Similarity Interference and Scrambling in Japanese

Lewis and Nakayama (1999) claim that the difficulty of comprehending sentences is a function, in part, of similarity-based interference, a limitation on working memory. By examining Japanese sentences, they found that the syntactic category, the syntactic position, and the consecutive occurrence of the same category all contribute to the difficulty of sentences. The present study examines the similarity interference hypothesis in Japanese scrambling sentences. According to the hypothesis, the scrambled sentences that reduce similarity interference would be considered easier than the unscrambled sentences. This prediction was borne out. However, the experimental results also showed that sentences with scrambled embedded objects were more difficult than unscrambled sentences. We argue that there are independently-motivated explanations for the difficulty of these sentences. Therefore, the results still support the similarity interference hypothesis, and indeed similarity interference plays a significant role in the difficulty of comprehending sentences.