Word Level Effective Text Entry On Mobile Devices

Despite the rapid growth of mobile phones, text entry on small devices remains a major challenge. Due to trade-offs in both size and compatibility, most mobile phones today are equipped with a 12-key keypad. This keypad is effective for dialing phone numbers but immediately becomes a performance bottleneck for editing contact lists, composing SMS messages or writing emails. Other input devices, such as mini QWERTY keyboards and touch screens are on the rise, but the 12-key keypad-based mobile phones are still found in the majority of handsets.

To enter the 26 alphabetic characters on the 12-key keypad, most mobile phones nowadays relay either on multiple key presses (MultiTap) or a built-in dictionary (Dictionary Based Disambiguation, i.e. T9). DBD input techniques such as T9 can efficiently enter words that are in the dictionary. However they suffer two major problems - 1. the resolution of complete ambiguity or collision (two or more words sharing the same numeric key sequence, such as 'book' and 'cool', 'of' and 'me') and 2. entering out-of-vocabulary (OOV) words.

we present a novel method called SHRIMP (Small Handheld Rapid Input with Motion and Prediction) which solves both the collision and OOV problems of DBD input. SHRIMP is a predictive text input method based on Vision TiltText and runs on camera phones equipped with a standard 12-button keypad. SHRIMP is as effective as conventional DBD when entering unambiguous, words within the dictionary. SHRIMP uses seamlessly integrated concurrent motion gestures to handle ambiguous dictionary words or OOV words without mode switching to achieve 1 KSPC. In the user study, we found the text entry speed of SHRIMP (12.1 wpm) is significantly faster than that of MultiTap (7.64 wpm) with less than 10 minutes of practicing.


  • SHRIMP received an Honorable Mention, CHI Best Paper Award at the upcoming CHI 2010! 03/01/2010
  • SHRIMP (Small Handheld Rapid Input with Motion and Prediction) has been accepted as a full paper by the upcoming CHI 2010! 12/18/2009



  • Jingtao Wang, Shumin Zhai, John Canny, SHRIMP - Solving Collision and Out of Vocabulary Problems in Mobile Predictive Input with Motion Gesture, In ACM CHI 2010, Atlanta, Georgia, April 10 -15, 2010. (TO APPEAR) Honorable Mention, CHI Best Paper Award

Contact Information