{ "site": "https://lab.atenu.org", "name": "Atenu Educational Technology Lab", "shortName": "AETL", "description": "AETL is a research and innovation lab developing technology-driven learning systems for African education — AI tutoring, digital assessment, simulations, and learning analytics.", "tagline": "Researching the Future of Learning", "parentOrganization": { "name": "Atenu Foundation", "url": "https://atenu.org", "identifier": "1700418-4" }, "articles": [ { "id": "mother-tongue-vs-english-stem-ethiopia", "url": "https://lab.atenu.org/research/mother-tongue-vs-english-stem-ethiopia/", "title": "The Language of Assessment: Performance Differences Between Mother-Tongue and English-Medium Items in STEM Subjects", "description": "Research review on how the language of test items affects student performance in STEM subjects, with evidence from Ethiopia, South Africa, PISA, and controlled item experiments. Practical implications for the ESSLCE under computer-based testing.", "articleType": "research-review", "pillar": { "slug": "digital-assessment-cbt", "url": "https://lab.atenu.org/pillars/digital-assessment-cbt/" }, "topics": [ "Language of assessment", "ESSLCE", "Mother-tongue instruction", "STEM achievement", "Test validity", "Computer-based testing", "Ethiopia" ], "publishedDate": "2026-05-22T00:00:00.000Z", "updatedDate": "2026-05-22T00:00:00.000Z", "readingMinutes": 14, "citations": [ { "text": "Abedi, J. (2006). Language issues in item development. In S. M. Downing & T. M. Haladyna (Eds.), Handbook of Test Development. Lawrence Erlbaum Associates.", "year": 2006 }, { "text": "Abedi, J., & Lord, C. (2001). The language factor in mathematics tests. Applied Measurement in Education, 14(3), 219–234.", "url": "https://doi.org/10.1207/S15324818AME1403_2", "year": 2001 }, { "text": "Cummins, J. (1979). Linguistic interdependence and the educational development of bilingual children. Review of Educational Research, 49(2), 222–251.", "url": "https://doi.org/10.3102/00346543049002222", "year": 1979 }, { "text": "Cummins, J. (2000). Language, Power and Pedagogy: Bilingual Children in the Crossfire. Multilingual Matters.", "year": 2000 }, { "text": "Curran, F. C., Pacheco, M. B., Boza, L., Harris-Walls, K., Tan, T., & Deig, A. (2024). Multilingual learners and elementary science achievement: Exploring trends and heterogeneity across subgroups. AERA Open.", "year": 2024 }, { "text": "Heugh, K., Benson, C., Bogale, B., & Yohannes, M. A. G. (2007). Final report: Study on medium of instruction in primary schools in Ethiopia. Commissioned by the Ministry of Education, Ethiopia.", "year": 2007 }, { "text": "Howie, S. J. (2003). Language and other background factors affecting secondary pupils' performance in mathematics in South Africa. African Journal of Research in Mathematics, Science and Technology Education, 7(1), 1–20.", "url": "https://doi.org/10.1080/10288457.2003.10740545", "year": 2003 }, { "text": "English medium instruction in multilingual contexts: Empirical evidence from Ethiopia. (2024). International Journal of Educational Development, 105.", "year": 2024 }, { "text": "On the relationship between bilingualism and mathematical performance: A systematic review. (2024). Education Sciences, 14(11), 1172.", "url": "https://doi.org/10.3390/educsci14111172", "year": 2024 }, { "text": "Schleppegrell, M. J. (2004). The Language of Schooling: A Functional Linguistics Perspective. Lawrence Erlbaum Associates.", "year": 2004 }, { "text": "Solano-Flores, G. (2008). Who is given tests in what language by whom, when, and where? Educational Researcher, 37(4), 189–199.", "url": "https://doi.org/10.3102/0013189X08319569", "year": 2008 }, { "text": "Test language effect in international achievement comparisons: An example from PISA 2009. (2015).", "year": 2015 }, { "text": "The differing effect of language factors on science and mathematics achievement using TIMSS 2015 data: South Africa. (2018). Research in Science Education.", "year": 2018 }, { "text": "The dynamics of multilingual assessment: Exploring the impact of linguistic accommodations on science achievement. (2024). Frontiers in Communication.", "year": 2024 }, { "text": "UNICEF. (2016). Language and Learning: The Ethiopia Case Study.", "url": "https://www.unicef.org/esa/media/1681/file/UNICEF-2016-Language-and-Learning-Ethiopia.pdf", "year": 2016 } ], "keyQuestions": [ { "q": "Does the language an exam is written in actually change a student's measured STEM achievement?", "a": "Yes, by educationally meaningful margins. In Ethiopia, students assessed in their mother tongue have scored on average 42.3% across STEM versus 36% for peers assessed in English. A 2024 Ethiopian study found English-medium assessment lowered Grade 7–8 mathematics scores by about 0.2–0.25 standard deviations. The same direction of effect appears in South Africa (TIMSS), across PISA countries, and in controlled item experiments in the United States." }, { "q": "Why is the language penalty larger in science than in mathematics or reading?", "a": "Science carries the heaviest load of academic-register language: dense technical vocabulary, long noun phrases, passive constructions, and logical connectors that compress reasoning into single sentences. Conversational fluency develops in ~2 years; the academic language register needed for science exams takes 5–7 years. A student can sound fluent and still lack the register the test demands." }, { "q": "What does this mean for Ethiopia's English-medium ESSLCE, now delivered as a CBT?", "a": "For a meaningful share of students, an ESSLCE STEM score is a blend of subject knowledge and English reading proficiency, which makes it a partly invalid measure of science or mathematics ability. The penalty falls hardest on rural and lower-income students, so a language-loaded examination can widen existing gaps rather than measure across them fairly. The CBT interface adds a separate, additive cognitive load that competes for the same attention." }, { "q": "Is the answer to switch the ESSLCE out of English?", "a": "No. English is the language of Ethiopian higher education and the labour market, and competence in it is a legitimate goal of secondary schooling. The argument is narrower: examinations should measure science and mathematics as accurately as possible, and unnecessary linguistic complexity in items works against that without serving any defensible purpose." }, { "q": "What is the most practical lever for fairer STEM assessment under an English-medium exam?", "a": "Item design. Reducing the non-essential linguistic complexity of a question — short sentences, active voice, common non-technical vocabulary, fewer embedded clauses — measurably raises scores for English learners while leaving native-speaker scores unchanged. This indicates the language load was a measurement artifact, not part of the construct." }, { "q": "What is Atenu Lab doing about this in the Ethiopian context?", "a": "We are designing a within-subjects item experiment: matched STEM item triples (English, mother-tongue, linguistically reduced English) randomly assigned on the Atenu platform, analysed with differential item functioning methods to produce a direct local estimate of the language penalty in current ESSLCE-style items." } ] } ], "updatedAt": "2026-05-23T12:53:30.423Z" }