Professional Documents
Culture Documents
Graduate School
Las Piñas City, Lucena City Group
Syntax Negation
In many cases, a negative word can be formed by adding the prefix un- to the positive form of a
word (as in unhappy and undecided). Other negative affixes (called negators) include a-, de-,
dis-, in-, -less, and mis-.
Î He is not a linguist.
Î am not rich
The conditional complementizer if presupposes that the condition may or may not be true:
Another example is the simple fact that something can be wrong, i.e. not right:
(36) Stealing is wrong.
Further examples include children playing pretend, irony and sarcasm, lies and deceit,
etc.
Thus, negation is undoubtedly in the language of thought (LOT) and therefore it must
also be in LF, as the feature [Neg], and in turn, therefore also in the syntax and hence, also in
NPIs may also be licensed by non-negative verbs; for example, wonder also licenses
NPIs but it selects an interrogative clause as its object: I wonder [whether he she knew / if she
knew].
the lexicon; recall that the LF representation is derived from a lexical array LA taken from the
lexicon. As universal (functional) categories project (e.g. tense and TP, see Sigurðsson 2003),
there is also a universal NegP. Due to recoverability in interpretation and learnability, at least
one or both of spec-Neg and Negº must be realized overtly (see chapter 2, section 2.4.2) (this is
probably what drives Jespersen’s Cycle, Jespersen 1917; see chapter 3, sections 3.2.5 and 3.2.6).
Smith et al. (1993) has tried to teach Christopher (an autistic man with extraordinary linguistic
abilities in the face of severe mental retardation) an artificial language which they call Epun.
This artificial language does not have a negation marker and thus deviate from the principles of
universal grammar UG. Instead, negation is expressed with word order: In negative clauses the
verb precedes the subject, and past tense is realized by fronting the object:
Though the system seems fairly simple, neither Christopher nor the normal controls were able to
learn it. On the other hand, this would most likely be easy for an artificial neural network to learn
as it operates on statistics and, unlike human language users, structure-independently. A human
language, i.e. one that satisfies the principles of UG, on the other hand, must have a negation
marker.
Types of Negation
This is a very brief introduction to negation. The point is to establish sentential negation which
will be one of the main topics of this dissertation. For a comprehensive analysis of negation, see
e.g. Horn (2001) and Haegeman (1995).
Basically, I shall argue that there are three different types of negation, namely (i)
sentential negation which is realized in the clausal spine as NegP positioned between FinP and
TP (sentential negation has wide scope: it negates the whole clause); (ii) constituent negation
which has narrow scope (it does not negate the clause, only constituents such as DPs and PPs);
(iii) meta negation which has neither constituent / narrow scope nor sentential / wide scope; It
has scopes over more than the clause, so in a sense it has the widest possible scope. These three
types of negation have different structural positions in the syntactic tree, but sentential negation
is part of the clausal spine; meta-negation is outside or above it, while constituent negation can
be anywhere else, for example below it on a DP object (see e.g. Zanuttini 1997 and Cormack &
Smith 2002 for analyses with multiple NegPs.) The scope of the negative operators is their c-
command domain:
Sentential Negation
The scope of negation can be tested with, for example, an opposite truth value test,
positive and negative tags, and negative polarity items.
A negative sentence has the opposite truth value of the corresponding sentence without
negation: If X is true, then the negation of X (¬X) is necessarily false; and if X is false, then ¬X
is necessarily true. For example, if (39)a is true, the (39)b is necessarily false, and vice versa
Opposite truth values is a necessary but not a sufficient condition on sentential negation.
Clauses with opposite truth values need not be a clause (X) and its negated counterpart (¬X).
For example, if (a) is true, then (b) must be false, and vice versa, but the cannot possibly
be described as a clause and it negated counterpart. This becomes even more clear when (c) is
taken into consideration. Only one of the three can be true at a time, but neither (b) or (c) means
the same as (d) which is the real negative counterpart of (a). Furthermore, if (a) is false and (d) is
true, it does not necessarily follow that either (b) or c are true; a, b, and c may all be false at the
same time.
a. Gunnar is in Oslo.
b. Gunnar is in Baghdad.
c. Gunnar is in Sweden.
(42) a. I will not get it right, and neither will you. (Negative)
Likewise, positive sentences take negative tag-questions, such as won’t I? or positive elliptic
conjuncts, such as and so will you, while such tags are incompatible with negative polarity:
Like the opposite truth values test, NPI licensing is a necessary but not sufficient condition, as
NPIs can be licensed in interrogative contexts as well, regardless of polarity:
To summarize, a negative sentence (i) is the counterpart of the same sentence without
negation, (ii) takes positive tag-questions (will I?) and negative conjuncts (and neither…), and
(ii) disallows negative tag-questions (won’t I?) and positive conjuncts (and so…), and (iv) allows
negative polarity items (NPIs).
Constituent Negation
Constituent negation has narrow scope compared to the wide scope of sentential
negation; it scopes over e.g. an NP, an AdvP, a small clause or a VP:
a. With [not [NP too many errors]], this should work. (NP)
Constituent negation fails in all the tests that sentential negation passed, and passes the
ones sentential negation failed.
A clause with constituent negation does not have the opposite truth value of the
corresponding clause without negation: (a) is not incompatible with (b) and they can both be true
at the same time. The true negative counterpart of (a) is (c).
Constituent negation does not license negative tags, neither positive tag-questions, as in
(a), nor negative tag-clauses, as in (b). This means that the negative operator clearly does not
scope over the clause.
In contrast, constituent negation is compatible with positive tags, both negative tag
questions, as in (a), and positive tag-clauses, as in (b). Again, this is a strong indication that the
clause as such is not negative.
Finally, unlike negation with sentential scope, constituent negation does not license NPIs:
Meta-negation
As stated above, meta-negation has a wider scope than the wide scope of sentential negation.
Meta-negation is not the same as sentential negation. It selects a CP (which can be either
declarative, as in (a) and (a), or interrogative, as in (b) and (b) or a PP, (c) and (c), not a TP or
even a FinP, as in (d) and (d):
(53) Da: a. [Ikke [CP at det ikke nytter noget]] Not that it not is.of.use any “Not that it
doesn’t help.”
b. [Ikke [CP om jeg gider at høre mere om det]] Not if I bother to hear more
about it “I simply don’t want to hear more about it”
Note that it selects a subordinate clause. This is particularly clear in the Danish examples
because the embedded word order is different from the one in main clauses. Finite verbs move to
Cº (second position) in main clauses but remain in situ in embedded clauses and thus follow
sentential adverbials and negation. Meta-negation cannot be fronted/topicalized sentential
negation because
(i) topicalization only takes place in (matrix and embedded) main clauses, and
(ii) because topicalization of negation is otherwise impossible in Danish and English.
Interestingly though, meta-negation passes the tests for sentential negation and fails the tests for
positive polarity:
GENERALIZATIONS
This is formed through adding word “not” after the first auxiliary verb in the sentence. As a
review, the auxiliary verbs in English are:
Do Have am shall
Does Has is must
Did had are might
was Can
were will
wound
Negation can be applied to all verb tenses that we have.
Remember, when we do not have an auxiliary verb in the sentence, we add the auxiliary verbs
do, does or did.
Example 1:
Affirmative Negative
You must study your lessons You must not (mustn’t) study your lessons
I had been to Canada I had not (hadn’t) been to Canada
She will be visiting an old friend She will not (won’t) be visiting an old friend
We can attend practice tomorrow We cannot (can’t) attend practice
tomorrow.
She has a beautiful garden She does not (doesn’t) have a beautiful
garden
They were having breakfast They were not (weren’t) having breakfast
References: