An article in the Sunday Times crossed my eyeballs this week. It’s called “Can song lyrics tell us how society is changing?” and the obvious answer is
But there’s a bit of linguistics in here (and no, it’s not good), so let’s dig in.
The article is by Tom Calver and he says “I have downloaded and analysed the full lyrics for the 40 biggest songs in every year going back to 1985, numbering 720,000 words from 1,600 tracks.” Ok, cool! That’s kind of a small corpus, but it covers all of what it is supposed to cover – lyrics from the top 40 songs in the UK charts from the last 40 years. Super.
Calver then gets into his analysis:
The first obvious finding is that music used to be more overtly fun. By assembling a list of “party” words — “club”, “bottle”, “VIP”, “DJ”, “dance” — we can track their frequency over time. The year I turned 18, 2011, was the partiest year on record, led by Mr Worldwide aka Pitbull (“give me everything tonight, for all we know, we might not get tomorrow”).
I’m guessing those aren’t the only words in his “party words” category. But he also doesn’t seem to provide his data, so we’re left wondering what the others are.
There are two important points to make here. First, this analysis needs a close reading of the concordances, or the lines of text which contain the search term. Think about the word bottle – is it always a “party” word? Merle Haggard has a song called “The Bottle Let Me Down,” which is about “a man who could no longer find solace from binge drinking to relieve his grief over a lost love” (Wikipedia). It didn’t make the top 40 in the UK, but I wouldn’t be surprised to find similar examples in the corpus. Think about Bruce Springsteen’s “Dancing in the Dark” (which came out in 1984 so it’s just outside of the corpus here). That song sure uses the word dancing a lot. But would you say that it is “overtly fun” or that it’s about partying? These analyses are not as simple as Calver makes them out to seem.
Second, Calver shows us how the percentage of “party” words shifts over the years. But he doesn’t give us any indication of whether these are meaningful shifts.
The differences here seem quite small – only around 1%. We have no way of telling whether these are meaningful differences because we have nothing to compare them to. Like, should we expect differences of 1% in the use of other words on a year-by-year basis? Show me the differences in usage of words such as the or in over the years – are they around 1%? Or the differences in other content words, such as love and call and thinking. The differences between years in Calver’s chart are usually less than 1%.
Calver gives some other charts about how lyrics have supposedly changed over time. For example, the category called “anticipation” goes from 10.8% in 1985 to 8.4% in 2025. That seems… concerning? Or maybe it’s a good thing? Tough to say.
I have no idea what Calver considers “positive” or “negative” words and he doesn’t tell us. Is bad a negative word? What about hot or cool? Close readings are really gonna matter here.
Calver then claims:
The tense chosen by songwriters is also quite revealing. Artists in the 1980s and 1990s — a period marked by the end of the Cold War and the seeming triumph of the western worldview — frequently looked forward. Words such as “will”, “forever” and “always” are common in songwriters’ vocabularies.
Today, though, music lives in the present: now, today and tonight. The year 2021 — featuring such hits as Joel Corry’s “I got a bed but I’d rather be in yours tonight” — is the least forward-looking of any (perhaps unsurprising, given most new songs that year were written during lockdown).
I’m not too sure what’s going on here. None of the words that Calver gives (will, forever, always, now, today, and tonight) are how we mark tense in English. In fact, we don’t use any word to indicate tense in English. Instead we mark it on the verb (run >> ran, jump >> jumped). The word will is used to point to the future but it is by no means the only way of doing so in English. You can just as easily use tonight to point to the future:
- I will leave.
- I leave tonight.
We’re told that Joel Corry’s song is in the least forward-looking year, but I don’t get it. Here are some of the lyrics of “Bed” (by Corry, RAYE and Guetta):
I got a bed, but I’d rather be in yours, yours
Not sure how “I (would) rather be in yours” doesn’t look forward to some desirable future.
And I got work in the mornin’
That is, I have to work tomorrow mornin’, which is in the future.
Somethin’ I’ll regret in the mornin’
A bit of both here. The singer is looking forward… to something they will regret. So it’s not really good.
Again, things aren’t as simple as they seem.
Then Calver makes a major error reading the top 40 tea leaves. He says:
But for me, the most striking pattern in the data hints at our growing self-obsession. Music used to be more obviously a social endeavour. Forty years ago, it was common for songs to be dominated by words like “we”, “us”, “ourselves”, “you”, and “together”.
Today, though, music is dominated by first-person pronouns. As Taylor Swift put it: “It’s me, hi, I’m the problem, it’s me.” This is more than just a passing fad: the psychologist C Nathan DeWall found that song lyrics since the early 1980s were becoming more socially disconnected with each passing decade.
This is not the linguistic analysis that Calver thinks it is. “Self words” and “social words” are not really linguistic categories, but let’s leave that aside for a moment. And let’s leave aside the fact that he doesn’t notice that we and us and ourselves are also first-person pronouns (because yikes). Calver seems to be saying that using first-person singular pronouns (I, me, myself) indicates a selfish outlook, while using first-person plural pronouns (we, us, ourselves) and second-person pronouns (you) indicates a communal or social outlook. This is the George F. Will School of Linguistic Analysis and it is bad (like, actually bad, not like so-bad-it’s-good bad).
Calver cites DeWall et al. (2011) to support his claim, but there’s a big problem. Mark Liberman pointed out (way back in 2011) that the DeWall study “uses misleading graphics to exaggerate its findings, and does a remarkably tone-deaf job of controlling for genre changes”. That is, the study did not find that song lyrics since the early 1980s were becoming more socially disconnected with each passing decade.
What about the link between first-person singular pronouns and “self-obsession”? Well, it’s a myth. Linguists have known for a long time that things are much more complicated than counting first-person singular pronouns as some kind of marker for a speaker’s narcissism. James W. Pennebaker, who has done a lot of work on this topic, explains the intricacies on Language Log here. A 2015 study by Carey et al. found there is no link between narcissism and first-person singular pronouns. So much for our growing self-obsession, at least as far as it being shown by song lyrics.
Then there are some head scratchers to finish the article off. Calver says:
Today’s songs are increasingly written to be listened to alone.
Huh?
They express little desire to belong to something bigger: instead, they narrate our immediate physical and psychological needs.
You’re gonna have to prove that one, bud.
Taylor Swift’s 2023 concert film was the highest grossing concert film ever. And she broke attendance records at stadiums around the world during her 2023-24 tour. It seems to me like people want to be part of something bigger.
—-
For what it’s worth, I reached out to Calver on Bluesky to ask him about these things, but he didn’t get back to me. I’ll update this post if he does.