ABSTRACT

There can be little doubt that the development of formal generative linguistic analysis has been the most important advance in the study of language in the last half century. These formal models predict the structure of linguistic output both accurately and compactly, including its open-ended generativity. The elegance and compactness of these formal theories are a function of good reverse-engineering. It would be odd if these theories exhibited poor predictive power, considering that they were consciously designed to yield results consistent with vast amounts of known (and self-generated) linguistic data. Thus it seemed reasonable to extrapolate from this model to the evolutionary and neurological origins of language. But after nearly a half-century of unsuccessful efforts to find the neurological and genetic counterparts to universal grammar or the algorithms comprising the principles and parameter-setting mechanisms of language, it may be time to reflect on why neurobiology and formal linguistics have not converged.