The language lab: a short history of a good idea that fell out of fashion

If you studied a language at a university between about 1958 and 1985, you spent hours in a room that looked like a small office with the walls taken out. Rows of booths, a headset in each, a tape deck under a smoked-glass lid, a console at the front where a technician could listen in on any booth without warning. That room was the language laboratory, and for a quarter of a century it was the most expensive thing a languages department owned.

Where it came from

The format was worked out in the United States during the Second World War. The Army Specialized Training Program needed soldiers who could speak, not read, Japanese, German, Italian, and dozens of other languages, and it needed them in months. The linguists it hired (Leonard Bloomfield among them) threw out the grammar-translation syllabus and built courses around mimicry and memorisation: a native informant said a sentence, the class said it back, and the day was spent turning a small stock of patterns over and over until the soldier could produce them without thinking.

After the war the method moved into universities under the name audio-lingual, and the tape recorder made it scalable. A department could record one native speaker once and let two hundred students drill against that voice every week. Charles Fries at Michigan and Robert Lado wrote the textbooks; Lado’s Language Teaching (1964) is still the clearest statement of what a drill is for. The Sputnik panic of 1957 put federal money behind it: the National Defense Education Act paid for thousands of labs, and by 1965 a lab was as standard in an American college as a chemistry bench.

France took it up in parallel through CREDIF at Saint-Cloud and the Voix et Images de France method, which is why the French labs (and the exercises in Parlessa, which follow their conventions) call the exercise a laboratoire and the units Modèle 1, Modèle 2.

What a session looked like

The tape did four things, in a fixed rhythm, for every item:

The stimulus. The voice asks a question or gives a cue: Tu as fait la valise ?
The pause. Silence, long enough for you to answer aloud according to the model you heard at the start of the exercise: Oui, je l’ai faite hier.
The answer. The voice gives the correct response.
The repeat pause. You say it again, this time matching the voice.

Then the next item. Eight to twelve items per exercise, six to eight exercises per session, a dialogue at the end to hear the structures in use. The worksheet in front of you showed the model and nothing else. You could not read the answers; you had to produce them.

The drills came in a few types that anyone who did the Michigan books would recognise: repetition (say what you hear), substitution (replace one slot: Je vais à la gare → chez le médecin → J’y vais), transformation (change something structural: affirmative to negative, present to passé composé, noun to pronoun), and question and answer. The transformation drill is the one that matters, because it is the one that most resembles what happens in a conversation: someone says something, and you have to reshape it under time pressure.

Why the rooms emptied

Three things killed the language lab, and none of them was that the drills did not work.

The theory changed. Chomsky’s 1959 review of Skinner’s Verbal Behavior made “habit formation” an embarrassing phrase in linguistics departments, and the audio-lingual method had been sold in exactly those terms. Communicative language teaching, from the mid-1970s, held that language is learned by using it for a purpose, and that pattern practice divorced from meaning produces parrots. There was truth in the criticism: a lab session of repeat, repeat, repeat with no idea what you were saying was a real experience for a lot of students, and the worst labs were run that way.

The rooms were terrible. Reel-to-reel decks broke. The technician at the console could interrupt you at any moment, which made the booth feel like an exam. The tapes were the same for every student regardless of level. Sessions were timetabled at 8 a.m. on a Thursday. The equipment cost more than the teachers, and when the maintenance budget went, the lab went with it.

Nobody measured. The famous Pennsylvania Project (1965–69) compared audio-lingual classes with traditional ones over two years and found no meaningful difference in outcomes, which was read as “labs don’t work” although the study’s own authors were more careful. Without a result to point to, the rooms were converted into computer clusters in the 1990s.

What was worth keeping

Strip away the room, the technician, the timetable and the theory that oversold it, and what is left is a piece of practice design that later research has been kind to:

Output practice. Swain’s work from the 1980s onward showed that comprehension alone does not produce fluent speech; learners need to be pushed to produce. The drill pushes.
Retrieval under time pressure. The pause is the point. Recalling a form from memory and saying it, with a clock running, strengthens it more than reading it, more than hearing it, more than typing it.
Immediate comparison. You hear the correct answer two seconds after your attempt, while your own version is still in your ear. That is the tightest feedback loop in language learning, and it needs no technology beyond a recording.
Spacing and recycling. A structure that appears in Lab 1 comes back in Lab 4 and Lab 9. The Michigan books did this by instinct; the spacing literature since Cepeda et al. (2006) says why it works.

The audio-lingual method’s mistake was to think the drill was the whole of language teaching. It is not. It is the practice half, the half that turns something you understand into something you can say on time. That half went missing when the labs closed, and the apps that replaced them (flashcards, gamified lessons, AI chat partners) each cover some other half.

Why we built it again

The cost structure of the lab has inverted. In 1965 the recording was cheap and the room was expensive. In 2026 the room is your headphones, and what costs money is getting the French right: a native reviewer for every line, two voices recorded properly, a syllabus that recycles structures on a schedule. That is what Parlessa spends on. The booth, the tape and the console are software, and the technician who could listen in is gone for good. Nobody is listening. That was always the part students hated, and it turns out it was never necessary.

Further reading: Robert Lado, Language Teaching: A Scientific Approach (1964); Wilga Rivers, The Psychologist and the Foreign-Language Teacher (1964); Philip D. Smith, A Comparison of the Cognitive and Audiolingual Approaches to Foreign Language Instruction: The Pennsylvania Foreign Language Project (1970); Merrill Swain, “The output hypothesis” (1985, 2005).

Try a lab free All posts