Empty Category Principle

In linguistics, the empty category principle (ECP) was proposed in Noam Chomsky's syntactic framework of government and binding theory. The ECP is supposed to be a universal syntactic constraint that requires traces to be properly governed.

Proper government is defined as follows:

• A properly governs B iff A theta-governs B or A antecedent-governs B.
  • A theta-governs B iff A governs B and A theta-marks B.
  • A antecedent-governs B iff A governs B and is coindexed with B.

The ECP is a way of accounting for, among other things, the empirical fact that it is generally more difficult to move up a wh-word from a subject position than from an object position.

The intermediate traces must be deleted because they cannot be properly governed; theta-government is impossible because of the position they occupy, Spec-CP; the only possible antecedent-governor might be an overt NP (a wh-word), but the Minimality Condition would always be violated because of the tensed I (which must be present in all matrix clauses), the tensed I would c-command the intermediate trace but it would not c-command the wh-word. So we have to say that intermediate traces must be deleted at LF so that they can avoid the ECP.

In the case of object extraction (the trace is a complement of VP), theta-government is the only possible option. In the case of subject extraction (the trace in Spec-IP), antecedent-government is the only possible option.

If the trace is in Spec-IP and we have an overt complementizer (such as that), the sentence is ungrammatical because the ECP is violated. The closest potential governor would be the complementizer, which cannot antecedent-govern the trace because it is not coindexed with it (and theta-government is impossible since trace is in Spec-IP).

For example, in the sentence Who do you think (that) John will invite? the ECP works in the following way (the structure is given for the embedded clause only):