fig? ‘|\\\|\‘i\l‘hHWWNWWINWI

   

M318

lllllllllllllllllHIlllHllllllllll'lllllllllllllllllllllllll

31293 01801 73

LIBRARY
Michigan State
University

 

 

 

 

This is to certify that the

thesis entitled
jafkmflse, B. st‘cmi’. ,JZ 9WCWVQ9
presented by

Ta L’ami 0611*

has been accepted towards fulfillment
of the requirements for

/(’{Ll<>i?i' Kg Am’degree in [I‘lngutSrtcS

 

 

bud, Lot, /§f_J,—’\

Major professor

Date /"'/é ~ 7?

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINB return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1198 WWW“

JAPANESE EXISTENTIAL STRUCTURES

By
Takami Oda

A THESIS

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

MASTER OF ARTS
Department of Linguistics

1998

ABSTRACT

JAPANESE EXISTENTIAL STRUCTURES

By
Takami Oda

The present study analyzes Japanese existential structure, A 112'
(Dative/Locative) B ga (Nominative) aru/iru (to exist), and explains various
constraints on its use, including scrambled ga ni aru/iru sentences and double
nominative ga ga aru/iru sentences. The Japanese structure of the form A ni
B ga aru/iru sometimes expresses the locative meaning, '8 is in A', and
sometimes the possessive meaning, 'A has 3'. Muromatsu (1994) proposes
two different structures to account for the two interpretations and tries to
explain a constraint on scrambled sentences, however I have found many
data which camot be explained by her analysis. while Muromatsu discuss the
m' ga aru/iru structures syntactically, she does not discuss when they can most
appropriately used. This study analyzes the constraints on m' ga aru

constructions which have been unexplained in previous analysis.

TABLE OF CONTENTS

1. Introduction .................................................................................................................. 1
2. Previous research on existentials ............................................................................. 5
2.1. Freeze (1992) on English have .................................................................... 6
2.2. Hornstein, Rosen and Uriagereka (1994) on 'integral' and 'spatial'
interpretation .......................................................................................... 10
2.3. Analysis on have ........................................................................................ 12
2.3.1. Problems with Freeze (1992) ...................................................... 12
2.3.2. Cowper (1989) and Ritter and Rosen (1993) on unified
theory of have ............................................................................ 14
2.4. Existential sentences in Japanese ............................................................. 16
2.4.1. Word order in Japanese .............................................................. 17
2.4.2. Muromatsu (1994) on integral and spatial sentences in
Japanese ........................................................................................ 20
2.5. Kato (1991) on topic-subject constructions ............................................ 22
2.6. Summary ...................................................................................................... 25
3. Locatives and possessives in Japanese .................................................................. 26
3.1. THEME ga/wa LOC 111' EXIST ................................................................... 26
3.2. LOC ga/wa THEME ga EXIST ................................................................... 32
3.3. Constraints on the LOC/POSS m' THEME ga EXIST structure .......... 37
3.3.1. Integral sentences ......................................................................... 40
3.3.2. Spatial sentences .......................................................................... 42
3.3.3. Summary ....................................................................................... 46
4. Conclusion .................................................................................................................. 48
Bibliography... ................................................................................................................. 51

III

 

INTRODUCTION

The present study analyzes Japanese existential structure, A m'
(Dative/ Locative) B ga (Nominative) aru/z'rul (to exist), and explains various
constraints on its use, including scrambled2 ga ni aru/iru sentences and double
nominative ga ga aru/iru sentences. As many linguists have suggested that
locatives, possessives and existentials are closely related to each other (Lyon
1967, 1968b, Clark 1978, Ultan 1978, Givon 1984), the Japanese structure of the II?
form A ni B ga aru/iru sometimes expresses the locative meaning, '8 is in A', and

sometimes the possessive meaning, 'A has B', as shown in the examples below.

(1) a. Reezookoni ringo ga aru. (locative) '4
fridge Loc apple NOM exist um
’An apple is in the fridge.’

 

b. John ni kodomo ga iru. (possessive)
DAT child NOM exist
'John has a child.’

Muromatsu (1994) considers the Japanese sentence (2) below ambiguous
between locative and possessive. She proposes two different structures, shown
in (3a) and (3b), to account for the two interpretations, and claims that the
ambiguity is due to the ’surface identity’ between the dative-marker 111' and the

locative ni.

(2) Kuruma ni engine-ga aru
car DAT / in engine-NOM exist
’There is an engine in the car.’

 

‘ The verb, am is used for inanimate objects and iru is used for animate beings.
2 Scrambling is a phenomenon in which preverbal major constituents are reordered (Shibatani
1990: p. 259).

 

(3) a. Kuruma ni engine ga aru.
car DAT NOM exist
SUBJECT PREDICATE

’The car has an engine.’

b. Kuruma ni engine ga aru.
car in NOM exist
PREDICATE SUBJECT
’In the car is an engine.

She also argues that the ambiguity disappears when engine is moved to the front.
According to her, sentence (4) means ’An engine is sitting in the back seat of the

car or elsewhere’ rather than ’An engine is an integral part of the car.’

(4) Engine ga kuruma ni aru.
engine NOM car in exist
’An engine is in the car.’

Muromatsu explains that this ”absence of ambiguity” is due to structural
differences. Thus, the ga (nominative) ni (Dative/Locative) order is allowed for a
locative sentence, but not for a possessive sentence. However, (4) is still
ambiguous between the locative meaning, ’In the car is an engine.’ and the
possessive meaning, ’The car has an engine’, according to all of the native
Japanese speakers I asked and to me. To us the possessive meaning is actually
the first reading we get, rather than the locative meaning. Although
Muromatsu’s claim of (4) is problematic, it is true that there are certain sentences
which do not allow ga ni word order. Consider the examples which allow ga ni
word order as in (5b) and (6b), and the sentence which does not allow ga ni order

as in (7b).

(5) a. Niwa - ni neko ga iru.
garden LOC cat NOM exist
'There is a cat in the garden.‘

 

b. Neko ga niwa ni iru.
cat NOM garden LOC exist
'There is a cat in the garden.‘

(6) a. Rakuda ni kobu ga aru.
camel DAT hump NOM exist
'There is a hump on a camel.’

b. Kobu ga rakuda ni aru.
hump NOM camel DAT exist
'A camel has a hump.‘

(7) a. John ni kodomo ga iru. h
DAT child NOM exist
'John has a'childf

b. ? Kodomo ga John ni iru.
child NOM DAT exist
'John has a child.’

 

r...“

In this study, I analyze the constraints on scrambled sentences, that is
THEME ga LOC ni EXIST. I also present other data which cannot be explained
by Muromatsu's analysis in terms of the structural differences. For example,
there is a constraint on what can appear in the double ga (nominative)
constructions. Consider (8a)-(10b). There are some sentences which allow ga ga
order as in (8b) and (9b), and there are some sentences which do not allow ga ga

order as in (10b).

(8) a. John ni kodomo ga iru.
DAT child NOM exist
'John has a child.’

b. John ga kodomo ga iru.
NOM child NOM exist
'It is John who has a child.‘

(9) a. Kuruma ni engine ga aru.
car LOC/ DAT NOM exist
'There is an engine in the car.‘

(10)

b. Kuruma ga engine ga aru.
car NOM NOM exist
'It is a car that has an engine.‘

a. Niwa ni neko ga iru.
garden LOC cat NOM exist
'There is a cat in the garden.’

b. ? Niwa ga neko ga iru.

garden N OM cat NOM exist
'It is a garden that has a cat in it.’

While Muromatsu (1994) discusses the m' ga aru/iru structures

syntactically, she does not discuss when they can most appropriately be used.

As shown in (11), (12), (13), and (14), there are some ungrammatical ni ga aru/iru

sentences.

(11)

(12)

(13)

(14)

"' Enpitsu . ni keshigomu ga aru.
pencil DAT / LOC eraser NOM exist
'There is an eraser on the pencil.’

* Inu ni kusari ga aru.
dog DAT leash NOM exist
'There is a leash on the dog.‘

* Juice ni koori ga aru.
LOC ice NOM exist
'There is ice in the juice.‘

* Kata ni kami-no-ke ga aru.

shoulder LOC hair NOM exist
'There is a hair on his / her shoulder.’

In this thesis I first argue that the Japanese basic word order of existentials

is LOC/POSS ni THEME ga aru/iru, and analyze why there are such constraints on

scrambled sentences, and double nominative sentences, as well as on the

 

LOC/POSS ni THEME ga aru/iru structure itself. I also discuss topicalized
sentences in relation to the constraints on Japanese existentials.

The organization of this thesis is as follows. Following this introductory
chapter, Chapter 2 reviews current research on existential structures in English
and in Japanese. I also discuss Freeze’s (1992) analysis on English have since
much of the current research on existential structures follows this analysis.
Chapter 3 examines Japanese existential structures and analyzes their constraints.
Finally, Chapter 4 states the conclusions and the implications for future research

on Japanese existential structures.

2. Previous research on existentials

This chapter provides a review of previous research on the existential
structure in both English and Japanese. Section 2.1 discusses Freeze’s analysis of
the English have. Section 2.2 reviews Hornstein, Rosen and Uriagereka (1994)
regarding English existentials, focusing on the two interpretations of the have
structure, ’locative’ and ’possessive’. Section 2.3 discusses problems with this
analysis, and sections 2.4 and 2.5 critically review Muromatsu (1994) and Kato
(1991). Muromatsu (1994), mentioned above, analyzes Japanese existential
structure, LOC/POSS ni THEME ga aru/iru, following Hornstein, Rosen and
Uriagereka (1994) on English existentials and Freeze (1992) on English have.

Section 2.6 is a summary of this chapter.

 

2.1. Freeze (1992) on English have

Freeze (1992) proposes a unified theory of the ’locative paradigm’, which

consists of the locative, the existential, and the possessive categories. In his

theory, (15) and (16) are all derived from a single D-structure shown in (17).

(15) English:
a. Locative: The book is on the bench.
b. Existential: There is a book on the bench.
c. Possessive: Lupe has a book.
(16) Russian:
a. Kniga byla na stole.
book.NOM.FEM was on table.LOC.
'The book was on the table.’

b. Na stole. byla kniga.
on table.LOC. was book.NOM.FEM.
'There was a book on the table.’

c. U menja byla sestra.

at lsg.GEN was sister. NOM
'I had a sister.‘

(17) Freeze (1992: p. 558)3

 

 

IP
/\
XP 1'
/\
j’ PP
/\
NP P'
/\
I I
[11’ [SPEC e] [1' [1] [PP [s ECNP [ 'P NP]]]]

 

3 Freeze (1992) omits VP node assuming the existence of the VP is understood.

 

With locative and existential sentences from several languages, Freeze
shows that the two constructions represent different orderings of the theme and

the locative constituents.

(18):
ORDER LANGUAGE LOCATIVE EXISTENTIAL
SVO Russian T cop L L cop T
Catalan T cop L e p cop T L
VOS Chamarro cop L T cop T L
Palauan cop L T cop p T L
VSO Tagalog cop L T cop T L
SOV Hindi T L cop L T cop

(T=Theme L=Location cop=copula p=proform e=empty category)

Freeze observes that there are some languages in which a proform (= p) is used
in the existential constructions of Catalan and Palauan. In such a language, the
relative order of the theme and the locative might not differ between the
existentials and the predicate locative sentences. Although he does not explain
how the proform affects this ordering, he claims that different orderings of the
theme and the locative are accounted for in terms of movement as shown in (19a)

and (19b). (p. 559)

(19) Freeze (1992:,p. 559)
a. Predicate locative: the theme moves to [Spec, IP]

b. Existential: the locative argument moves to [Spec, IP]

For example, (21a-b) are the derivations respectively of (20a), the predicate

locative, and (20b), the existential in Russian:

 

(20) SVO (Russian):

Underlying Structure. [1P e [I' byla [PP [NP kniga] [P' na stole]]]]

a. [IP [NP kniga]i [I' byla [PP ti [P' na stola]]]]
book.NOM.FEM was on table.LOC
'The book was on the table.‘

b. [IP [P' na stole]i [I' byla [PP [NP kniga] ti]]]‘1
on table.LOC was book.NOM
'There was a book on the table.‘

 

[1
(21) Freeze (1992: p. 559) E
1P i
X? I'
/\ U
1 PP *
/\ b

 

 

 

j“ NP
[IP [SPEC e] [I' [byla] [PP [NP kniga] [P'na stile]]]]

a. [IP [NP knigali [I' byla [PP ti [P' na stola]]]]
book.NOM.FEM was on table.LOC
'The book was on the table.‘

b. [IP [P' na stoleJi [I' byla [PP [NP kniga] ti 1]]
on table.LOC was book.NOM
'There was a book on the table.‘

For Freeze, have is a problematic element, and he proposes that have
should be interpreted as a copula. In his framework, the have predication
constitutes the third member of the locative paradigm. One argument for this is
that many languages do not have different copulas in be predications and have

predications.

 

 

4 Movement of X' is crucial in his analysis as we can see in (19b) and (20b). According to Freeze,
X'-level constituents are movable categories excluding the specifier.

Freeze shows S-structures for the locative, the existential and have
predication in order to make his claim clear: The order of the existential is the

same as that of the have predication, as shown in (22).

(22):

ORDER EXAMPLE LOCATIVE EXISTENTIAL HA VE

SVO Russian T cop L L cop T L copT
VSO Tagalog cop L T cop T L cop T L

VOS Yucatec T cop P' cop T L cop T L

SOV Hindi T L cop L T cop L Tcop

The have predication is accounted for in the same way as the existential: it is
derived from the D-structure in (21) by movement of the locative argument to
subject position5 However, Freeze does not discuss what motivates the
movement or how the theme gets definite or indefinite marking.

He claims that 'in some languages the derivation of the have predication,
moving the locative argument to subject position, leaves the preposition
immediately adjacent to Infl.’ (p. 586) Assuming have is a copula just like be, he
also states: ’For have copula languages, I propose that this state of affairs results
in P moving to Infl and incorporating into it, yielding what corresponds
morphologically to have.’ (p. 586) He concludes that in this sense why the
absence of P in have predication in a language with have copula is no longer
mysterious. He also provides a sample derivation of an English have sentence as
in (23a-b):

(23) a. Torni [IP]i a book ti t.
(has)

 

5 He suggests that the difference between the existential and the have predication depends on the
[+ human] and [- human] feature of the locative subject, not on the difference in the verbs
themselves. However, I do not agree with his claim because there are some counterexamples;

e. g., The bathroom has a shower.

'JJTC r-‘n..

 

(23) b.
IP
Tornj I’

[be+Pi] VP

(has) /\
abook PP

 

Following Freeze's have derivation, shown in (23b), Hornstein, Rosen, and In
Uriagereka (1994) analyze English existential structures, to which we now turn.
2.2. Hornstein, Rosen and Uriagereka (1994) on ’integral’ and ’spatial’ .

interpretation

English there sentences can express two meanings, ’locative’ and
’possessive’. Hornstein, Rosen and Uriagereka (henceforth HRU) (1994) discuss

the ambiguity of the following sentence:

(24) There is a Ford T engine in my Saab.
Sentence (24) can express both (25a) and (25b).

(25) a. My Saab has a Ford T engine.
b. (Located ) in my Saab is a Ford T engine.

According to HRU, sentence (25a) expresses the ’integral’ relationship between
my Saab and a Ford T engine, that is, a Ford T engine is an integral part of the Saab.
On the other hand, (25b) simply expresses the location of a Ford T engine. This

engine needs not to be a part of the Saab; it can mean that the Ford T engine

10

happens to be sitting in the back seat of the Saab, though (25a) cannot. HRU call
the (25a) reading the ’integral’ interpretation (II), and (25b) the ’standard’
interpretation (S1).

The central claim of their work is that the ambiguity stems from two
different grammatical structures. Underlying the interpretation of II is a small

clause shown in (26a), while SI is derived from (26b).

(26) a. II: [SC my Saab [a Ford T engineJJ q
subject predicate , "
b. SI: [SC a Ford T engine [ in my Saab]] 3
subject predicate 5
Following Freeze’s have derivation, HRU analyze the structures t;
[L

 

underlying the two different structures of (26a-b). The relationship between my
Saab and a Ford T engine is different in II and SI sentences. The difference

between the derivations of H and SI is shown in (27a-b) and (28a-b).6

O 0
(27) a. [[SPEC my Saab]i [infl be +[D/P] j ] [SPEC a Ford T enginek ej [DP POSS ti Agr
t ll]
k

My Saab has a Ford T engine.

b [[spec a Ford T engine] [infl is] [PP in my Saabl]
A Ford T engine is in my Saab.

 

6 HRU treat the Determiner as a Preposition. They state that ’Kayne suggests [the Determiner] is
in some sense prepositional (p. 71); however, their claim is not clear.

11

(28)

a. ' b.
IP IP
/\ - /\
my Saabi I’ a FT engine I’
/\ /\
[be+Dj°=have] DP is PP
A /\
a FT enginek D’ P’
/\ /\
e]. AgrP PU DP
/\ in /\
DP Agr’ my Saab
ti /\
Agr0 tk

Although HRU consider Freeze's (1992) analysis of have to be accurate,

there actually appear to be several problems in it. The next section discusses

problems with this analysis. This discussion is relevant to the present study

since the major analyses of existential structures, such as HRU (1994) and

Muromatsu, (1994) are based on it. In section 2.3.1 I point out the problems, and

in section 2.3.2 I discuss other have analyses by Cowper (1998) and Ritter and

Rosen (1993) to argue against Freeze’s analysis.

2.3. Analysis on have

2.3.1. Problems with Freeze (1992)

the incorporation of be and a preposition. As HRU (1994) and Muromatsu (1994)

According'to Freeze (1992), as discussed in section 2.1, have is yielded by

show, existential 'there' sentences indeed have two possible readings; locative

('predicate locative' in Freeze) and possessive (’haoe structure' in Freeze). HRU

12

 

(1994), following Freeze, posit the relationships between there sentences and have

sentences as shown in (28a)-(29b).

(28) a. [[spec there] [inf] are] [8 tentaclesk][[) /P on [DPposs octOpus AgrO thJ]
There are 8 tentacles on an octopus.

b. [[spec An octopush [inf] be+[D/P]0j ] [8 tentaclesk ej [DPposs ti AgrO
tklll '5‘}

An octopus has 8 tentacles.

 

(29) a. b.
IP IP
/\ /\
there I’ an octopusi I’
/\ /\ t..-
are DP [be+Di°=have] DP
/\ /\
8 tentaclek D’ 8 tentaclek D’
/\ A
on AgrP D AgrP
/\ ej /\
DP Agr’ ti Agr’
octopus/\ /\
Agr0 tk tk

There are several problems with this analysis. First, these researchers do
not discuss what the mechanisms are for there insertion in existential structures
((28a) and (2%)), what motivates it, or how the theme gets definite or indefinite
markings (the octopus vs. an octopus). Second, this analysis does not account for

a have+PP sentence like (30) which has both a preposition and have.

(30) The car has an engine in it.

13

Third, there are many kinds of have; not only locative and possessive, but also
causative, and experiencer as shown in (31). It is hard to show on the tree that

the derivation of causative and experiencer have is also 'be+P'.

(31) a. Ihad a good time. experiencer
b. I had him run. causative

 

Freeze does not discuss causative or experiencer types, however, if his analysis is

correct, we seem to have to admit an uneconornical analysis. That is, one type of

 

have is derived from 'be+P', but the others are generated in some other way. In
the next section, I present other analyses which contradict Freeze's derivation of

have.

2.3.2. Cowper (1989) and Ritter and Rosen (1993) on a united theory of have

Cowper (1989) and Ritter and Rosen (1993) argue for a unified analysis of
have; namely, the verb have has only one representation. Cowper (1989) regards
have as an auxiliary and as a main verb to be identical. It is commonly assumed
(Pollack 1988, Grimshaw and Mester 1988), she claims, that the auxiliaries have
and be do not assign theta-roles, and this property makes an auxiliary an
auxiliary. She thereby claims that sentences containing have exhibit a fairly wide

range of theta-roles to the arguments of have, as shown in (31) (p.86).

14

(31) a. Mary has a performance tomorrow night.

Agent

b. Sue has a headache today.
Experiencer

c. Sue has visitors today.
Goal

(1. Michael has a new car.
Possessor

With these examples, Cowper points out that the roles assigned to the arguments

of have seem to be determined almost completely by the arguments themselves. F1
Ritter and Rosen (1993) also argue that there is only one have, and that the F
verb have, lacking thematic content, imposes aspectual interpretation. They call
an element which has no thematic information a ’functor predicate’. They
g. _

 

compare causative make with have, and claim that the meaning of make is fixed in
the lexicon while the meaning of have is not. In their examples shown in (32) and
(33), they show the meaning of make as always causative while have may denote

relations including location, possession, experience, or creation.

(32) a. John made a new cabinet. cause (creation)
b. Bill made the roast. cause (change of state)
c. Ralph made the bed. cause (change of state)
d. The car made a noise. cause (creation)

(33) a. Sheila had a good time. experience
b. Harold had a dinner party. cause (creation)
c. John has a new cabinet. possession
d. The cabinet has a stereo in it. location

The interpretation assigned to the sentence with have thus depends on the
relation between its subject and its complement. As Chomsky (1986) has already
mentioned, the theta role for subjects is compositionally assigned by the whole

VP.

15

In my view, Freeze's analysis of have is uneconornical and inadequate
since there are many types of have, as shown in (33). Unlike the possessive and
the locative types, the causative and the experiencer types cannot be derived
from be. It is more economical to assume that have has only one representation,
and I consider that have of any type is not derived from be+P.

Above, I discussed the analysis of have because previous studies on
existential sentences (e.g. HRU (1994) and Muromatsu (1994)) follow Freeze's
analysis. Although I basically agree with HRU's and Muromatsu's analyses
positing structural differences, they are not sufficiently precise.

In the next section, I will turn to previous research on Japanese

existentials.

2.4. Existential sentences in Japanese

This section provides a review of previous research on the Japanese
existential structure, LOC/POSS ni THEME ga EXIST. In section 2.4.1., I present
ni ga aru/iru as a canonical word order of Japanese existentials. This is relevant
because I include scrambled ga ni sentences in my analysis. In section 2.4.2., I
analyze Japanese existential structures, assuming that the canonical word order

of existentials is ni ga aru.

16

mm:

T.--.—. . _._ .——.
l;

2.4.1. Word order in Japanese

Word order in Japanese is relatively free except for the main predicate
which appears at the end of a sentence.7 In this section I first discuss a variety of
word order in Japanese sentences. I then show that the canonical order is X ni Y
ga aru/iru ’Y exists at location X’, and that Y ga X ni aru/iru is a result of
scrambling.
Kuroda (1978) discusses general word order in Japanese. First, he posits a .3
rule called 'Subject ni-Raising', which makes an embedded subject a clausemate
of the matrix verb and assigns it the particle ni. Sentence (35) exemplifies this
rule applied in the so-called indirect, or adversative, passive construction (34)

(Kuroda 1978: 31). , ,._._

(34) Taro (Hanako sono tegami yom) rareta
that letter read PASSIVE-PAST

(35) Taro ga Hanako ni sono tegami o yomareta.
N OM DAT that letter ACC read-PASSIVE—PAST
'Toro had that letter read on him by Hanako.’

Kuroda assumes that the subject and the object are base-generated without

particles, and formulates the rule of Case Marking as shown in (36) (p. 34).

(36) Case Marking: Mark the first unmarked noun phrase with ga, and
mark any other unmarked noun phrase or phrases
with 0.

Assuming (36), he presents canonical sentence patterns as follows. (p. 35)

 

7 An element can actually appear to the right of a verb in Japanese; i.e. in the postposing

construction (See for example, Simon 1989). Sentences of this type will not be treated herein,
however.

17

(37) Canonical Sentence Patterns (Kuroda 1978):
I Transitive sentence pattern: NP ga NP 0.
H Ergative sentence pattern: NP ni NP ga.

III Intransitive sentence pattern: NP ga.

Kuroda notes that the transitive and the intransitive patterns are self-

explanatory, and explains that the ergative pattern is one which is illustrated by

 

”A
such sentences as (38) and (39).
(38) Taro ni (wa) kane ga aru. (= Kuroda’s (16))
DAT TOP money NOM exist
'Taro has money.‘
(39) Taro ni wa gengogaku ga wakaranai. (= Kuroda’s (17)) z - -

DAT TOP linguistics NOM understand-NEG
'Taro does not understand linguistics.’

According to him, the base structure of the ergative sentence (38) is (40), and

Case Marking assigns the particle ga to the unmarked noun phrase kane (money)
in (40).

(40) Taro ni kane aru. (= Kuroda’s (18))
DAT money exist
’Taro has money.’

With regard to locative sentences in Japanese, Kuno (1973) claims that a
sentence like (41a) represents the canonical order, and (41b), its scrambled
counterpart. An example (1a) is repeated below as (41a). Kuno (1973) calls

sentences of the type in (41 a-b) ’existential’.

18

(41)

a. reezooko ni ringo ga aru.
LOC apple NOM exist
'There is an apple in the fridge.‘

b. ringo ga reezooko ni aru.
NOM LOC exist
’There is an apple in the fridge.’

Kuno observes that there are many existential sentences which are ’canonical’

with the locative NP in sentence-initial position, but which become awkward or
even unacceptable when their theme8 precedes the locative NP. Kuno (1973)

explains (p. 352):

"MED

With respect to certain grammatical features, existential sentences with
the word order LOCATIVE + SUBJECT behave similarly to nonexistential
sentences with the word order SUBJECT + OBJECT, and existential
sentences with the word order S + L behave similarly to nonexistential
sentences with the word order 0 + S. Since there is overwhelming
evidence that S + O + V is the basic word order for nonexistential
sentences, L + S + EXIST is the basic word order for existential sentences.

The following are. Kuno’s examples (p. 355).

(42)

a. Yama ni ki ga aru.
mountain LOC tree NOM exist
’There are trees on the mountain.’

b.?ki ga Yama ni aru.

tree NOM mountain LOC exist
'Trees exist (are) on the mountain.‘

Consider Kuroda's ergative sentence (38) again, repeated below as (43a).

Let us call a sentence of this type ’possessive’, because a sentence like (43a) only

 

3 Kuno (1973) uses the term theme for t0pics. Here I use theme for objects just as I have so far in
this study, and [will continue to do so.

19

expresses possessive meaning Taro has money. It does not express locative
meaning there is money on/at him. I adopt Kuroda’s canonical sentence patterns,
and regard (43a) to represent the canonical order of a possessive sentence and

(43b) its scrambled counterpart.

(43) a. Taro ni ° kane ga aru.
DAT money NOM exist
'Taro has money.‘
b. Kane ga Taro ni aru.

money NOM DAT exist
'Taro has money.‘

In the next section, I discuss Muromatsu (1994) on Japanese existential
structures, assuming that ni ga aru/iru represents the canonical order and that ga
ni aru/iru is the result of scrambling for both locative and possessive

constructions.

2.4.2. Muromatsu (1994) on integral and spatial sentences in Japanese

Muromatsu considers the Japanese existential sentence (44) ambiguous.

(44) kuruma ni engine-ga aru
car in. engine-NOM exist
’There is an engine in the car.’

Following HRU, she proposes two small clauses, shown in (45a) and (45b), to
account for the two interpretations of (44), and states that the ambiguity is due to
the ’surface identity’ between the dative-marker ni and the postposition ni.

Muromatsu calls HRU’s integral interpretations ’integral’ as in (45a), and their

20

standard interpretations ’spatial’ as in (45b). For convenience, I adopt

Muromatsu’s terms.

(45) a. Integral: [SC kuruma [ engine]]

car engine
b. Spatial: [SC engine [kuruma ni]]
engine car in
(46) a. kuruma ni engine ga aru.
car DAT NOM exist

SUBJECT PREDICATE
’The car has an engine.’

b. kuruma ni engine ga aru.
car in NOM exist
PREDICATE SUBJECT

’In the car is an engine.

Muromatsu argues that ambiguity disappears when engine is moved to the
front. According to her, only a spatial reading is available in (47a) whose

structure is presented in (47b). Note that (47b) is the same as (45b).

(47) a. engine ga kuruma ni aru.
engine NOM car in exist
’An engine is in the car.’

b. Spatials: [SC engine [kuruma ni]]
engine car in

Muromatsu claims that engine in (46b) can be preposed because it is a subject of
the spatial small clause, but the engine in (46a) cannot be scoped out to the front
because it is a predicate of the integral small clause. That is the reason why the
ambiguity disappears in sentence (47a). However, (47a) to me, is actually

ambiguous between integral and spatial readings, as well as to all the native

21

speakers of Japanese I have checked with. Similarly, if the theme, engine , is
topicalized as in (48), the meaning of the sentence can either be integral or
spatial.
(48) engine wa kuruma ni aru.
NOM car in exist

1. ’In the car is an engine’. = Spatial

2. ’An engine is something that a car has.’ = Integral

Although I agree that there are two different structures, integral and
spatial, Muromatsu's analysis of the ga ni aru structure is inadequate as there are
data which cannot be explained by her small clause analysis. In what follows, I

discuss such data that the small clause analysis cannot explain.

2.5. Kato (1991) on topic-subject constructions

Kato (1991) examines the topic-subject construction, wa ga, which encodes
possessive relationships. Kato uses Mikami's (1975) classic examples to analyze

the topic-subject constructions in (49) (Kato 1991: 37).

(49) a. 200 no hana wa nagai
elephants’ GEN trunk TOP long
’Elephants’ trunks are long.’

b. Zoo wa hana ga nagai
elephants’ TOP trunk NOM long
’Elephants’ have long trunks.’

The choice of the structures depends on what the speaker’s attention is directed
to. According to Mikami, (49a) is about the trunks of elephants, while (49b) is
about elephants. Kato discusses Wierzbicka's (1982) examples of English have,

22

shown here as (50a) and (50b), with corresponding Japanese sentences (51a) and

(51b), respectively (p. 37).

(50) a. John’s eyes are blue.
b. John has blue eyes.

(51) a. John no me wa aoi. (= Kato’s (2a))
GEN eye TOP blue

’John’s eyes are blue.’

b. John wa me ga aoi. (= Kato’s (2b))
TOP eye NOM blue
’John has blue eyes.’

Wierzbicka observes that

in a have sentence, a predication made about the object cannot be
conceptually separated from a predication about the subject: what is said
about the object is a way of saying about the subject.’ ’The only explicit
predicate mentioned here [(in 50b)], i.e. blue, applies to the eyes, not to
John. Yet thn, not his eyes, is the sentence’s main topic. John’s eyes are
better described in a have sentence [(50b)] than in a be sentence [(50a)]

because the former uses the predicate ascribed to John's eyes ’blue’ to say
something about John. (p. 37)

Kato points out that the wa ga construction is similar to an English have sentence,
although there are some differences between an English have possessive and a
Japanese wa ga possessive structure. While English have sentences are used to
demonstrate any kind of possession, such as temporary possession, according to
Kato, wa ga constructions are used only when there are permanent, close
relationships between two NPs, such as an inalienable possession or a part-whole

relationship as in (52).

23

(52) John wa kodomo ga ookii.
TOP child NOM big
’John has a big child.’

Kato claims that the wa ga construction highlights the conceptually close
relationships between two NPs, and in this way it is similar to an English have
sentence. However, I have found many counterexamples which contradict
Kato's claim that the wa ga construction is used only when two NPs have close

relationships. For instance, consider (53).

(53) a. John no . fuku wa kyoo kitanai.
GEN clothes TOP today dirty
’John’s clothes are dirty today.’

b. John wa kyoo fuku ga kitanai.
TOP today clothes NOM dirty
’John has dirty clothes today.’

In this study, I also analyze the conditions under which wa ga
constructions are allowed with existential sentences and those under which they
' are not. While Kato examines wa ga constructions with adjectival and nominal
predicates, I examine wa ga constructions with existential predicates, aru/iru_
’exist’. Kato notes in her footnote (p. 48) that in the wa ga construction, the topic
marker wa can bereplaced with the subject marker ga, and says that the topic
marker and subject marker are sometimes interchangeable. However, Kato does
not discuss double ga (nominative) constructions to give enough evidence to
claim that the topic marker and subject marker are interchangeable. There, I
disagree with her claim and examine double ga (nominative) constructions with

existential predicates, aru/iru 'exist'.

24

2.6. Summary

This Chapter reviewed previous research on English have and existential
structures in English and Japanese. Freeze (1992) analyzes have as being derived
from 'be+P', and this analysis is adopted in HRU (1994) and Muromatsu (1994).
However, there are several problems with Freeze's analysis. First, the
mechanisms for there insertion in existential structures are not discussed.
Second, this analysis does not account for sentences which have both a
prepositional phrase and have. Third, there are many types of have, not only
locative and possessive types, but also causative and experiencer types, as
Cowper (1989) and Ritter and Rosen (1993) discuss.

With regard to existential structures in English and Japanese, HRU (1994)
and Muromatsu (1994) proposed that the two interpretations, 'locative' and '
'possessive', are due to two different underlying structures. Based on Freeze’s
analysis of have, HRU and Muromatsu propose small clause analyses. In
Japanese, however, there are some data which cannot be explained by positing
small clauses. In what follows, I analyze such data, and examine what
conditions exist for the occurrence of existential structures such as THEME ga/wa

LOC/POSS ni exist, LOC/POSS ga/wa THEME ga exist, and LOC/POSS ni THEME

ga exist9

 

9 I do not discuss LOC/POSS ni THEME wa exist because this wa is not a topic marker. It does not

expresses thematic meaning, but contrastive meaning. It is not our purpose to examine
contrastive wa in this study.

25

3. Locatives and possessives in Japanese

As discussed in Chapter 2, HRU (1994) and Muromatsu (1994) claim that
existential sentences have two interpretations; 'locative' and 'possessive' in
English and Japanese respectively. HRU and Muromatsu's small clause analyses
argue that the two interpretations come from two different structures. There are,
however, some Japanese data which cannot be explained by a small clause
analysis. This Chapter examines such data. In section 3.1., assuming that
LOC/POSS ni THEME ga exist is the canonical word order, I analyze the
scrambled counterpart THEME ga LOC/POSS ni exist structures, which
Muromatsu's analysis fails to explain. The topicalized sentence, THEME wa
POSS ni exist structure is also examined. Section 3.2. presents the analysis of the
double ga construction and the topic construction. Finally, section 3.3. discusses
when the LOC/POSS ni THEME ga exist structure is used. I point out that there
are some locative and possessive sentences that sound awkward in the ni ga

aru/iru structure and also examine possible constraints.

3.1. THEME ga/wa LOC ni exist

In this section, first I make the terms, ’integral’ and ’spatial’ clear, then

discuss the constraints on THEME ga/‘wa LOC ni sentences. In an integral NP, ni

26

N P2 ga aru/iru sentence, the relationship between the two NPs is one of whole

and part, and the relationship is permanent.

(54) Integral- NP1 ni NPzga aru/iru:
NP1: whole

NP2= part
There is a permanenuflammp between NP1 and NP2.

Rakuda ni kobu ga aru.
camel DAT hump NOM exist
'A camel has a hump.’

The whole/ part relationship does not hold in a spatial sentence, and the

relationship between two NPs is temporary.

(55) Spatial- NP1 ni NPZ ga aru/iru:

NP1: location, container
NP2= object, person

There is a W between NP1 and NP2.

Niwa ni neko ga iru.

garden LOC cat NOM exist

'A cat is in the garden.‘

Now let us turn to discuss the THEME ga LOC ni sentences. since the
canonical order of Japanese existential sentences is ni ga aru/iru, the ga ni aru
order would sound unnatural to native speakers of Japanese. However, if koto

’nominalizer’ is added at the end of the sentences, they become more natural. A

sentence that is first judged to be ungrammatical becomes well formed by the

addition of koto.

27

I"

T
\t

(56) a. John ni kodomo ga iru.
DAT child NOM exist
’John has a child.’

b. ? kodomo ga John ni iru.
child NOM DAT exist

c. kodomo ga John ni iru koto.

child NOM DAT exist that
’That John has a child.’

It is unclear why koto has such an effect on a sentence. However, it is true that

without adding koto there is a difference in the judgment of grammaticality
among these sentences. For example:
(57) a. ? kodomo ga John ni iru.
child NOM DAT exist
‘John has a child.‘
b. propeller ga helicopter ni aru.

NOM DAT exist
’The helicopter has a propeller.’

Although I do not discuss the effect of koto in this study, it would be a very
interesting subject for future research.

When an element is topicalized, it is usually preposed to the beginning of
a sentence and is marked by the particle wa (Kuno 1973). Therefore, if a subject is
preposed and marked by ga, the sentence sounds awkward. However, even if a

preposed element is marked by wa, some possessive sentences are unacceptable.

For example:

28

(58) ? Kodomo wa John ni iru.
child TOP DAT exist
'John has achild.‘

Muromatsu (1994) claims that integral sentences express an inalienable
possession. A sentence such as (56a) only has an inalienable possession reading.
Consider the unacceptable sentence (56b). Muromatsu argues that a sentence
like (56b) is ungrammatical because ga ni sentences are available only for a
spatial reading. However, this claim is very questionable. As I stated earlier, all
the native speakers of Japanese I asked judge that ga ni sentences to have both
integral and spatial meanings. Besides that, I have found many sentences
indicating inalienable possessions (a type of part/ whole relations), which are
acceptable with ga ni order. In (59b), for example, the first NP (subject) is marked
by ga, and the sentence is well formed. This is because (59b) is a scrambled

sentence, not a topicalized one.

(59) a. Rakuda ni kobu ga aru.
camel DAT hump NOM exist
'A camel has a hump.‘

b. Kobu ga rakuda ni aru.
hump NOM camel DAT exist
'A camel has a hump.‘

In sentence (56b), kodomo 'child' is intended to refer to John‘s child, not any
child in general. I argue that sentence (56b) is unacceptable in its intended
meaning because the relationship between an antecedent and an anaphor is
unclear. In sentence (56b), kodomo 'child' needs an antecedent which indicates
whose child s/he.is. That is, John has to c-command kodomo 'child' for it to mean
John's child, not a child in general. However, the c-commanding relationship

between an antecedent and an anaphor seems to hold only for kinship

29

 

T

relationships. On the other hand, if a kinship relationship is not in question as in
(59b), the antecedent-anaphor relationship does not have to be held. In (59b),
kobu 'hump' refers, of course, to a camel's hump, but kobu ‘hump' does not have
to be c-commanded by rakuda ‘camel‘.10 This is because if the sentence can be
interpreted as spatial, the antecedent-anaphor relationship does not have to be
held. Note that sentence (56a) cannot mean spatial, 'There is a child with/by John’,
but only integral, 'John has a child’ as opposed to sentence (59a) which may mean
'There is a hump on a camel '. For (56b) to be interpreted in the spatial meaning, it

requires phrases like to issho ni 'with‘ or no soba ni 'by' as in (60).

(60) Kodomo ga John to issho ni (no soba ni) iru.
child NOM with by exist
’A child is with (by) John.‘

In sentence (60), kodomo ’child’ may or may not be John's. When the sentence is
spatial like (60), kodomo 'child' does not have to be c-commanded by John for it to
mean John's child. Another interesting problem for further research is why the
antecedent-anaphor relationship strictly applies to kinship relations in the
inalienable possession structures.

Now let us. consider the THEME wa LOC/POSS ni aru/iru structure.
According to Kuno (1973), the particle wa is either thematic or contrastive. I will
only discuss thematic wa, since the contrastive reading does not concern the
purpose of this study. As Kuno observes, theme must be anaphoric or generic.
Sentence (58), repeated below as (61), when used in isolation, violates the
anaphoricity of the particle wa because there is no antecedent which indicates

whose child he / she is. The particle wa cannot be generic because kodomo 'child‘ in

 

‘0 The reason for the difference in the grammaticality judgement between sentences like (56b) and
those like (59b) is not clear. As I state here, a kinship relationship seems to be crucial, but it may
also have to do with human vs. other subjects.

30

sentence (61) is not intended to mean a child in general, but John’s child,
specifically.
(61) ? Kodomo wa John ni iru.

child TOP DAT exist
‘John has a child.‘

Now consider sentence (62).

(62) Engine wa kuruma ni aru.
TOP car LOC / DAT exis
1. 'The engine is in the car.‘
2. 'An engine is something that a car has.’

The topic of sentence (62) can be any engine that the speaker wants to
anaphorically refer to; e. g. the engine in the back seat of a car. In another sense,
engine is generic, as in An engine is something that a car has. The topic can thus be
interpreted anaphoric or generic.

This section presented the analysis of ga ni sentences and wa ni sentences.
Muromatsu's (1994) analysis cannot be maintained because, as I show in (59b),
some inalienable possession sentences allow ga ni order. Ga ni order is not
allowed in kinship relation sentences because it seems the theme needs to be

bound by its possessor to indicate the relationship between the antecedent and

anaphor.

31

3.2. LOC ga/wa THEME ga exist

In this section, I will discuss the constraints on LOC ga THEME ga
sentences. Consider sentences (63a-b) and (64a-b), which are all integral. Double
ga is allowed in (63b) and (64b). Sentences (65a-b) and (66a-b) are all spatial.

Neither of the double ga sentences (65b) nor (66b) are acceptable.

(63) a. Kuruma ni engine ga aru.
car DAT N OM exist
'There is an engine in the car.‘

b. Kuruma ga engine ga aru.
car NOM NOM exist
'It is a car that has an engine.‘

(64) a. John ni kodomo ga iru.
DAT child NOM exist
‘John has a child.‘

b. John ga kodomo ga iru.
NOM child NOM exist
‘It is John who has a child.‘

(65) a. Reezooko ni ringo ga aru.
fridge LOC apple NOM exist
‘There is an apple in the fridge.‘

b. * Reezooko ga ringo ga aru.
fridge NOM apple NOM exist
‘It is a fridge that has an apple in it.‘

(66) a. Heya ni otoko ga iru.
room LOC man NOM exist
‘There is a man in the room.‘

b. " Heya ga otoko ga iru.

room NOM man NOM exist
It is a room that has a man in it.'

32

As Kuno (1973: 364) observes, unless the predicate of a sentence
represents a state or a habitual / generic action, double ga sentences are awkward.
In possessive sentences, POSS ga THEME "ga aru/iru , the subject 'possessor‘ is
always characterizable by the rest of the sentence. Since the possessive sentences
always characterize the subject possessor and the predicate represents a state,
possessive sentences are always grammatical in the ga ga structure, as shown in
(63b) and (64b). In the locative sentences, (65b) and (66b), it is odd to
characterize a generic fridge or room as having an apple or a man in it respectively.
Therefore, these sentences are ungrammatical. On the other hand, it is plausible
to characterize that fridge and that room as having an apple or a man in it,
respectively; therefore sentences in (67) and (68) are grammatical.12
(67) sono reezooko ga ringo ga aru.

that fridge NOM apple NOM exist
’It is that fridge that has an apple.’
(68) sono heya ga otoko ga iru.

that room NOM man NOM exist
’It is that room that has a man.’

Unlike the Japanese demonstrative sono 'that', neither the English demonstrative
that nor the definite article the works in this way. Sentences (69) and (70) are

ungrammatical.

(69) * That fridge has an apple.

 

11See footnote 8.
12I owe this observation to Mutsuko Endo Hudson (Personal communication).

33

(70) * That room has a man.

However, if the predicate of a sentence represents a state or a habitual / generic

action, the sentence is well-formed, as shown in (71b).

(71) a. * The room has a man.

b. The room has a guard.

We can surmise that characterizability or generalizability constrains English
sentences like (71a-b). I will discuss more about English sentences with have later
in this section.

Now let us consider wa ga sentences. Ktmo (1973) states that relative

clauses are constrained by the characterizability. Consider his examples.

(72) a. Asa wa itsumo usugurai uchi ni okita.
morning TOP always semidark while got-up
'In the morning, I always got up while it was still semidark.‘
b. * Itsumo usugurai uchi ni okita asa
always semidark while got-up morning

'the morning such that I always got up while it was still semidark.‘
According to Kuno (1973: 257), sentence (72b) is unacceptable because one does
not characterize morning in general as something in which the speaker got up

while it was still dark. However, if asa 'morning‘ is a specific morning, the

sentence becomes acceptable, as shown in (73a-b).

34

(73) a. sono asa‘ wa usugurai uchi ni okita.
that morning TOP semidark while got-up
'In that morning (in those mornings), I got up while it was still semidark.‘

b. Usugurai uchi ni okita asa
semidark while got-up morning
'those mornings when I woke up while it was still semidark.‘
I contend that the same applies to wa ga aru/iru sentences. If one can easily

characterize the topic (NP marked by wa) as having the theme (NP marked by ga)

in it, the sentence is perfectly acceptable. If not, the sentence is unacceptable.

Consider (74a-b) and (75a-b).

(74) a.*Kabanwa hanger ga aru.
bag TOP NOM exist

" 'The bag has hanger.‘

b. Closet wa hanger ga aru.
TOP NOM exist

‘The closet has hanger.‘

(75) a. * Niwa wa panda ga iru.
garden TOP NOM exist
* ‘The garden has a panda.‘
b. Chuugoku wa panda ga iru.
China TOP NOM exist

'China has pandas.‘

By the same token, if one thinks a playground can be characterized by having

children in it, it is, acceptable to say (76).

(76) Kooen wa kodomo ga iru.
playground TOP child(ren) NOM exist
‘The playground has children.‘

35

Recall Kato's (1991) contention that am ga constructions highlight the
conceptually close relationships between two NPs. She claims that wa ga
constructions are similar to English have sentences, but that while the latter can
express any kind of possession like temporary possession, wa ga constructions
can only express a permanent relationship between two NPs. However, two NPs

do not have to be in permanent relationships. I repeat the example here in (77).

(77) a. John no fuku wa kyoo kitanai.
GEN clothes TOP today dirty
’John’s clothes are dirty today.’
b. John wa kyoo fuku ga kitanai.
TOP today clothes NOM dirty
’John has dirty clothes today.
Thus the conditions which allow for ma ga constructions are characterizability of
the topic by the rest of the sentence: if a topicalized element is characterizable by
having the theme in (or on) it, the sentence is acceptable. It seems that the
English LOC have THEME construction is similar to the LOC wa THEME ga
construction in Japanese in that it is also constrained by characterizability of the

subject by the rest of the sentence. Consider English sentences in (78a-b). These

examples correspond to Japanese sentences in (75a-b) respectively.

(78) a. * The garden has pandas.

b. China has pandas.

36

Just as Japanese wa ga sentences can be used if characterizability condition is met,
the same can be said for English have sentences. In the English LOC have THEME

sentence, however, it seems that the theme has to be something fixed at the

location.

(79) a. * The park has children.

b. The park has a swing.
(80) a. ” The waste basket has waste.

b. The waste basket has a trash bag.
(81) a. * The closet has clothes.

b. The closet has hangers.

In this section, I analyzed locative sentences, the LOC ga/wa THEME ga
aru/iru construction, in Japanese and the LOC have THEME in English. I contend
that characterizability is what allows for these locative sentences in both

languages. The locative have sentences in English observe the further constraints

that the theme must be a fixed feature.

3.3. Constraints on the LOC/POSS ni THEME ga exist structure

This section discusses the constraints on LOC/POS S ni THEME ga exist

sentences. As opposed to well formed sentences shown in (82) and (83), ni ga am

37

sentences are often awkward, as shown in (84)-(87). Note that the corresponding
English sentences are perfectly well formed with there, as shown in the

translations.

(82) Reezooko ni ringo ga aru.
fridge LOC apple NOM exist
'There is an apple in the fridge.‘

(83) Rakuda ni Kobu ga aru.

camel LOC / POSS hump N OM exist
'There is a hump on a camel.‘

(84) “ Enpitsu ni keshigomu ga aru.
pencil DAT/LOC eraser NOM exist
'There is an eraser on the pencil.‘
(85) "’ Inu ni kusari ga aru.
dog DAT leash NOM exist
‘There is a leash on the dog.‘
(86) * Juice ni koori ga aru.
LOC ice NOM exist
‘There is ice in the juice.‘
(87) * Kata ni kami-no-ke ga aru.

shoulder LOC hair NOM exist
‘There is a hair on his / her shoulder.‘

Although Muromatsu (1994) does not discuss when ni ga aru structures
are well formed, let us consider her examples again. As I remarked, she first
examines English examples using HRU's (1994) framework, and discusses
corresponding Japanese data. Below are HRU's example in (88) and

Muromatsu's Japanese example in (89).

38

(88) There is a Ford T engine in my Saab.
(89) Kuruma ni engine ga aru.
car DAT/LOC NOM exist
‘There is an engine in the car.

Actually, the exact counterpart of (88) in Japanese is (90), which Muromatsu not

examine.

(90) ? Watashi no Saab ni Ford T engine ga aru.
my DAT / LOC NOM exist
‘There is a Ford T engine in my Saab.‘
She does not discuss the sentence (90). I consider that sentence (90) is not

acceptable in integral reading. If a Ford T engine is a part of my Saab, native

speakers of Japanese would use another verb, as in (91).

(91) Watashi no Saab ni Ford T engine ga tsuiteiru.
my DAT NOM is attached
'My Saab has a Ford T engine.‘
Below I will analyze ni ga aru/iru sentences. An analysis of the constraints on ni
ga aru constructions is important because previous researchers have not
discussed this issue at all. Section 3.3.1 examines the constraints on ni ga aru

structures with integral reading, and section 3.3.2 examines those with spatial

reading.

39

3.3.1. Integral sentences

In this section, I will discuss the constraints on the ni ga aru structure with

integral reading.

(92) John ni chiisai kodomo ga iru.13
DAT small child NOM exist
‘John has a small child.‘
(93) a. Rakuda ni chiisai kobu ga aru.
camel DAT small hump NOM exist
‘A camel has a small hump.
(94) a. ? Kuruma ni chiisai engine ga aru.14
car DAT/LOC small NOM exist
'There is a small engine in the car,‘
(95) a. ? Yama ni chiisai ki ga aru.
Mountain LOC small tree NOM exist
’There are small trees on the mountain.’
All the above examples are acceptable without the adjective chiisai ‘small';
however, (94) and (95) sound awkward when the adjective is added. We see
from this that only generic statements are allowed when using ni ga aru
structures unless they demonstrate inalienable possession, as in (92) and (93). If
the sentence is about a specific NP as in (91), repeated here as (96a), it is also

unacceptable. If (96a) has integral reading, tsuiteiru ‘is attached‘ must be used as

in (96b).

 

‘3 This sentence consems inalienable possession which is one kind of integral sentences. Thus, the
two NPs, John and kodomo ’child’, hold permanent relationship; however, I do not know how is it
permanent for John to have chiisai kodomo ’small child’.

”This sentence is acceptable in spatial reading, but awkward in integral reading.

40

(96) a. ? Watashi no Saab ni Ford T engine ga aru.

my DAT / LOC NOM exist

‘There is a Ford T engine in my Saab.‘

b. Watashi no Saab ni Ford T engine ga tsuiteiru.

my DAT / LOC NOM attach

'A Ford T engine attaches to my Saab.‘
As shown in (96b), a ni ga aru sentence that sounds awkward can be made well
formed by replacing the verb aru with another verb like tsuiteiru ‘is attached‘, or
haitteiru 'is contained'. If an integral sentence is not a generic statement, the verb

must be tsuiteiru ‘is attached‘ and the like, not iru or aru, is used as shown in (97b)

and (98b).

(97) a. * Enpitsu ni keshigomu ga aru.
pencil eraser

b. Enpitsu ni keshigomu ga tsuiteiru.
pencil eraser attach
'There is an eraser on the pencil.‘

(98) a. * Inu ni kusari ga aru.
dog leash

b. Inu ni kusari ga tsuiteiru.
dog leash attach
'There is a leash on the dog.‘

To summarize, I show integral ni ga aru structures in table (1).

41

Table (1):

Integral (NP, ni NP2 ga aru, representing a permanent relationship)

 

Inalienable possession

Alienable possession

 

 

 

generic OK OK
Rakuda ni kobu ga aru. Kuruma ni engine ga aru.
camel DAT hump NOM exist car DAT NOM exist
’A camel has a hump.’ ”T he car has an engine.’
nongeneric OK ???

 

Rakuda ni chiisai kobu ga aru.
camel DAT small hump NOM exist
’A camel has a small hump.’

 

? Kuruma ni chiisai engine ga aru.
car DAT small NOM exist
’The car has a small engine.’

 

Integral sentences have to be either inalienable possession or generic statement,

or both. If the sentence is neither inalienable possession nor generic, it is

unacceptable.

3.3.2. Spatial sentences

Now let us consider spatial sentences.

(99) Niwa ni ike ga aru.
garden LOC pond NOM exist
‘There is a pond in the garden.‘

(100) Grand River ni cafe ga aru.

LOC NOM exist

‘There are cafes on Grand River.‘

42

 

Spatial ni ga sentences do not express generic meaning, but only specific facts. I
have found some unacceptable spatial sentences in the ni ga aru structure. For
example, if the spatial sentence leans more toward an integral meaning, the ni ga
aru sentence sounds bad. As I pointed out in section 3.1., an integral sentence
represents a permanent relationship between two NPs, more specifically, a
whole / part relationship. There are some sentences in which the relationship
between two NPs represents whole and part, but not a permanent type. In such
a case, a sentence is unacceptable with the ni ga aru structure. For instance,

consider sentences (101a-b).

(101) a. * Juice ni koori ga aru.
ice

b. Juice ni koori ga haitteiru.

ice contain
'There is ice in the juice.‘

In sentence (101b) koori ‘ice‘ is now in the juice. The reason why (101a) is ill
formed is that it cannot be interpreted in the integral sense since the NP2, koori
‘ice‘, is not permanently part of NP], juice. Consider the English counterpart
integral sentence (102a), which is also ungrammatical, and the spatial sentence

(102b), which is grammatical.

(102) a. * The juice has ice.

b. The juice has ice in it.

43

As in (101a), a spatial sentence is ill formed when the relationship between two
NPs is whole and part. Sentences (103) and (104) are other examples of spatial
sentences in which the relationship between two NPs is whole and part like an
integral sentence, but the relationship is not a permanent kind.
(103) a. * Juice ni alcohol ga aru.
LOC NOM exist
b. Juice ni alcohol ga haitteiru.
LOC NOM contain

'There is alcohol in the Juice.‘

(104) a. * Kabe ni gamu ga aru.
wall LOC chewing gum NOM exist

b. Kabe ni gamu ga tsuiteiru.
wall LOC chewing gum NOM attach
‘There is chewing gum on the wall.‘

There is another type of sentence which is different from any of the

sentences I have discussed thus far. Consider sentences (105) and (106).

(105) a. * Kata ni kami-no-ke ga aru.
shoulder LOC hair NOM exist

b. Kata ni kami-no-ke ga tsuiteiru.
shoulder LOC hair NOM attach
’There is a hair on the shoulder.‘

(106) a. * Hoppeta ni okome ga aru.
cheek LOC rice NOM exist

b. Hoppeta ni okome ga tsuiteiru.
cheek LOC rice NOM attach
‘There is rice on the cheek.‘

There is no doubt that sentences (105) and (106) are spatial because the two NPs,
kata 'shoulder' and kami-no-ke 'hair‘ in (105) and hoppeta 'cheek‘ and okome 'rice' in
(106), are not in the hold whole / part relationship. Sentences (105) and (106) are
locative, not possessive, sentences. It is apparent that spatial sense cannot be
obtained when the theme is just accidentally in the location, and the location is
where the theme usually does not belong. For example, consider other examples

(107a—c).

(107) a. Atama ni shirami ga iru.
hair/ head LOC lice NOM exist
'There is lice in my hair.‘

b. " Atama ni mushi ga iru.
hair/ head LOC bug NOM exist

c. Atama ni mushi ga tsuiteiru.
hair/ head LOC bug NOM attach
‘There is a bug in (someone‘s) hair.
Shirami ‘lice' can belong to one's atama 'head‘, but mushi ‘bug' does not. A bug can
only accidentally be there. The same can be said in (105) and (106). A hair is
only accidentally cited on someone’s shoulder in (105). In (106), a piece of rice is

accidentally on so'meone’s cheek, maybe because that person ate rice in a hurry.

Table (2) summarizes spatial ni ga aru structure.

45

Table (2)

Spatial (NP, ni NP 2 ga aru, representing temporary relationship)

 

nonwhole / part relationship

whole/ part relationship

 

temporary OK

Niwa ni neko ga iru.
garden LOC cat NOM exist
’A cat is in the garden.’

3(-

‘Juice ni alcohol ga aru.

LOC NOM exist

 

 

accidental *

"Hoppeta niokome ga aru.
cheek LOC rice NOM exist

 

 

 

Spatial sentences have to represent temporary relationship and nonwhole/ part

relationship. Sentences are ungrammatical if they represent either accidental

relationship or whole / part relationship.

3.3.3. Summary

We have seen in this section that NP, ni NP2 ga aru sentences exhibit the

following constraints.

1. Integral sentences always express a permanent relationship between the two

NPs. If they express inalienable possession, they can either be generic or

nongeneric statements, but if they express alienable possession, they must be

generic.

46

 

(108) a. Kuruma ni engine ga aru.
car DAT NOM exist
’The car has an engine.’

b. * Kuruma ni chiisai engine ga aru.
car DAT small NOM exist

(109) a. Integral: kuruma [ engine]]

car engine

[sc

kuruma ni engine ga aru.
car DAT NOM exist
SUBJECT PREDICATE

’The car has an engine.’

b. Spatial: [SC engine [kuruma ni]]
engine car in
kuruma ni engine ga aru.
car in NOM exist

PREDICATE SUBJECT
’In the car is an engine.

2. Spatial sentences always express a temporary relationship between the two

NPs. As such, the two NPs are never in a whole / Part relationship.

(110) Niwa ni neko ga aru.
garden DAT cat NOM exist
’A cat is in the garden.’

3. Spatial sentences in which the NPs are not a whole / part relationship are good

if the theme is temporary in the location, but not if it is just accidentally there.

(111) a. Atama ni shirami ga iru.
hair/ head LOC lice NOM exist
’There is lice in my hair.’

b. * Atama ni mushi ga iru.
hair / head LOC bug NOM exist

47

4. Conclusion

The present study has examined Japanese existentials, in terms of their
constraints. Earlier studies of Japanese existentials proposed a small clause
analysis to show that there are two different structures; one is locative and the
other is possessive. However, the small clause analysis has not satisfactorily
explained what prevents Japanese existentials from using ni ga aru sentences,
scrambled ga ni aru sentences, or double nominative ga ga aru sentences.

Section 3.1 analyzed scrambled sentences and concluded that scrambled,
ga ni word order, is not allowed in kinship sentences because the theme will not
be c-commanded by it's possessor. Following Kuno's (1973) analysis of
characterizability, in section 3.2, I showed that double nominative ga ga sentences
and topicalized wa ga sentences are constrained by the characterizability of the
subject NP. Finally in section 3.3, I analyzed the constraints on existentials
LOC/POS S ni THEME ga EXIST and concluded that sentences sound awkward in

the following cases.

1. Integral sentences always express a permanent relationship between the two
NPs. If they express inalienable possession, they can either be generic or
nongeneric statements, but if they express alienable possession, they must be
generic.

(112) a. Kuruma ni engine ga aru.

car DAT NOM exist
’The car has an engine.’

48

b. * Kuruma ni chiisai engine ga aru.
car DAT small NOM exist
2. Spatial sentences always express a temporary relationship between the two
NPs. As such, the two NPs are never in a whole / Part relationship.
(113) Niwa ni neko ga aru.
garden DAT cat NOM exist
’A cat is in the garden.’
3. Spatial sentences in which the NPs do not hold a whole / part relationship are
good if the theme is temporary in the location, but not if it is just accidentally
there.
(114) a. Atama ni shirami ga iru.
hair/ head LOC lice NOM exist
’There is lice in my hair.’

b. * Atama ni mushi ga iru.
hair/ head LOC bug NOM exist

Table (3) summarizes my findings regarding both integral and spatial

sentences in Japanese.

49

Table (3)

Integral (NP, ni NP, ga aru, representing a permanent relationship)

 

Inalienable possession

Alienable possession

 

generic OK

Rakuda ni kobu ga aru.
camel DAT hump NOM exist
’A camel has a hump.’

OK

Kuruma ni engine ga aru.
car DAT NOM exist
’T he car has an engine.’

 

 

nongeneric OK

Rakuda ni chiisai kobu ga aru.
camel DAT small hump NOM exist
’A camel has a small hump.’

 

 

???

? Kuruma ni chiisai engine ga aru.
car DAT small NOM exist
’The car has a small engine.’

 

Spatial (NP, ni NP, ga aru, representing temporary relationship)

 

nonwhole / part relationship

whole/ part relationship

 

temporary OK

Niwa ni neko ga iru.
garden LOC cat NOM exist
’A cat is in the garden.’

*

*Juice ni alcohol ga aru.
LOC NOM exist

 

 

accidental ’*

‘I-loppeta niokome ga aru.
cheek LOC rice NOM exist

 

 

 

The present study has analyzed the constraints on Japanese existential

structures. Although some issues remain, I hope that this study has shed light on

the facts surrounding Japanese existentials and their constraints. Further

research on the motivation for the constraints is needed in order to gain a more

comprehensive understanding of the Japanese existentials.

50

 

 

BIBLIOGRAPHY

BIBLIOGRAPHY

Abbott, Barbara. 1993. A Pragmatic Account of the Definiteness Effect in
Existential Sentences. Journal of Pragmatics 19. 39-55.

Chomsky, Noam. 1986. Knowledge of language: Its nature, origin, and use.
New York: Praeger.

Chomsky, Noam. 1995. The Minimalist Program, Cambridge Mass: MIT
Press.

Cowper, Elizabeth A. 1989. Thematic Underspecification: the case of have.
University of Toronto, Working Papers 10, 85-93.

Croft, William. 1991. Syntactic Categories and Grammatical Relations. The
University of Chicago Press.

Barker, Chris. 1994. Rethinking double genitives as partitives: Explaining an
anti-uniqueness implication (Referee Draft) Center for the Sciences of
Language, University of Rochester.

Freeze, Ray. 1992. Existentials and Other Locatives. Language. Volume 68.
553-595.

Givon, Talmy. 1984. Syntax: A Functional-Typological Introduction. VOL. 1.

Greenberg, Joseph H. 1978. Universals of Human Language. Stanford
University Press. VOL. 4

Harasawa, Itsuo. 1993. The usage of the Existential Verb iru and aru in Terms
of ”Dynamic and Static”. Nihongo Kyooiku Journal of Japanese
Language Teaching No. 80. 62-73

Hornstein, Norbert. Rosen, Sara. and Uriagereka, Juan. 1994. Integrals.
University of Maryland, Working Papers 2, 70-90.

Inoue, Kazuko. 1995. Causative Have and Experiential Have. English
Linguistics: Journal of the English Linguistic Society of Japan 12, 73-95.

Kato, Yoko. 1991. Possessive Constructions in Japanese. University of
Michigan, Working Papers in Linguistics, 35-49.

Kayne, Richard. 1994. The Antisymmetry of Syntax. Cambridge Mass: MIT
Press.

51

Kuno, susumu. 1973. The Structure of the Japanese Language Cambridge
Mass: MIT Press.

Kuroda, S.-Y. 1978. Case Marking, Canonical Sentence Patterns, and Counter
Equi in Japanese. Problems in Japanese Syntax and Semantics:
Kaitakusha, 30-51

Lasnik, Howard. 1992. Case and Expletives: Notes toward a Parametric
Account . Linguistic Inquiry 23, 381-405.

Levin, Beth. 1993. English Verb Classes and Alternations The University of
Chicago Press.

Makino, Seiichi. Japanese ‘BE'. The Verb 'Be' and Its Synonyms. D. Reidel
Publishing Company. 1-19.

Martin, Samuel E. 1975. A Reference Grammar of Japanese. New Haven and
London: Yale University Press.

Mikami, Akira. 1975. Mikami Akira Ronbun-shu. Tokyo: Kuroshio-
shuppan.

Muromatsu, Keiko. 1994. Integrals and Spatials in Japanese and the
Internal Structure of possessives. University of Maryland, working
papers 1,

Muromatsu, Keiko. 1994. Integrals in Japanese. University of Maryland,
working paper, 136-158.

Napoli, Donna Jo. 1993. Syntax: Theory and Problems Oxford University
Press.

Napoli, Donna Jo. 1989. Predication theory: A Case Study for Indexing
Theory. Cambridge: Cambridge University.

Pinker, Steven. 1989. Learnability and Cognition Cambridge Mass: MIT
Press.

Ritter, Elizabeth and Rosen, Sara Thomas. 1993. Deriving Causation. Natural
Language and Linguistic Theory 11: 519-555.

Sato, Paul T. 1985 Existence and Possession. Papers in Japanese Linguistics,
Volume 10. 34-47.

Shibatani, Masayoshi. 1990. The Languages of Japan Cambridge University
Press.

52

Simon, Mutsuko E. 1989. An analysis of the postposing construction in
Japanese. Doctoral dissertation. University of Michigan.

Szabolcsi, Anna. 1983. The Possessor that Ran Away from Home. The
Linguistic Review 3: 89-102.

Szabolcsi, Anna. 1994. The Noun Phrase. In: The Syntactic Structure of
Hungarian: Syntax and Semantics, Vol. 27. Academic Press. 179-274.

Tsunoda, Tasaku. 1996. The Possession Cline in Japanese and Other
Languages. The Grammar of Inalienability. Berlin; New York: M. de
Gruyter. 565-630.

Wierzblicka, Anna. 1982. Why can you have a drink when you can’t *have
an eat? Language 58: 753-799.

Yamagata, Ayako. 1998. The Stage-Level/Individual-Level Distinction: An
Analysis of -te-iru. Michigan State University, working paper 27,

Yamamoto, Akira. 1975. The Syntax and Semantics of Possession I: Japanese.
In proceedings of 1975 Mid-America Linguistics Conference, Lawrence:
Linguistics Dept., University of Kansas. 589-599.

Yamanaka, Nobuhiko. 1988 The disambiguation of ’Modifier Noun no
Noun’ construction in Japanese. Gengo kenkyuu: Journal of The
Linguistic Society of Japanese. Volume 94. 75-99.

Yoshikawa, Chizuko. 1995. Nichiee Hikaku Dooshi no Bunpoo. Kuroshio

53

"illiillliall“