Language & Communication · Language learning & multilingualism

How adults learn a second language

What has to happen for an adult to learn a language, how the main teaching methods compare, what the US diplomats' hour estimates assume, and what app studies can and can't show.

4 short lessons · about 17 minutes · 13 sources · updated November 2026

What you will learn

  • Explain why input a learner mostly understands is the raw material of language learning
  • Describe what speaking, conversation and noticing add that listening alone doesn't
  • Compare what grammar-translation and audio-lingual drills each train and leave out
  • Explain what communicative teaching and immersion add and why accuracy can lag

Lesson 1 of 4 · 5 min

What has to happen for a new language to stick

Before you readA linguist learning Portuguese in Brazil recorded his own conversations every month. When did a new grammar form start showing up in his speech?

Correct.Not quite. The answer is After he had consciously noticed it.

Forms he'd heard for weeks only showed up in his speech after he noticed them. Next you'll see what has to happen for heard language to become language you can use.

The linguist Richard Schmidt took a Portuguese course in Brazil and chatted with Brazilians on the side. With his colleague Sylvia Frota, he recorded a conversation every month. Their records raise a simple question. What has to happen for the language you hear to become language you can use?

Start with input, the language you hear and read. The linguist Stephen Krashen argued that you learn from comprehensible input, meaning messages you understand. The best input sits one step above your level. Krashen called this i+1, where i is what you already know.

Understanding comes first because context does the work. Say you know the words for coffee and please, and you hear a new verb between them. The situation tells you what the verb means, so it can stick. Speech far beyond you gives no such foothold, so it washes past. Krashen went further and said input is the only cause of learning, and speaking is just its result.

Check yourselfAccording to Krashen's input hypothesis, which kind of listening helps a language learner most?

Correct.Not quite. The answer is Mostly familiar speech with some new parts.

Krashen's i+1 is input just above your level, so context lets you work out the new parts. Speech you can't follow gives no foothold, and speech with nothing new teaches nothing.

Other researchers agree input matters but doubt it's enough. Merrill Swain proposed that output, what you say and write, does work of its own. When you try to speak, you hit gaps between what you mean and what you can say. Each sentence also tests a guess about the grammar, and the listener's reaction shows whether the guess held.

Michael Long added interaction. When a listener looks puzzled and asks "Sorry, you went where?", you both rework the sentence until it lands. That back and forth makes input easier to understand. It also hands you feedback while the mistake is fresh.

Schmidt's recordings supplied the last piece. Forms he had heard for weeks only appeared in his speech after he noticed them, often once someone pointed them out. Schmidt concluded in 1990 that noticing is what turns input into intake. He later softened this, saying noticing helps but isn't strictly required.

Try to recallSomeone has watched TV shows in a language for years and understands them well, yet speaks with the same grammar mistakes. What do output and noticing research suggest is missing?

They've had plenty of input but little output and feedback. Speaking would expose gaps between what they mean and what they can say, and a listener's reactions would test their guesses about grammar. Without that, nothing makes them notice their mistakes, so the mistakes stay.

Input you mostly understand supplies the raw language. Speaking and conversation expose your gaps and bring feedback, and noticing a form is what lets you start using it.

Lessons in this topic

  1. What has to happen for a new language to stick

    Learn why input you mostly understand is the raw material, and what speaking, conversation and noticing add to it.

    Essentials · 5 min

  2. From translation to role-play, four ways to teach languages

    Compare grammar-translation, audio-lingual drills, communicative teaching and immersion, and see what each one trains.

    Essentials · 4 min

  3. How many hours a new language really takes

    Use the US Foreign Service Institute's estimates to see why some languages take English speakers far longer, and what those numbers assume.

    Deeper · 4 min

  4. What language apps can prove, and why learners differ

    Read app studies with care, see what a human partner's feedback adds, and learn how aptitude and motivation each shape progress.

    Deeper · 4 min

Three quick stories from this topic

The 2,200-hour languages

US diplomats are expected to need about 2,200 class hours to reach working Japanese

The Foreign Service Institute puts Arabic, Mandarin, Cantonese, Japanese and Korean in its hardest group. Dutch, French, Italian and Spanish take 600 to 750 hours, because more of English carries over to them.

Full-time hours

The hours are full-time classes, about 25 a week. The hardest group's 2,200 hours fill 88 weeks.

Why the gap

Shared words, grammar and sounds transfer from the languages you know. The further a language sits from English, the less an English speaker can reuse.

Immersion's weak spot

Years of French immersion left Canadian pupils understanding well but speaking with errors

Taught school subjects in French, immersion students reached near-native listening and reading. Yet they fell behind native speakers in grammar when they spoke and wrote, and many rarely said more than a clause.

Where it started

Canada's first French immersion program began in Quebec in 1965. Pupils learned ordinary subjects through French.

What it changed

Merrill Swain argued that learners also need to speak and get feedback, because speaking shows them the gaps that listening hides.

The correction that rarely sticks

Teachers mostly correct by recasting, which rarely gets students to fix the error

In four French immersion classrooms, teachers mostly said a student's sentence back correctly. Students seldom repaired the error themselves afterward. Asking them to try again or giving a hint led to more self-correction.

What a recast is

A recast quietly hands over the right form, so the learner can hear it as agreement and carry on.

What works better

A 2010 review of 15 classroom studies found corrections that prompt the learner had larger effects than recasts.

Sources

  • Why understanding can come before speaking

    Understanding a language and speaking it are separate skills, and understanding usually comes first. See the evidence from children and new learners, and use it to judge your own progress.

    4 lessons · 17 min

  • How readers pick up new words

    Readers pick up new words from the sentences around them, but it takes a lot of reading for the words to stick.

    4 lessons · 22 min

  • How children learn to talk

    Babies go from crying to sentences in about three years with no lessons. See the stages from babble to grammar, what children's mistakes reveal, what they need from the people around them, and how far that varies.

    4 lessons · 16 min

  • How translation works

    Why word-for-word translation fails, how a text's purpose decides what faithful means, the old argument over moving the reader or the author, and how professionals draft and check their work.

    4 lessons · 16 min

  • Telling news from opinion and ads

    Learn the test that separates a factual claim from an opinion, what the labels above a story promise, and how to spot an ad made to look like reporting.

    4 lessons · 20 min

  • Why English spelling lags behind speech

    English letters stopped matching English sounds centuries ago. See how the vowels moved, how silent letters got in, what the mismatch costs new readers and why reform keeps failing.

    4 lessons · 20 min

All language & communication topics · Take the quiz