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

The Phenomenon O WN The man injured his child. The man injured someone else’s child. The man injured a child. O THER ’ S Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 1

The Phenomenon O WN The man injured his child. The man injured someone else’s child. The man injured a child. O THER ’ S Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 1

The P henomenon… …and the research question The X V -ed a Y . The man injured his child. O WN The man broke a finger. What determines this The man broke a nose. variation in the The man injured a child. directionality and strength of inferences about The man injured a son. utterance meaning? The man injured someone else’s child. O THER ’ S Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 1

Gricean inferences Quantity Informativeness Be brief. speaker behavior Be informative. John ate some of the cookies I’ll give you $5 if you mow the lawn listener inferences +> but not all of them +> but only if you do The X V -ed a Y . Grice (1967, 1975); Zipf (1949); Levinson & Atlas (1987); Levinson (2000); Horn (1984, 2004); Frank & Goodman (2012) Grice (1967, 1975); Zipf (1949); Levinson & Atlas (1987); Levinson (2000); Horn (1984, 2004); Fr Grice (1967, 1975); Zipf (1949); Levinson & Atlas (1987); Levinson (2000); H Grice (1967, 1975); Zipf (1949); Le Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 2

The Rational Speech Act (RSA) model O WN O THER ’ S The man injured a child. 1 1 Assumption 1: The man injured his child. 1 0 The “Lexicon” The man injured someone else’s child. 0 1 Frank & Goodman (2012) Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 3

The Rational Speech Act (RSA) model O WN O THER ’ S The man injured a child. 1 1 Assumption 1: The man injured his child. 1 0 The “Lexicon” The man injured someone else’s child. 0 1 𝐸 𝑏 = 1 Assumption 2: 𝐸 ℎ𝑗𝑡 = 1 Utterance costs 𝐸 𝑡𝑝𝑛𝑓𝑝𝑜𝑓 𝑓𝑚𝑡𝑓 ′ 𝑡 = 4 Frank & Goodman (2012) Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 3

The Rational Speech Act (RSA) model Utterance Cost O WN O THER ’ S The man injured a child. 1 1 1 The man injured his child. 1 1 0 The man injured someone else’s child. 4 0 1 L 0 S 1 L 1 Literal Listener Gricean Speaker Pragmatic Listener L 0 ~ lexic lexicon on * p * prio rior S 1 ~ S 1 ~ exp ~ exp exp(log( exp( λ *(l og(L 0 )-cost *(log( og(L 0 )-cost ost) ost)) )) L 1 ~ ~ prior prior * * S 1 O THER ’ S O THER ’ S 1 𝑞(𝑏|𝑃𝑈𝐼𝐹𝑆 ′ 𝑇) Interpretation Interpretation 𝑞(𝑏) O WN O WN 0 O WN O THER ’ S O WN O THER ’ S O WN O THER ’ S Frank & Goodman (2012) Prior Prior Prior Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 3

Predictions 1.Interpretations track priors 2.Baseline Q -implicature O THER ’ S towards OTHER’S Interpretation 3.Reduced Q -implicature in “headlines” (Prior only) 4.Strengthened Q -implicature The man broke a nose. The X V -ed a Y . Utterance The X V-ed a Y. X V-ed Y. where X ’s Y is unique # a brightest student X V -ed Y . ambiguous 1 0 The X V -ed a Y . # a US president his 1 1 ( X ’s Y unique) O WN someone else’s 4 4 D( a , OWN ) > D( a , OTHER ’ S ) O WN O THER ’ S Prior Frank & Goodman (2012); Hawkins (1991); Jäger (2012) Frank & Goodman (2012); Hawkins (1991) Frank & Goodman (2012) Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 4

Methodology The man broke a finger. Man broke finger. The man broke a nose. Man broke nose. The man injured a child. Father injured son. Experiment 1: Interpretations The father injured a son. Nurse broke finger. The nurse broke a finger. Man shaved leg. The man shaved a leg. Man shaved upper lip. Norming experiment: Priors The man shaved an upper lip. Woman shaved leg. The woman shaved a leg. Woman shaved upper lip. The woman shaved an upper lip. Man entered house. The man entered a house. Man broke neck. The man broke a neck. Tiger broke nose. The tiger broke a nose. Python broke nose. The python broke a nose. Python broke neck. The python broke a neck. Tiger broke neck. The tiger broke a neck. Man broke leg. The man broke a leg. Man broke back. The man broke a back. Man broke promise. The man broke a promise. Man broke cup. respo sponse nse ~ prior ior + + XYu XYuniq niquenes ueness + + rel relata atabilit bility + + hea headli dline ne + (1 (1 + + headli eadline ne | | item) tem) Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 5

man saving family teacher injuring student Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 6

Results Model Predictions Regression Results O THER ’ S O THER ’ S *** *** Model Coefficients Posterior (Prior only) The X V -ed a Y . ns X V -ed Y . The X V -ed a Y . ( X ’s Y unique) O WN O WN *** *** O WN O THER ’ S Prior Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 7

Discussion point 1 of 3: No effect of the prior? 3 possibilities: 1. Noisy measures O THER ’ S *** 2. Maybe RSA got it wrong? Norming experiment: Priors 3. Event priors vs. “Intention priors” *** Model Coefficients ns O WN *** *** Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 8

Discussion point 2 of 3: Support for RSA RSA excels at predicting Q -implicatures: O THER ’ S *** 1. Overall OTHER’S skew 2. Opposing trend in headline versions *** Model Coefficients 3. Enhanced Q -implicature where X ’s Y is unique ns O WN *** *** Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 9

Discussion point 3 of 3: Relatability – Q or I (or neither?) The X V -ed a Y . O THER ’ S The man injured a child. *** The father injured a child. *** Model Coefficients 2 possible reasons: • Ad hoc Q -implicature about referring expressions (e.g. man vs. father ) • I -driven inference from real-world knowledge ns about the event participants (cf. I almost bought a car today but the engine was too noisy.) O WN *** *** Hirschberg (1985); Clark (1975); Prince & Cole (1981); see also Cohen & Kehler (in prep) Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 10

Conclusion What we have learned Where to go from here • Forced-choice experiments and • Cross-linguistic validation of RSA mixed-logit models: great for • More research on I -driven studying interpretational preferences inferences • 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 Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 11

Thank you.

References Atlas, J., & Levinson, S. (1981). It-clefts, informativeness Horn, L. R. (1984). Toward a New Taxonomy for Pragmatic and logical form: radical pragmatics (revised Inference: Q-based and R-based Implicatures. In D. standard version). Radical Pragmatics . Schiffrin (Ed.), Meaning, Form, and Use in Context Clark, H. (1975). Bridging. In Proceedings of the 1975 (pp. 11 – 42). Georgetown University Press. workshop on Theoretical issues in natural language Horn, L. R. (2004). Implicature. In L. R. Horn & G. Ward processing . Stroudsburg, PA, USA: Association for (Eds.), Handbook of pragmatics (pp. 3 – 28). Computational Linguistics. Blackwell Publishing Ltd. Frank, M. C., & Goodman, N. D. (2012). Predicting Jäger, G. (2012). Game theory in semantics and pragmatic reasoning in language games. Science , pragmatics, in C. Maienborn, P. Portner & K. von 336 (6084), 998 – 998. Heusinger (eds.), Semantics. An International Grice, H. P. (1957). Meaning. The Philosophical Review , Handbook of Natural Language Meaning, Vol. 3, 377 – 388. Berlin: de Gruyter, 2487-2516. Grice, H. P. (1975). Logic and conversation. 1975 , 41 – 58. Kehler, A., & Cohen, J. (in prep). Conversational Elicitures. Hawkins, J. A. (1991). On (in)definite articles: implicatures Levinson, S. C. (2000). Presumptive meanings: The theory and (un)grammaticality prediction. Journal of of generalized conversational implicature . MIT Linguistics , 27 (02), 405. Press. Hirschberg, J. (1985). A Theory of Scalar Implicature Prince, E. F., & Cole, P. (1981). Toward a taxonomy of (Natural Languages, Pragmatics, Inference). given-new information (pp. 223 – 255). Dissertations Available from ProQuest . Zipf, G. (1949). Human behavior and the principle of least effort. Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11 11

RSA predictions by disambiguation costs Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11

Comparing RSA implementations Q/I Resolution in Indefinite Reference Till Poppels & Roger Levy (UCSD) of 11