The most challenging problem in the design of fuzzy rule-based classification systems is the construction of a fuzzy rule base for the target problem. Much research has focused on generating and adjusting antecedent fuzzy sets. In many cases, initial fuzzy sets, each of which has a linguistic meaning, are predefined by domain experts and are thus required to be maintained in order to ensure interpretability of any subsequent inference results. However, learning fuzzy rules using fixed fuzzy quantity space without any quantification will restrict the accuracy of the resulting rules. Fortunately, adjusting the weight of a fuzzy if-then rule may help improve classification accuracy without degrading the interpretability. There have been different proposals for fuzzy rule weight tuning through the use of various heuristics with limited success. This paper proposes an alternative approach using Particle Swarm Optimisation in the search of a set of optimal rule weights, which can entail high classification accuracy. The proposed method is initially tested on the iris data set with regard to different predefined fuzzy partitions of linguistic variables to assess its performance. Experimental results demonstrate that the proposed approach is not sensitive to the predefined fuzzy partitions, and can boost classification performance especially when a coarse fuzzy partition is given.