Based on the participants’ responses, the researchers created a model to predict how humorous a made-up word would be. Eliminating strings that sound like actual obscene words (like “dongl,” “shart,” “focky,” and “clunt”), which risk inducing giggles on their own, the researchers tested their model on 56 fluent English speakers, and found that unlikely words were reliably deemed funnier.

Nonsense words like “himumma,” “suppopp,” “finglam,” and “pachang” were deemed most funny, while “advical,” “rousent,” “menclar,” and “materal” were judged least.

Chris Westbury, lead author of the study, tells Quartz that the results evoke Schopenhauer’s idea that humor comes not just from something incongruous but from a violation of our expectations.