Listening or reading: what the research actually says about comprehension
Do we understand and remember a text we listen to as well as one we read? The question is fair, and the honest answer is neither 'it's the same' nor 'nothing beats reading'. Here is what a few studies say, and what they do not.
When you offer an audio version of an article, one objection keeps coming back: does listening really count? Do we understand and retain a text we hear as well as one we read, or is audio just a convenience that leaves fewer traces? The question deserves better than a slogan. Since our bias is quality, not a promise, here is what the research says, with its references, and above all where it stops.
A starting point: reading is already listening
A classic model of reading, the "simple view of reading" put forward by Gough and Tunmer in 1986, breaks reading comprehension into two factors: decoding (recognising the written words) and oral comprehension. In other words, once the words are recognised, understanding a text you read draws on the same resources as understanding the same text you hear. Listening comprehension is not a watered-down version of reading; it is a component of it. That starting point explains why experimental comparisons usually find less of a gap than people expect.
What direct comparisons find
A frequently cited study, by Rogowsky, Calhoun and Tallal, published in 2016 in the journal SAGE Open, had adults read, listen to, or do both at once with the same content, then measured their comprehension immediately and after a delay. The result is clear: no statistically significant difference in comprehension between the three conditions, neither right away nor later. On that text and that population, listening did neither better nor worse than reading.
A single result is not a law. That is why the meta-analysis by Virginia Clinton-Lisell, published in 2022 in the Review of Educational Research, carries more weight: it pulls together 46 studies comparing reading and listening, across nearly 4,700 participants. Its overall conclusion points the same way: the global difference between reading and listening is not significant. But, and this is where the nuance matters, it is not zero everywhere. Reading keeps a slight edge in two specific cases: when the reader sets their own pace, and for inferential comprehension, that is, drawing conclusions that are not spelled out on the page. For literal comprehension, the facts stated directly, the gap disappears.
What these studies do not say
These results have to be read for what they are, without stretching them. They most often involve texts of moderate length and audiences of students or literate adults: they do not settle long-term retention for every kind of content, nor the case of a very dense text that you reread several times. They also do not say that audio "replaces" reading. What they do say is more useful and more modest: for many texts, listening and reading lead to comparable comprehension, and the gap, where it exists, favours reading on texts where you need to go back and think.
That nuance is exactly what guides how we think about audio. A news article, a post, a letter, a story: linear texts that you follow from start to finish lend themselves well to listening, and that is where text-to-speech has the most value, a subject we lay out on our French text-to-speech page. A contract, a technical demonstration, a text you annotate: content where you stop, reread a passage, where moving at your own pace matters, is still better served in writing. Audio does not claim to take their place.
What actually helps listening
The research highlights a factor we see on the product side too: self-pacing, and therefore fatigue. Past a certain length, the ear drifts, and a long text listened to in one go loses comprehension, not because it is audio, but because attention sags. We cover this in our article on long texts and listening fatigue: splitting the text, marking pauses, offering chapters gives the listener back some of the control a reader has naturally.
And there are audiences for whom the question does not arise in the same terms. For someone with dyslexia, listening is not a convenience compared with reading: it is often the most direct route to meaning, because decoding, the first factor in the 1986 model, is precisely what costs them. We touch on this in our article on audio and dyslexia. There, listening does not compete with reading, it opens it up.
In short
The honest answer has three parts. For many texts, the studies find no significant gap between listening and reading. When a gap shows up, it leans toward reading on texts that call for going back and reasoning. And for some readers, audio is the best way in, not a fallback. That is why a text written for the ear listens better: what reads well and listens poorly, we listed in writing for the ear. Audio is not a universal substitute for reading, and it does not need to be in order to be worth it.
Give your articles a voice with WeDispatch
This blog is itself voiced by WeDispatch. Curious how it sounds on your content?
Book a demo