Recovering archival specters with digital historical methods

How might historians use digital research methods to study archival specters – people who lived, breathed, and made their mark on history, but whose presence in the archives and extant documents remains limited, at best, if not altogether lost? A number of scholars have written cogently and movingly about power and archival silences, and a growing number of tutorials provide pragmatic guidance on the use of computational techniques, but few studies exemplify how digital research methods may be used to address archival voids, nor do they model a complete research life cycle from sources to data to quantitative, qualitative, and visual analyses. Feminist literary theory, postcolonial studies, and critical information studies provide theoretical frameworks with which to “listen to the silences,” and read archives “against the grain”. Now, developments in topic modeling, text mining, data visualization, as well as statistical and social network analysis offer practical methods to investigate latent patterns in our available source materials. Therefore, this study responds to these needs by demonstrating how digital research methods may be used to explore the ways in which ethnicity, gender, and kinship shaped early modern Algerian society and politics. The approaches presented in this study have applications far beyond English, French and Arabic language sources and the history of the Middle East and North Africa. More broadly, these methods will be of use to scholars interested in identifying and studying relational data, demographics, politics, discourse, authorial bias, and social networks of both known, as well as unnamed, actors. Digital tools cannot metaphorically resurrect the dead nor fill archival gaps, but they can help us excavate the people-shaped outlines of those who might have filled these spaces.

The few extant fragments of information emerge from European and American travel accounts, consular records, nineteenth-century French scholarship, as well as French and Algerian chronicles of the provincial governors of Constantine, Algeria. The original sources are in French, English, and Arabic; the data mined from them have been translated into English for comparative and presentation purposes. Through the application of topic modeling, statistical hypothesis testing, text mining, data visualization, and network analysis, this project demonstrates how computational techniques may be used to read both colonial and Algerian sources “with” and “against the grain,” in order to foreground the experiences of some of history’s most marginalized people.

Quantitative approaches are not entirely new to historians who embraced them in the movement toward social history in the 1960s. However, by the 1980s, many scholars had largely concluded that computation had little to offer the field. With increasingly sophisticated computational methodologies that are now readily available and more accessible, I argue that it is time to revisit this assumption. Using documents related to Ottoman Algeria as a case study, I show how new digital research techniques have much to offer humanists, especially when sources are scant, difficult to find, and even more challenging to access.