The Internal Structure of Absolute Small Clauses

The In e na1 S uc u e o Absolu e Small Clauses
Luis Mpez
Co nell Uni e si y
This pape is an in es iga ion in o he unc ionai shuc u e o Absolu e Small Clauses
(ASCs), wi h
da a
based on Spanish. ASCs a e composed o a p edica e and an NP
and hey a e adjoined o a main clause wi hou a media ing complemen ize . I is
shown ha heu s uc u e is no 'speciai' bu a p ope subse o main clause s uc u e.
ASCs ha e some in e es ing es ic ions: (i) Indi idual Le el (IL) p edica es a e
disailowed. (ii) deep subjec s a e disailowed, and (iii) he o de NP+p edica e is
disailowed. The analysis o (i)-(iii) elies on ecen p oposais ega ding clause
s uc u e, in pa icula , (i) Diesing's Mapping Hypo hesis and he s uc u ai analysis
o he Indi iduallS age Le el dis inc ion and (ii) wha Koizumi
has
e med he Spli
VP Hypo hesis, acco ding o which some unc ionai ca ego ies a e gene a ed
be ween he lowe and he uppe VP shells. Addi ionaily, he ques ion o how he
case o he NP is checked is add essed, and i is sugges ed ha he ea u e
composi ion o he head Aspec in ASCs is minimaily di e en om ha o main
clauses. A discussion o I aiian ASCs ensues and i is a gued ha hey ha e he same
s uc u e as he Spanish ones.
1.
In oduc ion
The empi ical scope o his pape is Spanish Absolu e Small Clauses (hence o h ASCs).
ASCs a e adjunc s composed o a p edica e -usually, bu no necessa ily, a pas pa iciple-
and a noun ph ase, which is op ional in many cases. In (la) he p edica e ha heads he ASC
is a pa iciple, in (Ib,c) i is an adjec i e, in (lc,d) i is an ad e b (examples om He nanz
(1991)):
Ca alun
Wo king Pape s in Linguis ics
(Ca
WPL) 4.1 '(1994):
45-92
Uni e si a Au bnoma de Ba celona
(1)
a. Desapa ecidas las joyas, llama on a la policia.
disappea ed he jewels called3pl
o
he police
b. Limpias las mas, 10s soldados salie on a la calle.
cleaned he weapons he soldie s wen -ou
o
he s ee
c. Tenso el ges o, el ga o nos obse aba desde la silla.
ense he ges u e he ca
us
looked-a om he chai
d. Asi las cosas, uno no sabe qué pensa .
like- his he hings one no knows wha belie e
e. Lejos 10s buques, la poblacidn isleña pudo e o na a sus casas.
a he ba s he popula ion o -island could e u n o hei homes
He e I ocus mainly on pa icipial ASCs, hough he analysis p esen ed ex ends o he o he
ypes. Using he examples in (2),
I
ske ch he dis inc i e cha ac e is ics o his cons uc ion.
ASCs a e unaccusa i e o ansi i e:
(2)
a. Desapa ecidas las joyas, llama on a la policia.
disappea ed he jewels called-3pl o he police
b. Ro os 10s inculos amilia es po la i esponsabilidad de 10s pad es, 10s
b oken he bonds amilia by he i esponsibili y o he pa en s he
adolescen es
se
sumen en la dep esi6n.
adolescen s
SE
sink in o he dep ession
In bo h cases, he
NP
can be phone ically emp y. I so, i has o be con olled by some
cons i uen in he main clause, usuaily he subjec .
I
assume his emp y ca ego y o bep o:
(2) c. Si iada p oi du an e a ios meses, la ciudadi u o que endi se.
besieged o se e al mon hs he ci y had ha su ende
(2) d. decla ada p q en es ado de eme gencia po e ce a ez en pocos meses, el
decla ed in s a e o eme gency o hi d ime in ew mon hs he
gobie no decidi6 en ia e ue zos mili a es a @a ciudad de A~~cuc~o]~
go emen decided send ein o cemen s mili a y o he ci y o Ayacucho
When he p edica e is ansi i e, i admi s a by-ph ase:
(2) e. Des uido el li omi po la mancha de pe 6le0, 10s pescado es emig a on.
des oyed he sho e by he spo o oi1 he ishe men emig a ed
The NP mus be a 'deep' objec , ha is, he subjec o an unaccusa i e e b o he objec o a
ansi i e e b. 'Deep' subjec s a e disallowed (Pe lmu e (1978), Belle i (1981)):
(2)
.
*T abajado
Juan,
se
ue de paseo.
wo ked
Juan
SE
wen -3sg o walk
g. *Des uido el gene al, mi 6 la ciudad con o gullo.
des oyed he gene al looked-3sg he ci y wi h p ide
The o de pa iciple+NP is obliga o y. The sequence NP+pa iciple is ung amma ical:
(2)
h. *Los ínculos amilia es o os, 10s adolescen es se sumen en la dep esi6n.
he bonds amilia b oken he adolescen s
SE
sink in o he dep ession
,
Finally, he NP ecei es nomina i e Case:'
I
use a p onominal
NP
o show he mo phological Case di e ence. While ASCs wi h p onouns sound
somewha unna u ai, he e is a de ini e con as be ween (2i) and (2j).
(2)
i. Desmayado yo, la eunión no pudo comenza .
ain ed
I
he mee ing no could begin
j. *Desmayado mi, la eunión no pudo comenza .
ain ed me he mee ing no could begin
No ice ha ASCs do no show any ace o a ense mo pheme. Since cu en heo y (see o
ins ance Chomsky (1993)) links nomina i e Case o ini e Tense, he a ailabili y o
nomina i e in ASCs is a phenomenon ha needs o
be
accoun ed o .
Ano he cha ac e is ic o his ype o cons uc ions -no iced by He nanz and by S ump
(1985) o English, is ha he pa iciple mus
be
a S age Le el (SL) p edica e, i canno
be
an
Indi idual Le el (IL) p edica e (see Ca lson (1977) o a desc ip ion o he dis inc ion):
(3)
.
a. *Sabidas muchas ma emg icas, Ca los pudo en a en la uni e sidad.
known many ma hema ics Ca los could en e in he uni e si y
b.
*Hablado espaiiol, Susana consiguió un buen abajo.
spoken Spanish Susana go a good job
In his pape
I
will a gue o an analysis o ASCs ha answe s he ollowing ques ions:
Q1. wha he unc ional s uc u e o ASCs is;
Q2.
why ASCs mus
be
SL p edica es;
43. why he
NP
mus
be
a 'deep' objec ;
Q4.
why he o de V+NP is obliga o y;
Q5. how he NP ecei es nomina i e Case.
Following Belle i's, De Miguel's and He nanz's insigh s,
I
show ha ASCs a e composed o
he lowe pa o an o dina y clause. Fu he mo e
I
claim ha ASCs p o ide s ong e idence
o he iew ha some unc ional s uc u e has o
be
p ojec ed be ween he wo VP shells (in
he sense o La son (1988)), as has been a gued by T a is (1991), Bowe s (1992), Koizumi
(1993), Zagona (1993) and Collins and Th áinsson (1993). Following Koizumi,
I
call his he
spli -VP hypo hesis. I (4) ep esen s an o dina y clause, ASCs consis o only he
VE,
he
inne shell, and he unc ional p ojec ions associa ed wi h i (FP= unc ional p ojec ions):
ASCs
I call he unc ional p ojec ions p ojec ed by
V2
'inne ' unc ional ca ego ies, and he
unc ional p ojec ions p ojec ed by Vl, 'ou e ' ca ego ies.
The undamen al hypo hesis ha unde lies my discussion is ha he unc ional hie a chy is
ixed by
UG,
a ia ion limi ed o he ea u e composi ion o he unc ional ca ego ies (Bo e
(1983)). This has wo consequences o an analysis o ASCs. Fi s , he s uc u e o Spanish
ASCs has o be a p ope subse o he g a nma o Spanish clauses -i.e., he s uc u e o
ASCs canno
be
'speciai' wi h espec o o dina y clauses. Since s uc u e is no subjec o
in alinguis ic a ia ion, 'subg amma s' (Addis
(1993))
mus
be
de ined in o he ems.
I
will
a gue ha he ea u e composi ion o he unc ional ca ego y Aspec in ASCs is minimally
di e en om ha o main clauses.

Secondly, ASCs in o he languages should ha e he same s uc u e, hence he same p ope ies
as
ASCs in Spanish -excep wo d o de , which depends no on hie a chy bu on he s eng h
o ichness o ea u es, as gene ally assumed (Pollock (1989), and many o he s). In his
espec , I belie e ha he analysis ha I p esen he e ex ends o ASCs in English and I alian.
This pape is o ganized as ollows: in sec ion 2,
I
p esen he s uc u e ha
I
p opose o
ASCs, answe ing
Q1.
In sec ion
3,
I explain why ASCs mus be SL p edica es. In sec ion
4,
I
show how he spli VP hypo hesis accoun s o he absence o ex e na1 a gumen s in ASCs. In
sec ion 5, I discuss he wo d o de p oblem. In sec ion
6,
I
p esen an accoun o he licensing
o he NP, wi h which I answe
Q5.
Sec ion
7
summa izes my esul s so a . In sec ion
8,
I
discuss I alian ASCs and inally, in sec ion
9,
I a gue con a Belle i (1990), De Miguel
(1990) and He nanz (1991) ha he e is no CP in ASCs, in keeping wi h he hypo hesis ha
he s uc u e o ASCs is p ope ly con ained in he s uc u e o o dina y clauses.
2.
The S uc u e
o
ASCs
2.1.
No
'Ou e '
TP
in
ASCs
Belle i (1990), De Miguel(1990) and He nanz (1991) a gue ha he e is no Tense Ph ase in
ASCs.
I
ag ee ha he e is no 'ou e ' Tense in ASCs. Pa iciples in Spanish do no suppo any
ense mo phemes: when pa iciples a e cons i uen s o ull clauses, ense ea u es need he
suppo o an auxilia y. ASCs a e ung amma ical wi h an auxilia y wi h ini e ense,
as
shown
in (5). I is possible o ha e a ge undi e auxilia y wi h a pa iciple o ming a sen en ial
adjunc , as in (Sb), bu hen we ob ain a di e en ype o cons uc ion,
as
we shall see in
sec ion 2.5.
(5)
a. *Ha disuel o el sul u o en agua, pie de sus p opiedades co osi as.
has dissol ed he sulphu in wa e loses i s p ope ies co ossi e
b. Habiendo disuel o el sul u o en agua, el quimico son i6 a la audiencia.
ha ing dissol ed he sulphu in wa e he chemis smiled
o
he audience
Mo eo e , as
I
men ioned a he beginning o his pape , adjec i es and ad e bs can also head
an ASC, and nei he o hese ca ego ies a e assumed o p ojec a TP.
On he o he hand, De Miguel (1992) claims ha TP is he highes node in ASCs. This TP is
desc ibed as
an
abs ac mo pheme bu as a syn ac ically ele an ca ego y. He mo i a ion is
ha ASCs con ain he empo al in o ma ion ha he e en desc ibed in he ASC is an e io
wi h espec o he e en aking place in he main clause. She hus cha ac e izes he head o
TP as con aining he ea u e [+an e io ]. This ea u e would
be
esponsible o he a ie y o
ad e bial unc ions
is-a- is
he main clause ha an ASCs may ha e, since all o hem a e o
be unde s ood as empo ally an e io . Fo ins ance, in (6a) he ASC is o
be
ead as causal,
whe eas in (6b) i is condi ional (examples om De Migue1(1992:65), who a ibu es he i s
one o Alcina and Blecua and he second one o iáza o Ca e e ). Bo h causali y and
condi ionali y a e hus encapsula ed in he ea u e [+an e io ]:
(6)
a. Dada su poca se iedad, op amos po no eno a le el con a o.
gi en he li le se iousness choose-lpl o no enew3sgDa he con ac
b. Gas ada esa agua, no pod emos
ni
bebe .
was ed ha wa e no can-FUT-lpl nei he d ink
Since he TP ha De Miguel a gues o canno be iden i ied wi h he o e Tense o main
clauses ha has a cka mo phological co ela e,
I
sugges ha i is an 'inne ' Tense, no an
'ou e ' one. The claim ha he e is a unc ional node Tense in e na1 o he
VP
has been pu
on a d by T a is (1991) and Collins and Th 6insson (1993). Thus, De Miguel's a gumen o
a TP in ASCs would p o ide u he e idence o his hypo hesis.
2.2.
No
NegP
in
ASCs
Belle i (1990), De Miguel (1990)and He nanz (1991) co ec ly poin ou ha he e is no
NegP in ASCs ei he :
(7)
a.
*
No disuel o el sul u o en agua
...
no dissol ed he sulphu in wa e
b.
*
No disuel o, el sul u o
...
no dissol ed he sulphu
Following hese au ho s,
I
a ibu e he ung amma icali y o
(7)
o he absence o a NegP.
De Miguel
(1992)
claims ha he p ohibi ion agains nega ion in ASCs is o a seman ic, no a
syn ac ic na u e, de i ed om i s empo al na u e. Howe e , compa e he pa icipial absolu e
o
(8a)
wi h he ge undi e o (8b):
(8) a. *No disuel o el sul u o de la mane a adecuada, Juan hubo de
no dissol ed he sulphu o he way igh Juan had
o
pospone la conclusi6n del expe imen o has a el dia siguien e.
pos pone he conclusion o - he expe imen
-
un il he day ollowing
b. No habiendo disuel o el sul u o de la mane a adecuada, Juan hubo de
no ha ing dissol ed he sulphu o he way igh Juan had o
pospone la conclusi6n del expe imen o has a el dia siguien e.
pos pone he conclusion o - he expe imen un il he day ollowing
(8a) is ung amma ical, in con as wi h he g amma ical (8b).
I
canno ind any seman ic
ea u e ha dis inguishes (8a) om (8b) and ha can be held esponsible o he
ung amma icali y o
(8a).2
I
conclude ha lack o he necessa y s uc u e causes he
ung amma icali y o
(8a).
No ice ha he posi i e coun e pa o (8b) has a di e en in e p e a ion:
(i)
Habiendo disuel o el sul u o,
Juan
p osigui6
con
el expe imen o.
ha ing dissol ed he sulphu
Juan
wen -on wi h he expe imen
2.3.
No
CP
in
ASCs
Belle i (1990), De Miguel (1990) and He nanz (1991) claim ha he e is a CP in ASCs.
Since his poin has nume ous ami ica ions, I pos pone i o sec ion 9, whe e he hypo hesis
o a CP in ASCs is discussed in de ail. I will be shown ha he e is no con incing e idence
o a CP in ASCs.
2.4.
Ag OP
Pa iciples and adjec i es in ASCs ag ee wi h he NP. This is shown in (9), whe e he owel
su ix
-a
indica es eminine and
-s
indica es plu al:
(9)
a. Acabada la cena, Jesús se di igi6 a 10s ap6s oles.
inished- em he- em dinne Jesús SE add essed o he apos les
'Dinne inished, Jesus add essed he apos les.'
b. Acabadas las a eas, 10s niños salie on a la calle.
inished- em-pl he- em-pl homewo k he child en wen -ou o he s ee
'Homewo k inished, he child en wen ou .'
Following Kayne's (1985, 1989) analysis o pa icipial ag eemen in F ench, which in ol es
an Ag P, I conclude ha he pa iciple o ASCs p ojec s an Ag P.
Now we a e in a posi ion o answe Q1: he unc ional s uc u e o ASCs eaches as a
as
Ag OP and maybe o some TP.
~
-
Whe eas (i)
can
ha e bo h a empo al and a causal in e p e a ion, (8b) can only ha e a causal one (see Rigau
(1991) o a simila phenomenon wi h Ca alan in ini i es, also ci ed in De Miguel (1992)). I could
be
a gued
ha nega ion makes he adjunc clause
in
(8b) obliga o ily causal, and ha ASCs do no ha e his in e p e a ion
-howe e , (6a) shows ha ASCs can ha e a causal in e p e a ion. The ques ion is why (8a) canno
be
g amma ical wi h a causal in e p e a ion.
being [+pe ec i e]. In pa icipial absolu es, Asp is mo phologically ealized on he pa iciple.
See also Ca s ens and Kinyalolo (1989) on AspP and T a is (1991) and Zagona (1993) on he
exis ence o
an
'inne ' aspec . Finally, see Collins and Th áinsson (1993) on an 'inne ' Tense,
connec ed, as I men ioned in sec ion 2.1, wi h De Miguel's (1992) p oposal o a TP in ASCs.
The s uc u e ha
I
p opose o ASCs is he one con ained in he
box
o (16):
VP 1
n
NP V' ASC
A
T Ag OP
n
s
pec A g '
n
Ag AspP
n
ASP VP 2
n
NP V'
I
-
(16) gi es us a di ec answe o 43: 'deep' subjec s a e gene a ed in he uppe VP1 shell and
his is no p esen in ASCs. Also, i accoun s o he licensing o he by-ph ase and hence he
passi e appea ance ha ansi i e ASCs ha e,
as
can
be
seen in example (Ze), epea ed he e:
(2) e. Des uido el li o al po la mancha de pe óleo, 10s pescado es emig a on.
des oyed he sho e by he spo o oi1 he ishe men emig a ed

The ex e nal 0- ole in an o dina y clause is assigned by V1, he uppe e b. In a passi e
cons uc ion, he passi e mo phology supp esses he ex e nal 0- ole (Bake
e
al.
(1989),
among o he s), which in ou ems means ha V1 loses i s abili y o license i . Then he
ex e nal 0- ole may eappea as an oblique. In ASCs he uppe V1 is no inse ed, so he
ex e nal 0- ole is no licensed by a p edica e. I ollows ha ASCs ha e no 'deep' subjec s and,
as
a consequence, he agen ole can eappea
as
an oblique. Since he ex e nal a gumen is
no gene a ed, an ASC is e ec i ely iden ical o a unca ed passi e cons uc ion, he esul o
supp essing an ex e nal a gumen and o no gene a ing i in he s uc u e being he same.
Addis (1993) and F anco (1993) a gue ha Spanish ASCs a e passi e cons uc ions. Gi en
he analysis in (16), saying ha ASCs a e passi es is edundan because i would en ail
supp essing some hing ha was ne e he e in he i s place.'
In o de o con i m ha (16) is needed o accoun o he absence o 'deep' subjec s in ASCs,
le us y o see i
43
could be answe ed om a s uc u e in which Ag OP is p ojec ed abo e
VP1, as in (12a). Le us assume (12a') as he s uc u e o he clause. In (12a') he e is no
unc ional s uc u e be ween he wo VP shells. The cu ly b acke s ep esen he s uc u e o
ASCs as a gued o by Belle i (1990) and Hemanz (1991)
(I
omi any e e ence o a CP, TP
o AspP, o simpli y):
7
A passi e o med wi h a lip e b (i.e. a e b in which he pa ien ge s nomina i e Case and he expe ience
ge s da i e) is iU- o med,
as
shown in (i). As poin ed ou by an anonymous e iewe , ASCs o med wi h a lip
e b a e also ill o med, as shown in (ii):
(i)
*Lm
cla eles son gus ados po
Juan.
he ca na ions a e liked by
Juan
(ii) *Gus ados 10s cla eles po
Juan
...
liked
he ca na ions by
Juan
I
do no ha e an explana ion o he ung amma icalii y o (i) and (ii). In any case, he ac ha lip e bs a e
unaccep able in bo h ypes o cons uc ions does no en ail ha one is a sub- ype o he o he . As
I
explain
in
he
main ex , supp essing a 8- ole and no gene a ing a 0- ole a e dis inc p ocesses ha gi e ise o cons uc ions
wi h he same p opehes.
Assuming he s uc u e o ASCs ep esen ed in (12a1), we can p edic ha he subjec s o
ansi i e e bs a e no licensed because he e is only one Case posi ion
-
[Spec,Ag OP]-
and wo a gumen s. [Spec,Ag OP] is aken by he objec , and he subjec does no ha e a
place whe e i checks i s case and $- ea u es and canno
be
licensed.
Howe e , (12a') does no ule ou une ga i e p edica es: since hey do no ha e an objec ,
no hing p e en s hei only a gumen om being licensed in [Spec,Ag OP]. No hing explains
he ung amma icali y o (2 ).
As Collins (p.c.) poin s ou , Kayne's (1993) p oposal could
be
ele an a his poin . He
sugges s ha une ga i e p edica es con ain an emp y complemen ha aises o [Spec,Ag OP]
o check ea u es. Kayne p oposes his emp y objec . o make su e ha he subjec o an
une ga i e e b does no s op a [Spec,Ag OP] and igge s ag eemen on he pa iciple in a
ypical pe ec i e cons uc ion in Romance. Wi h he s uc u e p oposed in (16) his p oblem
does no a ise. The subjec o an une ga i e p edica e is gene a ed in [Spec,VPl], ha is, a a
posi ion highe han [Spec,Ag OP], he e o e he subjec o an une ga i e canno check
ea u es in [Spec,Ag OP] unless i lowe s, a mo emen disallowed o p incipled easons.
Mo eo e , he claim ha languages like English and F ench can license an emp y objec
needs elabo a ion, since so a all he a ailable e idence has poin ed ou in he opposi e
di ec ion (see, o ins ance, Rizzi (1986)).
I
conclude he e o e, ha Kayne's idea is no a
iable al e na i e o he analysis de ended in his pape .8
8
Inciden aily, Kayne's p oposal wn adic s he analysis o une ga i es de eloped by Hale
and
Keyse
(1991).
Hale
and
Keyse p opose iha une ga i es con ain
an
inw po a ed objec which, ollowing cmen assuinp ions
on ihe na u e o inco po a ion, en ails ha i mo check ea u es in [Spec,A@P].
Belle i (1990) assumes a s uc u e like (12a') and, acknowledging he p oblem p esen ed, she
o e s a solu ion. Howe e , he solu ion is based on assump ions di e en om mine. As I
men ioned abo e, she a gues ha he e is a CP domina ing Ag P in ASCs. Fu he , she claims
ha C con ains a Case ea u e. When he une ga i e pa iciple, also con aining a Case ea u e,
aises o
C
(a mo emen ha she a gues is necessa y o license he NP) a Case con lic a ises.
The assump ion o a CP iola es one o he basic hypo heses o his pape , namely, he no ion
ha he s uc u e o ASCs is a subse o he s uc u e o o dina y clauses. Mo e on his in
sec ion
8.
Mo eo e , Belle i mus assume ha une ga i e e bs ha e an accusa i e Case o assign. In
cu en concep ions o Case Theo y -as in Chomsky (1993)-, Case is ega ded as a ea u e
ha mus
be
checked. Unchecked ea u es gi e ise o c ashing de i a ions. Assuming his, i
une ga i e e bs ha e a Case ea u e ha is no checked, hen he de i a ions o une ga i e
e bs will ne e con e ge.
5.
Wo d O de
5.1.
English
and
F ench
In his sec ion
I
um o
44,
he wo d o de issue. As we saw in (2h), epea ed he e, he o de
pa iciple+NP is obliga o y. This con as s wi h he si ua ion in o he languages: F ench and
English pa icipial absolu es mus ha e he o de NP
+
pa iciple, as can
be
seen in (17):
(2)
h. *Los ínculos amilia es o os, 10s adolescen es se sumen en la dep esi6n.
he bonds amilia b oken he adolescen s
SE
sink in o he dep ession
(17) a. All hings conside ed, he si ua ion is no ha bad.
b. Dinne inished, we le o he ope a. ( om Ko mann (1991))
c. Ceci di , la kunion a pu commence .
his said he mee ing has could s a
(17) d.
Le
cha pa i, les sou is dancen .
he ca gone he mice dance
A good answe o Q4 should y o encompass (2h) and
(17).
Le us i s conside he English
and F ench cases. Main clauses in English and F ench ha e obliga o y p e e bal subjec s.
Following Chomsky
(1993),
his means ha hei 'ou e ' Tense p ojec s s ong N- ea u es.
As we saw in (14) and (IS), e idence seems o indica e ha he 'inne ' Tense also has s ong
N- ea u es ha a ac an objec NP by spell-ou in English. The e o e, he o de
NP+pa iciple in English ASCs is he esul o o e mo emen o he NP o [Spec,TP]. By
hypo hesis, he same analysis would ex end o F ench. The logic o his a gumen leads o he
conclusion ha he 'inne ' T in Spanish has weak N- ea u es and ha is why i does no aise.
5.2.
Spanish: he ThemelRheme Dis inc ion
As
I
discuss in mo e de ail in Sec ion
9,
Belle i, De Miguel and Hemanz p opose ha he NP
mo es ou o he VP o [Spec,Ag P], he pa iciple mo ing o a posi ion highe han Ag P.9
The o de pa iciple+NP can be ob ained wi h se e al di e en scena ios, gi en he s mc u e
p oposed he e. The NP could emain
in si u
and he pa ic~ple could aise o ei he Asp, Ag
o
T.
The NP could make a sho mo emen o [Spec,AspP] and he pa iciple s ill has wo
possible su ace posi ions. Finally, he scena io p e e ed by he ci ed au ho s
IS
one
n
which
he NP is in [Spec,Ag OP] and he pa iciple is in a highe posi ion.
I
do no ha e any hing o add conceming he su ace posi ion o he pa iciple. Le us assume
ha he p edica e ha heads an ASC has an ex a ea u e, call i [absolu e], ha igge s o e
A
head
T
o
De Miguel,
C
o
Hemanz and Belle i.
64
mo emen o Asp, whe e he ea u e is checked. A his poin , his ea u e is a me e
s ipula ion, bu in sec ion
6.
I show ha i is necessa y o a ull accoun o ASCs.
As o he NP, I p esen a piece o e idence ha sugges s ha he NP emains
in
si u
in
Spanish. In (18a), we can see ha an NP subjec in an o dina y clause can s and a quan i ie
when i aises and in (18b) we can see ha his is possible e en when he NP aises o a
posi ion o he igh o he e b -[Spec,TP], p obably. In ASCs, e en hough he subjec can
ha e a uni e sal quan i ie , as in (I%), his quan i ie canno
be
s anded, as in (18d). The
conclusion is ha he NP in
an
ASC does no aise ou o i s base gene a ed posi ion:
(18) a. Los homb es decidie on odos queda se en casa.
he men decided-3pl all s ay-SE a home
b. Saluda on 10s embajado es odos ai hijo del mona ca.
g ee ed3pl he ambassado s all o- he son o - he mona ch
c. Saludados odos 10s embajado es, el hijo del mona ca son i6 a las cáma as.
g ee ed ai1 he ambassado s he son o - he mona ch smiled o he came as
d. *Saludados 10s embajado es odos, el hijo del mona ca son i6 a las cáma as.
g ee ed he ambassado s all he son o - he mona ch smiled o he came as
Le us ake a look a he posi ion o subjec s in.o dina y clauses. O dina y clause subjec s in
Spanish do no need o aise o [Spec,Ag SP] in he o e syn ax. Conside he ollowing
g amma ical sen ences:
(19)
a. Vino Ped o.
came Ped o
'Ped o came.'
b. Tele one6 Ped o.
called Ped o
'Ped o cailed.'

(19) c. iComp 6 Ped o las pa a as?
bough Ped o he po a oes
'Did Ped o buy he po a oes?'
d. Las pa a as las comp 6 Ped o.
he po a oes hem bough Ped o
'Ped o bough he po a oes.'
e. Fue on a acados 10s ca gue os.
we e a acked he ca gos
The ca gos we e a acked.'
I seems ha Spanish subjec s do no aise in o [Spec,Ag SP] o e ly. I p opose ha Spanish
Ag S has weak N- ea u es and subjec s do no need o aise un il LF o check hei
mo phological ea u es.
I
sugges ha Spanish Ag O also has weak N- ea u es, so he NP
gene a ed in [Spec,VP2] does no need o aise o [Spec,Ag OP] un il LF. This hypo hesis
would explain why he NP o an ASC
can
s ay
in
si u.
Howe e , he subjec s o o dina y
clauses can always aise. Wha is le o explain is: (i) wha igge s NP mo emen in o dina y
clauses, (ii) why he subjec s o ASCs
mus
s ay
in si u.
Why do Spanish subjec s aise? One possibili y could be o assume ha [Spec,Ag SP]
op ionally p ojec s s ong ea u es in Spanish.10 An al ema i e is sugges ed in Con e as
(1976) and Suñe (1982). They show how he SV o de makes he subjec heme, whe eas he
VS makes i heme. Fo he pu poses o his pape ,
I
adop he no ion o Suñe (1982:4):
heme is he pa o a sen ence ha is asse ed, heme is he pa ha alls ou o he scope o
asse ion.
l0
Chomsky
(1993)
p oposes his o S anda d A abic. Howe e , he case o A abic is di e en om Spanish:
when he subjec is
in
si u
in
hbic, he e b does show poo e mo phology.
Le us see i wi h an example. In (20) he e a e wo possible con e sa ions. In (20a), he
subjec is p e e bal, hence heme, whe eas he e b is heme. This means ha he issue ha is
alked abou is wha Juan is doing, and we a e asse ing ha he is coming. In (20b), he
subjec is heme, so he issue is whe he Juan o someone else is coming:
(20) a. A: C eo que Juan iene.
'I
hink ha Juan is coming.'
B: No, se queda.
'No, he's s aying.'
a. B': *No, Ped o.
'No, Ped o.'
b.
A: C eo que iene
Juan
'I hink ha Juan is coming.'
B: *No, se queda.
'No, he's s aying.'
B': No, Ped o.
'No, Ped o.'
I sugges ha o e NP aising in Spanish o dina y clauses is igge ed by his seman ic
eason:
NP
subjec s aise in o de o make he dis inc ion be ween heme and heme: heme
being he posi ion(s) ex e nal o he VP, heme he posi ion(s) in e nal o he VP.ll
Ten a i ely,
I
sugges including he hemel heme dis inc ion in he Discou se Rep esen a ion
Theo y adi ion s emming om Heim (1982) and he Mapping Hypo hesis in Diesing
(1992).
Gi en a logical ep esen a ion o he o n Op, [P(x)] Q(x), he es ic i e clause is he non-
-
-
In he mos ecen e sion o he
P incipies
and Pa ame e s Theo y, ha o Chomsky
(1993).
mo emen is
only igge ed by mo phological eame checking. Seman ic o discou se ac o s a e no supposed o play a ole.
This en ails ha my p oposal in his sec ion does no i in o he Minimalis
P og am
as is. Howe e , i seems o
me ha Chomksy's s ong posi ion ill ha e o
be
e ised in o de o ein oduce a leas Quan i ie Raising back
in o he hw y. A emp ing o do his is ou o he scope o his pape .
asse ed and he nuclea scope is he asse ed pa .12 Syn ac ically, hema ic subjec s will
show up in he Spec o a high unc ional ca ego y, hema ic ones inside he
VP
p ojec ion.
My claim conce ning ASCs is ha , being oo small o a ipa i e s uc u e, hey canno
license any o he ope a o s (as o ins ance he gene ic ope a o ) ha ake scope in
a
quan i ica ion s uc u e. This has he ollowing se o p edic ions: (i)
IL
p edica es a e no
licensed in ASCs (see sec ion
2),
(ii) gene ic subjec s a e no licensed in ASCs, (iii) lacking a
mo i a ion o aise, he subjec s o ASCs should s ay
in
si u.
In he nex pa ag aphs I look a
he in e p e a ions o NPs in o dina y clauses in p e e bal and pos e bal posi ion and in
ASCs.
I
show how all hose p edic ions hold.
Le us i s conside he beha io o inde ini e subjec s in o dina y and small clauses.13 In
o dina y clauses, p e e bal inde ini es ge a gene ic eading, and hey canno ge an
l2 As h4olly Diesing poin s ou , no all discou se phenomena can be ep esen ed in a quan i ica ion s uc u e.
The e is, howe e , some o e lap, which I exploi he e.
l3 The discussion conce ning (21). (22). (24) and (25) should be conside ed a ough ske ch whe e some
a iables a e le ou . An impo an a iable, poin ed ou by he e iewe is pe ec i i y. inde ini e
NPs
canno be
gene ic in a pe ec ense, e en i placed be o e he e b:
(i)
IJn
espaiiol Ueg6 con e aso.
a Spania d ani ed wi h la e
The e iewe sugges s ha he ahsence o gene ic eadings in ASCs is ied o pe ec i i y -i has been no ed
ha ASCs mus ha e he pe ec i e aspec (see De hliguel (1990, 1992). Hemanz (1991)). Howe e , in a ull
dause, an
NP
wi h a de ini e de e mine
can
ha e a gene ic eading e en in he pe ec :
(ii) Desde iempos inmemo iales 10s espaíioles han, llegado a de
a
sus ci as.
om imes immemo ial ihe Spania ds ha e ani ed la e
o
hei appoin men s
Howe e , in he con ex o an ASC. NPs wi h a de ini e de e mine canno ha e a gene ic eading (see
(26)).
So,
he ac emains ha gene ic eadings a e no a ailable in ASCs and ha his is no a by p oduc o pe ec i i y. I
seems ha a a ie y o ac o s in iuence gene ic and exis en ial eadings in main clauses, some o which I ha e o
abs ac away om un i1 hey a e be e unde s ood.
68
exis en ial one, as seen in (21). On he o he hand, pos e bal inde ini es ge an exis en ial and
no a gene ic eading,
as
seen in (22):14
(21)
Un espaiiol llega con e aso.
a Spania d a i es wi h la e
*&
x
a Spania d
A
x is la e
Gen, [x is a Spania d] x is la e
(22) Llega con e aso un espaiiol.
a i es wi h la e a Spania d
&
x is a Spania d
A
x is la e
*Gen, [x is a Spania d] x is la e
Inde ini e subjec s in ASCs mus also
be
exis en ial. In example
(23)
I
do no p o ide he
main clause so ha he eade makes su e ha
I
a n no a o ing one in e p e a ion o e he
o he :
(23)
Llegado con e aso un espaiiol,
...
a i ed wi h la e a Spania d
&
x is a Spania d
A
x is la e
*Gen, [x is la e] x is la e
Exa nples (24) h ough (26) show he same poin : a plu al de ini e
NP
ecei es a gene ic
eading in p e e bal posi ion, bu no in pos e bal posi ion (whe e he eading hey ge is
mo e like a uni e sal quan i ie ):
l4
Rega dless o whe he he e a e o he cons i uen s be ween he e b and he subjec .
I
assume ha when he
VOS
o de ob ains, he subjec is igh adjoined o he
VP,
as in Bone
(1989).
Rhema ic eadings a e no ela ed
o one posi ion
in
pa icula , he only p o iso being ha hey be loca ed low enough o a oid being mapped on o
he es ic i e clause.
I seems ha he Case ha he p edica e o a ASC checks is some kind o de aul Case. This
could ce iainly be a gued o La in, since abla i e is a synch e ic Case ha emb aces
loca i es, oblique agen i es and ins umen als.
This ype o solu ion i s in o he p og am se ou in he in oduc ion o his pape . Though he
hie a chy o unc ional ca ego ies does no a y, he ea u es o he unc ional ca ego ies do
a y, gi ing ise o c osslinguis ic a ia ion and, as
I
ha e shown in his sec ion, also
in alinguis ic. The p oblem o a leas his pa icula ype o 'sub-g amma ' (Ha is
(1968))
-i.e. he ac ha ce ain cons uc ions seem o ha e a g amma ha in e sec s ins ead o
being p ope ly con ained in he g amma o he language as
a
whole- is esol ed by posi ing
ea u e a ia ion a he han s uc u e a ia ion.
We s ill ha e one mo e w inkle o i on ou . T ansi i e e bs a e assumed o ha e an
accusa i e Case o check. I e bs a e gene a ed wi h his Case, hen ASCs ha e wo Cases
a ailable: nomina i e Case o [absolu e] in Asp and he accusa i e Case o he e b. This
p oblem becomes mo e appa en han eal i we inco po a e Collins and Th &insson s
(1993)
Case heo y. D awing on Icelandic da a, hey conclude ha
V2
does no ha e a Case ea u e,
and ha accusa i e Case is ac ually bo ne by
V1,
which i sel p ojec s an Ag OP - hus, he e
a e wo Ag OPs in he clause, one p ojec ed by
V1,
whe e Case is checked, and he o he is
p ojec ed by
V2,
a place whe e + ea u es a e checked, bu no case because
V2
has no Case
o
check. ~he e o e, he pa iciple o ASCs, which is a
V2,
ne e has any accusa i e Case o
l6
AS
he e iewe poin s ou , some hing should be said abou wha
I
e med AFCs:
(i) Habiendo Juan disuel o el sul u o, el expe imen o pudo ealiza se en menos de dos ho as.
ha ing
Juan
dissol ed he sulphu he expe imen could accomplish-SE
in
less o wo hou s
The ques ions should
be
asked conce ning (i): (a) does he ge undi e in (i) ha e
an
[absolu e] ea u e?.
@)
does
he subjec o he
AFC
check Case agains his ea u e [absolu e]?
I
do no ha e any e idence in Spanish o
answe any o hose ques ions. Howe e , conside a ion o c osslingus ic
&ia
could gi e us a clue. La in and
Classical G eek ha e AFCs wi h ac i e pa iciples -a
ali
e ec s, equi alen o AFCs like (i) because hey admi
wo a gumen s. In hese cons uc ions he subjec
NP
exhibi s abla i e Case in La in, geni i e in G eek,
and
hese

7.
Mid-Way Conclusions
Now we a e in a posi ion o answe Q1-5. The ema kable p ope ies o ASCs ha e been
shown o be de i ed om hei 'smallness': hei unc ional s uc u e includes Ag OP and
excludes any o he highe ca ego ies, which has he ollowing consequences: (i) he subjec s
o IL p edica es lack a Spec posi ion whe e hey can be licensed, ollowing Diesing's (1992)
heo y; (ii) since he uppe e bal head is no gene a ed, 'deep' subjec s canno be gene a ed
ei he ; (iii) he wo d o de pa iciple+NP is obliga o y because he o de NP+p edica e in
Spanish is he esul o aising a hema ic subjec o a es ic i e clause in Heim's (1982) sense
and his is una ailable in ASCs. Finally, he NP can check Case wi h an ex a ea u e in Asp,
absen in o dina y clauses.
In he nex sec ion I es he hypo hesis ha ASCs should ha e he same s uc u e
c osslinguis ically agains he I alian da a p esen ed in Belle i (1990). Though he esul s
ob ained in his pa o my esea ch a e en a i e, I belie e ha his hypo hesis gi es us be e
empi ical p edic ions han assuming a di e en s uc u e -as a gued o , in a Rela ional
G amma amewo k, by Addis (1993).
a e he Cases ha he subjec o an ASC ge s. On he o he hand, he objec ge s accusa i e Case. (u) is a La in
example ( om Allen and G eenough (1991)) and (ui) is G eek ( om Smy h (1984)):
(u)
E0
impe i-um enen e. e en um imeo.
he-Abl-sg powe -Acc-sg holding-Abl-sg ou come em-lsg
Wi h him in powe ,
I
ea he ou come.'
(iii)
Tosau -a eipon -o:n -o:n pla ai-o:n, A khidam-os
so-many- hings-Acc-p1 ha ing-said-Gen-pl he-Gen-pl pla aean-Gen-pl A chidamus-Nom-sg
hupolab-o:n eipen.
answe ing-Nom-sg said
'Ha ing he Pla aeans said so
many
hings, A chidamus eplied.'
Since hese subjec s do no ge he expec ed nomina i e o he subjec o a main clause bu he Case o he
NP
o
an
ASC, i seems plausible o assume ha he same ea u e ha checks abla i e o geni i e in ASCs checks
abla i e o geni i e in AFCs.
Mu a is mu andis,
i could
be
a gued ha
in
(i), and in gene al in AFCs in any
language, he subjec is licensed by he ea u e [absolu e].
8.
I alian T ansi i e ASCs
Appa en ly, I alian unaccusa i e ASCs wo k he same way he Spanish do. (34) is an
exa nple:
(34) Am a o io, la eunione
6
cominciaia.
am ed I he mee ing is sia ed
Howe e , ansi i e ASCs di e signi ican ly. In he i s place, hey do no ha e he 'passi e
look' ha ansi i e ASCs ha e in Spanish: he NP ecei es accusa i e Case, as we can see in
(35) and by-ph ases a e disallowed, as
can
be
seen in (36) ( om Belle i
(1990)):
(35) Salu a a mel*io, si
acco o che c'e a mol a al a gen e.
g ee ed me I SI is ealized ha he e-we e many o he people
(36)
.
Salu a a me (*da Gianni), u i usci ono dalla saia.
g ee ed me by Gianni e e ybody le o - he oom
As
I
men ioned abo e, i is he case ha [absolu e] checks di e en cases c osslinguis ically,
he a ia ion being mo e o less unp ediciable.
Belle i a gues ha ansi i e ASCs a e con ol shuc u es, i.e., apa om he o e NP he e is
also ano he a gumen PRO. Acco ding o he heo y o clause s uc u e laid ou he e, his
ex a a gumen would ha e o
be
gene a ed in he Spec o VP1. Though i is heo e ically
possible ha I alian ansi i e ASCs had mo e s uc u e han he Spanish ones (Addis
(I=)),
I a gue he e ha hey do no . Le us e iew and c i ique Belle i's a gumen s o PRO.
Fi s , acco ding o Belle i, his PRO has o
be
con olled by he subjec o he main clause.
The ung amma icali y o (37a) con as s wi h (37b) because in (37a) he PRO is no
con olled:
(37) a. *Chiama o i1
axi,
smise de pio e e.
called he axi s opped3sg o ain
b. PRQ chiama o i1
axi,
Ma iai usci.
called he
axi
Ma ia wen -ou
Howe e , Rosen (1988) shows ha con ol in I alian ASCs can be pu ely p agma ic. In (38a),
i is p obably he judge who absol ed he de endan , so a case could
be
made o subjec
con ol, bu in (38b) he judge o ju y is absen om he linguis ic con ex :
(38) a. Assol o I'impu a o, i1 giudice se n'andb a p anzo.
absol ed he-de endan he judge SE NE-wen o lunch
b. Assol o l'impu a o, scopia ono gli applausi.
absol ed he-de endan bu s -ou -3pl he applauses
(38) weakens Belle i's a gumen because p agma ic con ol does no equi e a PRO subjec .
(39) is he Spanish equi alen o (37). In Spanish, howe e , he pe son who calls he axi is
no necessa ily Ma ia -i could be he ho el po e , o ins ance:
(39) a. *Llamado el
axi,
pa 6 de Ilo e .
called he axi s opped3sg o ain
b.
[e]i/j
llamado el
axi,
Mm'ai sali6.
called he
axi
Ma ia le
In I alian, like in Spanish, he e o e, ansi i e ASCs a e no con olled: he cons uc ion
allows o an a bi a y subjec o pe o m he ac ion - he mi o image o a e b like ea
which, in i s une ga i e usage, allows o any a bi a y hing o be ea en. As o he
unaccep abili y o (37a). I belie e i is due o a lack o ob ious seman ic connec ion be ween
he main and he absolu e clause and no o any g amma ical eason.
Belle i's second a gumen is ha he o e NP can
be
a e lexi e:
(40)
a. Elogia a solo se s essa, Ma ia es b del u o isola a.
p aised only he sel Ma ia emained comple ely isola ed
b. La a asi le mani, Ma ia
2
usci a.
washed-he sel he hands Ma ia is le
She claims he e has o
be
a PRO a gumen in he ASC ha binds he e lexi e. Howe e , his
is no he only analysis a ailable. I he e is no 'deep' subjec , ASCs do no con ain
an
accessible SUBJECT, he e o e he binding domain o he anapho is he main clause and no
he ASC. My I alian in o man s ce ainly ag ee ha
se s essa
('he sel ) has o be co e e en
wi h
Ma ia,
which con i ms my hypo hesis.17. Thus my claim is ha we do no need o posi
a PRO in ASCs any mo e han we need i in he adjunc ph ase
nea each o he
in he
sen ence
Nea each o he hey saw snakes
( hanks o Wayne Ha be (p.c.) o sugges ing his
exa nple o me). On he o he hand, he claim ha he e is a PRO in 'deep' subjec posi ion
aises an empi ical p oblem: why can' he e be une ga i e ASCs? The subjec o he
une ga i e could be gene a ed as PRO and a sen ence like (41) should be g amma ical. Bu
(41)
is ung amma ical in I alian, as i is in Spanish:
(41)
*PRO,
ele ona o, Giannii andb all appun amen o.
elephoned Gianni wen o- he-appoin men
The e o e, he easons adduced by Belle i o ha e a PRO in ansi i e ASCs do no seem so
con incing. On he o he hand, i seems ha he e a e good easons no o ha e a PRO, and
l7
I alian
ASCs a e he e o e anspa en o ex e na1 binde s, whe eas he Spanish ones a e no .
I
conside his a
consequence o he ac ha in Spanish absolu es he
NP
bea s nomina i e Case and nomina i e anapho s a e
disaiiowed.
80
s ick o ou analysis o ASCs wi hou a VP1. I seems, he e o e, ha my hypo hesis ha he
s uc u e o absolu e clauses does no
a y
c osslinguis ically can
be
main ained.
9.
The P oblem o
a
CP Node
in
ASCs
9.1
Belle i's A gumen s
o
a
CP
in
ASCs
In he emainde o his pape , I will discuss He nanz's ( o Spanish) and Belle i's ( o
I alian) claims ha he Ag P is domina ed by a
CP.
I
Hemanz's and Belle i's claim we e o
hold, i would se iously h ea en my hypo hesis ha he s uc u e o ASCs is p ope ly
con ained in he g amma o o dina y clauses. The su ace s uc u e ha hey o e o ASCs
is he ollowing:
I
NP
Ag '
n
AspP
n
i
"i'
A
The e a e h ee poin s ha need o
be
discussed wi h espec o (42):
1
1.
Whe he he e is di ec e idence o a CP in ASCs. This will occupy he es o his
subsec ion.

2.
Whe he p edica e aising can p o ide indi ec e idence o a CP in ASCs. This will
occupy subsec ions 9.2 and 9.3.
3. Whe he NP mo emen p o ides indi ec e idence o a CP in ASCs. This will occupy
subsec ion 9.4.
Belle i p o ides wo ypes o e idence o a CP in I alian ASCs. Fi s , she claims ha he ee
ela i e
quan o
('wha ') can in oduce an ASC,
as
in (43):
(43) Quan o comunica oci,
...
wha communica ed-us
Unde cu en assump ions on he licensing o wh-wo ds, he mos likely analysis o (43)
should include a [Spec,CP] ha would house
quan o.
Howe e ,
quan o communica oci
is an
in ini i al ela i e and canno
be
used
as
an adjunc .18 (44) is an example o i s use:
(44) Quan o comunica oci
ci
ha dis u ba o mol o.
wha communica ed-us us has dis u bed a-lo
Belle i's second a gumen is ha I alian ASCs can
be
in oduced by
anche
se
('e en i ') and
benchk
('al hough'), which she akes o
be
complemen ize s. This en ails ha he e mus be a
C
whe e hey can be gene a ed. In (45) a e wo examples:
(45) a. BenchC pa i a da sola, Ma ia si di e i.
al hough le by alone Ma ia SI had- un
b.
Anche
se
am a a in i a do, Ma ia non si scusb.
e en i am ed in la e Ma ia no SI apologize
Howe e , bo h
benché
and
anche se
may in oduce o he ypes o ph ases in which i is no
assumed he e o be a CP:l9
(46)
a. Ques o
spiega o in una manie a de aglia a bench6
[Ap
non chia a].
This is explained in a de ailed hough no clea manne .'
b. Ques o agazzo, anche se s upido, euscid a consegui e la lau ea.
This young man, e en i s upid, will manage o ob ain he BA.'
The examples in
(46)
show ha he e is no a one- o-one co espondence be ween hese wo ds
and he p esence o a
C
posi ion, he e o e hey do no p o ide a s ong eason o assume ha
he e is a CP in I alian ASCs. A ca e ul analysis o
anche se
('e en i ) and
benché
('al hough')
is ou o he scope o his pape . Pending u he esea ch,
I
sugges hey could
be
analyzed as
adjoined elemen s.
9.2.
Does he Pa iciple Raise in o
C?
He nanz and Belle i ag ee ha he pa iciple aises o C, hough hei a ionale o he
mo emen di e s. I hei claim we e demons a ed, hen we woul-d ha e indi ec e idence
ha he e is a
CP
in ASCs. Howe e ,
I
will a gue agains hei analyses.
He nanz shows ha he e canno
be
an ASC as complemen o a subo dina ing p eposi ion
like
po
(' o '),
pa a
('in o de o, o '),
sin
('wi hou '), in con as o he e y simila
in ini i als:
(47)
a. Sin comp a comida, no e a e as a eni .
b. *Sin comp ada comida, no e a e as a eni .
'Wi hou buying ood, do no da e come.'
l9
Thanks o Ca ol Rosen o p o iding hese examples.
(47)
c. Pa a comp a comida, e al supe me cado.
d. *Pa a comp ada comida, e al supe me cado.
'In o de o buy ood, go o he supe ma ke .'
e. Po hace a esu as, e quedas cas igado.
.
*Po hechas a esu as, e quedas cas igado.
'Fo doing mischie , you will be punished.'
He a gumen is ha , assuming ha hese p eposi ions occupy he CO posi ion, he eason why
hey a e ung amma ical in ASCs is because he pa iciple aises in o CO, he e o e wo
elemen s a e compe ing o he same posi ion. Howe e , unde he same asump ions, i could
al e na i ely be sugges ed ha i he e is no C in ASCs, he e canno be subo dina ing
p eposi ions ei he , which would nea ly accoun o he ung amma icali y o (47b,d, ). On he
o he hand, i is well known ha in Romance p eposi ions appea o selec o CPs a he han
occupy he Coposi ion:
(48)
a. No s& c6mo echaza lo sin que pa ezca g ose o.
'I don' know how o ejec him, wi hou ha i seems ude.'
b. No s&
c6mo disimula lo pa a que no pa ezca g ose o.
'I
don' know how o hide i o ha i doesn' seem ude.'
Mo eo e , he e may
be
o he easons why (47b,d, ) a e ung amma ical. Le us conside he
abo e p eposi ions a bi mo e ca e ully.
Sin
('wi hou ') can ake NPs as complemen s, bu no
adjec i es o ad e bs:
(49) a. Sin la comida, no e a e as a eni .
'Wi hou he ood, do no da e come.'
b. *Sin cansado, no e a e as a eni .
wi hou i ed no da e o come
(49) c. *Sin ápidamen e, no e a e as a eni .
wi hou quickly no da e o come
I seems ha
sin
equi es an NP complemen .
I
claim ha he in ini i al in (47a) con ains
nominal ea u es ha make i , in some in ui i e sense,
like
a noun. Since ASCs a e no
nominal bu adjec i al o ad e bial in na u e, he sou ce o he ung amma icali y o (47b) is a
iola ion o he selec ional p ope ies o he p eposi ion
sin.
Le us ake a look a
po
and
pa a
(bo h meaning ' o '). Selec ional p ope ies a e again
c ucial o accoun o he ung amma icali y o (47d) and (47 ).
Po
and
pa a
may ake
adjec i es as complemen s, bu only i hey o m IL p edica es, ne e i hey o m SL
p edica es:
(9)
a. Pa a on o, mi he mano.
o silly my b o he
'My b o he is an example o a silly pe son.'
b.
Po on o pe dis e el en.
o silly missed-2sg he ain
'Because you we e silly, you missed he ain.'
c. *Pa a u ioso, mi he mano.
o ang y my b o he
'My b o he is an example o an ang y pe son.'
d. *Po u ioso, e cas iga emos.
o ang y youpunish-FUT-lpl
'Because you we e ang y, we will punish you.'
I seems ha he gene aliza ion ha
po
and
pa a
canno selec o SL p edica es is clea .
Since ASCs mus
be
SL p edica es (see
(3)
and discussion in sec ion 2.4.), i ollows ha hey
Pese sky, D.
(1989)
'Language Pa icula P ocesses and he Ea liness P inciple', ms. MIT.
Pollock, J.-Y.
(1989)
'Ve b Mo emen , Uni e sal G amma and he S uc u e o IP',
Linguis ic Inqui y
20,365-424.
Pos al,
P.
(1974)
On Razking,
The MIT P ess, Camb idge.
Rigau, G.
(1991)
'La
legi imaci6 de les cons uccions empo als d'in ini iu', ms. Uni e si a
Au bnoma de Ba celona.
Rizzi,
L.
(1986)
'Nul1 Objec s in I alian and he Theo y o p o',
Linguis ic Inqui y
17,
501-
557.
Rosen, C.
(1988)
The Rela ional S uc u e o Re lexi e Clauses: E idence iom I alian,
Ga land, New Yo k.
Smy h, H.
(1984)
G eek G amma ,
Ha a d Uni e si y P ess, Camb idge.
Spo iche, D.
(1988)
'A Theo y o Floa ing Quan i ie s and I s Co olla ies o Cons i uen
S uc u e',
Linguis ic Inqui y
19,425-450.
S owell,
T.
(1981)
O igins o Ph ase S uc u e,
PhD disse a ion, MIT.
S ump, T.
(1985)
The Seman ic Va iabili y o Absolu e Cons uc ions,
Reidel, Do d ech .
Suñe ,
M.
(1982)
Syn ax and Seman ics o Spanish P esen a ional Sen ence-Types,
Geo ge own Uni e si y P ess, Geo ge own.
T a is,
L.
(1991)
'De i ed Objec s, Inne Aspec and he S uc u e o VP', pape p esen ed a
NELS
22.
Zagona,
K.
(1994)
'Pe ec i i y and Tempo al A gumen s' in
M.
Mazzola (ed)
Romance
Linguis ics: Theo y and Issues.
203
Mo ill Ha11
Co nell Uni e si y
Z haca
NY
14853
USA
E-mail: [email p o ec ed]