Hmmm... that would be nice start, but the resultswould indicate what the CS undergraduates themselves want (or think they want) in a programming language.

I think I'd more interested in how the general populace likes to control their computer: their user interface, or "programming language"; or whatever you'ld like to call it. We have some UI design concepts, but in general, they're poorly studied and user complaints that programmers "don't think like end users" are common. A good start would be trying to get a solid grasp on what people tend to really want.

Then we could try to design a language that most nicely bridges the gap between what the users actually want, and what the programmers can make the machine can do, and how those programmers like to program. No language I've seen has been designed from psychological principles with both the source creator and end user in mind; I'd be curious to see what the results could tell us, if anything...