Abstract: A system, method and computer-readable medium that previews words and commands recognized by a continuous stroke recognizer based on input into a virtual keyboard layout is presented. The system and method allows a user to see the word that will be recognized or output before the user commits the continuous stroke input to the system. Further, after recognition and output, the system presents the recognized output word and the closest word candidates in a correction display integrated with the virtual keyboard layout. Among other things, the user is provided with options to change the current output word by selecting another word candidate on the correction display, and to delete the current output word by selecting the representation of the current output word on the correction display.