Two reasons with substance. (You can dismiss the reasons of people who have nothing more intelligent to say than "Wow, regular expressions look like line noise!" You'll save a lot of time flipping the bozo bit on them right away.)

First, Perl is relatively old—not C old, but well established. It's had two great surges of popularity, first with system administrators in the Perl 1 - 4 days, and then with web programmers in the early to late '90s. Certainly Perl has moved on from those niches, but a language with a killer application has trouble shaking off that label.

Second, Perl 5 is long in the tooth and has had a lot of trouble demonstrating real language evolution. Part of that is by design, where CPAN and language extensibility didn't happen by accident, but part of that is because it's really difficult to change Perl 5 to remove warts or to add important features or to make the implementation smaller, faster, simpler, easier to port, or more maintainable.

The lack of any credible Perl 6 implementation after eleven and a half years doesn't help. ("Oh, but three people are using it for side projects, and all you have to do is check out a branch from source control for each of several projects in the stack and probably most of the test suite will run" doesn't match my criteria for "Could I do something useful I care about with this software?" and especially "How much busywork will I have to do to keep this running for the next six months?")