Linguistically Optimized Text Entry on a Mobile Phone * submitted to CHI 2001

We present an analysis of linguistically optimized text entry on mobile phones. This analysis compares the behavior of a linguistically optimized system with wordbased dis ambiguation methods. Through theoretical analysis, it is shown that in real-world situations in which typing errors are common and dictionaries are incomplete, the speed of text -entry using a word-guessing method degrades to the speed of a multi-press method. Linguistically optimized text entry speed, however, decays gracefully in the face of significant noise, and remains close to those of unambiguous text -entry methods.