scalar implicatures and adjectives can a decent student get into harvard? a study on gradable...

1
Scalar implicatures and adjectives Can a decent student get into Harvard? A study on gradable adjectives and scalar implicatures XPRAG2011 Barcelona Some students left the party Said: some and possibly all Implicated: some but not all It’s warm Said: It’s warm and possibly hot Implicated: it’s warm but not hot (via Quantity Maxim) IS THERE A COMMON THREAD BEHIND THE COMPUTATION OF SIs? HYPOTHESIS: the degree of contrast on the scale. The longer the distance between an element and the top of the scale, the more likely the upper-bound is to be drawn Self-paced reading experiment (subj = 23, items = 24 + 30 fillers) Andrea Beltrama, Ming Xiang University of Chicago Summary Ward et al. (2008) -Experimental study on 4 different kinds of scales: Cardinals, Ranked orders, Quantifiers/Modals and Adjectives Is it possible to give a unified account of how SIs are triggered by these types of scales? -The rate of SIs associated with adjectives is context sensitive: boosted when other scalar terms are evoked. -the interpretation of scales is sensitive to the context and the scale type. Horn (1989): SIs are triggered by entailment scales. Hot entails warm, warm implicates not hot. Hirschberg: Sis are triggered by partially ordered sets: there must be a contextually salient criterion to rank the elements Self paced-reading task (2nd sentence word by word) and acceptability question at the end. “How sensible is the second sentence in light of the first one”? 1-5 score “Excellence scale” Discreto Bravo Eccellente (Decent) (Good) (Excellent) Weak condition Marco è uno studente discreto. Per questo è stato preso ad Harvard Marco is a decent student. That’s why he has been accepted to Harvard Intermediate condition Marco è uno studente bravo. Per questo è stato preso ad Harvard Mark is a good student. That’s why he has been accepted to Harvard. Strong condition Marco è uno studente eccellente. Per questo è stato preso ad Harvard Mark is an excellent student. That’s why he has been accepted to Harvard. Contradictory condition Marco è uno studente Per questo è stato preso ad Harvard Mark is a bad student. That’s why he has been accepted to a) Does the degree of contrast between the current term and the higher one on the scale trigger SIs? Gradable adjectives as testing ground b) How are SIs processed? 2 possibilities: - Delayed process (Bott and Noveck 2004) - Immediately computed online (Nieuwland, Ditman & Kuperberg 2010 N=no value evoked M= multiple values evoked O= one value evoked Current study Follow up study If a strong contrast facilitates implicatures, the upper bound should show up with intermediate adjectives as well both in acceptability judgments and reading times. Enrichment of the context The competion for admission to college is extremely hard. Mark is a good student. That’s why he has been accepted to Harvard Vs The competion for admission to college is extremely hard. Mark is an excellent student. That’s why he has been accepted to Harvard References: Bott, L. & Noveck, I.A. (2004). Some utterances are underinformative: The onset and time course of scalar inferences. Journal of Memory and Language, 51(3), 437-457). Doran, Ryan, Rachel Baker, Yaron McNabb, Meredith Larson, and Gregory Ward, 2008,“On the Non-Unified Nature of Scalar Implicature: An Empirical Investigation,” in International Review of Pragmatics 1:1-38. Grice, Paul, 1975, "Logic and conversation". In Cole, P. and Morgan, J. (eds.) Syntax and semantics, vol 3. New York: Academic Press. Hirschberg, Julia. 1991. A Theory of Scalar Implicature. New York, NY: Garland Publishing. Horn, Laurence R. 1984. Towards a new taxonomy for pragmatic inference: Q- and R- based implicature. Meaning, Form, and Use in Context, ed. by Deborah Schiffrin. Washington D.C.: Georgetown University Press, 11-42 Nieuwland, M.; Ditman, T. & Kuperberg, G. 2010. On the incrementality of pragmatic processing: An ERP investigation of informativeness and pragmatic abilities. Journal of Memory and Language. 63:324-346. Acceptability rating: significant difference between weak and strong and weak and middle conditions, with weak adjectives receiving a lower score. Middle and strong adjectives seem to be acceptable to the same degree. The degree of contrast, therefore, appears to be a major factor influencing SIs’ computation. Reading time at the Critical Word the strong condition receives the lowest value. Weak and middle conditions, on the contrary, require longer RTs. This suggests that middle and strong conditions, in spite of similar judgments, are not processed in the same way. Also, there seems to be some early computation going at the CW in the weak condition, suggesting that SIs are computed early in the sentence (p approaching significance for Reading time at the response: as in RTs at the CW, the middle condition requires the longer RT, followed by the weak condition. This again suggests that the middle condition may not have been processed in the same way as the strong one, i.e. informativity may have been calculated for the middle condition after all. A cceptability judgm ents 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 w eak m iddle strong contrad R T atthe C W 0 100 200 300 400 500 600 700 800 w eak middle strong contrad ms R T atthe response 0 500 1000 1500 2000 2500 3000 3500 4000 4500 w eak middle strong contrad ms

Upload: drusilla-hill

Post on 13-Dec-2015

238 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Scalar implicatures and adjectives Can a decent student get into Harvard? A study on gradable adjectives and scalar implicatures XPRAG2011 Barcelona Some

Scalar implicatures and adjectives

Can a decent student get into Harvard? A study on gradable adjectives and scalar implicatures

 XPRAG2011

Barcelona

Some students left the party Said: some and possibly allImplicated: some but not all

It’s warm Said: It’s warm and possibly hot Implicated: it’s warm but not hot (via Quantity Maxim)

IS THERE A COMMON THREAD BEHIND THE COMPUTATION OF SIs?

HYPOTHESIS: the degree of contrast on the scale. The longer the distance between an element and the top of the scale, the more likely the upper-bound is to be drawn

IS THERE A COMMON THREAD BEHIND THE COMPUTATION OF SIs?

HYPOTHESIS: the degree of contrast on the scale. The longer the distance between an element and the top of the scale, the more likely the upper-bound is to be drawn

Self-paced reading experiment (subj = 23, items = 24 + 30 fillers)

Andrea Beltrama, Ming Xiang University of Chicago

Summary 1

) A strong contrast on the scale is a factor faciliating the computation of SIs.

2) SIs seem to be computed in real time. More subjects needed to confirm this.

3)Assessing (under)informativity in relation to a scenario requires extra time.

Ward et al. (2008)

-Experimental study on 4 different kinds of scales: Cardinals, Ranked orders, Quantifiers/Modals and Adjectives

Is it possible to give a unified account of how SIs are triggered by these types of scales?

-The rate of SIs associated with adjectives is context sensitive: boosted when other scalar terms are evoked.-the interpretation of scales is sensitive to the context and the scale type.

Horn (1989): SIs are triggered by entailment scales. Hot entails warm, warm implicates not hot.

Hirschberg: Sis are triggered by partially ordered sets: there must be a contextually salient criterion to rank the elements

Self paced-reading task (2nd sentence word by word) and acceptability question at the end. “How sensible is the second sentence in light of the first one”? 1-5 score

“Excellence scale”

Discreto Bravo Eccellente (Decent) (Good) (Excellent)

Weak conditionMarco è uno studente discreto. Per questo è stato preso ad HarvardMarco is a decent student. That’s why he has been accepted to Harvard

Intermediate conditionMarco è uno studente bravo. Per questo è stato preso ad Harvard Mark is a good student. That’s why he has been accepted to Harvard.

Strong conditionMarco è uno studente eccellente. Per questo è stato preso ad HarvardMark is an excellent student. That’s why he has been accepted to Harvard.

Contradictory conditionMarco è uno studente scarso. Per questo è stato preso ad HarvardMark is a bad student. That’s why he has been accepted to Harvard.

a) Does the degree of contrast between the current term and the higher one on the scale trigger SIs? Gradable adjectives as testing ground

b) How are SIs processed? 2 possibilities:- Delayed process (Bott and Noveck 2004)- Immediately computed online (Nieuwland, Ditman &

Kuperberg 2010

N=no value evoked

M= multiple values evoked

O= one value evoked

Current study

Follow up study If a strong contrast facilitates implicatures, the upper bound should show up with intermediate adjectives as well both in acceptability judgments and reading times. Enrichment of the context

The competion for admission to college is extremely hard. Mark is a good student. That’s why he has been accepted to Harvard

Vs The competion for admission to college is extremely hard. Mark is an excellent student. That’s why he has been accepted to Harvard

References: Bott, L. & Noveck, I.A. (2004). Some utterances are underinformative:  The onset and time course of scalar inferences. Journal of Memory and Language, 51(3), 437-457). Doran, Ryan, Rachel Baker, Yaron McNabb, Meredith Larson, and Gregory Ward, 2008,“On the Non-Unified Nature of Scalar Implicature: An Empirical Investigation,” in International Review of Pragmatics 1:1-38. Grice, Paul, 1975, "Logic and conversation". In Cole, P. and Morgan, J. (eds.) Syntax and semantics, vol 3. New York: Academic Press.Hirschberg, Julia. 1991. A Theory of Scalar Implicature. New York, NY: Garland Publishing. Horn, Laurence R. 1984. Towards a new taxonomy for pragmatic inference: Q- and R- based implicature. Meaning, Form, and Use in Context, ed. by Deborah Schiffrin. Washington D.C.: Georgetown University Press, 11-42 Nieuwland, M.; Ditman, T. & Kuperberg, G. 2010. On the incrementality of pragmatic processing: An ERP investigation of informativeness and pragmatic abilities. Journal of Memory and Language. 63:324-346.

Acceptability rating: significant difference between weak and strong and weak and middle conditions, with weak adjectives receiving a lower score. Middle and strong adjectives seem to be acceptable to the same degree. The degree of contrast, therefore, appears to be a major factor influencing SIs’ computation.

Reading time at the Critical Word the strong condition receives the lowest value. Weak and middle conditions, on the contrary, require longer RTs. This suggests that middle and strong conditions, in spite of similar judgments, are not processed in the same way. Also, there seems to be some early computation going at the CW in the weak condition, suggesting that SIs are computed early in the sentence (p approaching significance for weak vs strong)

Reading time at the response: as in RTs at the CW, the middle condition requires the longer RT, followed by the weak condition. This again suggests that the middle condition may not have been processed in the same way as the strong one, i.e. informativity may have been calculated for the middle condition after all.

Acceptability judgments

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

5

weak middle strong contrad

RT at the CW

0

100

200

300

400

500

600

700

800

weak middle strong contrad

ms

RT at the response

0

500

10001500

2000

2500

30003500

4000

4500

weak middle strong contrad

ms