THREE NOTES ON THE PERFORMATIVE ANALYSIS Asa Kasher The performative analysis of declarative sentences,,as suggested by Ross and defended extensively by others,^ rests on two" pillars: the underlying abstract representation and the rule of performative deletion. In what follows I point out a syntactic crack in the latter rule and a semantic creak made by the former representation. I also suggest that any amended version of the performative analysis should be couched in appropriate pragmatical terms. 1. A Syntactic Note The performative analysis postulates for every declarative sentence an underlying representation of the following form: (1) S NP I V NP NP +performative ^communication +linguistio ♦declarative _ When the embedded clause of (l) ends up in surface structure as an independent sentence, it is the performative deletion transformation that does the Job: Mnp +V ♦performative ♦c ommunication ♦linguistic ♦declarative S, NP li 4- 0 o o It 1 2 3 It has been argued that (2} is obligatory. This explains the difference between (3) and (4), both of which have the same under- lying representation. (3) There are no black tulips. (it) *I tell you that there are no black tulips. The difference between (It) and (5) ie explained in the same way, since, strictly speaking, (2) does not apply to structures which include any item in addition to the "standard" ones, i.e. a first-person subject, a verb of saying, a second-person indirect object and an embedded sentenc (5) I hereby tell you that there are no black tulips. There are, however, sentencesjrhose underlying representations both include nonstandard items and have undergone some performative deletion operation: