In Shona, the melody of a song tends to parallel the melody of speech. An acoustic comparison of sung and spoken lyrics for three Shona songs finds that, while the parallelism is statistically significant, the locus of parallelism cannot be predicted. The realization of linguistic tone during singing is a question which has periodically puzzled both linguists and musicologists. This paper is a preliminary examination of vocal music from Shona that provides further evidence towards the solution of this puzzle.