Listening alone achieves almost nothing

With a difficult audiobook, the reflex is often to slow it down. Roger Griffiths had intermediate learners listen to texts at 200, 150 and 100 words per minute. At the fast speed, comprehension collapsed. At the slow speed, however, they understood no more than at the normal one, and they misjudged the changes of speed. It was different in Yong Zhao’s study in 1997, where the learners were allowed to set the speed themselves. They understood more, and they did so by slowing it down. I read the two results together as saying that an audiobook slowed down all the way through achieves nothing, but that it helps to be able to step in yourself when you can no longer keep up.

An audiobook only helps if you understand most of it anyway. For reading, the rule of thumb since Marcella Hu and Paul Nation in 2000 has been that 98 per cent of the words should be known. For listening, the threshold is lower. In 2013, Hilde van Zeeland and Norbert Schmitt played spoken stories with different proportions of unknown words to listeners. With 90 per cent of the words known, most of them already understood enough, and at 95 per cent almost everyone’s results were close together. So one unknown word in twenty is the goal. If every sentence has three unfamiliar words in it, that is no longer listening but dictionary work. Then it is better to take an easier book, or you can learn the unknown words beforehand. According to a meta-analysis by Satsuki Kurokawa from 2025, by the way, material written for learners has about twice as large an effect on vocabulary as material for native speakers.

In the study by Brown, Waring and Donkaewbua, the words that occurred fifteen to twenty times were learnt far more often than those that occurred two or three times. In 2003, Rob Waring and Misako Takaki found that words that appeared fewer than eight times in the book were gone after three months. So the audiobook hands you the common words of a language for free. The rarer ones, and they are most of the words you meet in a novel, need a second channel. That is why WieGesagt lets you save a word and learn it later through repetition. What has to happen to the word after that for it to stay is another subject, which I explain in the post “Looking up is not learning”.

In 2010, Larry Vandergrift and Marzieh Tafaghodtari taught 106 learners of French a simple sequence, in which you consider before listening what the text will be about, then listen, then check against the text, and then listen once more. With the same listening time, these learners understood more than the comparison group, and the weaker listeners gained the most. Applied to an audiobook, that means you read the beginning of the chapter, guess what comes next, listen to the chapter, click back to the sentences you missed, and listen to it a second time.

It takes more effort than having the radio on in the background. But, as mentioned, nothing is left of the radio in the background after three months.