Language learning methods: what actually works, and why

The core methods, and what each one is for

These six do the heavy lifting. Each has a named body of research behind it, and each solves a different part of the problem. You do not need all six from a textbook. You need to know what each one fixes.

Comprehensible input (Krashen)

You learn a language by understanding messages slightly above your current level, what Stephen Krashen calls i+1: input that is mostly known words plus one new thing, so you can follow it without a dictionary. It suits anyone who freezes when native speakers talk, because it trains meaning before grammar rules. The catch is finding material at your exact level, which is where most learners stall. More on the comprehensible input page.

Spaced repetition (Ebbinghaus, now FSRS)

You forget new words on a predictable schedule, the forgetting curve Hermann Ebbinghaus measured in the 1880s. Spaced repetition brings a word back right before you would lose it, so each review costs less and the memory lasts longer. Modern apps run this on FSRS, the open-source scheduler inside Anki. It suits anyone drowning in vocabulary they keep re-learning. See spaced repetition for language learning.

High-frequency vocabulary first (Nation)

A small core of words does most of the work. Paul Nation's vocabulary research found the 2,000 most common word families cover roughly 80% of ordinary written text, and spoken language leans on an even smaller core. Learn those first, in frequency order, and you understand far sooner than if you memorise words at random. It suits beginners who want the fastest path to following real speech. See how many words you need to be fluent.

Sentence mining

Instead of studying word lists, you pull whole sentences you actually met, from a show, a message, a sign, and turn each into a flashcard with the new word in context. The immersion-learning community built the method around comprehensible input and spaced repetition, and it works because you remember words attached to a real moment, not a definition. It suits intermediate learners who understand a lot but cannot produce it yet. See sentence mining.

Shadowing (Arguelles)

You speak along with a native recording in real time, matching the rhythm and sounds as you go, a technique the linguist Alexander Arguelles is known for popularising. It trains pronunciation and listening at once, because your mouth has to keep up with a real speaker. It suits people who read fine but sound flat or freeze when speaking. Honest note: it feels awkward and does little for grammar. See shadowing for language learning.

Dual coding (Paivio)

A word paired with an image is easier to recall than the word alone, Allan Paivio's dual coding theory: your brain stores it twice, once as language and once as a picture. It suits concrete vocabulary you can actually see, and does little for abstract words, which is why good decks add an image only where one helps. See dual coding for language learning.

Methods for one specific problem

These are not competing theories, they are the pages for when you already know what is wrong.

• Why can't I understand native speakers? The wall is usually your ear, not the words. How listening and decoding training fixes it.
• The best AI for language learning When a chatbot tutor genuinely helps for speaking practice, and when it is a gimmick.
• Learn a language in 10 minutes a day Whether a small daily habit actually works, and how to make ten minutes count.
• How do polyglots learn languages? The habits that repeat across people who have done it several times.
• The fastest way to learn a language The honest version, with the myths removed.

How we put these together

The trap with a methods list is that reading it doesn't teach you anything. You still have to assemble the pieces: find level-appropriate input, schedule the reviews, pick the high-frequency words, add images only where they help. That assembly is the work, and it is why most people quit.

So we built the six that survive scrutiny into one habit. TangoLango runs high-frequency vocabulary first (Nation), spaced repetition on FSRS, comprehensible input at your level (Krashen), dual coding where a picture earns its place (Paivio), native audio on every card for listening training, and a tutor that starts in English and fades it out as you improve, the scaffolding Lev Vygotsky described. Ten minutes a day, one dialect at a time. If you want the method applied to a specific language, the guide to learning European Portuguese shows what it looks like in practice. And if you are here because an app let you down, why people are ditching Duolingo covers what a streak does and does not teach.

Frequently asked questions