Abstract
In philosophical circles, but not only there, Chomsky's views on natural language regularly fall a prey to misrepresentation. Very often the confusion involves the creative aspect of language use, an aspect of linguistic performance, which tends to be confounded with the notion recursivity, a property of the grammatical competence system. The present article clears away the most deep-seated confusions and proves that criticism of generative grammar based upon them cannot be upheld. In particular, it shows that the existence of metaphors, and deviations from rules more generally, reinforces rather than refutes Chomsky's theory so long as the crucial distinction between knowledge of language (competence) and language use (performance) is taken heed of. It is argued that theories which deny this distinction and view human language as a set of dispositions or as an ability are misguided. The final section illustrates some properties of Universal Grammar (UG) by means of data from the domain of negative polarity and subject-object asymmetries. Though these phenomena are far from being fully understood, the present level of our insight warrants the conclusion that they lend considerable support to the Chomskyan conception of UG as a highly complex and richly articulated linguistic genotype