Beyond Words - Language Blog

Arbitrary Sounds? Don’t Be Saussure

Scientific investigation is a slow process. Rarely are long-established ideas smashed with one individual research paper that causes everyone to turn their heads. The study of sound symbolism (also called phonosemantics) follows suit. One of the founding principles of modern linguistics, as put forward by Saussure, is the idea that sound is arbitrary: the combination of phonemes that make up the word ‘cat’ could just as easily be used to mean a dog, or a type of tree, or the feeling of elation.

This seems to hold true within individual languages. There’s nothing stopping one set of sounds from meaning any particular concept. Language is a constantly developing consensus.


But an increasing body of research is showing that there are tendencies for specific meanings to feature specific sounds more than others, in a way that cannot be explained by things like genetic relationships, areal features (features being spread across geographical areas), or borrowing. The latest, and perhaps largest, attempt to demonstrate this was published in PNAS with the title “Sound–meaning association biases evidenced across thousands of languages”. The authors looked at basic word lists from two thirds of the world’s six- or seven-thousand languages, and found statistically verifiable evidence that a number of key concepts, including physical properties and body parts, routinely are represented by words that use or avoid certain sounds.

It’s easy, when looking at the media snippets covering this paper, to say “English doesn’t have an ‘l’ in its word for tongue” and write off the research. The research, though, only points to a statistical tendency, and doesn’t pretend a 1:1 correlation. They are also very careful to be conservative, and not to generalize across the data.

Magnitude Sound Symbolism

Still, significant evidence for the cross-linguistic (between languages) phonosemanticity of vocabulary like body parts is fairly new. Much more attested, since Sapir’s work on Native American languages, is physical descriptors, or what is known as magnitude sound symbolism. Across the world, words indicating smallness, intimacy, and a bunch of related concepts tend to have vowels like ‘ee’, while words meaning largeness tend to have open back vowels like ‘ahh’. Our intuitions about this have been demonstrated with invented words.

Some languages, like Japanese, Korean, and a wide variety of African languages, have systems of sound symbolism that form their own class. These are commonly called ideophones, and research has shown that they tend to be colloquialisms, emotions and otherwise artful vocabulary. ‘Pika-pika’ (meaning ‘shiny’ in Japanese) is one that English speakers are likely to be familiar with for being the catchphrase of Pikachu. All languages, it seems, have onomatopoeia, but for these languages, such words aren’t reserved just for child-like speech. In Japanese, speakers can even invent words to describe particular emotions or sensations that are correctly interpreted by others.

The applications are interesting, with evidence suggesting that such words in Japanese are easier for children and non-native speakers in general to pick up, and so should be taught early on to make language acquisition faster.

I started investigating for this article based on a sentence in one write-up that described the research on two-thirds of the world’s languages as “a study that shatters a cornerstone concept in linguistics”. Honestly, it doesn’t shatter Saussure’s concept of the arbitrariness of sound. That concept has been slowly scraped away and refined to something less generalist, and this research just helps solidify the linguistics field of sound symbolism.

Paul Sutherland writes about endangered languages, sociolinguistics and related phenomena for ALTA Language Services. He is a linguist, photographer and writer with a passion for supporting endangered language communities. To this end, Paul has an MA in Language Documentation & Description from SOAS and has worked with groups including language archives, teaching material developers and UNESCO.