On the Functional Properties of AGR

ON THE FUNCTIONAL PROPERTIES OF AGR Gemma &gau Uni e si a Au bnoma de Ba celona This pape is concemed wi h he p ope ies o he Ag eemen node: Pe son and Numbe . We analyze he ole ha each o hese elemen s plays in he sen ence. I will be claimed ha Pe son is he Nomina i e Case assigne , whe eas Numbe ma ks he mos p ominen a gumen o he p edica e. Consequen ly, numbe ag eemen be ween he e b and an a gumen does no necessa ily mean ha he a gumen ecei es Nomina i e Case. The da a we will discuss belongs o wo Ca alan dialec s No hwes e n and Cen al. I will be p oposed ha numbe ag eemen is au onomous wi h espec o pe son ag eemen in No hwes e n Ca alan. In Cen al Ca alan, howe e , lack o speci ica ion o Pe son en ails lack o speci ica ion o Numbe . O. In oduc ion Func ional elemen s ha e adi ionaily been conside ed he exp ession o he ela ion ha holds be ween he subjec and he p edica e in a clausal s uc u e. I has long been accep ed ha , in many languages, he unc ional ca ego y AG~UBJE~ may con ain wo elemen s: PERSON and NUMBER. In his pape , I analyze he ole ha each o hese elemen s plays in he sen ence. I will be claimed ha PERSON (PERS) is he Nomina i e Case assigne , whe eas i NUMBER (NUM) ma ks he mos p ominen a gumen o he p edica e. In he Ca alan exa nples (1 a,b) he p ominen a gumen is he ex e nal en Pe e: (1) a. En Pe e menja pomes 'Pe e ea s apples' b. En Pe e plo a a 'Pe e c ied' Ca alan Wo king Pape s in Linguis ics (CWPL) 1991: 235-260 Uni e si a Au bnoma de Ba celona VO o 10 mo emen allows he elemen PERS o o mally ela e he e b and i s ex e na1 a gumen ia Nomina i e Case assignmen , while NUM ag eemen mani es s he syn ac ic ela ion be ween he e bal elemen and he mos p ominen a gumen in he sen en ial s nic u e. This ela ion is implemen ed by checking NUM speci ica ion wi h ha a gumen . NUM ag eemen appea s o be independen o PERS ag eemen in some Ca alan dialec s. In o he dialec s, howe e , NUM and PERS ag eemen in e ac wi h each o he : lack o PERS speci ica ion esul s in lack o NUM speci ica ion.The da a discussed in his pape belongs o wo Ca alan dialec s: No hwes e n and Cen al. 1, Case Res ic ions on he Syn ac ic Na u e o Ve bal A gumen s My analysis will adop a se ies o hypo heses. Fi s , I assume ha AGR [-PERS] is no a Case assigne . In his I will ollow Kayne (1989: n 1). This p oposal is compa ible wi h Raposo's (1987) sugges ion ha o AGR o assign Case, i mus i sel ecei e Case om Tense. Second, Nomina i e and Accusa i e Case es ic he syn ac ic na u e o he ecep o a gumen , he same way ha Pa i i e assignmen does. A gumen s ha ecei e Pa i i e Case a e NPs, whe eas a gumen s ha ecei e s uc u al (Nomina i e and Acusa i e) Case a e DPs (see To ego (1983) and Belle i (1988)). I will claim ha an a gumen can only ecei e one Case. As opposed o Belle i (1988), I a gue ha ag eemen be ween an unaccusa i e e b and i s in e nal a gumen is no ob ained by he conjoined assignmen o inhe en Pa i i e a D- s uc u e and s uc u al Nomina i e a S-s uc u e. On he con a y, I will sugges ha a i a gumen ecei es ei he inhe en o s uc u al Case, bu no bo h. Thi d, AGR [+NUMI is no necessa ily ela ed o Nomina i e Case assignmen . Thus, he NP au obusos 'buses' in (2 a) and he cli ic en '(00 i / hem in (2 b) ecei e only Pa i i e Case: 236 (2) a. Dem& no ci cula an au obusos omo ow will no ci cula e buses 'Buses will no ci cula e omo ow' b. Demh no en ci cula an omo ow will no (o ) hema ci cula e The in e na1 a gumen au obusos in (2 a) is an NP. The e o e, i is no compa ible wi h s uc u al Case. Howe e , in cases like (3) below, he only ype o Case compa ible wi h he in e na1 a gumen o he unaccusa i e e b --i.e. els au obusos ' he buses -- is s uc u al (Nomina i e). This is so because i is a DP: (3) Demh no ci cula an els au obusos omo ow will no ci cula e he buses The buses will no ci cula e omo ow' 1 .I. Quan i ied A gumen s I has been p oposed ha a DP is incompa ible wi h Pa i i e Case, whe eas NP is incompa ible wi h s uc u al Case. Unde his assump ion, he g amma icali y o (4) below, needs o be accoun ed o . Should he a gumen es au obusos ' h ee buses' ha e an NP o a DP analysis?: (4) Dem& ci cula an es au obusos omo ow will ci cula e h ee buses Th ee buses will ci cula e omo ow' I should be no ed ha es au obusos can ha e ei he a speci ic eading o a nonspeci ic eading. The speci ic in e p e a ion can be pa aph ased as in (5): (5) Demh ci cula an es dels au obusos omo ow will ci cula e h ee o he buses Th ee o he buses will ci cula e omo ow' I will p opose ha es au obusos in (4) may ha e wo di e en analyses. In one, i is a quan i ied NP, wi h a s uc u e oughly ep esen ed in (6): (6) [QUANT es [~p au obusos]] h ee buses This s uc u e, which does no con ain a DP is assigned Pa i i e Case, and ecei es a nonspeci ic in e p e a ion ('some h ee buses o o he s'). The ph ase can al e na i ely be analyzed as a quan i ied DP, possibly wi h an emp y head, and a nume al in a speci ie posi ion. This analysis is in onnally ep esen ed in (7), and co esponds o he speci ic in e p e a ion. The a gumen will be assigned s uc u al Case: (7) [QUANT es [~p e [Np au obusos I]] h ee buses We will now discuss some da a in suppo o ou claim ha he a gumen in (4) can ecei e wo al e na i e analyses, and wo ypes o Case (s uc u al and inhe en ). Conside he ollowing: (8) a. T es au obusos sembla en espa lla s h ee (o he) buses seemed damaged b. Dues pomes s6n d'aquell pome wo (o he) apples a e om ha apple ee The de i ed subjec s ecei e s uc u al (Nomina i e) Case. Acco ding o ou p oposal, hey mus be analyzed as DPs, and hei s uc u al ep esen a ion should co espond o (7). No e ha he de i ed subjec s in (8) can only ecei e a speci ic in e p e a ion, i.e. ' h ee o he buses', ' wo o he apples' espec i ely. We should also no e ha he a gumen s ha ecei e Accusa i e om a ECM e b a e always DPs. Fo ha eason, he a gumen una aqua el.la 'a wa e colo ' in a sen ence like (9) below has always a speci ic in e p e a ion, a eading ha makes possible o pa aph ase una aqua el.la by 'a pa icula wa e colo ', 'a ce ain wa e colo ', o 'one o he wa e colo s': (9) Conside o mol imp essionan una aqua el.la (I) conside e y imp essi e a wa e colo Tu kish, which exp esses Case mo phologically, o e s us an in e es ing p oo o he ela ion ha exis s be ween he na u e o he e bal a gumen and he ype o Case ha i is able o ecei e. Acco ding o Eng (1991), in e na1 a gumen s showing a mo phological ma k o Accusa i e a e obliga o ily in e p e ed as speci ic in Tu kish. In e nal a gumen s wi hou an explici mo phological ma k o Case --i.e. he ones ecei ing Pa i i e--, a e in e p e ed as nonspeci ic. The ollowing examples a e om Eng (1991): (10) a. Ali bi ki ab-l al& Ali one book-Acc bough 'A book is such ha Ali bough i ' b. Ali bi ki ap al& Ali one book bough 'Ali bough some book o o he ' 1.2. The Cli ic ENINE A his poin , i is necessa y o b ie ly discuss he cli ic en Ine 'o i l hem'. I we a gue ha Accusa i e and Nomina i e Case a e only assigned o DPs, while claiming ha Pa i i e is assigned o NPs, we mus cla i y which ype o Case he cli ic en Ine may o mus ecei e. We will dis inguish be ween a Pa i i e en lm and a Geni i e en Ine. Acco ding o Belle i (1988), en lne ecei es Pa i i e. This is so in he example (2 b) abo e, whe e he cli ic p onominalizes he NP au obusos 'buses'. Howe e , no all cases o en Ine cli iciza ion ins an ia e he Pa i i e Case assigned by a e b, because en lne doesn' necessa ily ecei e Case om a e b. Conside he ollowing examples: ( 1 1) a. To s en eco dem algun, dels seus poemes all us (o ) hem emembe some, o hislhe poems 'We all emembe some, o hislhe poems' b. Vaig oba els eus amics i només en aig econiixe dos: en Pe e i en Pau (I) ha e me you iends and only (o ) hem ha e ecognized wo: Pe e and Pau 'I ha e me you iends and I ha e only ecognized wo (o hem): Pe e and Pau' He e, he cli ic p onominalizes a pa o he di ec objec . This objec is seman ically pa i i e in each case (see En$ (199 1)). The objec en..algun Ien. .dos 'CL.. somelCL.. wo' deno es a pa o a de ini e speci ic se (hislhe poemslyou iends). The e b in (1 1) assigns Accusa i e o he di ec objec , a quan i ied DP. Thus, he cli ic en lne does no ecei e Case om he e b, bu om he quan i ie . I is he same Case ha he cli ic en Ine ecei es om an N. Conside he ollowing examples om Ba a (1987): (12) a. Se me n' escapen els de alls CL ( o)me (o )i escape he de ails The de ails o i escape me' (12) b. N' han c ema o es les p o es (o )i ha e bumed all he e idence They ha e bu ned all he e idence' These examples show ha he p esence o en Ine does no imply ha he e b has assigned Pa i i e o i s in e na1 a gumen . The cli ic can exp ess Geni i e Case assigned by a quan i ie o a Noun. 2. The P ope ies o PERS and NUM in No hwes e n Ca alan We will now u n o discussing he p ope ies o PERS and NUM. Conside he sen ences exempli ied in (13) om No hwes em Ca alan (see Soli (1973), (1987)), and compa e hem wi h he ung amma ical (14) om he same dialec : (13) a. Fal a més b a~os is needed mo e a ms 'Mo e a ns a e needed' b. A iba pa en s a i es ela i es '(some) ela i es am e' c. Ha ingu més u is es has come mo e ou is s 'Mo e ou is s ha e come' d. Passa mol s co xes is passing many ca s 'Many ca s a e passing by' (14) a. *Fal en més b agos a e needed mo e anns (14) b. *Ambenpa en s a i e ela i es c. *Han ingu més u is es ha e come mo e ou is s d. *Passen mol s co xes a e passing many ca s The g amma ical examples in (13) show ha he e is no ag eemen be ween he a gumen and he e b. In hese examples he e bs a e unaccusa i e and able o assign Pa i i e o hei in e na1 a gumen , an NP, which is a ecep o o Pa i i e Case. These a e impe sonal sen ences, in he sense ha AGR is no speci ied o PERS o NUM. The No hwes e n Ca alan sen ences in (15) below a e also impe sonal (see Sola (1973), (1987)): (15) a. Hi ha mol s hbmens he e is many men The e a e many men' b. Enguany es plan a h mol s a b es his yea CL(=people) will-~lan 3,d/~i~~ many ees 'People will plan many ees his yea ' c. Plou ains 'I ains' The examples in (13) and (15) show a de ec i e e bal mo phology. The e b is in lec ed o 3 d/sing., which is he mo phological exp ession o absence o Pe son and Numbe in Ca alan. Ou analysis p edic s ha a sen ence can no be impe sonal i he in e na1 a gumen o an unaccusa i e e b is a DP. This p edic ion is bo ne ou : (16) Vind h 10 pa e will-come he a he The a he will come' The in e na1 a gumen , a DP, is incompa ible wi h Pa i i e Case. The only a ailable Case o his DP is Nomina i e, gi en ha he e b is unaccusa i e. AGR mus hen con ain [+PERS]. The No hwes e n Ca alan examples (17) show ha he mos p ominen a gumen (i.e. he in e nal DP) mus ag ee wi h he e b: (17) a. Vind an els pa en s will-c0me3~d/~l~~ he ela i es The ela i es will come' b. Els pa en s ind an a ia he ela i es will-~0me3~d/~l~~ soon c. * Vind A els pa en s will-come3,d/,ing he ela i es d . Nosal es a iba em a ia we will-come~ s ~p~u soon The ung amma ical sen ence (17 c) shows ha he in e nal DP equi es Numbe ag eemen also, when i is he mos p ominen a gumen . Nomina i e Case is assigned o he in e nal DP subjec in he g amma ical examples abo e when he DP unde goes mo emen o a posi ion whe e i can be go e ned by AGR [+PERS]. Le us assume ha his posi ion is [Spec,VP]. The ollowing ep esen s he abs ac con igu a ion o exa nple (16), whe e de ails i ele an o he p esen discussion ha e been ommi ed: e e en ial con en . P o is allowed in a gi en s uc u e i i ag ees wi h a unc ional ca ego y able o assign ea u es o p o. In (30), NUM ag eemen wi h he e b ia AGR will be enough o eco e and iden i y Accusa i e p o: (30) a. Les ines es ja es poden ob i he windows al eady SE may open 'The windows may al eady be opened' b. [[Les ines esli [ja es podenpiu ob i p oi]] The e a e o he ac s showing ha he Case assigned o he in e nal a gumen in (27) and (28) is no Nomina i e. The e idence can be ound in sen ences simila o hose showing SEARB: hose wi h unaccusa i e p onominal e bs. Sola (1987) no es ha he e is a con as in g amma icali y be ween (3 1 a) and (3 1 b) in No hwes e n Ca alan: (3 1) a. Amb aquelles ne ades es an en onsa o es les eulades del poble wi h hese snows SE ~ollapsed3,d/~l, al1 he oo s o he own Tha snow caused all he oo s o own o collapse' b. *Amb aquelles ne ades es a en onsa o es les eulades del poble wi h hese snows SE collap~ed3~d/~i,~ a11 he oo s o he own The p onominal e b en onsa -se ' o collapse' is unaccusa i e, he e o e i is unable o assign Accusa i e Case. The only (s uc u al) Case a ailable o he DP is Nomina i e. The e o e, hese ypes o sen ences a e pe sonal, like he g amma ical Jo m'en onso, u 'en onses, e c. 'I collapse, you collapse', e c. We will now discuss an in e es ing ac in Cen al Ca alan. The e is no NUM ag eemen when he in e nal a gumen o a sen ence wi h SEARB is a de ini e cli ic, as in (32 a) and (33 a) below. NUM ag eemen is, howe e , possible in (34), whe e he cli ic is Pa i i e (i.e. non de ini e): (32) a. No se us ap ecia gai e no SE youp~~~ app ecia q dlsing e y much 'Y ou a e no app ecia ed e y much' b. *No se us ap ecien gai e no SE youp~~~ app e~ia e3~dl~l~~ e y much (33) a. No se'ls ha ac a bé no SE hem ha q dlsing ea ed well 'They ha e no been well ea ed' b. *No se'ls han ac a bé no SE hem ha q d/,lW ea ed well (34) a. No se'n . eu, de um no SE EN(=o i ) sees, o smoke 'Smoke, i is no seen' b. No se'n euen, de mi acles no SE EN(=o hem) see, o mi acles 'Mi acles, hey a e no seen' The beha io o de ini e and non de ini e cli ics can be explained i we accep he hypo hesis ha de ini e cli ics a e hemsel es AGROBJECT (see Fe nández-So iano (1989), Co din (1990)). Hence, hey may no es ablish a new ag eemen ela ion. Tha is, he cli ics us /els 'youl hem' in (32) and (33) espec i ely may no exp ess AGRSUBJECT and AGROBJECT ag eemen a he same ime. Conside now he da a in (35), which belongs o a subdialec o Cen al Ca alan, and whe e he de ini e cli ic exp esses AGROBJECT. The Pas pa iciple ag ees wi h he cli ic in Numbe and Gende : (35) a. No se les ha ac ades bé no SE hem~~~ has ea ed~~~g~m well TheyFEM ha e no been well ea ed' b. No se 1' ha ac ada bé no SE he has ea ed~,~~~~ well 'She has no been well ea ed' Se les~~g cons uc ions in Cen al Ca alan show a s ong p e e ence o p o o e a de ini e cli ic when a [-human] DP is p onominalized. P o, as opposed o he cli ic la 'he ' es ablishes NUM ag eemen , because p o needs o ag ee in o de o be iden i ied: (36) a. Aques es copes, no es poden en a p o amb aigua calen a hese cups ,no SE can3,dlplu cleanp o wi h ho wa e These cups canno be cleaned wi h ho wa e ' b. ??I?~ques es copes, no se les po en a amb aigua calen a hese cups no SE hema Can3,dlsing clean wi h ho wa e 5. The P edica e HAVER-HI Impe sonal sen ences wi h he e b ha e -hi (li .:' o ha e- he eCL'; ' he e be') show a beha io simila o hose wi h SEARB: (37) a. No hi ha pa no he ec~ has b ead The e is no b ead' b. Hi ha ia un home he a had a man 'The e was a man The cli ic hi ' he e' seems o ha e been inco po a ed o he e b ha e ' o ha e' in he lexicon, o o m he p edica e ha e -hi. Fo me ly, ha e ' o ha e' exp essed possession in Ca alan. The possessi e meaning o ha e s ill emains in some cases: (38) a. Aques a noia, no 1' hau hs pas his gi l, no he c~ will-ha~e2~d,~i~~ NEG. This gi l, you won' ha e he ' b. Aixb cos a d'ha e his is ha d o o ha e Thls is ha d o ha e' In sen ences o he ype (37), he cli ic hi beha es like a loca i e subjec (see To ego (1989)). In o he wo ds, hi blocks he ex e na1 a gumen o ha e ' o ha e' when i inco po a es in o his e bal o m in he lexicon. Hi also blocks Nomina i e Case assignmen , p e en ing he ealiza ion o a subjec . E idence ha hi is no a ' ee' loca i e, bu an inco po a ed cli ic, is shown by he ac ha he cli ic may no be subs i u ed by a eal loca i e exp ession: (39) a. Hi ha ia un noi HI had a boy 'The e was a boy' b. *All& ha ia un noi ' he e had a boy' c. *Un noi ha ia a Banyoles a boy had in Banyoles The inco po a ed cli ic hi impe sonalizes he e b by bloclung PERS, which is blocking Nomina i e assigmen acco ding o ou p oposal. The e o e, i is impossible o ha e NUM ag eemen be ween he e b and i s only a gumen in No hwes e n Ca alan: (40) a. Hi hau & pocs hbmens (No hwes e n) HI will-ha~e3~dl~i~~ ew men The e will be ew men' b. *HI hau an pocs hbmens HI wiii-ha~e3~d/~l~ ew men c. Hi ha ia pas issos de xocola a HI had3 dlsing chocola e cakes The e we e chocola e cakes' d . *HI ha ien pas issos de xocola a HI had3 d/plu chocola e cakes Ou hypo hesis p edic s ha sen ences (40 b, d) will be well o ned in Cen al Ca alan, and his is he case. The e is NUM ag eemen be ween he a gumen and he e b, e en hough he cons uc ion is an impe sonal sen ence. This is an in e es ing ac , gi en ha he cli ic hi blocks PERS and NUM in he majo i y o Ca alan dialec s. I will sugges ha his phenomenon occu s because hi is inco po a ed in o he e b a he lexical le el, which is no he case wi h SEARB. This hypo hesis appea s o be con i med when one conside s Spanish equi alen cons uc ions. The appea ance o y (=hi ) blocks PERS and NUM: (41) a. Hay mosqui os has-Y mosqui oes The e a e mosqui oes' b. *Hany mosqui os ha e-Y mosqui oes I y is non o e , NUM may no be abso bed in Spanish. The ollowing sen ences a e bo h gmma ical: (42) a. Había mosqui os had3 d/sing mosqui oes b. Habían mosqui os had3 aplu mosqui oes The e we e mosqui oes' 6. NUM Ag eemen I ha e shown ha NUM ag eemen be ween he e b and an a gumen doesn' necessa ily mean ha he a gumen ecei es Nomina i e Case. Nomina i e is assigned by PERS, no by NUM. Howe e , when PERS is speci ied and assigns Nomina i e, i selec s NUM obliga o ily (i.e. (23 a)). NUM ag eemen ob ains be ween he e b and i s mos p ominen a gumen . The e o e, when he e is a Nomina i e a gumen , i will also ag ee wi h he e b in NUM ea u es. Hence, he ung amma icali y o (43) ollows om he ac ha he e b ag ees in NUM wi h he objec , bu no wi h he subjec : (43) *Els in an s p e e eix aques joc he child en p e e s his game Now he ques ion is he ollowing: Wha causes he p esencelabsence o numbe ag eemen be ween he e b and i s in e na1 a gumen in Ca alan dialec s when he sen ence is [-PERS]? We wan o p opose ha numbe ag eemen wi h an in e nal a gumen depends on how p edica ion is licenced a LF in each dialec . Conside again he ollowing examples: (44) a. Vind ii pa en s (No hwes e n) will-~omg~d/~i~~ ela i es b. Vind an pa en s (Cen al) will-come~ ~p~w ela i es 'Rela i es will come' These examples sa is y hema ic equi emen s and he in e nal a gumen sa is ies Visibili y equi emen s in (45 a, b) because i is assigned Case. Cen al Ca alan appea s o s ill equi e ano he condi ion o he sen ence o be in e p e ed as a p edica i e s uc u e. We sugges ha sen ence (45 a) can no be in e p e ed in Cen al Ca alan. When a p edica e selec s one o mo e a gumen s, one o hem has o be o e ly ma ked as p ominen in his dialec . P ominence is exp essed by NUM ag eemen be ween he ensed e b and he a gumen . As was said, NUM ag eemen depends on he hie a chical ela ions be ween PERS and NUM, bu also on he hie a chical ela ions be ween a gumen s. The hema ic hie a chy needs also o be ' isible' in Cen al Ca dan. A LF, he hema ic condi ion in (45) excludes (44 a) in Cen al Ca alan: (45) P ominence Condi ion When a p edica e selec s one o mo e a gumen s, one o he a gumen s has o be in e p e ed as he p ominen a gumen o he p edica ion. P ominence is exp essed h ough numbe ag eemen be ween he e b and he a gumen . The hema ic hie a chy is espec ed. This condi ion is always ac i e in Cen al Ca alan, e en in impe sonal ([-PERS]) sen ences. By con as , he No hwes e n a ie y does no equi e (45) in impe sonal sen ences. NUM ag eemen is only equi ed when an a gumen is assigned Nomina i e Case in No hwes e n. We mus ecall ha [+PERS] always selec s [+NUMI in Ca alan dialec s, and ha [-PERS] selec s [-NUMI in No hwes em. Sen ences wi h a 'sel -su icien ' e b, ha is, a e b ha doesn' need an a gumen o exp ess p edica ion --like plou e ' o ain --, will no mani es [+NUMI: (46) a. Plou [-PERS,-NUM] ains b. *Plou [-PERS,+NUMsing] c. *Plouen [-PERS ,+NUMplu ] Wha excludes sen ences (46 b, c)?. Recall ha posi i e speci ica ion o PERS and NUM mus be checked wi h an a gumen and appea su ixed o he e b. A sen ence will be ill- onned i [+NUMI is no checked, he same way ha he sen ence is ill- onned i [Tense] is no amalgama 4 o he e b. The unc ion o [+PERS] and [+NUMI is ha o ma king ag eemen , hence, hese elemen s neeú o be exp essed in a e b as well as in an a gumen . A sen ence like (47) will be speci ied [+PERS,+NUM]. The ac ha i may ha e an a bi a y in e p e a ion does no ollow om he unc ional elemen s in he sen ence, bu ollows om o he ac o s: (47) a. p o uquen [+P~S,+NUMJ a la po a ( hey) a e knocking on he doo b. p o ~~~~~~~[+PERs,+NuMI a la po a ( hey=someone) a e knocking on he doo 7. Conclusion Suppose ha he e a e wo possible ways o assign [PERS] (o Nomina i e Case) and NUM o an a gumen : by go e nmen , i he a gumen is in [Spec,VP]; o by Spec-head ag eemen . The la e is possible o DPs, bu no o NPs. Conside he schema a (48) and (49): IP n I' h I VP [+PERS] [+NUMI I has been claimed ha he ac ha a subjec ag ees wi h he e b does no imply ha i mus necessa ily be assigned Nomina i e. I has been ou con en ion ha PERS and NUM igge di e en ypes o ela ions be ween a subjec and he e b. E idence o his claim can be ound in he wo Ca alan dialec s we ha e discussed. I is possible, and emp ing, o p opose ha hese wo unc ional elemen s p ojec on o wo sepa a e syn ac ic ca ego ies. Howe e , he ques ion o whe he PERS and NUM cons i u e sepa a e X-ba p ojec ions o no is no c ucial a his s age o ou esea ch. He e, ou goa1 has simply been o acqui e a be e knowledge o he ac o s in e ening in ag eemen ela ions. This esea ch has been suppo ed by a DGICYT g an PB 89-0324, awa ded o he UAB. Some e sions o his pape we e ead a a Wo kshop on Cen al Romance Languages, Uni e si é de Gen& e, July 1990; a he P ime Coloquio de G amá ica Gene a i a, UAM Mad id, Ma ch 1991; and a he Uni e si A di Venezia, May 1991. I would like o hank hese audiences o hei ques ions and commen s. I am also indeb ed o Joan Sola and o my colleages a he UAB Linguis ics Semina , o Ca me Picallo in pa icula , o hei sugges ions o p e ious d a s o his manusc ip . Re e ences Ba a, A. (1987) 'Enca a n'hi ha més. (En o n de en i alguns SNs geni ius', Llengua i Li e a u a 2,377-427. Belle i, A. (1987) 'Los inacusa i os como asignado es de caso', in Sin axis de las lenguas ománicas V. Demon e and M.F.Lagunilla (eds), El A que o, Mad id, 167-230. See also Linguis ic Inqui y 19.1, 1-34. Co din, P. (1990) 'Da i e Cli ic Doubling and Visibili y o Case in T en ino', unpublished ms. Uni . di T en o. En$, M. (1991) The Seman ics o Speci ici y', Linguis ic Inqui y 22.1 1-26. Fe nández So iano, 0. (1989) Reccidn y Ligamien o en español: Aspec os del pa áme o del suje o nulo, PhD disse a ion, U.A.M., Mad id. Kayne, R. (1989) 'Face s o Romance Pas Pa iciple Ag eemen ', in Dialec Va ia ion and he Theo y o G amma , P. Beninca (ed), Fo is, Do d ech , 85-103. Picallo, C. (1991) 'El posesi o y el p incipio "E í ese el P onomb e"', pape ead a he P ime Coloquio de G amá ica Gene a i a, M a lo es de la Sie a, Mad id. Raposo, E. (1987) 'Case Theo y and In l- o-Comp: The In lec ed In ini i e in Eu opean Po uguese', Linguis ic Inqui y 18, 85- 109. Solh, J. (1973) Es udis de Sin axi Ca alanal2, Edicions 62, Ba celona Sol&, J. (1987) Qües ions Con o e ides de Sin axi Ca alana, Edicions 62, Ba celona.