Explaining human performance in psycholinguistic tasks with models of semantic similarity based on prediction and counting: A review and empirical validation | Litlas