Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

PL4221

No description
by

Ashley Tong

on 16 September 2014

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of PL4221

2.5 year olds use
cross-situational consistency
to learn
verbs
under
referential uncertainty

Introduction
Experiment 1
General Discussion
Experiment 2
HYPOTHESIS
PARTICIPANTS
APPARATUS
STIMULI
Can infants identify the referents of
TRANSITIVE
verbs?

36 children - 18 & 18
2.5 years old
Native English speakers
Same as experiment 1
vs
PROCEDURE
Participants split into high/low vocabulary groups using MCDI
Random assignment -> List & Verb order
RESEARCH QUESTIONS
PARTICIPANTS
SET-UP
STIMULI
36 children - 18 & 18
2.5 years old
Native English speakers
Split into high/low vocabulary groups

Can 2.5 year olds
1) Attach
referential information
to
novel
INTRANSITIVE verbs

when multiple potential referents are present



2) Use
cross-situational consistency
to identify each verb’s intended referent

PROCEDURE
Analysis/Critique
strengths
weaknesses
THOUGHT QUESTIONS
1) What underlying issues could account for the difference in learning capabilities between high and low vocabulary children in word learning?
2) Other than preferential looking, what methodology could possibly be used to test for true verb learning for children of this age?
3) Would bilingual children perform differently in disambiguating referential uncertainty when learning new words? How? And why?

recommended readings






20"
12"
30"
Learning from mistakes
Interobserver reliability
Use of MCDI
Limited generalizability
Limited validity
Clear, systematic justification
- Using of concrete action verbs
- Age group to test verb learning
- Timing of analysis windows
- Using pre-controlled timings in Experiment 2
Simplifies control and standardization
colour video
4 female actors
4 toys
4 novel verbs - pimming, tazzing, nading, rivving
12 unique tokens (4 actors X 3 actions)


}
actor-toy pairing consistent throughout
- 2 independent coders
MCDI = Parental report of productive vocabulary
- Experiment deals with comprehension
- Potential social desirability bias (over-reporting)
EXPERIMENTAL LIST
Alternative: Peabody Picture Vocabulary Test (PPVT)
- Objective assessment of receptive vocabulary
- Overly simplified compared to real life situations
- Preferential looking paradigm to judge identification
- Additional test for true learning of verb needed
Summary of Key findings
1
Experiment 1:
Intransitive Verbs
2
Experiment 2:
Transitive Verbs
High Vocabulary
2.5 Year Olds
Low Vocabulary
2.5 Year Olds
Pimming
Nading
Tazzing
Rivving
Slide "S" on table
Lift and lower
Toss between hands
Spin on table
Spin on table
Toss between hands
Lift and lower
Slide "S" on table
*
To ensure there were no systematic differences between lists (children are not biased to certain actions/verbs)
Conclusions
2.5 year olds possess basic mechanism to exploit cross-situational consistency in novel verb learning
Limits on children's ability to encode/compare under referential ambiguity (low-vocab children in Exp. 2)
Adapted from Smith & Yu's (2008) paradigm

Create lexical entry for new verb
6 novel object labels

Lexical items in isolation


Separate training & test trials

Did not differentiate children of different language abilities
Attach referential information about possible candidates
4 novel action verbs

Novel verbs presented in sentences

Continuous study-test procedure


Sought to explore relationship between children's productive vocabulary & task performance
Retrieve and update information across trials + actor variation
SMITH & YU CURRENT STUDY
Identify verb-action pairings
- Difficulty occurred despite minimal referential ambiguity
Simplification from real life + source of implicit contrast
- May require additional sources of information for learning
Prior exposure to phonological form + syntactic properties (Swingley, 2007)
Implications for Research
Complements Past Literature
Future Research
Using cross-situational consistency to identify referents of novel object names under referential uncertainty (Smith & Yu, 2008)


Ability to store multiple types of information regarding novel verbs in created lexical representations (Yuan & Fisher, 2009)


Process of fast mapping and extended mapping (Carey, 1978)
Mechanism underlying cross-situational learning
True cross-situational learning
Conjecture-based cross-situational learning
Learning Points
Word learning differs depending on several factors
Word category (noun vs verb)
Word imageability (concrete vs abstract)
Level of vocabulary proficiency (high vs low)
Additional cues (syntactic, phonological)
Cross-situational consistency is one source of verb learning
Can be combined with distributed exposure to enhance word learning, especially for younger/lower-vocabulary group children
Coloured videos of 4 female actors
Novel solo actions paired with novel intransitive verb
12 unique tokens
created from 4 actors X 3 actions each
Audio that narrates sentences containing novel intransitive verbs
pimming nading rivving tazzing
ANALYSES
RESULTS&DISCUSSION
Lany, J., & Saffran, J. (2011). Interactions between statistical and semantic information in infant language development. Developmental Science, 14(5), 1207-1219.
Lany, J., & Saffran, J. (2011). Interactions between statistical and semantic information in infant language development. Developmental Science, 14(5), 1207-1219.
- Examines potential underlying mechanism behind difference in how high and low vocabulary children learn words
Results
Low Vocabulary Children
High Vocabulary Children
Successfully generalized when given
phonological cues
to new words' potential referents
Successfully used
distributional cues
to generalize, no evidence of using phonological cues
Dittmar, M., Abbot-Smith, K., Lieven, E., & Tomasello, M. (2011). Children aged 2 ; 1 use transitive syntax to make a semantic-role interpretation in a pointing task. Journal of Child Language, 38(5), 1109-1123.
Dittmar, M., Abbot-Smith, K., Lieven, E., & Tomasello, M. (2011). Children aged 2 ; 1 use transitive syntax to make a semantic-role interpretation in a pointing task. Journal of Child Language, 38(5), 1109-1123.
Criticism of preferential looking task
Child might focus on target event initially
Child gets bored
Child spends more time on distracter event instead
Mentions a study where children aged 30 months looked at familiar verbs at chance level, but looked at novel verbs above chance level
Results with pointing paradigm:
Children aged 25 months: Pointed significantly above chance at target for novel verbs only if final trial excluded (fatigue)
Children aged 31 months: Pointed significantly above chance at targets for both novel and familiar verbs (no effects of fatigue)
definitions:
VERBS
NOUNS
- action=event
- thing
NOUN PHRASE
NOUN PHRASE ARGUMENT
CONCRETE
ABSTRACT
: can be observed
(e.g. throw/punch)
: cannot be observed
(e.g. miss)
TRANSITIVE
INTRANSITIVE
: act upon something
- typically followed by noun/noun phrase that it is acting upon in statements
(e.g. I miss you/Punch the pillow)
: do not act upon something
- not followed by a direct object
(e.g. sleep/arrive)
Smith, L., & Yu, C. (2008). Infants rapidly learn word-referent mappings via cross-situational statistics.
Cognition, 106
, 1558-1568.
Yuan, S., & Fisher, C. (2009). "Really? She blicked the baby?": Two-year-olds learn combinatorial facts about verbs by listening.
Psychological Science, 20
, 619-626.
Carey, S. (1978). The child as word leaner. In M. Halle, J. Bresnan, & G. A. Miller (Eds.),
Linguistic theory and psychological reality
(pp. 264-293). Cambridge, MA: MIT Press.
: a phrase that consist of at least a noun
- can be a single noun only
: any of the noun phrases in a clause that are related directly to the verb
Swingley, D. (2007). Lexical exposure and word-form encoding in 1.5 year-olds.
Developmental Psychology, 43
, 454-464.
I miss you. (2 N-P Argument)
I sleep. (1 N-P Argument)
question:
How do children cope with
referential ambiguity
?
researchers' thoughts:
Additional
referential contexts
taken into account
Relevance =>
Cross situational consistency
=> Meaning
RESEARCH FINDINGS:
(GILLETTE, GLEITMAN, GLEITMAN & LEDERER, 1999)
HUMAN SIMULATIONS OF VOCABULARY LEARNING
QUESTION:
Can adults use cross-situational consistency to identify referents of words?
EXPERIMENT:
Participants shown video clips of mother-infant interaction
Video clips contain a common noun/verb [target word] uttered by mother
Original video clips:
with soundtrack
Edited video clips:
soundtrack removed
[target word] replaced by tone
A series of video clips were shown
Cross situational observation
Task: Guess referent of [target word]
Performance for verbs increased drastically when adults were given sentence structures in which verbs occurred.
(Gillete et al., 1999; Snedeker & Gleitman, 2004)
Similar results for 7 y/o children.
(Piccin & Waxman, 2007)
Adults managed to guess
45% of nouns
15% of verbs
IMPLICATIONS:

RESULTS:
Scene observations
are systematically less useful for learning verbs than for learning nouns.
Verb referents are more abstract than noun referents
Sentence structure cues
act as linguistic guidance to ensure effective verb learning.
Syntax is systematically related to verb meaning
Example:
Action involving 1 person:
Intransitive verb
[She laughs]
→ Action involving 2 people:
Transitive verb
2 N-P[She tickles her]
Toddlers can infer that a new transitive verb refers to an event involving 2 participant roles (instead of 1) & use the transitive word order accordingly
(e.g. Fisher, 2002; Naigles & Kako, 1993)
Many verbs occur in > 1 sentence structure
set of sentence structures in which each verb occurs yield more refined information about each verb's meaning
(e.g. Scott & Fisher, 2009, Snedeker & Gleitman, 2004)
SCENE OBSERVATIONS
SENTENCE STRUCTURE CUES
: SEMANTIC CONTENT
= A UNIQUE WORD MEANING
: SEMANTIC STRUCTURE
= number/type of participant-roles involved
= A CLASS OF VERB MEANINGS THAT CAN RANGE BROADLY IN SEMANTIC CONTENT
INFANTS RAPIDLY LEARN WORD-REFERENT MAPPINGS VIA CROSS-SITUATIONAL STATISTICS
(SMITH & YU, 2008)
QUESTION:
Can 12-14 m/o infants use cross-situational observation to learn novel nouns despite referential ambiguity?
EXPERIMENT:
1. Training trials
Pictures of 2 novel objects presented side by side
Each pair of pictures accompanied by 2 novel labels
Infants do not know which novel label pairs with which object
Each novel word consistently occurs with only 1 object
2. Test trials
Infant hears label with the referent and a distractor object
RESULTS:
Infant looks longer at the target object. (Learning has occured.)
IMPLICATIONS:
Infants kept track of which word consistently accompanied which picture.
12-14 m/o infants possess the
basic mechanisms
required to benefit from
cross-situational consistency
in the noun learning process
Ability to attach some information about potential object referents to each word's lexical entry
, even when referential context contains > 1 possible referent
Ability to update information across trials
to identify correct object-noun label pairings
challenges faced by young children:
challenge 1.
Encoding referential information under uncertainty is more difficult for verbs than for nouns.
Nouns: Static & Imageable
Verbs: May change over time & Non-imageable + Roles
To date, there is no evidence that children can attach event [action] referential information to a verb's lexical entry when multiple candidate referent events are present.
challenge 2.
Complexity of event referents may complicate the retrieval and comparison of referential options.
Nouns [Objects]:
Despite within-category variability of objects, young children do encounter repeated tokens of the same object category.
Verbs [Events]:
Young children do not encounter repeated tokens of the same action unless captured on video
In reality:
Need to compare different event tokens and abstract across irrelevant variation
Mixed evidence regarding children’s ability to abstract across irrelevant variation when learning verbs
THE STRUCTURAL ALIGNMENT AND COMPARISON OF EVENTS IN VERB ACQUISITION
(CHILDERS, 2008)
EXPERIMENT:
- 2.5 y/o children
1. Training trials
(A) Consistent action + Varying result -> Series -> Same novel verb
(B) Consistent result + Varying action -> Series -> Same novel verb
2. Test trials
Children were given a set of objects
Task:
To enact action named by novel verb
RESULTS:
Children's enactments preserved the consistent elements.
IMPLICATIONS:
Children understand that event components that occur consistently across situations are central to word's meaning.
= Children possess the basic comparison process required to benefit from cross-situational consistency.
HMM I'm smarter than you think I am! :p
FOCUSING ON THE RELATION: FEWER EXEMPLARS FACILITATE CHILDREN'S VERB LEARNING AND EXTENSION
(MAGUIRE, HIRSH-PASEK, GOLINKOFF & BRANDONE, 2008)
EXPERIMENT:
- 2.5 y/o children
Children see:
(A) 1 repeated event token [enacted by same actor]
(B) 4 distinct event tokens [enacted by different actors]
She's pimming her toy. And

she's nading her
toy
RESULTS:
1 REPEATED TOKEN
HEHE, I GET IT!!! :B
4 DISTINCT EVENT TOKENS
ERM... WAIT A MINUTE HMM :(
IMPLICATIONS:
Children may fail to identify consistent action across different event tokens
12 trials of continuous study-test procedure
On each trial..
2 event tokens presented simultaneously
2 intransitive sentences played to toddler at the same time as video
"Look, she's pimming. And
she
's
nad
ing.
Between trials..
3 sec blank-screen, silent interval
Counter-balanced the order in which the 2 verbs were presented within trials
Children assigned to one of the two verb orders
V.O. 1: “Look, she’s pimming.. And she’s nading.”
V.O. 2: “Look, she’s nading.. And she’s pimming.”
Left-right positioning of event tokens was counterbalanced within verb order

Details for audio
Coded where children looked (left screen, right screen, away) frame by frame from silent video
8 children’s data independently coded by second coder
Analysed children’s visual fixations within a 2.5-s time window that began at the onset of the verbs
"Look! She's pimming! And, she's rivving!"


30 analysis windows dropped: Children looked away from the screens for >67% of the 2.5s analysis window
Remaining 834 analysis windows: Computed children’s looking time to the target event and the distractor event
Omitted first analysis window for each verb 
Main analyses conducted on 2nd - 6th window

Distractor vs. Target events
Descriptive statistics:
overview of the time course of children’s looking patterns

2 x 4 x 2 x 2 mixed ANOVA
Block (1, 2): within
Verb (pim, nade, rivv, tazz): within
Event (target, distracter): within
Vocabulary group (high, low): between
Dependent Variable
: Children’s looking times
Target event
Distractor event
Target-advantage score
= Time to target – time to distractor

<- 2.5 analysis window->
<- 2.5 analysis window->
Statistical results
Trend
Implications
In summary:
Significant main effect of
event
Children looked reliably longer at the target event than at the distractor event
Children able to gather information about the events that co-occurred with each verb and update that info across trials
despite referential ambiguity
Significant main effect of
block
Children's tendency to look away more in block 2 than in block 1
No interaction between
target event and block
2.5 y/o children were selected as participants because they possess basic comparison process
There is a possibility that children cannot detect consistency across varying event tokens with the additional challenge of encoding and comparing events under referential ambiguity.
Children did not require exact repetition of event tokens in order to benefit from cross-situational consisteny
-
ULTIMATE QUESTION:
given the challenges faced by children in verb learning...
do you think that children can use cross situational consistency to learn verbs under referential ambiguity?
-
Children have learnt the verb referents already in the first block, and hence lost their interest in the second block
SPECIAL EXPERIMENTAL AIDS:
1. Relevant sentence context information
helps constrain range of interpretations children consider, thereby facilitating use of cross-situational consistently within that range
2. Use of concrete [throw] (as opposed to abstract [think]) verbs
results from human simulation paradigm
adults use scene information to identify concrete verbs much more often than abstract verbs
cross-situational observation should be most useful for learning concrete verbs, simply because referents are relatively easy to observe in scenes
ANALYSIS
RESULTS
HIGH VOCAB LOW VOCAB
12 Trials
Trial 1

Trial 2

Trial 3


Trial 4

Trial 5

Trial 6
Trial 7

Trial 8

Trial 9


Trial 10

Trial 11

Trial 12
Order randomized but with these constraints:
Same actor never appeared on both screens at once
No action occurred more than 2 trials in a row
Each action appeared an equal number of times on the right and left screens


REFERENCES
Sample Trial
No interaction between
vocabulary group and target-advantage scores
Children's vocabulary level did not affect their ability to benefit from cross-situational consistency in this expt
Syrett, K., Arunachalam, S., & Waxman, S. R. (2013; 2014). Slowly but surely: Adverbs support verb learning in 2-year-olds. Language Learning and Development, 10(3), 263-278. doi:10.1080/15475441.2013.840493
Syrett, K., Arunachalam, S., & Waxman, S. R. (2013; 2014). Slowly but surely: Adverbs support verb learning in 2-year-olds. Language Learning and Development, 10(3), 263-278. doi:10.1080/15475441.2013.840493
Arunachalam, S., & Waxman, S. R. (2010). Specifying the role of linguistic information in verb learning. In K. Franich, K. Iserman, & L. Keil (Eds.), Proceedings of the 34th Annual Boston University Conference on Language Development (pp. 11–21) Somerville, MA: Cascadilla Press.
Interaction of syntactic and semantic information in verb learning
2 y/o children were more successful in learning novel transitive verbs when frames contained content nouns as opposed to pronominal nouns (Arunachalam & Waxman, 2010)
Adverbs as a potential cue to aid toddlers in acquiring novel transitive verbs that occur in pronominal frames
Adverbs have to be concrete = provides semantic content (e.g. slowly vs. nicely/right now)
1st evidence that adverbs can facilitate verb learning in toddlers
Gillette, J., Gleitman, H., Gleitman, L., & Lederer, A. (1999). Human simulations of vocabulary learning. Cognition, 73, 135–176.
Childers, J. B. (2008). The structural alignment and comparison of events in verb acquisition. In B. C. Love, K. McRae, & V. M. Sloutsky (Eds.), Proceedings of the 30th annual conference of the Cognitive Science Society (pp. 681–686). Austin, TX: Cognitive Science Society
Maguire, M. J., Hirsh-Pasek, K., Golinkoff, R. M., & Brandone, A. C. (2008). Focusing on the relation: Fewer exemplars facilitate children’s verb learning and extension. Developmental Science, 11, 628–634
2 display trials


BLOCK 1




BLOCK 2
familiarise chilen with actors, toys, structure of sentences
"She has a toy. And
she
has a toy."



Continuous study-test procedure
6 trials per block
12 novel events presented with novel transitive verbs
8s display
3s blank, silent screen

Repeated 12 events
frame by frame coding
-left
-right
-away

interobserver reliability (98% corroboration)

target looking time
distractor looking time
target-advantage score
Results
Trend
Implications
Significant main effect of
event
Children looked reliably longer at the target event than at the distractor event
Children able to gather information about the events that co-occurred with each verb and update that info across trials
despite referential ambiguity
Significant main effect of
block
Children's tendency to look away more in block 2 than in block 1
No interaction between
target event

and

block
Children did not require exact repetition of event tokens in order to benefit from cross-situational consisteny
-
High-vocabulary children perform differently from low-vocabulary children
Children have learnt the verb referents already in the first block, and hence lost their interest in the second block
Interaction between
event
and
vocabulary group
no effect of sex, list & verb order
-performed similarly to children in expt 1
-appeared to have learnt/understand structure of task
-SUCCESSFULLY gathered info about verb's referent across trials using
cross situational consistency

Main effect of event
look longer at target than distractor event

Interaction of window and event
look more reliably at
Verb 2
target than
Verb 1
presenting
Verb 2

provides

additional constraint

She's .....____ing
..................
And she's ____ing
She's .....____ing
..................
And she's ____ing
-no clear preference for
verb 1
target/
verb 2
target
-no evidence that low-vocabulary children identified
verb's referents

significant preference for distractor event in block 2

2 x 4 x 2 x 2 anova
block verb event vocabulary
WHY?
increased complexity
transitive sentences = 2 N-P arguments
more complex semantic structure
greater language processing load
complicates encoding, retrieval & comparison of referential information


children need to abstract across actors, toys & actions
detecting consistency across contexts MORE CHALLENGING
reflects challenges under typical learning conditions
Full transcript