Classical NEG Raising: An Essay on the Syntax of Negation

1. Introduction

Introduction 1.1 Domain of Inquiry On one reading, English examples like (1a) seem to be paraphrases of corresponding examples like (1b): (1) a. Karen expected that the moon would not turn purple. b. Karen did not expect that the moon would turn purple. In addition to the interpretation of (1b) that is a paraphrase of (1a), (1b) has another interpretation where Karen had no expectations at all, perhaps because she had not thought about the matter, or because she had thought about the matter but was undecided. This a priori unexpected semantic similarity between structures with negation in the complement clause and those with negation in the main clause, found in many other languages as well, has been the subject of a great deal of work; see Horn 1971, 1975, 1978, 1989 [2001] for extensive references to earlier periods dating back as far as St. Anselm in the eleventh century. Beginning with Fillmore 1963, within the generative grammar tradition there developed in the 1960s and 1970s considerable agreement that the roughly equivalent reading of pairs like (1a,b) was a function of a syntactic phenomenon of negation (henceforth: NEG) raising, from the complement into the main clause. That is, the shared reading was taken to be associated with cases like (1b) because the main clause NEG originates in the complement clause and is raised into the main clause. This conclusion involved the assumption that the preraising structure determined the meaning. The original syntactic conception is nicely represented in the following passage: (2) (Horn 1971:120) “It is convincingly argued that the ungrammaticality of the sentences which result from substituting verbs such as demand or claim for want and believe — 1 4 Chapter 1 (3) a. *Chauncey doesn’t demand to die until he has touched fair Hermione’s lips again. b. *I don’t claim you have remembered to button your fly in years. reveals that NEG-raising is a minor rule applying to some predicates of opinion and expectation . . . , of intention . . . and of perceptual approximation . . .” The posited raising involved a transformational movement rule, and we will informally assume such an approach in this monograph. But NEG raising, as argued for in this work, could equally be given a nontransformational syntactic account.1 We refer to the phenomenon illustrated in (1) as Classical NEG Raising (hereafter: Classical NR). The modifier is motivated by the fact that, following Postal 2005, we posit a wide variety of NEG raisings distinct from that taken to be present in (1b). This more general appeal to NEG raising is discussed in some detail in chapter 3. While a syntactic approach to Classical NR was popular at the earlier period, not long after Fillmore’s (1963) article, acceptance of such an approach began to erode; see Jackendoff 1971, Pollack 1974. Increasingly, a syntactic view of Classical NR was challenged by pragmatic and semantic approaches.2 And even earlier adherents of a syntactic conception, such as Horn (1971:120, 127; 1972:228), came to reject it in favor of nonsyntactic views; see Horn 1978, 1989 [2001], Horn and Bayer 1984. Our goal in this monograph is to argue that this rejection was mistaken. A key matter is the well-known fact that Classical NR only exists with a particular highly restricted subclass of higher-clause predicates. Horn (1975, 1978, 1989 [2001:322–330]) argued extensively that this class has a universal semantic characterization, while stressing that it is nonetheless subject to parochial lexical restrictions excluding elements that otherwise fall under the universal characterization. We simply refer to a predicate that permits Classical NR as a Classical NR predicate (hereafter: CNRP). Cited members of this class in English include the predicates in (3), although some exhibit a certain amount of speaker variation: (3) advisable, advise, appear, believe, choose, expect, feel, feel (like), figure, find, guess (dialectal), imagine, intend, likely, look (like), mean, plan, reckon (dialectal), recommend, seem, sound (like), suggest, suppose, supposed, tend, think, turn out, used to (temporal form), want, wish Introduction 5 For concreteness, we will assume a syntactic version of Classical NR that has the form in (4): (4) Classical NR If NEG1 raises from one clause B into the next clause above B, call it clause A, then the predicate of clause A is a CNRP. A remark about the logic of formulation (4). It is designed to fit into our broader view that there is a general phenomenon of NEG...