According to research led by Pompeu Fabra University (UPF), sexism in the lyrics of the most popular songs in Spain has increased significantly over the last 20 years and especially in the last decade. This paper examines sexism in contemporary music in Spain by analysing the lyrics of more than 2,000 songs from between 1960 and 2022. 51% of the songs analysed contain lyrics with sexist expressions. It is a pioneering scientific study, due to the use of content analysis tools based on artificial intelligence techniques.

The study was recently published in an article inthe journal Cogent Arts & Humanities. The main author of the article is Laura Casanovas-Buliart and this study is the result of her end-of-degree project, supervised by Carlos Castillo, director of the Web Science and Social Computing (WSSC) group of the UPF Department of Engineering. The article is co-authored by Priscila Álvarez-Cueva, who was a Juan de la Cierva researcher at the University of Barcelona (UB) until the end of 2024 and is currently a lecturer at the Faculty of Communication of the Autonomous University of Barcelona (UAB). 

Laura Casanovas (UPF): “The results are shocking, as they reveal that although there have been advances in equality and the feminist struggle, the lyrics of many songs continue to perpetuate harmful stereotypes"

Regarding the conclusions of the study, Laura Casanovas (UPF) states: “The results are shocking, as they reveal that although there have been advances in equality and the feminist struggle, the lyrics of many songs continue to perpetuate harmful stereotypes. But, above all, I think this is a study that shows that artificial intelligence can be a great tool to analyse a large amount of data rigorously, and can be key to continue investigating and addressing this problem and other, similar ones in the future”. Carlos Castillo (ICREA and UPF) points out that the increase in chauvinism in songs in recent years is manifested above all in the "hypersexualization and objectification of women's bodies or through ideas related to possession and control by men".

The study pinpoints several causes of the increase in sexist discourses in the songs analysed. On the one hand, reference is made to the influence of the social and historical context on artistic production, considering that music reflects the values of the cultural framework in which it is inserted. The study recalls that Spanish society has not yet detached itself from the historical legacy of traditional gender stereotypes or the scourge of male violence, despite recent social and political advances in equality and the rise of the feminist movement, especially since 2018.

Beyond the persistence of traditional gender roles, the study points to another possible cause of sexism in song lyrics: the growing replacement of the radio by streaming platforms as the main channel for music consumption. On these platforms, no filter or selection criteria applied by those responsible for the radio’s musical spaces. Instead, all that is prioritized is related to popularity, which can have counterproductive effects. Platforms’ algorithms -which position what they offer users based on their level of consumption- also places these types of songs higher up in their lists. According to the research, this generates a “feedback loop”.

An automatic system has been trained to detect sexism in songs from manually tagged data

To conduct this study, one of the main challenges has been to examine such a large sample of songs (more than 2,000) almost without having previous data concerning the lyrics of these songs tagged for their level of sexism. Facing this challenge, the researchers have created a new method to make it viable and feasible to examine large volumes of data in a reasonable period of time. They have trained a computational model to automatically detect sexism in songs, based on manually tagged data, using machine learning techniques and artificial intelligence. To train the system, manually tagged data from a crowdsourcing initiative has been used, a kind of data that will need to continue to be expanded in the future to hone the current processing system. Thus, the research has combined human and machine intelligence.

Priscila Álvarez-Cueva, study co-author: “the research is also an opportunity for social science studies because it includes data processing as a methodological tool"

Priscila Álvarez-Cueva, study co-author, adds: “the research is also an opportunity for social science studies because it includes data processing as a methodological tool, which contributes significantly to the analysis of large amounts of data. This combination of disciplines is noteworthy for future studies that embrace the analysis of cultural production as a battlefield, deal with the context and existing power relations, and continue to delve into the problem of sexism and the objectification of the body, especially of women, but also from an intersectional approach”. 

According to Carlos Castillo (ICREA and UPF), "we could ask streaming platforms not to give as much visibility to songs that promote sexist behaviour", just as we make similar requests of social networks

The results of the research and the creation of the new data processing system can help to detect and monitor sexist biases in songs more easily and effectively. Carlos Castillo (ICREA and UPF) concludes: “Just as we ask social media platforms to reduce the visibility of radical or xenophobic content, we could ask streaming platforms not to give as much visibility to songs that promote chauvinistic behaviour”.

Reference article:

Casanovas-Buliart, L., Alvarez-Cueva, P., & Castillo, C. (2024). Evolution over 62 years: an analysis of sexism in the lyrics of the most-listened-to songs in Spain. Cogent Arts & Humanities, 11(1). https://doi.org/10.1080/23311983.2024.2436723