Sentence mining: the method behind every card you keep
Sentence mining is the method of building your flashcards from whole sentences you meet in real material, so each card teaches a word inside its grammar and context instead of stranded on its own. You find a sentence you almost understand, turn it into a card with the meaning and native audio, and review it with spaced repetition so it comes back right before you'd forget it.
- Cards built from real sentences, not stranded words
- Native audio on every sentence, 24 dialect-pure tracks
- FSRS scheduling, the same engine inside Anki
What sentence mining means
Sentence mining means you learn from sentences, not from word lists. Instead of memorising "comboio = train" on a bare card, you keep the whole line you actually read (o comboio das oito está atrasado) so the word arrives with its article, its verb, and the situation it belongs to. Your brain files it the way it will meet it again.
The idea got its name and its following from the Japanese-learning community in the late 2000s, when Khatzumoto's All Japanese All The Time blog pushed the "10,000 sentences" approach: mine and review enough real sentences and you absorb the patterns without translating in your head. It rests on a simple, well-supported idea from Stephen Krashen: you learn most from input one step past your current level, his i+1. A whole sentence you nearly understand is i+1 in its cleanest form.
How to do sentence mining
The method is four moves, and the order matters:
- Find an i+1 sentence. Read or listen to something a bit above your level and stop at a sentence where you know everything except one word or one piece of grammar. That gap is the whole point. A sentence with five unknowns is too hard to be a good card.
- Make the card. Put the sentence on the front, the meaning on the back, and add native audio so you train your ear, not just your eye. Most people can read a language long before they can follow someone speaking it fast.
- Review with spaced repetition. Feed the card to a scheduler that shows it again just before you'd forget. This is where the sentences turn into memory instead of a list you read once.
- Keep mining, daily. A few new sentences a day beats a hundred in one sitting. The habit is the method.
Why it works
Two things do the work. First, context: a word learned inside a sentence carries its grammar and its collocations, so you don't just know it, you know how to use it. Second, timing: spaced repetition matches review to your forgetting curve, so you spend effort only on what's about to slip. Modern schedulers like FSRS, the open-source engine inside Anki, predict that moment per card. Put comprehensible input and good scheduling together and you get the thing every method is chasing: words that stay.
The catch with the classic workflow
Done by hand, sentence mining is powerful and slow. The traditional stack is Anki plus a pop-up dictionary: you read, hit a word, look it up, copy the sentence, find audio, build the card. That's a couple of minutes per card, every card. Tools help, and they're good ones. Migaku mines from Netflix and YouTube in a click; LingQ mines from text you import; Refold maps the whole immersion routine. If your input is anime or novels, that toolchain is genuinely the better call. But all of them mine from media you sit down to consume. None mines the sentence you met by accident, out in your day.
Mining your real life, automatically
If you live where the language is spoken, your best material isn't a subtitle file. It's the parking sign, the pharmacy label, the voice note from the creche. TangoLango does automatic sentence mining from exactly that: photograph it or paste it, and the app finds the i+1 sentence, records it in a native voice for your dialect, and schedules it with FSRS. No deck to build. It's the same method the immersion crowd swears by, aimed at the language your week keeps demanding, and it runs in 24 dialect-pure tracks. If you want a tool rather than a technique, here's which sentence mining app to pick. For the input side of the idea, see comprehensible input, and for where mining sits next to the other approaches, start at the methods directory.
"Sentence mining fixed my language when nothing else did. The method is simple. Keeping it up by hand is the hard part, so I built the app that does the mining and leaves me the ten minutes of review."
Frequently asked questions
What is the sentence mining method?
It's a way of learning a language by turning whole sentences you meet in real material into flashcards, then reviewing them with spaced repetition. Each card teaches a word with its grammar and context, and the scheduler brings it back just before you'd forget. You mine a few sentences a day from what you read, watch, or run into, and the vocabulary builds itself around real usage instead of a word list.
Does sentence mining actually work?
Yes. Learning words inside sentences gives you context and grammar together, which sticks better than isolated pairs, and spaced repetition keeps what you mine. It leans on a solid principle, Krashen's idea that you learn most from input just past your level. The one requirement is consistency, so the friction of building cards by hand is the real enemy, not the method.
What does sentence mining mean?
Mining, as in digging useful sentences out of real material (a book, a show, a conversation, a sign) and keeping them as study cards. The "mine" is the sentence you almost understand; the value you extract is the one new word or structure it teaches. You then review those sentences with spaced repetition so they move into long-term memory.
Is sentence mining better than learning word lists?
For most people, yes, because a word list strands a word with no grammar and no situation, so you can recite it but freeze when you need it. A mined sentence carries the article, the verb form, and the moment it belongs to, so you recall it the way you'll use it. Word lists are fine for a quick vocabulary boost; sentences are what make the words usable.
Learn the Portuguese people actually speak here
Ten minutes a day, native audio, your own real-life sentences. Free for 7 days.
Start mining your first sentences (free for 7 days)