A Computation Analysis of Character Archetypes in the Works of Calderón de la Barca

The Siglo de Oro period of Spanish theater was marked by a rapid increase of production of theatrical pieces. These plays used clear patterns, notably for genre, plot and character even though there were no explicit drama conventions. Our study aims at characterizing such conventions in the work of Calderón de la Barca. We investigate the portrayal of character archetypes based on character speech, adopting a scalable reading approach that employs machine learning to aggregate empirical evidence from a large number of Calderón's works. Concretely, we develop a neural network model to predict the six main character archetypes. We analyze the predictions of this model, inspect the lexical material that determines these predictions, and visualize the representations that the model learns. We find that the model predicts character archetypes with some accuracy (f1 = 0.47), and that some character archetypes like criadas, are more standardized than others, like reyes.