The Headturn Preference Paradigm

1
The Headturn Preference Paradigm

• Barbara Höhle
• University of Potsdam
• and
• Free University of Berlin

2
The headturn preference paradigm

• First use
• Hirsh-Pasek, K. et al. (1987) Clauses are
  perceptual units for infants. Cognition 26,
  269-286.

3
(No Transcript)
4
Features of the paradigm

• acoustic stimuli are presented from one of two
  loudspeakers on the sides of the testing booth
• an at least 45 headturn should be necessary for
  the child to fixate the lamp
• times of turning away are not included into the
  calculation of the orientation time
• a trial is finished when the child looks away for
  more than two seconds

5
Features of the paradigm

• it is not possible to present single stimuli in
  this paradigm
• to create enough variation the acoustic stimuli
  should have a duration of 16 to 20 seconds
• this duration can be achieved either by
  repetitions of a single stimulus or by the use of
  a sequence of different words or sentences

6
Rationale of the Paradigm

• the child learns the contingency of her headturn
  and the stimulus presentation
• the more interesting the stimulus is for the
  child the longer the duration of the headturn
• results can be interpreted as reflecting a
  preference for one of the types of stimuli
  presented
• results also reflect the ability of the infants
  to discriminate the types of stimuli

7
The two basic versions of the method

• Without familiarisation
• experiment starts immediately with presentation
  of the test stimuli (warming-ups)
• with familiarisation
• the child is presented with a specific kind of
  stimulus at the beginnig of the experiment (e.g.
  two words) until a criterion has been reached
• immediately after the familiarisation the testing
  phase is started

8
What to use these versions for?

• without familiarization
• reflects the spontaneous preference of the child
  for one of the types of stimuli used
• gives an insight into the knowledge the child
  brings into the experiment
• e.g. presentation of grammatical and
  ungrammatical sentencs (Santelmann Jusczyk,
  1998)
• e.g. presentation of phonotactically legal and
  illegal pseudowords (Friederici Wessels, 1993)
• e.g. presentation of trochaic and iambic words
  (Jusczyk, Cutler Redanz, 1993)

9
What to use these versions for?

• with familiarization
• has first been used to study word detection
  (e.g. presentation of isolated words in the
  familiarization, presentation of passages with or
  without these words in the test phase Jusczyk
  Aslin, 1995)
• can also be used to study learning mechanisms
• (e.g. presentation of a continuous speech stream
  during familiarization, presentation of parts of
  the stream and new stimuli during testing
  (Saffran, Aslin Newport, 1996)

10
What to use these versions for?

• to study discrimination abilities
• (e.g. familiarization/habituation with one type
  of stimulus, testing with this old stimulus and
  an additional one familiarization with syllable
  /ba/, testing with /ba/ and /pa/, differences in
  orientation time show discrimination

11
Subjects

• no special criteria for subject selection
  (monolingual, fullterm etc.)
• ages tested 4 month-olds to 24 month-olds
• drop-out rates depending on age
• best for 6 to 9-month-olds (15 to 20)

12
Dos and dont dos

• make sure that the experimenter is blind with
  respect to the condition to which a given trial
  belongs
• make sure that the parents do not hear the
  acoustic stimuli
• make sure that the experiment is run in a quite,
  smoothly illuminated room
• make sure that the child does not carry toys or
  nipples into the experiment
• make sure that the child is healthy when coming
  to the experiment

13
Dos and dont dos

• tell the parents that they should chancel their
  date if they have the feeling that their child
  does not feel well
• if the child and the mother are seated in the
  experimental booth start the experiment
  immediately
• make sure that the types of stimuli you are using
  are presented from both sides for an equal number
  of times
• make sure that the number of stimulus
  presentations from both sides is balances

14
Dos and dont dos

• make sure to use different randomizations because
  there are very strong order effects in the
  results
• use warming-up trials at the beginning of the
  experiment (two to four) to make the child
  familiar to the procedure

15
Dos and dont dos

• dont play around with the child too much before
  starting the experiment
• there should not be many people around the child
  before the experiment starts
• dont put too many trials into one single
  experimental session (around 16 seems to be
  perfect)

16
Possible Outcomes Familiarity vs. Novelty-Effects

• Familiarity-Effect
• preference for the familiar stimulus
• e.g. grammatical over ungrammatical stimulus
• e.g. familiarized words over new words
• Novelty-Effect
• preference for the new stimulus
• e.g. new words over familiarized words
• e.g. new contexts over old contexts

17
Possible Outcomes Familiarity vs. Novelty-Effects

• Familiarity Effects (examples)
• word detection Jusczyk Aslin, 1992 Jusczyk et
  al. 1999
• grammaticality Santelmann Juzscyzk, 1990
• phonotactics Friederici Wessels, 1993 Jusczyk
  et al. 1993)
• Novelty-Effects (examples)
• word detection Saffran et al. 1996 Echols et
  al. 1997)
• grammaticality Höhle et al. (in press)

18
Modelling the direction of the preference (Ames
Hunter, 1988)

• initially infants show no preference for one of
  the stimuli
• at first a preference for the familiar stimulus
  arises
• during the experiment this preference changes to
  the novel stimulus
• whether a familiarity or a novelty effect is
  observed in an experiment depends on how long the
  children stay in these phases

19
An experiment on rhythm discrimination

• 4-month-olds
• familiarization with trochaic bisyllabic BA-ga
  for 30s
• during test phase presentation of trials
  consisting only of trochaic or of iambic
  bisyllabics

20
(No Transcript)
21
An experiment on rhythm discrimination
22
An experiment on rhythm discrimination
23
Modelling the direction of the preference (Ames
Hunter, 1988)

• there are several factors influencing how quick
  these phases are passed
• age of the children with growing age a novelty
  effect becomes more probable
• variability of the stimuli the more variable the
  stimuli the higher the probability for a
  familiarity effect
• complexity of the stimuli the more complex the
  stimuli the higher the probability for a
  familiarity effect

24
Evidence for the model

• word detection
• Juscyzk and coworkers real speech stimuli in
  different contexts familiarized for 45 s
  familiarity effect
• Saffran et al. sequences of simple CV-syllables
  repeated for 2 minutes during familiarization
  novelty effect

25
Using other dependent variables with this paradigm

• Headturn latency (Orientation Latency Procedure
  Gout, Christophe Dupoux, 2002)
• each trial consisted of familiarization stimulus
  (presented centrally)
• in the test phase two different kind of stimuli
  were presented laterally either from the same
  category as the familiarization stimulus (control
  trials) or from another category (test trials)
• dependent variable latency of headturn
  initiation from beginning of the presentation of
  the acoustic stimulus, headturn was reinforced by
  a flashing light
• result longer latencies for control than for
  test trials

26
Using other dependent variables with this paradigm

• headturn direction (Hirsh-Pasek et al. 1986)
• fixed presentation side for different types of
  stimuli
• child is familiarized to the presentation side at
  the beginning of the experiment
• at the beginning of a trial lights on both sides
  flash
• stimulus corresponding to that side is started
  when the child turns her head to the side
• child controls which kind of stimulus is
  presented by headturn
• results no systematic preference for one of the
  types of stimuli

27
Summary

• Pros
• different kind of domains can be studied
• wide age range
• relatively low drop-out rates
• high inter-rater reliability (90 - 98)
• Cons
• restricted number of trials restricted number
  of experimental conditions (normally two) gtgt
  between subjects designs
• no clear prediction of direction for a given
  experiment possible

28
(No Transcript)