Word Order in Colonial Brazilian Portuguese: Initial Findings

Abstract

Some recent studies have posed the hypothesis according to which the grammatical stage that precedes the cultured trend of Brazilian Portuguese is Colonial Brazilian Portuguese, not Classical Portuguese. Therefore, there are still few works systematically comparing these two varieties. This is the goal of the present paper, which focuses on word order. By undertaking a corpus-based study using the same textual genre from the contemporary authors Eusébio de Matos and António Vieira, we have looked for all word order patterns while paying special attention to the X*VS order (with one or more constituents preceding the verb and a postverbal subject), given that it is quite typical of Classical Portuguese as a V2-like grammar, unlike the modern Portuguese grammars. We have observed that, although the colonial text follows the general trends of the classical language, it starts to depart from a V2-like grammar because it shows a higher frequency of non-V2 orders and a preference for informationally marked constructions involving internal positions to the clause. From a parameter hierarchy viewpoint, the main conclusion is that such differences represent frequency divergences which are consistent with nano- or microparametric changes which took place later.

1. Introduction

With the development of research on the history of Brazilian Portuguese and the availability of new data sources, not only written by Portuguese citizens, but also by natives, some investigations are testing the possibility that the differentiation between Brazilian Portuguese and European Portuguese (respectively referred from now on as BP and EP to allude to their modern varieties) started earlier than previously expected. This hypothesis considers that even Colonial Brazilian Portuguese (from now on, ColBP)—the term we use to refer to the variety spoken by the descendants of the Portuguese colonizers—showed some spectrum of variation, which should have been even stronger in other population segments, supposedly due to more robust language contact.

The present work therefore puts into effect a research agenda on ColBP and applies it to a specific case study as follows: the description and analysis of word order in a text written by a Brazilian author from the 17th century, with special attention to the X*VS pattern, in which there are one or more preverbal constituents, the verb, and the subject. This choice is significant for the study of grammars displaying V-to-C movement (V2 grammars) because they show more frequent occurrences of X*VS cases, together with a more varied distribution of informational-structural values related to preverbal constituents.1 Our main research question is whether there are differences regarding the frequency of constructions involving a preverbal constituent between the grammar of ColPB and the grammar of Classical Portuguese (from now on, ClP).

To achieve this end, in this first stage, we intend to compare data with fronted constituents in main clauses from the following two sermons: Ecce Homo,2 written by Eusébio de Matos, a printed original document of 25,090 words edited by Ilma Alkimim (cf. Alkimim 2014) and the Sermons3 by Father António Vieira, a printed original document of 53,855 words edited by Father Gonçalo Alves (cf. Vieira 1907).

The reason for these choices is, in summary, due to the high degree of similarity between these two texts, which represent the same textual genre, and their authors, who had a similar background. Our interpretation is that these multiple similarities are welcome because they suggest that any relevant differences between word order in these texts should be related to the inception of new grammars in variation, especially if they are consistent.

The theoretical framework adopted for this research is generative diachronic syntax, according to which syntactic changes are triggered by mismatches in parameter setting in the human mind, arising from linguistic experience (Lightfoot 1999). Two complementary theories are adopted as follows: the concept of parameter hierarchies, which stems from the notion of syntactic microparameters (Roberts 2019)4, and the interface between Information Structure and Syntax mapped into projections at the left periphery of the clause (cf. Rizzi 1997).

This text is organized as follows. Section 2 discusses the main features of ColBP and its relevance for a more detailed understanding of the history of BP and beyond. Section 3 presents background information on word order patterns in ClP. Section 4 details the materials and methods followed in our corpus-based investigation. Section 5 presents the comparative results, both at a general level—involving a comparison of the various word order patterns in main, complement and adjunct clauses—and at a more specific level, regarding the possibilities involved in X*VS order. In Section 6, we include some qualitative observations related to the results. To conclude, Section 7 presents the final remarks.

2. Colonial Brazilian Portuguese: Characterization and Relevance

The divergence between the main modern varieties of Portuguese (BP and EP) is widely recognized and raises questions about the moment of its inception, its motivation, and development. The predominant hypothesis indicates that BP emerged during the Portuguese colonial period (1532–1822), specifically in the mid-18th century (Tarallo 1993; Mattos e Silva 2004; Galves 2007). This assumption was based on sociohistorical factors known to potentially accelerate linguistic changes, such as intense multilingualism, a high degree of miscegenation, and low access to formal education.

The intense linguistic contacts that characterized the colonial period represent a primordial factor for the hypothesis that in Colonial Brazil, especially during the 18th century, an authentic grammar was being developed. During Portuguese colonization in Brazilian territory, the settlers had to deal with a complex linguistic situation: numerous indigenous languages, which were present in the colony since the pre-colonial period; various African languages taken to its territory with enslavement; as well as some regional pidginized languages brought about by the contact with Portuguese. The various sociolinguistic scenarios were surely facilitated by the fact that, although the Portuguese immigrants comprised a demographic minority, they held a position of social supremacy, which allowed them to maintain frequent interactions with African (and, in many cases, also with indigenous) peoples, resulting in a mixed population with several social layers.5

These circumstances lasted until the beginning of the 18th century and represented a disadvantage for the use of the Portuguese language, which was often supplanted in favor of more ‘accessible’ languages aimed at communicating and catechizing non-European peoples, known as Indigenous General languages. Being widely adopted until the mid-1750s, inclusively by the Portuguese colonizers and their descendants, these creolized varieties of Tupian languages fell into demise because of a series of coordinated measures involved in linguistic imposition, which favored the Portuguese language.

Due to this process, scholars such as Mattos e Silva (2004) and Galves (2007) consider the 18th century as a crucial period for the emergence of BP grammar; with the adoption of Portuguese as a language of mandatory and universal use in Brazil, its acquisition occurred through an accelerated and imperfect process by indigenous peoples and by Africans living in Brazil and their descendants, as well as through a natural process by white individuals born in Brazil, culminating in the configuration of a grammatical system quite different from Portuguese grammar, which served as the norm during the colonization of Brazil.

This sociohistorical reality is aligned with grammatical analyses carried out based on texts written during the colonial period in Brazil. Tarallo (1993), who investigated a range of phenomena (the pronominal system in relation to subject and object positions, relativization strategies, and word order in both declarative sentences and direct and indirect questions) highlighted the following: “O português do Brasil existe como língua literária somente a partir dos anos 1700. Qualquer material anterior àquela data revelaria, pois, traços do português europeu e enviesaria os dados” [Brazilian Portuguese has existed as a literary language only since the 1700s. Any material prior to that date would therefore reveal relics of European Portuguese and would produce data bias] (Tarallo 1993, p. 65). Therefore, there are grounds to believe that, during the 18th century, two grammars were already interacting in Brazil: ClP, ‘transferred’ by the mouths of the colonizers since the 1500s, and a preliminary version of BP.6 Retracting the identification of Brazilian Portuguese prior to 1700 has not been deeply explored from the grammatical point of view until now but is compatible with previous studies. Barbosa (1999, p. 14), one of the first descriptions of the Portuguese language written in Brazil during the colonial period, already recognizes the necessity of distinguishing what he calls ‘Portuguese in Brazil’ from ‘Portuguese of Brazil’. In the same line, Ramos and Oliveira (2021, p. 19) state that the Portuguese colonizers fostered the creation of a new dialect in Brazil already at their third generation on the new soil.

With regard to word order in declarative sentences, Tarallo (1993) highlights that the decrease in the use of the verb-subject order is more striking between the end of the 19th century and the beginning of the 20th century. His data, based on the original results in Berlinck (1988), are presented in

Table 1. Frequency of use of verb-subject (VS) order in the history of Brazilian Portuguese.

Period	Percent	Occurrences: VS/Total
18th century (1750)	42%	203/486
19th century (1850)	31%	144/469
20th century (1987)	21%	63/1262

Source: Tarallo (1993, p. 71), translated with adaptations.

.

Before the 19th century, according to the author, there was only an “imitation” of the grammatical pattern of the Portuguese language in Portugal, in which the Brazilian data did not show any grammatical innovations. Other studies have addressed the 18th century in their corpora, but there has not yet been enough data to definitively identify the point of emergence of BP (Pagotto 1992; Corôa 2022).

From this perspective, the main approach to understanding the process of grammatical change in the development of BP has been a comparative analysis that contrasts the Portuguese language from Portugal in the 17th and 18th centuries with Brazilian texts written from the 18th century onwards. The colonial period is seen, therefore, as a transition between two grammars, whose linguistic research is limited to a specific time frame (18th century) to identify syntactic patterns that are close to standard BP. According to this traditional view, the period during which ClP and BP were used partially overlaps, as shown in Figure 1, adapted from Mattos e Silva (2004) and Galves (2007).

Figure 1 represents the traditional hypothesis for the emergence of BP, which already points to the relevance of the colonial period in this process. The vertical lines indicate the probable period of emergence of Portuguese grammars in Colonial Brazil. The horizontal lines indicate the probable maintenance and use of two grammars in the Brazilian territory: ClP and BP. Pagotto (1992), Carneiro (2005), Martins (2009), and Carneiro and Galves (2010), who analyzed syntactic aspects related to the order of clitic pronouns in Brazilian 19th-century texts, propose that ClP remained in use in the Brazilian territory until the mid-19th century. Nevertheless, it has mingled with BP since the 18th century according to Tarallo (1993), Galves (2007) and Mattos e Silva (2004). According to this scenario, there is no need to assume a different grammar in Colonial Brazil.

However, we are taking another stance regarding the study of Portuguese in the colonial period. Following Andrade (2020), Cardoso (2020), and Cardoso et al. (2023), we highlight the relevance of deepening the syntactic investigation into the Portuguese language in Colonial Brazil since the 17th century, based on a comparative analysis between texts written by Portuguese authors and by individuals born in Brazil during this period. This methodological position is based on the hypothesis that a grammatical variety existed between ClP and BP, which we call ColBP. This variety would have been acquired by the descendants of Portuguese citizens born in Brazil in the 17th century. Figure 2 depicts this hypothesis, in accordance with Cardoso et al. (2023).

The sociohistorical and linguistic evidence that supports such a perspective has been presented in Cardoso et al. (2023). A key aspect of the sociohistorical evidence in favor of the existence of ColBP consists in the assumption that linguistic contact had been present in Brazil since earlier periods, although to a lesser degree if compared to the part of the population acquiring Portuguese as a second language. This is the null hypothesis, as there had already been contact between indigenous, European, and African peoples and their descendants since the beginning of the colonization process, also with consequences among the European descendants. Historians who have investigated Brazil’s colonial period have highlighted the noteworthy contact between children of Portuguese and African people during childhood (Mattoso [1979] 2003), a period in which language acquisition occurs (Lightfoot 1999). Therefore, we did not identify a similar linguistic experience to that of Portugal that would support the assumption that ClP was the only variety grammatically acquired and spoken by white individuals during the first 100 years of Portuguese colonization, devoid of any process of change in its grammatical components, as assumed in the hypotheses formulated in Mattos e Silva (2004) and Galves (2007).

Therefore, we recognize the existence of ColBP as a linguistic variant of Portuguese acquired by the descendants of the Portuguese born in Brazil from the 17th century onwards, emerging after the introduction of ClP in Brazil and preceding the formation of ‘cultured’ BP (i.e., the variant of BP spoken by the most prestigious class of the population).7 However, although ColBP presents some grammatical differences in relation to the ClP grammar written in the 17th century, we consider that these changes in the grammatical system were less significant if compared to the changes observed in BP. This is mainly due to the general presence of regular linguistic transmission among the likely speakers of this grammatical variety (cf. Lucchesi 1994).

Due to the recent introduction of this proposal, there is yet little linguistic evidence favoring the existence of ColBP. Nevertheless, a comparative study on the placement of clitic pronouns, both in texts written by Portuguese descendants born in Brazil between the 17th and 18th centuries and in texts written by Portuguese citizens living in Brazil during the same time period (Cardoso 2020) has already revealed an important difference: the former data display more frequent proclisis in Variation Context 2, whereas the latter group showed more frequent enclisis, as shown in Figure 3.8

The difference illustrated in Figure 3 is not taken as innovative because proclisis is also possible in the ClP grammatical system, although it is not the predominant form in the relevant context. Nevertheless, the prevalence of proclisis in ColBP texts stands out in relation to the tendency towards enclisis in ClP texts, highlighting a distinctive feature of these colonial texts that may have led to a grammatical change at a later date. Interestingly, these findings have been confirmed by Corôa (2022), who examined a different set of texts written by Portuguese descendants born in Brazil in the 17th century. A marked preference for a proclitic placement was observed in sentences, including a main verb preceded by a subordinate clause.

The evidence highlighted above, which relate sociohistorical and grammatical aspects, requires the pursuit of the following new research directions that depart from the mainstream practices followed to date:

• Expanding the scope of investigation into the history of the Portuguese language in Brazil to include 17th-century texts;
• Carrying out comparative research between ClP and ColBP, without necessarily assimilating the latter to linguistic phenomena associated with BP.

The first measure involves the investigation of the grammar present in texts written by Portuguese descendants born in Brazil in the 17th century. Unlike the previous proposal on the history of BP (Tarallo 1993), we assume that these materials will not just reveal an “imitation” of the grammar brought by the Portuguese in the previous century. Instead, the study of 17th-century texts shall contribute to the identification of ColBP and the investigation of possible signs of the emergence of a new grammatical variety. The importance of this aspect, however, brings about other challenges, such as the rarity of documents written during this period and the difficulty in identifying their authors. These tasks demand interdisciplinary efforts that involve cooperation among experts in the areas of philology and paleography (Lose 2022).

The second measure involves comparative research between ClP and ColBP, with the aim of identifying the characteristics of the latter and valuing its particularities. Searching for features of BP in the data may limit our ability to discern the grammar of ColBP, considering that we treat these as different varieties. Therefore, we propose a comparative analysis focusing on grammatical phenomena known for their diachronic instability, susceptibility to idiosyncratic patterns, and parametric changes over short periods of time, so that ColBP may be studied as a variety to be described in its own right.

These phenomena tend to correspond to lower-scale parameters (micro- and nanoparameters), as defined by Roberts (2019), which may be illustrated by various grammatical rules, such as clitic placement and the expression of null subjects.9 Alternatively, only the frequency of some grammatical options may be affected, a process that clears the way for a parametric change to take place much later. The goal of this comparison is to identify linguistic patterns that, even on a small scale, diverge from ClP, thus providing arguments from internal history that help to identify the nature of ColBP.

Although we privilege a language-internal approach for linguistic change, we do not overlook external motivations that may play a role in the development of ColBP. In fact, it is probable that contact was the main source of innovations in this variety, even though it was much less widespread when compared to General Brazilian Portuguese. Even so, there are elements that suggest ColBP was also related to language shifts, as most of its speakers were not of pure European origin, i.e., they also expressed some degree of miscegenation, which could lead to interference, in which the aspects most susceptible to change are the lexicon and pragmatics. From the side of the Portuguese colonizers, we posit the existence of foreigner talk, typical of situations in which native speakers see learners of the dominant language as socially inferior. Both trends point to a process of language simplification, which may include, in the case of foreigner talk, the avoidance of contractions, shorter and less complex sentences, the retention of optional constituents, and the explicit marking of semantic relations, among other consequences (cf. Matras 2009).

3. Word Order Patterns in Classical Portuguese

The description of word order in ClP has been the origin of much debate regarding the parameter regulating verb movement. Some authors pursue the hypothesis according to which the verb was placed in I since Old Portuguese (cf. Martins 2019). Others have put forward that there was a change in verb position in the clause, from a larger movement (until C) to a narrower movement (until I) (cf. Galves and Kroch 2016). Here, we follow the position that, to some degree, the two proposals can describe most problems, but the latter more clearly identifies the change between older and modern stages of Portuguese. This being so, let us observe the main characteristics of ClP grammar regarding word order.

According to Galves (2020), who adopts Wolfe’s (2015) notion of ‘relaxed V2 grammar’, in ClP, “linear V2 is the dominant order, the pre-verbal field is not specialized for subjects, and subjects very frequently appear in post-verbal position” (p. 384). Unlike in strict V2 languages, V1 is a frequent order and V3 (and even V4) may also occur. According to Galves’ corpus-based research on ClP, there were 59% of V2 cases, among which 27% represent SV occurrences and 73% XV occurrences, in main declarative clauses with transitive verbs. This is consistent with a V2 type of grammar, in which not only is V2 frequent, but there is also a great predominance of non-SV sentences among the cases of V2.

Among the cases of V1, the following two different situations must be considered: absolute V1 and V1 preceded by a conjunction. The first situation may be derived by the presence of a null subject (i.e., rhematic or presentational V1) or by the presence of a postverbal subject, a situation in which there is topic continuity marked by a postverbal subject. Therefore, V1 is seen as totally compatible with a V2 type of grammar. Comparing V2 with V1, the latter represents 32% of the occurrences.

The occurrences of V > 2 are represented by various types of preverbal constituents as follows: fronted or left-dislocated objects, adjuncts represented by simple phrases or by dependent clauses (especially adverbial clauses), subjects and intercalated phrases (parentheticals). To explain their derivation, Galves (2020) follows Benincà’s (1995) proposal, according to which only fronted constituents count for V2. Nevertheless, higher positions in the phrase marker may be expressed by base-generated constituents. The proposed representation for the left periphery, considering a cartographic approach for syntax, is shown in (1) (from Galves 2020, p. 391):

(1)

[ForceP

[TopP* (Topic)

||

[FocP (Focus)

[KontrP (Kontr)

[FinP

[TP ]]]]]]

In this structure, the doubled vertical line separates the external domain, in which (left-dislocated or strongly contrastive) topics are base-generated from the internal domain, in which quantified foci and contrastive constituents occur.

Other questions relate to the differences between ClP and the other stages of the history of Portuguese in Europe. Galves and Kroch (2016) present some of the most important parameters distinguishing these stages, which may be summarized as follows:

• OldP > ClP: Multiple AgrS specifiers (yes > no)
• ClP > EP: C attracts V (yes > no)

Accompanying the latter shift, they also assume a change in the domain governing the restriction against clitic-first (changing from a prosodic to a morphosyntactic restriction). Results from recent works suggest, however, that word order changes between OldP and ClP are still to be further explored and may be better explained as microparameters operating at the left periphery (cf. Andrade and Galves 2019, p. 22, fn. 26), more specifically, those related to stylistic fronting (understood as vP-movement to Spec, FinP) and to clitic movement to Force (where the complementizer is positioned).

4. Materials and Methods

4.1. Materials

To carry out the research agenda described in Section 2, rigorous observational control and a detailed analysis are necessary to identify peculiar grammatical patterns between the two grammars under analysis. However, there are some challenges to face, primarily the scarcity of texts that meet our minimal prerequisites. Considering the social aspects of the time, such as limited access to schooling, few individuals born in the 17th century had the opportunity to learn to read and write. Furthermore, the time factor presents itself as a problem for studies of past synchronies as not every text was considered relevant to the point of being preserved for centuries, a fact that restricts the material available for analysis. This is reflected by the fact that the archives containing documents from the colonial period, for the most part, contain a set of manuscripts of administrative and/or literary content, preserved according to the relevant criteria of the time in which these materials were produced.10

The difficulty involved in the selection of the sources lies not only in the localization of the manuscripts, but is also related to the restricted types of the textual genres preserved.11 According to Kabatek (2008), issues such as the purpose of communication and the linguistic-discursive formulas present in certain genres can influence the grammar displayed in the written text, making it confusing to identify the grammar of the individual who wrote it and what corresponds to the writing traditions of the period. To alleviate such problems, we selected for this comparative research two manuscripts of the same textual genre, produced at the same time by contemporary individuals. Relevant information regarding the selected materials is summarized in

Table 2. Corpus selected for the linguistic analysis.

	Colonial Brazilian Portuguese	Classical Portuguese
Author	Eusébio de Matos (Brazil, 1629–1692)	António Vieira (Portugal, 1608–1697)
Textual Genre	Sermon (Ecce Homo)	Sermon (Sermões)
Edition	Conservative edition by I. Alkimim	Conservative edition by G. Alves
Number of words	25,090 words	53,855 words
Source	CE-DOHS	TBC

.

For the analysis of ColBP word order, we chose to analyze Ecce Homo, a collection of sermons delivered at the Jesuit College of Salvador da Bahia during the 17th century, whose first printed version was published in 1677 in Lisbon. Its digital version was made available on the website of Biblioteca Brasiliana Guita and José Mindlin.12 However, for this study, we chose the philological edition produced by Alkimim (2014), who used the first edition of the text as the basis to develop a critical edition. Although the author added critical and clarifying comments to the text, considering its context of production and reception, the edition was considered conservative as there were no orthographic, syntactic, or punctuation updates involving distinctive linguistic value (Alkimim 2014, p. 44).13 This material is part of a set of manuscripts selected to compose the colonial corpus of the Electronic Corpus of Historical Documents of the Hinterland (CE-DOHS, cf. Carneiro and Lacerda 2024).14

As a representative of 17th-century ClP, we have selected the sermons written by Father António Vieira. Although the first edition of this text was published in Lisbon between 1679 and 1695, for this work, we used the modernized edition published in the year 1907. This edition preserves aspects related to the punctuation and morphosyntax of the original text, and is made available in a parsed version at the Tycho Brahe Corpus of Historical Portuguese (TBC, cf. Galves et al. 2017).15

We observe that the materials chosen for our research corpus present notable similarities. The two texts, published in Lisbon in a similar period, i.e., between the 1770s and 1790s, have significant literary content and belong to the same textual genre. The selection of this corpus was not random; given the number of similarities, it becomes more difficult to attribute grammatical divergences to extralinguistic factors, such as the place and year of publication, or to the influence of discourse traditions used in that period.

Additionally, there are similarities between the authors, contributing to the need for rigorous control for our investigation. As highlighted by Alkimim (2014), both authors maintained a social relationship and were aware of each other’s work. Eusébio de Matos, born in Bahia in 1629, was the brother of the renowned poet Gregório de Matos and studied at the Jesuit College (Colégio da Bahia), where he later served as Professor of Theology and Philosophy. Father Antonio Vieira was born in Lisbon and arrived in Bahia at the age of six, but returned to Portugal when Eusébio de Matos was 11 years old, and did not return to Brazil until the 1680s. Although they did not live together, their work was mutually recognized, with Vieira making some public comments about the Brazilian preacher. According to Machado (1741), Vieira said the following about Matos: “Deus se apostara em o fazer em tudo grande, e não fora mais por não querer” [God had committed to making him great in everything, and he did not become even greater because he did not want to] (p. 766).

A third aspect regarding the similarities between the two texts consists in the authors’ literary style. According to Alkimim (2014), literary critics proposed that Eusébio de Matos was an “imitator” of Vieira, since his sermons carry a theme and style that mirror those of the famous Portuguese sermonist. A relevant example consists in the mention of the speed of the sun, which is found in both texts, as demonstrated below (examples from Alkimim 2014, p. 294 and Vieira 1907, p. 25, respectively):

(2)

a.

Ora

notem:

o

curso

do

Sol

material

now

notice

the

course

of.the

Sun

material

[…]

corre

em

cada

hora

runs

in

each

hour

trezentas

e

oitenta

mil

légoas

(ColBP)

three.hundred

and

eighty

thousand

leagues

‘Now notice: the course of the material Sun […] runs three hundred and eighty thousand leagues every hour.’

b.

O

Sol

[…]

corre

a

cada

hora

the

Sun

runs

in

each

hour

trezentas

e

oitenta

mil

légoas

three.hundred

and

eighty

thousand

leagues

(ClP)

‘The Sun […] runs three hundred and eighty thousand leagues every hour.’

Not only were the themes similar, but also certain stylistic structures present in Eusébio’s texts resemble Vieira’s style. Among these, Alkimim (2014, p. 100) highlights the use of the expression que muito, which means “why is it a wonder” or “how strange it is that”. This expression appears in both Eusébio’s and Vieira’s texts, with a greater frequency than with other literary authors of the same period (examples from Alkimim 2014, p. 100 and Vieira 1907, p. 27, respectively):

(3)

a.

que

muito

que

diga

eu,

que

how

much

that

say

I

that

com

aquela

Capa

está

Cristo

cubrindo

with

that

cloak

is

Christ

covering

nossas

culpas.

(ColBP)

our

faults

‘why should it be a wonder that I say Christ is covering our faults with that cloak?’

b.

que

muito

é

que

digam

e

informem

how

much

is

that

say

and

inform.3pl

[…]

que

lhes

sobejam

merecimentos

that

to.them

remain

merits

(ClP)

‘why should it be any wonder that they say and testify […] that their merits reach to the rooftops?’

Therefore, the only fundamental difference between the texts lies in the fact that they were written by individuals born in different places, who acquired different grammars, with other points of divergence mitigated due to the similarities highlighted above.

4.2. Methods

Since the selected texts are available in different formats, data retrieval involved different steps as follows: for ClP, we could run automated queries using the CorpusSearch 2 software (Randall et al. 2004), and for ColBP, we carried out manual searches for the desired word order combinations.

We studied word order in two stages. The first was more general and encompassed an investigation into the behavior of subject expression (null subjects, subject-verb order, and verb-subject order) and verb position (V1, V2, and V > 2) in three types of clauses: main clauses, complement clauses, and adjunct clauses. In the second stage, we selected only main clauses with X*VS order (i.e., clauses with a postverbal subject with one or more preverbal constituents) and classified them according to the following syntactic and informational aspects:

• clause type: main clause, complement clause, or adjunct clause;
• syntagmatic type of the fronted constituent: nominal phrase; prepositional phrase; adverbial phrase;
• syntactic function of the fronted constituent: subject, accusative complement, dative complement, oblique complement, adjunct, nominal predicate, dislocated topic, or parenthetical;
• discursive or informational function: discourse marker, vocative, complementizer, focalizing particle, V2 topic, contrastive topic, frame-setting topic, contrastive focus, or informational focus.16

The results underwent statistical analysis to verify their significance. Either a chi-square test or a G-test was applied for each table (with partitionings if required).17 Significant p-values are marked in bold. Residual analyses have been applied for significant tables, with the respective results shown in the footnotes.

In the next section, we present the quantitative results organized according to these descriptive levels and, in the subsequent section, we discuss the qualitative aspects observed from the data.

5. Results

5.1. General Word Order Patterns

In this subsection, we present the general results for word order in the three clause types studied. The results for main clauses in the two varieties are shown in

Table 3. Comparison of word order patterns in the main clauses of the corpus.

Variety	Verb Position/ Subject Expression	Verb-Subject	Subject-Verb	Null Subject
Colonial Brazilian Portuguese	V1	71 (49%)	0 (0%)	127 (53%)
	V2	58 (40%)	68 (61%)	92 (38%)
	V > 2	17 (11%)	43 (39%)	21 (9%)
	Total	146 (100%)	111 (100%)	240 (100%)
Classical Portuguese	V1	97 (42%)	0 (0%)	74 (49%)
	V2	122 (53%)	86 (61%)	64 (43%)
	V > 2	12 (5%)	56 (39%)	12 (8%)
	Total	231 (100%)	142 (100%)	150 (100%)
p-value		0.0015 *	0.9965 †	0.6960 †

* Chi-square test for independent samples. ^† G-test for independent samples.

.

From the results in Table 3, we observe that there are some differences regarding the preferences between V1, V2, and V > 2 positions, which are significant only in the verb-subject order.18 Unlike in ClP, ColBP shows a comparatively higher percentage of V1 with null subjects and more cases of V1 than of V2 in clauses with verb-subject order:

(4)

a.

Temos

o

exemplo

entre

mãos

have.1pl

the

example

between

hands

(V1; null S)

‘We have the example between (our) hands’

b.

Quiseram

os

antigos

pintar

a

justiça

wanted

the

ancient.ones

paint

the

justice

mais

rigurosa

(V1; VS)

more

strict

‘The ancient ones wanted to represent Justice as (being) more strict’

In ClP, V2 sentences are more frequent, equally with a preverbal or with a postverbal subject:

(5)

a.

[NP-S

as

outras

prophecias]

cumprem-se

the

other

prophecies

fulfill.3pl=inh

a

seu

tempo

(ClP, SV)

in

their

time

‘The other prophecies fulfill themselves in their (due) time’

b.

[PP

Na

parábola

das

dez

Virgens],

fallava

in.the

parable

of.the

ten

virgins

spoke.imperf

[NP-S

Christo

Senhor

nosso],

propria

e

Christ

Lord

our

proper

and

literalmente

do

dia

do

Juiso

(ClP, VS)

literally

of.the

day

of.the

Judgment

‘In the Parable of the Ten Virgins, Christ our Lord spoke, properly and literally, about the day of Judgment’

The results for complement clauses in the two varieties are shown in

Table 4. Comparison of word order patterns in the complement clauses of the corpus.

Variety	Verb Position/ Subject Expression	Verb-Subject	Subject-Verb	Null Subject
Colonial Brazilian Portuguese	V1	14 (63%)	0 (0%)	42 (65%)
	V2	7 (32%)	20 (74%)	19 (29%)
	V > 2	1 (5%)	7 (26%)	4 (6%)
	Total	22 (100%)	27 (100%)	65 (100%)
Classical Portuguese	V1	6 (38%)	0 (0%)	31 (80%)
	V2	10 (62%)	9 (75%)	5 (12%)
	V > 2	0 (0%)	3 (25%)	3 (8%)
	Total	16 (100%)	12 (100%)	39 (100%)
p-value		0.2238 †	0.9931 †	0.1532 †

^† G-test for independent samples.

.

Table 4 reveals that the frequency of V1 and V2 are similar between the two varieties, except with postverbal subjects; ColBP shows V1 more frequently than V2 in this context. However, this result, as well as the one found in the other columns, is not significant.

The results for adjunct clauses in the two varieties are shown in

Table 5. Comparison of word order patterns in the adjunct clauses of the corpus.

Variety	Verb Position/ Subject Expression	Verb-Subject	Subject-Verb	Null Subject
Colonial Brazilian Portuguese	V1	20 (49%)	0 (0%)	91 (58%)
	V2	17 (41%)	56 (79%)	62 (39%)
	V > 2	4 (10%)	15 (21%)	5 (3%)
	Total	41 (100%)	71 (100%)	158 (100%)
Classical Portuguese	V1	8 (40%)	0 (0%)	8 (42%)
	V2	11 (55%)	1 (20%)	4 (21%)
	V > 2	1 (5%)	4 (80%)	7 (37%)
	Total	20 (100%)	5 (100%)	19 (100%)
p-value		0.5839 †	0.5982 †	0.0002 †

^† G-test for independent samples.

.

Table 5 shows that there are similarities between adjunct clauses and other clause types previously discussed. Nevertheless, the only significant column was that corresponding to null subject clauses.19 In ColBP, there is a preference for less complex structures, with the hierarchy V1 > V2 > V > 2, as the examples below suggest (all showing a null subject and a porque ‘because’ connector):

(6)

a.

porque

lhe

acomodam

bem

as

penas

(ColBP; V1)

because

3sg.dat

accommodate

well

the

feathers

‘because the feathers accommodate well to it [love].’

b.

porque

[com

elas]

nos

explica

because

with

them

us

explains

o

quanto

por

nós

padeceu.

(ColBP; V2)

the

much

for

us

suffered

‘because with them [the wounds] he explains to us how much he suffered for us.’

c.

porque

[como

o

cair

é

pensão]

[…] [em

because

as

the

falling

is

burden

in

nossa

própria

fraqueza]

temos

algũa

desculpa

(ColBP; V3)

our

own

weakness

have.1pl

some

excuse

‘because, as falling is a burden, we have some excuse for our own weakness.’

In ClP sentences with a null subject, there are more V > 2 structures, comparatively:

(7)

a.

porque

[aos

pequenos]

concede-se-ha

misericordia

(ClP; V2)

because

to.the

little.ones

grant=pass=fut

mercy

‘because mercy will be given to the little ones.’

b.

porque

[conforme

cada

um

tem

o

coração],

because

according.to

each

one

has

the

heart

[assim]

prophetisa.

(ClP; V3)

so

prophesies

‘because according to (where) each one has their heart, so they prophesy.’

5.2. Preverbal Constituents with Postverbal Subjects in Main Clauses

In this subsection, we present the detailed results for X*VS in main clauses, exploring different aspects of this topic. The most basic aspect is related to general combinations of preverbal elements, as presented in

Table 6. Comparison of word order combinations in the X*VS clauses of the corpus.

Word Order Combinations	Colonial Brazilian Portuguese	Classical Portuguese
OV	8 (21%)	24 (15%)
AV	22 (60%)	118 (74%)
AAV	5 (14%)	18 (11%)
AOV	2 (5%)	0 (0%)
Total	37 (100%)	160 (100%)
p-value = 0.4904	(G-test for independent samples)

, in which the type and number of preverbal constituents is considered (O represents any type of object complement and A stands for any type of adjunct).20

In ColBP, there are more preverbal objects and preverbal adjuncts in isolation and more combinations of preverbal phrases (i.e., AAV and AOV) in comparison to ClP. Although the results for this table are not significant, it is still noticeable that (i) ColBP shows much fewer cases of X*VS data than ClP (37 versus 160 occurrences) and that (ii) ColBP shows two cases of AOV, whereas ClP displays none, despite the significantly greater amount of relevant data.

Table 7. Comparison of types of preverbal constituents in the X*VS clauses of the corpus.

Type of Preverbal Constituent	Colonial Brazilian Portuguese	Classical Portuguese
DP accusative complement	10 (10%)	5 (3.6%)
DP dislocated topic	0 (0%)	13 (9.3%)
DP subject	0 (0%)	1 (0.7%)
PP dative complement	2 (2%)	5 (3.6%)
PP adjunct	36 (39%)	64 (45.7%)
AdvP or Adverbial Clause	43 (46%)	50 (35.7%)
Vocative	1 (1%)	0 (0%)
Parenthetical	0 (0%)	2 (1.4%)
Nominal predicate	2 (2%)	0 (0%)
Total	94 (100%)	140 (100%)
p-value = 0.0069	(G-test for independent samples)

summarizes the types of preverbal elements, in which a more detailed combination of syntagmatic and functional information on preverbal constituents is offered. The total numbers are different if compared to Table 6 because a constituent may iterate and because some of them were not included before, noticeably those in bold in the left corner.

This table shows significant results, unlike the previous one. This suggests that the relevant factor groups are those capturing either more precise syntactic or informational-structural values.21 In relation to ClP, ColBP has more fronted accusative complements, and it shows no dislocated topics (either accusative or dative complements with a resumptive pronoun). This leads us to the next and last table.

Table 8. Comparison of types of marked constructions found in the X*VS clauses of the corpus.

Informationally Marked Construction	Colonial Brazilian Portuguese	Classical Portuguese
Clitic Left Dislocation	0 (0%)	6 (5%)
Hanging-Topic Left Dislocation	0 (0%)	1 (0.5%)
Contrastive Focus	23 (29%)	11 (8.5%)
Aboutness Topic (V2 Topicalization)	7 (9%)	20 (16%)
Frame-Setting Topic	47 (59%)	74 (58%)
Contrastive Topic	2 (3%)	15 (12%)
Total	79 (100%)	127 (100%)
p-value = 0.0124	(G-test for independent samples)

includes the types of informationally marked constructions, considering the classification in Götze et al. (2007) combined with that proposed in the literature on ClP (cf. Andrade 2015 for a review).22

The results for this table were significant.23 The differences indicate that ColBP

• does not show any cases of dislocation, unlike in ClP;
• significantly shows more cases of contrastive focus;
• shows fewer sentences with an aboutness topic or a contrastive topic.

In sum, ColBP displays a less pronounced variation in the types of marked constructions. These results may indicate a preference for the usage of clause-internal constituents at the left periphery in ColBP if we consider that they tend to express contrastive foci, whereas clause-external constituents indicate various types of topics (cf. contrastive topics), as the following examples from ColBP illustrate24

(8)

a.

Primeiramente

[com

aquelas

feridas]

representa

firstly

with

those

wounds

represents

Cristo

o

quanto

nos

ama

(contr. focus)

Christ

the

much

us

loves

‘Firstly, with those wounds, Christ represents how much he loves us’ (interpretation: ‘…it is with those wounds that Christ represents how much he loves us’)

b.

[A

Thomé]

mostrou

Cristo

cinco

Chagas;

since

Thomas

showed

Christ

five

sores

porém

a

nós

cinco

mil

(contr. topic)

but

to

us

five

thousand

‘To Thomas, Christ showed five sores; but to us, five thousand’

6. Discussions

The results shown above reveal, as expected, a great deal of similarity between the two texts, but there are some differences which we consider crucial for distinguishing between the two grammars.

Similarities are stronger in the domain of complement clauses, which do not reveal significant word order differences between the texts. Furthermore, the two texts display variation regarding the usage of double complementizers and of null complementizers, which are illustrated below only with ColBP sentences:

(9)

Pois

estai

certos,

fiéis,

[que]

se

não

thus

be.2pl

certain

believers

that

if

neg

correspondermos

de

outra

sorte

a

tão

correspond.fut.1pl

of

another

luck

to

such

grande

amor,

[que]

este

mesmo

amor

great

love

that

this

same

love

se

há

de

converter

em

indignação

pass

will

part

turn.into

in

indignation

‘Thus be certain, believers, that if we do not respond in another way to such great love, this same love will turn into indignation’

(10)

tantos

incêndios

sentia

no

coração,

so.many

fires

felt

in.the

heart

que

parece

∅

encerrava

no

peito

that

seems

(that)

held

in.the

chest

novo

Etna,

novo

Mongibello…

new

Etna

new

Mongibello

‘He felt so many fires in his heart, that it seems like he held a new Etna, or a new Mongibello in his chest…’

Differences occur in the main clauses of the corpus in the context involving the verb-subject order, which is exactly the combination detailed in Table 6, Table 7 and Table 8. Crucially, the V2 position is statistically more significant to distinguish between ColBP and ClP, and the former shows smaller numbers of V2 sentences. This may be theoretically interpreted as indicating a future demise of the V2 parameter in ColBP, although the data are still compatible with a V2 grammar. Another significant difference was found in the word order in adjunct clauses, and this may be interpreted in the following way: ColBP has a less complex left periphery (especially in dependent clauses), with a clear preference for V1 over the other categories.

Regarding the more specific comparisons shown in Table 7 and Table 8, they confirm that ColBP prefers to fill the lower (internal) positions in the left periphery, which are occupied by various constituents because of movement operations. Higher (external) positions tend not to be accessed in this grammar, and this is even stronger because there are no cases of dislocation constructions. Notice, crucially, that the use of internal positions to the clause triggers proclisis instead of enclisis.

The results do not allow us to affirm that a parametric change is at stake yet. However, they reveal frequency differences that are compatible with changes that will take place later:

• in the order between subject and verb, because BP shows a subject-verb grammar;
• in the order between clitic and verb, because BP shows a proclitic grammar.

EP, on the other hand, still shows more space for VS occurrences and a predominantly enclitic grammar.

7. Conclusions

The selected texts have many similarities, but also several points of divergence which cannot be interpreted as a mere reflection of personal styles. The main aspects of their differences, considering the adopted theory of syntactic change, are as follows:

• unlike in ClP, in which V2 is the preferred option, the ColBP text shows a greater number of V1 order (if compared to V2 and V3) in all clause types;
• unlike in ClP, there is a smaller use of structures with an external topic, especially those embodied in left-dislocation and contrastive-topic constructions.

With changes stemming from these differences, we expect to observe, either in other ColBP texts or in later texts from native Brazilian authors, the following characteristics:

• a higher frequency of subject-verb clauses;
• a higher frequency of proclisis.

Our hypothesis is that ColBP already showed higher frequencies of subject-verb and proclisis than ClP. The first characteristic may be observed from Table 3 if the totals of subject-verb are compared to those of verb-subject in the two texts (43% in ColBP versus 38% in ClP). The second point was already observed in a previous work (cf. Figure 3 and Cardoso 2020). These observations, which are merely made in terms of frequency at this point, may have catastrophic consequences for the later grammar of BP, following the observations of Lightfoot (1999), as empirically tested by Meisezahl (2024) for the acquisition of V2 grammars as depending on a high variability of constituents in first sentence position, with a high frequency of adjunct-verb orders in the primary linguistic data.

The relevance of frequency in the specific constructions of a language may impose a grammar change, to quote a paper by Hinterhölzl (2004). In other words, frequency changes may probably trigger micro- or nanoparameter changes in later linguistic stages. However, we would like to highlight that the results presented here should be considered not only from their quantitative significance, but also (and crucially) from their qualitative meaning as the pressure of language contact should also be considered among the speakers of ‘cultured’ Portuguese in Brazil. Further work on ColBP is needed to verify whether these results are confirmed in other texts and in other word order patterns.