There are definitely what I call “proof-of-concept” (POC, a.k.a. PITA) languages, which are designed by academics for academics (or for military contracts). They trust the computer so much, and the programmer so little, that an actual programmer can’t actually do a damn thing with it. :-) And then there are the handful “war-horse” languages such as Perl (and, in its own world, COBOL) that can be relied upon to ride into battle with you and to bring you out both victorious and alive.
As I have related before, I have also watched a very famous company that should have known better dive into a Ruby project to replace a Perl project ... only to discover, not only that they were “merely re-implementing something that already worked,” but that the replacement didn’t work because the actual implemented state of the Ruby libraries they were relying on were not as thoroughly implemented as those they were conditioned to rely-upon in Perl. The entire project turned into a death march. A pragmatic language implementation, I think, lives-or-dies more based on its available library base than by the properties of the language itself.