Here's how I've come to rationalize it. It's slightly different than some of the other responses here, and I cannot with confidence tell you that those other responses are wrong, but perhaps my rationalization will make sense to you, as well. Sound waves require a medium in which to travel. When we speak, that medium is our breath. That is, speech is inextricably linked to airflow (i.e., how and when we breathe). Mechanically, stuttering can be reduced to an attempt to speak without that airflow. I cannot make any claims as to the neurological reasons why the link between breathing and speaking is different in stutterers and non-stutterers, but I'm not sure that there would be a way to correct/treat that, anyway. When stutterers sing, their breathing patterns change relative to their breathing patterns when they are speaking; they are therefore better able to do so fluently. Whether this is due to different neurons firing, the nature of singing, or both, I'm not sure (but I would guess both). You can imagine that there are all kinds of circumstances where your breathing patterns might differ slightly from the one you fall into when you try to speak the way that you've "learned" to speak. Unfortunately, it's incredibly difficult, at least in my experience, to incorporate those breathing patterns into normal speech in any lasting way.