The sound of swearing

« previous post | next post »

Trigger warning:  I'm VHM and I do not approve of this message in its entirety.

Article by Elizabeth Preston in NYT (12/6/22):

"Curse Words Around the World Have Something in Common (We Swear)"

These four sounds are missing from some of the seven words you can never say on television, and the pattern prevails in other languages too, researchers say.

Starting with the second paragraph:

A study published Tuesday in the journal Psychonomic Bulletin & Review found that curse words in several unrelated languages sound alike. They’re less likely than other words to include the consonant sounds L, R, W or Y. And more family-friendly versions of curses often have these sounds added, just like the R in “shirt” or “fork.” The finding suggests that some underlying rules may link the world’s languages, no matter how different they are.

“In English, some of the worst words seem to have common phonetic properties,” said Ryan McKay, a psychologist at Royal Holloway, University of London. They’re often short and punchy. They also tend to include the sounds P, T or K, “without giving any obvious examples,” Dr. McKay said. These sounds are called stop consonants because they interrupt the airflow when we’re speaking.

My interest had started to flag, because I felt that the conclusions being drawn were two facilely deterministic, but when I hit the sentence about "the sounds P, T or K", I was electrified, because my Sinitic phonological being is intensely attuned to precisely these sounds.  In Sinitic languages, they are referred to as "entering tones".

A checked tone, commonly known by the Chinese calque entering tone, is one of the four syllable types in the phonology of Middle Chinese. Although usually translated as "tone", a checked tone is not a tone in the phonetic sense but rather a syllable that ends in a stop consonant or a glottal stop. Separating the checked tone allows -p, -t, and -k to be treated as allophones of -m, -n, and -ng, respectively, since they are in complementary distribution. Stops appear only in the checked tone, and nasals appear only in the other tones. Because of the origin of tone in Chinese, the number of tones found in such syllables is smaller than the number of tones in other syllables. In Chinese phonetics, they have traditionally been counted separately.

For instance, in Cantonese, there are six tones in syllables that do not end in stops but only three in syllables that do so. That is why although Cantonese has only six tones, in the sense of six contrasting variations in pitch, it is often said to have nine tones.

Final voiceless stops and therefore the checked "tones" have disappeared from most Mandarin dialects, spoken in northern and southwestern China, but have been preserved in the southeastern branches of Chinese, such as Yue, Min, and Hakka.

Tones are an indispensable part of Chinese literature, as characters in poetry and prose were chosen according to tones and rhymes for their euphony. This use of language helps the reconstruction of the pronunciation of Old Chinese and Middle Chinese since the Chinese writing system is logographic, rather than phonetic.


The loss of the entering tones in northern, especially Mandarin, languages, has always been an enigma to me, and I have often wondered how and why it happened.  We have a pretty good idea of roughly when it occurred — the late medieval period:

The voiceless stops that typify the entering tone date back to the Proto-Sino-Tibetan, the parent language of Chinese as well as the Tibeto-Burman languages. In addition, it is commonly thought that Old Chinese had syllables ending in clusters /ps/, /ts/, and /ks/[citation needed] (sometimes called the "long entering tone" while syllables ending in /p/, /t/ and /k/ are the "short entering tone"). Clusters later were reduced to /s/, which, in turn, became /h/ and ultimately tone 3 in Middle Chinese (the "departing tone").

The first Chinese philologists began to describe the phonology of Chinese during the Early Middle Chinese period (specifically, during the Northern and Southern Dynasties, between 400 and 600 AD), under the influence of Buddhism and the Sanskrit language that arrived along with it. There were several unsuccessful attempts to classify the tones of Chinese before the establishment of the traditional four-tone description between 483 and 493. It is based on the Vedic theory of three intonations (聲明論). The middle intonation, udātta, maps to the "level tone" (平聲); the upwards intonation, svarita, to the "rising tone" (上聲); the downward intonation, anudātta, to the "departing tone" (去聲). The distinctive sound of syllables ending with a stop did not fit the three intonations and was categorised as the "entering tone" (入聲). The use of four-tone system flourished in the Sui and Tang dynasties (7th–10th centuries). An important rime dictionary, Qieyun, was written in this period.

Note that modern linguistic descriptions of Middle Chinese often refer to the level, rising and departing tones as tones 1, 2 and 3, respectively.

By the time of the Mongol invasion (the Yuan dynasty, 1279–1368), former final stops had been reduced to a glottal stop /ʔ/ in Mandarin. The Zhongyuan Yinyun, a rime book of 1324, already shows signs of the disappearance of the glottal stop and the emergence of the modern Mandarin tone system in its place. The precise time at which the loss occurred is unknown though it was likely gone by the time of the Qing Dynasty, in the 17th century.


I had never thought that the loss of entering tones in Mandarin and other northern Sinitic languages may have relevance to swearing, never even dreamed of it, so I decided to read on in Preston's article:

Dr. McKay teamed up with his colleague Shiri Lev-Ari to learn whether this familiar pattern went beyond English. They wondered whether it might even represent what’s called sound symbolism.

Sound symbolism is when a word sounds like what it means. One type is onomatopoeia; for example, words that describe a cat’s meow or a rooster’s crow are similar across many languages. Globally, words having to do with noses often include [VHM:  "Sound–meaning association biases evidenced across thousands of languages" (PNAS 2016)] the nasal N sound, and words related to smallness often have an “ee” sound (as in “mini” or “teensy weensy”), like the squeaking of a small creature.

To look for patterns in swearing, the researchers asked fluent speakers of Hebrew, Hindi, Hungarian, Korean and Russian to list the most vulgar words they could think of. Once they’d compiled a list of each language’s most frequently used epithets, the researchers compared these with neutral words from the same language.

In these languages, they didn’t find the harsh-sounding stop consonants that seem common in English swear words. “Instead, we found patterns that none of us expected,” Dr. Lev-Ari said. The vulgar words were defined by what they lacked: the consonant sounds L, R, W and Y. (In linguistics, these gentle sounds are called approximants.)

Next, the scientists looked for the same phenomenon using speakers of different languages: Arabic, Chinese, Finnish, French, German and Spanish. The subjects listened to pairs of words in a language they didn’t speak, and guessed which word in each pair was offensive. In reality, all the words were invented. For example, the researchers started with the Albanian word “zog,” for “bird,” and created the pair of fake words “yog” and “tsog.” Subjects were more likely to guess that words without approximants, such as “tsog,” were curses.

Finally, the researchers combed through the dictionary for English swear words and their cleaned-up versions, also called minced oaths (“darn,” “frigging” and so on). Once again, the clean versions included more of the sounds L, R, W and Y.

“What this paper finds for the first time is that taboo words across languages, unrelated to each other, may pattern similarly,” said Benjamin Bergen, a cognitive scientist at the University of California, San Diego, who was not involved in the study.


Don Keyser, who called my attention to Preston's article, writes:

You can color me profoundly skeptical of these findings … I don't find this to be the case, at least noticeably, in even the languages I know well enough to know the curse words (Russian, Chinese, Japanese, French … and in a more limited way German and Spanish).

I'm reminded of the old saw about Cantonese — "making love in Cantonese sounds like making war in Mandarin."

And you can color me skeptical about that old Cantonese saw.


Selected readings


  1. Blolix said,

    December 7, 2022 @ 11:09 am

    There's a running gag in the TV comedy series "The Good Place" in which a "bad" person – played by Kristen Bell – who's wound up in "the good place" by mistake finds that whenever she tries to curse she utters an inoffensive sound-alike – usually "shirt" or "fork."

  2. Victor Mair said,

    December 7, 2022 @ 11:51 am

    I have a student named Ruolan. The first time I heard that name, I have to say that I found it exceptionally soothing.

    Sitting next to her is another student named Cai Ci / Ts'ai Tz'u. Although it is not in the -p, -t, -k category and I can pronounce it properly, I have to be conscientious when I say her name to get it right. It requires an effort, whereas Ruolan just rolls off the tongue.

  3. Chris Button said,

    December 7, 2022 @ 12:20 pm

    The loss of the entering tones in northern, especially Mandarin, languages, has always been an enigma to me, and I have often wondered how and why it happened.

    This blog post by John Wells covers the phonetics nicely:

    The only caveat would be that the Burmese final stops that he mentioned are all undifferentiated glottal stops now.

    Another reason why historical linguistics need to understand phonetics in order to create reasonable hypotheses.

  4. Haamu said,

    December 7, 2022 @ 12:46 pm

    “In English, some of the worst words seem to have common phonetic properties,” said [McKay]. They’re often short and punchy. They also tend to include the sounds P, T or K …

    I can't resist citing the standard vulgar insult in Klingon, petaQ (whose pronunciation might be better indicated by any of several variant spellings: "p'tahk", "p'takh", etc.). It's interesting that this is a language specifically constructed to sound unlike English or other terrestrial languages, and yet …

  5. DaveK said,

    December 7, 2022 @ 12:55 pm

    How do the researchers account for religion-based profanities like hell, Lord, Christ, etc.
    Speakers of any language/ adherents of any religion presumably have a wide variety of religious terms to choose from when blaspheming and it’s interesting to see which ones became most common.

  6. David P said,

    December 7, 2022 @ 1:06 pm

    P, T, K –
    Maybe that explains why captains get called that.

  7. Taylor, Philip said,

    December 7, 2022 @ 2:23 pm

    "… the standard vulgar insult in Klingon, petaQ" — ah, that would explain why my past attempts to invite visiting Klingons to a friendly game of pétanque have met with such (until now) seemingly inexplicable hostility …

  8. David Marjanović said,

    December 7, 2022 @ 2:59 pm

    …Oh, this is about approximants, not about trilled [r]. I was wondering!

    Still, the sheer frequency of блядь probably outweighs all of the world's approximant-free swearwords put together.

    Although it is not in the -p, -t, -k category and I can pronounce it properly, I have to be conscientious when I say her name to get it right. It requires an effort

    Of course. Aspirated affricates require an unusual amount of air pressure; that's why they're globally much rarer than aspirated plosives.

    (Aspirated fricatives are extremely rare.)

    whose pronunciation might be better indicated by any of several variant spellings: "p'tahk", "p'takh", etc.

    Oh bullshit. The word was, as so often, coined by Star Trek screenwriters who didn't bother to consult with Marc Okrand. Klingon doesn't have any reduced vowels, for example. Okrand was left to give the word an approximation in the actual Klingon sound system after it had appeared on screen.

  9. Haamu said,

    December 7, 2022 @ 3:38 pm

    the actual Klingon sound system

    Possibly the most distant front in the prescriptivist/descriptivist wars.

  10. Chas Belov said,

    December 7, 2022 @ 4:38 pm

    The Cantonese p/t/k connection was my first thought as well.

    Thinking of the Hokkien curse KNNBCCB, of course has K. CB, well, has B which is related to P. The corresponding Cantonese curse DNMMH, has D which is related to T but H doesn't fit the rule.

    Similarly, the Thai curse hia doesn't fit the rule.

  11. Jerry Packard said,

    December 7, 2022 @ 5:14 pm

    The distribution and frequency of certain consonant types over curse words in the languages of the world is most likely completely arbitrary and unpatterned. The use of phonesthemes likely has nothing to do with it. The loss of the ptk finals in Mandarin is likely a reflection of general syllable-final weakening in the Sinitic languages over the past couple thousand years as Matthew Chen has pointed out. That syllable-final weakening is in turn an example of the phonetic syllable-final weakening that happens in all languages over the passage of time. A simple example from English is the fact that the syllable-final p/b t/d k/g distinction is mostly a function of the length of the preceding vowel rather than the articulation of the final consonant itself.

  12. Andreas Johansson said,

    December 8, 2022 @ 3:16 am

    The Swedish mild swear jävla has the even milder variant jäkla, formed by people who weren't quite comfortable with invoking the Devil. According to this theory, almost any phonetic substitution ought have been likelier than replacing an approximant with a voiceless stop.

    (I'm not sure how aware people who are not linguistics nerds are today that jävla is derived from djävul "devil" – it's an archaic genitive plural – but the words are closer than they might appear cited like this, the "d" being silent and the "u" dropping in the plural.)

  13. Hiroshi Kumamoto said,

    December 8, 2022 @ 6:12 am

    See also:

  14. Kenny Easwaran said,

    December 8, 2022 @ 12:10 pm

    I'd be interested in knowing what classes of words they investigated. In English, the biggest taboo words these days aren't traditional "swear words" but instead are slurs. Notably, what is probably the most taboo word in English these days ends with an "r" and begins with a nasal consonant, and the only stop is voiced. Of course, a single work breaking the patterns isn't such a problem for the theory, and many other prominent slurs do stick to the pattern of having voiceless stops without approximants.

  15. Doug said,

    December 8, 2022 @ 5:05 pm

    @Kenny Easwaran

    Note I'm not a linguist and have not read the paper.

    But I followed the links from the NYT article to the paper to the data, and it seems that their list of English swear words is here:

    Note the list is not in English orthography and not in IPA, but in some phonetic representation unfamiliar to me. (Perhaps dictated by the software used in the study.)

    Also note that the pronunciations and the selection of swear words are from British English. As I'm American, it took me some time to realize that the first 2 words are probably forms of "asshole" and "ass". I still don't know what the fourth one is.

  16. Doug said,

    December 8, 2022 @ 5:27 pm

    So far as I can tell, there are no racial or ethnic slurs on the list. The words all seem to be traditional vulgar words referring to sex, anatomy, excrement, and damnation.

  17. Taylor, Philip said,

    December 8, 2022 @ 6:20 pm

    Doug, the fourth is "bell-end", being a slang term for the glans penis. But what is the second ?

  18. Doug said,

    December 8, 2022 @ 6:30 pm

    I presume the second one "Es" represents "ass", although the phonetic spelling suggests the wrong vowel for my pronunciation.

    (Similarly the second and third from the end have "E" in words with orthographic "a".)

  19. JOHN S ROHSENOW said,

    December 8, 2022 @ 8:56 pm

    Not directly on topic, but a friend introduced me to the ffl
    What the F: What Swearing Reveals About Our Language, Our … › What-Swearing-Reveals-La…
    As linguist and cognitive scientist Benjamin K. Bergen shows us, it also opens a new window onto how our brains process language and why languages vary …
    Rating: 4.4 · ‎109 reviews · ‎$43.95

  20. Yerushalmi said,

    December 20, 2022 @ 12:13 pm

    The first Hebrew curse word I thought of was "L'azazel", with both starts and begins with an L. But that is admittedly a relatively mild one.

  21. Daniel said,

    December 20, 2022 @ 3:14 pm

    When I was a teenager, I was pretty careful not to say swear words. However, I would get frustrated, and something would want to come out. For several years, my go-to expletive was simply p't'k'. Sometimes it would be shorter, just p't' or p'k', or longer pshhhk'. This also reminds me of the scoffing expletive: pfff. Expletives are interesting!

RSS feed for comments on this post