Classical NEG Raising: An Essay on the Syntax of Negation

12. Islands: Survey

Islands: Survey 12.1 Preview In this chapter, we survey a range of cases where island constraints block Classical NR. At issue are examples invoking the Complex NP Constraint, clause-internal topics, truth predicates, wh-islands, clause-internal clefts, pseudoclefts , and Negative Inversion. The clear generalization is that Classical NR is never possible from an island. Such a generalization is especially striking for cases where all known semantic conditions on Classical NR are met (e.g., for truth predicates), but Classical NR is still not possible. As known since Ross’s (1967 [1986]) establishment of the notion island, syntactic raising phenomena are subject to island constraints. Therefore, it is possible to account naturally for the above generalization under the assumption that classical NR is a syntactic raising phenomenon, as outlined in chapter 1. 12.2 Complex NP Constraint Cases Consider first clausal complements of nouns. Such nominal complement clauses are islands and therefore, on our syntactic account, these complements cannot be origins for Classical NR. The islandhood of clausal complements to nouns is illustrated in (1): (1) a. When do you believe that the moon will vanish? b. *When do you hold the belief that the moon will vanish? The following sentences indicate that when a strict NPI appears in a clause embedded under belief, its NEG cannot raise to the matrix clause: (2) a. I don’t believe that the moon will vanish until Thursday. b. I hold the belief that the moon will not vanish until Thursday. c. *I don’t hold the belief that the moon will vanish until Thursday. 12 112 Chapter 12 Similar facts hold for the complement of expectation: (3) a. What town do you expect that they will find rebels in? b. *What town do you have the expectation that they will find rebels in? (4) a. I don’t expect that they will find a living soul in that town. b. I have the expectation that they will not find a living soul in that town. c. *I don’t have the expectation that they will find a living soul in that town. That is, just as noun complements preclude interrogative extraction, they also do not allow strict NPIs whose licensers are outside of the complement clause.1 Other examples illustrating the same constraint are given in (5): (5) a. *Deborah did not have the expectation that Ted would stop at anything to get promoted. b. *Deborah did not entertain the thought that Carolyn would breathe a word about it. c. *Deborah did not hold the belief that Carolyn had told a living soul. d. *Deborah did not have the expectation that a damned/fucking thing would go wrong. e. *Deborah did not entertain the thought that Karen would lift a finger to help Sidney. f. *Karen did not hold the belief that Ted would learn squatA about Turkish politics. 12.3 Clause-Internal Topics As is well-known, topicalization internal to an embedded clause turns that clause into an island: (6) a. When does Leslie believe that Jim should call Irene? b. *When does Leslie believe that Irene, Jim should call? Correspondingly, Classical NR is blocked in such cases: (7) a. Leslie doesn’t believe that Jim should call Irene until tomorrow. b. *Leslie doesn’t believe that Irene, Jim should call until tomorrow. (8) a. Leslie doesn’t believe that a damned thing upset Lester. b. *Leslie doesn’t believe that Lester, a damned thing upset. Islands: Survey 113 (9) a. Leslie doesn’t believe that Carl said squatA to the chairman. b. *Leslie doesn’t believe that the chairman, Carl said squatA to. Since (6a,b) indicate that internal topicalization renders the complement clause an island, the fact that (7b), (8b), and (9b) containing strict NPIs are ungrammatical then argues that Classical NR is impossible out of that kind of island as well. 12.4 Truth Predicates In chapter 1, we outlined Bartsch’s (1973) pragmatic approach based on the excluded middle property. Truth predicates are particularly problematic for such an approach. In this section, we develop a criticism of a pragmatic or semantic account of the relevant properties of truth predicates, specifically true or the case. The discussion begins with key observations from Horn 1975:291–292, 1978:203: (10) a. It is not true/the case that 2 plus 2 is 8. b. It is true/the case/that 2 plus 2 is not 8...