Development of CDIs for rural Africa
K. Alcock
K. Rimba
A. Abubakar
P. Holding
Assessing vocabulary and
grammatical development
• “Clinical” testing
– e.g. Picture vocabulary tests, TROG, elicited speech
– Carter et al.
• Laboratory testing (HA!)
• Spontaneous speech samples
• Checklists
– Fenson et al.
– Bornstein et al. crosslinguistic
Bornstein et. al 2004
• Data from 7 languages
• Checklist similar to MacArthur/Bates but
shorter
• All have higher noun proportion. in 20mo
vocabulary
– Even Korean – unlike previous data
– Suggest child constraints lead to this
– But modified by language features
Two East African languages
• Kiswahili and Kigiriama
– Both spoken in coastal Kenya
– Town and. village families
Overall study
• Impact of HIV exposure on child
development
• Age 6 months to 3 years
• Language, motor and cognitive
development
• 200 families so far
Relevant sentence features
• Very variable word order
The child likes potatoes
Mtoto anapenda viazi – basic word order SVO
Anapenda viazi mtoto – common, VOS
Mtoto viazi anapenda – less common but possible, SOV
• Sentences can miss out subject
He likes potatoes
Anapenda viazi
• Rather like Italian in that respect
• Sentences can miss out subject
He likes (them)
Anapenda/Anavipenda
Data collection
• Adaptation of MacArthur CDI
• 175 families of children aged 8-15 mo
• Interview technique - not literate
Construction 1
• Macarthur-Bates CDI - younger (words &
gestures)
• Elimination of inappropriate words (snow,
penguin)
• Translation in parallel to both languages
Construction 2
• Adding words:
–
–
–
–
–
RAs experienced w. children
Mothers on staff & in community
Speech samples
Situational vocabulary (goat, maize porridge)
Function words
• Back translation & checking
Construction 3
• MacArthur-Bates CDI Older (Words and
Sentences)
• Same method with vocabulary
• Grammatical development:
–
–
–
–
–
Spontaneous samples
Examination of errors actually made
Enumeration of parts of speech children need to learn
Not possible to produce an order of learning in the time
But we will be able to tell what is learnt between 16-30
months
Validation
• Internal validity
• External validity
– Vocabulary and gestures – 9-15 months
– Vocabulary and grammar – 16-30 months
Production 8-15 mo
40
35
30
25
Lower conf interval
20
Mean
15
Higher conf interval
10
5
0
-5
8 mo
9 mo
10 mo
11 mo 12 mo
13 mo
14 mo
15 mo
Comprehension 8 - 15 mo
150
130
110
90
Lower conf interval
Mean
Higher conf interval
70
50
30
10
-10
8 mo 9 mo 10 mo 11 mo 12 mo 13 mo 14 mo 15 mo
Production 16-30 mo
895
795
695
595
Lower conf interval
495
Mean
395
Higher conf interval
295
195
95
30 m o
29 m o
28 m o
27 m o
26 m o
25 m o
24 m o
23 m o
22 m o
21 m o
20 m o
19 m o
18 m o
17 m o
16 m o
-5
Word combining 16-30 mo
1
0.9
0.8
0.7
0.6
0.5
Ever
0.4
0.3
0.2
0.1
0
16
mo
17
mo
18
mo
19
mo
20
mo
21
mo
22
mo
23 24
mo mo
25
mo
26
mo
27
mo
28
mo
29
mo
30
mo
Reliability and validity
• Internal reliability
– Alphas over .9 for most sections
– Alphas much over .9 for half of test
• Correlation with age
–
–
–
–
Total vocabulary (younger) .505
Gestures .679
Total vocabulary (older) .620
Grammatical affixes .469
• Effect of HIV
– When covary age, vocabulary approximately half that
of community children
Validity testing
• 9-15 months
• Challenge items
– Production, comprehension and gesture
– Low, medium, and high frequency
• Mother list of words
Challenge items
• No child produced any words in production
– Some incomprehensible sounds
• Comprehension and gesture more useful
– Significant correlations:
Gesture without prompt and gesture on CDI
Gesture without prompt and comprehension on CDI
Specific items tested in comprehension challenge
did not correlate with challenge comprehension
items
Mother lists of words
– Neither prompted nor spontaneous lists
correlate with:
•
•
•
•
Age
Gesture total
Language total
Challenge task
– Asked for words for actions/toys/things child
likes tends to give names of actions child can
do etc.
Conclusions for validity
• Test hugely internally valid
• Mothers better at observing gesture than
language?
• 10 challenge words very poor test?
• Mother spontaneous list paradigm poorly
understood
– CDI paradigm involves clearer explanation
– Reminder “can he say the word or does he just
understand the word”
Summary
• Reliability of test
• Validity
• Further work
– Spontaneous speech during cognitive testing session
• But no speech produced
– Spontaneous speech at home
– Written versus interview method for literate local
mothers
Descargar

First words in two East African languages