Cas e ー As s l gnme ntandPas s i veConns t r uc t i on
Hisao Yamamoto
Ingenerativegrammar,Case‑asslgnerSaregenerally assumed to be [‑ N] Categories, not [+ N ] Categories. However, there is a
・proposalthat[+N]categories assign Case,at least inherent Case・
・In this paper l discuss the system of Case asslgnment in English・
I also discuss the representation of the D‑Structure of passive construction,argulng that its head predicate is a passive participle generatedinthelexiconbytheattachmentofthepassivemorphemetoa baseverb.
1. Introduction
ln current generative grammar,for example,Goverment‑Binding theory (GB‑theory),NPs must have Case under the Case theory.
It is generally assumed that Case‑assigners are [‑ N ] Categories.
Furthermore,there is a condition called the Adjacency condition, which requires that a Case‑asslgnee be adjacent to its Case‑asslgner.
Butthereisaconstructioninbhicha verbtakestwoNP complements, thesecond ofwhich isnotadjacentto the head verb. For example, wehavethefollowlngSentenceinEnglish:
(1)JohngaveBillthebook ,
The second NP in (1) must have Case to satisfy the Case filter, but it is not adjacent to the verb gave,a Cace‑assigner. (1) is grammaticalcontraytoourprediction. Thuswemustaccountforhow t壬leNP inqlleStioncanbeasslgnedCaseundertheCasetheory.
Wealsohaveapassivesentencerelatedto(1) : (2)BillwasgiventhebookbyJohn
ln GB‑theory,passiveparticiples are assumed to have no ability to assign Case・ In (2),therefore,the NP the book cannot be assigned
〔185〕
186 Review .ofLiberalArts,No.74
Caseby
g乙 Ue n
eVen\ifitisadjacentto it. No otherCase‑asslgner is availablefortheNP.(2),then,Shouldberuled outasungrammatical by the Case filter,but it is in factgrammatical. This means that the object NP is asslgned Case in some way,satisfying the Case filter.In this artical,I investlgate the system of Case assignment, anddiscussthepropertiesofpassivecoristruction.
2. Framework
ln thispaperlassume the framework ofGB一七heory proposed by Chomsky (1981, 1985, 1986). In this theory Univ占rsal Grammar consistsofslユbtheories,whichinteractwitheachothertoexplainvarious linguisticphenomena.
Constituentstructuresaredetermi ned byX‑bartheory,underwhich wehave no category‑specific rewriting rule. So we do nothave the followingruleswhichrefertothecategoriesSandS':
(3)a.S'‑→COMP S S‑‑NP INFL VP
Instead,these clausalcategories are generated asC〝and I〟,respec‑ tively,intermsofX‑bartheory,asin (4):
(4) a.S′‑C〝‑[…[C,CI〝]] b.S‑Ⅰ〝‑[NP[Ⅰ′[vpV∴]]]
I use conventional notation for the maximal projection of lexical categories,including those of nonlexicalcategories. Thus VP‑Ⅴ〝, NP‑N",AP‑A",Pp二;P〝,IP‑Ⅰ",and CP‑C". Thecategory imme‑ diately dominated by XP is the specifier of X (or X′,or XP).
Move‑amovesa Luh‑Phrase to thespecifier position ofCP. The NP immediately dominated by IP is the spccifier ofa sentence in X‑bar theoreticalterms,Or the subject in traditionalterms. And so with theNPimmediatelydominatedbyanotherNPinthenominalconstruction・
The Np‑specifier is optlOnal whereas the IP‑S占ecifier is obligatory・
Thisdifferencebetween them willfollow from the Predication theory propos占dinWilliams(1980)and Rothstein (1983). Seealso Chomsky
CaseAssignmentandPassiveConstruction(HisaoYamamoto) J87 (1982,1985,1986),in which theobligatorinessoftheIp‑specifieris statedintermsoftheExtendedProjectionPrinciple.
In Chomsky(1986),itis argued ̲that the choice ofx'is forced when thereisa specifier,otherwise optlOnal. This optlOnality bfX′
is important in the currentGB‑theory,since the choice ofX′bears cruciallyonthedeterminationofbarrierhood,especiallywhenweconsider theMinimalityConditionongovernment. Thistopicwillbedealtwith inthenextsection.
ThechoiceofcomplementsisdetermindbytheProjectionPrinciple.
Their order is determind by the Case theory. In English, fur‑ ermore,a Case‑asslgneemustbeadjacentto itsCase‑asslgner',aS Was mentionedintheintroductoryremarks. Thisrequirement,thatis,the Adjecencycondition,Willbeowlngmainlytothepovertyofmorphological forms realized on lexical items. English is one such language・ Thustheungrammmaticalityof(5b)isexplainedbythiscondition:
(5) a. Johnputthebookonthetable b.*Jbhnputonthetablethebook
In (5b),the book must receive Case,butit is not adjacent to its Case‑asslgnerPZLt. So we have the violation of the Case filter,̲ which requlreS thatevery phonetically,realized NP be assigned C墾e・
Butthisfiltercan bedispensed with in favor of the Visiblity condi‑ tion of the♂ ‑Criterion. According to the♂ ‑Criterion,a♂ ‑role isasslngnedtoachainofanargument,andtoreceiveaβ‑role,a chain mustbevisible. A chaincanbevisibleonlywhenithasCase. There‑
fore we need the Case filter no more. Notice that predicative NPs were apparent'Counterexamples to the Case filter. In general these NPsappearinthenon‑Case一markedposition. Buttheymustbeassigned Case to satisfy the Case filter when they are phonetically realized.
Thisproblerp doesnotariseany more・ SincepredicativeNPsarenot arguments,they haveno relation to theVisibility condition,which is onlyrelevanttoarguments(Orargumen七一chainsin amorestrictsence).
Predicative NPs,though,must obey one of the licensing COnditions:
apredicatemusthaveasubjecttobelicensed.
Jgg Review ofLiberalArts,No.74
As,forsubcategorization,Weneedno traditionalsllbcategorization frams. Cf. Chomsky(1965). A lexicalcategoryselectsonlyaseman‑
tic category C,not a syntactic category. Thus if a verb s‑selects asemanticcategory,then itc‑Selectsa syntacticcategory thatisthe
"canonicalstru9turalrealization ofC" (CSR (C) )・ The verb
hi t
, for example,S‑selcts Theme;then it c‑selects NP,which is assumed tobetheCSR ofTheme.Ⅰn addition to tムesesubtheoriesofUG,Ialso assume the notion
"barrierhood"proposedinChomsky(1986). CllOmSky(1986,p.
1 4)
defines"barrier"asfollows:
(6) γisa BC for β iff γis not L‑marked and rdominates
β.
( 7)
γisabarrierfor♂ iff(a)or(ち):a. rimm叫 atelydominatesj,∂aBC forβ;
b. rュsaBC forβ,γ≠IP.
In(6)and(7),BC isa blocking category. The "defective" charactor OH P isextendedtoI'・ Chomsky(1986,p.42),furthermore,proposes theMinimalityConditioLn,whichholdsofgovernmentbutnotofmovement. ThusheextendstheconceptofbarriertoincludethefollowlngCase:
(8) γis a barrier for βifγis (a projection,the immediate projection)of∂,azero‑levelcategorydistinctfromβ・
Givenγ,βasin(8),acategoryαexcludedbyγwillnotgovernβ・l
In termsofthese subtheories and the notin "barrier",Iconsider thesystem ofCaseasslgnmentandthepropertiesofpassiveconstruction inthefollowlngSections.
3. CaseAssignment
Herewepayattention to the system ofCase assignmentin GB‑
theory. Chomsky(
1 9 81
,p.1 7 0)
assumes the following properties of Caseasslgnment:1)Chomsky(1986)defines"exclusion"and"L‑marking''asfollows:
(i) aexcludesPifnoヲegmentofadominatesP・
(ii)αLmarksβiffαlSalexicalcategorythat♂‑governsβ.
CaseAssignmentandPassiveConstruction(HisaoYamamoto) 789 (9) a. NPisnominativeifgovernedbyAGR.
b. NP is Objective ifgoverned by V with subcategoriza‑
tionfeature: NP (i.e.,transitive).
C. NPisobliqueifgovernedbyP.
d. NP isgenitivein[NP‑ Ⅹ′].
e・ NP ‑is inheremtly Case‑marked as determined by propertiesofits[‑N]governor.
English,moreover,has the Adjacency condition (maybe because of morphologicalpoverty)・ Now letusconsidertheseproperties.
Nominative Case is assigned to the specifier of tensed IP in thefollowlngCOnfiguration:
(10) [IPNP[Ⅰ'Ⅰ[vpV
…
]]]In(10),Ⅰ,thehead ofIP,governs its specifier position and assigns nominativeCase to NP. However,an adverbialcategory can appear between the specifier̀ position of IP and I, as IS shownin(ll):
(ll) a。.Johnoftengoesfishing
b.[IPJohnoften[Ⅰ'Ⅰ[vpgofishing]]]
In (llb),
John
isgoverned by I,but it is not adjacent to I,since theadverbo ft e n
intervenesbetween them. But(lla)isgrammatical, whichmeansthatJohn
isasslgned nominativeCasebyI. Onemight supposethattheAdjacenty condition should not work in English,Orthatitshouldbestatedintermsofconfiguration,notlinear relation.
Thenwecouldsaythatin (lib),
Jo hn
isconfigurationally adjacentto I,receivlngCasefrom Iwith no violation oftheAdjacency condition.This argument,howe.ver,does not hold of the Case VP‑internally assigned,that is,objective Case (and also oblique Case assigned by P). Theungrammaticality of(5b),repeatedhereas(12b),mustbe explained in termsofAdjacency condtion,since it otherwise satisfies alltheprlnCiplesofUG:
(12) a. Johnputthebookonthetable b.*Johnputonthetablethebook
In(12b),thePP
on t het abl e
intervenes bwtweenput
andt heb r ook.
190 Review ofLiberalArts,No.74
TheNP thebookisan argumentofput,and itoccupleStheposition governed by put. If we interpret the configprational adjacency as beingdefinedifltermsOfgovernment,thebookshouldbeadjacenttoand as+slgned Case byput. In fact this is impossible. The object NP,
therefore,must be assigned Case by being linearly adjacent to its Case‑asslgner.
Thenwehaveto assumethatEnglishhastheAdjacency condition on Case 卑SSlgnment. This condition must be stated in terms of linearlty. NominativeCaseasslgnmeentCan beexemptfrom it. This asymmetrybetweensubjectandobjectconcernlng adjacencymightarise from thefactthatthefomerisexternal and thelatterisinternal to VP. If an NP internal to VP is seperated from its Case‑ asslgnerby anadverb‑oranadjunct,forthematterofthat,Wemight take the NP for an argument exterllal to VP. Of collrSe, this mistakewollldbeowingtOthepoormorphologicalvariationofEnglish.
Nextweconsider(9b). Verbsassigh ObjectiveCaseto their NP complements,and in this case,the Adjacency conditionLOPerateS,aS justwesaw. ButinEnglish,wehavedoubleobjectconstruction,that is,(1),repeatedhereas(13):
(13) JohngaveBillthebook
In(13),thefirstNP objectcan be assigned Case bygave,since the former is adjacent to the latter. How,then,ca'n the second NP object be asslgned Case? The Adjacency condition is not satisfied, sincethebookis not linearly adjacent togave̲Ofcourse,the book cannot be assigned Case by AGR. AGR does not govein the book, alldCaseasslgnmentisperf()rmeduJldergovernment. SincethegoverllOr ofthebookisgoue,theplausible assumption is thatgavewillalso asslgnCasetotheb‑oohin somevay・ Letusexaminethispossibility・
InYamamoto(1986),Iarguedthatin(13),thesecond NP object isassigned Caseby a ̀̀smallVP",following the suggestion made by Chomsky(1981). Thusthestructureof(13)is(14):
(14) John[vp[vpgaveBill]thebook]
Is this structure v,alid? In (14),VP forms a kind of adjunction