The Featured Researcher for January 2025 was Ash Charlton, with a project titled, ‘Slavery and Race in the Encyclopaedia Britannica (1768-1860): A Text Mining Approach’
HEIs: University of Edinburgh, School of History, Classics & Archaeology
Supervisors: Professor Diana Paton & Professor Melissa Terras (University of Edinburgh); Dr Robert Betteridge & Dr Sarah Ames (National Library of Scotland)
My research is concerned with identifying and exploring legacies of race and slavery in the early Encyclopaedia Britannica (EB) through the first eight editions from 1768-1860. The EB was first published at the height of the transatlantic slave trade, and this early period covers rising abolitionist sentiment and the abolition of slavery in Britain. These editions contain over 100,000 pages of text which would be almost impossible to search by hand, so I use digital methods to find relevant information points across the editions, identify connections, and crucially see where there are gaps and missing information on slavery.
Finding and making datasets
The most important thing starting my research was to make sure I had a digital textual dataset that I could use. My research is in collaboration with the National Library of Scotland (NLS), who have openly available datasets of images and text of the first eight editions. Their version of the texts contained lots of unpredictable errors through the text recognition process, so I decided to create my own version using digitised images of EB from their Data Foundry. I transcribed and trained a custom AI-based recognition model with the automatic text recognition platform Transkribus to create higher quality versions of the text. My research also incorporates an openly available digital version of the seventh edition from the Nineteenth Century Knowledge Project.
Using digital methods to examine slavery
Once I had my datasets, the next question was what do I do with them? While there are many digital methods to choose from, I used some preliminary keyword searches to establish which words relating to slavery were used through the editions and how many times. These varied, from the obvious words like ‘slaves’ and ‘slavery’ to places associated with the transatlantic slave trade, goods produced with enslaved labour like sugar, tobacco and cotton. From there I realised that working out which encyclopaedia articles mentioned these words and made links to slavery was crucial for understanding how the topic was presented. Based on this I decided to use network analysis to visually map which articles contained the keywords and lead my focus in comparing the texts and how slavery was discussed (or not) through editions. This has helped me identify where there are allusions to slavery without explicit discussions in the text alongside the more obvious mentions to form a larger picture of how the topic was written about, and how this changed over time.
My research is the first exploring slavery across the early EB editions, and digital methods have helped me to identify and visualise information that would be impossible to find otherwise. My work also aims to provide a framework that can be used to research other topics comprehensively in EB, or to be used on similarly structured reference publications such as encyclopaedias and dictionaries.
