Compositional Semantics. Jacob Andreas

Transcription

1 Compositional Semantics Jacob Andreas

2 Problem 1 Each of the three girls has a platypus. Each of the three girls climbed the mountain. How many platypuses? How many mountains? 2

3 Problem 1 Each of the three girls has a platypus. 3

4 Problem 1 Each of the three girls climbed the mountain. 4

5 Problem 2 name type coastal There are 128 cities Columbia city no in South Carolina. Cooper river yes Charleston city yes 5

6 Problem 3 Barack Obama was the 44th President of the United States. Obama was born on August 4 in Honolulu, Hawaii. In late August 1961, Obama's mother moved with him to the University of Washington in Seattle for a year Is Barack Obama from the United States? 6

7 Compositional semantics It s not enough to have structured representations of syntax: We also need structured representations of meaning. 7

8 Compositional semantics It s not enough to have structured representations of syntax: We also need structured representations of meaning. Today: How do we get from language to meaning? 8

9 PART I What is meaning?

10 Meaning in formal languages a + b = 17 10

11 Meaning in formal languages a + b = 17 11

12 Meaning in formal languages a + b = 17 a =? b =? 12

13 Meanings are sets of valid assignments a + b = 17 {a=0, b=0} {a=3, b=10} {a=5, b=12} {a=17, b=0} {a=10, b=7} {a=5, b=5} 13

14 Meanings are sets of valid assignments a + b = 17 {a=0, b=0} {a=17, b=0} {a=3, b=10} {a=10, b=7} {a=5, b=12} {a=5, b=5} 14

15 Meanings are sets of valid assignments a + 3 = 20 - b {a=0, b=0} {a=17, b=0} {a=3, b=10} {a=10, b=7} {a=5, b=12} {a=5, b=5} 15

16 Meanings are functions that judge validity a + b = 17 {a=5, b=12} 16

17 Meanings are functions that judge validity a + b = 17 {a=3, b=10} 17

18 Lessons from math a + b = 17 The meaning of a statement is the set of possible worlds consistent with that statement. Here, a possible world is an assignment of values to variables. {a=3, b=10} 18

19 Meaning in natural languages Pat Sal. 19

20 Representing possible worlds Individuals Pat Sal Properties whale sad Relations loves contains 20

21 Example world Sam Pat Sal Lou 21

22 Example world loves Sam scared worried Pat contains Sal Lou shark happy 22

23 Different example world loves sad Sam sad Pat loves loves Sal Lou sad sad 23

24 Representing possible worlds Individuals Pat Sal Properties whale={lou}, sad={pat,sal} Relations ={(Pat,Sal),(Sal,Sam)} 24

25 25 Pat Sal. worried Pat Sal Sam Lou loves happy scared shark worried Pat Sal Sam Lou loves contains loves scared shark worried Pat Sal Sam Lou contains happy scared human worried Pat Sal Sam Lou contains loves happy scared dog Interpretations of sentences

26 26 Lou is a shark. worried Pat Sal Sam Lou loves happy scared shark worried Pat Sal Sam Lou loves contains loves scared shark worried Pat Sal Sam Lou contains happy scared human worried Pat Sal Sam Lou contains loves happy scared dog Interpretations of sentences

27 Interpretations of sentences Sam is inside Lou, a shark. loves loves Sam scared Sam scared worried Pat worried Pat loves contains Sal Lou shark Sal Lou shark happy Sam scared loves worried Pat Sal happy contains Lou human worried Pat Sal Sam scared contains Lou dog happy 27

28 KEY IDEA The meaning of a sentence is the set of possible worlds it picks out.

29 PART II How is meaning constructed?

30 Explicit representation is too hard Pat Sal. loves loves Sam scared Sam scared worried Pat worried Pat loves contains Sal Lou shark Sal Lou shark happy Sam scared loves worried Pat Sal happy contains Lou human worried Pat Sal Sam scared contains Lou dog happy 30

31 Meanings as functions Pat Sal loves Sam scared worried Pat Sal Lou whale happy 31

32 Meanings as logical statements Pat Sal (Pat, Sal) loves Sam scared worried Pat Sal Lou whale happy 32

33 Expressing functions with logic Pat Sal (Pat, Sal) 33

34 Meanings as logical statements Lou is a shark shark(lou) 34

35 Meanings as logical statements Sam is inside Lou, a shark 35

36 Meanings as logical statements Sam is inside Lou, a shark shark(lou) contains(lou, Sam) 36

37 Meanings as logical statements Nobody Lou 37

38 Meanings as logical statements Nobody Lou x. (x, Lou) 38

39 Meanings as logical statements Everyone who knows Sal is happy 39

40 Meanings as logical statements Everyone who knows Sal is happy x. knows(x, Sal) happy(x) 40

41 KEY IDEA Collections of possible worlds can be compactly represented with logical forms.

42 Compositionality of meaning Pat Sal (Pat, Sal) Lou is a shark shark(lou) Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Nobody Lou x. (x, Lou) 42

43 Compositionality of meaning Pat Sal (Pat, Sal) Lou is a shark shark(lou) Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Nobody Lou x. (x, Lou) 43

44 Compositionality of meaning Pat Sal (Pat, Sal) Lou is a shark shark(lou) Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Nobody Lou x. (x, Lou) 44

45 Compositionality of meaning A Sal le gusta Pat (Pat, Sal) Lou es un tiburón shark(lou) Sam está dentro de Lou, un tiburón shark(lou) contains(lou, Sam) A nadie le gusta Lou x. (x, Lou) 45

46 Compositionality of meaning a12 b5 c67 a8 (Pat, Sal) a12 b5 c0 a0 shark(lou) a12 b16 c12 c12 shark(lou) contains(lou, Sam) a53 x. (x, Lou) 46

47 KEY IDEA Pieces of logical forms correspond to pieces of language

48 Building a lexicon Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Pat: Pat Sal: Sal Sam: Sam Lou: Lou 48

49 Building a lexicon Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Pat: Pat Sal: Sal Sam: Sam Lou: Lou shark: λx.whale(x) Sal: Sal Sam: Sam Lou: Lou 49

50 Building a lexicon Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Pat: Pat Sal: Sal Sam: Sam Lou: Lou shark: λx.shark(x) Sal: Sal Sam: Sam Lou: Lou 50

51 Building a lexicon Sam is inside Lou, a shark shark(lou) contains(lou, Sam) Pat: Pat Sal: Sal Sam: Sam Lou: Lou shark: λx.shark(x) : λyx.(x, y) nobody: λf. x. f(x) Lou:... Lou 51

52 What do we do now? Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) 52

53 What do we do now? Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) 53

54 What do we do now? Ax.letter(x)? Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) 54

55 What do we do now? letter(λf.ax.f(x))? Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) 55

56 What do we do now???? Ax.letter(x)? Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) 56

57 What do we do now?????? Pat sent Lou urgently. Pat λyzx.sent(x,y,z) Lou??? 57

58 Semantic types Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) {Obj,Obj,Obj} Object Object Object Object Bool Bool Bool Bool 58

59 Semantic types & syntax Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP,NP,NP NP NP NP NP S S S S 59

60 Semantic types & syntax Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP ((S NP) NP) NP NP S (S NP) S NP 60

61 Categorial grammar Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP ((S\NP)/NP)/NP NP NP/(S/NP) S/NP 61

62 Parsing with a categorial grammar Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP ((S\NP)/NP)/NP NP NP/(S/NP) S/NP Ax.letter(x) NP 62

63 Parsing with a categorial grammar Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP ((S\NP)/NP)/NP NP NP/(S/NP) S/NP λzx.sent(x,lou,z) (S\NP)/NP Ax.letter(x) NP 63

64 Parsing with a categorial grammar Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP ((S\NP)/NP)/NP NP NP/(S/NP) S/NP λzx.sent(x,lou,z) (S\NP)/NP Ax.letter(x) NP λx.sent(x,lou,ax.letter(x)) S\NP 64

65 Parsing with a categorial grammar Pat sent Lou a letter Pat λyzx.sent(x,y,z) Lou λf.ax.f(x) λx.letter(x) NP ((S\NP)/NP)/NP NP NP/(S/NP) S/NP λzx.sent(x,lou,z) (S\NP)/NP Ax.letter(x) NP λx.sent(x,lou,ax.letter(x)) S\NP sent(pat,lou,ax.letter(x)) S 65

66 Semantics Synax! Pat sent Lou a letter 66

67 Semantics Synax! Pat sent Lou a letter 67

68 KEY IDEA Types in logic correspond to grammatical categories in language

69 Problem 1 Each of the three girls has a platypus. Each of the three girls climbed the mountain. x.girl(x) y.platypus(y) has(x, y) y.mountain(y) x.girl(x) climbed(x, y) 69

70 Problem 2 There are 128 cities in South Carolina name type coastal Columbia city no Cooper river yes Charleston city yes 70

71 Problem 2 There are 128 cities in South Carolina name type coastal Columbia city no Cooper river yes same(128, Charleston city yes count x. city(x) in(x, SouthCarolina) 71

72 Problem 3 Barack Obama was the 44th President of the United States. Obama was born on August 4 in Honolulu, Hawaii. In late August 1961, Obama's mother moved with him to the University of Washington in Seattle for a year Is Barack Obama from the United States? 72

73 Problem 3 Barack Obama was the 44th President of the United born(obama, Hawaii, August 4) States. Obama was born on August 4 in Honolulu, Hawaii. In late August 1961, Obama's mother moved with him to the University of Washington in Seattle for a year Is Barack from(obama, from United the United States) States? 73

74 Problem 3 Barack Obama was the 44th President of the United born(obama, Hawaii, August 4) States. Obama was born on August 4 in Honolulu, Hawaii. In late August 1961, Obama's mother moved with born(x, y, z) from(x, y) him to the University of Washington in Seattle for a year from(x, y) in(y, z) from(x, z) in(hawaii, United States) Is Barack from(obama, from United the United States) States? 74

75 KEY IDEA The meaning of a sentence is the set of possible worlds it picks out.

76 KEY IDEA Collections of possible worlds can be compactly represented with logical forms.

77 KEY IDEA Pieces of logical forms correspond to pieces of language

78 KEY IDEA Types in logic correspond to grammatical categories in language

79 BONUS ROUND What s missing?

80 Saying what we mean Q: How do you like my cooking? 80

81 Saying what we mean Q: How do you like my cooking? A: It s extremely interesting. 81

82 Saying what we mean Q: How do you like my cooking? A: It s extremely interesting. Q: Do you know what time it is? 82

83 Saying what we mean Q: How do you like my cooking? A: It s extremely interesting. Q: Do you know what time it is? A: Yes, I do. 83

84 Belief & possibility Sal might have seen a unicorn. Pat thinks Sal saw a unicorn. Pat wants to find a unicorn. 84

85 KEY IDEA Not all meaning is literal!

86 BONUS ROUND Historical Notes

87 Alfred Tarski Richard Montague

88 Learn more ling121: Logical Semantics Ted Briscoe s lecture notes: Mark Steedman, The Syntactic Process 88