This thesis investigates, in an exploratory manner, whether there are independent effects of word integration and prediction in human sentence processing data. We train a probabilistic language model to compute two information ...