My first substantial learning experience in Perl was in the Perl 4 days (even before web pages existed!); I was handed a few thousand lines of code arranged in several source files that worked together in a pipeline (output of script 1 goes into script 2 then script 3, etc, each of which did a specific text filtering/conditioning task). I was supposed to make it work on a different qulaity of input data.

I spent a lot of time looking up descriptions of core functions via perldoc perlfunc. It took me a while before I discovered the very handy "perldoc -f function_name" approach, but that was okay, because when I scrolled through all those other functions on my way to read about "split" or "substr" or whatever, I learned about a lot of useful things.

I used a lot of the other man pages as well (perlsyn, perlvar, perlop, anything else that looked interesting or relevant). Throughout it all, the thing that helped most in my case was Perl's adoption of -- and useful extensions to -- things I was already familiar with from using C, unix shell, and regular expressions (as in unix tools like 'sed' and 'grep').

But it was also very helpful to be starting with code that already did useful stuff, and having some idea of additional or different stuff that I wanted to make it do. Breaking code, and using the perl debugger to figure out how it broke, can be very instructive.