Anda di halaman 1dari 13

5 From Structure to Interpretation 0 Aim and scope of this chapter This chapter establishes a link between the domain

of syntax and that of semantics: we examine some of the interpretive aspects of syntax !e will try to establish relations between structure on the one hand and meanin" on the other In the course of the discussion we will postulate that in addition to #$structured and S$structure% sentences are associated with a level of representation called &o"ical Form '(F) This level was initially conceived to represent interpretive properties of *uantifiers +owever% by extendin" the concept it will be shown that the level of &o"ical Form is also relevant for the study of lan"ua"e typolo"ies This chapter is perhaps sli"htly more theoretical and less descriptive than the previous chapters It shows how% startin" from empirical ar"uments% the lin"uist can proceed to formulatin" more abstract concepts and models ( Introduction: syntax and interpretation ( ( Interpretation and semantics In very broad terms% the study of semantics involves the relation between the propositional content of a sentence and the world The notion ,world, has to be interpreted as meanin" the discourse world in which the speaker is located A declarative sentence such as '(a) contains a proposition or a claim about a state of affairs in the world: '() a -ary invited Thelma .nderstandin" '(a) boils down to knowin" the circumstances under which its propositional claim is true or false% '(a) is true if and only if the followin" conditions 'at least) obtain: there is a person called -ary and there is another person called Thelma% and there is a relation of invitin" between -ary and Thelma% in which -ary is the A/01T of the invitation and Thelma is the 2ATI01T The passive sentence '(b) is true under the same conditions as '(a) !e can say that '(a) and '(b) have the same truth conditions '() b Thelma was invited by -ary If meanin" were to be defined solely in terms of truth conditions then we mi"ht say that% thou"h structurally different% '(a) and '(b) ,mean the same thin", +owever% the reader may well feel that this conclusion is not warranted and that there are interpretive differences between '(a) and '(b) For instance% it could be felt that '(a) is a sentence about -ary and that '(b) is a sentence about Thelma !e return to this point presently As seen above% semantics concerns the relation between a sentence and the world Semantics is also concerned with the relation between propositions For instance% we can say that the proposition expressed by '(a) and that expressed by '(b) have the same truth conditions 3n the other hand% the proposition expressed by '(a) and that expressed by '(c) are contradictory '() c -ary did not invite Thelma There is no discourse world in which sentence '(a) and sentence '(c) can be simultaneously true The proposition expressed by sentence '(d) can also be related to sentences '(a) and '(b): '() d Thelma was invited 4learly '(a) and '(d) are not paraphrases of each other: '(d) may be true while '(a) is false For '(d) to be true it is sufficient that there be a person called Thelma% and that there be a relation of invitin" between some unmentioned person and Thelma% in which Thelma is the 2ATI01T of the invitation '(d) does not re*uire that there be a person called -ary and that -ary is the A/01T In '(a) the A/01T of the invitation is specified as -ary +owever% observe that whenever '(a) is true% '(d) is also true 2roposition '(d) can be inferred from '(a)% or% puttin" it differently% '(a) entails '(d) !e can present entailment as in '(e): '() e 5la675ld6 Since '(a) and '(b) have the same meanin"% '(b) also entails '(d): '() f 5lb6 7 5ld6 3bserve that the reverse does not hold: '(d) does not entail '(a)% for instance ( 8 Structure and interpretation In this chapter we examine the relations between "rammatical structure and interpretation !e will not provide an exhaustive description of the interpretive aspects of the syntactic representations we have elaborated% but we will concentrate on a few of them In this section we introduce some "eneral concepts for meanin" and structure In sections 8 and 9 we turn to some more concrete aspects of the interpretation of sentences Section 8 concerns the interpretation of functional nodes: section 9 concerns the interpretation of operators such as *uantifiers and wh$phrases ( 9 4ompositionality ( 9 ( The interpretation of the sentence is a function of its parts

&an"ua"e is a means of communication: that is% there is a relation between a form in a lan"ua"e and an interpretation% a meanin" 4onsider sentence '8a)% with the structure in '8b): '8) a -ary will invite Thelma

Mary will invite Thelma

Sentence meanin" is compositional: the meanin" of the sentence is determined by its constituents and the relations between them The meanin" of '8a) contains a number of components which we can relate to elements of the structure: '8) c Interpretive elements of '8b) Structure ;: invite
AGENT PATIENT5#2%;

Thelma 5#2%;26 Mary T: will 4 5$!+6

Interpretation there is an action of invitin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 Thelma the A/01T is the referent of the #2 Mary the action is situated in the future the proposition is presented as true

The meanin" of sentence '8a) is determined by the words it contains% but not solely by the words The meanin" of a sentence is also determined by the relations between the words For instance% '8d) contains the same words as '8a) but its meanin" is different: '8) d 542 5c $!+6 5I2Thelma will 5;2 invite -ary666

e Interpretive elements of '8d) Structure Interpretation ;: invite A/01T 2ATI01T 5#2%;6 -ary 5#2%;26 Thelma T: will 4 5$!+6 there is an action of invitin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 -ary the A/01T is the referent of the #2 Thelma the action is situated in the future the proposition is presented as true

'8f) also contains the same words as '8a) but a"ain its meanin" differs: (2) f. Will -ary invite Thelma< In '8f) the movement of will to 4 indicates that the sentence is interro"ative: 4 has the feature 5=!+6 '8) "

2.g

The interpretation of this sentence could be represented as follows: '8) h Interpretive elements of '8") Structure Interpretation ;: invite there is an action of invitin" AGENT involvin" an A/01T PATIENT5#2%; and a 2ATI01T Thelma 5#2%;26 the 2ATI01T is the referent of the #2 Thelma Mary the A/01T is the referent of the #2 Mary

T: will the action is situated in the future 4 5=!+6 the true value of the proposition is not fixed 4han"es in the structure of the sentence thus have an impact on its interpretation 4onsider also '8i): '8) i Thelma will be invited by -ary At first si"ht% '8i) does not differ in interpretation from '8a) A/01Thood is associated partly with the inflectional endin" on the passivi>ed verb invited and partly with the preposition by The 22 by -ary specifies the A/01T of invited '8) ? 0lements of the interpretation of '8i) Structure ;: invited
-ed: AGENT PATIENT5#2%;

Thelma P: by [DP,P'] Mary T: will 4 5$!+6

Interpretation there is an action of invitin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 Thelma the A/01T is specified the A/01T is the referent of the #2 Mary the action is situated in the future the proposition is presented as true

At this point the difference between '8a) and '8i) lies in the way in which the A/01T of the activity is spelt out There is another contrast% thou"h% which we hinted at in the discussion of the active@passive pairs '(a)@ '(b): while '8a) is felt to make a statement about -ary% '8i) is felt to make a statement about Thelma This contrast does not appear in the representation above Alon" the same lines% consider '8k) In '8k) the ob?ect #2 Thelma has been moved into the 42 domain -ore specifically% it is topicali>ed: '8) k Thelma% -ary will invite +ere a"ain% at first si"ht there is no difference in meanin" between '8k) and '8a): certainly the truth conditions of '8a) and '8k) are identical and '8k) can be said to express the followin": '8) ( Interpretive elements of '8k) Structure ;: invite
AGENT PATIENT5#2%;

Thelma [DP,PV] Mary T: will 4 5$!+6

Interpretation there is an action of invitin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 Thelma the A/01T is the referent of the #2 Mary the action is situated in the future the proposition is presented as true

Aet a"ain '8a) differs from '8k) in that '8a) is a sentence about -ary and '8k) is a sentence about Thelma !e return to this notion of ,aboutness, below ( 9 8 Idioms In section ( 9 ( we presented the view that the meanin" of the sentence is calculated on the basis of its parts The normal method of construin" the meanin" of the sentence is compositional: we combine the meanin"s associated with the words% phrases and syntactic confi"urations contained in the sentence But sometimes compositionality is not fully respected In a sentence containin" an idiomatic element% the interpretation of the sentence is not fully determined by the interpretation of its parts 4onsider% for instance% the sentences in '9): '9) a Cohn broke the vase

b Cohn repaired the vase c Cohn broke the ice In '9a) the meanin" of the sentence is determined compositionally: '9) d Interpretive elements of '9a) Structure ;: break
AGENT PATIENT5#2%;

the vase [DP,PV] John T: past 4 5$!+6

Interpretation there is an action of breakin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 the vase the A/01T is the referent of the #2 John the action is situated in the past the proposition is presented as true

In '9b) the event depicted is situated in the past time sphere% as before% and is one of repairin" The event involves the same ar"uments: ,'9) e Interpretive elements of '9b) Structure ;: repair
AGENT PATIENT5#2%;

the vase [DP,PV] John T: past 4 5$!+6

Interpretation there is an action of repairin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 the vase the A/01T is the referent of the #2 John the action is situated in the past the proposition is presented as true

Sentence '9c)% however% has two interpretations: a literal one and an idiomatic one In the literal interpretation the sentence depicts a breakin" event implicatin" Cohn as the A/01T and the ice as the 2ATI01T For instance% Cohn may have broken the ice which he will subse*uently use to keep fish cool '9) f &iteral interpretation of '9c) Structure Interpretation ;: break there is an action of breakin" AGENT involvin" an A/01T PATIENT5#2%; and a 2ATI01T the vase [DP,PV] the 2ATI01T is the referent of the #2 the vase John the A/01T is the referent of the #2 John T: past the action is situated in the past 4 5$!+6 the proposition is presented as true 3n the other hand% the sentence may also have an idiomatic interpretation in which the event described is not one of ,breakin" ice, but of startin" a conversation% succeedin" in communicatin" '9c) mi"ht be appropriate in a context in which Cohn is with some people and he speaks first In the idiomatic interpretation there need not be any ice in the situation depicted: the #2 the ice does not refer to the material which we normally indicate by that term Dather% the ;2 break the ice as a whole means ,start a conversation, The literal interpretation of the strin" broke the ice in '9c) is determined compositionally% but its idiomatic meanin" is not '9) " 'Approximate) Idiomatic readin" of '9c) Structure ;: break the ice AGENT [DP,PV] John T: past 4 5$!+6 Interpretation there is an action of breakin" the ice involvin" an A/01T the A/01T is the referent of the #2 John the action is situated in the past the proposition is presented as true

It is interestin" to consider% for a moment% the effect of the structure of the sentence on the two interpretations of '9c) !e can turn the sentence into a direct yes no !"estion or a wh$*uestion and preserve both readin"s:

'E) a #id Cohn break the ice< b !hen did Cohn break the ice< Similarly% the sub?ect of the sentence may be *uestioned% or relativi>ed% with both readin"s bein" preserved: '5) a !hich student broke the ice first< b Cohn% who first broke the ice% is a sociable fellow The effects on the interpretation of the sentence may be different when we carry out an operation on the clause which affects the ob?ect #2 the ice, the #2 which in the idiomatic readin" of the sentence loses its literal interpretation The effect of wh$movement of the ob?ect is that only the literal interpretation is preserved: 'F) a !hat did Cohn break< b !hich ice did Cohn break< c +ow much ice did Cohn break< Similarly% relative clause formation on the basis of the ob?ect and topicali>ation of the ob?ect result in the loss of the idiomatic readin": 'G) a The ice which Cohn broke 'had come all the way from 1orway)% b The ice% Cohn broke *uickly But it would not be true to say that any operation which affects the ob?ect #2 the ice destroys the idiomatic readin" 2assivi>ation of the sentence leads to a rearran"ement of constituents in which the #2 the ice becomes the "rammatical sub?ect .nder passivi>ation% both readin"s of '9c) are retained: 'G) c The ice was broken by Cohn Similarly% when the idiomatic expression break the ice is the ;2 of a non$finite clause% then the ob?ect #2 the ice can under"o passivi>ation and raise into a hi"her clause: 'G) d The ice seems to have been broken already 'Gd) retains the two interpretations The interpretation of idiomatic expressions is not fully predictable from the interpretation of their components 1or is the effect of syntactic operations on the idiomatic readin"s fully predictable 'Ha) has two interpretations: one a literal% fully compositional% interpretation% in which there is a past act of kickin" involvin" -ary as an A/01T and an ob?ect% the bucket% as a 2ATI01T: and a second% idiomatic% readin" in which the expression kick the b"cket is not interpreted compositionally and means ,die, 'H) a -ary kicked the bucket b Interpretive elements of 'Ha): literal interpretation Structure ;: kick A/01T 2ATI01T 5#2 ;6 the b"cket5#2 ;26 Mary T: past 4 5$wh6 Interpretation there is an action of kickin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 the b"cket the A/01T is the referent of the #2 Mary the action is situated in the past the proposition is presented as true

c Interpretive elements of 'Ha): idiomatic interpretation #tr"ct"re $nterpretation V: kick the b"cket there is an action o% 'kickin& the b"cket' P'T$()T[DP*VP involvin& a P'T$()T ] Mary the P'T$()T is the re%erent o% the DP Mary T: past the action is sit"ated in the past + [,wh] the proposition is presented as tr"e

The idiomatic expression $ or the idiom chunk% as it is sometimes called $ kick the b"cket is more fro>en than the idiom chunk break the ice in '9c) The transformations illustrated by the sentences in 'Hd) all entail loss of idiomaticity 'H) d !hich bucket did -ary kick< !hat did -ary kick< The bucket which -ary kicked The bucket% -ary kicked The bucket was kicked by -ary The bucket seems to have been kicked by -ary 3nly certain *uestions preserve the idiomatic interpretation:
'H) e !hen did -ary kick the bucket< !e conclude that idiom chunks are an exception to compositionality in that the meanin" of the idiom chunk cannot be read off from its parts: there is no obvious reason% for instance% why kick the b"cket should mean ,die, As we have seen% the effect of movement operations on idiom chunks is also not predictable: some idioms retain their idiomatic readin" under passivi>ation [break the ice-, others don,t .kick the b"cket-* ( E (conomy and interpretation ( E ( Full Interpretation The principle of compositionality says that the interpretation of the sentence is based on the interpretation of its components and the relations between them &eavin" aside the interpretation of idiom chunks% compositionality is the principle that re"ulates the relation between form and meanin" Decall that we propose that lin"uistic structures are also "overned by economy considerations For instance% we proposed in chapter 8 that when an element under"oes movement% it always moves to the nearest potential landin" site 4ombinin" the principle of economy with the principle of compositionality leads to a stricter formulation of compositionality .nder this view% it is not only the case that the meanin" of a sentence is determined by its components and by the relation between these components% but also that there must not be any redundant elements: all the components of the sentence must contribute to the interpretation of the sentence If the representation of a sentence contains superfluous elements which do not contribute to its interpretation then the representation is not optimal in terms of economy: there is waste The 2rinciple of Full Interpretation says that every symbol of the "rammar must be semantically interpreted As an illustration of the application of this principle consider 'I): 'I)a -ary will invite Thelma b J-ary will invite Thelma &ouise 'Ia) is the same as '8a) Its 'partial) interpretation is presented as before: 'I) c Interpretive elements of 'Ia) 'K 8c)

Structure ;: invite A/01T 2ATI01T 5#2 ; Thelma5#2 ;26 Mary T: will 4 5$!+6

Interpretation there is an action of invitin" involvin" an A/01T and a 2ATI01T the 2ATI01T is the referent of the #2 Thelma the A/01T is the referent of the #2 Mary the action is situated in the future the proposition is presented as true

!hen we try to assi"n a readin" to 'Ib) we have a problem% however: 'I) d Interpretive elements of 'Ib) Structure ;: invite A/01T 2ATI01T5#2 ; Interpretation there is an action of invitin" involvin" an A/01T and a 2ATI01T

Thelma 5#2 ;26 -ary5#26 &ouise T: will 4 5$!+6

the 2ATI01T is the referent of the #2 Thelma the A/01T is the referent of the #2 -ary < the action is situated in the future the proposition is presented as true

'Ib) is un"rammatical The reason is that it contains a #2% /o"ise, which cannot be mapped onto the interpretation of the sentence /o"ise lacks a thematic role If /o"ise were to be assi"ned the patient role of invite, then Thelma would lack a thematic role: 'I) d, Interpretive elements of 'Ib)
AGENT C-WH] involving an agent the proposition is presented as true

Louise[DP. P the patient is the re!erent o! the DP ] Louise "ar# [DP] the agent is the re!erent o! the DP "ar# patient[DP. and a patient $tru%ture &nterpretation T' (ill Thel)a ' invite the a%tion is situated in the !uture * there is an a%tion o! inviting

A second aspect of the 2rinciple of Full Interpretation is that the components of a sentence must be interpretable in their surface form This aspect of the principle can be appealed to in order to explain the contrast between Italian and 0n"lish as illustrated by '(0): '(0) a [pro] compro un "iornale [pro] buy a newspaper b 0[pro] buy a newspaper In Italian% the sub?ect of a finite clause may remain non$overt: we represent it as pro, a non$overt pronominal ( The non$overt sub?ect pro is not available in 0n"lish !e can say that '(0b) violates the 0mpty 4ate"ory 2rinciple8 which re*uires that every empty cate"ory be identified If we insert a non$overt element pro in the sub?ect position of a finite sentence in 0n"lish% then this element remains unidentified: 0n"lish lacks the rich a"reement which is re*uired to recover the features of the sub?ect The 0mpty 4ate"ory 2rinciple mi"ht be interpreted as an instantiation of the re*uirement that all the symbols of a representation be interpretable If the null sub?ect of 0n"lish '(0b) cannot be identified% then it is uninterpretable: we don,t know what its features are As '(0b) contains an uninterpretable symbol% it will violate the 2rinciple of Full Interpretation ( E 8 Luestions concernin" the application of the 2rinciple of Full Interpretation For most of the components of sentence '((a)% with the structure 'lib)% the application of the 2rinciple of Full Interpretation is by and lar"e unproblematical Cf. chapter 4, sectio 2.!. 1 Chapter 4, sectio 2.4.2.", (#!). '(() a -ary will invite Thelma

11b

!e can map the components of the sentence onto the interpretation: '(() c Interpretive elements of '((a) #tr"ct"re V: invite '1()T P'T$()T [DP,V'] Thelma [DP,VP] Mary T: will + [,23] $nterpretation there is an action o% invitin& involvin& an a&ent and a patient the patient is the re%erent o% the DP Thelma the a&ent is the re%erent o% the DP Mary the action is sit"ated in the %"t"re the proposition is presented as tr"e

+owever% the reader may well wonder what the contribution is of the functional node A/D in the structure !e turn to this *uestion in the next section Another *uestion concerns the role of notions such as case !e proposed that all #2s must have case and we ar"ued that a sentence such as 'lid) is un"rammatical because the #2 Mary lacks case: '(() d JIt was believed 5A/D2 -ary to have invited &ouise6 4ompare 'lid) to 'lie) in which the #2 Mary has moved to the matrix sub?ect position and can be assi"ned nominative case% 'lie) is only a partial representation: for convenience,s sake we omit some of the traces '(() e -aryi is believed 5A/D2 ti to have invited &ouise6 4ase is ar"ued to play a role in the structure of the sentence: case is re*uired for every #2 +ence case is a symbol in the representation of the sentence: in order to check if a sentence is "rammatical% we must check if the #2s have case So the case features of a #2 must be represented in the structure: they are part of the representation If all symbols in the representation of the sentence must be interpreted% then we may well wonder how case plays a role in interpretation In the mappin" between form and interpretation% we have discovered some components which are represented but which do not seem to have a meanin" correlate For '((c) we should substitute '((c,) in which the boldfaced elements have so far not been associated with an interpretation: '(() c, Interpretive elements of '((a) $nterpretation there is an action o% invitin& involvin& an '1()T and a P'T$()T the P'T$()T is the re%erent o% the DP Thelma the a&ent is the re%erent o% the DP Mary 5 5 5 the action is sit"ated in the %"t"re the proposition is presented as tr"e

#tr"ct"re V: invite '1()T P'T$()T[DP,V'] Thelma[DP*VP] Mary +'#(: )4M +'#(: '++ '16: s& T: will + [,23]

!e return to the role of A/D and of case in the interpretation of the structure in the next section Another problem for the 2rinciple of Full Interpretation is raised by the pair of sentences in '(8): '(8) a 5#2 Three more candidates6 arrived at the office b There arrived 5#2 three more candidates6 at the office '(8a) and '(8b) are intuitively felt to have very similar readin"s: however% they differ in their structure '(8b) contains an existential construction% in which the 5Spec A/D26 position is occupied by the expletive there and the #2 three more candidates occupies a lower A$position 9 In our discussion in chapter (% we proposed informally that expletives do not contribute to the meanin" of a sentence If this statement were literally true% then expletives would be superfluous elements: they would not be mapped onto the interpretation The

le"itimacy of an element in the representation of a sentence derives from the fact that it receives an appropriate interpretation By the 2rinciple of Full Interpretation% elements which do not receive an interpretation cause un"rammaticality: such elements should be eliminated from the representation of the sentence The 2rinciple of Full Interpretation thus makes the presence of the expletive there problematic !e will return to this issue below ( E 9 4ase and A/D 7*8*9*7 +ase $.4.".$.$ %I&I'I(IT) As pointed out above% it is not immediately obvious how the case feature of a #2 is mapped onto the interpretation of a clause 1ominative and accusative forms are overtly distin"uished only in the pronominal system 4ase plays a role in determinin" the "rammaticality of a clause 4ase may tri""er movement: A$ movement is case$driven If case plays a role in determinin" the "rammaticality of a structure% case must be available as a symbol in the representation of the structure: and if case is a symbol in the representation of the sentence% it follows by the 2rinciple of Full Interpretation that case contributes to interpretation In order to link case to interpretation we proposed in chapter lE that case is needed to mediate the assi"nment of thematic roles To express the link between thematic roles and case the metaphor of visibility is used: case renders ar"uments visible for the lexical head which assi"ns their thematic roles In '(9a)% for instance% Mary receives nominative case from the finite inflection By virtue of its case% the #2 Mary can be ,seen, by the ; invite: hence invite can assi"n a thematic role to Mary* '(9) a -ary has invited &ouise Similarly% in '(9b) the #2 Mary is the sub?ect of the non$finite clause: its case is determined by the "overnin" verb believe, which can case$mark the specifier of its complement: '(9) b I believe 5A/D2 5#2 -ary6 to have invited &ouise6 '(9c) is un"rammatical The #2 Mary in the embedded clause lacks case: neither the non$finite I .to- nor the passive verb 'believed- can assi"n case to Mary: '(9) c JIt was believed 5A/D2 -ary to have invited &ouise6 4ompare '(9c) to '(9d) in which Mary has moved to the matrix sub?ect position and is assi"ned nominative case by the finite inflection '(9d) is "rammatical: '(9) d -aryi is believed 5A4D2 t: to have invited &ouise6 There are two problems with '(9c) First% the non$finite verb invite is associated with two thematic roles If we assume that case is re*uired to make a #2 visible for the assi"nment of thematic roles% then in '(9c) invite is unable to assi"n the external thematic role 'a"ent) to the #2 Mary* As a result% one thematic role remains unassi"ned This situation leads to un"rammaticality The effect of the absence of case on the thematic structure of the clause is the same as if there were no external ar"ument at all in the structure: '(9) e JIt was believed to have invited &ouise 3n the other hand% if the #2 Mary does not receive case and if case is needed to render the #2 visible% then this #2 remains invisible This means that the #2 Mary could not be associated with a thematic role and that it could thus not be related to the lexical head invite* The #2 Mary would remain a symbol which plays no part in the interpretation So% if Mary lacks case% '(9c) violates the 2rinciple of Full Interpretation By proposin" that case renders #2s visible we also provide a more principled explanation for the 4ase$Filter:5 #2s need case in order to be visible ;isibility "uarantees interpretability: it allows #2s to function as ar"uments of lexical heads and to be assi"ned thematic roles &inkin" case with visibility for theta$role assi"nment entails that the 4ase$Filter as such does not have to be stated as an independent principle of "rammar The 4ase$Filter derives from the con?unction of 'i) the 2rinciple of Full Interpretation 'itself a reflex of 0conomy)% 'ii) the theta$criterion% the re*uirement that each ar"ument receive a thematic role and that each thematic role associated with a head be assi"ned to an ar"ument% 'iii) the hypothesis that case renders an 12 visible% and 'iv) the hypothesis that visibility is re*uired for theta$role assi"nment The proposal that case is re*uired to "uarantee visibility of the ar"ument leads to important repercussions in the theory we have been proposin" !e will not pursue all these ramifications here Three points will be considered: the conse*uences of the visibility condition for the status of the non$overt sub?ect of non$finite structures% 2D3%F for the relation between 12$movement and visibility% and for the relation between expletives% the 2rinciple of Full Interpretation and case -any of the topics developed below depend to a lar"e extent on theoretical decisions Alternative proposals may also be descriptively ade*uate Dather than aimin" at "ivin" a definitive account% our discussion serves to illustrate the kind of ar"umentation which may be used

$.4.".$.2 P*+ AN, CA&E (ii) Decall that we proposed that non$finite clauses in 0n"lish may have non$overt sub?ects G The non$overt sub?ect of the infinitival clause is represented as 2D3 In '(Ea) the infinitival sub?ect 2D3 is controlled by the matrix sub?ect $; in '(Eb) 2D3 is assi"ned an arbitrary interpretation '(E) a Ii wonder 542 whether 5A/D2 2D3i 5T2 t, to 5;2 ti buy a newspaper6666 b 542 !hether 5A/D2 2D3arb to 5T2 tarb 5;2 tarb buy a newspaper or not6666 is a bi" issue DecallH that unlike the empty cate"ory in the sub?ect position of a finite clause in Italian% which we represented as pro, the non$overt sub?ect of a non$finite clause does not alternate with an overt #2 An overt sub?ect cannot be inserted in the position which we take to be occupied by 2D3: '(5) a Thelma wondered 542 whether 5A/D2 P*+ (Jshe@ Jher) to trust Bill66 b 542 5A/D2 2D3 'Mme@ MCohn) to buy these books now66 is important c The *uestion was 542 whether 5A/D2 P*+ (Jwe @MCohn) to buy these books or not66 d It is important 542 5A/D2 P*+ (Jus @MCohn) to buy these books now66 4onversely% if an overt sub?ect is possible in an infinitival clause% then 2D3 is excluded: '(F) a 542 For 5A/D2 Cohn to buy this book66 would be surprisin"% b J542 For 5A/D2 P*+ to buy this book66 would be surprisin" In the discussion in chapter E we related the un"rammaticality of the overt sub?ect of the infinitive in '(5) to case theory In all of the sentences in '(5)% the sub?ect of the infinitive cannot be overt because it cannot be assi"ned case There is no case$assi"ner available I 2ursuin" our discussion we can now say that these examples violate Full Interpretation If the overt #2 is caseless it will not be able to receive a thematic role and hence will not contribute to the interpretation of the sentence In '(Fa)% on the other hand% an overt sub?ect is allowed: %or assi"ns accusative case to the sub?ect 1ow 2D3 is excluded '(Fb) In our discussion we tentatively concluded that the non$overt #2 2D3 occurs in a position which lacks case 2D3 and overt #2s are in complementary distribution !e offered two ways of interpretin" this situation 3ne approach was to say that 2D3 must not be assi"ned case% while overt 12s have to be assi"ned case Another proposal was to say that 2D3 is not really caseless% but rather that it is compatible with a minimal or >ero case assi"ned by a non$finite inflection The minimal case is insufficient to "uarantee the visibility of overt #2s but it is sufficient for the non$overt cate"ory 2D3 In the li"ht of our more recent discussion of the visibility re*uirement for #2s we are led to choose between the two options sketched above 2D3 is an ar"ument and receives a thematic role 4onsider for instance '(G): '(G) 52D3 to have invited &ouise6 was a mistake In '(G) 2D3 is the external ar"ument of invite* If ar"uments can receive a thematic role only if they are visible% and if visibility is a function of case% then we must conclude that 2D3 must be visible 2ursuin" this option we say that 2D3 is case$marked by the infinitival inflection to* !e then have to assume that to can assi"n a minimal case to 2D3% but that this case is inade*uate for overt #2s If we assume that 2D3 is assi"ned 'minimal) case in the specifier position of the infinitival A/D2% then we also have a motivation for the movement of 2D3 from its base thematic position to 5Spec%A/D26 2D3 cannot remain in its base thematic position% 5Spec%;26% because% by assumption% 5Spec ;26 is not a case position $.4.".$." %I&I'I(IT) AN, -+%E-ENT !e have discussed in some detail a movement operation referred to as 12$movement% in which an 12 'now #2) moves to an A$position !e will keep usin" the label 12$movement: it refers to the movement of the sub?ect #2 from the specifier of the lexical pro?ection 'for instance% 5Spec ;26) to the canonical sub?ect position% the movement of an ob?ect #2 to the canonical sub?ect position in passive sentences% and the raisin" of the sub?ect #2 of a non$finite clause or a verbless clause to a hi"her sub?ect position 12$movement is case$driven: that is% the #2 moves to attain a position in which case is assi"ned In '(H) we provide some illustrations of 12$movement In '(Ha) the sub?ect #2 Mary moves from its base$position% the specifier of ;2% to the canonical sub?ect position In '(Hb) the ob?ect #2 /o"ise moves from the ob?ect position to the sub?ect position: it moves via the intermediate specifier positions% 'i) the specifier position of the ;2 headed by invited, 'ii) the specifier position of the I2 associated with the passive morpholo"y% and 'iii)

the specifier position of the ;2 headed by be* In '(Hc) the sub?ect of the non$finite clause Mary moves to the matrix sub?ect position: a"ain it transits throu"h the intermediate specifier positions '(H)a 5A/D2 -aryi 5T2 ti 5;2 t, invited &ouise666 b 5A/D2 &ouisei was; 5T2 ti 5;2 ti ty 5I2 ti 5;2 ti invited ti 66666 c 5A/D2 -aryi 5T2 ti 5;2 ti seems 5A/D2 ti 5T2 ti to 5;2 ti like &ouise666666 !e will now briefly reconsider 12$movement in the li"ht of the discussion of visibility !e will base our discussion on '(Ha)% but it carries over to the other examples in '(H) In '(Ha) the verb invite takes two ar"uments% Mary and /o"ise, each of which is assi"ned a thematic role:

19

The external ar"ument of a verb ori"inates in the specifier position of the ;2 0ven thou"h the #2 Mary occupies its thematic position in the #$structure in '(I) it cannot be assi"ned a thematic role In the specifier position of the ;2% the #2 Mary is not assi"ned case: hence it is invisible for theta$role assi"nment The #2 Mary moves to the canonical sub?ect position in order to receive case !e co$index the moved #2 and its trace: '80) a 5A/D2 -aryi [T2 ti 5;2 ti invited &ouise666 In '80a) we have associated the #2 Mary in the specifier of A/D2 with its trace in the specifier of ;2 by means of co$indexation The two positions are related: Mary and its trace are linked The moved #2 and its trace form a chain (0 The hi"hest position of the chain receives case 'here nominative) &et us say that when one position of a chain receives case then the chain becomes visible If a chain is visible 'by virtue of case) then all the positions of the chain are visible Thus in '80a) not only is the #2 Mary visible but so is its trace in the specifier of ;2% and the verb can assi"n a thematic role to the specifier position of ;2 The chain in '80b) contains one case position and one thematic position:
+,-.

N/"&NAT& E

b 5A/D2 -ary%
AGENT
[TP ti [ P ti invited Louise]]].

$.4.".$.4 E.P(ETI%E& AN, CA&E In section ( E 8 we mentioned the problem raised by expletives for the 2rinciple of Full Interpretation In this section we see that there are additional problems raised by the fact that expletives are assi"ned case !e have related the case re*uirement of a #2 to its thematic role: case is needed to make a #2 visible for thematic role assi"nment If we pursue this option it mi"ht appear at first si"ht that expletives need not be assi"ned case as they are not assi"ned a thematic role: '8() a There arrived three more students at the party b I expect there to arrive three more students at the party In the examples in '8() the verb arrive is associated with an expletive sub?ect% there* In both examples the expletive occupies a case position: in '8(a) it occupies the canonical sub?ect position associated with nominative case% and in '8(b) it receives case from the verb e:pect which assi"ns case to the specifier of its complement That the expletive there occupies a case position in '8() is clearly shown by the fact that an overt #2 may occupy the same position: '88) a 5A4D2 The students arrived at the party6 b I expect 5A/D2 the students to arrive at the party6 If the expletive in '8() has case then it is visible for theta$role assi"nment And yet% by definition% expletives do not receive thematic roles As we saw in section ( E 8% expletives pose a problem: they are symbols in the representation while apparently not contributin" anythin" to the interpretation of the structure -oreover% we

now conclude that expletives receive case% a feature typically related to visibility% hence to interpretation In fact% expletives cannot appear in a position to which case is not assi"ned 4onsider '89a): '89) a 542 For 5A/D2 there to arrive too many students at the party66 is a problem In '89a) there occupies a case position: the complementi>er %or can case$mark the sub?ect of its complement clause If we remove %or from the structure% the sentence becomes un"rammatical: '89) b J542 5A/D2 There to arrive too many students at the party66 is a problem The data in '89) su""est that not only can the expletive there occur in a case position% in fact it -.ST occupy a case position 0xpletives re*uire case The conclusion that expletives re*uire case is paradoxical in the li"ht of the previous discussion: expletives are elements that do not receive a thematic role% while case is needed for a #2 to be visible% i e to be able to receive a thematic role The answer to the paradox is already present in our discussion in chapter 8 !hen dealin" there with existential constructions we raised the *uestion of how the post$verbal indefinite sub?ect in such constructions can satisfy the 4ase$Filter !e su""ested that the post$verbal indefinite sub?ect is co$indexed with the expletive .there- in the canonical sub?ect position and that it is the chain expletive$post$verbal sub?ect% i e there,too many st"dents in our example% which is assi"ned case The expletive is assi"ned case by I1F& and the case on the expletive licenses the post$verbal sub?ect #2 in its chain 3bserve that we now have postulated two types of chains: chains created by movement and chains created by co$indexin" an expletive with an ar"ument N 7*8*9*< '16 In the discussion in section ( E 8 we hi"hli"hted a problem for Full Interpretation: the *uestion arose as to the role of the symbol A/D in the interpretation of the sentence 4onsider the followin" examples: '8E) a The "irl reads the paper b The "irls read the paper !e propose that the sub?ect #2 occupies the specifier position of the hi"hest pro?ection of the I2 domain% A/D2 The sub?ect #2 enters into an a"reement relation with A/D '8Ec) is un"rammatical in the present tense because the number feature of the sub?ect does not match that of A/D: '8E) c The "irl read the paper The contrast between a sin"ular #2 and a plural #2 has an impact on the interpretation of a sentence: in '8Ea) the sin"ular sub?ect refers to one person in the universe of discourse% whereas in '8Eb) the sub?ect is plural% i e it picks out more than one referent Thus the distinction between sin"ular and plural #2s contributes to interpretation In our dia"rams in which we map form and meanin" we would have to include the mappin" of the nominal a"reement features sin"ular and plural of the #2 to the concept of number &et us propose that the functional node A/D associated with the verb is re*uired in the structure to license the a"reement features of the sub?ect #2 The specifier$head relation between A/D and the sub?ect makes the a"reement features of the #2 visible for interpretation There is a contrast between the function of nominal A/D features and inflectional A/D 1ominal a"reement contributes directly to interpretation: a sin"ular #2 is different in interpretation from a plural #2 ;erbal A/D formally licenses the nominal A/D features If we say that the contribution of the a"reement head is to make the a"reement features of the sub?ect visible for interpretation% then we have answered the *uestion as to the role of A/D in the representation But we have created a further problem If the number feature on the #2 must be made visible for interpretation then we would also expect that the number feature of the ob?ect #2 in '8E) must be visible The ob?ect #2 may be sin"ular% as in '8Ea$c)% but it may also be plural: '8E) d The "irl reads the papers +ow can we make these number features visible< A complete answer to this *uestion would raise a host of theoretical issues which "o well beyond the scope of this introduction to 0n"lish syntax: but let us "ive the reader some indications of the way to think about this problem If the a"reement features of the sub?ect #2 have to be licensed in a specifier$head relation with A/D% then ceteris parib"s we would expect that the a"reement features of the ob?ect #2 also have to be checked in a relation with an A/D head So we would have to propose that not only does a sentence contain an a"reement pro?ection for a sub?ect% which we can label A/Do2% but it also contains an a"reement pro?ection for an ob?ect% A/Do2:

25

In 0n"lish% A/Do is abstract: it does not have an overt reflex The form of the verb does not vary dependin" on the number of the ob?ect So far% we have proposed that the inflectional properties of A/Ds and T are ultimately associated with the verb .reads- by lowerin" onto ; By analo"y% we propose that A/Do is also lowered onto ; As a result of the association of ; and A/Do% the head A/Do will be able to "overn the ob?ect #2 the papers and check its number features In other lan"ua"es we find an overt reflex of the a"reement of the verb and its ob?ect 4onsider% for instance% the French examples in '8F): '8F) a Luel texte as$tu lu< which text have$you read$masc$s" b Luels textes as$tu lus< which texts have$you read$masc$pl c Luelle revue as$tu lue< which ?ournal have$you read$fem$s" d Luelles revues as$tu lues< which ?ournals have$you read$fem$pl The variation in the form of the participle l".e- ,s is determined by the number of the ob?ect Thus% ob?ect a"reement is checked by the a"reement features of the participle !e could say that dominatin" the pro?ection of the participle there is an a"reement pro?ection which can check the a"reement of the ob?ect 1ote that in French% the participle a"rees with the ob?ect in "ender as well as in number: te:te ',text,) is masculine% rev"e ',?ournal,) is feminine ( E E 0xpletive constructions !e have seen(8 that expletives raise a number of problems for the 2rinciple of Full Interpretation &et us return to the examples with expletive constructions in '(8) discussed in section ( E 8% repeated here as '8G): '8G) a 5#2 Three more candidates6 arrived at the office b There arrived 5#2 three more candidates6 at the office The *uestion we raised in the discussion was how to deal with the presence of the apparently meanin"less expletive there in these sentences% in the li"ht of the 2rinciple of Full Interpretation The paired examples in '8G) show that it cannot be the case that there receives a thematic role: in '8Ga) the verb arrive has apparently one ar"ument% three more candidates* !e assume that arrive is a one$ar"ument verb In '8Gb) the same verb is used with the same interpretation% so we maintain the ar"ument structure !hat then is the function of therel* Another *uestion which arises is why the expletive has to occupy a case position (9 A partial answer to the *uestions raised here can be "iven in the li"ht of the discussion in section ( E 9 In '8Ga) the #2 three more candidates is assi"ned a thematic role by arrive* In order to receive a thematic role it must be visible: in order to be visible% the #2 must have case The sub?ect of the finite clause receives nominative case Similarly% the #2 three more candidates is plural The plural morpheme is meanin"ful and must be licensed in the structure: this licensin" is done by the A/Ds node In '8Gb) the #2 three more candidates also has a thematic role and it too has a"reement features So the #2 must have case 'in order to be visible and to receive the thematic role) and it must have its a"reement features checked In '8Gb) expletive there occupies the canonical sub?ect position of the finite clause% i e the position in which nominative case is assi"ned The canonical sub?ect position is the specifier position of A/Ds Followin" the traditional analysis of expletive sub?ects we say that there is a place$holder for the sub?ect !e co$index there and the post$verbal sub?ect three more candidates; by co$indexation we establish a chain between the position of there and the position of the post$verbal sub?ect (E The chain composed of there and the post$verbal #2 contains a case position% 5Spec%A/D26% and it contains the thematic position of the sub?ect% the post$verbal position itself The post$verbal sub?ect #2 becomes visible by virtue of bein" in a chain with there, which occupies the canonical sub?ect position

Anda mungkin juga menyukai