Linguistic differences in Mother Tongue Reading
Linguistic differences in Mother Tongue Reading Performance in Uganda
Rachel Jordan RTI International CIES 2018
•
Role of linguistic differences on program impacts
•
Challenges of program implementation in 12 local languages
Language in Uganda • Over 40 officially recognized languages • 1993 Local Language Policy • 2007 Thematic Curriculum BANTU
USAID/Uganda School Health and Reading Program 7 year USAID funded Working through MoES systems 12 Local Languages Large scale- 3,761 schools in 37 Districts Early Grades Reading approach Expanded through USIAD/Uganda LARA GPE UTSEP Build Africa ILeap
USAID/Uganda School Health and Reading Program
Systems and policies are strengthened
P1 to P4 pupil books and teacher guides
Teachers trained and supported
Advocacy for local language literacy instruction
Rigorous assessment of foundational reading skills
Learners read more fluently and understand more of what they are reading in both local language and English
5
USAID/Uganda School Health and Reading Program Cluster 1
Cluster 2
Cluster 3
Ateso
Leb Acoli
Ngakarimojong
Leblango
Lugbarati
Lhukonzo
Luganda
Lumasaaba
Lugwere
Runyankore- Rukiga
Runyooro-Rutooro
Lusoga
Language Complexity and Learning to Read Differences in reading outcomes in European languages is associated with variations in syllabic complexity and orthographic depth (Seymour et al. 2003) Learners’ ability to recall words is related to word length (Blake et al., 1994) Agglutinating nature of some Bantu languages means words are longer, and oral reading fluency may be lower (Abadzi, 2012)
Language Complexity in Uganda Language
Language Family
Average Word Length
Tonal Markings?
# of non-Latin letters
Comparative rank of language complexity
Acoli
Nilotic
3.6
Mark tone
6
High
Ateso
Nilotic
5.5
Mark tone
7
High
Leblango
Nilotic
3.8
Mark tone
18
High
Lhukonzo
Bantu
8.3
16
High
Luganda
Bantu
6.4
2
Low
Lugbarati
Sudanic
4.2
21
High
Lugwere
Bantu
7.1
6
Low
Lumasaaba
Bantu
6.0
10
Med
Lusoga
Bantu
7.0
11
Med
Ngakarimojong
Nilotic
4.8
6
High
Runyankore-Rukiga
Bantu
6.9
0
Low
Runyoro-Rutooro
Bantu
6.3
8
Med
Mark tone
Early Grades Reading Assessment Subtask & Measure
ABC
What is measured
Letter Sound Fluency
Correct letter sounds per minute (clspm)
Ability to identify sounds of letters
Oral Reading Fluency
Correct words per minute
Ability to read connected text
Percentage correct
Ability to comprehend reading passages
Reading comprehension Reading 20+ correct words per minute
Dichotomous: Child read 20+ cwpm or not
Early Grades Reading Assessment
Schools sampled at Coordinating Centre level 14 treatment/ 14 control 30 P1-P4 pupils randomly sampled (alpha cohort) Boys and Girls equally represented 1 on 1 oral assessment Approx. 15 minutes per language
11
SHRP Impact Average Local Language Words Read Per Minute (wpm) Beginning of P1 to end of P4 25 20
Luganda Program
23
LugandaControl
22 20
Leblango Program 15
15 13
Leblango Control Ateso Program
10
Ateso Control Run-Rukiga Program
5
0
6
Run-Rukiga Control Beginning P1 Feb2013
End P1 Oct2013
End P2 Oct2014
End P3 Oct2105
End P4 Oct2016
SHRP Impact 0.95
0.93
Overall Average Effect Size, by Language
0.82 0.72
Cluster 1
0.59
Cluster 2
0.45
Cluster 3
0.39 0.3
0.28
0.21
0.13
0.09
Language Characteristics and Impact of SHRP Language Issues
Effect size R2
Language Family
Rank of language category
Average word length
# non-Latin letters
0.19
0.40*
0.0
0.11
Implementation Issues
Effect size R2
Socioeconomic Issues
Years of implementation
% schools receiving CCT visits
% teachers trained
% pupils wearing shoes
% attended pre-school
Average class enrolment
0.04
0.01
0.11
0.12
0.39*
0.06
Language Characteristics and Impact of SHRP
Effect size R2
Language Issues
Implementation Issues
Socioeconomic Issues
0.71
0.23
0.41
Implementation
“Calvin and Hobbes” by Bill Watterson
Language is Politically Sensitive
First Languageof P1 SHRP teachers in Karamoja Region Ngakarimojong
61%
Ateso
29%
Luo
4%
Leblango
4%
Runyankore-Rukiga
4%
Languages don’t sit still
Multiple languages isn’t “cheap”
English Titles P1 (1) 2,245 copies @ 0.54 USD
1,210 USD
Local Language Titles P1 (8) 278 avg copies per title @ 2.75 USD avg
5,686 USD
19
“Good schools teach
in English”
20
Implications & Conclusion Language complexity is a significant predictor of SHRP impacts • More research to guide program adaptation • Discussions around national assessments and benchmarks
Reading programs in heterogeneous language settings face complex challenges- some strategies for confronting these challenges: • • • •
Staff on the ground and at every school Work at local as well as national levels Work through the system, so the system owns the processes Continuous dialogue
Rachel Jordan RTI International CIES 2018
•
Role of linguistic differences on program impacts
•
Challenges of program implementation in 12 local languages
Language in Uganda • Over 40 officially recognized languages • 1993 Local Language Policy • 2007 Thematic Curriculum BANTU
USAID/Uganda School Health and Reading Program 7 year USAID funded Working through MoES systems 12 Local Languages Large scale- 3,761 schools in 37 Districts Early Grades Reading approach Expanded through USIAD/Uganda LARA GPE UTSEP Build Africa ILeap
USAID/Uganda School Health and Reading Program
Systems and policies are strengthened
P1 to P4 pupil books and teacher guides
Teachers trained and supported
Advocacy for local language literacy instruction
Rigorous assessment of foundational reading skills
Learners read more fluently and understand more of what they are reading in both local language and English
5
USAID/Uganda School Health and Reading Program Cluster 1
Cluster 2
Cluster 3
Ateso
Leb Acoli
Ngakarimojong
Leblango
Lugbarati
Lhukonzo
Luganda
Lumasaaba
Lugwere
Runyankore- Rukiga
Runyooro-Rutooro
Lusoga
Language Complexity and Learning to Read Differences in reading outcomes in European languages is associated with variations in syllabic complexity and orthographic depth (Seymour et al. 2003) Learners’ ability to recall words is related to word length (Blake et al., 1994) Agglutinating nature of some Bantu languages means words are longer, and oral reading fluency may be lower (Abadzi, 2012)
Language Complexity in Uganda Language
Language Family
Average Word Length
Tonal Markings?
# of non-Latin letters
Comparative rank of language complexity
Acoli
Nilotic
3.6
Mark tone
6
High
Ateso
Nilotic
5.5
Mark tone
7
High
Leblango
Nilotic
3.8
Mark tone
18
High
Lhukonzo
Bantu
8.3
16
High
Luganda
Bantu
6.4
2
Low
Lugbarati
Sudanic
4.2
21
High
Lugwere
Bantu
7.1
6
Low
Lumasaaba
Bantu
6.0
10
Med
Lusoga
Bantu
7.0
11
Med
Ngakarimojong
Nilotic
4.8
6
High
Runyankore-Rukiga
Bantu
6.9
0
Low
Runyoro-Rutooro
Bantu
6.3
8
Med
Mark tone
Early Grades Reading Assessment Subtask & Measure
ABC
What is measured
Letter Sound Fluency
Correct letter sounds per minute (clspm)
Ability to identify sounds of letters
Oral Reading Fluency
Correct words per minute
Ability to read connected text
Percentage correct
Ability to comprehend reading passages
Reading comprehension Reading 20+ correct words per minute
Dichotomous: Child read 20+ cwpm or not
Early Grades Reading Assessment
Schools sampled at Coordinating Centre level 14 treatment/ 14 control 30 P1-P4 pupils randomly sampled (alpha cohort) Boys and Girls equally represented 1 on 1 oral assessment Approx. 15 minutes per language
11
SHRP Impact Average Local Language Words Read Per Minute (wpm) Beginning of P1 to end of P4 25 20
Luganda Program
23
LugandaControl
22 20
Leblango Program 15
15 13
Leblango Control Ateso Program
10
Ateso Control Run-Rukiga Program
5
0
6
Run-Rukiga Control Beginning P1 Feb2013
End P1 Oct2013
End P2 Oct2014
End P3 Oct2105
End P4 Oct2016
SHRP Impact 0.95
0.93
Overall Average Effect Size, by Language
0.82 0.72
Cluster 1
0.59
Cluster 2
0.45
Cluster 3
0.39 0.3
0.28
0.21
0.13
0.09
Language Characteristics and Impact of SHRP Language Issues
Effect size R2
Language Family
Rank of language category
Average word length
# non-Latin letters
0.19
0.40*
0.0
0.11
Implementation Issues
Effect size R2
Socioeconomic Issues
Years of implementation
% schools receiving CCT visits
% teachers trained
% pupils wearing shoes
% attended pre-school
Average class enrolment
0.04
0.01
0.11
0.12
0.39*
0.06
Language Characteristics and Impact of SHRP
Effect size R2
Language Issues
Implementation Issues
Socioeconomic Issues
0.71
0.23
0.41
Implementation
“Calvin and Hobbes” by Bill Watterson
Language is Politically Sensitive
First Languageof P1 SHRP teachers in Karamoja Region Ngakarimojong
61%
Ateso
29%
Luo
4%
Leblango
4%
Runyankore-Rukiga
4%
Languages don’t sit still
Multiple languages isn’t “cheap”
English Titles P1 (1) 2,245 copies @ 0.54 USD
1,210 USD
Local Language Titles P1 (8) 278 avg copies per title @ 2.75 USD avg
5,686 USD
19
“Good schools teach
in English”
20
Implications & Conclusion Language complexity is a significant predictor of SHRP impacts • More research to guide program adaptation • Discussions around national assessments and benchmarks
Reading programs in heterogeneous language settings face complex challenges- some strategies for confronting these challenges: • • • •
Staff on the ground and at every school Work at local as well as national levels Work through the system, so the system owns the processes Continuous dialogue
