"Talk Like a Woman": Applying Sentiment Analysis to Convention Speeches

Since the first woman was elected to the United States Congress in 1917, scholars have inquired about the behavior of women in politics, focusing on legislative priorities and political relationships. Some theorized that men and women have different political "voices", but the question remained unanswered for decades due to the difficulty in quantifying speech. However, the recent emergence of sentiment analysis technology finally allows for the computational study of speech. Sentimentanalysis tools gather text data and match words with their corresponding tones, granting us a new ability to discern emotion from large amounts of speech data. In this project, I apply sentiment analysis to 92 United States party convention speeches from 1992-2020. I find that, while women's speeches are marginally more positive and joyful than men's speeches, the differences end there. Men and women were equally likely to invoke anger, anticipation, disgust, fear, sadness, surprise, and trust in their speeches. My findings indicate that men and women are equally emotional speakers, and prompt a larger discussion about the perceptions of women's speech in political settings.Plan II Honors Progra