I've seen Processing used in some intro courses for that reason, because the time-to-interesting-experimentation is fairly low (similar to the goal of Logo in the '80s, which was my own first language, although I think Logo was much better as a programming language). I don't think it's a common first language in intro-CS programs, but there are a number of curricula being built around Processing outside the context of CS majors.

Processing is a good candidate because it allows a novice to make something interactive and interesting in very few lines of code. The downside is that they wind up typing things about which they understand nothing...e.g void setup {···}. And it's not like void will really make sense after two weeks. On the other hand, placing the mastery in the executing artifact, not the underlying code, might be a good thing.

I think pyprocessing (https://code.google.com/p/pyprocessing/) makes an excellent middle ground here. It implements much of the Processing API in Python, without making an effort to hide the underlying principles of the language (like Processing seems to do for better and for worse).