Building a Better Smartphone Keyboard
Can KALQ succeed where QWERTY fails?
Let’s admit it: two-thumb typing sucks. Even dumb-phone users—T9, anyone?—know the frustration of picking up a friend’s Droid and trying to dash off a quick text. AutoCorrect #fails are practically a 21st-century art form. For as much as mobile apps have sped up our lives, there’s still nothing slower than touch-screen typing.
Enter KALQ, a new keyboard layout optimized for thumbs. Designed by a team of German computer scientists to replace the universally popular QWERTY keyboard on phones and tablets, KALQ dramatically increased the word-per-minute rate at which test subjects were able to type.
What’s surprising is not KALQ’s superior performance, but that we’ve put up with two-thumb QWERTY typing for as long as we have.
The trick to building a better keyboard, according to the researchers, is to maximize the number of alternate-thumb keystrokes—try typing “minimum” on your QWERTY iPhone and you’ll understand why—and minimize overall “thumb travel” distance. To optimize the layout for real-world use, the researchers relied on a collection of some 600,000 emails sent between Enron executives before the energy giant’s collapse. (The so-called “Enron Corpus,” made public during a subsequent federal investigation, is a treasure trove for computer scientists because it captures “email English” at its most casual and conversational.) After evaluating more than 5.6 million possible key configurations, the team landed on a single best design.
With KALQ, all vowels except “Y” are handled by the right thumb, while the left thumb is responsible for more letters overall; the result is that, in commonly used words, 62 percent of keystrokes alternate between sinistral (left) and dextral (right) digits. Instead of a spacebar, which requires the thumb to travel all the way to the bottom of the screen, two “blank” keys—one for each thumb—are found on the centrally located home-row. Frequently used letters are clustered where they can be easily reached; obscure letters are relegated to the margins.
In tests, non-native English speakers were asked to transcribe sentences as quickly as possible, first using a standard QWERTY keyboard, and then using KALQ. As was expected, KALQ performed poorly at the outset. After two weeks of practice, however, subjects who “trained” for an hour each day saw their performances improve dramatically. On QWERTY, mobile typists averaged 27.7 words per minute; on KALQ, they averaged 37.1 words per minute—a 34 percent increase. KALQ produced more accurate results, too, with an error rate of five percent, compared to QWERTY’s nine percent.
What’s surprising is not KALQ’s superior performance, but that we’ve put up with two-thumb QWERTY typing for as long as we have. QWERTY, of course, is a holdover from the Remington typewriters of the 1870s; today, it’s found on computers around the globe. As office folklore has it, the keyboard’s layout was designed to prevent the Remingtons from jamming by placing common letter pairs far away from each other, forcing the typist to slow down.
Alternatives do exist, such as Dvorak, a layout designed along the same principles as KALQ—maximize alternation, minimize finger movement—and patented in the 1930s. Dvorak has a small, hardcore fan-base, including the late Barbara Blackburn, long the Guinness Book of World Record’s “fastest typist.” (Blackburn discovered Dvorak after nearly failing a QWERTY touch-typing class in high school; at the height of her secretarial game, she could sustain speeds of 150 to 170 words per minute.) But apart from a cult of contrarians and computer science geeks (e.g. my brother), Dvorak has caught on about as well as Friendster.
KALQ, which will be made available, free, for Android devices in early May, is a step in the right direction for mobile culture. It won’t make texting on a phone as fast as, say, typing on a laptop, but it may ease the strain on users’ thumbs, and reduce the number of lewd AutoCorrect messages you accidentally send to mom. It’s a first-world-problem, sure, but that doesn’t mean it’s not worth solving.