Kalunga is a variety of Afro-Portuguese spoken in a rural community located in the state of Goiás, Brazil. In this study, we compare Kalunga with other varieties of Portuguese spoken in Angola, Brazil, Cape Verde, and Portugal and Portuguese-based creoles from a contact linguistics perspective. We investigate typological similarities, differences, and possible connections between these varieties. The results support previous sociohistorical and linguistic studies that reveal significant differences between Kalunga and standardized varieties of Portuguese, and the typological distinction between creoles, more vernacular varieties, and more standard varieties.
This paper compares varieties of language in the
Varieties of language in the Portuguese-speaking world.
Our data consist of feature value assignments based on speech data collected and analysed by specialists in the respective varieties. We use dialectal and typological features that address two sources. First, we have taken features discussed in previous descriptive studies of varieties of Portuguese, focusing on the linguistic phenomena identified in Kalunga (e.g.
In various disciplines, scholars have adopted the phylogenetic approaches of biology, making it possible to track evolutionary patterns and degrees of similarity among entities. Linguists have used phylogenesis to compare large amounts of data e.g.
Comparative studies of varieties of language in the Portuguese-speaking world have been carried out for only a few varieties, and consider a limited amount of data (e.g.
This study investigates the interrelations between varieties in the Portuguese-speaking world, based on empirical data. More specifically, we examine how Kalunga Portuguese, spoken by an Afro-Brazilian community, relates to a range of more standard and vernacular varieties of Portuguese, and to creole varieties with a Portuguese lexifier. It also highlights the relevance of data-driven studies in contact linguistics, and the importance of methodological consistency in comparative studies. All in all, it contributes to the classification and analysis of new language varieties, and to a better understanding of the outcomes of language contact in the Portuguese-speaking world, and more specifically, of the linguistically diverse situation in Brazil.
The paper is organized as follows: in Section 2 we discuss relevant concepts from the field of language contact, and we define the linguistic and sociohistorical background of the varieties studied. Section 3 presents the selection of features and coding used in our data set, and the phylogenetic method used to compare the data. Our analysis and results are discussed in Section 4, and Sections 5 and 6 present the discussion and conclusions.
There is no consensus in the literature on contact linguistics about the definition and classification of terms such as ‘vernacular variety’, ‘creole language’, and ‘standardized variety’. Without engaging in a deeper theoretical discussion on this matter, we present our definitions of the relevant terminology here, together with concise background information about the varieties compared in this study.
For convenience, we divide the varieties studied into three categories, based on their sociohistorical conditions:
Portuguese-based creoles
Portuguese vernacular varieties
Standardized spoken Brazilian Portuguese
The first category, (i) Portuguese-based creoles, refers to a very heterogeneous group of languages called ‘creoles’. In this study, we use the following approach to this term, inspired by Winford (
The second category, (ii) Portuguese vernacular varieties, includes a diverse set of spoken vernacular forms from various geolinguistic areas. Different processes of acquiring the Portuguese language were at work in the formation of these vernacular varieties. Accordingly, the outcomes of these processes are varieties that diverge from one another to varying degrees when viewed from a linguistic perspective. These varieties are used in everyday communication. Previous studies have include the following terms in the category of Portuguese vernacular varieties: ‘vernacular Brazilian Portuguese’ (cf.
In this second category, we have included urban and rural varieties of Brazilian Portuguese, and varieties of Portuguese spoken by communities that show a degree of isolation from mainstream Brazilian society, at a cultural and/or sociohistorical and/or geographical level. In addition to these Brazilian varieties, we include varieties of Portuguese spoken in Angola, Cape Verde, and Portugal. Subsequently, we present the geolinguistic areas of the Portuguese varieties examined in this study, and some aspects of their sociolinguistic.
(1) | Brazil | |
• | Afro-descendant communities |
Afro-indigenous communities
Rural and urban areas of Brazil
(2) | Cape Verde |
The
(3) | Angola |
(4) | Portugal |
The third category – iii) standardized spoken Brazilian Portuguese – refers to a variety of spoken Portuguese that resembles the ‘standard form’. It may be identified with the variety used by mainstream TV news programmes (
In short, the three language categories compared in this study represent a range of varieties used in the Portuguese-speaking world, with diverse profiles, in relation to standardization, social and geographical dimensions. In our comparison, we may be able to verify how much these three language categories converge and diverge, and how relevant this division into groups is.
For this study, we have chosen sixteen Portuguese language varieties, listed linguistic features thereof, and contacted professional linguists considered experts on the selected varieties. The database includes as many varieties of Portuguese and Portuguese-based creoles as possible that meet the following criteria: experts have conducted fieldwork on the variety, they have collected spoken-language data, and they are willing to code our tables of features and share language examples. We also made sure that the experts followed similar guidelines with respect to their methods of data collection and organization. This means, for instance, that the data collected must relate to spontaneous speech, that the data represent situational variation and is representative of the variety in question, and that the quality of their recordings was good enough to permit some acoustic analyses. Thus, we worked with speech data collected and analysed according to similar parameters. Moreover, the individual experts coded typological and dialectal features that were carefully developed on the basis of descriptive studies and previous studies of contact languages. All this means that we have a unique database that presents relative homogeneity in terms of data collection and data organization, consequently, an internally comparable dataset.
As we requested in our questionnaire, each specialist with whom we worked coded two tables of features of the language variety, and was asked to provide illustrative examples. The two tables of features are
By selecting these two data sets, we were able to obtain a good balance with regard to the number and the types of features used in our comparison. The WALS/APiCS features may allow us to place the varieties we investigated in a typological framework, whereas the specifically chosen features listed in the dialect questionnaire reveal typological and historical connections among the varieties of Portuguese. We do not include semantic/pragmatic phenomena in the Dialectal Features Table, because they have not yet been studied in Portuguese varieties within Corpus Linguistics as much as morphosyntactic and phonetic phenomena. For instance, with regard to lexical features, our data is based on spontaneous speech collected by the individual researchers, therefore it does not share a list of terms that include accurate semantic criteria that would allow for a comparative analysis of dialects. A set of meanings and their forms would have been much more difficult to extract, especially considering the wide range of topics of the data involved.
All the dialectal features we studied have binary values (presence–absence). They comprise different linguistic phenomena, such as agreement, negation, word order, and simplification of complex onsets. Agreement is a morphosyntactic phenomenon, and we include features concerning gender and number-agreement variation in a noun phrase (NP), person and number-agreement variation in a verbal phrase (VP), and the presence or absence of agreement in both NP and VP. Table
Features related to VP agreement, extracted and translated from Dialectal Features Table.
Verbal Agreement | Yes | No | |
---|---|---|---|
13. | Variation in the first-person singular marking on the verb |
||
(1) no son (Kalunga:
‘I don’t have a son.’ |
|||
(2) already the coffee (Kalunga:
‘I already made the coffee’ |
|||
14. | Variation in the first-person plural marking on the verb | ||
(3) no son (Kalunga:
‘We don’t have children’ |
|||
(4) already the coffee ‘We already made the coffee’ |
|||
15. | Variation in the third-person plural marking on the verb | ||
(5) no television (constructed example)
‘They don’t watch TV’ |
|||
(6) already the coffee (constructed example)
‘They already made the coffee’ |
|||
16. | No person and number marking on the verb | ||
(7) because- come (Kabuverdianu:
‘ ( |
|||
(8) sing know (Kabuverdianu:
‘ ( |
Agreement markers in the VP have been studied in several varieties of Portuguese, especially Brazilian varieties (e.g.
‘Negation’ is also a morphosyntactic phenomenon that is present in the Dialectal Features Table. Table
Examples of features related to Negation, extracted and translated from the Dialectal Features Table.
Negation | Yes | No | |
---|---|---|---|
20. | Frequent use of multiple negation markers in the same sentence. |
||
(9) go there ‘I don’t go there’ |
|||
(10) came nobody ‘Nobody came’ |
|||
21. | Negative indefinite pronoun ( |
||
(11) Nobody got job ‘Nobody could get a job’ |
|||
(12) In this place here nobody have future ‘In this place here, nobody has a future’ (Kalunga: |
|||
22. | Combined use of |
||
(13) had year that he nor came (Kalunga:
‘There were years when he did not even appear’. |
|||
23. | Use of |
||
(14) the rain destroyed the harvest ‘Did the rain destroy the harvest?’ |
|||
(15) destroyed because harvest we never had planted ‘It [the rain] didn’t destroy [the harvest] because we hadn’t planted the crops, right?’ |
In this set of features (20 to 23), we examine some constructions generally known as ‘multiple negations’, since previous experience indicated that the variety of negation markers among the varieties of Portuguese is significant. Feature 20 relates to the frequent use of multiple negation markers in the same sentence, and it represents a more general and common case of multiple negation in varieties of Portuguese. Feature 21, the negative indefinite pronoun as a subject followed by a negative particle, is not as widely distributed as Feature 20, yet our data shows the presence of Feature 21 in many varieties of vernacular Portuguese and a few creole languages. This was rather unexpected, as in the literature, this feature was not reported for many varieties of BP. Feature 22 seems to be uncommon among the selected varieties, and its presence is reported only for Kalunga Portuguese, Minas Gerais Portuguese, Cape Verdean Portuguese, and Kabuverdiano. Feature 23 had not been reported in the literature for Portuguese varieties, but our data shows that this feature is present in seven language varieties, namely Kalunga, Barreirão, Minas Gerais, Tembé dos Guamá, Libolo, Cape Verdean Portuguese, and Kabuverdianu.
The number of dialectal features is not equally distributed among the various grammatical categories in our Dialectal Features Table (Appendix 1). There are more features in some grammatical categories than in others (e.g. four features relating to negation and agreement in VP, and two features relating to topic constructions). This difference occurs because we selected features that were expected to represent variation among the Portuguese language varieties, according to previous dialectological and descriptive studies. Therefore, in order to avoid bias in our results, we wanted to determine whether these features are really significant when mapping differences and similarities among the varieties. For this purpose, during the data analysis we experimented with various combinations of features. For instance, we randomly removed some features, we removed some selected categories of features, and we removed features that (apparently) overlap in the same grammatical category, to test whether we would obtain a different result or not. The results remained essentially the same.
The more general typological data used in this study relies on features selected for WALS and APICS. Based on their set of shared features, we adjusted these features by translating their descriptions into Portuguese. We provided relevant examples, in order to facilitate the contributing researchers’ work. The features were coded using multi-state values. Because the dialectal features are binary, and the typological features are multi-valued, we do not combine dialectal and typological feature sets in the same analysis.
When applying phylogeny, computational methods allow the quantitative comparison of a significant amount of data, in order to investigate the relationships between entities. Different algorithms may be chosen to yield the desired type of comparison. In linguistics, the number of studies using computational phylogeny has recently increased (e.g.
In this study, we use SplitsTree4’s Neighbour-Joining and Neighbour-Net algorithms (
To build the graphic representation, first we have to encode the Dialectal and Typological features. This is done by answering ‘yes’ for the presence of a feature in a variety, and ‘no’ for the absence of a feature from a variety (for the Dialectal Features Table), or by choosing the value that best corresponds to the feature in the variety analysed (for the Typological Features Table). After encoding the features from both Dialectal and Typological Features Tables, the codes were converted into arbitrary numerical values. Missing data (where the participants did not know or did not have the information), were coded with a question mark (‘?’). These scores are not included in the calculation. These encodings lead to matrices containing the language varieties and the scores for the features.
The matrices with numerical values may be converted by the SplitsTree4 program into relative numerical distances between the compared entities. Then, the Neighbour-Joining algorithm converts these distances into a graphic in the form of a tree. The more similar the entities are, the closer together they are in a tree. The numerical distances help to visualize the closeness and remoteness calculated between the compared varieties. Besides the visual distances between the entities, the Neighbour-Net algorithm also shows different possible ways where a split may be made. The web-like network represents the conflicting signals in the form of multiple places where splits may be made.
An advantage of both algorithms in Splits Tree is that they provide an easy-reading general sense of how the different Portuguese varieties are related among themselves and in relation to Kalunga. Other cluster analysis algorithms, such as Principle Component Analysis, do not provide a well-displayed format of the groupings. A drawback of Splits Tree is that features responsible for the groupings are not informed. Therefore, in order to overcome this, we organized a heatmap of features based on the results provided by Splits Tree, to identify which features are more probably responsible to define the groupings. These features are discussed in Section 4.1 below.
Moreover, the foregoing algorithms have proven suitable for comparing language varieties when they descend from a common ancestor, as is the case with the varieties of Portuguese compared in this study (for references regarding the adequacy of distance methods, e.g.
In this section, we report the results of the phylogenetic analyses. In 4.1 and 4.2, we present the results of the analysis of the Dialectal and Typological features, respectively. A detailed description of all features mentioned in this section, and the relevant examples, may be found in Appendix 1 (Dialectal features) and Appendix 2 (Typological features).
Our first test relates to dialectal similarities among the 16 varieties of Portuguese, including Portuguese creoles. In Figure
Split network of a Neighbour-Net of the 57 dialectal features of the 16 language varieties.
At the top of Figure
Split tree of a Neighbour-Joining Tree of 45 morphosyntatic dialectal features of 15 language varieties.
Spoken Lisbon Portuguese appears relatively close to spoken Standardized Brazilian Portuguese (SBP) and Cape Verdean Portuguese (CVP), followed by Luanda Portuguese and then Libolo Portuguese. The position of SBP in the network shows that SBP tends to be closer to Lisbon Portuguese than to Brazilian vernacular Portuguese varieties, even though only spoken language data is considered in this study. The proximity of CVP to SBP and Lisbon Portuguese corresponds to the findings of recent studies that note the use of Portuguese mainly as an L2 variety in Cape Verde (a non-nativized variety), and to the use of European Portuguese as a referential variety for CVP speakers (
At the bottom of the network in Figure
In Figure
Analysing the data set in detail, we verify that there are not many features that clearly separate these groups of languages. The presence and absence of features are generally distributed among this set of language varieties. Only Feature 20, ‘Frequent use of multiple negation markers in the same sentence’ is shared by all the varieties, that is, it is irrelevant for the graphed results. Only six features (16, 37, 3, 29, 46, and 53) are present in, or absent from specific varieties. These features are discussed below.
Features 16 – ‘no person and number marking on the verb morphology’ – and 37 – ‘no grammatical gender in adnominals, as, determiners, pronouns and adjectives’ – are present in the creole varieties only. ‘No grammatical gender’ and ‘no inflectional morphology’ are referred to in the literature as prototypical features of creole languages, when compared to their lexifiers (e.g.
Features 3 – ‘the use of affricative /ʧ/ in context of /t/ plus approximant /r/’ as in
The two features not present in Lisbon Portuguese only are Features 46 –
With respect to Feature 46,
The phenomenon described in Feature 53, ‘element in topic position – to the left of the sentence – without having been moved from any position of the sentence’, also seems to be an outcome of language contact. It is absent only in Lisbon Portuguese,
Apart from the six features (16 and 37 for creoles, 3 and 29 for Kalunga, and 46 and 53 for Lisbon Portuguese) just described, there are other features that may be predictors of language groups, due to differences in the experts’ interpretations of the definition of each feature, and to the apparently less precise formulation of the features. For instance, Features 17, 18, and 19, which deal with specific properties of tense, mood, and aspect related to verbal morphology, are not present in either creole or standardized varieties of Portuguese. These features are present in vernacular varieties. Since creole languages usually have preverbal TMA particles, rather than inflectional verb morphology, Features 17, 18, and 19 are marked ‘no’ for creole varieties, meaning that they are not found in the creoles. The same features are coded as ‘no’ for standardized varieties, but for a different reason: the standardized varieties do have a more heterogeneous inflectional morphology than the vernacular varieties, and here, these features go in the opposite direction, that is, towards a more homogenous verb morphology. Therefore, the proximity of creole varieties and standardized varieties in the trees is partly due to the features that they do not share with vernacular varieties, rather than the features that these two groups share. In the same vein, the 12 phonetic and phonological features (Features 1 to 12) are present in vernacular varieties, but not in creoles and standardized varieties.
The feature groupings of the African and Afro-Indigenous varieties of Portuguese (Kalunga, Jurussaca, and Tembé do Guamá) suggest that these varieties have specific traits in common. However, they do not appear as a homogeneous group, probably because different linguistic processes were at work in the formation of these varieties, and/or they have developed in different ways. For instance, when comparing Kalunga, Tembé do Guamá, and SBP, we notice that both Kalunga and Tembé do Guamá show significant divergences from standardized varieties. However, Kalunga and Tembé do Guamá do not share the same differences with respect to SBP. When Kalunga and Tembé do Guamá do not share a set of features, we may consider two possibilities: i) Kalunga shares this set of features with SBP and the varieties of Goiás (São João and Barreirão), that is, Tembé is the particular case, or ii) Tembé do Guamá shares this set of features with SBP and Jurussaca (geographically close to Tembé), that is, Kalunga is the particular case.
Figure
Split tree of a Neighbour-Joining Tree based on 45 morphosyntatic features of seven Brazilian Portuguese varieties.
When comparing Kalunga with the other language varieties in Goiás (Barreirão and São João), we find 13 features that Kalunga does not share with the other two varieties. Of these 13 features, Kalunga shares only one feature with SBP, that is, Kalunga has 12 features from different grammatical categories that are neither shared with the varieties of Goiás nor with SBP. This shows that the rural variety of Barreirão is closer to SBP than Kalunga is, and, as expected, Kalunga shares more features with Barreirão than São João D’Aliança. This proves that there are various linguistic degrees of divergences between Kalunga (Afro-Portuguese), Barreirão (rural Portuguese), and São João (Portuguese from a rural town). Relative to SBP, Kalunga is the most distant, followed by Barreirão, and then São João.
The last tree (Figure
Split tree of a Neighbour-Joining Tree based on 48 Typological features, adapted from APICS/WALS of 16 language varieties.
It is not surprising that these general (typological) features present less marked differences, as all are varieties of Portuguese and Portuguese-lexifier creoles, and the selected features were designed to cover the typological variety of the languages of the world, whereas the dialectal features were selected to cover variation in varieties of Portuguese.
Comparative studies based on linguistic corpora that include a significant amount of data that allow us to verify the historical and typological relationships among entities are highly relevant to language contact studies. The use of computational models such as those used for this study helps us to visualize the relationships among language varieties, as they map similarities and differences among these entities. However, the results presented in the previous section, although interesting and significant, should not be analysed without considering the sociolinguistic contexts in which the languages are spoken (c.f.
Our language feature selection is certainly one of the greatest factors to have an effect on the results.
– there are three major clusters: creoles, standardized varieties, and vernacular varieties;
– Kalunga shares more features with the BVP varieties than with creoles and standardized varieties; generally, Kalunga shares more features with creoles than standardized varieties;
– Kalunga shares specific features with the language varieties spoken in the same geographical area of Goiás state, namely, Barreirão and São João; Kalunga shares more features with Barreirão (rural community) than with São João (country-side town);
– Kalunga is also close to Tembé do Guamá, which is an Afro-Indigenous Portuguese variety;
– Kalunga differs from Lisbon Portuguese and SBP more than the other vernacular varieties do;
– SBP is closer to Lisbon Portuguese than it is to any Brazilian vernacular variety, which may suggest that the spoken data from Lisbon is indeed close to standardized Portuguese; also, this suggests that (spoken) BP still follows the (spoken) European Portuguese language as standard;
– Cape Verdean Portuguese is close to SBP and Lisbon Portuguese;
– The BVP varieties cluster;
– The Minas Gerais variety is close to the Goiás varieties;
– There is no special relationship between Kalunga and the analysed varieties of Portuguese spoken in Angola and Cape Verde, in Africa.
One aspect that may have had an effect on the analyses is the way in which the researchers interpreted the linguistic features, and how they analysed the linguistic phenomena. Feature values that are essentially similar may be analysed and classified differently by researchers. Therefore, similar features may have been scored differently according to the researcher’s understanding. Moreover, the understanding of what is a categorical feature in the variety may vary from researcher to researcher. The potential impact of these observations of the outcomes has been reduced by our asking the contributors to provide examples or direct access to their data. After analysing the answers, it is sometimes possible to verify which features may be controversial, or may have led to different understandings among the researchers. In a few cases, we may observe a discrepancy between an example and the proposed score. In this study, we addressed that by double-checking such answers with the researcher.
As mentioned above, in this study we have considered only spoken language data, and we have attempted to use data that have been collected in compatible ways. However, with respect to methodology and for future works, we recommend that varieties of Portuguese, or areas where each variety of Portuguese is spoken, be more clearly delimited, as these play a role in how a language variety or a sociolinguistic area is defined. A more precise definition of varieties or areas may allow a more accurate comparative analysis, as we compare micro-areas such as those of Afro-Indigenous communities (Jurussaca and Tembé do Guamá), an Afro-Brazilian community (Kalunga), a rural community (Barreirão) with macro-areas, such as the Minas Gerais state in Brazil, and Lisbon and surrounding areas, in Portugal.
Extending the database by adding dialectal features and language varieties would allow a more fine-grained analysis of the relationship between linguistic varieties in the lusophone world, therefore we would be able to reach more informed conclusions in relation to the different categories of languages. For instance, it would be interesting to add to the comparison varieties of Portuguese spoken in Asia, more varieties of Portuguese spoken in Africa and Portugal, more Afro-communities in Brazil, such as in Bahia (where it is argued that there was a strong contact influence from African languages) and in the Rio Grande do Sul (where, apparently, the Afro-Brazilian communities were not as isolated from the rest of society as the other Afro-communities in Brazil). It would also be interesting to investigate varieties of Portuguese spoken close to the country’s borders, as a significant amount of language contact exists there.
In this study, we analyse the relationship between varieties of spoken Portuguese and Portuguese-based creoles, based on morphosyntactic and phonetic/phonological features. The results show differences between the varieties of Portuguese and the creoles, and reveal clusters of three major groups: creoles, more standard varieties, and more vernacular varieties. In general, these groupings are consistent with the group-language distinction made in Section 2 (i.e., Portuguese-based creoles, Portuguese vernacular varieties, Standardized spoken Brazilian Portuguese). The only exceptions are Lisbon Portuguese and Cape Verdean Portuguese, both of which appear in the results with SBP, suggesting that they could be categorized as standard varieties, instead of vernacular varieties. The results corroborate Raposo’s (
Kalunga appears close to the vernacular varieties of Portuguese, especially BVP varieties, and substantially distant from standardized varieties. In general, Kalunga is typologically closer to creoles than to standardized varieties, possibly owing to features that are outcomes of the specific contact situation from which Kalunga emerged (as discussed in Section 4.1). Also, Kalunga has specific features (also discussed in Section 4.1) that are not shared by other varieties, possibly features that are independent developments owing to the degree of Kalunga isolation.
This study finds interesting similarities and differences among varieties of Portuguese. This paper presents the first quantitative phylogenetic analysis of the relationships among a wide range of varieties of Portuguese and Portuguese Creoles, based on spoken data. This analysis helps us to understand the relationship between Kalunga and the Portuguese varieties and Portuguese-based creoles.
The additional files for this article can be found as follows:
Dialectal Features Table. DOI:
Typological Features Table. DOI:
We do not indicate Standardized Brazilian Portuguese on Map 1, since it is not a variety spoken in a specific geographic region of Brazil. See our definition of Standardized Brazilian Portuguese in Section 2.
According to the
Planned speech belongs to a type of written corpora, which is another language diasystem (
Since the more and less standard varieties of BP show no differences between the second person singular and third person singular, and since the second person plural has the same form as the third person plural in many varieties of BP, we did not analyse the 2
The glossing and free translation into Portuguese in (7) and (8) are ours.
By ‘same sentence’, we mean that the particle ‘no’ is part of the verbal negation, so we do not consider the particle ‘no’ when it is part of a different utterance.
In the Splits Manual (p. 10) (
For Feature 46, we refer to the use of relative pronouns in dependent relative constructions (DRC), and not to the relative pronouns in free relatives (FR). DRC are constructions that are semantically and syntactically related to an antecedent, as in
Feature 53 does not deal with topicalization (which involves the movement of an element to the left ‘periphery of the sentence’ without any resumptive element), nor with left dislocation (topicalization of an element that is resumed by another element). Both topicalization and left dislocation are widely mentioned in the literature on varieties of European Portuguese (cf.
Refer to Section 3.1 to see some actions, in order to shorten the effects of some specific features of the results.
We would like to thank all the participating researchers for their immeasurable contributions to this paper. Without them, this study would not have been possible: Carlos Figueiredo (UM), Ednalvo Campos (UEPA), Eduardo Ferreira dos Santos (UNILAB), Fernanda Ziober (VU), Francisco João Lopes (Instituto Camões), Gabriel Antunes de Araújo (USP), Heliana Mello (UFMG), Manuele Bandeira de Menezes (UNILAB), Mara Jucá (UEPA), Márcia Moretti (TV journalist), Maria de Lurdes Zanoli (USP), Nélia Alexandre (ULisboa), Shirley Freitas (UNILAB), Silvana Silva de Farias Araujo (UEFS), Tommaso Raso (UFMG).
The authors have no competing interests to declare.