To an extraterrestrial, human language would be nonsensical. For starters, there are thousands of languages, many of which sound nothing like each other, and each language has thousands of words that appear to have no rhyme or reason. By creating "alien languages" and teaching them to volunteers, researchers are trying to better understand what about a language makes it learnable and how languages might have evolved.
Words don't necessarily sound like or look like the object they connote. For example, the word "microorganism" is long, although the creatures themselves are tiny, whereas the huge whale has to make do with a very short moniker. Despite this, many languages do have what Padraic Monaghan of Lancaster University in the United Kingdom, calls "pockets of systematicity." Many English words beginning with "sn," for instance, tend to have something to do with the nose: sneeze, snort, snot. In many languages, vowels made with the back of the tongue, such as "o" and "ah," tend to appear in words that describe something big (boulder), whereas vowels made at the front of the mouth, such as "ee," often denote something smaller (flea). It's unclear why these "pockets" exist: whether they're accidents or are somehow tied to language learning. "Maybe they get people going" when they're learning a new language, Monaghan speculates.
In a presentation given at the British Science Festival here last week, Monaghan talked about his efforts to answer this question. His group used computer algorithms to create "alien languages," which they taught to human volunteers using words paired with pictures. Some of these languages were very systematic in the way they sounded—big things were always described by long words, for example. Another language was completely arbitrary. The volunteers had a hard time learning either of these. Then the researchers switched up the sounds within the words, creating a language with words that were half-arbitrary and half-systematic, such as describing big things with short words, but long vowel sounds. The participants learned this language more easily, which Monaghan believes suggests that both features are needed. Language has to be arbitrary, he suggests, because if words denoting similar objects all sounded alike, we would be more likely to confuse those objects. When the researchers scoured through a list of 5000 of the most common nouns and verbs in English and French and mapped the sounds that make up each one, they found that, on the whole, these words have both systematic sounds and arbitrary ones within each word.
Linguist Simon Kirby of the University of Edinburgh in the United Kingdom, who is also using computer-generated languages to study learnability, isn't surprised. "That's what's special about language as opposed to any other system of communication; you can understand it because it's made up of interchangeable parts," he says. He believes that our brains are designed to put these parts into categories.
Kirby, whose results were also presented at the meeting, explored how language evolves to become easier to learn. Kirby gave his volunteers long lists of random syllables and told them they were words that described pictures. The experience is not like learning a real language, Kirby says: because they're entirely random, those words are "incredibly hard" to remember.
But it got easier: By testing the volunteers on the new language and incorporating their answers into a slightly modified language for the next group, they found that words started to emerge that put the objects into categories. For example, the volunteers created words or sounds within words that indicated an objects' color or its shape: words ending in a "k" sound, for example, might be red. The human brain, Kirby concludes, prefers to categorize things and does this by making arbitrary language more systematic.
Kirby and Monaghan see this as evidence that learnability is selected for through a process much like natural selection. "You make mistakes, but those mistakes aren't random," Kirby says. They modify the language slightly, making it "easier for next person to learn."
Reardon, Sara. 2011. "Calculating the Language of Babel". Science Magazine. Posted: September 23, 2011. Available online: http://news.sciencemag.org/sciencenow/2011/09/calculating-the-language-of-babe.html