A child's job is to play, we should let them (Final Paper
Maritza Gomez - Sigma Xi 2013 Virtual Research Expo
1. The role of parental labeling in
early language development
Maritza S. Gomez
Undergraduate Student Researcher (Class of 2014)
Brain & Cognitive Sciences
Majors: Psychology, Linguistics
University of Rochester
2013 Sigma Xi Undergraduate
Research Showcase
March 18-23, 2013
3. External linguistic input required
* Infants require external linguistic input to
acquire their first languages
* Experienced language users provide infants
with particular sounds, words, and
grammatical rules for their target language
* Example: Word an infant will learn for “milk”
depends upon the language environment
(e.g., leche in Spanish,laitin French)
4. Parents as primary input providers
* Parental labeling
behavior plays a
very important role
in children’s early
language learning
* More parental
language input
thought to yield
faster word learning
(e.g., Hurtado et al. 2008, Fernald
et al. 2006)
5. Previous studies of word learning
* Previous work focused on how infants solve
mapping problem using distributional
statistics
* Infants employ cross-situational statistics to
rapidly learn word-object pairings
(e.g., Smith & Yu, 2008; Yu & Smith, 2011)
* Input is typically thought of as a running
stream of word- and object co-occurances
that unfolds over time—independently of
what the infant thinks or does
6. Proposed alternative possibility
* Parents could adjust what they say to
their infants as a function of their
infants’ changing knowledge states
* Could potentially give infants a boost
in process of learning first language
* Parents could be providing learning
material that is “just right” for efficient
learning
7. Research questions
* Is parental language
input dynamically
related to the infant’s
knowledge state?
* Are parents more
likely to label objects
for which their infants
already have some
existing knowledge?
9. Data collection
* Worked with
research team to
collect video data of
infants and parents
engaged in
naturalistic play
* Head-mounted
camera sets on
infant and parent
captured video
10. Volunteer subjects
* Parents with infants recruited
from the Rochester, NY area
* Ages 12- to 15-
months-old (M =
13.4, N =12)
11. Data collection
* In each 3-minute trial, parent and infant
provided with 4 to 8 toys
* Each toy set contained a mixture of both
familiar objects (e.g., ball) and unfamiliar
objects (e.g., lobster)
12. Labels and vocab surveys
* In each trial, parents naturally provided
labels for toys in the scene
* Each label was recorded
* Vocabulary surveys also collected from
parents after play sessions
* Surveys asked which words
infants understood
* Two analyses conducted to
test whether parents were
more likely to use labels their
infants understood
13. Study 1: Difference of
means
Difference-of-means test methods
t-test results
14. Study 1: Analysis
* Computed the mean number of labels a
parents for two different object
categories:
Known—objects for which infant knew
the label
Unknown—objects for which infant did
not know the label
* Significance of difference tested with
Welch’s t-test (independent samples)
15. Study 1: Results
Number of Labels Produced by Parent
10
* p< .05
8
6
4
2
0
Known Unknown
Child's Knowledge of Labels
16. Study 1: Results and conclusions
* Plot shows parents use significantly more
labels for objects that are known
(familiar) to their infants than for those
that are unknown (unfamiliar)
* But is this difference due to some factor
other than infants’ knowledge
(e.g., maybe the fact that older infants
knew more labels drove the effect)?
* Second analysis controls for a number of
relevant factors
18. Study 2: Analysis
General linear mixed model to test whether
child’s knowledge of the label predicts
number of parental labels, controlling for:
* Age
* Total number of objects present in trial
* Total number of labels parent produced in
trial
* Factor for random subject effects
20. Study 2: Results and conclusions
* Infants’ knowledge of the label was a
significant predictor of the number of
times a parent used that label
* Effect was significant, even controlling for
other factors (e.g., infants’ age, number
of objects present)
* Number of labeling events (control
variable) was also significant; parents
who produced lots of labels were more
likely to produce a particular object label
22. Summary and discussion
* Results evidence a dynamic relationship
between parental labeling behavior and
infant’s vocabulary knowledge
* Could either be the case that parents are
more likely to use labels that infants
understand, or it could be the case that
infants are quicker to learn labels that
parents use more often
* Future work in progress will disentangle
two possible explanations for correlation
reported here
23. Future directions
* Follow-up research will collect measures
of parental productions and children’s
vocabulary in the home
* This work will allow us to collect greater
quantity of data at multiple time points
* Aim is to determine to what degree
parents’ labeling behavior drives
children’s lexical development
24. Potential impact of research
* Research aims to better understand what
type of learning environments are most
conducive to language learning
* Work will yield normative baselines for
language development
* Baselines can be used to develop more
effective language therapies for children
with delayed language development
25. Acknowledgements
Thank you to:
* My research project advisors, Dr. Richard
N. Aslin and Celeste Kidd
* Rochester Baby Lab research staff,
especially lab manager Holly Palmeri and
postdocs Dr. Steven T. Piantadosi and
Dr. Lauren Emberson
* LocalLocal.tvfor video of infants in lab
* Volunteer parents and infants for
participating in this research