Why Digital History? What can text data show us?

Digital methods, such as text mining for historical analysis, are central to this research, offering insights into linguistic structures and patterns that traditional historical modes of investigation, such as close reading and contextualization, alone do not easily uncover. The colonial press printed and disseminated texts shaped by the social, cultural, and political priorities of white colonists, ship captains, merchants, enslavers, and administrators from numerous locales throughout the British Empire. These priorities influenced the content of what was published and the linguistic framing of black resistance. While traditional historical methods excel at interpreting individual narratives, they can fall short in identifying the broader structural patterns embedded in language over time.

Co-Occurrence of Nouns and Adjectives in Corpus One 1732-1742

The legend on the right labelled “cooc” stands for the number of co-occurrences. This word co-occurrence network displays th most frequent noun and adjective co-occurrences in the first ten years of the South Carolina Gazette’s publication included in Corpus One. On the bottom of the network visualization, we can see the co-occurrence of the terms “rebellious” and “negroes.”

Text mining techniques, including word frequency analysis and part-of-speech co-occurrence, allow us to pinpoint moments when colonial authors and printers differentiated between perceptions of when black conspirators could remain “our slaves” and when they could only be “the rebellious Negroes.” These linguistic shifts reflect how black resistance shaped white authors to rationalizations of their dependence on enslaved labor with the destabilizing force of black defiance. By aggregating and analyzing data systematically, digital methods expose traces of black political violence embedded within colonial language, enabling us to read against the grain of the texts. Moreover, digital methods afford further insights into temporal, regional, and colonial change. Regional variations, such as tactical selectivity of information and reticent vocabularies, underscore how local contexts influenced the colonial descriptions of collective enslaved resistance, while shared linguistic strategies—such as the pervasive use of passive voice—reflect broader white anxieties.

Example of the Output of a Concordance Search for the Term “Insurrection” in Corpus One (The South Carolina Gazette)

Concordances are a computational technique that allows researchers to view all instances of a given word, token, lemma, or phrase and its surrounding textual context.

Importantly, digital methods highlight that even within articles that expressed reluctance or unease about slavery and the treatment of enslaved men and women, white authors employed similar grammatical structures that reinforced and flattened ideas of black collective resistance. This suggests that these linguistic patterns undergirded the ideological stances of individual writers, pointing to a shared colonial grammar that could, at an operational level, transcend editorial intent and reinforce the broader racial and social orders set in motion by the transatlantic slave trade.

Distribution of Syntactic Dependency Relationships for Target Lemmas in Corpus One 1732-1775

This grouping of visualizations displays the relative frequency of specific syntactic dependency relationships (dependency grammars) for the lemmas “insurrection,” “negro,” “negroes,” “rebel,” “rebellious,” and “slave” in Corpus One between 1732 and 1749. The legend on the right labelled “dep_rel” stands for dependency relation. Each type of dependency relation, based on the Universal Dependencies annotation scheme, is color coded. Notably, between the 1730s the lemma “negroes” more frequently appeared in sentences as the syntactic object or oblique nominal. By the 1740s, “negroes” was more often employed as a nominal modifier, meaning that it modified another noun.

By scaling up the analysis of linguistic patterns, digital methods provide a comprehensive view of how language functioned as both a tool of colonial power and a record of disruption. This approach highlights structures of oppression and recovers the traces of black defiance that reshaped those structures. Digital analysis, therefore, opens new avenues for understanding the dynamic relationship between the structure of language, power, and historical black resistance. Thus, when employed carefully and creatively, computational methods can open novel avenues for understanding the dynamic relationship between language, power, and resistance.

Percentage of Active and Passive Voice in Corpus One 1732-1775

This figure highlights the percentage of active and passive voice sentences in Corpus One (1732-1775). For Corpus One, Corpus Two, and Corpus Three, the percentage of passive voice and active voice follow similar distributions.

Percentage of Active and Passive Voice in Corpus Two 1736-1775

Percentage of active and passive voice sentences in Corpus Two (1736-1775). A little over sixty percent (61.3%) of the sentences in the articles related to enslaved resistance published in the Virginia Gazette feature active voice, while just under forty percent (38.7%) contain passive voice.

This project underscores the the idea that digital methods provide an essential avenue for re-examining well-trodden archival sources - in particular those authored by oppressors, enslavers, and colonizers. By looking beyond narrative content, we can uncover the deeper ideological work performed by language—work that sought to flatten and distort enslaved resistance to fit colonial prerogatives. Understanding how language was used to diagnose and contain black resistance in the past equips us to critically engage with the persistence of these dynamics in contemporary narratives about race and resistance.

Pronoun Usage per 10,000 words in Corpus One 1732-1773

This figure illustrates the normalized usage of pronouns over the eighteenth century in Corpus One (South Carolina Gazette). The pronoun frequencies are relativized per 10,000 words to better address varying article lengths and years with relatively little text data. The distribution of pronoun frequencies reflects patterns text data over time, highlighting that in the 1750s the South Carolina Gazette printed little information regarding collective enslaved resistance after 1750.

Pronoun Usage per 10,000 words in Corpus Two 1736-1775

This figure highlights the normalized usage of pronouns over the eighteenth century in Corpus Two (Virginia Gazette). The pronoun frequencies are relativized per 10,000 words to better address varying article lengths and years with relatively little text data. The distribution of pronoun frequencies generally reflects patterns in text data over time, highlighting that in the 1740s and 1750s the Virginia Gazette published relatively little information regarding collective enslaved resistance.

The Hybridity of Mapping and Text Mining: Enslaved Resistance, Colonial Response, and Slave Trade Data

To further explore how digital methods afford us new perspectives on the past I used both coding in R (with leaflet and _shiny (https://www.shinyapps.io/)) as well as the open source mapping software Quantum Geographic Information System (QGIS). Below I have generated two static maps as well as two interactive timeline maps that combine slave trade data and location data from the South Carolina Gazette and the Virginia Gazette to bring the trafficking of enslaved people into relief alongside news of resistance in the eighteenth century.

Both map types remain prototypes and are not immune to the problematic, reducing nature of mapping. However, this project was inspired by the scholar Katherine Mckittrick’s book Demonic Grounds, the combination of text metadata and mapping attempts to provide a new way of understanding black geography. She argues that “existing cartographic rules unjustly organize human hierarchies in place and reify uneven geographies in familiar, seemingly natural ways.” By studying slavery through memories, writings, theories, and geographies, McKittrick addresses the idea that “locations of captivity initiate a different sense of place through which black women can manipulate the categories and sites that constrain them.” In particular, McKittrick suggests that space and place give black lives meaning in a world that has, for the most part, incorrectly deemed black populations and their attendant geographies as ‘ungeographic’ and/or philosophically undeveloped.” Thus how might the locations extricated from the grammars of colonial response to slave revolt, insurrection, or conspiracy help us to redefine the geographies of black life and resistance?

Overview of the Transatlantic and Intercolonial Slave Trades to South Carolina and Virginia alongside Reports Refercing Collective Enslaved Resistance printed in the South Carolina Gazette and Virginia Gazette in the Eighteenth Century

Geography: “a location through which a moving technology can create differential and contextual histories” (xii); “socially produced and therefore an available site through which various forms of blackness can be understood and asserted,” Katherine Mckittrick, Demonic Grounds, xix.

Below, I generated two static rudimentary maps that display general embarkation regions of transatlantic and intercolonial voyages weighted by the number of enslaved captives that arrived in South Carolinia and Virginia ports and the locations referenced in the South Carolina Gazette and Virginia Gazette’s reports regarding collective enslaved resistance. These maps can be re-created by following the University of California, Santa Cruz (UCSC) guide’s for mapping data from SlaveVoyages.org. To find the general guide for generating a map in QGIS see Mapping Data from SlaveVoyages.org. For creating a heatmap and displaying basic change over time with QGIS software (Quantum Geographic Information System) see the Heatmaps and Change over Time tutorial. These tutorials come from UCSC’s course LALS 194E - Unfree Migrations. I adapted these guides to meet my project’s objectives, including adding layers that could display the referenced locations from the colonial press. Moreover, these maps are only starting points—it is critical bear in mind that they dispay in a disarmingly whimsical manner, the violent and obscene trafficking of people from Africa and their descendants during the eighteenth century. Moreover, the data displayed should not be taken as complete or wholly representative of the scale of human trafficking during this period. Recall: data are people. What these maps do afford is a unique perspective on the forced migrations of enslaved captives alongside when and where news regarding collective enslaved resistance was printed in the South Carolina Gazette and Virginia Gazette.

Map of Transatlantic and Intercolonial Embarkation Locations for Voyages that Disembarked in South Carolina and the Referenced Locations of Collective Enslaved Resistance Printed in the South Carolina Gazette between 1732 and 1775

This map illustrates the violent trafficking of people from Africa and their descendants to the British North American colony of South Carolina between 1732 and 1775. The base map for this visualization and the following is a map do not display country borders or regions. This basemap as well as many others are freely available through the Natural Earth Vector GitHub repository. I have opted for a base map without country borders or regional divisions in order to not reinforce inaccurate representations colonial and postcolonial boundaries; the resourse African Regions provides a useful frame of reference and discussion from seminal digital humanists, offering a window into recent scholarship on the social, historical, and cultural complexities of spatial mapping and historical data. It is important to note that this map currently solely displays data by year because the data for specific dates that voyages began and ended are often incomplete (this is especially true of data regarding the date of enslaved captives purchase and embarkation from Africa and other Atlantic colonies). Using a heatmap, this map displays extant locations of transatlantic and intercolonial embarkation regions derived from the Slave Voyages database. The heatmap is weighted by the number of transatlantic and intercolonial voyages that embarked for ports in South Carolina, namely Charleston. This means that the lighter the visualization becomes (yellow) the greater the number of voyages concentrated in that area and the darker (purple) the fewer voyages. The red star markers highlight the locations mentioned in the South Carolina Gazette’s coverage of news related to enslaved resistance—including reports of occurred insurrections, conspiracy scares, legislation, editorials, news of rebels (both enslaved and self-liberated black communities), and shipboard uprisings. This map was generated using QGIS, the free and open-source counterpart to ArcGIS.

Interactive Map of Transatlantic and Intercolonial Slave Trades to South Carolina alongside Reports Refercing Collective Enslaved Resistance printed in the South Carolina Gazette in the Eighteenth Century

The blue dots on this map represent the principal place of purchase of enslaved people on voyages that eventually disembarked in South Carolina (1732–1775). The red dots represent locations referenced in reports published in the South Carolina Gazette about enslaved resistance (insurrections, conspiracies, plots, uprisings, or revolts).

Map of Transatlantic and Intercolonial Embarkation Locations for Voyages that Disembarked in Virginia and the Referenced Locations of Collective Enslaved Resistance Printed in the Virginia Gazette between 1736 and 1775

This map illustrates the violent trafficking of people from Africa and their descendants to the British North American colony of Virginia between 1736 and 1775. It is important to note that this map currently solely displays data by year because the data for specific dates that voyages began and ended are often incomplete (this is especially true of data regarding the date of enslaved captives purchase and embarkation from Africa and other Atlantic colonies). Using a heatmap, this map displays extant locations of transatlantic and intercolonial embarkation regions derived from the Slave Voyages database. This map displays the same variables as the previous.

Interactive Map of Transatlantic and Intercolonial Slave Trades to Virginia alongside Reports Refercing Collective Enslaved Resistance printed in the Virginia Gazette in the Eighteenth Century

The blue dots on this map represent the principal place of purchase of enslaved people on voyages that eventually disembarked in Virginia (1736–1775). The red dots represent locations referenced in reports published in the Virginia Gazette about enslaved resistance (insurrections, conspiracies, plots, uprisings, or revolts).

As a preliminary case study, this project advances the idea that colonial grammar was iteratively deployed to rationalize and contain perceptions of black resistance, embedding it into the fabric of the colonial worldview. By analyzing these linguistic patterns at scale, my research reveals how language reduced the complexity of collective black resistance into a distorted and flattened phenomenon. This work reminds us that the iterative nature of language can obscure as much as it reveals, perpetuating power structures while shaping historical narratives.

The colonial newspapers I analyzed help us to pull back the layers of linguistic strategies designed to erase the complexity of black resistance and justify its containment as well as the perpetuation of racial slavery in the colonial period. While my work does not yet recover lost voices in the violence of the past, it exposes the socio-linguistic machinery that sought to silence them.