March 2, 2016

Using cognitive science to help children read

Written by Radhika Iyengar, Earth Institute, Columbia University and Helen Abadzi, University of Arlington


Children gather at the village learning center just as the sun dives behind the hills of Mwandama, in southern Malawi. It is winter in the southern hemisphere and the temperature is mellow. About 40 children dressed in rags and looking small for their age nimbly sit on the ground, in front of a small dilapidated house. This is their classroom. This was the only space available to hold the class after school hours. They range from first to fifth graders, some of who are as old as 15. Though most attend school, they are illiterate in Chichewa and English.

The village learning center was sponsored by the Millennium Villages Project, a multi-sectoral initiative led by Columbia University that operates in the poorest areas of 10 African countries. The center is run by volunteers, who mostly have not gone past primary school. The reading activities appear “traditional,” but are in fact quite innovative. Cognitive science has been used to optimize students’ perception and memory so that they can identify letters and words rapidly and with only a modicum of feedback.


Schools in Malawi simultaneously teach reading in English and Chichewa in grade 1. English is taught irregularly, many letters vary, as a result children must learn letter names. From there they are expected to progress to reading entire words. Unlike English, Chichewa is spelled consistently, and letters always have the same sounds. We pull out a sheet with the Chichewa word baba written in big bold letters. What is this word? Most are silent, but one third grader gives the English letter names: bi- ay- bi- ay. And he is puzzled. He cannot make sense of the word.


If the child had known to pronounce the letter sounds b – a – b – a, he would have succeeded. With practice and some guidance in blending letters, the word would have made sense. But these steps are missing, and there is no one at home to help bridge the gap. Parents are unschooled and cannot help with homework. And teachers are of little help. They have classes of 70-120 students and can only afford to interact with the few students who can manage.


The Early Grade Reading Test sponsored by USAID shows that 73% of second graders across the country could not read a simple story in Chichewa. In Mwandama baseline data conducted in July 2014 show that students, particularly the younger ones hardly knew letter sounds. On average, first graders could only identify three letters correctly in one minute. (Midline data are to be collected in October 2014).


The invisible culprit is English spelling. English grammar is easy, but many words cannot be decoded or sounded out. Children must memorize word lists, learn complex rules and predict unknown spellings. So, much emphasis goes to teaching advanced forms of comprehension. Reading specialists are overwhelmingly from English speaking countries, so they advocate complex methods for languages like Chichewa.


We hope to challenge these assumptions with science. The vast majority of the world’s languages are like Chichewa; all but about nine spell words pretty consistently. Letters have only one sound. People learn most easily when units are small, shapes are simple and clearly paired to a stimulus. We all combine small units with practice into larger units and best do so when we see analogies. In reading these could be ba, be, bi. Practice makes us speed up; our reaction time drops very quickly initially, as learning curves research shows. Thus students could decode b-a-b-a and with practice do this faster and faster. Then working memory, which is what we use to think right now, would understand the written message. And with more practice, a part of the brain that automatizes reading would become activated, and reading would become effortless and enjoyable.


Teaching all combinations in English takes about three years, but learning the essential letters in most languages may only require 100 days of instruction. In eastern Europe this has been traditionally understood. In Cyprus, Greece, Albania, Serbia, children are expected to decode in the first half of grade 1, by Christmas. So why not children in Malawi?


Additional problems beset the very poor who have never seen books. Initially they cannot tell that a small-sized m is the same as a larger m. To do so, they need specific practice with very large, spaced letters that satisfy the brain’s visual system. Schoolbooks and blackboards have small and dense letters, and the very poor fail at this very low-level task. And they cannot go on.


To deal with these problems, specialized textbooks are needed with perhaps 4,000 words of text. Malawian grade 1 readers have only about 400 words. So we have prepared a “fat” textbook (divided in modules for easy transport), whose initial font sizes are around 100. Village workers are basically asked to do two things: Present one letter briefly drawn on a sheet of paper and just give its sound, show combinations, and then spend most of the time asking children to practice. They go from child to child and for a few seconds listen and correct.


What follows is almost a miracle. The children get their modules and start decoding letter by letter. They have not done it before, so some get distracted. Kids who know the reading basics are helping some. We approach a 12-year old who never learned and try a combination of two letters: a and m. The sound is mmmmmm. What is the next one? 7 seconds pass. aa. And the next? mmmmm. This time 5 seconds pass before the m is identified. Then 3 seconds. Then 1 second. And the 1 second speed is maintained. If this child keeps it up and learns a new letter every other day, he will have learned to read with reasonable speed in 4 months.  But the first 1-2 weeks are crucial. If the kids are absent or the teacher does not get them to practice, the unknown letters pile up, and the kids cannot learn them simultaneously or identify them fast enough.


And who will teach students as science suggests? Village education workers have at best a high school education, and have been trained for only 3 days for this role. Can they get students to practice, encourage them, and provide brief feedback for all? Can they attend to the back rows of silent students as well as the more knowledgeable kids? Initially few do; as research predicts, they teach the way their own teachers did. More training is scheduled. Videoclips are prepared to show these workers what they should do. Compared to lengthy training programs of governments and donors, this is quite efficient. It is important to keep training simple and not expect teachers to remember 45 optional activities or overburden them with heavy teacher guides (which no one reads!). These are some key ingredients to scaling-up this approach.


Meetings ensue with some officials, where videoclips and science are presented in simple paradigms. Officials have many questions, but they seem to understand. This method seems simple enough. The findings suggest that poor children remain illiterate due to issues that are really trivial, low-level: knowing no letter sounds or being unable to identify the shape of letters that have different sizes. These trivia doom children to a lifetime of illiteracy.


Yet, it is clear how to make all students learn how to read, not just the brightest. Therefore, when we say Education For ALL, we need to literally mean for ALL. In multilingual societies, kids ought to learn letter sounds using local languages that are spelled consistently. To avoid confusion, they should learn just oral English (or French or Portuguese) in grade 1 and read it only in later grades, after becoming fluent readers in a local language.


How much does such a program cost? Millions of dollars are not required. The Millennium Villages Project provided around US$50 monthly for the community volunteers. Some were given inexpensive bicycles to travel to their village clusters. The volunteers also have community based meetings to enroll children, conduct enrollment campaigns, meet with village elders etc. The supplemental book (only with small letters) costs around US$2.50. We have had three trainings from July 2014 to October 2014 on a shoe-string budget. Supervision from project staff is done by shared vehicles and with the help of the Government Primary Education Advisers. With a few thousand dollars the project has reached out to roughly 1,000 children in 27 classes. The cost-effectiveness of this approach is evident.


The Millennium Villages Project is at the forefront of the post-2015 sustainable development goals. Goal number 4 of is to “ensure inclusive and equitable quality education and promote life-long learning opportunities for all.” Now that access to school targets are largely on track, student learning must also be kept on track. Often the culprit will be trivial issues, that once identified could be clearly overcome. But that requires science – cognitive science. This should be the driving force behind the post-2015 targets.


International and government agencies need guidance from a swat team of cognitive scientists who understand poverty and development. They are few, but more can be trained. The principles can be disseminated along with explanations that lay people can understand. Additional field research can fill knowledge gaps. Then the vast majority of the students can learn the reading basics in about 100 instructional days, around the first four months of grade 1. Billions will be saved in budgets, and the Education for All will have been fulfilled.