What if lyrics shape a song’s success more than we realize? Analyzing 176 entries from the Bundesvision Song Contest, this study reveals how emotional arcs in lyrics—especially darker ones—independently influence audience votes. By blending narrative analysis with audio data, it uncovers how storytelling quietly steers musical impact and cultural decisions.
Have you ever experienced a moment like this? A song starts, and in just a few lines, you’re pulled into its world. The lyrics draw you in, and the melody carries you away, until it feels less like music and more like a story unfolding. Joy, heartbreak, and longing – sometimes even emotions we can’t quite name.
But does every song tell a story? Do lyrics truly convey a narrative, or do we simply imagine one where none exists? And if songs do carry stories, do those stories shape how we respond to them, maybe even deciding which songs grab our attention or which artists win competitions?
While narrative structures have been studied in books and films, music has often been overlooked in this discussion. However, with the advent of natural language processing and new text analysis methods, it is now possible to explore lyrics in ways that were previously impossible. Researchers have already shown that the narrative structure of novels and movies affects their success. Could the same be true for songs?
Why This Study
This study adopts a narrative-analytical approach to music. To accomplish this, we examine the Bundesvision Song Contest (BSC), Germany’s equivalent of the Eurovision Song Contest. The BSC offers a valuable opportunity to analyze how lyrics influence audience decisions. From 2005 to 2015, sixteen songs, each representing a German federal state, competed annually, with viewers awarding points. Since most of the songs were performed in German, the lyrics were easily understood by the voting audience. This makes the BSC an ideal setting to separate the role of words from that of music and to investigate how song narratives might matter.
How We Looked at It
Our analysis covers all 176 entries from 2005 to 2015. We combined two types of data:
- Audio features from Spotify, such as energy, danceability, and speechiness to control for musical features.
- Emotional arcs of lyrics classified using narrative analysis. Each song was assigned to one of several emotional archetypes and classified by its ending — either a ‘happy ending’ or a ‘sad ending’.
By combining these datasets, we could distinguish the effects of lyrical content from musical qualities. We then used panel models to examine how the two aspects of sound and story related to the points awarded, while controlling for factors like song duration, genre, running order, and home-state bias.
What We Found
Lyrics matter. Certain emotional arcs consistently correlated with higher scores, even when we considered musical features such as energy and danceability. Perhaps most notably, songs with negative emotional trajectories, those that shift through sadness or melancholy, often performed better. This pattern reflects the so-called ‘paradox of negative emotions’, the idea that darker themes can be more engaging and memorable than purely positive ones. Of course, audio features still played an important role in shaping audience response. However, the lyrical narrative added its own independent layer of influence. While no single emotional archetype guaranteed success, the findings suggest that, in a contest setting, some emotional arcs resonate more strongly with audiences than others.
Why It Matters
- For artists and producers, the findings highlight that crafting a song’s emotional arc is not just an artistic choice; it can influence how an audience responds.
- For cultural economists, it shows that lyrics are not mere embellishment but a factor with measurable economic relevance.
- In the context of media and branding, it highlights the potential of the interplay of music and narrative as a tool for shaping perception and engagement.
- More broadly, in the field of narrative economics, these findings show that cultural content (here, song lyrics) can shape decisions in ways that go beyond immediate entertainment value.
Outlook
The dynamics we observed in the BSC are likely just as relevant on today’s digital platforms, where songs are discovered, shared, and judged within seconds. Even short-form media, like TikTok, condense storytelling into snippets, yet emotional arcs can still be felt and recognized.
This raises both creative and ethical questions: if we know that certain narrative structures can influence audiences, should we intentionally design content to do so? The challenge for cultural economists and media researchers is to continue exploring how art and narrative intersect with economic behavior, and how these interactions evolve as the media landscape changes.
About the article
Rösch, J., Rauch, MJ. ‘Do songs tell stories?’ An empirical analysis of the effect of emotional arcs on success in a national song contest. J Cult Econ 49, 603–637 (2025). https://doi.org/10.1007/s10824-025-09532-6
About the authors
Jürgen Rösch is a junior professor of Digital Economies in the Media Management Department of the Faculty of Media at the Bauhaus University in Weimar
Maxi-Josephine Rauch is a research assistant at Bauhaus-Universität Weimar
About the image
Allegory of Music MET 50.189 1, Laurent de La Hyre, CC0, via Wikimedia Commons
