Resolving Quantity and Informativeness Implicature in Indefinite - - PowerPoint PPT Presentation

Resolving Quantity and Informativeness Implicature in Indefinite Reference

Till Poppels and Roger Levy July 18 XPRAG 2015

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

The Phenomenon

1

The man injured his child. The man injured someone else’s child. The man injured a child.

OWN OTHER’S

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

The Phenomenon

1

The man injured his child. The man injured someone else’s child. The man injured a child.

OWN OTHER’S

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

The Phenomenon…

1

OWN OTHER’S

The man injured a child. The man injured his child. The man injured someone else’s child. The man broke a finger. The man broke a nose. The man injured a son.

…and the research question

The X V-ed a Y.

What determines this variation in the directionality and strength

• f inferences about

utterance meaning?

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Gricean inferences

speaker behavior Be brief. Be informative. listener inferences

Quantity Informativeness

Grice (1967, 1975); Zipf (1949); Levinson & Atlas (1987); Levinson (2000); Horn (1984, 2004); Frank & Goodman (2012) John ate some of the cookies +> but not all of them I’ll give you $5 if you mow the lawn +> but only if you do

The X V-ed a Y.

2

Grice (1967, 1975); Zipf (1949); Le Grice (1967, 1975); Zipf (1949); Levinson & Atlas (1987); Levinson (2000); H Grice (1967, 1975); Zipf (1949); Levinson & Atlas (1987); Levinson (2000); Horn (1984, 2004); Fr

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Frank & Goodman (2012) OWN OTHER’S The man injured a child. 1 1 The man injured his child. 1 The man injured someone else’s child. 1

The Rational Speech Act (RSA) model

Assumption 1: The “Lexicon”

3

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Frank & Goodman (2012) OWN OTHER’S The man injured a child. 1 1 The man injured his child. 1 The man injured someone else’s child. 1 𝐸 𝑏 = 1 𝐸 ℎ𝑗𝑡 = 1 𝐸 𝑡𝑝𝑛𝑓𝑝𝑜𝑓 𝑓𝑚𝑡𝑓′𝑡 = 4

The Rational Speech Act (RSA) model

Assumption 1: The “Lexicon” Assumption 2: Utterance costs

3

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Frank & Goodman (2012)

The Rational Speech Act (RSA) model

3

OWN OTHER’S The man injured a child. 1 1 The man injured his child. 1 The man injured someone else’s child. 1

L0

Literal Listener

Prior

OWN OTHER’S L0 ~ lexic lexicon

• n * p

* prio rior Interpretation OWN OTHER’S S1 ~ ~ exp exp(log(

• g(L0)-cost
• st)

S1

Gricean Speaker Utterance Cost 1 1 4

Prior

OWN OTHER’S

𝑞(𝑏|𝑃𝑈𝐼𝐹𝑆′𝑇) 𝑞(𝑏)

1 S1 ~ exp exp(λ*(l *(log(

• g(L0)-cost
• st))

))

L1

Pragmatic Listener L1 ~ ~ prior prior * * S1

Prior

OWN OTHER’S Interpretation OWN OTHER’S

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Predictions

Prior

OWN OTHER’S

Interpretation

OTHER’S OWN

1.Interpretations track priors 2.Baseline Q-implicature towards OTHER’S 3.Reduced Q-implicature in “headlines” 4.Strengthened Q-implicature where X’s Y is unique

4

Utterance The X V-ed a Y. X V-ed Y.

ambiguous 1 his 1

1

someone else’s 4

4

The man broke a nose. # a brightest student # a US president D(a,OWN) > D(a,OTHER’S)

(Prior only) The X V-ed a Y. X V-ed Y. The X V-ed a Y. (X’s Y unique) Frank & Goodman (2012) Frank & Goodman (2012); Hawkins (1991) Frank & Goodman (2012); Hawkins (1991); Jäger (2012)

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Man broke finger. Man broke nose. Father injured son. Nurse broke finger. Man shaved leg. Man shaved upper lip. Woman shaved leg. Woman shaved upper lip. Man entered house. Man broke neck. Tiger broke nose. Python broke nose. Python broke neck. Tiger broke neck. Man broke leg. Man broke back. Man broke promise. Man broke cup. The man broke a finger. The man broke a nose. The man injured a child. The father injured a son. The nurse broke a finger. The man shaved a leg. The man shaved an upper lip. The woman shaved a leg. The woman shaved an upper lip. The man entered a house. The man broke a neck. The tiger broke a nose. The python broke a nose. The python broke a neck. The tiger broke a neck. The man broke a leg. The man broke a back. The man broke a promise.

Methodology

Experiment 1: Interpretations Norming experiment: Priors

respo sponse nse ~ prior ior + + XYu XYuniq niquenes ueness + + rel relata atabilit bility + + hea headli dline ne + (1 (1 + + headli eadline ne | | item) tem)

5

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

man saving family teacher injuring student

6

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Results

Model Coefficients

OWN OTHER’S ns *** *** *** ***

Prior

OWN OTHER’S

Posterior

OTHER’S OWN

Model Predictions Regression Results

7

(Prior only) The X V-ed a Y. X V-ed Y. The X V-ed a Y. (X’s Y unique)

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Discussion point 1 of 3: No effect of the prior?

Norming experiment: Priors 3 possibilities:

• 1. Noisy measures
• 2. Maybe RSA got it wrong?
• 3. Event priors vs. “Intention priors”

8

Model Coefficients

OWN OTHER’S ns *** *** *** ***

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Discussion point 2 of 3: Support for RSA

RSA excels at predicting Q-implicatures:

• 1. Overall OTHER’S skew
• 2. Opposing trend in headline versions
• 3. Enhanced Q-implicature where X’s Y is unique

9

Model Coefficients

OWN OTHER’S ns *** *** *** ***

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

The X V-ed a Y.

The man injured a child. The father injured a child.

Discussion point 3 of 3: Relatability – Q or I (or neither?)

2 possible reasons:

• Ad hoc Q-implicature about referring expressions

(e.g. man vs. father)

• I-driven inference from real-world knowledge

about the event participants (cf. I almost bought a car today but the engine was too noisy.)

Hirschberg (1985); Clark (1975); Prince & Cole (1981); see also Cohen & Kehler (in prep)

10

Model Coefficients

OWN OTHER’S ns *** *** *** ***

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Conclusion

What we have learned Where to go from here

• Forced-choice experiments and

mixed-logit models: great for studying interpretational preferences

• Q/I resolution is determined by

multiple interacting factors

• RSA captures the essence of

Q-implicature

• We don’t understand

Informativeness nearly as well

• Inference taxonomies may become

explanatorily obsolete

• Cross-linguistic validation of RSA
• More research on I-driven

inferences

11

Thank you.

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

References

Atlas, J., & Levinson, S. (1981). It-clefts, informativeness and logical form: radical pragmatics (revised standard version). Radical Pragmatics. Clark, H. (1975). Bridging. In Proceedings of the 1975 workshop on Theoretical issues in natural language

• processing. Stroudsburg, PA, USA: Association for

Computational Linguistics. Frank, M. C., & Goodman, N. D. (2012). Predicting pragmatic reasoning in language games. Science, 336(6084), 998–998. Grice, H. P. (1957). Meaning. The Philosophical Review, 377–388. Grice, H. P. (1975). Logic and conversation. 1975, 41–58. Hawkins, J. A. (1991). On (in)definite articles: implicatures and (un)grammaticality prediction. Journal of Linguistics, 27(02), 405. Hirschberg, J. (1985). A Theory of Scalar Implicature (Natural Languages, Pragmatics, Inference). Dissertations Available from ProQuest.

11

Horn, L. R. (1984). Toward a New Taxonomy for Pragmatic Inference: Q-based and R-based Implicatures. In D. Schiffrin (Ed.), Meaning, Form, and Use in Context (pp. 11–42). Georgetown University Press. Horn, L. R. (2004). Implicature. In L. R. Horn & G. Ward (Eds.), Handbook of pragmatics (pp. 3–28). Blackwell Publishing Ltd. Jäger, G. (2012). Game theory in semantics and pragmatics, in C. Maienborn, P. Portner & K. von Heusinger (eds.), Semantics. An International Handbook of Natural Language Meaning, Vol. 3, Berlin: de Gruyter, 2487-2516. Kehler, A., & Cohen, J. (in prep). Conversational Elicitures. Levinson, S. C. (2000). Presumptive meanings: The theory

• f generalized conversational implicature. MIT

Press. Prince, E. F., & Cole, P. (1981). Toward a taxonomy of given-new information (pp. 223–255). Zipf, G. (1949). Human behavior and the principle of least effort.

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

RSA predictions by disambiguation costs

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Comparing RSA implementations

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Regression results

+HEADLINE (p < .001) +X's Y unique (p < .001) +Relatable (p < .001) Intercept (p < .001) Prior (p = .13)

• 2

<< OTHER'S

• 1

1 2 OWN >>

Model Coefficients

Q/I Resolution in Indefinite Reference

Till Poppels & Roger Levy (UCSD)

• f 11

Logit transform