Presupposition PRESUPPOSITION - uni-konstanz.de

Presupposition PRESUPPOSITION
in DRT
2
Slides by Patrick Blackburn & Johan Bos
(malaga4.ppt) and slightly modified by Maribel

Romero (MR)

Ling391:
Advanced Computational Semantics

1

Presupposition What is presupposition?

!Presupposition vs. Entailment !It is hard to pin down precisely what
!Look at some examples of presuppositions are or how they
behave
presupposition
!Look at the typical problems associated !Presuppositions are a bit like
entailment but not quite…
with presuppositions
!Concentrate on a DRT based approach 4

due to Rob van der Sandt

3

Entailment Entailment

!Consider: !Consider:
Vincent has a car. Vincent has a red car.
A car is a vehicle.
!This entails:
!This entails: Vincent has a car.
Vincent has a vehicle.

5 6

Entailment and negation Entailment

!Entailments are typically not preserved !Consider:
under negation. Vincent has no car.
A car is a vehicle.
7
!This does not entail:
Vincent has a vehicle.

8

Entailment Presupposition

!Consider: !Consider:
Vincent does not have a red car. Vincent cleaned his car.

!This does not entail: !This entails:
Vincent has a car. Vincent has a car.

9 10

Presupposition Entailment or presupposition

!Consider: !We call implications preserved under
Vincent did not clean his car. negation presuppositions

!This entails: !We call implications not preserved
Vincent has a car. under negation entailments

11 12

Presupposition triggers Presupposition trigger test

!In English, presuppositions are usually !Consider the sentence:
triggered by lexical items Alex is a bachelor.

!There are several tricks to find out !This sentence implies that Alex is male.
whether a lexical item is a !But are we dealing with a
presupposition trigger or not
presupposition or entailment?
!These tests are:
14
" The negation test

" The conditional test

" The question test 13

Presupposition test Presupposition test

! Alex is a bachelor. ! Alex is a bachelor.
Does this presuppose: Alex is male? Does this presuppose: Alex is male?

15 ! Negation: Alex is not a bachelor.
Implies: Alex is male? YES

16

Presupposition test Presupposition test

! Alex is a bachelor. ! Alex is a bachelor.
Does this presuppose: Alex is male? Does this presuppose: Alex is male?

! Negation: Alex is not a bachelor. ! Negation: Alex is not a bachelor.
Implies: Alex is male? YES Implies: Alex is male? YES

! Conditional: If Alex is a bachelor, then ... ! Conditional: If Alex is a bachelor, then ...
Implies: Alex is male? YES Implies: Alex is male? YES

17 ! Question: Is Alex is a bachelor?
Implies: Alex is male? YES

18

Presupposition test Presupposition trigger test

! Alex is a bachelor. 19 !Consider the sentence:
Does this presuppose: Alex is male? Alex is a man.

! Negation: Alex is not a bachelor. !This sentence implies that Alex is male.
Implies: Alex is male? YES !But are we dealing with a

! Conditional: If Alex is a bachelor, then ... presupposition or entailment?
Implies: Alex is male? YES
20
! Question: Is Alex is a bachelor?
Implies: Alex is male? YES

! Conclusion:
being a bachelor presupposes being male.

Presupposition test Presupposition test

! Alex is a man. ! Alex is a man.
Does this presuppose: Alex is male? Does this presuppose: Alex is male?

21 ! Negation: Alex is not a man.
Implies: Alex is male? NO

22

Presupposition test Presupposition test

! Alex is a man. ! Alex is a man.
Does this presuppose: Alex is male? Does this presuppose: Alex is male?

! Negation: Alex is not a man. ! Negation: Alex is not a man.
Implies: Alex is male? NO Implies: Alex is male? NO

! Conditional: If Alex is a man, then ... ! Conditional: If Alex is a man, then ...
Implies: Alex is male? NO Implies: Alex is male? NO

23 ! Question: Is Alex is a man?
Implies: Alex is male? NO

24

Presupposition test Presupposition trigger test

! Alex is a man. !Consider the sentence:
Does this presuppose: Alex is male? Butch knows that Zed is dead.

! Negation: Alex is not a man. !This sentence implies Zed is dead.
Implies: Alex is male? NO !But are we dealing with a

! Conditional: If Alex is a man, then ... presupposition or entailment?
Implies: Alex is male? NO
26
! Question: Is Alex is a man?
Implies: Alex is male? NO

! Conclusion:
being a man does not presuppose being male. 25

Presupposition test Presupposition test

! Butch knows that Zed is dead. ! Butch knows that Zed is dead.
Does this presuppose: Zed is dead? Does this presuppose: Zed is dead?

27 ! Negation: Butch does not know that Zed is dead.
Implies: Zed is dead? YES

28

Presupposition test Presupposition test

! Butch knows that Zed is dead. ! Butch knows that Zed is dead.
Does this presuppose: Zed is dead? Does this presuppose: Zed is dead?

! Negation: Butch does not know that Zed is dead. ! Negation: Butch does not know that Zed is dead.
Implies: Zed is dead? YES Implies: Zed is dead? YES

! Conditional: If Butch knows that Zed is dead, then ... ! Conditional: If Butch knows that Zed is dead, then ...
Implies: Zed is dead? YES Implies: Zed is dead? YES

29 ! Question: Does Butch know that Zed is dead?
Implies: Zed is dead? YES

30

Presupposition test Presupposition triggers

! Butch knows that Zed is dead. !Presupposition triggers are not rare
Does this presuppose: Zed is dead? !English comes with a large variety of

! Negation: Butch does not know that Zed is dead. presupposition triggers
Implies: Zed is dead? YES
32
! Conditional: If Butch knows that Zed is dead, then ...
Implies: Zed is dead? YES

! Question: Does Butch know that Zed is dead?
Implies: Zed is dead? YES

! Conclusion: 31
knowing P presupposes P.

Possessives To regret

!Example: !Example:
Mia likes her husband.
Mia does not like her husband. Vincent regrets that he left Mia alone.
Vincent does not regret that he left Mia alone.
!Presupposition:
Mia has a husband. !Presupposition:

33 Vincent left Mia alone.

34

To like To answer

!Example: !Example:
Mia likes Vincent.
Mia does not like Vincent. Butch answered the phone.
Butch did not answer the phone.
!Presupposition:
Mia knows Vincent. !Presupposition:
The phone was ringing.

35 36

Only Again

!Example: !Example:
Butch escaped again.
Only Jules likes big kahuna burgers. Butch did not escape again.
Not only Jules likes big kahuna burgers.
!Presupposition:
!Presupposition: Butch escaped once before.
Jules likes big kahuna burgers.

37 38

To manage Third

!Example: !Example:

Butch manage to start the chopper. Butch lost for the third time.
Butch did not manage to start the chopper. Butch did not loose for the third time.

!Presupposition: !Presupposition:
Butch lost twice before.
Butch had difficulties starting the chopper.
40
39

Continue To win

!Example: !Example:
Germany won the world cup.
Butch continued his race. Germany did not win the world cup.
Butch did not continue his race.
!Presupposition:
!Presupposition: Germany participated in the world cup.
Butch interrupted his race.
42
41

Another To lie

!Example: !Example:
Butch lied to Marsellus.
Peter wants another beer. Butch did not lie to Marsellus.
Peter does not want another beer.
!Presupposition:
!Presupposition: Butch told something to Marsellus.
Peter had at least one beer.
44
43

Cleft construction Proper names

!Example: !Example:
Butch talked to Marsellus.
It was Butch who killed Vincent. Butch did not talk to Marsellus.
It was not Butch who killed Vincent.
!Presupposition:
!Presupposition: There is someone named Marsellus.
Someone killed Vincent.
46
45

Definite NP Dealing with Presupposition

!Example: !OK, so presuppositions are fairly
Butch talked to the boss. common. But what`s the big deal?
Butch did not talk to the boss.
!Problems related to presupposition:
!Presupposition:
There is a boss. " The Binding Problem

" The Denial Problem

" The Projection Problem

!Presupposition may convey new
information

47 " Accommodation 48

The Binding Problem The Binding Problem

!Example: !Example:
Butch nearly escaped from his
apartment. A boxer nearly escaped from his
apartment.
!Trigger “his apartment” presupposes
that Butch has an apartment. !Trigger “his apartment” presupposes
that a boxer has an apartment.
49
!But which boxer? A boxer? Any boxer?

50

The Denial Problem The Denial Problem

!Vincent does not like his wife. !Vincent does not like his wife.
!Vincent does not like his wife,
51
because Vincent does not have a wife!

52

The Denial Problem The Denial Problem

!Vincent does not regret killing Zed, !Vincent does not regret killing Zed,
because he did not kill Zed! because he did not kill Zed!

53 !Alex is not a bachelor,
because she is a woman!

54

The Denial Problem The Projection Problem

!Vincent does not regret killing Zed, !Consider:
because he did not kill Zed!
Mia’s husband is out of town.
!Alex is not a bachelor,
because she is a woman! !Presupposes that Mia is married.

!Butch did not lie to Marsellus, 56
because he did not tell him anything!

55

The Projection Problem The Projection Problem

!Consider: !Consider:

If Mia has a husband, then If Mia is married, then
Mia’s husband is out of town. Mia’s husband is out of town.

!Does NOT presuppose that Mia is !Does NOT presuppose that Mia is
married. married.

57 58

The Projection Problem The Projection ProblemMR

!Consider: !Consider:

If Mia dates Vincent, then John’s donkey is eating quietly in the stable.
Mia’s husband is out of town.
!Presupposes that John has a donkey.
!Does presuppose that Mia is married.
60
59

The Projection ProblemMR The Projection ProblemMR

!Consider: !Consider:

Either John has no donkey or John’s Either John is not a donkey-owner
donkey is eating quietly in the stable. or John’s donkey is eating quietly in the stable.

!Does NOT presuppose that John has a !Does NOT presuppose that John has a
donkey. donkey.

61 62

The Projection ProblemMR The Projection Problem

!Consider: !Complex sentences sometimes
neutralise presuppositions
Either John is out of hay or John’s donkey
is eating quietly in the stable. !`Complex` meaning here sentences
with conditionals, negation, or
!Does presuppose that John has a disjunction, modals
donkey.
!These sentences make it difficult to
63 predict whether a presupposition
projects or not

64

Accommodation Solutions

!Example: !There is a rich literature on
Vincent informed his boss. presupposition

!Presupposition: Vincent has a boss. !There are many different attempts to
!What if we don’t have a clue whether solve the problems related to
presupposition
Vincent has a boss or not?
!Accommodation: incorporating missed " Many-valued logics

information as long as this is not " Default logics
conflicting with other information 65
" Pragmatic theories

" Non-monotonic reasoning

66

Van der Sandt’s Theory Van der Sandt’s Theory

!Presuppositions are essentially !Presuppositions introduce new DRSs
extremely rich anaphoric pronouns that need to be incorporated in the
discourse context
!Presuppositions introduce new DRSs
that need to be incorporated in the !There are two ways to resolve
discourse context presuppositional DRSs:

!It is a good way of dealing with the " By binding
binding, projection, and denial " By accommodation
problems
68
67

Two birds with one stone Two birds with one stoneMR

!The presupposition as anaphora theory !Idea: In the same way that we find antecedents
handles anaphoric pronouns and to bind pronouns and anaphora (1), we find
presuppositions in essentially the same antecedents to “bind” presuppositions (2):
way
(1) If a farmer owns a donkey, he beats it.
Presupposition = Anaphora
(2) If Mia has a husband, then Mia’s husband is
Anaphora = Presupposition out of town.

69 !Note that the antecedents of anaphora and
presupposition need not be individuals, but can
be VP-properties, propositions, etc.

(3) Sue likes movies, and so does Joan.

(4) Ana stopped smoking. 70

One mechanism Presuppositions in DRT

!Essentially one mechanism to deal with !We need to carry out two tasks:
pronouns, proper names, definite " Select presupposition triggers in the lexicon
descriptions, etc. " Indicate what they presuppose

!The differences are accounted for in the way !We will use a new operator,
they can accommodate and bind the alpha-operator, !

" Pronouns do not accommodate !If B1 and B2 are DRSs, the so is B1!B2
" Proper names always accommodate globally !B1 is the presupposition of B2
" Definite descriptions can accommodate anywhere
72
71

Preliminary DRSs Presupposition in the lexicon

! She dances x ! She x
female(x)
! !p. female(x) ! p@x
dance(x)

! Mia dances x x
mia(x)
! dance(x) ! Mia !p. mia(x) ! p@x

! The woman dances ! ! The woman x ! p@x
x dance(x)
woman(x) !p. woman(x)

73 74

Indefinite vs. Definite NP The algorithm

!A woman x !After constructing a preliminary DRS
for an input sentences, we still have to
!p. woman(x) ; p@x resolve the presuppositions

!The woman x ! p@x !After resolution we will have an
ordinary DRS that we can use for our
!p. woman(x) inference tasks

75 !Resulting DRS needs to be consistent
and informative

76

Binding Presuppositions Binding Presuppositions

! Example: ! Example:
Vincent danced with a woman. Vincent danced with a woman.
The woman collapsed.

xye xye z

vincent(x) vincent(x) ( ! )woman(z)
dance(e) dance(e) collapse(z)
agent(e,x) agent(e,x)
with(e,y) with(e,y)
woman(y) woman(y)

77 78

Binding Presuppositions Binding Presuppositions

! Example: ! Example:
Vincent danced with a woman. Vincent danced with a woman.
The woman collapsed. The woman collapsed.

xye z xye z

( vincent(x) ; ( ! ))woman(z) ( vincent(x) ; ( woman(z) ! ))collapse(z)
dance(e) collapse(z) dance(e)
agent(e,x) agent(e,x) z=y
with(e,y) with(e,y)
woman(y) merge woman(y) pick antecedent
79 80

Binding Presuppositions Binding Presuppositions

! Example: ! Example:
Vincent danced with a woman. Vincent danced with a woman.
The woman collapsed. The woman collapsed.

xyez ; )collapse(z) xyez

vincent(x) move vincent(x) merge reduction
81 dance(e) 82
( dance(e) agent(e,x)
agent(e,x) with(e,y)
with(e,y) woman(y)
woman(y) woman(z)
woman(z) z=y
z=y collapse(z)

Accommodating Presuppositions Global accommodation

! Example: ! Example:
If Mia dates Vincent, then her husband is out If Mia dates Vincent, then her husband is out
of town of town

xy z xy z
mia(x) mia(x)
vincent(y) "( husband(z) ! out(z) ) vincent(y) "( husband(z) ! out(z) )
of(z,x) of(z,x)
date(x,y) date(x,y)

83 84

Global Accommodation Non-global accommodation

! Example: !Performing global accommodation is
If Mia dates Vincent, then her husband is out saying that something is presupposed.
of town
!But recall the projection problem.
xyz
!Presuppositions can be neutralised by
mia(x) " out(z) binding and non-global accommodation.
vincent(y)
husband(z) 85 86
of(z,x)

date(x,y)

Non-global Accommodation Non-global Accommodation

! Example: ! Example:
If Mia is married, then her husband is out of If Mia is married, then her husband is out of
town town
x x
mia(x) mia(x)

married(x) z out(z) ) married(x) z out(z) )

"( husband(z) ! 87 "( husband(z) ! 88

of(z,x) of(z,x)

Non-global Accommodation Preferences

! Example: !Binding is preferred to accommodation
If Mia is married, then her husband is out of !Global accommodation is preferred to
town
x local accommodation

mia(x) " out(z) 90
z
89
married(x)
husband(z)
of(z,x)

Van der Sandt’s Algorithm Accessibility and SubordinationMR

1. Generate a DRS for the input sentence, with all ! A DRS B1 is accessible from DRS B2 when B1 equals
elementary presuppositions marked by ! B2, or
when B1 subordinates B2
2. Merge this DRS with the DRS of the discourse so far
processed ! A DRS B1 subordinates B2 iff:

3. Traverse the DRS, and on encountering an !-DRS " B1 immediately subordinates B2
try to:
" There is a DRS B such that B1 subordinates B
1. Link (MR) or bind the presupposed information to an and B subordinates B2
accessible antecedent, or
! B1 immediately subordinates B2 iff:
2. Accommodate the information to a superordinated level of
DRS " B1 contains a condition ¬B2

4. Remove those DRSs from the set of potential " B1 contains a condition B2#B or B#B2
readings that violate the acceptability constraints
" B1 contains a condition B2 " B
91
" B1 " B2 is a condition in some DRS B 92

The acceptability constraints Free Variable Check

!DRSs should obey the binding rules !Consider the example:
!DRSs should not contain free variables Every man likes his car
!DRSs should be consistent and
!DRS obtained with global accommodation:
informative
!DRSs should also be locally consistent y

and locally informative car(y)
of(y,x)
That is: the resolved DRS should not contain
a subordinate DRS K whose falsity or truth is x " like(x,y)
entailed by a DRS superordinate to it. (MR, man(x)
from v.d.Sandt p. 367)
94
93

Free Variable Check Free Variable Check

!Consider the example: !Consider the example:
Every man likes his car Every man likes his car

!DRS obtained with global accommodation: !DRS obtained via intermediate accommodation:

y xy

car(y) man(x)
of(y,x) car(y)
of(y,x)
x " like(x,y) " like(x,y)
man(x)

95 96

Free Variable Check The projection problem solved

!Consider the example: !Recall our example:
Every man likes his car If Mia is married, then her husband is out of
town
!DRS obtained with local accommodation:
!Local constraints play a crucial role here!

x y xz
man(x)
" car(y) mia(x)
of(y,x) husband(z) of(z,x)

like(x,y) married(x) " out-of-town(z)

97 98

The projection problem solved The projection problem solved

!Recall our example: !Recall our example:
If Mia is married, then her husband is out of If Mia is married, then her husband is out of
town town

!Local constraints play a crucial role here! !Local constraints play a crucial role here!

Locally xz Locally x
uninformative informative mia(x)
mia(x)
husband(z) of(z,x) z

married(x) " out-of-town(z) married(x) " out-of-town(z)
husband(z)
99 of(z,x) 100

The projection problem solvedMR Denial

! Question: !Example:
Recall our previous examples: Vincent does not like his dog.
He does not have a dog!
(1) Either John is not a donkey-owner or his donkey is eating quietly
in stable. x
vincent(x)
(2) If Mia has a husband, then her husband is out of town.
(3) Either John does not have a donkey or his donkey is eating y

quietly in the stable. ¬ dog(y)
(4) If Mia dates Vincent, then her husband is out of town. of(y,x)
(5) Either John has run out of hay or his donkey is eating quietly in like(x,y)

the stable.

For each example, show how the acceptability constraints plus
the preference binding > global accomm. > local accomm.
determine the projection possibilities of the presuppositions at
issue. 101 102

The binding problem solved The binding problem solved

! Example: ! Example:
A boxer nearly escaped from his apartment. A boxer nearly escaped from his apartment.

!Preliminary DRS: !Preliminary DRS:

xz ! nearly-escaped-from(x,z) )) xz ! nearly-escaped-from(x,z) ))

( boxer(x) ;( apartment(z) ( boxer(x) ;( apartment(z)

of(z,x) of(z,x)

• Final DRS: xz

103 boxer(x) 104
apartment(z) of (z,x)
nearly-escaped-from(x,z)

Proper Names Proper Names

!Proper Names can be treated as !Example:
presupposition triggers Every man knows Mia.
She is Marsellus’ wife.
!Only global accommodation is
permitted for proper names "x y !
mia(y) know(x,y)
!This assures they will always end up in man(x)
the global (outermost) DRS, accessible
for subsequent pronouns 106

105

Proper Names Implementation

!Example: !The Curt system
Every man knows Mia. !Small fragment of English
She is Marsellus’ wife.
" Pronouns, presupposition triggers
y know(x,y)
mia (y) !Uses theorem prover

"x " Bliksem

man(x) !Uses model builder

" Mace

!Does all inference tasks

107 108