Upload
abel
View
15
Download
0
Tags:
Embed Size (px)
DESCRIPTION
TVJT : DISTRIBUTION OF SUBJECTS AS A FUNCTION OF AGE and YES RESPONSES. NUMBER OF TIMES CRITICAL STATEMENTS WERE ACCEPTED. Did the puppet say it WELL?. NELS 35. SCALAR IMPLICATURES IN CHILDREN: FAILURES OR SKILLFUL STRATEGIES? - PowerPoint PPT Presentation
Citation preview
Sentence Evaluation Task (SET): SOME ELEPHANTS HAVE TRUNKS
Do you agree?
CONCLUSIONS so far:
CRITICAL AGE FOR EMERGENCE OF SI:
at 6 children are like adults, deriving SI “not all”;
at 5 only half of the children are adult-like
TASK INFLUENCES SUBJECTS’ ANSWERS:
more “logic” responses in children and adults if non natural setting
BIMODAL DISTRIBUTION OF SUBJECTS:
subjects are consistent in their answers
CHILDREN PERFORMANCE DEPENDS ON SCALE:
SI related to discrete scales emerge earlier
0
0
5
SET: DISTRIBUTION OF CHILDREN AFTER TRAINING
NUMBER OF TIMES CRITICAL STATEMENTS WERE ACCEPTED
10
15
SUBJECTS
0 1 2 3 4 5
Fig. 3
SCALAR IMPLICATURES IN CHILDREN: FAILURES OR SKILLFUL STRATEGIES?
Francesca Foppolo & Maria Teresa Guasti - University of Milano-Bicocca
For correspondance: [email protected]
What is said [literal meaning]: SOME (= at least one), TWO (= at least two)
What could have been said instead [alternatives] : ALL, THREE
Scales: <some, all>; < n, (n+1)>, <, > where is informationally stronger than
Make your contribution as informative as it’s required (Maxim of Quantity)
What is conveyed [scalar implicature] : NOT ALL, NOT THREE
A: Some linguists work in MilanB: Where do the other linguists work?
some but not all linguists work in MilanC: Lyn has 2 childrenD: I think she wants a third
Lyn has exactly 2 children
MECHANISM OF SI COMPUTATION (Chierchia, 2002)
SIs are part of the recursive interpretation of a sentenceFor any expression with a scalar item, the strengthened interpretation is computed by adding an implicature(=negation of stronger alternatives) to its plain valuePlain/scalar value are comparedSI is adopted, only if it leads to a more informative statement
SCALAR IMPLICATURES (SI): examples & derivation
DISTRIBUTION OF CHILDREN
1 3 4 5 0 1 2 3 4 5
SUBJECTS
SET: DISTRIBUTION OF 7 YEAR-OLD-CHILDREN BEFORE TRAINING15
10
5
0 NUMBER OF TIMES CRITICAL STATEMENTS WERE ACCEPTED
Fig.2
Making the experimental goals clear
Training: A friend of mine calls this
AN ANIMAL
BUT... there is a better way to describe it:
THIS IS A PIG.
RESULTS 1 (SET)
CHILDREN (7 YEARS OLD) ARE MORE LOGICAL THAN ADULTS (Some = some even all)
TRAINING HAS AN EFFECT: CHILDREN BECOME LIKE ADULTS
ADULTS’ DERIVATION OF SI: 50%
Results: compare Fig.3 with Fig. 1&2EXPERIMENT 1 - Guasti et al. (2004): Replication of Noveck (2001)
5 year old children – 3 scales: <some, all>, <two, three>, <start, finish>Derivation of SI by children:DISCRETE SCALE (65%) LOGIC SCALE (12,5%)ASPECTUAL SCALE (10%)
Interestingly, subjects held abimodal distribution:6/10 always rejected the critical statements3/10 always accepted them(1/10 behave at chance)
SCALE matters
Papafragou & Musolino (2004) – testing different scales
Critical trial: an example
Exp: What’s happening in the story? Puppet: SOME OF THE SMURFS ARE GOING ON A BOAT
NUMBER OF TIMES CRITICAL STATEMENTS WERE ACCEPTED
TVJT: DISTRIBUTION OF SUBJECTS AS A FUNCTION OF AGE and YES RESPONSES
Did the puppet say it WELL?RESULTS 2 (TVJT)
CHILDREN AT 6 & 7 DERIVE SI AS MUCH AS ADULTS:
interestingly, adults’ & 7 year olds’ performance improved in this task: 9/19 adults & 16/18 children interpreted “some” logically in
SET, while only 1/12 adults and 2/15 did so in TVJT
ONLY 50% OF THE CHILDREN AT 4 & 5 DERIVE SI:
interestingly, they don’t behave at chance: half consistently accept, and half consistently reject the critical statements
AGE & TASK affect subjects’ performance
EXPERIMENT 2 – Developmental study using the Truth Value Judgment Task
Subjects N. Age15 712 612 512 412 Adults
Material 5 critical stories
6 story-fillers
Procedure Truth Value Judgement Task
DO CHILDREN DERIVE SCALAR
IMPLICATURES?
SET: DISTRIBUTION OF ADULTS (no training)15
10
5
0
0 1 2 3 4 5
NUMBER OF TIMES CRITICAL STATEMENTS WERE ACCEPTED
Fig.1
SUBJECTS
TURN PAGE
NELS 35
PRAGMATIC FAILURES
OR RESPONSE
STRATEGIES?
SET: DISTRIBUTION ADULTS(NO TRAINING)
Which features of the experimental design adopted in all the experiments so far could have encouraged the use of a strategy in responding? different groups of subjects were tested on different scales separately, so that each subject was tested on the same kind of statement 4 times in the course of the experiment (in P&M)
the critical stories had all the same underlying structure and the same outcome (some children remarked this similarity across stories, despite the presence of the fillers in between)
this fact could contribute to make the whole situation “artificial”, putting pragmatic norm aside
Significant effect of AGE (F(1, 78) =28.85, p<.0001) & ITEM (F(3, 234)=5.54, p<.001) and interaction among these factors (F(3, 234)=6.89, p<.0001)Children derivation of SI: two=97,5% significantly higher than other items: a piece of = 62,5% (p=.0003) some (subj.) = 70% (p=.006); some (obj.) = 75% (p=.03) (difference between some subj./obj. position: n.s.)Children performance on control items: above 92% for all the items except a piece of (p<.006): 80% correct
Derivation of SI
30
39
2825
40393940
0
5
10
15
20
25
30
35
40
1 2 3 4
Scalar items
Num
. of s
ubj.
that
der
ive S
I
Children
Adults
A piece of
Some (subj) Some (obj)Two
INTERESTINGLY: Subjects do not split in 2 groups anymore:
not necessarily the answer given to the 1st scalar item is reiterated for the others
5 year-olds performance with SI improve a lot in this experiment, compared to the
results obtained by P&M with the same task (but different design):
65% vs 97,5% for TWO;
12% vs 72,5 % for SOME
FINAL DISCUSSION Individual variability across items indicate that subjects were answering according to the item, as they normally do in conversations, and not resorting to a strategy The emergence of the ability in deriving SI is linked to the scalar item involved: different scales may be lexicalised at different stages in developmentHigh performance on controls indicate that children know the scalar items: this is not enough for SIYounger children are more sensible to the “anomalies” of the experimental setting and less ready to detect the “rules” of the game
RESULTS
Subjects n. 40 5-year-old children (Age range: 4,11-5,11; Mean Age: 5,4; SD: ,15) n. 40 Adults control
Materials 16 critical trials = underinformative statements containing different scalar items (in contexts that verify “three”, “all”, “whole” respectively):4 items with TWO – like Two Smurfs went on a boat 8 items with SOME OF - 4 items like Some of the clowns went fishing
4 items like The dwarf picked up some of the carrots4 items with A PIECE OF – like Cinderella decorated a piece of the tree32 control trials - to check the understanding of the scalar items involved in the inference
Procedure TVJT4 conditions were created, 12 statements each: 8 control + 4 critical
statements, one for scale 10 subjects were randomly assigned to each condition so that each subject was shown only one occurrence of each target itemto avoid a parallelism across stories, the structure of the stories used to test different scales varied a lot, and the same initial structure was used to test
different statements (with different outcomes)
1 characters+ 3/5 objects
5 characters+ 2 alternatives
this allowed a greater variability of the material presented to each subject and a comparison of performance across different scalar items WITHIN the same subject
The dwarf picked up SOME OF the carrotsUnderinformative
Batman boughtSOME OF the pearsTrue
Baloo closed TWO binsTrue
The owl lighted up TWO candlesFalse
ALL the soldiers rode a horseFalse
TWO clowns went fishingUnderinformative
SOME OF the Smurfs went by boatUnderinformative
ALL the monkeys had a biscuitTrue
A different experimental design: BLOCKING STRATEGIES
ReferencesBrain, M. and B. Roumain (1981). Children’s comprehension of “or”: Evidence for a sequence of competencies. Journal of Experimental Child Psychology 31:46-70.Chierchia, G., S. Crain, M. T. Guasti, A. Gualmini and L. Meroni (2001). The Acquisition of Disjunction: Evidence for a Grammatical View of Scalar Implicatures. In Proceedings of the 25th Boston University Conference on Language Development. Sommerville, Cascadilla Press.Chierchia, G. (2002) Scalar Implicatures, Polarity Phenomena, and the Syntax/Pragmatics Interface. Manuscript, University of Milan-BicoccaChierchia, G., M. T. Guasti, A. Gualmini, L. Meroni, S. Crain, F. Foppolo (2004). Semantic and pragmatic competence in children and adults’ comprehension of “or”. In Experimental Pragmatics, eds I.Noveck and D. Sperber, Palgrave.Crain, S. and R. Thornton (1998). Investigations in Universal Grammar. A Guide to Experiments on the Acquisition of Syntax and Semantics. The MIT Press, Cambridge, MA.808Foppolo F., Guasti M. T., Chierchia G. (2004) Pragmatic inferences in children’s comprehension of scalar items. Talk presented at Second Lisbon Meeting on Language Acquisition Lisboa, 1-4 June 2004Grice, P. (1975). Logic and Conversation. In P. Cole and J. Morgan (eds.) Syntax and Semantics 3: Speech Acts. Academic Press, New York. Grice, P. (1989) Studies in the Way of Words. Harvard University Press, Cambridge, MA.Guasti M. T., Foppolo F., Chierchia G. (2004) (in prep.) Scalar Implicatures in Child Language: failures or skilful strategies? Guasti M. T., Chierchia G., Foppolo F., Gualmini A., Meroni L.(2004) Why Children and Adults Sometimes (but not always) Compute Implicatures.To appear in Language and Cognitive ProcessesGuasti M. T. (2002) Language Acquisition. The Growth of Grammar. MIT Press. Cambridge, MA.Hirschberg, J. (1985). A Theory of Scalar Implicatures. Doctoral dissertation, University of Pennsylvania, Philadelphia, PA.Horn, L. (1972). On the Semantic Properties of Logical Operators in English. Doctoral Dissertation, UCLA, CA. Levinson, S. (2000) Presumptive meaning. Cambridge, MA :MIT Press.Noveck, I. (2001). When children are more logical than adults: Experimental investigations of scalar implicatures. Cognition, 78, 165-188.Papafragou A. and J. Musolino (2003) Scalar Implicatures at the Semantic-Pragmatics Interface. Cognition, 80, 253-282.Smith C. L. (1980) Quantifier and question answering in young children. Journal of Experimental Child Psychology, 30:191-205.e-mail: [email protected] 35 - Storrs, Oct 22-24 2004