Non-Recursive Human Languages?

Daniel raised the question last class about whether there is a non-recursive human language. Although I felt quite confident in class that such a thing could not exist, there have been some controversial claims about the existence of a non-recursive human language. The language in question is known as Pirahã, spoken by a remote tribe in the Brazillian Amazon. Dan Everett, now a professor of Linguistics at Illinois State University, studied the Pirahã language and spent several years living with the Pirahã (first as a missionary and later as a linguist).

Everett made several fascinating claims about Pirahã, but they are disputed by other linguists. In addition to being non-recursive, for example, the only counting words in Pirahã are one, two, and many. The most shocking claim, though, is the Pirahã does not use recursion. Many linguists, however, have disputed this claim (see Piraha Exceptionality: a Reassessment by Andrew Ira Nevins, David Pesetsky, and Cilene Rodrigues, March 2007).

The main technical question all this raises for us is what does it really mean for a language to be recursive? All real humans have a very limit stack depth in the actual language we use (it is very rare to have a stack depth of higher than 2 in a real utterance, and exceptionally rare for it to be higher than 3). If the stack depth is limited, then there is a non-recursive way to define the language: we can generate all possible English sentence structures with some bounded stack depth.

This New Yorker article gives an engaging account of Everett and the Pirahã.

Some scholars were taken aback by Everett’s depiction of the Pirahã as a people of seemingly unparalleled linguistic and cultural primitivism. “I have to wonder whether he’s some Borgesian fantasist, or some Margaret Mead being stitched up by the locals,” one reader wrote in an e-mail to the editors of a popular linguistics blog.

Towards the end of the article, it describes an experiment Fitch (a Chomskian who was a co-author of the paper I mentioned in class yesterday about recursion in human languages) does to see if the Pirahã can recognized the language anbn:

Fitch decided to test the girl on a higher level of the Chomsky hierarchy, a “phrase-structure grammar.” [this is similar to a context-free grammar] He had devised a program in which correct constructions consisted of any number of male syllables followed by an equal number of female syllables. Hauser, Chomsky, and Fitch, in their 2002 paper, had stated that a phrase-structure grammar, which makes greater demands on memory and pattern recognition, represents the minimum foundation necessary for human language.

The girl gazed at the screen and listened as the HAL-like computer voices flatly intoned the meaningless syllables. Fitch peered at the camera’s viewfinder screen, trying to discern whether the girl’s eye movements indicated that she understood the grammar. It was impossible to say.

If Fitch’s experiments were inconclusive on the subject of whether Chomsky’s universal grammar applied to the Pirahã, Jackson’s movie [King Kong] left no question about the universality of Hollywood film grammar.

Here’s a video of Daniel Everett on Recursion and human thought: why the Pirahã don’t have numbers:

Comments are closed.