Chomsky claims of late that recursion is the one formal property that singles out human language from sophisticated social signaling more generally.* I think he may have a point. It is not that recursion is that great per se. As we are well aware, we don’t make unlimited use of it – nothing that requires such an awesomely creative tool. With center embedding, we get lost after about the third recursive element.
But recursion is a sign that we are free of the restrictions of the meat. Animal “syntaxes” are, in fact, more akin to prosodic structures, with parts made out of the coordinative affordances of the vocal tract. They do not have recursion, as it would make no sense to have a unit like the syllable inside another syllable, or a foot inside a foot. These units are performative, and deeply engrained in the meat.
The recent work of Stan Dehaene, Liz Spelke and Susan Carey suggests something similar. There, language is seen as a fluid vehicle for passing meaning around between domains of organization that have long phylogenetic histories, and that are, thus, special-purpose. The emerging notion I am developing of the P-domain may help here.
*This has given rise to some discomfort with the claim that has recently arisen that the Amazonian Piraha tribe speak a non-recursive language – the jury is still out on that one.