can you learn a language just by listening?
partly, and the part is bigger than most people expect. listening on its own can build real knowledge of a language: where its words begin and end, what its sentences sound like, a good share of what they mean. what it does not reliably build on its own is speech. the clearest evidence is the people who grew up hearing a language every day and never spoke it. many of them understand a great deal and can say very little. so the honest answer is that listening can take you a long way into understanding a language and only a short way into speaking it.
what the ear does without being asked
in 1997 a team led by jenny saffran played children and adults a stream of invented speech while they did something else. the participants were not told it was a language, or that they should listen. they were making pictures on a computer. the stream had no pauses between its made-up words. the only clue to where one word ended was how often certain syllables followed each other.
afterwards both groups had learned the language's words, and the children did as well as the adults. the paper's title says it plainly: listening, and learning, out of the corner of your ear.
hold that finding at its real size. it was a small invented language, and what people learned was where the words were, not what they meant. still, it shows something worth knowing before you start. the ear is not idle when the mind is busy.
where meaning comes in
finding the words is one job. attaching meaning to them is another, and here sound alone does less.
a 2003 study by patricia kuhl and colleagues gave american babies of nine months twelve short sessions with mandarin speakers, who read to them and played with toys. those babies kept their ability to hear a mandarin sound distinction that babies hearing only english were losing. a second group got the same mandarin from recordings, on video or audio only. for them it had no measurable effect. these were infants, and the skill was telling sounds apart, so the result cannot be carried straight over to an adult with headphones. what it does suggest is that sound arriving with a situation around it, a person, a face, something happening, does more than sound arriving alone.
adults show a gentler version of the same pattern. studies of second-language learners listening to lectures, talk shows and interviews find that they do pick up some unfamiliar words from listening. the gains are smaller than from reading, and part of what is picked up fades within weeks.
understanding without speaking
the strongest case for listening, and the clearest view of its limit, comes from people linguists call receptive bilinguals: they understand a language and speak little or none of it. a study of heritage listeners of labrador inuttitut, in canada, found good though not perfect comprehension and basic vocabulary, alongside great difficulty producing speech. their grasp of the grammar was real but partial.
the authors give a reason that is worth keeping. when you listen, context fills the gaps in what you know. the situation, the speaker's face, the rest of the sentence all help. when you speak, there is nothing to fill them. you have to choose the ending yourself, and if you never learned it, you choose wrong or you stall.
merrill swain reached a related conclusion in canadian french immersion classrooms in the 1980s. students taught in french for years scored close to their french-speaking peers on some listening and reading tests, while their speaking and writing stayed different, with errors that persisted. she argued that understanding input is not enough on its own, and that producing language makes a learner notice what they do not yet know.
what listening builds, and how to tell
put those findings together and listening has a clear job. it builds recognition: the shape of the language, its sounds, its common words and phrases, the sense of what a sentence is doing before you could translate it. that is real knowledge, even while none of it can yet be said back.
it gives you a test you can run on yourself. after a week or two of listening to the same material, ask three questions. can you hear where one word stops and the next begins? can you tell a question from a statement without the words? is there a phrase you recognise before your mind has translated it? if the answers are moving towards yes, the listening is working.
it also tells you where to stop expecting more from it. if you want to speak, at some point you have to open your mouth: repeat lines aloud, say them back to yourself, talk to someone who will answer. and if what you are hearing has no meaning attached to it, no translation, no picture, no situation you can follow, it is likely to stay sound for a long time. the research is kindest to listening that comes with something to hold on to.
where driftspeak fits
driftspeak sits on the listening side of that line, on purpose. it is subtitles for your ears: lines from albums and films, a line at a time, in their language and then in yours. it gives each line a meaning to hold on to, which is the part sound alone lacks. it is not built to get you speaking, and it does not claim to.
words i kept is the place to start. it is thirteen short tracks of lines kept from records and films on the shelf, about nineteen minutes, and free in full in the app. the french track opens on "ce soir, je sors seule" ("tonight, i'm going out alone"), first as french, then as english.
listen to a sample of words i kept and see whether "ce soir, je sors seule" is still only sound by the time the english arrives.