In sleeping neonates. Using changes in neural responses t

1
Language Individual differences

• Daniel Messinger, Ph.D.

2
Language overview review

• What is the normative course of infant language
  development? 
• How do infant cries develop (directed and
  undirected)?
• What are the stages of development of non-cry
  vocalizations?
• What are some early milestones of verbal
  development (verbal development involves words)?

3
Perspective

• Last time
• Features of language that all infants develop
• Focus on production speech
• This time
• How infants differ in learning language
• Differences in learning to hear a first language
• Differences in learning to talk a first language
• Autism and deafness

4
Todays questions

• How does the ability to distinguish between
  non-native speech sounds change in the first
  year?
• What does this mean about development?
• Can distinctions between non-native sounds be
  taught?
• How is language experience associated with later
  child language competence and IQ?
• How is socioeconomic status associated with
  differences in language experience?
• What does cochlear implantation teach us about
  language development?

5
Consider the spoken tokens of doll.

• To a Hindi speaker, the difference between the
  d sounds in this doll versus your dolla
  phonetic contrast between a dental d?al versus
  a retroflex ?al, respectivelywould signal two
  possible word forms (either lentils or branch).
• In English, both of those d sounds signal just
  one possible word formphonetically labeled as an
  alveolar dal.

6
Different languages provide different phonetic
experiences
7
Whats going on?

• English-learning infants hear Hindi contrast
  better than English-speaking adults
• Almost as well as adult Hindi-speakers

8
Distinguishing between non-native speech sounds
in 1st year

• At birth, infants are capable of discriminating
  all phonetically relevant differences in the
  worlds languages
• They perceptually partition the acoustic space
  underlying phonetic distinctions in a universal
  way.
• By 6 months of age, infants raised in different
  linguistic environments show an effect of
  language experience.
• Their representations are becoming language
  specific

9
How does this develop?

• Infants lose this ability in the first year of
  life, especially toward one year of age

10
What this mean for development

• Very young infants can discriminate a wide range
  of phonetic contrasts in a variety of languages
• Between 1 12 months, infants
• increase knowledge of which syllables follow
  which in their native language
• but lose ability to make contrasts that do not
  occur in their native language
• /r/ vs. /l/ . /b/ vs. /v/ . Te vs. te, tu vs.
  too
• Development involves relatively permanent change,
  but not always improvement in all things.

11
Parallels in speech production

• Infant babbling shows little influence of native
  language.
• Once the infant forms his/her 1st words than the
  sounds produced conform more closely to those of
  the native language
• This corresponds to the stage at which infants
  begin to show language-specific sensitivity
  (10-12 months).

12
Possible roles of experience

• Induction prior experience with a language is
  necessary because perceptual capability depends
  entirely on environmental input
• Attunement experience makes possible the full
  development of a capability.
• Facilitation experience effects only the rate
  of development of a capability.
• Maintenance/loss the ease in which a capability
  is fully developed before the onset of
  experience, but experience is necessary to
  maintain the capability.
• Maturation development of a capability
  independent of experience

13
Perceptual Magnet Effect

• Instances of sounds that belong to a category are
  drawn toward the Prototype.
• Physical (acoustic) vs. perceptual maps
• the latter differ for speakers of different
  languages

14
Can distinctions between non-native sounds be
taught?

• Cheour has experimentally produced this
  development
• In sleeping neonates
• Using changes in neural responses to sounds as an
  outcome variable
• http//www.med.cornell.edu/news/press/2002/feb_22_
  newborn.html

15
How sleeping babies learn

• The babies had electrodes placed on their scalps,
  and speakers near their heads gently played a
  randomized sequence of two similar Finnish vowel
  sounds as they slept a "standard" sound, /y/,
  and a "deviant" sound, /i/.

16
Mismatch Negativity (MMN)

• when the brain hears the standard sound, there
  is a certain response in the brain, and when it
  hears the deviant sound, there is another
  response.
• Subtracting the responses to the deviant from the
  responses to the standard produces the MMN.

17
Language-specific phoneme representations in the
infant brain

• Development of language specific 'memory traces'
•  
• Infants 6 months - 1 year 
•  
• Estonian and Finnish languages 
•  
• Analyzing mismatch negativity (MMN)
• deviant vs. standard stimuli  
•  
• Fernandez

18
Training

• No initial MMN for any group (N15).
• Over the following night, for between
  two-and-a-half and five hours, the experimental
  group had a "training" session of exposure to the
  two sounds.
• /y/ vs. /i/.
• One control group did not have this exposure, and
  the other control group heard two different
  sounds, /a/ and /e/.

19
Results

• The experimental group showed significant
  mismatch negativity to the deviant sound.
• The babies had learned to distinguish between
  these two Finnish vowels.
• Persisted for at least 24 hours.
• The two control groups showed no MMN to the
  deviant sound.

20
(No Transcript)
21
(No Transcript)
22
Conclusion

• "We have shown that newborns can assimilate
  auditory information while they are sleeping,
  suggesting that this route to learning may be
  more efficient in neonates than it is generally
  thought to be in adults."
• Cheour
• Is this learning?

23
Conclusions

• Language-dependent memory traces emerge before
  the age of 12 months 
•  
•  MMN amplitude by the age of 1 year
•    for native phonemes
•    for non-native phonemes
•  
• The ability of infants to discriminate native
  speech sounds improves, while they lose some of
  their ability to discriminate non-native speech
  sounds 
•  
• Fernandez

24
A word on word learning

• 14-month-old infants can rapidly learn arbitrary
  associations between words and objects,
• this ability appears to develop at about 14
  months
• only 14-month-old infants formed word-object
  associations but appeared to do so only when the
  objects were moving.
• Although 8- to 12-month-olds did not form the
  associations, they appeared to process both the
  word and the object information.
• Werker, J. F. Cohen, L B. Lloyd, Casasola,
  Stager/ 1998 Developmental Psychology.
  Acquisition of word-object associations by
  14-month-old infants. 34(6) 1289-1309

25
Reviewing the power of language

• More maternal vocalizing at 1 month
• Associated with vocalizations at 8 24 months
  and with socioeconomic status
• Also predicts greater adolescent intelligence
• R2 .22 for gazing and maternal vocalizations

26
Overview

• Socioeconomic differences in how folks talk to
  their kids
• What impact might it have?
• How is language experience associated with later
  child language competence and IQ?

27
Socioeconomic status differences in language
experience are associated with later child
language competence and IQ

• Meaningful differences in the everyday
  experiences of young American children. Hart
  Risley (1995). Baltimore, MD Brookes Publishing
  Co
• Some text from summaries by Susan Brunner, Dahra
  Jackson, and Amy Vaughan

28
Participants

• Longitudinal project from 9-10 months infant age
  up until 2-2 ½ years later
• 42 families observed for one hour every month, at
  home, in natural settings
• recruited from birth announcements, friends and
  families at University pre-school, WIC meetings,
  and state records
• all but 8 families were intact, all but one had a
  male figure involved
• 13 upper SES, 10 middle SES, 13 lower SES, and 6
  families on welfare all well-functioning

29
Data collection

• Observers transcribed and audio-recorded all
  verbalizations and interactions that would have
  an effect on another person never interacted
  with child, but responded to parents
• Observers assigned to families for entire study,
  when possible, and similar to family in terms of
  background
• no drop-outs after first year, reliability on
  coding and observations was adequate
• words coded as part of speech, episodes coded by
  type, and speaker coded dictionaries compiled
  for each speaker (all on computer)

30
Commonality

• Despite how strikingly different the families
  were in how much talking and interaction
  typically went on in the home, just socializing
  during everyday activities was sufficient for all
  children (regardless of SES) to learn to talk by
  age 3.

31
42 Families and the Differences Among Them

• differences observed in family language style
  parents language seemed to reflect the number
  and variety of behaviors they had for dealing w/
  their children
• some families talked more than others, and this
  was variable within families from month-to-month,
  but stable over the 3 years
• birth order and family size affected the amount
  of talk each child received, but did not affect
  the total amt. of talk

32

• SES seemed to make the biggest contribution to
  both amount of talk and time spent in
  interactions, with hi SES at an average of 482
  wds/hr and 48 mins/hr, and welfare families at
  197 wds/hr and 17 mins/hr

33
Language and SES (class)

• Children from all backgrounds have the same kinds
  of everyday language experiences.
• But more economically advantaged children differ
  in the amount of these experiences it is the
  frequency that matters.
• More opportunities for learning language occur
  when children engage in many and varied
  interactions with other people families tend to
  be consistent in the opportunities they provide
  for their children.

34
Talk that teaches talk

• THEY JUST TALKED
• parents talked beyond what was needed to provide
  care
• THEY LISTENED
• To add information and prompt elaboration
• THEY TRIED TO BE NICE
• When enforcing a rule
• THEY GAVE CHILDREN CHOICES
• THEY TOLD CHILRESN ABOUT THINGS
• Things worth noticing or remembering (Halloween)

35
Quantity of language Nouns, adjectives, and
adverbs to child
36
Being positive

• Repetitions, extensions, expansions,
  confirmations, praise, approval over all feedback
  (including imperatives, criticisms, etc).

37
Relating things and events

• Nouns, modifiers, and past-tense verbs divided by
  number of utterances per hour

38
Can you. . . ?

• Proportion of yes/no questions over yes/no
  questions and imperatives

39
Responsiveness

• Ok I see
• of responses not preceded by an initiation

40
How language experience is associated with later
child IQ

• Parenting Language diversity feedback tone
  symbolic emphasis guidance style
  responsiveness
• Predicts between and within SES groups

41
Language experience makes the difference
42
Implications for intervention

• To intervene with vocabulary growth rate
  increase the experiences available to the
  children
• Limited success ultimately the growth rates
  increased only temporarily.
• Could easily increase the size of the childrens
  vocabularies, could not accelerate the
  developmental trajectory.
• Removing barriers and offering opportunities and
  incentives is not enough to overcome the past,
  the transmission across generations of a culture
  of poverty.

43
Is environmental influence global or specific?

• We know that there are differences in language
  development across SES
• Mothers are primary source of language-experience
• Does maternal speech mediate the relation between
  SES and child vocabulary development?

The Specificity of Environmental Influence
Socioeconomic Status Affects Early Vocabulary
Development Via Maternal SpeechErika Hoff
44
Maternal speech fully mediates relationship
between SES and child vocabulary!

• SES -gt 5 of variance in child vocabulary
• SES significantly associated with maternal speech
• MLU -gt 22 of variance in child vocabulary
• When removed, only 1 of variance explained by
  SES
• Sothere are 2 processes going on
• 1. SES affects maternal speech
• Childrearing beliefs
• Time availability
• 2.Maternal speech affects language growth
• Provides data for childs word-learning
  mechanisms
• Longer utterance -gt more variance in word types
  (richer vocabulary)
• Longer utterance -gt more info about meaning
• Longer utterance -gt richer syntax

45
Support environmental specificity model

• Vocabulary development depends on specific
  properties of language experience
• Implies that enriching language experience can
  increase vocabulary development for low-SES kids

46
SES ? Parenting ? Child language
Hoff (2003)
47
Automated Vocal Analysis of Naturalistic
Recordings from Children with Autism, Language
Delay, and Typical DevelopmentOller et al., 2010

• Used a fully automated system to assess
  childrens vocalizations from naturalistic
  recordings
• Predict linguistic development and differentiate
  between typically developing children and those
  with language-related disorders
• Recordings collected from those with typically
  developing language, language delay, autism
• All-day recorder worn by children

48
Recordings

• Signal processing software identified
• Consecutive child vocalizations speech-related
  child utterances (SCUs)
• These divided into speech-related vocal islands
  (SVIs) analysis of rhythmic/syllabic
  articulation and voice
• 12 infrastructural acoustic features in 4
  groupings
• Rhythm/syllabicity, low spectral tilt/high pitch
  control, high bandwidth/low pitch control, and
  duration
• Presence or absence of each feature

49
Results

• Automated acoustic modeling tracked development
  (predicted age for typically developing children)
  and differentiated groups
• Primary factor childs control of
  infrastructural features of syllabification
• Differentiated between children with and without
  a language disorder with higher accuracy than
  between the two language disorder groups (autism
  and language delay)

50
(No Transcript)
51
Classification
52
Niparko et al., 2010
53
Earlier implantation, earlier language gains
54
Receptive Language
55
Expressive Language
56
Other findings

• Higher parent-child interactions and higher
  socioeconomic status were associated with greater
  rates of language learning.
• Bilateral implantation was not associated with an
  increase in language acquisition.
• Gender was not associated with an increase in
  language acquisition.