Since this parent post is getting downvoted, maybe I should have elaborated: the claims in the title of this article ("proof of X") seem to be a lot stronger than the actual research paper it references ("suggests that X").

This subject is something that has been discussed for over a century (to be honest I'm not sure how much it's been considered seriously by linguists in recent years, but hey, I remember it being brought up back in LING201).

The title of the article just seems a bit extreme to me, as if the debate around linguistic relativity is over now that someone ran a counter over some bilingual dictionaries. It's an interesting approach, and maybe it can give some direction into where to look, but I think we'd need a lot more than numerical analysis on dictionaries to prove something about language, and we need to account for other causes of correlations.

Eg, bilingual dictionaries (which this research analyses) are likely to be compiled by people who are aware of these claims about their language. If you're creating a dictionary for a language that is known for having "X words for snow", you'll put more effort into listing many words for snow than many words for taste. Note that bilingual dictionaries often exist for language learning purposes, so they intentionally won't paint a complete picture of the language.