The Effects of the COVID-19 Pandemic on the Pop Music Scene - A Data Science Research Project
The Effects of the COVID-19 Pandemic on the Pop Music Scene - A Data Science Research Project
In the following article (that’s been about 2 years in the making!), I examine the pop music scene during the pandemic and explore how such a world-changing event shaped the pop music landscape
Hi all, today I have an incredibly special article to share!! This one has been in the works since around the spring of 2022, and here it is!! A data science research project where I used sentiment analysis to analyze any trends between the COVID-19 pandemic and the overall mood of the hit music. But let me start with what I did to investigate this: I pulled the song lyrics for the year-end Hot 100 lists from 2019-2022 from Genius.com - 2020-22 because, obviously, these are peak pandemic years, 2019 because this was the last pre-COVID year and I figured we would need a baseline on how the mood of the pop music was then - and performed sentiment analysis on them.
We already know that pop music can reflect world events, to quote YouTube music critic Mark Grondin in his best hits of 2009 list: “it was impossible to not see 2009 as a year of reckless abandon: the economy had crashed, most of my generation was broke, and if we didn't have money, we were going to party as if we did - it might have seemed bleak, but I think a lot of us were riding the contact high that came from a new president and a desperate desire to believe in hope…”. This reflects that the overall mindset of the general public in 2008-09 was that even if the economy crashed and they were broke, they might as well make the most of what little they had.
So let’s transition into what my hypothesis was going into this: I thought that if 2019 was our baseline, 2020 would be our lowpoint and drastically lower than 2019. And then as the pandemic progressed, the sentiment would slowly get more and more positive. So in that regard, I expected 2019 to have the most positive sentiment and then 2022 to have the next highest, then 2021 second lowest/second most negative, and 2020 to have the most negative.
Let’s go onto the code I used. I used the Genius API to scrape the lyrics for each year and then I used nltk (Python) to do sentiment analysis on those lyrics. So what are the results?
There were some quite surprising results here. For one, my “control”, 2019, somehow had the highest percentage of negative words among the years I sampled. And to my surprise, 2022 had the highest positive percentage, not just among the pandemic years, but overall! Now, I have a theory as to why this was the case: I vaguely remember reading somewhere in 2019 that mental health in adults was on the decline (I cannot track down where I found this out, I’m sorry), so the music got a lot darker in 2019. Now for why the negativity was higher in 2019 than 2020, think back to Grondin’s quote, in 2008-09, the general public was trying to make the best out of a really awful situation (in that case, the 2008 Recession). I think a lot of that mentality carried into 2020. In 2019 no one knew what was about to happen, so they didn’t have any desire to make the most out of what they had. 2020 was a very overwhelming year for current events, the COVID pandemic first hitting, the BLM protests, and countless more. Even if I do remember some of the pop music in 2020 making references (both intentional and unintentional) to the pandemic (“If The World Was Ending”, “stuck with u”, and not really pandemic related but it’s still pretty explicitly tied to the current events in 2020 - “The Bigger Picture”), I can’t imagine anyone would’ve wanted to be constantly reminded of such an overwhelming event when they’re trying to distract themselves from said event. I don’t think anyone would want to use music as a distraction and then have that music be about the thing they’re trying to distract themselves from. And I think this lines up relatively well with some of the songs I remember making the year-end list. There were some pretty happy-sounding tunes there, like “Roxanne”, “Ballin’”, and “Watermelon Sugar”, just to name a few. 2022 having the highest positive percentage makes a lot of sense since this was the year a lot of the COVID restrictions were lifting up, and the mood wasn’t resetting to the equilibrium from prior to the pandemic, it was more because in comparison to the last two years being in lockdown, it felt so much better, people probably had a mindset of “We’re finally free! Anything is better than that hell”. With 2021 having the highest negative percentage, I’d say it’s because after 2020, people were expecting 2021 to be all sunshine, rainbows, and lollipops in comparison to the hellscape that was the previous year. This probably left people feeling really underwhelmed or disappointed. Now for the consistent 80% 's on the neutral word count...I’d attribute that to the amount of junk (data quality) that came into my txt file when I scraped the lyrics more than anything else.
So what’s the conclusion/key takeaway from all of this? Well, I’ll say that this project taught me a lot about data science and how I can use it for similar projects in the future. It also really taught me that with how decentralized music and pop culture as a whole have become, it’s really not just black and white when using the sentiment of pop songs to analyze how people felt overall. It can be a great tool, but it shouldn’t be the only parameter (as evidenced by the wild range of data across all the years and all three categories of sentiments). No one turns on MTV anymore to see the hottest new songs or whatever. Everyone deals with overwhelming events in their own ways, whether it be listening to music (which is even further decentralized, people listen to whatever genres they like), this was clearly evidenced by the low percentages in the positive and negative categories.
.png)
This is a really cool project idea! Interesting that the margin between positive and negative was relatively small in 2019 and especially 2021. Thanks for sharing your findings :)
ReplyDelete