A further difference between generativism and Bloomfieldian and post-Bloomfieldian structuralism - though in this respect generativism is closer to Saussurean structuralism - relates to the distinction that Chomsky draws between competence and performance. A speaker's linguistic competence is that part of his knowledge - his knowledge of the language-system as such - by virtue of which he is able to produce the indefinitely large set of sentences that constitutes his language (in Chomsky's definition of a language as a set of sentences. Performance, on the other hand, is language-behaviour; and this is said to be determined, not only by the speaker's linguistic competence, but also by a variety of non-linguistic factors including, on the one hand, social conventions, beliefs about the world, the speaker's emotional attitudes towards what he is saying, his assumption bout his interlocutor's attitudes, etc. and, on the other hand, the operation of the psychological and physiological mechanisms involved in the production of utterances.
The competence-performance distinction, thus drawn, is at the very heart of generativism. As presented in recent years, it relates to mentalism and universalism in the following way. A speaker's linguistic competence is a set of rules which he has constructed in his mind by virtue of his application of his innate capacity for language-acquisition to the language-data that he has heard around him in childhood. The grammar that the linguist constructs for the language-system in question can be seen as a model of the native speaker's competence. To the extent that it successfully models such properties of linguistic competence as the ability to produce and understand an indefinitely large number of sentences, it will serve as a model of one of the faculties, or organs, of the mind. To the extent that the theory of generative grammar can identify, and construct a model for, that part of