From whong to quingel: the science of funny words

Without scrolling down, have a think about which word in these pairs you find funnier.

Quingel v Heashes
Prousup v Mestins
Finglam v Cortsio
Witypro v Octeste
Rembrob v Sectori
Pranomp v Anotain
Fityrud v Tessina

If you’re like most people, it will be the first in each pair. Researchers led by Chris Westbury at the University of Alberta found that 56 English-speaking subjects rated those on the left as being more funny than those on the right. What’s amazing is that these nonsense words (NWs) were not designed by a human being with their potential for comedy in mind. They were produced by a computer program using a simple algorithm.

It looks like Westbury et al, whose paper appears in the January 2016 edition of Memory and Language, have developed a reliable method of machine-generating humour. How on earth did they do it?

Well, what they’ve found is a strong inverse correlation between funniness and a property called entropy. This is a way of expressing how usual the letters in the NW are – so the less commonly they’re used in English, the lower the total entropy of the NW. To put it another way, the less “wordy” these NWs are, the more they strike us as humorous.

Read more: The Guardian

Leave a Reply

Your email address will not be published.

fifteen − 6 =