The study was aimed to investigate the difference between prosody shadowing and contents shadowing, to find out whether the attention distribution between prosody and content influences subjects in their shadowing performance. An experiment was conducted to explore how working memory (WM) capacity and attention distribution influence shadowing performance of Japanese for advanced Japanese learners among Chinese native speaker. The shadowing performance was measured by two factors:understanding of the content, the accuracy and fluency of oral reproduction. Two main findings are presented as follow: (1) the results of the shadowing performance did not show significant difference between the subjects with a large WM capacity and those with a small one. (2) attention distribution was a slightly significant factor affecting understanding of the content. Subjects performed better in contents shadowing than in prosody shadowing. The results showed that when the shadowing materials were in a low difficulty, WM capacity did not affect shadowing performance. Further, there was a high possibility that in the contents shadowing task, regardless of WM capacity, oral reproduction of the inputted speech information would be processed with comparatively shallow semantic manner.