Data Processing

We had access to 169,222 documents, including 57,669 petitions, from the Portuguese Overseas Archives in Lisbon (Arquivo Histórico Ultramarino de Lisboa). Our goal was to identify and extract key information about the senders and recipients of these petitions, such as their locations, role in the empire (e.g., king, governor, soldier, member of the society, etc.), ethnicity, societal group (if any), and any other information that we could attribute to the actors. To accomplish this, we used techniques from natural-language processing (NLP)—a branch of artificial intelligence, machine learning, and linguistics. NLP can be used to analyze, process, and comprehend large sets of documents.

First, we used a technique called regular expressions, which allows the identification of patterns in a text based on a sequence of characters. Although many of the documents we analyzed do not present a well-structured pattern because of their old age, we were able to isolate pieces of text with information about senders and recipients. Moreover, we identified the documents involving minoritized people in the empire, such as the population of color, Indigenous people, people from religions other than the Catholic Church, and low-ranked military members.

Next, we used entity recognition, an NLP technique, to locate and extract entities mentioned in a text and then to classify them into defined categories, such as person, location, organization, products, and events. To address the particularities of historical documents in Portuguese, we created a sample of 2.5 percent of the total number of documents to train the tool using its machine-learning function. The tool classified entities as PERSON, LOCALIZATION, and ORGANIZATION. We also created two new categories for the classification of entities: ROLE, in which we identified the roles of the actors in the Portuguese Empire, and AFFILIATION, to extract degrees of kinship between those involved in the communication. In December 2019, we improved our dataset by manually annotating four thousand records and by classifying them into new categories: Female Name (FEM), Male Name (MALE), royal and colonial political institutions (ORGANIZATION), Military units (MILIT), religious institutions and brotherhoods (RELIG), noble and land title (TITLE), occupation (OCC), and localization (LOC). By doing so, we successfully reached 93.1 percent accuracy for our dataset.

Did the Subaltern Speak? SNA for Extracting Silenced Voices

The documents stored at the Portuguese Overseas Archives allowed us to define a group of petitions that were sent to Lisbon by enslaved people, freedmen, Indigenous people, mulattos, pardos, and others living in Portuguese colonies in the first half of the eighteenth century. Those petitions are our relational data, because they connected the population of color within the colonies to public agents or institutions in Portugal. Our attribute data reflect the special characteristics that belonged to petitioners as individuals or groups, based on gender, social status, or religion. We strongly believe that social position within Portuguese colonial networks should not be analyzed only through dichotomous relations and measures of exclusion. We determined that those social actors of the early modern population of color could be best studied using four categories, which were important at the time and contributed to racial and class discrimination in its colonies:

• skin color or/and origin: negro/preto (Black), pardo (Black), índio (Indigenous), mameluco, branco (white), mestiço (mestizo), gentio da Guiné (gentile from Guinea)
• gender: female, male
• religion: cristão (Christian), cristão-novo (New Christian), judeu (Jewish), mouro/moursico (Moorish), etc.
• social status: isente de escravidão (free from slavery), forro (manumission of an enslaved person), escravo (enslaved person), liberto (freedman), etc.

By recognizing the population of color not as a homogeneous group but as an endogenously and exogenously diverse group whose social attributes perfectly reflected members’ social categories, we combined SNA with the theory of intersectionality. The first served as an analytic tool, and the second provided a theoretical background, both enabling a framework to better understand how aforementioned social divisions of color, gender, religion, or social status could differently position the population of color in the early modern world, especially in relation to colonial discrimination. Building on Patricia Hill Collins and Sirma Bilge’s argument that “intersectionality proposes a more sophisticated map of social inequality, that goes beyond class-only accounts,” we believe that combining intersectionality with SNA can provide a more complex understanding of colonial discrimination.5 Therefore, we examined not only which categories magnified social exclusion but also which categories granted some Black people the right to have a voice and let them speak loudly.

While analyzing our dataset, we discovered some peculiarities in the official correspondence exchanged between Lisbon and its colonies. We tried to understand which marginalized groups were the subject of official correspondence in the years 1642–1833 among the population of color, people living on the border of the law, the so-called Shadow Empire, or people persecuted by the Inquisition or accused of immoral acts. Among all the possible marginalized groups of the Portuguese colonial empire, the population of color was the most-represented one and the most debated among different social agents, including themselves. However, even if a graph may show direct connections between those groups and the monarch, we must remain aware that, practically, such a relationship was impossible. Because of low levels of literacy among the vast majority of colonized people (not to mention the marginalized within those societies), it is highly likely that people of color would have resorted first to the literate royal officials in the colonies, who then provided them with support. Yet it still reveals the possible political channels between the colony and the metropole, keeping in mind the “invisible social networks” and invisible agents who made this contact possible, such as lawyers, procurators, and other officials of the judicial authorities.

Figure 13.1 displays information on the numbers of separate documents that referred to the population of color, divided first between ethnicity/race (Indigenous and African) and second by social status. Note that the two categories can intersect because the Africans and the Indigenous Brazilians could be both enslaved and freed.

There are three possible approaches to creating and visualizing the social networks of the people of color and their position in the colonial structure in the early modern Portuguese Empire:

1. The first quantitative approach refers to the cooccurrence networks that provide potential relationships between the population of color and colonial authorities based on the correspondence networks. It is based on the terms referring to the Brazilian Native peoples and the African communities (see Supplemental Figures 13.1 and 13.2). We did not consider the anonymous groups of enslaved people who were treated as the merchandise in the transatlantic slave trade; instead we focused on those who were named and existed in such social networks with their own identity. Supplemental Figure 13.1 presents the cooccurrence based on the ethnicity/“race” of the Indigenous Brazilian peoples and Africans, and Supplemental Figure 13.2 presents the cooccurrence of the same groups according to their social status—enslaved or freed.6

   In our dataset, we identified 684 nodes, which represent the official correspondence exchanged between various social agents representing colonial institutions or the society and that referred to the Brazilian Native peoples and the Africans, who were mentioned in such documents by first name or by first name and surname. Among them, Africans represented 73.68 percent of all records, whereas Native peoples represented 26.32 percent.

   We applied the same approach in the next visualization (Supplemental Figure 13.2) showing the occurrence of the enslaved and freed in the official correspondence between Lisbon and its Atlantic colonies between 1642 and 1833. In this graph, we identified 609 nodes that represent the enslaved and freed African and Indigenous people mentioned by first name or by first name/surname in the official correspondence. The enslaved constituted 84.73 percent of such documents, whereas the freed comprised 15.27 percent. Both Supplemental Figures 13.1 and 13.2 reflect quantitative research, complemented by the graph in Supplemental Figure 13.4, which is based on qualitative historical research that takes into consideration all personal relations mentioned by the people of color themselves.

2. The second approach is based on qualitative research, revealing various relations between the population of color and the administrative structure. People of color themselves used identifiable categories to determine their own position and to achieve their purpose. Such a graph is focused more on particular social actors, such as the king, the Overseas Council, or local governors. Supplemental Figure 13.3 shows the social position of the population of color within the Atlantic colonies of the Portuguese Empire under the reign of John V of Portugal (1706–1750).7

   People of color were aware of ethnic/racial differences or status.8 Regardless of their position, whether it was internally defined or imposed from the top, what can be observed from the graph itself is that the people of color were not only directly connected to the Lisbon metropole but also deeply immersed in multiple personal or official networks.

3. The third approach is also based on qualitative research: it refers to the people of color’s personal narratives included in the petitions they sent to the colonial administration. It is important to determine all the possible social actors mentioned in such documents and define their personal relations with the population of color. Such relations are impossible to detect automatically; hence historical analysis is required. Their narratives, as well as their own perceptions of the world and of themselves, can say a lot about their position in the colonial system and relations with non-enslaved agents. In Supplemental Figure 13.4, we extracted all social relations mentioned by the Africans and Indigenous Brazilian peoples in their petitions sent to the King John V during 1705–1755.9 In this graph, there are 242 social agents (nodes) and 340 relations (edges) between them. Whites formed the dominant group, which represented 73.14 percent of all individuals mentioned by both groups. Black people represented 18.18 percent of existing social relations, and Indigenous people, 8.68 percent. Contrary to what we observe in Supplemental Figure 13.1, where the official channels included the Indigenous Brazilians and the Africans together, the personal relations between both groups were not mentioned.

Motivation and Strategies of the Population of Color: A Qualitative Approach

It is worth keeping in mind that living social networks should not be conflated with network visualization or the graphs themselves, because they are only representational elements. Therefore, the subject matter of the analysis is not the characteristics of the subjects analyzed, but the wide range of relations between them—from simple interpersonal relationships to economic or political connections. The graphs of historical social networks may show some tendencies and dynamics of building relationships with others but do not reveal the details of the personal relationships nor the strategies; therefore, a mixed-methods approach that includes qualitative analysis is essential.10

Analysis of the petitions sent by people of color identified some of the correspondences’ principal characteristics. First, the letters were sent in the name of the representatives of different social groups: enslaved people, forros (freedmen), pretos (Black people), mulattos, or Indigenous peoples. Differences, however, are seen in the body of the letters and the locations from where they were sent. Women who were correspondents in those documents lived exclusively in different parts of Brazil, although many had been brought as enslaved people relatively recently from Africa to the Portuguese America. In contrast, the letters sent by men came from all of the Atlantic Portuguese colonies, such as Brazil, Guinea, Cabo Verde, São Tomé, and Angola. The content of the letters also differed: those written by women mainly concerned the processes of liberation, domestic relations, and arbitration in the conflicts with their masters, whereas the letters written by men addressed both personal and professional issues.

To explore existing social relationships, we needed a qualitative approach. Consequently, we defined and studied the following elements of the petitions: social actors, function, localization, rhetoric, structures and techniques of writing letters, patterns, the relevant third party involved, narration of the problem, and social attributes. We assumed that the Portuguese Empire was not a set of isolated regions but was connected by multiple social networks. These networks created a dynamic basis for social interactions between various levels, enabling social dialogue. By adopting a network perspective in our research, we also assumed that the individuals were connected, their outcomes were significantly related, and all the actors’ actions were intentional.11 Moreover, we claimed that the Portuguese Empire should not be seen merely as a strongly hierarchical pyramid-like society, but as an embedded world in which individual choices should be analyzed in relation to their environment. For this reason, a qualitative approach, based on a thorough analysis of historical documents, is absolutely crucial.

Our qualitative analysis shed further light on relationships between enslaved people and those who enslaved them based on documents from the period of 1706–1750. The graphs illustrate invisible patterns and provide new research questions; they demonstrate that there were different relationships among people of color and between people of color and others. Such was the case of Rosa Preta, an enslaved woman from the Gulf of Guinea who lived in the Brazilian city of Salvador da Bahia. In 1725, she asked King John V of Portugal to evaluate her manumission. Rosa expressed her hope that the monarch would order Vasco Fernandes César de Meneses, Viceroy of Brazil, to pronounce her freedom. While enslaved, she had been subjected to violence and “illicit treatment,” which became a fundamental argument to apply for her freedom. She complained that she was abused, handcuffed, and beaten by her master, suffering many injuries. The request was sent to the king so that he could directly appeal to the colonial authorities and, more specifically, to the viceroy of Estado do Brasil. This case is of great relevance because it shows that, inside the enslaved community there must have been some network that enabled the flow of information, so that a newly arrived enslaved African like Preta, who was unaware of the proper political and legal channels and had only a poor command of the Portuguese language, was able to take advantage of such information to apply for her freedom.12

Moreover, in several cases, the enslaved had relationships that went beyond the microworld of senzalas (living quarters for the enslaved) to engage local institutions. Such was the case of the Indigenous enslaved girl, Francisca Lopes de Sousa, who appealed to the king for her freedom in 1732, stating that she was a white honorable woman and the daughter of a powerful man in the city. Another enslaved woman, Rosa Cafusa, while requesting her freedom, asked that she continue to live and work in the house of her enslavers, with whom she maintained relatively friendly relations, because of her fear of the local authorities.13 These examples demonstrate that people of color were indeed embedded in the colonial structure and interacted with other, non-enslaved people; their agency should be interpreted not only within a class struggle characterized by oppression and resistance but also within a broader cultural and relational context.

A significant case that showed that social networks of enslaved people exceeded local and regional relationships and extended across the ocean is that of João José.14 This Black freedman, born as a free man in the city of São Cristóvão of Havana (Cuba), appealed to Lisbon in search of justice. He requested that King John V instruct the governor and the judges of the island of São Tomé to restore his freedom and land and provide compensation for damages and injuries suffered in the Islands of São Tomé. While working on a Castilian ship, he was imprisoned by English pirates who took him to the Islands of São Tomé, the Portuguese colony in the South Atlantic. Unfortunately, being unable to prove that he was a free man, he was arrested and sold to Manuel Luis Coelho, who during one of his trips to the Brazilian port of Rio de Janeiro, granted João José his freedom. After Coelho’s death, João José was arrested again in the Islands of São Tomé by Coelho’s brother, Domingos Luís Coelho, who kept him in a private jail for four months. Finally, he was sold to a French captain and taken to France, where he managed to escape. In France he was advised by some confessors, as he described them in his letter, who were not enslaved (although they were Black and fugitive) that he had the right to appeal directly to the Portuguese king. Going from port to port, he finally managed to get to the court of King John V. It should be noted that João José chose the judicial form of solving the problem, rather than escaping, and that the information about such a potential solution was circulating within other European countries.

These examples, as well as our analysis of the relations between the population of color and Lisbon authorities, demonstrate some strong tendencies to create networks and to live in communities in the former Portuguese colonies. Relationships to the king, the Overseas Council, and the most active colonial officials in the colonial structure can be analyzed in two ways. First on the macro scale, the king and his administration were in the center of all relationships, bringing together and connecting all colonial subgroups. Lisbon was at the center of all changes occurring inside the empire, because the king was so powerful in the network and had an extremely high rate of connectivity with the colonies. Metropolitan institutions greatly influenced colonial societies and authorities, controlled the access and flow of resources, and were essential for the diffusion of new standards, laws, and norms. Second, on the micro scale, which is revealed by qualitative historical analysis, communications between enslavers and enslaved people show that the king was seen not only as an arbiter, mediator, and a “good father” in a typical domestic relationship but also as the sovereign and the last instance in the relations of authority between the center and its colony. In the petitions presented earlier, the king played the role of a protector who defended his servants and “loyal vassals” from local injustices and oppressive authorities.

Final Considerations

Historical network research is very challenging, because it requires trying different approaches and constantly seeking ways to improve those tools. Graphs cannot provide definitive answers but can help researchers formulate better and deeper research questions. In our case, we worked with more than 170,000 correspondence registers. To understand the principles ruling the colonial world, we had to look at certain elements of the network from a broader perspective. As Supplemental Figures 13.1 and 13.2 show, this approach made it possible to extract social groups that we may not have been able to perceive otherwise.

Another challenge, which is not found in contemporary social network studies, is posed by the fragmented nature of historical sources, which are most useful for studies of correspondence, kinship, social movements, or economic history.15 Of course, there is also the issue of the availability of documents; many of those stored in the Lisbon archives before 1755, the year of a major earthquake, were likely destroyed. Additionally, analysis of the graphs of social networks should be backed up by a solid qualitative study that considers aforementioned invisible social networks.

Regarding populations of color, SNA can contribute a great deal to historiographical research, because it can extract the “voices” that often go unheard and demonstrate that, enslaved or not, people of color spoke loudly about their needs, problems, and fears and were, to some extent, able to decide about the course of their present and future lives. Some engaged directly the institutions of the metropole and used elements of the Lisbon bureaucratic system for their own purposes. Consequently, even an enslaved person in Brazil could be a subject rather than only an object of the historical changes of early modern colonial relations. These results may contradict traditional historiography that concentrated on people of color being reduced to the condition of a thing, or a submissive group deprived of their rights, dependent merely on those who enslaved them to be the only ones to make decisions about their lives and properties, and being just an instrument of work not capable of any education or morality.16

Two important questions remain unanswered: How did individuals from the margins of colonial society have access to information, and who connected the smaller communities scattered throughout Brazil? Because we cannot find these answers in the given documents, we should take into consideration factors other than network concepts and look to historical events. For example, Domingo Álvares, the African healer and priest, crisscrossed Brazil between 1730 and 1750 against his will; he was jailed in Pernambuco, sold to owners in Rio de Janeiro, and finally sent to Portugal. Throughout this time he maintained his rebellious character and continually opposed the legitimacy of the Portuguese imperial power. His life was also proof of the movement of enslaved people between different senzalas and, with that movement, the circulation of much information and news. It is believed that Domingo created a community based on traditional African hierarchies and allowed both the enslaved and freemen to join it. The structure of that community indicates that those marginalized in colonial societies demonstrated an inclination to create and to live in smaller communities.

As the results of our research demonstrate, even though the enslaved, freedmen, mulattos, pardos, and Indigenous people were socially, politically, and economically constrained, they were able to fight for their rights and, to some extent, to establish contact with Lisbon. Perhaps the roles played by these groups in the history of the Portuguese Empire were much more central, crucial, and interdependent than we could ever imagine. SNA, with its quantitative, qualitative, and mixed-methods dimensions, as our research shows, offers much to the study of the African diaspora by expanding our understanding of colonial relationships.

Notes

Research presented in this article is a part of The Portuguese Overseas Identity project; see https://mape.netlify.app/. Agata Błoch thanks the (Polish) National Science Centre (grant 2017/27/N/HS3/01104) and the Tadeusz Manteuffel Institute of History of Polish Academy of Sciences for support. The authors also thank Jessica Joia, Bartosz Lemiec, and Monika Pawluczuk for their contributions to data preparation and processing. We are also grateful to Lina Oliveira Lopes-De Matos and his students from University Marie Curie-Sklodowska in Lublin for their help in annotating the source documents.

1. Note that the abolition of slavery in Portugal was not tantamount to prohibiting the transatlantic slave trade, which was abolished only in the early nineteenth century, whereas slavery was not abolished in the entire area of the Portuguese Empire until 1869. Yet, in Brazil, slavery continued until the law called “Lei Áurea” (nº 3.353) came into force on May 13, 1888, which in two articles definitively abolished slavery.

   Return to note reference.

2. Lacombe et al., Rui Barbosa e a queima dos arquivos.

   Return to note reference.

3. Schwartz, Segredos Internos.

   Return to note reference.

4. Pimentel, Viagem ao fundo das Consciências. A escravatura na época moderna.

   Return to note reference.

5. Collins and Bilge, Intersectionality Key Concepts, 15–16.

   Return to note reference.

6. Supplemental Figure 13.1 depicts the cooccurrence of the Indigenous Brazilian peoples and the African communities in the official correspondence between Lisbon and its Atlantic colonies between 1642 and 1833, and Supplemental Figure 13.2 depicts the co-occurrence of the enslaved and the freed in the official correspondence between Lisbon and its Atlantic colonies between 1642 and 1833. These data visualizations were not reproducible for print but may be found in the Manifold edition of The Digital Black Atlantic, which will be available at https://dhdebates.gc.cuny.edu.

   Return to note reference.

7. Supplemental Figure 13.3 depicts the social networks of the population of color within the Atlantic Portuguese Empire based on their narratives between 1706 and 1750. This data visualization was not reproducible for print, but may be found in the Manifold edition of The Digital Black Atlantic, which will be available at https://dhdebates.gc.cuny.edu.

   Return to note reference.

8. The first half of the eighteenth century is an important period of study, because at the same time as African, European, and Indigenous communities were being rooted in the Portuguese colonies, the slavery model based on strong social classification and discrimination policies still remained in force.

   Return to note reference.

9. Supplemental Figure 13.4 depicts social relations mentioned in the narratives of the Africans and Brazilian Native peoples between 1706 and 1755. This data visualization was not reproducible for print but may be found in the Manifold edition of The Digital Black Atlantic, which will be available at https://dhdebates.gc.cuny.edu.

   Return to note reference.

10. Martinez et al., “Combining Qualitative Evaluation and Social Network Analysis”; Domínguez and Hollstein, Mixed Methods Social Networks Research.

    Return to note reference.

11. Robbins, Doing Social Network Research, 4–5.

    Return to note reference.

12. “Requerimento de Rosa Preta do gentio da Costa da Mina ao rei.”

    Return to note reference.

13. “Requerimento da Rosa Cafusa ao rei.”

    Return to note reference.

14. “Requerimento de João José ao rei.”

    Return to note reference.

15. Rollinger et al., “Editor’s Introduction.”

    Return to note reference.

16. Filho, Depoimentos de escravos brasileiros, 57; Montesquieu, Do Espírito das Leis, 221; Malheiro, A escravidão africana no Brasil, 24–25.

    Return to note reference.

Bibliography