摘要：In recent years, it has been reported that WM (working memory) is concerned with word generation, but many points regarding the relationship between the individual differences of WM capacity and the patterns of word generation remain unclear. This study is to investigate these unclear points by using three types of word fluency task with different strategies. Subjects were 28 healthy native speakers of Japanese with an average age of 22.46 - 3.26 years old. The experimental design included the following two tasks: (1) Japanese reading span test: Subjects were divided into the high-span, middle-span, and low-span WM capacity groups on the basis of the proportion of words recalled correctly; and (2) Word fluency task (category, letter, and verb conditions): In this task, subjects gave as many words as possible that were associated with each stimulus word within 60 s; four stimulus words in each category were used with 12 words in all. As a result, it has become clear that the high-span group with larger WM capacity can generate more words in number than the other groups and is superior in word generating capacity to them and that the low-span group with smaller WM capacity is the highest in the proportion of words generated in the first 15 s when their vocabulary is abundant, while it is the lowest during 45-60 s when their vocabulary dries up. The results suggest that WM capacity is involved in the ability to extract appropriate vocabulary from long-term memory information in immediate response.