Discussion of a case where direct psycholinguistic evidence appears to weigh in favor of a model of the mental grammar, viz. an 'abstract performative grammar', rather than any model that 'autonomous linguistics' might construct.