A State-Restricted Hidden Markov Model for Authorship Attribution of the Deutero-Pauline and Pastoral Epistles

Josiah Leinbach${}^a$, Xuwen Zhu${}^b$ and Shuchismita Sarkar${}^a$

${}^a$Bowling Green State University, ${}^b$The University of Alabama

The New Testament contains thirteen epistles attributed to the Apostle Paul, all of which were traditionally accepted as authentically Pauline by early Christian theologians. Since the 19th century, however, many scholars have questioned Paul’s authorship of certain epistles due to differences in vocabulary and writing style compared to the undisputed Pauline epistles. In particular, two clusters of epistles, known as the Deutero-Pauline Epistles (Ephesians, Colossians, and 2 Thessalonians) and the Pastoral Epistles (1 Timothy, 2 Timothy, and Titus) have been subject to the most doubt. This study presents a novel state-restricted hidden Markov model that constructs a constrained state space for the undisputed Pauline epistles and an unrestricted state space for other epistles. The model jointly analyzes all thirteen epistles and some additional biblical texts, employing a novel first-order Markov emission for transitions between parts of speech to classify sentences based on Pauline and non-Pauline style detection. Then, informed by New Testament scholarship, the result of the model is interpreted and the possibility of Pauline authorship for the Deutero-Pauline and Pastoral Epistles has been examined.

Keywords: hidden Markov model, stylometry, authorship attribution. (Use at most 3 keywords)