麻豆传媒

(Image: Michael Heissner/Getty Images)

Sarcasm may be called the lowest form of wit, but it still exercises psychologists, linguists and computer scientists hoping to build automatic irony detectors

IT WAS a classic case of mistaken identity. At the start of his first year at university, Oren Tsur got an unexpected email. It asked if he might consider chairing a prestigious event based on his contribution the year before. Tsur can鈥檛 remember his exact reply. 鈥淚 answered jokily,鈥� he says 鈥� in a way he was sure would make it clear they had the wrong guy.

Only the organisers didn鈥檛 get the joke. And they remained undeterred, despite Tsur鈥檚 increasingly sarcastic replies. Finally they asked what new research he might share at the upcoming event. Still baffled, he deliberated before finally writing: 鈥淚鈥檓 working on detecting sarcasm and irony in written text.鈥�

As a computer scientist who studies language processing, Tsur has since gone on to do just that. In an era of instant, often impersonal electronic communication, understanding why we often say the opposite of what we mean, and how other people pick up on it, is becoming more crucial than ever. At the same time, we are gaining new insights into how we produce and process irony and its more cutting cousin, sarcasm. It might be the lowest form of wit, but it also seems to be a higher form of brain workout.

One thing is for sure: irony and sarcasm are pervasive. , a psychologist at the University of California, Santa Cruz, has found that nearly 1 in 10 remarks between friends is not intended at face value. That starts early: developmental studies show that we can grasp ironic sentences around age 5, and even younger children make gestures such as slapping their forehead in mock despair. And although its use may vary between nations and regions 鈥� 鈥� sarcasm is present in all cultures.

鈥淧eople will often say, I went to Japan and nobody understood my sarcasm, I don鈥檛 think Japanese people are sarcastic,鈥� says , a cognitive psychologist at the University of Calgary in Alberta, Canada, who has examined the way children develop a sense of sarcasm. 鈥淥f course, that鈥檚 not true at all. There鈥檚 lots of really good irony research coming out of Japan, it鈥檚 just the forms are slightly different.鈥� Each culture seems to have its own cues. In English, for example, an ironic remark generally starts with a lowered voice, but for Cantonese speakers the opposite is true: elevated pitch signals that a comment shouldn鈥檛 be taken literally.

Evolutionary psychologists have suggested many possible explanations for our love of sarcasm. It might allow us to test that others have a similar outlook or level of intelligence, or perhaps enable us to vent frustrations and criticise one another without being overtly hostile. In a series of studies in 2011, researchers in Israel examined the different ways you might complain to customer service representatives. They found that whereas outright hostility hindered the reps鈥� ability to come up with creative solutions, .

The result is all the more surprising, given that we must jump through many more mental hoops to understand a sentence whose meaning relies on more than its words. Neuroscientists have shown that ironic and sarcastic sentences often take associated not just with verbal understanding but also with interpreting intention and social information. 鈥淚f all of a sudden someone starts missing jokes, that鈥檚 a sign that something might be going awry in their brain,鈥� says Kate Rankin, a neuropsychologist at the University of California, San Francisco. Not clocking ironic intention can be an early sign of some forms of dementia, and people with autism also struggle with sarcasm 鈥� although Jonathan Tarbox, a behavioural psychologist with the Autism Research Group in Thousand Oaks, California, has found that a , even to the point where they start using irony themselves.

鈥淢issing jokes can be a sign that something is going awry in the brain鈥�

Great for insomniacs

No matter how well practised we are, at some point most of us will find ourselves skewered by a barb we failed to understand. As Tsur discovered, the written word is particularly treacherous territory. Perhaps that鈥檚 because we are not privy to the important cues that come so naturally in speech: you can鈥檛 roll your eyes in an email. Such difficulties eventually led Tsur to design an algorithm he hoped would automatically detect sarcasm. His first attempt 鈥� based on how well the star ratings provided by purchasers on Amazon matched the language they used in their reviews 鈥� was not a success. But when he teamed up with two colleagues, they cracked it. The trick was providing the computer with the right examples.

The team hand-picked a few hundred clear examples of sarcastic Amazon reviews 鈥� 鈥淭his book is great for insomniacs鈥�, for example 鈥� and fed them into a machine-learning algorithm. By the end of its education, it could correctly identify 78 per cent of sarcastic Amazon reviews and 83 per cent of sarcastic tweets on Twitter. Tsur thinks the algorithm fared better on Twitter because of its 140 character restriction. 鈥淭witter lacks context. So, when people are trying to be sarcastic, they make it clearer.鈥� In some cases, users even added #sarcasm to their tweets.

Tsur, , has largely left sarcasm research to explore other areas of language processing, but others have picked up the baton, with many more . People who have difficulty picking up the subtle nuances of speech might benefit, too: at the City University of New York has developed software that can in more than 80 per cent of cases.

There are many possible applications, from marketers hoping to keep track of public reactions to their products to people who want to spend time in a different culture and learn a new language; eventually, it may be essential for computer translation software. After Tsur published the research about his algorithm, he also got several emails from people with autism who thought it could be a useful aid.

But we are unlikely to have automatic sarcasm detectors built into the web any time soon. Ambiguity is part of the richness of language. Perhaps that is why people have also resisted long-standing attempts to introduce . 鈥淚t鈥檚 a bad idea 鈥� it鈥檚 like explaining a joke,鈥� says Tsur, because it takes away the pleasure of sussing it out yourself. Or as one commenter debating these 鈥渟arc marks鈥� deadpanned, 鈥淭hat is the best idea ever. I can鈥檛 imagine life without it.鈥�

This article appeared in print under the headline 鈥淪arcastic? Moi?鈥�