26.01.2014 Views

Session 08

Session 08

Session 08

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Introduction to Semantics<br />

<strong>Session</strong> 8<br />

Quantification<br />

Cornelia Endriss<br />

Cognitive Science Program<br />

University of Osnabrück<br />

cendriss@uos.de


Outline for Today<br />

1. Homework from 2 weeks ago<br />

2. Quantificational DPs: Type and Lexical Semantics<br />

• First Try: Individuals<br />

• Second Try: Sets<br />

• Finally: Sets of Sets<br />

3. Quantificational Determiners<br />

4. Formal Properties of Quantifiers<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

2


Homework<br />

Homework<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

3


Homework<br />

1. The intensifier very combines with adjectives as the following DP<br />

illustrates:<br />

the very intelligent student<br />

Which class of adjectives can be intensified by very? Why can<br />

others not be intensified by very? Figure out the semantic type of<br />

very and try to give a lexical semantics for it. For instance, you can<br />

think of a very intelligent student as intelligent compared to the<br />

intelligent students.<br />

2. H&K p.79: Exercise 1 (escalator in South College) on the definite<br />

determiner.<br />

3. H&K p.80: Exercise 2 (apples in a row) on the definite determiner.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

4


Very<br />

Which class of adjectives can be intensified by very?<br />

The so-called gradable ones. According to the classification from<br />

H&K’s textbook the non-intersective non-intensional ones:<br />

(a) the very tall boy, the very intelligent girl.<br />

I.e. not the intersective ones:<br />

• (b) #the very red ball, #the very round circle<br />

but watch out for metalinguistic uses<br />

• Note also:<br />

(c) the very wet coat, the very full cupboard<br />

those so-called absolute standard adjectives are still gradable<br />

and hence modifiable by very<br />

I.e. not the intensional ones:<br />

(d) #the very former president<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

5


Very<br />

Figure out the semantic type of very and try to give a lexical<br />

semantics for it. For instance, you can think of a very intelligent<br />

student as intelligent compared to the intelligent students.<br />

• Type: <br />

• Idea: very is an adjective-duplicater.<br />

• Adjective doubling is common in many languages<br />

Lit:<br />

the mean mean dog =<br />

‘the very mean dog’<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

6


Very<br />

’very÷ = λR λQ . R(R(Q))<br />

’mean÷ = λPλx. P(x) & the meanness of x is above the average meanness of<br />

the elements of {y : P(y)}<br />

’very÷(’mean÷) = λQ λPλx. P(x) & the meanness of x is above the average<br />

meanness of the elements of {y : P(y)}(λPλx. P(x) & the meanness of x is<br />

above the average meanness of the elements of {y : P(y)}(Q))<br />

= λQ λPλx. P(x) & the meanness of x is above the average meanness of the<br />

elements of {y : P(y)}(λx. Q(x) & the meanness of x is above the average<br />

meanness of the elements of {y : Q(y)})<br />

= λQ λx. Q(x) & the meanness of x is above the average meanness of the<br />

elements of {y : Q(y)} & the meanness of x is above the average meanness<br />

of the elements of {y : Q(y) & the meanness of y is above the average<br />

meanness of the elements of {y : Q(y)}}<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

7


Very<br />

’very÷(’mean÷)(’dog÷) = λQ λx. Q(x) & the meanness of x is above the<br />

average meanness of the elements of {y : Q(y)} & the meanness of x is<br />

above the average meanness of the elements of {y : Q(y) & the meanness of<br />

y is above the average meanness of the elements of {y : Q(y)}} (λz. dog(z))<br />

= λx. [λz.dog(z)(x)] & the meanness of x is above the average meanness of<br />

the elements of {y : λz. dog(z)(y)} & the meanness of x is above the average<br />

meanness of the elements of {y : [λz. dog(z)(y)] & the meanness of y is<br />

above the average meanness of the elements of {y : [λz. dog(z) (y)]}}<br />

= λx. dog(x) & the meanness of x is above the average meanness of the<br />

elements of {y : dog(y)} & the meanness of x is above the average<br />

meanness of the elements of {y : dog(y) & the meanness of y is above the<br />

average meanness of the elements of {y : dog(y)}}<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

8


Very<br />

= λx. dog(x) & the meanness of x is above the average meanness of the<br />

elements of {y : dog(y) & the meanness of y is above the average meanness<br />

of the elements of {y : dog(y)}}<br />

= λx. dog(x) & the meanness of x is above the average meanness of a mean<br />

dog<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

9


Very<br />

Note that many solutions that seem easier do not work, e.g.:<br />

’very÷ c<br />

= λPλx. x’s P-property is above s, where s is the standard<br />

made salient by the context of all y which are P in c.<br />

This solution does not work, because we do not know the “Pproperty”,<br />

we simply know the extension of P.<br />

Hence all solutions that assume ’very÷ to be of type <br />

are doomed to fail.<br />

Otherwise the set of very intelligent students would necessarily<br />

have to be the same as the set of very tall students, if the sets of<br />

intelligent creatures and the set of tall creatures happen to coincide<br />

in the world and the context under discussion.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

10


Escalator: H&K<br />

What does our current theory predict?<br />

Undefined because semantics of one<br />

of the daughter nodes undefinded<br />

Undefined because semantics of one<br />

of the daughter nodes undefinded<br />

Undefined because semantics of one<br />

of the daughter nodes undefinded<br />

’the÷(’escalator in South College÷) =<br />

λP∈D and there is ex. 1 x st. P(x)=1.<br />

(the) y such that P(y) = 1 (λx.e-i-SC(x))<br />

is undefined<br />

because P is the empty<br />

set. Hence<br />

presupposition failure.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

11


Escalator: Russel<br />

What would Russel predict?<br />

i’John uses the escalator in South College÷ = 1<br />

’John uses the escalator in South College÷ =<br />

∃x.e-i-SC(x) ∧ ∀y[e-i-SC(y) → x=y] ∧ use(john,x) = 0<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

12


Escalator: intuitions<br />

What is empirically correct?<br />

Debatable. Depends on the context. There are contexts (maybe<br />

contrastive ones) where one could say something like John did<br />

not use the escalator in South College, because there is none.<br />

So here it seems that the definite determiner is not<br />

presuppositional or at least the entire sentence does not carry<br />

an existence presupposition. So here the current theory would<br />

make false predictions.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

13


Apples: intuitions<br />

b 1 b 2 b 3 b 4 b 5 b 6<br />

(i) (ii) (iii) = undefined<br />

(i) the leftmost apple in the row<br />

(ii) the leftmost dark apple in the row<br />

(iii) the apple that is both leftmost in the row and dark<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

14


(i) the leftmost apple in the row<br />

DP <br />

’the÷(’leftmost apple in the row÷) =<br />

the unique y such that y is an apple in the row<br />

and y is leftmost<br />

the <br />

NP <br />

PM( x.x is an apple in the row,<br />

x.x is leftmost) =<br />

x.x is an apple in the row and x is leftmost<br />

leftmost N' <br />

x.x is an apple in the row<br />

apple in the row<br />

Assumption: ’leftmost÷ = λx. x is leftmost<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

15


(i) the leftmost apple in the row<br />

b 1 b 2 b 3 b 4 b 5 b 6<br />

(i) (ii)<br />

(iii) = undefined<br />

̌<br />

(i) the leftmost apple in the row<br />

(ii) the leftmost dark apple in the row<br />

(iii) the apple that is both leftmost in the row and dark<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

16


(ii) the leftmost dark apple in the row<br />

the<br />

<br />

DP<br />

leftmost<br />

<br />

’the÷( x. x is leftmost and x is dark and x is an apple<br />

in the row) = the unique y such that y is an apple<br />

in the row and y is leftmost and y is dark<br />

NP<br />

PM( x.x is an apple in the row and x is dark,<br />

x.x is leftmost) = x.x is an apple in the row and<br />

x is dark and x is leftmost<br />

N'<br />

PM( x.x is an apple in the row,<br />

x.x is leftmost) =<br />

x.x is an apple in the row and x is dark<br />

dark N' <br />

<br />

x.x is an apple in the row<br />

apple in the row<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

17


(ii) the leftmost dark apple in the row<br />

b 1 b 2 b 3 b 4 b 5 b 6<br />

(i) (ii)<br />

(iii) = undefined<br />

̌<br />

<br />

(i) the leftmost apple in the row<br />

(ii) the leftmost dark apple in the row<br />

(iii) the apple that is both leftmost in the row and dark<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

18


(iii) the apple that is both leftmost in<br />

the row and dark<br />

b 1 b 2 b 3 b 4 b 5 b 6<br />

(i) (ii)<br />

(iii) = undefined<br />

̌<br />

<br />

̌<br />

(i) the leftmost apple in the row<br />

(ii) the leftmost dark apple in the row<br />

(iii) the apple that is both leftmost in the row and dark<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

19


What we derive so far<br />

• our one-place predicate treatment of “leftmost” is<br />

appropriate for (i) and (iii)<br />

• but “the leftmost dark apple in the row” (ii) has the<br />

same truth-conditions as “the apple that is both<br />

leftmost in the row and dark” (iii) and thus goes<br />

against our intuitions<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

20


Lexical Semantics of leftmost<br />

• It has to be a function of type D <br />

• ’leftmost÷ = λP c D . [λx c D e . P(x) and x is the leftmost<br />

element among the elements of {y: P(y)}]<br />

• “the leftmost dark apple in the row” denotes the leftmost apple<br />

in the row of dark apples<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

21


(ii) the leftmost dark apple in the row<br />

the unique y st. y is an apple in the row and y is dark and y is the leftmost<br />

element among the elements of {y: y is an apple in the row and y is dark}]<br />

DP P x.P(x) and x is the leftmost element<br />

among the elements of {y: P(y)}<br />

( x.x is an apple in the row and x is dark)<br />

the NP = x.x is an apple in the row and x is dark and<br />

<br />

x is the leftmost element among the elements<br />

of {y: y is an apple in the row and y is dark}<br />

leftmost<br />

<br />

N'<br />

x.x is an apple in the row and x is dark<br />

dark<br />

<br />

N' <br />

x.x is an apple in the row<br />

apple in the row<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

22


(ii) the leftmost dark apple in the row<br />

b 1 b 2 b 3 b 4 b 5 b 6<br />

(ii) (ii)<br />

(iii) = undefined<br />

̌<br />

(i) the leftmost apple in the row<br />

(ii) the leftmost dark apple in the row<br />

(iii) the apple that is both leftmost in the row and dark<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

23


Quantification<br />

Quantification<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

24


More DPs<br />

Do all DPs behave like proper names or definite descriptions?<br />

S <br />

DP <br />

VP <br />

Fred<br />

Fred<br />

walks<br />

λx ∈ D e<br />

.x walks<br />

’Fred walks ÷ = 1 iff λx ∈D e<br />

. x walks(Fred)<br />

iff Fred ∈ walk<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

25


More DPs<br />

Do all DPs behave like proper names or definite descriptions?<br />

S <br />

DP <br />

VP <br />

D <br />

NP <br />

the man walks<br />

λf∈D the unique… λx ∈ D e .x is a man λx ∈ D e .x walks<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

26


More DPs<br />

A woman<br />

Somebody<br />

Everybody<br />

Few people<br />

Nobody<br />

At most two people<br />

– an arbitrary individual?<br />

– every individual? A set?<br />

– an arbitrary small set?<br />

– no individual? The empty set?<br />

– an arbitrary set of<br />

at most two people?<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

28


Quantification<br />

Quantificational DPs<br />

as Individuals<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

29


Quantifiers as Individuals?<br />

John lives in Osnabrück<br />

u John lives in Germany<br />

(i)<br />

’ John ÷ ∈ D e<br />

(ii) ’ lives in Osnabrück ÷ ⊆ ’ lives in Germany ÷<br />

(iii) F char<br />

(x) = 1 iff x ∈ F<br />

Nobody lives in Osnabrück<br />

^ Nobody lives in Germany<br />

At most two people live in OS<br />

^ At most two people live in Germany<br />

Few people live in Osnabrück<br />

^ Few people live in Germany<br />

Intuitively invalid, but could be derived if the DPs were of type e!<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

30


Quantifiers as Individuals?<br />

John is in this room<br />

u It’s not the case: John is outside this room<br />

(i)<br />

(ii)<br />

’ John ÷ ∈ D e<br />

’ is in this room ÷ ∩ ’ is outside this room ÷ = ∅<br />

(iii) F(x) = 1 iff x ∈ F<br />

A woman is in this room<br />

^ It’s not the case: a woman is outside this room<br />

Somebody is in this room<br />

^ It’s not the case: somebody is outside this room<br />

Three people are in this room<br />

^ It’s not the case: three people are outside this room<br />

Intuitively invalid, but could be derived if the DPs were of type e!<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

31


Quantifiers as Individuals?<br />

Tautological (= always true) statement:<br />

(i)<br />

(ii)<br />

John is over 30 years old or John is under 40 years old<br />

’ John ÷ ∈ D e<br />

’ be over 30 years old ÷ ∪ ’ be under 40 years old ÷ = D e<br />

(iii) F(x) = 1 iff x ∈ F<br />

Everybody is over 30 years old or everybody is under 40 years old<br />

At most three people are over 30 years old or at most three people are under 40<br />

years old<br />

Intuitively non-tautological, but would be derived as tautology if the<br />

DPs were of type e!<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

32


Quantification<br />

Quantificational DPs<br />

as Sets<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

33


Quantifiers as Sets?<br />

The problems indicate: DPs do (in general) not denote an individual of<br />

type e. So maybe they denote sets of individuals (i.e. they are of type<br />

)?<br />

’ Fred ÷ = { Fred }<br />

’ everything ÷ = D e<br />

’ nothing ÷ = ∅<br />

’ at most 2 people ÷ = an arbitrary set of at most 2 people<br />

Compositional rules (or the semantics of verbs) has to be changed<br />

accordingly. Assume:<br />

’S ÷ = ’ DP VP ÷ = 1<br />

iff ’ DP ÷ ⊆ ’ VP ÷<br />

Alternatively, change the semantics of the<br />

VP:<br />

’ lives in Osnabrück ÷ =<br />

λX ∈ Pow(D). X ⊆ {y: y lives in Osnabrück}<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

34


Does this proposal yield adequate results for the following<br />

sentences?<br />

S<br />

Quantifiers as Sets?<br />

DP<br />

N<br />

Fred<br />

VP<br />

V<br />

lives in Osnabrück<br />

S<br />

{ Fred } ⊆ {y: y lives in Osnabrück }<br />

iff Fred lives in Osnabrück <br />

DP<br />

Everybody<br />

VP<br />

V<br />

D e ⊆ {y: y lives in Osnabrück }<br />

iff every individual lives in Osnabrück <br />

lives in Osnabrück<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

35


Quantifiers as Sets?<br />

Does this proposal yield adequate results for the following<br />

sentences?<br />

DP<br />

Nobody<br />

S<br />

VP<br />

V<br />

∅ ⊆ {y: y lives in Osnabrück }<br />

always true! <br />

lives in Osnabrück<br />

DP<br />

At most 2 people<br />

S<br />

VP<br />

V<br />

an arbitrary set of at most 2 people<br />

⊆ {y: y lives in Osnabrück }<br />

also always true! <br />

live in Osnabrück<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

36


Quantifiers as Sets?<br />

What about unintuitive inferences?<br />

Nobody lives in Osnabrück ^ Nobody lives in Germany<br />

Still predicted!<br />

∅ ⊆ {y: y lives in Osnabrück } ⊆ {y: y lives in Germany }<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

37


Quantification<br />

Quantificational DPs<br />

as Sets of Sets<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

38


Solution: Change Type of DPs<br />

So far:<br />

VP (DP )<br />

S<br />

<br />

DP<br />

<br />

VP<br />

<br />

… how about:<br />

DP (VP )<br />

S<br />

<br />

DP<br />

<br />

VP<br />

<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

39


What does it mean to change the type of DPs<br />

from to ?<br />

What sort of things are of the type ?<br />

functions are properties of properties,<br />

i.e. 2nd order properties<br />

For instance:<br />

At least 2 people walk d<br />

Everybody walks d<br />

Properties of Properties<br />

the property of walking is a property<br />

that at least 2 people have.<br />

the property of walking is a property<br />

that everybody has.<br />

The walk property is in the set of properties that 2 people<br />

have/everybody has.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

40


Functional View vs. Set View<br />

A Generalized Quantifier (GQ) is a function from functions to truth<br />

values:<br />

’ nothing ÷ = λf ∈ D <br />

. there is no x∈D e<br />

such that f(x) = 1<br />

’ something ÷ = λf ∈ D <br />

. there is some x∈D e<br />

such that f(x) = 1<br />

’ everything ÷ = λf ∈ D <br />

. for all x∈D e<br />

: f(x) = 1<br />

A different view of the same thing: A GQ is a set of sets<br />

’ nothing ÷ = { X ∈ Pow(D) : X = ∅ }<br />

’ something ÷ = { X ∈ Pow(D) : X ≠ ∅ }<br />

’everything ÷ = { X ∈ Pow(D) : X = D }<br />

Definitions for truth and falsity have to be amended accordingly:<br />

’ S ÷ = ’ DP VP ÷ = 1 iff ’ VP ÷ ∈ ’ DP ÷<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

41


Quantifiers as Sets of Sets<br />

GQs as sets of sets:<br />

’ no woman ÷ = { X ∈ Pow(D) : woman ∩ X = ∅ }<br />

’ nobody ÷ = { X ∈ Pow(D) : people ∩ X = ∅ }<br />

’ a man ÷ = { X ∈ Pow(D) : man ∩ X ≠ ∅ }<br />

’ somebody ÷ = { X ∈ Pow(D) : people ∩ X ≠ ∅ }<br />

’ every dog ÷ = { X ∈ Pow(D) : dog ⊆ X }<br />

’ everybody ÷ = { X ∈ Pow(D) : people ⊆ X }<br />

’ at least 2 people÷ = { X ∈ Pow(D) : |X ∩ people| ≥ 2 }<br />

’ at most 2 people÷ = { X ∈ Pow(D) : |X ∩ people| ≤ 2 }<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

42


Problematic Cases Solved<br />

At most 2 people live in Osnabrück.<br />

’ live in Osnabrück ÷<br />

’ at most 2 people ÷<br />

= { X ∈ Pow(D) : |X ∩ people| ≤ 2 }<br />

Only true iff at most 2 people<br />

and no more live in Osnabrück. <br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

43


Problematic Cases Solved<br />

Nobody lives in Osnabrück.<br />

’ live in Osnabrück ÷<br />

’ Nobody ÷<br />

= { X ∈ Pow(D) : X ∩ people = ∅ }<br />

Not tautological, but only true iff<br />

nobody lives in Osnabrück. <br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

44


Solving the Inference Problem<br />

At most 2 people live in Osnabrück ^<br />

At most 2 people live in Germany<br />

’ live in Osnabrück ÷<br />

’ at most 2 people ÷<br />

’ live in Germany ÷<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

45


Solving the Inference Problem<br />

Somebody is in this room<br />

^ It is not the case: somebody is outside this room<br />

’ be in this room ÷<br />

’ somebody ÷<br />

’ be outside this room ÷<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

46


Quantification<br />

Quantificational<br />

Determiners<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

47


Quantificational Determiners<br />

Given that quantifying DPs have lexical entries like these:<br />

’ nothing÷ = λf ∈ D . there is no x∈D e such that f(x) = 1<br />

’ no woman ÷ = λf ∈ D . there is no x∈D e such that x is a woman<br />

& f(x) = 1<br />

What are the lexical entries for determiners like no, a, every?<br />

S <br />

DP <br />

VP <br />

D <br />

no<br />

NP <br />

man sleeps<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

48


Quantificational Determiners:<br />

Functional View<br />

’ no ÷ = λf ∈ D . [λg ∈ D . . there is no x ∈ D e such that<br />

f(x) = 1 & g(x) = 1]<br />

’ a ÷ = λf ∈ D . [λg ∈ D . there is some x ∈ D e such that<br />

f(x) = 1 & g(x) = 1]<br />

’ every ÷ = λf ∈ D . [λg ∈ D . for all x ∈ D e such that<br />

f(x) = 1, g(x) = 1]<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

49


Derivation of Truth Conditions<br />

Computing the truth conditions for A man left<br />

S <br />

there is some x ∈ D e such that x is a man and x left<br />

DP ><br />

λg ∈ D . there is some x ∈ D e such that x is a man & g(x) = 1<br />

D <br />

a<br />

λf ∈ D . [λg ∈ D .<br />

there is some x ∈ D e<br />

such that f(x) = 1 & g(x) = 1]<br />

NP <br />

man<br />

Py ∈ D e . y is a man<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

VP <br />

left<br />

Pz ∈ D e . z left<br />

50


Quantif. Ds: Relational View<br />

Recall the ‘set of sets’ view on generalized quantifiers, e.g.<br />

’ something ÷ = { X ∈ Pow(D) : X ≠ ∅ }<br />

The same works for GQs corresponding to DPs in general:<br />

’ no woman ÷ = { X ∈ Pow(D) : woman ∩ X = ∅ }<br />

’ a man ÷ = { X ∈ Pow(D) : man ∩ X ≠ ∅ }<br />

’ every dog ÷ = { X ∈ Pow(D) : dog ⊆ X }<br />

Under this view, determiners denote relations between sets:<br />

’ no ÷ = { ∈ Pow(D) % Pow(D): A ∩ B = ∅ }<br />

’ a ÷ = { ∈ Pow(D) % Pow(D): A ∩ B ≠ ∅ }<br />

’ every ÷ = { ∈ Pow(D) % Pow(D): A ⊆ B }<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

51


Functional and Relational View<br />

Truth conditions under the relational and the functional view:<br />

’ a man left ÷ = 1 b relational view<br />

iff ∈ ’ a ÷<br />

iff ∈ { ∈ Pow(D) % Pow(D): A ∩ B ≠ ∅ }<br />

iff man ∩ left ≠ ∅<br />

iff there is some x ∈ D e<br />

such that x ∈ man and x ∈ left<br />

iff there is some x ∈ D e<br />

such that x is a man and x left<br />

iff λg ∈ D <br />

[there is some x ∈ D e<br />

such that<br />

x is a man & g(x) = 1](left)<br />

iff λf ∈ D <br />

[λg ∈ D <br />

[there is some x ∈ D e<br />

such that<br />

f(x) = 1 & g(x) = 1](man)(left)<br />

iff (’ a ÷(’ man ÷))(’ left ÷) = 1<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

b functional view<br />

52


Notations<br />

• People usually refer to a quantificational DP such as every<br />

woman/someone/most horses (of type ) as<br />

Generalized Quantifier (GQ) or simply as quantifier.<br />

• The quantificational determiner (of type<br />

) is sometimes also called a quantifier.<br />

• We say that in<br />

Every horse sleeps<br />

horse is the restrictor of the quantificational determiner<br />

every and sleeps is the nucleus or (nuclear) scope.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

53


Quantification<br />

Formal Properties of<br />

Quantifiers<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

54


A determiner D can be classified according to certain formal properties of its<br />

meaning under the relational view ’ D ÷.<br />

Symmetry<br />

A determiner D is called symmetric<br />

iff for all sets A and B: if ∈ ’ D ÷ then ∈ ’ D ÷<br />

For instance: is a symmetric?<br />

Informal check via natural language:<br />

Symmetry<br />

a German won an academy award<br />

an academy award winner was German<br />

and hence<br />

for particular A (being German) and B (being AA winner)<br />

Formal proof (for general A and B):<br />

suppose A and B are sets such that ∈ ’ a ÷<br />

hence A ∩ B ≠ ∅ and likewise B ∩ A ≠ ∅<br />

therefore ∈ ’ a ÷<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

55


Symmetry<br />

Another example: Is every symmetric?<br />

Informal check via natural language:<br />

every German nominee won an AA<br />

every AA winner was a German nominee<br />

for particular A (being German nominee) and B (being AA winner)<br />

From the first sentence one cannot conclude to the second<br />

Formal proof (for general A and B):<br />

suppose A and B are sets such that ∈ ’ every ÷<br />

hence A ⊆ B,<br />

but then B ⊆ A only if A = B,<br />

therefore v ’ every ÷ unless A = B<br />

Conclusion: a is symmetric, but every is not.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

56


Monotonicity<br />

Monotonicity<br />

Different DPs allow for different inference patterns.<br />

(a) A bachelor smoked cigars u A man smoked cigars<br />

(b) A bachelor smoked cigars u A bachelor smoked<br />

(c) A bachelor smoked cigars ^ A young bachelor smoked cigars<br />

(d) A bachelor smoked cigars ^ A bachelor smoked Cuban cigars<br />

(a) Every bachelor smoked cigars ^ Every man smoked cigars<br />

(b) Every bachelor smoked cigars u Every bachelor smoked<br />

(c) Every bachelor smoked cigars u Every young bachelor smoked cigars<br />

(d) Every bachelor smoked cigars ^ Every bachelor smoked Cuban cigars<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

57


Monotonicity<br />

Monotonicity<br />

(a) No bachelor smoked cigars ^ No man smoked cigars<br />

(b) No bachelor smoked cigars ^ No bachelor smoked<br />

(c) No bachelor smoked cigars u No young bachelor smoked cigars<br />

(d) No bachelor smoked cigars u No bachelor smoked Cuban cigars<br />

Important Observation:<br />

In the (a) cases, the first argument of the determiner was enlarged:<br />

bachelor ⊆ man<br />

In the (b) cases, the second argument of the determiner was enlarged:<br />

smoke cigars ⊆ smoke<br />

In the (c) cases, the first argument of the determiner was reduced:<br />

bachelor r young bachelor<br />

In the (d) cases, the second argument of the determiner was reduced:<br />

smoke cigars r smoke Cuban cigars<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

58


Monotonicity<br />

Monotonicity<br />

Determiners are classified according to the inference patterns they<br />

allow.<br />

Let A, B, C be sets such that A ⊆ B . Then a determiner D is called<br />

left upward monotone<br />

if ∈ ’ D ÷ then ∈ ’ D ÷<br />

left downward monotone<br />

if ∈ ’ D ÷ then ∈ ’ D ÷<br />

right upward monotone<br />

if ∈ ’ D ÷ then ∈ ’ D ÷<br />

right downward monotone<br />

if ∈ ’ D ÷ then ∈ ’ D ÷<br />

(a) cases<br />

(b) cases<br />

(c) cases<br />

(d) cases<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

59


Monotonicity<br />

Monotonicity<br />

The examples above illustrate:<br />

a is left upward and right upward monotone<br />

every is left downward and right upward monotone<br />

no is left downward and right downward monotone<br />

Determiners that are neither upward nor downward monotone<br />

in one argument are called non-monotone in that argument.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

60


Homework<br />

To be handed in by Saturday next week (June 14th).<br />

– Read Chapters 6 and 7 of Heim & Kratzer’s textbook and go through the<br />

lecture notes again.<br />

– H&K p. 198: Exercise (calculating truth-conditions of (7’)).<br />

– It is shown in the textbook (pp.159-160) that the Aristotelian<br />

understanding of the determiners all, some, no differs from the "modern"<br />

understanding. Peter Strawson (cf. p. 161 of the textbook) proposed a<br />

view of the semantics of these determiners that is intended to save the<br />

Aristotelian interpretation.<br />

(a)<br />

(b)<br />

(c)<br />

(d)<br />

Describe the Aristotelian and the modern semantics of the<br />

determiners all, some, no (using material from the textbook).<br />

Show where precisely there is a conflict.<br />

Describe Strawson's idea for solving the conflict.<br />

Give your own assessment of Strawson's position and discuss any<br />

consequences you find interesting.<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

61


Thank you!<br />

Thank you!<br />

03.06.20<strong>08</strong> Slides based on semantics textbook from I. Heim & A. Kratzer<br />

and lecture notes from M. Krifka<br />

62

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!