On the Analysis of Tough-Movement

Question

Tough-movement was said to involve object-to-subject movement back in the day. The object of the embedded clause is raised to the matrix clause:

Mary1 is tough to please t1

Later, the analysis of tough-movement was revised in G&B. It no longer involved object-to-subject movement but the movement of an operator to SpecCP and the generation of PRO in SpecIP of the infinitival clause (control).

Mary is tough Op1 PRO to please t1

What made linguists rethink the analysis of tough-movement? Was object-to-subject movement inadequate for some reason?

Answer 1

Lasnik & Fiengo (1974) "Complement Object Deletion" (section 2) argue against Tough Movement, and Chomsky (1977) "On WH-Movement" p. 102 adopts the base-generated analysis and accepts the correctness of that approach without comment. L&F begin by presenting arguments against arguments against tough-deletion. For instance, an earlier argument for TM was based on the grammaticality of "Headway is easy to make on problems like these", where the noun headway – which it was assumed only exists as the object of "make headway" is raised. L&F raise an objection to the datum, and then counter with the ungrammaticality of parallel idiom-objects raisings like "*Tabs were easy to keep on Mary", "*Heed is important to pay to such warnings", which under the raising analysis should be grammatical. There is also an amusing argument based on tough movement and nominalization, centering around the assumed ungrammaticality of "John's easiness to please", which in fact is ungrammatical for some people and grammatical for others, and as a person who finds this grammatical I find the assertion that this is ungrammatical to be as incomprehensible as others find the assertion that this is grammatical.

Their section 2.3 presents a series of empirical arguments against TM; for instance, "John is being easy to please" would have to come from "*To please John is being easy". (Their arguments are this "progressive" argument, the "try" argument, the indefinite NP argument, the intentionality argument, and the modal argument"). Overall the argument does not appear to be theory-driven, i.e. they aren't arguing that we have to get rid of TM because it violates some valuable grammatical principle, it just that raising doesn't work.

[EDIT]

On re-reading and recollecting, I think there actually is an underlying theoretical principle, the Specified Subject Condition, which TM violates. While L&F do not structure the argument to make SSC be the driving force, they do take advantage of the opportunity to point to mention that SSC would be runned afoul of in certain TM cases. I think the operator and control analysis is from Chomsky (1977).

Answer 2

I'm not quite familiar with the details but back then, I suppose, it had to do with theta-role assignment. That is, if you assume the movement of Mary from the object position of the embedded clause to the subject position of the matrix clause, it means Mary (DP) is already assigned a patient theta-role and it can't be assigned more by T (is) after movement, because it would violate the "one role per DP" principle. Again, I'm not very sure because it would mean there's something wrong with the usual movement too. There's a bunch of accounts for this in MP, but I guess it's not the purpose of this question.