How Student Feedback Shaped New Features for English Language Learners in Math

By: Olivia Martens

What if teachers had the resources to give every student the individualized support they need to succeed? For many it’s an ideal that seems out of reach.

That’s because teachers today are responsible for keeping students with diverse academic backgrounds and learning styles motivated and engaged. This task becomes monumental when considering English language learners (ELLs) who are not only acquiring new skills but also a new language.

Aiming to find a solution to help increase motivation and engagement of ELLs, Kyron Learning partnered with the AIMS Collaboratory to conduct a four-cycle research study funded by the Gates Foundation to measure motivation and engagement in the middle school math classroom during the 2024-2025 school year. Throughout the study, Kyron worked alongside 22 teachers and nearly 1,000 students from 17 schools across the United States, to explore how AI-powered instruction could support English Language Learners. For more details about the study design, research findings, and the AIMS Collaboratory, check out this blog.

While the study focused heavily on student motivation and engagement, it also provided valuable insights into how students and teachers interacted with Kyron Learning’s platform. Those findings directly influenced several feature updates introduced throughout the study timeline.

Why Iterative Research Matters in EdTech

One of the most important components of the study design was the commitment to an iterative research process. Throughout each research cycle, feedback on the platform was collected through student surveys, teacher interviews, usability reflections, and lesson engagement data. This helped identify potential learning barriers for ELL students and guided decisions about which features could best support learners’ needs and motivation to learn moving forward.

For some students, conversing in English can be difficult, while mathematical vocabulary can add an additional layer of cognitive demand. Because of this, multilingual learners benefit from additional contextual supports, comprehension scaffolds, visual reinforcement, and embedded clarification tools that help them process information at their own pace.

These realities remained at the center of decision-making throughout the study and directly influenced when and why new features were introduced. Rather than designing features based on assumptions, the study prioritized observation, iteration, and participants’ voices to guide product development. As one teacher reported, “The lessons have engaged every student regardless of their deficiency. They are excited about getting into the lessons and they are responding more and with greater confidence.”

From Feedback to Features

These feature updates were introduced across four research cycles. Cycle 1 served as the baseline implementation cycle, where students interacted with AI-powered instruction that provided conversational support and immediate feedback.

After each cycle, participant feedback and identified ELL language support helped determine which feature to implement in the subsequent cycle. Rather than flashy additions, the study focused on features that could better support learner comprehension, accessibility, pacing, and engagement.

Across the study timeline, three major features were introduced: closed captioning, highlighted transcripts, and clickable definitions. (See illustration below.)

From Cycle 1 to Cycle 2: Supporting Comprehension Through Closed Captioning

After Cycle 1, one of the clearest patterns observed involved the increased cognitive load experienced by many multilingual learners during audio-heavy instruction. Students replayed lesson segments, slowed down videos, and revisited explanations multiple times to process unfamiliar academic vocabulary and better understand mathematical explanations presented in English.

The decision to introduce closed captioning supported ELLs who benefit from spoken and written language being presented simultaneously. Moreover, it aligned with findings from Venturini et al. (2022), who noted that “captions support comprehension, vocabulary development, and literacy, while providing learners with multiple pathways to access information within a universally designed learning environment.” This product enhancement also assisted students with hearing impairments.

The feature gave learners greater control over pacing and comprehension. In fact, for many, captions did not simply improve accessibility; they changed how confidently students engaged with the lesson itself. As middle school English language learner put it (translated here from Spanish), “I liked the subtitles because they make the subject matter easier to understand.”

From Cycle 2 to Cycle 3: Making Learning Easier to Follow with Highlighted Transcripts

As the study progressed into Cycle 2, feedback revealed that many students, especially multilingual learners, still struggled with longer explanations of mathematical concepts. To address this challenge, Kyron introduced highlighted transcripts for Cycle 3.

This feature allowed students to view the full lesson transcript alongside the instructional video while simultaneously highlighting the exact portion currently being spoken. Students could also click anywhere within the transcript to jump directly to that specific point in the video, creating a more interactive and navigable lesson experience.

This feature helped create even clearer connections between spoken language and written content while also improving lesson navigation and pacing support. Rather than replaying entire video segments to relocate missed information, students could quickly revisit specific explanations as needed.

Research findings later showed that multilingual learners disproportionately used highlighted transcripts compared to non-ELL students, suggesting that these supports may have been especially beneficial for language processing and comprehension. One student participant reflected a common view with this feedback: “I liked how [the Kyron transcript] would give me a definition when I did not know the word they were using.”

From Cycle 3 to Cycle 4: Reducing Friction with Clickable Definitions

For both English Language Learners and non-ELL students alike, encountering unfamiliar academic vocabulary can create major interruptions during instruction. As students worked through increasingly complex mathematical concepts, many still found themselves pausing lessons to look up definitions or to ask for clarification outside the lesson environment, thereby losing cognitive momentum.

To help reduce this friction, Kyron introduced clickable definitions during Cycle 4. With a list generated automatically with the lesson, built directly into the highlighted transcript feature introduced in Cycle 3, students could now click underlined vocabulary words to view definitions in real time as they continue through the lesson.

This proved especially valuable for multilingual learners, as it helped maintain learning momentum while simultaneously building contextual vocabulary understanding. “What I liked about the definitions in the transcripts,” one student participant told us, “was that they helped me get a better understanding of the different things I was learning.”

Cross-Cycle Themes & Key Learnings

Over the course of the study, several consistent themes emerged:

• Meaningful instructional support does not always need to be flashy. While gamification and highly visual tools often dominate conversations in educational technology, many students simply needed stronger embedded support throughout the learning process. Features like captions, highlighted transcripts, and clickable definitions helped reduce friction by giving students access to support exactly when and where they needed it.
• Reducing cognitive load is especially important for multilingual learners. Learning new mathematical concepts while simultaneously processing unfamiliar academic language can quickly become overwhelming. Providing embedded clarification and pacing support helped students remain more engaged throughout lessons that might otherwise have felt inaccessible.
• Participant-driven iteration is an invaluable part of guiding product development. Teacher interviews and surveys, as well as student surveys, played a major role in guiding product development throughout the research process. The most effective educational technology is not built around assumptions. In our experience, it is built around observation, iteration, and considering teacher and student feedback.

Ultimately, this study was never simply about adding new features between research cycles. It was about understanding learners’ needs, especially through the lens of multilingual learners, and increasing learners’ motivation and engagement in the middle school math classroom.

The goal throughout the project remained consistent: providing timely, learner-centered, read-along supports to help students confidently persevere through new mathematical skills and concepts while also supporting the educators guiding that learning process.

Student motivation and engagement improve when support appears at the moment it is needed most. Whether through embedded questions, synchronized transcripts, or contextual vocabulary support, each read-along support update illustrates the same guiding belief: students learn best when instructional support is timely, intentional, and designed around how people actually learn.

Olivia Martens is a project manager at Kyron Learning.