I've never understood why the common line is that any good hacker can pick up a new language and be just as effective as someone who has used it for year in a short time.

I've been thrown into overdue projects in languages I hadn't used before, and have been able to start fixing bugs almost immediately... but "tweaking existing code" is a far cry from implementing new functionality, which in turn is far below making architectural decisions.

Languages exist in ecosystems that develop over years; knowing those ecosystems in depth takes a lot of time.