During the 2003 SARS outbreak, almost all of the suspected cases in the United States were people who had recently traveled to parts of the world where SARS was spreading, like China. In July of that year, Sakae Inouye, a researcher at Otsuma Women's University in Tokyo, brought up a puzzling fact: There had been more Japanese tourists in China than Americans, and yet Japan had somehow escaped SARS—with zero officially reported cases.
Inouye's guess as to why, which he wrote about in a letter to the medical journal The Lancet, was that SARS might be more easily transmitted in certain languages, namely Chinese and English, because of how much breathing in and out is needed to produce those language's common sounds.
In Chinese, "the consonants p, t, k, q, ch, and c, when placed in front of vowels, are pronounced with a strong breath,” he wrote. In English, p, t, and k are also pronounced by exhaling breath—but not in Japanese. He also noted that the “p” sound isn’t used as frequently in Japanese.
He suggested that a Chinese person would mostly speak English to an American traveler—“I believe American tourists would, hence, be exposed to the infectious droplets to a greater extent than would Japanese tourists.”
This idea—that disease transmission could change based on language—has been raised again during our current global pandemic with SARS-CoV-2.
On May 14, Japan began to lift its state of emergency, about two weeks earlier than it had planned to. Their new COVID-19 cases have dropped near .5 per 100,000 people, and they accomplished this without strict social distancing or widespread testing seen in other countries. "Did Japan Just Beat the Virus Without Lockdowns or Mass Testing?" a recent headline at Bloomberg wondered. Even in Tokyo, the most populated city in the world, case numbers have fallen to single digits.
There have been speculations circulating online as to why Japan got off so easy. Hitoshi Oshitani, a virologist and public health expert at Tohoku University, told Science magazine that they identified clusters of infections to stop them from spreading and urged people to avoid “the 'three Cs'—closed spaces, crowds, and close-contact settings in which people are talking face-to-face.” Other factors may have led to Japan’s success: An already existing mask-wearing culture and rapid response from contact tracers.
But the question of language has reemerged: Maybe something about speaking the Japanese language produces less viral particles to pass onto others. In May, a clip from Japanese television started circulating on Twitter, garnering over 40,000 likes. It shows a woman saying, “This is a pen," first in Japanese, and then in English, with a white cloth hanging in front of her face. When the woman speaks English, the phrase causes the cloth to flap in the wind emitted from her mouth; when she speaks Japanese, it stays nearly entirely still.
One of the ways the coronavirus is thought to be transmitted between people through respiratory droplets, which can fall and land on surfaces, and then be brought to people’s faces by touch. Coughing and sneezing are the poster children of transmitting these disease-containing particles, which are often large enough for the eye to see.
The phrase “The rainbow is a division of white,” produces many more particles than “A sign from the gods to foretell war.”
But normal speech can also produce smaller particles that we can’t see, yet are big enough to carry pathogenic viruses. The question of how well COVID-19 can spread through the air on these small particles, or aerosols, has been up for debate. Case reports of transmission at choir practices, restaurants, and call centers suggest that it is possible for SARS-CoV-2 to hitch a ride on aerosol particles and stay suspended in the air.
Two recent studies have found that normal talking can emit thousands of small particles that can linger in the air for over 10 minutes. And other recent work has found that different speech sounds produce more of these particles than others. For example, more particles are emitted by saying phrases that have a lot of vowels.
“Saying eee releases more than saying ahhh,” said William Ristenpart, a chemical engineer and an expert in transport phenomena at The University of California Davis. In a study by Ristenpart and his colleagues from January, they found that saying the phrase “The rainbow is a division of white,” produces many more particles than “A sign from the gods to foretell war.”
Still, we're not yet able to pinpoint a particular language as being more or less risky for transmission. Ristenpart said that the volume at which a person speaks might overpower any individual differences of language. He and others found that some people, despite their language or volume, seem to emit much more particles than others while they talk. Taken altogether, the variations of different kinds of talking, breathing, and singing need to be paid closer attention to and investigated, multiple experts have said, especially as public health recommendations are made around mask-wearing, how far apart and where people can gather (and talk), and what kinds of speaking or singing activities can be done indoors—no matter what the language.
A widely referenced CDC study from May found that at a choir practice in Washington state, one person who had COVID-19 infected 52 other people over the course of a 2 1/2 hour rehearsal—two died.
It wasn't the first time singing—a kind of loud, prolonged talking—has been implicated in the spread of disease. In the 1960s, researchers looked into how singing might help spread tuberculosis. They found that the number of airborne droplets produced by singing was six times more than the number produced by talking, and about the same as coughing.
Watch more from VICE:
More recently, a study in the Journal of Aerosol Science found that counting from 1 to 100 releases six times as many particles as one cough. In 2009, a study found that coughing continuously for 30 seconds released half as many particles as saying “aah” for 30 seconds.
These particles are thought to come from the mucosal layer coating the respiratory tract, and they are very small: about 1 micron in diameter (a micron is a millionth of a meter; about .00004 inches). We know these minuscule particles can carry other pathogens. Research from 2016 and 2018 found infectious influenza in particles emitted by influenza-infected people who weren’t coughing or sneezing, but just breathing naturally.
These particles could be an even higher risk for infection, because they linger in the air longer (since they’re so light) and could penetrate further into the respiratory tract. And since speech is more common than coughing or sneezing, it has more of a chance to release particles.
A study from May in The Proceedings of the National Academy of Sciences found that talking can propel thousands of tiny aerosol droplets into the air, which can evade gravity and stay floating in the air for 8 to 14 minutes.
The researchers asked people to say the phrase “stay healthy” while measuring the droplets they emitted. There were around 2,600 small droplets produced per second. Though the subjects in the study weren’t sick themselves, using information from past work, they estimated that one minute of loud talking could make 1,000 virus-containing droplets.
“For asymptomatic people, saliva droplets created from simple speech – not sneezing or coughing – likely account for the majority of droplets a person expels, but more research is needed to determine the number of virus particles that are present in speech-generated droplets,” said Lisa Yuan, a spokesperson on behalf of the study's authors at the National Institute of Diabetes and Digestive and Kidney Diseases.
Different kinds of talking haven't been as closely examined, let alone different languages. In the past, research has instead focused on comparing talking to coughing, Ristenpart said—not comparing different phones, the basic sound components of speech.
In January, Ristenpart and his colleagues looked at how different phones could affect particle emission. They asked people to say single sounds and disyllabic words (baba, papa, dada), and had them read a section of text called the Rainbow passage, commonly used in linguistics research—where the aforementioned phrases “The rainbow is a division of white,” and “A sign from the gods to foretell war," come from.
They found that some phones are correlated with higher particle productions. The vowel /I/—which is the sound made in the words need and sea—makes more particles that /ɑ/—the sound in the words saw and hot. Vowels tend to be spoken more loudly, and the noises are made without any obstruction in the vocal tract—which means there’s no barrier to particles being emitted.
Languages all have different phones that are used most commonly, as well as different rates of speech. Ristenpart and his colleagues wrote in their paper that their research suggests that “airborne transmission of respiratory pathogens could be modulated by the phonetic characteristics of the language spoken.”
Differences in phones in the Japanese language isn't enough to fully explain why their COVID-19 cases have decreased so rapidly. Ristenpart stressed that there are many, many factors that determine transmission besides the way someone speaks. It’s not just how many particles are released, but where you are, what the environmental air flow is like, how long you’re exposed to the infectious particles, how many particles there are, and then how the individual's immune system responds if those particles end up in their bodies.
“So many things go into that,” Ristenpart said. “Maybe the choice of language spoken plays some role, or maybe Japanese people on average tend to stand a little bit further apart, talk more quietly, or maybe they don't speak as much.”
Even Inouye, who first published about how Japanese could emit less particles, offered a different hypothesis a year later, in a letter to Improbable Research: "My new hypothesis: The Japanese do not shake hands, but bow," he wrote.
Pathogen spread might be slightly influenced by language, but also by certain people who just produce more particles when they talk. Research from 2019 also found that a small number of people were “speech superemitters,” who regularly produced much more particles than the others, despite what language they were using or how loud they were talking.
“We called that the loud-mouth hypothesis,” Ristenpart said. “Some people speak pretty loud, pretty often, and they’re going to emit more. And then some people, for reasons we don’t understand, emit a heck of a lot more than others.”
There could be some unknown physiological reasons that makes someone a speech superemitter, and we need more research to find out if that contributes to make someone a “super spreader”—individuals that infect many others.
“It would be wise to avoid extended face-to-face conversations with other people unless you are far apart and in a well-ventilated space, including outdoors."
Still, Ristenpart thinks that language—what sounds are made as well as the volume at which they're spoken—should be an epidemiological consideration that’s factored in, especially when considering why or why not a certain country has high or low infection rates.
In a recent editorial in Science, scientists from National Sun Yat-sen University in Taiwan and University of California San Diego wrote that they believed a large proportion of the spread of COVID-19 "appears to be occurring through airborne transmission of aerosols produced by asymptomatic individuals during breathing and speaking."
They argued that for society to open up, measures needed to be taken that recognized and addressed this aerosol transmission—like asking everyone to wear a mask. The World Health Organization recommendations for 6 ft distancing and hand washing are based on studies from the 1930s, which showed that large droplets from coughs and sneezes fall quickly. “However, when these studies were conducted, the technology did not exist for detecting submicron aerosols,” the authors wrote. “Given how little is known about the production and airborne behavior of infectious respiratory droplets, it is difficult to define a safe distance for social distancing."
Ultimately, it's much less risky to meet outdoors, at a minimum of 6 ft, because that's because “when outdoors, speech droplets rapidly disperse into the space around us, so provided we practice physical distancing, the probability of transmission likely becomes lower," Yuan said.
“It would be wise to avoid extended face-to-face conversations with other people unless you are far apart and in a well-ventilated space, including outdoors,” Linsey Marr, a professor of civil and environmental engineering at Virginia Tech, told The New York Times.
Yet in May, the Trump administration released guidelines for the reopening of churches and other houses of worship. They removed warnings about singing that originally recommended considering “suspending or at least decreasing use of choir/musical ensembles and congregant singing, chanting, or reciting during services.”
This is concerning, given the explanation the CDC gave for why the people at the Washington choir practice were so susceptible to infection: “Members had an intense and prolonged exposure, singing while sitting 6-10 inches from one another, possibly emitting aerosols.
It exposes another difference from Japan that has more to do with public health policy than what's being said. Even at Tokyo Disney Resort and Universal Studios Japan, new safety measures recommend visitors to try not to scream on roller coasters.
We still don’t know for sure which activities we do where we exhale, like breathing, talking, coughing, yelling, singing, or sneezing, contribute the most risk when it comes to transmitting disease—for COVID-19 or future disease outbreaks.
Ristenpart has one idea, though: “One possible, very easy, public health measure would be just to recommend everybody to use their library voices."
Follow Shayla Love on Twitter.