Hypothesis Generation in a Computational Model for Visual Word Recognition