Logical Form & Predicate Logic

Jean Mark Gawron

Linguistics
San Diego State University
[email protected]
http://www.rohan.sdsu.edu/∼gawron

Logical Form & Predicate Logic – p. 1/30

Predicates and Arguments

Logical Form & Predicate Logic – p. 2/30

 Logical Form

The Logical Form of English sentences can be represented by formulae of Predicate

Logic,

What do we mean by the logical form of a sentence?

We mean: we can capture the truth conditions of complex English expressions in

predicate logic, given some account of the denotations (extensions) of the simple
expressions (words).

Logical Form & Predicate Logic – p. 3/30

 The Leading Idea: Extensions

Nouns, verbs, and adjectives have predicates as their

translations.

What does ’predicate’ mean?

’Predicate’ means what it means in predicate logic.

Logical Form & Predicate Logic – p. 4/30

 Intransitive verbs

Translating into predicate logic

(1) a. John walks.

b. walk(j)

(2) a. [[j]] = John

b. “The extension of ’j’ is the individual John.”
c. [[walk]] = {x | x walks }
d. “The extension of ’walk’ is the set of walkers”

(3) a. [[John walks]] = [[walk(j)]]

b. [[walk(j)]] = true iff [[j]] ∈ [[walk]]

Logical Form & Predicate Logic – p. 5/30

 Extensional denotations

• Remember two kinds of denotation. For work with logic, denotations are always
extensions.
• We call a denotation such as [[walk]] (a set) an extension, because it is defined by
the set of things the word walk describes or “extends over”
• The denotation [[j]] is also extensional because it is defined by the individual the
word John describes.
• Next we extend extensional denotation to transitive verbs, nouns, and sentences.

Logical Form & Predicate Logic – p. 6/30

 Transitive verbs

(4) a. John loves Mary

b. love(j, m)

(5) a. [[j]] = John

b. [[m]] = Mary
c. [[love]] = {hx, yi | x loves y}
d. “The extension of ’love’ is the set of pairs of individuals x and y such that x
loves y.”

(6) a. [[John loves Mary]] = True iff h[[John]], [[Mary]] i ∈ [[love]]

b. [[love(j, m)]] = true iff h[[j]], [[m]]i ∈ [[walk]]

Logical Form & Predicate Logic – p. 7/30

 Nouns

• A simple noun predication

(7) a. Fido is a dog.
b. dog(f)
• But what about ...?
(8) A dog barked.
Different meaning of dog? Hopefully not!

Logical Form & Predicate Logic – p. 8/30

 Adjectives

• A simple adjectival predication

(9) a. Fido is happy.
b. happy(f)
• But what about ...?
(10) A dog is happy.
Different meaning of happy? Hopefully not!

Logical Form & Predicate Logic – p. 9/30

 Assuming both nouns and verbs are
predicates

• Fido is a dog.
dog(f)
• Fido barked.
bark(f)
• A dog barked.
dog(x) ∧ bark(x)

BOTH predicates present: some non-specific x is a dog and barked.

Logical Form & Predicate Logic – p. 10/30

 Assuming both nouns and adjectives
are predicates

• Fido is a dog.
dog(f)
• Fido is happy.
happy(f)
• A dog is happy.
dog(x) ∧ happy(x)

BOTH predicates present: some non-specific x is a dog and is happy.

Logical Form & Predicate Logic – p. 11/30

 A new use of ∧

(11) a. A dog barked.

b. dog(x) ∧ bark(x)
c. a dog is happy.
d. dog(x) ∧ happy(x)

• We’re using ∧ even though the word and hasn’t occurred in either sentence.
• ∧ is going to turn out to have a lot more uses in our logical translations than just
as a translation of and
• Other sentential logical connectives will also turn up in surprising places

Logical Form & Predicate Logic – p. 12/30

 Sentential Connectives (revisited)

We still use sentential connectives from statement logic for English sententential
connectives (where they work!)

(12) a. John doesn’t love Mary

b. ¬love(j, m)

(13) a. John loves Mary and Fred loves Sue.

b. love(j, m) ∧ love(f, s)

Logical Form & Predicate Logic – p. 13/30

 Relational Nouns

• Alex is Bill’s henchman.

What kind of a predicate does the noun henchman correspond to?

a. henchman(a) ∧ ??
b. henchman(a, b )

• Relational nouns: friend, enemy, mother, father, brother, sister, husband, wife,
owner, bottom, promise, blame

Logical Form & Predicate Logic – p. 14/30

 Logical Form and surface syntax

Brigitte is taller than Danny. ...is taller than... .

Alex is Bill’s henchman. ... is ...’s henchman.
Fiji is near New Zealand. ... is near ... .

TALLER (b, d )
HENCHMAN (a, b )
NEAR (f, n)

Logical Form & Predicate Logic – p. 15/30

Quantifiers

Logical Form & Predicate Logic – p. 16/30

 Preview

Our proposal thus far is on the right . . .

but

. . . but there are some serious problems treating nega-

tion.

. . . So we introduce quantifiers.

Logical Form & Predicate Logic – p. 17/30

 Extending the proposal in a natural
way

(14) a. John loves someone.

b. Love(j, x)

Logical Form & Predicate Logic – p. 18/30

 Other one-place predicates

(15) a. John drives a Buick.

b. Drive(j, x) ∧ Buick(x)

Logical Form & Predicate Logic – p. 19/30

 The problem: negation

(16) a. John doesn’t drive a Buick.

b. ¬[Drive(j, x) ∧ Buick(x)]

Does this mean the right thing?

For some unspecific x, it’s not the case both that x
is a Buick and John drives x.

Logical Form & Predicate Logic – p. 20/30

 Scope

• The meaning we’ve got for (16) is that there’s some specific Buick (say, Fred’s)
that John doesn’t drive.
• Maybe that’s a reading for (16), but it’s surely not the most natural one.
• The meaning we want: It is NOT the case that there’s a Buick that John drives.
• The problem is that at the moment we havent even got a way of writing down the
most natural reading, on which the scope of the negation claim includes the
existence claim.

Logical Form & Predicate Logic – p. 21/30

 Restating the problem with truth
conditions

(16b) gives the wrong truth-conditions for (16a)

(17) a. Suppose B11 and B12 are both Buicks. John drives B11 and John doesn’t
drive B12.
b. Then there is an x such that it’s not the case both that x is a Buick and John
drives x. Namely B12. While B12 is a Buick John doesnt drive it.
c. So the logical formula (16b) comes out true in these circumstances.
d. But the English sentence (16a) is not true in these circumstances. John
shouldn’t be driving ANY Buicks, yet he’s driving B11.
e. The logical formula (16b) misdescribes the truth conditions of (16a) .
f. This is the semantic analogue of the grammar mis-describing the
grammaticality of a sentence.

Logical Form & Predicate Logic – p. 22/30

 A Solution

(18) a. John drives a Buick.

b. ∃x[Drive(j, x) ∧ Buick(x)]
c. ∃x[Drive(j, x) ∧ Buick(x)] is true iff there is some entity b such that
[Drive(j, b) ∧ Buick(b)] is true.
d. True whenever John drives any entity that is a Buick
e. False only if there is NO entity that is a Buick that John drives
f. [[∃xφ(x)]] = true iff there is some entity b such that [[φ(x)]]b/x = true

Logical Form & Predicate Logic – p. 23/30

 Negation

(19) a. John doesn’t drive a Buick.

b. ¬∃x[Drive(j, x) ∧ Buick(x)]
c. ¬∃x[Drive(j, x) ∧ Buick(x)] is true iff it is not the case that there exists some
entity b such that [Drive(j, b) ∧ Buick(b)] is true.
d. Previously: ¬[Drive(j, x) ∧ Buick(x)] is true iff there exists some entity b
such that it is not the case that [Drive(j, b) ∧ Buick(b)] is true.

Logical Form & Predicate Logic – p. 24/30

 Revising previous analyses

Other Fixes
(20) a. A dog is happy.
b. ∃x[Dog(x) ∧ Happy(x)]
c. A dog barked.
d. ∃x[Dog(x) ∧ Bark(x)]
e. Fido is a dog.
f. ∃x[Dog(x) ∧ x = f]

Logical Form & Predicate Logic – p. 25/30

 Scope: the new analysis

• By introducing ∃x we formally marked the scope of an existence claim.

• In ∃xφ(x), we call φ(x) the scope of the existential.
• In ¬φ we call φ the scope of the negation.
• ¬∃x[drive(j, x) ∧ Buick(x)]

Operator Scope

∃x [drive(j, x) ∧ Buick(x)]
¬ ∃x[drive(j, x) ∧ Buick(x)]

• We say: the scope of the negation is wider than the scope of the existential.

Logical Form & Predicate Logic – p. 26/30

 Every and All

We use ∀x to mean “for all x”

(21) a. [[∀xφ(x)]] is true iff for every x, φ(x) is true.

b. So we need to look at a large number of cases; Each needs to turn out true.
c. How many cases? All of them. Every entity in the universe.
d. Every dog is a mammal is a claim about every dog, not every entity in the
universe.
e. How do we represent that fact?

Logical Form & Predicate Logic – p. 27/30

 Wrong semantics

(22) a. Every dog is a mammal.

b. ∀x[Dog(x) ∧ Mammal(s)]
c. This requires every entity in the universe to be a dog and a mammal.
d. Paraphrase: Everything is a dog and a mammal.
e. We make no distinction between the truth conditions of every dog is a
mammal (true) and Every mammal is a dog. (false)

Logical Form & Predicate Logic – p. 28/30

 Right semantics

(23) a. Every dog is a mammal.

∀x[dog(x) → mammal(s)]

For every x: if x a dog, then x is a mammal.

b. Every mammal is a dog.

∀x[mammal(x) → dog(s)]

For every x: if x a mammal, then x is a dog.

Logical Form & Predicate Logic – p. 29/30

 Working through the truth-conditions

p q p→q
T T T
T F F
F T T
F T T

• x not a dog dog(x) → mammal(x) T

F F
F T
x a mammal and dog dog(x) → mammal(x) T
T T
x a nonmammal and dog dog(x) → mammal(x) F
T F
• For every individual, being a dog implies being a mammal.

Logical Form & Predicate Logic – p. 30/30