The Negation Operator
Negation of simple clauses is formed with the negative particle not. There are two kinds of negation: simple sentence negation and contrastive negation.
Let us start with sentence negation. At S-structure the negative particle is adjoined to the first auxiliary verb in the clause:
Negation is an operator. Its function is to claim that the proposition it modifies is false. In (2), for example, the speaker is claiming that the proposition that Mary went to the movies is false. In the logical system developed here, Neg is an operator forming a Neg-proposition with its argument an M-proposition:
In the first pass to the lexicon[NEG] is bound and weak. It's host is T:
Following a binary analysis, NEG contains the feature [+Neg]. [+Neg] is negative, while [-Neg] is positive. It is almost always empty phonetically. When two people are having a disagreement, one may say:
It appears that so may be the positive marker. If so, it is the only case I can think of in English where[-Neg] has a phonetic form. o that (5) I think not does not means the same thing as I don't think. Not in (5) is negating the understood complement which is empty here.
The logical structure of b) is the following:
Negation is a weak operator in that it does not trigger movement. The S-structure for (2) is the following:
The S-structure for the first clause in (3) is the following:
Why is Neg adjoined to T. The only answer at this time is that English evolved in such a way that T became the hose for the sentential negative particle ([Neg not]. If Neg is generated at D-structure adjoined to V, then there can be no explanation accounting for its position. There is reason to believe that operators form heads of projections as we have seen. Suppose we assume that there is a projection of Neg dominated by NegP, the position of the N-proposition. NegP now dominates CP which in turn dominates TP:
There may be a semantic explanation for this hierarchy as must be the case for NegP, since it must take scope over the sentence it negating.
In the first pass to the lexicon we find out that Neg is bound and weak and is marked for the host T. We will formally distinguish between clausal negation and contrastive negation with the feature [Contrastive]:
Any declarative statement that we make can be affirmative (unmarked normally) or negative (usually marked with Neg). If (5) is correct, then Neg originates in NegP and then it must lower to the verb. Neg is a weak particle which means that it cannot stand alone unless the remainder of the sentence is omitted:
Here, not is the only form left after the remainder of the sentence is omitted:
If the argument of not is not omitted, then not must lower:
It is a parameter of English that the only acceptable host for the particle Neg can be an inflected auxiliary (strong) verb. This explains why not does not lower to NP (Mary). It cannot lower to go because go is neither an auxiliary verb nor is it inflected here. If there is no auxiliary verb, Neg cannot lower to the main verb:
The functional (dummy) auxiliary verb do must be inserted to function as a host for tense which licenses Neg:
The infinitive provides some further evidence. The negative particle cannot lower to the uninflected auxiliary (or a the main verb) in an infinitival clause:
(13 To have been loved or not to have been loved is the question.
) a.
b. *To have been loved or to have not been loved is the question.
Although the negative particle is adjoined to the inflected word, it is tense that licenses the lowered negative particle. Not is strong only when it cannot lower. The strengthening of not is a last resort.
The lowering of Neg to T brings up a new problem. Although Neg is adjoined to T, it does not form the same kind of bond with T that an inflectional does. First note that not is spelled as a separate word whereas inflectional endings never are:
The difference appears to be in the kind of boundary between T and Neg. It is stronger than the morpheme boundary. The symbol `#' is often used to mark this boundary, indicating a word boundary. But we emphasize here that `#' marks adjunction. `##' is used to mark independent words as if we were to rewrite ) with boundaries:
Normally, only one word in the X#Y construction can be stressed. However, the second word can be stressed in an emphatic context:
The next question is how to represent this structurally. This seems not to be a structural problem but one of morphology and word concatenation. Looking at some other languages, the agreement inflectional morphemes are the periphery of inflection, and if anything is adjoined to the word containing agreement morphemes, the bond is weaker--that is it is `#'. Unless it becomes absolutely necessary to mark it, we shall defer this question for now.
As we mentioned above Neg blocks the lowering of T. Neg, however, remains strong. It does not raise to Neg in the syntax:
The hostless T must wait until the second pass to the lexicon where the appropriate dummy verb is selected. If T is a sister to a main verb, {DO} is selected:
The complete structure after the lowering of Neg is:
The feature clitic means that the features associated with it are realized as a distinct clitic form in the syntax. T, then, splits into T plus Neg. The agent raises in search for Case:
In the second pass to the lexicon, the bound features are spelled out as [ EMPTY ], {do} is inserted, {do}+[+Past] is spelled out as di+d, Neg as not, TO as to, [+Def] as [ EMPTY ] preceding Mary and as the preceding movies, MARY as Mary MOVIE+[+Pl] as movie+s:
Constrastive negation is where any node (phrasal node?) is the complement of not, and the negated phrase is conjoined to a positive phrase by the conjunction but. Not is a word, Neg, and c-commands the phrase that it takes scope over:
(23 Mary may go not to the movies but to a play.
) a.
b. Mary may go [ not to the movies ] but to a play.
Here, the PP is contrastively negated. Although traditional grammar. The string not to the movies but to a play forms a constituent as evidence in preposing:
(24 Not to the movies but to a play did Mary go.
) a.
b. Not to the movies did Mary go but to a play.
b) shows that but a to a play is a constituent. It is moved to the end of the clause by Right Node Movement. The effect of RNM is to put the constituent into a marked focus position. Not to the movies is also a constituent as evidenced in a cleft construction:
It appears that but takes two arguments:
But is the head of the construction. What remains unclear is why to a play forms the immediate complement of but, and why not to the movies occurs to its left. We tentatively propose the following S-structure for is the following:
As in the case of NPs, we assume that operators modifying PPs are adjoined to PP as indicated in .
Of course, Neg may take NP as an argument as well as PP
(28 Bob bought not this red car but that blue one.
) a.
b. It was not Sebastian but his brother who got tipsy at the party.
Or it may take a AP or a CP as an argument:
(29 Cosmo was not sad but happy when he retired.
) a.
b. Eloise was not just overjoyed but exuberated when she won the lotto.
(30 It is not that I hate you, but that I like you why I did it.
) a.
b. It was not that his wife left him that bothered Jason, but that she ran off his incompetent nephew.
And Neg may take a VP as an argument:
(31 What John did was not sell the car but wreck the car.
) a.
b. Sam must not wash the car but repaint the car.
The difference between clause negation and contrastive negation is that in the latter case, a specific incomplete proposition is negation, whereas in clause negation the entire proposition is claimed to be false, but not which part. For example:
(32 Zelig didn't buy a red tie for his brother.
) a.
b. Zelig bought not a red tie but a blue tied for his brother.
c. Zelig bought a red tie not for his brother but for his father.
Contrastive staress does not appear to occur in subject position:
(33 ?*Not Zelig but Malissa bought a red tie her brother.
) a.
b. It wasn't Zelig but Malissa that bought a red tie for her brother.
The cleft construction is used to mark the subject in contrastive negation.
This is difficult to determine. CONTRAST is an operator that can modify an eventuality or an object. CONTRAST takes two arguments. Theta roles have never been cooked up for these. The basic logical structure for contrast is the following:
The arguments of CONTRAST can be either an object (NP) or an eventuality, but each argument must match the other; that is, if one is an object, the other must be an object. The subscripts are used to ensure that the two objects or two eventualities are not the same.
The logical structure for a) is the following:
In the first pass to the lexicon CONSTRAST is linked to the Conunction containing bur; the objects are assigned NP, and NEG, Neg:
For some unknown reason, either NP may raise and adjoin to the top NP. The first NP raises to form a):
The remaining rules discussed previously apply and in the final pass to the lexicon, Conj is spelled out as but if Conj is c-commanded by Neg.
If it precedes Neg, CONTRAST must be empty:
To go to theta roles click here.