A swear word is like a linguistic punch in the nose. Virtually every language and culture has them—and virtually every language and culture formally disapproves of them. But that doesn’t stop them from being used widely, loudly, and lustily.
What gives a swear word its power is partly its meaning—typically referring coarsely to bodily parts and functions—and partly its sound. In English, for example, studies have shown that swear words contain a higher ratio of so-called plosive sounds—including P, T, and K. Profane English monosyllables are especially likely to end in a plosive rather than begin with one. In German, profanity is also heavy on plosives, as well as on short vowel sounds.
What’s been less well explored is which sounds don’t wind up in curses—which ones soften the sound of a word so that it can’t pack the angry, cathartic power that common curse words do. Now, a new study in the journal Psychonomic Bulletin & Review has taken on that question and concluded that if you want to clean up the language, the best way is to lean on words that contain what are known as approximants—sounds that include the letters I, L, R, W, and Y, formed by passing air between the lips and the tongue, which are not touching when the sound is pronounced. Across multiple languages, the new paper showed, words that contain approximants are broadly judged less profane than words that contain other, more aggressive sounds.
The study, conducted by psychologists Shiri Lev-Ari and Ryan McKay of Royal Holloway, University of London, recruited 215 native speakers of six languages—Arabic, Chinese, Finnish, French, German, and Spanish—and presented them with words with which they were not familiar from 20 distinct languages. Though some of the speakers’ own languages were included in the list (Arabic, Chinese, and German), there was a good reason none of the subjects recognized any of the words: all of them were actually pseudo-words, based on real words in the multiple languages but changed slightly, both to include an approximant and not include an an approximant.
The Albanian word zog, for example, which means bird, was changed to the nonsense words yog, which contains an approximant, and tsog, which doesn’t. The Catalan word soka (or rope) was changed both to sola (with an approximant) and sotsa (no approximant).
Participants in the study—which was titled “How good is your ‘sweardar’?”—were not told that the pairs of words they were presented were not real words. Instead, they were told that one was a curse word in an unnamed foreign language and one was not a curse word; they were then asked to guess which was which. In total, the subjects were presented with 80 word pairs each, and in 63% of those cases, they chose the word that did not contain an approximant as the likely obscene one. Significantly, those results held true even for the French speakers, whose language does include curse words that contain approximants, but who still found the pseudo-words less offensive if they included approximants.
“Our findings reveal that not all sounds are equally suitable for profanity,” the authors wrote, “and demonstrate that sound symbolism is more pervasive than has previously been appreciated.”
In a second portion of their study, Lev-Ari and McKay examined “minced oaths” in the English language—words like “darn” and “shucks” that are used in place of their coarser alternatives. They collected 67 minced oaths that were variations on 24 swear words. (Some words have multiple minced oaths associated with them—”frigging,” “freaking,” and “effing,” for example.) Overall, they found that approximants were 70% likelier to be found in the minced oaths than in the swear words.
In a third portion of their paper, the researchers recruited 100 other volunteers, 20 apiece fluent in one of five languages—Hebrew, Hindi, Hungarian, Korean, and Russian—and asked them to provide a list of the most vulgar words in their language that they could think of. Lev-Ari and McKay included only words submitted by at least two participants, and wound up with a list of 141 curse words. The participants then rated each swear word in their own language on a scale of 0 to 100, from least to most offensive, and on another scale from least common to most commonly used. Yet again, approximants were underrepresented in the most offensive words behind plosives, fricatives (a consonant like F or V produced by forcing air through a narrow opening in the lips or throat), and other categories of sounds.
Exactly why approximants are considered less offensive than other sounds isn’t clear, but the researchers cited a body of existing work that certain phonemes, letters and sounds are closely associated with both word meaning and imagery. Multiple studies, for example, have shown that smaller objects are assigned words that are spoken in a higher frequency than larger objects. Another found that when people were shown drawings of both spiky and curved shapes, they chose jagged-sounding nonsense words like “takete” and “kiki” for the spiky images and softer sounding “moluma” and “bouba” for the curvy ones. Yet another compared curse words to lullabies and carols and found that while the curse words contained a disproportionate share of plosives, the songs contained what are known as sonorant consonants—like L and W—which are produced without turbulent air flow in the vocal tract.
“The connection between the sound and meaning of a word is arbitrary,” Lev-Ari and McKay write. “Nonetheless swear words have sounds that render them especially fit for their purpose.”
More Must-Reads From TIME