A Diachronic Overview of the Prepositional Accusative in Portuguese

Abstract

One of the several differences between Modern European Portuguese (EP) and Modern Brazilian Portuguese (BP) is the prepositional expression of complements licensed by the preposition a. While in EP the preposition a occurs in several contexts, this element has been substituted by other strategies in BP, as is extensively discussed in the literature. The aim of this paper is to investigate the historical behavior of a-marked prepositional accusatives (PP-ACC) in Portuguese. In order to do so, a search was conducted for PP-ACCs in the Historical Portuguese Corpus Tycho Brahe. The results showed an increase of PP-ACCs in the 17th century, followed by a decrease in the 18th century. Thereafter, unmarked accusatives (NP-ACC) were analyzed in the corpus, which resulted in 7756 sentences, contrasting with 624 PP-ACCs in the same contexts. This result shows that the a-marked accusative is far less common than bare accusatives in Historical Portuguese. Psych verbs, however, behaved differently, showing a constant increase in PP-ACCs. In EP, the preposition a still introduces Experiencer arguments in structures with some psych verbs (O vinho agradou ao João—lit. ‘The wine pleased ‘to’ John’). In BP, the preposition a has disappeared in psych predicates (O vinho agradou Ø o João—‘The wine pleased John’). In both Modern EP and BP, most PP-ACCs have become typical unmarked direct objects. In the context of psych verbs, however, structural accusative assignment has shifted to structural dative Case in Modern EP, so as to ascertain the interpretation of the Experiencer in the internal argument via the preposition a. While in Modern BP, the argument is not overtly marked since it receives inherent accusative case in the derivation.

1. Introduction

The argument structures found in Modern European Portuguese (EP) and Modern Brazilian Portuguese (BP) have undergone several changes historically (cf. Galves 2001, 2007, 2020, a.o.), especially regarding the prepositional expression of complements licensed by the preposition a. In EP, the preposition a is used to introduce indirect dative arguments, as in (1a) and (2a) below. In BP, synchronic data shows that the preposition a found in EP has been replaced by other prepositions with a more transparent thematic role, such as para and de in several contexts ((1b) and (2b)) (Torres Morais and Salles 2010; Torres Morais and Berlinck 2018). Additionally, some predicates with psych verbs encode their Experiencer argument with the preposition a in EP, but the same complement is unmarked in BP (cf. 3):

(1)	a.	A	Maria	enviou	uma	carta	ao João	/enviou-lhe	uma	carta.
		The	Maria	sent	a	letter	Pa(to) the João	/sent-CL.DAT.3SG	a	letter.
	b.	A	Maria	enviou	uma	carta	ao	/para o João	/ele.
		The	Maria	sent	a	letter	Pa(to)	/Ppara(to) the João	/him.
(2)	a.	A	Maria	roubou	o	relógio	ao João	/roubou-lhe	o relógio.
		The	Maria	stole	the	watch	Pa(to).the João	/stole-CL.DAT.3SG	the watch.
	b.	A	Maria	roubou	o	relógio	do João	/dele.
		The	Maria	stole	the	watch	Pde(of).the João	/his.
(3)	a.	O	vinho	agradou	aos convidados		/agradou-lhes.
		The	wine	pleased	Pa(to).the guests		/pleased- CL.DAT.3PL
	b.	O	vinho	agradou	(*a) os convidados		/os agradou	/agradou eles.
		The	wine	pleased	Pa(to).the guests		/CL.ACC.M.3PLpleased	/pleased they.ACC1

The comparison between both varieties is based on the fact that they share a common historical background, as the language that arrived in Brazil in the 16th century was Classical Portuguese (CP) (Galves 2007). Over the years, CP has become EP and BP, as these varieties have parted ways in several syntactic and morphological aspects, such as the different strategies used to encode their arguments (cf. Galves 2001). Therefore, the texts produced by Portuguese authors from the 16th century on are essential to the study of both Portuguese varieties.2

Specifically, the main goal of this paper is to investigate the contexts in which accusative arguments were a-marked historically, as example (4) from the 17th century illustrates. Synchronic data from both BP and EP show that accusative direct objects selected by dicendi verbs such as chamar, ‘call’, are never marked.3

(4)	(…)	chama	a	Pedro	e	André,	e	aos filhos	do	Zebedeu.
		call	Pa(to)	Pedro	and	André,	and	Pa(to).the sons	of.the	Zebedeu
	‘(…)	call Pedro and André, and Zebedeu’s sons.’
		17th century (V_004,202.1819)4

The literature on Portuguese variation and change highlights that the great historical modifications in the structure of BP began to be noticed in the 18th century; some became stable in the 19th and 20th centuries, while others are still in the process of variation in current BP (Kato et al. 2009; Torres Morais and Berlinck 2018; Galves 2020). Furthermore, Galves (2020) attests that even though there are several studies with an empirical basis about the 19th and 20th centuries, there are fewer works on the 17th and 18th centuries. For that reason, an investigation on Historical Portuguese from the 16th to the 19th century is very relevant to understanding the contexts and possible reasons for the changes attested in EP and BP. Accordingly, to achieve the main aim of this paper, 19 texts from the Historical Portuguese corpus, organized by the Tycho Brahe Project at Unicamp, were analyzed.5 These texts cover precisely the period between the 16th and 19th centuries. The data collected provided clear evidence, especially, for psych verbs, which showed a distinct behavior in licensing arguments with the preposition a throughout the centuries, as we will discuss in more detail in the results section.

As shown in examples (1) to (3), the preposition a introduces indirect dative arguments in Modern EP. This element has been analyzed as a dummy marker that licenses the dative argument. According to Torres Morais (2007), there are two pieces of evidence for the a-DP dative status in EP: firstly, the possibility that the argument will be displaced to the topic position, maintaining the preposition (Ao João, a Maria enviou uma carta, ‘To João, Maria sent a letter’). Secondly, the fact that the a-DP in EP always alternates with dative clitics (lhe/lhes), as attested to in the examples above.

In Modern BP, on the other hand, the dative clitic lhe(s) has been replaced by other strategies, such as 3rd person pronouns preceded by contentful prepositions, such as de/para ele(s)/ela(s), ‘of/to him/her/them’, as exemplified above (Calindro 2015, 2016; Torres Morais and Berlinck 2018; Bazenga and Rodrigues 2019).6 This is evidence that case assignment in BP is different from EP, since the morphological case in the form of the clitic lhe has been lost and the dummy preposition a has been replaced by contentful prepositions that assign an oblique case to the argument they introduce or has disappeared, as in the context of psych verbs (cf. 3b) (Calindro 2015, 2020).

As for psych verbs specifically, their structure is transitive (cf. 3), not ditransitive as in (1) and (2). The direct object in (3) is the Experiencer of the event, overtly marked in EP (3a) but not in BP (3b). Most of the literature on variation and change in Portuguese has addressed ditransitive verbs with Goal indirect objects introduced by the preposition a in EP and para in BP. Transitive contexts with prepositional accusatives seem to be less explored (Ramos 1992; Gibrail 2003; Calindro 2017). Even less discussed are syntactic issues related to the licensing of the preposition a in psych predicates with Experiencer arguments (cf. Figueiredo Silva 2007; Carneiro and Naves 2010). This fact, coupled with the corpus analysis, led us to a second goal, which is to analyze the reasons why EP and BP mark Experiencers in a distinct way. We are assuming that, with psych verbs, since the Experiencer argument is generated as an internal argument, a conflict is displayed between case and theta assignment. To solve this divergence, the argument is a-marked, in some contexts, in predicates with psych verbs in EP. In BP, on the other hand, the Experiencer argument enters the derivation with inherent accusative case, and the Theme, such as vinho, ‘wine’, in (3b) is raised to receive the nominative case (in Chomsky’s (1981) terms).7

Finally, in the literature, a-marked accusative arguments are sometimes addressed as PP-ACC and/or are treated as differential object marking (DOM), so, in the next section, we present an overall view of how these phenomena are related and addressed in the studies about Portuguese, followed by a discussion on psych verbs. In Section 3, the research methods are exposed. Section 4 shares the data results and a discussion on the categorical status of a-marked objects. Based on these considerations, our proposal is put forward in Section 5, followed by the final remarks in the last section.

2. PP-ACC and DOM in Historical Portuguese

It seems not to be clear in the literature how and why accusative a-marking/DOM appeared in Romance. According to Bossong (1991), many languages employ what he calls DOM in order to syntactically encode subsets of direct objects. This phenomenon is present in several Romance languages, such as Spanish, Catalan, Romanian, and Italo Romance varieties (Andriani 2015). With the exception of Romanian, the DOM marker is homophonous to the dative used to introduce indirect objects, as the preposition a in Portuguese. Therefore, the data presented previously indicate the possibility of a-marked objects in Portuguese being instances of DOM. On the surface, DOM seems to share features with prototypical prepositions, hence this phenomenon may also be found in the literature about Romance as PP-ACC. According to Fábregas (2013), DOM only partially shares features of prototypical prepositions. Additionally, Gerards (2020) points out that the diachronic data on DOM poses a problem, as it has been attested that a-marking first appeared with [+human] personal pronouns—Isso agradou a ela, ‘lit’. This pleased ‘to’ her (Döhla 2014); that is, specifically, one of the few elements that display overt case marking in all Romance languages and which, for that reason would not need additional Case marking.

Overt marking seems to emerge when the argument is autonomous from the verb, i.e., with self-constituent objects. In fact, Bossong (1991) discusses that being dependent or independent of the verbs is a factor to be taken into account when analyzing instances of DOM. This leads us to the type of verb the argument is associated with. For example, in Spanish, there are two verbs that always select a marked argument: buscar (to look for) and querer (to want) (Bossong 1991, pp. 158–59).8

Actually, Bossong (1991, p. 160) claims that DOM is related to three domains: i. the domain of inherence − [+deictic] < [+proper] < {[+parent]] < [+human] < {[+person]] < [+animated] < [+discrete]; ii. the domain of reference − [+individuality] < [+/−referentiality]/[+/−definiteness]; and iii. the domain of constituence − [+/−independent existence]/[+/−pragmatic constituency]. The author argues that it is common to use a mixture of these dimensions when analyzing overtly marked arguments. For Spanish and Romanian, Bossong claims inherence is the dominant factor required for DOM to occur. In this paper, as further explained in the methods section, we analyzed the data via verb classes, so we will consider what the author calls constituence—the argument being dependent or not on the verb—the main factor for our analysis at this point of the research. López (2016) also claims that syntactic configuration is one of the aspects that trigger DOM in Romance, along with features of the verbs and features of the object.

Synchronic EP data shows overtly marked arguments in predicates with some psych verbs (cf. 3a). Hence, the main question that emerged from the examples presented in the introduction was when and how this marking appeared historically and whether structures with verbs other than psych verbs also present a-marked arguments. Therefore, as will be explained in more detail in the methodology and results sections, the first search in the corpus was for all the accusatives marked by the preposition a that had been labeled by the Tycho Brahe group as PP-ACCs. The results confirmed that psych verbs behaved differently from other verb groups. This was indeed a very relevant context to address in order to understand how a-marked arguments behave in Portuguese. Nevertheless, before addressing the types of verbs present in the corpus, another question required consideration—should we consider the a-marked arguments instances of PP-ACCs or DOM in Portuguese?

Given the DOM characteristics put forth by the aforementioned authors, PP-ACC seems to be part DOM, which can be seen as a wider phenomenon, as it is related to several factors. Gerards (2020, p. 1) observes that the Romance-specific term PP-ACC may be misleading and inferior, so the author suggests a “categorically agnostic label flagging DOM”. This seems to be a valid solution; however, it is not the aim of this paper to solve this issue. What concerns us at the moment is the historical change Portuguese has undergone, considering that our results showed a decrease in a-marked elements in general, which is in conformity with the synchronic Portuguese data, as the vast majority of direct objects are unmarked accusatives. Moreover, to address the specific differences regarding PP-ACC and DOM, it is important to analyze the features of the objects, as suggested by López (2016) and Bossong (1991), who consider the domains of inherence and reference when analyzing DOM arguments, as mentioned above. It is not in the scope of this paper, however, to analyze the specific features of the a-marked arguments, such as definiteness, referentiality, and animacy (to name a few features mentioned by Bossong (1991)). We intend to do so in future work.9

As shown in the introduction, argument marking changes depending on the verb class that the predicate belongs to. This first observation guided the research towards the analysis of all a-marked arguments, followed by the analysis of the different verb classes the marked object appeared with. Then, it led to the analysis of psych verb predicates specifically, since the pattern presented in this context differs from the other ones, as already explained. Therefore, alongside the disappearance of PP-ACCs in BP, one of the aims of this research was to understand the context in which the Experiencer argument ceased to be marked in all psych predicates in BP, as exemplified in (3b). In the following section, the properties of psych verbs will be addressed.

Returning to PP-ACC/DOM, historically, overtly marked arguments in Romance appeared with topicalized personal pronouns used anaphorically and emphatically (Döhla 2014, p. 272):

Proto-Romance—11th century
(5)	a	míbe	ṭu	no(n)	queréś.
	Pa(to).	me	you	neg	love
	‘(but) me you don’t love’
	(adapted from Corriente 1997, p. 319 apud Döhla 2014, p. 272)

In Modern Spanish, where, differently from Portuguese, DOM is a core phenomenon, definite animate DPs, for instance, are always associated with DOM. Manzini and Franco (2016) claim that all definite arguments display DOM in Spanish, as these arguments are attached VP-internally via DOM. On the other hand, DOM with indefinite animate arguments depends on the verb it is associated with, as some verbs prefer DOM. The authors argue that some action, psych, and perception predicates, which have non-affected objects, seem to have more transparent v-V structures. These are exactly the classes with more a-marked cases in the corpus analyzed here, which will be discussed in the results section.

In Spanish, the development of the preposition a as an accusative marker used to be seen in the literature as a singular historical event that demanded a specific historical explanation. Bossong (1991) points out, however, that this manifestation is one of the most common ways case marking changes in natural languages. According to the author, if subject and object are no longer distinguishable in the system, morphological marking is then substituted by a new differential system, which may in turn “become non-differential again by the continuous extension of the sphere of positive object marking; at this point of the evolution, the life cycle of case marking may start anew” (Bossong 1991, p. 152), which seems to be the case of the data analyzed in this paper. Therefore, a-marking was a way to overtly case-mark atypical direct objects, but as Portuguese progressed, this marking was no longer necessary, because SVO order became even more common than VOS, for instance, as we will discuss further in this paper.

Specifically, when it is not possible to distinguish between subject and object by formal means, two things may happen. First, such as what happened from Latin to Romance, when morphological marking is eliminated, it may just disappear, as in French, which has replaced morphological marking with structural Case marking (Bossong 1991, p. 146). The same is true for Nominative and Accusative in Portuguese in general, as we see further in this paper. The second option is to resort to what Bossong calls a new grammemic marking, such as a-marking.

According to Döhla (2014), DOM in Modern EP and BP is considered to be a marginal phenomenon. The author claims that DOM occurred in Old Portuguese due to three main reasons: i. parallelism (6)—when one a-marked object is followed by another one; ii. left dislocation—to emphasize the topicalized patient argument (7); and VSO order (8):

Old Portuguese

(6)

(…)

tendes

em

vossa

ajuda

muy

certos

a

mym

e

ao

Conde d’Ourem.

(you) have

in

yours

help

very

sure

Pa(to)

me

and

Pa(to)

.the Count of Ourem

‘you have me and the Count of Ourem to help you.’

(7)	aos	proues	e	mjnguados	sostinha.
	Pa(to).	the poor	and	neglected	support
	‘(he/she) supported the poor and the negleted.’

(8)	amando	mais	as	maes	a	seos	filhos.
	loving	more	the	mothers	Pa(to)	their	children
	‘the mothers loving more their children.’
	(Delille 1970, pp. 36, 39, 42 apud Döhla 2014, pp. 274–75)

As Döhla, Ramos (1992), and Gibrail (2003) also verified, there was a higher occurrence of a-marked arguments with VSO in the Historical Portuguese data they analyzed.10 As mentioned before, there seems to be an overlap between what is understood as DOM and what is understood as PP-ACC. These authors, for example, apply the name PP-ACC to the same examples Döhla refers to as DOM.

The numbers presented by Ramos and Gibrail show an increase in PP-ACCs in the VSO context in the 18th century. The authors claim that object marking in VSO sentences aims to disambiguate the syntactic function of NPs. One hypothesis is that as Portuguese started to present a more fixed SVO order (Galves 2020), object marking ceased to be necessary.

As mentioned before, our intent in this section was to present previous works on a-marked arguments in Historical Portuguese. Since these works address a-marked direct objects both as PP-ACC and DOM, it was important to give an overall view of these phenomena. Given what was presented in this subsection, there is indeed an overlap between the concepts of PP-ACC and DOM, as both strategies are responsible for overtly case-marking direct objects and may be related to the syntactic configuration and features of the verbs (López 2016). So, as specific features of the a-marked NPs in the data analyzed will not be discussed in this paper, we will continue to address a-marking arguments as PP-ACC. We leave the discussion on DOM vs. PP-ACC for future work, otherwise there would be too many variables to be considered here.11

Psych Verbs

In this paper, one of the main focuses is on structures with psych verbs where the Experiencer argument is still marked in Modern EP but unmarked in Modern BP.

In predicates with psych verbs, the Experiencer argument must be a person, hence [+animate] and [+human], who experiences a mental state (Belletti and Rizzi 1988). It can be either the subject (9) or the object (10) of the event. Additionally, the Theme argument, which is the content or the object of the mental state, may also be the subject (11a) or the object (11b). Even though these verbs fall into the same category, Belletti and Rizzi (1988) separate them into the three types, exemplified in (9) to (11), due to their distinct theta grids.

Italian
(9)	Gianni		teme	questo.
	Gianni		fears	this
(10)	Questo		preoccupa	Gianni.
	This		worries	Gianni
(11)	a.	Questo	piace	a	Gianni.
		This	pleases	Pa(to)	Gianni
	b.	A	Gianni	piace	questo.
		Pa(to)	Gianni	pleases	this
		(Belletti and Rizzi 1988, pp. 291–92)

In Portuguese, sentence (12) is equivalent to (9). The examples in (13) show that there are also two possible mappings of the structure of some psych predicates (Carneiro and Naves 2010); (13a) is equivalent to (10), but the Theme argument can also be the content of the subject mental state, as we can see in (13b). In addition, the category of the verb assustar, ‘frighten’, also has an agentive component in Portuguese, since the [+human] subject can act intentionally, as in Maria assustou João, ‘Maria frightened John’ (Naves 2005, p. 122).

(12)	JoãoEXP	teme	a	aranhaTHEME.
	João	fears	the	spider
	‘John fears the spider.’

(13)	a.	A	aranhaAGENT	assusta	o	JoãoEXP.
		The	spider	frightens	the	João
		‘The spider frightens João.’
	b.	O	JoãoEXP	se	assusta	com	a	aranhaTHEME.
		The	João	CL.3SG	frightened	with	the	spider
		‘João gets frightened with the spider.’

Among the many issues related to psych verbs, recall that we are specifically interested in the category of psych verbs such as agradar, ‘please’ (cf. 3)12, for three distinct reasons: firstly, the argument structure of these verbs includes object marking by the preposition a in EP, but it does not in modern BP; secondly, the data from the corpus analyzed showed that this context displays a-marked arguments throughout the four centuries; finally, the constituent order in these predicates is different in EP and BP, as shown specifically in examples (15) and (18):

(14)	O	filme	agradou	ao	João.	(EP)
	The	movie	pleased	Pa(to)	João
(15)	Ao	João	agradou	o	filme.	(EP)
	Pa(to)	João	pleased	the	movie
(16)	O	filme	agradou	-lhe.		(EP)
	The	movie	pleased	-CL.DAT.3SG
(17)	O	filme	agradou	o	João.	(BP)
	The	movie	pleased	the	John
(18)	*O	João	agradou	o	filme.	(BP)
	The	João	pleased	the	movie
(19)	O	filme	agradou	-o.		(BP)
	The	movie	pleased	-CL.ACC.3SG.
(20)	O	filme	agradou	ele.		(BP)
	The	movie	pleased	he.ACC13

Observe that the structures with agradar in EP, with a prepositional-marked Experiencer argument, are the same as with piacere in Italian (cf. 11). In BP, however, the Experience is unmarked in the structures with agradar, similarly to (10) in Italian. Moreover, if we substitute the DP Theme filme, ‘movie’, in (17) for a [+animate] DP, such as João agrada Maria, ‘João pleases Maria’, there is the possibility of an agentive interpretation, which is not possible in (17) with the Theme filme, ‘movie’, but possible in (13a) with assustar, ‘frighten’. Thus, agradar would be part of a fourth class that combines the characteristics of (13a) and (17), which does not exist in Italian. In addition, the argument in BP is pronominalized with accusative elements with the abstract accusative case (cf. 19 and 20), while EP shows a dative counterpart (16).

Having presented some considerations on PP-ACC, DOM, and psych verbs crosslinguistically, in the next section we present the methodology employed to deal with the historical data at hand.

3. Materials and Methods

In this section, the data collected from the Tycho Brahe Corpus is presented. The corpus consists of 88 texts written by Portuguese and Brazilian authors who were born between 1380 and 1978. Out of the 88 texts, 19 texts had already been syntactically annotated at the time of the data collection (today there are 27 texts annotated syntactically), which adds up to a total of 819,932 words (Galves et al. 2017).

The main purpose of analyzing Portuguese historical data was to examine the contexts in which a-marked direct objects (DOs) presented variation between the 16th and 19th centuries. At first, a search was carried out for all occurrences of prepositional accusatives (PP-ACC-marked DOs) in the corpus, in sentences such as (4) (renumbered as (21)), below:

(21)	(…)	chama	a	Pedro	e	André,	e	aos	filhos	do	Zebedeu.
		call	Pa(to)	Pedro	and	André,	and	Pa(to).	the sons	of.the	Zebedeu
	‘(…)	call Pedro and André, and Zebedeu’s sons.’
		17th century (V_004,202.1819)

The relevant information about the texts examined in this paper is in

Table 1. Analyzed Texts.

	Text Code	Author	Date of Birth	Text Title	Number of Words
16th Century	c_007	Diogo do Couto	1542	Décadas (livro de crônicas)	47,605
	g_001	Manuel de Galhegos	1597	Gazeta	28,839
	g_008	Pero Magalhães de Gandavo	1502	História da Província de Santa Cruz	22,944
	p_001	Fernão Mendes Pinto	1510	Perigrinação	47,580
	s_001	Luis de Sousa	1556	A vida de Frei Bertolameu dos Mártires	53,986
17th Century	b_001	André de Barros	1675	Vida do apostólico padre Antonio Vieira	52,055
	b_008	José da Cunha Brocado	1651	Cartas	35,058
	c_002	Maria do Céu	1658	Vida e Morte de Madre Helena da Cruz	27,419
	v_002	Padre Antonio Vieira	1608	Cartas	57,088
	v_004	Padre Antonio Vieira	1608	Sermões	53,855
18th Century	a_001	Matias Aires	1705	Reflexões sobre a vaidade dos homens	56,479
	a_004	Marquesa de Alorna	1750	Cartas	49.900
	c_001	Cavaleiro de Oliveira (Francisco Xavier)	1702	Cartas, Cavaleiro Oliveira	51,234
	c_005	José Daniel Rodrigues da Costa	1757	Entremezes de Cordel	24,252
19th Century	a_003	Marquês de Fronteira e d’Alorna	1802	Memórias do Marquês de Fronteira e d’Alorna	54,588
	b_005	Camilo Castelo Branco	1826	Maria Moisés	20,548
	g_004	Almeida Garret	1799	Teatro	44,252
	o_001	Ramalho Ortigão	1836	Cartas a Emília, Ramalho Ortigão	32,441

:

All authors mentioned above were Portuguese.14 The 19 annotated texts add up to 39,761 sentences distributed as follows: 8930 in the 16th century; 8948 in the 17th century; 10,967 in the 18th century; and 10,916 in the 19th century.

In addition to a large amount of data available in the Tycho Brahe Corpus, the syntactic annotation of the corpus makes it possible to use the Corpus Search computational tool (available on the project website), which allows searching for specific syntactic contexts through queries.15 In the 39,761 sentences from the corpus, 624 sentences exhibited PP-ACCs. The results showed a vast variety of verbs accompanied by an overtly marked accusative argument.

After looking for all PP-ACCs, the data were divided into different verb classes, based on studies on Spanish and Catalan (Pineda 2012, 2017) and Portuguese (Cançado 1996, 2013). Pineda’s work, for instance, focuses on a similar ongoing change in Catalan, but from dative to accusative; these studies were used as reference in order to divide the data collected into different groups, so that it would be possible to run another search for NP-ACCs in the same type of predicates in which the PP-ACCs were found. This division resulted in 11 distinct verb classes: contact; dicendi; psychological; social interaction;16 transfer; transfer related to values; reverse transfer; transfer of knowledge; related to places; quantity; and other types of verbs. See some examples from the 17th century data below:

Contact Verbs
(22)	Para	tirar	toda	a	duvida,	oiçamos	ao	mesmo	Christo.
	To	solve	every	the	doubt	listen.1PL	Pa(to).	the same	Christ
	‘To solve every doubt, let’s listen to the same Christ.’
	(V_004,70.149)

Dicendi Verbs
(23)	mandou	chamar	a	Dom	Duarte	de	Castellobranco	seu	cunhado,
	asked	call	Pa(to)	Dom	Duarte	de	Castellobranco	his	brother-in-law,
	marido	de	sua	segunda	irmã	Dona	Luiza	de	Mendonça.
	husband	of	his	second	sister	Dona	Luiza	de	Mendonça
	‘(someone) sent for Dom Duarte de Castellobranco his brother-in-law,
	husband of his second sister Dona Luiza de Mendonça.’
	(C_002,139.107)

Psych Verbs

(24)

Tambem

lhe era

duro deixar

nos

claustros

adonde

se

criara,

Too

her was hard

to leave

in.the

cloisters

where

CL.3rd

was raised

companheyras,

e

amigas

de

tantos

anos,

e

na

corte

a

uma

irmã

colleagues,

and

friends

of

many

years,

and

in.the

court

Pa(to)

a

sister

a

quem

tanto

amaua.

Pa(to)

whom

so much

loved

‘It also hard for her to leave colleagues and friends of many years in the

cloisters where she was raised, and in the court a sister who she loved so much.’

(C_002, 214.1039)

Verbs related to places

(25)

(…)

a

communicaçaõ

com

Deos;

o

retiro

de

toda

a

creatura,

alguma

doente,

the

communication

with

God;

the

retreat

of

all

the

creature,

some

sick,

a

quem

visitaua

precizada

da

charidade

com

as

mãos

e

olhos

prezos.

Pa(to)

whom

visited

in need

of.the

charity

with

the

hands

and

eyes

fixed.

‘(…)

the communication with God; the retreat of all creatures, some sick,

who (he/she) visited with the hands and fixed eyes.’

(C_002,165.427)

Quantity related verbs
(26)	Resta	nos	o que	a	tudo	excede	a	notícia	das	gentes,
	Remain	us	what	Pa(to)	everything	exceeds	the	news	of.the	peoples,
	que	habitam	uma,	e	outra					margem;
	that	inhabit	one,	and	other					margin
	‘It is left for us everything related to the news about the peoples that live one
	margin and the other.’
	(B_001_PSD, 88.700)

Verbs of Transfer

(27)

(…)

mandou

Dom

Jorge

Mascarenhas,

Marquês

de

Montalvão,

e

Vice-Rei

sent

Dom

Jorge

Mascarenhas,

Marquis

of

Montalvão,

and

Vice-King

daquele

Estado,

a

seu

filho

Dom

Fernando

Mascarenhas

no

ano

de

1641

of.that

State,

Pa(to)

his

son

Dom

Fernando

Mascarenhas

in.the

year

of

1641

‘Dom Jorge Mascarenhas, Marquis of Montalvão, and Vice-King of that

State sent his son Dom Fernando Mascarenhas in the year of 1641.’

(B_001_PSD,19.162)

Verbs related to transfer of values

(28)

Que

esta

Praça

Diuina

assim

como

alumea

aos

peccadores,

premeya

That

this

Square

Diuina

just

as

enlights

Pa(to). the

sinners

awards

aos

Justos,

a

huns

abrindo

os

olhos,

a

outros

enchendo

o

coraçaõ.

Pa(to).the

Just,

Pa(to)

some

opening

the

eyes

Pa(to)

others

fulfilling

the

heart

‘May this Divine Square just as it enlights the sinners, awards the Just, opens

the eyes of some and fulfills the heart of others.’

(C_002,205.928)

Verbs of reverse transfer
(29)	O	Padre	António	Vieira	com	excessivas	expressões		recebeu	nos	braços
	The	Priest	António	Vieira	with	excessive	expressions		received	in.the	arms
	aos	dois	Padres,	como	a	irmãos,	como	a	filhos,	e	como
	Pa(to).the	two	Priests,	as	Pa(to)	brothers,	as	Pa(to)	children,	and	as
	a	heróicos	companheiros	de	sua	glória;
	Pa(to)	heroic	fellows	of	his	glory;
	‘The Priest António Vieira with excessive expressions received two priests in the
	arms, as brothers, as children, and as heroic fellows of his glory.’
	(B_001_PSD,193.1528)

Verbs of transfer of knowledge
(30)	Com	estes	foi	um	Índio	Cristão	antigo,	a	quem	instruíram
	with	these	went	an	Indian	Christian	ancient,	Pa(to)	whom	instruct
	os	Padres,	e	adestraram	na	forma	do	Baptismo,	para	que
	the	Priests,	and	train	in.the	way	of.the	Baptism,	in order to	that
	nos	casos	precisos	os	instruísse,	e	baptizasse;
	in.the	cases	necessary	them	instruct,	and	baptized;
	‘An ancient Christian Indian, who was instructed by the Priests, and trained in
	the ways of the Baptism, went with them, in order to instruct and baptize the
	cases needed.’
	(B_001_PSD,196.1545)

Verbs of social interaction
(31)	Teue	grandissimos	dezejos	da	solidaõ,	e	de	imitar	nella
	There	were enormous	desires	of.the	loneliness,	and	of	imitate	in.her
	aos	antigos	Annacoretas;
	Pa(to)	old	anchorites;
	‘There were enormous loneliness desires, and of imitating the old anchorites.’
	(C_002,206.949)

Other types of verbs

(32)

Que

esta

Praça

Diuina

assim

como

alumea

aos

peccadores,

premeya

That

this

Square

Diuina

just

as

enlights

Pa(to).the

sinners

awards

aos

Justos,

a

huns

abrindo

os

olhos,

a

outros

enchendo

o

coraçaõ.

Pa(to).the

Just,

Pa(to)

some

opening

the

eyes

Pa(to)

others

fulfilling

the

heart

‘May this Divine Square just as it enlights the sinners, awards the Just, opens the

eyes of some and fulfills the heart of others.’

(C_002,205.928)

In the next section, the details of the data collected will be presented, in order to analyze the behavior and the contexts in which PP-ACC cases were found in Historical Portuguese.

4. Results and Discussion

The search conducted in the corpus resulted in a total of 624 sentences with PP-ACCs, distributed throughout the centuries as shown in Figure 1: 17

The data show that PP-ACC occurrences increased in the 17th century; however, there followed an immediate fall in the 18th century (a preliminary analysis of this data was presented in Calindro 2017).

Gibrail (2003), who also analyzed the Tycho Brahe Corpus, attests to a high number of PP-ACCs with the NP ‘God’, as well as the combination guardar a Deus, ‘keep/save God’. In fact, in the data analyzed here, there were 96 examples of this structure in the 17th century, so these examples were excluded from this analysis. Moreover, it seems that such construction has become a crystallized expression even in Modern BP, in which sentences like adorar a Deus, lit. ‘worship ‘to’ God’, are still present in vernacular BP with an a-marked DO; hence, it does not show the variation and change process being addressed here.18

In the data we are concerned with, one explanation for the decrease in the PP-ACC cases in the 18th century may be the change in word order in Portuguese. Galves (2020, p. 21) attests to a rise in the VS order in the 17th century, followed by its fall in the next century. In fact, Ramos (1992) and Gibrail (2003) showed that PP-ACC was more productive in VS contexts.

Therefore, to verify whether word order influenced the occurrence of PP-ACCs in the corpus analyzed in this paper, the data was divided into the following contexts: SVO, VOS. Interestingly, the results showed that the order of the constituents does not seem to be relevant for a-marking the argument, as they presented very similar percentages, especially in the 16th (13% for both VS and SV) and the 17th century (15% for VS and 14% for SV). As mentioned before, we may assume that the accusatives ceased to be marked when the fixed SVO became more common in Portuguese, so we intend to look in more detail at the data collected in the future.

At the moment, two other facts are striking in the corpus analyzed: i. the quantity of PP-ACCs (624) in comparison with NP-ACCs (7756); ii. the increase in a-marked arguments in the 17th century, followed by an immediate fall in the 18th century, and the few cases found in the 19th century as well.

Ramos (1992), Gibrail (2003), and Döhla (2014) argue that Portuguese was greatly influenced by Spanish in the 17th century; therefore, these authors assume that this may explain why there was an increase in PP-ACCs in this particular century. Indeed, DOM is still a very productive phenomenon in Modern Spanish (López 2012, 2016; Fábregas 2013, a.o.). According to Gerards (2020, p. 2): “Spanish is the Romance language with the most advanced (…) DOM system”. Even Gil Vicente, a canonical Portuguese author, wrote plays in Spanish.

However, it does not mean this influence reached all the Portuguese population at that time. Galves (2020) observes that the written variety encountered in the sources available for diachronic empirical research corresponds to the written and spoken language of the dominant classes, who were probably the only ones who had access to literature. This register was presumably quite far from the oral language of the common people, who were mostly illiterate. Actually, the corpus we are dealing with is even further away from popular variety, insofar as it is composed of renowned authors, such as Camilo Castelo Branco and Almeida Garrett, but, unfortunately, this is the data we have available when it comes to Portuguese from the 16th to the 19th centuries.

Inasmuch as Spanish may have played a part in the increase of PP-ACCs in the 17th century, the smaller quantity of a-marked DOs in comparison to unmarked ones strongly suggests that there are language internal issues responsible for this marking.

Since the data was divided into 11 verb classes, as mentioned in the methods section, each context was analyzed separately. The data showed that the quantity of a-marked DOs with psych verbs and social interactional verbs has remained a dominant factor throughout the centuries when compared to all the other nine verb classes combined, as we can see in Figure 2 (see also Calindro 2017). From the 328 overtly marked arguments in the 17th century, 33% (109) are sentences with psych verbs, 17% (57) are sentences with social interaction verbs, and 49% (162) are the other verb classes. Hence, the two former verb classes seem to be more relevant contexts for a-marked arguments to occur in. This is compatible with the fact that in Modern EP the Experiencer argument of some psych verbs is still introduced by the preposition a (cf. 6).19

When we analyze only the context of psych verbs, we notice a steady increase, proportionally, in PP-ACCs throughout the centuries, from 7.25% to 33.3% of the total data, as we can see in

Table 2. PP-ACC/NP-ACC with psych verbs throughout the centuries.

	16th	17th	18th	19th
PP-ACC	7.2% (76)	15.8% (110)	28% (35)	33.3% (3)
NP-ACC	92.3% (975)	84.2% (586)	72% (90)	66.7% (6)
Total	1051	696	125	9

, when we compare pych verbs with PP-ACC and NP-ACC, followed by examples (33) and (34) from the 18th century, which illustrate PP-ACC and NP-ACC associated to psych verbs, respectively:20

(33)	(…)	(ela) poderá muito bem agradar aos espectadores com um bom quadro.
		may very well please Pa(to).the viewers with a good painting.
		‘(she) may very well please the viewers with a good painting.’
		(C_001, 140.1)
(34)	(…)	e os discursos gerais não podem ofender os particulares que são discretos.
		and the speeches public not can offend the private citizens who are discreet
		‘and the public speeches cannot offend the private citizens who are discreet.’
		(C_001, 20.18)

In the context of psych verbs, the Experiencer argument has referential properties typical of subjects. Thus, in order to guarantee its referential properties, a-marking was preserved in EP, while in BP it became unmarked. As psych verbs are two-place predicates, due to economy reasons, the fixed SVO order, attested by Galves (2020), seemed to be enough for the Experiencer to be read as accusative and not nominative in BP, so a-marking was not necessary anymore. We will return to this discussion in the next section.

5. Proposal

The differences in licensing indirect arguments in BP and EP led us to analyze the context in which the objects were traditionally labeled prepositional accusatives in Historical Portuguese. Recall that the starting point of this paper was the differences between Modern BP and EP, regarding a-marked objects in psych predicates (cf. 3). On the surface, a-marking has remained in psych predicates in EP (cf. 3a) and has disappeared in the same context in BP (cf. 3b).

As briefly discussed in the introduction, in Modern EP, the preposition a is a dummy dative marker that enters the derivation to assign dative Case to this element (for more details, cf. Torres Morais 2007; Figueiredo Silva 2007; Calindro 2015, 2016, 2020), similar to Spanish (Cuervo 2003) and other Romance languages. The dative case is confirmed by the possibility of the argument being displaced, as exemplified in (14), and for the alternation between the marked object (a-DP) and the dative clitic lhe. This alternation is no longer part of Modern BP, as the clitic lhe has been substituted by other strategies, as shown in examples (1b), (2b) and (3b) in the introduction (Calindro 2015); thus, the arguments introduced by para receive the structural oblique Case, while the complements of psych verbs are unmarked accusative, i.e., typical direct objects.

The examples in BP, EP, and Italian show crosslinguistic evidence that dative and accusative complements entail a different distribution. According to Manzini and Franco (2016), datives arise in the syntax to reflect a distinct mapping of Participant internal arguments in the event structure when necessary, as the presence of the dative may be due to the activation of a split eventive structure (Svenonius 2002). Manzini and Franco (2016) make an important distinction between Goal datives in ditransitive predicates, as the datives we saw in (1a), and DOM datives. The former are constituents required by the event, while the latter are constituents required by referential properties of the internal argument (similar to Bossong’s and López’s claims addressed previously in Section 2). Thus, based on Svenonius, Manzini, and Franco, DOM datives reflect neither a morphological regularity nor an accident. DOM may be a way to overtly assign case to internal arguments that show referential properties typical of subjects, for instance (cf. Bossong 1991; Gerards 2020).

The data analyzed here showed this may be precisely the case with psych verbs, as the Experiencer argument in these structures remained a-marked over the years when it does not enter the derivation with its inherent case alongside its thematic role. Additionally, its case seems to have shifted from Accusative to Dative. This shift can be explained away using nanosyntax. In this approach, each syntactic–semantic feature is an independent head that projects (Baunaz and Lander 2018, p. 5); thus, cases can be decomposed into more primitive features and hierarchically organized (Caha 2009; Hardarson 2016), as we can see in (35):

(35)
	(Caha 2009, p. 23)

In short, Caha (2009) observes that syncretic morphology in case assignment is not incidental across languages, as exemplified by the syncretism in Russian colored in gray in

Table 3. Syncretism in Russian.

	Table, pl.	Student, sg.	We	Bridge, sg.	Good, f.sg
NOM	STOL-Y	student-∅	my	most-∅	xoroš-aja
ACC	STOL-Y	STUDENT-A	NA-S	most-∅	xoroš-uju
GEN	stol-ov	STUDENT-A	NA-S	most-a	XOROŠ-EJ
PREP	stol-ax	student-e	NA-S	MOST-U	XOROŠ-EJ
DAT	stol-am	student-u	na-m	MOST-U	XOROŠ-EJ
INS	stol-ami	student-om	na-mi	most-om	XOROŠ-EJ

Source: Caha (2009, p. 13).

.

The crosslinguistic data analyzed by Caha (2009) shows a contiguity relation between cases. Nominative pronouns and nominals can be syncretic with accusatives, for instance, but they will not be syncretic with genitive pronouns if the accusative has a different form. In this sense, Hardarson (2016) proposes a slightly different hierarchy of case features from Caha’s (cf. 36), which shows a direct contiguity between accusative and dative (without the interference of the genitive):

(36)	NOM > ACC > DAT > GEN > LOC > ABL/INS>…

According to Hardarson, the dative shares features directly with the accusative, but it is more specific than the accusative, as it is higher in the hierarchy and therefore composed of more features (cf. 36), including, for instance, accusative features.

In the context of psych verbs in EP, it seems that the historical PP-ACC underwent a specialization, which resulted in its accusative features being incorporated into the dative. Consequently, the Experiencer argument, which shows referential properties typical of subjects, displays an a-marked dative case in Modern EP to be mapped as an internal participant in the structure.21

In BP, on the other hand, the PP-ACC argument with psych verbs is an unmarked accusative (cf. 3b). In order to account for this difference between EP and BP, let us return to the comparison of psych verbs in Italian and BP. In the argument structure of psych verbs, when the Experiencer is the complement of V’, the external argument position is available, as we can see in the representation of Italian proposed by Belletti and Rizzi (1988) and adapted by Figueiredo Silva (2007, p. 93) in (37):

(37)

For convenience, we repeat the examples with psych verbs in Italian, EP, and BP below:

Italian
(38)	Gianni		teme	questo.
	Gianni		fears	this
(39)	Questo		preoccupa	Gianni.
	This		worries	Gianni
(40)	a.	Questo	piace	a	Gianni.
		This	pleases	Pa(to)	Gianni.DAT
	b.	A	Gianni	piace	questo.
		Pa(to)	Gianni.DAT	pleases	this
	(Belletti and Rizzi 1988, pp. 291–92)

In (39), the Theme questo is generated in the complement position of V, as well as the Experiencer Gianni, so the external argument position remains available. In (39), the Theme questo moves to receive Nominative Case, the unmarked Experiencer Gianni remains as an internal argument. Example (40b) shows that the marked Experiencer a-Gianni may move to a higher position, probably in the left periphery of the sentence (in Rizzi’s (1997) terms), but not to SpecIP, which is occupied by the Theme questo.22 In EP, the dislocation of the dative to the left periphery is possible as well, since the Experiencer carries the dative Case by being overtly marked (see 42), as we have just discussed.

(41)	O	filme	agradou	ao	João.	(EP)
	The	movie	pleased	Pa(to)	João.DAT
(42)	Ao	João	agradou	o	filme.	(EP)
	Pa(to)	João.DAT	pleased	the	movie
(43)	O	filme	agradou	-lhe.		(EP)
	The	movie	pleased	-CL.DAT.3SG

For BP, Figueiredo Silva (2007, pp. 93–94) argues that internal arguments licensed with psych verbs have inherent accusative case associated with the Experiencer theta role—João in (44) and (45). Subsequently, the Theme argument, o filme, ‘the movie’, in (44) and (45), may move to check its nominative case in SpecIP. Therefore, (45) is ungrammatical because the DP Theme O João already bears inherent ACC and cannot receive NOM in SpecIP.

(44)	O	filme	agradou	o	João.	(BP)
	The	movie	pleased	the	John.ACC
(45)	*O	João	agradou	o	filme.	(BP)
	The	João.ACC	pleased	the	movie
(46)	O	filme	agradou	-o/ele.		(BP)
	The	movie	pleased	-CL.ACC.3SG./ACC

It is remarkable that the only possible way to displace the Experiencer argument in BP to the left periphery would be to double it by a personal pronoun that will fill the accusative position, as in (47)23. It is interesting, though, that this position cannot be occupied by accusative clitics such as o/a, ‘him/her’; it may only be occupied by pronouns such as ele, ‘he’, which used to be only nominative but has undergone a change, and now it may also display accusative case, as we have seen before in (3b).24

(47)	O	João,	o	filme	agradou	ele.
	The	João	the	movie	pleased	he.ACC

Furthermore, historically, BP has become quite resistant to the inversion of constituents (Galves 2020; Martins 2011), which also explains the ungrammaticality of (45). Therefore, we assume that the fixed SVO order is also a key factor on the loss of marked accusative in BP, as the internal argument receives inherent accusative case when entering the derivation in the complement position of V.

6. Conclusions

In this paper, we explored the variation and change of a-marked arguments in Historical Portuguese, with a specific focus on its occurrence in historical texts from the 16th to the 19th centuries. The results of the data analysis demonstrated that the cases of PP-ACCs increased in the 17th century, followed by a decrease in the 18th century. The number of PP-ACC occurrences is much less frequent than of NP-ACCs, though, which has led other authors to treat a-marked arguments as a marginal phenomenon in Portuguese. Some authors deal with this phenomenon using the label PP-ACC, while others adopt the concept of DOM, as this is a widespread phenomenon in other Romance languages, such as Spanish. However, in order to accurately address the differences and/or similarities of PP-ACCs and DOM in Portuguese, it would have been necessary to address features of the objects, such as animacy, definiteness, and referentiality of the a-marked arguments. These important characteristics will be taken into account in future work, as they were not part of the scope of this paper.

Finally, when the verb classes found in the data were examined, it became evident that the set of psych verbs showed a different path from other contexts, as the quantities even increased along the centuries (see Table 2). In EP, there are still a-marked arguments in psych predicates, as opposed to the zero morpheme in BP. The analysis led to the conclusion that, in these cases in EP, the accusative became the structural marked dative, which guarantees the thematic reading of the Experiencer to the internal argument of psych verbs. Meanwhile, in BP, to avoid an ambiguous interpretation, the argument displays inherent accusative case; hence, the Experiencer remains in the accusative position and the Theme rises to receive nominative, so overtly marking the Experiencer is no longer necessary.