The Autocrats' Speech: Exploring Xi Jinping, Putin, and Kim Jong Un Addresses

• In Event: Chinese Politics under Xi Jinping's Rule

Abstract

Sometimes, the delivery of a statement can be more meaningful than its content. While many studies of political speeches mainly use transcripts and conduct text analysis, we explore audio as an independent package of data vectors. Specifically, we focus on audio features such as speaking rate, pitch, intensity, waveform, and amplitude as indicators of speaker confidence, sincerity, and conviction. Our analysis focuses on audio samples from authoritarian leaders who have been in power for extensive periods and whose speech is already the subject of extensive text-based analysis. Specifically, we examine a large corpus of audio speeches delivered by China’s Xi Jinping and Russia’s Vladimir Putin, and North Korea’s Kim Jong Un. How will the phonetic indicators of these speeches change with form, content, environment, and time? While the substance of their public speeches is known to be carefully edited and choreographed, we show that their deliveries reveal meaningful variation. In each of these cases, for instance, delivery speed has declined alongside consolidation of power. Likewise, we show that pitch and intensity vary in predictable ways depending on whether the content being delivered is scripted or improvised as well as whether the subject is uncomfortable for the speaker. However, the way they emphasize certain content varies, either by raising the pitch, increasing the intensity, or slowing the rate. We conclude with examples of how our approach to audio analysis can be paired with other research methods and data generating strategies. This study demonstrates the potential of using audio-as-data approach for cross-lingual political analysis.

Authors

• Dimitar D Gueorguiev, Syracuse University

• Minju Yeo

• Jun Zhang