Contrasting Playwrights through Sonic Signatures

Well-written characters in drama and literature are often imbued with distinctive voices through differences in word choice, sentence structure, and even the sound of their words. In past work, we have shown that it is possible to detect phonetic differences between classes of Shakespeare characters (e.g., protagonists, antagonists) computationally, using Naive Bayes classifiers. In this talk, we will discuss our early attempt to use such classifiers to draw contrast between different playwrights. We lay out a corpus of early modern playwrights on an axis measuring the predictive accuracy of our model. Additionally, we present visual tools for identifying which phonemes are most distinctive for different characters and authors.