Language basis of reading comprehension difficulties

advertisement
Language Basis of Reading Comprehension
Difficulties
Hugh W. Catts, Ph.D.
University of Kansas
NIU 2013
Outline
• Model of reading comprehension
• Language basis of comprehension difficulties
• Evidence for language basis
• Early identification
Reading Comprehension
The construction of a mental model or
meaning representation based on a printed
text and one’s prior knowledge
Mental
Model
Mental Model
Meaning Representation
Coherent Understanding
Situation Model
Text
Knowledge
Context
• Purpose
• Motivation
• Mental Alertness
Mental
Model
Standard of Coherence
Text
Knowledge
Knowledge
Miafirst
ensimmäisen
jokaMental
Mia
let loose a joukkueen,
team of gophers.
The
muodostuu
plan
backfiredgophers.
when a Kun
dog koira
chased them
Model
suunnitelman
laukaukseksi
away.
She thenepäonnen
threw
party
butdid
theMia put the
1. aWhere
pannut
toimeen.
Sentheir
jälkeen
hän
guests
failed
to bring
motorcycles.
järjestivät juhlat
vieraita
ei olenot
Furthermore,
her mutta
stereogophers?
system was
niiden
moottoripyöriin.
Lisäksicalls
hänen
loud
enough.
Obscene phone
gave
stereojärjestelmä
eithe
oleWhy
tarpeeksi
suuri.want the guests
her
some hope until2.
number
didwas
Mia
Text
Knowledge
HävytöntäIt puheluita
piti
joitakin
niin
changed.
was
the installation
of
the
to bring their motorcycles?
kauan numero
annettiin
blinking
neon lights
acrossmuuttaa.
the street that
Asennus
neonvalot
finally
did oli
thevilkkuva
trick. She
framed kadun
the ad
3.
What
did
the
ad
toiselle
loppujen
lopuksi
from
the puolelle
classifiedettä
section
and now
hasoli.
it say?
Hän muotoili
ad siitä nyt
hanging
on herkysymystään
wall.
luokiteltujen ja joka hänen.
GETTING RID OF
BAD NEIGHBORS
Mental
Model
Text
Knowledge
Mental
Model
Text
Language
Speech
Language
Comprehension
Knowledge
Simple View of Reading
(Gough & Tunmer, 1986; Hoover & Gough, 1990)
Reading Comprehension
Word
Recognition
x
Language
Comprehension
SEMANTICS & GRAMMAR
(vocabulary, syntax)
TEXT PROCESSING
(text structures, cohesion)
BACKGROUND KNOWLEDGE
(facts, concepts, etc.)
VERBAL REASONING
(problem solving, inferencing)
METACOGNITION
(comprehension strategies)
PHONOLOGICAL AWARENESS
(syllables, phonemes, etc.)
DECODING (alphabetic principle,
spelling-sound correspondences)
SIGHT RECOGNITION
(of familiar words)
Adapted from Scarborough, H. S. in Neuman, S.B. & Dickinson, D. K. (2001). Handbook of Early
Literacy Research. New York: Guilford Press.
Variance Explained
in Reading Comprehension
1
0.8
0.6
0.4
0.2
0
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Variance Explained
in Reading Comprehension
100%
80%
60%
40%
68%
20%
0%
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Variance Explained
in Reading Comprehension
100%
80%
60%
40%
68%
20%
50%
40%
0%
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Variance Explained
in Reading Comprehension
100%
80%
9%
21%
60%
40%
68%
20%
50%
36%
40%
0%
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Unique & Shared Variance in Reading
Comprehension
100%
80%
9%
21%
60%
40%
36%
41%
38%
20%
27%
0%
2nd Grade
Word Recognition
38%
12%
4th Grade
8th Grade
Shared
Listening Comprehension
Simple View (SEM)
Construct
First Grade
Second Grade
Third Grade
Word Recognition
.81
.48
.48
Listening Comp
.22
.57
.60
Preliminary results from the LARRC Consortium
Language Basis of Reading
Comprehension
Specific Language Impairment (SLI)
• A disorder that involves delayed onset and protracted
development of language (including morphosyntax,
semantics, phonology, or pragmatics) relative to other
areas of development
• Generally identifiable during the preschool years (3 to 5
years of age) Tager-Flusber & Cooper (1999)
Iowa Longitudinal Study
• Identified 225 children
with LI (123 SLI, 102 NLI)
and 379 without LI in
kindergarten (age 5-6
years)
• Drawn from an
epidemiologic sample of
over 7000 children
• Followed in 2nd, 4th, 8th,
and 10th grades
• Word Recognition
Word Identification
Word Attack (WRMT-R)
• Reading Comprehension
WRMT-R Passage Comp Gray
Oral Reading Test-3
Diagnostic Achievement
Battery (QRI, 8th,10th)
Percentage of Reading Disorders
(Reading Comprehension > 1 SD below the mean)
Percentage of Reading Disorders
(Reading Comprehension > 1 SD below the mean)
100
90
80
70
60
50
40
30
20
10
0
2nd
4th
8th
10th
Grade Grade Grade grade
SLI
NLI
Normal
Percentage of Reading Disorders
(Reading Comprehension > 1 SD below the mean)
Relative Risk
100
90
80
70
60
50
40
30
20
10
0
2nd
4th
8th
10th
Grade Grade Grade grade
SLI
NLI
Normal
2nd
4th
8th
10th
SLI
5.0
4.8
5.3
5.5
NLI
7.8
8.2
7.9
9.2
SLI
NLI
Poor Comprehenders
Deficits in reading comprehension but
normal decoding
“specific comprehension deficit”
Poor Comprehenders
• Used arbitrary cut-offs that were subject to error
• Artificially created subgroups where none exist
Latent Class Analysis
• “Classifies” subjects on the bases of multiple measures of
reading comprehension and word recognition and provides
probability of class membership
• Constrained groups on the basis of 8th grade scores
Constraints
8th RC
Poor comprehender
Poor decoder
Generally poor
Generally good
Average
8th WR
-1
1
-1
1
1
-1
-1
1
free to vary
Latent Class Analysis
• “Classifies” subjects on the bases of multiple measures of
reading comprehension and word recognition and provides
probability of class membership
• Constrained groups on the basis of 8th grade scores
• Classified students on basis of 8th and 10th grade scores
Average Class Probabilities
Class
1
2
3
4
5
1
0.82
0.00
0.05
0.00
0.13
2
0.00
0.80
0.02
0.05
0.13
3
0.02
0.00
0.92
0.00
0.06
4
0.00
0.00
0.00
0.97
0.03
5
0.03
0.01
0.04
0.04
0.88
Class Size
(weighted)
Poor Comprehenders
Poor
Decoders
Generally Poor
6.5%
3.7%
19.4%
6.5 +19.4 = 25.9%
6.5/25.9 = 25.1%
Generally High
25.7%
Average
44.8%
Kindergarten Language Abilities
1.5
1
0.5
Vocabulary
0
Grammar
PC
-0.5
-1
-1.5
PD
Poor
Good
Average
Narration
History of Language Impairments (K)
Poor Comprehenders
Poor Decoder
Generally Poor
Generally Good
Average
27.4% (22.5 SLI, 4.9 NLI)
0%
47.9% (20.1 SLI, 27.8 NLI)
.3% (.3% NLI)
7.4% (5.1 SLI, 2.3 NLI)
Poor Comprehenders
• Often did not have a reading problem until later
in school
• Normal readers in 2nd grade
• Problems emerged in 4th grade
• Has been described “Fourth grade slump”
Late-Emerging Poor Reader
Late-Emerging Poor Readers
• 493 participants from Iowa study
• Multiple measures of reading comprehension and word
reading at 2nd, 4th, 8th, and 10th grades
• Used Latent transition analysis
Classes
Grade 2
TD
RD
TD
TD
TD
RD
RD
RD
→
→
→
→
→
→
→
→
Grade 4
TD
RD
RD
TD
TD
TD
RD
RD
→
→
→
→
→
→
→
→
Grade 8
TD
RD
RD
RD
TD
TD
TD
RD
→
→
→
→
→
→
→
→
Grade 10
TD
RD
RD
RD
RD
TD
TD
TD
Late-Emerging
•
•
•
•
•
About 70% were normal readers at all grades
17% were poor readers with early and persistent deficits
13% had late emerging deficits (42% of all poor readers)
Emerged by 4th grade and tended to be stable thereafter
Most of these children had comprehension problems (65%)
and many had a history of language impairments (46%)
Can we predict who will be a lateemerging poor reader or a poor
comprehender based on earlier
language skills?
Early Identification
Means “predicting the future”
“It is tough to make predictions, especially
about the future.”
Yogi Berra
Yogi Berra
“It’’s like deja-vu all over
again.”
“Nobody goes there anymore,
it’s too crowded.”
“Always go to other people’s
funerals, otherwise they won’t
come to yours.”
Prediction of RD
Catts, H., Fey, M.E., Zhang, X., & Tomblin, J.B. (2001). Estimating risk for
future reading difficulties in kindergarten children: A research-based model
and its clinical implications. Language, Speech, and Hearing Services in
Schools, 32, 38-50.
Adlof, S. A., Catts, H.W., Lee, J. (2010). Kindergarten predictors of second vs.
eight grade reading comprehension impairments. Journal of Learning
Disabilities, 43, 332-345.
Catts, H., Nielsen, D., & Bridges, M. (in progress). Early identification of
reading disabilities within a RTI framework.
Predicting reading outcomes
• classified children from Iowa sample as good and poor
readers (-1SD)
• Based on 2nd grade or 8th grade reading comprehension
• Kindergarten (spring) measures to predict outcomes
Kindergarten Predictors
•
•
•
•
•
•
•
•
•
•
Letter identification
Phoneme deletion
Rapid naming
Sentence imitation
Vocabulary production/comprehension
Grammatical understanding
Grammatical completion
Narrative production/comprehension
Mother’s education
Nonverbal IQ
Predicting reading outcomes
• Grouped children from Iowa sample into good and poor
readers (-1SD)
• Based on 2nd grade or 8th grade reading comprehension
• Kindergarten predictors (spring)
• Logistic regression (best-subsets variable-selection)
Second Grade Predictors
Sentence imitation
Letter identification
Mother’s education
Rapid naming
Phoneme deletion
Reading Comprehension
Sensitivity = 92% Specificity = 80%
AUC = .90
Eight Grade Predictors
Grammatical completion
Nonverbal IQ
Sentence imitation
Phoneme deletion
Mother’s education
Narrative production
AUC = .87
Reading Comprehension
Catts, Nielsen, & Bridges (in progress)
• 366 kindergarten children
• Administered a screening battery at the
beginning of the year
• Included a range of language measures
• Measured reading outcomes at the end of 1st-3rd
grades
WRMT:R Passage Comprehension
Step
1
Variables
Word recognition (2nd)
R2
.576
p
<.001
WRMT:R Passage Comprehension
Step
Variables
1
Word recognition (2nd)
2
Receptive Vocabulary (K)
3
Expressive Vocabulary (K)
4
Receptive Narration (K)
5
Expressive Narration (K)
6
Sentence Imitation (K)
R2
.576
R2
change
p
<.001
WRMT:R Passage Comprehension
Step
Variables
R2
R2
change
p
1
Word recognition (2nd)
.576
<.001
2
Receptive Vocabulary (K)
.673
.097
<.001
3
Sentence Imitation (K)
.682
.001
<.003
Gates-MacGinitie
Step
1
Variables
Word recognition (2nd)
R2
.445
R2
change
p
<.001
Gates-MacGinitie
Step
Variables
R2
1
Word recognition (2nd)
.445
2
Receptive Vocabulary (K)
.520
3
Sentence Imitation (K)
4
Expressive Vocabulary (K)
5
Expressive Narration (K)
6
Receptive Narration (K)
R2
change
p
<.001
.075
<.001
Language Intervention
• At-risk children
• Randomly assigned to intervention condition
• Vocabulary and narrative intervention
• Pre- and post-test measures
WRMT:R Passage Comprehension (N=113)
WRMT:R Passage Comprehension (N=112)
Step
Variables
R2
R2
change
p
1
Word recognition (2nd)
.621
<.001
2
Vocabulary (pre)
.689
.068
<.001
3
Narration (pre)
.691
.002
>.05
WRMT:R Passage Comprehension (N=112)
Step
Variables
R2
R2
change
p
1
Word recognition (2nd)
.621
<.001
2
Vocabulary (pre)
.689
.068
<.001
3
Narration (pre)
.691
.002
>.05
4
Vocabulary (post)
.700
.009
.07
5
Narration (post)
.719
.019
<.01
Gates-MacGinitie (N= 112)
Step
Variables
R2
R2
change
p
1
Word recognition (2nd)
.364
<.001
2
Vocabulary (pre)
.462
.098
<.001
3
Narration (pre)
.471
.009
>.05
Gates-MacGinitie (N= 112)
Step
Variables
R2
R2
change
p
1
Word recognition (2nd)
.364
<.001
2
Vocabulary (pre)
.462
.098
<.001
3
Narration (pre)
.471
.009
>.05
4
Vocabulary (post)
.495
.024
<.05
5
Narration (post)
---
---
---
Thank you
Iowa Longitudinal Study
Evidence for Simple View
• Aaron, Joshi, & Williams
• Catts, Adlof, & Hogan, 2005
• Hoover & Gough, 1990
P (RD) > P (RD)
90%
Non-RD
RD
Percentage of Reading Disorders
(Word Recognition > 1 SD below the mean)
Relative Risk
100
90
80
70
60
50
40
30
20
10
0
2nd
4th
8th
10th
Grade Grade Grade grade
SLI
NLI
Normal
2nd
4th
8th
10th
SLI
3.1
3.3
3.7
2.9
NLI
5.6
5.9
5.8
4.7
Poor Comprehenders
80
70
60
50
Typical
40
NLI
SLI
30
20
10
0
2nd Grade
4th Grade
8th Grade
10th grade
.90
True Positive
.80
False Positive
Mental
Model
Text
Listening
Comprehension
Language
Speech
Knowledge
Variance Explained
in Reading Comprehension
100%
80%
60%
40%
68%
20%
50%
0%
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Variance Explained
in Reading Comprehension
100%
80%
9%
60%
40%
68%
20%
50%
40%
0%
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Variance Explained
in Reading Comprehension
100%
80%
9%
21%
60%
40%
68%
20%
50%
40%
0%
2nd Grade
Word Recognition
4th Grade
8th Grade
Listening Comprehension
Knowledge
Mental
Model
The procedure is actually quite simple.
First you arrange things into different
groups. Of course, one pile may be
sufficient depending on how much there
is to do. If you have to go somewhere
else due to lack
of facilities that is the
Text
next step, otherwise you are pretty well
set. It is important not to overdo things.
That is, it is better to do too few things at
once than too many.
WASHING
Knowledge
CLOTHES
Download