It’s not just your imagination. Spanish and Japanese speakers talk fast, each squeezing a full eight syllables into every second of speech, on average. Meanwhile, English in the U.S. averages closer to six, while Hungarian dips closer to five.
But as it turns out, this rate of speech is largely irrelevant. Because fundamentally, no matter how fast or slow the world’s languages sound, they each evolved to convey the same amount of measurable information every second: roughly 39 bits.
It’s a finding that you can appreciate more by studying the sharp graphs below. (They were recently rediscovered by FlowingData, though they’re originally from research published in Science Advances in 2019.)
On the left, you see a column of how fast people say words across the world. And on the right, you see a column of the actual information rate of this communication. While the left column has incredible variance because some languages are spoken faster than others, the right column lines up because they average out to contain the same measurable information per second.
How did scientists figure this out? They translated the exact same written passages into 17 different languages, as closely as possible. Then researchers recorded native speakers reading these passages around the world.
“If you record just one person in each language they may be a slow or fast speaker,” says Christophe Coupé, an assistant professor at University of Hong Kong who coauthored the paper. “But if you take a number of people and put them in the same conditions, with a shared baseline, then you compute and average the value . . . you see? It’s not only about the speech rate; it’s about the initiation density. When you cross the two, you find there is some balance in information rate, which is what really matters.”
Another way of thinking about it: People from across the world, speaking languages that are all constructed differently, read their passages in roughly the same amount of time.
Coupé points out that as obvious and logical as this conclusion may seem, it’s been a radical discovery for the linguistics community since his team began publishing similar findings about a decade ago. At the turn of the 19th century, Eurocentrism plagued researchers who believed the success of Chinese civilization was at odds with their seemingly simplistic speech—which largely consisted of shorter words with fewer syllables and pared-back syntax.
Of course we know now that European ears weren’t appreciating the tonal variances that differentiate many Chinese words. But even the very basis of this argument was grounded in racism that outweighed logic.
“The question in the end to us is, Are all languages equally effective or not?” Coupé says.
The truth is that cultures around the globe naturally designed their own distinctive approach to language. And from what we can tell, they’re equally effective . . . even if the individual words aren’t all equally quick to pronounce, even for native speakers.
In Japan, words are constructed with a consonant-then-vowel cadence that flows out of one’s mouth rapidly, while across Eastern Europe, vowels mash together in long sequences that simply take more time to untangle off the tongue.
“I’ll give you a word in Serbian: stršljen. It’s five consonants in a row!” Coupé says, pointing out that stršljen translates in English to the two-syllable word “hornet.” “So the complexity is higher [in Serbian], but you produce the words slower.”
As all these sounds play out into actual spoken communication, the disparate approaches average out. A reasonable, but nearly impossible to prove, inference is that cultures across the globe seem to have independently created languages that all sit within a sweet spot of our brain’s processing power. We like to receive verbal information at 39 bits every second, regardless of what that information sounds like. (How much is 39 bits? That’s a complicated concept to grok outside computer science. Technically it means 39 distinct pieces of yes or no information, “but of course, this does not translate very well the meaning of sentences and texts,” Coupé says.)
There are some caveats to the research. While Coupé’s team measured the sheer factual information conveyed in passages, it’s fair to say that communication is about a lot more than sharing quantifiable data. When I ask Coupé about metrics that his study didn’t measure—elements that might convey sarcasm, love, or less-quantifiable expression—he admits that the best methods in linguistics still fall short.
“What you’d really like to access, as directly as possible, is meaning. The problem is meaning isn’t easy to capture with numbers,” he says. “Information is not exactly meaning . . . though they are closely related.”
He also notes that while language’s information rate is similar across the globe, research has consistently found that men speak faster than women. Does that mean that men are faster thinkers than women? No. As Coupé’s previous research has shown, men read passages aloud faster than women. But when men and women read in their minds silently, they read at the same speed.
“It’s not that women are worse readers or speakers,” Coupé says. “[Instead], there are some interesting cultural influences and factors at hand.”