In this paper we present a new model for predicting text entry speed on a 12-button mobile phone keypad. The proposed model can predict the performance of novice users. Like other models for text entry, the proposed model includes a movement component based on Fitts' law and a linguistic component based on letter digraph probabilities. It also adds cognitive delay times before key presses and takes into account the fact that Fitts' law cannot model multiple presses of the same key accurately. Finally, we compare the prediction of our model to previously published experimental results, demonstrate that it fits observed results for novices very well, and list some observations about learning.