Predicting human similarity judgments with distributional models: The value of word associations.