Lexical array

Formal device, introduced in Chomsky (1998); a lexical array is the set of lexical items that will be used in a derivation. The notion is similar to numeration; the only difference is that the elements in a numeration bear indices, while the elements in a lexical array do not. The empirical evidence for the existence of lexical arrays comes from sentences like the following:

(i) a.  Therei is likely [A ti to be          [a proof discovered]]. b. * Therei is likely [A a proof to be         [ ti discovered]].

At stage A of the derivation, T's EPP-feature must be checked. This could be done by either merger of there, or movement of a proof. Chomsky argues that (b) is ungrammatical because a proof has been moved before the expletive is inserted, and this violates Merge-over-Move. This is indirect evidence for the existence of lexical arrays, because the principle of Merge-over-Move doesn't make sense if we don't have a restricted set of accessible elements; if the lexicon would be fully accessible during the derivation, and we have Merge-over-Move, we would expect every sentence to have an expletive satisfying the EPP, since merger of an expletive is cheaper than moving the subject. See also lexical subarray.

Link
Utrecht Lexicon of Linguistics