Directorate-General for Research & Innovation logo Horizon: the EU Research & Innovation magazine | European Commission logo
Receive our editor’s picks

Settee or sofa? Meet the words earmarked for extinction

The MOTHERTONGUE project treats the evolution of words like the evolution of genes. Image: Shutterstock/ Pressmaster
The MOTHERTONGUE project treats the evolution of words like the evolution of genes. Image: Shutterstock/ Pressmaster

‘Davenport’, ‘settee’, ‘couch’ – less common words are more likely to go out of fashion than precise terms, according to researchers who are borrowing techniques from genetics to study the evolution of language. Their results could help inform the way that computers communicate.

‘We take it for granted that everyone calls the first month of the year “January” but there is still disagreement about what to call a sofa,’ said Professor Mark Pagel, principal investigator of the MOTHERTONGUE project, which is studying the evolution of language around the world.

It’s because, for example in English, the precise meaning of the word ‘January’ makes it more important that everyone uses the same word, so other competing words quickly fall out of use.

However, since the meaning of ‘sofa’ is neither well-defined nor particularly important for communication, alternative words may gain more momentum than others through usage, and the team’s most recent work suggests that cultural pressures also play a role.

Hence American English speakers prefer ‘couch’, while several other words remain in minority usage, such as ‘settee’ and ‘davenport’.

The project, which is funded by the European Research Council (ERC) and finishes in 2016, studies these changes by treating the evolution of words like the evolution of genes, where less efficient traits become extinct through natural selection.

‘The word “dirty” in the Indo-European languages changes every 600 years or so.’

Professor Mark Pagel, principal investigator of the MOTHERTONGUE project

In some cases, the words may be equally good, as with ‘sofa’ and ‘couch’, whereas in others one word is simply better, as was the case with ‘January’.

‘The word “dirty” in the Indo-European languages changes every 600 years or so,’ said Prof. Pagel.

Researchers can also use their theories of how language changes to date historic texts. ‘We also used the same approaches of studying rates of word evolution to estimate a date for Homer’s great epic work The Iliad,’ said Prof. Pagel. ‘We date this work on the basis of vocabulary similarities and differences to modern Greek and to the extinct Hittite languages. Our estimated date of 762 BCE conforms closely to linguistic scholarship.’

Follow the sounds

As spellings change, the best way to track words is through the way they sound. To do this, the ERC-funded EVOLAEMP project is using computer programmes that can analyse phonetically spelled words and trace their usage in different languages such as the German ‘wasser’ to English ‘water’, showing how sounds often remain similar as languages change. It’s similar to the techniques that biologists use to track proteins through different organisms.

‘Basically we treat sequences of sounds as if they were sequences of proteins,’ said Prof. Gerhard Jäger, principal investigator. Though there are many features of language that change over time, ‘sound change and word replacement are a good starting point because they are fairly easy to identify and there are so many instances of them.’

This automated, phonetic approach requires meticulously collected word lists from a many different languages, of which there are estimated to be up to 7 000 in the world. In addition to the 40-word lists from 6 000 languages collected by researchers at the Max Planck Institute of Evolutionary Anthropology in Leipzig, the team is also producing 1 000-word lists from 100 Asian and European languages.

‘Forty words per language … is enough to get broad patterns, but for fine details you need more data,’ Prof. Jäger said. By the time the project ends in 2018, the team hopes to create a theoretical framework to track language evolution.

Robot chat

One reason that researchers are studying the evolution of language is that their findings can be used to improve communication between computers, which has important applications for the Internet of Things.

‘Our research could have an enormous impact on how autonomous systems will communicate with each other in the future,’ said Professor Luc Steels, coordinator of the LANGEVO project, which finishes in 2015.

The project is looking at how a person’s genetic make up is linked to their ability to create and learn language. The idea is to provide a clearer picture of how cultures, individual abilities and genetics guide language evolution, in order to work out the most efficient way to design a computer language.

‘We try to understand by what strategies humans come up with new words, meanings, and grammatical constructions, how these inventions can propagate through learning between language speakers, and how a population can reach a consensus without a central coordinator,’ he said.

As a result of all of this, researchers are increasingly certain that even though individual words will change, the sounds and grammar of future languages are unlikely to. That means that sentences will still be ordered as subject-verb-object ‘She wrote the letter’ or subject-object-verb   ‘She the letter wrote’ in most languages. ‘Alas,’ said Prof. Pagel, ‘we cannot say exactly what the languages or words will be.’

More info