Early speech

1
Early speech language skills etiological links
to reading

• Emma Hayiou-Thomas
• Department of Psychology
• University of York

2
Language Reading

• Enormous body of research linking phonological
  skills reading
• Especially decoding
• Less clear how non-phonological language skills
  are related to reading
• More important for comprehension

3
Quadrant model of SLI dyslexiaBishop
Snowling, 2004
4
Behavioural genetics Phonology reading

• Phonological awareness reading
• Colorado twin study, rg .70
• Gayan Olson, 2001
• Phonological STM reading
• Reading problems more heritable in combination
  with nonword repetition deficits
• Bishop, 2001 Bishop, Adams Norbury, 2004

5
Behavioural genetics Phonology reading

• Longitudinal evidence
• Phonological awareness in preschool has
  substantial genetic influence
• Stable through kindergarten
• Generalises to kindergarten reading
• Byrne et al., 2005 Byrne et al, 2006

6
Behavioural genetics Broader oral language
reading

• Language, IQ, phonological awareness, and reading
  (6-7 years)
• Covariance genetically mediated
• Hohnen Stevenson, 1999
• Semantics, syntax, phon awareness, rapid naming,
  letter knowledge (preschool)
• Both genetic and (shared) environmental
  covariance
• All influences on oral language also influenced
  literacy measures
• Additional genetic and environmental influences
  specific to literacy variables
• Samuelsson et al., in press

7
Study aims

• Phenotypic associations between oral language
  reading skills
• Phonology ? Decoding
• Semantics/Syntax ? Reading comprehension
• Etiology of the associations
• Genetic link for phonology decoding
• Genetic AND shared environmental link for
  semantics/syntax reading comprehension

8
Twins Early Development Study

• Based at the Social, Genetic, Developmental
  Psychiatry Centre (SGDP) Institute of
  Psychiatry
• Sampling frame all twins born in England and
  Wales in 1994, 1995 1996 (15,000 twin pairs)
• Ages 2, 3, 4, 7, 9, 10and on!
• Language reading, cognitive behavioural
  measures

9
Participants

• 4 ½ years
• Sub-sample N 1672
• Home visit
• NB over-represented for low language
• 7 years
• Whole sample N 9757
• 10 years
• Whole sample N 5777
• Children with both language and reading data
• 4 ½ 7 199 MZ pairs, 190 DZ same-sex pairs (N
  800)
• 4 ½ 10 105 MZ pairs, 77 DZ same-sex pairs (N
  350)

10
Speech Language at 4½

• Speech
• Goldman-Fristoe test of articulation
• Nonword Repetition Task
• Language
• Semantics
• McCarthy Word Knowledge
• McCarthy Verbal Fluency
• Morphosyntax
• BAS Comprehension
• Renfrew Action Pictures Grammar
• And also
• Bus Story
• Phonological Awareness
• McCarthy Verbal Memory

11
Reading at 7 10

• Decoding
• 7 years
• Test of Word Reading Efficiency (TOWRE)
• Word and nonword reading subtests
• Administered over the phone
• Comprehension
• 10 years
• PIAT Reading Comprehension
• Sentence-level comprehension
• Administered on the web

12
Previous findings Early language

• 9 measures ? 7 General Language 2 Phonology
• General Language
• Moderately heritable (.34)
• Shared environment also important (.50)
• Phonology
• More heritable (.56)
• Shared environment not very important (.15)
• Overlap between language and phonology
• Substantial but not total (rg .64, rc 1.00)

13
Previous findings Reading

• Decoding
• TOWRE at 7
• Highly heritable (.70)
• Modest role for shared environment (.15)
• Comprehension
• PIAT Reading Comprehension at 10
• Moderately heritable (.41)
• Shared environment also important (.30)

14
Phenotypic associations Predicted
15
Phenotypic associations Actual
16
Bivariate heritability environmentality
DECODING
17
Bivariate heritability environmentality
COMPREHENSION
18
Bivariate heritability environmentality
DECODING
COMPREHENSION
19
Speech Language ? Decoding

• Early language ? Decoding
• Moderate phenotypic association (.43)
• Half genetic, half environmental
• Early phonology ? Decoding
• Moderate phenotypic association (.35)
• All genetic

20
Speech Language ? Reading comprehension

• Early language ?
  Reading comprehension
• Moderate phenotypic association (.47)
• Half genetic, half environmental
• Early phonology ?
  Reading comprehension
• Modest phenotypic association (.26)
• All genetic

21
Stability across ages domains

• Strength of phenotypic association
• Language at 4 ? reading at 7 (r .43)
• Language at 4 ? reading at 10 (r .47)
• Pattern of genetic and environmental contribution
  from early speech language to reading is the
  same for decoding at 7 and comprehension at 10

22
Speculative Summary

• Genetic core of early phonological skills
• Continues into decoding
• Through decoding also affects reading
  comprehension
• Genes environments for semantics/syntax
• Both the genes AND environments influence reading
• Both decoding and comprehension

23
For the (imminent) future

• Atypical development!
• SLI Dyslexia

24
Thankyou!

• Maggie Snowling
• Dorothy Bishop
• Robert Plomin
• Nicole Harlaar
• Yulia Kovas
• Philip Dale

25

• and a lovely audience!

26
Logic of the Twin Method I

• How similar are the two members of a twin (or any
  other sibling) pair?
• If very similar, evidence for familiality
• Genes and/or shared environment
• How to untangle genes shared environment?

27
Identical twins (monozygotic, MZ)
28
Non-identical, fraternal twins (dizygotic, DZ)
29
Logic of the Twin Method II

• MZ monozygotic twins
• Share 100 of DNA
• DZ dizygotic twins
• Share 50 of varying DNA (on average)
• MZ and DZ twins share their environment to the
  same extent (Equal Environments Assumption)

30
Logic of the Twin Method III

• MZs and DZs are both similar to each other ?
  shared environment
• MZs are more similar to each other than DZs ?
  genes

31
Sources of variance

• Genes/Heritability
• Effects of alleles at all gene loci that affect
  the trait
• Shared environment
• any environmental influences that contribute to
  the similarity between co-twins
• e.g. family environment
• Nonshared environment
• any aspect of environmental influence that makes
  co-twins different from each other
• e.g. child-specific event, such as illness

32
Estimating components of variance

• Heritability
• 2 X difference in MZ and DZ twin correlations.
• Shared environment
• Subtract heritability from MZ correlation
• Nonshared environment
• Subtract MZ correlation from 1
• Includes measurement error

33
Twin correlations - example

• MZ twin correlation .80
• DZ twin correlation .50
• h2 (MZ DZ) X 2 .60
• c2 MZ h2 .20
• e2 1 MZ .20

34
Model-fitting analysis

• Structural equation modelling
• More comprehensive way of estimating variance
  components based on principles of twin method
• Allows investigation of sex differences in the
  genetic and environmental influences
• Allows multivariate analysis

35
ACE
36
For the (imminent) future

• Atypical development

37
Speculative Summary

• Genetic core of phonological skills
• Continues into dyslexia
• Also a risk factor for poor comprehension but
  only to the extent that poor decoding is a risk
  for poor comprehension
• Part of SLI heritable phenotype?
• Genes environments for semantics/syntax
• Both the genes AND environments influence reading
• Both decoding and comprehension
• The other half of SLI?