How to Learn a Foreign Language: What Cognitive Science Actually Recommends

People search for language-learning advice more than almost any other self-improvement topic, and the volume of apps, courses and tips can make the basic question harder to answer: what actually works? A look at the underlying research in memory science and second-language acquisition (SLA) narrows the field to a handful of methods with real evidence behind them.

Step 1: Review with spaced, active retrieval instead of re-reading

The single best-documented technique for vocabulary retention is spaced repetition: reviewing a word or phrase at increasing intervals, timed just before you are likely to forget it. A 2019 study published on the U.S. National Library of Medicine's PMC platform analyzed about 5.2 million word-pair reviews from Duolingo users and built a scheduling model (called MEMORIZE) that adjusts review timing to each learner's actual forgetting rate. The study found that learners whose review intervals matched the model's optimal schedule had lower forgetting rates than those on fixed or rule-of-thumb schedules, and that the benefit grew the longer a learner used the system.

Spacing only works well when it is combined with retrieval, not passive re-reading. In a widely cited 2008 study by cognitive psychologists Jeffrey Karpicke and Henry Roediger, published in the journal Science, students learned 40 Swahili-English word pairs under different conditions: some items were dropped from further study once recalled correctly, others were dropped from further testing. One week later, students who kept retrieving (testing themselves on) the words recalled about 80% of them, while those who had only kept re-studying the same words after their first correct recall remembered roughly a third. The researchers concluded that repeated studying after a word was already learned added almost nothing, while repeated retrieval was responsible for nearly all of the long-term gain.

In practice, this means flashcard apps that force you to produce the answer (rather than just re-read a word list) and that space reviews over days and weeks will generally outperform cramming or passive review of notes.

Step 2: Get input you can mostly understand, not input that is too hard

Linguist Stephen Krashen's Input Hypothesis, one of the most discussed frameworks in second-language acquisition, holds that people acquire a language by understanding messages that are slightly above their current level — a concept he labels "i+1." As summarized in a research explainer hosted by the Texas Gateway educational platform, Krashen states that "mere input is not enough; it must be understood," and that the most effective input is "interesting, even compelling; so interesting that you forget you are listening to or reading another language." Under this model, material that is too difficult to follow, or input a learner isn't actually paying attention to, does little for acquisition even if it is technically in the target language.

For a practical learner, this translates into choosing graded readers, slowed or subtitled audio, or TV shows and podcasts pitched just above one's comfort level, rather than diving into unmodified native-speed content before having a basic vocabulary and grammar foundation, or over-relying on material that is too easy to push growth.

Step 3: Produce the language — speaking and writing, not just listening and reading

Comprehensible input is necessary but, according to a competing and complementary framework, not sufficient. Linguist Merrill Swain developed the Output Hypothesis, which holds that producing a language — not just receiving it — also plays a distinct role in acquiring it. According to a summary on Wikipedia of Swain's work, producing language — by speaking or writing — serves three functions: a "noticing" function, where learners become aware of gaps between what they want to say and what they can say; a hypothesis-testing function, where attempting to speak lets learners test grammatical rules and get feedback; and a metalinguistic function, where producing language helps learners reflect on and internalize its structure. As the same summary describes it, the hypothesis holds that learning takes place when learners encounter a gap in their linguistic knowledge of the second language, and that noticing this gap can prompt them to modify their output and learn something new about the language.

The practical implication is that passive study — watching videos or reading without ever speaking or writing the language — leaves a predictable gap in production ability that only practice in speaking or writing appears to close.

How long it actually takes: the FSI benchmark

One of the most frequently cited real-world timelines comes from the U.S. Foreign Service Institute (FSI), which trains American diplomats and has decades of data on how long learners need to reach "Professional Working Proficiency" (level S-3/R-3 on the U.S. government's scale, roughly comparable to upper-B2/C1 on the Common European Framework). FSI groups languages into categories by the classroom hours this typically requires for English-speaking adult learners in full-time instruction, according to figures published by FSI-affiliated course providers.

These figures describe full-time, classroom-based instruction with native-speaking teachers aimed at professional proficiency — not casual self-study aimed at conversational ability, which can be reached far faster for easier languages. The categories reflect how linguistically and structurally distant a language is from English, not how "hard" it is in any absolute sense.

Putting the steps together

1. Set a specific, realistic target (e.g., conversational ability vs. professional proficiency) and expect the FSI-scale hours above as a rough reference point if your target language is distant from your own.
2. Use spaced-repetition flashcards that require you to produce the answer, not just recognize it, and review on an expanding schedule rather than cramming.
3. Choose input (reading, audio, video) that you can follow with effort but not input you must constantly look up — material pitched at "i+1," slightly above your current level.
4. Build in regular speaking or writing practice — with a tutor, exchange partner, or even self-recorded speech — so you notice and close gaps that pure input doesn't reveal.
5. Track your own forgetting: if you consistently remember a word or structure, lengthen the interval before reviewing it again; if you keep forgetting it, shorten the interval, which is the core logic behind adaptive spaced-repetition algorithms.

None of these methods is a shortcut around time on task. The common thread across the spaced-repetition research, the testing-effect studies and the FSI hour estimates is that consistent, effortful, correctly-timed practice — not any single app or trick — accounts for most of the measurable difference in how fast people learn a new language.