If you visited Iceland and asked someone what they called the smelling organ in the middle of their face, they'd tell you, nev. In Japan, it's hana. To Sar speakers in southern Chad it's kon, and among the Zuni tribe of the southwestern United States. it's noli. In fact, you could go to more than 1,400 places around the world, question speakers of more than 1,400 different languages, and hear 1,400 words that contain the sound "n." But all of them mean the same thing: nose.
That's one of the findings of a sweeping study in the Proceedings of the National Academy of Sciences Monday, which found evidence of strong associations between the sounds in words and the ideas they represent in completely unrelated languages from all corners of the world. Despite a long-standing assumption in linguistics that the sounds we pick to signify certain concepts are arbitrary, the researchers argue that at least some associations are more universal than you'd think.
"Most models for how words come into our lexicon are predicated on this assumption that the sound doesn't tell you anything about what it represents," said Jaime Reilly, a cognitive psychologist and speech pathologist at Temple University who was not involved in the study. "So the really neat thing about this paper is it sort of questions whether that arbitrariness assumption actually holds across all words."
"It's going to end up being a very important study," he said.
A century ago, Ferdinand de Saussure — one of the founders of modern linguistics — wrote about the arbitrary relationship between signals (or the words we use) and the signified (the concepts we're trying to describe). For example, there is nothing inherent about the term "cat" that calls to mind the fluffy, mildly standoffish felines we keep in our homes. That we can say "cat" and other English speakers know what we're talking about is a result of convention.
It's a hallmark of what makes human languages unique among animal communication systems, said Morten Christiansen, a language scientist at Cornell University and one of the authors of the PNAS study. "In other animal systems there’s more a direct relationship between what a signal means and what comes across."
Humans, on the other hand, are sophisticated enough to come up with tens of thousands of otherwise meaningless sounds and be able to keep track of exactly what they stand for.
Still, there have been hints that the sounds we choose for certain concepts aren't entirely random. A series of studies starting in 1929 have documented what's called the "bouba/kiki" effect: People from societies across the world almost universally associate round shapes with the made-up word "bouba" and spiky shapes with the non-word "kiki." Within languages, research has shown that sounds can become associated with ideas — for example, English words having to do with sight, like "glance," "glimmer" and "glare," all start with the sound "gl." And Reilly was co-author on a recent study that found that English speakers can distinguish between words for concrete terms and those representing abstract concepts, even when the words come from languages they don't understand.
This notion that vocal sounds carry meaning in and of themselves, and that meaning can be mapped onto the ideas they're used to represent, is called "sound symbolism." Though linguists have acknowledged its existence for decades, "it's been sort of marginalized as not being super contributory to how languages evolve," Reilly said.
Christiansen's work pushes sound symbolism back to the forefront. He and his team, which included statisticians, neuroscientists, physicists and computer scientists, examined the words for 100 concepts in more than 6,000 languages in search of commonalities. They weren't looking for universal rules — just examples of associations that popped more often than you'd expect due to pure chance.
"We took a big data approach," Christiansen said. "We were trying to see if, for a given concept, are people across the world more likely to use a particular sound in association with this concept."
To eliminate the influence of geographic relationships, shared sounds would only be considered symbolic if they appeared in languages from at least three of the world's six large and relatively isolated areas: North America, South America, Eurasia, Africa, Papua New Guinea and the Pacific islands, and Australia. The researchers also controlled for a false discovery rate of about 5 percent in order to weed out the possibility of false positives.
In the end, they found 30 concepts that were strongly associated with certain "signals." For example, words for "round" were extremely likely to contain the letter "r," including kokoro (from the indigenous Colombian language Ika), biribil (from the Basque language of northwestern Spain) and aridonda (from the Chamorro language spoken in the Mariana islands). "Sand" was associated with the "s" sound.
Some concepts were negatively associated with particular sounds — "skin" is especially likely to not contain "m" or "n," and "eye" probably won't contain the sound "a." Also, not all of the rules apply to all languages — in many cases, English was an outlier. So if you're reading this article, you might be surprised to learn that most words for "dog" contain the sound "s," and "you" usually doesn't have an "o" or "u" in it.
"What we think is the really exciting part, is independent of linguistic lineage, independent of whether languages are historically related to one another and independent of geographical location we see these signals showing up over and over again across the world," Christiansen said.
Exactly how associations evolved is fodder for several lifetimes of future research, he added. Christiansen and his colleagues considered whether it could be a result of evolution from some prehistoric protolanguage spoken by the earliest humans, but various statistical analyses ruled that possibility out.
That leaves "something psychological and universal as a factor that gives a slight edge to a few sound-meaning associations," Johanna Nichols, a prominent linguist and professor emerita at the University of California in Berkeley, wrote in an email.
It seems there is something biological about our association of certain sounds with ideas. For a few of Christiansen's concepts, the explanation appears straightforward. The "n" in nose (and nev and hana and kon and noli) is nasal, requiring that we speak through the very organ we're trying to describe. Things that are small make high-pitched, squeaking sounds, just like the "i" in tiny, sagheer (Arabic) and liten (Norwegian). For other associations, the link between signal and signified is more opaque.
This doesn't mean that Saussure was completely wrong about arbitrariness. Though words like nose may share some commonalities, most of the terms in the average adult's 80,000-word vocabulary represent abstract concepts that aren't easily represented with concrete sounds. When words for the same concept do resemble each other, it is usually a sign that they share an etymological root.
"It is still true that for the most part the relation of a word's meaning to its consonants and vowels is arbitrary. That is what makes historical comparison possible in the first place," Nichols wrote. For example, the fact that the words for "two" in Romance and Slavic languages almost always start with d "isn't due to some inherent naturalness but is a quirk likely to be inherited," she continued — evidence that the languages share a family history.
"That still stands," Nichols said, "but we now have to add those grains of salt."