Listening with Eyes, Reading with Fingers: Redesigning Foundational Numeracy Assessment to Include the Children Most Left Behind

What does it take to ensure that a foundational learning assessment does not leave children with disabilities behind? The image of a visually impaired enumerator administering ICAN to a child with visual impairment at Kangundo DEB Primary School in Machakos, Kenya, offers a glimpse of the answer. It requires reimagining how each question reaches the child: a printed chart showing 13 apples, 9 bananas and 5 oranges becomes a collection of real objects; a spatial task answered visually in seconds requires a learner to trace the boundaries of a tactile card before comparing distances; and Braille-literate children may choose to read embossed instructions independently rather than have them read aloud. These moments capture the central challenge of inclusive assessment: changing how children encounter and respond to a task without changing the numeracy skills it is designed to measure.

The Global Mandate and the Assessment Gap

Across the Global South, the international education community relies on foundational learning assessments to track SDG 4.1.1(a); measuring whether children can read a simple text and solve basic arithmetic. Global frameworks designed by the UNESCO Institute for Statistics (UIS) and implemented through standard tools like the UNICEF’S Multiple Indicator Cluster Surveys (MICS-FLM) and citizen-led (CLA) frameworks such as the People’s Action for Learning PAL Network‘s International Common Assessments of Numeracy (ICAN) and Reading (ICAR) form the bedrock of global education data.

However, standard diagnostic tools, administered orally, one-on-one, present systemic barriers for children with disabilities. When an assessment protocol assumes sight and sound, children with visual or hearing impairments are frequently omitted from household surveys. Bridging this gap requires moving beyond minor accommodations toward fundamentally redesigning tasks, stimulus materials, and delivery mechanisms to fit diverse learning realities.

Our Guiding Principle: Universal Design for Learning (UDL)

Throughout this work, PAL Network and its partners use Universal Design for Learning (UDL). UDL means planning for every kind of learner from the start, rather than adding fixes later. In an assessment, this means that we can change how a question reaches a child and how the child answers, but never what it measures. A child reading a chart using Braille takes the same numeracy assessment as a child reading it with their eyes. Sightsavers, our global technical partner, advises on applying these principles throughout the adaptation.

Partnerships for Inclusive Measurement: Global Expertise, Local Roots

Making an assessment inclusive requires disability expertise, assessment expertise, contextual adaptation and government ownership. PAL Network’s work brings these capabilities together through a partnership model that connects global technical guidance with local implementation and national education priorities. The partners below are those we work with directly on this adaptation:

• Sightsavers (Global Technical Partner): Sightsavers is an international non-profit that works to prevent avoidable blindness and to promote the rights of people with disabilities, including access to inclusive education. Since 2018, it has worked across geographies, including ITA in Pakistan to adapt learning assessments for children with visual and hearing impairments, including an earlier version of ICAN. Its ICAN guidance notes are the foundation of the current adaptation. In this round of adaptation, Sightsavers reviews the adapted tools, training materials, field protocols and safeguarding provisions, and will contribute to the interpretation of the pilot findings from a disability-inclusion perspective.
• Idara-e-Taleem-o-Aagahi (ITA), Pakistan: A PAL Network member and long-standing partner of Sightsavers, ITA implemented ICAN at national scale in Pakistan in 2025 and leads the adaptation and testing in Pakistan, working closely with special education institutions and respective government departments.
• The Action Foundation (TAF), Kenya: TAF is the implementing partner in Kenya. Drawing on its experience with children and youth with disabilities, TAF adapts the ICAN tool to the Kenyan context (including Kenyan Sign Language), mobilises children and schools, and trains and supervises the enumerators who carried out the pre-pilot.
• Government Partners: Ministries of Education and special education departments in Kenya and Pakistan open the door to schools and help ensure that the adapted tools fit national priorities for inclusive education.

Together,  the partners connect technical standards with the realities of children, schools and education systems. Across all our assessments, PAL Network also uses the Washington Group questions on child functioning so that foundational learning results can be broken down by disability type.

Field Realities & Methodological Nuances: The ICAN-SED Adaptation (Pre-Pilot)

The process of adapting the International Common Assessment of Numeracy (ICAN) for children with Sensory and Educational Disabilities (SED) illustrates the deep operational, linguistic, and psychometric nuances required to improving accessibility while preserving validity.

In standard household environments, administering the 36-item ICAN tool takes roughly 8 to 12 minutes per child. Under the ICAN-SED adaptation, administration times expand significantly, up to 45 to 60 minutes, due to the tactile exploration of physical manipulatives, Braille reading, and total communication sign approaches.

1. Visual Impairments: From Flat Paper to Tactile Reality

Field insights from a pre-pilot at Kangundo DEB Primary School in Machakos, Kenya, led by TAF with technical support from Sightsavers and building on Sightsavers’ ICAN guidance for visual impairment (and comparative implementations in Pakistan with ITA) highlight the physical transformation required for visual impairment protocols:

• Real-World Manipulatives (Realia): In a standard paper ICAN item evaluating data management (e.g., item N5/N6), a child reads a printed chart displaying 13 apples, 9 bananas, and 5 oranges. For a child with total vision loss, the flat page is replaced with actual physical items —13 real apples, 9 bananas, and 5 oranges—or tactile representations produced using PIAF (Stereo Copying) and concrete counting materials like bottle tops, beads, and coins. PIAF (“Pictures in a Flash”) heats charts printed on special “swell” paper. Only the black lines and shapes rise, so the flat chart becomes a raised picture children can trace with their fingertips. The copy keeps the same layout as the printed ICAN booklet. These adaptations preserve the original numeracy task while allowing the child to access the information through touch.

• Navigating Spatial Boundaries: Sighted children process spatial tasks (e.g., item N2: identifying which object is furthest from a tree) visually in seconds. A child with visual impairment must explore a custom tactile setup, a card featuring a central straw (tree) and glued round shapes at varying distances, mapping the outer boundaries first before locating relative positions. This tactile card follows the design set out in Sightsavers’ ICAN guidance for visual impairment. The task remains the same, but the learner needs additional time to construct the spatial relationship through touch.

• Autonomous Reading vs. Oral Prompting: Sighted enumerators read instructions aloud while the child looks at the visual sheet. However, in inclusive testing, Braille-literate learners often insist on reading the embossed instructions themselves to process the context fully. Evaluators must support this self-directed reading using hand-under-hand guidance rather than directing the child’s hands or reading the stimulus for them.

2. Hearing Impairments: The Four Administration Modes & “Sign Leakage”

Adapting assessments for deaf or hard-of-hearing children requires balancing construct validity with communication access. Building on Sightsavers’ ICAN guidance for hearing impairment, ICAN-SED establishes four distinct administration methods: Method 1 (Independent), Method 2 (Read Aloud), Method 3 (Signed), and Method 4 (Total Communication).

• The “Sign Leakage” Dilemma: A major challenge during signed administration (Method 3) is preventing accommodations intended to provide access to unintentionally compromise the validity of the assessment. For example: tracing a three-sided shape in the air when asking “Which shape is a triangle?” (item N12) inadvertently provides the answer. To solve this, pre-printed Visual Label Cards (e.g., cards displaying terms like “smallest”, “inside the box”, “straight line”) are held up so the learner reads the mathematical term without the enumerator gesturing the answer.
• Exact Signed English vs. Natural Sign Language (KSL/PSL): Translating questions word-for-word using Signed Exact English (SEE) confuses children who use natural sign languages like Kenyan Sign Language (KSL) or Pakistani Sign Language (PSL). The field pre-pilot revealed that assessors must be trained to deliver instructions using natural KSL/PSL syntax while keeping the underlying mathematical construct intact.
• Language Exposure Gaps: Many children in low-resource settings have not received formal training in standardized sign language or Braille. When children communicate using informal or home signs, evaluators must involve a Learner Support Assistant (LSA) or sibling to bridge communication, while carefully documenting these barriers.

3. Capturing Nuance digitally: SurveyCTO Integration

To ensure the adaptation does not read as a mere activity log, the digital architecture built into SurveyCTO (the digital platform used for data collection) explicitly separates Construct Competency from Access Barriers. Capturing this distinction prevents an access failure from being recorded as a learning failure. This is part of the contextual information we collect for all our pilots:

• Disentangling Item Difficulty from Access Difficulty: The digital platform requires enumerators to log whether an incorrect answer was due to a child’s mathematical ability or an access barrier (such as an unfamiliar sign, an unclear tactile sheet, or an uncalibrated hearing aid).
• Tracking Planned vs. Actual Accommodations: Section C of the contextual questionnaire captures the discrepancy between planned accommodations (e.g., large print or sign interpreter) and what was actually available on the day of testing. This particular section helps with preparation for the larger pilot later in January 2027 and corrects some of the assumptions we might have made during the pre-pilot preparations.
• Stop and Pause Rules: To prevent fatigue, adaptive stop rules are enforced: if a learner cannot answer easy arithmetic operations (N27–N30), the digital platform automatically skips higher-level operations (N31–N36). The children were also allowed pauses/breaks during the assessment for various reasons. For example: some young children who are living with visual impairment have very short attention span; combined with adapted material length and in some cases the children want to read (via braille material) the instructions themselves; they get tired and consequently require breaks. The pauses help them process the information that’d otherwise become overwhelming, and we allowed that as part of the adaptation process.

Key Takeaways for Scaling Inclusive Assessment

1. The assessment measures the same numeracy skills while only the format changes: Based on the UDL principles, we change how questions are presented (raised charts, real fruit, printed word cards) but never make the underlying mathematics easier or harder. This means results for children with disabilities can be compared fairly with results for all other children.
2. Standardized protocols prevent “Prompt Leakage”: Written guidelines and printed visual label cards help enumerators prevent prompt leakage in the form of signing, gesturing or hinting at the answer while asking the question.
3. Digital documentation of how each child was assessed is essential.: We record metadata on the communication mode used with each child, assistive device availability (glasses, hearing aids, Braille materials) and the testing conditions. This makes sure the learning data reflects what the child can actually do, rather than reflecting failures in the system around them.

By refining these tools, starting with foundational numeracy (ICAN) and expanding to reading (ICAR), alliances across the Global South prove that inclusive assessment is not only possible, but essential for equitable education. Lessons from the pre-pilot will inform a larger pilot in early 2027, including further testing of validity, administration protocols and feasibility at scale. The findings will then guide preparations to incorporate the adapted materials into large-scale implementation in 2028.