Gold's Paradox

Gold (1967) proved that text learning was a technique that was not powerful enough to teach a system a complex grammar, such as a transformational grammar.  However, human children learn complex natural languages via text learning (Pinker, 1979).  The fact that this learning is possible, though Gold proved that it was not, is called Gold’s paradox.  Gold’s paradox is important to cognitive science because it motivated new studies of human language learning, in an attempt to find a way around the paradox.  Researchers like Wexler and Culicover (1980) were able to propose new constraints on the nature of language, such as the universal base hypothesis, that enabled them to prove that with these constraints one could learn a transformational grammar via text learning.

References:

1. Gold, E. M. (1967). Language identification in the limit. Information and Control, 10, 447-474.
2. Pinker, S. (1979). Formal models of language learning. Cognition, 7, 217-283.
3. Wexler, K., & Culicover, P. W. (1980). Formal principles of language acquisition. Cambridge, MA: MIT Press

(Added October, 2009)