Exploring Modality and Its Impact on Comprehension in Education

Beth A. Rogowsky, Barbara M. Calhoun, Paula Tallal; C.A.N. Knoop-van Campen, D. ter Doest, Ludo Verhoeven, Eliane Segers; Robin K. Pettit, Marjorie Kinney, Lise McCoy•Sources (3):Does Modality Matter? The Effects of Reading, Listening, and Dual Modality on Comprehension The effect of audio-support on strategy, time, and performance on reading comprehension in secondary school students with dyslexia A descriptive, cross-sectional study of medical student preferences for vodcast design, format and pedagogical approach

HighlightsBalancedharry & lynda hosts

Lynda: Picture a teenager with dyslexia, headphones in, eyes tracking a screen while a voice reads the same words aloud. That image feels like obviously good design — audio filling in where decoding breaks down. But here's the question underneath it: does the way information enters your head actually change what stays there? Harry: Three papers take that question seriously, and they keep arriving at the same uncomfortable conclusion. The scores do not change. Lynda: Uncomfortable, in what way? Harry: Rogowsky, Calhoun, and Tallal randomly assigned ninety-one college-educated adults to one of three conditions: digital audiobook, e-text, or dual modality. The same material was used, a chapter from the nonfiction book Unbroken. A written comprehension test was conducted immediately after, and then again two weeks later. Mean scores were nearly identical, around thirty-one out of forty-eight across all three groups. The main effect of modality was not significant, and neither was gender. Lynda: That's the result I keep wanting to argue with. Everyone has a strong feeling about how they learn best. Harry: That feeling is real. The null result is also real. They controlled for individual comprehension aptitude as a covariate, and that covariate was remarkably significant, with an F-value of fifty-six point eighty-six. The instrument was sensitive enough to detect that stronger readers performed better. It just couldn't detect any difference between modalities. At two weeks, the same story emerged, with means of roughly twenty-eight to twenty-nine across all three conditions, and modality remained not significant. Lynda: Rogowsky and colleagues do flag the limits. They note that adult readers of narrative nonfiction tend to rely more on written tests. Harry: Right — and that's exactly where Knoop-van Campen and colleagues come in. They ran the same modality question with secondary school students, twenty-one with dyslexia and twenty-two typically developing controls, but they added something that Rogowsky didn't have: an eye tracker. Lynda: Which allows you to see the reading, not just the outcome. Harry: Precisely. The researchers computed what they called a disparity score, which is the standard deviation of weighted fixation durations across three paragraphs. High disparity indicates selective reading, where readers jump to what matters. Low disparity suggests intensive reading of the entire text, from start to finish. Audio support pushed readers toward intensive reading in open-ended tasks. Disparity dropped with audio, showing an F-value of 8.61 and a p-value of below 0.006. Reading time also increased for both groups. Lynda: But what about comprehension scores? Harry: Still no difference. That's the echo of Rogowsky in a completely different population. The tools changed, the strategies changed, and the outcome did not move. Lynda: Which is either reassuring or alarming, depending on your perspective. Harry: For a student with dyslexia, audio access is an equity issue, regardless of whether scores improve. However, the strategy finding is important. If audio is always guiding you linearly through a text, you may never develop the selective reading skills that more challenging academic tasks require. Knoop-van Campen flagged this directly. Lynda: Now, the vodcast paper feels like it's from a different planet. It discusses medical students, surveys, and design preferences. Harry: It connects more tightly than it looks. Pettit, Kinney, and McCoy surveyed three cohorts of medical students at A.T. Still University. Across all cohorts, they found that clear explanations, organization, conciseness, high-yield content for board exams, and the ability to speed up playback were the most helpful. The least helpful features included music and objects moving on screen. The preferred length was 27 to 28 minutes. Lynda: That efficiency preference, which includes speed controls and no animation, indicates that students are treating audio-visual content as a pure delivery mechanism. Harry: And Pettit and colleagues point out something that ties all three papers together. When transcripts were provided, some students skipped the vodcasts entirely. The researchers ask whether those students are disadvantaged and whether reading and watching produce the same outcome. That is Rogowsky's question, dressed in medical school clothes. Lynda: Three studies, three populations, and three methods. The comprehension scores continue to remain unchanged. Harry: But what does move is strategy, behavior, and equity. Modality shapes how you process, even when it doesn't change your score. The part we still can't measure well is whether those processing differences accumulate over years. That remains the open question none of these papers can answer. Lynda: Which means the conversation is not over. Harry: Not even close. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.

Picture a teenager with dyslexia, headphones in, eyes tracking a screen while a voice reads the same words aloud. That image feels like obviously good design — audio filling in where decoding breaks down. But here's the question underneath it: does the way information enters your head actually change what stays there?

Three papers take that question seriously, and they keep arriving at the same uncomfortable conclusion. The scores do not change.

Uncomfortable, in what way?

Rogowsky, Calhoun, and Tallal randomly assigned ninety-one college-educated adults to one of three conditions: digital audiobook, e-text, or dual modality. The same material was used, a chapter from the nonfiction book Unbroken. A written comprehension test was conducted immediately after, and then again two weeks later. Mean scores were nearly identical, around thirty-one out of forty-eight across all three groups. The main effect of modality was not significant, and neither was gender.

That's the result I keep wanting to argue with. Everyone has a strong feeling about how they learn best.

That feeling is real. The null result is also real. They controlled for individual comprehension aptitude as a covariate, and that covariate was remarkably significant, with an F-value of fifty-six point eighty-six. The instrument was sensitive enough to detect that stronger readers performed better. It just couldn't detect any difference between modalities. At two weeks, the same story emerged, with means of roughly twenty-eight to twenty-nine across all three conditions, and modality remained not significant.

Rogowsky and colleagues do flag the limits. They note that adult readers of narrative nonfiction tend to rely more on written tests.

Right — and that's exactly where Knoop-van Campen and colleagues come in. They ran the same modality question with secondary school students, twenty-one with dyslexia and twenty-two typically developing controls, but they added something that Rogowsky didn't have: an eye tracker.

Which allows you to see the reading, not just the outcome.

Precisely. The researchers computed what they called a disparity score, which is the standard deviation of weighted fixation durations across three paragraphs. High disparity indicates selective reading, where readers jump to what matters. Low disparity suggests intensive reading of the entire text, from start to finish. Audio support pushed readers toward intensive reading in open-ended tasks. Disparity dropped with audio, showing an F-value of 8.61 and a p-value of below 0.006. Reading time also increased for both groups.

But what about comprehension scores?

Still no difference. That's the echo of Rogowsky in a completely different population. The tools changed, the strategies changed, and the outcome did not move.

Which is either reassuring or alarming, depending on your perspective.

For a student with dyslexia, audio access is an equity issue, regardless of whether scores improve. However, the strategy finding is important. If audio is always guiding you linearly through a text, you may never develop the selective reading skills that more challenging academic tasks require. Knoop-van Campen flagged this directly.

Now, the vodcast paper feels like it's from a different planet. It discusses medical students, surveys, and design preferences.

It connects more tightly than it looks. Pettit, Kinney, and McCoy surveyed three cohorts of medical students at A.T. Still University. Across all cohorts, they found that clear explanations, organization, conciseness, high-yield content for board exams, and the ability to speed up playback were the most helpful. The least helpful features included music and objects moving on screen. The preferred length was 27 to 28 minutes.

That efficiency preference, which includes speed controls and no animation, indicates that students are treating audio-visual content as a pure delivery mechanism.

And Pettit and colleagues point out something that ties all three papers together. When transcripts were provided, some students skipped the vodcasts entirely. The researchers ask whether those students are disadvantaged and whether reading and watching produce the same outcome. That is Rogowsky's question, dressed in medical school clothes.

Three studies, three populations, and three methods. The comprehension scores continue to remain unchanged.

But what does move is strategy, behavior, and equity. Modality shapes how you process, even when it doesn't change your score. The part we still can't measure well is whether those processing differences accumulate over years. That remains the open question none of these papers can answer.

Which means the conversation is not over.

Not even close. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.

Related lectures