I've spent quite a bit of time recently branching out into new programming languages, writing significant works in each one. And each time I do so, I come back to my day job and realize just how limiting it is to only work in one "blessed" language. There are things which are can be eloquently expressed in Rust or Haskell which do not translate nicely over to Python. Sometimes I can do in a single expression what it would take an entire module of confusing metaprogramming to achieve in Python!
Sadly, that street runs both ways. Some of these new languages want to be so safe that they hobble you down with the chains of explicit types and memory protection; making you write pages of boilerplate to do some of the simplest things.
It's true, when writing code, your world is limited by your language. The flip side of that is also true: by choosing a language, you define the world in which you live.
When writing code, your world is limited by your operating system too. Heck, we're even limited by the hardware, which hasn't changed conceptually since it was created in the 1940s. We're all under the spell of "instructions to the beat of a clock".
"The Sapir-Whorf hypothesis is the theory that an individual's thoughts and actions are determined by the language or languages that individual speaks. The strong version of the hypothesis states that all human thoughts and actions are bound by the restraints of language, and is generally less accepted than the weaker version, which says that language only somewhat shapes our thinking and behavior." 
And this can help for pedagogical purposes, too. Another example is infinitesimals in calculus, which were dismissed in favor of Weierstrass's epsilon-delta definitions over a hundred years ago, but found to be more understandable to students in more recent studies: http://academic.brcc.edu/johnson/Projects/Calculus%20with%20...
Chinese characters are more like the four components of the word "nonbiological" (non, bio, logos, al) than like the word "nonbiological" itself but, unlike English, there are more cases where the components can also stand alone.
Ten thousand is on the high side for the number of characters an educated, native Chinese speaker would usually know (much less use), but the number of what Westerners might consider "words" known by such a Chinese speaker would typically be much higher than 10,000, with most of them being composites of multiple characters.
As for interpretations of quantum mechanics that is a topic for 1970s popular books by drugged out authors and for the marginal fringe of physics journals.
However, one should also remember that language isn't just words. You can refer to colors by motion (pointing at a chart) or by proxy (using a code like #34F278). So we are limited by our language, but our language is not so limited as to make us fools for not having a word or two.
I thought that some of the work on the Himba tribe's perception of colour was, tentatively, showing just that. Their lack of linguistic distinction between green and blue could explain that, in experiments, they took longer to differentiate between shades of blue and green than an English speaker would. It doesn't mean they don't see it though, just that it's a more complex task for them.