This seems clear enough, but when we look carefully at Chomsky's use of the term `competence' we quickly notice that it must be defined relative to the ideal speaker-hearer. Speech that reflected competence with no hitch would contain none of the errors associated with performance. And indeed as this unfolds, it becomes clear that it is this--this set of rules which if applied in an ideal situation would produce perfect speech (performance)--which Chomsky has the linguist trying to uncover or discover or master or ... Thus,
competence is the ideal speaker-hearer's knowledge of his language
Linguistics is the study of competence, but to get to this, we must sift through the data provided by actual languages in performance and take account of the linguistically irrelevant factors of performance in order to get to the speaker-hearer's underlying competence. Thus, we study performance, taking careful note of all the various factors which can interfere with a speaker-hearer's application of his ``underlying'' competence, with the aim of uncovering that competence..
In light of this, when we read Quine we wonder why he didn't see this, thus sparing his linguist all the bother of accounting for the various factors that form part of performance in his theory of language, the formation of his manual of translation, rather than looking at them only in order to eliminate them from a theory that captures only competence. Of course, the reasons aren't too difficult to discern. I want to continue a bit farther with Chomsky's description of linguistics before returning to this.