the role of veridicality and factivity in clause selection
play

The role of veridicality and factivity in clause selection Aaron - PowerPoint PPT Presentation

NELS 48 University of Iceland 1 University of Rochester, Department of Linguistics Institute for Data Science 2 Johns Hopkins University, Department of Cognitive Science 1 The role of veridicality and factivity in clause selection Aaron Steven


  1. factivity/veridicality for all clause-embedding verbs to show that... Use experiments that measure (i) syntactic distribution and (ii) 1. ...selection doesn’t directly traffic in these properties. 2.1 ...only analyzing frequent verbs. 2.2 ...lack of attention to confounding event structural 7 Today’s talk Focus Two intentional properties— factivity and veridicality —that are argued to determine selection of interrogatives & declaratives Claims 2. ...apparent correlations between selection and factivity and veridicality arise from... variables like transfer and stativity .

  2. Introduction Background: veridicality & factivity Measuring syntactic distribution Measuring veridicality and factivity Results and analysis Conclusion 8 Outline

  3. Introduction Background: veridicality & factivity Measuring syntactic distribution Measuring veridicality and factivity Results and analysis Conclusion 8 Outline

  4. Introduction Background: veridicality & factivity Measuring syntactic distribution Measuring veridicality and factivity Results and analysis Conclusion 8 Outline

  5. Introduction Background: veridicality & factivity Measuring syntactic distribution Measuring veridicality and factivity Results and analysis Conclusion 8 Outline

  6. Introduction Background: veridicality & factivity Measuring syntactic distribution Measuring veridicality and factivity Results and analysis Conclusion 8 Outline

  7. Introduction Background: veridicality & factivity Measuring syntactic distribution Measuring veridicality and factivity Results and analysis Conclusion 8 Outline

  8. Background: veridicality & factivity

  9. Jo knew that Bo was alive Jo proved that Bo was alive Factivity A verb v is factive iff np v s presupposes s Kiparsky & Kiparsky 1970, Jo didn’t know that Bo was alive Jo didn’t prove that Bo was alive Bo was alive b. Bo was alive a. (2) Karttunen 1971b et seq 10 Bo was alive b. Bo was alive a. (1) tunen 2012, Spector & Egré 2015 a.o. Veridicality and factivity Veridicality A verb v is veridical iff np v s entails s Karttunen 1971a, Egré 2008, Kart-

  10. Jo proved that Bo was alive Factivity A verb v is factive iff np v s presupposes s Kiparsky & Kiparsky 1970, Jo didn’t know that Bo was alive Jo didn’t prove that Bo was alive Bo was alive b. Bo was alive a. (2) Karttunen 1971b et seq 10 Bo was alive b. a. (1) tunen 2012, Spector & Egré 2015 a.o. Veridicality and factivity Veridicality A verb v is veridical iff np v s entails s Karttunen 1971a, Egré 2008, Kart- Jo knew that Bo was alive → Bo was alive

  11. Factivity A verb v is factive iff np v s presupposes s Kiparsky & Kiparsky 1970, Jo didn’t know that Bo was alive Jo didn’t prove that Bo was alive Karttunen 1971b et seq Bo was alive b. Bo was alive a. (2) 10 b. a. (1) tunen 2012, Spector & Egré 2015 a.o. Veridicality and factivity Veridicality A verb v is veridical iff np v s entails s Karttunen 1971a, Egré 2008, Kart- Jo knew that Bo was alive → Bo was alive Jo proved that Bo was alive → Bo was alive

  12. Jo didn’t know that Bo was alive Jo didn’t prove that Bo was alive 10 a. Bo was alive tunen 2012, Spector & Egré 2015 a.o. (1) a. b. b. Bo was alive Karttunen 1971b et seq (2) Veridicality and factivity Veridicality A verb v is veridical iff np v s entails s Karttunen 1971a, Egré 2008, Kart- Jo knew that Bo was alive → Bo was alive Jo proved that Bo was alive → Bo was alive Factivity A verb v is factive iff np v s presupposes s Kiparsky & Kiparsky 1970,

  13. Jo didn’t prove that Bo was alive 10 Karttunen 1971b et seq Bo was alive tunen 2012, Spector & Egré 2015 a.o. (1) a. b. b. a. (2) Veridicality and factivity Veridicality A verb v is veridical iff np v s entails s Karttunen 1971a, Egré 2008, Kart- Jo knew that Bo was alive → Bo was alive Jo proved that Bo was alive → Bo was alive Factivity A verb v is factive iff np v s presupposes s Kiparsky & Kiparsky 1970, Jo didn’t know that Bo was alive → Bo was alive

  14. 10 Karttunen 1971b et seq b. tunen 2012, Spector & Egré 2015 a.o. (1) a. a. b. (2) Veridicality and factivity Veridicality A verb v is veridical iff np v s entails s Karttunen 1971a, Egré 2008, Kart- Jo knew that Bo was alive → Bo was alive Jo proved that Bo was alive → Bo was alive Factivity A verb v is factive iff np v s presupposes s Kiparsky & Kiparsky 1970, Jo didn’t know that Bo was alive → Bo was alive Jo didn’t prove that Bo was alive ̸→ Bo was alive

  15. 11 see also George 2011, Uegaki 2012, 2015; cf. Beck & Rullmann 1999, Spector & Egré 2015 b. see also Karttunen 1977a,b, Groenendijk & Stokhof 1984 et seq (3) a. a. b. (4) Veridicality/factivity and responsivity Responsivity (Lahiri 2002) A verb is responsive iff it takes interrogatives and declaratives Jo knew that Bo was alive. Jo knew whether Bo was alive. Generalization A verb is responsive iff { factive (Hintikka 1975) / veridical (Egré 2008) } Jo knew {that, whether} Bo was alive. Jo thought {that, * whether} Bo was alive.

  16. 12 Predicted correlation Responsivity Factivity/Veridicality

  17. Measuring syntactic distribution

  18. Ordinal (1-7 scale) acceptability ratings for 1000 clause-embedding verbs 50 syntactic frames 14 Measuring syntactic distribution MegaAttitude dataset (White & Rawlins 2016)

  19. Ordinal (1-7 scale) acceptability ratings for 1000 clause-embedding verbs 50 syntactic frames 14 Measuring syntactic distribution MegaAttitude dataset (White & Rawlins 2016)

  20. Ordinal (1-7 scale) acceptability ratings for 1000 clause-embedding verbs 50 syntactic frames 14 Measuring syntactic distribution MegaAttitude dataset (White & Rawlins 2016) ×

  21. 15 MegaAttitude verbs

  22. Solution know + NP _ed {that, whether} S tell + NP _ed NP {that, whether} S bother + NP was _ed {that, which NP} S b. Someone was bothered {that something, which thing} happened. c. Someone was told {that, whether} something happened. 16 Someone knew {that, whether} something happened. a. (5) Construct semantically bleached frames using indefinites is sufficiently general to many verbs Automate construction of a very large set of frames in a way that Sentence construction Challenge

  23. know + NP _ed {that, whether} S tell + NP _ed NP {that, whether} S bother + NP was _ed {that, which NP} S Someone knew {that, whether} something happened. Someone was bothered {that something, which thing} happened. c. Someone was told {that, whether} something happened. b. 16 a. (5) Construct semantically bleached frames using indefinites is sufficiently general to many verbs Automate construction of a very large set of frames in a way that Sentence construction Challenge Solution

  24. tell + NP _ed NP {that, whether} S bother + NP was _ed {that, which NP} S 16 c. Automate construction of a very large set of frames in a way that is sufficiently general to many verbs Someone was bothered {that something, which thing} happened. Construct semantically bleached frames using indefinites (5) a. Someone knew {that, whether} something happened. b. Someone was told {that, whether} something happened. Sentence construction Challenge Solution know + NP _ed {that, whether} S

  25. bother + NP was _ed {that, which NP} S 16 Someone was told {that, whether} something happened. Automate construction of a very large set of frames in a way that is sufficiently general to many verbs Someone was bothered {that something, which thing} happened. Construct semantically bleached frames using indefinites (5) a. Someone knew {that, whether} something happened. b. c. Sentence construction Challenge Solution know + NP _ed {that, whether} S tell + NP _ed NP {that, whether} S

  26. 16 Someone was told {that, whether} something happened. Automate construction of a very large set of frames in a way that is sufficiently general to many verbs Someone was bothered {that something, which thing} happened. Construct semantically bleached frames using indefinites (5) a. c. Someone knew {that, whether} something happened. b. Sentence construction Challenge Solution know + NP _ed {that, whether} S tell + NP _ed NP {that, whether} S bother + NP was _ed {that, which NP} S

  27. 16 Someone was told {that, whether} something happened. Automate construction of a very large set of frames in a way that is sufficiently general to many verbs Someone was bothered {that something, which thing} happened. Construct semantically bleached frames using indefinites (5) a. c. Someone knew {that, whether} something happened. b. Sentence construction Challenge Solution know + NP _ed {that, whether} S tell + NP _ed NP {that, whether} S bother + NP was _ed {that, which NP} S

  28. Measuring veridicality and factivity

  29. ...you will be given a statement and a question related to that statement. Your task will be to respond yes , maybe or maybe not , or no to the question, assuming that the statement is true. (cf. Karttunen et al. 2014) 18 Task

  30. 19 Task

  31. 20 Task

  32. • 348 verbs only in the active frame • 142 only in the passive frame • 27 in both 1,088 items randomly partitioned into 16 lists of 68 21 Stimuli 517 verbs from the MegaAttitude based on their acceptability in the [NP _ that S] and [NP was _ed that S] frames

  33. • 348 verbs only in the active frame • 142 only in the passive frame • 27 in both 1,088 items randomly partitioned into 16 lists of 68 22 Stimuli 517 verbs from the MegaAttitude based on their acceptability in the [NP _ that S] and [NP was _ed that S] frames

  34. • 348 verbs only in the active frame • 142 only in the passive frame • 27 in both 1,088 items randomly partitioned into 16 lists of 68 22 Stimuli 517 verbs from the MegaAttitude based on their acceptability in the [NP _ that S] and [NP was _ed that S] frames

  35. • 348 verbs only in the active frame • 142 only in the passive frame • 27 in both 1,088 items randomly partitioned into 16 lists of 68 22 Stimuli 517 verbs from the MegaAttitude based on their acceptability in the [NP _ that S] and [NP was _ed that S] frames

  36. • 348 verbs only in the active frame • 142 only in the passive frame • 27 in both 1,088 items randomly partitioned into 16 lists of 68 22 Stimuli 517 verbs from the MegaAttitude based on their acceptability in the [NP _ that S] and [NP was _ed that S] frames

  37. • 348 verbs only in the active frame • 142 only in the passive frame • 27 in both 1,088 items randomly partitioned into 16 lists of 68 22 Stimuli 517 verbs from the MegaAttitude based on their acceptability in the [NP _ that S] and [NP was _ed that S] frames

  38. Passive 23 b. happened. Someone wasn’t bothered that a particular thing b. Someone was bothered that a particular thing happened. a. (8) Someone wasn’t told that a particular thing happened. Someone was told that a particular thing happened. a. (7) Someone didn’t think that a particular thing happened. b. Someone thought that a particular thing happened. a. (6) Stimuli Active

  39. 23 b. happened. Someone wasn’t bothered that a particular thing b. Someone was bothered that a particular thing happened. a. (8) Someone wasn’t told that a particular thing happened. Someone was told that a particular thing happened. a. (7) Someone didn’t think that a particular thing happened. b. Someone thought that a particular thing happened. a. (6) Stimuli Active Passive

  40. 23 b. happened. Someone wasn’t bothered that a particular thing b. Someone was bothered that a particular thing happened. a. (8) Someone wasn’t told that a particular thing happened. Someone was told that a particular thing happened. a. (7) Someone didn’t think that a particular thing happened. b. Someone thought that a particular thing happened. a. (6) Stimuli Active Passive

  41. • 10 ratings per item... • ...given by 10 different participants 24 Participants 160 unique participants through Amazon’s Mechanical Turk

  42. • 10 ratings per item... • ...given by 10 different participants 24 Participants 160 unique participants through Amazon’s Mechanical Turk

  43. • 10 ratings per item... • ...given by 10 different participants 24 Participants 160 unique participants through Amazon’s Mechanical Turk

  44. Results and analysis

  45. 26 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  46. 27 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  47. 28 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  48. 29 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  49. 30 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  50. 31 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  51. 32 Raw responses know prove think V(p) ¬V(p) o e s o e s o e s e e e n b n b n b y y y y y y a a a m m m

  52. Transformation (roughly) Normalize Map each verb to single two-dimensional point by assigning -1 to no , 0 to maybe , and 1 to yes , then take the mean. Use ridit scoring to normalize for how often a particular partic- ipant gives a particular response. (Similar to z -scoring.) 33 Normalization

  53. Normalize Map each verb to single two-dimensional point by assigning -1 to no , 0 to maybe , and 1 to yes , then take the mean. Use ridit scoring to normalize for how often a particular partic- ipant gives a particular response. (Similar to z -scoring.) 33 Normalization Transformation (roughly)

  54. 34 Normalized responses ¬p ← V(p) → p ¬p ← ¬V(p) → p

  55. Map each verb to single two-dimensional point by assigning -1 to no , 0 to maybe , and 1 to yes , then take the mean. Use ridit scoring to normalize for how often a particular partic- ipant gives a particular response. (Similar to z -scoring.) 35 Normalization Transformation (roughly) Normalize

  56. 36 Normalized responses ¬p ← V(p) → p ¬p ← ¬V(p) → p

  57. 37 Normalized responses ¬p ← V(p) → p Nonveridicals decide think promise hope wish believe Frame NP _ that S a NP was _ed that S a ¬p ← ¬V(p) → p

  58. 38 Normalized responses Factives know love surprise hate find_out ¬p ← V(p) → p Nonveridicals decide decide think think promise promise hope hope wish wish believe believe Frame NP _ that S a NP was _ed that S a ¬p ← ¬V(p) → p

  59. 39 Normalized responses Veridicals Factives show verify know know love love find surprise surprise prove hate hate show find_out find_out ensure indicate ¬p ← V(p) → p Nonveridicals decide decide decide think think think promise promise promise hope hope hope wish wish wish believe believe believe Frame NP _ that S a NP was _ed that S a ¬p ← ¬V(p) → p

  60. 40 Normalized responses Veridicals Factives show show verify verify know know know love love love find find surprise surprise surprise prove prove hate hate hate show show find_out find_out find_out ensure ensure indicate indicate ¬p ← V(p) → p Nonveridicals decide decide decide decide think think think think promise promise promise promise hope hope hope hope wish wish wish wish believe believe believe believe fabricate Antiveridicals pretend mislead misinform Frame fake NP _ that S a NP was _ed that S a ¬p ← ¬V(p) → p

  61. 41 Normalized responses Veridicals Factives show show show verify verify verify know know know know love love love love find find find surprise surprise surprise surprise prove prove prove hate hate hate hate show show show find_out find_out find_out find_out ensure ensure ensure indicate indicate indicate ¬p ← V(p) → p Nonveridicals decide decide decide decide decide think think think think think promise promise promise promise promise hope hope hope hope hope wish wish wish wish wish believe believe believe believe believe fabricate fabricate hallucinate Antiveridicals Antifactives? pretend pretend mislead mislead misinform misinform Frame fake fake NP _ that S a NP was _ed that S a ¬p ← ¬V(p) → p

  62. 42 Normalized responses ¬p ← V(p) → p ¬p ← ¬V(p) → p

  63. 43 Normalized responses Veridicals Factives verify verify verify verify verify show show show show show know know know know know know love love love love love love surprise surprise surprise surprise surprise surprise find find find find find prove prove prove prove prove hate hate hate hate hate hate show show show show show find_out find_out find_out find_out find_out find_out ensure ensure ensure ensure ensure indicate indicate indicate indicate indicate Veridicality Nonveridicals decide decide decide decide decide decide decide promise promise promise promise promise promise promise think think think think think think think hope hope hope hope hope hope hope wish wish wish wish wish wish wish believe believe believe believe believe believe believe fabricate fabricate fabricate fabricate hallucinate hallucinate Antiveridicals Antifactives? pretend pretend pretend pretend mislead mislead mislead mislead misinform misinform misinform misinform Frame fake fake fake fake NP _ that S a NP was _ed that S a Factivity

  64. Do factivity/veridicality positively correlate with question-taking? 44 Relating factivity, veridicality, and question-taking Question

  65. 45 Correlation: factivity and question-taking Acceptability of [_ CP[+Q]] Factivity

  66. Intuition For a particular verb, maximum acceptability over all frames that contain an interrogative complement. If a verb is acceptable in some frame that contains an interrog- ative complement, it is acceptable with interrogatives. 46 Measure of question selection Acceptability of [ CP[+Q]]

  67. For a particular verb, maximum acceptability over all frames that contain an interrogative complement. If a verb is acceptable in some frame that contains an interrog- ative complement, it is acceptable with interrogatives. 46 Measure of question selection Acceptability of [ CP[+Q]] Intuition

  68. 47 Correlation: factivity and question-taking Acceptability of [_ CP[+Q]] Factivity

  69. 48 Correlation: factivity and question-taking Acceptability of [_ CP[+Q]] Factivity

  70. 49 Correlation: factivity and question-taking Acceptability of [_ CP[+Q]] Factivity

  71. 50 Correlation: veridicality and question-taking Acceptability of [_ CP[+Q]] Veridicality

  72. 51 Correlation: veridicality and question-taking Acceptability of [_ CP[+Q]] Veridicality

  73. 52 Correlation: veridicality and question-taking Acceptability of [_ CP[+Q]] Veridicality

  74. Two hypotheses How could we have gotten the direction of correlation so wrong? 1. Previous analyses were biased by verb frequency. 2. Our analysis missed subregularities due to verb class. 53 What’s going on? Question

  75. How could we have gotten the direction of correlation so wrong? 1. Previous analyses were biased by verb frequency. 2. Our analysis missed subregularities due to verb class. 53 What’s going on? Question Two hypotheses

  76. Finding #1 Finding #2 If we look at only the most frequent verbs, the correlations flip. There are subregularities, but they don’t validate the purported correlation. 54 Two findings

  77. Finding #2 If we look at only the most frequent verbs, the correlations flip. There are subregularities, but they don’t validate the purported correlation. 54 Two findings Finding #1

  78. 55 Correlation: factivity with all verbs Acceptability of [_ CP[+Q]] Factivity

  79. 56 Correlation: factivity with high-frequency verbs Acceptability of [_ CP[+Q]] know write tell showshow think say see find require Factivity

  80. 57 Correlation: veridicality with all verbs Acceptability of [_ CP[+Q]] Veridicality

  81. 58 Correlation: veridicality with high-frequency verbs Acceptability of [_ CP[+Q]] know write tell show show think say see find require Veridicality

Recommend


More recommend