Have you got rhythm? George and Ira Gershwin’s famous 1930 song boasted that they had. Today, it has been discovered that if you speak in any of 48 languages, your cadence is – amazingly – the same as all people around the world, produced at a remarkably consistent rate of once every 1.6 seconds.
This low-frequency rhythm is stable across languages, cultures, and ages, suggesting a universal cognitive mechanism of human communication.
The findings shed new light on how the human mind structures language in time, according to a new study at the Hebrew University of Jerusalem (HUJI). The researchers suggest that their findings could have important implications for neuroscience, language learning, and speech technology.
A natural conversation flows like a dance – pauses, emphases, and turns arriving just in time. The study has discovered that this isn’t just intuition; the brain follows a biological rhythm that is embedded in our speech.
According to the study, which was led by Dr. Maya Inbar, Prof. Eitan Grossman, and Prof. Ayelet Landau, human speech across the world pulses to the beat of what are called Intonation Units (IUs), which are short prosodic phrases, most often groupings of words.
Human speech follows a universal rhythm every 1.6 seconds, study finds
They are marked by boundaries set by changes in intonation and timing, such as final slowing or lengthening. These units organize the flow of speech and help listeners understand sentences by signaling structure and meaning.
The research has just been published in the journal Proceedings of the National Academy of Sciences under the title “A universal of speech timing: Intonation units form low frequency rhythms.”
“The importance of IUs in communication has been studied in qualitative and experimental studies, most commonly within individual languages and without an explicit comparative methodology,” Inbar explained to The Jerusalem Post.
“Testing whether IUs assume all these roles universally is a fascinating yet complex research program. To carry it out, we first realized that an important bottleneck needs to be lifted: the reliance on manual annotations of IUs.
“Therefore, before studying the structure of IUs in time, we began by devising an automatic annotation protocol. We then successfully validated its performance compared to traditional manual annotations across languages and applied it to a larger corpus of speech recordings.”
THE TEAM analyzed 668 recordings in 48 languages spanning every continent and 27 language families. They did not include Yiddish, Chinese, Amharic, or Japanese, but they did encompass many widely spoken as well as rarely spoken tongues.
They found that in every language – from English and Russian to endangered languages in remote regions such as Totoli (belonging to the Austronesian language family spoken by only 25,000 people of Central Sulawesi, Indonesia) – people naturally break their speech into chunks in a similar temporal fashion, a new unit beginning every 1.6 seconds or so.
“We focused on spontaneous, informal speech recordings, aiming to capture speakers’ naturalistic behavior,” Inbar noted.
“These findings suggest that the way we pace our speech isn’t just a cultural artifact; it’s deeply rooted in human cognition and biology. We also show that the rhythm of intonation units is unrelated to faster rhythms in speech, such as the rhythm of syllables, and thus likely serves a different cognitive role.”
This is an important discovery because IUs play a critical role in helping listeners follow conversations, take turns speaking, and absorb information. They also offer children crucial cues for learning language.
In fact, the low-frequency rhythm they follow mirrors patterns in brain activity linked to memory, attention, and deliberate action, thus highlighting the deep connection between how we speak and how we think.
“This study not only strengthens the idea that intonation units are a universal feature of language,” said Grossman from HUJI’s linguistics department, “but also shows that truly universal properties of languages are not independent of our physiology and cognition.”
Landau, who holds appointments at HUJI’s psychology department and the cognitive and brain sciences department, also lectures at the experimental psychology department at University College London.
“Understanding this temporal structure helps bridge neuroscience, linguistics, and psychology. It may help explain how we manage the flow of information in the dynamic natural environment, as well as how we bond socially through conversation,” she suggested.
“As we move toward more human-like AI speech, better treatments for speech disorders, and deeper insights into neurological function, this research offers a powerful reminder: beneath the beauty and diversity of the world’s languages lies a shared rhythm, one that beats roughly every 1.6 seconds.”
INBAR RECENTLY graduated from the Hebrew University. She participated in the joint HUJI-Tel Aviv University program in linguistics and was a fellow in the PhD Honors Program at the Jack, Joseph, and Morton Mandel School for Advanced Studies in the Humanities, as well as in the Azrieli Graduate Studies program.
In 2016, she joined Landau’s Brain, Attention and Time Lab, dedicated to the cognitive neuroscience of attention and time perception.
Landau received a BA in psychology and philosophy and went on to study a master’s degree in neuropsychology at HUJI, followed by a doctorate in cognition, brain, and behavior in the psychology department at the University of California at Berkeley.
She was a postdoctoral researcher at the Ernst Strüngmann Institute for Neuroscience in Germany and has since been awarded numerous grants from the European Research Council, the James S. McDonnell Foundation, Joy Ventures, DFG, the Israeli Science Foundation, and the National Institute for Psychobiology in Israel.
Both Inbar and Landau play at least one musical instrument.
Grossman, who does not, told the Post that “when one makes a linguistic comparison, one always collects samples. There are all kinds of sampling biases and properties of how big the sample is and maybe geography.
“Ideally one needs as wide a range as possible in order to make generalizable claims. The kind of spoken data we relied on in this study is annotated carefully and thus hard to get. It would have been great to include more of the over 7000 languages of the world, but even artificial intelligence is not currently in a position to do it.”
Grossman was a postdoctoral researcher in the Ramses Project (and later a visiting professor) at the University of Liège, a Kreitman Fellow at Ben-Gurion University of the Negev, and spent 2010-2012 at the Martin Buber Society of Fellows in Jerusalem.
He also spent some time as a senior research fellow at Berlin’s Humboldt University. His research revolves around the questions, “Why are languages the way that they are, and how do they become the way they are.”