Skip to content

Overview

The Protestant Reformation, or Reformation, is addressing the 16th century schism within Western Christianity. The movement was initiated by Martin Luther and supported by John Calvin and other early Protestant thinkers. This event is ideologically defined by an intense objection to the policies, rituals, and ecclesiastical organization of the Roman Catholic Church which resulted in the creation of new denominations of Protestantism that spread throughout Europe and eventually the rest of the world. This project focuses on the manifestation of the Protestant Reformation in England. Here, under the rule of Henry VIII, Reformation ideals inspired the establishment the Church of England with the monarch as God’s deputy on earth. Our focus is between the years of 1509-1649, encapsulating the rule of Henry VIII to Charles I. The project ends with the rule of Charles I, as the English Civil War prompted debates about the finite details of Protestantism, since many different Protestant sects began to rise and gain traction in public thought.

Naive Bayes Classification

We started our study with the following question: is change in religious thought evident in the rhetoric of Crown Acts and Declarations during the years of the English Reformation and is this sentiment traceable? We wanted to see if we could trace religious change across different monarchs, ranging from Henry VIII to Charles I. To do so we used NaiveBayesClassifier from the textblob library in Python in order to classify whether or not each sentence in each document was Catholic, neutral, Protestant in sentiment on a scale from -1 to 1. It is apparent that Protestant sentiment increased over time, but took a steep decline under Mary Tudor during the Catholic counter reformation, as expected. This analysis was only performed on our INIT subset of 25 crown issued religious documents. With more data, our accuracy might be increased. With optimization our model's accuracy was approximately 87%.

Tracing protestant sentiment over time
NaiveBayes Classification algorithm on crown-issued religious documents from Henry VIII to Charles I on predicting Protestant sentiment.

We were also interested in looking at deference to the crown and how that changed over time. We define deference to be rhetoric where the crown commands total authority over its people. The same methodology was followed as above. If we consider the gaussian-smoothed curve, there is a spike in deference during Mary's reign, during the Catholic counter-reformation. This model's accuracy was 93%.

protestant-crowndef
NaiveBayes classification algorithm on crown-issued documents from Henry VIII to Charles I on predicting Protestant sentiment (orange) and crown deference (red). The graph is smoothed to account for variance, which are both shown in gray.

 

Important Features (SVM)

Using the SVM implementation from sklearn in Python, we created a model to show the most important features, that is, prominent words, to classify each subset of CCP data (crown, church, people). The following graphs show the weight of different features on our model. Note: these texts do not account for variations in spelling.

svm-church
Most important features for church class in CCP texts.
svm-crown
Most important features for crown class in CCP texts.
ccp-people
Most important features for people class in CCP texts.

 

Word Vectorization

Next we wanted to focus on rhetoric. That is, we wanted to trace specific words and their relationship through time. To do so we considered the Word2Vec model using the gensim package in Python with two different corpuses of data. We chose to plot two different relationships, 'god' and 'church' (for CCP and EEBO) and 'predestination' and 'church' (for EEBO). These CCP and EEBO notebooks allow you to input your own terms.

God & Church

“God” is perhaps one of the most important terms in religion. The vectorized distance between “God” and “church” gives insight into a very complex relationship. These graphs seem to indicate there is a separation between God and church. This may be indicative of how the English church was used as a method of the state, with rhetoric implying the church may not be as religious as it claims.

god-church-eebo
Vectorized distance between 'God' and 'church' in EEBO texts.
god-church-ccp
Vectorized distance between 'God' and 'church' in CCP texts. The lack of complexity in the graph shows the impact of a smaller dataset.

 

Predestination & Church

The term “Predestination” was chosen because of its complex development and role within Protestant doctrine. Predestination is defined as the divine foreordaining of God’s people. This idea, which originated as part of Calvin’s teachings, was widely accepted and preached as an early doctrine of the Church of England. However, this was not an idea that was publicly preached by the Catholic church in England. Thus, there is evident change in the distance between the terms “predestination” and “church” when the monarch of England was Catholic.

predest-church-eebo
Vectorized distance between 'predestination' and 'church' in EEBO documents. This was not performed on CCP documents because the word 'predestination' did not occur in the texts of every monarch.
bar-diff-eebo
A different representation of vectorized distance between 'predestination' and 'church' over time, by monarch, in our EEBO texts.

 

Interactive Scatterplots

Using our w2v models we were able to create interactive graphs where the user can choose terms of interest, a chosen model, and see a scatter of words in vector space. This can be interesting in that one can study a variety of topics. In particular, we were interested in looking at communion. The sacrament of Communion, or the Eucharist, holds different meaning the Protestant and Catholic church. In the Catholic church, the doctrine of transubstantiation, or the actualization of Jesus Christ’s body and blood in the Eucharist, is upheld. In the Protestant faith, the bread and wine of communion act as representations of the body and blood of Christ, in a hope to strengthen one’s spirit. The meaning of the ritual of communion was dependent on the religious sentiment of the monarch and thus, the country. By interacting with the scatter graph we could visualize the change in terms in vector space that related to communion.

Here is the notebook to look at EEBO models by year; here is the notebook to build EEBO w2v models by year.

Here is the notebook to look at EEBO models by monarch; here is the notebook to build EEBO w2v models by monarch.