Every sound is made up of pitch, intensity and length (P, I and L). These universal parameters work together to give a sound its sensation. This paper presents a case of using P, I and L, and a hypothetical measure, "S[subscript c]" ("Stress Composite"), to appraise the effect of prosodic training. The main question this paper explores is whether or not a training activity can cause P, I and L to vary in certain ways. The research is set in a "singing classroom", which is an ideal context for learners to exercise their P, I and L. The research instruments included a perception test and a production test. Two major findings were yielded: First, while the learners judged pitch variation to be important, they relied on length variation when encoding prosody. Second, singing did not alter the fact that length variation was a dominant encoder, and pitch only came second to length. These findings can lead to several interpretations. They may indicate that singing could affect prosody in other ways, but not how P, I and L are varied in the voice; or, indeed, they may point to a "normal" way of encoding speech. The current method of analysis has implications for prosodic assessment. The mismatched results between the learners' perception and production of P, I and L will be explained, and the potential use of the S[subscript c] measure is discussed.