guessing the upcoming icons

[vidma]

• because of lack of space - for tablets only, or can we compact everything to gain additional row?
• the guesses could be based on some bi-gram word probability (dependent on the ones already inputed),e.g:
  • I -> want
  • I want -> to eat | to sleep