Readwyn 0.1.0

Is listening to an audiobook the same as reading?

For understanding and remembering a book, mostly yes. A 2021 review of 46 studies found no reliable difference in comprehension between listening and reading, although reading did better when the reader set the pace and when the test asked for inferences rather than plain facts.

The experiments, 2010 to 2016

The early classroom and lab comparisons did not favour the ear. David Daniel and William Douglas Woody gave students either a podcast or the text of the same primary source and then quizzed them. The podcast group scored lower, and the students' first preference for podcasts changed straight after the quiz. Their paper, in Teaching of Psychology in 2010, judged podcasts better for extra material than for the main content of a course.

At the University of Waterloo in 2013, Trish Varao Sousa, Jonathan Carriere and Daniel Smilek expected a listener, with no page to follow, to drift off without noticing, so their main measure was attention. The 211 students in that study took in three excerpts of about 1,800 words from Bill Bryson's A Short History of Nearly Everything: one read silently, one read aloud, and one heard as Bryson's own recording. Up to ten times in each excerpt, at random intervals of 30 to 90 seconds, a box popped up to ask whether the student's mind had been wandering. Listeners said yes at about half of the probes, silent readers at about a third, and students reading aloud least often. Listening also gave the worst scores on the 10-question true or false test after each excerpt, and less interest in the material than reading aloud.

Then a 2016 study by Beth Rogowsky, Barbara Calhoun and Paula Tallal found no gap at all. Its 91 participants got the preface and one chapter of a non-fiction book as a digital audiobook, as e-text on a tablet, or as both at once, sat a written test straight away, and sat the same test two weeks later. Neither test showed a statistically significant difference between the three groups, and men and women recalled the same amount.

The brain and the meta-analysis

Brain scans from 2019 suggest why the results can be so close. In Jack Gallant's lab at Berkeley, people lay in an fMRI scanner while they listened to, and while they read, several hours of the same stories. The patterns of meaning across the cortex were highly correlated between listening and reading, and a model trained on one modality predicted brain responses in the other. The paper, by Fatma Deniz, Anwar Nunez-Elizalde, Alexander Huth and Gallant, calls the two representations "almost identical".

The review behind the first paragraph is Virginia Clinton-Lisell's meta-analysis, which the Review of Educational Research published online in December 2021. It pooled 46 studies with 4,687 participants of all ages. Across all of them, the gap between reading and listening was an effect size of 0.07, too small to be reliable (p = .23). The differences showed up in the subgroups. When readers could go at their own speed, reading had a small advantage (g = 0.13); when the experimenter set the pace, it did not. Going at your own speed means slowing down on a hard sentence or looking back a paragraph, which a recording played straight through doesn't allow. Text-to-speech on an e-reader lets you choose the speed and rewind, which gets part of the way there. Reading also helped more on questions of inference (g = 0.36) than on literal recall, where the two were level. There was some sign, too, that listening and reading came closer still in languages whose spelling matches their sounds closely (g = 0.001) than in languages with irregular spelling (g = 0.10). Clinton-Lisell's own conclusion is practical: accessible teaching needs both a written and an audio version of the material.

Talking books in the 1930s

Books made for listening are much older than any of these studies. In the United States, the Pratt-Smoot Act became law on 3 March 1931 and set up a national library service for blind adults at the Library of Congress. By 1933 the American Foundation for the Blind had a player ready, in two kinds: one driven by a spring and one combined with an electric radio. The Library's history quotes a description of the talking book as "the recording on a disc of the voice of a good reader", and the format was built for length. Records cut at 150 grooves to the inch and played at 33⅓ rpm held about thirty minutes of reading each, so a book of 60,000 words fitted on eight or nine double-sided twelve-inch records.