What the research actually found
Some of the most striking evidence comes from how communities talk about space. Speakers of Kuuk Thaayorre, an Aboriginal language of Pormpuraaw in northern Australia, do not use body-relative terms like 'left' and 'right' for everyday space; they use fixed compass directions. In work by Lera Boroditsky and Alice Gaby, these speakers stayed oriented with remarkable accuracy, and they laid out time along the same east-to-west axis, rearranging it as they physically turned to face a new direction. Language habits tracked a genuinely different way of organizing the world.
Color offers a cleaner laboratory test. Russian marks an obligatory split between lighter blue ('goluboy') and darker blue ('siniy'). In a 2007 study led by Jonathan Winawer, Russian speakers were reliably faster to tell two blues apart when the pair straddled that boundary than when both fell inside one category. English speakers, who have a single word for blue, showed no such advantage. Tellingly, the effect vanished when speakers did a distracting verbal task, suggesting language was quietly assisting perception in real time.
The strong version is wrong
The old 'strong' Sapir-Whorf idea, that language determines thought and traps you inside its categories, does not survive contact with evidence. People routinely think about things their language handles awkwardly, learn new distinctions, and see colors their language does not name. The decades-long color-naming debate settled on a middle ground: perception has universal constraints, and language modulates rather than dictates it.
Caution is warranted about individual findings, too. A famous claim held that grammatical gender colors how people picture objects, so that German speakers describe a 'bridge' (feminine in German) as beautiful and slender while Spanish speakers (masculine) call it strong. Later high-powered and preregistered replications, including a 2022 registered replication of Phillips and Boroditsky, found the effect weak, language-dependent, or absent. It is a good reminder that relativity effects can be fragile and need careful testing.
The honest modern picture
What remains is a well-supported 'weak' linguistic relativity: language is one influence on thought among many, shaping which distinctions come easily and fast rather than what is thinkable at all. Because some of these effects can be switched off by a verbal distraction, they look less like a permanent cage and more like a habitual tool the mind reaches for when words are available.
This is why bilingual people are so interesting: they can carry more than one set of habits and shift between them. The takeaway is neither the romantic claim that language builds separate worlds nor the dismissive claim that it changes nothing. Your language gently tilts the playing field of attention and memory, and over a lifetime small tilts add up, without ever fencing thought in.