What Do Pop Songs Reveal About Us? New Research at Dickinson Seeks to Find Out
Xiaolu Wang used an innovative method to analyze the emotional tone of the lyrics in more than 200,000 Chinese-language pop songs. His findings challenge what we think we know about pop lyrics and national mood. Photo by Dan Loh.
New research probes connections between lyrics and national mood
by MaryAlice Bitts-Jackson
What, if anything, can the emotional tone of pop-song lyrics tell us about society? A Dickinson professor is on the case.
Assistant Professor of International Business & Management Xiaolu Wang analyzed the lyrics in more than 260,000 Chinese-language pop songs, released across five decades. His study is the first large-scale sentiment analysis of Chinese pop lyrics and one of the first such studies to deploy AI technology. It challenges a prevailing idea about how directly emotions in pop-music lyrics might reflect shifts in national mood.
Wang published his findings in the Journal of Cultural Analytics. He also co-published an article about his innovative use of AI and lexicon-based analysis in the International Journal of Digital Humanities. That article was co-written by Evan Wong ’24, a former Dickinson computer-science major now working at Deloitte.
A personal connection
The sentiments in Western pop lyrics have been analyzed in previous studies. One research team measured the emotional tones in 232,574 English-language lyrics from 1960 to 2007 and observed that the emotions skewed more negative over time. The researchers postulated that this steadily lowering mood reflected a decline in public mental health.
Wang's focus on Chinese-language lyrics is informed by his past: He grew up in China at a time of sweeping sociopolitical changes, making Chinese-language pop songs from Taiwan and Hong Kong newly accessible. But his methodology is forward-facing. Originally, the professor intended to use conventional lexicon-based sentiment analysis, but when ChatGPT was released, he decided to combine a lexicon method with AI analysis.
Digging in
First, Wang extracted the lyrics of 264,851 songs from a Chinese-language music database. He then fed the lyrics into ChatGPT. The AI chatbot analyzed the text and assigned overarching emotions to each song—for example, sadness, determination or hope—and Wang validated the output with random manual checks.
Wang matched ChatGPT’s words to an emotion lexicon, mapping its valence (positive, negative, neutral) and intensity (1-5). He identified a strong, recurring 34–35-year emotional cycle and developed a physics-inspired mathematical model to analyze it.
General listeners’ average emotional preferences remained relatively stable across time, he found, with the average sentiment of lyrics oscillating around that stable mean. Wang likens this process to a pendulum swinging away from and toward the center.
The research reveals how emotional dynamics in Chinese-language lyrics are the result of an ongoing push-and-pull between creators and audiences in the pop market. Creators try to align the sentiments with listener preference, but sometimes, they overshoot their mark—hence the swings.
East-West divide
How, then, to explain the increasingly dampened mood and emotional complexity other researchers identified in the English-language market, post-1960s? These changes, after all, largely moved in one direction.
Wang argues that the emotional pendulum in English-language lyrics moves much more slowly back to the center because of differences between the English- and Chinese-language pop-music systems. Singer-songwriters dominate in the English-language pop market, and their work typically focuses on authentic personal expression. The Chinese-language pop-song market, on the other hand, is a “song factory” model characterized by an assembly line among composers, lyricists and singers. It therefore responds more sensitively to market feedback.
Wang also notes that larger trend lines often hide underlying waves, and he suggests that cyclic patterns in cultural sentiment may be more common than many researchers assume.
Today, Wang’s working with two Dickinson students to study astrology’s growing popularity in contemporary China. They’re using big-data methods to analyze vlogs, blogs and social media posts and to understand the broad sociocultural forces behind astrology's appeal. It’s another exciting opportunity to shed light on a cultural aspect of Wang’s homeland—and another opportunity for his student-assistants to sharpen their own research, coding and data-analysis skills.
TAKE THE NEXT STEPS
Published December 5, 2025