Study shows comprehension intervention, working memory predict math achievement in English learners

LAWRENCE — When a class is assigned to solve a math problem about a baker making cookies in the morning, selling them throughout the day, baking more and determining how many are left at the end of the day, the students are not just deploying math, but reading skills as well. And many students are reading in a second language. 

New research from the University of Kansas has found that using an intervention designed to boost reading comprehension can boost math achievement and that working memory can also predict students’ word problem-solving abilities.

Michael Orosco, professor of educational psychology at KU, led a study in which researchers worked with 75 third grade students in an elementary school. Of the group, 43 were Hispanic children with mathematical learning difficulties, and 32 were peers placed in a control group. The former received comprehension strategy instruction, and all were tested before and after the intervention to gauge their math abilities, as well as a profile that tested their working memory.

Results showed that both the CSI intervention and higher working memory were significant predictors of mathematical achievement.

The intervention involved teachers working with students to improve vocabulary, language comprehension, text structure, meaning and other variables in reading. Students in both groups were also tested to gauge their working memory, or how well they could hold multiple ideas while working on word problems.

“This is work that needs to be done to help give our students the best education we can,” Orosco said. “With this study, I wanted to continue working on problem-solving with the ideas of the science of reading and helping students better understand text to solve problems and also hold multiple ideas in their heads while doing so.”

With the CSI intervention, teachers modeled the aforementioned concepts of reading, starting with simple problems and increasing the difficulty as students’ skills improved. Students then practiced independently. Post tests showed that as reading comprehension skills improved, math achievement did as well.

Written with Shahlo Mamedova of KU and Naheed Abdulrahim of the University of Nebraska Kearney, the study was published in the journal Frontiers in Psychology.

Research has traditionally focused on improving math and reading for native English speakers, but Orosco said the study and previous research he has conducted in similar areas is increasingly showing it is effective with English learners as well. He has also researched cultural influences on learning.

Orosco, who also directs KU’s graduate certificate in Mind, Brain and Education, said the findings also support the idea that focusing on aspects like working memory and the science of reading can make interventions like comprehension strategy instruction more effective.

“We used working memory as a moderator in this study. Problem-solving is dependent on your working memory and being able to hold multiple concepts while working toward a solution. Students with higher working memory tend to adapt quicker and be better problem solvers,” Orosco said.

The study’s implications also show the potential of building scaffolds to assist students in areas like reading comprehension and math, given that using the same approach for every student is not effective, Orosco said. By better understanding how certain interventions help certain students, researchers and educators can build supports that tailor assistance for each student at the level they could most benefit from. 

While the study did not consider artificial intelligence, Orosco said theoretically it shows that if used properly, AI could comb through such research and help build scaffolds to tailor individualized supports based on scientific evidence.

Such an approach would not be a way to replace teachers but to assist by helping translate research into action that educators would model for students and work through with them, he said.

“AI could assist in providing individualized instruction,” Orosco said. “We have a model now for how to boost and support reading comprehension, but the technology could help develop scaffolds to help each individual student improve their ability to address problems. Human minds are made to interact with each other. Students still need to be addressed by a human teacher and learn to creatively solve problems, AI doesn’t ‘think’ or have critical thinking or our problem-solving skills.”