Analyzing Antislavery Texts

As part of my dissertation research into how and why antislavery writers took up racialist ideas, I studied eighteenth-century antislavery texts from the Delaware Valley, especially Quaker ones, and found a shift from purely religious arguments to those including Enlightenment discourse in the 1750s. While this addition of Enlightenment language is not by itself a novel finding, I used statistical text analysis to demonstrate the magnitude of this shift, giving it greater support than qualitative methods alone.

As the first colony-turned-state to legislate gradual slave emancipation in 1780, historians have described Pennsylvania as a crucial step in abolishing slavery in the United States, Britain, and beyond. Pennsylvania Quakers are generally regarded as playing a crucial role in developing the area’s antislavery sentiment, and transatlantic Quaker networks contributed to spreading these feelings and the rhetoric used to argue for them.

Questions remain, however, about how widespread Quaker antislavery sentiment was, when and about how and when their rhetoric developed. While the impact of individuals such as John Woolman and Anthony Benezet have long been acknowledged, Brycchan Carey has argued that the debates within the larger Quaker community played a significant role in developing their collective antislavery rhetoric over time.

Using statistic analysis on eighteenth-century antislavery texts from Pennsylvania and the Delaware Valley revealed a quantifiable shift to supporting existing religious arguments with newer Enlightenment ones in the mid-1750s. Specifically, texts can be compared by examining the different frequency of words appearing in them. In the principal component analysis (PCA) graph to the right, texts closer together share similar word frequencies, while those further apart differ more. You’ll note that texts before 1755 cluster to the left, while later texts are all on the right.

I’ve also overlayed the relative “pull” of the differing frequencies of different words (the red arrows) to reveal why texts differ. The words whose differing rates of appearance had the greatest impact on differentiating the texts were “every”, “slavery”, “liberty”, “love”, “lord”, “upon”, “god”, “spirit”, “colonies”, and “trade”. The more religious terms, such as “god”, “love”, and “spirit”, “pulled” pre-1755 texts to the left, while more observational language, such as “slavery”, “liberty”, and “colonies”, “pulled” post-1755 texts to the right. I interpret this to mean that later texts included more observational, Enlightenment-inspired language related to the actual conditions of enslaved people in addition to the earlier arguments about the sinfulness of “manstealing.”

These findings are still preliminary and more work is needed to check whether this shift in antislavery discourse is also present in other kinds of writing. Nevertheless, such a measurable shift in discourse suggests a quick and radical change took place in the 1750s.

Principal component analysis (PCA) graph of eighteenth-century antislavery texts from the Delaware Valley.
Omohundro Institute 2017 conference presentation