On Brazilian Portuguese 3 rd person object full pronouns

It has been observed that certain overt 3 rd person object pronouns, such as ‘it’ in paycheck sentences, propositional clitics in Portuguese and English, 3 rd person non-propositional clitics in Catalan, Spanish, Slovenian and Serbian/Croatian may allow sloppy readings. Additionally, it is well-known that Brazilian Portuguese (BP) has lost 3 rd person object clitics, which were replaced either by [-animate] null objects or by [± animate] full pronouns. The latter, differently from personal 3 rd person clitics in certain languages, do not allow strict/sloppy readings, regardless of their animacy feature specification. Nevertheless, animacy is relevant for the use of 3 rd person object full pronouns in BP since only the [+animate] ones can be focused. I propose that the BP does not fit in the tripartite division of pronouns into ‘strong/weak/clitic’. I assume that full pronouns in Romance underwent a reanalysis from demonstratives, and that BP full pronouns have a structure that precludes them from licensing strict/sloppy readings. As for the animacy distinction in focalization contexts, I assume that [+animate] full pronouns in BP are specified for [Person] and must move out of VP to check that feature. Thus, they escape the relevant vP phase and become available for checking [focus] by a low Focus head.


Introduction
An interesting property observed of certain pronouns in the literature refers to the possibility of strict/sloppy readings.For example, certain overt object pronouns allow sloppy readings, as is the case of pronouns in 'paycheck sentences' (Karttunnen 1969), shown in (1) (example (3) in Gardent 1997: 189), where it refers to Peter's paycheck: (1) John spent [his paycheck] but Peter saved it.
Additionally, as shown by Chierchia (1984), the propositional pronoun it allows both strict and sloppy readings as can be seen in ( 2): (2) John wants [to become president] and Mary wants it too.Strict reading = Mary wants John to become president Sloppy reading = Mary wants herself to become president Likewise, the propositional (neuter) clitic o 'it' in European Portuguese, as in (3), also allows strict/sloppy readings. 1  (3) O Pedro disse que vai sair logo e o Ivo também o disse the Pedro said that go leave soon and the Ivo also it said 'Pedro said that he is going to leave soon and Ivo said it too.' Strict reading = Ivo said that Pedro is going to leave soon Sloppy reading = Ivo said the he, Ivo, is going to leave soon This is also true for the propositional (neuter) ho in Catalan (4) (Quer & Rosselló 2013: 356): 2 1 As for Brazilian Portuguese, according to Cyrino (1994Cyrino ( /1997)), the propositional (neuter) clitic o, as the one in sentence (3), was the first clitic to be lost, and, differently to what happened to personal (non-propositional) clitics, it was not replaced by the full pronoun ele, as we will see in this paper.
Since there is no neuter full pronoun in BP (nor in European Portuguese)as opposed to Spanish ello, which, incidentally, can only occur in subject position -, only propositional ellipses are possible in BP (as well as in European Portuguese), allowing both strict and sloppy readings, as can be seen in (i).(i) O Pedro disse que ele vai sair logo e o Ivo também Ø/*ele disse the Pedro said that he go leave soon and the Ivo also said 'Pedro said that he is going to leave soon, and Ivo said (it) too.' Strict reading = Ivo said that Pedro is going to leave soon Sloppy reading = Ivo said he, Ivo, is going to leave soon 2 Quer & Rosselló (2013:356, ex. (29)) actually point out to a mixed reading for (4).In other words, the sentence has four possible readings as shown in (i): (i) En Perei diu que el seu fill el va veure; en Jaumej també ho diu.Strict reading = Jaumej hisi hei.CL Sloppy reading = Jaumej hisj hej.CL Mixed reading 1 = Jaumej hisj hei.CL (4) En Pere diu que el seu fill el va veure; en Jaume també ho diu. the Pere says that the his son he.CL PAST see the Jaume also it says 'Pere says that his son saw him and Jaume says it too.' Strict reading = Jaume says that Pere saw Pere's son Sloppy reading = Jaume says that he (Jaume) saw his own son On the other hand, personal (i.e., non-propositional) 3 rd person clitics do not usually allow strict/sloppy readings, except for certain cases in some languages, such as Catalan (5), Spanish (6), Slovenian (7) and Serbian-Croatian (8), which allow these readings in particular contexts: (5) (ex. (40)  In Brazilian Portuguese (henceforth, BP), it is well-known that personal 3 rd person object clitics o/a were lost and have been replaced either by [-animate] null objects (9) 3 or by full pronouns ele/ela 'he/she' (Cyrino 1994(Cyrino /1997)), the latter being possible for both [± animate] antecedents (10): 4 Mixed reading 2 = Jaumej hisi hej.CL In Spanish, strict and sloppy readings of the clitic lo in the corresponding sentence are also possible.However, the possibility for mixed readings is not that clearaccording to Francisco Ordoñez (p.c.), mixed reading 1 is a question mark, and mixed reading 2 is very difficult.

4
A reviewer points out that in some instances, null objects whose antecedents are [+animate] might be possible in BP, as in the examples provided below (reviewer's judgements): (i) a.A Lia chutou o rapaz depois que beijou / de ter beijado Ø. the Lia kicked the boy after that kissed of have kissed 'Lia kissed the boy after (she) kissed (him)/ having kissed (him).' (9) a.A Lia comprou o vestido depois de experimentar Ø.
[-animate] the Lia bought the dress after of try 'Lia bought the dress after trying (it) on.' b. *A Lia chutou o rapaz depois de beijar Ø.
[-animate] the Lia bought the dress after of try it 'Lia bought the dress after trying it on.'b.A Lia chutou o rapaz depois de beijar ele [+animate] the Lia kicked the boy after of kiss he 'Lia kicked the boy after kissing him.'However, as Cyrino (1994Cyrino ( /1997) ) shows, in contrast to null objects, object full pronouns do not allow strict/sloppy readings, as seen in ( 11 Full pronouns in object position, nevertheless, are restricted to certain contexts, depending on their animacy specification, as observed by Galves (2001:163) the Lia took the boy and after humiliated expelled chased-away 'Lia took the boy and afterwards humiliated (him)/expelled (him)/ chased (him) away.'I don't share the reviewer's intuitionthese sentences out of context do not sound natural to me, and I would need a full pronoun there.However, it's true that certain null objects are possible with animate antecedents in some contexts (see, for example Cyrino 2019), and, in fact, corpora studies do not show 100% of null objects with inanimate antecedents.Therefore, it might be that the reviewer's observation relates to other possible causes, such as topicality or other pragmatic effects, but these factors still await further investigation.

5
A reviewer points out that in some cases full pronouns could allow sloppy readings, for example, as in sentences as (i): (i) ?O Ivo gastou o dinheiro/prêmio com viagem e o Pedro gastou ele the Ivo spent the Money prize with trip and the Pedro spent it com roupa.with clothes 'Ivo spent the money/prize with trips and Pedro spent it with clothes.'Although I agree that pragmatics may be involved in the case of a sloppy interpretation in (i) as suggested by the reviewersay, when the sentence is uttered in a context where there were two winners in a certain lottery -, I think that these are special situations.In my judgement, the strict readingand not the sloppy readingis the first interpretation one gets in this sentence, and that is why I would mark (i) with a question mark (or even #), to indicate that, out of the blue, no sloppy reading of ele is obtained.

Weak Clitic
In order to explain the difference related to animacy in ( 12)-( 13), Galves follows Cardinaletti & Starke (1999) and suggests that object full pronouns in BP can be classified into strong and weak.Animacy seems to be a property relevant for the strong/weak distinction pronouns proposed by Cardinaletti & Starke (1999:156): strong pronouns cannot refer to [-animate] antecedents; weak (deficient) pronouns can refer to [-animate] antecedents, but only when the latter are prominent discourse antecedentsin other words, weak pronouns are "referentially deficient".Strong pronouns must refer to new information or focus, whereas deficient pronouns are anaphoric.Hence, according to Galves, only [+animate] full pronouns in object 6 Notice that instances of the full pronouns ele/ela when the antecedent is not in the discourse, as in an example provided by a reviewer (i), are not the focus of this paper: (i) However, the proposed stipulation that [+animate] are strong and [-animate] are weak pronouns does not explain why there is a distinction with respect to focus, as seen in ( 15)-( 16).
In an interesting paper, Manzini (2014) proposes that we don't need the weak vs. strong divide for full pronouns, since independently motivated categories (such as Case, focus, etc.) may be sufficient to yield the attested distributions of non-clitic pronouns.Following her insight, I assume that crossing distribution + morphology is not necessary to account for the distribution of [± animate] full pronouns (ele/ela 'he/she') in object position in BP.I propose that the syntactic properties of [+animate] pronouns, which are different from [-animate] ones, are sufficient to explain the attested focus-related differences.Hence, no double categorization of object full pronouns (as strong/weak) in BP is necessary.
Given that BP has lost [± animate] personal 3 rd person clitics o/a, it could be reasonable to suppose that their substitute, 3 rd person object full pronouns ele/ela 'he/she', would have the same properties as their 'antecessor'.However, regarding object full pronouns, as we have seen above, there is a [± animate] distinction which applies in contexts of focalization.This distinction doesn't carry on to the possibility strict/sloppy readings phenomena, since differently from propositional clitics and from 3 rd person clitics in certain languages seen above, full pronouns do not allow ambiguous readings regardless of their animacy features. 7 In this paper, I focus on the syntax of 3 rd person object full pronouns in BP, and I advance a proposal to explain their behavior.In other words, I explore answers to the following two questions: (i) why object full pronouns do not allow strict and 7 Note that the availability of strict/sloppy readings for propositional pronouns and certain 3 rd person (non-propositional) clitics in some languages is not the focus of this work.However, the topic has been continuously investigated in the literature, as can be seen in the references given in this paper.sloppy readings regardless of animacy; (ii) why only [+animate] full pronouns can be focalized.
The paper is organized as follows.In Section 2, I advance an analysis for the structure of 3 rd person object full pronouns that will be able to account for their appearance in the language and their lack of strict/sloppy readings.Next, in Section 3, I explore the feature composition and the syntax of full pronouns in order to explain their properties in BP regarding focalization.Section 4 concludes the paper.

On the structure of 3 rd person object full pronouns in BP
In this section, I first look at a proposal for the structure and emergence of 3 rd person full pronouns in Romance, and I advance an account for the structure of 3 rd person object full pronouns in BP.Then, I present the hypothesis for their emergence in the language, which was facilitated by the loss of 3 rd person clitics.

The structure of full pronouns in BP
According to Giusti (2001), from Latin to Romance there was a reanalysis of the demonstrative ille 'this' into a full pronoun.She bases her proposal on the work by Renzi (1997), where he analyzes the categories demonstrative, personal pronoun, and article as a bundle of semantic and syntactic features.These elements have different structural positions due to the partial difference/similarity in features they share.For example, whereas the deitic feature of demonstratives is included ('>') in definite, which is included in anaphoric, the anaphoric feature is in complementary distribution ('/') with ostensive, and the adjectival feature, with pronominal status, as seen in ( 18) (Giusti 2001:158) Giusti (2001) considers deitic as an independent feature since it can be found in association with non-nominal categories such as verbs and adverbs.It is the presence/absence of this feature which differentiates demonstrative vs. pronouns/ articles.She includes this insight in Giusti (2015), as we will see below.
Giusti also observes that the feature definite, present in the three categories is what unifies them synchronically and what allowed the diachronic development relating them.Giusti did not consider the features pronoun and adjective in her analysis since these features are related to the categorial status of the three items.Hence, the author proposes the semantic and syntactic bundle of features for demonstratives (20) and pronouns (21) (Giusti 2001: 165) Renzi (1997, apud Giusti 2001), the development of the demonstrative into the pronoun and the definite article was the result of a loss of semantic features.Giusti, however, proposes that full pronouns in Romance underwent a morphological reanalysis from demonstratives; both demonstratives and pronouns check their features in the same position, namely Spec, DP, as in ( 22).Additionally, there was a reanalysis of a pronoun in Spec,DP into a head in D, which resulted in the development of the article, and this reanalysis was also responsible for the further development of the strong pronoun into the clitic pronoun (23a-23b). 88 Because there was no lexical material in the head in (23a) and in the specifier (23b) "the two structures may have coexisted in the language for several generations.By the time the first syllable had completely disappeared, [(23a)] had also disappeared in favour of [(23b)].It is reasonable to suppose that the trigger for the "new" analysis was the phonological (23) a.
DP 3 Hence, the article and the 3 rd person clitic are in the same structural position.Since BP lost 3 rd person clitics, I assume that the full pronouns ele/ela that replaced the personal (i.e., non-propositional) clitics are demonstratives, but in complementary distribution with the latter.In this respect, I follow Giusti (2015), who shows that full pronouns provide a referential index to the nominal expression, saturating an open position associated with an (empty) N: the presence of this referential index turns the pronouns into a referential expression.
Crucially, in Giusti's (2015:159) proposal, [Person] features are not part of the extended projection of N, but are merged as specifiers of N: "pronouns are not (necessarily) the head of full DPs, but are IndPs [Indexical Phrases, SC] which can stand alone or be merged with N."She proposes a syntactic configuration which includes an ι-operator in demonstratives, personal pronouns, and proper names that occupy the specifier of a nominal expression, as in (24) (ex (14), Giusti 2015:56), in which the article may or may not appear: 9 weakening of (IL)LE.On the other hand, the existence of the "new" analysis may have quickened the loss of the first syllable."(Giusti 2001:167).
9 Assuming examples such as (i) (ex.(58a) in Giusti 2015:155), Giusti proposes that these elements combine with N and ultimately merge at the left edge of the DP.(i) Voi (*i) professori credete che sia facile ma per noi (*gli) studenti è difficile.'You (the) professors believe that it is easy, but for us (the) students it is difficult.'Although at first sight these examples seem to be only possible with 1 st and 2 nd person plurals, Giusti mentions that these constructions are also possible with 3 rd person pronouns in contrastive contexts, such as noi filologi contro loro linguists 'we philologists vs. them linguists.'The same is true for BP, as witness the grammaticality of nós filólogos contra eles linguistas.Interestingly, no plural or contrastive contexts are needed in BP, as the grammaticality of Eu artista e você músico estamos na mesma situação!Lit.'I artist and you musician are in the same situation!'shows.
(24) Furthermore, Giusti proposes that articles have functional features related to noun morphology, that is, Case, Number and Gender; they do not have inherent semantic features.Hence, articles are different from other determiners as pronouns in that only the latter have indexical features: [Person] is not a part of a feature bundle with N because, according to Giusti, in no language the article is inflected for Person. 10 In this paper, I assume that 3 rd person object full pronouns in BP have a referential index (parallel to overt demonstratives), and they are merged in a DP-like structure/Indexical Phrase, as in ( 25). 11This structure will account for the properties they show in the language, as we will see below.In the next subsection, I turn to my account for the emergence of 3 rd person object full pronouns in BP.

On the emergence of 3 rd person object full pronouns in BP
In BP, the emergence of full pronouns ele/ela for [±animate, +specific] (nonpropositional) antecedents as the consequence of the loss of 3 rd person clitics is welldocumented in several diachronic works.The innovation already appears in corpora in the 19 th century and its occurrence increases thereafter (Cyrino 1994/1997, Marques de Sousa 2017, Duarte et al. 2021, among others).Cyrino (2010Cyrino ( , 2012) ) hypothesizes that this replacement of the personal 3 rd person clitic by a full pronoun was possible given that causative/perception verbs in 10 Giusti (2015:110): "…I take Person, reference, or deixis, to be separate features from those that belong to the bundle of N (which denotes a property).Person is part of the indexical element that saturates the open position of N, turning N from a property into a referential expression (Higginbotham 1985(Higginbotham , 1987)).Like all other arguments, this element must be merged as a specifier (in the lexical layer […] or directly in DP according to Cinque 2005, Adger 2013) and checked in the nominal Left Edge (SpecDP)."11 Giusti (2015:157) assumes that full pronouns are in (Indexical Phrases, IndPs) that start as low modifiers: ] In (i), "N/D is the phasal head and can be overt or non-overt, according to the inflectional properties of N as well as the Concord requirements of the pronouns." At this point it is not crucial for my analysis whether full pronouns are at the specifier of a (empty) nominal or whether they are moved there.It is crucial, though, that 3 rd person object full pronouns in BP in fact do not co-occur with an article (nor with a clitic: the latter has been lost, and there is no clitic doubling int BP, as will be pointed out below).
Portuguese allow inflected infinitives as complements (Martins 2004, Guasti 2006), as seen in ( 26).The agreement marker in the inflected infinitive shows that the embedded full pronoun subject ele/ela is clearly nominative.
(26) O professor mandou/viu eles/elas apagarem o quadro.the teacher had saw he.PL/she.PL erase.3PL the board 'The teacher had/saw them erase the blackboard.' The uninflected infinitive is also possible in such sentences.In European Portuguese (EP), the embedded subject must be an accusative 3 rd person clitican ECM construction (27) -; there is no agreement in the verb to allow nominative Case.
(27) O professor mandou/viu-os/as apagar o quadro.the teacher had saw-he/she.3CL.PL erase the board 'The teacher had/saw them erase the blackboard.'However, in BP, as opposed to EP, a 3 rd person full pronoun may show up in that context instead of a 3 rd person clitic, as seen in ( 28): (28) O professor mandou/viu eles/elas apagar o quadro.the teacher had saw he.PL/she.PL erase the board 'The teacher had/saw them erase the blackboard.' Cyrino (2010Cyrino ( , 2012)), considering that sentences with causatives and perception verbs are abundant in the linguistic data the child encounters in the acquisition process, proposes that these contexts were the source for the emergence of full pronouns in object position and the consequent replacement of 3 rd person clitics.In other words, precisely the possibility of structures like (26), where full pronouns are possible, provided the necessary evidence for a diachronic reanalysis.The reanalysis was helped by two other concomitant phenomena that might have strengthened such change: (i) the well-known weakening of subject-verb agreement in BP (Duarte 1995, among others), which allowed sentences as (28); (ii) the emergence of null objects and consequent the decrease of 3 rd person clitics (see Cyrino 1994Cyrino /1997)).
Considering this scenario, I assume that, since BP lost 3 rd person clitics (D), represented as the struck-through o/a in (29a), the next possible replacement for the clitic available in the pronoun structure was the higher DP/IndPpron, ele/ela.In other words, the structure in (25) above, here repeated as (29b), is the result of the loss of 3 rd person clitics in BP.As I pointed out above, 3 rd person object full pronouns in BP must have a feature composition that includes an indexical feature. 12Besides, as I will show in the next section, object full pronouns in the language may include or not a [Person] feature, depending on their animacy specification.

Feature composition of 3 rd person object full pronouns in BP
In this section, I propose answers to the questions raised in the introduction, namely, (i) why object full pronouns do not allow strict and sloppy readings regardless of animacy; (ii) why only [+animate] full pronouns can be focalized.

Accounting for no sloppy readings in BP 3 rd person object full pronouns
As seen above in (11), 3 rd person object full pronouns in BP only allow strict readings, regardless of their [animacy] feature specification (see also Cyrino 2021b) See also Cerqueira (2015Cerqueira ( , 2019)), who, assuming a geometry of features framework, proposes that BP object full pronouns include [+specificity] and [+definiteness] in their feature composition.
This might be regarding a specific property of full pronouns in BP, if we compare them to other pronominal elements that allow sloppy readings, such as the paycheck and propositional pronouns and clitics, along with other non-propositional 3 rd person clitics in certain languages (as seen above in the introduction), which do allow sloppy readings. 13 In order to explain strict/sloppy readings with 3 rd person clitics in Spanish, 14 as in (32), Cyrino & Ordoñez (2018) have assumed that clitics are always doubled (see also Bosković 2018 and footnote 10).The authors modify Uriagereka's version and propose that the clitic is truly a D head, and it has a DP as its complement (see also Giusti 2001Giusti , 2015)).In the case of possessive DPs as antecedents, the complement DP may be either deleted (33b) or pronounced (33a), in which case the DP would be double by the determiner.Hence, the authors do not think that the DP is a pro (pace Uriagereka 1996) but that it is a deleted DP (indicated in (33b) by the strikethrough).( 32 The same happens with doubled possessive DPs as (34) shows.The insertion of the a is only done later in the derivation (Ordóñez & Roca, 2018).In this case, elision of the DP is not possible since it is animate and it will move out of the DP for DOM reasons, the D being not able to license the ellipsis.In fact, differently from the case in Catalan seen above in (5), in Spanish sloppy readings with clitics are harder with [+animate] antecedents, as shown in (35): 15, 16   13 Note that Bosković (2018) relates the possibility for sloppy readings of clitics in some Slavic languages to the fact that these languages do not have DPs.The author proposes a clitic doubling structure, whereby the clitic licenses the NP (argument) ellipsis, giving rise to the possible sloppy reading.However, as mentioned in this paper, the possibility for sloppy readings for personal 3 rd person clitics does occur in languages that have DPs, such as Catalan and Spanish.Because of lack of space, this topic is not explored in this paper (see note 4).14 See also Ruas & Ordoñez (2021).
15 See Cyrino & Ordoñez (2018) for an account of the impossibility of strict/sloppy readings of clitics referring to [+animate] antecedents.

16
A reviewer suggests that the problem with ( 35) is that it includes a full object pronoun él.In this case, the structure of the full pronoun might also be responsible for the ungrammatical sloppy reading, as is the case in BP.But the point raised in this section is related to the structure for the DP involving double clitics/determiners, seen above in ( 33)-(34).Interestingly, in the sentence below, where no full pronoun is present and the antecedent As for BP, I attribute the unavailability of strict/sloppy readings of 3 rd person object full pronouns to the fact that they are inherently indexical, as seen in Section 2 above.As we will point out below, BP does not have clitic doubling, as is the case in EP; hence full pronouns are DPs like other referential pronouns, and they have the structure I proposed in (25) above.Incidentally, they have the same form as demonstratives in Latin (ille), and I attribute to that fact the lack of sloppy readings for BP full object pronouns, since they refer strictly to an antecedent/referent as demonstratives do.
Below, I turn to the availability of focus being restricted to [+animate] 3 rd person full object pronouns in BP.I attribute that restriction to the fact that only these pronouns have a specified [Person] feature.

On the specification of [PERSON] features for BP 3 rd person object full pronouns
Assuming Richards's (2008) analysis of the feature composition of pronouns, Cyrino (2016)  The insertion of a (DOM) is only done later in the derivation (Ordóñez & Roca, 2019).
For Richards (2008:140), [Person] is an exclusive (syntactic) property of animate nominals, "since a person specification on inanimates is redundant (it is always 3) and thus plausibly left unspecified".Hence [-animate] nominals bear only number (and gender) featuresaccording to Richards, inanimate nominals are 'defective' in the agreement system (in the sense of Chomsky 2001).Therefore, animacy in syntax for him is the result of [±Person] feature checking. 18 Cyrino (2016, 2021b) proposes that animacy is expressed in syntax as the result of the movement of a category specified either with a [+Person] or a [-Person] feature to the specifier of a functional category that has [Person] and [EPP] features, a functional head dubbed F[Person] by the author.Accordingly, F[Person] is located below vP and above InnAsp (Travis 2010, MacDonald 2008).The movement triggered by the [± Person] feature-checking reflects a DOM-like requirement (see Torrego 1998, Lopez 2012, Rodriguez-Mondoñedo 2007, Ordoñez & Roca 2019, among others). 19 In this paper, I focus on the fact that animacy is relevant for 3 rd person object full pronouns in BP.Although moving out of VP to check animacy features, BP 3 rd person object full pronouns are not doubled by clitics 20 and are not marked for DOM as in the Spanish case seen above in (36) (see also Cyrino & Irimia, to appear).
However, since there is a contrast related to animacy for object full pronouns in BPnamely, only [+animate] ones can be focalized -, I assume that 3 rd person object full pronouns in BP do carry the features seen in Table 1.In other words, animate full pronouns ele/ela 'he/she' have a [-Person] feature, and inanimate ones do not have a specification for that feature, that is, they are 'person-less'.Hence, [+animate] (i.e.[+Person] or [-Person]) full pronouns move to the specifier of F [Person] triggered by its EPP feature.This movement is represented in the structure in (37b) for a sentence as (37a): 18 Incidentally, it is important to consider that animacy is relevant for several syntactic phenomena such as: (i) Differential Object Marking in several languages (Fábregas 2013, among many others)animates are specially marked; (ii) occurrence of verbal sufixes in Blackfoot (Bliss 2010)there is a dedicated order whereby transitive animates precede transitive inanimates; (iii) Double Object Constructions (Larson 1988, Levin 2008, Demuth et. al 2005) there is a specific order, whereby animates precede inanimates; (iv) Agreement (Baker 1996, Hualde 1989, Woolford 2000) in Mohawk and KiRimi, object agreement is only possible with animates (inanimates incorporate in Mohawk); (v) Person-Case Constraint (Ormazabal & Romero 2007, 2013) in leísta dialects, the clitic le is a marker of animacy, since it is used not only for dative arguments but also for accusative animate arguments), lo is used if the object is inanimate; (vi) null objects in BP (Duarte 1986, Cyrino 1994/1997, a.o.).19 Note, however, that, as argued by Cyrino & Irimia (2019, to appear), in BP there are just a few restricted cases of DOM in object DPs.The authors show that BP DOM, which is possible in certain configurations, is triggered by the presence of nominals with a complex structure, which contain discourse-related features beyond Case and which need licensing by discourse-related Topic projections in the left periphery, as is the case with a language such as Romanian.20 This is not the case in European Portuguese.See the example from Magro (2021: 33, ex. (17) Now, if this is correct, we may understand why it is the case that only [+animate] full pronouns can be focalized in BP. 21In a nutshell, I propose that [+animate] 3 rd person object full pronouns in BP move to a position outside a relevant vP-phase, escapes Transfer and move to a higher position to check a focus feature.On the other hand, [-animate] full pronouns in BP stay in situ, they cannot be focalized because they are trapped inside a relevant vP-phase and they cannot move further. 22 Before I flesh out my proposal, let's start by laying out some assumptions.I assume a dynamic definition of phase heads (Carnie 2005): (39) Phases must contain: (i) a sole argument; (ii) a predicative element (either V or v) that introduces the argument; (iii) a temporal operator (a functional category) that locates the predicate and the argument in time and space (that is, Asp or T). 2321 It is important to emphasize at this point that my analysis in this paper focuses on full pronouns in object position in BP, which includes an indexical feature as seen in the structure in (29b).A related question, the answer to which I will not develop here, is: why is it the case that object DPs, or demonstrative pronouns, but not full pronouns, can be focalized irrespectively of their animacy?A reviewer suggests that inanimate full pronouns are deficient in that they are not furnished with discourse features.This could explain why DP objects or demonstratives may move to a low left periphery; crucially, that movement is not related to animacy, but to discourse features.I leave the development of this discussion for further work.

22
See also Sheehan & Cyrino (2018, 2022), who present a proposal to explain why long passives are blocked in the complement of causatives and some perception verbs: depending on the size of the complement, the internal argument cannot escape a v-related phase.[EndP [End' End [v goal [v' v... ]]] (iii) Agent Phase [TP [T' T [vp agent [v' v... ]] See also Boskovic (2014Boskovic ( , 2016) ) for a dynamic proposal for phases.
The relevant phase heads for the phenomena under study are v and InnAsp, marked in boldface in (40).
(40) [vP v [F[person] Additionally, I assume the Phase Impenetrability Condition in Chomsky (2001), the so-called PIC2, which states: (41) The domain of the phase head H becomes inaccessible at the point where the next phase head is merged.
PIC2 differs from PIC1 (Chomsky 2000) in providing a 'window of opportunity' for checking operations from the outside of a phase-head before the next phase head is merged.
Another ingredient is the assumption of a low left periphery.Belletti (2004), in a seminal work successfully developed in the recent literature (Mioto 2003, Jiménez-Fernández 2009, Lacerda 2015, 2020, Cépeda & Cyrino 2020 a.o.), argues that the verb phrase is endowed with a fully-fledged periphery of discourse-related structural positions, see (42), in parallel with the high left periphery.
(42) [FocP [TopP [vP ]]] With these preliminary assumptions in mind, we turn to my proposal to explain the difference between BP [+animate] and [-animate] 3 rd person full pronouns in object position with respect to focus.As said above, [+animate] (i.e., [+Person] or [-Person]) DPs/full pronouns move to the specifier of F[Person] triggered by its EPP feature.However, [-animate] (i.e., 'person-less') 3 rd person full pronouns stay in situ.
By PIC2, before the next phase v is merged, InnAsp is not yet a phase.F[person] can probe to check its features and ele/ela [±Person], that is [+animate], raises.In other words, after the InnAsp phase is merged, movement of ele/ela[+animate] can take place before the next potential phase head, v, is merged.Consider the structure in (43), where this movement of ele (/ela) is represented: Crucially, according to PIC2, v will only be a phase head after the next phase, C, is merged.Therefore, ele[+animate] can be probed and move up to the low Focus head in BP.This movement has the result that ele[+animate] can be focalized (and even move further to C, see Cyrino 2019Cyrino , 2021a)).
Figure 1 below shows the derivation of focused [+animate] pronouns in BP under PIC2: X can agree with YP since YP is not spelled out until Z is merged.
This analysis also explains why ele/ela [-animate] cannot be focalized.The pronoun is trapped inside the phase, and it cannot raise to the low FocP, since the complement of the InnAsp phase will have been transferred to the interfaces at the relevant point of the derivation (after v is merged).
In this subsection, I focused on the contrast related to animacy in certain object full pronouns constructions in BP: only [+animate] ones can get (contrastive) focus.I show that what appears to be a single lexical element (ele/ela), undergo, in fact, different syntactic operations according to their different animacy feature make up.I advanced an explanation for this difference in terms of phase theory: given that vP is only a phase after C is merged, [+animate] full pronouns in object position in BP first move up to a functional category below vP to check their [Person] features, escaping the lower phase, InnAsp; from there they can move to FocP to check focus features.In contrast, [-animate] full pronouns are "trapped" inside the InnAsp-phase and cannot move further as they are inaccessible to operations from the outside of the phase; as a result, they cannot be focalized.

Conclusion
In this paper, I show that the loss of 3 rd person clitics might have paved the way for the appearance of object full pronouns, which have the referentiality of (their Latin 'antecessors') demonstratives and are, thus, indexical phrases.In this way, I was able to account for the lack of strict/sloppy readings for these pronouns.
Interestingly, full object pronouns are also possible in languages like EP (ele/ela), Spanish (él, ella) or Catalan (ell, ella), but they have some restrictions not present for the object full pronouns in BP: (i) they refer only to [+animate] antecedents; (ii) they must be doubled by clitics; (iii) they are DOM marked.In contrast, BP object full pronouns ele/ela can refer either to [+animate] or to [-animate] antecedents, just like their 3 rd person clitic 'antecessors', and they are not clitic doubled.
Nevertheless, there is one environment where the [± animate] distinction is relevant for BP object full pronouns: in contexts of focalization.In order to explain this fact, I relate this distinction to the well-studied phenomenon of Differential Object Marking (DOM), assuming with the literature that [+animate] full pronouns move out of VP in languages like Spanish.Although they are not generally overtly marked, BP [+animate] full pronouns have DOM syntax, and that is why they can be focalized.Because of their [Person] features, they must move to a position outside the InnAsp phase, where they can be probed and move up to a low Focus head.Inanimate full pronouns, on the other hand, remain in situ, trapped inside that phase.
My aim in this paper was to contribute to the understanding of BP full pronouns in object position, as well as to the study of other languages in which pronouns have properties related to phenomena such as animacy, DOM marking, restrictions in focalization and in strict/sloppy readings.

Figure 1 .
Figure 1.Derivation of focalized [+animate] full pronouns in BP under PIC2 in Quer & Rosselló 2013: 357)En Pere estima [la seva mare] i en Joan també l' estima.det Pere loves.the his mother and the Joan also her.CL-loves 'Pere loves his mother and Joan also loves her.' . demonstrative and full pronouns are maximal projections (DPs) in the specifier of another DP, whose head is the article (D).
I/II/II/IV/V/VI person iv.[it projects an extended projection] She offers the structure in (22) (ex.(2) in Giusti 2001:159), whereby Ivo pôs seu anel no cofre, mas o Pedro guardou ele na gaveta.the Ivo put his ring in-the safe but the Pedro kept it in-the drawer 'Ivo put his ring in the safe, but Pedro kept it in the drawer.' ) María lavó [su bici]i, y Juan también lai lavó.María washed her bike and Juan too it.CL washed 'María washed her bike and Juan also washed it.' la= Maria's bike (sloppy reading) la= Juan's bike (strict reading) (33) a. [DP la [DP su bici ]] (Spell out doubled) b. [DP la [DP su bici ]] (Elided doubled) (35) Pedro vio a [su padre] y María también lo vio a él.Pedro saw DOM his father and María too him.CL saw DOM he 'Pedro saw his father and María saw him too.' él = Pedro's father (strict reading) él = *Maria's father (sloppy reading)

Table 1 .
has proposed that [+animate] objects in BP are different from [-animate] ones.She claims that [+animate] full pronouns and object DPs have a specified [Person] feature whereas [-animate] full pronouns and object DPs lack that feature.Consider Table 1.Feature composition of person-animacy