Reexamining semantic clustering: Insight from memory models

Abstract

It has been repeatedly argued that semantically related words should not be learned together because learning is impeded. However, the results of past studies are not uniform, some providing favorable results for semantic clustering, and some seem to suggest different types of similarity affect memory in different ways. The types of similarity that truly causethe problem therefore need to be examined more carefully. Focusing on visual features, which are commonly observed across different models of working memory, a study was conducted to examine if learners have difficulty memorizing a group of words that describe items with common physical features. The study compared the learning of three types of word sets: unrelated, semantically related, and physically related. While no statistically significant difference was observed between semanticallyrelated and unrelated sets, the scores for physically related sets were significantly lower than those for the other two types. This suggests the possibility that the impeding effect of semantic clustering reported in the past could be partly due to the precise nature of semantically similar words, which sometimes share visual features.