On the object level, your belief-as-stated is not conclusively known. Everett sub 1986 believed that there were words for "one", "two" and "many"; this belief was updated in 2008 when one speaker in an n=4 study used the word for "one" when there were six things presented to them.

On the meta-level, none of Everett's results (as far as I know) have been replicated by an independent anthropologist, which means that your belief-as-stated has one point of failure. Given the surprising nature of his results, we should demand strong evidence that his results are true and not due to, e.g., cultural/linguistic misunderstandings. In fact, the linguistics community has indeed questioned the data closely.