Preschoolers' pragma/c generaliza/ons based on scalar adjec/ve use Amanda Pogue, Michael K. Tanenhaus, & Chigusa Kurumada
University of Rochester
Problems that children face in reference comprehension: 1. Children must iden/fy that there are many ways to refer, and that some of these choices differ in the amount of informa/on that they convey 2. Need to understand what causes these differences in informa/vity 3. Need to learn when it is appropriate to aMribute these uMerance choices speaker-specifically
Evidence suggests that children can: • Iden/fy when there is not enough info provided in an uMerance by age 31 • Produce under-informa/ve uMerances at age 3, but can appropriately adjust when asked for more info2 1,2 • Children produce more appropriate referen/al expressions when given feedback to their aMempts • Can iden/fy an unreliable labeller and use this info to make judgments of the reliability of a labeller for future labeling events3,4
Experiment 1
Experiment 2
Can children use the input to detect an informa/vely unreliable speaker?
Can children update their predic/ons about future likely expressions?
*
0.75
0.50
Click on the big bucket.
Click on the bucket.
Feedback Condition (n=16)
No Feedback Condition (n=16)
Non−Modified Scalar−Modified
Non−Modified Scalar−Modified
Non−Modified Scalar−Modified
1.00
0.25
0.75
0.50
0.25
0.00
Modifying speaker (informa/ve)
0.00
Non-Modifying speaker (under- informa/ve)
Test Trial Type
- Overall effect of instruc/on type (β =0.79344, p<.05) - No effect of feedback (β = 0.01645, p > .9)
No Feedback Feedback Condition
-
Performance between groups differs (p < .006) - Feedback: differs from chance (p <.001) - No Feedback: not different from chance (p >. 5) - The feedback on this task does not tell a child why they might have chosen the incorrect item Current Direc/ons: - Looking at the amount of certainty children show on individual trials - Inves/ga/ng whether we can predic/vely code whether a child likely does or does not know the difference between the speakers
Adults
Test (12 trials): who do you think said this hint?
Scalar−Modified Trials
Test (1 trial): which speaker would you like to hear the next hint from?
Adults (No Feedback, n=32)
Elmo’s friend said, “Click on the (small) flower.” Do you think it was Dave or Sally?
1.00
1.00
0.75
0.75
0.50
0.25
0.00 0.000.00
0.25 0.50 0.25 0.75 Non−Modified Trials
Children
0.50
0.25
0.00
Feedback Condition (n=16) No Feedback Condition (n=16)
1.00 0.50 0.00 Non−Modified Trials
0.25 0.750.50 0.75 Non−Modified Trials
1.00 1.00
- A lot of individual differences, some kids approaching adult like generaliza/ons
Main Findings: 1. Feedback seems to help children discriminate speakers with respect to their informa/ve language use. 2. Successful iden/fica/on seems to play a role in later retrieval of speaker-specific informa/on 3. Children struggle to fully generalize their expecta/ons
Adults
Scalar−Modified Trials
Proportion of Responses Given for the Scalar Modifying Speaker
1.00
Feedback condiBon (n=16): correct 100% on the informa/ve speaker trials, 50% on the underinforma/ve speaker trials No Feedback (n=16): not told about the correctness of their selec/ons Exposure (20 trials):
Proportion of Responses Given for the Modifying Speaker
*
Feedback condiBon (n=25): received a s/cker if they were correct: 100% on the informa/ve speaker trials, 0% on the under-informa/ve speaker trials No Feedback (n=24): received s/ckers for correct guesses at the end of the experiment Exposure (12 trials):
Current Work & Future DirecBons: 1. Rela/onship between certainty during training and the likelihood of generalizing speaker-specifically 2. Rela/onship between own ability to produce appropriately modified referring expressions on reliability detec/on 3. Children’s knowledge on the role of adjec/ve use as a contras/ve cue
References: 1. Deutsch, W., & Pechmann, T. (1982). Social interac/on and the development of definite descrip/ons. Cogni?on, 11, 159–184. 2. MaMhews, D., Lieven, E., & Tomasello, M. (2007). How toddlers and preschoolers learn to uniquely iden/fy referents for others: A training study. Child Development, 78, 1744– 1759. 3. Koenig, M. A., & Harris, P. L. (2005). Preschoolers mistrust ignorant and inaccurate speakers. Child Development, 76, 1261–1277. 4. Scofield, J., & Behrend, D. a. (2008). Learning words from reliable and unreliable speakers. Cogni?ve Development, 23, 278–290.