The general condition for use of some form of Neg in describing things appears to be another type of contrast - contrast between what was expected or predicted by the speaker and what he actually observed. The unfulfilled expectation may concern the absence or altered condition of an entity, in which case Neg would be expressed in a noun phrase; or it may concern the absence or altered character of an expected action or state, in which case Neg would appear in the verb phrase. The forms that Neg can take are many (cf. the ‘classic’ paper on negation by Klima, 1964) and the rules by which Neg forms combine with each other are very subtle (cf. a paper by Baker, 1969, on double negatives). Linguists have noted that ‘positive polarity’ and ‘negative polarity’ expressions tend to occur in pairs: for example, always/never, some/any, either/neither, always/never and certain verbs like expect/doubt (note that I expect that you will find some gold and I doubt that you will find any gold are entirely acceptable but I expect that you will find any gold and I doubt that you zvill find some gold are questionable at best). With regard to predicates like I am surprised that, I am relieved that and It is lucky that, Baker (1969) says that ‘speaking intuitively, we can say that each of these predicates expressed a relation of “contrariness” between a certain fact and some mental or emotional state’ (p. 33). Thus, I am surprised that anyone could sleep but hardly *I am surprised that someone could sleep.

As a test of the unfulfillment prediction with respect to presence or absence of expected entities, we may compare events #16 (THE PLATE IS EMPTY) and #32 (THERE IS NOTHING ON THE TABLE) with events # 15 (A BALL, A SPOON AND A POKER CHIP ARE ON THE PLATE) and #26 (SOME POKER CHIPS ARE ON THE TABLE). For the control events (#15 and # 26) none of the sentences produced contained Neg in any form, but for the test events (#16 and #32) there were 62% and 85 % Neg respectively. Most Neg in #16 were the adjectival form, empty :; in #32 Neg divided about equally between adjectives (bare, naked) and the negative nominal form, nothing. To test the unfulfillment prediction with respect to absence of an expected action, we may compare events # 24 (THE BALL MISSES THE TUBE) and #25 (THE BALL WILL MISS THE TUBE) with control events #22 (THE BALL HIT THE TUBE) and #23 (THE BALL WILL HIT THE TUBE). Again, there were no Neg of any type for the control events, whereas the test events yielded 26% and 30% Neg respectively—usually either not with a positive verb (not hit) or a negative verb (miss), although one without hitting was offered.