In this paper, we present a LPC-based Analysis-Prosody Modification-Synthesis (APS) approach for auditory user feedback in the SPELL pronunciation teaching system. The basic idea of the approach is to demonstrate to students correct pronunciation by his own voice, given that the pronunciation difference would be clearer than simple play-back of the teacher's voice, due to the elimination of differences between the personalities of the two voices involved. The auditory user feedback by this approach is accomplished by three processing steps: analysis, prosody modification and synthesis. Listening tests revealed that this approach is very encouraging for teaching prosody patterns of a foreign language. It is independent of speaker and language, and very simple to implement. One important aspect of the approach is that it offers a potential tool for students to imitate his (her) own voice with a different and often unfamiliar prosody pattern. It might also be interesting to explore its efficiency in terms of pedagogical methodology.