…I find it a bit surprising no one has ever analyzed perl.

What would you do with the knowledge if you had a perfectly detailed index/concordance before you? I have a difficult time imagining any use for it and therefore would only be surprised if someone had sunk their time into developing it. I’d like to be surprised though.

I could do it on CPAN, but we all know that is not representative code…

I have only once in 15 years written any Perl professionally that is publicly available—happens to be on the CPAN—and this is standard. So there is no valid corpus of representative code for anyone to examine as huge tracts of the data you find relevant are private.