the desk and the ear

what the desk is for

in 2008 jeffrey karpicke and henry roediger gave forty university undergraduates a list of forty swahili words, each paired with its english meaning. once someone had produced a word's meaning correctly, the researchers changed what happened to that word next. for some people it kept coming back to be read again. for others it kept coming back as a question. a week later, the extra reading had made no difference to what people could recall. the extra questions had made a large one. and when the same people were asked to predict how they would do, their predictions bore no relation to how they did.

a replication in 2015 by nicholas soderstrom and colleagues found the main result held, and also that rereading did help once the spacing between repeats was matched. what survives both papers is the part worth keeping: pulling a word out of memory does something that seeing it again does not. that is the desk's real work, and nothing else does it as well.

notice what the experiment never asked of anyone. every swahili word arrived on a screen, in writing. nobody heard one spoken, and nobody had to catch one in the middle of a sentence said at the speed people talk, so whatever the desk built, it was recall of a word they had been shown.

what the ear lays down

the clearest evidence for the other kind of time comes from people who got a great deal of it without trying. in 2002 terry kit-fong au and her colleagues at ucla screened about two hundred university students in second-year spanish classes, looking for anyone who had overheard spanish regularly as a small child: informal spanish, from native speakers, several hours a week for at least three years before the age of six. they found eleven. most of them could say some words and common phrases. they were compared with classmates who had had no regular contact with spanish until around fourteen.

on accent, the overhearers were clearly ahead. acoustic measurements of how they said certain consonants put them close to native speakers, and native speakers who listened to recordings rated their accents as reliably more native-like. on grammar, they were no better. asked to make the endings of nouns and adjectives agree, or to pick out the mistakes in spoken sentences, they did virtually the same as classmates who had never overheard any spanish at all.

hold that at its real size. these were children, the listening went on for years, and eleven is a small group. an evening with a record is not a childhood. what the paper does show is the shape of what listening leaves: the sound of a language, laid down deep enough to last into adulthood, and very little of its grammar.

side by side

put the two papers side by side and they describe partners. the overhearers' years of listening had not made them speakers. at university they were sitting in a spanish class like everyone else, being asked to make their endings agree. and that class, with all its practice, had not given their classmates the accent the overhearers carried in from childhood. each kind of time had done work the other had not.

this is why listening looks weak when it is judged by the desk's measure. ask of an hour with a record how many words you can now recall on demand, and the answer will often be very few, and the hour will look wasted. it was not built for that. ask instead whether the language sounds less like noise than it did, whether a line sounds familiar before you know what it means, whether you would put it on again, and the hour is answering a question it can answer.

giving listening an occasion

an occasion is the opposite of a gap filler. it is the difference between listening to a language because the desk feels like too much today and listening to it the way you would put on an album: on purpose, for a stretch of time that belongs to it, with something you chose because you wanted it.

in practice that can mean a time in the week that belongs to listening alone, the way a film night belongs to the film. it can mean choosing what you listen to for the pleasure of it, a record or a film you would want anyway, rather than for how useful it looks. and it means leaving the desk's questions at the desk. the cards will test you soon enough; the listening does not need to test you as well.

where each one stops

the desk will not give you the sound of a language at speed. the ear will not give you recall on demand, or the grammar to build a sentence of your own, and if you want to speak, you will need the desk, or a person who answers, or both. neither is a stage you finish before moving to the other. they run side by side for as long as you keep the language.

there is a simple way to see your own balance. for one week, put everything you do with the language into one of two columns. the first holds every moment you pulled a word back out: a card, a quiz, a sentence tried on a tutor. the second holds every moment the language came in at its own speed and you let it. if one column is full and the other nearly empty, the empty one is the next thing to add.

six songs, one evening

driftspeak belongs in the second column. it is subtitles for your ears: lines from albums and films, a line at a time, in their language and then in yours. it asks you to pull nothing back out. playlist #178 is six songs about wanting someone, in four languages, about twenty-one minutes in all, and its first three tracks have free samples.

give the empty column one evening: listen to a sample of playlist #178, whose third track opens on 오늘 밤은 내 거야 ("tonight is mine."), and leave the cards in the drawer while it plays.