Gender in Shakespeare: Automatic Stylistic Analysis of Shakespeare's Characters

The problem of automatically determining the gender of the author of a literary document (i.e. fiction/non fiction), an electronic document (i.e. text collected from e-mail), an informal text (i.e. essays) etc. has been studied and have achieved reasonable accuracy in the range of 70-80%. These works have proven that noticeable differences exist, which discriminates author gender using simple lexical and syntactic features. Our aim here is to extend this research on automatically determining gender of literary characters (i.e. from plays) from the playwright’s word use. The earlier mentioned works and this work is different because real men and real women are involved as the authors of various forms of documents where as in plays, literary character’s gender is characterized by playwright’s word usage. Here we have proved Shakespeare used language differently for his male and female characters, and we have studied the top discriminating features from both gender characters. We used Sequential Minimal Optimization (SMO) for classification of gender character by importing their speeches. We used function words (FWs), Bag of Words (BoWs) as feature selectors and achieved 67%, 74% accuracy respectively. We observed style has played the role in discriminating gender of Shakespeare’s literary characters. We also found that, there is similarity with the previous research on classifying gender author of literary books (Argamon et al.2003).