Sifting through two-thirds of the world’s languages, scientists have discovered a strange pattern: Words with the same meanings in different languages often seem to share the same sounds — even when those two languages are completely unrelated.
The findings, published in Proceedings of the National Academy of Sciences, run counter to a long-held idea in linguistics and could complicate the work of researchers trying to trace the history and evolution of the world’s languages.
If you look at basic words across unrelated languages from disparate parts of the world, you’ll typically find they sound nothing alike, the study authors point out. For example, “ptitsa,” “ndege” and “tori” all mean “bird” – in Russian, Swahili and Japanese, respectively.
“The idea that there is essentially no relation between sound and meaning has [existed for] over 100 years now,” said lead author Damian Blasi, a language data scientist at the University of Zurich in Switzerland and the Max Planck Institute for the Science of Human History in Germany. “And it strikes at something really intuitive.”
Researchers often discount the idea that sounds might have some relationship to the meaning of their words in part because it encourages half-baked thinking that could lead to flawed science, Blasi said.
“I think the reason why some of these ideas have a bad reputation is because people have proposed a lot of very sloppy theories,” Blasi said.
Certainly, you might find related words within a language that sound alike – think glance, glimmer and glare in English, which all have to do with vision and begin with a “gl.” But that doesn’t mean you’d find the same “gl-” cluster across other languages as well.
Still, Blasi and his colleagues realized there wasn’t a whole lot of data backing up this claim either way. But advancements in computing and modern statistical methods now mean that, instead of comparing a few — or a few dozen — languages, the scientists could do what generations of linguists before them could not: analyze thousands of language data sets at once.
The team studied nearly two-thirds of the world’s 6,000-plus languages using word lists covering about 100 shared basic concepts, checking to see if similar sounds kept cropping up. (For example, “snow,” while it’s a basic concept, is not shared, since many languages arose in places that never have any; but “rocks” are a ubiquitous feature of natural human habitats.)
Of course, plenty of languages share words that have similar sounds because they’re either “descended” from the same original language (such as Spanish “hospital” and French “hopital,” both of which arose from Latin) or because they’ve borrowed heavily (as English did after the French invaded in the Norman Conquest of 1066). The researchers had to make sure to rule out sound patterns that were similar simply because two languages were related.
The problem is, some researchers disagree about which languages are related, and how. So the scientists based their analyses on two different linguistic family trees, looking for patterns that were clear enough to survive under both models.
The researchers found, to their surprise, a number of sound/meaning relationships that cropped up across unrelated languages.
For example, words for “tongue” often tend to have an “l” or a “u” (such as the Spanish “lengua”). Words for “nose” often have an “n” sound. Words for “round” often have an “r,” and “small” is associated with “ee” sounds. These preferences weren’t universal – far from it – but they appeared over and over again.
Researchers are already fielding possible explanations for why these patterns crop up: perhaps “l” is associated with tongues and “n” is associated with noses because those body parts play a role in making those sounds, Blasi said. It might be that other associations are due to synesthesia — an ability to link perceptions from one sense to another (for example, seeing colors when hearing music).
But the scientists avoided trying to pin down explanations for the phenomena they observed.
“We don’t take any stance with respect to the origin,” Blasi said. “We just say, ‘What is the data saying? Is there any strong association, yes or no?’ Well, it seems that there is. Why? That’s a different question.”
It’s far too easy to come up with unfounded theories to explain such patterns, Blasi said. He even tested this problem out at a conference a few years ago by making up links between words and sounds – “fire” and the sound of “t,” for example – and would randomly ask people to explain these associations. People found a reasonable explanation every time – for a relationship that did not actually exist.
Instead, scientists may have to probe these preferences experimentally, Blasi said. Perhaps you could make up words for two different “alien” languages, and test people to see if they found it easier to remember words with the preferences found in this paper.
The findings may have implications for linguists who look for shared sounds (and predictable changes) in current and older written languages to try to reconstruct their ancient, long-gone ancestors. Blasi’s study shows that some of those shared characteristics between “sister” languages may not be inherited from a “mother” language; instead, they could have arisen independently, simply because humans tend to like certain sounds with certain words.
“That calls into question some of the attempts that people have put forward in order to determine the prehistory of many linguistic families,” Blasi said.
This could throw a curveball into the work of researchers tracing our linguistic heritage.
“The more we look into languages, the more we learn that they are extremely complex, and that we have to take them seriously,” he said.
MORE IN SCIENCE