Early identification and prevention of difficulties in the acquisition of basic scholastic skills –
reading and math
Heikki Lyytinen
Agora Centre & Department of Psychology
University of Jyväskylä
Niilo Mäki Institute
Finland
Stockholm, 25. October, 2012
Overview of the content of this presentation
• Learning to read highly transparent orthography• The developmental association between spoken
and written language skills – highlights from the Jyväskylä Longitudinal study of Dyslexia (JLD)
• Introduction to the JLD: goals and design• Overview of the predictive relations and their modelling• Early identification of children in need of support
• Supporting reading acquisition
.. Learning the connections between spoken and written language using a learning game - Graphogame
The concept of reading skill• Basic reading skill – ability to pronounce
written words accurately• Literacy – readiness to comprehend fluently
written language; requires– sufficient mastery of the spoken language meant
to be learned to read– accurate and fluent basic decoding skill– a lot of reading to acquire functional reading skill– appropriate vocabulary knowledge, motivation
and strategy to comprehend the written language
Development of nonword reading accuracy during 1st Grade in different orthographies
(Scottish data up to 2nd grade)
0
20
40
60
80
100
TP0 TP1 TP2 TP3 TP4 TP5
% C
orr
ec
t
Greek
Portuguese
French
Finnish
Scottish
Spanish
Results from COST A8, Seymour, et al. 2003
Before school and then in equal time steps to the end of the 2.grade
Reading acquisition and the consistency of the connections between spoken and written language
• If the reading instruction is organized as it should… the time child needs for the acquisition of the basic reading skill is the shorter – the smaller the number of connections one has to
learn (Finnish, Swedish.. < 30 letter-sounds)– the more consistent are the connections, ie. the
fewer complexities/alternatives need to be learned and
– the more optimally built phonics instruction is available for successfully engaging the child in training the skill
Jyväskylä Longitudinal study of Dyslexia (JLD; 1993-)
The JLD research groupMikko Aro, Timo Ahonen, Kenneth Eklund, Tomi Guttorm, Jarmo Hämäläinen,
Ritva Ketonen, Marja-Leena Laakso, Seija Leinonen, Matti Leiwo, Paavo Leppänen, Paula Lyytinen, Kurt Muller, Anna-Maija Poikkeus, Anne
Puolakanaho, Ulla Richardson, Paula Salmi, Asko Tolvanen, Minna Torppa, Helena Viholainen
..when biological factors compromise reading acquisition..
The goals of the JLD to identify (from children at familial risk for dyslexia)
•precursors of dyslexia
•predictors of compromised acquisition
•developmental paths leading to dyslexia
The last step: the development of preventive measures
DEFINING FAMILIAL RISK IN THE JYVÄSKYLÄ LONGITUDINAL STUDY OF
DYSLEXIA (JLD)
SELECTION CRITERIA FOR THE AT-RISK FAMILIES: parents
• At least one parent with diagnosed dyslexia from multiple criteria• Reported dyslexia among at least one of the first degree
relatives• IQ at least 85 (Raven matrices)• No reported language problems in childhood or later• No neurological or psychiatric symptoms• No hearing problems
For detals, see Leinonen et al. Reading and Writing, 2001
N=
108
N=
107
N=
107
N=
107
N=
107
N=
107
N=
95
N=
96
N=
94
N=
95
N=
93
N=
93
18 month
2 years
2½ years
3½ years
4½ years
5years
5½ years
N=
107
N=
93
Collection of the data continues
N=
108
N=
107
N=
112
Neo-nata
l
6 month
14 mont
h
N=
96
N=
94
N=
94
6½ years
N=
107
N=
93
Igrade
IIgrade
N=
108
N=
92
IIIgrade
N=
107
N=
92
I Screening II Screening III Screening
Short questionnaire administered
at the maternity
clinics
N=8427 parents
Compre-hensive
questionnaire
N=3146 parents
Assessment of parents’
reading and spelling
skills
N=410 parents
AT -RISK GROUP
N=117 infants
CONTROL GROUP
N=105 infants
Born at the
hospitals of Central Finland during
01.04.93-31.07.96
N= 9368 infants
Number of children who have attended the last originally agreed assessment phase at the 3rd grade
AT -RISK GROUP
N=108 children
CONTROL GROUP
N=92children
N = 1515
N=
108
N=
92
CLASSMATES N = 2859
VIIgrade
N=
85
N=
66
VIIIgrade
N=
101
N=
81
N = 1537
IXgrade
N=
76
N=
73
N = 1508
Criteria of dyslexia among children
Reading speed• Reading 3 and 4 syllabic words and non words – mean response time• Reading text – read words / minute• Reading non word text – read words / minute• Reading word list, standardized test – correctly read words in two minutesReading and writing accuracy• Reading 3 and 4 syllabic words and non words – correct / 40• Reading text – percentage of correctly read words• Reading non word text – percentage of correctly read words• Spelling words and non words – correct / 18
Measures taken at the end of the 2th grade at the averate age of 8 y. 9 months
CriteriaA child was diagnosed as dyslexic, if he / she scored below or at
the 10 percentile of the contol group in at least1. Three out of four speed measures OR2. Three out of four accuracy measures OR3. Two speed AND two accuracy measures
The reading status of children born at familial risk for dyslexia
at school age
• Expectation of the genetic influences– > 1/2 would be affected (due to 1 parent’s dyslexia)
• The observed result: 42 /107– compromised initial reading acquisition 48 / 107– severe, persistent reading disorder 42 / 107
N=108
N = 38
1st gr 2nd gr
At risk group
Children with reading disability
3rd gr 8th gr
N=92
1st gr 2nd gr
Control group
3rd gr 8th gr
N = 38 N = 36 N = 42
N = 10 N = 9 N = 10 N = 12
Age Variable
7 - yrs Reading accuracy & speed D
5 - yrs Naming speed P & D
4 - 6 yrs Phonological manipulation P & D
5 - 6 yrs Letter knowledge P & D
5 - yrs Verbal memory P & D
3 - 6 yrs Phonological sensitivity P & D
3 - 5 yrs Inflectional skills P & D
2 - 3 yrs Articulation accuracy P
2 yrs Maximum sentence length P & D
6 mth Speech perception P & D
Birth ERP to speech sound P & D
IDENTIFYING & PREDICTING RISKa summary of significant measures
P = Predictors D = Differences between groups
Lyytinen et al., Annals of Dyslexia, 2004; Dyslexia, 2004; Sage Handbook of Dyslexia, 2008
From: F3, F4, C3, C4, P3, P4 (Ag/AgCl-electrodes), referred to ipsilateral mastoid
Bandpass: 0.5-35 Hz, sampling rate 200 Hz
METHODS – ERP recording
Prediction for the very early ERPs
Predictors: ERP responses to speech sounds which significantly differentiate children with and without risk
Criterion measures: The most important language measures that predict early reading skills and early reading
Newborn ERPs in the at-risk group
/ba/ /da/ /ga/
Guttorm, et al. (2005) Cortex 41, 291-303.
F4
C3 C4
P3 P4
EogL
500ms
EogR
5 µV _+
F3
540-630 ms
Writing letters 6.5 y -.336*
Reading 6.5 y -.329*
Audio-phonemic associations 5.5 y-.451*
Boston Naming Test 5.5 y-.311*
Word identification 6.5 y-.308*
Word identification 6.5 y-.339**
Letter identification 6.5 y-.339*
Phoneme Deletion 6.5 y-.339*
Naming 1st Phoneme 6.5 y-.342*
Rapid Naming: Faults 6.5 y.414*
Word identification 6.5 y-.415**
3.5 4.5 5 5.5 6.5
Age (years)
0
5
10
15
20
25
30
0,85
2,68
3,74
6,21
14,03
3,09
10,41
13,57
16,59
25,41
Reading acquisition fails during 1. grade
Reading acquisition normal during 1. grade
The letter knowledge of 3.5-6.5 year olds (JLD) and reading acquisition
Lettter
names known
Lyytinen et al., (2007)Nordic Psychology
Reading composite 2. gr.
Reading composite, 1. gr.
IQ, 5 y.
Letter knowledge 5.5 y.
Letter knowledge, 5 y.
Letter knowledge, 4.5 y.
Letter knowledge, 3.5 y.
Rapid naming, 6.5 y.
Rapid naming, 5.5 y.
Phonological skills, 5.5 y.
Phonological skills, 4.5 y.
Phonological skills, 3.5 y.
Pseudoword repetition, 3.5 y.
Receptive speech, 2.5 y.
-3 -2 -1 0 1
z-score (mean = 0, sd =1)
Individual profiles of the prediction measures of the JLD children whose reading acquisition was most severely compromised
Lyytinen, et al.Scand. J. ofPsychology,2009.
From the JLD-follow-up from birth to school age of reading-related development
Precursors/predictors: conclusions• Familial background increases the risk of dyslexia
substantially – relatively the more so, the more severe reading difficulties are attended
• Speech perception is predictive from 6 months and does so at school age still after controlling for other known predictors
• Among at risk children very early language delays can be informative, both in the expressive and receptive language domains but receptive language may be more important
• Poor letter name learning predicts without false negatives (false positives should be accepted)
• Naming fluency predicts the most persistent difficulties …also when the phonological skills revealed by traditional assessment tools fail to predict
How to support reading acquisition among at risk
childrenAn enjoyable learning game:
Graphogame
GraphogameThe task: Catch the letter that matches the sound you hear!
Competitor’s catcher
Player’s catcher
Falling letters
Correctly chosen letters
Player’s resultsCompetitor’s results
Mouse pointer
Programming: Tuomo Hokkanen
GraphoMath
A learning environment for L1 and L2 spoken and written languages: Graphogame
Introduces reading skill of any wanted language
Teaches the phonetic basis of language with the help of written language– Tunes the speech perception for the use of a
wanted language– Helps in training correct pronunciation of the
sounds of a language– Introduces spoken words (vocabulary) via
written language
How and where Graphogame works
• Applies phonics: trains the connections between spoken and written items in optimized order:
from easy to differentiate to more similar phonemic units…
from letter-sounds to syllables and words• Adapts automatically to child’s actual skill level• Guarantees experience of success (80%)
The cumulative number of learned items
Hours of playing
Exemplary learning curves of 4-8 year olds (N=726)
Modelling: Janne Kujala
After 2. trainingsession
After 1. trainingsession
Before training
Cor
rect
ly a
sse
mb
led
soun
dst
22
20
18
16
14
12
10
Order of training sessions:
1.Math game – 2.Letter-sound g.
1.Letter-sound g.– 2. Math game
Ability to assemble sounds on the basis of letters
Screeningtest
(N=166)
CARRIgroup(n=25)
Main-streamgroup (n=116
)
RRI group (n=25)
Subtest2
Subtest3
Subtest4
Subtest5
Post test
Follow-up 1
IQ Estimatio
n
Remedial reading intervention (RRI)(T1-T6)
Remedial reading intervention and computer-
assisted instruction (CARRI) (T1-T6)
Screening
August Grade 1
Groupping
September
Grade 1
T1OctoberGrade 1
T2Decembe
r Grade 1
T3JanuaryGrade 1
T4March
Grade 1
T5May
Grade 1
T6August
Grade 2
T8August Grade 3
CARRI group = Computer assisted remedial reading intervention groupMainstream group = Mainstream reading instruction groupRRI group = Remedial reading intervention group
T7May
Grade 2
Follow-up 2
Saine et al., Child Development, 2011
(=1/4 of the remedial reading support session)
Saine et al., Child Development, 2011
0
5
10
15
20
25
30
35
T5 T6 T7 T8
Num
ber
of W
ords
Spe
lled
Cor
rect
lySpelling
RRI
CARRI
Mainstream
Saine et al., Child Development, 2011
Successful preventive practice
Effective if not used too early, starting just before child enters school– practiced > 1 times per day in subsequent days– short < 12 minutes’ sessions– playing in so ”active” form as possible (by e.g. repeating
the sounds)– the task of parents: to show that they are happy when
child plays– playing long enough (2-20 hours)
See: www.lukimat.fi (where Finnish children play) or www.graphogame.com for description and demo in English
For explanations, see Lyytinen et al., 2009
Illustration of the game developed byJanne Kujala
GG training of <5 hours affects brain
Pre-Post GG: Children (n=15) before and after playing with Graphogame
BA18/19
LG-FG, IFG
No difference Condition differences
Post-pre interaction between groups playing Graphogame vs Mathgame (same with numbers): p<0.005
Condition differences Increased activation in occipito-temporal areas
Words-False fonts
HL and UR in collaboration with Swiss colleagues Daniel Brandeis, Sylvia Brehm
Brem et al., PNAS, 2010, 107(17), 7939-7944.
Potential assessment use of Graphogame
• Dynamic assessment:– Online follow-up of the proceedings of the
training of the letter-sound connections– Application of the observed results to guiding
the next steps of the practice towards contents still in need of further practice
i.e. integrating assessment and intervention as made in the response-to-intervention model …note, the cycle of refocussing the intervention can happen in seconds
Supporting Finnish children
• All children entering school this autumn are screened – for familial risk and/or letter knowledge < 7
Motivating children showing such a risk
to play dynamic assessment version of Graphogame > resistance to training with optimal phonics
Children showing resistance (1-2%) followed individually for testing the limits of the Graphogame
Note: percentiles 3-15% trained successfully
GRAPHOLEARN model
• Ekapeli/Graphogame used under the responsibility and funding of the Ministry of Education in Finland
• > 200.000 users (2006-)• Centralized automatized
feedback from our servers• Could work as main model
for implementations
elsewhere as well
The basic principles of Graphogame development for a new writing system
• Careful study of the written language environment with local experts for developing and implementing an appropriate content
• Evidence-based documentation of the efficiency of the game of the content before any distribution
• Distribution and use under the responsibility of the local Ministry of Education after research has shown its efficiency in a new orthographic/cultural environment
For more.., please,• Call: +358 50 552 4892• Have a look of our research: heikki.lyytinen.info• Ask for reprint(s): [email protected]
• The service for Finns: http://www.lukimat.fi/• ..in English: http://www.graphogame.com• See also grapholearn.info for the whole approach
Thank you for your attention!