Man, I feel like a woman! Understanding the positioning of women in music

How lyrics from the top female artists have changed throughout the years.

Back in 1997 Shania Twein sang “Man I Feel Like A Woman!” …but what exactly does it mean to feel like a woman?

At Musixmatch we are on a mission to enhance the experience of listening to music and we know that words matter.

So on the occasion of International Women’s Day, we’ve decided to analyse the lyrics from the most famous songs by female artists from each decade, starting from the 60s until now. The aim? Understand the changes of the women in music through their words.

Using a data sample of more than 150 English lyrics, we created a profile for each decade, showing the evolution of the writing style of the female protagonists that have marked, the history of music, one word at a time.

Covering a broad variety of genres and singing styles, our data includes lyrics by artists such as the unsurpassed eclectic mind of Nina Simone, the Queen of Pop Madonna, Blondie, Tracy Chapman and also current stars like Adele, Doja Cat, Lady Gaga and Cardi B.

The Method

Understanding of a written text, by machines, as well as the Information Retrieval from it, remain widely studied sub-problems of Natural Language Processing.

For this analysis we used a combination of both traditional Natural Language Processing techniques, together with the latest, more advanced Deep Language Models which enable computers to understand and simulate human language by exploiting a large amount of textual data.

Our intelligent models are tuned on lyrics specifically and are trained to solve a wide range of lyrics-related tasks.

As a result, we have created a unique pipeline for processing lyrics: we start from the raw text and extract all kinds of higher-level information, in the form of lyrics-based music metadata such as the one we are taking into consideration in the following paragraphs.

The Analysis Parameters

The analysis is based on a set of 10 different parameters, for each of which we have examined the variation throughout the different decades.

1. Structure Factor

It’s a value that accounts for the complexity of the structure of the lyrics.

This parameter considers the basic song parts such as verse, chorus, and bridge up until extra parts such as pre-chorus, intro and outro, and hooks.

As the below radar chart shows, while the song structure seems to have been quite ordinary up until the 80s, with a slight increase in its complexity between the 90s and the 2000s, the big difference has been recorded in the last decade.

This particular change can be directly related to the creation and rise of music streaming services. Through the years these services have changed the way we traditionally listened to music and music has adapted to meet the needs of the listeners, renewing the structure to catch the listener’s attention from start to finish.

2. Rhyming Factor

It is an indicator of the average number of rhymes in the lyrics.

As we can see from the graph, the rhyming factor has increased over time, with higher peaks in the 90s and the last decade (2010–2020), while we record the lowest value during the 2000s.

The 90s were a decade characterized by the presence of different music genres, with pop artists such as Britney Spears and Christina Aguilera sharing the charts with rock bands such as The Cranberries and The Cardigans and experimental musicians like Björk. The greater rhyming factor can be considered as a consequence of the genre variety that distinguished these years.

Meanwhile, the higher value observed in the last decade is mainly due to a greater presence of women in a male-dominated music genre: RAP.
2020 especially was a year that saw a high number of female rappers such as Cardi B, Doja Cat, and Meghan Thee Stallion dominating the mainstream music and registering top positions in multiple charts around the world.

3. Vocabulary

The aim of this section is to analyze the complexity of the vocabulary used in the lyrics and it groups the following parameters:

• Vocabulary Size: considers the total number of unique words used in the lyrics of a decade.
• Lyrics Length: number of words contained in the lyrics.
• Word Rareness: a value that represents the rareness of a word in the English language, based on the usage of the words within written texts.

By observing the graphs we can see that all three parameters have grown over the years. While the length of the lyrics has been growing consistently, the Vocabulary and Word Rareness has had some fluctuations between the 70s and the 90s.

We can indeed say that the Word Rareness tends to follow a similar trend as the Vocabulary Size. This can be logically interpreted as a higher percentage of chance to find rare words when the number of words used is higher.

Does this mean that songs are getting longer?

It doesn’t. As previously mentioned, the rise of streaming has not only influenced the way music is written and produced, but also its duration. If the structure has changed to hook the listener’s attention from the beginning and listening to at least the first 30 seconds of the song (a stream is only paid if the song is played for at least 30 seconds!) the duration has instead decreased in the years as an attempt to generate more streams if the same song gets listened in a loop.

But why are we seeing a higher volume of words in the graphs if the songs are shorter, on average?

Firstly, the Lyrics Length parameter only refers to the number of words in the lyrics and doesn’t consider the duration of the songs.

Secondly, we interpreted these changes as the effect of:

• Pop and rock no longer being the major genres in the mainstream music world but sharing the scene with genres such as RAP, Trap or R’n’B.
• a higher presence of women in these genres, especially RAP, where lyrics are more complex and tend to contain more words in a shorter period of time compared to Pop and Rock Hits.

4. Slang, Profanity and Explicitness parameters

Has the vocabulary changed through the decades, and how, specifically?

Thanks to our AI, we were able to retrieve information about:

• Slang: the percentage of slang words in lyrics.
• Profanity: the percentage of profanity within a lyrics.
• Explicitness: if the lyrics contain explicit themes.

As we get close to the recent decade, it seems the use of profanity and explicit content is following a similar trend, especially a high increase in the usage of slang. Lyrics are more commonplace and a reflection of everyday language expression.

I got new rules; I count ‘em
I got new rules; I count ‘em
I gotta tell them to myself
I got new rules; I count ‘em
I gotta tell them to myself
(Dua Lipa — New Rules)

It is also important to observe that we have a higher value of profanity and explicitness in the 2000s more than in recent years, which we expected to find during the analysis.

However, it’s worth pointing out that the analysis was conducted on a small data set and it would be worthwhile to conduct a similar analysis on a broader set of lyrics.

Nonetheless, it is still interesting to see that since the 90s, there has been a constant increase in all of the three parameters.

Lyrics Clouds

What are the recurring words of each decade then?

To understand it, we have created these lyrics clouds.

Lyrics clouds visually highlight the most frequent terms used in the lyrics of each decade: the bigger the word, the higher its frequency in the songs.

As expected “Love” is the word we consistently find in each decade. It’s interesting to observe how the rest of the words in each cloud are gradually changing into more sexually explicit and/or materialistic topics (money, diamonds, boss).

Conclusion

This brief analysis aimed to show the changes throughout the decades of the most popular songs by female artists, the songs that have influenced millions of women of different generations and from different parts of the world.

At Musixmatch we strongly believe in the importance that lyrics have in communicating feelings and concepts, while also being a reflection of social changes.

We have seen that the theme of love remains the common factor of each decade, as expected.

But while the themes might be consistent, the ways of treating them adjust and evolves. These shifts are partially dictated by the changes in the music industry and also by the way we listen to music. We’ve observed that the Lyrics Structure has become more complex, with a peak in its complexity most notably registered in the last decade.

While on the other hand, changes like the ones we’ve observed in the Rhyming Factor and the Vocabulary are also a mirror of the way women are positioning themselves in the music industry. A marked, higher presence in genres that were historically male-dominated such as RAP and with songs that speak more explicitly about what they want and what they feel.

We are excited to see what the future of women in music holds and maybe, repeat this same analysis in the years ahead to see what has changed.

In the meantime…happy International Women’s Day!

Acknowledgments

This blog started from an idea of the Musixmatch Diversity and Inclusion committee and uses the AI technology developed by our internal AI team.

Written by Silvia Olivieri (Operation Support), M. Stella Tavella (AI Engineer) , Luca Antognetti (AI Intern).

Proofreading: Laurell Boyers (Studio Co-ordinator/English Content Editor), Anna Romani (Product Manager).

Illustration: Niche Chathong (Product Designer)

A special thanks to all the great women of Musixmatch! You rock!