Psy1306 Psy1306 Language and ThoughtLanguage and Thought
Lecture 5Lecture 5
Spatial LanguageSpatial Language
Some Discussion Some Discussion PointsPoints
Language learningLanguage learning– Construction of new concepts?Construction of new concepts?
After learningAfter learning– Change in saliency of a Change in saliency of a
dimension/distinction?dimension/distinction? Second language learningSecond language learning
Language AcquisitionLanguage Acquisition
Successful comprehension of any intended Successful comprehension of any intended linguistic expression cannot be achieved without linguistic expression cannot be achieved without some commonality of thought.some commonality of thought.
Universalist view of word Universalist view of word learninglearning
Language has to Language has to be learnable.be learnable.
Languages are Languages are the way they are the way they are constrained by constrained by the cognitive the cognitive abilities of its abilities of its learners.learners.
Universalist view of Universalist view of word learningword learning Child has concept Child has concept
before the word before the word is learnedis learned
Word learning Word learning consists of consists of mapping mapping concepts to concepts to soundssounds
Concepts Concepts unchangedunchanged
Evidence for a Evidence for a universalist accountuniversalist account Pattern of word learning similar across languagesPattern of word learning similar across languages
– Common paths in vocabulary acquisition (e.g. Common paths in vocabulary acquisition (e.g. in-on-underin-on-under in spatial words)in spatial words)
– Some distinctions acquired early and with few exposures Some distinctions acquired early and with few exposures ((upup).).
Grammatical morphemes show limited Grammatical morphemes show limited crosslinguistic variation, suggesting innate crosslinguistic variation, suggesting innate limitationslimitations
Children’s errors are similar across languages Children’s errors are similar across languages suggesting common underlying conceptssuggesting common underlying concepts
Studies of infants indicate some concepts are Studies of infants indicate some concepts are available prelinguisticallyavailable prelinguistically
Ring ON Pole
Cup ON Saucer
Telephone ON Wall
Lady ON TV
Moustache ON Face
Bowerman & Choi’s Bowerman & Choi’s ChallengeChallenge
IM? UM?
AUF?
AN?
IM?
Considerable crosslinguistic variationsConsiderable crosslinguistic variations Children quickly learn language specific patternChildren quickly learn language specific pattern
By 18-months-old (if not earlier)By 18-months-old (if not earlier)
Bowerman & Choi’s Bowerman & Choi’s ChallengeChallenge Is it plausible that Is it plausible that
prelinguistic prelinguistic children have all children have all the spatial the spatial concepts of all the concepts of all the world’s languages world’s languages prior to word prior to word learning?learning?
(alternative (alternative unclear)unclear)
Alternative 1: Alternative 1: Dimensions in Dimensions in Conceptual SpaceConceptual Space
Prelinguistic children Prelinguistic children have have multidimensional multidimensional concept spaceconcept space
All natural language All natural language spatial concepts are spatial concepts are areas (or vectors) in areas (or vectors) in this spacethis space
Word learning Word learning consists of mapping consists of mapping words to these words to these vectorsvectors
Alternative 1: Alternative 1: Dimensions in Dimensions in Conceptual SpaceConceptual Space Feature theoryFeature theory In: +containmentIn: +containment Auf: Auf:
+support +support +horizontal+horizontal
(some problems (some problems with classical with classical feature theory)feature theory)
Alternative 2: Alternative 2: Linguistic DeterminismLinguistic Determinism Language learning leads to Language learning leads to
concept acquisitionconcept acquisition Appears to be Whorf’s viewAppears to be Whorf’s view How? Where do the concepts How? Where do the concepts
come from?come from?
Fodor’s challengeFodor’s challenge
What is concept learning?What is concept learning?– All explicit theories treat it as hypothesis testing All explicit theories treat it as hypothesis testing
(HT)(HT)– Collect data, form hypothesis, testCollect data, form hypothesis, test
HT presupposes that the learner has the concepts to HT presupposes that the learner has the concepts to form the hypothesisform the hypothesis
Thus concept *learning* is impossibleThus concept *learning* is impossible All concepts must be formed out of existing All concepts must be formed out of existing
conceptsconcepts– Perhaps by combining set features that are Perhaps by combining set features that are
innate or triggered by experience (classical innate or triggered by experience (classical theory)theory)
To what extent does language “map onto preexisting concepts or by its own structure lead children to create new ones? Have preverbal children learned all the major spatial relations that various languages express? Or have they learned only a subset and do languages teach them to attend to new ones they have not analyzed on their own?”
(Mandler, 1996, p. 379)
What do pre-linguistic infants know?
Empirical QuestionEmpirical Question
Looking for EvidenceLooking for Evidence
Initial Conceptual Structure & Cognitive
Predispositions
LanguageAcquisition
New Conceptual Structure & Cognitive
Predispositions
Conceptual Precursors to Conceptual Precursors to Language Language (S. Hespos & E. Spelke, 2004)(S. Hespos & E. Spelke, 2004)
Experiment with preverbal infants Experiment with preverbal infants (5 month-olds)(5 month-olds)
Investigated two forms of spatial Investigated two forms of spatial conception: conception:
1. Loose/Tight contact relation1. Loose/Tight contact relation
2. Containment/Support relation2. Containment/Support relation
Conceptual Precursors to Conceptual Precursors to Language Language (S. Hespos & E. Spelke, 2004)(S. Hespos & E. Spelke, 2004)
How does one learn a conceptual How does one learn a conceptual distinction?distinction?– In linking different events by the same In linking different events by the same
semantic phrase, speakers perceive semantic phrase, speakers perceive implicit similarities in the actions…implicit similarities in the actions…
– Or, sensitivity to conceptual distinctions Or, sensitivity to conceptual distinctions (of any arbitrary language) emerge at a (of any arbitrary language) emerge at a pre-linguistic level, and are diminished or pre-linguistic level, and are diminished or enhanced by exposure to languageenhanced by exposure to language
Conceptual Precursors to Conceptual Precursors to Language Language (S. Hespos & E. Spelke, 2004)(S. Hespos & E. Spelke, 2004)
* *
*habituate
Test
Conceptual Precursors to Conceptual Precursors to Language Language (S. Hespos & E. Spelke, 2004)(S. Hespos & E. Spelke, 2004) Experiment 3Experiment 3
– Infants habituated to tight and loose containing Infants habituated to tight and loose containing cylinderscylinders
– Using a divider separating the inserted object and Using a divider separating the inserted object and entered object, test trials would show the cylinder entered object, test trials would show the cylinder moving across, with either the lower cylinder moving across, with either the lower cylinder moving with it, or remaining stationarymoving with it, or remaining stationary
Observed longer looking times whenObserved longer looking times when– loose-habituated saw cylinders moving togetherloose-habituated saw cylinders moving together– Tight-habituated saw lower cylinder remain stillTight-habituated saw lower cylinder remain still
Conceptual Precursors to Conceptual Precursors to Language Language (S. Hespos & E. Spelke, 2004)(S. Hespos & E. Spelke, 2004)
“surprise” look longer
Conceptual Precursors to Conceptual Precursors to Language Language (S. Hespos & E. Spelke, 2004)(S. Hespos & E. Spelke, 2004) Do you agree:Do you agree: These findings suggest a linguistic influence These findings suggest a linguistic influence
on conceptual distinction is selective at best; on conceptual distinction is selective at best; they can reduce sensitivity for distinctions they can reduce sensitivity for distinctions unmarked in a language – but it unmarked in a language – but it does notdoes not produce the conceptproduce the concept
Universalist PositionUniversalist Position
Infants’ knowledge of tight and Infants’ knowledge of tight and loose, support and containmentloose, support and containment– Needham and Baillargeon 1993 Needham and Baillargeon 1993 – Baillargeon, Needham, DeVos 1999Baillargeon, Needham, DeVos 1999– Huttel and Needham 2000Huttel and Needham 2000– Aguir and Baillargeon 1999Aguir and Baillargeon 1999e.g. 5.5 months, infants know containers have e.g. 5.5 months, infants know containers have
bottoms; 8.5 months, width and compressibility bottoms; 8.5 months, width and compressibility determines fit into a container. determines fit into a container.
Criticism of Hespos & Criticism of Hespos & SpelkeSpelke(and of Baillargeon and colleague’s infant studies)(and of Baillargeon and colleague’s infant studies)
Objects used in studies are very Objects used in studies are very similar perceptually.similar perceptually.
Could children abstract the Could children abstract the spatial relation when given spatial relation when given events involving very diverse events involving very diverse types of objects?types of objects?
Lexicalizations of Lexicalizations of IN IN and and ON ON in English and in English and KoreanKorean
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Preferential Looking ParadigmPreferential Looking Paradigm 9-14 months-olds 9-14 months-olds
(PRELINGUISTIC!!!)(PRELINGUISTIC!!!) Use very different looking objectsUse very different looking objects
Set-upSet-up Scene 1-6: same relation (e.g., all loose-Scene 1-6: same relation (e.g., all loose-
containment examplescontainment examples Tests familiar vs. novel (e.g., loose-Tests familiar vs. novel (e.g., loose-
containment vs. tight-containment)containment vs. tight-containment)– Does the child look at one side more than the other?Does the child look at one side more than the other?
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003) Experiment 1 (English speakers)Experiment 1 (English speakers)
– Group 1: Tight-fitting containment Group 1: Tight-fitting containment habituationhabituation
– Group 2: Loose-fitting support habituationGroup 2: Loose-fitting support habituation
Test Group 1 & 2:Test Group 1 & 2:– Tight containment vs. Loose supportTight containment vs. Loose support
Adults were also given an oddity task, Adults were also given an oddity task, and then asked to give their reasoningand then asked to give their reasoning
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003) Results:Results:
– Observed Observed developmentaldevelopmental shiftshift in in preference across the infant age groupspreference across the infant age groups
Possibly, younger infants were continuing to Possibly, younger infants were continuing to process the scene while the older group had process the scene while the older group had completed processing the familiar scenecompleted processing the familiar scene
– Adults looked longer at the novel scene than Adults looked longer at the novel scene than the familiarthe familiar
In fact, they looked longer at first novel relation In fact, they looked longer at first novel relation than second, suggesting the second instance was than second, suggesting the second instance was already somewhat familiaralready somewhat familiar
– 78% correctly identified the oddity task, 78% correctly identified the oddity task, without reference to tight/loose fitwithout reference to tight/loose fit
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Experiment 2 (English speakers)Experiment 2 (English speakers)– Same as experiment 1, except Same as experiment 1, except
habituation featured tight and loose habituation featured tight and loose containmentcontainment
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003) Predictions for infants and adults?Predictions for infants and adults?
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003) Results of test trials:Results of test trials:
– Infants: Infants: showed a main effect for scene. All groups showed a main effect for scene. All groups (9-, 10-, 14-months) looking longer at (9-, 10-, 14-months) looking longer at familiar than the novel scene, with familiar than the novel scene, with no no development difference development difference as seen beforeas seen before
– Adults: Adults: showed no preference to either scene.showed no preference to either scene.12 out of 32 picked the “correct” odd one 12 out of 32 picked the “correct” odd one out. only 7 gave tight/loose justificationout. only 7 gave tight/loose justification
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Experiment 3Experiment 3– Subjects were now Korean-raised Subjects were now Korean-raised
infants and adult Korean speakersinfants and adult Korean speakers– Identical to experiment 2Identical to experiment 2
This study would offer contrast to This study would offer contrast to a language with the relevant a language with the relevant distinctiondistinction
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Predictions:Predictions:– Adults would succeed on both the Adults would succeed on both the
looking and oddity task, correctly looking and oddity task, correctly identifying and describing the odd identifying and describing the odd relationrelation
– Infants would behave similarly to the Infants would behave similarly to the English-environment onesEnglish-environment ones
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)
Results:Results:– Main effect for scene:Main effect for scene:
Infants looked significantly longer at Infants looked significantly longer at the familiar than novel scenes, as the familiar than novel scenes, as well as the Korean-speaking adultswell as the Korean-speaking adults
– 80% of the Korean-speaking adults 80% of the Korean-speaking adults gave correct responses in the oddity gave correct responses in the oddity tasktask
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003) Contrast experiments 2 and 3Contrast experiments 2 and 3
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)What did the study show in comparing What did the study show in comparing
English and Korean infants?English and Korean infants? Infants can abstract spatial relations Infants can abstract spatial relations
across scenes with varying objectsacross scenes with varying objects Infants show no initial cross-linguistic Infants show no initial cross-linguistic
difference, as they learn their difference, as they learn their language.language.
Understanding spatial relations: Understanding spatial relations: Flexible Infants, Lexical Adults Flexible Infants, Lexical Adults (McDonough, Choi, and Mandler, (McDonough, Choi, and Mandler, 2003)2003)What did the study show in comparing What did the study show in comparing
English and Korean adults?English and Korean adults? Adults show cross-linguistic differenceAdults show cross-linguistic difference They continue to show sensitivity to a They continue to show sensitivity to a
distinction made in their language, but distinction made in their language, but not to one not made in their language.not to one not made in their language.– E.g. English speakers insensitive to E.g. English speakers insensitive to
tight/loose.tight/loose.– (see also Hespos & Spelke)(see also Hespos & Spelke)
Hespos & Spelke’s Hespos & Spelke’s adultsadults
Analogy to Categorical Analogy to Categorical PerceptionPerception
Choi et al. have shown that 9-month-old Choi et al. have shown that 9-month-old infants have equal facility to make, e.g., infants have equal facility to make, e.g., English versus Korean spatial distinctions, English versus Korean spatial distinctions, while by 18 months they are tuned into the while by 18 months they are tuned into the local language-specific distinctions. By the local language-specific distinctions. By the time we reach adulthood, just as we find time we reach adulthood, just as we find alien language distinctions hard to hear, so alien language distinctions hard to hear, so English-speaking adults have lost the ability English-speaking adults have lost the ability to make Korean distinctions even in to make Korean distinctions even in nonlinguistic implicit categorization. nonlinguistic implicit categorization. (Levinson, p. 27)(Levinson, p. 27)
A Development StoryA Development Story
According to Bowerman and Choi, According to Bowerman and Choi, – First, infants notice different properties of First, infants notice different properties of
specific spatial situations, possibly in specific spatial situations, possibly in abstract forms; some properties embedded abstract forms; some properties embedded in context (Lego attachment)in context (Lego attachment)
– Stimulant to comparison is the redundant Stimulant to comparison is the redundant use of a word; as it is heard, the child use of a word; as it is heard, the child attempts, maybe subconsciously, to align attempts, maybe subconsciously, to align the referent situations and derive the the referent situations and derive the common linkcommon link
(Process dubbed as “Structure Alignment”)(Process dubbed as “Structure Alignment”)
A Development StoryA Development Story
– As semantic categories are formed, As semantic categories are formed, the speaker is more skilled at the speaker is more skilled at making rapid judgmentsmaking rapid judgments Korean speakers implicitly see tightness Korean speakers implicitly see tightness
of fit in comparison with previous of fit in comparison with previous experience; there is a standing experience; there is a standing readiness of judgment (Slobin, same readiness of judgment (Slobin, same volume)volume)
– Likewise, unencoded sensitivities Likewise, unencoded sensitivities diminish over timediminish over time
DiscussionDiscussion
What do you What do you think of the think of the analogy?analogy?
Kay & Kempton’s Kay & Kempton’s Naming StrategyNaming Strategy
Kay & Kempton’s Kay & Kempton’s Naming StrategyNaming Strategy
put on
put in
put on
A B
C
kkitanohta
kkita
Crosslinguistic Differences
Example of the typediscussed by Melissa Bowerman andcollaborators (Choi, DeLeon)
put on
B
kkita
put in
C
kkita
put on
A
nohta
Throw away the picture least like the other two.
Picture Selection
395 Pictures of Actions involving two objects.Either putting 2 objects together or taking apart 2 objects.
1. English speakers or Korean speakers asked to describe the pictures in their native language.
2. Then the pictures were coded for whether the relevant preposition (on or in) or the relevant verb (nohta or kkita) were used.
3. Based on the coding 10 pictures from each of three categories were selected.
Selected 10 pictures for each categoryCoding for the selected pictures
% of Korean Speakers Using “nohta” and “kkita”
Picture Description Task
0
10
20
30
40
50
60
70
80
90
100
A B CCategory
Per
cen
tag
e o
f S
s
NOHTAKKITA
% of English SpeakersUsing “on” and “in”
Picture Description Task
0
10
20
30
40
50
60
70
80
90
100
A B CCategory
Per
cen
tag
e o
f S
s
ONIN
BC
A
Nohta - Put On Kkita - Put On Kkita - Put In
Category A: On/Nohta
Category B: On/Kkita
Category C: In/Kkita
Practice:12 Practice Triads. Identical for each Ss.
Actual Test:
30C3 = 4060 possible triads. 500 triads takes 45 min to 90 min.9 Ss x 500 triads/Ss = 4500 Triads. 4500 Triads > 4060 Triads.Hence 9 Ss make 1 Meta Ss. For reliability, 2 Meta Ss per language (= 18 Ss per language).
Ss Recruited:18 Korean Ss (ESL students, grad students & spouses at Penn)18 English Ss (Penn students)
Actual Task: Which one is least like the others?
Pair of Picture ...
.
..
28 other pictures
0101...010000...00
Similarity Score = Sum of # times a 3rd picture isthrown away (0 to 28)
Cluster Analysis
Creating Similarity Matrix -- How Similar Are Two Given Pictures
Non-Overlapping Clusters for Korean SsColor CodeA: Put On/NohtaB: Put On/KkitaC: Put In/Kkita
P21
bu
lb/l
amp
P24
ea
rph
on
e/w
alkm
an
P10
rec
ord
/pla
yer
P5
icep
ack/
he
ard
P1
blo
ck/b
lock
s
P9
gin
ge
rbre
ad/t
ray
P4
bu
rger
/gri
ll
P27
bo
okm
ark
/bo
ok
P20
bo
ok/
shel
f
P28
len
s/ey
eP
22 s
lid
e/c
aro
us
el
P12
ke
y/c
hai
n
P30
fil
m/c
amer
a
P23
pie
ce/j
igsa
w
P13
bea
d/s
trin
g
P25
ke
y/k
eyh
ole
P14
ham
mer
/sti
ck
P18
bla
de/
razo
r
P17
sa
usa
ge/
stic
k
P15
rin
g/p
ole
P19
nu
t/b
olt
P29
lem
on
/gla
ss
P11
pu
pp
et/h
and
P16
rin
g/f
ing
er
P3
bo
ok
on
bo
oks
P6
tran
spa
ren
cy/O
H
P2
ute
ns
il/n
apki
nP
7 m
ou
se/
pad
P8
cher
ry/i
ce c
ream
P26
den
ture
/mo
uth
17.9
35.7
53.6
71.4
89.3
Cluster 1 Cluster 22-1 2-2
100
.0
Similarity
P10
rec
ord
/pla
yer
P5
icep
ack/
he
ard
P1
blo
ck/b
lock
s
P9
gin
ge
rbre
ad/t
ray
P4
bu
rger
/gri
ll
P27
bo
okm
ark
/bo
ok
P20
bo
ok/
shel
f
P28
len
s/ey
e
P22
sli
de
/ca
rou
sel
P12
ke
y/c
hai
nP
30 f
ilm
/cam
era
P23
pie
ce/j
igsa
w
P13
bea
d/s
trin
gP
21 b
ulb
/lam
p
P25
ke
y/k
eyh
ole
P14
ham
mer
/sti
ckP
18 b
lad
e/ra
zor
P17
sa
usa
ge/
stic
kP
15 r
ing
/po
leP
19 n
ut/
bo
ltP
29 l
emo
n/g
lass
P11
pu
pp
et/h
and
P16
rin
g/f
ing
er
P3
bo
ok/
bo
ok
sP
6 tr
ansp
are
ncy
/OH
P2
ute
ns
il/n
apki
nP
7 m
ou
se/
pad
P8
cher
ry/i
ce c
ream
P26
den
ture
/mo
uth
P24
ea
rph
on
e/w
alkm
an
17.9
35.7
53.6
71.4
89.3
Non-Overlapping Clusters for English Ss
Cluster 1 Cluster 22-1 2-2
Color CodeA: Put On/NohtaB: Put On/KkitaC: Put In/Kkita
Similarity
Description
For ABC Triads, % Pictures Thrown Out
0
10
20
30
40
50
60
70
A B CCategory
Per
cen
tag
e o
f P
ictu
res
Korean Speakers
Triad Task
English Speakers
0
10
20
30
40
50
60
70
A B CCategory
Per
cen
tag
e o
f P
ictu
res
For ABC Triads, % Pictures Thrown Out
% of Korean Speakers Using “nohta” and “kkita”
Picture Description Task
0
10
20
30
40
50
60
70
80
90
100
A B CCategory
Per
cen
tag
e o
f S
s
NOHTAKKITA
% of English SpeakersUsing “on” and “in”
Picture Description Task
0
10
20
30
40
50
60
70
80
90
100
A B CCategory
Per
cen
tag
e o
f S
s
ONIN
ppayta
A
take offppayta
B
take out
Complaint by some of the first study.“nohta”-put on is VERY different than “kkita”-put on. English speakers of course can see this big difference, and so act like Korean speakers.
Try replication with:
kkenayta
C
take out
Sample pictures selected based on the rating.
Category A: Off/Ppayta
Category B: Out/Ppayta
Category C: Out/Kkenayta
1-10: Off/Ppayta11-20: Out/Ppayta21-30: Out/Kkenayta
Clusters for English Ss
P1
bead
/str
ing
P2
ring/
finge
r
P3
key/
ring
P4
shea
th/k
nife
P5
dres
s/ha
nger
P6
pin/
tie
P7
cup/
cup
P8
cap/
cap
P9
Thi
mbl
e/th
umb
P10
hai
r ba
nd/h
ead
P11
see
d/ap
ple
P12
key
/key
hole
P13
plu
g/so
cket
P15
ear
phon
e/w
alkm
an
P16
pen
cil/s
harp
ener
P17
den
ture
/mou
thP
18 c
onta
ct le
ns/e
ye
P14
cor
k/ b
ottle
P19
boo
kmar
k/bo
ok
P20
nee
dle/
arm
P22
sod
a/sh
elf
P23
boa
t/ tu
b
P21
laun
dry/
was
her
P24
lette
r/m
ailb
ox
P25
chi
cken
/pan
P26
tom
ato/
bow
l
P27
fold
er/c
abin
et
P28
egg
/nes
tP
29 O
nion
/bag
P30
clu
b/go
lf ba
g
Cluster 1ECluster 2E
2E-1 2E-21E-2
1E-1
18.3
45.5
72.8
100.
0
Similarity
Clusters for Korean Ss
P9
Thi
mbl
e/th
um
b
P1
bea
d/st
ring
P2
ring
/fin
ger
P3
key/
ring
P4
shea
th/k
nife
P6
pin/
tieP
5 dr
ess
/han
ger
P7
cup/
cup
P8
cap/
cap
P10
ha
ir b
and/
head
P20
ne
edle
/arm
P11
see
d/a
ppl
e
P12
key
/ke
yhol
e
P15
ea
rpho
ne/w
alkm
an
P17
de
ntu
re/m
outh
P19
bo
okm
ark/
book
P13
plu
g/so
cket
P16
pe
ncil/
shar
pene
r
P18
con
tact
lens
/eye
P14
cor
k/ b
ottle
P21
lau
ndry
/ w
ashe
rP
24 le
tter
/ma
ilbox
P25
chi
cken
/pan
P29
On
ion/
bag
P22
sod
a/s
helf
P23
bo
at/
tub
P26
tom
ato/
bow
lP
27 f
olde
r/ca
bin
et
P28
eg
g/n
est
P30
clu
b/go
lf ba
g
Cluster 2K
Cluster 1K
1K-1 1K-2(same as 1E-2)
18.3
45.5
72.8
100
.0Similarity
1-10: Off/Ppayta11-20: Out/Ppayta21-30: Out/Kkenayta
Casasola et al. Casasola et al.
Could children abstract the spatial Could children abstract the spatial relation when given events involving relation when given events involving very diverse types of objects?very diverse types of objects?– Relations:Relations:
Containment Containment from both tight-in and loose-in from both tight-in and loose-in eventsevents
Tight-fit Tight-fit from both tight-on and tight-in from both tight-on and tight-in eventsevents
SupportSupport from both tight-on and loose-on from both tight-on and loose-on eventsevents
StimuliStimuli 3 Sets:Containment eventsTight-fit eventsSupport events
*Groups Habituated on Different Events, but tested on Same Events.
By Spatial Relationship: By Spatial Relationship: English speaking childrenEnglish speaking children
10-months-olds 18-months-olds
**
“Categorize” by containment only (not support or tight-fit) at both ages.“Categorize” in the sense of children looking longer to novel relation over old one.
Casasola & Bhagwat Casasola & Bhagwat (in press)(in press)
* *
Korean 10 months-old = English 10 months-olds: categorize by containment18 months-old Korean children <> English 18 months-old:
• K children categorize by tight-fit, and E children don’t• E children categorize by containment still, K children don’t
Same study as Casasola & Cohen, except with Korean children habituated on containment and support.
Casasola & Bhagwat Casasola & Bhagwat (in press)(in press) Now test English speaking 18 months-olds Now test English speaking 18 months-olds
with new paradigm, but with a novel word with new paradigm, but with a novel word “Toke”“Toke”During HabituationDuring Habituation– See: tight-fit eventsSee: tight-fit events– ““Look. She put it toke. See? She put it toke.”Look. She put it toke. See? She put it toke.”
During TestDuring Test– ““Look at that.”Look at that.”
18 months-old English speaking children 18 months-old English speaking children now categorize by tight-fitnow categorize by tight-fit
Acquisition of Acquisition of Second Language Spatial Second Language Spatial SemanticsSemantics
Edward Munnich’s DissertationEdward Munnich’s Dissertation
ENGLISH
SPANISHAANDUTCH
FINNISH
ON IN -LLA -SSA
ENINOP
First Stream: Crosslinguistic variation (Bowerman, 1996)
Second stream: Early Second stream: Early commitmentcommitment
• Early in life, we commit to L1 structure• Difficult to master L2 structure later in life
• Phonology (e.g. Werker, 1994; Kuhl et al., 1992)
• Morphosyntax (e.g. Johnson & Newport, 1989)
Hypotheses for spatial Hypotheses for spatial semanticssemantics
Continued Access: Learned at any time in life
Early Commitment: Optimally learned early in life
Experiment
• Question - Early commitment in spatial semantics?
• Approach - Examine spatial semantics of L2 learners,
varying in Age of Acquisition (AOA)
- Compare to morphosyntax, using Johnson & Newport’s (1989) methods
Participants
• • Control: 20 adult native English Control: 20 adult native English speakersspeakers
• Experimental: 60 adult non-native • Experimental: 60 adult non-native English speakersEnglish speakers
AGE OF ACQUISITION OF
ENGLISH
FIRST
LANGUAGE
Early
(0-7)
Mid
(8-13)
Late
(14+)
Korean 10 10 10Spanish 10 10 10
Design
• • Morphosyntax Morphosyntax replicationreplication
• Spatial Semantics • Spatial Semantics - Open Ended- Open Ended- Ratings- Ratings
Morphosyntax StimuliJohnson & Newport (1989)
Task: Judge grammaticality of sentences Task: Judge grammaticality of sentences
1(a) The bird has fallen from the oak 1(a) The bird has fallen from the oak tree tree (b) * The bird has fall from the oak (b) * The bird has fall from the oak treetree
2(a) The women played basketball2(a) The women played basketball (b) *The women played the (b) *The women played the basketballbasketball
Rules tested:Rules tested:• auxiliaries • auxiliaries • subcategorization• subcategorization• determiners • determiners • yes/no questions• yes/no questions• plurals • plurals • word order• word order
Spatial Semantic Stimuli
• • 20 contrasts20 contrasts10 “on” vs. “in”,10 “on” vs. “in”,5 “on” vs. “over”, 5 “on” vs. “over”, 5 “on vs. “under”5 “on vs. “under”
• 4 sentences per contrast• 4 sentences per contrast
Total: 80 sentencesTotal: 80 sentences
Open Ended Spatial SemanticsFill in the blanks with spatial terms
There is a blue ball __ the container. There is a blanket __ the man.
Spatial Semantic RatingsRate uses of spatial terms
Response: There is a blue ball in the container.
Model A: There is a blue ball on the container.Model B: There is a blue ball in the container.
There is a blanket on top of the man.
There is a blanket over the man.There is a blanket on the man.
Spatial Semantic RatingsScoring: Example 1
Given
Open Ended Response: There is a blue ball in the container.
Ratings
Model Response Code Answer CorrectA: on worse 0 0 YESB: in just as good 1 1 YES
Spatial Semantic RatingsScoring: Example 2
Given
Open Ended Response: There is a blue ball in the container.
Ratings
Model Response Code Answer CorrectA: on better 1 0 NOB: in just as good 0 1 NO
Morphosyntax: Early Morphosyntax: Early commitmentcommitment(Johnson & Newport, 1989)(Johnson & Newport, 1989)
Pro
port
ion
Cor
rect
Age of Acquisition
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
r= - 0.77
Replication in morphosyntaxReplication in morphosyntax
Pro
port
ion
Cor
rect
Johnson and Newport (1989) Present Experiment
Age of Acquisition
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
Korean r= - 0.81
Spanish r= - 0.76
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
r= - 0.77
Open Ended Spatial Open Ended Spatial SemanticsSemantics
Pro
port
ion
Cor
rect
Morphosyntax
Age of Acquisition
Korean
Spanish
r= - 0.76
Open Ended
Korean
Spanish
r= - 0.46
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
Spatial Semantic RatingsSpatial Semantic Ratings
Pro
port
ion
Cor
rect
Morphosyntax
Age of Acquisition
Korean
Spanish
r= - 0.76
Ratings
Korean
Spanish
r= - 0.41
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 5 10 15 20 25 30 35 40
early latemid
AOA effects in spatial AOA effects in spatial semanticssemantics
1. People relative to object-regions vs. supporters
• e.g. “in the seats” vs. “on the sofa”
AOA effects in spatial AOA effects in spatial semanticssemantics
2. Indentations vs. images relative to surfaces:
• e.g. “crack in” vs. “paint on pavement”:
AOA effects in spatial AOA effects in spatial semanticssemantics
3. Top members vs. other members relative to set:
• e.g. “red book on stack” vs. “in stack”
AOA effects in spatial AOA effects in spatial semanticssemantics
4. Locations relative to political vs. geographic entities:
• e.g. “airport in the city” vs. “on the island”
AOA effects in spatial AOA effects in spatial semanticssemantics
5. Substances higher than vs. touching top surface
• e.g. “chocolate over” vs. “on the orange”
AOA effects in spatial AOA effects in spatial semanticssemantics
6. People relative to mass transit vs. small vehicles:
• e.g. “on the ferry” vs. “in the canoe”
Lack of effects in spatial Lack of effects in spatial semanticssemantics
1. Objects contained by vs. supported by containers
• e.g. “ball in” vs. “on the container”
Lack of effects in spatial Lack of effects in spatial semanticssemantics
2. Objects folded vs. spread underneath other objects
• e.g. “foil under1” vs. “under2 the bowl”
Returning to hypothesesReturning to hypotheses
• Continued Access: Learned at any time in life
• Early Commitment: Optimally learned early in life
Conclusion: Early commitment in spatial semantics, as in morphosyntax and phonology.
ColorColor