For anyone who is interested in what an ideal language would look like, particularly in respect to brevity vs. informativeness I'd highly suggest looking into Terry Regier's work: http://lclab.berkeley.edu/
I worked in his lab on one of many projects showing that most human languages use a near optimal trade-off in various semantic domains (so far - color, kinship, containers, and spatial relations). His work also includes some of the best evidence for some language dependent forces in cognition interacting with some universal ones.