Click here for free access to our latest coronavirus/COVID-19 research, commentary, and news.

Support nonprofit science journalism

Science’s extensive COVID-19 coverage is free to all readers. To support our nonprofit science journalism, please make a tax-deductible gift today.

Olena Yakobchuk/Shutterstock

What is love? It depends which language you speak

Falling in love is never easy. But do it in a foreign language, and complications pile up quickly, from your first fumbling attempts at deep expression to the inevitable quarrel to the family visit punctuated by remarks that mean so much more than you realize. Now, a study of two dozen terms related to emotion in nearly 2500 languages suggests those misunderstandings aren’t all in your head. Instead, emotional concepts like love, shame, and anger vary in meaning from culture to culture, even when we translate them into the same words.

“I wish I had thought of this,” says Lisa Feldman Barrett, a neuroscientist and psychologist at Northeastern University in Boston. “It’s a very, very well-reasoned, clever approach.”

People have argued about emotions since the ancient Greeks. Aristotle suggested they were essential to virtue. The stoics called them antithetical to reason. And in his “forgotten” masterpiece, The Expression of the Emotions in Man and Animals, Charles Darwin wrote that they likely had a single origin. He thought every culture the world over shared six basic emotions: happiness, sadness, fear, anger, surprise, and disgust.

Since then, psychologists have looked for traces of these emotions in scores of languages. And although one common experiment, which asks participants to identify emotions from photographs of facial expressions, has led to many claims of universality, critics say an overreliance on concepts from Western, industrialized societies dooms such attempts from the start.

To find out how a concept like love varies from language to language, Joshua Conrad Jackson, a Ph.D. student in cultural psychology at the University of North Carolina in Chapel Hill, tried a new approach using statistics. He teamed up with Johann-Mattis List, a computational linguist at the Max Planck Institute for the Science of Human History in Jena, Germany, who manages the Database of Cross-Linguistic Colexifications (CLICS). CLICS uses data from field linguists and anthropologists to catalog relationships between concepts and the words that represent them in nearly 3000 languages. Importantly, CLICS can take words that represent more than one concept, like “dull,” and reveal other words that express the same concepts in all languages in the database. That kind of concept mapping was just what Jackson was looking for. “It was a match made in heaven,” he says.

Networks of emotional concepts vary significantly among language families. Here, the size of the circles represents the number of shared words per concept, and the weight of the lines represents the strength of the connections between concepts. (Click here to see the full networks for five major language families.)

Jackson et al./Science

Over 2 years, Jackson and List assembled a team of statisticians, psychologists, and linguists to analyze the CLICS data in the largest ever study of its kind. They started with 24 emotional concepts (see chart) and used multiple statistical methods to map how they were related to different words in 2474 languages in 20 language families. The more words the two concepts had in common, the closer their relationship. For example, the concepts love and pity are both expressed by the subtle Hawaiian expression “aloha.” (That connection between “pity” and “love” seemed to be particularly strong in the Austronesian language family.)

After aggregating the data, the researchers visualized these connections in 21 networks—one for each of the different language families and one that aggregated all findings into a universal network. When the researchers analyzed the networks, they found that the links between emotional concepts differed even more than expected across language families. And when they repeated the process with 13 concepts related to color—which are already known to be relatively culture specific—they found that the emotional concepts had three times as much variability by language family. This means, the group writes today in Science, that emotional terms vary widely in their meaning from language to language. For example, Persian uses one term, ænduh, to express both grief and regret, but the Dargwa word for grief, dard, also expresses not regret, but anxiety.

What’s more, geographically close language families have more closely aligned networks than distant ones, suggesting that culture—either through shared experiences or ancestry—may be responsible for the evolution of some of these terms.

The work is “ambitious” and “convincing,” says Alexandre François, a field linguist with CNRS, the French national research agency in Paris. But François, who coined the term “colexification” more than 10 years ago, worries the study ignores emotional concepts that aren’t represented in English—or terms that don’t translate smoothly among languages. For example, he says, speakers of Mwotlap in the South Pacific island nation of Vanuatu have no perfect equivalent of the English word “love.” The closest match is a verb tam which means, variably, empathy, generosity, and hospitality. That’s a good term for loving one’s neighbors, François says, but it doesn’t cover romantic love (which is expressed by another verb, similar to “need,” as in, “I need you”).

At the very least, Jackson hopes his study offers some insight for language learners. “Just learning the words isn’t learning the language,” he says. You also need to learn the context.