Paper Reading #1: Personalized Input: Improving Ten-Finger Touchscreen Typing through Automatic Adaptation

Introduction
Personalized Input: Improving Ten-Finger Touchscreen Typing through Automatic Adaptation is a paper coauthored by Leah Findlater of the College of Information Studies at University of Maryland and The Information School DUB Group of University of Washington and Jacob O. Wobbrock, also from The Information School DUB Group at University of Washington. Findlater is a CHI professor at the University of Maryland specializing in personalized adaptation. Wobbrock is an associate professor at the Information School focusing on new user interface technologies and input interaction techniques.

Summary

Image taken from the paper demonstrating the effects of the adaptive software

Findlater and Wobbrock used a variety of classification procedures to formulate an adaptive touch-screen keyboard designed to increase typing speed using a touch model (unlike the language-based model used by Apple iOS). Based on observations from previous experiments and a pilot experiment conducted prior to the paper they developed two keyboards using the same underlying structure, the only difference being the visual representation of the keyboard. One keyboard updates the visual to show the changes made to the underlying input interpretation and one does not.

They concluded that the NonVisual-Adaptive keyboard out-performed both the Visual-Adaptive keyboard and the Conventional keyboard. They infer, based on prior experiments, that this is due to the increased cognitive load that the Visual-Adaptive keyboard places on the user when the keyboard visual changes. This change is believed to distract the user, negating the benefits of the underlying adaptation.

Related Works
Un-cited related works include:

1. From Plastic to Pixels: In Search of Touch-Typing Touchscreen Keyboards by Leah Findlater & Jacob O. Wobbrock
2. Manual Text Input: Experiments, Models, and Systems by Poika Isokoski
3. Adaptive personalized interfaces—A question of viability by Saul Greenberga & Ian H. Wittena
4. On-line Personalization of a Touch Screen based Keyboard by Johan Himberg, Jonna Häkkilä, Petri Kangas & Jani Mäntyjärvi
5. Achieving Accessibility through Personalization by Lynne Coventry, Graham Johnson & Antonella De Angeli 
6. Typing on Flat Glass: Examining Ten-Finger Expert Typing Patterns on Touch Surfaces by Leah Findlater, Jacob O. Wobbrock & Daniel Wigdor
7. CATKey: Customizable and Adaptable Touchscreen Keyboard with Bubble Cursor-Like Visual Feedback by Kentaro Go & Yuki Endo
8. HybridPointing: Fluid Switching between Absolute and Relative Pointing with a Direct Input Device by Clifton Forlines, Daniel Vogel & Ravin Balakrishnan
9. High Precision Touchscreens: Design Strategies and Comparisons with a Mouse by Andrew Sears & Ben Shneiderman
10. User Preference and Performance with Three Different Input Devices: Keyboard, Mouse, or Touchscreen by Chi-Hui Lin & Kathy J. Schmidt

These papers cover various subjects covered in Personalized Input, from touchscreen input to various applications of similar adaptive input methods. I found From Plastic to Pixels, Achieving Accessibility through Personalization, and HybridPointing especially intriguing.

Evaluation

Findlater and Wobbrock evaluated the successes and failures of their keyboards and underlying model using a variety of measures. Testing was conducted using 12 participants over the course of 3 90-minute sessions. Quantitative measurements centered around words-per-minute typed and error rates. The NonVisual-Adaptive keyboard consistently improved typing speeds by at least 12.9%. The Visual-Adaptive keyboard showed no benefits compared to the Conventional keyboard and error rates were unaffected. In addition to these measurements, subjective measures were also taken concerning the participants' efficiency/ease/preference, comfort/naturalness, and least frustration. The NonVisual-Adaptive keyboard was consistently rated highest in efficiency/ease/preference, while the Visual-Adaptive keyboard earned the most votes for comfort/naturalness. The measure of least frustration was deemed inconclusive. Additionally, the Visual-Adaptive keyboard was negatively considered high in obtrusiveness.

Discussion

Overall I found the paper to be quite interesting. There is a great demand for touch screens that users can reliably touch-type on. Their methodology and reasoning was well documented, as were their conclusions. My only significant complaint would be their decision to rely on experiments and experiences outside the direct scope of the paper to formulate certain parameters without experimenting with variations in these parameters. However, they do mention this as a confounding effect in closing, so I don't feel that it jeopardizes the validity of their study. I would be very interested in seeing any follow-up research conducted on this subject and definitely plan on reading more of Findlater & Wobbrock's research.