Recursive syntactic pattern learning by songbirds
Abstract:
Linguistic theory has long asserted that language minimally requires “context-free” syntactic structures with a formal complexity beyond the “finite-state” patterns said to specify all non-human communication systems. Recognition of the recursive syntactic structures produced by context-free grammars implies specific computational abilities that have been hypothesized as unique to human language. I will show that European starlings, a species of songbird, can learn to recognize acoustic patterns defined either by a context-free or by a finite-state pattern. Starlings successfully classify novel patterns defined by these grammars, and reliably exclude agrammatical patterns - ruling out the use of a finite-state solution strategy. These results motivate revision of uniqueness claims regarding language, and establish an animal model for the biological mechanims of complex syntactic processing.Pressmeddelandet från University of Chicago sammanfattar det hela rätt bra: ‘Uniquely human’ component of language found in gregarious birds.
En intressant kritik av artikeln skriven av en lingvist-student finns här "A little bird told me".
Update: Erik Stattin skriver lite om fåglarna och språket på mymarkup.net.


