When you have one data source and only a handful of data points and that data is not reproducible, you're not doing science. Everett might have some interesting hypotheses, but what he's doing isn't what you could call peer-reviewable. Linguists who have aspirations towards hard science will, for that reason alone (not to mention the theoretically tenuous/unpopular nature of his claims) dismiss him pretty quickly.
While it is far more likely that Everett is wrong, science has to start somewhere, and Pirahã people exist on this planet and other people could go check the story. You can absolutely prove him wrong.
Coming up with crazy ideas based on some strange observations is a part of science too.
And questioning that, pointing out flaws and questioning methodology, is all science, too.
It's just convenient that Everett had the only 'in' with the community and so is the de facto expert. Once the data source is available, we'll know more.
