Sentiment analysis of popular song lyrics in the US from the 1990s and 2010s

Abstract

Popular music is significant in today’s society, reflecting our emotions and cultural trends. While nowadays listeners often focus mostly on melodies and rhythms of a song, they do not pay attention to the messages that these songs are trying to convey. From a linguistic point of view, analyzing the sentiments in popular music offers valuable insights into societal values, cultural shifts, and music preferences. This study aims to conduct a quantitative corpus study in order to even further broaden the field of using sentiment analysis to study song lyrics by looking at the popular songs in the United States from the 1990s and 2010s. In order to do so, the SentiStrength software was used with the goal of highlighting its capabilities in this field. A specialized corpus consisting of songs that had reached the number one spot on the Billboard Hot 100 during those decades was compiled for analysis of these sentiments. The research question for this study intended to understand how the results of the sentiments of both decades differed. The BA thesis is divided into three sections: the introduction, literature review, and empirical analysis. It begins with an introductory section that defines the term “popular music”, discusses the reasons for conducting this study, and provides an overview of the upcoming sections. The literature review focuses on defining sentiment analysis alongside its classifications and limitations, the importance of song lyrics and popular music, and discusses previous works on similar topics. The empirical analysis section focuses on the compilation of the song corpus, explains the nature and functionality of SentiStrength, followed by a presentation of the results of the study and finally a discussion section. The entire thesis ends with a conclusion which summarizes the key points brought out in the course of the study.