The issue of quantitative association rules in large databases is discussed in this paper. In order to soften partition boundary of the domain, the relational fuzzy c-means algorithm is adopted to determine two parameters of normal fuzzy numbers, then the normal fuzzy number model is adopted to partition the domain of the quantitative attributes and a series of linguistic value association rules are generated. The mining method of the linguistic value association rules is also provided. Because the abstract concepts can be well expressed with the linguistic values, the mined association rules are more abstract and easy to understand.