We study the design of split keyboards for fast text entry with two thumbs on mobile touchscreen devices. The layout of KALQ was determined through first studying how users should grip a device with two hands. We then assigned letters to keys computationally, using a model of two-thumb tapping. KALQ minimizes thumb travel distance and maximizes alternation between thumbs. An error-correction algorithm was added to help address linguistic and motor errors. Users reached a rate of 37 words per minute (with a 5% error rate) after a training program.

KALQ is a split keyboard for touchscreen devices. The position of the keyboard on the display and the assignment of letters to keyslots were informed by a series of studies conducted with the aim of maximizing typing performance. KALQ is used by gripping the device from its corners. Trained users achieved an entry rate of 37 wpm (5% error rate). This is an improvement of 34% over their baseline performance with a standard touch-QWERTY system. This rate is the highest ever reported for two-thumb typing on a touchscreen device. The following factors are exploited in the design of KALQ:

1. Grip: Grasping the tablet with its corner in the valley created by the thenar and hypothenar eminence yields ~4% faster tapping performance than does a random grip. Moreover, the associated keyboard layout occludes the display the least.
2. Hover-over technique: When typing with two thumbs, the idle thumb should not stay put but approach its next target and hover over it to minimize travel distance. We estimate that this typing strategy saves about 10–20% on MT in alternating taps. See an image here.
3. Optimization of letter assignment: KALQ was optimized computationally from a model of best-performance two-thumb typing. As a result, it maximizes alternating taps and minimizes same side travel distances. Our model predicted a benefit of 4% over a comparable quasi-QWERTY layout. However, this prediction was made assuming the same typing technique and grip. Without the technique and grip, the benefit will be larger.
4. Error correction: We developed an error-correction technique that adapts well-known techniques to the unique motor and linguistic aspects of two-thumb typing. Although our users’ error rates were not improved by the online version of our corrector, offline analyses showed that with better parameters, the error rate can be decreased by 1.3 percentage points.

The layout has the following properties:

1. The division of work is almost equal, at 54% and 46% for the right and left thumb, respectively.
2. Alternation is rapid: 62% of the taps are switches.
3. Travel distances are short: On average, the left thumb moves 86 px, the right 117.
4. The spacebars are centrally located.
5. The right thumb handles all vowels except y. The clustering of vowels around the spacebar favors quick switches and minimizes travel distance. The right thumb is responsible for 64% of same-side taps.
6. The left thumb has most of the consonants, exploiting its ability to hover above the next button sooner. It has most first letters of words and most of the consonants.

Download KALQ for Android devices

KALQ (official) in Google Play. KALQ replaces the in-built keyboard of an Android device. Please follow the Google Play page for updates.

Media Coverage

Additional Materials

Frequently Asked Questions

• Why should I use KALQ?
  Such a decision requires weighing of the involved costs and benefits, like most decisions in life. On the one hand, you will lose about 4-8 hours of your life training a new keyboard. On the other hand, you will be able to type significantly faster. For users who spend more time in typing, the improvement in typing speed may overcome the training overhead.
• How much faster can I be?
  This depends on your present level of skill and willingness to practice. Users in our study improved 34% and reached a rate of 37 words per minute. Note that this involves not only switching the layout. It involves changing your grip and learning a typing technique where the other ("idle") thumb is moving toward its next target and hovers over it while the other is typing. Note that you can also get faster with Qwerty if you train, but your maximum speed will remain impeded by the fact that finger travel distances are longer.
• Can I use KALQ on [my device]?
  Yes and No. A version will be released for Android smartphones. iPhone and Windows Phone are not presently supported. But if you know somebody willing to take the effort, let us know.
• Is it difficult to learn?
  It will 4-8 hours of practice to get to a level that allows everyday use. And this means real practice. On the other hand, learning to do touchtyping with 10 fingers on a regular keyboard also takes time (20-30 hours).
• I wonder why you did [a particular feature in design]?
  Please see the paper, it justifies all design choices.
• Is KALQ patented? Can I use/edit it for my application?
  KALQ is not patented, you can use it and edit the layout however you like.
• KALQ is for right-handed users. Can left-handed users use it?
  The keyboard layout should be mirrored then. We will add this feature to the Android version.
• How does KALQ work with hands of different sizes?
  The Android version will have a feature to scale the keyboard according to the span of the thumb.
• How does KALQ work with screens of different sizes?
  The keyboard design does not allow overlap in the split halves. Any screen size that allows non-overlapping design is accommodated. You can check this with a simple test: grab the device from its edges and sweep with your thumbs. If the thumbs meet, the screen is too small for KALQ. In our experience 7 inches and above are fine for most.
• Will you release a version for [my language]?
  Using our approach, one could optimize a layout for any given language. Our code can be used with MATLAB for this purpose. You should reserve about 1-2 weeks of computation time to find a good layout, though. We are working on a faster optimization method.

Links

Citation

  
@inproceedings{oulasvirta2013improving,
        title={Improving two-thumb text entry on touchscreen devices},
        author={Oulasvirta, Antti and Reichel, Anna and Li, Wenbin and Zhang, Yan and Bachynskyi,
                Myroslav and Vertanen, Keith and Kristensson, Per Ola},
        booktitle={Proceedings of the 2013 Annual Conference on Human factors in Computing Systems 
                   (CHI'13)},
        year={2013},
        organization={ACM Press}
}