The Sapir-Whorf hypothesis is both one of the most famous and most misrepresented theories about the nature of human thought. It claims (in what is technically the weak version) that how we speak about the world influences the way we think about the world. Cognitive scientists generally agree with this version of the hypothesis. Empirical evidence shows that there are, in fact, concepts that exist more or less equally across cultures, regardless of the words used to express them. Linguistic studies demonstrate that children, given a broken, partial language (called a pidgin), use their innate language abilities to develop it into a fully-formed language, grammatically mature, able to discuss all of the universal topics. This brand-new language is called a creole.
Language's influence on thought can depend heavily on the circumstances. There are arenas of our lives that are ruled by instinct, such as your tendency to jump at hearing a bang. In actions that lie close to the metal, so to speak, such as your impulses and instincts, language has a relatively small effect on your thoughts. But in the arena of symbols and interpretation, and especially when dealing with abstract ideas, the structure of your native language is quite complicit in shaping your thoughts on the subject. Oftentimes, with abstractions, words become our only way to picture it and relate to it in our minds. For this reason, in literature, when ideas are flying fast and furious, most of them being unfamiliar and strange to the reader, the details of the language that the author uses will have an immense influence on the reader's thoughts.
Sapir-Whorf applies to potential speech acts. It's what you would have said about an object (had someone asked you) that affects how you think about that object. With literature, it's much more direct: language is actually being set in motion.