Linguist Greenberg proposed a genetic relationship between the Macro-Je, Macro-Panoan, and Cariban families. However, although initially greeted with excitement, Greenberg's classification for the Americas was not accepted by many linguists who worked with the languages in question once they had a look at his evidence.
Rodrigues (2000) finds evidence instead for including Tupian, but not Panoan.
Eduardo Ribeiro of the University of Chicago, who has worked with Macro-Je and Tupian languages, has found further that these families share irregular morphology with each other and with Cariban, but not with Panoan.
Shared grammatical irregularities are strong supporting evidence for putative language families, as they are unlikely to be borrowed or to be due to chance.