Exploring the pathways of our lexicon
When Autocorrect is Not So Correct
My latest On Language column for The New York Times Magazine explores a topic that any owner of smartphone knows too well: the often bizarre behavior of autocorrect, which can "miscorrect" what you type into unexpected and outrageous output.
As Farhad Manjoo wrote in Slate last year, "Autocorrect gets no respect. Every day, you dash off dozens of messages on your mobile phone, and most of the time, you do it wrong—you mistype, misspell, or make some other kind of error that's bound to cause you great embarrassment. In the vast majority of cases, your phone steps in to save the day." But, as New York Times tech-blogger David Pogue notes, when autocorrect gets it wrong, "it can be really wrong."
Now the cataloging of the most embarrassing smartphone miscorrections has become something of a cottage industry, with the wildly popular website Damn You Auto Correct leading the way. I interviewed DYAC's founder Jillian Madison for the column, and she sounded a bit frazzled by the hundreds of submissions that she receives every day. (There's even a DYAC app for the iPhone that lets you submit screen grabs of egregious exchanges directly to the site.) DYAC is spawning a book, to be published by Hyperion in March, so there's clearly an endless appetite for the unintentional humor that our phones are providing us.
I've had a long-standing interest in the linguistic trouble that "helpful" computer algorithms can get us into. Automated spellchecking in word-processing programs can often create howlers, though perhaps with less frequency than smartphone autocorrect. The so-called "Cupertino effect" has been an unfortunate element of spellchecking ever since Microsoft Office 97 couldn't recognize unhyphenated cooperation and instead replaced it with Cupertino, the name of a California town. I first learned about the Cupertino effect in 2006 from a former writer for the European Union. You can still manage to find documents online from the EU and other international organizations that have the word Cupertino where cooperation is intended, such as this NATO document that has the line, "The Cupertino with our Italian comrades proved to be very fruitful." (I spoke about the Cupertino effect and other editorial miscorrections on the National Public Radio show Radiolab last year, in an episode entitled, "Oops.")
With the advent of smartphones with virtual keyboards, particularly Apple's iPhone and the Android phones from Google, the Cupertino effect is running amok on an unprecedented scale. And there are many obstacles faced by designers of smartphone autocorrect features that earlier spellchecker designers didn't have to deal with. For instance, as I mention in the column, there's the problem of words with letters repeated for emphasis, a common feature of text-ese. Thanks to autocorrect's tendency to look for fat-finger errors, yeahhhh can get changed to uranium because y is close to u on the virtual keyboard, e is near r, and h is in the vicinity of i, u, and m. You can test this out yourself: try typing wheeeeeee with seven e's into your phone and you might get autocorrected to shredders; add one more e and whereafter may instead be the suggested substitution.
Despite all of this, the situation with smartphone autocorrect is destined to get better. Early word-processing spellcheckers had a lot of hiccups, but their utility has been enhanced with innovations like the "contextual speller" rolled out for Microsoft Office 2007, which can catch correctly spelled words that are erroneous in context (I've called such stealth errors "miss steaks," based on a famous spellchecker poem). As the storage capacity of smartphones increases, they'll be able to work off of such contextual data to provide more accurate correction of our input. Until then, however, let's all enjoy the laughs that wayward autocorrect can generate.
You can hear me talk about autocorrect goofs in my recent appearance on the NPR show "Talk of the Nation" here.