This thesis is to explore the speech recognition difference between Chinese and Japanese characters. In this behavioural experiment, twenty Chinese and twenty Japanese nation speakers are selected to identify vowels and three affecting factors (word frequency, cross language or not, target vocabulary) are designed of which each has two variables. After data analysis and discussion, two conclusions are drawn: 1. L1 and L2 are not only mutually independent but also mutually affect in speech representation and the impact that L1 has on unskilful L2 is greater than that unskilful L2 has on L1; 2. high-frequency words contributes to phonetic recognition.