Earticle

The purpose of this paper is to review Chomsky’s (1998, 1999) multiple agree, looking for problems and suggesting solutions. When there are defective probes, the goal cannot delete its uninterpretable feature and it continues to be active until it searches for the probe of complete ϕ-feature. So when a multiple agree happens, more and more pairs of agree can be generated, and Defective Intervention Constraint does not apply to multiple agree by means of the Maximization Principle. Defective Intervention Constraint seems to contradict the Maximization Principle. To get rid of Chomsky’s problems, I’d like to suggest Multiple Agree Principle as follows. (1) Multiple Agree Principle Multiple Agree must apply to all the defective probes and goals between a complete probe and a complete goal in a single simultaneously derivational operation. If we apply multiple agree like this, we can solve Chomsky’s (1998, 1999) problems. Because a multiple agree applies to all the defective probes and goals simultaneously, there is no chance for the intervening defective probes and goals to be inactive beforehand. And we can keep the Maximization Principle because a defective probe and goal can get an agree with a complete probe and a complete goal at the same time. The number of agreed pairs can be decreased because the multiple agree applies in a single operation.