Popular in Course
Popular in Linguistics
This 56 page Class Notes was uploaded by Americo Abshire on Friday October 30, 2015. The Class Notes belongs to LINGUIST 601 at University of Massachusetts taught by Rajesh Bhatt in Fall. Since its upload, it has received 17 views. For similar materials see /class/232234/linguist-601-university-of-massachusetts in Linguistics at University of Massachusetts.
Reviews for Intr
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 10/30/15
Introduction to Transformational Grammar LINGUIST 601 October 16 2008 EPP versus Case 1 What does Case do 0 Determine the morphological shape of argument DPs morphological case 0 Regulate the distribution of overt DPs abstract case Three Stages in the use of Case 1 Stage 1 Lecture on Government and Binding onwards The Extended Case Filter Np a if a has no Case and a contains a phonetic matrix or is a variable Chomsky l981l75 a Overt DPs are defective They need to be case licensed b Particular heads do caselicensing in particular configurations c Case is the motivation for movement If a DP needs case it moves to a location where it can get case If such a movement is illegal or there is no such location the structure is ungrammatical 2 Stage 2 Minimalist Inquiries Chomsky 1998 Chomsky 1999 a Overt DPs are defective They need to be case licensed b Particular heads do caselicensing via Agree c Case is not the motivation for movement though in principle it could be d Movement is motivated by independent EPP features Implicit assumption made in Stages 1 and 2 even though the abstract case feature does not need to have an overt realization if there is overt realization of case it is determined by the abstract case feature Despite their differences Stage 1 and Stage 2 require that DPs need to be caselicensed If the DP is not caselicensed the structure containing it crashes Classic cases explained by Case Licensing 3 Distribution of overt subjects in infinitival clauses case is taken to not be generally avail able in the SpecTP of a nonfinite clause a It is unfortunate John to be sick It is unfortunate for John to be sick b John to be sick is unfortunate For John to be sick is unfortunate c John tried Bill to leave John wants Bill to leaveJohn believes Mary to be innocent d John believes Mary to be innocent Mary is believed ti to be innocent It is believed Mary to be innocent The idea is that PRO the null pronoun that often appears in the subjects of infinitival clauses not being an overt DP does not require case A further assumption that is needed is that PRO cannot be governed Together these assumptions derive to a significant extent the distribution of PRO 4 Complements of AN a John is fond of olives b Kate is the queen of Bethesda 5 Location of Overt DPs a My computer was stolen ti b Heather appeared t in Bill s dream c Johni seems that t is sick 2 Stage 3 Case is not in the Syntax 21 Abstract Case is not enough Abstract Case does not directly determine Morphological Case Marantz 1991 McFadden 2004 Bobaljik 2005 It is not clear that given syntactic structure we need an abstract mediating fea ture but see Legate 2008 who argues for the continued relevance of abstract case for the realiza tion of morphological case 22 Getting rid of DP Licensing via Case Abstract Case does not regulate the licensing or location of Overt DPs McFadden 2004 6 a Issues of location are determined by EPP considerations b Case Licensing reduces to unrelated properties regulating i the distribution of overt and covert complementizers ii conditions on the associate of it it needs a CP associate 7 10 Distribution of Complementizers a b c d lwould like for him to buy the book lbelieve that he bought the book For him to buy the book would be preferable That he bought the book was unexpected idea 7c needsfor not for case reasons but for whatever reason 7d requires that seem only takes a nonfinite TP complement a F7 P It seems for John to be sick A subcategorization of seem is not satisfied control infinitivals have to be CPs It seems PRO to be on edge lately A subcategorization of seem is not satisfied It seems John to be sick A subcategorization of seem is satisfied but it cannot have a TP associate Unlike believedlikely the clausal complement of seem cannot appear in subject position likely can take both nonfinite CPs and TPs 31 F7 P P D It is likely for John to win A it has CP associate It is likely John to win A it needs CP associate has TP associate For John to win is likely A complement of likely can raise to subject position John to win is likely A complement of likely can raise to subject position but then overt C is needed John is likely Tp t to win A TP can t raise Hence its subject is available for raising What blocks quot It is likely PRO to win unfortunate takes only CPs E F7 P P D It is unfortunate for John to have to leave A it has CP associate It is unfortunate John to leave so early A it needs CP associate has TP associate Question what rules out a null C0 CP structure as in It is unfortunate that John left so early It is unfortunate PRO to leave so early A it has CP associate control infinitivals have to be CPs For John to have to leave is unfortunate A complement of unfortunate can raise to subject position John to win is unfortunate A complement of unfortunate can raise to subject position but then overt C is needed f PRO to have to leave so early is unfortunate A control complement of unfortunate can raise to subject position g There is unfortunate to be a party tonight A unfortunate can only take infinitival CP complements 11 believe takes infinitival TPs and not infinitival CPs a Gina believes Ron to be innocent A believe has TP complement b Gina believes for Ron to be innocent A believe has CP complement c Roni is believed ti to be innocent A believed has TP complement d It is believed for Ron to be innocent A if complement is TP it has wrong associate A if complement is CP believed has CP complement e It is believed PRO to be innocent A PRO needs CP believed cannot handle a CP 23 The Essential Contrasts 12 it can never have an infinitival associate with an overt subject without afor a It seems for John to win b It is likely for John to win1 c It is unfortunate for John to have to leave so early d It is believed for John to win H 5quot possibility offorless infinitival with overt subject depends upon embedding predicate a Jeremy believes quotfor Mary to be innocentquotJeremy believes PRO to be innocent b Jeremy wants quotfor Mary to winJeremy wants PRO to win c Jeremy tried for Mary to winJeremy tried PRO to win 14 it can have a PROsubject infinitival associate depending upon the matrix predicate a It seems PRO to win b It is likely PRO to win c It is unfortunate PRO to have to leave so early d It is believed PRO to win A explanation probably lies in the domain of implicit arguments and control 1A more natural example perhaps If no curs are present it is twice as likely for there to be no curs ut the next time instant us it is for there to be one cur 24 Challenges for Case Theory Mysterious Licensors 15 a John remembered for Frank buying the beer b For Frank buying the beer was unexpected c For Frank being too sick to move John had to buy the beer Default Case 16 Me and John are going to the store PseudoPassives 17 This bed has been slept in ti References Bobaljik J 2005 Where s 15 Agreement as a postsyntactic operation in M van Koppen ed Leiden Papers in Linguistics Vol XX Leiden University Leiden Chomsky N 1981 Lectures on Government and Binding Foris Dordrecht Chomsky N 1998 Minimalist Inquiries The Framework MIT Occasional Papers in Linguistics 15 MITWPL Cambridge MA Chomsky N 1999 Derivation by Phase MIT Occasional Papers in Linguistics 18 MITWPL Cambridge MA Legate J A 2008 Morphological and Abstract Case Linguistic Inquiry 391 557101 Marantz A 1991 Case and Licensing in G Westphal B A0 and HR Chae eds Proceedings ofESCOL 91 Cornell University Ithaca NY Cornell Linguistics Club 2347253 McFadden T 2004 The position of morphological case in the derivation a study on the syntax morphology interface Doctoral dissertation University of Pennsylvania Philadelphia PA Introduction to Transformational Grammar LINGUIST 601 November 19 2004 Binding Binding Theory determines the interpretation and distribution of pronouns and anaphors It is formulated in terms of three principles Condition A which applies to anaphors Condition B which applies to pronouns and Condition C which applies to name and other referential expressions Reexpressions We have already discussed Condition C 1 Condition C a pronoun cannot refer to an Reexpression that it cecommands We now focus on Conditions A and B of the binding theory 1 Condition A Condition A governs the distribution and interpretation of anaphors Anaphors are dependent nominal elements which must have a sentenceeinternal antecedent Unlike pronouns they can not generally refer to a sentenceeexternal contextual element Most languages have two kinds of anaphoric elements 2 a reflexives himself herself themselves myself ourselves yourself b reciprocals each other The discussion here will focus largely on re exives Most of what we will propose for re exives will also apply to reciprocals but reciprocals introduce additional complexity which we will not get into here 11 Properties of Anaphors o Anaphors unlike pronouns must have an antecedent within the sentence 3 a Himselfarrived b He arrived The ungrammaticality of 3a can be plausibly attributed to an inability of himselfto appear in a nominative position 4 does not run into this problem 4 a For himself to leave now would be good b For him to leave now would be good 0 Anaphors must have featureecompatible antecedents 5 a Stephini likes himselfi b Claudiai likes himselfi 6 a Susani believes himselfi to be a genius b Susan believes him to be a genius Pronouns do not require a sentenceiintemal antecedent However if they do have a sentence internal antecedent then it must be featureicompatible with the pronoun 7 a Vladislavi thinks that hem is a genius b Vladislavi thinks that shewMA is a genius o The antecedent of the anaphor must cicommand the anaphor 8 a Stephin si mother likes himselfi b Stephin si mother likes himi c That Stephini is always ignored irritates himselfi 1 That Stephini is always ignored irritates himi Pronouns do not have a cicommand requirement The antecedents of the pronouns in 8b 1 do not cicommand them 0 The antecedent of the anaphor cannot be too far from the anaphor 9 a Stephini thinks that Claudia likes himselfi b Claudia thinks that Stephini likes himselfi The antecedent ofa pronoun cannot be too close to the anaphor 10 a Stephini thinks that Claudia likes himi b Claudia thinks that Stephini likes himi 12 Binding Domains The intuition is that anaphors must have a binder that is close enough while pronouns may not have a binder that is too close ll NP1 binds another NP iff NP1 cicommands NP and NP1 and NP are coiindexed An NP is bound iff there is an NP st NP binds NP 12 a Condition A An anaphor must be locally bound b Condition B A pronoun must not be locally bound c Condition C An Riexpression can not be bound What does locally bound mean In particular what constitutes local local in the context of the binding theory binding domain Thus the binding conditions can be restated as 13 a Condition A An anaphor must be bound in its binding domain b Condition B A pronoun must not be bound in its binding domain c Condition C An Riexpression can not be bound 14 Binding Domain Attempt l the binding domain of an NP is the smallest TP that contains it 13 Problems with 14 14 goes quite far in capturing the examples we have seen so far In fact it explains every single example discussed in this handout up until this point It can also explain cases of ambiguity like the following 15 They pointed the guns at each other Despite this it is ultimately inadequate ECM provides one environment where 14 makes incorrect predictions 16 a Vladislavi believes Tphimselfi to be a genius b Vladislavi believes Tphlmg to be a genius By 14 the binding domain ofthe subject of the embedded TP is the embedded TP Therefore we incorrectly predict that 16a should be ungrammatical and that 16b should be grammatical One might think 17 to also be a counterexample 17 a Johni tried Tp to kill himselfi b Johni tried Tp to kill himi Given our assumptions it isn t a counterexample Why We might have the intuition that what is going wrong in 16 is that we are dealing with a non nite TP and that therefore we should reformulate 14 as follows 18 Binding Domain Attempt 2 the binding domain of an NP is the smallest nite TP that contains it However ECM proves to be a problem once again 19 a Ristoi considers Liina to be fond of himselfi b Ristoi considers Liina to be fond of himi By 18 the binding domain of himselfhim in 19 is the entire sentence 7 the entire sentence is the minimal nite clause that contains the anaphorpronoun Hence 19a is incorrectly predicted to be good and 19b to be bad 0 Reverse engineering points out that we need to distinguish between the subject of an ECM in nitive and the object of an ECM in nitive The binding domain of the subject of the ECM in nitive seems to be larger than that of the object of the ECM in nitive 14 Reformulating Binding Domains 0 Binding domain of the subject of the ECM in nitive includes the clause of the ECM verb 0 Binding domain of the object of the ECM in nitive includes only the ECM in nitive 20 Binding Domain Attempt 3 the binding domain of an NP is the smallest clause that contains i the NP ii its caseimarker and iii a higher subject 2 The Role of Subjects 20 our last de nition of binding domain makes reference to the notion subject By subject we mean an NP in the SpecTP The new de nition allows us to handle properly the examples in 21 and 22 21 Subjects Sebastieni considers Tphimselfi to be intelligent F7 E Sebastieni considers Tphlmg to be intelligent P Sebastieni believes that Tp himselfi is intelligent fl Sebastieni believes that Tp hei is intelligent 22 Objects Sebastieni considers TpMiguel to be fond of himselfi F7 P Sebastien considers TpMigueli to be fond of himselfi P Sebastieni believes that TpMiguel is fond of himselfi fl Sebastien believes that TpMigueli is fond of himselfi 21 Subjects Across Categories Above we de ned subject as an NP in SpecTP This assumption requires revision We already assume that subjects do not originate in SpecTP They are merged in lower 297 positions from where they raise to SpecTP to get case 23 a Roland might VpRoland visit Klaus b Roland might be Ap Roland nice c Roland might be pp Roland in Amsterdam 1 Roland might be Np Roland a star lfwe supply case to the various 297positions then overt subjects can appear there 24 a lwatched Vp Roland visit Klaus b l ndconsider Ap Roland nice c lwant pp Roland in Amsterdam 1 I consider Np Roland a star So we will extend subject to mean an NP that occupies the SpecXP position of any XP 22 Binding Domains don t have to be TPs The notion of binding domain made reference to the smallest clause that contains the pronounanaphor Now that we are talking about subjects across categories we will need to replace smallest clause by smallest XP The evidence 25 VP a Keni watched Vp Rolandj hit himselfjMA b Keni watched Vp Rolandj hit himicj 26 AP a Keni considers AP Rolandj fond of himselfjMA b Keni considers AP Rolandj fond of himicj 27 Binding Domain Attempt 4 the binding domain ofan NP is the smallest XP that contains i the NP ii its caseimarker and iii a subject 23 Optional Subjects NPs allow for optionality with respect to subjects ie some NPs but not all have subjects So NPs constitute a binding domain or not depending upon the presence or absence of a subject Our theory makes correct predictions with regard to the cases in 28 28 a Marci will believe any positive description of himselfi b Marci will believe Annie s description of himselfi c Marci will believe Annie s description of himW 1 Annie will believe Marc si description ofhimselficj e Annie will believe Marc si description ofhiij What are the judgements about 29 Marci will believe any positive description of himi Our theory so far correctly predicts that 28a is grammatical For the same reason it should also predict that 29 is ungrammatical This is indeed the judgement reported in the literature However there seems to be a dialect split here For some speakers 29 is ungrammatical while for others it is ok Other examples are also found of environments where there is no complementarity between prof nouns and anaphors These are for the most part cases where the anaphorpronoun appears inside an NF 30 a Artemisi lost a beautiful picture of herselfi that I had given her b Artemisi lost a beautiful picture of heri that I had given her 31 a They heard stories about each otherithemselvesi b Theyi heard stories about themi Complementarity is not lost with all NPs 32 a Jacobi took a picture of himselfihimi b Jacobi saw a picture of himselfihimi Speakers pretty 39 nd 1 in in take ar 39 but there seems to be optionality for at least some speakers otherwise The generalization seems to be that complei mentarity holds in semiiidiomatic environmenst like take a picture or tell a story but not generally1 In 3032 the relevant NP is in object position Nonicomplementary distribution also emerges when the NP is in subject position 5 33 Jonahi thinks that Tp a beautiful picture of himselfi is hanging on the outside wall of the gym 0 Jonahi thinks that Tp a beautiful picture of himi is hanging on the outside wall of the gym Various scholars have taken cases where the complementarity between anaphors and pronouns breaks down to be instantiating a different module of grammar They propose that anaphors that appear in these environments are Iogophors which they argue have distinct properties see Reinhart and Reuland 1993 Classical binding theory see Chomsky 1981 Chomsky 1986 does not make distinctions be tween anaphors in NPs and anaphors elsewhere The following extensions were proposed to handle the special issues raised by anaphors in NPs 3 Binding Theory Extensions 31 Accessible Subjects Contrast 33 which is repeated here as 34 with 35 34 E Jonahi thinks that Tp a beautiful picture of himselfi is hanging on the outside wall of the gym 0 Jonahi thinks that Tp a beautiful picture of himi is hanging on the outside wall of the gym 35 E Jonahi thinks that Tp himselfi is intelligent 0 Jonahi thinks that Tp hei is intelligent We will focus on the contrast between 34a and 35a 36 Binding Domain Attempt 4 the binding domain ofan NP is the smallest XP that contains i the NP ii its caseimarker and iii a subject 20 The binding domain for himselfin both 34a and 35a is the embedded TP Hence we predict that both should be ungrammatical This is correct for both 35a but incorrect for 34a Now consider how the binding domain for himselfis determined for 34a and 35a 37 a For himselfin 34a caseemarker of himself of subject a beautiful picture of himself 1Implicit Arguments have been argued to play a role here See Williams 1985 b For himselfin 35a caseemarker of himself 10 subject himself The problem intuitively is that the anaphor is either contained in the subject or is the subject Such a subject is not accessible to the anaphor This suggests the following revision of the binding theory 38 Binding Domain Attempt 5 the binding domain of an NP is the smallest XP that contains i the NP ii its caseemarker and iii an accessible subject But don t we predict that 35a should be good We do But this is not a problem given that we have another way of ruling out 35a himselfis accusative but appears in a nominative position To properly test the predictions made by 38 we would need a language which has nominative anaphors o himselfin accusative subject positions does ne but this is not surprising 39 a Matti would like for himselfi to win b Matti considers himselfi to be competent 0 We also need to say that each other cannot appear in nominative positions 40 a Elena and Artemish know that each otheri isare wonderful b Elena and Artemish know that theyi are wonderful The binding domain of each other in 40a is the entire clause and so it has to be blocked by some thing other than binding theory 32 Binding Theory Compatibility We are still left with the breakdown of complementarity between the distribution of anaphors and pronouns The problem of complementarity also arises with possessive pronouns 41 a Maya likes heri husband b Mayai thinks that ldan dislikes heri husband c ldan and Doriti like each other si booksZ her as a possessive pronoun can be bound locally and also nonelocally By the de nition of binding domain sketched so far in 4la the binding domain of the pronoun should be the whole TP This incorrectly predicts that 4la should be ungrammatical due to a violation of Condn C One possibility is tojust say that the possessive her is systematically ambiguous between a pro noun her and an anaphor her own Thus 4la would involve the anaphor and 41b would involve the pronoun This line of reasoning receives some support from the fact that English doesn t seem to have a simplex possessive re exive pronoun 42 a Thomasi saw his sister 2What are thejudgements concerning ldan and Don39t think that Maya likes each other s books b Thomas saw himself si sister c Thomas saw self si sister Other languages do not have this lexical gap and there we nd what our binding theory predicts 43 Hindi a Thomasine apnii behiniko dekhiaa ThomasiErg selff s sisteriAcc seeinv Thomasi saw hisim sister Thomasine usikii behiniko dekhiaa ThomasiErg heiGenf sisteriAcc seeva 0 Thomasi saw hisiMA sister This line of reasoning may be correct but it does not account for the lack of complementarity in anaphorpronoun distribution that we found with anaphorspronouns inside NPs These were cases where both a pronoun and an anaphor can felicitously appear 44 a Bruce thinks that Tp a beautiful picture of himselfhimi is hanging on the outside wall of the gym b Theyi heard stories about themthemselvesi lfwe reverse engineer from 44 and the basic binding conditions as we know them we are forced to the conclusion that contrary to our assumption anaphors and pronouns are not required to have the same binding domain In both 44a b the binding domain ofthe anaphor would include the matrix subject but the binding domain of the pronoun would not ie the binding domain of the anaphor would be bigger than the binding domain of the pronoun Why would anaphors and pronouns have different binding domains Further why would the binding domain of the anaphor be larger than the binding domain of the pronoun The following answer suggests itself anaphors need to be locally bound while pronouns need to be locally free 0 So we want to give the anaphor a binding domain where it in principle has a chance to get bound ie there is an accessible subject 0 The binding domain for the pronoun can be more conservatively de ned 7 any subject like element accessible or not will do Expletive subjects satisfy the in principle 45 a Migueli said that it seemed to himselfi that we were trying to speak Dutch b Migueli said that it seemed to himi that we were trying to speak Dutch 46 Binding Domain Final Attempt for now a For Anaphors the binding domain of an NP is the smallest XP that contains i the NP ii its caseimarker and iii an accessible subject b For Pronouns the binding domain of an NP is the smallest XP that contains i the NP ii its caseimarker and iii a subject It was mentioned earlier that there is speaker variation concerning the acceptability of 101mg saw a picture ofhimi It is possible that this variation can be related to what counts as a subject in 46 The speakers who permit coreference would be analyzing the determiner as a subject while those who do not allow for coreference will not allow it to satisfy the subject requirement 4 Digression Garden Path Sentences 47 a The daughter of the king s son likes himself Christina Willis pc b The horse raced past the barn fell c I drove my aunt from Peoria s car References Chomsky N 1981 Lectures on Government and Binding Foris Dordrecht Chomsky N 1986 Knowledge of Language its Nature Origin and Use Praeger New York NY Reinhart T and E Reuland 1993 Re exivity Linguistic Inquiry 244 6577720 Williams E 1985 PRO and the subject of NP Natural Language and Linguistic Theory 33 2977315 Introduction to Transformational Grammar LlNGUlST 601 November 4 2008 Passives 1 Passives 1 a Active Dave invited Roumi b Passive Roumi was invited by Dave 11 Some components of the passive 2 a DemotionDeletion of the external argument b Promotion of the Direct Object Additional Properties in English 3 31 There is a passive participle fen which happens to be homophonous with the past participle F7 The passive participle combines with the auxiliary be which is therefore sometimes called the passive auxiliary n The demoted external argument is optional but if it is overtly realized it appears with the preposition by The external argument thought not syntactically projected as an argument is very much around in the passive This distinguishes the passive from related unaccusative ergatives 4 a The tar was being melted There was someone who was melting the tar b The tar was melting The water could be melting because it was very hot 5 Hindi passive of kauf transitive cut syntactically agentive E yeh per Ramdwaaraa kal kaataa gayaa thaa this treeMSg Ramby yesterday cutvaMSg GOvaMSg be stMSg This tree was cut by Ram yesterday F7 kaf intransitive cut encyclopedically agentive but not syntactically agentive yeh per Ramdwaaraa kal kataa thaa this treeMSg Ramby yesterday cutvaMSg bePstMSg This tree cutimT yesterday no accurate English translation the passive auxiliary in Hindi is jun go In general transitive verbs can be passivized crosslinguistically 0 Languages vary with respect to intransitive predicates 1 AN 3 0 some languages allow for passivization of unergative intransitives suggesting that 2b may not be obligatory in these languages possibly related to variation wrt EPP on T0 39 oo few if any languages allow for p of definitional nature of 2a 0 the Syntactic Treatment of the Passive Burzio s Generalization a 1 head that does not assign a Orole to its specifier does not license accusative case 3 VUnacc b vAG without a specifier ie the passive vAG configuration note Burzio s Generalization only apples to v s Otherwise T0 would notbe able to license case a b c d NPuCase7 VAC NPluCaseiilll enPass vAG NPuCase7 be enPass vAG NPuCase7 T uNquot be enPass vAG NPuCase7 f Cases like the following suggest that there might be a uNquot feature on the intermediate heads have be mm 8 a The apples might all have been eaten b The apples might have all been eaten c The apples might have been all eaten d The ungrarnrnaticality of 8d needs an explanation See suggestion in Sportiche 1988 The apples might have been eaten all Why do we need a eenPuss How do we force a eenPuss 9 Selectional Properties of eenPuss T0 eenPer and fingPr0g a T0 vP headed by vam or a vAG with a specifier have and be can be thought of as instances of vUWCC b eenPer vP headed by vam or a vAG with a specifier c fingPr0g vP headed by vam or a vAG with a specifier d eenPuss vP headed by vAG without a specifier A semantic generalization 10 a vP headed by vam or a vAG with a specifier saturated F7 vP headed by vAG without a specifier unsaturated 11 a T0 fenPer fingPr0g saturated vPs The train arrivedhas arrivedis arriving John readhas readis reading the paper The paper readhas readis reading F7 fenPass unsaturated vPs The egg was boiled The train is arrived Iohn is read the paper The fact that fenPass saturates an unsaturated predicate led some authors cf Baker et al 1989 to argue that the external argument Orole and the associate case are both assigned to this fen thus accounting for CaseAbsorption and 9absorption 12 Further Selection a have selects for fenPer b be selects for everything else So be is only very loosely a passive auxiliary This is good given the existence of the apple eaten by 0hn The presence of the vAG in the passive allows for indirect specification of the agent through a by phrase While there may be many ways in which the byphrase indirectly specifies the agent the one thing we would like to make sure given the preceding discussion is that the byphrase should not saturate the external argument of vAg If it did we would incorrectly not get fenPass1 13 Three Kinds of Passive Participles 13 a Eventive Passive only eventive The door was closed at 5pm by John F7 Resultative Passive stative with a previous event The doors are closed as the result of a previous event The cake is attened The metal is hammered n AdjectivalStative Passive purely stative no event These doors were built closed see Embick 2004 for details 1Some additional facts to keep in mind i not all instances of byphrases are relevant here We are only interested in agentintroducing byphrases thus the byphrase in the unaccusative The water will drain all by itself is not an exception to our correla onbetween the presence absence of VAG in a verbal structure and the possibility of a byphrase ii the aforementioned correlation between VAG and the possibility of a byphrase is limited to verbal contexts Nominalizations allow for byphrases even though there are reasons to believe that they do not involve the projection of a VAG eg the destruction of Carthage by the Romans See Marantz 1997 The adjectival passive sometimes differs in form from the other two passives 14 15 16 a The door was opened at 5pm F7 The door is opened as a result of an opening event These doors were built openopened P a There was a door opened at 5pm F7 There are several doors opened right now There are several doors open right now P Embick s characterization a Eventive Passive Asp VAC NPll b Resultative Passive ASPRO VFIENT l NPll c Stative Adjectival Passive Asps0 NP 0 For the resultative and the stative passive to work it should be possible to associate a state with the root 0 My characterization of the structural location of the NP differs from that of Embick but as he indicates not much depends upon this 0 The semantic characterization provided for the insertion of the eeriPass does not apply to the resultative passive and the stative passive To make sure that we get the surface form fen we need additional morphological assumptions For a worked out proposal see Embick 2003 2 Expletives Expletives can appear with 17 E all passives because of be There were several apples eaten at the count fair F7 more generally everything that takes be as an auxiliary There was a man eating an apple at the county fair There were several firemen availablein the room many unaccusatives arrive accumulate appear materialize Suddenly there arrived an undead creature from Athol P P but not all unaccusatives break sink 7There sank a ship Levin l99388 91 notes that the verbs that allow for there subjects can be broken down into the following subclasses 18 Verbs of Existence blaze bubble cling coexist tower wind writhe F7 E Verbs of Spatial Configuration crouch dangle hang kneel stretch swing Meander Verbs cascade climb crawl cut weave wind P 0 Verbs of Appearance accumulate appear arise stem supervene surge 17 Verbs of lnherently Directed Motion ran and roll verbs amble climb crawl creep strut swim trudge walk and more An important distinction to keep in mind is that Verbs of Change of State see Levin 1993240 248 do not permit there even though they are unaccusatives Some examples break Chip rip shatter split tear bend crease rample wrinkle bake blanch roast toast etc Levin 1993 points out that verbs that allow for there subjects differ in where they allow for the postnominal NP to appear 19 arrive vs ran a i There arrived three gentlemen from Verona ii There arrived from Verona three gentlemen making three gentlemen heavy can make 19aii better b i There ran a raggedy looking cat into the room ii There ran into the room a raggedy looking cat The contrast between arrive and ran does not follow from our system and suggests that further distinctions might need to be made between arrive and run A simple treatment of the above facts is suggested by Freeze 1992 who proposes that verbs with a locative component in their meaning optionally select for there in their specifiers Freeze s solution is not widely adopted because of the conceptual difficulties associated with selecting a semantically vacuous element But then we are left without an explanation for the ungrammaticality of 17 among other things Freeze s explanation might shed light on contrasts such as the following 20 a There have arrived several people from Verona b 7quot There have several people arrived from Verona c There were several people arrested by the police today d 7There were arrested several people by the police today References Baker M K Johnson and l Roberts 1989 Passive Arguments Raised Linguistic Inquiry 202 2197252 Ernbick D 2003 Locality Listedness and Morphological Identity Studiu Linguisticu 573 1437170 Embick D 2004 On the Structure of Resultative Participles Linguistic Inquiry 353 35amp392 Freeze R 1992 Existentials and Other Locatives Language 68 5537595 Levin B 1993 English Verb Classes and Alternutions A Preliminary Investigation The University of Chicago Press Chicago Marantz A 1997 No Escape from Syntax Don t Try Morphological Analysis in the Privacy of Your Own Lexicon in A Dirnitriadis L Siegel C Surek Clark and A Williams eds Proceedings of the 21st Penn Linguistics Colloquium UPenn Working Papers in Linguistics Philadelphia 2017225 Sportiche D 1988 A Theory of Floating Quantifiers and its Corollaries for Constituent Struc ture Linguistic Inquiry 194 4254149 Introduction to Transformational Grammar LINGUIST 601 November 1 2004 Passives 1 Passives l a Active Dave invited Roumi b Passive Roumi was invited by Dave 11 Some components of the passive 2 a DemotionDeletion of the external argument b Promotion of the Direct Object Additional Properties in English 3 5 There is a passive participle E which happens to be homophonous with the past participle 0 The passive participle combines with the auxiliary E which is therefore sometimes called the passive auxiliary P The demoted external argument is optional but if it is overtly realized it appears with the preposition The external argument thought not syntactically projected as an argument is very much around in the passive This distinguishes the passive from related unaccusative ergatives 4 a The tar was being melted There was someone who was melting the tar b The tar was melting The water could be melting because it was very hot 5 Hindi passive of kaat transitive cit syntactically agentive P yeh per Ramidwaaraa kal kaatiaa gaiyaa thaa this treeMSg Ramiby yesterday cutinvMSg GOinvMSg bePstMSg This tree was cut by Ram yesterday 0 kit intransitive cit encyclopedically agentive but not syntactically agentive yeh per Ramidwaaraa kal kaltiaa thaa this treeMSg Ramiby yesterday cutinvMSg bePstMSg This tree cutim yesterday no accurate English translation In general transitive verbs can be passivized crosslinguistically 0 Languages vary with respect to intransitive predicates 7 0 some languages allow for F of uueigati e 39 39 39 CO 39 O that 2b may not be obligatory in these languages possibly related to variation wrt EPP on T o few if any languages allow forr 39 39 39 of 39 39 U the de nitional nature of 2a 12 Syntactic Treatment of the Passive 6 Burzio s Generalization a v head that does not assign a 87role to its speci er does not license accusative case 8 VUmutzz b VAC without a speci er ie the passive vAG con guration note Burzio s Generalization only apples to v s Otherwise T0 would not be able to license case a NPuCase7 b VAC NPuCase7 c 7enPass VAC NPuCase7 d be 7enPass VAC NPuCase7 e T uN be 7enPass VAC NPuCase7 f Cases like the following suggest that there might be a uN feature on the intermediate heads have bi VAG 8 a The apples might all have been eaten b The apples might have all been eaten c The apples might have been all eaten d quotThe apples might have been eaten all The ungrammaticality of 8d needs an explanation See suggestion in Sportiche 1988 Why do we need a fen Pass How do we force a fen Pass 9 Selectional Properties of 7enPass T 7enPerf and fingProg a T vP headed by vwwm or a VAC with a speci er have and bi can be thought of as instances ovamm 0 7enPerf vP headed by vwwm or a VAC with a speci er o fingProg vP headed by vwwm or a VAC with a speci er fl fen Pass vP headed by VAC without a speci er A semantic generalization 10 a vP headed by vwwc or a VAC with a speci er saturated b vP headed by VAC without a speci er unsaturated 11 a T 7enPeri fingProg saturated vPs The train arrived has arrived is arriving John readhas readis reading the paper The paper readhas readis reading 0 en Pass unsaturated vPs The egg was boiled The train is arrived John is read the paper The fact that en Pass saturates an unsaturated predicate led some authors cf Baker et al 1989 to argue that the external argument 87role and the associate case are both assigned to this 2 thus accounting for CaseiAbsorption and 297absorption 12 Further Selection a have selects for 7enPerf b b selects for everything else So E is only very loosely a passive auxiliary This is good given the existence of the apple eaten by John The presence of the VAC in the passive allows for indirect speci cation of the agent through a g phrase While there may be many ways in which the byiphrase indirectly speci es the agent the one thing we would like to make sure given the preceding discussion is that the iphrase should not saturate the external argument of VAG lfit did we would incorrectly not get 7enPass1 13 Three Kinds of Passive Participles 13 a Eventive Passive only eventive The door was closed at 5pm by John 0 Resultative Passive stative with a previous event The doors are closed as the result of a previous event The cake is attened The metal is hammered P AdjectivalStative Passive purely stative no event These doors were built closed see Embick 2004 for details 1Some additional facts to keep in mind i not all instances ofbiyrphrases are relevant here We are only interested in agentrintroducing biyrphrases 7 thus the biyrphrase in the unaccusative The water will drain all by itself is not an exception to our correlation etween the presenceabsence of vAG in a verbal structure and the possibility of a rphrase ii the aforementioned correlation between vAG and the possibility of a biyrphrase is limited to verbal contexts Nominalizations allow for rphrases even though there are reasons to believe that they do not involve the projection of a vAG eg thie destruction ofCarthage by the Romans See Marantz 1997 The adjectival passive sometimes differs in form from the other two passives 14 15 16 a The door was opened at 5pm 0 The door is opened as a result of an opening event 9 These doors were built openopened 5 There was a door opened at 5pm 0 There are several doors opened right now 9 There are several doors open right now Embick s characterization a Eventive Passive Asp0 VAC NPll b Resultative Passive ASPR0 VFIENT NPll c Stative Adjectival Passive Asps0 NP 0 The roots V in the resultative and the stative passive have to be stative 0 My characterization of the structural location of the NP differs from that of Embick but as he indicates not much hangs upon this 0 The semantic characterization provided for the insertion of the Pass does not ape ply to the resultative passive and the stative passive To make sure that we get the surface form fen we need additional morphological assumptions For a worked out proposal see Embick 2003 2 Expletives Expletives can appear with 17 a all passives because ofbie There were several apples eaten at the count fair b more generally everything that takes E as an auxiliary There was a man eating an apple at the county fair There were several remen available in the room c many arrive appear materiali e Suddenly there arrived an undead creature from Green ed d but not all unaccusatives break sink 777There sank a ship Levin 19938891 notes that the verbs that allow for there subjects can be broken down into the following subclasses 18 a Verbs of Existence blaze bubble cling coexist tower wind writhe 0 Verbs of Spatial Con guration crouch dangle hang kneel stretch swing c Meander Verbs cascade climb crawl cut weave wind 1 Verbs of Appearance accumulate appear arise stem supervene surge e Verbs of lnherently Directed Motion E and gllverbs amble climb crawl creep strut swim trudge walk and more An important distinction to keep in mind is that Verbs of Change of State see Levin l9932407 248 do not permit there even though they are unaccusatives Some examples break chip rip shatter split tear bend crease rumple wrinkle bake blanch roast toast etc Levin 1993 points out that verbs that allow for there subjects differ in where they allow for the postnominal NP to appear 19 mvs m a i There arrived three gentlemen from Verona ii 77There arrived from Verona three gentlemen b i There ran a raggedy looking cat into the room ii There ran into the room a raggedy looking cat The contrast between arrive and run does not follow from our system and suggests that further distinctions might need to be made between arrive and run A simple treatment of the above facts is suggested by Freeze 1992 who proposes that verbs with a locative component in their meaning optionally select for there in their speci ers Freeze s solution is not widely adopted because of the conceptual dif culties associated with selecting a semantically vacuous element But then we are left without an explanation for the ungrammaticality of l 7 among other things Freeze s explanation might shed light on contrasts such as the following 20 a There have arrived several people from Verona b 7 There have several people arrived from Verona c There were several people arrested by the police today 1 777There were arrested several people by the police today References Baker M K Johnson and l Roberts 1989 Passive Arguments Raised Linguistic lnguiry 202 2197252 Embick D 2003 Locality Listedness and Morphological Identity tudia Linguistica 573 1437170 Embick D 2004 On the Structure of Resultative Participles Linguistic Inquiry 353 3557392 Freeze R 1992 Existentials and Other Locatives Language 68 5537595 Levin B 1993 English Verb Classes and Alternatinns39 APreliminary 39 sity of Chicago Press Chicago 0 The Univeri Marantz A 1997 No Escape from Syntax Don t Try Morphological Analysis in the Privacy of Your Own Lexicon in A Dimitriadis L Siegel C SurekiClark and A Williams eds Proceedings of the 21st Penn Linguistics C J 39 UPenn Working Papers in Linguistics Philadelphia 2017225 Sportiche D 1988 quotA Theory of Floating Quanti ers and its Corollaries for Constituent Struci ture Linguistic lnguiry 194 4257449 Introduction to Transformational Grammar LINGUIST 601 September 27 2004 Complement Selection and the Structure of the VP 1 C selection S selection and L selection 11 Categorial Selection 1 Ceselection categorial selection 7 certain heads impose particular demands on the category of the XP they combine with These demands are referred to as ceselection Some things we could code using ceselection 2 Imowcan take NPs indicatives SS and interrogative S s a John knows Npthe time b John knows 3 that the world is full of noises c John knows 3 what the time is 3 ask can take NPs and interrogative S s but not indicative S s a John asked me Npthe time b John asked me 3 that the world was full of noises c John asked me 3 what the time was 4 wondercan only take interrogative S s not NPs or indicative S s a Paul wonders Npthe time b Paul wonders 3 that the world is full of noises c Paul wonders 3 what the time is 5 a A adjectives require PP complements PP fond of the tall student NP fond the tall student N fond tall student AP fond tall b N nouns require PP complements PP queen ofthe blue isle NP queen the blue isle N queen blue isle AP queen que c P prepositions typically require NP complements NP on the brown table N on brown table AP on brown PP on below the brown table We can build ceselection into our system by adding uninterpretable categorial features on heads A head which ceselects subcategorizes for an XP will have an uninterpretable X categorial fea ture indicated as uX For a syntactic derivation to succeed ie converge all the uninterpretable features must be deleted by a matching categorial feature on its complement 12 Semantic Selection Semantic selection is the idea that predicates impose quot 39 on their by imposing constraints on the semantics of the complement For example for the verbs in 4 we could have something like the following 6 a know complement must be a question or a proposition b ask wonder complement must be a question S7selection seems particularly helpful in cases Where an argument of a particular sort is needed but its category is not xed 7 put selects for a location a Bill put the book on the table b Bill put the book under the table c Bill put the book there 1 Bill put the book away e Bill put the book One can imagine theories that only have c7selection theories that only have s7selection and also theories that have both c7selection and s7selection cf GrimshaW 1979 Certain authors have argued that s7selection is the most basic form of selection and that certain aspects of c7selection can be derived from the semantic properties of the relevant head Theo7 ries that attempt to eliminate c7selection in favor of s7selection need and have explanations for contrasts between ask and wonderWhich have similar s7selectional needs 8 a John asked me the time b John wondered the time 13 Lexical Selection Sometimes particular heads Will select for particular lexical items notjust particular categories This is called Lexical7selection by Pesetsky l 991 9 a verbs i depend rely7 on ii hope 7 for iii toy7 With 0 nouns i love 7 for of ii desire 7 for of P adjectives proud ashamed 7 of similar7 to different 7 from 2 consistent 7 with L7selection displays considerable idiosyncrasy Lexical items that are semantically close can 17 select different prepositions There is also unpredictable crosslinguistic variation in this domain L7selection is also found With clausal complements C 8 a 39 She liked the concerto She liked hearing the concerto iii She liked to hear the concerto 0 39 She enjoyed the concerto She enjoyed hearing the concerto She enjoyed to hear the concerto ll a i He succeeded in convincing her ii He succeeded to convince her b 1 He managed in convincing her i39 He managed to convince her The consensus in the literature seems to be that we need seselection augmented with leselection What would traditionally be put under ceselection can be derived from seselection certain princie ples that govern how certain meanings are canonically realized syntactically and other indepen dent properties of the lexical item See Pesetsky 1991 for details 2 S in X theory 21 VP in X etheory In 12 the NP the city is the complement of the verb destroy l 2 The Romans Vp destroyed Np the city What goes into the Spec VP position Let us consider the case of nominalizations 13 The Romans destruction of the city NP NA The Romans N PP destruction A P NP of the city In 13 we see that all the arguments of N occur within its own phrase maximal projection A natural idea within the spirit of crossecategorial symmetry is to extend the idea that all the arguments of a head should occur within its own phrase maximal projection This gives us the tree in 15 for l4 14 The Romans destroyed the city 15 XP NP X39 The Romans X VP 7 A Npapezzifier V lt The Romansgt V NPzzommmem destroy the city But there is no The Romans inside the VP Yes there isn t because now it is in the Spec XP Questions 1 What is XP 2 Why does the NP move from Spec VP to Spec XP We will answer Question 1 in this section and Question 2 in the next What about VP Several authors have argued that what we have treated above as a VP 1 6 Vp The Romans yr destroy the city has additional structure and involves another head called V pronounced little v 1 7 Ma The Romans M v0 Vp destroy the city The v0 head is taken to introduce the external argument 7 the argument that comes in at the end and tends to be interpreted as an agent see Kratzer 1996 for arguments in support of this position For our current purposes the difference between 16 and l 7 will not be relevant So in the rest of this handout when I will say VP it will actually correspond to what is often called VP When we start talking about passivization and unaccusativity the difference will become more substantive and the relevance of the decomposition of the VP will become clear 22 What is XP XP S However ifXP S what is X the head ofXP S We know from our discussion ofX theory that every XP has to have a head of the same kind ie XP cannot be headed by Y This property is know as endocentricity Now let us consider how we have been analyzing cases like 18 18 Mark believes that Laetitia should kiss Ophelia S Comp S that NP 7 Laetitia Modal VP should V NP kiss Ophelia Neither S nor S are endocentric How can we reformulate S and S so that they are endocentric and t within the X ischema From their distributions we know that VP 75 S So V0 should not be the head of S 0 Approach 1 The element which we have labeled Modal could be head of S This is the most natural approach because i heads are atmost lexical items there may be heads that are smaller than words ii Out of the immediate constituents of S Modal is the only lexical item What then about cases which don t have a modal such as 19a b 19 a Mark believes that Laetitia is kissing Ophelia right now b Mark believes that Laetitia kissed Ophelia yesterday 19a is easier to take care of We canjust create a class ofAuXiIiaryverbs which includes all modal verbs be and one kind of have But cases like 19b pose a greater challenge One option is to say what we have been saying up until now ie the following structure 20 S NP VP Laetitia A V NP kissed Ophelia The immediate problem for this representation is that there is no candidate for the head of S The immediate constituents of S are NP and VP and neither of them are heads So we seem to be stuck At this point we should look back to cases of VPipreposing VPitopicalization We had seen there that some kinds of VPs could be preposed but others couldn t 21 a Vp Kiss Ophelia Laetitia did b Vp Kissed Ophelia Laetitia c Vp Kissed Ophelia Laetitia did What distinguishes the grammatical Zlb from the ungrammatical 21a The grammatical case involve a VP without tensepersonnumber marking The verb appears in its bare form The VP in the ungrammatical case involves a verb marked for tensepersonnumber marking In the case at hand the verb kissed is marked for Past Tense We want to distinguish between these two kinds of VPs and yet also retain a link between them 0 Approach 2 We postulate node In ection where the tensein ectional information associated with a verb could be stored This node ln ection will head a phrase ln ection Phrase lP which will be equivalent to an S IP is also often referred to as the Tense Phrase TP 22 Laetitia kissed Ophelia 23 IP NP 1 Laetitia 10 VP ed A preztifier V Laetitia V sztomplemem kiss Ophelia 0 Clearly there has to be a way for the fed suf x under I0 and the verb kiss to combine In ection Proposal An I0 and a V0 that heads the complement VP of the I0 combine in the phonological output ie when you try to pronounce the above tree the 10 node and the V0 node combine and are pronounced together Cases like Zlb are bad because they would involve topicalization of an I and only full phrases can be moved around Zlc is bad because there are two sets oftenseagreement markings oat ing around while there is only one I0 to supply the information 23 X rules for IP Since we have adopted X itheory the form of the rules will be quite familiar 24 a Iquot gt NPsmm 139 b 139 gt 1 VPCmnplwnem We can now give a new de nition of the notion subject ofa sentence The subject ofa sentence is the NP that occurs in the Spec lP In English and many other languages the subject of a sentence agrees with the verb What it means for a subject to agree with its verb is illustrated in 25 25 a John eats pizza b John eat pizza c I eat pizza 1 l eats pizza This relationship between the verb and its subject can be stated extremely locally within our new system as the re ex if the Speci erihead relationship 26 Agreement Rule Copy the personinumber features of the NP in Spec IE on the l The personinumber features of an NP are also referred to as its features phiifeatures The features of some pronouns are shown below 27 a l lst person singular b we lst person plural you 2nd person lo she 3rd person singular they 3rd person plural 39D In English gender is not part of the verbal agreement system so she and he can be taken to have the same features but in languages where they are part of the agreement system they would also need to be represented 24 What can go under 1quot ln sentences without any auxiliary element the in ection is all there is in 1 However other elements can also appear under I l mustshouldcould eat some waf es Modals 28 a b I am eating some waf es be c l have eaten some waf es today auxiliary have 1 I did not eat the waf es auxiliary do e I want to eat waf esin nitival to Earlier we saw two syntactic processes 7 VP Topicalization and pseudocleft formation 7 which were sensitive to the presence of in ection Another grammatical process that is sensitive to the presence of in ection is VP Ellipsis VP Ellipsis in English involves a silent tenseless VP together with an overt realization of l 29 Jerry shouldn t leave town Bill should Vpleavetown UP Tyrone isn t eating waf es today but Ken is XPiseaungapples Max hasn t nished his homework but Jose has Vp mshedhishomewmk 9 Ana doesn t want to leave but Mona wants to Vpleave fl Chunghye doesn t like unicorns but Maribel does Vp likeumcoms Chunghye doesn t like unicorns but Maribel Vp likesumcmns 57 Each ofthe elements in 28 has distinct properties Let us consider them individually To provide contrast we will start by looking at one verbal element that cannot occur in 1 namely a main verb 241 Main Verbs Main verbs have nonitensed forms past participles present participles and in nitival forms 30 Talvin ate the pizza a b Talvin has eaten the pizza past participle P Talvin is eating the pizza present participle fl Talvin wants to eat the pizza in nitival form When negated or questioned a form ofthe verb do is needed Otherwise the sentence is ungrami matical 31 a Talvin didn t eat the pizza b Talvin eatn t the pizza c Did Talvin eat the pizza 1 Eat Talvin the pizza e Why did Talvin eat the pizza f Why ate Talvin the pizza Cases such as these can be explained by noting that the presence of the negation disrupts the local relationship needed by the l0 and the V0 in order to combine together 1 Crucially main verbs stay in V0 and do not move to l The verb do comes in and saves the day by giving a realization to the suf x in l which could not have been pronounced on its own This process is called do support 242 Modals Modals are set apart by the fact that they can never occur in nonitensed environments 32 a Talvin wants to mustshouldcould win this game b To mustshouldcould play baseball is fun Modals invert in questions and precede negationZ 33 a Mustshouldcould Talvin win this game b Why mustshouldcould Talvin win this game c Talvin mustshouldcould not win this game The facts follow if we assume that Modal verbs are always generated in a Tensed I Since modal verbs are generated in 1 they can realize whatever features I0 has and the I0 does not need to be close to V0 1The structure is something like I0 Negation PWhere do adverbs go 2In fact dorsupport is not a possibility here 243 Auxiliaries be and havepwfm Unlike Modals auxiliaries can occur in noni nite environments 34 a Talvin wants to be popular b Talvin wants to have been popular However like modals auxiliaries invert in questions and precede negation 3 35 a Is Talvin winning this game b Has Talvin won this game c Why is Talvin winning this game 1 Why has Talvin won this game e Talvin isn t winning this game f Talvin hasn t won this game The above examples suggest that the auxiliaries havebe are generated in V0 like main verbs but can move up to l0 unlike main verbs 4 o to only occurs in itense l 0 do only occurs in tense I0 when the l0 is unable to combine locally with V05 25 S in X theory 5 as it stands is an exocentric projection ie it is not headed by a head of its own category Actually things are even worse It is quite unclear whether 539 has a head 36 S Comp S that Tim is nuts A neat solution and one that is compatible with X itheory is to take Comp as the head of S In fact this solution is forced upon us since the only potential head among the immediate constituents of S is Comp We cannot look inside the SlP for a head because the IP is a complete phrase by itself Assuming the IP to be a complement of Comp we have the following tree 37 Speci er C Comp lP that Tim is nuts 3Dorsupport is not a possibility here either Consider what happens with have to possessive have 4What does the tree look like now 5What about the other do What goes into the Spec CP We will answer this question when we discuss Whemovement questions relative clauses etc We now move to discuss why the NP in Spec VP moves to Spec 1P The answer lies in the domain of Case Theory and AeMovement References Grimshaw J 1979 Complement Selection and the Lexicon Linguistic Inquiry 102 2797326 Kratzer A 1996 Severing the External Argument from Its Verb in J Rooryck and L Zaring eds Phrase Structure and the Lexicon Kluwer Pesetsky D 1991 Zero Syntax Vol2 7 ln nitives unpublished manuscript MIT available from website Introduction to Transformational Grammar LlNGUlST 601 November 30 2006 In nitival Complementation 1 Control So far we have looked more or less exclusively at finite clauses We now turn our attention to infinitival clauses We will look at the following three cases 1 a Control the subject of the infinitival is a null pronoun called PRO David tried PRO to dance PRO to dance with David is fun Raising the subject of the infinitival moves to a higher subject position Makotoi appears ti to be happy Ioeyi seems ti to be exhausted ECM the subject of the infinitival is an overt NP lbelieve Angela to be innocent Minjoo wants him to stay F7 P 1 PRO and Control 2 a Dave tried to write a paper about causatives b Iunko decided to visit UMass in December c Bernhard plans to teach a seminar in the Fall 0 The embedding predicate is a two place predicate someone who triesdecidesplans some thing and something that is trieddecidedplanned o The infinitival clause cannot have an overt subject 3 a Dave tried Dave to write a paper about causatives b 7ka decided Iunko to visit UMass in December c Bernhard plans Bernhard to teach a seminar in the Fall 11 Some motivation for PRO Given our assumptions about Oroles and the EFF we need to assume that there is a null subject in the infinitival clauses under discussion What could this subjectbe Some null DPs 4 a DPtracecopies of DPs that we do not pronounce b m a null pronoun that needs case c PRO a null pronoun that does not need case 0 An important theoretical consideration 5 Ocriterion An argument DP must receive a Orole and may receive only one 6role The only one Orole requirement rules out movement and hence DPtracescopies of DPs in the infinitival subject in the infinitival subject position mcould be a possibility see Borer 1989 butm as generally conceptualized can only appear in positions where it can get case and where overt DPs may also appear If we keep to the only one O role part of the 6criterion we need to postulate a new kind of entity a null pronoun called PRO which can satisfy the EPP requirement receive a Orole and which does not need case 6 a Iunkoi decided PROZ to visit UMass in December b Fred promised Alex PROZ to finish his paper by Monday c Fred persuaded Alex PROZ to finish her paper by Monday The co indexing indicates what DP controls the PRO If the proposal that this relationship is essentially a semantic one is correct then these indices do not need to be represented in the syntax As always once we postulate a null entity we have to make sure that it appears only where we want it to appear This will be an important aspect of our discussion 12 Control as Raising Control and Raising are generally taken to involve different modules of the grammar control theory and movement respectively Control Theory however remains poorly understood and there has been an active line of work originating in Hornstein 1999 which tries to derive Control via movement To derive Control via movement the only one Orole part of the 6criterion must be given up Note that deriving Control via Movement is not the same thing as saying that Raising and Con trol are the same There is no denying that there are many many properties on which control constructions differ from raising constructions Landau 2003 suggests that since control constructions are very different from raising construc tions we want to keep the derivation of the two distinct In their reply to Landau 2003 Boeckx and Hornstein 2004 point out that there are in fact certain parallels between control and raising and claim that the nonparallels can be derived from independent differences between raising and control 13 Different Kinds of Control Control construction can be subdivided along several dimensions Some of these subclasses seem more amenable to a movement analysis than others 131 Subject vs Object Control 7 Ditransitive control predicates a subject control Fred promised Alex to finish his paper by Monday b object control Fred persuaded Alex to finish her paper by Monday There are many object control predicates but very few subject control predicates This has led some researchers to propose a Minimal Distance Principle which forces object control cf Rosen baum 1967 MDFviolating subject control is taken to involve a special structure where the Minimal Distance Principle is in fact respected cf Larson 1991 The conclusion most other researchers have reached is that while the availability of a control relationship is a property of the syntax the exact identity of the controller subject vs object follows from the semantics of the embedding predicate cf Dowty 1985 Culicover and Jack endoff 2001 Iackendoff and Culicover 2003 among others The intuition expressed by Dowty is that there could not be a verb that had the same meaning as promise but which was object control and vice versa Facts from acquisition add an additional twist to this discussion MDP violating subject control verbs like promise seem to be acquired much later see Boeckx and Horn stein 2003 132 Obligatory Arbitrary and Optional Control Most of the cases of control seen so far involve obligatory control ie the subject of the infinitival clause can only be interpreted as dependent on an argument of the embedding predicate for its interpretation 8 Obligatory Control a lnfinitival nonwhComplements i Angela tried to disinvite him ii Fred promised Alex to finish his paper by Monday iii Fred persuaded Alex to finish her paper by Monday b lnfinitival Adjuncts i Andre read Rushdie s article about Coetzee to make a presentation in his class ii Roumi went to Tromsoe to talk to Sylvia Not all instances of control are obligatory in some cases the FRO seems to lack an obvious controller1 and takes on a genericarbitrary interpretation these cases are referred to as FROaTb 1But see Epstein 1984 and Bhatt and Izvorski 1997 who argue that even in these cases there is an implicit controller Arbitrary control is diagnosed by its ability to bind oneself and the availability of a paraphrase that involves the pronoun 9 Arbitrary Control a unique argument of embedding predicate i PROarb to walk along Paradise Pond in the Fall is fun ii PROMy to behave oneself in public is important iii It is not allowed PROarb to perjure oneself part of a whCP F7 i lVLinjoo knows how PROM to behave oneself in public ii Tim wonders how PROM to protect oneself from creditors In some cases such as when the infinitival clause is embedded in a wh CP arbitrary control is not the only option 10 optional control a Minjooi knows how PRO to behave herself in public b Tim wonders how PRO to protect himself from creditors For obvious reasons these cases are referred to as involving optional control Landau 2000 Landau 2003 makes a further distinction noting that the PRO subject of initial adjuncts can be interpreted as nonarbitrary and yet not controlled by the matrix subject 11 NonObligatory Control NOC a Mary was baf ed Even after PRO revealing her innermost feelings Iohn remained untouched b Mary lost track of John because PROM having been angry at each otherm he had gone one way and she another P Having PRO just arrived in town the main hotel seemed to Bill to be the best place to stay He argues that the whinfinitival cases of arbitrary control are really special cases of partial control and should not be mixed with cases of NOC like the above In the whinfinitival cases even the arbitrary PRO must include the subject in its reference True disjoint reference is not allowed 12 a John wondered who PRO to introduce his fianceehimi to b Iohni asked how PRO to talk to Mary himi about oneself 133 Partial vs Exhaustive Control There is also a class of cases where the matrix predicate provides only part of the reference of the subject of the infinitival clause see Landau 2000 for details 13 Partial Control We thought that a He wanted PROPr to meet in the lobbydo the dishes together b The chair preferred PROFr to gather at 6 c Bill regretted PROFr meeting without a concrete agenda d Mary wondered whether PROFr to apply together Other predicates do not allow partial control 14 a Bevi began PROFr to do the dishes together b The chair managed PROi to gather at 6 134 Implicit Control The controller of PRO can be an implicit argument ie an argument that does not seem to be syntactically projected Languages differ in the extent to which they allow for implicit arguments to control a PROZ 15 Unaccusat39ive vs Passives a No implicit argument No Control quot The ship sank PROZ to collect the insurance b Implicit agent Control The ship was sunk PROZ to collect the insurance 16 lmplicit accusatives vs lmplicit datives a lmplicit Accusatives This leads one PROZ to draw the following conclusion b lmplicit Datives Iohn said shouted to the Visitors PROZ to return later Unlike English both are good in Italian See Rizzi 1986 2T1 y 39 quot 39 mlledBach F quot 39 vrlihnto L39 L L39 39 quot Bachs C quot hold as luugwe 39 to 39 L t D 439 439 quott ue implith 14 Some Properties of Control Constructions Setting aside cases of NOC For that see Landau 2001 o The controller can never be an expletive 17 a Therei tried PROZ to annoy David b lti hopes PROZ to win with explet ive it This is a definitional property of control 0 The PRO is always a subject 18 a Hei tried PROZ to annoy David b Hei tried David to annoy PROi c Hei tried PROZ to be annoyed at David 0 The controller of PRO needs to be an argument of the predicate to which the in nitival clause is attached 19 a Hei thinks that Ij tried PROMj to annoy David b Hei thinks that lj persuaded Mildredk PROM c to leave 20 ccommand follows from argument requirement and the fact that the clausal complement is the innermost argument a Hisi parentsj tried PROMj to annoy David b L persuaded Mildredj s motherk PROM c to leave 0 PRO cannot be a real expletive 21 a For Mary to dance would be fun b PRO to dance is fun c For there to be a party tonight would be fun d PRprl to be a party tonight would be fun e For it to seem that Mary is a nonsmoker she will have to get new rugs f PRprl to seem that Mary is a nonsmoker she will have to get new rugs 22 from Lasnik 1992244 a There having been a robbery there was an investigation b There was a crime without there being a victim c PROZ having witnessed the robbery Iohni aided the investigators d Harry was a witness without PROZ being a victim But PRO can function as weather pseudoambient it 23 a It can hail without it snowing b It can hail without PRO snowing References Bhatt R and R lzvorski 1997 Genericity Implicit Arguments and Control paper presented at SClL V11 available at ftplingupennedustudentpapersbhattPROarbps Boeckx C and N Hornstein 2003 Reply to Control is not Movement Linguistic Inguiry 342 2697280 Boeckx C and N Hornstein 2004 Movement under Control Linguistic lnguiry 353 431452 Borer H 1989 Anaphoric AGR in O Iaeggli and K Safir eds The Null Subject Parameter Kluwer Dordrecht 697109 Culicover P W and R Iackendoff 2001 Control is not Movement Linguistic lnguiry 323 4937512 Dowty D 1985 On recent analyses of the semantics of control Linguistics and Philosophy 8 2917331 Epstein S D 1984 Quanti erm and the LE Representation of PROarb Linguistic lnguiry 153 4997505 Hornstein N 1999 Movement and Control Linguistic lnguiry 301 6996 Iackendoff R and P W Culicover 2003 The Semantic Basis of Control in English Language 793 5177556 Landau l 2000 Elements of Control Structure and Meaning in lnfinitival Constructions Kluwer Dordrecht Landau l 2001 Control and Extraposition The Case of SuperEqui Natural Language and Linguistic Theory 191 1097152 Landau l 2003 Movement out of control Linguistic lnguiry 343 471498 Larson R 1991 Promise and the theory of control Linguistic lnguigy 221 1037139 Lasnik H 1992 Two Notes on Control and Binding in R K Larson S latridou U Lahiri and IT OO eds Control and Grammar Studies in Linguistics and Philosophy 48 Kluwer Academic Publishers Dordrecht Reidel 2357252 Rizzi L 1986 Null Objects in Italian and the Theory of m Linguistic lnguiry 173 5017558 Rosenbaum P 1967 The grammar of English predicate r con u uctiou MIT Press Cambridge MA Introduction to Transformational Grammar LINGUIST 601 November 12 2004 In nitival Complementation 2 ECM and Raising 1 Exceptional Case Marking Two distinct structures are possible for the string in l 1 NF V NP to VP a ECM NP V NP to VP I believe him to be innocent b Control NP V NP PRO to VP I persuaded David PRO to dance Implications of the the proposed structures 2 a ECMimexm NPtO Vpembeddedl l Vmaww is a two place predicate He expected me to take syntax 1 He expected that I would take syntax 2 NF does not get a 87role from Vmawm gt NP can be expletive if VP8mbedded permits gt NP can be a nonireferential phrase licensed by VPWMMM 3 NF gets case from outside the embedded in nitival clause gt the case relationship crosses a TP boundary 13 Contmli Vmim NP PR0 to Vpembeddedl l Vmaww is a three place predicate He persuaded me to take syntax 1 He persuaded me that I should take syntax 2 NF gets a 87role from Vmawm gt NP cannot be an expletive gt NP cannot be a nonireferential phrase licensed by VPWMQMM 3 NF gets case from Vmawm gt the case relationship does not cross a TP boundary 11 ECM vs Control wrt Expletives and Idiom Chunks 111 Expletives We know that expletives1 cannot appear in object positions 3 expletives a There is a man in the garden b A man is thereawt in the garden not expletive 112 Idiom Chunks nonreferential NPs Certain phrasal constituents receive a special idiomatic interpretation when they appear together with certain other lexical items These combinations are called idiom chunks and for the idiomatic reading to be available the phrasal constituent cannot be an argument of another predicate 4 idiom chunks the cat out of the bag i The cat is out of the bag 9 ii l persuaded the cat to be out ofthe bag take advantage i Lane took advantage of Andrew Advantage was taken of Andrew 0 ii l persuaded advantage to be taken of Andrew 113 Applying the Diagnostic Now what is relevant here is that expletives and idiom chunks can appear as objects of certain embedding predicates but not others ECM predicates allow for expletives and idiom chunks while control predicates do not 5 expletive there a ECM I want there to be 50 chairs in room by noon tomorrow The police allowed there to be looting in the Muslim quarter of the city He expects there to be someone waiting for him at the airport terminal 0 Control 1 persuaded there to be 50 chairs in room by noon tomorrow The police ordered there to be looting in the Muslim quarter of the city He advised there to be someone waiting for him at the airport terminal 1This is especially clear with expletive there The facts with clausal expletive it are murkier due to the existence of cases like He can 39t stand it that they were mean to David See Postal and Pullum 1988 for critical discussion 6 idiom chunks a ECM I don t want advantage to be taken of David He believes the cat to be out of the bag They expected the fur to y They expected the chickens to come home to roost They expected the shit to hit the fan 0 Control 1 persuaded advantage to be taken of David He allowed the cat to be out of the bag with idiomatic reading They persuaded the fur to y They persuaded the chickens to come home to roost They persuaded the shit to hit the fan Something to keep in mind is thatjust because a predicate is an ECM predicate it does not fol low that it will allow for an expletiveidiom chunk NP The relevant NP is licensed within the embedded in nitival and it is the embedded in nitival VP that determines what kinds of NPs are possible The matrix predicate simply plays no role 7 ECM predicate a Expletive The police allowed there to be looting in the Muslim quarter of the city The police allowed there to have a man eat an apple The police allowed there to sink a ship ldiom Chunk I don t want advantage to be taken of David 1 don t want advantage to be read by David 1 don t want advantage to be seen by David 0 Another way of thinking about it is that the possibility of expletivesidiom chunks with ECM predicates depends upon the potential wellformedness of the embedded in nitival The ungrami matical cases in 7 all involve embedded in nitival clauses that are not wellformed 12 Embedded Passivization Passivization of the embedded in nitival substantially changes the meaning in the case of control in nitives but not in the case of ECM in nitives 8 Control a He persuaded the doctor PRO to examine David b He persuaded David PRO to be examined by the doctor 9 ECM a He wants the doctor to examine David b He wants David to be examined by the doctor 13 Two Kinds of ECM We noted that in ECM in nitives the subject of the in nitival gets case from a head that is outside the embedded in nitival There are two possible heads that could supply case 10 a the embedding predicate i I expect there to be unhappiness about this ii He believes him to be uncaring b the in nitival Complementizer for i For there to be a party tonight would be excellent ii He intended for his parents to be present 131 The location of for The fact that there is possible in lObi indicates that the structure is as indicated and not for example ll For there to be a party tonight would be excellent Such a structure might actually be present for He intended for his parents to be present but it is not the only structure possible l 2 Two structures for He intended for his parents to be present a He intended for his parents to be present b He intended for his parents to be present lZa has for his parents forming a constituent 7 thus we expect that it might be possible to move it around and indeed we nd that we can in fact move it aound l 3 a It was for his parents that he intended to be present b For his parents he intended to be present The possibility of l 3 indicates that 12a is a possible structure However the sentences in l 3 have only one of the interpretations available to 12 the interpretation where the parents are the bene ciary of his intended action of being present 7 he will be the one who is present not his parents ie he controls the PRO subject in the in nitival clause in 12a A similar point is made by pseudoclefting 14 a What he intended was for his parents to be present gt forces the structure in 12b b What he intended for his parents was to be present gt forces the structure in 12a 0 for can function as an in nitival Complementizer which assigns accusative case 132 The possibility of ECM o ECM by for is always available as long as the local syntactic context permits a forCP ie for in Standard English always allows for ECM it being an independent question whether a forCP is permitted or not 0 ECM by versz whether a verb can function as an ECM predicate seems to be somewhat id iosyncratic For a predicate to be an ECM predicate it must take a small enough in nitival complement and it must assign accusative case 15 E regret that7CPok for7CP ECM NPok i I regret that he is no longer here ii I regret for him to no longer be here iii I regret him to no longer be here iv I regret this outcome hope that7CPok for7CPok ECM NP i I hope that it doesn t snow this week ii I hope for him to get well soon iii 1 hope him to get well soon iv 1 hope for a favorable outcome believe that7CPok for7CP ECMok NPok i I believe that she is innocent ii I believe for her to be innocent iii I believe her to be innocent iv I believe her account want prefer that7CPok for7CPok ECMok NPok i lwant that he leave ii lwant for him to leave iii lwant him to leave iv 1 want his immediate departure 0 P 0 Often the licensing of accusative case can be diagnosed by the possibility of an NP in place of the in nitival complement But as regret shows the possibility of an NP object does not guarantee ECM ie ECM predicates allow for accusative NP objects but the reverse does not follow Also there is no 171 correlation between the possibility of for7CPs and ECM in nitivals cf hope vs believe 14 ECM vs Control 0 de se interpretations 16 a Control only de se David wants PRO to win the election b ECM both de se and de re David wants himself to win the election 2There are no adjectival ECM predicate 7 this would follow from the general inability of adjectives to assign case There are also no ditransitive ECM predicatesie ECM counterparts ofpezsuade I don t know why this is so o Adjacency effects Control clauses can be separated from their predicates by intervening adverbs While ECM clauses cannot l 7 a Control David tried yesterday PRO to book a ticket for Holland b ECM David expected yesterday him to be late These adjacency facts are part of a more general pattern Which requires adjacency in English between an accusative licensing complex head vAg and the NP that gets accusative The source of this adjacency requirement remains an open question made more interesting by the fact that it seems to be a language speci c restriction not applying for example in French 18 I saw him yesterdayl saW yesterday him No comparable restrictions seem to apply to CPs 2 Raising In principle two distinct structures are possible for the string in 19 19 NP V to VP a Raising NP V th to VP Hei seems ti to like David b Control NP V PRO to VP Hei tried PROi to like David 21 Raising vs Control Implications of the the proposed structures 20 8 RaiSingi NPi mem tNP t0 Vpembeddedl l Vmaww is a one place predicate Hei seems ti to like David 1 It seems that he likes David 2 NF does not get a 87role from Vmawm gt NP can be expletive if VP8mbedm permits gt NP can be a nonireferential phrase licensed by VPWWMM 3 NF gets case from outside the embedded in nitival clause gt the case relationship crosses a TP boundary 13 Control NPi mem PROi t0 mebmml l VWW is a two place predicate He wants to take syntax 1 He desires that he take syntax 2 NF gets a 87role from Vmawm gt NP cannot be an expletive gt NP cannot be a nonireferential phrase licensed by VPembewed 3 NF gets case from VWW gt the case relationship does not cross a TP boundary 21 Expletives and Idiom Chunks a There seems to be a man in the garden b There is likely to be a farmer harvesting pumpkins somewhere right now c The cat appears to be out of the bag 1 The chickens happen to be coming home to roost Expletivesldiom Chunk subjects are only possible if the embedded VP permits them 22 a There seems to be raining b There is likely to be a farmer harvest pumpkins somewhere right now c quotThe cats appear to like David with idiomatic reading 1 The chicken happens to have hit the fan with idiomatic reading Raising predicates can be stacked on top of each other suggesting that raising can take one far 23 a Joey appears to have turned out to have left b Roland happens to apppear to seem to be sick But this is not something that distinguishes raising from control since control predicates can also be stacked 24 a Joey wants PRO to try PRO to get the McGill position b lwant PRO to persuade Joey PRO to apply for the McGill position 22 The class of Raising Predicates An incomplete list of raising predicates 25 a appear seem b happen turn out c be likely be unlikely be certain To this list we can add modals and auxiliaries even though the movement over these doesn t have to involve crossing a TP boundary Only predicates that do not assign a 87role to their speci er can be raising predicates ie raising predicates are all unaccusatives For this reason raising predicates involves raising and not ECM Note that unaccusative syntax is a necessary but not suf cient condition for a predicate to be a raising verb The unaccusative must be able to combine with an in nitival complement in the rst place and the in nitival complement must be small enough 26 Unaccusatives that do not allow Raising3 a be truefalse that7Conk for7CPzquot Raisingquot It is truefalse that David is leaving It is truefalse for David to be leaving Davidi is truefalse ti to be leaving 0 stinksbe possiblenecessary that7Conk for7Conk Raising lt stinksis possiblenecessary that John will win It stinksis possiblenecessary for ohn to win Johni stinksis possiblenecessary Lg to win 0 A prediction ECM predicates and Raising predicates seem to share many properties differing primarily in the case297domain o ECM 87role to subject case to subject of in nitival complement 0 Raising no 87role to subject no case to subject of in nitival complement Given Burzio s Generalization these two properties are related We would expect that ifwe were to passvize an ECM predicate we would have a Raising predicate on our hands and this seems to be the case 27 a expect Johni is expected ti to arrive at 5pm Therei is expected ti to be a party tonight 0 believe 30 missilesi are believed ti to be missing Therei are believed ti to be around 30 missiles missing 3 Formal Treatment Parameters of Variation 0 size of in nitival complement CP vs TP 0 the source nature of case matrix T matrix VAG C0 3I used to put probable in this group but it seems that some people sports fans in particular can say things like However he is probable to play against the Sabres on Wednesday reports the Philadelphia Inquirer Further at least one descriptive grammar online advice common inveighs against the practice of combining probable with an in nitive suggesting that suf ciently many people say this in the rst place 31 Control 28 V op C0 TP PRO T to w 0 Does PRO need case Can PRO ever have case 0 How is the distribution of PRO regulated 311 PRO cannot bear case An important intuition PRO cannot appear in positions where an overt NP can for the most part Since overt NPs require case this distribution would follow if PRO could not tolerate case This intuition seems surprisingly correct 29 a No PRO in object of active He triedwants David to annoy PROi b PRO can be merged VPeinternally as in a passive He triedwants PROi to be annoyed at David 30 a try He tried PRODavid to leave b believe He believes DavidPRO to be innocent 31 for a For all the fugitives to be apprehended by the dementors would be great b C0 PRO to be apprehended by the dementors would be great c For PRO to be apprehended by the dementors would be great4 312 Why a CP 1 In many languages see Landau 2003488 overt complementizers are possible in control cone texts 2 The presence of the CP layer is taken to protect the PRO from potential case assigners 7 case relationships do not cross CP boundaries irrespective of whether they are established via Agree or Move 3 Postulating a null CP layer also helps us with potential exceptions to the case generalization 4As recently as 1584 for to in nitives were still possible in English appearing for example as a verse in the populat folk song Greensleeves Greensleeves was all my joy Greensleeves was my delight Greensleeves was my heart of gold And who but my Lady Greensleeves I have been ready at your hand to grant whatever you would crave I have both wagered life and land Your love and good will for to have They continue to be possible in Irish English though the location of for in Irish English is probably not the same as in Standard English See Henry 1992 and Henry 1995 for details 32 also prefer a He wantsexpects David to be a good person in his next life b He wantsexpects PRO to be a good person in his next life The idea is that expectwant can take both CP and IP complements 33 also prefer a He wantsexpects IpDavid to be a good person in his next life b He wantsexpects Cp C0 TpPRO to be a good person in his next life Earlier stages of the theory referred to wan expect as S ideleting predicates re ecting the belief that the control predicate was basic and that the ECM predicate was derived from it through a process that deleted the CP layer 313 How to block PRO from getting case 0 Strategy 1 PRO has no case feature Case assigners in general have case features that need to be deleted 0 The CP layer idea allows for PRO to appear in all the places where it should It also blocks other NPs or NPitrace from appearing in these positions 0 If PRO is inserted in a position where case is available it will fail to check the case feature of the case assigning head leading to ungrammaticality 0 Strategy 2 Null Case 7 PRO and only PRO has a special kind of case called Null Case which is assigned by the null C0 that appears with control in nitives o The C0 in the CP layer allows for PRO to appear in all the places where it should It also blocks other NPs or NPitrace from appearing in these positions Pronouns inserted here will get Null Case and will be realized as PRO 0 If a pronoun is inserted in a position where some other case is available it will come out with a noninull case ie it will not be pronounced as PRO One nice thing about the null case proposal is that the fact that case relationships cannot be estab7 lished across a CP boundary now falls out from the de nition of Agree 34 X Casel C null YPCasel 32 ECM and Raising The proposals for ECM and Raising now follow straightforwardly We have already ruled out the possibility of a CP complement We are left with the following structures which involve TP complementation 35 a ECMvAG V Tp NP to He believes Tp David to be innocent NP gets case from matrix VAC and raises to embedded SpecT0 for EPP reasons 10 b Raising Npg T0 VUNACC V 7 th t0 David seems Tp ti to be innocent NP raises to embedded SpecT for EPP reasons gets case from matrix T0 and raises to matrix T0 for EPP reasons Violations like Superraising and SuperiECM now follow from the de nition of Agree 36 a Superraising John seems that it appears to be happy Johni seems Cpthat Tpit appears Tp ti to be happy b SuperiECM He believes that it seems David to be happy John believes Cpthat Tp it seems Tp David to be happy We can also handle cases like the following if we assume that in English nite T must assign nominative and that case can only be assigned to an NP that doesn t already have case 37 Davidi seemed that ti left 4 Interactions between Passvization and Control ECM and Rais ing 41 Passivization and Control 411 Subject Control Predicates Most subject control predicates cannot be passivized This is known as Visser s Generalization 38 a It was preferred PRO to go b It was wanted PRO to go c It was tried PRO to go The ungrammaticality of 38 cannot be attributed to the inability of implicit arguments to con trol 7 we know that implicit arguments can control PRO The ship was sunk PRO to collect the insurance Visser s Generalization is brought out particularly nicely by the predicate promise Promise can function as a subject control predicate but also as a ditransitive verb 39 a He promised David PRO to stay b He promised David a new beginning Interestingly the in nitival complement taking promise cannot be passivized while the ditransie tive promise can be 40 a David was promised PRO to stay b David was promised a new beginning But Visser s F quot 39 is not an 39 quot 39 decid 39 an exception to it 41 It was decided PRO to leave 412 Object Control Predicates Object control predicates seem to passivize quite happily 42 a He was ordered PRO to leave b He was persuaded PRO to leave c He was permitted PRO to leave and so for instruct allow encourage The possibility of passivization is not surprising because these predicates satisfy general condi tions on passivization and the control relationship is not disrupted by passivization 42 Passivization and ECM Several ECM predicates can be passivized 43 a He believes David to be innocent b David is believed t to be innocent The embedded clause may also be passivized leading to an interesting range of interactions 44 a He believes David to have eaten the apple b He believes the apple to have been eaten t by David c David is believed t have eaten the apple 1 The apple is believed t to have been eaten by David Not all ECM predicates passivize 45 a David was wanted t to leave He wanted David to leave b David would have been preferred t to be less possessive of him I would have preferred David to be less possessive ofhim The failure of passivization and other interpretive properties of wan prefer have been used to argue that the in nitival complement of wantprefer might be a CP headed by a null for inspired by the idea that wan prefer can take for7CP complements But then we need to say something more to explain the ungrammaticality of the following 12 46 a It was wanted for David to leave b It would have been preferred for David to be less possessive of him The source of the ungrammaticality of 46 is mysterious but is probably related to the fact that 47 doesn t seem that great either 47 a It was wanted for David to leave b It would have been preferred for David to be less possessive of him Finally at least one ECM predicate that also takes for7CP complements can be passivized 48 a Makotoi was expected ti to win b I can t expect him to accept everything c I can t expect for him to accept everything The picture all of this suggests is as follows 49 a hope overt f0r7CPs covert f0r7CPs TPs b want prefer overt f0r7CPs covert f0r7CPs quotTPs c expect overt f0r7CPs covert f0r7CPs TPs d believe overt f0r7CPs covert f0r7CPs TPs 43 Passivization and Raising Raising predicates are all unaccusatives It is therefore unsurprising that they do not passivize 50 David is seemedappeared to be happy They can embed passive complements though 51 The articlei seems ti to have been written ti by David References Henry A 1992 ln nitives in a foreto dialect Natural Language and Linguistic Theory 102 2797301 Henry A 1995 Belfast English and Standard English Dialect Variation and Parameter Setting Oxford University Press Oxford Landau l 2003 Movement out of control Linguistic Inquiry 343 4717498 Postal P M and G K Pullum 1988 Expletive Noun Phrases in Subcategorized Positions Linguistic Inquiry l94 6357670
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'