What infants know about syntax but couldn

1
What infants know about syntax but couldnt have
learned experimental evidence for syntactic
structure at 18 monthsJeffrey Lidz, Sandra
Waxman, Jennifer Freedman

• Ashley David
• Sara McTaggart
• Annie Porter
• Ellen Ramsey

2
Two major views

• Nativist view
• Grammar acquisition depends on innate structures
  in addition to input
• Major support poverty of the stimulus
• General learning view
• Linguistic input is sufficient in explaining the
  childs acquisition of grammar
• Uses general purpose learning mechanisms

Background Information
3
Support for the Nativist View
Poverty of the stimulus Looking at NP structure
and Anaphors
Possibility of two representations of NP
(containing a det, adj, and noun) Flat
Nested Anaphors Anaphoric elements substitute
only for constituents. Example Ill play with
this red ball and you can play with that one.
What is one referring to?
ball
red ball
Background Information
4
Noun Phrase Representations
Support for the Nativist View

• Flat structure
• hypothesis
• NP
• det adj No
• the red ball

• Nested structure hypothesis
• NP
• det N
• adj N
• No
• the red ball

Background Information
5
Flat vs Nested structure in children

• One is anaphoric to N and this is only possible
  in the nested structure
• BUT
• Even if a child used the flat structure, finding
  evidence that they were wrong would be difficult
  because every situation where one N red ball
  true also makes one Nball true

Background Information
6
Flat vs Nested structure in children
continued

• Consider the following situation which would be
  needed to provide evidence against the flat
  structure
• Sally has a red ball but Julie doesnt have one.
• Imagine that Julie has a ball, but its a blue
  ball.
• In this case, interpreting one as referring to N
  is false since one would be referring to ball,
  but Julie has a ball, just not a red ball.

Sally
Julie
Background Information
7
Flat vs Nested structure in children
continued

• A child coming across this type of situation
  would have to come to the conclusion that their
  flat structure hypothesis was false and would
  have to change to the nested NP structure in
  order to have correct grammar.
• These types of situations are rare. So, if a
  learner started with the flat structure, it is
  possible that they would never get the evidence
  to lead them to the correct structure.
• Because there is no evidence of English speakers
  having a flat structure grammar, the idea is not
  considered. We assume the nested structure.

Background Information
8
Support for a nested structure

• Hamburger Crain (1984) found that children do
  represent the NP with a nested structure AND that
  they know one is anaphoric to N
• BUT
• Still not enough to describe how learners
  structure begins- as flat structure that matures
  into nested structure OR as a nested structure
  from the start

Background Information
9
Corpus analysis
Support for the Nativist View

• Examined linguistic input of (2) children using
  CHILDES (found 792 anaphoric uses of one)
• Experimenters were interested in looking at the
  possibility of one as anaphoric to N not being
  available to learners
• Examined structures of antecedent
• Main idea If there are a lot of instances of
  one referring unambiguously to N- it is possible
  that input is significant for child to learn
  syntactic properties of one (general learning
  view)
• Results In 95 of the cases, the antecedent did
  not provide unambiguous indication of one
  referring to N
• Conclusion Linguistic input is insufficient for
  learner to know that one is anaphoric to N
  (nested structure) which supports the nativist
  view

10
Experimentation

• Research question Do infants know that one is
  anaphoric to N and thus have a nested NP
  structure?
• Prediction They do have a nested structure
  and will therefore interpret one as anaphoric to
  N.

11
Subjects
Experimentation

• 24 English-speaking children (12 male 12 female)
• Age 16m 23d 18m15d (ave 18m3d)
• Selected because just beginning to produce more
  than one-word utterances
• Two taken out because of failure to complete
  experiment

12
Materials
Experimentation

• Auditory stimuli
• Sentences questions recorded in infant-friendly
  voice to be used as audio track with video in
  testing
• Used objects whose names are normally recognized
  by infants of 13 months (a bottle, a car, a
  shoe, a bear)
• Visual stimuli
• Constructed using computer drawing program
• Brightly coloured large to attract attention
• Of equal salience
• (Two) objects displayed on split-screen
  (side-by-side)

13
Set-up
Experimentation
Records infant looking times locations
Camcorder
Presents audio-visual materials
61 in. LCD screen
20 in.
20 in.
12 in.
12 in.
18 in.
18 in.
80 in.
Projects images
Camcorder
Child (sitting on booster seat)
14
Procedure
Experimentation

• Playroom-
• parents were explained experiment, signed consent
  form, and completed vocabulary checklist
• Testing room-
• Parents instructed not to interact with child-
  sat behind child and wore visor that blocked view
  of screen so not to influence childs direction
  of gaze

15
Procedure (continued)
Experimentation

• Familiarization Phase
• Image of single object presented 3x
• Appeared alternating fashion on L or R screen
  accompanied by a recorded voice
• Object was presented with a NP that included a
• det adj noun (example Look! A yellow
  bottle.)
• Testing Phase
• Two objects appeared simultaneously side-by-side
  (on either side of the screens midline)
• Both images were from the same category as the
  familiarization object BUT only one was the same
  colour

16
Procedure (continued)
Experimentation

• Control condition-
• In the test phase, the child heard a neutral
  phrase
• Now look. What do you see now?
• Anaphoric condition-
• In the test phase, the child heard a phrase
  containing the anaphoric one
• Now look. Do you see another one?
• Each condition consisted of 4 trials
  (familiarization and test phase)
• Children were randomly assigned to either
  condition
• Test phase lasted 8 seconds
• Entire experiment lasted 3 minutes 46 seconds

17
Familiarization Phase
Example of Control Condition
Look! A brown bear!
18
Familiarization Phase
Example of Control Condition
Wow! A brown bear!
19
Familiarization Phase
Example of Control Condition
Look! A brown bear!
20
Test Phase
Example of Control Condition
Now look. What do you see now?
21
Familiarization Phase
Example of Anaphoric Condition
Look! A yellow bottle.
22
Familiarization Phase
Example of Anaphoric Condition
Wow! A yellow bottle.
23
Familiarization Phase
Example of Anaphoric Condition
Look! A yellow bottle.
24
Test Phase
Example of Anaphoric Condition
Now look. Do you see another one?
25
Coding Predictions
Experimentation

• Coding
• Durations of looking time to the L or R test
  image were coded afterwards by a single coder who
  watched the video of the session
• Another coder coded 25 of the data
• Inter-coder reliability- 96
• Predictions
• Control condition- linguistic stimulus does not
  favour one image over the other so expect child
  to look longer at novel image
• Anaphoric condition- where child looks should
  indicate their type of structure
• If represent with flat structure- either image
  could be referents of the noun
• Linguistic stimulus is uninformative
• Should see same as control (look to novel image)
• If represent with nested structure- should have
  preference for image that matches one as
  anaphoric to N (look to familiar object)

26
Results
Experimentation

• Control condition
• Preference for novel image
• Anaphoric condition
• More attention to familiar image than novel image
• Subjects in the anaphoric condition were more
  likely to look at the familiar image than were
  the subjects in the control condition
• Conclusion
• By 18 months of age infants interpret one as
  anaphoric to N and therefore represent NPs with
  a nested structure

What does this imply???
27
Figure 1. Mean looking time (in seconds) to the
two test images in each condition
Experimentation
28
Other possible interpretation of results
Experimentation

• It could be the case that infants treat one as
  anaphoric to N
• BUT
• when they are asked to pick out another image
  that denotes, for them, N- they have a
  preference for the familiar image (familiarity
  bias)
• SO
• Conducted a control experiment to make sure this
  was not the case.

29
Control Experiment

• Same as Experiment with the following exception
• Test phase, two conditions
• Noun condition- children asked question that
  contained only the noun presented during
  familiarization (N)
• Example Do you see another bottle?
• Adjective-Noun condition- children asked question
  that had adj-noun combination heard in
  familiarization (N)
• Example Do you see another yellow bottle?

30
Predictions and Results
Control Experiment

• Noun Condition (where both are nouns)
• If children had a familiarity bias- would expect
  to see children looking longer at the familiar
  image
• If children do not have a familiarity bias- would
  expect child to look at either image in the noun
  condition
• Adj-Noun condition
• Would expect child to look longer at the familiar
  object since only that image corresponds with the
  N

What found
31
Figure 2. Mean looking time (in seconds) to the
two test images in the control experiments
Control Experiment
32
Interpretation of results
Control Experiment

• Because no familiarity bias was found- conclude
  that children choose familiar object (in
  anaphoric condition) because it matches their
  nested NP structure

33
Discussion

• Learners just beginning to combine words already
  have a rich syntactic representation
• Support for poverty of the stimulus
• Learners never consider that an element could be
  anaphoric to N but behave like adults who
  consider one as anaphoric to N
• Support for nativism- but experimenters still
  acknowledge role of input

END