1. I put this down not to be quarrelsome but as an instance of what can happen when even the best of analyses tries to carry its procedures too far. In hisSyntactic StructuresChomsky proposes that we can account for the falling intonation of interrogative-word questions on the basis of a sequence of two transformations, Tqand Twl. The effect of the first is “in part to convert the intonation from one of these to the other [i.e., to reverse the terminal fall and terminal rise],” and since the second has the same effect, and in the transformational history is applied after the first, there is a double reversal, with the result that the intonation of interrogative-word questions is like that of declaratives. The proposal takes no account of the many interrogative-word questions (e.g., all the reclamatories) and yes-no questions (all the assertives with rising intonation) in which the opposite of what is assumed is true. Chomsky's error stands as a curious example of how two different interpretations of the same bit of grammatical imagery can lead to opposite conclusions. He and Jespersen both picture questions as “doing something to” statements, and both conceive of certain questions as being “added to” other questions (for Jespersen,Modern English GrammarIV, §18.7 [2]; this gives a “question raised to the second power”). But where Chomsky's train of thought runs something like this, “What a question does to a statement isreverse its direction—therefore, another added question wouldreverseit again and give a lowered pitch,” Jespersen's train runs like this, “What a question does to a statement israise its pitch—therefore, another added question wouldraiseit some more and give an extra-high pitch.” Chomsky goes off on the tangent of interrogative-word questions. Jespersen goes off on the tangent of echo questions. Both are right, and both are wrong.