Chomsky's claim is that humans, unlike other animals, have the capacity to use recursion not that all humans use that structure. So the Piraha objection is irrelevant unless you can demonstrate the Pirahas cannot understand recursive language.
My understanding is that Everett claims that Chomsky claims that not only do humans have the capacity to use recursion but that it is an essential feature of all our languages and for us to have languages (whether this is an accurate representation of Chomsky's views, I do not know; Pinker has stated that Chomsky is hard to pin down.) Everett further states that only Piraha children who have learned a recursive language as their first language are able to speak it recursively; adults, it seems, may learn another vocabulary, but will not use it recursively. Therefore, according to Everett, we have a human population (those who have not learned another language in childhood) with a functioning language, who display no capacity for recursion, even when exposed to it.
This seems to me to leave a sliver of hope for what is alleged to be Chomsky's theory, but I would prefer to see more data than more argument over the current data.
It's not clear what Everett is claiming in his rebuttal. He says Pirahas have no genetic impediment from learning recursive language (Portugese); which supports rather than undermines Chomsky's hypothesis.
Then he points out that Pirahas that learn Piraha first before learning Portugese don't use recursive Portugese. What is that observation supposed to suggest when taken together with the previous observation? That you lose the ability to use recursive language when you first learn a linear one? Or simply that you tend to transliterate when you learn a second language?
