2
SSC EPC EPC/PSC PSC Other
H-ga H-ga 17 0 2 4 0H-ga H-ga H-o 21 0 1 1 0H-ga H-ga H-ni 20 0 1 2 0H-ga H-ga H-ni H-o 18 0 2 3 0
H-ga I-ga 22 0 0 0 1H-ga I-ga I-o 20 0 2 0 0H-ga I-ga H-ni 23 0 0 0 0H-ga I-ga H-ni H-o 20 1 0 0 1
(Uehara & Bradley, 2002; 274)
Their results are that SSC is the most preferred structure with 88% of the responses,
and there are only one EPC completion and ten completions for PSC (Table: 4). PSC was
blocked in the fragments of the right hand in Table: 3. However this interpretation can be
possible in the left hand column in Table: 3.
36 Chapter 3. Clause boundaries in Japanese processing
In sum, the strong preference is to treat NP-ga as a subject following the pattern of
SSC. They conclude that ga is not ambiguous for the parser’s structure building, and it is
treated as a subject.
We have seen in this section that case markers, especially nominative case markers, play
a crucial role for sentence parsing. It is assumed that individual NP properties such as
animacy information also affect human sentence processing. We will see how lexical items
involving animacy information affect the processing in the next section.
3.5 Animacy information
Previous work indicates that other properties associated with individual NPs, such as an-
imacy, also need to be considered in the strategy of LACB (Muraoka & Sakamoto, 2004).
They claim that a parser refers to tentative theta-role information in addition to the case
marking information to create the clause boundaries. They conducted a self-paced reading
experiment using the following sentences as in (65).
(65) a. Rooka-de shunin-ga kodomo-ga nigate-da-to buchoo-ga
corridor-loc manager-nom children-nom hard to deal with - that head-nom
Aihara-san-tachi-ni hanashite-ita
Aihara-title-and others-dat was telling
‘In the corridor, the head was talking to Aihara and others that it was hard for
the manager to deal with children. ’
b. Rooka-de shunin-ga keisan-ga nigate-da-to buchoo-ga
corridor-loc manager-nom calculation-nom hard to deal with - that head-nom
Aihara-san-tachi-ni hanashite-ita
Aihara-title-and others-dat was telling
‘In the corridor, the head was talking to Aihara and others that it was hard for
the manager to deal with calculation. ’
c. Aihara-san-tachi-ni shunin-ga kodomo-ga nigate-da-to
Aihara-title-and others-dat manager-nom children-nom hard to deal with - that
buchoo-ga Rooka-de hanashite-ita
head-nom corridor-loc was telling
‘The head was talking to Aihara and others in the corridor that it was hard for
the manager to deal with calculation.’
Section 3.5. Animacy information 37
d. Aihara-san-tachi-ni shunin-ga keisan-ga nigate-da-to
Aihara-title-and others-dat manager-nom calculation-nom hard to deal with - that
buchoo-ga Rooka-de hanashite-ita
head-nom corridor-loc was telling
‘The head was talking to Aihara and others in the corridor that it was hard for
the manager to deal with calculation.’
(Muraoka & Sakamoto, 2004)
They evaluated three hypotheses: insertion of a left clause boundary based on case
marking information, tentative theta-role information, and use of both case marking and
tentative theta-role information. If a parser uses only case marking information to insert
the left clause boundary, clause boundary is inserted when the NP+/−a ga is encountered,
and it is reanalyzed when the Pred-to appears in (66c) and (66d), because there is no verb
which takes NP-ni NP-ga NP-ga as arguments. On the other hand, in (66a) and (66b),
since clause boundary is not reanalyzed, the reading time on the word Pred-to is expected
to show the pattern exhibited in (67a).
(66) a. Adv / NP-ga / NP[+a]-ga / Pred-to / NP-ga / NP-ni / V
b. Adv / NP-ga / NP[−a]-ga / Pred-to / NP-ga / NP-ni / V
c. NP-ni / NP-ga / NP[+a]-ga / Pred-to / NP-ga / Adv / V
d. NP-ni / NP-ga / NP[−a]-ga / Pred-to / NP-ga / Adv / V
(67) a. (66c) = (66d) > (66a) = (66b)
b. (66a) = (66c) > (66b) = (66d)
c. (66c) > (66a) = (66b) = (66d)
If the parser creates a clause boundary based on only tentative theta-role information, it is
inserted when the NP+/−a ga is input and reanalyzed when the Pred-to is input in (66a)
and (66c), because they suppose that NP-gais assigned the Actor role and a clause does
not include more than one word which is assigned the Actor role. Following this strategy,
in (66b) and (66d), there is no reanalysis on the word Pred-to. Then the reading time on
the word Pred-to is expected as (67b). If the parser uses information of both case marking
and tentative theta-role, the clause boundary is inserted when the word NP+/−a ga is read
38 Chapter 3. Clause boundaries in Japanese processing
in (66c), and the parser reanalyzes it when the Pred-to is read. On the other hand, since
there is no reanalysis in (66a), (66b) and (66d), the reading time is expected to pattern as
(67c).
Their results confirm the third hypothesis that the parser utilizes both case marking and
animacy information. Therefore, they claim that the strategy of the left clause boundary
insertion needs not only case marking information but also animacy information.
3.6 Summary
In this chapter, we saw what factors cause difficulty in processing Japanese sentences by
looking at some recent works that are related with the present work.
From the perspective of incrementality of Japanese sentence processing, we began with
Yamashita (1978)’s work which examined scrambled sentences. It is suggested that one
crucial source of information during this process is case marking. In particular, because of
their special syntactic properties, nominative NPs have been claimed to indicate the begin-
ning of a new clause, although nominative markers have a variety of syntactic properties
(Miyamoto, 2002, Uehara & Bradley, 2002). We saw the work of Uehara and Bradley (2002)
suggesting that each NP-ga is processed with different clauses in the construction of the
nominative case marker repetition, such as NP-ga NP-ga, that is, each NP-ga is a single
clause subject.
It is also suggested that animacy is an important factor in determining whether a nom-
inative NP is interpreted as a subject or an object (Muraoka & Sakamoto, 2004).
Chapter 4 Saturation of nouns
4.1 Introduction
In the following sentences, there are sequences of proper noun phrases with a case marker
in (68a) and (68b), while there is a common noun phrase in (68c). Although they are all
sequences of noun phrases, they convey different information.
(68) a. Tanaka-ga Yamamoto-o
Tanaka-nom Yamamoto-acc
b. Ken-ga Naomi-o
Ken(male)-nom Naomi(female)-acc
c. Isha-ga kanja-o
doctor-nom patient-acc
- kanbyooshita.
took care of
‘NP1 took care of NP2.’
Proper nouns such as Tanaka without any context only present a name as an individual
person who is named Tanaka. Furthermore, even if another proper noun such as Yamamoto
follows it, we can just expect only the person named Tanaka does something for Yamamoto
in (68a) before a verb follows it. In the example of (68b), there are also proper noun phrases
and also they are person’s names. They contain slightly more information (gender) than
those in sentence (68a). On the other hand, common nouns such as doctor can imply more
information than proper nouns, therefore we can make some kind of guess about which
verbs are to come, such as kanbyoosuru ‘take care of ’ from the sequence of the common
nouns such as (68c) doctor-nom patient-acc.
Inoue and Den (1999) show that verb predictability1 from the proper noun phrases is
low by dint of a sentence completion task. The verb predictability of both the subject
proper nouns and the object proper nouns is also low, and the processing load for proper
nouns is high.
1the possibility of prediction of a kind of verb from the noun phrases (Inoue and Den, 1999)
39
40 Chapter 4. Saturation of nouns
It is usually assumed that an ambiguity in the relative clause sentences such as (69)
causes a garden path effect in Japanese. If the word string of Tanaka-ga Yamamoto-o
tasuketa is wrongly considered as a single clause, the garden path effect is caused at the
place Suzuki-o because it should be reanalyzed as a relative clause.
(69) Tanaka-ga Yamamoto-o tasuketa Suzuki-o hometa.
Tanaka-nom Yamamoto-acc helped Suzuki-acc praised
‘Tanaka praised Suzuki who helped with Yamamoto.’
The following sentence has the same structure as (69); it is a relative clause sentence,
where the proper noun phrases in (69) are changed to common noun phrases. Sentences
such as (70) has a smaller garden path effect at the position of boy-acc (Den and Inoue,
1997).
(70) Fukoo-ga hahaoya-o sagashita shoonen-o osotta.
bad luck-nom mother-acc searched for boy-acc befell
‘Bad luck befell a boy who searched for (his) mother.’
In the test sentences of the sentence processing experiments, it is sometimes claimed
that the animacy of the nouns affects parsing because of their thematic roles. For instance,
Trueswell et al. (1994) conduct an eye-movement experiment, and examine the effect of
animacy with structurally ambiguous sentences such as (71).
(71) a. The defendant examined by the lawyer turned out to be unreliable.
b. The defendant that was examined by the lawyer turned out to be unreliable.
c. The evidence examined by the lawyer turned out to be unreliable.
d. The evidence that was examined by the lawyer turned out to be unreliable.
(Trueswell et al., 1994: 318)
They found that sentences with animate subjects (defendant) are harder to interpret
than inanimate ones (evidence). The defendant examined is thought to be given a main
clause analysis in which the defendant is subject and typically agent of examined. On the
other hand, in (71c) The evidence is inanimate and is given a theme role by the verb.
Gunji (1995) suggests that nouns such as “relational nouns” should be used carefully in
the example sentences. The “relational nouns” refer to nouns which should have a relation
Section 4.1. Introduction 41
with other nouns such as brother or head (Harada, 1991). We will see this type of nouns
later.
The suggestion of Gunji (1995) is that the interpretation of “exhaustive listing” is easy
to obtain in sentence (72b); on the other hand, it is hard to obtain such an interpretation
for sentence (72c) if the relation between Ken and Naomi is absent.
(72) a. Ken-ga taido-ga warui.
Ken-nom behavior-nom bad
Ken’s behavior is bad.’
b. Ken-ga otooto-ga taido-ga warui.
Ken-nom brother-nom behavior-nom bad
‘Behavior of Ken’s brother is bad.’
c. Ken-ga Naomi-ga taido-ga warui.
Ken-nom Naomi-nom behavior-nom bad
(lit.) ‘As for Ken, behavior of Naomi is bad.’
(Gunji, 1995; 158)
However, experiments related with the sentence processing often do not make mention
of differences in the noun phrases in the test items, including the fact that proper noun
phrases and common noun phrases are mixed. They usually try to figure out the ambiguity
of the sentence structure following the garden path effect or the scrambling effect (see also
related discussion in Inoue, 2006).
The genitive constructions such as Mary’s teacher, and the team of Mary are challenging
constructions for argument or modifier distinctions in noun phrases. It is assumed that a
genitive phrase is always an argument of some relation, but the relation does not always
come from the head noun. That is, some genitives are arguments and some are modifiers
(Partee, 1983, 1997; Barker, 1995).
For instance, in English, there are some words preceded by genitive-marked NPs (Tom’s)
or possessive pronouns (e.g., his, her). For instance, kinship terms such as mother, or body
parts such as hands need a possessor as in (73).
(73) a. Tom’s mother / father
b. his hands / head
42 Chapter 4. Saturation of nouns
In the next section, we will assume that some nouns in Japanese share the same property
with English noun phrases as discussed above. That is, Japanese counterparts to nouns such
as mother and father, and body parts like hands and head also need a possessor pronoun,
even if it is implicit.
Pronouns such as she or he are commonly expressed in English sentences. However,
these overt pronouns are rarely used in Japanese sentences; instead, what are referred to
as “zero pronouns” are used in Japanese. Moreover, while the possessor may be indicated
with a genitive marked NP, possessive pronouns are also often left implicit and indicated
with zero pronouns.
In this chapter, we will look into some kinds of common noun phrases related to our
present experiment. First we will examine the type of nouns which is called “relational
noun” in the terms of Harada (1991), Gunji (1995), and also Kikuchi and Shirai (2006),
and also similar types of nouns such as “houwa meishi” ‘saturated noun’ using Nishiyama’s
(1990) term. Then we will look into our definition of such kinds of nouns from the perspec-
tive of sentence processing: what we call “saturation” in our experiment; saturated nouns
and unsaturated nouns.
• Relational noun (Harada, 1991; Gunji, 1995; Kikuchi and Shirai, 2006)
• Houwa ‘Saturated’/ Hihouwa ‘Unsaturatead’ (Nishiyama, 1990)
• Saturated / Unsaturated (present work)
In order to investigate the relation between noun phrases, first, we will see some examples
of relational nouns in the NP1 no NP2 construction in Harada (1991) and also Kikuchi and
Shirai (2006).
Next, we will also see the distinction between “saturated nouns” and “unsaturated
nouns” from Nishiyama (1990)’s proposal.
Then we will see the relation between saturation and argument structure from Gunji
(2002)’s analysis. That suggests that some nouns have their own arguments in the same
way as verbs do.
Finally, we will define how we treat saturation and relational nouns in our experiments
and also we will show some examples of our items in the experiments.
4.2 Relational nouns
The nouns which require some kind of relationship with another nouns are sometimes called
“relational nouns” (Harada, 1991; Gunji, 1995; Gentner, 2005).
Section 4.2. Relational nouns 43
Harada (1991) observes how the relational nouns are interpreted in the construction of
NP1 no NP2. He claims that noun phrases have to be syntactically modified by the preceding
words if the relations between NP1 and NP2 are the <whole-part> or <attribute>.
We will examine a classification which Harada shows for the construction NP1 no NP2.
The first examples are the types of <person> and <his/her part> as shown in (74).
(74) a. nagai kami-no josei
long hair-gen lady
‘a lady with long hair’
b. * kami-no josei
hair-gen lady
‘a lady with hair’
c. suraritoshita ashi-no onnanoko
slender legs-gen girl
‘a girl with slender legs’
d. * ashi-no onnanoko
legs-gen girl
‘a girl with legs’
e. futoi ude-no dansei
thick arm-gen man
‘a man with a thick arm’
f. * ude-no dansei
arm-gen man
‘a man with a arm’
(Harada, 1991: 8)
These are grammatical when the hair has its modifier such as long as in (74a); on the
other hand, it is ungrammatical if the noun hair does not have its modifier as in (74b). The
other examples seem to follow the same pattern.
This restricttion behaves in the same way in English as in the following examples (75).
(75) a. blond-haired lady
b. *haired lady
44 Chapter 4. Saturation of nouns
c. long-legged boy
d. *legged boy
e. bug-eyed monster
f. *eyed-monster
(Harada, 1991: 8)
The grammaticality of the examples in (76) also exhibits the same restriction as the
previous examples. They are of the types of <person> and <his/her attributes>.
(76) a. sugoi taijuu-no josee
immense wieght-gen woman
‘woman with immense weight’
b. * taijuu-no josee
wieght-gen woman
‘woman with weight’
c. iyana seikaku-no hito
unpleasant character person
‘person who has unpleasant character’
d. * seikaku-no hito
character person
‘person who has character’
e. surudoi kankaku-no hyooronka
keen sense commentator
‘commentator who has keen sense’
f. * kankaku-no hyooronka
sense commentator
‘commentator who has sense’
(Harada, 1991: 9)
The examples shown above seem to have the same restriction as the previous examples;
the grammaticality changes depending on whether they have their modifiers or not.
Section 4.2. Relational nouns 45
Harada (1991) claims that, however, when they are of the types of <wearing things>
or <having (owning) things>, the phenomenon is not perfectly consistent compared to the
above relations. For instance, examples of the types of <person> and <wearing> are shown
in (77).
(77) a. kuroi uwagi-no dansee
black jacket-gen man
‘man with black jacket’
b. ?? uwagi-no dansee
jacket-gen man
‘man with jacket’
c. chairo-no kutsu-no hito
brown-gen shoes-gen person
‘person with brown shoes’
d. ?? kutsu-no hito
shoes-gen person
‘person with shoes’
e. marui megane-no gakusee
round glasses-gen student
‘student with round glasses’
f. ?? megane-no gakusee
glasses-gen student
‘student with glasses’
g. nagai sukaato-no onnanoko
long skirt-gen girl
‘girl with long skirt’
h. ?? sukaato-no onnanoko
skirt-gen girl
‘girl with skirt’
(Harada, 1991: 10)
46 Chapter 4. Saturation of nouns
As we see, it is hard to judge their grammaticality with this type of <person> and
<wearing>. The following are further examples that indicate this kind of grammaticality
judgment difficulty.
(78) a. ?? uwagi-no dansee
jacket-gen man
‘man with jacket’
b. ? kooto-no dansee
coat-gen man
‘man with coat’
c. suutu-no dansee
suit-gen man
‘man with suit ’
d. wafuku-no dansee
kimono-gen man
‘man with kimoto’
(79) a. ?? kutsu-no dansee
shoes-gen man
‘man with shoes’
b. geta-no dansee
geta-gen man
‘man with geta’
Harada explains this wavering of judgment in terms of common sense and linguistic and
non-linguistic ones. Moreover, he claims that <person> and <having (owning) thins> are
of the same type as <person> and <wearing>. Some instances of <person> and <having
(owning) thins> are the following.
(80) a. shiroi sutekki-no dansee
white cane-gen man
‘man with white cane’
b. ? sutekki-no dansee
cane-gen man
‘man with cane’
Section 4.2. Relational nouns 47
c. akai porushe-no josee
red Porshe-gen woman
‘woman who owns red Porsche’
d. porushe-no josee
Porshe-gen woman
‘woman who owns Porsche’
e. kuroi booenkyoo-no josee
black telescope-gen woman
‘woman who has telescope’
f. ? booenkyoo-no josee
telescope-gen woman
‘woman who has telescope’
He also suggests that the sentences become natural expressions when the part of a person
is optional and is not expected for everyone to have, such as hige ‘beard’ as shown in (81)
(81) a. hige-no dansee
beard-gen man
‘bearded man’
b. nikibi-no otokonoko
pimple-gen boy
‘boy with pimpls’
c. sobakasu-no onnanoko
freckle-gen girl
‘girl with freckles’
There is further research on the construction NP1 no NP2; for instance, Kikuchi and
Shirai (2006) analyze this construction using the relational nouns.
They claim that noun phrases depend not only on the semantic properties of NPs but
also on contextual information shown in the NP1 no NP2 construction, as in (82). They
propose that there are two kinds of no in NP1 no NP2 constructions: in one case NP1 is
the complement of NP2, and in the other NP1 is used as an adnominal phrase that modifies
NP2. Then they claim that determining the relation R is crucial, and it is heavily context
dependent.
48 Chapter 4. Saturation of nouns
(82) a. Naomi no haha
Naomi mother
“Naomi’s mother”
b. machi no hakai
city destruction
“destruction of a/the city”
(Kikuchi and Shirai, 2006: 1)
No is a postposition in Japanese and its grammatical function is similar to that of the
preposition ‘of’ and the possessive marker ‘’s’ in English. In the first case such as (82a)
and (82b), the NP2 has to have an argument, if NP1 is its complement. Kikuchi and Shirai
(2006) define such an NP2 as a verbal noun which can make a light verb with suru (e.g.,
hakai-suru) as in (82b), or a function/relation noun denoting an entity which depends on
another entity such as haha ‘mother’ as in (82a).
On the other hand, in the second case, no’s part-of-speech is adnominal, and it has an
argument as in (83a) – (83d).
(83) a. Toyota no kuruma
Toyota car
“a car made by Toyota”
b. bijin no haha
beautiful woman mother
“one’s beautiful mother,” “a/the beautiful woman’s mother”
c. Sheekusupia no hon
Shakespeare book
“Shakespeare’s book,” “a book on Shakespeare,” “a book written by Shakespeare”
d. Pari no ie
Paris house
“a house in Paris”
e. Naomi no chiimu
Naomi team
“Naomi’s team,” “the team that Naomi expects to win,” etc.
Section 4.2. Relational nouns 49
(Kikuchi and Shirai, 2006: 1)
They analyze typical possessive structures such as Naomi no neko ‘Naomi’s cat’, in
which Naomi no is an adnominal phrase that modifies neko ‘cat’ as in (85). Therefore, the
two kinds of no are treated separately, because in the former case no can appear as a gap
in a long-distance dependency construction as is (84), but in the other it does not behave
in the same way as shown in (85).
(84) a. ie no mae ga akichi da
house fromt nom vacant cop
‘ (the) house’s front lot is vacant’
b. φi mae ga akichi da to Naomi ga shinji-teiru iei
house nom vacant cop comp Naomi nom believe house
‘ (the) house whose front lot Naomi believes is vacant’
c. (kono) sakana no ryori ga taihen da
(this) fish cooking nom hard cop
‘ (this) fish is hard to cook’
d. φi ryori ga taihen da to Naomi ga omot-teiru sakanai
cooking nom hard cop comp Naomi nom believe fish
‘ (the) fish which Naomi thinks is hard to cook’
(85) a. Naomi no neko ga Taro ni kamitsui-ta
Naomi cat nom Taro goal bit
‘ Naomi’s cat bit Taro’
b. ?? φi neko ga Taro ni kamitsui-ta Naomi
cat nom Taro goal bit Naomi
‘ Naomi, whose cat bit Taro’
(Kikuchi and Shirai, 2006: 3)
In sum, if NP1 is an argument of NP2 as a complement, it provides complementation.
On the other hand, if NP1 is an adjunct, it will provide modification.
(86) a. complementation: NP1 is an argument of NP2
b. modification: NP1 is an adjunct for NP2
50 Chapter 4. Saturation of nouns
4.3 Unsaturated and saturated nouns
In the following sentences, noun phrases at the position of NP2 in the construction of NP1
no NP2 , for instance shuyaku ‘leading actor’, share common properties. It can be assumed
that these nouns need some kind of possessor. It means that they belong to someone or
something. Nishiyama (1990, 2003) discusses how these nouns cannot define their extension
by themselves and are not complete in their semantics, unless a parameter such as ‘X’s’
of shuyaku is provided. For example, the sentences such as ‘Are you a partner?’ does not
make sense unless it is defined whose partner you are. From this perspectives, Nishiyama
(1990) calls these nouns “hihouwa” ‘unsaturated’ nouns, which need to have a parameter
in the form of X no ‘X’s’.
(87) a. kono shibai-no shuyaku
this play-gen leading actor
‘a leading actor of this play’
b. dai juuyonkai shopan konkuuru-no yuushoosha
number fourteenth Chopin contest-gen winner
‘a winner of a Chopin contest in the number fourteenth’
c. Taroo-no jooshi
Taro-gen boss
‘Taro’s boss’
d. kono daigaku-no sooritsusha
this University-gen founder
‘a founder of this University’
e. Genjimonogatari-no sakusha
Genji-story-gen writer
‘ a writer of a Genji story’
f. seisei bunpoo riron-no kenkyuusha
generative grammar theory-gen researcher
‘researcher of generative grammar theory’
g. jiyuuminshutoo-no kanbu
the Liberal Democratic Party-gen executive
‘a executive in the Liberal Democratic Party ’
Section 4.3. Unsaturated and saturated nouns 51
h. Yooko-no aite
Yoko-gen partner
‘Yoko’s partner’
i. ano hon-no hyooshi
that book-gen cover
‘a cover of that book’
(Nishiyama, 2003: 33)
In the following examples, the second NPs need not have a possessor intrinsically.
Nishiyama (1990, 2003) classifies the nouns such as haiyuu ‘actor’ as a “houwa ” ‘satu-
rated’ nouns. They don’t require a parameter and are complete in their semantics by
themselves. In his discussion, ‘Yoko owns a necklace’ is something like ‘a necklace belongs
to Yoko’, which defines NP2 having some relation to NP1 in (88).
(88) a. Yooko-no kubikazari
Yoko-gen necklace
‘Yoko’s necklace’
b. Hokkaidoo-no haiyuu
Hokkaido-gen actor
‘actor in Hokkaido’
c. tonari-no heya-no oto
next to room-gen sound
‘sound from the room next to’
(Nishiyama, 2003: 16)
In this type, the NP2 is also a saturated noun, and has a sense that NP1 de aru NP2
‘NP2 which is NP1’ in respect to this NP no NP construction as in (89).
(89) a. korerakanja-no daigakusee
cholera patient University student
‘University student who is a cholera patient’
b. pianisuto-no seijika
pianist-gen politician
‘politician who is a pianist’
52 Chapter 4. Saturation of nouns
c. Hokkaidoo shusshin-no haiyuu
Hokkaido from-gen actor
‘actor from Hokkaido’
(Nishiyama, 2003: 19)
He claims that the distinction between saturated and unsaturated nouns is semantically
motivated, and is regulated in the lexicon. That is, it is defined for each word whether it is
a saturated or unsaturated noun.2 In the next section, however, we will see some evidence
that their syntactic behaviors are different.
4.4 The relation between Saturation and argument structure
It is sometimes suggested that the nouns also have their own arguments similarly to verbs.
Grimshaw (1990) claims that the “complex event nouns” have their own argument structures
in the same way as verbs. For instance, examination has its arguments such as in the
examination of the patient by the doctor. The following are a few examples in Japanese
from Kageyama (1993: 187), where the gerunds which can combine with suru ‘do’ are
examples of the complex event nominals in Japanese.
(90) a. misejimai - suru ‘closing up the store’
b. ashibumi -suru ‘stamping’
c. yasuuri - suru ‘going cheap’
(Kageyama, 1993: 187)
The phrases in (91) such as senzo-no ‘ancestor’s’ and issakunen nakunatta chichioya-no
‘father who died last year’, and furui ‘old’, mikageishi-no ‘granite-gen’ and yamaoku-ni aru
‘deep in the mountains-loc is’ as in (92) both seem to modify haka ‘grave’. However, the
phrases shown in (92) are modifiers of haka and optional, the phrases shown in (91) seem
to be arguments of haka because they provide indispensable information for haka.
(91) a. senzo-no haka
ancestor-gen grave
‘an ancestor’s grave’
2He also classifies other types of noun phrase based on the NP1 no NP2 construction. However, sinceother classifications are not our main concern, we will not examine them in detail, and concentrate on thedistinction in terms of saturation.
Section 4.4. The relation between Saturation and argument structure 53
b. issakunen nakunatta chichioya-no haka
last year died father-gen grave
‘grave of a father who died last year’
(92) a. furui haka
old grave
b. mikageishi-no haka
granite-gen grave
c. yamaoku-ni aru haka
deep in the mountains-loc is grave
Kageyama suggests that not only complex event nouns, such as those in (90), but nouns
which presents a relative concept, such as haka, have arguments. A piece of evidence is
shown in the examples (94) and (93) below.
(93) a. * furui hakamairi
old visiting grave
b. * mikageishi-no hakamairi
granite-gen visiting grave
c. * yamaoku-ni aru hakamairi
deep in the mountains-loc is visiting grave
(Kageyama, 1993: 334)
(94) a. senzo-no hakamairi
ancestor-gen visiting grave
‘visitiong an ancestor’s grave’
b. issakunen nakunatta chichioya-no hakamairi
last year died father-gen visiting grave
‘visiting grave of a father who died last year’
(Kageyama, 1993: 333)
54 Chapter 4. Saturation of nouns
Hakamairi ‘visiting grave’ cannot take words such as furui ‘old’ or mikageishi-no ‘granite-
gen’ as its modifiers as in (93). On the other hand, it can apparently take the person who is
buried in the grave as its modifier as shown in (94) above.3 He suggests that the relationship
between father and grave is crucial: grave has to have as its argument the person who is
buried. Therefore, grave also has its own argument, and it is called bunri fukanoo shoyuu
‘inalienable possession’ by Kageyama (1993). He claims that the argument of haka, which
is inside of the complex word hakamairi, is inherited to the complex word and behaves as
an argument of hakamairi.
Similar examples of ‘inalienable possession’ are shown in (95)
(95) a. otoosan-no katatataki
father-gen hitting shoulder
‘hitting father’s shoulder’
b. fuku-no shiminuki
cloth-gen removing a stain
‘removing a stain from cloth’
c. tejina-no taneakashi
magic-gen reveal the trick
‘reveal the trick of magic’
d. ryoori-no udejiman
cooking-gen proud of skill
‘proud of cooking skill’
e. akaji-no anaume
deficit-gen cover
‘cover a deficit’
f. senpai-no shirinugui
senior-gen clear up mess
‘clearing up senior’s mess’
g. danjo-no enmusubi
man and woman-gen form a connection
‘forming a connection with man and woman’
3It is usually the case that morphological integrity (cf. Kageyama (1993: 10, 325)) prohibits this kind ofmodification of part of a word.
Section 4.4. The relation between Saturation and argument structure 55
h. shukun-no katakiuchi
lord-gen revenge
‘revenge for lord’
i. hanayome-no oironaoshi
bride-gen changing dress
‘changing bride’s dress’
j. shoohin-no zaikoseiri
product-gen stocktaking
‘stocktaking of products’
(Kageyama, 1993: 334)
Gunji (2002) suggests that there are many more nouns for this type. For typical in-
stances, body parts, kinship terms, social positions or roles. These nouns do not make any
sense when they are used alone in the sentences, and they require their arguments in a
configuration with genitive marked NPs.
(96) a. Taro-no oyagawari
Taro-gen surrogate parent
‘surrogate of Taro’s parent’
b. musuko-no yomesagashi
son-gen seeking for wife
‘seeking for son’s wife’
c. oyagaisha-no yakuinsoujishoku
parent company-gen resignation of the all directors
‘the resignation of the all directors in a parent company’
d. shinnenkai-no kanjidaikoo
New Year party-gen agent of the organizer
‘agent of the organizer for New Year party’
(Gunji, 2002: 43)
Gunji (2002) formalizes this type of noun in Kageyama (1993)’s examples in the notation
of Head-driven Phrase Structure Grammar (HPSG: Sag, Wasow & Bender 2003) to present
56 Chapter 4. Saturation of nouns
their argument structures. We will see the formalization of senzo-no hakamairi ‘visiting
ancestor’s grave’ following Gunji’s analysis.
In general, each verb has its own lexical specification of argument structure. For in-
stance, transitive verbs such as mairu ‘visit’ take two arguments: the person who visits
somewhere and the place visited, as shown in (97).
(97)
phon mairu
synsem
syn
[
head verb
]
arg-st 〈 NPi , NPj 〉
sem see (i, j )
synsem represents the syntactic and semantic information. arg-st stands for Argument
Structure where arguments are aligned in the reverse order of obliqueness (i.e., subject first,
object second, etc.). The syn stands for its syntax and sem represents its semantics.
He assumes that complex words also have heads and they are at the right hand of the
words. They have their internal heads in the same way as phrases. Henceforth, in the
example of hakamairi ‘visiting grave’, mairi ‘visiting’ is a morphological head of the word,
indicated by mh, as shown in the tree structure as (98).
(98)NP
NP
senzo-no
N
m
haka
mh
mairi
(Gunji, 2002: 44)
It is proposed that mairi ‘visiting’ has the same argument structure as its verbal coun-
terpart mairu ‘visit’. The following is the structure of mairi. The verb mairu takes NPi
and NPj as its arguments, subject and object respectively, and makes a sentence such as
NPi visit NPj . He suggests that the noun mairi also takes two arguments in the same way
as the verb form; therefore, the meaning of the the noun phrase becomes NPi’s visiting of
NPj as in (99).
Section 4.4. The relation between Saturation and argument structure 57
(99)
phon mairi
synsem
syn
[
head noun
]
arg-st 〈 NPi , NPj 〉
sem visit (i , j)
(Gunji, 2002: 45)
It is also assumed that haka ‘grave’ has arguments which indicate whose grave it is.
Therefore, haka takes two arguments representing NPl is NPk’s grave, and its semantic
value represents that l is a k’s grave, as shown in (100).
(100)
phon haka
synsem
syn
[
head noun
]
arg-st 〈 NPl , NPk 〉
sem tomb-of (l , k)
(Gunji, 2002: 46)
Therefore, the noun hakamairi takes two arguments since it can be used in the form as
visiting NPk’s grave by NPi; hence its representation will be the following:
(101)
phon hakamairi
synsem
syn
[
head noun
]
arg-st 〈 NPi , NPk 〉
sem visit (i , j) ∧ tomb-of (j , k)
(Gunji, 2002: 47)
The arg-st of hakamairi is inherented from that of haka and mairi. In more detail,
(102) a. hakamairi and the head mairi share the first argument (a person who visits).
b. hakamairi and haka share the second argument (whose tomb).
c. The index of the first argument of haka and the index of the second argument of
mairi are coindexed.
58 Chapter 4. Saturation of nouns
d. The semantics of hakamairi is a conjunction of the semantics of haka and mairi.
(Gunji, 2002: 49)
(103)senzonohakamairi
arg-st 〈 1 NPi 〉
sem visit (i , j) ∧ tomb-of (j , k)
3 senzono[
sem k
]
hakamairi
arg-st 〈 1 , 3 〉
sem visit (i , j) ∧ tomb-of (j , k)
2 hakaj
arg-st 〈 NPj , 3 〉
sem tomb-of (j , k)
mairi
arg-st 〈 1 , 2 〉
sem tomb-of (i , j)
(Gunji, 2002: 48)
In sum, senzo is an argument for haka, however mikageishi-no is a modifier. Therefore,
Gunji (2002) concludes that complex nouns consisting of such nouns having their own
argument will follow this pattern.
Following the notion above, they can be used as a test for whether they are modifiers
or arguments of the nouns.
4.5 ‘Saturation’ in our experiments
In this section, we will see how we use the term saturation in our experiments.
As we have seen there are various related terms for this type of noun. Saturation, which
was introduced in section 4.3, is more restricted conception than relationality, since the
unsaturated nouns are included in relational nouns. The summary for relational nouns (86)
in section 4.2 is repeated here again as (104).
(104) a. complementation: NP1 is an argument of NP2
b. modification: NP1 is an adjunct for NP2
We can pigeonhole the mixed denotations, which we saw so far, as shown in (105). We
will adopt argumenthood here and call them as it.
Section 4.5. ‘Saturation’ in our experiments 59
(105) a. complementation vs. modification (Harada, 1991, Kikuchi and Shirai, 2006)
b. “hihouwa” (unsaturated) vs. “houwa” (saturated) (Nishiyama, 1990)
c. argument vs. modifier (Kageyama, 1993, Gunji, 2002)
We use only typical unsaturated nouns to be able to obtain clear results. For instance, we
do not use the nouns such as kubikazari, kodomo (they are saturated nouns in Nishiyama’s
classification) that can belong to the person if the words are preceded by person names.
(106) a. Naomi-no kubikazari
b. Naomi-no kodomo
c. Naomi-no hon
The following are sample groups of our experimental items.
(107) a. Unsaturated: kinship terms, inalienable body parts, internal character of the
person, etc.
• ryooshin (‘parents’)
• musume (‘daughter’)
• onshi (‘(former) teacher’)
• senpai (‘senior’)
• deshi (‘apprentice’)
• mokuhyoo (‘goal’)
• shoorai (‘future’)
• kangae (‘idea’)
• konomi (‘taste’)
• taichoo (‘physical condition’)
b. Saturated: professions, entity, scenery, etc.
• manzaishi (‘comic duo’)
• kangofu (‘nurse’)
• maikosan (‘apprentice geisha’)
• pianisuto (‘pianist’)
60 Chapter 4. Saturation of nouns
• sakkyokuka (‘composer’)
• niji (‘rainbow’)
• otsukisama (‘moon’)
• zasoo (‘weeds’)
• hoshizora (‘starlit sky’)
• supootsu (‘sports’)
4.6 Conclusion
In this section, we discussed the importance of treating the noun phrases carefully in sen-
tence processing experiments. In order to classify the common noun phrases, we looked
at some examples of relational nouns in the NP1 no NP2 construction in Harada (1991),
Gunji (1995) and Kikuchi and Shirai (2006). Then we looked into the distinction between
“saturated nouns” and “unsaturated nouns” from Nishiyama (1990)’s proposal. In order
to examine the phenomenon for relational nouns, we also looked at the relation between
saturation and argument structure from Kageyama (1993)’s and Gunji (2002)’s analysis of
relational nouns. It is suggested that some nouns have their own arguments in the same
way as verbs.
Finally, we defined how we are going to treat saturation and relational nouns in our
experiments.
Chapter 5 Experiment 1
5.1 Introduction
Assuming that people process sentences incrementally trying to understand the meaning of
each incoming segment immediately before reaching the end of the utterance, the question
arises as to how a sequence of NPs is interpreted before reaching a verb that can take them
as arguments (or as adjuncts). It is of interest to understand the kinds of information that
are used in this process and how soon they are used. The goal of this section is to provide
evidence for the claim that each individual NP is not interpreted in isolation based on its
individual properties such as case and animacy alone, but rather that readers try to create
interpretations including all NPs taking into consideration the possible relations that may
hold between them.
Previous work indicates that case markers can affect the expectations of what will follow
in the sentence. For example, after listening to a fragment containing a dative NP, Japanese
speakers are more likely to make predictive eye movements towards the drawing of an
object that is a potential direct object (Kamide, Altmann & Haywood, 2003). Although
plausibility is likely to play a role in this process (by favoring a candidate entity that is
plausible as a direct object), the case marker is unarguably crucial. This is confirmed by
completion questionnaire data which indicate that a dative NP is commonly followed by
another NP (about 33% of the time), whereas an accusative NP seldom elicits another NP
(2%; Miyamoto & Nakamura, 2005).
Moreover, nominative markers are important in determining how many clauses are ex-
pected to be in a sentence being read (Miyamoto, 2002; Uehara & Bradley, 2002). Further
work indicates that other properties associated with individual NPs, such as animacy, also
need to be considered in this process (Muraoka & Sakamoto, 2004). These points are
summarized below.
• the number of the arguments which verbs take from the information of case markers
• nominative case marker indicates clause boundary
• animacy information is also crucial
• saturation effect
61
62 Chapter 5. Experiment 1
There are two ways in which case marking and animacy can be used during the under-
standing of a sequence of NPs. In one model, each NP is interpreted exclusively based on its
case marker and its individual properties (which we will refer to as intra-nominal properties;
e.g., animacy). Hence, a nominative-marked human NP would be the prototypical agent,
an accusative-marked object would most likely receive the role of theme, and so forth.
In a second model, although intra-nominal properties are important, they are interpreted
within the context in which the NP occurs. In other words, the interpretation of a NP
depends on its relations to other NPs (inter-nominal relations), so that the same NP (e.g.,
an animate NP) may assume a certain role in are context but a different role in another
context.
• intra-nominal properties: case markers, animacy
• inter-nominal relations: saturation, possessor relations
We investigated how native Japanese speakers expect sequences of NPs to continue
by asking them to complete fragments into full sentences. Although the present results
cannot indicate how soon inter-nominal associations such as possessor relations affect the
interpretation of NPs, we will argue that they nevertheless provide evidence for the use of
such relations during the processing of sequences of NPs, therefore favoring the second type
of model described earlier.
For instance, if participants read following fragments such as (108), some possible can-
didates are the sentences such as (109).
(108) Naomi-ga konya-no hoshizora-ga ...
Naomi-nom tonight starlit sky-nom
⇓
(109) a. Same clause:
[ Naomi-ga konya-no hoshizora-ga sukida ]
Naomi-nom tonight starlit sky-nom like
‘Naomi likes tonight starlit sky.’
[ Naomi-ga konya-no hoshizora-ga mitai ]
Naomi-nom tonight starlit sky-nom see-want
‘Naomi wants to see tonight starlit sky.’
Section 5.1. Introduction 63
b. Different clauses:
[ Naomi-ga [ konya-no hoshizora-ga totemo kireidane ] to itta. ]
Naomi-nom tonight starlit sky-nom very beautiful comp said
‘Naomi said that tonight starlit sky is very beautiful.’
If we change the second NP to what we call the unsaturated inanimate NP in the
beginning of the fragments in (110), the examples of their answer will be sentences such as
in (111).
(110) Akira-ga kangae-ga ...
Akira-nom idea-nom
⇓
(111) a. Same clause:
[ Akira-ga kangae-ga subarashii. ]
Akira-nom idea-nom great
‘Akira’s idea is great.’
[ Akira-ga kangae-ga aru. ]
Akira-nom idea-nom has
‘Akira has an idea.’
b. Different clauses:
[ Akira-ga [ gapi kangae-ga subarashii ] buka-oi hometa. ]
Akira-nom idea-nom great subordinate-acc praised
‘Akira praised his subordinate whose idea is great.’
From the perspective of the incrementalily in Japanese sentence processing, we can
assume that people start to interpret the sentences even before the verb shows up. Some of
the crucial information are case markers, animacy, estimating the arguments which verbs
take, and clause boundary. We think the information which people use is not only intra-
nominal properties, but also inter-nominal properties, namely saturation of the nouns and
argument structure of nouns.
In this chapter, we will see whether the saturation affects our interpretation of the sen-
tences. In order to figure out what kinds of information are used in the sentence processing,
we conduct a sentence completion task in our experiment.
As we have seen in the chapter 3, the nominative marker ga is indicator of a left clause
boundary (Miyamoto, 2002) and ga is a subject of a different clause. Animacy information
64 Chapter 5. Experiment 1
(human noun or entity noun) also affects the processing (Muraoka & Sakamoto, 2004).
In this experiment, we hypothesis the saturation information also affects the processing.
Therefore what we will sort out is the following:
(112) a. Nominative marker ga is a single clause subject, and the indicator of the left clause
boundary.
b. Animacy information affects the interpretation of the sentences.
c. Saturation or possessor interpretation affects the interpretation of the sentences.
5.2 Method
Recent work indicates that in sentences containing the sequence NP-ni NP-ga NP-ga, the
animacy of the second nominative NP is important in determining whether a new clause
boundary is inserted between the two nominative NPs (Muraoka & Sakamoto, 2004). In
the present experiment, we manipulated animacy as well as possible possessor relations in
order to explore this phenomenon in more detail.
5.2.1 Participants
Thirty-two native speakers of Japanese, undergraduate students at Kobe Shoin Women’s
University, volunteered to participate in the experiment.
5.2.2 Materials
There were eight types of fragments following a 2×2×2 design, which participants were asked
to complete into sentences. The first factor was whether the fragment contained a dative
NP or not. The second nominative NP in each fragment was manipulated according to two
other factors, namely animacy and saturation (whether the noun requires a possessor).
We will use the term saturated to refer to nouns that in general do not presuppose
an owner (e.g., professions such as pianisuto ‘pianist’), and the term unsaturated will be
used to refer to nouns that usually belong to somebody (e.g., inalienables such as body
parts and kinship terms; see chapter 4, for more details). In English, unsaturated nouns
are usually preceded by a genitive marked NP (John’s hand) or a possessive pronoun (his
hand). In Japanese, the possessor can be indicated with a genitive marked NP, while
possessive pronouns are often left implicit and will be indicated with the zero pronoun pro
for convenience sake. The exact syntactic representation of the possessor in the multiple
nominative constructions investigated is not crucial for our claims (e.g., it has sometimes
been claimed to be a trace left by movement in the transformational grammar tradition,
but see Vermeulen, 2005, for a recent discussion against such an approach).
Section 5.2. Method 65
The sentences in (113) are a sample set of fragments that do not contain a dative marked
NP (the ga-ga conditions for short; see Appendix A for the complete list of fragments used).
(113) a. Unsaturated / Animate
Akira-kun-ga hahaoya-ga
Akira-title-nom mother-nom
b. Unsaturated / Inanimate
Akira-kun-ga mokuhyoo-ga
Akira-title-nom goal-nom
c. Saturated / Animate
Yooko-chan-ga ano pianisuto-ga
Yoko-title-nom that pianist-nom
d. Saturated / Inanimate
Yooko-chan-ga konyano hoshizora-ga
Yoko-title-nom tonight starlit sky-nom
The first nominative NP in each fragment is always a human proper name.
For the Unsaturated conditions illustrated in (113a) and (113b), unsaturated nouns
are employed as the second nominative NP, which moreover is animate in (113a) (ha-
haoya ‘mother’) and inanimate in (113b) (mokuhyoo ‘goal’). In contrast, in the Saturated
conditions exemplified in (113c) and (113d), saturated nouns are employed as the second
nominative NP, which is animate in (113c) (ano pianisuto ‘that pianist’) and inanimate in
(113d) (konya-no hoshizora ‘tonight’s starlit sky’).
A dative marked NP (a human proper name such as Michiko-chan) was added to begin-
ning of each one of the ga-ga conditions in order to obtain fragments with an initial dative
NP and two nominative NPs (the ni-ga-ga conditions). A sample set is provided in (114).
(114) a. Unsaturated / Animate
Michiko-chan-ni Akira-kun-ga hahaoya-ga
Michiko-title-datAkira-title-nom mother-nom
b. Unsaturated / Inanimate
Michiko-chan-ni Akira-kun-ga mokuhyoo-ga
Michiko-title-datAkira-title-nom goal-nom
c. Saturated / Animate
Yoshimoto-kun-ni Yooko-chan-ga ano pianisuto-ga
Yoshimoto-title-dat Yoko-title-nom that pianist-nom
66 Chapter 5. Experiment 1
d. Saturated / Inanimate
Yoshimoto-kun-ni Yooko-chan-ga konya-no hoshizora-ga
Yoshimoto-title-dat Yoko-title-nom tonight-gen starlit sky-nom
We would like to determine whether saturation affects the way readers complete such
fragments, or whether intra-nominal factors such as animacy are so important that satura-
tion has no effect on the interpretations that readers are willing to consider. One possible
way in which saturation can affect the completions is in terms of the number of clauses.
Because unsaturated NPs imply an owner, and the first nominative NP can be such an
entity in the fragments above, we predict that the two nominative NPs in the Unsaturated
conditions are more likely to be interpreted jointly, leading to an increase in the number
of completions in which these two NPs are in the same clause. If this is correct, then
this would be evidence for the claim that readers utilize inter-nominal relations (such as
possessor-possessee relations) in order to interpret a sequence of NPs.
Twelve sets of four fragments with unsaturated nouns (as in (113a, b) and (114a, b)
and twelve sets of four fragments with saturated nouns (as in (113c, d) and (114c, d)) were
created.
There were 28 fillers of various types in addition to the experimental fragments. Some
of the filler fragments required at least two predicates in order to yield a complete sen-
tence, while other fragments could be completed with just one predicate. None of them
manipulated possessor relations or had sequences of multiple nominative NPs.
5.2.3 Procedure
The 12 sets of unsaturated nouns and 12 sets of saturated nouns were subdivided into four
lists following a Latin Square design. Each list was interspersed with 28 filler fragments
and pseudo-randomly ordered avoiding as much as possible that test items of the same type
followed in succession. Each participant saw one list with 52 items (24 test fragments and
28 fillers) and was instructed to complete each fragment into a sentence in black or blue
ball-point pen. Fragments were presented in Japanese fonts followed by a straight line as
in the example below, which corresponds to (114a), and another example.
(115) 美智子ちゃんにあきら君が母親が
Michiko-title-dat Akira-title-nom mother-nom
(116) 伊藤さんが孫が
Itoo-title-nom grandchild-title-nom
Section 5.2. Method 67
Each questionnaire was preceded by two pages of instructions. The experiment took
participants about thirty minutes to complete.
5.2.4 Analysis
Completions were classified into 15 classes after ambiguous, ungrammatical and incomplete
answers were removed (classes X, Y and Z; see Table 5 for a schematic representation of the
types of completions; class (J2) in the table was not used for this experiment). Completions
of types (A) to (E2) are the cases in which the two nominative NPs are in the same clause,
whereas in completions of types (F) to (L) the two NPs are assumed to be in different
clauses.
Analysis of coding
We were primarily interested in whether the two nominative NPs were interpreted as
part of the same clause or not, as this is an indication of the kind of semantic roles that are
being assigned to them. We will nevertheless report the numbers for the 15 classes in order
to provide an idea of the variety of completions obtained. We will show some examples
of completions that were collected from participants in the next paragraph. However, we
will not discuss the differences between the classes in detail as they can be argued to be
somewhat subjective and their significance uncertain.
68 Chapter 5. Experiment 1
Table 5: Schematic representations for the classes used to analyze the completions.The column Clause indicates whether the two nominative NPs were classified as partof the same clause or not. The boxed constituents correspond to the two nominativeNPs. RC = Relative clause. Poss = Possessor.
Code Clause Description
A Same [ possessor subject V ]B Same [ subject object V ]C Same [ subject subject V ]D Same [ [ possessor subject V ] subject V ]D2 Same [ [ subject object V ] subject V ]E Same [ prospeaker [ possessor subject V ] V ]E2 Same [ prospeaker [ subject object V ] V ]F Different [ subject [ subject V ] V ]G Different [ subject [
RCsubject gapi(object) V ] NPi V ]
G2 Different [ subject [RC
gapi(possessor) subject V ] NPi V ]H Different [ subjecti [ proi(subject) object V ] V ]I Different [ subjecti [ proi(poss) subject V ] V ]J Different [ subject [ prospeaker(poss) subject V ] V ]K Different [ subject [ prosomebody(poss) subject V ] V ]L Different [ subject [ subject V(existential)] V ]X — Eliminated: Completely ambiguousY — Eliminated: UngrammaticalZ — Eliminated: Incomplete sentence
The following are examples of completions that were provided by participants for each
class. In all examples provided, the fragment and the completion are separated by a slash /.
Square brackets indicate clausal structure.
In type (A), the two nominative marked NPs are in the same clause and the first NP-ga
is the possessor of the second NP-ga as in (117). Sentences in this class include only one
predicate (see Uehara & Bradley, 2002, on the external possession construction).
(117) a. [Akira-kun-ga sakuhin-ga / kowareta.]
Akira-title-nom art-work-nom broke
‘Akira’s art work broke.’
Section 5.2. Method 69
b. [Fukuda-san-ga yume-ga / fukuranda.]
Fukuda-title-nom dream-nom swelled
‘Fukuda’s dream swelled.’
In type (B), the two NP-ga phrases are arguments of the same two-place predicate as
part of a single clause as in (118). The first NP-ga is the subject and second is the object
(see the nominative object construction in Uehara & Bradley, 2002)
(118) a. [Iida-san-ga zassoo-ga / sukida.]
Iida-title-nom weed-nom like
‘Iida likes weed.’
b. [Kawasaki-san-ga musume-ga / hoshii-rashii.]
Kawasaki-title-nom daughter-nom seem to want
‘Kawasaki wants to have (her) daughter.’
In class (C), the two NP-ga phrases are used as parallel (conjoined) subjects of the same
predicate in a single clause as in (119) (see the parallel subject construction in Uehara &
Bradley, 2002). Sentences in this class would probably be considered ungrammatical by
most speakers (as indicated by the question mark preceding the example), but they are
consistently produced by participants, therefore we included them in the analyses.
(119) ?? [Satoo-san-ga mago-ga / issho-ni undookai-de hashitteita.]
Sato-title-nom grandchild-nom together sports meeting-loc were running
‘Sato, (his/her) grandchild were running together at the sports meeting.’
Sentences in (D) include more than one clause but the two nominative NPs are in the
same clause. An NP in the completion is the subject of the matrix verb. An example is
shown in (120).
(120) a. [[Kobayashi-kun-ga taido-ga / kawatta]-to mawari-no hito-ga itta.]
Kobayashi-title-nom behavior-nom changed-that around people-nom said
‘The people around him said that Kobayashi’s behavior has changed.’
b. [[Kayo-chan-ga seikaku-ga / maruku-natta]-to otonatachi-wa iu.]
Kayo-title-nom character-nom get well-rounded-that adults-nom say
‘Adults say that Kayo’s character got well rounded.’
Class (D2) is similar to (D); the two nominative NPs are part of an embedded clause
and an NP in the completion is the matrix subject. The first nominative NP is subject and
the second NP is object as shown in (121).
70 Chapter 5. Experiment 1
(121) [[Kobayashi-kun-ga tonari-no hito-ga / suki-da]-to-iu-koto-wa minna shitte-iru.]
Kobayashi-title-nom next to person-nom like-that everybody knew
‘Everybody knew that Kobayashi likes the person next to him.’
Class (E) includes sentences in which the clause boundary is somewhat ambiguous, but
the interpretation with the two NP-ga phrases in the same clause is more natural as in
example (122). These kinds of sentences seem to have an empty pronoun pro as the matrix
subject which usually corefers with the speaker of the sentence.
(122) a. [pro [Suzuki-senpai-ga koodoo-ga / okashii]-to omotta.]
Suzuki-senior-nom behavior-nom strange-that thought
‘(I) thought that Suzuki’s behavior was strange.’
b. [pro [Nonaka-kun-ga konomi-ga / okashii]-to kiita.]
Nonaka-title-nom taste-nom strange-that heard
‘(I) heard that Nonaka’s taste is strange.’
Class (E2) is similar to (E), the two nominative NPs are part of an embedded clause
and the matrix subject is implicit (probably the speaker). The first nominative NP is the
subject of second NP which is object as shown in (123).
(123) [pro [Kayo-chan-ga Ogura-san-no seikaku-ga / sukirashii]-koto-o shitta.]
Kayo-title-nom Oguratitlecharacter-nom seem to like-that found
‘(I) found that Kayo seemed to like Ogura’s character.’
For the remaining classes, the two nominative NPs are part of different clauses.
In class (F), the two nominative phrases are in different clauses and each one is inter-
preted as the subject of a different predicate as exemplified by (124). The predicate in the
embedded clause is usually an adjective or an intransitive verb.
(124) a. [Naitoo-kun-ni Iida-san-ga [zassoo-ga / haetekita]-to itta.]
Naito-title-dat Iida-title-nom weed-nom are growing said
‘Iida said to Naito that weeds were growing.’
b. [Matsumoto-sensei-ga [Itoo-san-no ryooshin-ga / densha-ni notteiru]-no-o mita.]
Matsumoto-title-nom Ito-title-’s parents-nom train-loc be taking saw
‘Matsumoto saw that Ito’s parents were taking a train.’
Section 5.2. Method 71
For (G), the two nominative phrases are in different clauses, and the second one is part
of a relative clause as shown in (125).
(125) a. [Ogawa-kun-ga [tonarinohito-ga / gapi yondeiru] shinbuni-o nozokikonda.]
Ogawa-title-nom neighbor-nom is reading newspaper-acc looked into
‘Ogawa looked into a newspaper which neighbor was reading.’
b. [Shimizu-san-ga [imooto-ga / gapi motteiru] akusesariii-o nusunda.]
Shimizu-title-nom sister-nom have accessories-acc stole
‘Simizu stole accessories which his sister has.’
A class for (G2) is similar to (G), the embedded clause is a relative clause but the second
NP is subject and the gap is a possessor.
(126) a. [Yamazaki-kakarichoo-ga [RC gapi kangae-ga / subarashii ] bukai-o hometa.]
Yamazaki-title-nom idea-nom great subordinate-acc praised
‘Yamazaki praised his subordinate whose idea is great.’
b. [Kayo-chan-ga [RC gapi seikaku-ga/ ii ] kareshii-o tsukutta.]
Kayo-title-nom character-nom nice boyfriend-acc made
‘Kayo made a boyfriend whose character is nice.’
In class (H), the clause boundary is potentially ambiguous, but the more natural reading
is for the two NP-ga phrases to be parts of different clauses. The embedded pro is co-indexed
with the first NP-ga, which is the subject of the matrix clause, while the second NP-ga is
the object of the embedded clause as in (127).
(127) a. [Iida-sani-ga [ proi kangofu-ga / sukida]-to itteita.]
Iida-title-nom nurse-nom like-that said
‘Iida said that (she) likes the nurse.’
b. [Hashimoto-sani-ga [ proi Terada-kun-no egao-ga / sukida]-to itta.]
Hashimoto-title-nom Terada-title smile-nom like-that said
‘hashimoto said that (he) likes Terada’s smile.’
A possible alternative representation to (127) is (128), where both nominative NPs are
in the embedded clause and pro is the subject of the matrix clause. This interpretation is
similar to class (E). However, it is more natural to interpret this sentence as in (127) with
pro co-indexed with the first NP-ga Iida.
72 Chapter 5. Experiment 1
(128) [pro [Iida-san-ga kangofu-ga / sukida]-to itteita.]
Iida-title-nom nurse-nom like-that said
‘(I or somebody) said that Iida likes the nurse.’
Completions of type (I) are similar to the ones in (H), as the embedded pro is co-indexed
with the first NP-ga, but in (I) pro is the possessor of the second NP-ga, which in turn is
the subject of a one-place predicate as in (129).
(129) a. [Kishida-kuni-ga [ proi choonan-ga / umareta]-to hookokushita.]
Kishida-title-nom eldest son-nom born-that reported
‘Kishida reported that (his) eldest son was born.’
b. [Kuroda-senseii-ga [ proi sakuhin-ga / kanseishita]-to yorokonda.]
Kuroda-title-nom art-work-nom completed-that was pleased
‘Kuroda was pleased that (his) art work was completed.’
Completions of type (J) and (K) are similar to the ones in (I), but the embedded pro is
co-indexed with the speaker of the utterance in (J) as exemplified in (130), whereas in (K)
pro is co-indexed with somebody other than the speaker (or some other NP in the sentence)
as in example (131).
(130) [Kuroda-sensei-ga [ pro sakuhin-ga / iine]-to hometekureta.]
kuroda-professor-nom work-nom good-that praise
‘Professor Kuroda praised (me by saying) that (my) work of art was good.’
(131) [Yamazaki-kakarichoo-ga [ pro kangae-ga / amai]-to donatta.]
Yamazaki-subsection chief-nom idea-nom optimistic-that yelled
‘Subsection chief Yamazaki yelled that (somebody’s) idea was (too) optimistic.’
For completions of type (L), the two nominative NPs are in different clauses and the
embedded predicate is an existential predicate.
(132) a. [Akira-kun-ga [ pro mokuhyoo-ga / aru]-to itteita.]
Akira-title-nom goal-nom exist-that said
‘Akira said that (he) has a goal.’
b. [Fukuda-san-ga [ pro yume-ga / nai]-to nageiteita.]
Fukuda-title-nom dream-nom not exist-that grieved
‘Fukuda grieved that (he) does not have a dream.’
Section 5.3. Results 73
Completions in class (X) are completely ambiguous as they could be equally interpreted
as having the two nominative NPs in the same clause or in different clauses as in (133).
(133) Kobayashi-kun-ga senpai-ga / sukina-koto-wa shitteiru.
Kobayashi-title-nom senior-nom like-that-top know
‘Kobayashi knows the things which (his) senior likes.’ or
‘(I) know that Kobayashi likes (his ) senior.’
Completions of type (Y) are ungrammatical as in (134), in which itameru ‘hurt’ is a
transitive verb and its direct object (‘heart’ or ‘feelings’) is more commonly assigned the
accusative marker.
(134) * Nakamura-san-ga kokoro-ga / itameteiru.
Nakamura-title-nom heart-nom hurt
Completions in (Z) were incomplete sentences usually missing the matrix predicate as
in (135)
(135) * Okamoto-san-ga Hawai-no kikoo-ga / sugokuii.
Okamoto-title-nom Hawaii’s weather-nom great
‘Okamoto. . . Hawaii’s weather is great.’
In the remainder of the paper, the term fragment will refer to the fragment that was
provided to participants. And the term completion will be used for the completion written
by participants.
The data were arcsine transformed before being submitted to analyses of variance
(ANOVAs). Because of concerns with violating the homogeneity of variance assumption,
log-linear analyses (G2) and one-tailed Fisher probability exact tests (FPET) are reported
for the crucial results instead of the ANOVAs, with overall similar conclusions.
5.3 Results
There were 768 sentence completion responses (24 fragments × 32 participants) of which
18 (2.34%) were of types (X), (Y) and (Z) and were not included in further analyses. All
percentages reported hereafter exclude these cases.
A total of 750 sentence completion responses were analyzed to examine their clause
boundary placement, in particular whether the two nominative NPs in the fragments
were analyzed as part of the same clause (same-clause interpretation) or different clauses
(different-clause interpretation).
74 Chapter 5. Experiment 1
There were fewer completions (37 cases) in which the two nominative NPs belonged to
the same clause than to different clauses (706 cases) confirming previous results according to
which readers prefer to interpret each nominative NP as part of a separate clause (Miyamoto,
2002; Uehara, 1997; Uehara & Bradley, 2002).
A log-linear analysis was conducted with saturation, animacy and case (with or without
NP-ni) as factors. There was a 3-way interaction (G2(4)= 15.26, P < 0.01). However, case
did not interact with either of the other two factors (P s>0.4), but there was an interaction
between saturation and animacy (G2(2) = 14.08, P < 0.001). The remaining effects will
be discussed separately for the ga-ga conditions and the ni-ga-ga conditions.
Table 6: Number of completions with NP-ga NP-ga in the same clause of both NP-gaNP-ga and NP-ni NP-ga NP-ga conditions
Saturated Unsaturated
NP-ga NP-ga 10 27
NP-ni NP-ga NP-ga 1 6
0
5
10
15
20
25
30
sat unsat sat unsat
ga ga ni ga ga
Nu
mb
er o
f co
mp
leti
on
s
animate
inanimate
Figure 1: Number of completions with NP-ga NP-ga in the same clause of both NP-gaNP-ga and NP-ni NP-ga NP-ga conditions
Section 5.3. Results 75
5.3.1 Results for the NP-ga NP-ga fragments
When the dative NP was absent (i.e., the fragment only contained the two nominative NPs
as in the examples in (113)), results were as follows.
Table 7 provides the number of completions for each one of the classes in Table 5.
Saturation has a clear effect. However, because the numbers (especially in the Same clause
columns) are small, and as observed earlier the differences between some of these classes
can be argued to be subjective, we will not discuss these results in detail. See Table 8 for
the results according to animacy and saturation.
Table 7: Number of completions in each of the 15 classes for the NP-ga NP-gafragments in Experiment 1.
Same clause Different clauses Eliminated
Codes A B C D E F G G2 H I J K L X Y Z
Unsaturated/Animate 0 0 2 1 0 1 1 0 15 69 6 1 0 0 0 0Unsaturated/Inanimate 16 0 0 2 6 3 0 2 10 35 4 7 7 1 2 1
Saturated/Animate 0 5 2 0 0 62 3 0 23 1 0 0 0 0 0 0Saturated/Inanimate 0 2 1 0 0 68 0 0 23 0 1 0 0 0 0 1
Table 8: Number of completions (and percentages) for the ga-ga sequences accordingto whether the two nominative NPs were in the same clause or in different clausesin Experiment 1.
Same clause Different clauses Subtotals
Unsaturated/Animate 3 (3.1%) 93 (96.9%) 96Unsaturated/Inanimate 24 (26.1%) 68 (73.9%) 92
Saturated/Animate 7 (7.3%) 89 (92.7%) 96Saturated/Inanimate 3 (3.2%) 92 (96.8%) 95
Subtotals 37 (9.8%) 342 (90.2%) 379
That saturation and animacy interacted was confirmed by two types of analyses. First,
the proportion of same-clause completions for the Unsaturated conditions (Inanimate over
Animate: 24/3) was larger than for the Saturated conditions (3/7; FPET: P < 0.01).
Second, the proportion of same-clause to different-clause completions was larger in
the Unsaturated/Inanimate condition (24/68) than in the Unsaturated/Animate condition
(3/93; FPET: P < 0.00001); in contrast, the Saturated/Inanimate condition (3/92) and
the Saturated/Animate condition (7/89) did not differ (FPET: P < 0.169). In other words,
76 Chapter 5. Experiment 1
3
24
73
93
68
8992
0
10
20
30
40
50
60
70
80
90
100
Unsat/Anim Unsat/Inan Sat/Anim Sat/Inan
Nu
mb
er o
f co
mp
leti
on
s
same clause
different clauses
Figure 2: Number of completions (and percentages) for the ga-ga sequences accordingto whether the two nominative NPs were in the same clause or in different clauses inExperiment 1.
animacy only has an effect on clause boundary insertion when the second nominative NP
is unsaturated.
Table 9: Number of completions with NP-ga NP-ga in the same clause
Saturated Unsaturated
Animate 7 (7.3%) 3 (3.1%)
Inanimate 3 (3.1%) 24 (26.1%)
5.3.2 Results for the NP-ni NP-ga NP-ga fragments
When the fragment included a dative NP and two nominative NPs as in the examples in
(114), none of the differences reached statistical significance (P s>0.2; see Tables 10 and 11).
This is because there were few same-clause interpretations for the ni-ga-ga fragments overall
although the trends were the same as in the ga-ga fragments. This is probably because the
presence of the dative NP makes it harder for a single predicate to complete the fragment.
Section 5.3. Results 77
33
24
7
0
5
10
15
20
25
30
saturated unsaturated
Num
ber
of
com
ple
tions
animate
inanimate
****
****
NSNSNSNS
Figure 3: Number of completions with NP-ga NP-ga in the same clause
Table 10: Number of completions in each of the 15 classes for the NP-ni NP-gaNP-ga fragments in Experiment 1.
Same clause Different clauses Eliminated
Codes A B C D E F G G2 H I J K L X Y Z
Unsaturated/Animate 0 0 2 0 0 0 0 0 2 62 0 25 1 0 4 0Unsaturated/Inanimate 1 0 0 1 2 3 0 2 4 39 0 33 6 2 2 1
Saturated/Animate 0 0 1 0 0 74 1 0 16 1 0 0 0 0 3 0Saturated/Inanimate 0 0 0 0 0 82 0 0 13 0 0 0 0 0 0 1
78 Chapter 5. Experiment 1
Table 11: Number of completions (and percentages) for the ni-ga-ga sequences ac-cording to whether the two nominative NPs were in the same clause or in differentclauses in Experiment 1.
Same clause Different clauses Subtotals
Unsaturated/Animate 2 (2.2%) 90 (97.8%) 92Unsaturated/Inanimate 4 (4.4%) 87 (95.6%) 91
Saturated/Animate 1 (1.1%) 92 (98.9%) 93Saturated/Inanimate 0 (0.0%) 95 (100%) 95
Subtotals 7 (1.9%) 364 (98.1%) 371
Table 12: Number of completions with NP-ni NP-ga NP-ga in the same clause
Saturated Unsaturated
Animate 1 (1.1%) 2 (2.2%)
Inanimate 0 (0.0%) 4 (4.4%)
1
2
0
4
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
saturated unsaturated
Nu
mb
er o
f co
mple
tion
animate
inanimate
NSNSNSNS
Figure 4: Number of completions with NP-ni NP-ga NP-ga in the same clause
Section 5.4. Discussion 79
5.4 Discussion
The results provide evidence for the importance of case markers as indicators of clause
boundaries (Miyamoto, 2002), as nominative NPs were preferentially interpreted as part of
separate clauses (replicating Uehara, 1997; Uehara & Bradley, 2002). The dative NP also
had observable effects as its presence decreased the number of nominative NPs in the same
clause.
More crucially, the main manipulation in terms of saturation led to reliable results in
the ga-ga fragments. Saturated nominative NPs led to fewer same-clause interpretations
than unsaturated ones. Moreover, although animacy also had an effect, it was modulated
by saturation so that its effect were most visible with the unsaturated nouns.
This is the kind of result that would be expected if inter-nominal relations are taken
into consideration during the interpretation of NP sequences.
A comparison with previous results is informative as well. It has been reported that
reading time differences due to animacy are observed at a stative predicate after a ni-
ga-ga sequence but not after a ga-ga sequence (Muraoka & Sakamoto, 2004), whereas in
the present experiment the differences were primarily between the ga-ga sequences. The
contrast is interesting because in general reading time experiments and questionnaire data
tend to have comparable results. There are, however, a number of differences between the
items used and once we factor them out the results are consistent. The items are presented
again here as (136) (Muraoka & Sakamoto, 2004).
(136) a. Rooka-de shunin-ga kodomo-ga nigate-da-to buchoo-ga
corridor-loc manager-nom children-nom hard to deal with - that head-nom
Aihara-san-tachi-ni hanashite-ita
Aihara-title-and others-dat was telling
‘In the corridor, the head was talking to Aihara and others that it was hard for
the manager to deal with children. ’
b. Rooka-de shunin-ga keisan-ga nigate-da-to buchoo-ga
corridor-loc manager-nom calculation-nom hard to deal with - that head-nom
Aihara-san-tachi-ni hanashite-ita
Aihara-title-and others-dat was telling
‘In the corridor, the head was talking to Aihara and others that it was hard for
the manager to deal with calculation. ’
80 Chapter 5. Experiment 1
c. Aihara-san-tachi-ni shunin-ga kodomo-ga nigate-da-to
Aihara-title-and others-dat manager-nom children-nom hard to deal with - that
buchoo-ga Rooka-de hanashite-ita
head-nom corridor-loc was telling
‘The head was talking to Aihara and others in the corridor that it was hard for
the manager to deal with calculation.’
d. Aihara-san-tachi-ni shunin-ga keisan-ga nigate-da-to
Aihara-title-and others-dat manager-nom calculation-nom hard to deal with - that
buchoo-ga Rooka-de hanashite-ita
head-nom corridor-loc was telling
‘The head was talking to Aihara and others in the corridor that it was hard for
the manager to deal with calculation.’
(Muraoka & Sakamoto, 2004)
First, the ga-ga sequences in the previous study were preceded by a locative (e.g., rooka-
de ‘in the corridor’ in examples (9a, b) in Muraoka & Sakamoto, 2004) whereas in our
experiment there was no adjunct preceding the ga-ga fragments. The locative is important
because it can bias the expectations for the kind of predicate that will come next (for
example, a locative is usually not compatible with stative predicates).
But perhaps most important is that the previous study used saturated nouns (e.g.,
keisan ‘calculation’ in examples (9b, d) in Muraoka & Sakamoto, 2004) in which case, its
lack of difference in the ga-ga fragments is compatible with the present study given that we
see no animacy effect in the ga-ga conditions when the second NP-ga was saturated (the
difference in animacy in the ga-ga conditions are between the unsaturated nouns).
In sum, the comparison between the present study and previous reading time results
indicates that they are compatible once we factor in saturation effects and, therefore, it is
likely that the effects detected here should be observable in reading time measurements as
well.
5.5 Conclusion
We showed our result from the completion task in this chapter. There were fewer com-
pletions (37 cases) in which the two nominative NPs belonged to the same clause than to
different clauses (706 cases) in the result. That confirms previous results according to which
readers prefer to interpret each nominative NP as part of a separate clause (Miyamoto, 2002;
Uehara, 1997; Uehara & Bradley, 2002).
Section 5.5. Conclusion 81
However, case (in NP-ni NP-ga NP-ga and NP-ga NP-ga) did not interact with either
of the other two factors.
There was an interaction between saturation and animacy. Animacy only has an effect
on clause boundary insertion when the second nominative NP is unsaturated. That is, there
was no animacy effect in the ga-ga conditions when the second NP-ga was saturated (the
difference in animacy in the ga-ga conditions are between the unsaturated nouns). It thus
seems that Animacy is modulated by Saturation of Nouns.
The present results are compatible with a model in which NPs are interpreted together
in complex associations, taking into account not only local information such as case marking
and animacy, but also what we called inter-nominal relations, in other words, relations that
hold between NPs.
Chapter 6 Experiment 2
6.1 Introduction
Because the saturated NPs and unsaturated NPs in Experiment 1 are different in a number
of ways (e.g., ‘mother’ compared to ‘that pianist’; ‘goal’ compared to ‘tonight’s starlit sky’),
it is difficult to argue that factors other than the possessor relation are not contributing to
the effects reported. Therefore, in this experiment, we used only unsaturated head nouns
and manipulated whether they were modified or not by a possessor NP. In this way, we can
determine the effects of possessor relations more precisely by keeping other factors constant.
6.2 Method
6.2.1 Participants
Twenty-four native speakers of Japanese, undergraduate students at Kobe Shoin Women’s
University, who had not taken part in the first experiment, volunteered to participate.
6.2.2 Materials
There were 20 sets of four fragments in a 2×2 design, which participants were asked to
complete into sentences. The first nominative NP was always a human proper name. The
first factor manipulated was animacy, as the second nominative NP was animate as in (137a)
and (137c) or inanimate as in (137b) and (137d). The second factor was saturation which
was determined by the presence of a genitive NP (either a proper name or jibun ‘oneself’)
preceding the second NP-ga as its possessor (as in (137c) and (137d)) or its absence (as in
(137a) and (137b); see Appendix B for the complete list of fragments used).
(137) a. NP-ga NP-ga (Unsaturated / Animate)
Akira-kun-ga hahaoya-ga
Akira-title-nom mother-nom
b. NP-ga NP-ga (Unsaturated / Inanimate)
Akira-kun-ga mokuhyoo-ga
Akira-title-nom goal-nom
c. NP-ga NP-no NP-ga (Saturated / Animate)
Akira-kun-ga Yumi-chan-no hahaoya-ga
Akira-title-nom Yumi-title-gen mother-nom
82
Section 6.2. Method 83
d. NP-ga NP-no NP-ga (Saturated / Inanimate)
Akira-kun-ga Yumi-chan-no mokuhyoo-ga
Akira-title-nom Yumi-title-gen goal-nom
6.2.3 Procedure
The 20 test sets were distributed into four lists following a Latin Square design, and 32 filler
fragments were included in a total of 52 fragments per list. Other than that the procedure
was the same as in the first experiment. The experiment took participants about 30 minutes
to complete.
6.2.4 Analysis
The analysis was the same adopted for the previous experiment, except that here the com-
pletions were classified into 16 classes. Apart from the classes used in Experiment 1, class
(J2) was added for cases in which the possessor jibun coreferred with the speaker (in gen-
eral, jibun is preferentially interpreted as coreferring with the first nominative NP in the
fragments; see example (138) also Table 13 for the complete list of classes used).
Class (J2) was added in the analysis of Experiment 2 for cases in which the possessor
jibun coreferred with the speaker as in the sentence below (in general, jibun is preferentially
interpreted as coreferring with the first nominative NP in the fragments).
(138) [Kuroda-sensei-ga [jibun-no sakuhin-ga / ichiban umai]-to ittekureta.]
Kuroda-teacher-nom oneself work of art-nom the best great-that said
‘Kuroda said that (my) art work was the best.’
84 Chapter 6. Experiment 2
Table 13: Schematic representations for the classes used to analyze the completions.The column Clause indicates whether the two nominative NPs were classified as partof the same clause or not. The boxed constituents correspond to the two nominativeNPs.
Code Clause Description
A Same [ possessor subject V ]B Same [ subject object V ]C Same [ subject subject V ]D Same [ [ possessor subject V ] subject V ]D2 Same [ [ subject object V ] subject V ]E Same [ prospeaker [ possessor subject V ] V ]E2 Same [ prospeaker [ subject object V ] V ]F Different [ subject [ subject V ] V ]G Different [ subject [
RCsubject gapi(object) V ] NPi V ]
G2 Different [ subject [RC
gapi(possessor) subject V ] NPi V ]H Different [ subjecti [ proi(subj) object V ] V ]I Different [ subjecti [ proi(poss) subject V ] V ]J Different [ subject [ prospeaker(poss) subject V ] V ]J2 Different [ subject [ jibun-nospeaker(poss) subject V ] V ]K Different [ subject [ prosomebody(poss) subject V ] V ]L Different [ subject [ subject V(existential)] V ]X — Eliminated: Completely ambiguousY — Eliminated: UngrammaticalZ — Eliminated: Incomplete sentence
Section 6.3. Results 85
6.3 Results
Out of the 480 sentence completion responses obtained (20 fragments × 24 participants),
there were 15 (3.125%) of types (X), (Y) and (Z) and they were not included in further
analyses. All percentages reported hereafter exclude these cases.
The remaining 465 sentence completion responses were analyzed to examine their clause
boundary placements in the same way as in Experiment 1.
There were more completions in which the two nominative NPs belonged to differ-
ent clauses (408 instances) than completions in which they were in the same clause (57
instances), replicating once more the preference that readers have for interpreting each
nominative NP as part of a separate clause. See Table 14 for the results for each class and
Table 15 for the results according to animacy and saturation.
Table 14: Number of completions in each of the 16 classes in Experiment 2.
Same clause Different clauses Eliminated
Codes A B C D D2 F G G2 H I J J2 K L X Y Z
Unsaturated/Animate 2 2 9 0 1 2 3 0 16 71 3 0 3 4 1 2 1Unsaturated/Inanimate 27 0 0 8 0 2 0 10 18 29 10 0 9 4 2 1 0
Saturated/Animate 0 3 3 0 0 91 3 0 11 0 1 0 0 0 1 5 2Saturated/Inanimate 0 1 0 0 1 102 0 0 15 0 0 1 0 0 0 0 0
Table 15: Number of completions (and percentages) according to whether the twonominative NPs were in the same clause or in different clauses in Experiment 2.
Same clause Different clauses Subtotals
Unsaturated/Animate 14 (12.1%) 102 (87.9%) 116Unsaturated/Inanimate 35 (29.9%) 82 (70.1%) 117
Saturated/Animate 6 (5.4%) 106 (94.6%) 112Saturated/Inanimate 2 (1.7%) 118 (98.3%) 120
Subtotals 57 (12.3%) 408 (87.7%) 465
Trends similar to those observed for the ga-ga sequences in the previous experiment were
obtained here. In particular, the proportion of same-clause completions for the two Unsat-
urated conditions (Inanimate over Animate: 35/14) was larger than for the two Saturated
conditions (2/6; FPET: P < 0.05).
86 Chapter 6. Experiment 2
14
35
62
102
82
106
118
0
20
40
60
80
100
120
140
Unsat/Anim Unsat/Inan Sat/Anim Sat/Inan
Nu
mb
er o
f co
mp
leti
on
s
same clause
differentclauses
Figure 5: Number of completions (and percentages) according to whether the two nomina-tive NPs were in the same clause or in different clauses in Experiment 2.
Moreover, in the two Unsaturated conditions, the proportion of same-clause to different-
clause completions was larger for the Inanimate (35/82) than for the Animate condition
(14/102; FPET: P < 0.001). But for the two Saturated conditions, the proportion for the
Inanimate (2/118) and the Animate conditions (6/106) did not differ (P>0.118).
Table 16: Number of completions with NP-ga NP-ga in the same clause
Saturated Unsaturated
Animate 6 (5.4%) 14 (12.1%)
Inanimate 2 (1.7%) 36 (29.9%)
6.4 Discussion
We replicated the results of the first experiment, but in the present case with the same
nominative NPs across the conditions and by manipulating saturation through the insertion
of a possessor NP. The result provides evidence that it is saturation, and not some other
property of the NPs used, that is leading to the effects in clause-boundary interpretations
reported.
Section 6.4. Discussion 87
14
6
36
2
0
5
10
15
20
25
30
35
40
saturated unsaturated
Num
ber
of
com
ple
tions
animate
inanimate
****
****
NSNSNSNS
Figure 6: Number of completions with NP-ga NP-ga in the same clause
One concern with Experiments 1 and 2 is that some speakers of Japanese consider the
occurrence of two nominate NPs in the same clause to be marked or even ungrammatical.
This could have affected the completions, although there is no reason to believe that it
would have interacted with the effects of saturation and animacy reported. Moreover, the
properties of this type of construction have been extensively researched under the assump-
tion that they are grammatical (Kuno, 1973; Kuroda, 1992; Shibatani, 1977; Shibatani &
Cotton, 1976-77; Vermeulen, 2005; also see chapter 2 in this thesis).
The difficulty with multiple nominative NPs does not seem to be restricted to cases
in which they occur in the same clause. Reports indicate that there is also difficulty even
when the nominative NPs are part of different clauses in the sentence and that the difficulty
decreases if the first nominative marker is replaced with the topic marker (Babyonyshev &
Gibson, 1999; Miyamoto, 2002; Uehara, 1997). It is possible that there is a dispreference
for multiple nominative NPs because of phonological confusability (Baddeley, 1992, on the
phonological loop and the acoustic similarity effect; also similarity-based interference in
Nakayama, Vasishth & Lewis, 2006). Results from Korean tend to support this assumption
as sentences with multiple nominative markers are easier to understand if the nominative
markers used are phonologically distinct (Hagstrom & Rhee, 1997; Lee and Nakayama,
88 Chapter 6. Experiment 2
2003; but see Uehara & Bradley, 2002, Experiment 2).
Although the dispreference for multiple nominative NPs in the same clause is unlikely
to affect the saturation effects investigated, a corpus count and a questionnaire study were
conducted in order to address this possibility.
6.5 Corpus analysis
Sixteen sentences that have two adjacent nominative NPs in the same clause were automat-
ically retrieved from the Kyoto Corpus. According to a manual check the following eleven
are relevant. Each sentence is preceded by its identification label in the corpus and the type
of the second nominative NP according to saturation and animacy (the two nominative
NPs are underlined). Note that in seven of the relevant cases, the first NP-ga contains the
comparative expression no hou-ga, suggesting that there are indeed very few cases of two
successive nominative NPs in the same clause of the types used in the first two experiments.
(139) a. S-ID:950104186-030 Unsaturated/Inanimate
「十代でデビューしたころは、結婚が最終ゴールと考えていた時期がありましたけど、
結婚願望も変わってきました。もう私に普通の主婦はできませんよね。女優という仕事が奥が
深いことが、ますます分かってきましたし、好きな仕事。私って、あまり器用じゃ
ないので、自然に任せて、結婚を別に急ごうとは思いません」
b. S-ID:950105076-006 Unsaturated/Inanimtate
これから作る従業員十六人の子会社のほうが格が上とは世にも不思議な格付けだ。
c. S-ID:950107147-001 Unsaturated/Inanimate
東京都武蔵野市内で保護された若い女性が、記憶がはっきりせず関係者を捜してい
る。
d. S-ID:950110031-069 Unsaturated/Inanimate
世界では、アメリカ人より日本人の方が信用があるところがけっこう多いので、こ
れは損をする。
e. S-ID:950111272-008 Unsaturated/Inanimate
アンケートに基づく低密度者と高密度者のライフスタイル比較では、
「クラブ活動などでの運動歴」が低密度者が平均一・六年なのに対し、高密度者は倍
以上の三・八年。
f. S-ID:950116142-023 Saturated/Inanimate
全日本アマチュア野球連盟の山本英一郎副会長は「太陽の恵みに慣れっこな南の地
Section 6.5. Corpus analysis 89
方より、雪に閉ざされる北国の方が野球熱が高いということだろう」と風土の影響
を指摘。
g. S-ID:950117208-012 Saturated/Inanimate
抑えたほうが悲しみが伝わる。
h. S-ID:950121046-019 Unsaturated/Inanimate
もちろん今回の方がメリハリが利いているところもある。
i. S-ID:950521035-022 Unsaturated/Inanimate
文部省の管理主義が度が過ぎるため、このような雑務があまりに多く、教師が生徒
と語り合う時間は極めて少なくなっているのが、今の学校の実態である。
j. S-ID:950603049-022 Unsaturated/Inanimate
従来の講義型より、この参加・実践型の方が効果がありそうだ。
k. S-ID:950928047-023 Unsaturated/Inanimate
米国では千五百億ドルかかったが、日本の不良債権の方が規模が大きいのだ。
The following are the five cases that were found to be irrelevant according to a manual
check.
(140) a. S-ID:950110181-003 この連載をテキストに使った大学もあり、執筆した記者は「3
時間待ちの3分間診療に、大学病院への信仰とおごりが。意識改革が先決」と。
b. S-ID:950110198-003 大学は理科系が八七・六%、文科系が七五・九%と理科系の方
が内定率が高く、男女別では男子八四・六%、女子六九・四%。
c. S-ID:950206068-003例えば北海道小樽市のある小学校では、毎月二回、土曜日を「総
合学習」にあて、「バナナの旅」「チョコレートの一粒が」「ナタデココブームが去っ
て」などと名付けた学習を行い、子どもたちに平和や人権、環境問題を考えさせた
という。
d. S-ID:950513048-005 この条約の当初の目的は、核兵器を保有する大国が、核兵器保
有国が増えることを防止することにあった。
e. S-ID:950905049-012 だが、日の丸・君が代問題ばかりに論議が集中することが果た
して望ましかったのだろうか。
90 Chapter 6. Experiment 2
6.6 Conclusion
In this chapter we presented our results of experiment 2. We replicated the results of the
first experiment, however in the present case with the same nominative NPs across the
conditions and by manipulating saturation through the insertion of a possessor NP. The
result provides evidence that it is saturation, and not some other property of the NPs used,
that is leading to the effects in clause-boundary interpretations reported.
The effect of animacy was shown to be modulated by the possible associations that can
be created between NPs, namely possessor-possessee relations. Also the second model was
supported.
Chapter 7 Experiment 3
7.1 Introduction
The corpus count suggests that two nominative NPs in the same clause are rarely seen in
newspaper sentences. As observed earlier, whatever the reason for such a difficulty with
multiple nominative NPs, it should not interact with the findings of the two experiments
reported earlier. Nevertheless, a third questionnaire was conducted by replacing the first
nominative marker in the fragments of Experiment 2 with the topic marker in order to
verify if the findings in the previous two experiments would still stand.
7.2 Method
7.2.1 Participants
Twenty-four native speakers of Japanese, residents in the Kansai area, volunteered to par-
ticipate in the study.
7.2.2 Materials
The 20 sets of items of Experiment 2 were used with the nominative marker of the first NP
replaced with the topic marker wa sa shown in (141) (see Appendix C for the complete list
of fragments used).
(141) a. Unsaturated / Animate
Akira-kun-wa hahaoya-ga
Akira-title-top mother-nom
b. Unsaturated / Inanimate
Akira-kun-wa mokuhyoo-ga
Akira-title-top goal-nom
c. Saturated / Animate
Yooko-chan-wa ano pianisuto-ga
Yoko-title-top that pianist-nom
d. Saturated / Inanimate
Yooko-chan-wa konyano hoshizora-ga
Yoko-title-top tonight starlit sky-nom
91
92 Chapter 7. Experiment 3
7.2.3 Procedure and analysis
The procedure and the filler items used were identical to the ones for Experiment 2. The
classes used in the analyses were the same as in Experiment 1 and 2 (see Table 13) except
that all references to the first nominative NP should be replaced with this NP having the
wa marker instead.
ANOVAs are reported as the homogeneity of variance assumption was not violated. The
data were arcsine transformed before being submitted to ANOVAs which were conducted on
participant means (F1) and on item means (F2). Analyses using the untransformed numbers
revealed the same trends as the arcsine analyses reported. The percentages reported are the
actual percentages (not the arcsine transformed scores) obtained in the participant analyses.
7.3 Results
Out of the 480 sentence completion responses obtained (20 fragments × 24 participants),
there were 5 (1.04%) of types (X), (Y) and (Z) and they were not included in further
analyses. All percentages reported hereafter exclude these cases.
The remaining 475 sentence completion responses were analyzed to determine whether
the topic NP and the nominative NP were part of the same clause or different clauses.
7.3.1 Clausal structure
There were fewer completions in which the topic NP and the nominative NP belonged to
the same clause (222 instances) than completions in which they were in different clauses
(253 instances). But the proportion of same clause to different clauses was clearly larger
in the present experiment (222/253) than in Experiment 2 (57/408; FPET: P < 0.00001),
suggesting that the nominative NP is more likely to be part of the same clause when the
first NP has the topic marker rather than the nominative marker.
See Table 17 for the results for each class and Table 18 for the results according to
animacy and saturation. The main effects of saturation and animacy, as well as their
interaction were all reliable (P s < 0.01) as reported next.
There was a main effect of saturation as the Saturated conditions (i.e., the ones with the
possessor phrase) led to fewer same-clause completions than the Unsaturated ones (without
the possessor phrase; F1(1,23) = 45.19, P < 0.001; F2(1,19) = 49.81, P < 0.001). There
was also a main effect of animacy as the Inanimate conditions led to more same-clause
completions than the Animate ones (F1(1,23) = 45.88, P < 0.001; F2(1,19) = 20.94, P <
0.001).
Moreover, there was an interaction between animacy and saturation (F1(1,23) = 10.6,
P < 0.01; F2(1,19) = 10.17, P < 0.01) as the Unsaturated/Inanimate condition had more
Section 7.3. Results 93
Table 17: Number of completions in each of the 16 classes in Experiment 3.
Same clause Different clauses Eliminated
Codes A B D E F G G2 H I J L X Y Z
Unsaturated/Animate 9 40 0 0 11 2 0 4 47 2 3 1 0 1Unsaturated/Inanimate 89 4 1 4 0 0 3 3 14 1 0 0 1 0
Saturated/Animate 1 31 0 0 80 2 0 4 0 0 1 1 0 0Saturated/Inanimate 0 43 0 0 70 0 0 6 0 0 0 0 1 0
Table 18: Number of completions (and percentages in the ANOVAs by participants)according to whether the topic NP and the nominative NP were in the same clauseor in different clauses in Experiment 3.
Same clause Different clauses Subtotals
Unsaturated/Animate 49 (42.1%) 69 (57.9%) 118Unsaturated/Inanimate 98 (82.5%) 21 (17.5%) 119
Saturated/Animate 32 (27.3%) 87 (72.7%) 119Saturated/Inanimate 43 (36.7%) 76 (63.3%) 119
Subtotals 222 (46.7%) 253 (53.3%) 475
49
98
32
43
69
21
87
76
0
20
40
60
80
100
120
Unsat/Anim Unsat/Inan Sat/Anim Sat/Inan
Nu
mb
er o
f co
mp
leti
on
s
same clause
different clauses
Figure 7: Number of completions (and percentages in the ANOVAs by participants) accord-ing to whether the topic NP and the nominative NP were in the same clause or in differentclauses in Experiment 3.
94 Chapter 7. Experiment 3
same-clause completions than the Unsaturated/Animate (F1(1,23) = 44.9, P < 0.001;
F2(1,19) = 33.7, P < 0.001) but the two Saturated conditions did not differ (F1(1,23) =
2.77, P>0.1; F2(1,19) = 1.33, P>0.26).
Table 19: Number of completions with NP-wa NP-ga in the same clause
Saturated Unsaturated
Animate 32 (27.3%) 49 (42.1%)
Inanimate 43 (36.7%) 98 (82.5%)
32
98
49
43
0
20
40
60
80
100
120
saturated unsaturated
Num
ber
of
com
ple
tions
animate
inanimate
****
****
NSNSNSNS
Figure 8: Number of completions with NP-wa NP-ga in the same clause
7.3.2 Roles of the nominative NP
A second type of analysis was conducted in order to determine the effects of saturation and
animacy on the grammatical role (subject or object) of the nominative NP, regardless of
number of clauses involved.
In classes (B), (D2), (E2) and (H), the nominative NP (in Table 13, it is the second
NP-ga) has the role of an object, whereas in the remaining classes it has the subject role.
However, in this experiment, there were no instances of (D2) or (E2); therefore the object
Section 7.3. Results 95
role completions are restricted to (B) and (H). Table 20 summarizes the results.
Table 20: Number of completions according to the role of the nominative NP inExperiment 3.
Object Subject
Unsaturated/Animate 44 (37.3%) 74 (62.7%)Unsaturated/Inanimate 7 (5.8%) 112 (94.2%)
Saturated/Animate 35 (29.8%) 84 (70.2%)Saturated/Inanimate 49 (41.7%) 70 (58.3%)
Subtotals 135 (28.4%) 340 (71.6%)
There was a main effect of saturation as the Saturated conditions had fewer subject-role
completions than the Unsaturated conditions (F1(1,23) = 8.06, P < 0.01; F2(1,19) =
4.79, P < 0.05). There was also a main effect of animacy as the Inanimate conditions
led to more subject-role completions than the Animate conditions but the difference was
marginal in the participant analysis (F1(1,23) = 3.88, P = 0.061; F2(1,19) = 5.91,P <
0.05).
The interaction between saturation and animacy was also reliable (F1(1,23) = 29.62,
P < 0.001; F2(1,19) = 18.78, P < 0.001) as the Saturated/Animate condition had
more subject-role completions than the Saturated/Inanimate condition in the participant
analysis (F1(1,23) = 5.27, P < 0.05; F2(1,19) = 2.2, P = 0.155), whereas the Unsatu-
rated/Animate condition had fewer subject-role completions than the Unsaturated/Inanimate
condition (F1(1,23) = 19.03, P < 0.001; F2(1,19) = 39.77, P < 0.001). In other words,
the effects of animacy are once more modulated by saturation. In fact, animacy has oppo-
site effects depending on saturation. When saturated, animate nouns tend to be subjects
more often than inanimate nouns (although the difference is only reliable in the participant
analysis); however, when unsaturated, inanimate nouns tend to be subjects more often than
animate nouns. This last result is particularly interesting given that inanimate nouns as a
general rule are associated with the object role, but this clearly is not the case here.
In what follows, we discuss these results concentrating on classes (A) and (B). In class
(A), the NP-wa is the owner of the NP-ga which in turn is the subject of a one-place
predicate. In class (B), NP-wa is the subject and NP-ga is the object of a stative predicate.
In other words, in both classes, the two NPs are in the same clause but their grammatical
roles differ.
For the Saturated conditions in all three experiments, we only obtained the completion
96 Chapter 7. Experiment 3
0
20
40
60
80
100
120
Unsat/Anim Unsat/Inan Sat/Anim Sat/Inan
Nu
mb
er o
f co
mp
leti
on
s
Object
Subject
Figure 9: Number of completions according to the role of the nominative NP in Experiment3.
shown next for class (A) (in the examples, the fragment and the completion are separated
by a slash /).
(142) Saturated/Animate example for class (A)
Fujita-buchoo-wa jibun-no buka-ga / takusan imasu.
Fujita-director-top self-gen subordinates-nom many are
‘As for director Fujita, his own subordinates are many.’
When instead of jibun-no, the possessor is a person’s name, there are no completions of
type (A) with Saturated fragments. As observed by an anonymous reader, this is unsur-
prising given that there is no grammatical construction in which two distinct entities (the
NP-wa and the genitive phrase) can own the NP-ga. But even if class (A) in isolation is of
limited interest, when we consider it together with the distribution in class (B) it becomes
clear that saturation and animacy are interacting in interesting ways. Although completions
of class (B) are possible with all four types of fragments in this study, we see a small number
of such cases in the Unsaturated/Inanimate condition (4 instances) in comparison to the
other three conditions (40 for the Unsaturated/Animate, 43 for the Saturated/Inanimate,
31 for the Saturated/Animate).
Section 7.3. Results 97
In other words, unsaturated inanimate nouns are rarely assigned the object role and are
instead interpreted as the subject of the clause as attested by the large number of Unsat-
urated/Inanimate fragments (89 instances) in class (A), a class in which the completions
often have the NP-wa as a general topic and the NP-ga as its attribute as in (143a, b). The
preference for an inanimate noun to be interpreted as a subject in this case can only be
understood if one takes into consideration how the NP-ga is associated with the NP-wa and
the type of relation that holds between them.
(143) Unsaturated/Inanimate examples in class (A)
a. Kuroda-sensei-wa sakuhin-ga / koseiteki-da.
Kuroda-prof-top work-nom unique-is
‘Professor Kuroda’s work is unique.’
b. Kawasaki-san-wa seishin-ga / tafu-de-aru.
Kawasaki-title-top mind-nom tough-is
Lit.: ‘Kawasaki’s mind is tough.’ (‘Kawasaki is mentally tough.’)
The following is one of the few examples in which the NP-ga is an object (i.e., one of
the four instances in class (B) in Table 17).
(144) Unsaturated/Inanimate example in class (B)
Kuroda-sensei-wa sakuhin-ga / ki-ni-iranakatta.
Kuroda-prof-top work-nom be-not-pleased
‘Professor Kuroda was not pleased with the work.’
Compare those results with the Unsaturated/Animate completions. Although it is possi-
ble for them to lead to completions in class (A), as shown in (145a) below, the more common
pattern is class (B), namely, to have the nominative NP as the object of a so-called stative
predicate, as in (145b).
(145) a. Unsaturated/Animate example in class (A)
Kuroda-sensei-wa okusama-ga / utsukushii.
Kuroda-prof-top wife-nom beautiful
‘Professor Kuroda’s wife is beautiful.’
b. Unsaturated/Animate example in class (B)
Akira-kun-wa hahaoya-ga / daisukidesu.
Akira-title-top mother-nom adore.
‘Akira adores (his) mother.’
98 Chapter 7. Experiment 3
Thus, even when the animate NP-ga is interpreted as belonging to somebody (e.g.,
Akira’s mother), it seems to be independent enough that it can be interpreted as a direct
object (somebody that can be liked, disliked, adored, and so on). But with an inanimate
NP that belongs to somebody (e.g., ‘professor Kuroda’s work’), the tendency observed in
this study is to interpret it as an attribute of the NP-wa, and state a quality of it through
an adjective (as in (143a, b)). As should be clear, this is not simply an animacy effect, but
rather an interaction between animacy and saturation given that it arises from the distinct
ways in which animate and inanimate nouns behave in relation to the possessor NP-wa.
This result once more suggests that the animacy of the nominative NP is not inter-
preted in a vacuum and is dependent on inter-nominal relations such as possessor relations.
Although an inanimate NP is commonly a prime candidate to be interpreted as a direct
object (as suggested by the larger number of Saturated/Inanimate fragments in class (B)
compared to the Saturated/Animate fragments), it can be assigned different roles (e.g., as
a subject) if it can be associated with a prior entity in a possessor-possessee relation.
7.3.3 Length effects
An issue that needs consideration is that the saturated conditions were longer than the
unsaturated conditions in all three experiments. One could argue that longer NPs increase
memory load, thus it is the length of the NPs and not saturation or animacy that affected
clause boundary placement. One problem with this explanation is that extra memory load
should steer readers away from complex multi-clausal interpretations, in which case, readers
should avoid putting the two NPs in different clauses in the longer Saturated conditions,
the opposite of what we observed in the experiments.
An alternative way in which length can affect clause boundary insertion is in terms of
implicit prosodic contours that may be imposed on fragments read in silence (Fodor, 1998).
It has been shown that when the initial NP of a fragment is long, it is more likely for a pause
(and other prosodic phrasal markers) to be inserted immediately after it, thence increasing
the likelihood for an embedded clause boundary to be inserted at that point (Hirose, 2003).
However, in the experiments reported here, a pause after a long NP-ga is irrelevant because
the clause boundary (when present) should be inserted before not after this NP. In other
words, the prosodic contours that could be conceivably created by the long NP-ga do not
coincide with the clause boundary position.
In any case, in order to verify the extent of possible, “length effects”, linear correlation
analyses were conducted for the data in Experiment 3 using the arcsine transformed scores
for the ANOVAs (similar trends were observed for the first two experiments). Phrasal length
Section 7.4. Discussion 99
of the fragments was measured in two ways, by number of characters and number of moras.
For each item in each condition, the average number of same clause interpretations was
paired up with the length of the nominative NP (including the possessor NP if there was
one in the fragment) and a reliable correlation was found (using number of characters: r =
-0.496, P < 0.0001; using number of moras: r = -0.49, P < 0.0001). In other words,
the longer the fragment, the fewer the number of same clause completions. Similar trends
were observed when the inanimate conditions (characters: r = -0.66, P < 0.0001; moras:
r = -0.673, P < 0.0001) and the animate conditions (characters: r = -0.41, P < 0.01;
moras: r = -0.393, P < 0.05) were considered separately. Since the Saturated conditions
tended to be longer, this correlation is to be expected as the two factors (saturation and
length) are confounded. That this correlation is the result of saturation and not of length
is suggested by the lack of reliable correlations in the following two analyses.
First, no correlation was found when the conditions with possessor (characters: r =
-0.004, P>0.98 ; moras: r = -0.129, P>0.42) and without possessor (characters: r =
-0.087, P>0.59; moras: r = 0.089, P>0.58) were analyzed separately. In other words,
when saturation level is kept constant, we see no effect of length.
Second, for each level of animacy of an item, the Saturated condition and Unsaturated
condition only differed in that the former contained an extra phrase, namely the genitive-
marked possessor NP. Therefore, a correlation was conducted taking the length of the
possessor NP and the difference in same-clause completions between the Saturated and
the Unsaturated conditions for each level of animacy. However, no correlation was found
(characters: r = -0.14, P>0.38; moras: r = -0.0002, P>0.99).
In short, although we see overall negative correlations between fragment length and
number of same clause interpretations, this seems to occur because the Saturated fragments
tended to be long and the Unsaturated fragments tended to be short. When we keep
saturation constant, we see no effect of length. Moreover, there were no reliable correlations
when the length differences between Saturated and Unsaturated conditions were considered.
Thus, length is unlikely to explain the effects observed in the experiments.
7.4 Discussion
Although we obtained remarkably larger number of same-clause completions in this experi-
ment compared to the first two experiments, the claim that saturation has a critical effect on
NP interpretation is once more confirmed. The interaction between saturation and animacy
in particular is in line with the results of the previous two experiments.
In general, it is preferred that the two nominative NPs are in the embedded clause if
100 Chapter 7. Experiment 3
the sentences have a stative predicate, namely they require the matrix clause. The pattern
of NP-wa NP-ga is the predicate for stative sentences are more likely to become a single
sentence, and it is difficult for them to be in the same clause if there is a matrix clause as
in (146).
(146) a. [[ Naomi-ga Ken-ga suki na ] koto-ga minna-o odorokaseta. ]
Naomi-nom Ken-nom like -comp-nom everyone-acc be surprised
‘Everyone is surprised by the fact that Naomi likes Ken.’
b. [Naomi-wa Ken-ga suki da. ]
Naomi-top Ken-nom like
‘Naomi likes Ken.’
c. * [[Naomi-wa Ken-ga suki na ] koto-ga minna-o odorokaseta. ]
Naomi-top Ken-nom like -comp-nom everyone-acc be surprised
‘Everyone is surprised by the fact that Naomi likes Ken.’
Therefore, this fact might induce the result of a large number of same-clause completions
in this experiment compared to the first two experiments.
Previous results according to which a topic NP and a nominative NP are preferentially
taken to be part of different clauses (e.g., Miyamoto, 2002; Uehara, 1997) need to be
interpreted with the qualification that they were obtained with saturated animate nouns.
With unsaturated inanimate nouns, we see a reversal in preference with more same-clause
completions (98 out of 119 completions).
The present results together with Experiment 2 suggest that there is a strong preference
for interpreting a nominative NP as a subject, in light of the large number of different
clause interpretations in Experiment 2, which naturally follows if each nominative NP is
preferentially assigned the subject role. But, as observed earlier, there is also the possibility
that phonological confusability leads to the preference to have each NP-ga in a separate
clause. There is a clear need to investigate in more detail the nature of the difficulty with
multiple nominative NPs. But for present purposes, the crucial observation is that this
difficulty does not interact with saturation and animacy effects, which are observed both in
NP-ga-NP-ga sequences as well as NP-wa-NP-ga sequences.
7.5 Conclusion
The results of three questionnaire studies suggest that readers interpret NPs by taking into
consideration not only the individual properties of each NP but also possible inter-nominal
Section 7.5. Conclusion 101
relations, that is relations that can hold between NPs. We believe that this is another step
toward understanding how readers process NPs without the mediation of an overt predicate.
The present experiments replicate previous studies by providing evidence that case markers
and in particular nominative markers are critical in order to determine clause boundaries
(Miyamoto, 2002; Uehara & Bradley, 2002), and that animacy also has an effect (as in
Muraoka & Sakamoto, 2004). However, the effect of animacy was shown to be modulated
by the possible associations that can be created between NPs, namely possessor-possessee
relations.
The present results are compatible with a model in which NPs are interpreted together
in complex associations, taking into account not only local information such as case marking
and animacy, but also what we called inter-nominal relations, in other words, relations that
hold between NPs. Given that the fragments presented to test contained no predicate, the
result suggests that such relations affect NP interpretation even before a predicate is seen.
However, the conclusion is tentative at this point. As discussed in the introduction, there is
the question of the kinds of information that are used during sentence comprehension and
there is the question of how soon such types of information are used. Due to limitations
inherent to the questionnaire methodology used, the experiments reported here can only
address the first question, namely the results indicate that possible inter-nominal associ-
ations (such as possessor-possessee relations) influence the way how sequences of NPs are
interpreted. In the future, it would be of interest to investigate how soon these types of
relations come into play as the sentence is read.
Although questionnaire studies have the disadvantage of not being informative in rela-
tion to temporal aspects of processing, they can in some other ways provide more informa-
tion compared to on-line measurements. Reading-time experiments can indicate whether a
sentence type is read faster than another, but it is difficult to make detailed claims based on
the magnitude of the difference. In contrast, based on the completion studies reported here,
we can for example not only claim that each nominative NP is preferentially interpreted as
part of a different clause, but we can also claim that the magnitude of this preference is very
strong (in the weakest case the preference was 69.58% in the Unsaturated/Animate condi-
tion of Experiment 2, going up to 98.33% in the Saturated/Inanimate condition), although
it becomes considerably weaker or can even be reversed when the first nominative marker
is replaced by a topic marker (Experiment 3). Another advantage of questionnaires is that
they can provide an idea of the alternative interpretations that readers are considering when
reading sentence fragments. In the present case, the completions allowed us to investigate
in detail the kinds of interpretations that Japanese speakers are considering after reading a
102 Chapter 7. Experiment 3
sequence containing two NP-ga phrases and, in particular, under what circumstances hear-
ers are less willing to insert a clause boundary. The results also allowed us to show that the
bias to interpret an inanimate noun as a direct object can be overridden when a possessor
relation is established.
Chapter 8 Conclusion
8.1 Summary
In chapter 2, we looked at some properties of a nominative marker ga which are relevant
to our present experiments. In general, the nominative marker -ga in Japanese marks a
subject, and but can also mark the object. We discussed examples of the sentences which
contain ga marked noun phrases. We saw the properties of descriptive ga, exhaustive-listing
ga and objective ga, and also looked at the ga ga constructions with the subjective ga and
the objective ga. We explained how some of the ga ga constructions can change to ni ga
constructions and wa ga constructions.
In chapter 3, we saw what factor causes a difficulty in processing of Japanese sentences
by looking at some recent works that are related with the present work.
From the perspective of incrementality of Japanese sentence processing, we began with
Yamashita (1978)’s work, which examined a scrambling effect in Japanese. However, she did
not find a scrambling effect. She suggested that one crucial source of information during
this process is case marking. In particular, because of their special syntactic properties,
nominative NPs have been claimed to indicate the beginning of a new clause, although
nominative markers have many different syntactic functions (Miyamoto, 2002, Uehara &
Bradley, 2002). Miyamoto (2002) proposed that the nominative marker is the indicator of
the clause boundary, and proposed the Local Assignment of Clause boundaries. We saw
how the work of Uehara and Bradley (2002) suggests that each NP-ga is processed with
a each different clause, in the construction of nominative case marker repetition, such as
NP-ga NP-ga, that is, NP-ga is a single clause subject.
It is also suggested by Muraoka & Sakamoto (2004) that animacy is an important factor
in determining whether a nominative NP is interpreted as a subject or an object.
In chapter 4, we first discussed about the importantance of treating noun phrases care-
fully in sentence processing experiments. In order to classify the common noun phrases,
we looked into the distinction between “saturated nouns” and “unsaturated nouns” in
Nishiyama (1990)’s proposal.
Then we examined the relation between saturation and ARG-ST from Gunji (2002)’s
analysis of relational nouns. That suggests that some nouns have their own arguments the
same as verbs do. In order to examine the phenomenon for relational nouns, we also looked
103
104 Chapter 8. Conclusion
at some examples of relational nouns in the NP1 no NP2 construction in Harada (1991),
and also in Kikuchi and Shirai (2006). We then defined how we treat “saturation” and
relational nouns in our experiments. We use the term saturation; however the actual items
included typical relational nouns in the experiments.
In chapter 5, we showed our results on the sentence completion task in this chapter.
There were fewer completions in which the two nominative NPs belonged to the same clause
than to different clauses in the result. That confirmed previous results according to which
readers prefer to interpret each nominative NP as part of a separate clause (Miyamoto,
2002; Uehara, 1997; Uehara & Bradley, 2002).
Animacy only had an effect on clause boundary insertion when the second nominative
NP is unsaturated. That is, no animacy effect was observed in the ga-ga conditions when the
second NP-ga was saturated. (The difference in animacy in the ga-ga conditions was between
the unsaturated nouns. Therefore, it seemed that Animacy is modulated by Saturation.
The present results are compatible with a model in which NPs are interpreted together
in complex associations, taking into account not only local information such as case marking
and animacy, but also what we called inter-nominal relations, in other words, relations that
hold between NPs.
In chapter 6, we presented our results of experiment 2. We replicated the results of
the first experiment, however in the present case with the same nominative NPs across the
conditions and by manipulating saturation through the insertion of a possessor NP. The
result provided evidence that it is saturation, and not some other property of the NPs used,
that is leading to the effects in clause-boundary interpretations reported.
The effect of animacy was shown to be modulated by the possible associations that
can be created between NPs, namely possessor-possessee relations. In the way, the second
model was supported.
In the chapter 7, we presented our results of experiment 3. We replicated the results of
the previous two experiments, but we obtained a large number of same clause interpretation.
However, we also could find the saturation effect here.
The results are compatible with a model in which NPs are interpreted together in com-
plex associations, taking not only local information such as case marking and animacy,
but also what we called inter-nominal relations, in other words, relations that hold be-
tween NPs. Given that the fragments contained no predicate, the result suggests that such
relations affect NP interpretation even before a predicate is seen.
Appendix A Fragments used in Experiment 1
The following are the nouns used in Experiment 1. Items 1 to 12 were the Unsaturated
items, and items 13 to 24 were the Saturated ones.
1. 黒田先生が / 奥様が / 作品が / 岡田さんに
Kuroda-sensei-ga / okusama-ga / sakuhin-ga / Okada-san-ni
Kuroda-title-nom / wife-nom / work of art-nom / Okada-title-dat
(a) Ga-ga/Unsaturated/Animate: Kuroda-sensei-ga okusama-ga
(b) Ga-ga/Unsaturated/Inanimate: Kuroda-sensei-ga sakuhin-ga
(c) Ni-ga-ga/Unsaturated/Animate: Okada-san-ni Kuroda-sensei-ga okusama-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Okada-san-ni Kuroda-sensei-ga sakuhin-ga
2. 松本先生が / 両親が / 気質が / 本田さんに
Matsumoto-sensei-ga / ryoushin-ga / kishitsu-ga / Honda-san-ni
Matsumoto-title-nom / parents-nom / temperament-nom / Honda-title-dat
(a) Ga-ga/Unsaturated/Animate: Matsumoto-sensei-ga ryoushin-ga
(b) Ga-ga/Unsaturated/Inanimate: Matsumoto-sensei-ga kishitsu-ga
(c) Ni-ga-ga/Unsaturated/Animate: Honda-san-ni Matsumoto-sensei-ga ryoushin-
ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Honda-san-ni Matsumoto-sensei-ga kishitsu-
ga
3. あきら君が / 母親が / 目標が / 美智子ちゃんに
Akira-kun-ga / hahaoya-ga / mokuhyoo-ga / Michiko-chan-ni
Akira-title-nom / mother-nom / goal-nom / Michiko-title-dat
(a) Ga-ga/Unsaturated/Animate: Akira-kun-ga hahaoya-ga
(b) Ga-ga/Unsaturated/Inanimate: Akira-kun-ga mokuhyoo-ga
(c) Ni-ga-ga/Unsaturated/Animate: Michiko-chan-ni Akira-kun-ga hahaoya-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Michiko-chan-ni Akira-kun-ga mokuhyoo-ga
105
106 Appendix A. Fragments used in Experiment 1
4. 河崎さんが / 娘が / 精神が / 後藤部長に
Kawasaki-san-ga / musume-ga / seishin-ga / Gotoo-buchoo-ni
Kawasaki-title-nom / daughter-nom / spirit-nom / Goto-chief-dat
(a) Ga-ga/Unsaturated/Animate: Kawasaki-san-ga musume-ga
(b) Ga-ga/Unsaturated/Inanimate: Kawasaki-san-ga seishin-ga
(c) Ni-ga-ga/Unsaturated/Animate: Gotoo-buchoo-ni Kawasaki-san-ga musume-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Gotoo-buchoo-ni Kawasaki-san-ga seishin-ga
5. 岸田君が / 長男が / 将来が / 中島部長に
Kishida-kun-ga / choonan-ga / shoorai-ga / Nakajima-buchoo-ni
Kishida-title-nom / eldest son-nom / future-nom / Nakajima-chief-dat
(a) Ga-ga/Unsaturated/Animate: Kishida-kun-ga choonan-ga
(b) Ga-ga/Unsaturated/Inanimate: Kishida-kun-ga shoorai-ga
(c) Ni-ga-ga/Unsaturated/Animate: Nakajima-buchoo-ni Kishida-kun-ga choonan-
ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Nakajima-buchoo-ni Kishida-kun-ga shoorai-
ga
6. 鈴木先輩が / いとこが / 行動が / 丸山さんに
Suzuki-senpai-ga / itoko-ga / koodoo-ga / Maruyama-san-ni
Suzuki-title-nom / cousin-nom / action-nom / Maruyama-title-dat
(a) Ga-ga/Unsaturated/Animate: Suzuki-senpai-ga itoko-ga
(b) Ga-ga/Unsaturated/Inanimate: Suzuki-senpai-ga koodoo-ga
(c) Ni-ga-ga/Unsaturated/Animate: Maruyama-san-ni Suzuki-senpai-ga itoko-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Maruyama-san-ni Suzuki-senpai-ga koodoo-
ga
7. 中村さんが / 師匠が / こころが / 北村先生に
Nakamura-san-ga / shishoo-ga / kokoro-ga / Kitamura-sensei-ni
Nakamura-title-nom / master-nom / mind-nom / Kitamura-title-dat
(a) Ga-ga/Unsaturated/Animate: Nakamura-san-ga shishoo-ga
(b) Ga-ga/Unsaturated/Inanimate: Nakamura-san-ga kokoro-ga
(c) Ni-ga-ga/Unsaturated/Animate: Kitamura-sensei-ni Nakamura-san-ga shishoo-
ga
107
(d) Ni-ga-ga/Unsaturated/Inanimate: Kitamura-sensei-ni Nakamura-san-ga kokoro-
ga
8. 山崎係長が / 恩師が / 考えが / 池田さんに
Ymazaki-kakarichoo-ga / onshi-ga / kangae-ga / Ikeda-san-ni
Yamazaki-chief-nom / (former) teacher-nom / idea-nom / Ikeda-title-dat
(a) Ga-ga/Unsaturated/Animate: Ymazaki-kakarichoo-ga onshi-ga
(b) Ga-ga/Unsaturated/Inanimate: Ymazaki-kakarichoo-ga kangae-ga
(c) Ni-ga-ga/Unsaturated/Animate: Ikeda-san-ni Ymazaki-kakarichoo-ga onshi-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Ikeda-san-ni Ymazaki-kakarichoo-ga kangae-
ga
9. 田中先生が / 弟子が / 頭が / 西村さんに
Tanaka-sensei-ga / deshi-ga / atama-ga / Nishimura-san-ni
Tanaka-title-nom / apprentice-nom / head-nom / Nishimura-title-dat
(a) Ga-ga/Unsaturated/Animate: Tanaka-sensei-ga deshi-ga
(b) Ga-ga/Unsaturated/Inanimate: Tanaka-sensei-ga atama-ga
(c) Ni-ga-ga/Unsaturated/Animate: Nishimura-san-ni Tanaka-sensei-ga deshi-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Nishimura-san-ni Tanaka-sensei-ga atama-ga
10. 野中君が / 上司が / 好みが / 平井さんに
Nomura-kun-ga / jooshi-ga / konomi-ga / Hirai-san-ni
Nomura-title-nom / boss-nom / taste-nom / Hirai-title-dat
(a) Ga-ga/Unsaturated/Animate: Nomura-kun-ga jooshi-ga
(b) Ga-ga/Unsaturated/Inanimate: Nomura-kun-ga konomi-ga
(c) Ni-ga-ga/Unsaturated/Animate: Hirai-san-ni Nomura-kun-ga jooshi-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Hirai-san-ni Nomura-kun-ga konomi-ga
11. 藤田部長が / 部下が / 体調が / 小山さんに
Fujita-buchoo-ga / buka-ga / taichoo-ga / Koyama-san-ni
Fujita-cheif-nom / subordinate-nom / physical condition-nom / Koyama-title-dat
(a) Ga-ga/Unsaturated/Animate: Fujita-buchoo-ga buka-ga
(b) Ga-ga/Unsaturated/Inanimate: Fujita-buchoo-ga taichoo-ga
(c) Ni-ga-ga/Unsaturated/Animate: Koyama-san-ni Fujita-buchoo-ga buka-ga
108 Appendix A. Fragments used in Experiment 1
(d) Ni-ga-ga/Unsaturated/Inanimate: Koyama-san-ni Fujita-buchoo-ga taichoo-ga
12. 小林君が / 先輩が / 態度が / 永井さんに
Kobayashi-kun-ga / senpai-ga / taido-ga / Nagai-san-ni
Kobayashi-title-nom / senior-nom / attitude-nom / Nagai-title-dat
(a) Ga-ga/Unsaturated/Animate: Kobayashi-kun-ga senpai-ga
(b) Ga-ga/Unsaturated/Inanimate: Kobayashi-kun-ga taido-ga
(c) Ni-ga-ga/Unsaturated/Animate: Nagai-san-ni Kobayashi-kun-ga senpai-ga
(d) Ni-ga-ga/Unsaturated/Inanimate: Nagai-san-ni Kobayashi-kun-ga taido-ga
13. 矢島さんが / この番組の タレントが / この絵画の 虹が / 萩原監督に
Yajima-san-ga / kono-bangumi-no tarento-ga / kono-kaiga-no niji-ga/
Yajima-title-nom / this TV program’s personality-nom / this picture’s rainbow-nom/
Hagiwara-kantoku-ni
Hagiwara-title-dat
(a) Ga-ga / Saturated / Animate: Yajima-san-ga kono-bangumi-no tarento-ga
(b) Ga-ga / Saturated / Inanimate: Yajima-san-ga kono-kaiga-no niji-ga
(c) Ni-ga-ga / Saturated / Animate: Hagiwara-kantoku-ni Yajima-san-ga kono-bangumi-
no tarento-ga
(d) Ni-ga-ga / Saturated / Inanimate: Hagiwara-kantoku-ni Yajima-san-ga kono-
kaiga-no niji-ga
14. 近藤さんが / あの 漫才師が / この スポーツが / 小沢君に
Kondoo-san-ga / ano manzaishi-ga / kono supootsu-ga/
Kondoo-title-nom / that comic duo-nom / this sports-nom/
Ozawa-kun-ni
Ozawa-title-dat
(a) Ga-ga / Saturated / Animate: Kondoo-san-ga ano manzaishi-ga
(b) Ga-ga / Saturated / Inanimate: Kondoo-san-ga kono supootsu-ga
(c) Ni-ga-ga / Saturated / Animate: Ozawa-kun-ni Kondoo-san-ga ano manzaishi-ga
(d) Ni-ga-ga / Saturated / Inanimate: Ozawa-kun-ni Kondoo-san-ga kono supootsu-
ga
109
15. 藤原君が / この 芸人が / 今夜の お月さまが / 木下さんに
Fujiwara-kun-ga / kono geinin-ga / konya-no otsukisama-ga /
Fujiwara-title-nom / this entertainer-nom / tonight moon-nom/
Kinoshita-san-ni
Kinoshita-title-dat
(a) Ga-ga / Saturated / Animate: Fujiwara-kun-ga kono geinin-ga
(b) Ga-ga / Saturated / Inanimate: Fujiwara-kun-ga konya-no otsukisama-ga
(c) Ni-ga-ga / Saturated / Animate: Kinoshita-san-ni Fujiwara-kun-ga kono geinin-
ga
(d) Ni-ga-ga / Saturated / Inanimate: Kinoshita-san-ni Fujiwara-kun-ga konya-no
otsukisama-ga
16. 飯田さんが / 看護婦が / 雑草が / 内藤君に
Iida-san-ga / kangofu-ga / zasoo-ga /
Iida-title-nom / nurse-nom / weeds-nom/
Naitoo-kun-ni
Naitoo-title-dat
(a) Ga-ga / Saturated / Animate: Iida-san-ga kangofu-ga
(b) Ga-ga / Saturated / Inanimate: Iida-san-ga zasoo-ga
(c) Ni-ga-ga / Saturated / Animate: Naitoo-kun-ni Iida-san-ga kangofu-ga
(d) Ni-ga-ga / Saturated / Inanimate: Naitoo-kun-ni Iida-san-ga zasoo-ga
17. 吉原君が / この プロレスラーが / 隅田川の 堤防が / 田口さんに
Yoshihara-kun-ga / kono puroresuraa-ga / Sumidagawa-no teiboo-ga /
Yoshihara-title-nom / this pro(fessional) wrestler-nom / Sumidariver’s dike-nom/
Taguchi-san-ni
Taguchi-title-dat
(a) Ga-ga / Saturated / Animate: Yoshihara-kun-ga kono puroresuraa-ga
(b) Ga-ga / Saturated / Inanimate: Yoshihara-kun-ga Sumidagawa-no teiboo-ga
(c) Ni-ga-ga / Saturated / Animate: Taguchi-san-ni Yoshihara-kun-ga kono puroresuraa-
gaが
(d) Ni-ga-ga / Saturated / Inanimate: Taguchi-san-ni Yoshihara-kun-ga Sumidagawa-
no teiboo-ga
110 Appendix A. Fragments used in Experiment 1
18. 川上社長が / あの 舞妓さんが / この 八重桜が / 早川さんに
Kawakami-shachoo-ga / ano maikosan-ga / kono yaezakura-ga /
Kawakami-president-nom / that apprentice geisha-nom / this cherry blossoms-nom/
Hayakawa-san-ni
Hayakawa-title-dat
(a) Ga-ga / Saturated / Animate: Kawakami-shachoo-ga ano maikosan-ga
(b) Ga-ga / Saturated / Inanimate: Kawakami-shachoo-ga kono yaezakura-ga
(c) Ni-ga-ga / Saturated / Animate: Hayakawa-san-ni Kawakami-shachoo-ga ano
maikosan-ga
(d) Ni-ga-ga / Saturated / Inanimate: Hayakawa-san-ni Kawakami-shachoo-ga kono
yaezakura-ga
19. 陽子ちゃんが / あの ピアニストが / 今夜の 星空が / 吉本君に
Yooko-chan-ga / ano pianisuto-ga / konya-no hoshizora-ga /
Yoko-title-nom / that pianist-nom / tonight starlit sky-nom/
Yoshimoto-kun-ni
Yoshimoto-title-dat
(a) Ga-ga / Saturated / Animate: Yooko-chan-ga ano pianisuto-ga
(b) Ga-ga / Saturated / Inanimate: Yooko-chan-ga konya-no hoshizora-ga
(c) Ni-ga-ga / Saturated / Animate: Yoshimoto-kun-ni Yooko-chan-ga ano pianisuto-
ga
(d) Ni-ga-ga / Saturated / Inanimate: Yoshimoto-kun-ni Yooko-chan-ga konya-no
hoshizora-ga
20. 内田さんが / あの バレリーナが / この美術館の 日本庭園が / 中川先輩に
Uchida-san-ga / ano bareriina-ga / kono-bijutsukan-no nihonteien-ga /
Uchida-title-nom / that ballerina-nom / this museum’s Japanese garden-nom/
Nakagawa-senpai-ni
Nakagawa-title-dat
(a) Ga-ga / Saturated / Animate: Uchida-san-ga ano bareriina-ga
(b) Ga-ga / Saturated / Inanimate: Uchida-san-ga kono-bijutsukan-no nihonteien-ga
(c) Ni-ga-ga / Saturated / Animate: Nakagawa-senpai-ni Uchida-san-ga ano bareriina-
ga
111
(d) Ni-ga-ga / Saturated / Inanimate: Nakagawa-senpai-ni Uchida-san-ga kono-
bijutsukan-no nihonteien-ga
21. 吉田さんが / この 手品師が / 今日の 夜空が / 須藤君に
Yoshida-san-ga / kono tejinashi-ga / kyoo-no yozora-ga /
Yoshida-title-nom / this magician-nom / tonight sky-nom/
Sudoo-kun-ni
Sudoo-title-dat
(a) Ga-ga / Saturated / Animate: Yoshida-san-ga kono tejinashi-ga
(b) Ga-ga / Saturated / Inanimate: Yoshida-san-ga kyoo-no yozora-ga
(c) Ni-ga-ga / Saturated / Animate: Sudoo-kun-ni Yoshida-san-ga kono tejinashi-ga
(d) Ni-ga-ga / Saturated / Inanimate: Sudoo-kun-ni Yoshida-san-ga kyoo-no yozora-
ga
22. 西山さんが / この 作曲家が / 鳥取県の 砂浜が / 石田先生に
Nishiyama-san-ga / kono sakkyokuka-ga / Tottoriken-no sunahama-ga /
Nishiyama-title-nom / this composer-nom / Tottori-prefecture’s beach-nom/
Ishida-sensei-ni
Ishida-title-dat
(a) Ga-ga / Saturated / Animate: Nishiyama-san-ga kono sakkyokuka-ga
(b) Ga-ga / Saturated / Inanimate: Nishiyama-san-ga Tottoriken-no sunahama-ga
(c) Ni-ga-ga / Saturated / Animate: Ishida-sensei-ni Nishiyama-san-ga kono sakkyokuka-
ga
(d) Ni-ga-ga / Saturated / Inanimate: Ishida-sensei-ni Nishiyama-san-ga Tottoriken-
no sunahama-ga
23. 斉藤さんが / この 登山家が / ここの 並木道が / 長谷川先生に
Saitoo-san-ga / kono tozanka-ga / kokono namikimiti-ga /
Saito-title-nom / this alpinist-nom / this avenue-nom/
Hasegawa-sensei-ni
Hasegawa-title-dat
(a) Ga-ga / Saturated / Animate: Saitoo-san-ga kono tozanka-ga
(b) Ga-ga / Saturated / Inanimate: Saitoo-san-ga kokono namikimiti-ga
112 Appendix A. Fragments used in Experiment 1
(c) Ni-ga-ga / Saturated / Animate: Hasegawa-sensei-ni Saitoo-san-ga kono tozanka-
ga
(d) Ni-ga-ga / Saturated / Inanimate: Hasegawa-sensei-ni Saitoo-san-ga kokono
namikimiti-ga
24. 岡本さんが / あの お年寄りが / ハワイの 気候が / 鈴木君に
Okamoto-san-ga / ano otoshiyori-ga / Hawai-no kikoo-ga /
Okamoto-title-nom / that old person-nom / Hawaii’s climate-nom/
Suzuki-kun-ni
Suzuki-title-dat
(a) Ga-ga / Saturated / Animate: Okamoto-san-ga ano otoshiyori-ga
(b) Ga-ga / Saturated / Inanimate: Okamoto-san-ga Hawai-no kikoo-ga
(c) Ni-ga-ga / Saturated / Animate: Suzuki-kun-ni Okamoto-san-ga ano otoshiyori-
ga
(d) Ni-ga-ga / Saturated / Inanimate: Suzuki-kun-ni Okamoto-san-ga Hawai-no
kikoo-ga
Appendix B Fragments used in Experiment 2
The following are the nouns used in Experiment 2 . The actual fragments used are provided
as follows.
1. 黒田先生が / 奥様が / 作品が / 自分の
Kuroda-sensei-ga / okusama-ga / sakuhin-ga / jibun-no
Kuroda-title-nom / wife-nom / work of art-nom / self-gen
(a) Unsaturated/Animate: Kuroda-sensei-ga okusama-ga
(b) Unsaturated/Inanimate: Kuroda-sensei-ga sakuhin-ga
(c) Saturated/Animate: Kuroda-sensei-ga jibun-no okusama-ga
(d) Saturated/Inanimate: Kuroda-sensei-ga jibun-no sakuhin-ga
2. 松本先生が / 両親が / 気質が / 伊藤さんの
Matsumoto-sensei-ga / ryoushin-ga / kishitu-ga / Itoo-san-no
Matsumoto-title-nom / parents-nom / temperament-nom / Ito-title-gen
(a) Unsaturated/Animate: Matsumoto-sensei-ga ryooshin-ga
(b) Unsaturated/Inanimate: Matsumoto-sensei-ga kishitsu-ga
(c) Saturated/Animate: Matsumoto-sensei-ga Itoo-san-no okusama-ga
(d) Saturated/Inanimate: Matsumoto-sensei-ga Itoo-san-no kishitsu-ga
3. あきら君が / 母親が / 目標が / ゆみちゃんの
Akira-kun-ga / hahaoya-ga / mokuhyoo-ga / Yumi-chan-no
Akira-title-nom / mother-nom / goal-nom / Yumi-title-gen
(a) Unsaturated/Animate: Akira-kun-ga hahaoya-ga
(b) Unsaturated/Inanimate: Akira-kun-ga mokuhyoo-ga
(c) Saturated/Animate: Akira-kun-ga Yumi-cahn-no hahaoya-ga
(d) Saturated/Inanimate: Akira-kun-ga Yumi-cahn-no mokuhyoo-ga
4. 河崎さんが / 娘が / 精神が / 森田課長の
Kawasaki-san-ga / musume-ga / seishin-ga / Morita-kachoo-no
Kawasaki-title-nom / daughter-nom / spirit-nom / Morita-title-gen
113
114 Appendix B. Fragments used in Experiment 2
(a) Unsaturated/Animate: Kawasaki-san-ga musume-ga
(b) Unsaturated/Inanimate: Kawasaki-san-ga seishin-ga
(c) Saturated/Animate: Kawasaki-san-ga Morita-kachoo-no musume-ga
(d) Saturated/Inanimate: Kawasaki-san-ga Morita-kachoo-no seishin-ga
5. 伊藤さんが / 孫が / 髪型が / 上田社長の
Itoo-san-ga / mago-ga / kamigata-ga / Ueda-shachoo-no
Ito-title-nom / grandchild-nom / hairstyle-nom / Ueda-president-title-gen
(a) Unsaturated/Animate: Itoo-san-ga mago-ga
(b) Unsaturated/Inanimate: Itoo-san-ga kamigata-ga
(c) Saturated/Animate: Itoo-san-ga Ueda-shachoo-no kamigata-ga
(d) Saturated/Inanimate: Itoo-san-ga Ueda-shachoo-no kamigata-ga
6. 岸田君が / 長男が / 将来が / 田村さんの
Kishida-kun-ga / choonan-ga / shoorai-ga / Tamura-san-no
Kishida-title-nom / eldest son-nom / future-nom / Tamura-title-gen
(a) Unsaturated/Animate: Kishida-kun-ga choonan-ga
(b) Unsaturated/Inanimate: Kishida-kun-ga syoorai-ga
(c) Saturated/Animate: Kishida-kun-ga Tamura-san-no choonan-ga
(d) Saturated/Inanimate: Kishida-kun-ga Tamura-san-no syoorai-ga
7. 清水さんが / 妹が / 言動が / つよし君の
Shimizu-san-ga / imooto-ga / gendoo-ga / Tsuyoshi-kun-no
Shimizu-title-nom / sister-nom / speech and action-nom / Tuyoshi-title-gen
(a) Unsaturated/Animate: Shimizu-san-ga imooto-ga
(b) Unsaturated/Inanimate: Shimizu-san-ga gendoo-ga
(c) Saturated/Animate: Shimizu-san-ga Tsuyoshi-kun-no imooto-ga
(d) Saturated/Inanimate: Shimizu-san-ga Tsuyoshi-kun-no gendoo-ga
8. 鈴木先輩が / いとこが / 行動が / 西川君の
Suzuki-senpai-ga / itoko-ga / koodoo-ga / Nishikawa-kun-no
Suzuki-title-nom / cousin-nom / action-nom / Nishikawa-title-gen
(a) Unsaturated/Animate: Suzuki-senpai-ga itoko-ga
115
(b) Unsaturated/Inanimate: Suzuki-senpai-ga koodoo-ga
(c) Saturated/Animate: Suzuki-senpai-ga Nishikawa-kun-no itoko-ga
(d) Saturated/Inanimate: Suzuki-senpai-ga Nishikawa-kun-no koodoo-ga
9. かよちゃんが / 恋人が / 性格が / 小倉さんの
Kayo-chan-ga / koibito-ga / seikaku-ga / Ogura-sun-no
Kayo-title-nom / boyfriend / girlfriend-nom / character-nom / Ogura-title-gen
(a) Unsaturated/Animate: Kayo-chan-ga koibito-ga
(b) Unsaturated/Inanimate: Kayo-chan-ga seikaku-ga
(c) Saturated/Animate: Kayo-chan-ga Ogura-san-no koibito-ga
(d) Saturated/Inanimate: Kayo-chan-ga Ogura-san-no seikaku-ga
10. 福田さんが / ペットが / 夢が / 吉村君の
Fukuda-san-ga / petto-ga / yume-ga / Yoshimura-kun-no
Fukuda-title-nom / pet-nom / dream-nom / Yoshimura-title-gen
(a) Unsaturated/Animate: Fukuda-san-ga petto-ga
(b) Unsaturated/Inanimate: Fukuda-san-ga yume-ga
(c) Saturated/Animate: Fukuda-san-ga Yoshimura-kun-no petto-ga
(d) Saturated/Inanimate: Fukuda-san-ga Yoshimura-kun-no yume-ga
11. 中村さんが / 師匠が / こころが / 大沢先輩の
Nakamura-san-ga / shishoo-ga / kokoro-ga / Oosawa-senpai-no
Nakamura-title-nom / master-nom / mind-nom / Oosawa-title-gen
(a) Unsaturated/Animate: Nakamura-san-ga shisyoo-ga
(b) Unsaturated/Inanimate: Nakamura-san-ga kokoro-ga
(c) Saturated/Animate: Nakamura-san-ga Oosawa-senpai-no shishoo-ga
(d) Saturated/Inanimate: Nakamura-san-ga Oosawa-senpai-no kokoro-ga
12. 山崎係長が / 恩師が / 考えが / 武田さんの
Yamasaki-kakarichoo-ga / onshi-ga / kanngae-ga / Takeda-san-no
Yamasaki-cheif-title-nom / (former)teacher-nom / idea-nom / Takeda-title-gen
(a) Unsaturated/Animate: Yamasaki-kakarichoo-ga onshi-ga
(b) Unsaturated/Inanimate: Yamasaki-kakarichoo-ga kangae-ga
116 Appendix B. Fragments used in Experiment 2
(c) Saturated/Animate: Yamasaki-kakarichoo-ga Takeda-san-no onshi-ga
(d) Saturated/Inanimate: Yamasaki-kakarichoo-ga Takeda-san-no kangae-ga
13. 田中先生が / 弟子が / 頭が / 西田さんの
Tanaka-sensei-ga / deshi-ga / atama-ga / Nishida-san-no
Tanaka-title-nom / pupil-nom / head-nom / Nishida-title-gen
(a) Unsaturated/Animate: Tanaka-sensei-ga deshi-ga
(b) Unsaturated/Inanimate: Tanaka-sensei-ga atama-ga
(c) Saturated/Animate: Tanaka-sensei-ga Nishida-san-no deshi-ga
(d) Saturated/Inanimate: Tanaka-sensei-ga Nishida-san-no atama-ga
14. 野中君が / 上司が / 好みが / 宮下さんの
Nonaka-kun-ga / jooshi-ga / konomi-ga / Miyashita-san-no
Nonaka-title-nom / boss-nom / taste-nom / Miyashita-title-gen
(a) Unsaturated/Animate: Nonaka-kun-ga jooshi-ga
(b) Unsaturated/Inanimate: Nonaka-kun-ga konomi-ga
(c) Saturated/Animate: Nonaka-kun-ga Miyashita-san-no jooshi-ga
(d) Saturated/Inanimate: Nonaka-kun-ga Miyashita-san-no konomi-ga
15. 藤田部長が / 部下が / 体調が / 自分の
Fujita-buchoo-ga / buka-ga / taichoo-ga / jibun-no
Fujita-chief-title-nom / subordinate-nom / physical condition-nom / jibun-title-gen
(a) Unsaturated/Animate: Fujita-buchoo-ga buka-ga
(b) Unsaturated/Inanimate: Fujita-buchoo-ga taichoo-ga
(c) Saturated/Animate: Fujita-buchoo-ga jibun-no buka-ga
(d) Saturated/Inanimate: Fujita-buchoo-ga jibun-no taichoo-ga
16. 小林君が / 先輩が / 態度が / 石塚さんの
Kobayashi-kun-ga / senpai-ga / taido-ga / Ishizuka-san-no
Kobayashi-title-nom / senior-nom / attitude-nom / Ishizuka-title-gen
(a) Unsaturated/Animate: Kobayashi-kun-ga senpai-ga
(b) Unsaturated/Inanimate: Kobayashi-kun-ga taido-ga
(c) Saturated/Animate: Kobayashi-kun-ga Ishizuka-san-no senpai-ga
117
(d) Saturated/Inanimate: Kobayashi-kun-ga Ishizuka-san-no taido-ga
17. 橋本さんが / 後輩が / 笑顔が / 寺田君の
Hashimoto-sun-ga / koohai-ga / egao-ga / Terada-kun-no
Hashimoto-title-nom / junior-nom / smile-nom / Terada-title-gen
(a) Unsaturated/Animate: Hashimoto-sun-ga koohai-ga
(b) Unsaturated/Inanimate: Hashimoto-sun-ga egao-ga
(c) Saturated/Animate: Hashimoto-sun-ga Terada-kun-no koohai-ga
(d) Saturated/Inanimate: Hashimoto-sun-ga Terada-kun-no egao-ga
18. 山本部長が / 友人が / 発言が / 佐々木君の
Yamamoto-buchoo-ga / yujin-ga / hatsugen-ga / Sasaki-kun-no
Yamamoto-chief-title-nom / friend-nom / comment-nom / Sasaki-title-gen
(a) Unsaturated/Animate: Yamamoto-buchoo-ga yujin-ga
(b) Unsaturated/Inanimate: Yamamoto-buchoo-ga hatsugen-ga
(c) Saturated/Animate: Yamamoto-buchoo-ga Sasaki-kun-no yujin-ga
(d) Saturated/Inanimate: Yamamoto-buchoo-ga Sasaki-kun-no hatsugen-ga
19. 渡辺部長が / 依頼人が / 振る舞いが / 土屋さんの
Wtanabe-buchoo-ga / irainin-ga / furumai-ga / Tsuchiya-san-no
Watanabe-chief-title-nom / client-nom / behavior-nom / Tsuchiya-title-gen
(a) Unsaturated/Animate: Wtanabe-buchoo-ga irainin-ga
(b) Unsaturated/Inanimate: Wtanabe-buchoo-ga furumai-ga
(c) Saturated/Animate: Wtanabe-buchoo-ga Tsuchiya-san-no irainin-ga
(d) Saturated/Inanimate: Wtanabe-buchoo-ga Tsuchiya-san-no furumai-ga
20. 小川君が / 隣の人が / 服装が / 久保さんの
Ogawa-kun-ga / tonarinohito-ga / fukusoo-ga / Kubo-san-no
Ogawa-title-nom / neighbor-nom / dress-nom / Kubotitle-gen
(a) Unsaturated/Animate: Ogawa-kun-ga tonarinohito-ga
(b) Unsaturated/Inanimate: Ogawa-kun-ga fukusoo-ga
(c) Saturated/Animate: Ogawa-kun-ga Kubo-san-no tonarinohito-ga
(d) Saturated/Inanimate: Ogawa-kun-ga Kubo-san-no fukusoo-ga
Appendix C Fragments used in Experiment 3
The following are the nouns used in Experiment 3. For the fragments of Experiment 3, the
nominative marker -ga of the first NP was replaced with the topic marker -wa.
1. 黒田先生は / 奥様が / 作品が / 自分の
Kuroda-sensei-wa / okusama-ga / sakuhin-ga / jibun-no
Kuroda-title-top / wife-nom / work of art-nom / self-gen
(a) Unsaturated/Animate: Kuroda-sensei-wa okusama-ga
(b) Unsaturated/Inanimate: Kuroda-sensei-wa sakuhin-ga
(c) Saturated/Animate: Kuroda-sensei-wa jibun-no okusama-ga
(d) Saturated/Inanimate: Kuroda-sensei-wa jibun-no sakuhin-ga
2. 松本先生は / 両親が / 気質が / 伊藤さんの
Matsumoto-sensei-wa / ryoushin-ga / kishitu-ga / Itoo-san-no
Matsumoto-title-top / parents-nom / temperament-nom / Ito-title-gen
(a) Unsaturated/Animate: Matsumoto-sensei-wa ryooshin-ga
(b) Unsaturated/Inanimate: Matsumoto-sensei-wa kishitsu-ga
(c) Saturated/Animate: Matsumoto-sensei-wa Itoo-san-no okusama-ga
(d) Saturated/Inanimate: Matsumoto-sensei-wa Itoo-san-no kishitsu-ga
3. あきら君は / 母親が / 目標が / ゆみちゃんの
Akira-kun-wa / hahaoya-ga / mokuhyoo-ga / Yumi-chan-no
Akira-title-top / mother-nom / goal-nom / Yumi-title-gen
(a) Unsaturated/Animate: Akira-kun-wa hahaoya-ga
(b) Unsaturated/Inanimate: Akira-kun-wa mokuhyoo-ga
(c) Saturated/Animate: Akira-kun-wa Yumi-cahn-no hahaoya-ga
(d) Saturated/Inanimate: Akira-kun-wa Yumi-cahn-no mokuhyoo-ga
4. 河崎さんは / 娘が / 精神が / 森田課長の
Kawasaki-san-wa / musume-ga / seishin-ga / Morita-kachoo-no
Kawasaki-title-top / daughter-nom / spirit-nom / Morita-title-gen
118
119
(a) Unsaturated/Animate: Kawasaki-san-wa musume-ga
(b) Unsaturated/Inanimate: Kawasaki-san-wa seishin-ga
(c) Saturated/Animate: Kawasaki-san-wa Morita-kachoo-no musume-ga
(d) Saturated/Inanimate: Kawasaki-san-wa Morita-kachoo-no seishin-ga
5. 伊藤さんは / 孫が / 髪型が / 上田社長の
Itoo-san-wa / mago-ga / kamigata-ga / Ueda-shachoo-no
Ito-title-top / grandchild-nom / hairstyle-nom / Ueda-president-title-gen
(a) Unsaturated/Animate: Itoo-san-wa mago-ga
(b) Unsaturated/Inanimate: Itoo-san-wa kamigata-ga
(c) Saturated/Animate: Itoo-san-wa Ueda-shachoo-no kamigata-ga
(d) Saturated/Inanimate: Itoo-san-wa Ueda-shachoo-no kamigata-ga
6. 岸田君は / 長男が / 将来が / 田村さんの
Kishida-kun-wa / choonan-ga / shoorai-ga / Tamura-san-no
Kishida-title-top / eldest son-nom / future-nom / Tamura-title-gen
(a) Unsaturated/Animate: Kishida-kun-wa choonan-ga
(b) Unsaturated/Inanimate: Kishida-kun-wa syoorai-ga
(c) Saturated/Animate: Kishida-kun-wa Tamura-san-no choonan-ga
(d) Saturated/Inanimate: Kishida-kun-wa Tamura-san-no syoorai-ga
7. 清水さんは / 妹が / 言動が / つよし君の
Shimizu-san-wa / imooto-ga / gendoo-ga / Tsuyoshi-kun-no
Shimizu-title-top / sister-nom / speech and action-nom / Tuyoshi-title-gen
(a) Unsaturated/Animate: Shimizu-san-wa imooto-ga
(b) Unsaturated/Inanimate: Shimizu-san-wa gendoo-ga
(c) Saturated/Animate: Shimizu-san-wa Tsuyoshi-kun-no imooto-ga
(d) Saturated/Inanimate: Shimizu-san-wa Tsuyoshi-kun-no gendoo-ga
8. 鈴木先輩は / いとこが / 行動が / 西川君の
Suzuki-senpai-wa / itoko-ga / koodoo-ga / Nishikawa-kun-no
Suzuki-title-top / cousin-nom / action-nom / Nishikawa-title-gen
(a) Unsaturated/Animate: Suzuki-senpai-wa itoko-ga
120 Appendix C. Fragments used in Experiment 3
(b) Unsaturated/Inanimate: Suzuki-senpai-wa koodoo-ga
(c) Saturated/Animate: Suzuki-senpai-wa Nishikawa-kun-no itoko-ga
(d) Saturated/Inanimate: Suzuki-senpai-wa Nishikawa-kun-no koodoo-ga
9. かよちゃんは / 恋人が / 性格が / 小倉さんの
Kayo-chan-wa / koibito-ga / seikaku-ga / Ogura-sun-no
Kayo-title-top / boyfriend / girlfriend-nom / character-nom / Ogura-title-gen
(a) Unsaturated/Animate: Kayo-chan-wa koibito-ga
(b) Unsaturated/Inanimate: Kayo-chan-wa seikaku-ga
(c) Saturated/Animate: Kayo-chan-wa Ogura-san-no koibito-ga
(d) Saturated/Inanimate: Kayo-chan-wa Ogura-san-no seikaku-ga
10. 福田さんは / ペットが / 夢が / 吉村君の
Fukuda-san-wa / petto-ga / yume-ga / Yoshimura-kun-no
Fukuda-title-top / pet-nom / dream-nom / Yoshimura-title-gen
(a) Unsaturated/Animate: Fukuda-san-wa petto-ga
(b) Unsaturated/Inanimate: Fukuda-san-wa yume-ga
(c) Saturated/Animate: Fukuda-san-wa Yoshimura-kun-no petto-ga
(d) Saturated/Inanimate: Fukuda-san-wa Yoshimura-kun-no yume-ga
11. 中村さんは / 師匠が / こころが / 大沢先輩の
Nakamura-san-wa / shishoo-ga / kokoro-ga / Oosawa-senpai-no
Nakamura-title-top / master-nom / mind-nom / Oosawa-title-gen
(a) Unsaturated/Animate: Nakamura-san-wa shisyoo-ga
(b) Unsaturated/Inanimate: Nakamura-san-wa kokoro-ga
(c) Saturated/Animate: Nakamura-san-wa Oosawa-senpai-no shishoo-ga
(d) Saturated/Inanimate: Nakamura-san-wa Oosawa-senpai-no kokoro-ga
12. 山崎係長は / 恩師が / 考えが / 武田さんの
Yamasaki-kakarichoo-wa / onshi-ga / kanngae-ga / Takeda-san-no
Yamasaki-cheif-title-top / (former)teacher-nom / idea-nom / Takeda-title-gen
(a) Unsaturated/Animate: Yamasaki-kakarichoo-wa onshi-ga
(b) Unsaturated/Inanimate: Yamasaki-kakarichoo-wa kangae-ga
121
(c) Saturated/Animate: Yamasaki-kakarichoo-wa Takeda-san-no onshi-ga
(d) Saturated/Inanimate: Yamasaki-kakarichoo-wa Takeda-san-no kangae-ga
13. 田中先生は / 弟子が / 頭が / 西田さんの
Tanaka-sensei-wa / deshi-ga / atama-ga / Nishida-san-no
Tanaka-title-top / pupil-nom / head-nom / Nishida-title-gen
(a) Unsaturated/Animate: Tanaka-sensei-wa deshi-ga
(b) Unsaturated/Inanimate: Tanaka-sensei-wa atama-ga
(c) Saturated/Animate: Tanaka-sensei-wa Nishida-san-no deshi-ga
(d) Saturated/Inanimate: Tanaka-sensei-wa Nishida-san-no atama-ga
14. 野中君は / 上司が / 好みが / 宮下さんの
Nonaka-kun-wa / jooshi-ga / konomi-ga / Miyashita-san-no
Nonaka-title-top / boss-nom / taste-nom / Miyashita-title-gen
(a) Unsaturated/Animate: Nonaka-kun-wa jooshi-ga
(b) Unsaturated/Inanimate: Nonaka-kun-wa konomi-ga
(c) Saturated/Animate: Nonaka-kun-wa Miyashita-san-no jooshi-ga
(d) Saturated/Inanimate: Nonaka-kun-wa Miyashita-san-no konomi-ga
15. 藤田部長は / 部下が / 体調が / 自分の
Fujita-buchoo-wa / buka-ga / taichoo-ga / jibun-no
Fujita-chief-title-top / subordinate-nom / physical condition-nom / jibun-title-gen
(a) Unsaturated/Animate: Fujita-buchoo-wa buka-ga
(b) Unsaturated/Inanimate: Fujita-buchoo-wa taichoo-ga
(c) Saturated/Animate: Fujita-buchoo-wa jibun-no buka-ga
(d) Saturated/Inanimate: Fujita-buchoo-wa jibun-no taichoo-ga
16. 小林君は / 先輩が / 態度が / 石塚さんの
Kobayashi-kun-wa / senpai-ga / taido-ga / Ishizuka-san-no
Kobayashi-title-top / senior-nom / attitude-nom / Ishizuka-title-gen
(a) Unsaturated/Animate: Kobayashi-kun-wa senpai-ga
(b) Unsaturated/Inanimate: Kobayashi-kun-wa taido-ga
(c) Saturated/Animate: Kobayashi-kun-wa Ishizuka-san-no senpai-ga
122 Appendix C. Fragments used in Experiment 3
(d) Saturated/Inanimate: Kobayashi-kun-wa Ishizuka-san-no taido-ga
17. 橋本さんは / 後輩が / 笑顔が / 寺田君の
Hashimoto-sun-wa / koohai-ga / egao-ga / Terada-kun-no
Hashimoto-title-top / junior-nom / smile-nom / Terada-title-gen
(a) Unsaturated/Animate: Hashimoto-sun-wa koohai-ga
(b) Unsaturated/Inanimate: Hashimoto-sun-wa egao-ga
(c) Saturated/Animate: Hashimoto-sun-wa Terada-kun-no koohai-ga
(d) Saturated/Inanimate: Hashimoto-sun-wa Terada-kun-no egao-ga
18. 山本部長は / 友人が / 発言が / 佐々木君の
Yamamoto-buchoo-wa / yujin-ga / hatsugen-ga / Sasaki-kun-no
Yamamoto-chief-title-top / friend-nom / comment-nom / Sasaki-title-gen
(a) Unsaturated/Animate: Yamamoto-buchoo-wa yujin-ga
(b) Unsaturated/Inanimate: Yamamoto-buchoo-wa hatsugen-ga
(c) Saturated/Animate: Yamamoto-buchoo-wa Sasaki-kun-no yujin-ga
(d) Saturated/Inanimate: Yamamoto-buchoo-wa Sasaki-kun-no hatsugen-ga
19. 渡辺部長は / 依頼人が / 振る舞いが / 土屋さんの
Wtanabe-buchoo-wa / irainin-ga / furumai-ga / Tsuchiya-san-no
Watanabe-chief-title-top / client-nom / behavior-nom / Tsuchiya-title-gen
(a) Unsaturated/Animate: Wtanabe-buchoo-wa irainin-ga
(b) Unsaturated/Inanimate: Wtanabe-buchoo-wa furumai-ga
(c) Saturated/Animate: Wtanabe-buchoo-wa Tsuchiya-san-no irainin-ga
(d) Saturated/Inanimate: Wtanabe-buchoo-wa Tsuchiya-san-no furumai-ga
20. 小川君は / 隣の人が / 服装が / 久保さんの
Ogawa-kun-wa / tonarinohito-ga / fukusoo-ga / Kubo-san-no
Ogawa-title-top / neighbor-nom / dress-nom / Kubotitle-gen
(a) Unsaturated/Animate: Ogawa-kun-wa tonarinohito-ga
(b) Unsaturated/Inanimate: Ogawa-kun-wa fukusoo-ga
(c) Saturated/Animate: Ogawa-kun-wa Kubo-san-no tonarinohito-ga
(d) Saturated/Inanimate: Ogawa-kun-wa Kubo-san-no fukusoo-ga
Appendix D Completions provided by
participants for experiment 1
The following are the completion sentences that were provided by participants for the frist
experiment.
1. 黒田先生が 奥様が//
[ D ] 倒れたので午後の授業は休講になった。
[ I ] 遊んでいると仲間に入れてもらった。
[ I ] きれいだとほめちぎっていた。
[ I ] 「入院した」とおっしゃった。
[ I ] たおれたと家に帰った。
[ I ] きれいだと言った。
[ J ] 大変よ!と私に教える。
[ K ] いらっしゃいましたと伝えた。
2. 黒田先生が 作品が//
[ H ] 上手にできないと悩んでいる。
[ H ] 欲しいといっている。
[ I ] できないとうなる。
[ I ] きたないと言った。
[ J ] 痛むので丁寧に扱えと言った。
[ J ] 素晴らしいと皆に紹介してくれた。
[ J ]すばらしいので満天をやるとほめて下さった。
[ L ] ないと言った。
3. 岡田さんに 黒田先生が 奥様が//
[ I ] 美人だと自慢した。
[ K ] 急用で電話をかけてこられたのを伝えた。
[ K ] きとくですと伝えた。
[ K ] 呼んでいると言われた。
[ K ] 亡くなったと伝えた。
[ K ] 大変怒っていることを伝えた。
[ K ] 大変だといった。
[ K ] すばらしいとほめた。
4. 岡田さんに 黒田先生が 作品が//
[ A ] 認められた。
[ H ] 気に入らないと言っていた。
[ I ] 未完成であることを指てきされた。
[ I ] 仕上がったと伝えた。
[ I ] ダメだったと言った。
[ K ] 完成していないことに注意した。
[ K ] まだ出来ていないなら放課後も残るように
と言った。
[ K ] とても上手になったとほめていた。
5. 松本先生が 両親が//
[ F ] どれほど大切であるか教えてくれた。
[ I ]お見合いをすすめてくるとぐちを言っていた。
[ I ] 事故にあったと真っ青になりながら私に告
げた。
[ J ] 来てくだされば・・・とつぶやいていた。
[ J ] 悪いのではないと私をしかった。
[ J ] 呼んでいると、私に言った。
[ J ] 悪いと言った。
[ J ] 来たことを私に教えてくださった。
6. 松本先生が 気質が//
[ A ] あらい。
[ A ] あらいです。
[ F ] あらいのはまぁ良いと言った。
[ I ] 変わったと言った。
[ I ] 良いと言っていた。
[ K ] あわないとなげいている。
[ K ] 嫌だとはきすてる。
[ X ] 悪いので帰ろうといった。
123
124 Appendix D. Completions provided by participants for experiment 1
7. 本田さんに 松本先生が 両親が//
[ I ] 旅行していると伝えた。
[ I ] 旅行に行っていると話した。
[ K ] きとくなので早く帰りなさいと言った。
[ K ] 来てますよと伝えた。
[ K ] 学校へなぐり込んで来るようだと言って
いた。
[ K ] 来られないのならこちらから出向こうと
言った。
[ K ] あぶないと伝えた。
[ K ] 一大事だから帰ってくるように言われてる
と言われた。
8. 本田さんに 松本先生が 気質が//
[ E ] あらいと文句をいいたい。
[ I ] あらいと言っていた。
[ I ] 荒いと言われた。
[ K ] 荒いと言った。
[ K ] 変わったと言った。
[ K ] あらいと注意した。
[ K ] 荒すぎることを注意した。
[ K ] おかしいと言った。
9. あきら君が 母親が//
[ I ] 父親と離婚するんだと聞き泣いていた。
[ I ] 手伝いをしろと言っていると言った。
[ I ] 死んでいることを隠していた。
[ I ] 病気で入院していると言った。
[ I ] 急病だと言った。
[ I ] いなくなったのでさがし回っていた。
[ I ] こわいと泣いている。
[ I ] ふとっているといつももん句を言う。
10. あきら君が 目標が//
[ E ] 決まらない気がする。
[ F ] あることは良いことだが、それにとらわれ
すぎてもいけないと言った。
[ F ] 高ければ高い程いいと言っていた。
[ L ] なくなったとグチってきた。
[ L ] ないとなげいていた。
[ L ] ないとなげいていた。
[ L ] なくなって落ち込んでいた。
[ Y ] 達成させた。
11. 美智子ちゃんに あきら君が 母親が//
[ C ] 私が皆があめをやる。
[ H ] きらいだと言った。
[ I ] しんだと言った。
[ I ] 倒れたといって泣きついていた。
[ I ] かぜを引いたと言った。
[ I ] いないと泣きそうになって言った。
[ I ] ご飯食べにおいでって言ってたよと話して
いた。
[ I ] 大変だとつげた。
12. 美智子ちゃんに あきら君が 目標が//
[ G3 ] はっきりしない人間はダメだと言った。
[ H ] 達成しなかったと泣きついてきた。
[ I ] 決まったと言った。
[ I ] 高すぎて達成できないとこぼしていた。
[ L ] あると言った。
[ L ] あなたと結婚することだと言っちゃった。
[ L ] ないと言い切った。
[ X ] ないのよ、と弱音をはいた。
13. 河崎さんが 娘が//
[ H ] 心配だと言う。
[ I ] テストに合格したと喜んでいた。
[ I ] 家出をしたことに心を痛めておられる。
[ I ] 結婚したと泣いていた。
[ I ] 勉強しないのよーとグチをこぼしていた。
[ I ] 可愛すぎると親バカ発言をしていた。
[ I ] どんな誕生日プレゼントが欲しいのだろう
と頭を悩ませていた。
[ I ] いなくなったとあわてている。
14. 河崎さんが 精神が//
[ A ] おかしくなったらしい。
[ I ] 病んでいると医者に言われた。
[ I ] おかしくなりそうだと残業中にぐちった。
[ I ] 病んで病院に通っているときいた。
125
[ I ] 不安定だと言っていた。
[ I ] 病んでいますねと言われていた。
[ I ] 不安定なことを気にしていた。
[ I ] やられそうだとぼやいた。
15. 後藤部長に 河崎さんが 娘が//
[ I ] おみ合いするようすすめた。
[ I ] かわいいことを自慢した。
[ I ] 事故にあったと伝えた。
[ I ] かわいいことを自慢していた。
[ I ] 結婚することになったと報告していた。
[ I ] 小学校に入学した嬉しさを語った。
[ I ] 受験で大変なんですとお金を借りていた。
[ I ] 退院したと喜んで言っていた。
16. 後藤部長に 河崎さんが 精神が//
[ I ] 異常だと怒られた。
[ I ] もう限界ですと相談した。
[ I ] 不安定だと言った。
[ I ] みだれやすいので営業はやめてほしいと
いった。
[ I ] やんでいると言った。
[ K ] なっとらん!!と怒った。
[ X ] 健全すぎるとのたまう。
[ Y ] ちょっとおかしいのではないか。
17. 岸田君が 長男が//
[ H ] 嫌いだと言っていた。
[ H ] うらやましいと言った。
[ H ] うらやましいと言った。
[ H ] ほしいと言った。
[ I ] 大変なんだと悩んだ。
[ I ] 生まれたと報告した。
[ I ] 殺したと自白する。
[ I ] 産まれたといって喜んでいた。
18. 岸田君が 将来が//
[ H ] 心配だと言っていた。
[ H ] 不安だとぼやいていた。
[ H ] 有望だと思っている
[ H ] 不安だと言った。
[ H ] 見えない、とふさぎこんでいる。
[ H ] 怖いと言った。
[ I ] あまり有望すぎて各部門からひっぱりだこ
になっている。
[ I ] 決まっていると言っていた。
19. 中島部長に 岸田君が 長男が//
[ I ] グレたと言っていた。
[ I ] 病気だと言った。
[ I ] 病気ですと伝えた。
[ I ] 病気なので仕事の休みをとらせてもらえる
ようたのんだ。
[ I ] 生まれたと伝えた。
[ I ] うまれるといった。
[ K ] 生まれておめでとうと言っていた。
[ K ] 大学に合格されたそうでと話しかけた。
20. 中島部長に 岸田君が 将来が//
[ H ] 見えないんですと相談していた。
[ I ] きまっていないことを相談した。
[ I ] 不安なんです。と辞職届けを出した。
[ I ] あぶないと言われた。
[ I ] 不安で自分がどうすれば良いかわからない
と相談をした。
[ I ] 有望であるとほめられた。
[ K ] 危ないよと言っていた。
[ L ] ないことをそうだんした。
21. 鈴木先輩が いとこが//
[ H ] かわいくて…とえいえん話された。
[ I ] 結婚したと言っていた。
[ I ] 結婚したと言っていた。
[ I ] 転校してくるから仲良くやってくれと頼ん
できた。
[ I ] テレビに出ると、喜んでいる。
[ I ] 明日結婚するんだと嬉しそうに言った。
[ I ] 来ると言った。
[ I ] 死んだのでおそう式に行くと言っていた。
126 Appendix D. Completions provided by participants for experiment 1
22. 鈴木先輩が 行動が//
[ A ] いやらしいのだ!
[ A ] あやしい。
[ A ] いまいち分からない。
[ A ] あやうい。
[ E ] あやしいので気になっている。
[ G3 ]おかしい彼を見張っているようにと言った。
[ I ] 早いといった。
[ K ] おそい!とどなった。
23. 丸山さんに 鈴木先輩が いとこが//
[ I ] 丸山さんに似ていると言った。
[ I ] 3 人いるとばくろした。
[ I ] 昨日をしていたことを知っているか?とた
ずねた。
[ I ] 今度結婚するにあたって自分はそのブライ
ダルアシスタントをたのまれたんだがどうしたらいい
のかわからないと相談している。
[ I ] かしこいと自まんする。
[ I ] 結婚したことを報告した。
[ I ] 結婚するので金を貸してくれと頼んだ。
[ Y ] 出会った。
24. 丸山さんに 鈴木先輩が 行動が//
[ K ] 怪しすぎると注意した。
[ K ] 気にくわないと言った。
[ K ] おかしいと言っていた。
[ K ] 変だよと言っていた。
[ K ] おかしいともんくをいった。
[ K ] おそいと注意した。
[ K ] 少し変だと注意していた。
[ K ] 変と言った。
25. 中村さんが 師匠が//
[ I ] 末期を告げたと言った。
[ I ] 教えてくれないと思った。
[ I ] 厳しいことに弱音を吐いた。
[ I ] たおれたとさわいでいた。
[ I ] 大変だと叫んでいる。
[ I ] とても厳しいと言っていた。
[ I ] 一番すごいと周りに自慢していた。
[ I ] 悪いと言った。
26. 中村さんが こころが//
[ H ] よくわからないと言っていた。
[ H ] 病んでるとつぶやいた。
[ I ] とても晴れ晴れすると言っていた。
[ I ] 痛いと言った。
[ I ] あたたかいのだと幸せそうに言っている。
[ I ] ゆがんでいる、と言った。
[ Y ] 痛めている。
[ Z ] 丘の住人である限り、このいさかいがたえ
ることはない。
27. 北村先生に 中村さんが 師匠が//
[ H ] ほしいとねだった。
[ I ] 怒っていると言った。
[ I ] 厳しすぎると相談した。
[ I ] もう僕におしえることはないと言うんだ…。
とさびしそうにいった。
[ I ] すごい人だと伝えた。
[ I ] すでに集まっていると教える。
[ I ] こわいと言った。
[ Y ] これから食事に行こうと誘った。
28. 北村先生に 中村さんが こころが//
[ F ] 広いのはすばらしいことだといった。
[ H ] 病んでいると相談してみた。
[ I ] 痛むのはなぜだろうと相談をもちかけた。
[ I ] 暴れまくっていると伝えた。
[ I ] いたいと言った。
[ K ] 君のその目に表れていると言った。
[ K ] 読みたいと言っていた。
[ K ] ないねと言った。
29. 山崎係長が 恩師が//
[ C ] 私が、皆がリストラにあった。
[ I ] 退職することを耳にした。
[ I ] 入院したので見舞いに行くと言っていた。
[ I ]入院なさっていて見舞いに行くと話していた。
[ I ] 日本に帰国するため明日会社を休むそうだ。
[ I ] 訪ねて来るのでお菓子を買いに行った。
[ I ] 急死したと伝えられておどろいた。
[ I ] 素敵だと自まんする。
127
30. 山崎係長が 考えが//
[ A ] 古い。
[ A ] まとまったみたいだ。
[ A ] 批評された。
[ I ] 甘いと社長にしかられた。
[ K ] 合わないと言った。
[ K ] 甘いとどなった。
[ L ] あると言った。
[ L ] あると言っていた。
31. 池田さんに 山崎係長が 恩師が//
[ I ] 急病で倒れたので帰りますと言っていた。
[ I ] ぜひ会いたいと言っていたことを告げた。
[ I ] 亡くなったと言っていた。
[ I ] 死んだことを聞いた。
[ K ] 来ていることを教えた。
[ K ] 亡くなったと伝えた。
[ Y ] 同じである。
[ Y ] お土産を持ってきた。
32. 池田さんに 山崎係長が 考えが//
[ G3 ] 合わない部下は首にしろ!!と言った。
[ K ] 甘いと怒っていた。
[ K ] 変だと教える。
[ K ] まちがっていると思うと話した。
[ L ] あると話した。
[ L ] あるのできてくれといった。
[ Y ] 伝わった。
[ Z ] プランを伝える。
33. 田中先生が 弟子が//
[ H ] ほしいと言った。
[ H ] 欲しいと言っていた。
[ H ] 好きだと言った。
[ I ] わるいと思ったのでしかった。
[ I ] 悪さをするとなげく。
[ I ] 言うことを聞いてくれないと困っている。
[ I ] いなくて困っているらしい。
[ I ] いなくなったといっていた。
34. 田中先生が 頭が//
[ I ] 痛いと言って自習にして帰っていった。
[ I ] 痛いとわめく。
[ I ] ハゲていることを気にしていた。
[ I ] かゆいと言っている。
[ I ] 痛いと言って保健室へ行った。
[ I ] 痛いと急に言い出した。
[ I ] 痛いと言った。
[ J ] いいねとほめてくれた。
35. 西村さんに 田中先生が 弟子が//
[ I ] 不良だと嘆いた。
[ I ] あなたのことを好きらしいでと言った。
[ I ] 悪口を言っていると聞いた。
[ I ] いなくなったと言った。
[ I ] いなくなったと言った。
[ I ] お世話になったことの礼を言った。
[ I ] 病気になったと言った。
[ I ] 最近なまいきなんだと話していた。
36. 西村さんに 田中先生が 頭が//
[ F ] どうやったらはげないのかたずねていた。
[ F ] 良いだけでは駄目だ体力をつけろと言って
いた。
[ I ] ハゲてると言っていた。
[ I ] ハゲてると言われた。
[ I ] 痛いと言った。
[ K ] いたいなら保健室へ行きなさいと言って
いた。
[ K ] いいとほめていた。
[ K ] 良いねとほめていた。
37. 野中君が 上司が//
[ H ] 嫌いだと言っていた。
[ H ] ムカつくとかげ口を言っていた。
[ H ] 嫌いであると真剣に言っていた。
[ I ] だらしなくて嫌になるとぐちを言っていた。
[ I ] 口うるさくてストレスがたまるいっぽうだ
とぐちを言った。
[ I ] 倒れたと叫んできた。
[ I ] おこっていると、怖がった。
[ I ] 悪いから会社も悪い。と言った。
128 Appendix D. Completions provided by participants for experiment 1
38. 野中君が 好みが//
[ A ] 多い。
[ A ] 変だ。
[ E ] 広いと知る。
[ E ] わからない。
[ E ] おかしいときいた。
[ I ] むつかしくて誰もえらべないとうぬぼれた。
[ K ] わからないと言った。
[ K ] かたよっていると怒っていた。
39. 平井さんに 野中君が 上司が//
[ I ] たおれたと伝えた。
[ I ] 厳しすぎるとこぼした。
[ I ] いびってくることを相談した。
[ K ] 呼んでいたと伝えた。
[ K ] 怒っていることを告げ口した。
[ K ] おこっていると告げた。
[ K ] 怒っていると教えた。
[ L ] いやだと言っていたことを言った。
40. 平井さんに 野中君が 好みが//
[ E ] あなたみたいな子らしいでと言った。
[ I ] うるさいなと言われた。
[ I ] 合うねと話してるのを聞いたと伝えた。
[ I ] 違うことを伝えた。
[ I ] 優しい人と答えているのを聞いたそうだ。
[ I ] 合うねと話しかけた。
[ I ] わるいことを伝えた。
[ I ] 全然違うと言い争った。
41. 藤田部長が 部下が//
[ G ] おかした失敗に対して怒った。
[ I ] 従ってくれないと嘆いた。
[ I ] かなり仕事が出来るのであせっていた。
[ I ] 頼りないとこぼしていた。
[ I ] 悪いと責任をのがれた。
[ I ] 頼りないと怒った。
[ I ] たよりないことに頭を痛めていた。
[ I ] だらしがないと怒った。
42. 藤田部長が 体調が//
[ A ] 悪いらしい。
[ A ] 悪いらしい。
[ D ] 悪いらしいので女性社員は喜んだ。
[ E ] ますます悪化したように見える。
[ G3 ] すぐれないものは次の会議には出席しな
いようにと告げた。
[ I ] おもわしくない、と言う。
[ I ] すぐれないと言って帰った。
[ I ] 悪いから休むと言った。
43. 小山さんに 藤田部長が 部下が//
[ C ] 私が皆が怒るところだ。
[ I ] ケガをしたと伝えた。
[ I ] した失敗について話した。
[ I ] ちゃんと仕事しないとぐちった。
[ I ] なってないとグチった。
[ I ] いうことをきかないとグチっていた。
[ I ] なまけていると告げ口した。
[ I ] 失敗したとぐちった。
44. 小山さんに 藤田部長が 体調が//
[ I ] 悪いので会社を休むと伝えた。
[ I ] 悪いから休むと伝えた。
[ I ] 悪いらしいと伝えた。
[ I ] 悪いので後は任せると言った。
[ I ] すぐれないと言っていた。
[ I ] 悪くなったと伝えた。
[ I ] 悪いと伝えた。
[ K ] 悪そうだと心配そうに声をかけた。
45. 小林君が 先輩が//
[ C ] 私が、かけっこしている。
[ H ] 尊敬できると誇りに思っている。
[ H ] 好きだと教えてくれた。
[ H ] 嫌いだと言っている。
[ I ] 失恋したことをみんなに言いふらしていま
した。
[ I ] 厳しいと言っていた。
[ I ] 悪いことはわかりきっていると思った。
[ I ] いないと言っている。
129
46. 小林君が 態度が//
[ A ] おかしい。
[ A ] おかしかった。
[ D ] 変だとみんな言った。
[ I ] 大きいと思われた。
[ I ] 悪いと怒られた。
[ I ] 変わったと周りの女の子に言われた。
[ I ] 悪いことを注意されていた。
[ I ] 悪いと怒られた。
47. 永井さんに 小林君が 先輩が//
[ I ] こわいと相談された。
[ I ]かっこいいと目をキラキラさせていっていた。
[ I ] 悪かったんだとグチをこぼしていた。
[ I ] かばんを持たせた。
[ K ] 呼んでいたよと言っていた。
[ K ] 応援に来てくれるから大丈夫だと励ました。
[ K ] キミを好きみたいだよと言っていた。
[ K ] 嫌いだというぐちを聞いた。
48. 永井さんに 小林君が 態度が//
[ D ] 違うとみんなうわさした。
[ I ] 横柄だと注意された。
[ I ] 変だと言われた。
[ K ] 悪いと言った。
[ K ] 悪いと注意した。
[ K ] 悪いと言った。
[ K ] 悪いとつげた。
[ K ] 悪いと言いつける。
49. 矢島さんが この番組の タレントが//
[ F ] 面白くないと文句を言った。
[ F ] 今最も売れているんだと話していた。
[ F ] ショーモないと言った。
[ F ] 悪いのだと言い張った。
[ H ] 大好きだと言っていた。
[ H ] きらいといっていた。
[ H ] 嫌いなので見たくないと嫌がった。
[ H ] 好きだと言っていた。
50. 矢島さんが この絵画の 虹が//
[ F ] 美しいと言った。
[ F ] きれいだと、思った。
[ F ] 重要なポイントだと教えてくれた。
[ H ] 印象に残ったらしい。
[ H ] とても好きみたいでずっと見ている。
[ H ] 描けた理由を説明して下さった。
[ H ] とても好きだと言っていた。
[ J ] 一番よく描けているとほめていた。
51. 萩原監督に 矢島さんが この番組の タレントが//
[ F ] だめだと言った。
[ F ] 会いたがっていると説明した。
[ F ] なんだがすごい人だと話した。
[ F ] さっきそこの通りを歩いていたんだと教
えた。
[ F ] ふさわしいと提案した。
[ F ] なまけていると悔しそうに言った。
[ H ] 大大大好きと話す。
[ H ] 好きと言った。
52. 萩原監督に 矢島さんが この絵画の 虹が//
[ F ] すばらしいのだよ、と力説する。
[ F ] きれいだと言った。
[ F ] 映画のイメージにぴったりだと言った。
[ F ] 映画に出てきたのとそっくりだと言った。
[ F ] もう少しグラフィックで何とかならないの
かとグチっていました
[ F ] イマイチだと申し立てた。
[ F ] 美しいと話した。
[ F ] 一色たりないと不満を言った。
53. 近藤さんが あの 漫才師が//
[ F ] 面白いと教えてくれた。
[ F ] おもしろいと言った。
[ F ] 左手に持っているものは何かしら?と問う。
[ F ] 今日のネタはおもしろくないと言っていた。
[ F ] 実は付き合ってるとうわさした。
[ F ] 昨日街を歩いているのを見かけたそうだ。
[ H ] 気に入らないんだと怒っていた。
[ H ] 好きだと言っていた。
130 Appendix D. Completions provided by participants for experiment 1
54. 近藤さんが この スポーツが//
[ F ] 今ブームのヨガよと言った。
[ F ] あぶないと訴えた。
[ F ] おもしろいと話していた。
[ F ] マイナーであることを残念そうに話した。
[ H ] 得意だと聞いた。
[ H ] 苦手だと、何度も主張した。
[ H ] 嫌いと言った。
[ H ] 嫌いだと言ったので理由を聞いた。
55. 小沢君に 近藤さんが あの 漫才師が//
[ F ] おもしろいと言っていた。
[ F ] とてもおもしろいと聞いた。
[ F ] 面白い事をするらしいと教えた。
[ F ] 今度地元に来るとはしゃいできかせていた。
[ F ] 出てくると場の雰囲気が変わるねと言った。
[ F ] 面白いと言っていた。
[ H ] 好きだと言っていた。
[ H ] 見たいと言っていた。
56. 小沢君に 近藤さんが この スポーツが//
[ F ] サッカーだ!!とおしえた。
[ F ] 今一番人気があると教えてくれた。
[ F ] 今、熱いんだよ!と教えた。
[ F ] たのしいとさそった。
[ H ] 大嫌いとさけぶような気がする。
[ H ] 好きだと熱弁していた。
[ H ] 好きだと言った。
[ H ] 好きだと言った。
57. 藤原君が この 芸人が//
[ B ] 好きだった。
[ B ] 好きだ。
[ F ] 面白いとすすめた。
[ F ] つまらないとテレビを消した。
[ F ] 兄だと私に伝えている。
[ F ] 売れたのは時代が変わったおかげだよと
言った。
[ H ] 好きだといっていた。
[ H ] 好きだと言っていた。
58. 藤原君が 今夜の お月さまが//
[ F ] 人の顔のように見えると電話してきた。
[ F ] きれいだと言った。
[ F ] 魅惑的だと、私に言った。
[ F ] とりわけきれいだと言った。
[ F ] 丸いねと言った。
[ F ] いちばんきれいだと言った。
[ F ] きれいだと思った。
[ H ] 大好きだといった。
59. 木下さんに 藤原君が この 芸人が//
[ F ] あの変顔な人なんだよ言った。
[ F ] 結婚したらしいと言った。
[ F ] 絶対売れると念を押した。
[ F ] おもしろいよと言った。
[ F ] おもしろいと勧めた。
[ F ] おもしろくないと言った。
[ F ] おすすめだと言っていた。
[ F ] おもしろくないことをののしった。
60. 木下さんに 藤原君が 今夜の お月さまが//
[ F ] とても赤くて怖いと言った。
[ F ] 満月なら夜桜を見にいこうと誘った。
[ F ] 満月だと言った。
[ F ] きれいだと見せた。
[ F ] 沈む頃に会いに行く。と言っていた。
[ F ] まんまるいねと言った。
[ F ] キレイだねと言った。
[ F ] すごく綺麗であったらいいねと話していた。
61. 飯田さんが 看護婦が//
[ F ] 来るたびに嫌味を言うのは好きな子をいじ
める小学生と同じで素直になれないだけなんです。
[ F ] 足りない。とぼやいていた。
[ F ] 知り合いにいるといっていた。
[ F ] よくないと言っていた。
[ F ] おもしろい病院に入院していたと話して
いた。
[ H ] 嫌いだと言った。
[ H ] 好きと言った。
[ H ] 好きだと言っていた。
131
62. 飯田さんが 雑草が//
[ B ] 気になるみたいだ。
[ B ] すきだ。
[ C ] みんな成長している。
[ F ] 多くて困っていると話していた。
[ F ] はえているからぬいてみた。
[ F ] のびるのは実に早い。と興味深そうに私に
言った。
[ F ] 多いのでかってほしいとたのんだ。
[ H ] 苦手だと言って抜いている。
63. 内藤君に 飯田さんが 看護婦が//
[ F ] トイレに行ったと教えた。
[ F ] いないと言った。
[ F ] 呼んでいたことをつげている。
[ F ] 君に色目を使っているよ、と言う。
[ F ] 呼んでいると教えた。
[ F ] やさしかったので将来私もああなりたいと
語っていた。
[ F ] 仲良くしていると知らせた。
[ F ] メールアドレスを渡してきたとうれしそう
に話している。
64. 内藤君に 飯田さんが 雑草が//
[ F ] はえてきたと言った。
[ F ] 生えているのでぬいてとたのんだ。
[ F ] 食べれるって知ってたとたずねられた。
[ F ] 増えてきたからと草むしりを頼んだ。
[ F ] 生えているよと教えた。
[ F ] 生えてるので刈ってくるよう命令した。
[ F ] じゃまだから抜いてくれと言った。
[ H ] うっとうしいと言った。
65. 吉原君が この プロレスラーが//
[ B ] 好きらしい。
[ F ] 自分の父だとカミングアウトした。
[ F ] 強いと教えられた。
[ F ] 一番強いと言った。
[ F ] 負けたと言った。
[ F ] 世界チャンピョンだと教えていた。
[ H ] 好きなのだと恋人に熱く語っていた。
[ H ] 好きだと言っていた。
66. 吉原君が 隅田川の 堤防が//
[ F ] 壊れたことを町中に知られた。
[ F ] こわれたと言った。
[ F ] 決壊したと言って大さわぎになった。
[ F ] 危ないと言っていた。
[ F ] 崩壊しかけていると言っていた。
[ F ] けっかいしたとさけんだ。
[ F ] このままでは危ないと言うことを示唆した。
[ F ] 建て直されると、知った。
67. 田口さんに 吉原君が この プロレスラーが//
[ F ] 弱いんだと聞いた。
[ F ] 勝ったら 1 万円やると言った。
[ F ] 強いとすすめた。
[ F ] 父のかたきだとばらす。
[ F ] 強いんだといった。
[ F ] 優勝したと聞いた。
[ F ] かっこいいと説明した。
[ H ] 好きとおしえた。
68. 田口さんに 吉原君が 隅田川の 堤防が//
[ F ] こわれたとおしえた。
[ F ] 壊れて洪水になっていることを知らせた。
[ F ] 破壊したと伝えた。
[ F ] 古くなっていると伝えた。
[ F ] 良いとおしえた。
[ F ] 万が一崩壊したときの被害予想図と避難経
路の資料を提出するように求めた。
[ F ] 危険だと知らせた。
[ F ] こわれたからすぐに来てと電話してきた。
69. 川上社長が あの 舞妓さんが//
[ C ] おどった。
[ F ] 倒れたと血そうを変えて言った。
[ F ] やっぱり一番だな、とニヤけた。
[ F ] かわいいと喜んでいた。
[ F ] 良いねえと指名していた。
[ H ] 気に入ったと仰っていた。
[ H ] 気に入ったと言った。
[ I ] いいので呼んで来いと得意先の人を困らせ
ている。
132 Appendix D. Completions provided by participants for experiment 1
70. 川上社長が この 八重桜が//
[ F ] 咲いたら花見をしようと提案した。
[ F ] きれいといった。
[ F ] 美しいと気に入った様子だった。
[ F ] キレイだと言った。
[ F ] この辺で一番キレイだと言った。
[ H ] 一番好きだと言った。
[ H ] ほしいと言った。
[ H ] お気に入りだったと聞いた。
71. 早川さんに 川上社長が あの 舞妓さんが//
[ F ] キレイなんだと教えていた。
[ F ] かわいらしいと言った。
[ F ] よろしくと言っていたことを告げた。
[ H ] 気に入ったと言われた。
[ H ] 気に入っていることを伝えた。
[ H ] 好きだと顔を赤らめて言われた。
[ Y ] 「あなたの服装おかしい」と言った。
[ Y ] 踊らせた。
72. 早川さんに 川上社長が この 八重桜が//
[ F ] 舞い落ちる時、写真をとろう。
[ F ] きれいだと言った。
[ F ] さくときあなたと一緒にいたいと告白した。
[ F ] 象徴しているものを伝える。
[ F ] 美しいと説明した。
[ F ] キレイだとほめていた事を伝えた。
[ F ] 母のかたみなのだと話した。
[ F ] 満開になる日を教えた。
73. 陽子ちゃんが あの ピアニストが//
[ B ] すきだ。
[ B ] 大すきだ。
[ F ] すごい人だとほめていた。
[ F ] 母だと私に教える。
[ F ] 本当にコンテストで一位をとったのか不思
議だ。とボソっと言った。
[ F ] すばらしい演奏をしたと感激した。
[ F ] CDを出したと言っていた。
[ H ] 大好きだと言ったから僕は思いきってコン
サートに誘った。
74. 陽子ちゃんが 今夜の 星空が//
[ F ] きれいだと言った。
[ F ] 曇ってみえにくいと嘆いている。
[ F ] キレイねと言った。
[ F ] とても素敵だとメールをしてきた。
[ F ] 美しすぎると涙を流した。
[ F ] 異常だよと言っている。
[ F ] きれいだねと言った。
[ F ] 不気味だと不安がっていた。
75. 吉本君に 陽子ちゃんが あの ピアニストが//
[ F ]とてもやさしい音色を奏でることを話し合っ
ていた。
[ F ] テレビに出ていたと言っている。
[ F ] 私の母だと言った。
[ F ] すばらしいとすすめた。
[ F ] すてきと言った。
[ H ] 好きだとおしえてあげた。
[ H ] 好きだとアピールしていた。
[ H ] 好きなのだとこっそり話した。
76. 吉本君に 陽子ちゃんが 今夜の 星空が//
[ F ] きれいだと言った。
[ F ] きれいだから見てみてよと電話した。
[ F ] 晴れていたら写真にとって送るねと言った。
[ F ] 綺麗だから一緒に見ようと誘っていた。
[ F ] ずっとつづけば良いのに・・・と言っていた。
[ F ] キレイだと言った。
[ F ] とてもきれいだとメールで教えてもらった。
[ F ] ステキだからと、見せる。
77. 内田さんが あの バレリーナが//
[ C ] 同一人物らしい。
[ F ] きっと優勝するわと言った。
[ F ] 笑ってると言った。
[ F ] とても素敵だと言っていた。
[ F ] 本当は男だと聞いた。
[ F ] 友達だということに気付いた。
[ H ] 好きだと言っていたことを思い出した。
[ H ] 気に入らないと言った。
133
78. 内田さんが この美術館の 日本庭園が//
[ F ] とてもキレイなことを知っている。
[ F ] きれいだと言った。
[ F ] すばらしいとほめた。
[ F ] 素晴しいとほめていた。
[ F ] アメリカで絶賛されたと喜んでいた。
[ F ] 日本一だとほめた。
[ F ] すばらしいと絶賛した。
[ F ] スバラシイと強調する。
79. 中川先輩に 内田さんが あの バレリーナが//
[ F ] かわいいねと言った。
[ F ] 実は私の知り合いなんですと言った。
[ F ] おどっている曲目をたずねた。
[ F ] 外国へ行くことを教えてくれたことを聞
いた。
[ F ] 自分の?人だと言った。
?
[ F ] キレイね、と話しかけた。
[ H ] 好きだと言った。
[ I ] かつて恋していた人だと語った。
80. 中川先輩に 内田さんが この美術館の 日本庭園が//
[ F ] どれだけすばらしいかを語っていた。
[ F ] キレイねと言った。
[ F ] あの有名な庭園だよと言っていた。
[ F ] すばらしいことを語っていた。
[ F ] 美しいんだと教えた。
[ H ] お気に入りだと話した。
[ H ] 好きだと言っていた。
[ H ] 好きだといった。
81. 吉田さんが この 手品師が//
[ F ] ウソくさいと言った。
[ F ] 本当に上手であると絶賛していた。
[ F ] だめだとおい出した。
[ F ] さのサギの人だよと言った。
[ F ] インチキっぽいと話した。
[ F ] うさんくさいと言った。
[ F ] 歩いているのを見たと言っていた。
[ F ] カリスマだと言っていた。
82. 吉田さんが 今日の 夜空が//
[ F ] 今世紀最大の流星群ショーになるだろうと
語った。
[ F ] きれいだねと言ってきた。
[ F ] きれいだなあ、と言った。
[ F ] 一番すてきだねと言った。
[ F ] 美しいと喜んでいた。
[ F ] きれいだと言った。
[ F ] 綺麗なら明日は晴れだと言っていた。
[ H ] 今までで一番好きだと言っていた。
83. 須藤君に 吉田さんが この 手品師が//
[ C ] 手品セットをあげた。
[ F ] にせだと教えてくれた。
[ F ] よく当たると教えていた。
[ F ] にくいなあとつぶやく。
[ F ] いいと言った。
[ F ] アフリカで有名なんだ!!と自慢気に言った。
[ F ] すばらしいので見せたがっている。
[ Y ] おそろしい人なんだ。
84. 須藤君に 吉田さんが 今日の 夜空が//
[ F ] キレイだよ、と言った。
[ F ] とてもきれいだと言った。
[ F ] きれいだねと言った。
[ F ] きれいだと言った。
[ F ] きれいだから一緒に見に行こうと口説かれ
ていた。
[ F ] 二人でみるさいごの夜空よ、という。
[ F ] キレイだねと言った。
[ H ] 忘れられないと電話しています。
85. 西山さんが この 作曲家が//
[ F ] 誰だか知りたがっている。
[ F ] 亡くなったことを教えてくれた。
[ F ] 偉大であるのはそのありきたりな人生が曲
に与えた影響力がすごいからだと力説した。
[ G ] 作る曲は嫌いだね、という。
[ H ] お気に入りなんだと話した。
[ H ] キライと言った。
[ H ] 一番好きだと言っていた。
[ H ] 好きと言っていた。
134 Appendix D. Completions provided by participants for experiment 1
86. 西山さんが 鳥取県の 砂浜が//
[ F ] なくなったとうそをついた。
[ F ] きれいだと言っていた。
[ F ] きれいだと教えた。
[ F ] とてもキレイだったと言っていた。
[ F ] きれいで美しいと言っていた。
[ F ] きれいだったと話していた。
[ F ] 美しいことを友達に話していた。
[ H ] 好きと言った。
87. 石田先生に 西山さんが この 作曲家が//
[ F ] どういった人物かたずねた。
[ F ] すばらしいんだと熱く語った。
[ F ] 亡くなったと、伝えた。
[ F ] 有名だよと言った。
[ H ] 一番好きだと言っていた。
[ H ] 好きだということを伝えた。
[ H ] 好きなことを伝えた。
[ H ] 私は大好きなんですと熱く語った。
88. 石田先生に 西山さんが 鳥取県の 砂浜が//
[ F ] 広がっていることを告げた。
[ F ] 今、危ないんですと言った。
[ F ] なくなればいい!とさけぶ。
[ F ] きれいかどうか聞いた。
[ F ] きれいだといった。
[ F ] きれいだと言った。
[ H ] 見たいとおねだりした。
[ Z ] 全員集合してください。
89. 斉藤さんが この 登山家が//
[ F ] ヒゲをはやした例のおやじだと説明した。
[ F ] 有名だと言った。
[ F ] 素晴しいと絶賛していた。
[ F ] エベレストを上れるわけがないと言った。
[ F ] 実は有名な人だといっていた。
[ F ] はじめて山を登った時の写真があるんだと
いった。
[ G ] スキなのはにんじんと教える。
[ H ] 好きだと言った。
90. 斉藤さんが ここの 並木道が//
[ F ] 秋になるときれいだと言っていた。
[ F ] 色づいたねと言った。
[ F ] 私の並木道ですと言った。
[ F ] すごくきれいだと言っていた。
[ F ] きれいよと教えてくれた。
[ H ] 好きだと言っていたのを覚えている。
[ H ] 好きだと言ってたことを告げる。
[ H ] 好きだと言った。
91. 長谷川先生に 斉藤さんが この 登山家が//
[ F ] 有名な事をおしえた。
[ F ] 一番タフだと言っていた。
[ F ] あの有名な方ですねと言った。
[ F ] 私の父ですと言った。
[ F ] すごいんだと話していた。
[ F ] かっこいいと言った。
[ F ] 有名な方ですよと話していた。
[ G ] 登った山が近くにあることを話した。
92. 長谷川先生に 斉藤さんが ここの 並木道が//
[ F ] 秋と春がきれいなんです。と言った。
[ F ] 美しいことを伝えた。
[ F ]桜で満開になる頃は君も卒業ですねと語った。
[ F ] きれいであると聞いた。
[ F ] つぶれると、聞いた。
[ F ] 秋になるとキレイですよと教えていた。
[ H ] 好きだと言った。
[ H ] 一番落ちつくんですと言っていた。
93. 岡本さんが あの お年寄りが//
[ F ] かわいそうに見えた。
[ F ] 今にも倒れそうだと言っていた。
[ F ] かわいそうだと言った。
[ F ] 一人暮らしなので世話をしている。
[ F ] 辛そうだと言って席をゆずった。
[ F ] 今にも倒れそうと心配している。
[ F ] 困っていることに気付いた。
[ G ] 作る煮物はとても美味しいと言っていた。
135
94. 岡本さんが ハワイの 気候が//
[ F ] すばらしいと言った。
[ F ] 自分には合わないと言っていた。
[ H ] 一番好きだと思った。
[ H ] 好きなので今度ハワイに行く。
[ H ] とても好きだと言っていた。
[ H ] きになると言った。
[ H ] 気になるみたいとつぶやく。
[ Z ] すごくいい。
95. 鈴木君に 岡本さんが あの お年寄りが//
[ F ] 困っていると教えた。
[ F ] この前の事件の被害者だと教えてやった。
[ F ] 苦しそうにしているよ、と耳うちした。
[ F ] おばあちゃんに似ていると言った。
[ F ] 困っていると言っていることを教えた。
[ F ] さっき車にはねられそうになってたことを
教えてもらった。
[ F ] 大変そうだから手伝おうと提案した。
[ F ] 君と寄んでいるよと伝えた。
96. 鈴木君に 岡本さんが ハワイの 気候が//
[ F ] とてもあたたかいよと教えられてた。
[ F ] 爽やかだと言っていた。
[ F ] よろしくないと報告した。
[ F ] 君には合っているよと話した。
[ F ] 良かったと言っていた。
[ F ] 気持ちよかったと言った。
[ F ] 例年より暑くなっていると伝えた。
[ H ] あわないと言った。
Appendix E Completions provided by
participants for experiment 2
The following are the completion sentences that were provided by participants for the second
experiment.
1. 黒田先生が奥様が//
[ A ] ご出席なさっている。
[ C ] そろってスーパーで買い物をされていた。
[ H ] 大好きだとおっしゃっていた。
[ I ] 亡くなったことで今もふさぎこんでいる。
[ I ] 子供を産んだと嬉しそうな顔で私たちに話
してくれた。
[ K ] 来ていますよとおっとに伝えた。
2. 黒田先生が 作品が//
[ A ] こわれた。
[ I ] いまいちだと言っていた。
[ I ] 完成したと喜んだ。
[ I ] 完成したと言っていた。
[ I ] いまいちだと一人言を言っていた。
[ J ] 良いとほめてくれた。
3. 黒田先生が 自分の 奥様が//
[ F ] 女優だという事を自慢してきた。
[ F ] 病気であることを告白した。
[ F ] 浮気しているのではないかと探偵に依頼
した。
[ F ] 病気で亡くなったと告げた。
[ F ] 最近韓流にはまっていると愚痴って言って
いた。
[ Y ] スーパーで見たと言っていた。
4. 黒田先生が 自分の 作品が//
[ F ] とてもすばらしいとほめた。
[ F ] 美しいと自我自賛した。
[ F ] すばらしいと思っていた。
[ F ] 一番だということを自慢してきた。
[ F ] 「すばらしい」とぜっさんしている。
[ J2 ] クラスで一番うまいと言ってくれた。
5. 松本先生が 両親が//
[ C ] 私の事を言う。
[ C ] 話をしている。
[ I ]海外旅行に行くので一緒に行くと言いだした。
[ I ] 医者であるとじまんしてきた。
[ J ] 育て方をまちがえていると指摘した。
[ K ] 来られたよと告げる。
6. 松本先生が 気質が//
[ A ] はげしい。
[ A ] あらくなった。
[ G3 ] 悪い子供だとため息をついた。
[ G3 ] 激しい人には気をつけろとおっしゃった。
[ G3 ] 荒い人は嫌いだと言っていたが、実は松
本先生が一番気質が荒い。
[ X ] 激しいと言っていた。
7. 松本先生が 伊藤さんの 両親が//
[ F ] 教師だということをいろんな生徒に言った。
[ F ] 電車に乗っているのを見た。
[ F ] 怒っていることをしらせた。
[ F ] 共働きで大変ですねと言っていた。
[ F ] 離婚したことを伊藤さんから聞いた。
[ F ] 亡くなった事を伝えた。
136
137
8. 松本先生が 伊藤さんの 気質が//
[ F ] なかなか良い所までいっていると語って
いた。
[ F ] おかしいと見抜いた。
[ F ] 荒いことを気にかけているらしい。
[ F ] あらいと言った。
[ F ] 良いと褒めていた。
[ F ] 変だと言いふらした。
9. あきら君が 母親が//
[ A ] 殴られた。
[ H ] 怖いと私に訴えた。
[ I ] 参加日に来ることを嫌がっている。
[ I ] たたいて泣いている。
[ I ] 急に倒れたことを話してくれた。
[ I ] しんどそうだと言っている。
10. あきら君が 目標が//
[ A ] ありません。
[ H ] 達成したことをおしえてくれた。
[ H ] 断たれたと告げる。
[ I ] 総理大臣になることと大きなことを言って
いた。
[ L ] ナイとさけんでる。
[ L ] ないと悩んでいた。
11. あきら君が ゆみちゃんの 母親が//
[ C ] みんなそれを見ていた。
[ F ] 病気なことを知った。
[ F ] いなくなったと泣き出した。
[ F ] たおれたと知らせにやって来た。
[ F ] とてもキレイだと言っていた。
[ F ] 買いものしているのを見つけた。
12. あきら君が ゆみちゃんの 目標が//
[ F ] 壮大なことに驚いた。
[ F ] 自分と同じだと知って喜んだ。
[ F ] 高くておどろいた。
[ F ] テストで一番になる事だと言っていた。
[ F ] 甘いとアドバイスした。
[ H ] うらやましいと言った。
13. 河崎さんが 娘が//
[ B ] 欲しいらしい。
[ I ] いなくなったとさわいでいる。
[ I ] もうすぐ結婚するのだと社内で言っていた。
[ I ] 8 才になるとうれしそうに言っていた。
[ I ] 今年成人式なんだと言っていた。
[ I ] 最近反抗して大変だと言っていた。
14. 河崎さんが 精神が//
[ D ] 狂いそうな人だと思った。
[ I ] 不安定なことを気にしているらしい。
[ I ]安定する薬を処方してもらっているのを見た。
[ I ] 安定していないと気付いた。
[ J ] やんでいるんだと私に忠告した。
[ J ] 異常をきたしていると教えてくれた。
15. 河崎さんが 森田課長の 娘が//
[ B ] かわいくてしょうがないらしい。
[ F ] かわいいと言っていた。
[ F ] 結婚式で泣いているのを知った。
[ G ] 使うためのペンを買ってあげた。
[ H ] かわいくて結婚したいと話していた。
[ H ] 嫌いと言った。
16. 河崎さんが 森田課長の 精神が//
[ F ] まっすぐなことをほめていた。
[ F ] すばらしいと言っていた。
[ F ] 少し弱いと思ったそうだと耳にした。
[ F ] 安定していないのを批判した。
[ F ] 異常だと驚いていた。
[ F ] おかしいと言った。
17. 佐藤さんが 孫が//
[ B ] かわいくて、多くの写真をかざっていた。
[ C ] 一緒に運動会で走っていた。
[ H ] かわいくて仕方がないと私に言った。
[ I ] 遊んでいるのをながめていた。
[ I ] 今来ているんですよと近所の人に話した。
[ I ] 可愛いと自慢している。
138 Appendix E. Completions provided by participants for experiment 2
18. 佐藤さんが 髪型が
[ A ] 変わっていて、変だった。
[ A ] ぐるぐるになった。
[ H ] 気にくわないとぐちをこぼした。
[ I ] すてきとほめられた。
[ I ] なかなかきまらないと言っていた。
[ J ] 良くにあうとほめてくれた。
19. 佐藤さんが 上田社長の 孫が//
[ F ] 家出をしたとウワサした。
[ F ] 女優だと初めて知った。
[ F ] 美しい人だと褒めていた。
[ F ] 頭がいいと言った。
[ F ] スイミングスクールに通っていると教えて
くれた。
[ F ] お誕生日なのでプレゼントを買いに行った。
20. 佐藤さんが 上田社長の 髪型が//
[ F ] 「ハゲ」てると言った。
[ F ] 毎日同じだと言っていた。
[ F ] ヅラだと言われた。
[ F ] 変だと笑った。
[ F ] かつらだったことに気付いた張本人だ。
[ F ] 変だと言っていた。
21. 岸田君が 長男が//
[ F ] 両親の老後の面倒をみなくてはいけないと
いうことを私におしえてくれた。
[ H ] うっとおしくていやだと話していた。
[ I ] 生まれたと電話してきた。
[ I ] とても良い兄であることを紹介した。
[ I ] 倒れたと言っている。
[ Z ] 次男が三男が四男が...。
22. 岸田君が 将来が//
[ A ] 見えない。
[ H ] 心配だと話していた。
[ H ] 不安で仕方がないと先生に相談している。
[ H ] ふあんだといって先生にそうだんする。
[ H ] 不安だと嘆いていた。
[ I ] 楽しみだとみんなから言われている。
23. 岸田君が 田村さんの 長男が//
[ B ] 見たいそうだ。
[ F ] 大学を卒業したことに祝いの言葉をくれた。
[ F ] すごいと言っていた。
[ F ] 大学に受かったと言っていた。
[ F ] 立派な人だとほめていた。
[ F ] 結婚することを言い回っている。
24. 岸田君が 田村さんの 将来が//
[ F ] 有望だと期待している。
[ F ] バラ色のものになるだろうと言っている。
[ F ] あぶないと占った。
[ F ] すばらしいものになるようにと願った。
[ H ] 楽しみだと言っていた。
[ H ] 不安だと気付く。
25. 清水さんが 妹が//
[ G ] 持っているアクセサリーをぬすんだ。
[ I ] 最近うるさいと私に愚痴ってきた。
[ I ] 風邪だと思った。
[ I ] 最近色気付いてきたと言っている。
[ I ] 一番かわいいと思っている。
[ I ] 病気になったことを打ち明けた。
26. 清水さんが 言動が//
[ A ] みんなをまどわす原因となった。
[ A ] 違う。
[ A ] 怪しいとみんなが気にしていた。
[ A ] 流行語になった。
[ J ] おもしろいねと言ってくれた。
[ K ] 腹がたつといっていた。
27. 清水さんが つよし君の 妹が//
[ C ] 二人で帰っていた。
[ F ] 公園に向かって歩いて行ったことを教えた。
[ F ] ピアノを習っているのを聞いてびっくりし
ていた。
[ F ] 本屋でアルバイトしているのを目撃した。
[ F ] とても可愛いと言っているのをきいた。
[ F ] 交通事故にあったと聞いた。
139
28. 清水さんが つよし君の 言動が//
[ F ] おかしいと思っていることがみんなにば
れた。
[ F ] きついと注意していた。
[ F ] 変だと注意した。
[ F ] おかしいことに気がついた。
[ F ] よくないと注意した。
[ H ] 気に入らないとぼやいていた。
29. 鈴木先輩が いとこが//
[ I ] 来ているから、今日は急いで帰ると言った。
[ I ] 来ていると言っていた。
[ I ] 知り合いの人におごってもらったと言って
いたと言った。
[ I ] 来ることを告げて帰ってしまった。
[ I ] かわいいと自慢していた。
[ L ] いないことを悩んでいる。
30. 鈴木先輩が 行動が//
[ A ] 不信だ。
[ D ] なんだかおかしいとうたがっていた。
[ D ] おかしいと思った。
[ G3 ] 怪しいと思っている人物に会いにいった。
[ G3 ] おかしい私を注意した。
[ K ] あきらかにおかしいと予想した。
31. 鈴木先輩が 西川君の いとこが//
[ F ] 自分と同じサークル仲間だと言っていた。
[ F ] 鈴木という名だと言った。
[ F ] 犯人だったことをつきとめた。
[ F ] 実は兄弟だったことを知った。
[ H ] 好きだと言っていた。
[ H ] 知り合いだと知っている。
32. 鈴木先輩が 西川君の 行動が//
[ F ] おもしろいと言っていた。
[ F ] どうやらお笑い芸人の「レギュラー」の西
川君にものすごく似ていると、色々な人に言いふらし
ている。
[ F ] あやしいと語った。
[ F ] 正しいと言った。
[ F ] おかしいと指摘した。
[ F ] おかしいと言っていた。
33. かよちゃんが 恋人が//
[ C ] そろって両親にあいさつをしに来た。
[ H ] できたことを嬉しそうに話した。
[ I ] 最近冷たいと悩んでいた。
[ I ] 来年遠くへ引っ越すと悲しんでいた。
[ I ] 変だと言った。
[ Y ] 誕生日の日にプレゼントをあげていた。
34. かよちゃんが 性格が//
[ A ] いいと評ばんだ。
[ D ] 気に入らないと言った子がいる。
[ D ] いい事を言いふらした。
[ D ] 丸くなったと大人たちは言う。
[ G3 ] 良い彼氏をつくった。
[ I ] 悪いと悩んでいた。
35. かよちゃんが 小倉さんの 恋人が//
[ F ] 他の女の人と歩いているのを見たらしい。
[ F ] 私をたたいたの!と号泣していた。
[ F ] うわきしているこ所を目撃した。
[ F ] いいねと言った。
[ F ] 死んだと思いこむ。
[ F ] さっききれいな人と歩いていたと言って
いた。
36. かよちゃんが 小倉さんの 性格が//
[ D2 ] 好きらしいことを知った。
[ F ] 悪いと言っていた。
[ F ] 変だと思った。
[ F ] 悪すぎると言っていた。
[ H ] 合わないとぐちっていた。
[ H ] 好きだと言っていた。
37. 福田さんが ペットが//
[ H ] かわいくてしかたがないと話していた。
[ I ] 死んだことで泣いていた。
[ I ] 死んだと言った。
[ I ] 逃げたと困っていた。
[ J ] かわいいとじまんする。
[ Y ] ころした。
140 Appendix E. Completions provided by participants for experiment 2
38. 福田さんが 夢が//
[ A ] ふくらんだ。
[ F ] 叶うためには信じることよ!と自信満々に
言っているのを聞いた。
[ I ] 叶ったと嬉しそうに話した。
[ I ] 消えたと話していた。
[ L ] ないと、なげいていた。
[ L ] あると言って空を見あげていた。
39. 福田さんが 吉村君の ペットが//
[ F ] とても可愛いと言っていた。
[ F ] 死んだことを知らせるメールを送ってきた。
[ F ] かわいいと皆に言い広めた。
[ F ] 鳴き声がうるさいと文句を言っていた。
[ F ] 元気であることを知らせた。
[ F ] かわいいと言っていた。
40. 福田さんが 吉村君の 夢が//
[ F ] 無謀すぎると密かに思った。
[ F ] とてもステキだと感動していた。
[ F ] 叶わないと告げた。
[ F ] あまりにリアルでおどろいていた。
[ F ] パイロットだとみんなに言った。
[ F ] 現実になったら素敵といっている。
41. 中村さんが 師匠が//
[ C ] 大あわてしている。
[ G ] 作った絵を破いた。
[ H ] 恐いとおびえていた。
[ I ] 言っていたコトを思い出していた。
[ I ] 亡くなったことを嘆いていた。
[ I ] 最近ピリピリしていると恐がっていた。
42. 中村さんが こころが//
[ H ] 痛いとなげいている。
[ H ] いたむとうったえてきた。
[ H ] やんでいることを伝えた。
[ H ] 病んでいると悩んでいた。
[ H ] 病んでいると相談してきた。
[ I ] こわれちゃうと甘えてきた。
43. 中村さんが 大沢先輩の 師匠が//
[ F ] とても偉大な人であると改めて思ったと
言った。
[ F ] 亡くなったと言っていた。
[ F ] 病気だと伝えた。
[ F ] こないのだとうったえた。
[ F ] 書道の展覧会に出品しているのを見つけた。
[ Y ] こわがっていた。
44. 中村さんが 大沢先輩の こころが//
[ F ] 寛大な人だと言っていた。
[ F ] せまいと言っていた。
[ F ] 広いと思った。
[ F ] 変わっていたことに傷ついた。
[ H ] よめることを告白した。
[ H ] 分からないとなげいていた。
45. 山崎係長が 恩師が//
[ F ] 人生で大切な役割を示すと言った。
[ I ] 亡くなったことに大変悲しんでいた。
[ I ] 訪ねてきたと話してくれた。
[ I ] 亡くなったことで泣いていた。
[ I ] 急死したと悲しんでいた。
[ K ] 探している事を伝えた。
46. 山崎係長が 考えが//
[ A ] 堅いので代わりに部長に言ってみた。
[ G3 ] 正しい部下をほめた。
[ H ] まとまらなくてこまっている。
[ I ] 古いと先方からお怒りをうけた。
[ J ] 甘いと私に言った。
[ X ] たいていあんをしてきた。
47. 山崎係長が 武田さんの 恩師が//
[ F ] 亡くなったことを伝えてくれた。
[ F ] 武田さんをバカにしていたとウソをついた。
[ F ] とてもすばらしいとほめた。
[ H ] 知り合いで今度会うと言っていた。
[ Y ] 会った。
[ Z ] 意味不明...。。。
141
48. 山崎係長が 武田さんの 考えが//
[ F ] いいとさんしょうした。
[ F ] 正しいと言った。
[ F ] おかしいと指摘していた。
[ F ] おもしろいと取り入れた。
[ F ] まちがっているのではないかと指摘した。
[ F ] 一番今の会社の政策には合っているだろう
と判断した。
49. 田中先生が 弟子が//
[ H ] かわいいと思っている。
[ I ] あそびに行ってしまったと怒った。
[ I ] 成長している様子を自慢気に話した。
[ I ] 成長しているのを実感すると感嘆していた。
[ I ] 自分の思い通りにはいかないと悩んでいる
のを相談に乗っていた。
[ L ] いることを、誇りに思っている。
50. 田中先生が 頭が//
[ A ] いい事は有名だ。
[ F ] おかしいと困ると言った。
[ G3 ] 悪い生徒をなぐったらしい。
[ H ] 痛いとうったえていた。
[ H ] いたいと言って保健室に行った。
[ H ] 痛いと言っていた。
51. 田中先生が 西田さんの 弟子が//
[ F ] 入院したと教えてくれた。
[ F ] 結婚したと思った。
[ F ] たずねてきたと伝言した。
[ F ] あっちにいたと教えてくれた。
[ F ] すばらしいとほめていた。
[ H ] 欲しいと言ってきた。
52. 田中先生が 西田さんの 頭が//
[ F ] ハゲていることを言った。
[ F ] はげていると言っていた。
[ F ] ハゲてると思った。
[ F ] 良くなっていることを聞いた。
[ F ] 茶髪になったことに気付いた。
[ F ] 良いとほめていた。
53. 野中君が 上司が//
[ H ] にくいと悩んでいた。
[ H ] ムカツクと言う。
[ H ] 好きなのに、と言われた。
[ I ] いないことに気付いた。
[ I ] とても良い人だとうれしそうに話している。
[ I ] うるさいと言う。
54. 野中君が 好みが//
[ A ] かたよっている。
[ D ] おかしいと一緒に話していて思った。
[ H ] うるさいと豪語していた。
[ I ] 変わったのだと言っていた。
[ K ] おかしいと友達をひはんする。
[ Y ] 知りたい。
55. 野中君が 宮下さんの 上司が//
[ F ] 行った不正を知っていた。
[ F ] 優しいのでうらやましいと言っていた。
[ F ] もうすぐ退職することを聞いた。
[ F ] 男前だといううわさを聞いたらしい。
[ G ] いない所で話している。
[ H ] こわいと言っていた。
56. 野中君が 宮下さんの 好みが//
[ B ] 知りたかった。
[ F ] 少しおかしいと言っていた。
[ F ] おかしいと言いふらした。
[ F ] 悪いとからかった。
[ H ] わからないと悩んでいる。
[ H ] よく分からないといっている。
57. 藤田部長が 部下が//
[ I ] 罪を犯したと気付いた。
[ I ] 沢山いて助かると言っている。
[ I ] 1 人も出勤していない事に驚いた。
[ I ] ミスをしたのをかばった。
[ I ] 大変出来がよいことを自慢している。
[ I ] あまり働かないので社長に相談した。
142 Appendix E. Completions provided by participants for experiment 2
58. 藤田部長が 体調が//
[ A ] 悪いようだ。
[ A ] よろしくないそうだ。
[ I ] すぐれないと社長につげ口した。
[ I ] 悪いので会社を休んだ。
[ I ] すぐれないと保健室にかけこんだ。
[ I ] すぐれないと言った。
59. 藤田部長が 自分の 部下が//
[ F ] 犯したミスを一生懸命相手に謝っている。
[ F ] なやんでいるので相談に乗ってあげていた。
[ F ] 無能だと嘆いていた。
[ F ] 仕事に燃えているのを喜んだ。
[ F ] どこかへ行ってしまったとなげいていた。
[ H ] 心配だとぐちをこぼした。
60. 藤田部長が 自分の 体調が//
[ F ] よくないと言っていた。
[ F ] 悪いと言っていた。
[ F ] 悪いと部下に言っていた。
[ F ] よくない事を隠している事をみやぶった。
[ F ] 良いと医師から聞いて安心した。
[ F ] 良くないといって早退した。
61. 小林君が 先輩が//
[ H ] 恐いと言っていた。
[ H ] こわいと私に言った。
[ H ] 怖いと言った。
[ I ] 来れないことにがっかりしていた。
[ J ] さがしていたと教えてくれた。
[ X ] すきなことは知っている。
62. 小林君が 態度が//
[ A ] ころっとかわっていた。
[ D ] 変わったと周りの人が言った。
[ G3 ] 悪い人を注意した。
[ I ] 悪いと両親にしかられていた。
[ I ] 悪いと先生にしかられている。
[ K ] 気にくわねぇーんだよっ!と急に発狂した。
63. 小林君が 石塚さんの 先輩が//
[ B ] 嫌いです。
[ F ] 付き合ってくれと言ってくると言う。
[ F ] とても美しいと言っていた。
[ F ] ステキだと言っていた。
[ G ] 唯一気に入った後輩だ。
[ Y ] おこられた。
64. 小林君が 石塚さんの 態度が//
[ F ] あやしいと言った。
[ F ] いつもと違うと不安がっていた。
[ F ] おかしい、とまっ先に気がついた。
[ F ] 悪いと怒っている。
[ H ] きに入らないと言った。
[ H ] 気に入らないとおこっていた。
65. 橋本さんが 後輩が//
[ C ] 共に先生に謝っていた。
[ H ] 好きだと言っていた。
[ I ] 来ないと言った。
[ I ] 来ましたと言って連れてきた。
[ I ] なまけていたので、喝を入れた。
[ I ]自分の悪口を言っているのを聞いてしまった。
66. 橋本さんが 笑顔が//
[ A ] 魅力的だ。
[ G3 ] かわいい女の子が好きと言った。
[ H ] 苦手だと前に言っていた。
[ I ] すてきとほめられた。
[ I ] ステキだと言われて、照れていた。
[ J ] 可愛いとほめてくれた。
67. 橋本さんが 寺田君の 後輩が//
[ F ] さっきコンビニにいたと教えてくれた。
[ F ] ケガをしていると教えてくれた。
[ F ] 来ていたことを教えてくれた。
[ F ] いなくなったと報告を受けた。
[ H ] 結構タイプなのと言っていた。
[ Y ] たたいた。
143
68. 橋本さんが 寺田君の 笑顔が//
[ F ] ステキと言った。
[ F ] 醜いと言いふらした犯人だ。
[ F ] きもちわるいと言われた。
[ F ] すてきだと言っていた。
[ H ] 好きだと言った。
[ H ] 好きと言っていた。
69. 山本部長が 友人が//
[ C ] あいさつをした。
[ C ] みんなして笑う。
[ I ] 別荘をもっていて今度遊びにいくと自慢し
ていた。
[ I ] 亡くなったので欠席するという旨を伝えて
きた。
[ I ] 結婚することに喜んだ。
[ L ] いないと相談してきた。
70. 山本部長が 発言が//
[ A ] 苦しかった。
[ I ] まちがいであったと謝罪した。
[ J ] ないと怒られた。
[ K ] 少ないのではないかと部下を怒った。
[ K ] おかしいと怒った。
[ K ] おかしいと、佐藤くんにののしった。
71. 山本部長が 佐々木君の 友人が//
[ F ] 少ない事を心配した。
[ F ] 訪ねてきていると教えてくれた。
[ F ] 優秀だと聞いておどろいた。
[ F ] 自分 (部長) の友人でもあったことに驚い
ていた。
[ F ] 会社に来ているのを告げた。
[ H ] うらやましいとなぜか言っていた。
72. 山本部長が 佐々木君の 発言が//
[ F ] 一番正しいと言った。
[ F ] おかしいを主張した。
[ F ] 活気的だと誉めていた。
[ F ] 常識外だと非難する。
[ F ] 軽率だと注意した。
[ F ] 失礼だと説教した。
73. 渡辺部長が 依頼人が//
[ I ] 来ないことに腹を立てていた。
[ I ] 逃げたと勘違いした。
[ I ] むずかしいとつぶやいた。
[ I ] あの有名芸能人だったので驚いていた。
[ I ] 殺人者だと感付いた。
[ L ] いないと言っている。
74. 渡辺部長が 振る舞いが//
[ A ] 最近妙だ。
[ A ] きれいであると社長がほめていた。
[ A ] なんだか怪しい。
[ A ] おかしかった。
[ A ] はらだたしい。
[ K ] 悪いとおこった。
75. 渡辺部長が 土屋さんの 依頼人が//
[ C ] お互いに困惑していた。
[ F ] 来ましたと言って部屋につれてきた。
[ F ] ゆくえ不明だと言った。
[ F ] 実は土屋さんの夫であることをうちあけた。
[ F ] 見つかったと報告してきた。
[ F ] あやしいことを教えてくれた。
76. 渡辺部長が 土屋さんの 振る舞いが//
[ F ] 上品だとほめていた。
[ F ] ゆうがだと言っていた。
[ F ] 適切でないと注意した。
[ F ] すごく良いとほめた。
[ F ] あまりにも目にあまるものがあったので社
長に報告した。
[ H ] 気に入らないとぐちをこぼした。
77. 小川君が 隣の人が//
[ D2 ] 好きだということはみんな知ってる。
[ G ] 読んでいる新聞をのぞきこんだ。
[ H ] 気になると言った。
[ I ] 引越してきたことに期待した。
[ I ] おみやげをくれたと言って喜んだ。
[ I ] 倒れていると警察に通報した。
144 Appendix E. Completions provided by participants for experiment 2
78. 小川君が 服装が//
[ I ] へんだと周りから言われていた。
[ I ] なっとらん!と怒られていた。
[ I ] だらしないことを先生に注意された。
[ J ] 地味すぎると私に言った。
[ J ] 変だと教えてくれた。
[ K ] 変になっていく気がすると佐々木を見て
言った。
79. 小川君が 久保さんの 隣の人が//
[ F ] 誰か知らない。
[ F ] どんな人なのか久保さんに聞いた。
[ H ] 大好きだといっていた。
[ J ] 好きだと言われた。
[ X ] 不倫をしている相手だ。
[ Z ] ラムちゃんになった。
80. 小川君が 久保さんの 服装が//
[ F ] あんまり似合ってないと私に言った。
[ F ] おもしろいと言っていた。
[ F ] 素敵だと言っていた。
[ F ] かわいいねと話していた。
[ F ] はでだと言った。
[ F ] 変だと言った。
Appendix F Completions provided by
participants for experiment 3
The following are the completion sentences that were provided by participants for the third
experiment.
1. 黒田先生は奥様が//
[ A ] 美しい
[ B ] とても自慢だ
[ B ] とっても好きです
[ I ] 病気になったと心配していた
[ I ] フィットネスに通っているのを知らない
[ I ] とてもきれいなのを自慢している
2. 黒田先生は作品が//
[ A ] かたい
[ A ] 個性的だ
[ B ] 気にいらなかった
[ I ] 展示されることになり喜んでいる
[ I ] 拘りの魂である事を自慢した
[ J ] 見事だとほめてくれました
3. 黒田先生は自分の奥様が//
[ B ] キライ
[ B ] 好きだ
[ F ] 美人だと自まんする
[ F ] 働いていることを快く思っていない
[ F ] ケーキが好きなことを知っている
[ H ] 気に入っているのでやさしくしている
4. 黒田先生は自分の作品が//
[ B ] 完成した
[ F ] すばらしいと思い込んでいる
[ F ] 賞をとったと大喜びだ
[ F ] 一生の宝物だと確信した
[ F ] 輝いているように見えた
[ F ] すばらしいと自負した
5. 松本先生は両親が//
[ I ] 生徒なのでびっくりしている
[ I ] 元気なことがうれしいといつも言っている
[ I ] 元気になったと思った
[ I ] 健全であることを喜んでいる
[ J ] 在宅か聞いてきた
[ X ] 知り合いです
6. 松本先生は気質が//
[ A ] よい
[ A ] 荒い
[ A ] あらいがいい先生だ
[ A ] 荒いです
[ A ] 荒くむずかしい
[ A ] とてもやさしい人です
7. 松本先生は伊藤さんの両親が//
[ B ] すきなのよ
[ B ] 大好きです
[ F ] 元気なのか心配だ
[ F ] 離婚すると聞いて驚いた
[ F ] 離婚すると聞いて驚いた
[ F ] 相談に来られたので教室にいます
8. 松本先生は伊藤さんの気質が//
[ B ] 理解できない
[ B ] 知りたい
145
146 Appendix F. Completions provided by participants for experiment 3
[ F ] 激しいことを心配している
[ F ] あらいので気に入っていない
[ F ] 荒いと思っている
[ H ] 気に入って指導に熱を入れた
9. あきら君は母親が//
[ B ] 嫌いだ
[ B ] 大好きです
[ B ] 大好き
[ B ] 好きだ
[ I ] うるさく言うのにうんざりしている
[ I ] 元気でいることを願っている
10. あきら君は目標が//
[ A ] 大きすぎる
[ A ] ない
[ B ] 達成できた
[ B ] きまらない
[ B ] 見つからない
[ I ] 10 秒を切ることだが今は 15 秒です
11. あきら君はゆみちゃんの母親が//
[ B ] 大好きです
[ F ] 病気だと聞いた
[ F ] やさしいお母さんだと思っている
[ F ] 美人でうらやましがった
[ F ] いないことを知った
[ G ] くれたおかしを食べている
12. あきら君はゆみちゃんの目標が//
[ F ] 霊能者になる事だと言われて呆れた
[ F ] 総理大臣だと知った
[ F ] 立派だと感激していました
[ F ] 何なのかわからない
[ F ] 高すぎることを心配している
[ F ] 低いと笑う
13. 河崎さんは娘が //
[ A ] 結婚するらしい
[ B ] きらいじゃけ
[ B ] 可愛くて仕方がない
[ B ] 大嫌いだ
[ F ] 入院したので休んでいる
[ I ] かわいくて仕方ないと言っていました
14. 河崎さんは精神が//
[ A ] 不安定
[ A ] やんでいる
[ A ] 複雑な構造をしている
[ A ] タフである
[ A ] 安定せずにねこんでいる
[ H ] 入っていないので教育されねばならない
15. 河崎さんは森田課長の娘が//
[ B ] 好きになった
[ B ] 気に入りました
[ F ] 大変美しいと思っている
[ F ] 同じ会社に入ってくるときいた
[ F ] 自分の部下になって困っている
[ F ] 自分の娘と友達なので家族ぐるみでよく遊
びに行く
16. 河崎さんは森田課長の精神が//
[ F ] 弱っているので病院へ行くように進めた
[ F ] 病んでいると感じている
[ F ] 崩壊していることが分かった
[ F ] 強靭で尊敬した
[ F ] 強いことを知った
[ F ] 不安定なのを心配している
17. 佐藤さんは孫が //
[ B ] 大好きだ
[ B ] かわいくてしかたがない
[ B ] かわいくてしかたがない
[ B ] かわいくて仕方がない
[ B ] とてもかわいかった
[ H ] 大変かわいいと思っている
18. 佐藤さんは髪型が//
[ A ] 変わっている
[ A ] ライオンのたて髪によく似ている
[ A ] きまらない
147
[ A ] はじけている
[ H ] 気にいらないのでやり直してもらった
[ H ] 気になって何も手につかない
19. 佐藤さんは上田社長の孫が//
[ B ] 好み
[ B ] 好きだ
[ F ] 優秀だと知っている
[ F ] 入社してくることを聞いた
[ F ] 大きくなったのにおどろいた
[ H ] 大好きでよく遊んであげている
20. 佐藤さんは上田社長の髪型が//
[ B ] 気に入った
[ B ] 気に入りません
[ B ] 大変気に入っている
[ F ]風になびくのを見て気になってしょうがない
[ F ] とてもユニークだと思う
[ F ] 変わったことに気付かないようだ
21. 岸田君は長男が //
[ A ] 今年東大に入学したそうだ
[ A ] 生まれた
[ F ] 大学に合格したと大喜びだ
[ I ] 医者であると言った
[ I ] 犯人だと確信した
[ I ] 成人になるのを楽しみにしている
22. 岸田君は将来が //
[ A ] 心配だ
[ A ] 輝いている
[ A ] 明るい
[ G3 ] 楽しみな学生だ
[ G3 ] 楽しみな人だ
[ G3 ] とても楽しみな青年です
23. 岸田君は田村さんの長男が//
[ B ] 大好きだ
[ B ] 大好きではない
[ B ] お気に入りのようです
[ F ] 万引きをしたのをだまっていた
[ F ] 13 人兄弟の末っ子だと聞いていた
[ G ] 通う学校の教師になった
24. 岸田君は田村さんの将来が//
[ B ] 見える
[ B ] 心配でならなかった
[ B ] 心配だ
[ F ] どうなるか気になっている
[ F ] 明るいと思っている
[ H ] 気になり相談相手になった
25. 清水さんは妹が //
[ B ] かわいい
[ B ] 好きです
[ F ] 優秀だと思っている
[ F ] 幼いのでよく世話をしている
[ F ] 病気なのを知らない
[ I ] かくれている所を知っている
26. 清水さんは言動が//
[ A ] おかしい
[ A ] 一致しないです
[ A ] 不安定だ
[ A ] はげしすぎる
[ E ] のんびりなのでペースが合わない
[ I ] 乱雑だが本当はとてもやさしく気がきく人
です
27. 清水さんはつよし君の妹が//
[ F ] 美人だったので驚いた
[ F ] バレーを習っていると聞いた
[ F ] お兄ちゃんのゲームをかくしたことを知っ
ている
[ F ] いたずら好きなので困っています
[ F ] 歩いているのを見てしまった
[ F ] 人見知りだと思っている
28. 清水さんはつよし君の言動が//
[ B ] とても気になる
[ B ] 信じられんのやろか?
[ B ] 気に食わなかった
148 Appendix F. Completions provided by participants for experiment 3
[ B ] 気にいらない
[ F ] まちがっていると感じた
[ F ] ムカつくと思った
29. 鈴木先輩はいとこが//
[ A ] 医者だ
[ B ] こわい
[ F ] いないことに気付いた
[ F ] 合格したのをよろこんだ
[ F ] 男前だと自慢している
[ F ] 来年進学すると聞きつけてお祝いに行った
30. 鈴木先輩は行動が//
[ A ] 早い
[ A ] あらっぽい
[ A ] てきぱきしている
[ A ] 早い
[ A ] へんだ
[ Y ] 見えない
31. 鈴木先輩は西川君のいとこが//
[ B ] 気に入りました
[ F ] ひとまわりちがうのでショックを受けている
[ F ] 駅でゴミを拾っているのを見た
[ F ] 働き口を探していると聞いている
[ F ] 自分のいとこでもあると言っていた
[ F ] 同級生なのに気付いた
32. 鈴木先輩は西川君の行動が//
[ B ] 大変気になります
[ B ] どうしても許せなかった
[ B ] 気にくわない
[ F ] 落ちつきがないと指摘した
[ F ] おかしいと気付いた
[ F ] 挙動不審なのに気付いた
33. かよちゃんは恋人が//
[ A ] スポーツ選手でかっこいい
[ B ] 新しくできた
[ I ] たよりなので不満に思っている
[ L ] いたらいいなと思っています
[ L ] ずっといなくて 1 人ぼっちだ
[ Z ] いない歴 30 年
34. かよちゃんは性格が//
[ A ] 明るい
[ A ] ひん曲がっている
[ A ] とってもいい
[ A ] ややこしい
[ D ] 良いとみんなに評判である
[ I ] いいのですかれている
35. かよちゃんは小倉さんの恋人が//
[ B ] 知りたい
[ B ] 恐い
[ B ] 気になってしょうがない
[ B ] 気にいらない
[ F ] 自分に似ているかも知れないと思っている
[ F ] やさしいのでうらやましく思っている
36. かよちゃんは小倉さんの性格が//
[ B ] うらやましい
[ B ] とても気に入っている
[ B ] 好きです
[ F ] 寛大で尊敬している
[ F ] 変だと言った
[ F ] やさしすぎるので心配になった
37. 福田さんはペットが//
[ B ] 好きだ
[ B ] きらいです
[ G ] 大好きなベットで寝ている
[ H ] かわいくていつも一緒にいるらしい
[ H ] 大好きで飼いたいが飼える住宅ではない
[ I ] かなしそうな表情をするとかなしくなる
38. 福田さんは夢が//
[ A ] 大きい
[ A ] ない
[ A ] あった
[ I ] かなうまで努力するそうです
[ I ] かなうといいなと思っています
[ I ] かなってうれしいと思っている
149
39. 福田さんは吉村君のペットが//
[ B ] 大好きです
[ F ] 暴れてかみつくのを見た
[ F ] アナコンダであった事を不審に思った
[ F ] にげ出したのでいっしょにさがした
[ F ] 可愛いので好きだ
[ F ] 犬である事がやっと分かった
40. 福田さんは吉村君の夢が//
[ B ] 気になる
[ F ] 成就しにくいものだと思っている
[ F ] かなうよう心から応援している
[ F ] かなうよう協力した
[ F ] かなえばいいと思っている
[ F ] かなうことを願っている
41. 中村さんは師匠が//
[ B ] こわい
[ B ] キライ
[ B ] こわい
[ G ] 読んでいた本をかりて読んだ
[ H ] 尊敬できる人だと感じている
[ I ] 亡くなってかなしんでいる
42. 中村さんはこころが//
[ A ] うきうきしている
[ A ] 広い
[ A ] 広いです
[ A ] 広いです
[ A ] とてもあたたかい人です
[ I ] 清らかで誰にでも優しい
43. 中村さんは大沢先輩の師匠が//
[ F ] 有名人だと知った
[ F ] 訪ねて来たと報告した
[ F ] 円楽だと知らなかった
[ F ] テレビに出ているのを見た
[ F ] 以外にもスポーツが出来ると聞いてビック
リしていた
[ F ] 自分の知っている人だったのでおどろきま
した
44. 中村さんは大沢先輩のこころが//
[ B ] ほしいのだ
[ B ] 気にしている
[ B ] 気になる
[ B ] わかっていない
[ F ] 病んでいると思った
[ F ] 実は鉄の魂なのではないかと密かに思った
45. 山崎係長は恩師が//
[ A ] オリンピック選手だ
[ B ] 大好きだ
[ B ] 好きなの
[ I ] 亡くなったと聞いて泣いた
[ I ] リストラされるのを聞いた
[ I ] 昨夜家に来たと言っていた
46. 山崎係長は考えが//
[ A ] あまい
[ A ] 甘かった
[ A ] 甘い
[ A ] まとまっていない
[ A ] 甘い
[ I ] 浅いので仕事がすすまない
47. 山崎係長は武田さんの恩師が//
[ F ] 自分の友人だと知って驚いた
[ F ] 彼の恩師でもあったのでうれしかった
[ F ] 自分の父であることを告げた
[ F ] 自分の知りあいだと教えてくれた
[ F ] 自分の後輩だと言った
[ X ] 知り合いだった
48. 山崎係長は武田さんの考えが//
[ B ] 理解できなかった
[ B ] 分からなかった
[ B ] 手に取るように分かる
[ F ] 一番正しいと思っている
[ F ] 良いと仕事にとり入れた
[ F ] あまりにも現実ばなれしているのでびっく
りした
150 Appendix F. Completions provided by participants for experiment 3
49. 田中先生は弟子が//
[ B ] 大変気にいってます
[ I ] 自分をうらぎると思った
[ I ] 横領していることを知った
[ I ] 練習しないのに腹を立てた
[ I ] 自分を越えるだろうと喜んだ
[ I ] 成長したことを嬉しく思っている
50. 田中先生は頭が//
[ A ] かたい
[ A ] 3 つある
[ A ] ボーリング王のように輝いている
[ A ] あまりよろしくない
[ E ] おかしいと思う
[ I ] いたいと言っておられた
51. 田中先生は西田さんの弟子が//
[ B ] かわいくてしかたがない
[ B ] 好き
[ F ] ついてきているのを見た
[ F ] 男の人であるので呼び出した
[ F ] 多数であることを不思議に思っている
[ F ] あまりにごうまんなので西田さんにこうぎ
した
52. 田中先生は西田さんの頭が//
[ B ] じゃまになった
[ F ] ハゲていることが気になって仕方ない
[ F ] とてもきれると思った
[ F ] 金髪なのでせめて茶髪にするよう指導して
いる
[ F ] うすくなったと言った
[ F ] ぬれていると言った
53. 野中君は上司が//
[ B ] きらいです
[ B ] 好きです
[ B ] 大好きだ
[ I ] 言うことには従順である
[ I ] うそをついていると思った
[ I ] ミスをしたので何とかフォローしようとし
ている
54. 野中君は好みが//
[ A ] かたよっている
[ A ] 少しかわっている
[ A ] はっきりしている
[ A ] 悪い
[ A ] うるさい
[ E ] 多いので何が食べたいのかわかりません
55. 野中君は宮下さんの上司が//
[ F ] 父の友人であることを知った
[ F ] おいしいと思う
[ F ] ウソをついていると思った
[ F ] 脳梗そくで倒れたと後から知らされた
[ F ] 浮気しているのを目撃した
[ F ] 私の高校時代の同級生であることを知りま
した
56. 野中君は宮下さんの好みが//
[ B ] 理解できない
[ B ] わからない
[ B ] 分からない
[ B ] わかった
[ F ] 自分とにていてうれしかった
[ F ] 明るいタイプであることを知らない
57. 藤田部長は部下が//
[ B ] 信用できない
[ B ] いまいち信用できない
[ B ] キライ
[ F ] 失敗するのをおそれている
[ F ] 困っているのを助けた
[ L ]多くいるのでいつもいそがしそうにしている
58. 藤田部長は体調が//
[ A ] 思わしくないらしい
[ A ] すぐれないようです
[ A ] すぐれない
[ A ] 悪い
[ A ] 良くなった
[ I ] 万金なのにズル休みをしたそうだ
151
59. 藤田部長は自分の部下が//
[ A ] たくさんいます
[ B ] 心配だった
[ F ] 失敗したことをカバーした
[ F ] 自分を尊敬していることにおどろいた
[ F ] 優秀なので仕事をまかせても安心である
[ L ] いてうれしかった
60. 藤田部長は自分の体調が//
[ B ] 気にかかるのね
[ F ] 悪いと言った
[ F ] 万全であることを自慢しています
[ F ] すぐれないので検査に行った
[ F ] 悪いことが理解できない
[ F ] 良くないと分かっていたが病院に行かな
かった
61. 小林君は先輩が//
[ B ] 大の苦手である
[ B ] あこがれだ
[ B ] 怖い
[ B ] 大好きでたまらない
[ I ] 入院している病院に行った
[ I ] 結婚をするのでお祝いをした
62. 小林君は態度が//
[ A ] よくない
[ A ] わるい
[ A ] 悪い
[ A ] 横柄だ
[ A ] 変だ
[ I ] 良いのでほめられている
63. 小林君は石塚さんの先輩が//
[ B ] 好きだ
[ F ] おすぎに似ているのでピーコのものまねを
していた
[ F ] オリンピックに出ると信じている
[ F ] 自分の妻であったことを知った
[ H ] 苦手で同席するのをさけている
[ H ] 怪しくて仕方ない
64. 小林君は石塚さんの態度が//
[ B ] 気に入らなかった
[ B ] 気に入らないそうだ
[ F ] よそよそしいのが気になった
[ F ] 悪いので反省するよう忠告した
[ F ] 生意気だと思っている
[ H ] 腹立つと言っている
65. 橋本さんは後輩が//
[ A ] いなかった
[ B ] 気になってしかたありません
[ B ] きらいだ
[ I ] 上達していると褒めた
[ I ] きそくを守らないのでおこっている
[ I ] 聞いてくれたお別れ会に出席した
66. 橋本さんは笑顔が//
[ A ] ひきつっている
[ A ] 気持ち悪い
[ A ] さわやかである
[ A ] だれよりもステキだ
[ A ] かわいい
[ I ] 怖いとよく言われている
67. 橋本さんは寺田君の後輩が//
[ B ] ずいぶん気になる
[ B ] きらいだ
[ B ] 気にいっている
[ B ] キライ
[ B ] 気になっている
[ F ] 実は同じ年であると知った
68. 橋本さんは寺田君の笑顔が//
[ B ] 目に浮かぶ
[ B ] 忘れられない
[ B ] 大好きです
[ B ] 気に入った
[ F ] 輝いているので思わず赤面した
[ F ] 消えるのを恐れた
152 Appendix F. Completions provided by participants for experiment 3
69. 山本部長は友人が//
[ A ] 多い
[ I ] もうすぐやってくると言いました
[ I ] 総理大臣であることを公表した
[ I ] 通るのを見た
[ I ] 数年ぶりに訪ねてきてくれて喜んだ
[ J ] 来るといつもごちそうしてくれる
70. 山本部長は発言が//
[ A ] 部下の反感を買いやすい
[ A ] 過激だ
[ A ] きびしいです
[ A ] おかしい
[ A ] おかしい
[ A ] 厳しい
71. 山本部長は佐々木君の友人が//
[ B ] 大の苦手だ
[ B ] 気になる
[ B ] きらいなの?
[ F ] 実は自分の弟だった事を初めて知った
[ F ] 結婚すると聞いてよろこんだ
[ F ] 娘の恋人だと知った
72. 山本部長は佐々木君の発言が//
[ B ] 気にいらない
[ F ] まとを得ていることで大きくうなずいた
[ F ] 失礼にあたるとして注意した
[ H ] 気に入ってほめた
[ H ] 気に入らず呼び出した
[ Y ] 痛かった
73. 渡辺部長は依頼人が//
[ B ] 信頼できない
[ I ] 誰かを知っている
[ I ] 誰だか知らないので気がついていない
[ I ] 来るのを待った
[ I ] 知人であることを知り驚いていた
[ I ] 知人とわかっておどろいた
74. 渡辺部長は振る舞いが//
[ A ] ていねいた。
[ A ] はでだ
[ A ] 上品だ
[ A ] りっぱです
[ A ] とても上品だ
[ E ] ハデで見ていてこっけいです
75. 渡辺部長は土屋さんの依頼人が//
[ F ] 私だと知らなかった
[ F ] 自分の母であることにおどろいた
[ F ] 若く驚いた
[ F ] 信用できる誠実な人だと思った
[ F ] 知人であると説明した
[ F ] 意外な人物で驚いた
76. 渡辺部長は土屋さんの振る舞いが//
[ B ] 気にいらんけぇのぅ
[ B ] 気にいらない様子です
[ B ] 気にいらない
[ F ] きれいだと思った
[ F ] 失礼だと感じた
[ H ] 癪に触ると同僚にもらした
77. 小川君は隣の人が//
[ B ] 好いちゅう
[ I ] のぞいているとうたがった
[ I ] 病気であるのに気がつかなかった
[ I ] うるさくて毎日困っている様子です
[ I ] 今日越して来た事を知った
[ I ] 夜中に出て行くのを見かけた
78. 小川君は服装が//
[ A ] 1 人だけへんだ
[ A ] 落ち着いている
[ A ] 乱れがちだ
[ A ] おしゃれだ
[ A ] おしゃれなので人気がある
[ A ] 奇抜だ
153
79. 小川君は久保さんの隣の人が//
[ B ] 気に入っている
[ F ]風変わりな人だと聞いて見てみたいと思った
[ F ] 怪しいと思った
[ F ] うるさいと不満を言った
[ F ] 具合が悪そうなのを久保さんに教えた
[ F ] 手品をしていたので負けずに落語をした
80. 小川君は久保さんの服装が//
[ F ] とても似合うと思った
[ F ] 春らしいのでかわいく思った
[ F ] 上下サイズが違うよと言った
[ F ] 変だと言った
[ F ] はでだと感じている
[ F ] 大人びていておどろいた
Bibliography
Abney, S. (1989). A computational model of human parsing. Journal of Psycholinguistic
Research, 18, 129–144.
Altman, G., & Steedman, M. (1988). Interaction with context during human sentence
processing. Cognition, 30, 191–238.
Amano, S., & Kondo, T. (1999). Nihongo no goitokusei. (Vols. 1-6, NTT database series).
Sanseido, Tokyo.
Babyonyshev, M., & Gibson, E. (1999). The complexity of nested structures in Japanese.
Language, 75, 423–450.
Barker, C. (1995). Possessive Descriptions. CSLI Publications, Stanford.
Berwick, R. C., & Fong, S. (1995). Madama Butterfly Redux: Parsing English and Japanese
with a Principles and Parameters Approach. In Mazuka, R., & Nagai, N. (Eds.),
Japanese Sentence Processing, pp. 177–208. Lawrence Erlbaum, Hilsdale, NJ.
Bever, T. G. (1970). The cognitive basis for linguistic structures. In R, H. I. (Ed.), Cognition
and language development, pp. 277–360. Wiley and Sons, Inc, New York.
Choi, H.-W. (2001). Binding and Discourse Prominence: Reconstruction in Focus Scram-
bling. In Legendre, G., Grimshaw, J., & Vikner, S. (Eds.), Optimality-Theoretic Syn-
tax, pp. 143–169. MIT Press.
Chomsky, N. (1981). Lectures on Government and Binding. Foris, Dordrecht.
Crain, S., & Steedman, M. (1985). On not being led up the Garden Path: The Use of Con-
text by the psychological Syntax Processor. In Dowty, D., Karttunen, L., & Zwicky,
A. (Eds.), Natural Language Parsing. Psychological, Computational, and Theoretical
Perspectives, pp. 320–358. Cambridge University Press, Cambridge, England.
Den, Y., & Inoue, M. (1997). Disambiguation with verb-predictability: Evidence from
Japanese garden-path phenomena. In Proceedings of the nineteenth annual conference
of the Cognitive Science Society, pp. 179–184 Hillsdale, NJ. Lawrence Erlbaum.
Farmer, A. K. (1980). On the interaction of morphology and syntax. Ph.D. thesis, MIT.
154
Bibliography 155
Farmer, A. K. (1984). Modularity in Syntax: A Study of Japanese and English. MIT Press,
Cambridge.
Ferreira, F., & Clifton, C. (1986). The independence of syntactic processing. Journal of
Memory and Language, 25, 348–368.
Fodor, J. D. (1998). Learning to parse?. Journal of Psycholinguistic Research, 27, 285–319.
Frazier, L. (1987). Syntactic processing: Evidence from Dutch. Natural Language & Lin-
guistic Theory, 5, 519–560.
Frazier, L., Clifton, C., & Randall, J. (1983). Filling gaps: Decision principles and structure
in. sentence comprehension. Cognition, 13, 187–222.
Gentner, D. (2005). The development of relational category knowledge.. 245-75.
Gorrell, P. (1995). Syntax and Parsing. Cambridge University Press, Cambridge.
Gunji, T. (1987). Japanese Phrase Structure Grammar. Reidel Publishing Company.
Gunji, T. (1995). Ninchi kagakuno kiso, chap. 5, pp. 128–168. Iwanami shoten, Tokyo.
Gunji, T. (1999). On Lexicalist Treatments of Japanese Causatives. In Levine, R. D.,
& Green, G. M. (Eds.), Studies in Contemporary Phrase Structure Grammar, pp.
119–160. Cambridge University Press.
Gunji, T. (2002). Tangoto bunno koozoo. Iwanami shoten, Tokyo.
Hagstrom, P., & Rhee, J. (1997). Differences between Korean and Japanese processing
overload. Journal of Psycholinguistic Research, 26, 189–206.
Hale, K. (1980). Remarks on Japanese Phrase Structure: Comments on Papers on Japanese
Syntax. MIT Working Papers in Linguistics, 2, 185–203.
Harada, Y. (1991). ‘No’ ni tsuiteno jakkanno koosatsu. Gengo kyooiku kenkyuusho kiyoo,
42, 1–27.
Hirose, Y. (2003). Recycling prosodic boundaries. Journal of Psycholinguistic Research, 32,
167–195.
Hoji, H. (1985). Logical form constraints and configurational structures in Japanese. Ph.D.
thesis, University of Washington.
Iguchi, Y., & Miyamoto, E. T. (2006a). The interpretation of nominative NPs is modulated
by possessors relations. In 23rd Annual Meeting of the Japanese Cognitive Science
Society, pp. 118–119.
156 Bibliography
Iguchi, Y., & Miyamoto, E. T. (2006b). Possessor Relations and the Interpretation of
Nominative NPs in Japanese. In Proceedings of the 5th International Conference on
Cognitive Science, pp. 115–116.
Inoue, A. (1991). A comparative study of Parsing in English and Japanese. Ph.D. thesis,
University of Connecticut, Storrs.
Inoue, A., & Fodor, J. D. (1995). Information-paced Parsing of Japanese. In R., M., & N.,
N. (Eds.), Japanese Sentence Processing, pp. 9–63. Lawrence Erlbaum, Hilsdale, NJ.
Inoue, M. (2006). Nihongono bunrikaini okeru aimaiseino kaishooto horyuu. Cognitive
Studies: Bulletin of the Japanese Cognitive Science Society, 13, 353–368.
Inoue, M., & Den, Y. (1999). Influence of verb-predictability on ambiguity resolution in
Japanese. In Proceedings of the 2nd International Conference on Cognitive Science and
the 16th Annual Meeting of the Japanese Cognitive Science Society Joint Conference,
pp. 499–502.
Kageyama, T. (1993). Bunpooto gokeisee. Hitsuji shoboo.
Kamide, Y., Altmann, G. T. M., & Haywood, S. L. (2003). The time-course of prediction in
incremental sentence processing: Evidence from anticipatory eye movements.. Journal
of Memory and Language, 49, 133–156.
Kikuchi, R., & Shirai, H. (2002). Nihongo-meishiku-no imikaisyaku-no kentou. In 19rd
Annual Meeting of the Japanese Cognitive Science Society, pp. 78–79.
Kikuchi, R., & Shirai, H. (2006). Analysis and Interpretation of the Japanese Postposition
No. In Saint-Dizier, P. (Ed.), Syntax and Semantics of Prepositions, Springer, pp.
245–261. Kluwer Academic Pub.
Kim, Y. (1999). The effects of case marking information on Korean sentence processing.
Language and Cognitive Processes, 14, 687–714.
Kuno, S. (1973a). The structure of the Japanese language. MIT Press, Cambridge MA.
Kuno, S. (1973b). Nihon Bunpoo Kenkyuu. Taishukan, Tokyo.
Kuroda, S.-Y. (1978). Case-marking, canonical sentence patterns and counter equi in
Japanese. In Hinds, J., & Howard, I. (Eds.), Problems in Japanese syntax and se-
mantics, pp. 30–51. Kaitakusha, Tokyo.
Kuroda, S.-Y. (1992). Japanese Syntax and Semantics: Collected Ppers. Dordrecht, Kluwer.
Bibliography 157
Kurohashi, S., & Nagao., M. (1997). Kyoto University text corpus project (in Japanese).
In Proceedings of the Third Annual Meeting of The Association for Natural Language
Processing, pp. 115–118.
Lee, S.-H., & Nakayama, M. (2003). Effects of syntactic and phonological similarity in Ko-
rean center-embedding constructions. In Poster presented at the 16th Annual CUNY
Conference on Human Sentence Processing MIT and Northeastern University.
Lewis, R., & Nakayama, M. (2002). Syntactic and positional similarity effects in the pro-
cessing of Japanese Embeddings. In Nakayama, M. (Ed.), Sentence processing in East
Asian languages, pp. 54–85. CSLI, Stanford.
Marslen-Wilson, W. D. (1975). Sentence perception as an interactive parallel process. Sci-
ence, 189, 226–228.
Mazuka, R. (1991). Processing of empty categories in Japanese. Journal of Psycholinguistic
Research, 20, 215–232.
Mazuka, R., & Itoh, K. (1995). Can Japanese Speakers Be Led Down the Garden Path?.
In Reiko, M., & N., N. (Eds.), Japanese Sentence Processing, pp. 295–329. Lawrence
Erlbaum Associates, Hillsdale.
Mazuka, R., Itoh, K., & Kondo, T. (2002). Cost of scrambling in Japanese sentence pro-
cessing.. In Nakayama, M. (Ed.), Sentence processing in East Asian languages, pp.
112–131. CSLI, Stanford.
Mazuka, R., & Lust, B. (1990). On parameter setting and parsing: Predictions for cross-
linguistic differences in adult and child processing. In Frazier, L., & De, Villiers, J.
(Eds.), Language processing and language acquisition. Kluwer, Dordrecht, Holland.
McRae, K., Hare, M., Elman, J., & Ferretti, T. (2005). A basis for generating expectancies
for verbs from nouns. Memory and Cognition, 33, 1174–1184.
Mitchell, D. C., Corley, M., & Garnham, A. (1992). Effects of context in human sentence
processing: Evidence against a discourse-based proposal mechanism. Journal of Ex-
perimental Psychology: Learning, Memory, and Cognition, 18, 69–88.
Miyagawa, S. (1980). Complex Verbs and the Lexicon. Ph.D. thesis, University of Arizona.
Miyamoto, E. T. (2002). Case markers as clause boundary inducers in Japanese. Journal
of Psycholinguistic Research, 31, 307–347.
Miyamoto, E. T. (2006). Understanding sentences in Japanese bit by bit. Cognitive Studies:
Bulletin of the Japanese Cognitive Science Society, 13, 247–260.
158 Bibliography
Miyamoto, E. T., & Nakamura, M. (2005). Unscrambling some misconceptions: a comment
on Koizumi and Tamaoka. Gengo Kenkyu, 128, 113–129.
Muraoka, S., & Sakamoto, T. (2004). Tougokaiseki-ni-okeru setsukyoukai-no goijouhou-
no eikyou. In 21st Annual Meeting of the Japanese Cognitive Science Society, pp.
206–207.
Nakayama, M. (1999). Sentence Processing. In Tsujimura, N. (Ed.), The Handbook of
Japanese Linguistics, pp. 398–424. Blackwell, Malden, MA.
Nakayama, M., Vasishth, S., & Lewis, R. (2006). Difficulty of certain sentence constructions
in comprehension. In Nakayama, M., Mazuka, R., & Shirai, Y. (Eds.), Handbook
of East Asian Psycholinguistics, Volume 2, Japanese. Cambridge University Press,
Cambridge, UK.
Nemoto, N. (1999). Scrambling. In Tsujimura, N. (Ed.), The Handbook of Japanese Lin-
guistics, pp. 121–153. Blackwell, Malden, MA.
Nishiyama, Y. (2003). Meishiku-no imiron-to goyouron. Hitsujishobou, Tokyo.
Nishiyama, Y. (1990). Kakiryooriwa hiroshimaga honbada koubunni tsuite - houwa meishi
to hihouwa meishiku-. keiougijyuku daigaku bunkakenkyuusyo kiyoo, 22, 169–188.
Partee, B. (1999). Uniformity vs. versatility: the genitive, a case study 1983/1997. In Johan,
van, B., & Alice, ter, M. (Eds.), The Handbook of Logic and Language, pp. 464–470.
Elsevier.
Prince, A., & Paul, S. (1993). Optimality Theory: Constraint Interaction in Generative
Grammar. Rutgers University and University of Colorado.
Pritchett, B. (1992). Grammatical competence and parsing performance. The University of
Chicago Press, Chicago.
Pritchett, B. L. (1991). Noun phrases in Japanese and English: A study in syntax, learn-
ability and acquisition. Ph.D. thesis, University of Connecticut, Storrs, CT.
Rayner, K., Carlson, M., & Frazier, L. (1983). The interaction of syntax and semantics
during sentence processing. Journal of Verbal Learning and Verbal Behavior, 22, 358–
374.
Ross, J. R. (1967). Constraints on Variables in Syntax. Ph.D. thesis, Massachusetts Institute
of Technology.
Saito, M. (1985). Some Asymmetries in Japanese and Their Theoretical Implications. Ph.D.
thesis, MIT.
Bibliography 159
Sakamoto, T., & Walenski, M. (1998). The processing of empty subjects in English and
Japanese. In Hillert, D. (Ed.), Syntax and Semantics Vol. 31 Sentence Processing: A
Crosslinguistic Perspective, pp. 95–111. Academic Press, San Diego.
Shibatani, M.and Cotton, C. (1976-1977). Remarks on double nominative sentences. Papers
in Japanese Linguistics, 5, 261– 278.
Shibatani, M. (1977). Grammatical relations and surface cases. Language, 53, 789–809.
Suh, S. (1994). The syntax of Korean and its implications for parsing theory. Ph.D. thesis,
University of Marylnad, College Park, MD.
Takahashi, C. (1994). Case, Agreement, and Multiple Subjects: Subjectivization in Syntax
and LF. In Akatsuka, N. (Ed.), Japanese/Korean Linguistics, pp. 394–441. CSLI,
Stanford.
Tanenhaus, M., Spivey-Knowlton, M., Eberhard, K., & Sedivy, J. (1995). Integration of
visual and linguistic information in spoken language comprehension. Science, 268,
1632–1634.
Tanenhaus, M. K., Carlson, G. N., & Trueswell, J. C. (1989). The role of thematic struc-
tures in interpretation and parsing. Language and Cognitive Processing, Parsing and
Interpretation, 4, 211–234.
Tanenhaus, M. K., Spivey-Knowlton, M., Eberhard, K. M., & Sedivy, J. C. (1995). Integra-
tion of visual and linguistic information in spoken language comprehension. Science,
268, 1632–1634.
Taraban, R., & McClelland, J. (1988). Thematic roles in on-line sentence processing. In
Miller, A., & Powers, J. (Eds.), Proceedings of the Fourth Eastern States Conference
on Linguistics, pp. 282–295. The Ohio State University, Columbus, OH.
Taraban, R., & McClelland, J. (1990). Parsing and comprehension: A multiple-constraint
view. In Balota, D., Flores d’Arcais, G. B., & Rayner, K. (Eds.), Comprehension
processes in reading, pp. 231–263. Lawrence Erlbaum Associates, Hillsdale, NJ.
Trueswell, J., Tanenhaus, M., & Garnsey, S. (1994). Semantic influences on parsing: use of
thematic role information in syntactic ambiguity resolution. Journal of Memory and
Language, 33, 285–318.
Uehara, K. (1997). Judgments of processing load in Japanese: the effect of NP-ga sequences..
Journal of Psycholinguistics Research, 26, 255–263.
160 Bibliography
Uehara, K., & Bradley, D. C. (2002). Center-embedding problem and the contribution of
nominative case repetition. In Nakayama, M. (Ed.), Sentence Processing in East-Asian
Languages, pp. 257–287. CA: CSLI, Stanford.
Vermeulen, R. (2005). Possessive and adjunct multiple nominative constructions in
Japanese. Lingua, 115, 1329–1363.
Weinberg, A. (1993). Parameters in the theory of sentence processing: Minimal committ-
ment theory goes east. Journal of Psycholinguistic Research, 22, 339–364.
Weinberg, J. (1987). Parameters in the Theory of Sentence Processing: Minimal Commit-
ment Theory Goes East. Journal of Psycholinguistic Research, 22 (3), 339–364.
Whitman, J. (1979). Scrambled, over easy, or sunny side up. In Proceedings of the Fifteenth
Annual Meeting of the Chicago Linguistic Society, pp. 342–352.
Yamashita, H. (1997). The effects of word-order and case marking information on the
processing of Japanese. Journal of Psycholinguistic Research, 26, 163–188.
Yatabe, S. (1996). Long-distance scrambling vis partial compaction. MIT Working Papers
in Linguistics, 29.
Recommended