...

Carol Petersen (University of Maryland) [email protected]

by user

on
Category:

travel

30

views

Report

Comments

Transcript

Carol Petersen (University of Maryland) [email protected]
Carol Petersen (University of Maryland)
[email protected]
Control in Subjunctive Clauses in Brazilian Portuguese: Evidence for Tense Defectiveness
It is standardly assumed referential null subjects in BP are different from those of typical prodrop languages. Null subjects in BP indicatives are allowed only in embedded clauses, and
display all the diagnostics of obligatorily controlled (OC) PRO. Based on Hornstein’s (2001)
Movement Theory of Control, they have been analyzed as traces of A-movement (Ferreira 2009,
Rodrigues 2004). Ferreira, e.g., proposes finite Ts in BP are ambiguous in that they can enter the
derivation specified as -complete or -incomplete, respectively yielding standard nominative
Case marked embedded subjects or OC null subjects derived by movement (1). By contrast, null
subjects in BP subjunctives do not always behave in this way. Control structures involving
subjunctives are allowed in the complement clauses of dubitative and factive-emotive predicates
(2)-(3) but banned from subjunctive complements of volitional predicates (4). Significantly,
subjunctive complements of volitional predicates in BP (and many other Romance and Slavic
languages) also show subject obviation, where the pronominal subject of the subjunctive clause
must be disjoint in reference with the matrix subject (5). Thus, given that finite indicative clauses
in BP may constitute a porous domain for A-movement due to their defective -specification (T+,
-), the restrictions on finite control into subjunctive clauses seem rather unexpected. With this
picture in mind, in this paper I address the following issues: [1] why is OC not allowed
exclusively in obviative subjunctives in BP? [2] What do BP data reveal about obviation effects?
I follow Hornstein’s (2010) approach to Principles A/B, where reflexivization is a result of
movement, with reflexives spell out of traces. When movement is not possible, a pronoun is
inserted to get the bound reading. For Hornstein, Move (i.e. reflexivization) is cheaper than
Construe (pronominalization). Obviation then falls out as a consequence of economy of
derivation. I argue that BP facts provide empirical evidence that obviative subjunctives are
underlying infinitives since they compete derivationally. I propose that sentences like (4), (5) and
(6) involve untensed domains that share the same syntactic properties, regardless of their
morphological distinctions (for which I will provide an account in the talk). More specifically,
they are defective Tense domains (T-) that need to agree in Tense value with the matrix clause.
Once (5a) and (6) share the same numeration, they compete. The unvalued embedded T in the
relevant step of the derivation is not allowed to Case mark its subject, being a defective probe
(Chomsky 2001), which allows the DP to move to the matrix clause for Case requirements. Thus,
the movement alternative is the only convergent result starting from a numeration with a single
DP for a subject and it surfaces as an infinitive (6). The same considerations hold for (5b), the
only difference being that (5b) starts with a numeration with two distinct DPs. After getting
valued by the matrix T, the embedded T can value the Case of its subject, which leads to a
nominative embedded subject and subjunctive morphology on the verb.
Back to questions [1] and [2], BP data show that a -defective value in T does not necessarily
lead to OC. In obviative contexts, infinitives and subjunctives compete as they are both Tense
defective and share the same underlying numeration. Once the infinitive wins the competition, it
licenses OC (cf. (6)), whereas the subjunctive will be chosen just in case it does not compete with
an infinitive, i.e. when they have different numeration (more than one DP for the subject
positions; I will show that Case licensing fails for the embedded infinitive subject DP); hence the
unacceptability of (4). Thus, this proposal accounts for the tense dependency in subjunctive
complements of obviative predicates, as repeatedly attested in the literature (e.g. Raposo 1985,
Kempchinsky 1986). I show that radically different tense properties of two subjunctive
complement types in BP provide strong support for this analysis. Specifically, I show that the
obviative/non-controlled subjunctive (a) sentences of (7)-(10) patterns with infinitive
complements in that it shows morphological anaphoricity, obligatory Sequence of Tense
embedded tense interpretation, unavailability of Double Access Reading (DAR) and transparency
with respect to polarity item licensing. In contrast, non-obviative subjunctive complements that
allow OC (2)-(3) show the opposite pattern (b) sentences of (7)-(10)), behaving like indicative
complements with respect to the same properties, which confirms their independence in Tense.
Tense defectiveness is the key property that ties both obviative subjunctives in Romance and noncontrolled subjunctives in BP.
The paper concludes by discussing more supporting data for the competition analysis of
movement/infinitive and bound-pronoun/subjunctive structures that come from de se/non-de se
readings of obviative pronouns. Specifically, I will show that languages that display two series of
pronominals (pro and ello/lui), like Spanish and Italian, show strong obviation with weak
pronominals (pro) - i.e. neither de se nor non-de se readings are possible - and weak obviation
with strong pronominals (ello/lui), where non-de se readings can be obtained. By contrast, BP has
homophonous versions for these pro-forms and only displays weak obviation. I will show that
these facts can be straightforwardly explained once we consider the differences in the pronominal
status of the relevant elements and the (im)possibility of bound variable readings (cf. Hornstein
and Pietroski 2010).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
a. Joãoi disse que
elei/j/Maria vai viajar
Joãoi said that hei/j/Mary will travel
b. Joãoi disse que ti vai viajar
[TP [o João]i T-complete [vP ti disse [CP que [TP ti T-incomplete [vP ti vai viajar]]]]]
‘John said he/Mary was going to travel.’
Joãoi duvida que ti ganhe a corrida
John doubts that
win-SUBJ the race
‘Johni doubts that hei would win the race.’
chegado
tarde na
reunião
[O professor]i lamenta que ti tenha
The teacher regrets that
has-SUBJ get-PAST.PART late in.the meeting
‘The teacheri regrets that hei got late to the meeting.’
da
corrida
*O Joãoi quer/deseja que ti participe
the John wants/wishes that
participate-SUBJ of.the race
‘John wants/whishes to participate on the race.’
a.*Joãoi quer que elei leia
um livro por semana
b. Joãoi quer que elej leia
um livro por semana
Johni wants that he*i/j reads-SUBJ one book of week
‘Johni wants him*i/j to read one book every week’.
Joãoi quer ti/*j ler
um livro por semana
John wants
read-INFINITIVE one book of
week
‘Johni wants to read one book every week’
a. João quer
que Maria *ganhasse/ganhe
a corrida
John wants-PRES that Mary win-IMPERF/win-PRES the race
b. Pedro lamenta
que Ana acordasse/acorde
tão cedo
Peter regrets-PRES that Ana wake up-IMPERF/wake up-PRES so early
a. Ana exigiu que o Pedro estudasse mais naquela época
(SOT/*shifted reading)
Ana demended that Peter study-IMPERF more in.that time
b. Pedro lamentou que Ana acordasse tão cedo naquela época (SOT/shifted reading)
Peter regreted that Ana wake up-IMPERF so early in.that time
a.*Pedro queria que Ana esteja
grávida.
(*: DAR)
Peter wanted that Ana is-PRES.SUBJ pregnant
b. Pedro lamentou que Ana esteja
doente.
(OK: DAR)
Peter regreted that Ana is-PRES.SUBJ ill
a. A Maria não quer [que o Pedro converse com ninguém](OK:NPI)
the Mary not wants that the Peter talk-SUBJ with nobody
b. *A Maria não lamenta [que o Pedro saia
com ninguém] (*:NPI)
the Mary not said that
the Peter go.out-SUBJ with nobody
Fly UP