I think sound design is one of the few domains that really really benefits from visual programming languages. You really can't work effectively without 1. instant feedback and 2. the ability to quickly alter the signal path. I know there are live sound textual programming environments, but anything more complicated than changing individual values is going to break the creative flow if you have to create new files, modify imports, etc. Doing this by dragging around nodes and edges in a graph is intuitive enough to allow you to stay focused on the act of creation instead of the implementation details. This, and the fact that DSP is fundamentally a data flow problem, make visual programming environments the ideal UI for advanced sound design.
I feel that text is actually more effective, even though I agree there's much to commend about visual dataflow languages for music. But all the dragging around boxes and connections can become irksome, not to mention the energy spent aligning stuff and trying to prevent the (inevitable) dish of spaghetti that you get above a small level of complexity. Other general advantages of text: faster editing (leverage the features of the IDE, keyboard vs mouse, etc.), easier to organize and comment code, generally easier to debug, version control, and some languages are much more flexible and powerful than anything graphical. See Supercollider, and its clients using various general purpose languages, csound, Faust, and many others.
Max and PD are great though, with a nicer learning curve, fantastic projects done with them and a thriving community of users. Each to his/her own...
