CAS LX 522 Syntax Week 14b. PRO and control t lkely Th satfes the EPP n both clauses. The man clause has n Spec. The embedded clause has the trace n Spec. Th specfc nstance of A- movement, where we move a subect from an embedded clause a hgher clause generally called subect rang. DP V + V t A lkely AP DP Reluctance Reluctance Now, consder:. Th looks very smlar lkely. Can we draw the same knd of tree for t How many -roles does assgn Reluctant has two -roles assgn. One the one feelng the reluctance (Experencer) One the proposton about whch the reluctance holds () Leave has one -role assgn. To the one dong the leavng (Agent). n, what -role does get Reluctance n, dong the leavng, gets Agent from. showng the reluctance, gets Experencer from. And we have a problem: appears be gettng two -roles, n volaton of the -crteron. Reluctance. Reluctant assgns ts - roles wthn AP as requred, moves up Spec n the man clause by Spellout. But what gets the -role from, and what satfes the EPP for the embedded clause V + A 1
Reluctance. There must be somethng there, gettng the -role and satfyng the EPP. But we can t see t. t s a phonologcally empty (Ø) DP. We wll call t PRO. V + A Reluctance. There must be somethng there, gettng the -role and satfyng the EPP. But we can t see t. t s a phonologcally empty (Ø) DP. We wll call t PRO. V + A DP m PRO t m Reluctance [PRO ]. PRO does not get Case. * Bll. n fact, PRO cannot get Case. * for V + for Bll PRO refers (lke a pronoun or an anaphor). A DP m PRO t m f there s a PRO, how do we know [PRO m ] lkely [ ]. These two sentences look very much alke when faced wth a sentence that looks lke th, how do we know whch knd t f there s a PRO, how do we know Best method for fndng PRO: Count the - roles. f there appear be fewer arguments than -roles (n a grammatcal sentence), there must be a PRO. Another way try wth doms lke The cat out of the bag or The cat s got your ngue or The g up. doms For somethng have an domatc nterpretaton (an nterpretaton not lterally dervable from ts component words), the peces need be very close gether at the pont of orgnal Merge. t lkely that the g up. t lkely that the cat out of the bag. t lkely that the cat has your ngue. 2
doms t ok f the peces of the dom move away after ther orgnal Merge, we can stll get the domatc nterpretaton: [The cat] lkely have your ngue. [The cat] lkely be out of the bag. [The g] lkely be up. The mportant thng that they are gether orgnally (the -role needs be assgned by the predcate the noun) doms f we break up the peces, then we lose the domatc nterpretaton and can only get the lteral meanng. The cat thnks that out of the bag. The cat thnks that has your ngue. Wth PRO sentences ( control sentences ), we also lose the domatc readng. #The cat be out of the bag. #The cat attempted have your ngue. #The g tred be up. doms Control The reason for th that the domatc subect and the domatc predcate were never gether The cat [PRO be out of the bag] The cat attempted [PRO have your ngue] The g tred [PRO be up] Unlke wth rang verbs: [The g] lkely [ be up] PRO smlar a slent pronoun; t gets ts referent from somewhere outsde ts sentence. n many stuatons, however, PRO forced co-refer a precedng DP, unlke a pronoun. Bll thnks that he / a genus. Bll PRO /*. We say that PRO controlled (here by the matrx subect). Subect and obect control PRO arb There are actually two dfferennds of control verbs, those whose subect controls an embedded PRO and those whose obect does. Fnally, there a thrd use of PRO, n whch t gets arbtrary reference and means somethng lke someone/anyone. [PRO arb ] would be a mtake. Bll [PRO ] a subect control predcate John persuaded Bll [PRO ] persuade an obect control predcate The condtons on whch nterpretaton PRO can/must get are referred as Control Theory, although th day the underlyng explanaton for Control remans elusve. 3
Control theory For now, what control theory consts of ust markng the theta grds of specfc predcates (persuade, ) wth an extra notaton that ndcates when an argument a. Experencer Control theory Predcates that have a marked are control predcates. When the the external argument, t a subect control predcate, otherwe t an obect control predcate. Experencer persuade Agent Theme persuade Agent Theme k k The PRO conundrum Back when we talked about Bndng Theory, we sad that DPs come n one of three types, pronouns, anaphors, and R-expressons. PRO a DP, so whch knd t t gets ts reference from elsewhere, so t can t be an R- expresson. t sometmes forced gets referent from an antecedent, lke an anaphor and unlke a pronoun. But that referent outsde ts clause, meanng t can t be an anaphor (the antecedent would be o far away for Prncple A). Plus, t s not always forced (PRO arb ), lke a pronoun. The PRO conundrum Back when we talked about Bndng Theory, we sad that DPs come n one of three types, pronouns, anaphors, and R-expressons. PRO a DP, so whch knd t Concluson: t doesn t seem be any one of the three. t doesn t seem fall neatly under Bndng Theory hence, we need Control Theory deal wth the dtrbuton and nterpretaton of PRO. The PRO conundrum Control Theory These werd propertes of PRO are sometmes taken be the cause of another generalzaton about PRO (the PRO theorem ) PRO cannot get Case. That, PRO forbdden from any poston where Case would be assgned t (hence, t cannot appear n Spec of a fnte clause only a nonfnte clause) Despte the fact that PRO does not submt Bndng Theory, there are some bndngtheory-lke requrements on control of PRO. PRO only oblgarly controlled by a c- commandng. [Bll s mother] [PRO /* ] 4
PRO: One possble pece of support Let s thnk back Bndng Theory. Prncple A says that anaphors must be bound wthn ther bndng doman, and we take bndng doman be the clause. *Bll wants [ meet hmself] However, now consder: Bll buy hmself a gft. Bll promed buy hmself a gft. Why are these allowed PRO: One possble pece of support Bll [PRO buy hmself a gft] Bll promed [PRO buy hmself a gft] *Bll promed [PRO buy herself a gft] *Bll promed [PRO buy hm a gft] Bll promed [PRO buy her a gft] *Bll [PRO buy hm a gft] Whle t s true that Bll outsde of the bndng doman of hmself, and hence Bll cannot be the antecedent for hmself, PRO n the bndng doman and ts reference controlled. PRO: recap Although we can t see that PRO there, all of our theoretcal mechanms pont ts beng there. EPP says that clauses need a subect. The -crteron says that there must be exactly as many arguments as -roles. Bndng Theory ndcates somethng present nsde embedded clauses. f the rest of our theory rght, t seems that PRO must be there. talan subects Many languages have the property that when the subect undersod (often n the cases where n Englh we would use a pronoun subect), t can be ust left out entrely. For example, talan: Parlo. Parl. speak-1s speak-2s speak You speak talan subects So what about the EPP and the -crteron Clearly speak assgns a -role, and presumably the talan Spec needs be flled as well. Th sounds lke a famlar queston should we hypothesze that the subecn these sentences PRO Lttle pro There one mportant dfference between the talan null subect and PRO, namely the null subecn talan appears n a poston that gets Case. o parlo. speak-1s speak Snce PRO cannot appear n a Case-marked poston, we have take th be somethng smlar but dfferent: Lttle pro. 5
Lttle pro Lttle pro really ust a regular pronoun, only null. t doesn t have the fancy control propertes exhbted by PRO, t appears n Case-marked postons. Languages seem be dvded n those whch have lttle pro and those whch don t, often correlatng wth the amount of agreement on the verb (rch agreement makes t more lkely that a language wll have pro). Languages wth pro are often called prodrop languages or null subect languages. 6