Bedre leseforståelse: Hvilke tiltak har effekt?

advertisement
Improving reading comprehension:
Effects from interventions
Monica Melby-Lervåg
My talk
1. The foundation of reading comprehension
2. Reading comprehension and dyslexia
3. To examine the effects from an intervention
4. Effects from interventions directly targeting
reading comprehension
5. Effects from interventions targeting
underlying components of reading
comprehension (general cognitive processes,
decoding, vocabulary).
1. The foundation of reading
comprehension
Background
Reading comprehension = Word decoding × Linguistic
Comprehension
Morphology?
Inference
skills?
working
memory?
Syntax?
The study
198 unselected Norwegian speaking children
Assessment scheme
Middle
of 2nd
grade
End of
2nd
grade
Middle End of
of 3rd
3rd
grade grade
Middle Middle
of 6th of 7th
grade grade
Lervåg & Melby-Lervåg, work in progress
Inf. Skills
Residual
Inference
Skills
Morph.
Gen..
Residual
Morpheme
Generation
Syntac.
Skills
Residual
Syntactic
Skills
Vocab.
Width
Residual
Vocabulary
Width
Vocab.
Def.
Residual
Vocabulary
Definitions
Word
Decoding
Linguistic
Comprehension
3.55**
1.02**
2.21**
.735*
.622*
Listening
Recall
TOWRE B
TOWRE A
.614**
Working
Memory
Backward
Digit Recall
.058*
Reading
Comprehension
Initial Status -.20**
NARA T1
NARA T2
Reading
Comprehension
Early Growth
NARA T3
NARA T4
Reading
Comprehension
Later Growth
NARA T5
NARA T6
2. Reading comprehension and dyslexia
FAMILY RISK OF DYSLEXIA
Study
Meta-analysis of studies examining reading
comprehension and underlying skills in children
with dyslexia
(Snowling & Melby-Lervåg, submitted)
A systematic search detected 123
studies that used a family risk
methodology to study reading
disorders
Effect size
Effect size
Cohens d
Example d = -1.00
Toddlers (1- 3.5 years)
Results
Group difference d (STD units)
2
1.5
Family risk children with
dyslexia vs controls not at
–risk
Family risk children
without dyslexia vs
controls not at –risk
1
0.5
0
-0.5
-1
Articulatory accuracy
Expressive language
Receptive language
Preschool
1.6
Family risk children
with dyslexia vs
controls not at –risk
Group difference d (STD units)
1.4
1.2
1
Family risk children
with out dyslexia vs
controls not at –risk
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
Articulatory
accuracy
Receptive
vocabulary
grammar
Phoneme
awareness
Rapid
naming
Primary school
4
Family risk children with
dyslexia vs controls not at
–risk
Group difference d (STD units)
3.5
3
2.5
Family risk children with
out dyslexia vs controls
not at –risk
2
1.5
1
0.5
0
-0.5
-1
Expressive
vocabulary
Phoneme
awareness
Rapid
naming
Word
decoding
Reading comprehension
3. To examine the effects from an
intervention
Chose a
group of
children
Pretest
Intervention
Posttest
Chose a
group of
children
Pretest
Intervention
Posttest
Chose a
group of
children
Randomize the
children in a
training and an
intervention
group
Pretest
Pretest
Intervention
No
intervention
/irrelevant
intervention
Posttest
Posttest
Study:
A syntehesis of meta-analyses
Melby-Lervåg, Lervåg & Hulme, work in progress.
Method
Systematic search for reviews of educational interventions that have used a
quantitative summary of results after 1998
The meta-analysis had to examine an intervention that could in some way inform
about amelioration of difficulties related to:
Decoding, reading comprehension, language skills, mathematic skills, general
learning disorders, attention/hyperactivity, other behavioral/emotional problems
or bullying.
The meta-analysis had to provide a mean effect size of an academic achievement
or behavioral outcome that was based on a group design (i.e. meta-analyses
purely based on single case studies were excluded)
70 meta-analyses included, 3145 single studies
Melby-Lervåg, Lervåg & Hulme, work in progress.
Mean effect size in meta-analyses
Differences in mean effect size for
different designs
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
RCT
QED with
control
group
No control
group
Only 233 of the 3145 intervention studies
were randomised controlled trials.
Serious methodological weaknesses,
studies not suited to conclude about
intervention effects
5. Effects from interventions targeting reading
comprehension or underlying components of
reading comprehension
A. Interventions targeting reading
comprehension directly
Clarke, Snowling, Truelove og Hulme (2010)
Compared three interventions for 160 children in 4th grade.
Selected on the basis of a reading comprehension screening of
1200 children.
Intervention:
1. Linguistic
comprehension
2. Reading comprehension
3. Combined
-Vocabulary (60
new words)
-Narratives
-Oral language use
-Listening
comprehension
-Meta-cognitive strategies (repeated reading,
A combination
thinking aloud, visualisation)
(50/50 %).
-Use of these strategies when reading text and in
questions
-Inferences, rcognise and use
- Narratives and txt production
Results
Figure from the paper:
B. Interventions targeting reading
comprehension indirectly through Domain
General Cognitive Skills
Effects from computerised working
memory training
Study
Redick, Melby-Lervåg & Hulme (work in
progress).
2012: 23 studies
New study: 82 studies with 102 independent
experiments
Inf. Skills
Residual
Inference
Skills
Morph.
Gen..
Residual
Morpheme
Generation
Syntac.
Skills
Residual
Syntactic
Skills
Vocab.
Width
Residual
Vocabulary
Width
Vocab.
Def.
Residual
Vocabulary
Definitions
Word
Decoding
Linguistic
Comprehension
3.55**
1.02**
2.21**
.735*
.622*
Listening
Recall
TOWRE B
TOWRE A
.614**
Working
Memory
Backward
Digit Recall
.058*
Reading
Comprehension
Initial Status -.20**
NARA T1
NARA T2
Reading
Comprehension
Early Growth
NARA T3
NARA T4
Reading
Comprehension
Later Growth
NARA T5
NARA T6
Results
Decoding
Studies
Treated
controls
Untreated
controls
Mean effect size d immediatly after
training
Verbal abilities
Studies
Treated
controls
Untreated
controls
Mean effect size d immediatly after
training
Reading comprehension
Studies
Treated
controls
Untreated
controls
Mean effect size d immediatly after
training
Similar findings for auditory processing training
C. Interventions targeting reading
comprehension indirectly through
decoding/phonological awareness
Numerous of well controlled studies have shown
that phonological awareness in combination
with letter knowledge training can improve
word decoding skills………..
Unfortunatly, not that many have
reported transfer effects to standardised
tests of reading comprehension
10 studies met inclusion criteria for word decoding. Effects were moderate:
0.47 SD better (95% CI 0.06 to 0.88)
Only three studies reported data on transfer effects to reading comprehension:
0.14 SD better (95% CI -0.46 to 0.74)
RCTs that combine phonological
awareness/letter knowledge and vocabulary
intervention shows promising effects on reading
comprehension
Wolff, 2011
(d = 0.41, lasted at
follow up)
D. Interventions targeting reading
comprehension indirectly through
vocabulary/linguistic comprehension
Linguistic comprehension intervention
Three times a week, (2 x 45 minutes, 1 x 10 min
individually).
Narrative skills
Dialogical reading
Expressive
language tasks
Vocabulary
instruction
115 second language learners randomised in two groups. Training group
received 20 weeks of intervention
Rogde, Melby-Lervåg & Lervåg (submitted)
Vocabulary embedded in the training program
20
18
16
14
12
10
8
6
Group CONTROL
4
Group TRAINING
d = 0.53** immediatly after
training,
d = 0.44* follow up
2
1
0
0
2
Distal measures: Do the effects of training transfer to standardized tests
of expressive language?
20
15
18
14
13
16
12
11
14
10
12
9
8
10
7
8
6
5
6
Group CONTROL
4
3
Group TRAINING
Group TRAINING
2
2
d = 0.51** immediatly after
training,
d = 0.28 (p = 0.064) follow up
2
1
0
0
2
1
1
0
0
Group CONTROL
4
Distal measures: Do the effects of training transfer to standardized tests
of receptive language?
75
60
70
55
65
50
60
55
45
50
40
45
35
40
30
35
25
30
20
25
Group CONTROL
Group CONTROL
20
15
15
10
Group TRAINING
10
d = 0.02 immediatly after
training,
d = 0.06 follow up
2
1
2
1
0
0
0
5
5
0
Group TRAINING
My talk
1. The foundation of reading comprehension
2. Reading comprehension and dyslexia
3. To examine the effects from an intervention
4. Effects from interventions directly targeting
reading comprehension
5. Effects from interventions targeting
underlying components of reading
comprehension (general cognitive processes,
decoding, vocabulary).
Take-home message 1
From family risk studies of dyslexia it is clear that children
with dyslexia have impared word decoding and reading
comprehension skills. From an early age they also have
poor phonological awareness and often also broader
language skills.
From longitudinal studies we know that vocabulary,
grammar skills and word decoding are uniqly related to
the growth in reading comprehension.
Thus, for interventions to be succsessfull they should
focus on these areas.
Take-home message 2
Effects on reading comprehension can be
obtained by interventions either focusing on
reading comprehesion directly (strategies etc),
through linguisitc comprehension/vocabulary or
through decoding/phonological awareness.
But…
Effects are not easily obtained. Requires hard
and systematic work over time.
Take-home message 3
Reading comprehension is the ultimate goal of
literacy. Therefore more studies should focus on
this and measure this with standardised tests that
has good psychometric properties.
There is a great need in education for high quality
randomised intervention studies. Many studies
have used poor designs, too few participants and
interventions lacking a theoretical and empirical
rationale. This has given us misleading results and
lead us astray.
Thank you for the attention!
Foto: Kathrine Nordli, «Airborne»
Download