do 9-month-old infants expect distinct words to refer to kinds? kathryn dewar

Post on 04-Feb-2016

20 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Do 9-month-old Infants Expect Distinct Words to Refer to Kinds? Kathryn Dewar. - PowerPoint PPT Presentation

TRANSCRIPT

Do 9-month-old Infants Expect Distinct Words to Refer to

Kinds?

Kathryn Dewar

“Imagine a language with only proper names. A new word that names a dog must refer to that particular dog and nothing else. Learning this language would require the ability to track individuals over time, but it wouldn’t require any ability to generalize, to recognize how collies are different from terriers or how dogs are different from tables.

Such languages don’t exist, of course.” (Bloom, 2000)

Early Words: Proper Names or Count Nouns?

Proper names such as Fido refer to specific individuals

Count nouns such as dog refer to kinds of individuals or object categories

Controversial issue:Whether infants interpret these early words as

count nouns that refer to kinds or as proper names that refer to individual objects

What is the assumption of an early word-learner? Does this word refer to THAT (individual) object? (PN) Does this word refer to that KIND of object? (CN)

Are Early Words Proper Names that Designate Individuals?

• Locke: – At this early stage, children know only

proper names

• Anecdotal Evidence:– Only own rubber ducky is called duck

( label is not generalized to other members of the kind “duck”)

Are Early Words Count Nouns that Refer to Kinds?There are 3 lines of research used to

support this claim:Generalization StudiesCategorization StudiesObject Individuation Studies

While these studies are suggestive, ambiguity remains

Will a newly learned word for an object be extended to other members of the object’s kind?

Woodward, Markman, & Fitzsimmons (1994)13-month-olds & 18-month-olds• Children heard a new label 9 times in a 5 min

training session & their comprehension was later assessed

• Children generalized this newly learned label to novel exemplars of the training category

• Generalized to objects that differed in colourWhat about younger children?

Are Early Words Count Nouns that Refer to Kinds?

Generalization Studies

Are Early Words Count Nouns that Refer to Kinds?

Categorization Studies Does labeling objects with novel nouns highlight object categories?

Balaban & Waxman (1997)9-month-olds “A RABBIT”

Are Early Words Count Nouns that Refer to Kinds?

Categorization Studies Does labeling objects with novel nouns highlight object categories?

Balaban & Waxman (1997)9-month-olds• Familiarized to slides of animals (rabbits) while hearing either labeling information (word condition) or tones (tone condition)• On test trials, a new exemplar from the familiar category (a rabbit) was paired with a novel animal (a pig)Infants in the word condition showed greater attention to noveltyIt seems that the noun label facilitated categorization

A Brief Detour…Object Individuation StudiesThe process that assigns segregated

objects seen on different occasions to single or multiple objects.

Example: You see a dog at the park. The dog runs into the woods and a while later, a dog runs out of the woods

Did you see the same dog or two different dogs???QuickTime™ and a

TIFF (Uncompressed) decompressorare needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.-OR-

Object Individuation StudiesXu & Carey (1996)

Occluder opened to reveal…

-OR-

UNEXPECTEDEXPECTED

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Object Individuation StudiesXu & Carey (1996)

10-month-olds: did not look longer at the unexpected vs. the expected outcome– Did not use PROPERTY/KIND information to

conclude there was two distinct objects involved

12-month-olds: looked longer at the unexpected vs. the expected event– Did use PROPERTY/KIND information for

object individuation

Back on Track… Are Early Words Count Nouns that

Refer to Kinds? Object Individuation Studies Can language assist infants in the task of object individuation?Xu (2002):

9-month-olds• What if the objects were given distinctive

noun labels with each emergence?

Object Individuation StudiesXu (2002)

Occluder opened to reveal…

-OR-

UNEXPECTEDEXPECTED

“Look, a DUCK!”

“Look, a BALL!”

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.QuickTime™ and a

TIFF (Uncompressed) decompressorare needed to see this picture.

Are Early Words Count Nouns that Refer to Kinds?

Object Individuation StudiesXu (2002)Infants succeed at 9 months if given contrastive

labelsGiving the objects noun labels facilitates object

individuation

Xu suggests that the reason language helps is because distinct words refer to different KINDS of things2 distinct noun labels = 2 kinds of objectsBut, the evidence is also consistent with the

“early words as proper names” hypothesis

Early Words: PNs or CNs? Remaining questions from the

previous research…Did infants interpret the distinct words as referring to distinct kinds of objects (TYPES) or distinct individual objects (TOKENS)

Do they expect: 2 objects (could be identical, as long as there’s

2) 2 different objects (different in property? Or

kind?)

These are the very questions we address in the current set of studies

The Interactive Bit: What are Adults Expectations?

“I see a ZAV!I see a ZAV!”

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

???

The Interactive Bit: What are Adults Expectations?

“I see a FEP!I see a WUG!”

???QuickTime™ and a

TIFF (Uncompressed) decompressorare needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Study 1Questions to be addressed…. Do infants expect distinct labels refer to

different objects? -AND- Do infants expect one repeated label refers to

identical objects“I see a ZAV!”

“I see a ZAV!”

“I see a FEP!”

“I see a DAK!”

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Study 1 Procedure:Familiarizations

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Identical ObjectOutcome

Different ObjectOutcome

Study 1 Procedure:Familiarizations

8 familiarization trials: 4 identical object outcomes 4 different object outcomes

4 kinds of object-pairs shown: Familiarization trials 5-8 are a repeat of trials 1-4 The 4 object-pairs presented during familiarization are presented during the test trials The same object-pairs are used for both phases in order to:• Diminish the novelty of the object-pairs• Give infants a sense of what is inside the box (what they can expect to see)

Study 1 Procedure:Test Trials

“I see a FEP! I see a FEP!”QuickTime™ and a

TIFF (Uncompressed) decompressorare needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

EXPECTEDoutcome

UNEXPECTEDoutcome

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

ONE LABEL

Study 1 Procedure:Test Trials

“I see a DAX! I see a WUG!”

EXPECTEDoutcome

UNEXPECTEDoutcome

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

TWO LABELS

Study 1: Experimental Design

EXPECTED UNEXPECTED

UNEXPECTED EXPECTED

ONE LABEL(Fep, Fep)

TWO LABELS(Dax, Wug)

IDENTICALOBJECTS

DIFFERENTOBJECTS

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Study 1 Results:Familiarizations

12.61sSD = 4.91

14.04sSD = 6.03

Identical Objects

Different Objects

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

n.s. t (23) = -1.14, p = .27

Study 1: Results

EXPECTED

5.41s (SD=3.57)

UNEXPECTED

11.20s (SD=10.04)

UNEXPECTED

10.02s (SD=7.86)

EXPECTED

8.38s (SD=9.53)

ONE LABEL(Fep, Fep)

TWO LABELS(Dax, Wug)

IDENTICALOBJECTS

DIFFERENTOBJECTS

(n=24)

Study 1: Results

EXPECTED

5.41sUNEXPECTE

D

11.20sUNEXPECTE

D

10.02s

EXPECTED

8.38s

ONE LABELTWO LABELS

IDENTICALOBJECTSDIFFERENT

OBJECTS

No Main Effect of number of labels (one vs two) f (1,23) = 2.58, p = .12

No Main Effect of object outcome (identical vs different)

f (1,23) = .49, p = .49Interaction between number of labels & object outcome

f (1, 23) = 5.06, p = .03

Paired-sample t-testsWilcoxon signed ranks test

t (1, 23) = -2.93, p < .01t (1, 23) = 1.13, p = .14p = .02 p = .05 (1 tailed)

Study 1What is still not yet known…

In this first study: Pairs of objects were either identical or

completely different dissimilar colour, shape, texture, material, etc

However…

For the different object pair: Are all property differences created equal?

Is any dissimilarity between the “different objects” sufficient?

Must the objects simply be different in some respect or is the way in which the objects differ important?

Why all property differences might NOT be created equal Different words are used to designate

different KINDS of things Objects differing along a property dimension

that does not effect kind membership are not given distinct labels

However, the property difference of SHAPE is closely connected with kind membership (Rosch et al., 1976; Soja, Carey & Spelke, 1991)

Generally speaking, objects that differ in shape are usually different kinds of things & are marked by different labels

Predictions… Hearing two distinct labels should suggest

two different kinds of objects in the box

Infants will look longer to the unexpected outcomes if:– The property difference between the objects

implies a difference in kind (e.g. SHAPE)

Infants will NOT look longer to the unexpected outcome if:– The property difference between the objects does

NOT imply a difference in kind membership (e.g. COLOUR)

Study 2: SHAPE

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Study 2: SHAPE

Used the same procedure as Study 1, except objects were either:Identical --or--Identical in every property but SHAPE

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

“I see a ZAV!I see a ZAV!”

“I see a FEP!I see a WUG!”

Study 2 Procedure:Test Trials

“I see a ZAV! I see a ZAV!”

EXPECTEDoutcome

UNEXPECTEDoutcome

ONE LABEL

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Study 2 Procedure: Test Trials

“I see a DAX! I see a WUG!”

EXPECTEDoutcome

UNEXPECTEDoutcome

TWO LABELS

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Study 2: Experimental Design

EXPECTED UNEXPECTED

UNEXPECTED EXPECTED

ONE LABEL(Fep, Fep)

TWO LABELS(Dax, Wug)

IDENTICALOBJECTS

DIFFERENTOBJECTS QuickTime™ and a

TIFF (LZW) decompressorare needed to see this picture.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Study 2 Results:

Familiarizations

12.93sSD = 3.75s

12.52sSD = 3.42s

Identical Objects

Different Objects

n.s.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

t (15) = .44, p = .66

Study 2: Results

EXPECTED

6.89s (SD=3.41)

UNEXPECTED

10.04s (SD=8.84)

UNEXPECTED

10.34s (SD=6.06)

EXPECTED

5.72s (SD=3.72)

ONE LABEL(Fep, Fep)

TWO LABELS(Dax, Wug)

IDENTICALOBJECTS

DIFFERENTOBJECTS

(n=16)

Study 2: Results

EXPECTED

6.89sUNEXPECTE

D

10.04sUNEXPECTE

D

10.34s

EXPECTED

5.72s

ONE LABELTWO LABELS

IDENTICALOBJECTSDIFFERENT

OBJECTS

No Main Effect of number of labels (one vs two) f (1, 15) = .47, p = .51No Main Effect of object outcome (identical vs different)

f (1, 15) = .11, p = .75Interaction between number of labels & object outcome

f (1, 15) = 8.16, p = .01

Paired-sample t-tests

t (15) = -2.14, p = .02t (15) = 2.03, p = .03(1 tailed)

Study 2 (Shape): Discussion Results mirrored those of the first study (replication)

EXPECTED

8.38sUNEXPECTED

10.02s

UNEXPECTED

11.20sEXPECTED

5.41s

ONE LABELTWO LABELS

IDENTICALOBJECTS

DIFFERENTOBJECTS

EXPECTED

5.72sUNEXPECTED

10.34s

UNEXPECTED

10.04sEXPECTED

6.89s

ONE LABELTWO LABELSStudy 1 Study 2

Infants seem to expect that:A distinct (repeated) label denotes identical objectsTwo distinct labels denote two different (shaped) objects

Different-shaped objects pairs were equivalent to completely different object pairs

What we STILL do not know… Are all property differences created equal?• Shape differences between object pairs are just

as good as using totally different kinds of objectsShape differences are salient cues to kind membership

The results of the shape study are NOT enough!• Does not demonstrate that infants expect distinct

labels to refer to distinct KINDS Need to show that another property difference (that is

independent of kind membership) does not elicit this pattern of looking

Must demonstrate that not just any property difference between the object pairs will create this result

Returning to our Predictions…

Hearing two distinct labels should suggest two different kinds of objects in the box

Infants will look longer to the unexpected outcomes if:– The property difference between the objects

implies a difference in kind (e.g. SHAPE)

? Infants will NOT look longer to the unexpected outcome if:– The property difference between the objects does

NOT imply a difference in kind membership (e.g. COLOUR)

Study 3: COLOUR

Study 3: COLOUR

Used the same procedure as previous studies, except objects were either:Identical --or--Identical in every property but

COLOUR

“I see a ZAV!I see a ZAV!”

“I see a FEP!I see a WUG!”

Study 3 Procedure:Test Trials

“I see a ZAV! I see a ZAV!”

EXPECTEDoutcome

UNEXPECTED?outcome

ONE LABEL

Study 3 Procedure: Test Trials

“I see a DAX! I see a WUG!”

EXPECTED?outcome

UNEXPECTEDoutcome

TWO LABELS

Study 3: Experimental Design

EXPECTED UNEXPECTED

UNEXPECTED? EXPECTED?

ONE LABEL(Zav, Zav)

TWO LABELS(Dax, Wug)

IDENTICALOBJECTS

DIFFERENTOBJECTS

Study 3 Results: Familiarizations

11.24sSD = 7.12

9.68sSD = 4.11

Identical Objects

Different Objects

n.s. t (15) = .92, p = .37

Study 3: Results

EXPECTED

6.83s (SD=6.04)

UNEXPECTED

6.09s (SD=4.15)

UNEXPECTED?

8.09s (SD=6.50)

EXPECTED?

10.31s (SD=8.93)

ONE LABEL(Zav, Zav)

TWO LABELS(Dax, Wug)

IDENTICALOBJECTS

DIFFERENTOBJECTS

(n=16)

Study 3: Results

EXPECTED

6.83sUNEXPECTE

D

6.09sUNEXPECTE

D?

8.09s

EXPECTED?

10.31s

ONE LABELTWO LABELS

IDENTICALOBJECTSDIFFERENT

OBJECTS

No Main Effect of number of labels (one vs two)f (1, 15) = .20, p = .66

Main Effect of object outcome (identical vs different) (p=.02)

f (1, 15) = 7.61, p = .02

6.46sSD = 4.22

9.20sSD = 6.24

No Interaction between number of labels & object outcomef (1, 15) = 1.88, p = .19

Study 3: Results This pattern of results differs from that of

Study 1 & Study 2

EXPECTED

8.38sUNEXPECTED

10.02s

UNEXPECTED

11.20sEXPECTED

5.41s

ONE LABELTWO LABELS

IDENTICALOBJECTS

DIFFERENTOBJECTS

EXPECTED

5.72sUNEXPECTED

10.34s

UNEXPECTED

10.04sEXPECTED

6.89s

ONE LABELTWO LABELS

EXPECTED

10.31sUNEXPECTED

8.09s

UNEXPECTED

6.09sEXPECTED

6.83s

ONE LABELTWO LABELS

Study 1Completely Different

Study 2Shape

Study 3Colour

Not just ANY property difference between the “different object pair” produces the expectation that the objects should be marked by distinct labels

A difference related to kind membership (SHAPE) produced the effectA difference unrelated to kind membership (COLOUR) did not

Returning to our Predictions Again…Hearing two distinct labels should suggest

two different kinds of objects in the box

Infants will look longer to the unexpected outcomes if:– The property difference between the objects

implies a difference in kind (e.g. SHAPE)

Infants will NOT look longer to the unexpected outcome if:– The property difference between the objects does

NOT imply a difference in kind membership (e.g. COLOUR)

Comparing All Three Studies Conducted a repeated-measures ANOVA with “study” as

the between-subjects factor There was a 3-way interaction (study X word X

outcome)

EXPECTED

8.38sUNEXPECTED

10.02s

UNEXPECTED

11.20sEXPECTED

5.41s

ONE LABELTWO LABELS

IDENTICALOBJECTS

DIFFERENTOBJECTS

EXPECTED

5.72sUNEXPECTED

10.34s

UNEXPECTED

10.04sEXPECTED

6.89s

ONE LABELTWO LABELS

EXPECTED

10.31sUNEXPECTED

8.09s

UNEXPECTED

6.09sEXPECTED

6.83s

ONE LABELTWO LABELS

Study 1Completely Different

Study 2Shape

Study 3Colour

Looking only at Study 1 & Study 2:– No three-way interaction (Result pattern is the same)

Looking only at Study 1 & Study 3:– Three-way interaction (Result pattern is different)

Looking only at Study 2 & Study 3:– Three-way interaction (Result pattern is different)

Familiarizations: Establishing Possible Object

Outcomes Study 1 & 2 (totally different & different-shape):– Object-pairs alternate between

identical and different kinds– Possible Outcomes: either pairs of the

same kind or pairs of different kinds of objects

– Labeling a CUE to which object outcome will be revealed

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Familiarizations: Establishing Possible Object Outcomes

Study 3 (different colour):–Different-colour pairs are seen as the same kind of objects–Here, both identical & different-colour pairs represent the SAME kind of object –Possible Outcome: pairs of the same kind of object–Labeling adds no predictive information –Looking-pattern reflects baseline preferences

Early Words: Proper Names or Count Nouns?

It’s been claimed that early nouns are:• Proper names (designate INDIVIDUALS) --as opposed to--• Count nouns that refer to object categories (KINDS)

If early labels simply pick out individuals:• You would not expect differential looking between

identical and different object outcomes (both represent 2 individuals)

• All three studies should have the same looking-time pattern

Early nouns are likely count nouns that refer to kinds

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.QuickTime™ and a

TIFF (Uncompressed) decompressorare needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Special Thanks to:Dr. Fei Xu

Laura KerlinVashti GarciaAnjula Joshi

Stephanie DenisonHenny Yeung

top related