The language of colour, kinship and climate

Monday, Apr 1, 2019, 03:12 AM | Source: Pursuit

Charles Kemp

The number one thing I’m interested in is how people learn, think and reason. I’ve explored this question by building computational models of cognition and then running experiments to test them.

Melbourne has always been strong in mathematical psychology and recently started the Complex Human Data Hub, which focuses on computational analyses of human behaviour.

Languages provide a near-optimal trade-off between complexity and informativeness. Picture: Getty Images

I’m a University of Melbourne graduate who moved to the United States for my PhD. I spent 16 years there - ten of them as an academic at Carnegie Mellon University. That was a great fit for me because it’s a place where computation affects every field. The historians, the English professors, the philosophers - many of them think computationally in one form or another. I came back to Melbourne to join a group of researchers who focus on exactly the kind of psychology I’m interested in.

If you look at languages across the world, you see striking similarities and differences. A really interesting difference is the number of colour terms that different languages use. English has 11 basic colour terms – red, blue, orange and so on – but some languages have only two basic colour terms to differentiate between warm and dark colours.

The World Color Survey includes colour naming data for more than a hundred languages from around the world and is based on asking people who speak different languages to provide names for a standard set of colour chips.

English has two terms for grandparents. Chinese has four terms that include distinctions such as your mother’s mother and your father’s mother. Some languages have reciprocal kinship categories, so the term I would use for my grandfather is the same word he would use for me.

Working in this area is fascinating because I collaborate with people from different backgrounds. My longstanding collaborator is Professor Terry Regier from the University of California, Berkeley.

He has a PhD in computer science and is a Professor of Linguistics and Cognitive Science at Berkeley’s Language and Cognition Lab. Recently Terry and I have been fortunate to work with Noga Zaslavsky and Professor Naftali Tishby, who are computer scientists from Hebrew University.

Some languages have only two basic colour terms to differentiate between warm and dark colours. Picture: Getty Images

We all want to understand what makes a good system of categories. For example, if you think about colour, it would be possible to have different names for every hue of colour you could imagine but you would have to remember thousands of names. On the other hand, having one term for colour would be easy to learn but not especially useful.

A good system of categories needs to be simple yet informative. There’s a principle called informativeness, which is about communicative precision. We propose that languages provide a near-optimal trade-off between complexity and informativeness. Languages strike a balance between giving us the information or facts we need without overwhelming us with unnecessary detail.

It’s fascinating to think about how languages work and why we see the patterns we do. There’s an intellectual thrill when you take a data set that seems complex on the surface and then develop a computational model and find an approach that lets things fall into place with a clean underlying order.

The question of how we can better understand other cultures and languages is also important. This is particularly true in places where people from many different backgrounds come together. Moving back to Australia while doing this work is interesting because of the amazing linguistic diversity in this part of the world.

I’ve been thinking about season naming across languages. I’m doing this work along with Terry Regier and Dr Alice Gaby, a linguist from Monash University.

A current challenge for us is expanding on previous analyses of colour and kinship to see how well our theory works with other kinds of categories. This is interesting from an Australian perspective because Aboriginal languages often have different sets of season categories.

English has two terms for grandparents, but other languages like Chinese have four. Picture: Getty Images

In many places around the country, the European seasons don’t fit the local climate very well. We’re interested in exploring the relationship between Aboriginal seasonal classifications and the local environment. A book called Sprinter and Sprummer - Australia’s Changing Seasons by Professor Timothy J. Entwisle, suggests we adopt more suitable seasonal categories that are in tune with our plants, animals and climate. The CSIRO has also done great work in organising and collecting data on Aboriginal seasonal terms.

But a lot of languages are in danger of dying out. There are around 7 000 languages in the world and it’s thought that around half of them will disappear over the next century.

Some of the most important work I know focuses on documenting and preserving the knowledge that is locked away in these endangered languages.

Banner: Shutterstock