This may be going off topic, but Diana Deutsch, of auditory illusions fame, found that if a spoken phrase is played over again, multiple times, listeners begin to hear it as sung rather than spoken. A quick google search will reveal some sites with examples that you can try for yourself.
In fact Diana Deutsch has pursued this further:
If people listen to a phrase once, they think it's spoken
If they listen to it ten times, they think it's sung
If sopranos listen to it ten times and repeat it, they sing it back
If sopranos listen to it once, and repeat it, they say it back
If sopranos listen to the sung version only once, that the other sopranos produced after hearing the (spoken) version ten times, they are able to reproduce the version as sung, exactly.
Deutsch concludes that (1) there is actually not necessarily any physical difference between speech and song - the same sound-waves can be interpreted both ways depending only on the state of the observer; (2) if the brain has separate bits looking at speech and song, it's striking how the two can analyse the same input and come to wildly different conclusions; and (3) it's amazing how plastic we are, and how quickly we can reorganise the way we process something, so that we interpret it quite differently.
So, tongue-slightly-in-cheek, it might be that when classifying prose and song, the only difference is how many times you've heard it...