Methods and electronic devices are provided for processing prediction candidates in connection with virtual keyboards and touch-sensitive displays. A display area of the touch-sensitive displays a virtual keyboard. Within the virtual keyboard is a plurality of keys, each of at least some of the keys corresponding to one or more different characters of a plurality of characters. The methods and devices may detect a selection of one or more of the keys. The methods and devices may generate one or more sets of characters based on the one or more characters corresponding to the selected key. The methods and devices may display the sets of characters within or substantially within the virtual keyboard. The methods and devices may detect a gesture associated with one of the displayed sets of characters, and, in response to detecting the gesture, remove the one of the displayed sets of characters from the virtual keyboard.