April 26, 2017
People have been telling love stories for thousands of years. But in 2004, a new romantic subgenre was born—in the form of the New York Times’ wildly popular “Modern Love” column.
A typical “Modern Love” column is no more representative of how the average person falls in love than Romeo and Juliet. Naturally, the stories that appear in the paper tend to be dramatic. (Deadly diseases and trips to the emergency room are recurring features.) And the columns are disproportionately written by professional writers, which means the stories are evenly paced, and cleanly structured, in a way that love often isn’t.
Still, the column can reveal a lot about our cultural attitudes toward romance and heartbreak. As graduate students in economics and computer science, we decided to use statistics to analyze every “Modern Love” column published over the past 10 years—with the goal of identifying patterns in how romantic narratives take shape. Here’s what we learned.
1) Dating may be harrowing, but it makes for the best stories
The New York Times tags each article with its main topics, revealing the incredible number of ways to write about love.
Dating proves to be a particularly fruitful topic, with online dating a favorite subject. Fourteen columns mention match.com. Tinder gets six mentions; OKCupid appears in three; and Hinge, eHarmony, and JDate all get nods.
2.) The column prefers to stay demure when it comes to sex
Many columns deal with trials of true love: mental disorders, death and dying, cancer, infertility, crime and criminals, and adultery. But it turns out that “Modern Love” columns are quite innocent in another sense: they average only half a kiss per column, and the majority of the columns never explicitly mention “sex” at all. (Of course, people often allude to sex in convoluted ways that are difficult for a computer to detect, but we searched for common synonyms, like “make love,” as well).
“Modern Love” editor Daniel Jones says this comes as no surprise: “Our news standards don’t allow for much in the way of describing sex acts in personal essays, so sex happens, yes, but off-screen,” he notes via email.
That said, a few columns use the word “sex” a lot. (Are you only reading this to find those columns? Shame on you; here you go.) All three of these columns are by women, although two columns by men are close behind. All five columns center on the amount of sex the authors are having. The three women discuss having less sex than what they take to be the societal norm. One man writes about having more sex than average, while the last discusses how sex life oscillates because of his wife’s medication for Parkinson’s disease.
3.) Men are more likely to focus on other men
About 80% of “Modern Love” columns are written by women. While 79% of female writers use more male pronouns than female pronouns, the split is much more even among male writers—only 64% use more female pronouns than male pronouns.
At first we thought this might be because gay men were writing about romance more frequently—and, indeed, male writers use the word “gay” much more frequently than female writers do (and more frequently than female writers use the word “lesbian”). But when we started reading columns from the male writers that used mostly male pronouns, most of them were not about romantic love; many of them were about fathers. Strikingly, women mention their daughters twice as often as they mention their sons, while men mention their sons twice as often as they mention their daughters.
Jones says he has a theory about the gender split: “Men are often really hesitant to criticize women in love stories, which can lead to them not writing about women at all,” he writes. “Whereas women are less likely to hold back when it comes to writing about men (or criticizing them).”
4) “Modern Love” columns follow clear narrative arcs
We mathematically traced the arcs of people’s love stories by plotting where in the essay certain words occur. The beginnings of columns feature characters (“boyfriend”, “husband”) and set the scene (“college,” “beauty school”). As essays progress, they become more emotionally intense, using more sad language (as measured by LIWC scores, a standard approach).
But near the end, authors shift from using “she/he” to the more romantic “we.”
They stop talking about the past (using phrases like “met” and “years ago”) and look to the present and future (“now,” “I will”). Suggesting some form of personal growth or understanding, the authors also use more words indicating insight and certainty (eg, “realization”) as the end draws near. And at the very end, love blossoms; of the tens of thousands of words used in “Modern Love” essays, “love” is the one that spikes most significantly at the end.
5) There are a lot of ways to talk about loss
One column uses twice as many sad words (such as “grief” and “tears”) as any other. The author, Allison Amend, goes to a funeral, gets dumped by her boyfriend, and gets diagnosed with ovarian failure—all in one day. The column that uses the most anxious words (eg, “scared”) is Amy O’Leary’s piece about learning to admit her anxiety. Second place goes to a woman whose honeymoon in Paris is almost ruined by her anxiety.
But some sad stories use no sad language at all. Cindy Chupack’s column, about getting a divorce from a man who realizes he’s gay, fools the algorithm into thinking the story itself isn’t sad because it uses funny language. Cindy indeed mentions that she toyed with stand-up comedy during her divorce. Our algorithm could have laughed along with her whole set without picking up on any underlying hurt. When we tell a story about heartache, we don’t always do so straightforwardly; there are all kinds of ways to communicate loss.
6) Computers can’t write romance
In a final endeavor, we tried to train a computer program to write its own “Modern Love” columns after reading every column ever published. Its early attempts were rough: “Thene and yot oge a tat my hid trat that I soven the rast?” it pleaded. (To be fair, many people we know are similarly incoherent when talking about love.)
But eventually, our program learned to write credible beginnings to essays. “I loved him…” we prompted, and it produced a slightly disturbing constellation of continuations:
I loved him back, leaving a ragged triangle of bite marks on my hand.
I loved him so wildly I could be made legal.
I loved him for the weekend as well, and I drank apple martini ingredients like hummingbird saliva or snake testicles.
We apologize for our program’s prurience. But remember: its only exposure to “love” is through these 500 stories. It’s perhaps the equivalent of a very young child whose only exposure to love has come through princess movies and picture books. It may never be able to come up with a clear way to explain how love feels—at least until it meets another computer program that makes its subprocesses freeze for one beautiful, inexplicable moment.