Analysing and comparing lyrical trends between male-led, female-led, and mixed group songs
Song Lyrics and Gender
Research Question
How does the lyrical sentiment differ between male, female, and mixed group artists in the top 50 most popular Billboard songs in the US from 1960-2021?
Abstract
Music is universally celebrated and has a strong emotional presence in our lives. Its ubiquitous importance has inspired us to explore how lyrics express certain sentiments. By conducting comprehensive quantitative analyses and presentation of lyrical data, we have aimed to produce an interdisciplinary and relevant research investigation. To form our dataset, we used information from Billboard’s "The Hot 100" Songs, the Make Music Equal Pronoun & Gender Dataset, Genius, and Spotify. For clarity and concision, the data is presented in the form of a variety of graphical visualisations. We found that songs tend to have extremely positive or extremely negative sentiments in all emotional categories, and the pattern of these sentiments did not differ significantly between male, female, and mixed artist songs. The findings from our project led us to accept our null hypothesis: male and female-led songs do not differ in their lyrical sentiment.
This research can provide some insight into better understanding the influence of gender on linguistic expression through music. The data and analysis of this project are also relevant in uncovering the evolution of language and emotional expression in popular music. A detailed exploration of this could be an interesting next step for deeper historical analysis.
Aims
-
Identify and quantify any differences in the lyrical sentiments expressed by the most successful male and female artists
​
-
Use the Natural Language Toolkit (NLTK) to conduct quantitative research into a field that is usually analysed qualitatively, and expose people to NLTK techniques
Objectives
-
Find key differences between the sentiments most commonly expressed by women and men
​
-
Investigate patterns related to other lyrical elements, including word repetition, song length, filler to emotive word ratio, and use of unique words
​
-
Ensure code is user-friendly by adding comments
​
-
Present findings clearly and with as little bias as possible
Our research will provide a way to compare the lyrical sentiment between gender and genres, in addition to showing the changes over time. We will display the techniques and key elements of our code, but we will also go a step further than previous projects to show exactly how the sentiment analysis differentiates between positive, neutral, and negative lyrics. The NRC lexicon will be used, which classifies words according to their assigned sentiment and emotion. Essentially, we not only want to perform analysis on the overall lyrical sentiment, but also illuminate the processes of the Natural Language Toolkit.
OR
scroll to get to know us...
Get to know us
We are a group of 2nd year BASc undergraduates at UCL, working together to explore the relation between song lyrics and gender as part of our Quantitative Methods: Data Science and Visualisation module assessment.
​
​
YVETTE HOMERLEIN
"Hi! I'm Yvette, I'm a BASc Cultures major focussing on psychology, English literature, and architecture. For this project, I researched similar data science projects in the field, organised the project schedule, created the graph visualisations, and wrote the accompanying analysis."
MIA SHEPHERD
"Hey, I'm Mia, I'm a BASc Health and Environment major. I am particularly interested in biochemistry and neurobiology. I was responsible for the contextual elements of the project: the introduction, literature review, limitations, conclusion."
CHISTIAN ILUBE
"My name is Christian, and I major in Sciences and Engineering and minor in Cultures. In this project, I decided which sentiment analysis functions we would use and wrote a program to display lyrics with individual words colour-coded, which I used to assess the strengths and weaknesses of this approach."
ADAM PLEWCZYSKI
"Hi, I'm Adam, I major in Sciences and Engineering. In this project, my main responsibility was to extract and process the initial dataset using various programming techniques and to prepare it for further analysis."
SAUMYA SONI
"Hey! I'm Saumya, I'm a BASc Health and Environment major, in which I focus primarily on psychology and health related modules. My focus on this project was website design and structure, compiling the text, and writing the abstract. "