Abs ac
Null Subjec Romance Languages a e languages in which he V ea u e o T is necessa ily s ong.
Taking so much o g an ed, and gi en he mo phological cha ac e is ics o his kind o languages,
all hei p ope ies ollow: V aising deposi s a T a se o [– in e p e able] φ- ea u es ha a e
e ased h ough co e aising by he [+in e p e able] φ− ea u es o he subjec DP, ei he pho-
ne ically ealized o p o (p o: a no a ional abb e ia ion o a se o ea u es composed by he ca -
ego ial ea u e D and φ- ea u es). In e ed subjec (DP) and null subjec (p o) a e, he e o e, P
in e nal subjec s. The p e e bal subjec , in con as , is a DP me ged a T — he e o e a non-a gu-
men — in a de i a ion wi h a p o as he in e nal subjec . S a ing om his basic p oposal, he
analysis o some new da a poin s o he spu ious cha ac e o Nomina i e Case in hese languages
and sugges s ha p o is only allowed in Spec, P.
Key wo ds: Null Subjec Languages, [–in e p e able] V- ea u es, de i a ional app oach, p o,
Nomina i e.
Resum. Un en ocamen minimalis a del pa àme e del subjec e nul
Les llengües de subjec e nul omàniques, com a a el ca alà, l’espanyol i l’i alià, són llengües que
enen el e V de T necessà iamen o . Donan això pe descomp a i a eses les ca ac e ís iques
mo ològiques de la conco dança e bal d’aques ipus de llengües, el conjun de p opie a s que
les ca ac e i zen se’n segueix au omà icamen : l’ele ació de V diposi a a T un joc de e s φ
[–in e p e able] que són esbo a s, mi jançan aslla no mani es , pels e s φ[+in e p e able] de
l’SD subjec e, onè icamen eali za o p o (p o: una ab e ia u a no acional pe a un a conjun
de e s compos del e ca ego ial D i e s φ). Subjec e in e i (SD) i subjec e nul (p o) són,
doncs, subjec es in e ns a l’S . El subjec e p e e bal, en can i, és un SD ajun a a T —I, doncs,
un no a gumen — en una de i ació amb p o com subjec e in e n. Aplican aques a p opos a bàsi-
ca a l’anàlisi d’algunes dades empí iques no es, s’apun a el ca àc e espu i del cas nomina iu en
aques es llengües i se sugge eix que p o només és pe mès a l’Espec d’S .
Pa aules clau: Llengües de Subjec e Nul, e s de V [–in e p e able], en ocamen de i acional,
p o, nomina iu.
* I am g a e ul o he audience o Semina i de G amà ica Teò ica o he Uni e si a Au ònoma de
Ba celona o hei commen s on a p e ious e sion o his pape . Thanks a e due specially o
Ca WPL 8, 2000 97-128
A Minimalis App oach o he Null Subjec
Pa ame e *
Joana Rosselló
Uni e si a de Ba celona. Depa amen de Lingüís ica Gene al.
G an Via de les Co s Ca alanes, 585. 08007 Ba celona.
ossello@ il.ub.es
Recei ed: No embe 20 h 2000
Accep ed: Desembe 14 h 2001
Ca WPL 8 001-137 29/5/2001 17:30 Página 97
1. In oduc ion
As a as I know, he e is ye no uly minimalis p oposal accoun ing o he basic
p ope ies displayed by Null Subjec Languages (NSLs),1namely hose s a ed in
(1). Such a p oposal should be able o combine wo achie emen s: on he heo y-
in e nal side, i should no appeal o mo e han one In l ca ego y and i should be
s ic ly de i a ional; on he empi ical side, i should accoun a leas o he h ee
p ope ies in (1) wi hou eso o any mechanism which is ex aneous o he (human)
language Compu a ional Sys em (CSHL).
(1) a. P e e bal Subjec
En Joan balla a.
he Joan dance-PAST-3SG
‘Joan danced.’
b. Pos e bal Subjec
Balla a en Joan.
dance-PAST-3SG he Joan
‘Joan danced.’
c. Null Subjec
Balla a.
dance-PAST-3SG/1SG
‘He/she/I danced.’
Le us begin by conside ing how he cu en accoun s ail o achie e hese wo
goals. On he one hand, i has o be poin ed ou ha he e is no p oposal being able
o explain he p ope ies in (1), his ailu e being due ei he o omission —as in
Ba bosa (1997), Manzini & Sa oia (1997), O dóñez & T e iño (1999) and Pollock
98 Ca WPL 8, 2000 Joana Rosselló
Jaume Solà o his in aluable help in he p epa a ion o he English e sion. The esea ch ha
unde lies his wo k has been suppo ed by g an PB96-1199-C04-02 om he Minis e io de
Educación y Cul u a and by g an 1999SGR00113 om he Gene ali a de Ca alunya.
1. I wan o poin ou ha in his pape he exp essions «subjec », «p e e bal subjec », «Null Subjec
Pa ame e » e c. a e used in a a he in o mal way, and ha by «Null Subjec » languages (NSLs) we
in end hose in which he non- ealisa ion o he subjec as a phone ically o e DP is possible
because o e bal ag eemen . We he e o e exclude Chinese-s yle NSLs, and include bo h NSLs
wi h uni o m ag eemen (NSLsUA) —such as Ca alan, Spanish o I alian— and NSLs wi h al e na e
ag eemen (NSLsAA) —such as A abic, Cel ic languages, e c.
Table o Con en s
1. In oduc ion
2. The e bal φ- ea u es in NSLs
3. Towa ds a minimalis app oach
4. The EPP ea u e o T is no uni e sal
5. A minimalis app oach o he Null
Subjec Pa ame e
6. Impossible p e e bal (o null)
subjec s
7. Conclusions
Re e ences
Ca WPL 8 001-137 29/5/2001 17:30 Página 98
(1997)—, o o he conclusion ha he p ope ies in ou iad (1a)-(1c) a e no de i -
able om he mechanisms o CSHL —as in Picallo (1998) and Dona i & Tomaselli
(1997). So, in Picallo (1998), since he subje p eposing ope a ion ha would yield
(1a) cons i u es an economy iola ion when compa ed wi h (1b), i could no be
he esul o ope a ions d i en by CSHL o sa is y ea u e checking, bu he esul
o a discou se-d i en ope a ion.2
On he o he hand, we a e dealing wi h app oaches which a e nei he com-
ple ely no uly de i a ional: he e is o en a mix u e o ep esen a ional and de i a-
ional c i e ia. So, o ins ance, when i is said ha he p e e bal subjec is disloca ed
(in he sense o sen ence-ex e nal), i emains a mys e y how ha kind o subjec has
been de i ed and, hus, whe he i en e s ea u e checking, i i has o. Mo e gene ally,
he lack o explici ness abou he de i a ions unde lying each cons uc ion ex ends
o all he membe s o he iad.
Finally, i seems wo h men ioning ha in almos all he e e ences on his opic
appeal is made o mo e han one in lec ional ca ego y. So, some o hem posi
AGRP and TP (Alexiadou & Anagnos oupoulou (1996, 1998), Benedic o (1993),
Nash & Rou e e (1996), Pla zack (1994), Dona i & Tomaselli (1997)); and some
o he s posi wo IPs wi h di e en ea u es (Manzini & Sa oia (1997), Ca dinale i
(1994, 1997)).
Taking se iously he s a e o a ai s his su ey shows, i appea s as mo e han
jus expedien o pa e he way o an accoun acco ding o he minimalis guide-
lines ci ed abo e. In pa icula , I will de elop a sys em in which he [–in e p e able]
ea u es play a key ole in he de i a ion.
In wha ollows, I will show ha he igge o he p ope ies lis ed in (1a)-
(1c) is he bunch o [–in e p e able] φ− ea u es o he ensed e b in NSLs (sec-
ion 2). Then (sec ion 3), I will p esen he basic guidelines o my p oposal. One o
i s esul s (sec ion 4) conce ns he non-uni e sali y o he EPP, which I pu sue,
while a guing, howe e , agains he cu en p oposals acco ding o which he e is
no Spec,T posi ion in NSLs wi h Uni o m Ag eemen (NSLUA) —see no e 1.
Agains hese iews, a new se o da a is p esen ed which obliges us o be indul-
gen wi h he EPP in i s classical e sion, his is o say, as obliga o y p esence o
Spec,T. A meaning ul and, un il now, unno iced con as be ween English and
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 99
2. Picallo (1998) assumes ha he same nume a ion unde lies (1a) and (1b). As we will see (§ 4.1),
i can be a gued ha his assump ion is mis aken. Dona i & Tomaselli (1997) conclude ha nei-
he he Ex ended P ojec ion P inciple (EPP), no he Null Subjec Pa ame e can be de i ed unde
he elemen a y mechanisms o CSHL. The ac ha hei wo k c ucially elies on wo in lec ional
heads —a handicap o be a oided in he p esen p oposal— allows us o lea e i aside wi hou ha -
ing o deal wi h he a gumen s ha lead hem o such a conclusion. Ne e heless i mus be no iced
ha Dona i and Tomaselli aim a explaining he syn ac ic beha iou o subjec s in English, F ench
and Ge man; p o-d op languages —ou NSLUA—, and VS languages such as I ish —an ins ance
o NSLAA.
Unlike all he p oposals men ioned so a , Ka o’s (1999) does no incu any o hese p oblems.
Bu i assumes ha e bal ag eemen is ca ego ially a D(e e mine ) and, as such, an independen
lexical i em which en e s he de i a ion as a ue a gumen . The p oblems wi h such an app oach
will be add essed la e in he ex (§ 2, 4.1).
Ca WPL 8 001-137 29/5/2001 17:30 Página 99
Ca alan is o e ed ha shows i is mis aken o assume ha , beyond he ac ha
English has no p o, in bo h languages he e a e p e e bal subjec s (a gumen s o
exple i es) o he same kind. In he ollowing sec ion (sec ion 5), he basic p o-
posal al eady ske ched in sec ion 3 is p esen ed in a mo e explici way and compa ed
wi h he classical ones, namely hose assuming ha Spec,T is always illed.
Me ge wi h T (o abo e T) yielding he p e e bal subjec in NSLs is also exam-
ined. Finally (sec ion 6), wo almos unno iced con as s be ween p e e bal and
pos e bal subjec s lead me o explo e wo issues: he possibili y ha Nomina i e
Case is spu ious in NSLs, and he cons ained dis ibu ion o p o, ha would only
appea in Spec, P. Such an app oach o Nomina i e Case yields a new insigh in o
he concu ence o Pa i i e Case and ag eemen (wi h he e b) and makes i pos-
sible o es ablish a clea sepa a ion be ween cli ics and ag eemen a ixes.
To conclude his in oduc ion, a wa ning no e is in o de . Gi en he limi a ions
o his pape , i would be impossible o deal wi h wo mo e issues ha cons i u e,
in ac , he backg ound o my p oposal. The i s one, ha I ake o g an ed, is he
le -disloca ed s a us o p e e bal subjec s in NSLs (see Solà 1992, O dóñez &
T e iño 1999, Alexiadou & Anagnos oupoulou 1998), al hough in he accoun o
be p esen ed he e i simply means ha he p e e bal subjec is a non a gumen ha ,
in a simila way o exple i es, me ges in a non-θ-posi ion, in ac , o en in he same
posi ion he θ-subjec aises o in non-NSLs (see sec ion 5). The second issue is
ha he pu po ed e idence ha p o is p e e bal (in he sense ha i appea s in
Spec,T) —see, o ins ance, Rizzi (1990), Ca dinale i (1994, 1997)— is much less
compelling han i is usually assumed o be: pa allelism be ween sen ences wi h
p o as subjec and sen ences wi h a lexical p e e bal subjec is also expec ed in a
amewo k whe e lexical (and non ocused) p e e bal subjec s a e no hing bu sen-
ences wi h a null subjec (p o) in θ-posi ion and an (op ionally added) le -dislo-
ca ed subjec , in ac a non-a gumen me ged in a non-θ-posi ion. Wi h hese wo
ema ks in mind, we hence o h concen a e on he p oposal o accoun o he
p ope ies in (1a)-(1c).3
2. The e bal φ- ea u es in NSLs
The p esen p oposal, as all hose in he pas conce ning NSLs, ela es he p ope -
ies in (1) o he « ichness»4o e bal ag eemen . So a , he e o e, no hing is new.
Since Ta aldsen (1980) and Rizzi (1982), all he p oposals eso , in simila ways,
o « ich ag eemen ». In he REST amewo k (Chomsky, 1981), his guiding idea
was leshed ou as INFL ha ing he ea u e [+p onominal]. As a as I know, how-
100 Ca WPL 8, 2000 Joana Rosselló
3. I lea e ou o he p esen pape showing how he beha iou o copula i e sen ences in NSLsUA,
can be conside ably cla i ied wi hin he amewo k ad oca ed he e. As is well known, his kind
o cons uc ion shows ag eemen o he igh :
(i) Sóc jo.
am I
‘I ’s me.’
4. On how ichness is o be unde s ood, see Ka o (1999), who sys ema ises p e ious p oposals.
Ca WPL 8 001-137 29/5/2001 17:30 Página 100
e e , he e is as ye no p oposal in which he [–in e p e able] e bal ea u es (num-
be , (gende ), (pe son)), when so cha ac e ized, play he ole hey a e assumed o
play in he wo kings o NSLs, i is o say, in igge ing displacemen . Thus, Ka o
(1999) and O dóñez & T e iño (1999), o ins ance, gi e a gumen s a us o he
a ix con aining he [–in e p e able] ea u es o V, ins ead o conside ing i a eply
o he [+in e p e able] φ- ea u es o he co esponding DP a gumen . In his way,
he φ- ea u es o V become, ipso ac o, [+in e p e able]. Alexiadou &
Anagnos oupoulou (1998), and Pollock (1997) do no see he a ix as an a gumen ,
bu hey also assume (al hough no explici ly) ha he φ- ea u es o V a e [+in e -
p e able]: hey mus indeed be when he a ix is aken o belong o he ca ego y o
p onouns, as hey p opose (see Alexiadou & Anagnos oupoulou 1998: 517). Since
hey conside he ag eemen a ix a p onoun wi h φ- ea u es, i u ns ou ha V
aising o T checks no only he V- ea u e o T bu also he D- ea u e o T (EPP),
i.e., he la e checking akes place by adjunc ion o he head (T) ins ead o subs i-
u ion in Spec,T.
Independen ly o he empi ical p oblems ied wi h such iews —see §6—, i
mus be poin ed ou ha on he heo y-in e nal side i is indeed impossible o hink
o φ- ea u es o V as [+in e p e able]. The eason is ha o achie e his in e p e-
a ion one mus ejec he S ong Lexicalis Hypo hesis (SLH), which seems o be
inhe en o Minimalis P og am (MP). Indeed, acco ding o SLH, he a ay o lex-
ical choices p o ided by he nume a ion a e lexical i ems al eady in lec ed ha
p o ide each choice wi h an unique ca ego y.5This being so, he e is no way o
escape he conclusion ha φ- ea u es o V a e [–in e p e able]: o achie e he oppo-
si e conclusion i would be necessa y o assume ha hese ea u es a e in a a gumen
and his, in u n, would lead o in oducing V and he ag eemen a ix sepa a ely,
which is jus wha he SLH p ohibi s.
The e a e, mo eo e , gene al g ounds o cas ing doub s on he accoun s ha
g an in lec ional a ixes (cli ics a e o be kep apa ) he s a us o an a gumen .
We e hey a gumen s, i should be explained why Al e na e Ag eemen (AA) lan-
guages (see no e 1), as A abic o Cel ic ones, wo k as de ined by G eenbe g’s
Uni e sal 33 and no con e sely. Le us see why. Uni e sal 33 s a es ha «When
numbe ag eemen be ween he noun and e b is suspended and he ule is based on
o de , he case is always one in which he e b is in he singula .» This s a emen ,
ansla ed o ou e ms, amoun s o saying ha when he ag eemen be ween e b
and subjec is suspended, i is he e b ha loses (all he) φ- ea u es, whe eas he
φ- ea u es o he subjec DP emain unchanged. No e ha i he in lec ional a ix was
an a gumen , one could expec languages suspending he ag eemen be ween e b
and subjec he o he way ound, ha is, ha ing he e b wi h all he φ- ea u es and
he DP wi hou . As an illus a ion, we should ha e languages wi h he pa e n
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 101
5. No ice ha o Pollock (1997), Alexiadou & Anagnos oupoulou (1998), e c., he e bal o m would
no only be V, bu also D. This in e p e a ion is, in my iew, incompa ible wi h he SLH, since i
would make a ailable o compu a ion he lexical, ca ego ial, inne pa s o wha in syn ax coun s
as a single ca ego y —conceding i is legi ima e, in he i s place, o pos ula e he D ca ego y on
he basis o me e φ- ea u es.
Ca WPL 8 001-137 29/5/2001 17:30 Página 101
Wo k-3PL he man meaning ‘The men wo k’, ins ead o he exis ing pa e n in his
kind o AA languages, namely Wo k-∅ he men. As a as I know, his beha iou
is no a es ed ac oss languages. We conclude, he e o e, ha he φ- ea u es which
en e in o he subjec e b ag eemen a e p ima ily ea u es o he DP subjec (in
which hey a e [+in e p e able]) and, he e o e, ha i is he e b indeed ha ag ees
wi h he subjec , and no he subjec whi he e b, as e e s a ed in good adi-
ional g amma .6
3. Towa ds a minimalis app oach
Gi en he p e ious heo y-in e nal conside a ions and his las empi ical obse -
a ion, i seems ha one can ake as well ounded he main assump ion o he p e-
sen p oposal, namely he [–in e p e able] na u e o he φ- ea u es o V. Now, i V
emains in he posi ion in which i is me ged ini ially, hese ea u es will no be
able o a ac he φ- ea u es o SUBJ in he Spec, P basic posi ion, simply because
he e SUBJ s ands abo e V. I is hen necessa y ha V aises o e ly, as in ac i
does in all NSLs.7We hen ha e a s ong V ea u e o T and a se o [–in e p e able]
φ- ea u es on he e b, aising along wi h i o T. Wha e ases hem?8In he p e-
sen p oposal —whe e he p e e bal subjec , as we will show, is no a ue a gu-
men —, e asu e o such ea u es has o be accomplished co e ly by he φ- ea u es
o he so called pos e bal subjec ( he ue in e nal subjec in he case o une ga-
i e o ansi i e e bs, o he DP sis e o V in unaccusa i e e bs). Hence, such an
102 Ca WPL 8, 2000 Joana Rosselló
6. The e is a ce ain idealisa ion in his a gumen a ion, i we conside cases like:
(i) Ho hem p epa a els ajudan s.
i -ha e-1PL p epa ed he assis an s
‘We he assis an p o esso s ha e p epa ed i .’
(ii) Els nens enca a no ho sabeu.
he child en s ill no i -know-2PL
‘You child en don’ know ye .’
o he examples in (18). In ac , in hese examples he e is mo e in o ma ion in he e bal ending
(ex a pe son in o ma ion) han in he igge ing DP. Wha e e he p ecise analysis o hese cases,
hough, we can easonably main ain he a gumen in he ex , howe e quali ied, on he g ounds
ha a misma ch in numbe (V-plu al / DP-singula ) is ne e a es ed.
7. I can be aken as an empi ical ac ha in NSLs V aising is o e . This is shown by, among o he
ac s, ad e b dis ibu ion (La Ma ia a so in la migdiada, he Ma ia does o en he sies a). Is ha ,
hough, wha we should expec , conside ing he [±in e p e able] s a us o he ea u es en e ing
he compu a ion? I seems he answe is yes. Indeed, i we assume, as seems easonable, ha he
V- ea u e o T is [+in e p e able], a e b which has no aised o e ly would be le wi h no [–in e -
p e able] a ac o o ge aised co e ly, so ha i s [–in e p e able] φ- ea u es could no possibly be
e ased and he de i a ion would c ash a he C-I in e ace. I would ollow, hen, ha he p esence
o e bal ag eemen in NSLs mus be associa ed o a s ong V- ea u e o T. Following his line o
easoning, we could e en conclude ha a ibu ing s eng h o a/some ea u es(s) is no anymo e a
me e s ipula ion, bu a he a consequence o he in e play be ween he [±in e p e able] ea u es
wi hin he sys em. See oo no e 19 o u he in o ma ion.
8. Chomsky (1998) a ibu es φ- ea u es o T i sel . See, howe e , he ollowing discussion in he ex
and oo no e 19.
Ca WPL 8 001-137 29/5/2001 17:30 Página 102
e asu e is d i en by ba e a ac ion o he φ- ea u es o he DP, ha adjoin o he
[T V+φ-T] head, wi hou pied-piping o he DP.
I will now y o lay he ounda ions o he p esen accoun by showing he
sho comings o an al e na i e way o e asing he φ- ea u es o V. This possibili y,
explo ed in Chomsky (1998), assumes ha T is supplied i sel wi h φ- ea u es along
wi h V and D/N- ea u es. Then, depending on whe he he V ea u e o T is (a)
s ong o (b) weak, we will ha e he ollowing. I (a), he φ- ea u es o V could be
e ased as ee ide s when V is a ac ed by he s ong V- ea u e o T. Na u ally,
also he φ- ea u es o T would be e ased in his way. I (b), and i he D- ea u e o
T is s ong (EPP), he [–in e p e able] φ- ea u es o T would be e ased by he co -
esponding [+in e p e able] φ- ea u es o DP, aised as ee ide s along wi h he
DP a ac ed by he s ong D- ea u e o T. The [–in e p e able] φ- ea u es o V
would emain ye o be e ased, and would ha e o aise as ee ide s along wi h
he V ea u e, in an ins ance o pu e ea u e- aising, wi hou pied-piping o he whole
ca ego y. Once in T-adjoined posi ion, hey would inally be e ased by he φ- ea u es
o he DP in Spec,T.
So a , (b) — which could apply o languages like English— would yield he
in ended esul , namely he e asu e o he φ- ea u es o V by means o he co e-
sponding ones in he DP in Spec,T. S ill, he e would be an appa en ly supe lu-
ous ope a ion: wha would he checking (and e asu e) o he φ- ea u es o T be
used o ?
In con as , in case (a) —which would apply o any language wi h o e V ais-
ing—, he esul u ns ou o be unwelcome, independen ly o he s eng h o he
D- ea u e o T, since, e en in he case ha D is s ong, he Ex ension Condi ion
equi es ha he aising o V p ecedes he o e aising o he DP subjec , and so
he checking (and e asu e) o φ- ea u es o V (and T, in ac ) akes place in he
con igu a ion o adjunc ion o he head T. Indeed, his esul is unsa is ac o y no
only wi h espec o he in ended goal (only he ea u es o a DP should be able
o check he associa ed φ- ea u es o V) bu also wi h espec o he empi ical e i-
dence: by allowing e asu e o φ- ea u es o V by he φ- ea u es o T, we should
expec , con a y o ac , any ea u e misma ch be ween e b and subjec o be pos-
sible.9
Gi en his empi ical sho coming, i seems ha he assump ion ha T has i s
own φ- ea u es mus be u ned down, and ha we can keep o he idea ha T has only
V- and D/N- ea u es. So φ- ea u es would always be a p ope y o lexical i ems.
Upon e lec ion, posi ing φ- ea u es o T is eminiscen o posi ing an AGR ca e-
go y, which, as is well known, was inally elimina ed because i was no in e -
p e able a he C-I in e ace and mo eo e consis ed solely o φ- ea u es and s eng h
(see Chomsky, 1995, § 4.10). Fu he mo e, i should be no iced ha , by elimina -
ing he possibili y o φ- ea u es in T, case (b) beha es as desi ed: he φ ea u es o
DP will check (and e ase) he associa ed φ- ea u es o V wi hou eso o he supe -
luous ope a ion men ioned abo e.
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 103
9. See no e 6.
Ca WPL 8 001-137 29/5/2001 17:30 Página 103
One u he mo e can specula e on he ollowing lines: concep ually, i is no
e y elegan ha [–in e p e able] ea u es a e checked (and e ased) by also [–in e -
p e able] ea u es. Keeping o he minimalis guideline acco ding o which [–in e p e-
able] ea u es a e esponsible o displacemen , i seems mo e desi able ha [–in e -
p e able] ea u es a ac [+ in e p e able] associa ed ea u es.10 In any case, he
p esen p oposal adhe es his kind o iew.
To summa ize, he φ- ea u es o V can only be e ased by he ea u es o a DP.
In pa icula , in NSLsUA hese ea u es a e e ased by he co esponding DP a gu-
men (ei he o e o p o) wi hin P. So e asu e is always accomplished by ba e
ea u es, as an ins ance o head adjunc ion wi hou pied-piping. In ac , his is
s aigh away sugges ed by (1b), i.e. by he possibili y o a pos e bal subjec , and
is equi ed o de i e he null subjec (p o) sen ences like (1c). In e ec , on mini-
malis assump ions o aise p o o e ly in an ins ance o subs i u ion in Spec,T is no
a all a sound mo e, since pied-piping applies only when i is equi ed by he A-P
in e ace. Keeping o minimalis guidelines, p o has o be conside ed a se o ba e
ea u es, D and φ- ea u es.
So a , he possibili y o an exple i e p o (p oEXPL) has no been ye uled ou .
Such an elemen , i exis ing, would ha e o me ge in he in lec ional phase, gi en
i s non-a gumen na u e. Tha p oEXPL does no exis will be a gued o in he nex
sec ion. Suppose now ha his a gumen a ion is co ec . In his case, a sen ence
wi h a pos e bal subjec would yield an EPP iola ion (EPP unde s ood as check-
ing/e asu e o D- ea u e o T by means o a DP in Spec,T); u he mo e, a i s
glance i would seem ha in his way we p eclude (as Picallo 1998 does) he pos-
sibili y ha he ba e wo kings o CSHL can accoun o he double possibili y con-
ce ning he subjec posi ion, ei he p e e bal o pos e bal. Wi hou p oEXPL he
nume a ion unde lying sen ences wi h a pos e bal subjec would be le wi h
he same lexical i ems as, in his amewo k, he nume a ion unde lying sen ences
wi h a p e e bal subjec . And, wha is wo se, in ac one could ne e ob ain a p e-
e bal subjec since, as sen ences wi h pos e bal subjec s show, he e is no hing
compelling he subjec o aise.
Ne e heless, such a de ea ing conclusion is no wa an ed. Le us see why. In
he i s place, he disappea ance o p oEXPL does no imply as a inescapable con-
sequence ha he nume a ion o a sen ence wi h a p e e bal subjec mus be iden-
ical o ha eeding a sen ence wi h a pos e bal subjec . To ob ain wo di e en
nume a ions wi hou eso ing o p oEXPL, i su ices o ein e p e , in minimalis
e ms, he empi ically well g ounded assump ion ha in NSLs he so called p e-
e bal subjec is in ac le -disloca ed, hus associa ed o a sen ence-in e nal p o,
104 Ca WPL 8, 2000 Joana Rosselló
10. Ce ainly, e asu e o Case as a [–in e p e able] ea u e bo h in nominal ecei e s (DP/NP) and in
assigne s (T o Nomina i e; V o Accusa i e —and o he inhe en Cases) does no i his pa -
e n. We could a gue, hen, ha he essen ial poin in ou p oposal is ha he e mus be an asym-
me y be ween he wo membe s o he checking ma ch, one op ion being ha one such an asymme y
is [–in e p e able]/[+in e p e able]. Ano he op ion, ele an o Case, would be he Case-asign-
e /Case- ecei e asymme y. This way o p ese ing he asymme y becomes impossible in Chomsky
(1998), whe e s uc u al Case in nominals is aken o be he e lec ion o a se o [–in e p e able]
φ- ea u es.
Ca WPL 8 001-137 29/5/2001 17:30 Página 104
whe eas he pos e bal subjec is in i s θposi ion. This ein e p e a ion could e y
easily adop he ollowing o m: he p e e bal subjec is indeed a non a gumen
DP me ged abo e he θdomain, ei he in he In lec ion phase (like a ue exple-
i e) o abo e, in he Complemen ize phase (like a ue le -disloca ed elemen )
—see oo no e 19. I i is like an exple i e, i has o ha e an associa e ha is he
ue a gumen . In his case, howe e , he associa e is no lexical, bu a p o occu-
pying a θposi ion in P. No ice ha unde his iew a sen ence wi h a p e e bal
subjec mus ha e a p o in he nume a ion, whe eas a sen ence wi h a pos e -
bal subjec canno ha e any. Bo h p o (ei he associa ed wi h a p e e bal subjec
o no —in a null subjec sen ence) and lexical pos e bal subjec appea in hei
θposi ion. To conclude his i s ough app oxima ion o he p esen p oposal, i
is wo h no icing ha i has a clea ad an age o e p e ious ones assuming he le -
disloca ed na u e o p e e bal subjec in NSLs. In e ec , unlike i s p edecesso s, he
p esen p oposal need no eso o addi ional s uc u e o s a e he non a gumen al
na u e o he p e e bal subjec . Unde he de i a ional iew o MP, his na u e can
be s a ed h ough he de i a ion i sel .
In acco dance wi h wha we ha e jus seen, i appea s ha i would ha e been
he w ong mo e o conclude ha he in e play be ween p e e bal and pos e bal
subjec in NSLs can no ollow om he e y wo kings o CSHL. Mo eo e , i is
in ac a su p ising ou come ha e en in his amewo k he idea ha he p e e -
bal subjec is somehow ex aneous o CSHL can be exp essed. Upon ca e ul con-
side a ion, we will see ha al hough he p e e bal subjec en e s he de i a ion as
an exple i e (when me ged wi h T, a leas ) i does no ully wo k as such: an exple-
i e has o check he D- ea u e o T, while ou p e e bal subjec does no . The
checking/e asu e o he φ- ea u es o V and o he ca ego ial D/N- ea u e o T is
always achie ed co e ly by he ea u es o he DP/NP pos e bal subjec . In his
connec ion, we ha e o bea in mind ha when in he cou se o he de i a ion we
de i e T, no hing mo e is needed in NSLs: i espec i e o he addi ional me ging o
ou p e e bal (exple i e) subjec , he de i a ion would con e ge. In ligh o his, i
seems sensible o conclude ha a his poin , T (T0) (in NSLs), has all he ea u es
equi ed o he de i a ion o con e ge. And he ul ima e eason o his conclu-
sion is o cou se ha in NSLs he e is an a gumen al p o, as sugges ed by plen y o
e idence. I is a nice esul , as I ha e jus said, ha he op ional na u e o he p e-
e bal subjec can be accoun ed o by he e y mechanisms o CSHL.
I could be a gued, howe e , ha he a gumen jus de eloped is allacious
because in NSLs no only he p e e bal subjec is op ional bu also he pos e bal
one. So i seems a i s glance, a leas o NSLsUA. Fo NSLsAA, ins ead, he a gu-
men s ands, since in hese languages he me e d opping o a pos e bal subjec
would lea e V dep i ed o φ- ea u es (acco ding o U33 o G eenbe g) and hus
unable o legi ima e a null subjec (p o). Should we conclude ha he a gumen in
he p e ious pa ag aph can apply only o NSLsAA? No a all: he a gumen , con-
a y o appea ances, is also alid o NSLsUA. I is a ac , which has gone unno-
iced, ha in NSLsUA a pos e bal subjec can no always be le ou , whe eas a
p e e bal one always can. Mo e speci ically, he e a e sen ences ha become
ung amma ical when dep i ed o hei s pos e bal subjec , whe eas no hing pa -
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 105
Ca WPL 8 001-137 29/5/2001 17:30 Página 105
5. A minimalis app oach o he Null Subjec Pa ame e
The pu pose o his sec ion is o cha ac e ize, as succinc ly as possible, he
essen ial plo o he hypo hesis de ended he e, so as o compa e i wi h he essen-
ial plo o he (by now classical) p oposal, acco ding o which NSLs a e jus like
English, excep o he possibili y o pos e bal subjec s and he a ailabili y o a
phone ically null p onoun.20 In o de o es ablish he compa ison mo e clea ly, I
will make use o a minimally o malised p esen a ion.
5.1. Two concep ions. A compa ison
The idea I am pu ing o wa d, as implemen ed in he amewo k o Chomsky
(1995), would be he ollowing:
(9) a. P e e bal subj. [TP DP [T 0max FFS– V+φ–T] [ P p oS ...] ]
Ex.: En Joan balla a.
he Joan danced
b. Pos e bal subj. [TP [T 0max FFS– V+φ–T] [ P DPS …] ]
Ex.: Balla a en Joan.
danced he Joan
112 Ca WPL 8, 2000 Joana Rosselló
(i) a. *Que en Pe e ho sap?
WH he Pe e i -knows
‘Does Pe e know?’
b. Que ningú ho sap?
WH nobody i -knows
‘Doesn’ anybody know?’
(ii) a. *Ja en Pe e ho sap.
al eady he Pe e i -knows
‘Pe e al eady knows’
b. Ja ningú ho sap.
al eady nobody i -kows
‘Nobody knows anymo e.’
In iew o his ac s, I would p opose ha in Ningú queda (nobody s ays/*is-le ) o Ningú
al a (Nobody is-absen /*is-le ) p e e bal ningú, like he abo e men ioned DPs, is di ec ly adjoined
o a oo node o he sen ence (hence he only agen i e eading o he p e e bal posi ion), bu ,
unlike DPs, i would no be adjoined o T, bu o NEGa ion, which would be uni e sally lowe han
T, acco ding o he e idence om languages whe e nega ion is no a e bal cli ic. Admi edly, his
amoun s o saying ha s ong V- ea u es o T ail o be checked. I would say, howe e , ha , in
acco dance wi h he discussion in sec ion 3 (and oo no e 7), checking only equi es ha he e b
mus be ou side he P (θ-domain), and mo e in o he checking domain, so ha me ge wi h NEG
would sa is y his equi emen . I his easoning is on he igh ack (which would need much
mo e ca e ul examina ion), we should u he e ine he idea ha NSLs ha e a s ong V- ea u e in
T. Ten a i ely, we could assume ha in NSLs he i s in lec ional ca ego y abo e P has a s ong
V- ea u e. No ice ha such a solu ion is in acco dance wi h he philosophy o he MP, in he sense
ha equi emen s mus be sa is ied locally and in he mos economical way.
20. He e I will igno e V-mo emen .
Ca WPL 8 001-137 29/5/2001 17:30 Página 112
c. Null Subj. [TP [T 0maxFFS– V+φ–T] [ P p oS …] ]
Ex.: Balla a.
danced
[FF: Fo mal Fea u es]
We appeal o Me ge o de i e he p e e bal subjec , and o P oc as ina e, hence
o Mo e F(ea u es), o de i e bo h he pos e bal subjec and he (also pos e bal)
null subjec . Le us now compa e (9) wi h he ‘English-like’ e sion (10):
(10) a. [TP subjec i[T’ V [VP i V]]]
b. [TP p oEXPL [T’ V [VP DP V]]]
c. [TP p oi[VP i .. ]]
[ aces, indices and ba s a e a no a ional con en ion he ein as in (9)]
Now, e en i we we e o make he assump ion (which ne e heless can be p o ed
o be clea ly alse) ha he empi ical side does no a ou ei he op ion, he a ou ed
op ion om he poin o iew o he compu a ional ope a ions equi ed o de i e
he iad would no doub be he one de ended he e, namely (9). Le us see why. In
(10), bo h he p e e bal subjec and he null subjec (he e aken o be p e e bal)
equi e he applica ion o Mo e in o e syn ax, an ope a ion ha is complex and
hence «las eso » —in Chomsky (1998)21 i is he esul o combining he wo
o he possible ope a ions Me ge and Ag ee—, while i can be dispensed wi h in
(9). The only mo emen in (9) is me ely one o ea u es (Mo e F), sa is ying
P oc as ina e —in Chomsky (1998) Mo e F becomes Ag ee, wi hou mo emen
o any kind.
5.2. Me ge
The p oposal in (9) in oduces an occu ence o Me ge in ol ing a DP which,
despi e no being an exple i e, is pu ely me ged wi h T. This applica ion o Me ge
has he ollowing p ope ies: a) i does no iola e any equi emen , since he DP is
non-a gumen al; i he e o e complies wi h Chomsky’s (1995, 1998: (6)) poin ha
«pu e me ge [me ge which is no pa o mo e] in θ posi ion is equi ed o (and
es ic ed o) a gumen s»; b) i is p e e able because i ende s unnecessa y he
mo e complex op ion Mo e; c) i es ablishes a ela ion wi h i s co ela e p o which
is no pa o he checking mechanisms p ope ; ac ually, nei he he one no he
o he con ains any [–in e p e able] ea u es, which in Chomsky’s (1998) e ms is a
necessa y condi ion o a checking ela ion o be es ablished —we a e now assum-
ing ha elimina ing Nomina i e Case in NSLsUA is he igh mo e. This would
easonably p o ide an accoun o he ac ha his DP is always an op ional adden-
dum as a as s ic ly compu a ional ope a ions a e conce ned; d) i cap u es he
well known di e ences be ween p e e bal subjec s in NSLs and English-s yle lan-
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 113
21. I ha e chosen o use he heo e ical machine y in Chomsky (1995: Chap. 4), since he p oposal in
Chomsky (1998) is no well-de ined as i s ands. See Rosselló (1999).
Ca WPL 8 001-137 29/5/2001 17:30 Página 113
guages, by a ibu ing hem o he de i a ion and no o he ep esen a ion. Agains
a ce ain classical ep esen a ional iew ha held ha he p e e bal subjec is dis-
loca ed, ou side C’/CP, he de i a ional iew p oposed he e allows us o dispense
wi h supe luous s uc u e, since he p e e bal subjec in (1a) would occupy he
same posi ion as he English subjec ;22 e) i opens up new lines o he unde -
s anding o p ope ies and phenomena ha ha e p o ed esis an o ea men s ha
assume ha he p e e bal DP in (1a) is ob ained by mo emen . So, o ins ance,
(i) in e sion e ec s in in e oga i es —a s ill unsol ed opic o NSLsUA— could
be deal wi h as a case o ela i ized minimali y; (ii) mo e sa is ac o y pa hs
could open up o deal wi h cases o «app oxima e» ag eemen (Els lingüis es no
sabem es — he linguis s no know-1PL no hing ‘We linguis s do no know any-
hing—; Els es udian s ho em o nosal es — he s uden s i -do-1PL e e y ing we
‘We es uden s do e e y ing ou sel es’—; (Nosal es)ho em o les dones —(we)
i -do-1PL e e y hing he women ‘(Among us) we women do e e y hing’—) and o
shed ligh on he con ol and binding beha iou illus a ed in examples like Els
jo es ho aconsegui eu sense es o ça - os/*se gai e — he young i -achie e-FUT-
2PL wi hou o-s ain-you sel es/* hemsel es much ‘You he young will achie e
i wi hou much e o —; (iii) a well ounded explana ion could be p o ided o
G eenbe g’s uni e sal 33 (hence o he cha ac e is ic pa e n in NSLsUA): he p e-
e bal subjec cooccu s wi h maximally speci ied e bal ag eemen , p ecisely he
ag eemen ha shows up wi h null-subjec sen ences in hose languages; his would
ollow i his «subjec », unlike pos e bal ones, is a DP me ged wi h a sen ence
con aining a phone ically emp y p onoun; and inally (i ) i p o ides a be e unde -
s anding o he «loss» o in e p e a ions (o e en o g amma icali y, in some cases)
igge ed by p e e bal (as opposed o pos e bal) subjec posi ion in ce ain con-
s uc ions (in ol ing a copula i e e b—see (2c) s. (3c); o an unaccusa i e e b
—see (2a) s. (3a)). In wha ollows, I will only minimally de elop pa o his las
poin , which was al eady in oduced in sec ions 3 and 4.
6. Impossible p e e bal (o null) subjec s
In § 3 and 4.2 we had occasion o see ha p e e bal non ocused subjec s (o null
subjec s) enjoy less eedom o appea ance han pos e bal (o ocused/wh p e-
e bal) ones (see (2), (3), (7) and (8)). To s a wi h, such a si ua ion is incompa -
ible wi h he idea ha he p e e bal subjec is an S -in e nal subjec ha has mo ed
(an almos inescapable analysis in a amewo k assuming he S -In e nal Subjec
Hypo hesis)23. The da a, howe e , ha e aken us much u he . Speci ically, he
114 Ca WPL 8, 2000 Joana Rosselló
22. Me ge can ake place in highe o lowe posi ions, hough. See no e 19.
23. The e a e app oaches ha do no challenge his poin , which simply ollows om adop ing he
In e nal Subjec Hypo hesis, and s ill hold he iew ha he pos e bal subjec as an in e ed subjec .
How? By de i ing i om a p e e bal subjec . In his kind o app oach (Ka o 1999) he e is a p e-
e bal subjec o igina ing as an S -in e nal subjec ; and he e is an o e pos e bal subjec which is
he esul o in e ing he de i ed p e e bal subjec . Besides he sel -e iden complica ion, he p ob-
lem is ha his amewo k also p edic s, con a y o ac , ha be ween a p e e bal and a pos e bal
subjec he e canno be any mo e di e ences han hose ela i e o in o ma ion packaging.
Ca WPL 8 001-137 29/5/2001 17:30 Página 114
iden ical beha iou , in all such cases, o pos e bal subjec s and ocused/wh p e e bal
leads us o an ob ious conclusion: i ocusing o he p e e bal subjec is he only
s a egy o p ese ing he pos e bal eading o he subjec , hen Nomina i e Case
canno be a ailable o he a iable bound by he ocus/wh ope a o i , in acco -
dance wi h he o hodox iew, he Nomina i e Case posi ion is he p e e bal sub-
jec posi ion. To ake an example, we could no expec ha a couple o Majo can sen-
ences like Qui queda? (‘Who s ays?’ / ‘Who is le ?’) and Na MARIA queda
(‘MARIA s ays’ / ‘Ma ia is le .’) could ha e he s a i e-quan i ica ional eading
(‘Who is le ?’, ‘MARIA is le !’), besides he agen i e one (‘Who s ays?’ ‘MARIA
s ays!’) —which in Ba celonan in ol es a p onominal e b: Qui es queda?, La
MARIA es queda!—, while he s a i e-quan i ica ional eading does no ob ain in
Na Ma ia queda (‘Ma ia s ays’ / *‘Ma ia is le .’), which can be in e p e ed only
agen i ely —and is simply impossible in Ba celonan: *La Ma ia queda.
No ice ha he e is no way o escape his conclusion. I would no be legi ima e
o assume ha wh mo emen o he subjec akes place om he θ-me ge basic posi-
ion24 and ha hen co e checking o Nomina i e Case applies o he ace (copy).
To be p ecise, his escape-s a egy would no be such since i would only be applic-
able o he ins ances whe e i would no be necessa y (checking could also apply
o e ly, economy conside a ions aside): i would apply only in cases whe e he pu -
po ed p e e bal Nomina i e Case posi ion is in ac associa ed o he wh eading. The
eason o his limi a ion is his: co e aising o me ge DP ea u es wi h he T
in lec i e head places hese checking ea u es in a s uc u al posi ion which mus
ha e essen ially he same p ope ies as he [Spec, T] posi ion. And his leads again
o he w ong p edic ion ha i we ge he non-agen i e eading o Qui queda? we
should also ge i o Na Ma ia queda.
Now, his conclusion ollows om he iew ha assimila es co e mo emen
o DP/NP ea u es o o e mo emen o DP/NP, which is p ecisely he posi ion
aken by Chomsky (1995: § 4.4.4), in consis ence wi h his backg ound assump-
ion ha he Spec,T posi ion is o med by ei he a subjec DP aised om he P-
in e nal θ-posi ion, o by me ge o an exple i e. I we keep o his assump ion, he
assimila ion is concep ually plausible and empi ically enable. Wha is no enable,
howe e , is he assump ion i sel , namely ha he only way o o m a Spec,T is he
one con empla ed in Chomsky (1995):
(11) Spec o T is c ea ed by aised DP o an exple i e me ged wi h T.
I (11) was no he only op ion, u he possibili ies could be explo ed o ge
Nomina i e Case in sen ences o he kind Qui queda? and La MARIA queda in he
s a i e-quan i ica ional in e p e a ion, so as o p ese e he idea ha an a gumen is
a non- i ial Chain (Case, θ). Le us see why (11) canno be he only op ion.
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 115
24. I is mos ema kable and symp oma ic ha om a a ie y o da a and o independen heo e ical
cons uc s ( he Emp y Ca ego y P inciple in he Go e nmen and Binding pe iod and he Checking
heo y in he MP de i a ional app oach) one is o ced o he conclusion ha wh ex ac ion in NSLs
akes place om a pos e bal posi ion.
Ca WPL 8 001-137 29/5/2001 17:30 Página 115
I (11) was he only op ion, i would be legi ima e o assimila e, in he way sug-
ges ed abo e, he posi ion o adjunc ion o he T head wi h he Spec,T posi ion,
and he e o e a pos e bal subjec (always associa ed o an exple i e in Spec,T)
should ha e he same p ope ies as a p e e bal subjec , since he ea u es o he
exple i e-associa e ( he pos e bal subjec ) would me ge wi h T co e ly. In my
iew, his assump ion is di ec ly challenged by he pai s (2)-(3) and (7)-(8). And,
i we go u he o es he possibili ies o he subjec o ac as a con olle — he
es used by Chomsky (1995)—, i does no seem o me enable ha in NSLs he p e-
e bal o pos e bal subjec posi ion does no make a di e ence: con ol is possi-
ble in (12) while i sounds unna u al in (13):
(12) a. En Pe e e pe con ola .
he Pe e comes o con ol
b. Un pacien ha en a pe mi a .
a pa ien has come-in o look
(13) a. ?*En énen dos pe e el sopa (, de cuine s).25
EN-come wo o make he dinne (o cooks)
b. ?* No n’en a gai e a mi a l’exposició (, de gen ).
no EN-come-in many o look-a he-exhibi ion (o people)
c. ?* N’han a iba un mun pe es udia el ema (, d’expe s).
EN-ha e a i ed a lo o s udy he opic (o -expe s)
Con a y o he uni o m e alua ion ha has been gi en in he MP o Ca dinale i’s
examples adduced by Chomsky (1995: 274), my iew is ha i makes a di e ence
whe he he pos e bal subjec in ol es ex ac ion o cli ic en (i . ne) o i s ays in
ull. In he la e case, in ac , he examples sound less unna u al:
(14) a. Vénen dos cuine s p o essionals pe e el sopa .
come wo cooks p o essional o make he dinne
b. No en a gai e gen a mi a l’exposició.
no come-in many people o look-a he-exhibi ion
c. Ha(n) a iba un mun d’expe s pe es udia el ema.
ha e a i ed a lo o expe s o s udy he opic
In his connec ion, he ollowing pai — om Majo can— is also e y illumi-
na ing:
116 Ca WPL 8, 2000 Joana Rosselló
25. One can de ise e en mo e de ian examples. In a sen ence like No n’hi ha p ou/gai es pe e el
sopa (no EN-THERE-ha e enough/many o make he dinne ‘The e a e no enough/many o
hem o make he dinne ’), he cli ic en canno be he con olle o he subjec in he pu pose clause,
which can only ha e an a bi a y eading. En, in such sen ences, poin s o ano he e e en (gues s,
kind o ood, e c.) which can be in e p e ed as a p e equisi e o he ac ion exp essed in he pu pose
sen ence.
Ca WPL 8 001-137 29/5/2001 17:30 Página 116
(15) a. No en queda cap pe isi a (de pacien )/ (?*de me ge).
no EN-is-le /*s ays any o isi (o pa ien )/ (?*o doc o )
‘(As o pa ien s) none is le o be isi ed.’
*‘(As o doc o s), none s ays o do hei isi s.’
b. No queda cap pacien / cap me ge pe isi a .
no is-le /s ays any pa ien / any doc o o isi
‘No pa ien is le o be isi ed’
‘No doc o s ays o do hei isi s.’
In (15a), whe e ex ac ion o clí ic en ende s he unaccusa i e s uc u e he
only op ion o he main clause, he only a ailable in e p e a ion is he s a i e-quan-
i ica ional one, and subjec con ol o he in ini i e becomes impossible: he main
clause a gumen is in e p e ed as co e e en wi h he objec o he in ini i e, so ha
en is in e p e ed as «pa ien » and no «doc o », as shown by i s disloca ed co ela e
—on he p agma ically unma ked eading.
In (15b), ins ead, whe e cli ic en is no ex ac ed, he main clause s uc u e is
ambiguous: on he s a i e-quan i ica ional eading, i beha es as in (15a), wi h cap
pacien as he p agma ically app op ia e op ion o he V sis e posi ion; on he
agen i e (‘s ay’) eading o queda , ins ead, subjec con ol o he in ini i e becomes
a ailable, he in ini i al e b being used in ansi i ely. In his case, o cou se, he
p agma ically p e e ed in e p e a ion chooses cap me ge as subjec o une ga i e
queda (‘s ay’). This beha iou is no idiosync a ic o Majo can, as e idenced by
he ambigui y o s anda d Ca alan sen ences like Abans al a es u i a a al o jo
(‘Be o e, YOU we e missing/absen and now I am’), Ja no al a ningú (‘Nobody is
missing/absen anymo e’), A a al a la Ma ia (Now MARIA is missing/absen ’).
Le us now conside he con as in (16):
(16) a. (Dos indi idus) *han ingu pe eu e.
(Two guys) ha e come o see
b. Han ingu dos indi idus pe eu e.
Ha e come wo guys o see
‘Two guys ha e come o be seen.’
F om (16a) i can be in e ed ha a p e e bal o null subjec is unable o p o-
ide he con en o he obliga o y objec o eu e ‘see’ (in Ca alan, he s a i e in an-
si i e in e p e a ion o eu e —namely ‘be able o see’— equi es he cli ic hi: eu e-
hi) and hus ha con ol is obliga o y: he p e e bal (o null) subjec canno ail o
be a con olle . This is no he case, ins ead, wi h (16b), whe e dos indi idus can be
he an eceden o he in ini i e objec .26 We ha e, in sum, a u he example o di -
e en ial beha iou depending on he p esence o a p e e bal o a pos e bal sub-
jec , namely a u he coun e example o assimila ing one o he o he unde (11).
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 117
26. The s a us o (16b) is claimed o be dubious by a e iewe o his pape . The con as be ween
(16a) and (16b) holds, in any case.
Ca WPL 8 001-137 29/5/2001 17:30 Página 117
The posi ion de ended he e is well known: (11) is no he only possibili y o c e-
a ing Spec,T. Speci ically, NSLs can c ea e i by me ging wi h T a DP co ela ed wi h
he p o elemen , which s ays in Spec, P posi ion. F om his s andpoin , new pa hs
open up o us o explo e new possibili ies on how Nomina i e Case can be ob ained
— emembe ha his possibili y is linked o he non-uni e sali y o (11). I also
emains o be a gued o ha he posi ion o p o is Spec o P. In wha ollows,
we will y o ie up hese wo issues (Nomina i e Case and p o posi ion) and, co -
espondingly, make sense o he abo e da a, going beyond he me e assump ion
ha he p e e bal subjec posi ion is ob ained by di ec me ge wi h T.
Conside he sen ences in (17) (whe e he e b queda is in ended in he
Majo can use: ‘be le ’/‘s ay’).
(17) a. Quedes u. / Fal es u.
s ay/a e-le you a e-absen /a e-missing you
a’. Queda na Ma ia. / Fal a la Ma ia.
s ays/is-le he Ma ia is-absen /is-missing he Ma ia
b. Tu quedes. / Tu al es.
you s ay/*a e-le you a e-absen /*a e-missing
b’. Na Ma ia queda. / La Ma ia al a.
he Ma ia s ays/*is-le he Ma ia is-absen /*is-missing
c. Quedes. / Fal es.
s ay-2SG/*a e-le -2SG a e-absen -2SG/*a e-missing-2SG
The sen ences in (17a/a’) a e ambiguous — hey allow bo h he s a i e-quan-
i ica ional (unaccusa i e) eading and he agen i e (une ga i e) eading—, while
hose in (17b/b’), wi h a non- ocused p e e bal subjec only allow he agen i e
(une ga i e) eading. The la e si ua ion is also ue o (17c). Wha is o be empha-
sised is ha in (17a/a’) he unaccusa i e in e p e a ion is no linked o he occu ence
o cli ic en ex ac ion on he pos e bal subjec , no e en o he quan i ica ional o
ba e-NP s a us o he pos e bal subjec : he unaccusa i e in e p e a ion is possible
wi h any DP, e en a p onominal one, p o ided i is phone ically o e .
In a way, he esul is again su p ising: why should a null subjec (p o) no be
able o licence he unaccusa i e in e p e a ion, while a o e p onominal subjec
is? Should we pe haps appeal o an ex ag amma ical ac o , o he e ec ha he
unaccusa i e in e p e a ion should be necessa ily associa ed o ocus in e p e a-
ion on he a gumen ? This is unlikely: a sen ence like En queden (EN-a e-le
‘Some a e le ’) is pe ec ly well- o med, and i only has he unaccusa i e in e -
p e a ion, wi hou any ocusing on he a gumen .27
Le us see i we can do any p og ess on he side o Case heo y, which is he
o he unsol ed p oblem we a e le wi h. I he unaccusa i e in e p e a ion s ems om
118 Ca WPL 8, 2000 Joana Rosselló
27. I does no seem o any help o a ibu e o lexical p onouns a special and exclusi e s a us —as in
Rigau (1987), who ea s hem as quan i ie s in A’ posi ion— in o de explain he ac s in (17).
The eason is ha he e lexical p onouns beha e like nominal DPs —(17a) beha es like (17a’) and
(17b) beha es like (17b’).
Ca WPL 8 001-137 29/5/2001 17:30 Página 118
he V-sis e posi ion, while he une ga i e one s ems om he Spec, P posi ion,
we could possibly appeal o he incompa ibili y o he o me posi ion wi h
Nomina i e Case, which is equi ed by p o in (17c). Now, his does no seem a
sensible op ion when he sen ences in (17a) hemsel es a e unaccusa i e —in one
o hei in e p e a ions— and hei subjec s ea u e in Nomina i e Case, a leas
appa en ly. I eso ing o Nomina i e Case is o no use o explain he con as
be ween (17a/a’) —wi h wo eadings— and (17c) —wi h only he s a i e-quan-
i ica ional eading— and, besides, he only plausible means o checking Nomina i e
Case h oughou (17) is o be co e aising o he DP ea u es o T (which, in ac ,
a e independen ly equi ed by he φ- ea u es o V in T), i is emp ing o explo e
wha would happen i in NSLsUA we dispense wi h Nomina i e Case h oughou .
Since i would be a mo e economical op ion, o he ex en ha we dispense wi h
one (appa en ly supe luous) mechanism, ha o Nomina i e Case checking, i
seems wo h o conside a ion.
6.1. On he spu ious cha ac e o Nomina i e Case in NSLsUA
Once conside ed se iously, elimina ion o Nomina i e Case in NSLsUA seems
desi able. I would appea ha Nomina i e Case is supe luous in a language whe e
he [–in e p e able] φ- ea u es o he ini e e b aised o T will always a ac he
ea u es o he pos e bal subjec , whe he an o e DP o p o. In o he wo ds, in a
NSL Nomina i e Case is no necessa y, while in languages like English Nomina i e
Case is a necessa y esou ce in addi ion o he EPP ea u e (o s ong D- ea u e o
T). I is, in e ec , necessa y since in cons uc ions wi h a pos e bal subjec , which
is he associa e o a pu e exple i e (such as he e), he only means o sys ema ical-
ly achie ing a con e gen de i a ion is assuming ha he exple i e consis s solely o
he ca ego ial ea u e D and ha he associa e, ins ead, comes equipped wi h φ- ea-
u es and Nomina i e Case —bo h necessa y.28
I , ins ead, he e a e no exple i es, and V aised o T has [–in e p e able]
φ- ea u es, Nomina i e Case would become edundan , since aising o he o -
mal ea u es o he pos e bal subjec and, consequen ly, subjec ag eemen a e
gua an eed. A a mo e concep ual le el, elimina ing Nomina i e Case in NSLsUA
is a desi able mo e oo. Nomina i e Case is he exp ession o a Spec-head ag ee-
men ela ion. Now, i he ela ion be ween X and a DP (wi h FF1) in Spec o X
is equi alen o he ela ion [XFF1-X] (in he case o pu e ea u e checking wi hou
pied-piping), « eal» (mo phologically ealized) ag eemen in φ- ea u es (as in
Venim nosal es/p o —come-1PL we/p o—, Vénen els nens/p o —come3PL he
child en/p o—, Ho dius u/p o —i -say2SG you/p o) can be conside ed an imple-
men a ion o his abs ac ealiza ion in i s mos economical o ma , namely pu e
ea u e checking by adjunc ion o he T head wi hou pied-piping ([T FF1-T]).
One u he i ue o his iew is ha i does no do any iolence (i a he lends
suppo ) o he idea ha an a gumen always o ms a non- i ial chain wi h i s head
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 119
28. See he co esponding a gumen a ion in Chomsky (1995: § 4.5.3)
Ca WPL 8 001-137 29/5/2001 17:30 Página 119
in a checking posi ion wi h [–in e p e able] ea u e(s) —in he case unde consid-
e a ion he φ- ea u es o V— and i s oo in a θ-posi ion . Finally, i is o be poin -
ed ou ha me e φ- ea u e aising gi es he same e ec as he EPP in a co e way:
aising o D/N29 o T.30 Mo e specula i ely, i could be he case ha , a he han
he EPP —in Chomsky’s (1995) o mula ion—, wha was obliga o y was he ans-
mission o p edica ion o he in lec ional s e ch. Now, such a ansmission can be
achie ed bo h in he con igu a ion [TP DP T] (s anda d EPP) and in he con igu a-
ion [TFFD/N Τ].
Wi hou Nomina i e Case, sen ences like Qui queda (who s ays/is-le ) and Na
MARIA queda (The MARIA s ays/is-le ) can be de i ed by di ec mo emen o he
qu/ ocus ope a o . This mo emen can be o igina ed ei he in he Spec, P posi ion
(une ga i e eading) o in he sis e -o -V posi ion (unaccusa i e eading). The ope -
a o - a iable (ope a o -a gumen ) con igu a ion would be de i ed in wo s eps: i s ,
by o e aising we ge he ope a o ; second, by co e aising o he φ- ea u es o he
copy o he [–in e p e able] φ- ea u es o he e b, we ge he a gumen .
In e es ingly, he e is some independen e idence in suppo o dispensing wi h
Nomina i e Case in NSLs. We will conside i a e comple ing he analysis o he
sen ences in (17) and dealing in pa icula wi h he issue o he in e p e a ion o
(17c).
6.2. On he posi ion o he null subjec (p o)
Following he conclusion eached in he p eceding sec ion, in he examples in (17)
he e is no Nomina i e Case. Le us conside how he co esponding de i a ions
would p oceed a he ele an s age. In (17a/a’) he double in e p e a ion is ob ained
depending on which he θ-posi ion is o he a gumen (la/na Ma ia / u): i i is
Spec, P, he in e p e a ion is agen i e —in Ba celonan, as poin ed ou , his equi es
he p onominal o m o he e b, queda -se; i i is he V-complemen posi ion, he
in e p e a ion is s a i e-quan i ica ional. As o he ea u e-checking mechanism,
he e should be no p oblem: φ- ea u es o V co e ly a ac he φ- ea u es o he
a gumen , whe he in one o he o he posi ion.
The cases in (17b/b’) and (17c) should in p inciple be analogous o hose in
(17a/a’), he only di e ence being ha now i is p o —a Caseless amalgam o he
ca ego ial ea u e D- and φ- ea u es (pe son, numbe , gende )— ha mus check
he φ- ea u es o V in T.31 We a e aced, howe e , wi h he al eady no ed con as :
120 Ca WPL 8, 2000 Joana Rosselló
29. I he o hodox EPP amoun s o he s ong D- ea u e o T, he e sion de ended he e would no be
comple ely equi alen : on he basis o sen ences like Passen coses (happen hings), Fal a ca è (is-
lacking co ee) (see also he analysis o (20a), (20c)), i can be pe ec ly a gued ha i is an N (NP)
and no a D (DP) ha e ases he numbe ea u es o he aised e b (in hese cases, hi d pe son
can be conside ed a non-pe son).
30. O o an in lec ional (checking) ca ego y highe o ha o θ-me ge. See oo no e 19.
31. I is o be poin ed ou ha hese da a do no seem o be explainable om app oaches ha conside
ag eemen as an a gumen (o a leas a de e mine ), such as O dóñez & T e iño (1999), Ka o
(1999), Alexiadou & Anagnos opoulou (1998) e c. They all seem o p edic a uni o m beha iou
h oughou (17), since ag eemen does no a y a all ac oss he di e en cases in (17).
Ca WPL 8 001-137 29/5/2001 17:30 Página 120
bo h (17b/b’) and (17c) can only ha e he agen i e in e p e a ion, ha we associ-
a e wi h he Spec, P posi ion. Does his mean ha p o can only yield con e gen
de i a ions in his posi ion? Tha ’s wha we should in e . Le us conside wha
would happen o he wise. A sen ence like Els nens mi en ( he child en wa ch, ‘The
child en a e wa ching’), which con ains a e b ha does no obliga o ily ca y
Accusa i e Case (and so can yield an absolu e, in ansi i e in e p e a ion), should
be ambiguous be ween his absolu e in e p e a ion and he one yielded by he sen-
ence Els nens són mi a s ( he child en a e wa ched),32 con a y o ac . The p ob-
lem ha a ises i he dis ibu ion o p o is no es ic ed is ha we would ha e no
means o know whe he i is p esen o absen in a gi en de i a ion.
In a wo d, a p o, exclusi ely consis ing o [+in e p e able] ea u es om he
poin o iew o he C-I sys em, bu phone ically emp y, mus be iden i iable in an
op imal way. And his implies ha i mus be iden i iable unequi ocally om he
poin o iew o he in e p e a ion and locally om he poin o iew o he ea-
u e-checking ela ions ha i is in ol ed in and iden i ied by. Now, he local iden-
i ica ion equi emen would amoun o occupying he highes posi ion in he
θ-domain, i.e. Spec, P.
I is impossible o e iew he e all he consequences o he p esen p opos-
al.33 Le me jus say some hing on i s possible consequences on how we iew
he ela ion be ween an objec cli ic (di ec objec and da i e) and i s co e-
sponding θ-posi ion. I would like o poin ou ha , acco ding o wha he SLH
es ablishes, cli ics should be kep a pa om in lec ion because o he simple
A Minimalis App oach o he Null Subjec Pa ame e Ca WPL 8, 2000 121
32. Mo e explici ly, i mi a (‘look’, ‘wa ch’) has (o can assign) Nomina i e Case and p o is Caseless,
s a ing om a nume a ion wi h he i ems mi en¸ p o, T and C we could bo h ge an une ga i e P,
[SVp o mi en], and an unaccusa i e VP, [SV mi en p o], bo h ul ima ely con e ging. I u ns ou ,
hough, ha only he o me de i a ion con e ges, since mi en only yields he une ga i e
in e p e a ion. The o he one, [SV mi en p o], which would be in e p e a i ely equi alen o a pas-
si e —a e wa ched— wi hou passi e mo phology, is no ob ained. How can we p e en [SV mi en
p o] om con e ging? Wha we p opose he e is ha language design mus limi p o o Spec, P,
p ecisely o a oid he inde e minacy ha o he wise would a ise.
33. This p oposal on he dis ibu ion o p o implies ha unaccusa i e sen ences wi h a null (non pa -
i i e) subjec ha e p o in Spec, P. I don’ hink his is necessa ily a p oblem. I would conside
explo ing he ollowing idea: lexically unaccusa i e e bs can une ga i ize in he syn ax (by means
o aising by subs i u ion o Spec, P) This op ion would become obliga o y when he a gumen is
p o. This would accoun o he e y clea con as in agen i i y pe cep ible be ween Vénen (come-
3PL , ‘They come’) and En énen (EN-come-3PL, ‘Some come’) —c . Vénen pe eu e’ (‘They
come o see you’) s. *En énen pe eu e’ (‘Some come o see you’). O e DPs (excluding ba e
NPs), ins ead, appa en ly would be able o appea bo h in Spec, P and in V-sis e posi ion. (see
(17a/a’). In ac , one could specula e ha , con a y o wha has been p e iously assumed, he
Majo can e bs al a (‘be missing’/‘be absen ’) and queda (‘be le ’/‘s ay’) a e no lexically
ambiguous: a he wha happens is ha he p ocess o syn ac ic une ga i iza ion yields as a esul
a mo e nea con as on he in e p e a i e side. The eason could lie in hei lexical meaning. So, he
meaning o unaccusa i e al a implies «non p esence», while ha o une ga i e al a would imply
« olun a y, con olled non p esence». The ac ha in non Majo can Ca alan (as in Spanish) queda
does no allow une ga i iza ion would ha e he ollowing in e p e a ion: une ga i iza ion in ol es
a mo emen ope a ion ha can be spa ed i ano he lexical i em is a ailable, in his case he p onom-
inal e b queda -se, which is inhe en ly une ga i e.
Ca WPL 8 001-137 29/5/2001 17:30 Página 121
Le in, B.; Rappapo Ho a , M. (1995). Unaccusa i i y. A he Syn ax-Lexical Seman ics
In e ace. Camb idge, Mass.: MIT P ess.
Longoba di, G. (1987). «Las o aciones copula i as en la eo ia sin ác ica ac ual». In
Demon e, V.; Fe nández, M. (eds.) (1987). Sin axis de las lenguas ománicas.
Mad id: El A que o.
Manzini, R.; Sa oia, L.M. (1997). «Null subjec s wi hou p o». WPLUCL 9: 303-313.
Mo o, A. (1997). The Raising o P edica es. P edica i e Noun Ph ases and he Theo y
o Clause S uc u e. Camb idge: Camb idge Uni e si y P ess.
Nash, L.; Rou e e , A. (1996). «Func ional heads and P oxies: an asymme ic se ».
Glow Newsle e 36: 50-51.
O dóñez, F.; T e iño, E. (1999). «Le disloca ed subjec s and he p o-d op pa ame e :
A case s udy o Spanish». Lingua 107: 39-68.
Picallo, C. (1998). «On he Ex ended P ojec ion P inciple and he Null Exple i e
Subjec ». P obus 10: 219-241.
Pie a, C. (1987). «Sob e la es uc u a de las cláusulas de in ini i o». In Demon e, V.;
Fe nández, M. (eds.). (1987). Sin axis de las lenguas ománicas. Mad id: El A que o.
Pla zack, C. (1994). «Null Subjec s, Weak AGR and Syn ac ic Di e ences in
Scandina ian». WP in Scandina ian Syn ax 53: 85-106.
Pollock, J.-Y. (1997). Langage e Cogni ion. In oduc ion au p og amme minimalis e de
la g ammai e géné a i e. Pa is: P esses Uni e si ai es de F ance.
Rigau, G. (1987). «Sob e el ca ác e cuan i icado de los p onomb es ónicos en ca alán».
In Demon e, V.; Fe nández, M. (eds.). (1987). Sin axis de las lenguas ománicas.
Mad id: El A que o.
Rizzi, L. (1982). Issues in I alian Syn ax. Do d ech : Fo is.
— (1990). «Residual Ve b Second and he Wh-c i e ion». Technical Repo s in Fo mal
and Compu a ional Linguis ics. Uni e si y o Gene a. Also in Belle i, A.; Rizzi, L.
(eds.) (1996). Pa ame e s and Func ional Heads: Essays in Compa a i e Syn ax.
Ox o d: Ox o d Uni e si y P ess.
Rosselló, J. (1999). «Un en ocamen minimis a del Pa àme e del Subjec e nul». Repo s
de Rece ca del G up de G amà ica Teò ica de la UAB, GGT-06 (h p://gg .uab.es).
Solà, J. (1992). Ag eemen and Subjec s. Uni e si a Au ònoma de Ba celona. Doc o al
hesis.
Ta aldsen, K.T. (1980). On he Nomina i e Island Condi ion, Vacuous Applica ion, and
he ha - ace il e . Indiana Uni e si y Linguis ics Club.
128 Ca WPL 8, 2000 Joana Roseselló
Ca WPL 8 001-137 29/5/2001 17:30 Página 128
