“O bella ciao, bella ciao, bella ciao, ciao, ciao”

I grew up in the North-West of Italy, in a small town where the Alps meet the Apennines, and where, now over 50 years back, my grandfather, like many grandfathers of friends of mine, fought as a Partisan in World War II, in the civil conflict that split Italians in two factions, the Resistance and the National Fascist army. Being my area quite proud of its “rebellious” nature, I was raised singing chants from the resistance, a gentle reminder of our local heritage, a political statement on a conflict that I never witnessed. It made me think: music is eternal, timeless, but it is also often timely - we make music about things as they happen, and then these songs crystallize in time… or do we? Within the realm of music about war, do we make music during a war or after it happens? Are some wars more sung about in popular culture? If so, is it because they were more deadly? Or because they lasted longer? In the following analysis, I use data from several Wikipedia pages on anti-war music and on related conflicts to do an exploration of the factors that affect the creation of music about war.

Data Collection

The first step involved data collection from Wikipedia´s list of Anti-War songs . From the relevant lists, I only retained those that had a meaningful sample of songs (>1). Following some cleaning and wrangling, we got a dataframe containing 1054 songs, with information about the song title, artist, release date, and related conflict. We can start looking at the count of songs per each list in the Wikipedia page.

The second step in the analysis consisted in obtaining war-related information. Particularly, I was interested in the duration of the conflicts, death tolls, date of the conflicts, location, and combatants. A suitable place to obtain such data was again Wikipedia´s list of wars by death toll , where I assumed I would have been able to find most of the wars in the previous dataframe. Luckily, most wars were contained in a table called “modern”, thus I was able to quickly obtain data from the rows and columns that interested me. For all those I was unable to find in this page, I manually googled the information and inserted into the new dataframe on information about wars. Some conflicts were not so easily coded: for instance, my anti-war songs list puts the Yugoslav War as one category, when de-facto it is comprised of several smaller conflicts, with individual dates and death tolls. As a result, I summed information of the smaller conflicts to obtain a matching category. Same was done for the sub-list “9/11, Gulf War, Iraq War and the War on Terror”. I considered splitting the song list, but that would have inolved manually coding which conflict each song in the list related to, with potential overlaps. Again, some data wrangling and cleaning was required.

Once I obtained my two datasets, I merged them using the war name as a key id.

Data Analysis

The analysis was run in R, and visualisations were created through several packages, from the well-known ggplot2, to streamgraph and plotly to incorporate some interactivity on the output page.

The first plot is a plotly html embedded graph that shows the linerange of when the war took place, and distributes songs around the timeline as they got released. By hovering over the dots, information about the song is revealed.

From this graph, we can see that the patterns are quite different for each conflict. For example, the Yugoslav War had most music about it made during the years of the conflict, whereas on the opposite side of the spectrum the American Civil War became popular again in music nearly 100 years after it happened, suddenly revamping. Similarly, music about World War I and World War II has been made during, and well long after the conflicts ended, perhaps indicating them as the most “timeless” conflicts of them all. Clusters of songs can be seen around the Cold War and the Vietnam War, indicating a lot of music about them was made within the same short period of time.

Here below instead we can look at the conflict´s size and location and the number of songs dedicated it. Global conflicts such at WWI and WWII have been sung about more than conflicts that took place in just some countries.

Finally, I have included some interactive visualizations of the music-making overtime. The first one is a plot of how many songs were made daily per each conflict. The plot was made using R´s interactive package streamgraph. It shows that some conflicts have dominated some decades more than others: particularly the Vietnam War, the Cold War and the more recent Iraq War, Gulf War, War on Terror and 9/11. Following is a plot showing the cumulative version of this plot, again visualised through streamgraph. I here included name labels of the conflicts to provide an indication of when the conflicts happened. It shows mainly that most music included in the Wikipedia list is released from the 1960s onwards, despite some conflicts taking place much earlier. Going back to our original question, is music made during or after a war? If we were to really trust our data, we would conclude that there has been a boom in anti-war music-making following the sixties, and we could speculate that perhaps jazz, blues and classical music were unsuited genres to speak about war? Could the rise of rock, folk, punk, disco, hip-hop and pop have lead to more topical music, more political music? I guess this can be the foundational question for an extension of this project.

Yearly total of songs per conflict

Cumulative number of songs per conflict

This piece was originally published as an assignment within the Computational Social Science Master Program at the Institute for Analytical Sociology, Linköping University, Sweden.