So much is looking at things in a different way, and a quick way to do that is by picking something across the metaphorical language town, rather than the language next door.
The nature of how those languages model execution, their basic paradigms, data structures and design patterns all fall within a fairly narrow scope. Stepping outside those trappings lets you see them with a bit more clarity because you're forced to stretch your brain to accommodate something altogether new and get fresh perspective on the problems the languages are solving and how they solve them.
You won't necessarily gain any perspective if you're writing the same exact kind of code in languages X and Y, and you can gain perspective by writing an entirely different kind of code in a language you're used to. It just tends to follow that when you learn a new language, you'll learn to do things in the style of that language.
Knew Java, C++, Matlab and Perl before I learned Python.
A couple years ago, I learned R... I still use that a few times a week. Don't think it's improved my python programming.
More recently learned Clojure and Javascript. I think those count as meaningfully different from Python. I program differently in those languages, but I'm skeptical they've made me a better python progammer.
Maybe it hasn't, the whole idea is basically acquiring tacit knowledge and understanding through osmosis. That's why its hard to quantify how it should work.
I would suggest http://learnyouahaskell.com/
I'm a better iOS developer today, because I learned how awesome blocks can be from Ruby. Same goes for meta-programming.
Languages usually have features that span everything you do in that language. Getting comfortable with blocks in a language that uses them as a primary/native data type shows me how useful they can be in a language that uses them more cumbersomely.
That's how I've benefitted from learning multiple languages. As well as the fact that I will avoid Java/.NET because they and I don't agree! :D
And suggesting a 1000 hour investment is a lot different than suggesting a 100 hour investment.
But I've never noticed this in my experience. Moreover, whenever I come back to a language after any time away from it, I'm amazed how much time I spend remembering (or looking up) superficial differences.
Stuff like looking up whether I get the length of a vector with "len" or "length" (for a language I already knew) just feels like a distraction.