Text Entry

Work so far

My text entry work has three sides. Firstly, I have evaluated text entry methods experimentally. Secondly, I have modeled user performance with them. Thirdly, to do the experiments, I have constructed systems that allow me to quickly build text entry method implementations. Below is a short summary of my work. For details see the list of publications.


I have run experiments to collect data on user performance on text entry methods including: A minimal device-independent text entry method, Quikwriting (with stylus, gamepad, and keyboard), Desktop QWERTY keyboard, Handwriting, QWERTY sof-keyboard, Menu-augmented QWERTY soft-keyboard, numeric entry on a touchpad, text entry on wristwatches.


I have studied aspects of text entry performance modeling including: Character drawing time with a stylus, Learning rate, Error in learning rate predictions, maximal expert text entry rate on keyboards.


I developed a prototyping architecture for text entry techniques. The latest version is already several years old. It is a linux-based modular system written in C++ to a Java-based architecture where text entry method implementations, can be loaded and unloaded. The purpose of the system was to enable "compile once - run everywhere" -like behavior for the text entry modules. Having something like this makes the effort of implementing multi-platform text entry methods much smaller. So far, however, I am the only one to enjoy this benefit. Implementations of the server-side code that enables the platform-independent modules to run exists for Windows and Linux only. It is also a pain to set up and in no way ready for end user installation.

Plans for the future

On early smartphones the markets for text entry enterpreneurs were not open. My point of view during my doctoral thesis work wsas that a multi-platform server that could run the same text entry method implementations on desktop computers and mobile devices would change this. iOS and Android now both have interfaces for text entry techniques that can be installed by the user. This is excellent and improved the opportunities for developers. However, the skill transfer problems still exist between platforms and device generations. Smartphones are replaced so quikly that nobody seriously considers learning a new text entry system if there is a risk that it cannot be used on the next device. There are no signs of world-wide consensus fow how to solve this. Thus, my dream for unversal compatibility that would make it worthwhile for users to invest in learning better text entry techniques remains unfulfilled. At some point in the long run the smartphone is perhaps no longer the dominant form of computing. Perhaps the next revolution will offer new opportunities for setting up text entry user interfaces so that they make better sense for the user.

Running text entry experiments is fun. You know so much more about the performance of the system after running a longitudinal study. I plan to do this whenever I have the resources and a promising system turns up. I think that it is just great to have some impartial data to consider along with talk of the inventors and marketing people.

So if you have a text entry idea, feel free to contact me. I enjoy learning about new systems and sometimes I may even have a useful suggestion or two for further development.