Testing the Predictions of Surprisal Theory in 11 Languages