AGENTIVE
Q: P(e,X,y)⇒ P(e,X=Z,ガ
Japanese,such as′ α′θrν ̀make something stand up'and 4α rαbθrν ̀stand something in a row',are forlned by the sufflxation of― θ―which induces the argument substinltion at the level of lexical semantic strLICture of their corresponding intransitive verbs,such as′ α′ν
̀stand up'and αrαbν ̀stand in a row'。
(6.44)ao Tar00‑ga guraundo―ni tat―ta.
Taro―Nonl ground― ]Dat stand―Past
̀̀Taro stood on the ground."
b. Taroo―ga guraundo―ni hata―o tat―e―ta.
Taro―Nom ground―
Dat iag―
Acc stand―E―Past̀̀Taro put up a flag on the ground."
If this obseⅣation is valid, it may lead us to posれ1late that the rl119 of argument substitution, grounded on our cognitive faculty9 is universally a/pplicable to a certain extent,although its lnorpho― syntactic effect lnay vary froln language to language(ci Kageyama and Yumoto 1997).
6。 1。4。
Argument Binding
The forth, and the last, lexical semantic rLlle On arguments is thè̀binding"of argumentso Following Kageyama(1996),who adVOCates that the same phenomenon as binding arguments in syntax is also operative in the lexical semantic representation,I propose that the lexicon has the rLlle Of̀̀argument binding"in(6。 45),whiCh applies to the qualia strLICmre Ofpredicates.
(6.45)∠
rgν θ″BJ4清″g
in(6。45)is notiOnally the same as Kageyama's̀̀anti― causat市 ization",but its outcome in the syntax will be incontrovertibly different.
Kageyama(1996)argueS that the nlle like argument binding is responsible to the causative/inchoative altemation given in(6。46).
(6.46) Cα νsα′JソC/多zεttθα′Jソθ∠ルθrηα′Jθ″ a. Janet broke the cupe
b. The cup broke. (Levin 1993:29)
Kageyama's clailn is that arguments that are bound in the lexicon will bè̀suppressed"
in the sense that they become syntactically inerto Consequently9 if the l■ lle of argument binding applies to the extemal argument of a causative predicate,the bound extemal argument will be suppressed lexicall光 so that the only intemal argument can be realized in the syntax,giving rise to the inchoative variant ofthe altematione
CrLICiall光 however,there sccms to be no con宙 ncing reason in Kageyama's(1996) linking strategy why the bound argument does not have any syntactic realization.In fact, unlike passives,the inchoative variant of the causative/inchoative altemation does not have a PRO in the syntax。
(6.47)a.
b.
(6.48)a.
b.
*The door opened by John.
The door was opened by John.
(Kageyama 1996:143)
*The buildings bumed to coHlectinsurance.
The buildings were bumed to coHiectinsurance.
(Keyser and Roeper 1984:407)
In order to capture the casc,Kageyama's approach is forced to assume that suppression by argument binding is different in its quality from that in the case of passivizatione ln effect,he needs to presuppose different levels of representations,LCS and argument strtlcture,for two distinct levels at which argument suppression is applied.In section 3.2, we have already established the mechanisln of the causative/inchoative altemation in
tems of selectional inapping via event― headednesso ln our approach,therefore,there is no need to depend on different l■ echanislns for capturing the argument realization of causative/inchoative pairs of sentences.
In view ofthe circumstance,I rather suggest that the rLlle Of argument binding is responsible to the phenomena that include reflexivization"in the syntax.For example, when the agent argument of a causative verb binds its theme argument,nollllal reflexive sentences are produced。
(6。
49)a.JOhn cut himsell
bo They killed thelnselves.
c. Mary washed herseli d. I dressed mysel■
In(6。49),the agent argument ofthe verb binds semantically the theme argument,so that
the latter will be realized in syntax as a re■ exivё pronoun.
It is also possible that the theme argument of a causative verb binds semantically its agent argument. This scems what is happening in the so― called re■exive diathesis altemations,including the viJual re■ exive altemation in(6。 50).
(6。
50)
万r′ναJ Rσ7θε′JソθИルθr4α′Jθη a. The butcher cuts the lneatebo This lneat cuts itsell (Levin 1993:84)
Recall the lexical semantic representation ofε ν′in(6。51)。
(6.51) cut
QUALIA=CONST=i:cutlery
FORMAL=be(e2*,y9璽
)AGENTIVE=act(el*,X,i)
The rtlle of argument binding binds the agent argument(x)in the agentive quale with the theme argument(y)in the fom■ al quale,where the fo11ller Causes a change of state in the latter,so that it gives rise to a peculiar interpretation as if a theme brings about its change of state by itsel■ The argument realization of the vi■ual reflexive variant in
(6。50b)will be as follows.
(6.52)a.QA:aCt(el*,X,i)=⇒ QA:aCt(el*9X=y9i)
b・ QA:aCt(el*,X=y9i)― → [ソPX=y[ν ′ソVP]]
QF:bC(e2*,光
cul)→
[vP y[v′ V璽]]c. 1/P ←―
QA:aCt(el*,X=y9i)＼
the lnea場=ノ
ソ′
/＼
ソ
VP←
QF:be(e2*,勇璽
)/＼
the meaち
V′
V cut /＼
By viJhe of argument binding in(6。 52a),twO arguments with the identical semantic valuc are m町〕ped to the syntax,as in(6。 52b)。 Of the two instances of ttθ θα′in
(6.52c),the 10Wer occulcnce will be spelled―out as the reflex市e pronoun(ioe.,Jおθ:り according to the syntactic constraints on argument binding.
An ilnmediate consequence of this analysis is that verbs with a silnple event
strtlcture cannot participate in the vilttual reflexive altematione
(6。53)a.*JOhn danced hilnsel■
b.*The ambassador aHived hilnsel■
c.*The washing machine was sold itsel■
(Fellbaum 1989:126)
Predictably9 the result syntactic strLICture of argument binding needs a complex verb phrase as in(6。52c).TherefOre,trLle unergatives,truc unaccusatives and passives are all excluded from the altemation.
In our approach,it can also be correctly predicted that argument binding does not apply to verbs that strongly imply the existence of an agente
(6。54)a.*SOft WOod saws itsel■
b.*These rolls slice thelnselves.
c。 *Smooth surfaces paint over thelnselvese
(Fellbaum 1989:128)
Since verbs in(6.54)semantiCally require the active involvement of an agent, in the sense that a theme cannot be understood as an entity that perforlns the action by itselt the agent arguments of these verbs may resist to be bound by other arguments at the level oflexical semantics.
FurthelHnore,the rLlle Of argument binding gives a silnple account for the fact that inteHnediary instruments can never be the sutteCt Ofvilual reflexive variants.
(6.55) *ThiS pipe smokes itsel■
(Ci ThiS pipe slnokes easily.)
(Fellbaum 1989:129)
Since intellllediary instrLlrnents appear in the agentive quale,they cannot bind the agent argument in the qualia strLICture,though they can be the sutteCt Of middle sentences when the agent argument is demoted via argument demotion and the theme argument is abstracted away via argument abstraction.
The virtual renexive altemation seems to have other special properties that rnay not receive an instant solution in our approach(sec Fellbaun■ 1989).It iS also supremely difflcult to limit the set of verbs that participate in this altemation(ci Le宙
n1993).In
short,the mle of argument binding is lnerely a necessary condition for the forlnation of the vi■ual re■exive variant in(6。50b).In particular,we do not have any satisfactory explanation to the restrictions on the applicability ofargument bindingo The exact namre
ofthis altemation awaits further research.
6。2.Operations on Qualia
Another aspect of the clailn that argument altemations without a logical semantic shift do not involve an extension of the semantic strtlcmre Of predicates resides in the variety of argument linking itsell ln this section,three lexical operations that affect on the qualia stluclLlre of predicates are introduced,all of which should be counted as a generative device that provides a hndamental patteHl ofargument realization.
6。 2。 1。 Constitutive Specirlcation
ln section 2。3。2.3,we saw that arguments in the constitutive qualia of predicates can be realized in the syntax as trLle attuncts.The typical casc of such arguments is time and location associated with the action denoted by the verb.
(6.56)a.John sleptlate on Tuesday.
bo Mary saw Billin Boston. (PuStaOVsky 1995:66)
In(6。56),arguments that provide a property of the event are realized in the sentence in order to specify the situation being described.We call this linking pattem̀̀constimt市 e speciflcation",since the infoHnation expressed as trtle attunCtS needs to be specifled in the constitutive quale of a predicate.
Arguments in the constimt市 e quale ofthe direct ottect Can be realized in syntax in allnost the same mannen This operation gives a straightfolward explanation to the body―part possessor ascension altemation,given below。 4
(6。
57) 3ο
冴ソーPαr′Pθssθssθr/scθ ″sJθη∠Jたr α′Jθκ a. Selina touched the horse's back.bo Selina touched the horse on the back. (Levin 1993:71)
Consider the lexical semantic representation ofゎ νεtt in(6.58):
(6.58) touch
QUALIA=FORMAL=be(e2*,y9touched) AGENTIVE=act(el,X)
It is now obvious that ifthe valuc ofthe theme argument(y)of′ θνεtt is deteⅡnined by the NP ttθ ttθだθζ bαεた,the nollllal transit市 e sentence as in(6。
57a)comeS Out.When,
on the other hand,the value ofthe theme argumentis assigned by the NP′ 乃θ ttθだθ,the VP′θνε乃″θ ttθだθ in(6.59)will be cOmposed。(6。59) touCh thC horse
QUALIA=CONST=y:the horse
EQuALIA=CONST=Inane,hoo鳥
neCk,back,ctc。 コFORMAL=be(e2*,y:the horse,touched) AGENTIVE=act(el,X)
Hcre,the theme argument in the constimt市e quale(i.e。 ,励θ ttθだθ)has its own qualia
stmcl比lre including its constinltive quale.Suppose that constitutive speciflcation applies to this semantic strtlcmre ofthe vPo Then,the embedded constinltive quale of′ 乃θ ttθだθ can be a target ofthe operation nowo Thus,various speciflcation of the body― part of′乃θ 乃θだθ can be possible.
(6。60)Selina tOuched the horse on the{mane/hoo7neck/back/foot/nose/head}.
This analysis corectly predicts that the specifled body part of the horse should have an appropriate semantic relation with the ottectO In effect,an attunct PP must be a part ofthe ottect,but not vice versa.
(6。61)#Selina touched the{mane/hooyneck/back/foot/nose/head}on the horse.