Skip to main content

The best way to learn a language

There’s no shortage of opinions on the best way to learn a language. Apps promise fluency in three months. Courses sell structured curricula. Polyglots swear by immersion. The research has a clearer and somewhat uncomfortable answer: the main driver of language acquisition is understanding a large amount of real language — not memorising, not drilling, not gamifying.

Here’s what the evidence actually says, and what the fastest learners have in common.

Start a conversationFree: one 3-minute conversation. 18+.

Comprehensible input is the engine

Stephen Krashen’s input hypothesis — that we acquire language primarily by understanding messages slightly above our current level — is decades old and still the most robustly supported model we have. Learners who consume high volumes of understandable authentic content consistently outperform those who focus on grammar drills and memorisation.

“Comprehensible” is the key word. Raw native content at full speed isn’t comprehensible to most learners — it needs scaffolding: a transcript, a dictionary, the ability to replay. Add that scaffolding and the authentic content becomes the engine of acquisition.

Consistency over intensity

Language acquisition is a long game. An hour every day for a year will nearly always outperform an intense month followed by nothing. Your brain needs sleep-cycle consolidation and spaced repetition to build lasting memories — neither of which happens in a binge.

Short, daily sessions with real native content — even 15–20 focused minutes — compound dramatically. The learners who make the fastest progress are the ones who make language a daily habit, not an occasional project.

What the best learners actually do

Fast language acquirers share a few habits: they spend most of their time listening and reading in authentic content, not studying grammar; they tolerate ambiguity instead of needing to understand every word; they seek out content they genuinely enjoy so they keep coming back; and they test themselves regularly to move passive recognition to active recall.

They also speak before they feel ready. Input builds what you know; output shows you what you can retrieve in time. Learners who never speak often understand a great deal and still freeze when someone asks them a question.

Putting the two halves together

For the input half, real native video with a transcript is the highest-value free hour; Sygmatic, a free sister app from the same maker, does this for Spanish, French and German. For the output half, you need conversations — with a tutor, an exchange partner, or on your own.

banter is the on-your-own option: a short spoken conversation with an AI character in an everyday scene, in Spanish, French, German, Italian or Japanese, with a Banter Score and your own lines corrected at the end. A couple of those a week, on top of daily listening, covers both halves.

Now say it out loud

Pick a scene with an AI conversation partner who stays in character. Your own lines come back corrected at the end.

Free: one 3-minute conversation. 18+.

Frequently asked questions

What’s the single most impactful thing I can do?
Listen to a lot of real native speech you can almost understand — every day. Everything else (grammar study, output, vocabulary drilling) is useful but secondary to building a large foundation of comprehensible input.
Are some languages faster to learn than others?
Yes — the Foreign Service Institute estimates 600–750 hours to professional working proficiency in languages close to English (Spanish, French, Italian), versus 2,200+ hours for languages like Japanese or Arabic. But the method stays the same: authentic input, consistent practice, gradual progression.

Or see every guide.