An analysis of Inalienable possession constructions in Japanese1)

Osaka Keidai Ronshu, Vol. 64 No. 1 May 2013
103
An analysis of Inalienable possession constructions
1)
in Japanese
Mitsue Motomura
1. Introduction
In a transitive / intransitive pair in Japanese, a possessor of the object appears as the subject in
the transitive sentence (1a) while it stays inside the object in the intransitive sentence (1b).
These sentences are often called inalienable possession constructions.
( 1 ) a. Taro-ga
kami-o kit-ta.
T-nom
hair-acc cut-past
‘(i) Taro got a haircut. / (ii) Taroo cut (someone’s / his own) hair.’
b. Taro-no
kami-ga
kire-ta.
T-gen
hair-nom
cut-past
‘Taro’s hair snapped.’
The transitive / intransitive pattern is not limited to those with inalienable possession relations, but it is very common in Japanese as well as in other languages. Here is another example
of transitive / intransitive pair in Japanese. Pairs like (2) are often called causative / inchoative
verbs.2)
( 2 ) a. Taro-ga
kabin-o
T-nom
base-acc
‘Taro broke the base.’
b. Taro-no
kabin-ga
T-gen
base-nom
‘Taro’s base broke.’
kowashi-ta
break-past
koware-ta
break-past
1) Content of this paper was presented at the 7th International Workshop on Theoretical East Asian Linguistics held on February 18
19, 2012 at Hiroshima University as well as at the 22 nd Japanese / Korean Linguistics Conference held on October 1214 at the National Institute for Japanese Language and
Linguistics (NINJAL). I thank those who gave me valuable comments and questions at these presentations. I am also grateful to Norbert Hornstein and Howard Lasnik for their comments as well as
Shinsuke Homma, Maki Kishida, Takeo Kurafuji, Nao Orita, and Asako Uchibori for their comments and
data judgments. I am responsible for all the mistakes.
2) For discussions of Japanese causative constructions, see Kuroda (1965), Inoue (1976), Terada (1990),
Pesetsky (1995), Harley (1995, 2008), Miyagawa (1999) among many others.
104
Osaka Keidai Ronshu, Vol. 64 No. 1
(1a), however, shows different characteristics from (2a). I will call verbs of (1a) type ‘exter
nal possessor (ext-poss)’ verb and those of (1b) type ‘internal possessor (int-poss)’ verb
adopting Vergnaud and Zubizarreta’s (1992) terminology.
One characteristic is that the subject and the object of (1a) show an inalienable possession
relation. That is the subject is the possessor of the body-part object. No such relation is implied in (2a); it is neutral with respect to the possessor of the object. It could be Taro’s or
someone else’s.
Another characteristic of inalienable possession pairs appears in interpretation of the subject.
It has been proposed that the -role of the subject of the causative sentence in (2a) is agent /
causer (Pesetsky 1995), but the default interpretation of (1a) is (1ai), where the subject is
not the actual agent of hair cutting.3) The agent reading (1aii) is also possible given the appropriate context although it is not the first reading that comes to mind. Thus, (1a) is ambiguous
between a non-agentive reading and an agentive reading.
The third characteristic is observed in a contrast between (1a) and (3a).
( 3 ) a. Taro-ga
zibun-no
T-nom
self-gen
‘Taro cut his own hair.’
b. Taro-ga
Hanako-no
T-nom
H-gen
‘Taro cut Hanako’s hair.’
kami-o
hair-acc
kit-ta.
cut-past
kami-o
hair-acc
kit-ta.
cut-past
When a reflexive anaphor, zibun, appears as a possessor of the object as in (3a), the subject
is interpreted as an agent just as in (3b). Thus the sentence is interpreted as Taro cutting his
own hair. On the other hand, such shift in interpretation of the subject does not occur with the
causative verb in (2a). Compare (2a) with (4a-b).
( 4 ) a. Taro-ga
zibun-no
kabin-o
T-nom
self-gen
base-acc
‘Taro broke his base.’
b. Taro-ga
Hanako-no kabin-o
T-nom
H-gen
base-acc
‘Taro broke Hanako’s base.’
kowashi-ta
break-past
kowashi-ta
break-past
Whether an overt possessor of the object appears or not, the subjects of these sentences are
all interpreted as agent (or causer).4)
Fourth, relative clauses of ext-poss sentences show a subject-object asymmetry.
3) In this paper, I use the term ‘agent’ or ‘agentive’ to include causative interpretation. Therefore, the term
‘non-agentive’ does not include causative interpretation.
4) However, we will see below that acceptability of an overt possessor of the object varies among ext-poss
verbs.
An analysis of Inalienable possession constructions in Japanese
105
( 5 ) a. Taro-ga
kitta
kami
(agent / *non-agent)
T-nom
cut
hair
‘The hair that Taro cut’
b. kami-o
kitta
Taro5)
(agent / non-agent)
hair-acc
cut
T
‘Taro, who got a haircut / cut the hair’
As mentioned above, the ext-poss sentence in (1a) is ambiguous between a non-agent reading
and an agent reading. However, when the object is relativized as in (5a), the non-agent reading
disappears. On the other hand, (5b), which relativizes the subject, retains both readings although the non-agent reading is still natural or default. Since the causative sentence in (2a) is
not ambiguous as the subject is always interpreted as agent, no such contrast is obtained as
shown in (6).
( 6 ) a. Taro-ga
kowashi-ta kabin
T-nom
break-past
base
‘the base that Taro broke’
b. kabin-o
kowashi-ta Taro
base-acc break-past
T
‘Taro, who broke the base’
Other examples of inalienable possession pairs are given below.
( 7 ) a. Taro-ga
hone-o
ot-ta.
T-nom
bone-acc break-past
‘(i) Taro broke his bone. (ii) Taroo broke (someone’s / his own) bone.’
b. Taro-no
hone-ga
ore-ta.
T-gen
bone-nom break-past
‘Taro’s bone broke.’
c. Taro-ga
zibun-no hone-o ot-ta.
T-nom
self-gen
bone-acc break-past
‘Taro broke his own bone.’
5) (5b) and (6b) are cases of non-restrictive relative clauses. The result is the same with restrictive relative clauses as in (i) and (ii).
(i) kami-o
kitta
otoko
(agent / non-agent)
hair-acc
cut
man
‘a man who got a haircut / a man who cut his own hair’
(ii) kabin-o
kowashi-ta
otoko
(agent / *non-agent)
base-acc break-past
man
‘a man who broke the base’
The same holds with other verbs in (7)
(11).
106
Osaka Keidai Ronshu, Vol. 64 No. 1
d. Taro-ga
otta
hone
(agent / ??non-agent)6)
T-nom
broke
bone
‘the bone that Taro broke’
e. hone-o
otta
Taro
(agent / non-agent)
bone-acc broke
T
‘Taro, who broke his / the bone’
( 8 ) a. Taro-ga
me-o
samashi-ta.
T-nom
eye-acc
wake-past
‘Taro woke up.’
b. Taro-no
me-ga
same-ta.
T-gen
eye-nom
wake-past
‘Taro woke up.’
c. *Taro-ga zibun-no
me-o
samashi-ta.
T-nom
self-gen
eye-acc
wake-past
‘(lit.) Taro woke his own eye.’
d.*Taro-ga
samashita
me
T-nom
woke
eye
‘(lit.) the eyes that Taro woke’
e. me-o
samashita
Taro
eye-acc
woke
T
‘Taro, who woke up’
( 9 ) a. Taro-ga
koshi-o
itame-ta.
T-nom
back-acc
hurt-past
‘Taro hurt his back.’
b. Taro-no
koshi-ga
itan-da.
T-gen
back-nom hurt-past
‘Taro’s back hurt.’
c.*Taro-ga
zibun-no
koshi-o
itame-ta.
T-nom
self-gen
back-acc
hurt-past
‘Taro hurt his own back.’
d.??Taro-ga itame-ta
koshi7)
T-nom
hurt-past
back
‘the back that Taro hurt’
6) It seems that the non-agent reading is available in the following sentence.
(i) Taro-ga
otta
hone-wa
migi
ashi-da.
T-nom
broke
bone-top
right
leg-cop
‘The bone Taro broke is right leg.’
I will leave this problem for future research.
7) (9d) seems okey in the following sentence as in (7d) mentioned in footnote 6.
(i) Taro-ga
itame-ta
koshi-wa
sukkari
yokunatta.
T-nom
hurt-past back-top
completely
got.well
‘The back that Taro hurt got well completely’
An analysis of Inalienable possession constructions in Japanese
107
e. koshi-o
itame-ta
Taro
back-acc hurt-past
T
‘Taro, who hurt his back’
(10) a. Taro-ga
ashi-o
sukume-ta.
T-nom
leg-acc
duck-past
‘Taro ducked his legs (for fear).’
b. Taro-no
ashi-ga
sukun-da.
T-gen
leg-nom
duck-past
‘Taro’s legs ducked (for fear).’
c. *Taro-ga
zibun-no
ashi-o
sukume-ta.
T-nom
self-gen
leg-acc
duck-past
‘Taro ducked his own leg.’
d.*?Taro-ga sukume-ta
ashi
T-nom
duck-past
leg
‘the leg that Taro ducked’
e. ashi-o
sukume-ta
Taro
leg-acc
duck-past T
‘Taro, who ducked his leg (for fear)’
(11) a. Taro-ga
mi-o
chijime-ta.
T-nom
body-acc
shrink-past
‘Taro shrank his body (for fear).’
b. Taro-no
mi-ga
chijimat-da.
T-gen
body-nom
shrink-past
‘Taro’s body shrunk (for fear).’
c. *Taro-ga
zibun-no
mi-o
chijime-ta.
T-nom
self-gen
body-acc shrink-past
‘Taro shrank his body.’
d.*Taro-ga
chijime-ta
mi
T-nom
shrink-past body-acc
‘the body that Taro shrank’
e. mi-o
chijime-ta
Taro
body-acc shrink-past T
‘Taro, who shrank his body’
Note that the (a) examples of (8)
(11) are not ambiguous. As Hasegawa (2007; 67) notes,
the subjects of these sentences are interpreted as ‘some sort of patient, experiencer or
affectee.’ Given that the subject of some ext-poss verbs is agent while that of others is nonagentive, it seems necessary to clarify what role the subject of each ext-poss verb takes.
In this paper, I will clarify the availability of agentive readings and try to give a unified account to these sentences. First, I will summarize Hasegawa’s (2007, 2009) proposal and point
out some problems in section 2. In section 3, I will apply four tests to see if the subjects of extposs verbs show some kind of agentivity or intentionality. Based on the results of the tests, I
Osaka Keidai Ronshu, Vol. 64 No. 1
108
will present derivations of ext-poss and int-poss sentences and account for the characteristics
of ext-poss verbs in section 4. Section 5 will discuss remaining issues and section 6 will be conclusion.
2. Hasegawa (2007, 2009)
It has been a widely held view that a functional verb or a ‘little v,’ which is distinct from a lexical verb, is responsible for licensing of an external argument and the objective Case (Chomksy
1995 ; Collins 1997, Kratzer 1996). Postulation of ‘little v’ can explain Burzio’s generalization
(Burzio 1986 : 178) that ‘all and only the verbs that can assign -role to the subject can assign
(accusative) Case to an object.’
Unlike this view, Hasegawa (2001, 2004, 2007, 2009) claims that the little v has more properties than what the above view claims. That is, it is a locus where [±External Role (±ER)]
and [±Object Case (±OC)] are discharged / licensed.8) Thus, according to her, four types of
verbs are available given the combination of the four features of the little v, as listed in (12).
(12)9)
+External Role (+ER)
−External Role (−ER)
+Object Case
(+OC)
a. transitive
b. unaccusative transitive
−Object Case
(−OC)
c. unaccusative agentive intransitive d. unaccusative intransitive
Recall that with respect to the interpretation of the subject, (1a) and (7a) are ambiguous
while (8a)(11a) are not. (1a) and (9a) are repeated here.
(13) a. Taro-ga
kami-o
kit-ta.
(=(1a))
T-nom
hair-acc
cut-past
‘(i) Taro got a haircut. / (ii) Taroo cut (someone’s / his own) hair.’
b. Taro-ga
koshi-o
itame-ta.
(=(9a))
T-nom
back-acc hurt-past
‘Taro hurt his back.’
Hasegawa claims that the subject of (13ai) is an intentional entity having responsible for his
haircut even though he is not an agent of the action based on the fact that an agent oriented adverb can appear with this reading as shown in (14).
8) Hasegawa (2004 : 313) does not deny Burzio’s generalization but takes it as ‘a mere surface generalization on what kinds of sentences are given rise to in language’ rather than a realization of some grammatical principle or condition.
9) (12) is from Hasegawa (2009 : 444).
An analysis of Inalienable possession constructions in Japanese
(14)
Taro-ga
wazato
kami-o
mijikaku
T-nom
deliberately
hair-acc
shortly
‘Taro got a short haircut deliberately.’
109
kit-ta.
cut-past
Thus she concludes that (13ai), as well as (13a
ii), is a type of transitive verb (12a) having
a little v with [+ER, +OC] features. Based on the tei(ru) fact discussed in Takezawa
(1991)10) and the fact that the subject is the possessor of the object, Hasegawa proposes (15a)
for (13ai) and (15b) for (13a
ii).
(15)11) a.
TP
b.
T
DP1
vP
T
VP
DP2
T
DP
vP
v
t1
TP
v
t1
v[+ER,+OC]
VP
V
DP2
hair
v[+ER,+OC]
V
cut
t1
T
cut
e1
hair
In (15a), the subject is base-generated inside the object as its possessor and moves to TP
through spec. vP, where it gets an external -role, which is agent in a broad sense. The inalienable possession relation between the subject and the object is a consequence of this possessor raising construction. On the other hand, no such relation is observed in (13a
ii) given that
the subject is base-generated in spec.vP as shown in (15b). This is a case of standard causative
sentence.
In the case of (13b), the subject does not have any intention of hurting his back. Hasegawa
classifies this type of verbs as unaccusative transitive (12b) with [−ER, +OC]. (16) is the
derivation of (13b).
10) According to Takezawa (1991), when tei(ru) is attached to an unaccusative verb, the sentence expresses a result state. When tei(ru) is attached to (13a), it can express a result state as in (ia).
(i) Taroo-ga
kami-o
kit-tei-ru.
T-nom
hair-acc cut-TEI-pres
(a) Taroo got a haircut. (the result state) (b) Taroo is cutting the hair. (progressive)
This indicates that the verb of (13a
i) is a kind of unaccusative verb with a derived subject.
For discussions of tei(ru) facts in inalienable possession constructions, see Hasegawa (2009 : 448),
Homma (2004), and Motomura (2012).
11) (15) is based on Hasegawa (2009 : section 5.1).
Osaka Keidai Ronshu, Vol. 64 No. 1
110
(16)12)
TP
T
DP1
vP
VP
DP2
T
v[−ER,+OC]
V
hurt
t1
back
In this case, the subject is base-generated inside the object as its possessor and moves directly
to spec. TP. Since the subject does not get the external -role at spec. vP, it does not obtain
the agent reading resulting in experiencer interpretation. Hasegawa attributes the lack of agent
reading to the feature specification of [−ER].
It seems plausible to claim that the possessor subject of the ext-poss sentence moves from
inside the object since it can account for the inalienable possession relation and the tei(ru)
fact. It can also account for the contrast between an ext-poss sentence (1a) and the sentences
with overt possessors in (3a
b). However, I believe that Hasegawa’s analysis is not enough
to handle some of the data presented in section 1. Her analysis can not explain the contrast between the ext-poss sentence (1a) and the standard causative sentence (2a). It cannot account
for the subject object asymmetry in relative clauses observed in (5), either. Furthermore, she
does not discuss which verb shows agentivity and which verb does not.
3. Tests for intentionality
In this section, I will apply four tests to ext-poss verbs to see if the subject can be interpreted
as agent or as having some sort of intentionality. The three are from Jaconsen (1992 : chapter
4) : an adversity passive test, an imperative test, and a potential test. The last test is ni iku
attachment test.
3. 1 The adversity passive test
Japanese has two types of passive sentences (Harada 1973, Kuno 1973, Howard & NiyekawaHoward 1976). One is an English type passive, called direct passive, as in (17a); the other is
called adversity passive (or indirect passive) as in (17b).
12) (16) is based on Hasegawa (2007 : section 3.2).
An analysis of Inalienable possession constructions in Japanese
111
(17) a. Saihu-ga
doroboo-ni
nusum-are-ta
wallet-nom thief-dat
steal-pass-past
‘The wallet was stolen by a thief.’
b. Taro-ga
doroboo-ni
saifu-o
nusum-are-ta.
T-nom
thief-dat
wallet-acc
steal-pass-past
‘Taro was (adversely) affected by a thief’s stealing the wallet.’
In the adversity passive sentence (17b), there is an extra argument, Taro, which is not an argument of the verb, nusum ‘steal.’ This type of sentence is often analyzed as having a bi-clausal
structure with a matrix verb -rare and an embedded verb.13)
As Jacobsen (1992) discusses, it seems that the embedded subject of an adversity passive
sentence must be an intentional entity. Compare (18) with (19).14)
(18) a. Tanaka-wa supiido ihan-de
keisatsu-ni untenmenkyosho-o tor-are-ta.
T-top
speed violation-for police-dat driver’s.license-acc take-pass-past
‘Tanaka had his license taken away for speeding by the police.’
b. Morita-wa ohuro-ni haitte-iru tokoro-e okyakusan-ni ko-rare-ta
M-top
bath-in
enter-perf place-loc guests-dat
come-pass-past
‘Morita had guests arrive on him when he was in the bath.’
(19) a. *Kurai
heya-de kinko-o sagashite-iru
tokoro-o denki-ni
dark
room-in safe-acc search-prog
comp-acc lights-dat
barete-shimat-ta.
tuk-are-te
go.on-pass-ger be.discovered-past
‘As I was looking for the safe in the dark room, the lights went on on me, and I was
discovered.’
b.*Chosho-ni ure-rare,
kane-ni
mooka-rare,
arigata-meiwaku-da.
book-dat
be.sold-pass.inf money-dat be.made-pass.inf thanks-nuisance-cop
‘It’s a mixed blessing, the way (my) books keep selling on me and the money
keeps on being earned.’
In (18a
b), the embedded subjects, keisatsu ‘the police’ and okyakusan ‘guests’ are agent, and
the sentences are perfect. On the other hand, those in (19a
b), denki ‘lights,’ chosho ‘books,’
and kane ‘money’ are inanimate thus non-intentional entities, and the sentences are unacceptable.
Furthermore, as Inoue (1976) notes, unaccusative verbs such as agar ‘rise’ are often ambiguous between an intentional reading and a non-intentional spontaneous reading.15)
13) Whether the direct passive and the indirect passive have different structures or not is controversial, however. See Hoshi (1999) for an overview of the passive constructions in Japanese. Also see Motomura
(forthcoming) for an analysis of adversity passive constructions.
14) (18) and (19) are Jacobsen’s (1992 : 9495) (4)
(7)) with minor modification.
15) (20) and (21) are cited from Jacobsen (1992 : 95) with minor modification.
112
(20)
Osaka Keidai Ronshu, Vol. 64 No. 1
Kodomo-ga
suteeji-ni agat-ta.
(intentional / spontaneous)
child-nom
stage-at
rise-past
‘A child climbed / ended up on the stage.’
In (20), the child may climb up on the stage with his / her own intention, or he / she may end
up being on the stage by some force other than his / her own.
When this verb is embedded in the adversity passive construction, the only reading available
is the agent reading as shown below.
(21)
Shicho-wa
enzetsu-no
sanaka-ni
kodomo-ni
suteeji-ni
agar-are-te
mayor-top
speech-gen
middle-in
child-dat
stage-at
rise-pass-ger
yowat-ta.
be.dismayed-past
‘In the middle of the speech, the mayor was dismayed by a child climbing up the
stage.’
(intentional / *spontaneous)
These facts suggest that the embedded subject of adversity passive constructions must be an
intentional entity.16)
Here are the results of the adversity passive test applied to ext-poss verbs.17, 18)
16) There are cases that do not seem to obey this restriction. As Jacobsen (1992) discusses, the following
sentences show that the embedded subjects are not intentional entities : in (i), the subject is ‘wind’
which is a natural phenomenon, and in (ii), it is a ‘dead father.’
(i) (Watashi-wa)
kaze-ni
huk-are-ru-to
suzushiku-nat-ta.
(I-top)
wind-dat
blow-pass-pres-cond
cool-become-past
‘I felt cooler when I was blown on by the wind.’
(ii) (Watashi-wa)
chichi-ni
shin-are-te
ooini
yowat-ta.
(I-top)
father-dat
die-pass-ger
greatly
be.dismayed-past
‘I was greatly dismayed at my father’s passing.’
Jacobsen (1992 : 96 with minor modification)
I follow Jacobsen and assume that natural phenomena such as wind can be interpreted as intentional
entities while the case of (ii) is truly exceptional. The imperative test in the next section also supports
the claim that ‘wind’ is treated as an intentional entity in Japanese. See Jacobsen (1992 : chapter 4 and
chapter 6) on this issue.
17) The judgment is based on the interpretation where an inalienable possession relation holds between the
embedded subject and the embedded object.
18) The sentences such as (22b) and (22d) might not be so bad to some people with appropriate contexts
as shown below.
(i) ?Taro-ga
Hanako-ni
hone-o
or-are-te
komatteru.
T-nom
H-dat
bone-acc
break-pass-ger
be.in.trouble
‘Taro is in trouble by Hanako’s breaking the bone.
(ii) ?Taro-ga
Hanako-ni
koshi-o
itame-rare-te
komatteiru.
T-nom
H-dat
back-acc hurt-pass-past
be.in.trouble
‘Taro is in trouble by Hanako’s hurting her back.
Suppose that Taro is a women’s soccer team coach, and Hanako is one of the main members, but she
broke her bone or hurt her back, so she is not able to play in a tournament. Taro is in trouble because
An analysis of Inalienable possession constructions in Japanese
113
(22) Ext-poss verbs
a. Taro-ga
Hanako-ni
kami-o
kir-are-ta.19)
T-nom
H-dat
hair-acc
cut-pass-past
‘Taro was affected by Hanako’s getting a haircut.’
b.*?Taro-ga
Hanako-ni
hone-o
or-are-ta.
T-nom
H-dat
bone-acc
break-pass-past
‘Taro was affected by Hanako’s breaking the bone.
c. Taro-ga
Hanako-ni
me-o
samas-are-ta.
T-nom
H-dat
eye-acc
wake-pass-past
‘Taro was affected by Hanako’s waking up.
d.*?Taro-ga
Hanako-ni
koshi-o
itame-rare-ta.
T-nom
H-dat
back-acc
hurt-pass-past
‘Taro was affected by Hanako’s hurting her back.
e.*Taro-ga
Hanako-ni
ashi-o
sukume-rare-ta.
T-nom
H-dat
leg-acc
duck-pass-past
‘Taro was affected by Hanako’s ducking her leg.’
f. *Taro-ga
Hanako-ni
mi-o
chijime-rare-ta.
T-nom
H-dat
body-acc
shrink-pass-past
‘Taro was affected by Hanako’s shrinking her body.’
These results show that some of the ext-poss verbs do not allow an intentional entity as their
subject while others do. I will summarize the results at the end of this section.
3. 2 The imperative test
Jacobsen (1992 : 97) notes that ‘imperatives require that the verb express some sort of activity
under the volitional control of the addressee, and also that the addressee be the sort of entity
that is capable of entering into human dialogue.’ In other words, the addressee of imperatives
must be an intentional entity.
In the following examples, verbs, tabe- ‘eat’ (23a), ir- ‘be, stay’ (23b), huk- ‘flow’ and hur‘fall’ (23c), can appear in imperative constructions, but tuk- ‘go on’ (23d) and koware- ‘break’
(23e) can not.20), 21)
of her injury. Although the sentences may sound OK, it seems that the utterer of these sentences feels
as if Hanako did it on purpose even though he or she knows Hanako did not do it on purpose. It may be
that Hanako is treated as a volitional entity in these cases.
19) As the translation indicates, this sentence is acceptable even if Hanako is not the agent. That is, Hanako
is not the person who did the cutting.
20) The sentences in (23a
d) are from Jacobsen (1992 : 97
98 with minor modification).
e) would be acceptable in fantasy situations where the lights or the clock are alive and have their
21) (23d
own will.
114
Osaka Keidai Ronshu, Vol. 64 No. 1
(23) a. Isoide
gohan-o
tabe-ro.
hurry-ger
meal-acc
eat-imp
‘Eat meal quickly !’
b. Konban
juuji-ni
heya-ni
ir-o.
tonight
ten.o’clock-at
room-in
be-imp
‘Be in your room at ten o’clock tonight !’
c. Kaze-yo,
huk-e.
Ame-yo,
fur-e.
wind-voc
flow-imp
rain-voc
fall-imp
‘Blow, wind ! Fall, rain !’
d.*Denki-yo,
tuk-e.
light-voc
go.on-imp
‘Lights, go on !’
e.*Tokei-yo,
koware-ro.
clock-gen
break-imp
‘The clock, break !’
These data suggest that the sentence is acceptable when the addressee is an intentional entity.
When we apply the imperative test to ext-poss verbs, we get the following results.
(24) Ext-poss verbs
a. Kami-o
kir-e.
hair-acc
cut-imp
‘Get a haircut !’
b.*Hone-o
or-e.
bone-acc
break-imp
‘Break your bone !’
c. Me-o
samas-e.
eye-acc
wake-imp
‘Wake up !’
d.*Koshi-o
itame-ro.
back-acc
hurt-imp
‘Hurt your back !’
e.*Ashi-o
sukume-ro.
leg-acc
duck-imp
‘Duck your leg !’
f. *Mi-o
chijime-ro.
body-acc
shrink-imp
‘Shrink your body !’
As with the cases of adversity passive, we get mixed results depending on the verbs.
An analysis of Inalienable possession constructions in Japanese
115
3. 3 The potential test
A potential construction is another place where we can test whether the subject is intentional
or not. A potential particle, -rare / -e, can attach to intentional verbs but not to non-intentional
verbs as the following examples show.
(25) a. Taro-wa
T-top
‘Taro can
b.*Taro-wa
T-top
‘Taro can
karee-raisu-ga / o
umaku
tsukur-e-ru.
curry-rice-nom / acc
well
make-pot-pres.
cook curry rice well.’
chichioya-ni
ni-rare-ru.
father-dat
resemble-can-pres
resemble his father.’
Furthermore, consider the following sentences.
(26) a. Juuhassai
ijoo-de
nai-to
ano
eigakan-ni
hair-e-nai.
eighteen
over-cop
neg-cond
that
theater-loc
enter-pot-neg
‘You can’t enter that movie theater unless you are eighteen or over.’
b.*Kono
ringo-wa
zenbu
sono
hako-ni
hair-e-nai.
these
apples-top
all
that
box-loc
enter-pot-neg
‘These apples can’t all enter the box.’
c. Kono
ringo-wa
zenbu
sono
hako-ni
hair-anai.
these
apples-top
all
that
box-loc
enter-neg
‘These apples won’t all fit into that box.’
Jacobsen (1992 : 98)
A verb hair- ‘enter / fit into’ is ambiguous between an intentional meaning and a spontaneous
meaning. When it is followed by the potential particle as in (26a), however, the subject is interpreted as having an intention of entering. On the other hand, when the subject is inanimate as
in (26b), the potential particle cannot follow the verb. The meaning of being able to fit into
something is expressed without the potential particle as in (26c). Given these, we can conclude that the potential particle can attach to intentional verbs but not to non-intentional
verbs.22)
Applying the potential test to ext-poss verbs, we get the following results.
22) Restriction of the potential particle may be narrower than intentional verbs. It may be that it can attach
to agentive verbs as the following sentence does not seem to be acceptable.
(i) *Kaze-ga
minami-kara
fuk-e-ru.
wind-nom
south-from
blow-pot-pres.
‘The wind can blow from the south.’
116
Osaka Keidai Ronshu, Vol. 64 No. 1
(27) Ext-poss verbs
a. Taro-wa
sono
biyooin-de
kami-ga / o
kir-e-ru.
T-top
the
hair.salon-at
hair-nom / acc
cut-pot-pres
‘Taro can get a haircut at the hair salon.
b.*Taro-wa
hone-ga / o
or-e-ru.23)
T-top
bone-nom / acc
break-pot-pres
‘Taro can break the bone.’
c. Taro-wa
5-ji-ni
me-ga / o
samas-e-ru.
T-top
5-o’clock-at eye-nom / acc
wake-pot-pres
‘Taro can wake up at 5 o’clock.’
d.*Taro-wa
koshi-ga / o
itame-rare-ru.
T-top
back-nom / acc
hurt-pot-pres
‘Taro can hurt his back.’
e.*Taro-wa
ashi-ga / o
sukume-rare-ru.
T-top
leg-nom / acc
duck-pot-pres
‘Taro can duck his leg.’
f.*Taro-wa
mi-ga / o
chijime-rare-ru.
T-top
body-nom / acc
shrink-pot-pres
‘Taro can shrink his body.’
3. 4 The -ni iku ‘go to’ attachment test
The final test is to attach -ni iku ‘go to’ to verbs. The phrase -ni iku can attach to intentional
verbs (28) but not to non-intentional verbs (29).
(28) a. Taro-ga
T-nom
‘Taro went
b. Taro-ga
T-nom
‘Taro went
c. Taro-ga
T-nom
‘Taro went
(29) a.*Taro-ga
T-nom
‘Taro went
b.*Taro-ga
T-nom
‘Taro went
gohan-o
tabe-ni
it-ta.
meanl-acc
eat-NI
go-past
to eat meal.’
hashiri-ni
it-ta.
run-NI
go-past
to run.’
Hanako-no uta-o
kiki-ni
it-ta.
H-gen
song-acc listen-NI
go-past
to listen to Hanako’s songs.’
Hanako-no uta-ga
kikoe-ni
it-ta.
H-gen
song-nom
hear-NI
go-past
to hear Hanako’s songs.’
Hanako-ni
ni-ni
it-ta.
H-dat
resemble-NI
go-past
to resemble Hanako.’
23) This sentence is acceptable with a reading where Taro is an agent and breaks someone’s bone or with
an idiomatic reading of hone-o oru ‘take (great) pains, make efforts.’
An analysis of Inalienable possession constructions in Japanese
117
We get the following results from the application of the -ni iku test to ext-poss verbs.
(30) Ext-poss verbs
a. Taro-ga
kami-o
kiri-ni
it-ta.
T-top
hair-acc
cut-NI
go-past
‘Taro went to get a haircut.
b.*Taro-ga
hone-o
ori-ni
it-ta.
T-top
bone-acc break-NI
go-past
‘Taro went to break his bone.’
c. Taro-ga
me-o
samashi-ni
it-ta.
T-top
eye-acc
wake-NI
go-past
‘Taro went to wake up.
d.*Taro-ga
koshi-o
itame-ni
it-ta.
T-top
back-acc hurt-NI
go-past
‘Taro went to hurt his back.
e.*Taro-ga
ashi-o sukume-ni
it-ta.
T-nom
leg-acc duck-NI
go-past
‘Taro went to duck his leg.’
f. *Taro-ga
mi-o
chijime-ni
it-ta.
T-nom
body-acc
shrink-NI
go-past
‘Taro went to shrink his body.’
(31) summarizes the results of the four tests obtained in this section.
(31)
Adversity
passive
Imperative
Potential
-ni iku
a) get a haircut
b) break one’s bone
*?
*
*
*
c) wake up
d) hurt one’s back
*?
*
*
*
e) duck one’s legs
*
*
*
*
f) shrink one’s body
*
*
*
*
The results show that the ext-poss verbs of (31a) and (31c) allow intentional subjects while
those of (31b) and (31d)(31f) do not. I will call the former type, intentional ext-poss verbs,
and the latter, non-intentional ext-poss verbs. The results of (31a) and (31c) are especially interesting since the subjects of these verbs are not agentive, yet they are treated as intentional
entities. The results confirm Hasegawa’s treatment of (13a
i), which is (31a). (31a) and
(31c) fit into Hasegawa’s (12a) slot, and (31b) and (31d)
(31f) into (12b) slot. A revised
table is given in (32).
118
Osaka Keidai Ronshu, Vol. 64 No. 1
(32)
Intentional [+ER]
transitive / Transitive
intransitive [+OC]
alternating
verbs
Intransitive
[−OC]
Non-intentional [−ER]
a. Intentional ext-poss verbs b. Non-intentional ext-poss verbs
e. g.) get a haircut, wake up e. g.) break one’s bone, hurt
one’s back, duck one’s legs,
c. Hybrid verbs
d. Int-poss verbs
e. g.) rise / wake up, bend e. g.) one’s hair snaps, one’s bone
down
breaks
(32c) slot is occupied by a hybrid type discussed in Motomura (2012). The relevant sentences are given in (33)(34). In particular, (33b) and (34b) are cases of the hybrid type
(32c). Motomura (2012) argues that (33b) and (34b) are derived with a little v of [+ER,
−OC].24)
(33) a. Taroo-ga karada-o
T-nom
body-acc
‘Taroo rose up.’
b. Taroo-ga oki-ta.
T-nom
rise-past
‘Taroo rose / woke up.’
(34) a. Taroo-ga koshi-o
T-nom
hip-acc
‘Taroo bent down.’
b. Taroo-ga kagan-da.
T-nom
bend-past
‘Taroo bent down.’
okoshi-ta.
raise-past
kagame-ta.
bend-past
To complete the table, int-poss verbs of (1b) and (7b)(11b) are all classified to (32d).
I have tested intentionality of the subject of ext-poss verbs and classify inalienable possession verbs into the four types in (32) based on Hasegawa’s [±ER, ±OC] features. In the
next section, I will propose derivations of those types of inalienable possessor constructions.
4. A proposal
In this section, I will present a modified analysis of Hasegawa’s proposal and account for the
four characteristics specific to ext-poss verbs.
4. 1 Intentional ext-poss verbs
The properties of ext-poss verbs are recapitulated here:
24) (33a) and (34a) are classified as intentional ext-poss verbs in (32a).
An analysis of Inalienable possession constructions in Japanese
119
(35) a. The subject is an inalienable possessor of the object.
b. Interpretation of the subject is ambiguous between non-agentive and agentive.
c. When a reflexive anaphor appears as a possessor of the object, the non-agent reading disappears.
d. When the object is relativized, the non-agent reading disappears whereas when the
subject is relativized, both readings are retained.
The subject and the object of (1a), repeated below, show an inalienable possession relation
(=(35a)), and the subject is interpreted as either non-agentive (36ai) or agentive (36aii)
(=(35b)).
(36)
Taro-ga
kami-o
kit-ta.
T-nom
hair-acc
cut-past
‘(i) Taro got a haircut. / (ii) Taroo cut (someone’s / his own) hair.’
Recall, however, that the test results in section 3 indicate that the subjects of both readings
are intentional entities and are classified as intentional ext-poss verbs in (32a). Furthermore,
when an overt reflexive appears as in (3a), repeated in (37), the non-agent reading disappears
(=(35c)).
(37)
Taro-ga
zibun-no kami-o
T-nom
self-gen
hair-acc
‘Taro cut his own hair.’
kit-ta.
cut-past
I will first discuss these three properties (35a)(35c) in this section. Discussion of (35d) will
be given in the next section.
Derivations of the two readings of (36) are given in (38).
Osaka Keidai Ronshu, Vol. 64 No. 1
120
(38) a. intentional (non-agentive) (=(36
i))
b. intentional (agentive) (=(36
ii))
TP
TP
T
DP1
Taro
Taro
vP
t1
T
DP1
-possessor / agent
T
v
VP
NP
vP
t1
v
-agent
v[+ER,+OC]
VP
V
DP2
-possessor
N
v[+ER,+OC]
V
cut
t1
T
cut
pro
3
NP
hair
t3
-possessor
N
These are slight modification of Hasegawa’s derivations in (15). Following Vergnaud and
Zubizarreta (1992), I assume that the inalienable possessor, DP1, is base generated as the possessor of N. Furthermore, I assume that the object of ext-poss verbs is generated either as NP
or as DP. When it is generated as NP as in (38a), DP1 must move out of the NP for Case ;
otherwise, it would remain Caseless based on the assumption that a genitive Case is licensed
in spec. DP (Miyagawa 1993, Ochi 2001). This possessor raising ensures the inalienable possession relation between the subject and the object (=(35a)). DP1 first moves to spec. vP,
where it gets its second -role, agent, which is a cover term for agent, causer, and intentional
entity. Then it moves to spec. TP, where it gets nominative Case. By allowing movement into
a
-position advanced by Hornstein (2001, 2009) among others, we can account for the dual
status of the subject as a possessor and an intentional entity at the same time.
On the other hand, the object is generated as DP in (38b). The possessor, pro3, gets genitive
Case at spec. DP2 and stays there. DP1 is merged at spec. vP as agent and moves to TP for
nominative Case. This is a derivation of standard transitive sentence. In both (38a) and (38b),
[+ER, +OC] features of the little v ensure the agent subject and the accusative object. Thus
availability of NP and DP as the object leads to the two readings of (36) (=(35b)).
Zibun can replace pro3 in (38b), but there is no such pro in (38a). Therefore, (38b) is the
only derivation available to (37). This means that the subject of (37) is interpreted only as an
agent (=(35c)).
Recall that the ambiguity of (36) is a special property of inalienable possession verbs. I proposed above that the non-agentive reading of (36i) is derived by the verb’s ability to select NP
as its object. This is based on Vernaud and Zubizarreta’s (1992 : 612) Correspondence Law ;
An analysis of Inalienable possession constructions in Japanese
121
(39) The Correspondence Law : When a DP or an NP denotes, the DP denotes a token and
the NP denotes a type.
They wrote ;
‘whereas tokens may or may not be directly associated with objects in the world, types
may not be so associated. A type may be associated with an object in the world only indirectly, by instantiation as a token . . . Each token is an instantiation of some type.’
According to this law, the NP object in (38a) is a type, not a token, and the object is not a real
substance ‘hair.’ Therefore, it forces us to interpret the sentence as a hair-cutting event takes
place, but the subject is not the agent of the event because there is no real hair to be cut by the
subject. As a result, the subject is interpreted as an intentional entity having intention of a haircut. This is the reading of (36i).
On the other hand, DP is a token ; therefore, when the verb selects DP as its object in (38b),
it can be associated with a real object in the world, namely, ‘(someone’s) hair.’ Thus the subject can be interpreted as an agent, and the sentence expresses a hair cutting event by the subject.
4. 2 Subject object asymmetry in relative clauses
We have seen in section 1 that a non-agentive reading disappears when the possessee object
of ext-poss verb is relativized whereas both agentive and non-agentive readings are retained
when the possessor subject is relativized. The relevant examples are given here.
(40) a. otoko-ga
kitta
kami
(agent / *non-agent)
man-nom
cut
hair
‘The hair that a man cut’
b. kami-o
kitta
otoko
(agent / non-agent)
hair-acc
cut
man
‘a man who got a haircut / cut the hair’
In order to account for this contrast, we need to discuss two issues : the basic structure from
which these relative clauses are derived and structures of relative clauses in Japanese.
First, the non-agentive reading of (40b) is derived from the intentional (non-agentive) extposs construction in (41a) whereas the agentive reading of (40ab) comes from (41b).
(41) a. Otoko i-ga
[NP t i kami]-o
kit-ta.
man-nom
hair-acc
cut-past
‘A man got a haircut.’
b. Otoko-ga
[DP pro kami]-o
kit-ta.
T-nom
bone-acc
cut-past
‘Taro cut (someone’s) hair.’
(non-agentive)
(agentive)
Osaka Keidai Ronshu, Vol. 64 No. 1
122
The second issue is the structure of relative clauses in Japanese. According to the base generation analysis of relative clauses in Japanese (Kuno 1973), the gap in a relative clause is pro
as shown in (42a). Since pro is a type of pronoun, which is treated as DP, the gap of the
relativized object should be DP as in (42b).
(42) a. [IP otoko-ga
man-nom
b. [IP otoko-ga
man-nom
pro i
[DP pro]i
kitta]
cut
kit-ta]
cut
kami i
hair
kami i
hair
Therefore, the structure from which (40a) is derived should be (41b) where the object is DP.
Since (41b) is the structure for the agentive reading, only the agentive reading is available
when the object is relativized in (40a).
On the other hand, when the subject is relativized, there are two structures : a non-agentive
reading (43a) and an agentive reading (43b).
(43) a. [IP [DP pro]i
[NP
ti
b. [IP [DP pro]i
[DP
pro j
kami]-o
hair-acc
kami]-o
hair-acc
kit-ta]
cut
kit-ta]
cut
otoko i
man
otoko i
man
Since the gap is the subject, (43a) and (43b) are both possible structures. Hence, both readings are available in the case of subject relativization.
4.3 Non-intentional ext-poss verbs and int-poss verbs
Verbs such as (7a) and (9a)(11a) are classified as non-intentional ext-poss verbs in (32b),
and all the int-poss verbs (1b) and (7b)(11b) occupy (32d) slot. Relevant sentences are repeated below.
(44) a. Taro-ga
koshi-o
itame-ta.
(=(9a))
T-nom
back-acc hurt-past
‘Taro hurt his back.’
b. Taro-no
koshi-ga
itan-da.
(=(9b))
T-gen
back-nom hurt-past
‘Taro’s back hurt.’
Derivations of the two sentences are illustrated below.
An analysis of Inalienable possession constructions in Japanese
(45) a.
TP
b.
T
DP1
123
TP
T
DP1
Taro
vP
VP
NP
T
Taro’s back
v[−ER,+OC]
V
VP
t1
hurt
t1
vP
T
v[−ER,+OC]
V
hurt
N
back
In (45a), the object is NP, and the possessor of the object, DP1, moves out of the NP and raises
to TP. In this case, the little v has [−ER], so the subject cannot be interpreted as agent or
intentional entity. Recall that the non-intentional ext-poss verbs in (32b) all reject the
intentionality tests in section 3. We can take this property as a consequence of little v having
[−ER].
Finally, (45b) is the derivation of an int-poss verb (44b). In this case, the possessor stays
inside DP1 and the whole DP1 moves to TP. Since the little v has [−ER, −OC], neither external argument nor accusative Case is available. In other words, this is a case of standard
unaccusative.
5. Remaining issues
There are at least two problems in the analysis presented above. Although I do not have any
concrete answers, I will illustrate those problems with potential solutions to them.
5. 1 A problem of a genitive marker no
In the above sections, I have claimed that genitive Case is checked / assigned at spec. DP. Since
ext-poss verbs do not select DP but NP, the possessor of the object must move out of the NP
since no Case is available inside the NP. However, the genitive particle, -no, seems to be able
to appear internal to NP in some cases as shown in (46).
(46) a. Taro-ga
kami-no
ke-o
kit-ta.
(agentive / non-agentive)
T-nom
hair-gen
hair-acc
cut-past
‘(i) Taro got a haircut. (ii)Taroo cut (someone’s) hair.’
b. Taro-ga
ashi-no
hone-o
ot-ta.
(agentive / non-agentive)
T-nom
leg-gen
bone-acc break-past
‘(i) Taro broke his bone. (ii) Taroo broke (someone’s) bone.’
These are problematic for the current analysis since I assume that the genitive Case is as-
124
Osaka Keidai Ronshu, Vol. 64 No. 1
signed / checked structurally at spec. DP position.
Consider the following sentences where the possessor noun and the possessee noun are not
adjacent.
(47) a. Taro-ga
ashi-no
hosoi
hone-o
ot-ta.
(agentive / ??non-agentive)
T-nom
leg-gen
thin
bone-acc break-past
‘Taro broke his leg’s thin bone.’
b. Taro-ga
kami-no
shiroi
ke-o
kit-ta.
(agentive / *non-agentive)
T-nom
hair-gen
white
hair-acc
cut-past
‘Taroo cut (someone’s / his own) white hair.’
In these cases, the non-agentive readings seem to disappear, which suggests that the
possessee object is DP. Then, the -no particle of the possessor nouns in (47) is a structural
genitive Case.
On the other hand, the -no particle in (46) might be a morphological case which is licensed
only when the possessor noun is adjacent to the possessee noun.
5. 2 Passives
The non-agentive sentence (48) cannot be passivized unlike other standard causative verbs.
(48b) is not acceptable with a non-intentional reading.
(48) a. Taro-ga
kami-o
kit-ta.
T-nom
hair-acc
cut-past
‘Taro got a haircut.’
b.*Kami-ga
Taro-ni
kir-are-ta.
hair-nom T-dat
cut-pass-past
‘The hair was cut by Taro.’
Presumably, a direct passive requires an agent subject transitive verb, and an intentional
subject is not enough for a direct passive.
6. Conclusion
In this paper, I tried to account for the characteristics specific to external possessor verbs. The
crucial assumption is that the object of ext-poss verbs can appear as either NP or DP. Inalienable possession relation is a consequence of having NP as the object. This assumption allows
us to account for the contrasts between ext-poss verbs and standard causative transitive verbs.
It is also important that derivations proposed here allow movement into a -position advanced
by Hornstein (2001, 2009) among others. By allowing a single nominal expression to have
more than one -role, we can account for the dual status of the subject of ext-poss verbs as a
possessor and an agent / intentional entity.25)
25) Nakamoto (2010) proposes a control approach to inalienable possession constructions in French. The
An analysis of Inalienable possession constructions in Japanese
125
References
Burzio, L. (1986) Italian syntax : a government-binding approach. Kluwer : Dordrecht.
Chomsky, N. (1995) Minimalist program. MIT Press : Cambridge, MA.
Collins, C. (1997) Local economy. MIT Press : Cambridge, MA.
Harada, S. (1973) Counter-equi NP deletion. In Annual Bulletin 7. Research Institute of Logopedics
and Phoniatrics, University of Tokyo. Reprinted in N. Fukui (ed.), Syntax and meaning : S. I.
Harada collected works in linguistics, 181215. Taishukan : Tokyo.
Harley, H. (1995) Subjects, events, and licensing. Doctoral dissertation. MIT : Cambridge, MA.
Harley, H. (2008) On the causative construction. In S. Miyagawa and M. Saito (eds.), the Oxford
53. Oxford University Press : Oxford.
handbook of Japanese linguistics, 20
Hasegawa, N. (2001) Causatives and the Role of v : agent, causer, and experiencer, in K. Inoue and
N. Hasegawa (eds.), Interdisciplinary Research, Proceedings of the COE International Symposium :
1
35. Kanda University of International Studies : Chiba, Japan.
Hasegawa, N. (2004) ‘Unaccusative’ transitives and Burzio’s generalization : reflexive constructions
in Japanese. Proceedings of the Workshop on Altaic Formal Linguistics, vol. 1. MIT Working Papers
in Linguistics, 46 : 300
314.
Hasegawa, N. (2007) The possessor raising construction and the interpretation of the subject. In S.
Karimi, V. Samiian and W. K. Wilkins (eds.), Phrasal and clausal architecture : Syntactic derivation
and interpretation : 62
99. John Benjamins : Amsterdam.
Hasegawa, N. (2009) Chokusetsu judo-bun-to shoyuu judo-bun : little-v to shiteno “rare” to sono
sosei [Direct passives and possessor passives : “rare” as a little-v and its features (translation by
M. M.)]. In Goi-no imi-to bunpo [Meaning of lexicons and grammar (translation by M. M.)], 433
454. Kuroshio : Tokyo.
Homma, S. (2004) Derivation of the experiencer subject of transitves verbs. In Tsukuba English
Studies, vol 22, 247
262.
Hornstein, N. (2001) Move ! A minimalist theory of construal. Blackwell : Oxford.
Hornstein, N. (2009) A theory of syntax : minimal operations and Universal Grammar. Cambridge
University Press : Cambridge.
Howard I. and A. Niyekawa-Howard (1976) Passivization. In M. Shibatani (ed.), Japanese Generative Grammar. Syntax and semantics 5. Academic Press : New York.
Inoue, K. (1976) Henkei Bunpoo to Nihongo [Transformational grammar and Japanese]. Taishukan :
Tokyo.
Jacobsen, W. (1992) The transitive structure of events in Japanese. MIT Press : Cambridge, MA.
Kratzer, A (1996) Severing the external argument from its verb. In J. Rooryck and L. Zaring (eds.),
Phrase structure and the lexicon, 109137. Kluwer : Dordrecht.
Kuno (1973) The structure of Japanese language. MIT Press : Cambridge, MA.
Kuroda, S. -Y. (1965) Generative grammatical studies in the Japanese language. Doctoral dissertation. MIT : Cambridge, MA.
Miyagawa, S. (1993) Case-checking and Minimal Link Condition. In MITWPL 19, Papers on Case
and agreement II, C. Phillips (ed.), 213
254. MIT Working Papers in Linguistics : Cambridge,
idea is that the possessor external to its corresponding possessee position moves to a Case position
through the extra -position available in the derivation. He adopts the movement approach to control advanced by Hornstein (2001), where movement into a theta-position is allowed.
126
MA.
Miyagawa, S. (1999) Causatives. In N. Tsujimura (ed.), the handbook of Japanese linguistics, 236
268. Blackwell : Oxford, UK.
Motomura, M. (2012) Transitive-intransitive pairs with inalienable possession relations in Japanese. In the proceedings of the 13th Tokyo conference on psycholinguistics, Hitsuji-shobo : Tokyo.
Motomura, M. (forthcoming) Proposals concerning derivations of indirect passives and possessor
passives in Japanese. In the proceedings of the 37 th conference Kansai linguistics society.
Ochi, M. (2001) Move F and ga / no conversion in Japanese. Journal of East Asian Linguistics, 13 :
247
286.
Pesetsky, D. (1995) Zero syntax. MIT Press : Cambridge, MA.
Shibatani, M. (1990) The languages in Japan. Cambridge University Press : Cambridge.
Takezawa, K. (1991) Judo-bun, noukaku-bun, bunri-fukano shoyuu koubun-to teiru no kaishaku
[Passives, ergatives, inalienable possessor constructions and interpretation of teiru (translation
by M. M.)]. In Y. Nitta (ed.), Nihongo-no voice-to tadosei [ Japanese voice and transitivity (translation by M. M.)], 59
81. Kuroshio : Tokyo.
Terada, M (1990) Incorporation and argument structure in Japanese. Doctoral disserataion, University of Massachusetts, Amherst.
Vergnaud, J. -R. and M. L.Zubizarreta (1992) The definite determiner and the inalienable constructions in French and in English. In Linguistic Inquiry 23, 595652.