Tolerance and Prejudice: Digital Humanities and Sermons from 1750-1850

Jacob Honeycutt

For my digital humanities research project for the Religion and the Enlightenment class, I was tasked with examining published English-language sermons on the topic of tolerance and toleration between the years 1750 and 1850. Due to the nature of the English-speaking world at this time—which mostly consisted of the United Kingdom of Great Britain and Ireland and the the English American colonies which then became the United States of America, with a much smaller portion of material coming from places like Canada and Australia—virtually all of these sermons were preached by Protestant pastors. Having looked at quite a few Protestant sermons from England and the United States during this period and knowing that anti-Catholicism characterized nearly everything Protestant ministers preached on, including even the topics of liberty and tolerance, I added on another dimension to the research. In addition to finding sermons on tolerance between 1750 and 1850, I also sought out to find anti-Catholic sermons between 1750 and 1850.

To do this research, I utilized a few methods. First off, my materials came from, and were collected and cleaned in, Gale Digital Scholar Labs, a database and software produced by Gale and accessible through the Baylor Libraries. Gale Digital Scholar Labs allows one to identify historic primary sources and place them in collections, and for all of these sources, it provides OCR—i.e., an electronic transcript of the words in the document that works as a text file. In searching for my content sets, I used keywords like “tolerance” and “toleration” conjoined with “sermon” to find sermons on tolerance, and derogatory keywords like “popery,” “papist,” and “Romish” conjoined with “sermon” to find anti-Catholic sermons. From there, Gale Digital Scholar Labs also lets one create a pre-configuration that cleans the OCR before one creates a text file. I personally set up this pre-configuration so that it removed all punctuation, special characters, ASCII characters, numbers, and all non-body content. Additionally, a very helpful feature in this cleaning pre-configuration was Gale gives one the ability to remove “stopgap words,” which it provides a preset list for, including articles, prepositions, transition words, filler words, and other such words. This was very helpful as I was not looking for how frequently preachers used the word “and,” for example. After cleaning the texts, I then downloaded them and put them through the Jupyter notebook created by Baylor tech guru, Joshua Been. This notebook had the ability to do two forms of text data mining, keyword analysis and sentiment analysis, and I did both analyses for each of my content sets.

Doing this analysis resulted in several findings. First off, the keyword analyses confirmed the main thing that I expected to find coming into the project: that in these sermons, anti-Catholic rhetoric was closely tied to rhetoric around tolerance. In fact, in my sermons on tolerance set, anti-Catholic keywords such as “popery” came up frequently, while in my anti-Catholic sermons set, the keyword “tolerance” came up frequently. Moreover, the contrast in the political-related language used in the two sets was illuminating. Political republicanism, which tolerance, as a concept, was generally associated with,  In the sermons on tolerance set, positive words associated with political republicanism were shown as frequent, including “virtue,” “truth,” “reason,” “liberty,” and “conscience,” whereas in the anti-Catholic sermon set, political-related words were more dark and authoritarian in nature, such as  “power,” “authority,” “decrees,” “measures,” “persecution,” and “hierarchy.” Some of these terms are terms perhaps emendating out of the Enlightenment, like “reason,” for example,” but in my opinion, these terms—both the positive and negative ones—indicate more surely the influence of British Whig political philosophy. 

Other helpful findings were provided by doing sentiment analyses. One of the most helpful of these was the ability to determine the sentiment associated with each of the various locations discussed in the sermons. In the sermons on tolerance set, for example, the most frequent location mentioned with positive sentiment included “Rhode Island,” which makes sense because it was arguably the first place in the English-speaking world to allow freedom of (Protestant) religious expression. Other top locations included Maryland, which had, early on, provided tolerance to Catholics, and Holland, which had a reputation as a tolerant country. The only location with clear negative sentiment in this set was “Rome,” which, again, testifies to the connection between anti-Catholicism and discussions of tolerance—Roman Catholicism was seen as authoritarian and intolerant. 

Doing a sentiment analysis on the anti-Catholic sermons both showed locations that had positive and negative sentiments and revealed terms that were associated negatively in conjunction with Catholicism. However, one limitation here is that the locations that had positive sentiment in the anti-Catholic sermons seem to include too many locations that should have negative sentiment. The Protestant countries “Britain” and “Sweden” appeared high on the list, but after that, other locations with high sentiment included “France,” “Spain,” “Ireland,” and “Rome.” I am not sure why these locations would have high sentiment in sermons with anti-Catholic themes. However, the top negative sentiment locations were exactly what one might expect, with some results overlapping with the positive sentiment locations: “Rome,” “Rome,” “Rome,” “Hungary,” “France,” “Paris,” and “Ireland.” Additionally, looking at low sentiment language, aside from just locations, used in the anti-Catholic sermons shows some prominent themes that Protestant ministers employed in their critique of Catholicism. These words include: “veneration,” “reliquies,” “mediators,” “popish,” “purgatory,” “power,” “pardon,” and “priest.” So, the sentiment analysis of anti-Catholic sermons revealed exactly what one might expect—Catholic countries, and especially Rome, were viewed negatively, and Protestant ministers still, in 1750-1850, had issue with some of Catholic doctrines and practices that Protestants had always had issue with.

Overall, this digital humanities project was illuminating and helpful. I received results regarding sermons on tolerance and anti-Catholic sermons, as well as the relationship between the two of those sets, that were consistent with what I expected as someone with prior knowledge on Protestant sermons between 1750 and 1850. However, in my analysis, the most useful function of the digital humanities tools I used—keyword analysis and sentiment analysis through a Jupyter notebook—is the ability to now attach certainty, based on data, to ideas that previously were educated hunches. Any historian seeking to give more solid proof for the claims they are making should consider using these digital humanities tools to do so.