Keywords

1 Introduction

This research has received partial support from CNPq (grants 303195/2019-3 – first author – and 304954/2021-7, second author). We would like to thank Janayna Carvalho, Cláudia Coelho, Renato Lacerda, Ezekiel Panitz, Cilene Rodrigues, and Andrés Saab for valuable comments on a previous version of this chapter.

Since the seminal work by Pontes (1987), the literature on Brazilian Portuguese syntax (henceforth BP) has paid special attention to the so-called “topic-subject” constructions, that is, constructions where an apparent topic functions as a subject, controlling verbal agreement (see, e.g., Galves (1987, 1998), Kato (1989), Lunguinho (2006), Avelar and Galves (2011), Munhoz and Naves (2012), de Andrade and Galves (2014), Nunes (2016, 2017), and Kato and Ordóñez (2019)). These include, among others, constructions involving “possessor raising,” as in (1a), and “locative agreement,” as in (1b).

(1)

a.

[Os

relógios]

quebraram  

o

ponteiro.

  

the

watches

broke-3pl

the

arm

  

“The arms of the watches broke.”

 

b.

[Essas

gavetas]

cabem

muita

coisa.

  

these

drawers

fit-3pl

many

thing

  

“Many things can fit in these drawers.”

In (1a), the plural possessor os relógios “the watches” associated with the noun ponteiro “arm” surfaces in a preverbal position and triggers third-person plural agreement on the verb.Footnote 1 In (1b), in turn, the plural locative argument associated with the unaccusative verb caber “fit” is not headed by the expected preposition em “in” and is realized in a preverbal position, also triggering verbal agreement.Footnote 2

(i)

a.

The watches had their hands broken.

 

b.

European Portuguese:

  

Os

relógios tiveram

seus ponteiros

quebrados.

  

the

watches had

their hands

broken

  

“The watches had their hands broken.”

(ii)

a.

Japanese:

  

Kono tokei-wa

hari-ga

oreta.

 
  

clock-top

hand-nom

broke

 
  

“The watch had its hand broken.”

 

b.

Brazilian Portuguese:

  

O

relógio, o

ponteiro

quebrou.

  

the

watch the

arm

broke

  

“The watch had its arm broken.”

(i)

[Essas cidades]

chovem

muito no

verão.

 

these cities

rain-3pl

much in-the

Summer

 

“It rains a lot in these cities during Summer.”

Details aside, the prevailing view in the literature is that these constructions arose in BP as it ceased to be a canonical pro-drop language due to the weakening of its verbal agreement inflection and became a topic-prominent or discourse-oriented language.Footnote 3 From this perspective, as BP is in the course of becoming a non-pro-drop language, its [Spec,TP] tends to be overtly filled. In the case of unergative and transitive verbs, the external argument moves to this position, generally excluding the possibility of VS order commonly found in canonical Romance pro-drop languages. In the case of verbs lacking external arguments, VS is still allowed, as the subject may be licensed VP-internally. This in turn paves the way for elements other than the standard subject to occupy [Spec,TP] in constructions without an external argument. Given that BP also displays characteristics of a topic-prominent language, it is commonly held that topics came to be allowed to occupy [Spec,TP], yielding “topic-subject constructions” like the ones in (1).

Although the factors mentioned above certainly play a role in the emergence of constructions like (1) in BP, in this chapter we argue that they are not the primary causes of this new development in BP as they relate to portions of structure that are too high in the clausal domain. Assuming the general framework of the Agree-based model (Chomsky 2000, 2001), we argue instead that “topic-subjects” are regular subjects in the sense that they are derived by A-movement of a DP from a position within vP to [Spec,TP]. In other words, the relevant diachronic changes that gave rise to “topic-subject” constructions in BP actually involve changes in its vP and DP layers.

The chapter is organized as follows. In Sect. 2, we make some qualifications regarding the putative association of constructions like the ones in (1) with topics, showing that focus constructions display the same pattern of agreement seen in (1). In Sect. 3, we discuss diachronic changes affecting Case-licensing within vP and DP in BP, which yielded “topic-subject” constructions as a byproduct. In Sect. 4, we show why A-movement of the “topic-subject” to [Spec,TP] does not violate minimality. Section 5 discusses some fine-grained distinctions noted in Sect. 3 regarding person asymmetries and resumption in “topic-subject” constructions. Finally, Sect. 6 concludes the chapter.

2 “Topic-Subjects” Are Not Topics, But Subjects

Before we tackle the derivation of “topic-subject” constructions proper, a couple of comments are in order. “Topic-subject” constructions are not, strictly speaking, restricted to topics. The underlined constituents in (2) below, for example, involve different types of foci and the resulting constructions are as well formed as the ones in (1). The sentence in (3), in turn, shows that “topic-subject” constructions can be appropriate answers for out-of-the-blue questions, showing that they do not necessarily convey a categorical judgment (in the sense of Kuroda (1979)), as we would expect if they were true topic constructions.Footnote 4

(2)

a.

Focus with só “only”

  

Só esses relógios

quebraram

oponteiro.

     
  

only these watches

broke-3pl

the

arm

    
  

“Only these watches had their arms broken.”

 

b.

Focus with  nem “not even”

  

Nem essas gavetas

cabem

muita

coisa.

    
  

nor these drawers

fit-3pl

many

thing

    
  

“Not even these drawers are large enough.”

 

c.

Contrastive focus

  

OS RELÓGIOS

acabaram

a

bateria (não os celulares)

    
  

the watches

finished-3pl

the

battery not the cell.phones

    
  

“ The batteries of the watches (not the cell phones) are dead.”

 

d.

Aggressively non-D-linked wh-constituents

  

Que diabo de carro

vai

fundir

o

motor

depois

de

passar

  

what devil of car

goes

melt

the

engine

after

of

pass

  

pela revisão?

       
  

through-the revision

  

“What kind of car has its engine stopped soon after it leaves the garage?”

 

e.

D-linked wh-constituents

  

A: –

Que cidades chovem

muito

no

verão?

   

which cities rain-3pl

much

in-the

Summer

   

“In which cities does it rain a lot during Summer?”

  

Information focus

  

B: –

Rio e São Paulo

chovem

muito no

verão.

   

Rio and São Paulo

rain-3pl

much in-the

Summer

   

“It rains a lot in Rio and São Paulo during Summer.”

(3)

A:

– O que aconteceu?

  

what happened

  

“What happened?”

 

B:

– O

celular

acabou

a

bateria.

  

the

cell.phone

finished

the

battery

  

“The battery of the cell phone is dead.”

These observations are relevant for two reasons. First, they attenuate the putative connection between constructions such as the ones in (1) and topics. To put it in different words, (2) and (3) show that whatever makes constructions like (1) grammatical in BP, it should be structural in nature, rather than informational (see Lacerda (2020) for relevant discussion). In what follows, we will keep using the term topic-subject only for presentation purposes.

The second reason why the grammaticality of the sentences in (2) is relevant is that the types of focalization seen in (2a–d) cannot be base-generated (see, e.g., Cinque (1990)), as independently shown by the fact that the focalized constituent is sensitive to islands and is incompatible with a resumptive pronoun:Footnote 5

(i)

Que livro

todo

mundo

que

leu

(ele)

resolveu

mudar

de

vida?

 

which book

every

world

that

read

it

resolved

change

of

life

 

“Which book is such that everyone who read it decided to change his life?”

(4)

a. *

Só esse relógio

a

Maria

conversou

com

o

cara

  

only this watch

the

Maria

talked

with

the

guy

  

que queria vender

(ele).

     
  

that wanted sell

it

     
  

“Maria talked with the guy that wanted to sell only this watch.”

 

b. *

Nem essa gaveta

o

João

ficou

contente

depois

de

  

nor this drawer

the

João

stayed

content

after

of

  

consertar (ela).

      
  

fix it

      
  

“John did not get happy even after fixing this drawer.”

 

c. *

O RELÓGIO

a

polícia

prendeu

o

ladrão

 
  

the watch

the

police

arrested

the

thieve

 
  

que

roubou

(ele)

não

o

celular.

 
  

that

stole

it

not

the

cell.phone

 
  

“The police arrested the thieve that stole THE WATCH, not the cell phone.”

 

d. *

Que diabo de carro

o

João

se

arrependeu

depois de

  

what devil of car

the

João

himself

repented

after of

  

comprar (ele)?

     
  

buy-subj

it

    
  

“What kind of car did John regret having bought?”

This indicates that the focalized constituents in (2a–d) are not base-generated and have reached their surface position via movement. All things being equal, the same conclusion should also apply to the classic “topic-subject” constructions in (1). In other words, the grammaticality of constructions such as (2a–d) refutes the recurring idea that the subject of “topic-subject” constructions is directly merged in [Spec,TP]. Thus, the null hypothesis is that the DPs that trigger verbal agreement in constructions such as the ones in (1) reach the standard subject position ([Spec,TP]) via movement, pretty much like the ones in (2a–d). From [Spec,TP], they may eventually move to higher A’-positions in the left periphery – an issue that is orthogonal to the derivation of “topic-subject” constructions itself.

Another fact that corroborates this conclusion was observed by Galves (1998), who noted that “topic-subject” constructions are incompatible with resumptive pronouns, thus contrasting with regular topic constructions, as illustrated in (5).

(5)

a.

[Os

relógios],

quebrou

o

ponteiro

deles.

  

the

watches

broke-3sg

the

arm

of-them

 

a’. *

[Os

relógios]

quebraram

o

ponteiro

deles.

  

the

watches

broke-3pl

the

arm

of-them

  

“The arms of the watches broke.”

 

b.

[Essas

gavetas],

cabe

muita

coisa

nelas.

  

these

drawers

fit-3sg

many

thing

in-them

 

b’. *

[Essas

gavetas]

cabem

muita

coisa

nelas.

  

these

drawers

fit-3pl

many

thing

in-them

  

“Many things can fit in these drawers.”

The lack of verbal agreement with the topic in (5a) and (5b) signals that they instantiate a standard topic construction and the compatibility with the associated pronoun shows that the topic can be base-generated. In turn, the ungrammaticality of the “topic-subject” constructions in (5a’) and (5b’), which display verbal agreement with the alleged topic, is to be expected if the subject cannot be generated where it surfaces and must therefore compete with the associated pronoun for the relevant position within vP.

So, the interim conclusion we reach is that, unless proven otherwise, the DP that triggers verbal agreement in a “topic-subject” construction in BP is not necessarily a topic (although it may later move to a topic position) and comes to occupy the subject position by moving from a vP-internal position. From the perspective we are exploring here, this indicates that the availability of “topic-subject” constructions is not related to some special property to be ascribed to the upper part of the clausal domain.

3 Changes Within vP and DP

Our starting point is Kato and Ordóñez’s (2019) proposal regarding the diachronic source for the emergence of “topic-subject” constructions in BP, based on a comparison between BP and Dominican Spanish, a language that has also been analyzed as becoming non-pro-drop (see, e.g., Toribio (1996), Ordóñez and Olarrea (2008), and Kato (2012a)). The authors show that although the two languages exhibit common properties with respect to phenomena related to the loss of null subjects, Dominican Spanish does not allow “topic-subject” constructions, as illustrated in (6) below. Kato and Ordóñez also show that the grammatical versions of (6) in Dominican Spanish involve CLLD with a third-person dative clitic, instead of “topic-subjects,” as can be seen in (7).

(6)

Dominican Spanish (Kato and Ordóñez 2019: (27) and (25))

 

a. *

El reloj

rompió

las

agujas.

  

the clock

broke.3sg

the

needles

  

Intended: “The hands of the clock broke.”

 

b. *

Estos

bosques

llueven

mucho.

  

these

forests

rain.3pl

a.lot

  

Intended: “It rains a lot in the forests.”

(7)

Dominican Spanish (Kato and Ordóñez 2019: (30) and (28))

 

a.

A este reloj

se

le

rompió

 

la

aguja.

  

to this clock

refl

dat.3sg

broke.3sg

 

the

needle

  

“The clock’s hand broke.”

 

b.

A

estos

bosques

les

llueve

mucho.

  

to

these

forests

dat.pl

rain.3sg

a.lot

  

“In these forests, it rains a lot.”

Interestingly, as Kato and Ordóñez observe, BP has lost its series of third-person pronominal clitics, as well as its third-person possessive pronouns, as illustrated in the chart above, adapted from Kato (1993b) (Table 1).Footnote 6

Table 1 Third-person accusative and dative clitics and possessive pronouns in BP in the eighteenth and twentieth centuries
Full size table

This simplification in the pronominal paradigm of BP led the authors to conclude that the real trigger for the emergence of “topic-subject” constructions in BP was the impoverishment in its clitic and possessive systems, and not in its verbal agreement paradigm. Based on the work by Barros (2006) and Torres Morais (2016), Kato and Ordóñez propose that until the nineteenth century, the functional skeleton of the extended projection of vP in BP included a dative phrase (in the sense of Landau (1999)), whose specifier was licensed with dative Case and its head was realized as a dative clitic, yielding CLLD constructions analogous to (7). Once third-person dative clitics became null, their associates could no longer be licensed with dative Case and had to move to [Spec,TP] to have their Case licensed, yielding “topic-subject” constructions.

In this chapter we assume the gist of Kato and Ordóñez’s analysis, modifying some details of its technical implementation. In particular, we show below that the standard assumption that dative clitics are θ-marked in their argument position before undergoing cliticization suffices to account for the facts under discussion and we provide a novel analysis of “topic-subject” constructions with locatives, as they do not seem to have a clitic structure as their diachronic source.Footnote 7 We follow Barros (2006), Torres Morais and Salles (2016), Gonçalves and Miguel (2019), and Kato and Ordóñez (2019) in assuming that until the nineteenth century, the extended projection of vP in BP included a projection that was able to assign dative Case, yielding external possessor constructions in the sense of Vergnaud and Zubizarreta (1992), as illustrated by the sentences in (8) below, by the nineteenth-century author Machado de Assis. In (8a), the dative element in bold is an R-expression and in (8b), a clitic pronoun.

(8)

 

(Nineteenth-century BP; Torres Morais and Salles 2016: (33b–c))

 

a.

Clara

não

tinha

sequer

tempo

de

remendar

a

roupa

  

Clara

not

had

hardly

time

to

mend

the

clothes

  

ao marido.

  

the husband.dat

  

“Clara hardly had time to mend her husband’s clothes.”

 

b.

Clara

estendeu

a

mão

ao

marido

como

a

amparar-lhe

  

Clara

extended

her

hand

the

husband.dat

as

to

give-3sg.dat

  

o

ânimo.

  

the

support

  

“Clara extended her hand to her husband, as a way to give him support.”

Once vP became unable to assign dative Case in BP, external possessor constructions with an R-expression such as (8a) simply died out in the grammar. Let us then consider how DP1 in the abstract transitive vP structure depicted in (9) below can be licensed, after BP lost the dative licensing projection at the vP level, as well as its third-person dative clitics and third-person possessive pronouns.

(9)

[TP T [vP DP3 [v’ v [VP V [DP2 D … [NP N DP1]]]]]]

In (9), DP2 is arguably Case-licensed by v, and DP3 by T. Given that the extended projection of v is no longer able to license dative Case in BP, the only possibility for DP1 to have its Case licensed is to receive inherent Case by the noun that θ-marks it (see Chomsky 1986). If the inherently Case-marked DP1 in (9) is a first- or second-person pronoun, it may be realized as a possessive pronoun, as illustrated in (10a) below, or as a dative clitic, as illustrated in (10b).

(10)

a.

A Maria segurou

[a [{minha/sua}i

[mão ti]]]

  

the Maria held

the my/your

hand

 

b.

A Maria {me/te}i

 

segurou [a [mão ti]]]

  

the Maria me.dat/you.dat

 

held

the hand

  

“Maria held my hand.”

On the other hand, if the inherently Case-marked DP1 in (9) is a third-person pronoun, neither possibility is available, as shown in (11a) and (11b) below, for BP has lost the genitive and dative forms for third-person pronouns. A third-person pronoun occupying the position of DP1 in (9) may, however, be realized preceded by the dummy preposition de, as shown in (11c), a possibility that is also available for R-expressions, as shown in (11d), but not for first- and second-person pronouns, as shown in (11e).

(11)

a. *

[A

Maria]i

segurou

a

sua k

mão.

  
  

the

Maria

held

the

his

hand

  
 

b. *

A

Maria

lhe

segurou

a

mão.

  
  

the

Maria

him.dat

held

the

hand

  
 

c.

A

Maria

segurou

a

mão

dele.

  
  

the

Maria

held

the

hand

of-him

  
  

“Maria held his hand.”

 

d.

A

Maria

segurou

a

mão

do

Pedro.

 
  

the

Maria

held

the

hand

of-the

Pedro

 
  

“Maria held Pedro’s hand.”

 

e. *

A

Maria

segurou

a

mão

{de

mim/de

você}.

  

the

Maria

held

the

hand

of

me/of

you

  

“Maria held {my/your} hand.”

The contrast between (11c,d) and (11e) may be described as showing that de is only allowed if the expression it licenses does not have an independent form for the realization of inherent Case (genitive or dative, in this particular scenario). In other words, de-insertion is a last resort strategy for the realization of the inherent Case assigned by the noun in (9).

Notice that the asymmetries seen in (10) and (11) between first and second persons, on the one hand, and third person, on the other, were handled based solely on the Case properties and Case realizations within DP2 in (9), in a way quite independent from the properties of v. This leads us to expect that these asymmetries should not be restricted to transitive verbs, as in (9), but could also be found with unaccusative verbs. Kato and Ordóñez (2019) show that this is indeed the case. Take the monoargumental unaccusative structure in (12) below, for instance. Like what we saw above, DP1 in (12) is assigned inherent Case and is realized as a possessive pronoun or a dative clitic if it is a first- or second-person pronoun (see (13a–b)) or by an oblique form preceded by the preposition de if it is a third-person pronoun (see (13c)).

(12)

[TP T [vP v [VP V [DP2 D ... [NP N DP1]]]]]

(13)

a.

Ainda

não

nasceu

a

{minha/sua}

barba.

  

still

not

was.born

the

my/your

beard

 

b.

Ainda

não

{me/te}

nasceu

a

barba.

  

still

not

me.dat/you.dat

was.born

the

beard

  

“I/you have not grown a beard yet.”

 

c.

Ainda

não

nasceu

a

barba

dele.

  

still

not

was.born

the

beard

of-him

  

“He has not grown a beard yet.”

Given that unaccusative structures do not involve an external argument, DP1 in (12) may also be Case-licensed in BP by moving to [Spec,TP], yielding “topic-subject” constructions. Interestingly, “topic-subject” constructions also display sensitivity with respect to person, with third-person pronouns (and R-expressions) yielding the best outputs, as shown in (14).Footnote 8

(14)

a. %*

Eu

ainda

não

nasci

a

barba.

 
  

I

still

not

was.born.1sg

the

beard

 
  

“I have not grown a beard yet.”

 

b. %??

Você

ainda

não

nasceu

a

barba.

 
  

you

still

not

was.born

the

beard

 
  

“You have not grown a beard yet.”

 

c.

{Ele/o

João}

ainda

não

nasceu

a

barba.

  

he/the

João

still

not

was.born

the

beard

  

“{He/João} has not grown a beard yet.”

A similar state of affairs is found with unaccusative structures with two internal arguments, as sketched in (15) below. Given that unaccusative verbs are not associated with structural Case (Burzio’s Generalization), at least one of the arguments of V in (15) must be assigned inherent Case. Suppose that V assigns inherent dative Case to DP1. DP1 then surfaces as a dative clitic if it is a first or second person, but as an oblique form preceded by the preposition a if it is a third-person pronoun, as shown in (16). (15) may also give rise to a “topic-subject” construction, again displaying person sensitivity, with third-person pronouns and R-expressions being the best results, as shown in (17).

(15)

[TP T [vP v [VP DP2 V DP1]]]

(16)

a.

Me/te

faltou

sorte.

  

me.dat/you.dat

lacked

luck

  

“I was/you were unlucky.”

 

b.

Faltou sorte

a

ele.

  

lacked luck

to

he

  

“He was unlucky.”

(17)

a. *

Eu

faltei

sorte.

 
  

I

lacked.1sg

luck

 
  

“I was unlucky.”

 

b. %??

Você

faltou

sorte.

 
  

you

lacked

luck

 
  

“You were unlucky.”

 

c.

Ele

faltou

sorte.

 
  

he

lacked

luck

 
  

“He was unlucky.”

 

d.

Meus

times

faltaram

sorte.

  

my

teams

lacked.3pl

luck

  

“My teams were unlucky.”

The data in (14) and (17) seem to show that inherent Case assignment by N and V is obligatory when the relevant DP is a first- or second-person pronoun, thus blocking its movement to [Spec,TP] (see (14a,b)/(17a,b)), but optional when the relevant DP is a third-person expression, optionally allowing it to move to [Spec,TP], yielding a “topic-subject” construction (see (14c)/(17c,d)). This conceptually odd result may however be disentangled if we observe that the relevant difference is likely to be related to the loss of third-person possessive pronouns and dative clitics in BP. In other words, inherent Case assignment may be taken to apply obligatorily in the cases discussed above and the additional possibility available to third-person expressions may have to do with the realization of inherent Case. Under the standard assumption that an inherent Case is linked to a θ-role, we tentatively propose that if the target of inherent Case assignment cannot morphologically realize the Case it has received, two different repair strategies can be employed (with equal derivational cost): (i) the expression is realized with default Case morphology and a linker is added to encode the dependency relation with respect to the θ-marking head, or (ii) the expression “repels” the Case associated with the θ-role in the sense that it does not incorporate the Case morphology/specification associated with the θ-role, thus remaining active for the purposes of Case and agreement.

The possibility in (i) is illustrated by sentences such as (11c), (13c), and (16b). Crucially, with the loss of third-person dative and accusative clitics and third-person possessive pronouns in BP, the nominative form came to be used and licensed in all syntactic positions, for nominative is the default Case in BP. In other words, what we described in (11c), (13c), and (16b) as a third-person pronoun in its oblique form seems to be more adequately described as the default form preceded by a linker (the same applies to the R-expression in (11d), for instance). As for the possibility (ii), it is exemplified by “topic-subject” constructions such as (14c) and (17c,d), where the argument of N in (14c) and V in (17c,d) repels the “unrealizable” Case assigned and moves to [Spec,TP], where it gets licensed with nominative Case in the standard way. Importantly, a similar derivation is not available to first- and second-person pronouns (see (14a,b) and (17a,b)) because the last resort saving strategy is not applicable, for the inherent Case assigned can be morphologically realized.

Assuming that something along these lines may be on the right track, let us now examine some details of the derivation of “topic-subject” constructions.

4 The Role of Inherent Case in “Topic-Subject” Constructions

We saw in Sect. 3 that the loss of the projection licensing dative Case at the vP level, coupled with the loss of third-person dative clitics and third-person possessive pronouns, considerably reduced the possibilities for Case licensing within vP in BP. However, given that we are talking about the vP level, structural Case is not the only possibility for Case licensing. Crucially, vP is also a thematic domain and, therefore, a domain where inherent Case may be available. Our proposal is that with the reduction of structural Case assignment possibilities at the vP level, BP came to make pervasive use of inherent Case within the verbal domain, extending it to the adjectival and nominal domains, as well.

This innovation developed is interesting in that it provides a single solution for the two general puzzles posed by “topic-subject” constructions in BP: (i) how the relevant DPs are Case-licensed and (ii) why the relevant A-movement involved in these constructions does not violate minimality (see Nunes (2017)). Let us consider each of them in turn.

Take the alternation in (18) and (19), for example.

(18)

a.

Quebrou

o

ponteiro

dos

relógios.

  

broke-3sg

the

arm

of-the

watches

 

a’.

[TP proexpl [vP v [VP quebrar [DP o [NP ponteiro [DP os relógios]]]]]]

   

   |____↑inherent Case |____↑inherent Case

 

b.

[os

relógios]

quebraram

o

ponteiro.

  

the

watches

broke-3pl

the

arm

  

“The arms of the watches broke.”

 

b’.

[TP [os relógios]i T [vP v [VP quebrar [DP o [NP ponteiro ti]]]]]

    

  |____↑inherent Case

(19)

a.

Cabe

muita

coisa

nessas

gavetas.

  

fit-3sg

many

thing

in-these

drawers

 

a’.

[TP proexpl T [vP v [VP [muita coisa] [cabe [DP essas gavetas]]]]]

   

    inherent Case↑___||____↑inherent Case

 

b.

[essas

gavetas]

cabem

muita

coisa.

  

these

drawers

fit-3pl

many

thing

  

“Many things can fit in these drawers.”

 

b’.

[TP [DP essas gavetas] T [vP v [VP [muita coisa] cabem ti]]]]

    

    inherent Case↑____|

In (18a’), the verb quebrar and the noun ponteiro both assign inherent Case to their arguments, which are then licensed in situ. The inherent Case assigned by the noun is morphologically realized with the help of the preposition de (see Sect. 3). Recall that a given expression may also have the option of repelling a given inherent Case if it cannot morphologically realize it (see Sect. 3). This option is exercised in (18b’), where the DP os relógios “the watches” repels the inherent Case assigned by ponteiro “arm” and then moves to [Spec,TP], where it triggers verbal agreement and is assigned nominative Case. As for (19a), the verb caber “fit” assigns inherent Case to both of its arguments, which surface in situ, and the inherent Case assigned to the locative is realized with the help of the preposition em “in.” If the complement exercises the option of repelling the inherent Case assigned by caber, as sketched in (19b’), it may get licensed by moving to [Spec,TP], triggering verbal agreement and receiving nominative Case in the standard way (cf. (19b)).

One could say that the absence of em in (19b) is not a strong argument for taking em in (19a) as a marker of inherent Case, for the verb caber “fit” selects a location for its complement and em can be used as a true preposition with a locative meaning. There is however independent evidence that the preposition em in BP may be used as a realization of inherent Case in contexts unrelated to a locative meaning. Discussing the role of em in marking the complement of verbal nominalizations with the suffix -ada in BP, as shown in (20) below, Scher (2004:198) points out that em cannot be inserted to Case-mark the subject of a small clause, as illustrated by the contrast in (21). Given that os papeis “the papers” is an argument of the verb in (21a) but not in (21b), Scher interprets the contrast in (21), showing that em in these contexts is a realization of inherent Case.Footnote 9

(i)

Andrade and Galves (2014: 118, 137):

 

(ii)

Andrade and Galves (2014: 120, 138):

a.

A

mesa

quebrou

o

pé.

 

a.

Esse

carro

cabe

muita

gente.

 

the

table

break-past-3sg

the

foot

  

this

car

fit-3sg

many

people

 

“The table leg broke.”

  

“Many people fit in this car.”

b.

      

b.

     
 

  

(20)

A

Maria

deu

uma

organizada

nos

dados.

 

the

Maria

gave

a

organize-nomz

in-the

data

 

“Maria has organized the data.”

(21)

a.

O

João

deu

uma

classificada

nos

papeis.

 
  

the

João

gave

a

classify-nomz

in-the

papers

 
  

“João has classified the papers.”

 

b. *

O

João

deu

uma

classificada

nos

papeis

como

  

the

João

gave

a

classify-nomz

in-the

papers

as

  

interessantes.

  

interesting

  

“João has classified the papers as interesting.”

Also relevant to the present discussion is the fact that directional verbs in BP underwent a diachronic change replacing the preposition a “to” by the preposition em “in” as the head of their complements (see, e.g., Wiedemer 2013), as illustrated in (22) below. This change seems to have connected the double role played by em as an independent true preposition with locative meaning and as a marker of inherent Case assignment (see (21)), according well with our proposal that BP underwent a diachronic change that greatly expanded the use of inherent Case in its grammar.

(22)

a.

O

João

foi

no

mercado.

   
  

the

João

went

in-the

market

   
  

“João went to the market.”

 

b.

A

Maria

já

chegou

em

casa.

  
  

the

Maria

already

arrived

in

house

  
  

“Maria has already arrived home.”

 

c.

O

Pedro

veio

na

festa.

   
  

the

Pedro

came

in-the

party

   
  

“Pedro came to the party.”

 

d.

A

Maria

levou

o

filho

no

cinema

hoje.

  

the

Maria

took

the

son

in-the

movies

today

  

“Maria took her son to the movies today.”

Let us now consider the connection between inherent Case and minimality, by examining the English data in (23).

(23)

a.*

[Mary seems to himk [t to like Johnk]]

 

b.

[Maryi seems to him [ti to be nice]]

 

c.*

[To him] seems t [Mary to be nice]

The sentence in (23a) displays a Principle C effect, suggesting that the preposition does not prevent the pronoun from c-commanding into the embedded clause. That being so, the fact that it does not block movement of the embedded subject on (23b) becomes rather puzzling. Chomsky (1995:306) observes that the experiencer should get inherent Case within the lexical VP shell and suggests (fn. 77) that the preposition to is adjoined to the experiencer DP as a kind of Case-marker, explaining why it does not affect c-command. Building on these premises, Nunes (2008a) proposes that inherent Case renders a given element inert for the purposes of A-movement, as shown in (23c), removing it from computations of Relativized Minimality.Footnote 10 To put it in different terms, inherent Case makes an element transparent for A-movement across it. This is exactly what happens in (18b’) and (19b’). The DPs o ponteiro “the arm” in (18b’) and muita coisa “many things” in (19b’) do not block A-movement of the “topic-subject,” for they have become transparent after receiving inherent Case.Footnote 11

(i)

a.

Ela

beijou

o

João

no

rosto.

  

she

kissed

the

João

on-the

face

  

“She kissed João on the cheek.”

   

__

   

| ↓inherent Case

 

b.

[vP v [VP beijou [DP1 [DP2 o João]i [D’ o rosto ti]]]]

  

 |_________________↑structural Case

As shown in Nunes (2017), this proposal also provides a straightforward account of “mixed” and “extralong” “topic-subject” constructions such as (24) and (25).

(24)

a.

[Esses

porta-malas]

cabem

muita

coisa

na

lateral.

 
  

these

car-trunks

fit-3pl

many

thing

in-the

lateral

 
  

“Many things can fit on the side of the trunk of these cars.”

 

b.

[TP [DP esses porta-malas]i T [vP v [VP [DP muita coisa] [cabem [DP a lateral ti]]]]]

     

inherent Case↑______||______↑inherent Case

(25)

a.

[Esses

barcos]

diminuíram

o

tamanho

da

hélice

do

  

these

boats

diminished-3pl

the

size

of-the

fan

of-the

  

motor.

  

engine

  

“These boats had the size of the fans of their engine reduced.”

 

b.

[TP [DP esses barcos]i T [vP v [VP diminuíram [DP o tamanho [DP a hélice [DP o motor ti]]]]]]

    

|__↑inherent Case  |__↑inherent Case |__↑inherent Case

From the perspective taken here, apparently complex structures like the ones in (24) and (25) receive the same analysis as (18b) and (19b): given that the potential interveners receive inherent Case, they become transparent for A-movement across them, as sketched in (24b) and (25b), in the same way the pronoun in (23b) does not block A-movement of the embedded subject. In (24b), the verb caber assigns inherent Case to both the theme in its specifier and the locative in its complement (which is realized as em in the morphological component) and they become inert for the purposes of intervention. The argument of lateral, which has exercised the option of repelling inherent Case, can then move to [Spec,TP] to get Case-licensed without incurring in a minimality violation. As for (25b), the verb diminuir “diminish” and the nouns tamanho “size” and hélice “fan” assign inherent Case to their arguments and the DP esses barcos “these boats” repels the inherent Case assigned by the noun motor “engine.” Esses barcos then receives nominative Case after moving to [Spec,TP] and the inherent Case assigned by the nouns is realized as the preposition de, yielding the “topic-subject” construction in (25a), which involves an “extralong” instance of A-movement, but all the potential interveners have become inert after receiving inherent Case.

The amplification of the use of inherent Case in BP had widespread consequences in the grammar. Take the contrast in (26) below, for example.

(26)

a.

[Esse

livro]i,

todo

mundo

[que

gosta Øi]

vira

pacifista.

  

this

book

every

world

that

likes

becomes

pacifist

  

“Everybody who read this book becomes a pacifist.”

 

b.

Todo

mundo

gosta

*(de)

filme

de

detetive.

  

every

world

likes

of

movie

of

detective

  

“Everyone likes detective movies.”

The sentence in (26a) has a null object within a relative clause island, suggesting that it is a pro linked to the base-generated topic, rather than a trace. In turn, (26b) shows that the verb gostar “like” is not a Case assigner, requiring the preposition de in order to license its complement. That being so, one wonders how pro in (26a) is Case-licensed. To account for this sort of problem, Ferreira (2000) has proposed that pro in BP may be defective in not having a Case feature. However, Kato (2010, 2012b) shows that this cannot be what is behind the grammaticality of sentences such as (26a), for the apparently exceptional licensing of pro is lexically conditioned. The complement of the verb rir “laugh,” for instance, must also be preceded by the preposition de, as shown in (27). However, whereas gostar licenses a null object linked to a topic, rir doesn’t, as shown in (28).

(27)

a.

Ela

não

gostou

*(d)o

palhaço.

  

she

not

liked

of-the

clown

  

“She didn’t like the clown.”

 

b.

Ela

não

riu

*(d)o

palhaço.

  

she

not

laughed

of-the

clown

  

“She didn’t laugh at the clown.”

(28)

a.

[Aquele

palhaço]i,

ela não

gostou

pro i

  

that

clown

she

not

liked

  

“That clown, she didn’t like him.”

 

b. *

[Aquele

palhaço]i,

ela

não riu

pro i

  

that

clown

she

not laughed

 
  

“That clown, she didn’t laugh at him.”

Kato and Nunes (2009) argue that the difference between these two verbs is that rir selects for a PP headed by de, whereas gostar assigns inherent Case to its complement, which is realized as de if the complement is phonetically realized. Interesting empirical evidence for this proposal is the fact that gostar licenses an inherently Case-marked reflexive/reciprocal clitic, but rir doesn’t, as shown in (29) (see Nunes (2008b)).

(29)

a.

Eles

se

gostam

muito.

  

they

refl.cl

like

much

  

“They like each other a lot.”

 

b. *

Eles

se

riram

bastante.

  

they

refl.cl

laughed

much

  

“They laughed a lot at each other.”

Kato and Nunes (2009) also show that the salient resort to inherent Case in the grammar of BP also accounts for the existence of what Tarallo (1983) called PP-chopping relatives in the language. In his seminal work on relative clauses in BP, Tarallo postulated three types of relativization strategies, each of which with a different sociolinguistic status: the standard strategy with movement of a PP, as illustrated in (30a) below; the resumptive strategy with an overt resumptive pronoun, as illustrated in (30b); and the chopping strategy, with no visible presence of the relevant PP, as illustrated in (30c).

(30)

a.

a

pessoa

com

quem

eu

conversei

 
  

the

person

with

who

I

talked

 
 

b.

a

pessoa

que

eu

conversei

com

ela

  

the

person

that

I

talked

with

her

 

c.

a

pessoa

que

eu

conversei

  
  

the

person

that

I

talked

  
  

“the person I talked to”

For Tarallo, the chopping version in (30c) involves a null resumptive pronoun and the preposition is deleted in the phonological component, as BP does not allow preposition stranding. However, Kato (1993a) observes that if the chopping strategy involved preposition deletion because prepositions cannot be left stranded in BP, the counterpart of (30a) given in (31a) below should be grammatical under the derivation sketched in (31b), where the relative pronoun quem undergoes A’-movement and the stranded preposition is deleted in the phonological component.

(31)

a. *

a

pessoa

quem

eu

conversei

  

the

person

who

I

talked

  

“the person I talked to”

 

b.

a pessoa [quemi eu conversei com ti]

Kato and Nunes (2009) reanalyze Tarallo’s proposal in terms of phonetic realization. According to them, the verb conversar “talk” assigns inherent Case to its complement, which is realized as the preposition com “with” if the argument is phonetically realized. This is transparent in the case of (30b), as the overt resumptive pronoun is realized in its thematic position. In (30a), the verb assigns inherent Case to the relative pronoun before it moves to [Spec,CP] and it then surfaces accompanied by the preposition. In (30c), the preposition is not phonetically realized because the complement of the verb (pro) has no phonetic realization. Finally, under the assumption that the relative pronoun quem cannot be base-generated in its surface position (see Kato and Nunes (2009) for arguments and relevant discussion), it must have merged with the verb before moving to the left periphery, and once it receives inherent Case from conversar, it must surface with the preposition, explaining why (31a) is not acceptable.

It is thus not surprising that verbs of movement that came to take the preposition em preceding their locative complement (see (22)) freely allow chopping relatives, as illustrated in (32) below. This is exactly what we should expect if em may be a realization of inherent Case, as proposed above.

(32)

a.

o

mercado

que

o

João

foi

  
  

the

market

that

the

João

went

  
  

“the market João went to”

 

b.

o

lugar

que

a

Maria

chegou

  
  

the

place

that

the

Maria

arrived

  
  

“the place Maria arrived at”

 

c.

a

festa

que

o

Pedro

veio

  
  

the

party

that

the

Pedro

came

  
  

“the party Pedro came to”

  
 

d.

o

cinema

que

a

Maria

levou

o

filho

  

the

movies

that

the

Maria

took

the

son

  

“the movie theater Maria took her son to”

Kato and Nunes (2009) also observe that their reanalysis of Tarallo’s (1983) account of chopping relatives also extends to free relatives in BP. Lessa de Oliveira (2008) notes that free relatives in BP can be of the “chopping” variety, as illustrated in (33) below, whose embedded verbs require a PP complement, as shown in (34).

(33)

a.

Eu

vou

visitar

quemi

você

simpatiza

muito.

  

I

go

visit

who

you

sympathize

much

  

“I’m going to visit who you like a lot.”

 

b.

Eu

encontrei

o que

você

gosta.

  
  

I

found

what

you

like

  
  

“I found what you like.”

(34)

a.

O

João

simpatiza

*(com)

a

Maria.

  

the

João

sympathizes

with

the

Maria

  

“João likes Maria.”

 

b.

O

João

gosta

*(de)

romances.

 
  

the

João

likes

of

novels

 
  

“João likes novels.”

If “chopping” free relatives like (33) also involve inherent Case assignment to an object pro, one would expect contrasts such as the one in (28) to be replicated in free relative clauses. Kato and Nunes (2009) show that this prediction is borne out, as shown in (35).

(35)

a. *

O

João

sempre

critica

quem

ele

ri.

  

the

João

always

criticizes

who

he

laughs

  

“João always criticizes whoever he laughs at.”

 

b.

O

João

sempre

critica

quem

ele

gosta.

  

the

João

always

criticizes

who

he

likes

  

“João always criticizes whoever he likes.”

As seen above, gostar assigns inherent Case, but rir doesn’t. Hence, the embedded object position in (35b) can be licensed (if it is pro), but not the embedded object position in (35a) (regardless of whether it is pro or a trace).

Let us finally examine another domain where we can directly see the connection between inherent Case and A-minimality. Galves (1987) has observed that tough-predicates in BP display a very distinctive behavior. In addition to the standard interpretation as the object of the embedded predicate, the subject of a tough-construction such as (36), for instance, may also be interpreted as the embedded subject.

(36)

O

João

é

difícil

de

elogiar.

 

the

João

is

difficult

of

praise-inf

 

Tough-interpretation: “It is hard to praise João.”

 

Raising interpretation: “João rarely praises someone.”

Nunes (2008a) notes that the adjectival predicates that allow the crosslinguistic uncommon subject reading optionally take a preposition before its infinitival complement, as illustrated in (37a) and (38a) below. Interestingly, the optionality does not hold constant across different constructions. If the preposition is present, the infinitival cannot move to the matrix subject position, as shown in (37b) and (38b). Conversely, the embedded subject can only move to the matrix subject position if the preposition is present, as illustrated in (37c) and (38c).

(37)

a.

É

difícil

[(d)esses

jornalistas

elogiarem

alguém].

 
  

is

difficult

of-these

journalists

praise-inf-3pl

somebody

 
  

“It is rare for these journalists to praise someone.”

 

b.

(*D)esses

jornalistas

elogiarem

alguém

é

difícil.

 
  

of-these

journalists

praise-inf-3pl

somebody

is

difficult

 
  

“For these journalists to praise someone is very rare.”

 

c.

[esses

jornalistas]i

são

difíceis

*(de) t i

elogiarem

alguém.

  

these

journalists

are

difficult

of

praise-inf-3pl

somebody

  

“These journalists rarely praise someone.”

(38)

a.

Não

estava

previsto

(para)

as

aulas

começarem

amanhã.

  

not

was

predicted

for

the

classes

start-inf-3pl

tomorrow

 

b.

(*Para)

as

aulas

começarem

amanhã

não

estava

previsto.

  

for

the

classes

start-inf-3pl

tomorrow

not

was

predicted

  

“It was not expected that the classes should start tomorrow.”

 

c.

As

aulas

estavam

previstas

*(para)

começarem

amanhã.

 
  

the

classes

were

predicted

for

start-inf-3pl

tomorrow

 
  

“The classes were not expected to start tomorrow.”

Nunes (2008a) argues that these prepositions are actually realization of an inherent Case optionally assigned by the impersonal predicates to their infinitival complement, much like what Chomsky (1995) has proposed for the preposition to preceding the experiencer of raising constructions like (23). Like what we saw with the prepositioned experiencer in (23b), the prepositioned infinitivals in (37b) and (38b) cannot undergo A-movement because they have already been Case-marked. The inherent Case also renders the infinitival transparent for A-movement from within it, in the same way we saw in “topic-subject” constructions like (25); hence, A-movement from within the infinitival (37c) and (38c) can only take place if the preposition is present.

To summarize, with the weakening of structural Case licensing at the vP level in BP seen in Sect. 3, the grammar came to explore to its limits the other possibility for Case licensing, namely, inherent Case assignment. This expansion of the use of inherent Case reverberated across different domains in the grammar, yielding as byproducts apparent Caseless null objects, chopping relative clauses, hyper-raising constructions out of infinitivals, and topic-subject constructions.

5 Further Issues on Person Asymmetries and “Resumption” in “Topic-Subject” Constructions

As we saw in Sect. 3, Kato and Ordóñez (2019) observe that “topic-subject” constructions display a sensitivity to the value of the feature [person]. Whereas third-person pronouns (and R-expressions) function as good candidates as “topic-subjects,” first- and second-person pronouns in general do not yield well-formed results or exhibit considerable variation across speakers (see fn. 9), as seen in (14), repeated here in (39).

(39)

a. %*

Eu

ainda

não

nasci

a

barba.

 
  

I

still

not

was.born.1sg

the

beard

 
  

“I have not grown a beard yet.”

 

b. %??

Você

ainda

não

nasceu

a

barba.

 
  

you

still

not

was.born

the

beard

 
  

“You have not grown a beard yet.”

 

c.

{Ele/o

João}

ainda

não

nasceu

a

barba.

  

he/the

João

still

not

was.born

the

beard

  

“{He/João} has not grown a beard yet.”

This asymmetry seems to be lexically conditioned, though. In (40) below, for instance, there is no difference of acceptability among the different persons in the “topic-subject” versions.

(40)

a.

Inchou

o

meu

pé.

  

swelled

the

my

foot

 

a’.

Eu

inchei

o

pé.

  

I

swelled.1sg

the

foot

  

“My foot got swollen.”

 

b.

Inchou

o

seu

pé.

  

swelled

the

your

foot

 

b’.

Você

inchou

o

pé.

  

you

swelled

the

foot

  

“Your foot got swollen.”

 

c.

Inchou

o

pé

dele.

  

swelled

the

foot

of-he

 

c’.

Ele

inchou

o

pé.

  

he

swelled

the

foot

  

“His foot got swollen.”

Rodrigues (2020) in fact argues that possessor raising constructions in BP have a subclass in which the moving element moves directly to [Spec,TP] and another one in which the moving element lands in an intermediate position, where it receives the θ-role of affected entity. According to her, the contrast between the two sentences in (41) below can be accounted for if esturricar “burn” is a member of the second subclass, and accordingly, it requires that the moving element be affected, hence the pragmatic oddity of (41b), for the pig is dead and cannot be affected in the relevant sense. In turn, cair “fall” belongs to the first class and is not subject to this pragmatic restriction, for the possessor moves directly to [Spec,TP]; hence, (42) is acceptable even if the falling event occurs after Lincoln’s death.

(41)

(Rodrigues 2020, glosses and translation added)

 

a.

Eu

esturriquei

o

dedo

na

frigideira.

  

I

burnt

the

finger

in-the

frying.pan

  

“I burned my finger on the frying pan.”

 

b.#

O

porco

esturricou

a

costela

na

frigideira.

  

the

pork

burnt

the

ribs

in-the

frying.pan

  

“The pork’s ribs got burned.”

(42)

(Rodrigues 2020, glosses and translation added)

 

O

Lincoln

caiu

os

dentes

(depois

de

morto).

 

the

Lincoln

fell

the

teeth

after

of

dead

 

“Lincoln’s teeth fell off (after he was dead).”

What is relevant for our purposes is that the subclasses identified by Rodrigues seem to correlate with the person asymmetry noted by Kato and Ordóñez (2019). Specifically, verbs that are not associated with an affectedness θ-role in Rodrigues’s sense are the ones that display person sensitivity. The verb cair “fall,” for example, admits possessor raising with an R-expression, as seen in (42), or a third-person pronoun, as shown in (43a), but not with a first-person pronoun, as shown in (43b). By contrast, verbs that arguably involve the affectedness θ-role such as esturricar “burn” and inchar “swell” do not display person restrictions, as seen in (41a) and (40).

(43)

a.

Elas

caíram

o

cabelo.

  

they.f

fell-3pl

the

hair

  

“Their hair fell out.”

 

b.%*

Eu

caí

o

cabelo.

  

I

fell-1sg

the

hair

  

“My hair fell out.”

From an abstract point of view, the behavior of these two types of possessor raising in BP resembles the patterns of control and raising of DPs marked with quirky Case in Icelandic. The embedded main verbs of (44) below, for example, assign quirky dative to their complements. Dative morphology is preserved in standard raising constructions, as shown in (44a), but not in control constructions, as shown in (44b).

(44)

Icelandic:

 

a.

Mönnunum/*Mennirnir

virðist

báðum

hafa

verið

hjálpað.

  

men.the.dat/*nom

seems

both.dat

have

been

helped.dflt

  

“The men seem to have both been helped.”

  

(Sigurðsson 2008)

 

b.

Hann/*Honum

vonast

til

að

verða

bjargað

af

fjallinu.

  

he.nom/*dat

hopes

for

to

be

rescued.dflt

of

the.mountain

  

“He hopes to be rescued from the mountain.”

 
   

(Andrews 1990)

As is well known, quirky Case appears to involve a mixture of inherent and structural Case (see footnote 11). On the one hand, it behaves like inherent Case in establishing a connection between a specific θ-role and a specific piece of morphological information; on the other hand, it behaves like structural Case in its need to be licensed by a ϕ-complete probe. Assuming the Movement Theory of Control, Boeckx et al. (2010a, b) accounted for contrasts like the one in (44), by assuming that the additional θ-role assignment present in control breaks the connection between morphology and θ-role in inherent Case. The derivation of (44b), for example, proceeds along the lines sketched in (45) (with English words for convenience).

(45)

a.

Assignment of quirky Case:

  

[rescued he[θ1-DAT] from the mountain]

 

b.

Movement to [Spec,TP]:

  

[TP he[θ1-DAT] to be rescued t from the mountain]

 

c.

Applications of Merge:

  

[vP vθ2 [VP hopes [CP C [TP he[θ1-DAT] to be rescued t from the mountain]]]]

 

d.

Movement and θ-assignment:

  

[vP he[θ2+θ1] [v’ v [VP hopes [CP C [TP t to be rescued t from the mountain]]]]]

 

e.

Movement to [Spec,TP] and nominative Case assignment:

  

[TP he[θ2+θ1]-Case:NOM [T’ T [vP t [v’ [VP hopes [CP C [TP t to be rescued t from the mountain]]]]]]]

In (45a) the verb assigns quirky dative Case to the pronoun, which then moves to the embedded TP, yielding (45b). After the matrix v enters the derivation in (45c), the embedded subject moves to [Spec,vP] and receives an additional θ-role, as shown in (45d). Boeckx, Hornstein, and Nunes argue that the assignment of an additional θ-role in (45d) ends up breaking the previously established connection between θ1 and dative Case, for there are two θ-roles associated with the pronoun. Accordingly, the dative specification is deleted. The pronoun then undergoes standard movement to the specifier of a finite TP, triggering verbal agreement and receiving nominative Case (see (45e)).

A similar derivation appears to be found in the subclass of possessor raising constructions that Rodrigues (2020) argues involves an additional affectedness θ-role, with one proviso. Thus far, we have followed the standard wisdom according to which an element marked with inherent Case cannot undergo A-movement. This was exemplified above by (23c) in English and (37b) and (38b) in BP. It should be noted that the cases discussed both in English an BP involve A-movement for Case/agreement/EPP reasons, and if one assumes the Movement Theory of Control, A-movement also encompasses movement driven by θ-reasons, as seen in (45c–d). We would like to propose here that inherent Case actually renders a given element inert for the purposes of Case/agreement/EPP related movement, but not for θ-related movement. That being so, consider the abstract representation in (46) in BP, where N has assigned inherent Case to DP1 (see Sect. 3).

(46)

[vP vθ2 [VP V [DP2 D [NP N DP1-[θ1-inherent Case]]]]]

Suppose DP1 is a first- or second-person pronoun. All things being equal, DP1 should surface as a possessive pronoun (see (10a)) or a dative clitic (see (10b)). Things are not equal in (46), though, for there is still a θ-role to be assigned. If inherently Case-marked elements can move for θ-related reasons, as proposed above, DP1 can move to [Spec,vP] and receive a second θ-role, as illustrated in (47a) below. Like what we saw with quirky Case in control constructions (see (45d)), assignment of an additional θ-role to DP1 removes the previously established association between θ1 and inherent Case morphology. DP1 then moves to [Spec,TP] and receives nominative Case, yielding sentences such as (40a’), (40b’), or (41a). Nothing essentially changes if DP1 is a third-person pronoun (see (40c’)). To put it in general terms, the availability of an additional θ-role in (46) obliterates person distinctions as far as possessor raising is concerned.

(47)

a.

[vP DP1-[θ2+θ1] [vθ2 [VP V [DP2 D [NP N t]]]]

 

b.

[TP DP1-[θ2+θ1]-Case:NOM] [ T [vP t [vθ2 [VP V [DP2 D [NP N t] ] ] ] ] ] ]

Let us finally consider the data in (48) and (49).

(48)

a.

Eu

inchei

o

(meu)

pé.

  
  

I

swelled.1sg

the

my

foot

  
  

“My foot got swollen.”

 

b.

Eu

arranhei

o

(meu)

braço.

  
  

I

scratched.1sg

the

my

arm

  
  

“My arm got scratched.”

 

c.

Ele

quebrou

o

braço

(dele)

no

jogo.

  

he

broke.3sg

the

arm

of-him

in-the

game

  

“He broke his arms during the game.”

(49)

a.

Eles

ainda

não

nasceram

a

barba

(*deles).

  

they

still

not

were.born

the

beard

of-they

  

“They haven’t grown a beard yet.”

 

b.

Os

bebês

cresceram

o

cabelo

(*deles).

 
  

the

babies

grew.3pl

the

hair

of-they

 
  

“The babies’ hair has grown.”

The class of verbs that do not display person asymmetries optionally allow the realization of a possessor within the internal argument, as seen in (48), whereas the class of verbs that require that “topic-subjects” be third person do not, as seen in (49). Notice that the verbs in (48) belong to the subclass that has an additional affectedness θ-role, but not the ones in (49). This entails that if the possessor gets independently licensed within the internal argument in (48), another element could in principle be independently merged in the specifier of vP and be assigned the affected θ-role, yielding the versions of (48) with the possessive pronoun overtly realized. This possibility is not available in (49) even if the possessor is realized with inherent Case, for there is no additional θ-role to be assigned to the “topic-subject.”

This correlation between person sensitivity and “resumption” also seems to account for variation among speakers with respect to specific lexical items. Take the verb furar “puncture,” for example. Sentences such as (50a) below, with the “topic-subject” being an R-expression, are uniformly judged well formed by BP speakers, whereas sentences such as (50b) (under the relevant non-agentive “topic-subject” reading) find variation among speakers (the first author, for instance, admits them, but the second author doesn’t). Interestingly, speakers who allow (50b) also allow the corresponding sentences with a possessive pronoun in (51). This can be captured if the difference is reduced to whether or not the grammar of individual speakers encodes the unaccusative verb furar with an additional affectedness θ-role. If it does, all the sentences in (50) and (51) will be allowed; if it doesn’t, only (50a) is permitted.

(50)

a.

O

carro

furou

o

pneu

ontem.

  

the

car

punctured

the

tire

yesterday

  

“The car had a flat tire yesterday.”

 

b. %

Eu

furei

o

pneu

ontem.

 
  

I

punctured-1sg

the

tire

yesterday

 
  

“I had a flat tire yesterday.”

(51)

a. %

Eu

furei

o

meu

pneu

ontem.

  
  

I

punctured-1sg

the

my

tire

yesterday

  
  

“I had a flat tire yesterday.”

 

b. %

Eu

furei

o

pneu

do

meu

carro

ontem.

  

I

punctured-1sg

the

tire

of-the

my

car

yesterday

  

“My car had a flat tire yesterday.”

The data in (48)–(51) thus provide independent support to both Rodrigues’s (2020) distinction between two subclasses of possessor raising constructions in BP and our account of the person asymmetries observed by Kato and Ordóñez (2019).

6 Conclusion

“Topic-subject” constructions have received a lot of attention in the syntactic literature on BP. Details aside, the prevailing view is that the emergence of this type of construction in the grammar of BP is somehow related to the fact that it is no longer a canonical pro-drop language. Following Kato and Ordóñez (2019), we argued in this chapter that although BP’s becoming a non-pro-drop language has surely contributed to this innovation in the grammar, it is certainly not its primary cause. Developing Kato and Ordóñez’s insight, we proposed that with the weakening of structural Case licensing at the vP level and the loss of third-person dative clitics and third-person possessive pronouns, BP came to make extensive use of inherent Case. This ended up amplifying the possibilities for a Caseless DP to move to [Spec,TP], trigger verbal agreement, and receive nominative Case, as potential interveners have been rendered inert for minimality computations after receiving inherent Case (see Chomsky (1995) and Nunes (2017)). The general conclusion is that “topic-subject” constructions in BP are in fact a byproduct of changes at the vP level and its pronominal system.

Although there remain technical details to be further worked out, we believe that a major achievement of our proposal is that it provides a unified analysis for a series of distinct phenomena in BP that are apparently unrelated to “topic-subject” constructions such as the change of the preposition associated with directional verbs, the pervasive use of chopping relatives, and the emergence of hyper-raising out of infinitivals licensed by prepositions.