Learn vocabulary from reading
without losing the words.

Where the words go

Think of the last time you looked up a word in a book. You tapped it, read the definition, nodded, and carried on. Three pages later you could not have said what it meant. Most lookups go like that, because nothing about the lookup asks your memory to do anything, and nothing is left to ask it later.

The words you meet while reading are the right ones to learn. They are in material you chose, in the sense the author used. What fails is the step after: the word was never written down, or it was written down alone on a list that nobody opens again.

A method that holds up

Pick material where you already know most of the words. Refold, a language-learning community, describes its sentence-mining routine in its phase 2B page as pausing when you meet a sentence with one or two unknown words, looking them up, checking you understand the sentence, and saving it. That is a sensible ceiling. If every line needs three lookups, the text is too hard and you will end up reading the dictionary.

Then four habits. Look the word up in the sense that sentence is using, not in a list of every meaning. Keep the sentence with the word. Review a few saved words each day instead of a pile each month. And keep the number of new words low. Ten a day at most is a reasonable ceiling, because every new word becomes a review you owe later, and a backlog is the quickest way to quit.

Sentence cards or word cards

People argue about this, and both sides have a point. A word card shows the word and asks for the meaning. A sentence card shows the word inside the line you found it in. The blogger behind Deciding on a card type writes that vocab cards are much faster to review, and that reading a sentence can take a long time, which means more time in the flashcard app and less time reading. The same post says many words do not work well as an isolated card, particularly ones whose definition is ambiguous, and that the context around a word can end up easier to remember than the word itself.

That last point is the main risk with sentence cards: you recognise the sentence and not the word. The author's answer was a hybrid, showing the word first and keeping the sentence as a fallback. Refold's page does not split the two types in the part reviewed for this page, and counts mined sentences as ordinary flashcards.

A fill-in-the-blank card sits between the two. The sentence is there for the sense, and the gap forces you to produce the word instead of recognising the line.

What FSRS is, in two sentences

FSRS, the Free Spaced Repetition Scheduler, is an alternative to Anki's older SM-2 scheduling, and the Anki manual says it works by more accurately working out how much you are likely to forget. You choose a desired retention, which defaults to 90% in Anki and which the manual calls a good balance of retention and workload.

For you as a reader that means a card is shown when you are about to forget it, not on a fixed schedule. Anki lets you switch to it in deck options. If you are searching for an FSRS app that does not make you think about settings, the next section is one.

Making the lookup leave a trace

This is the gap LeafyApp was built for. Press ⌥A anywhere on a Mac and drag a box over the word, in a PDF, a game, a paper, a subtitle or an app that will not let you select text. macOS text recognition reads that patch of screen, and an AI explains the word as that sentence uses it. The word is saved automatically with its sentence, source and date. There is nothing to type.

The review is built in. Cards are fill-in-the-blank, made from the saved sentence, scheduled with FSRS-5, and a few saved words come around each day. It works as a vocabulary notebook that reviews itself, and you do not have to install anything else. Folders let you keep one book apart from another.

If you already live in Anki

Keep it. LeafyApp can send words to Anki through the free AnkiConnect add-on (Anki 23.10 or later), which it can install for you in Settings, Anki, Set up. Give a folder a deck, add words one at a time or switch on Send every word, and pick the card layout with a preview. The library also exports to CSV, Excel, JSON and PDF. More on the Anki side is in the Anki card maker page and, for sentence collecting in general, sentence mining on a Mac.

It does not run on iPhone or Windows, so if you review on your phone, Anki is still the way to do it. LeafyApp is a Mac app (Apple Silicon, macOS 14 and up). The picture is read on your Mac and thrown away. Only the word and its sentence go out so the AI can explain it, with nothing attached that says who you are, and the library stays on your Mac. See the privacy page.

Questions people ask

Can you really learn vocabulary from reading?
Yes, if the words are kept and reviewed. Reading gives you words in context, but a lookup that is not saved is forgotten quickly, so the useful habit is to keep the word with its sentence and see it again a few days later.

How do I learn words in context?
Save the whole sentence, not just the word, and review it with the word blanked out. LeafyApp saves the sentence automatically when you look a word up and builds fill-in-the-blank cards from it.

Is there a spaced repetition app for Mac that handles the cards for me?
Anki is the standard free option and runs on a Mac. LeafyApp has a built-in review that uses FSRS-5 on sentences you saved while reading, and can also send words to Anki. Both are covered above.

What is FSRS?
The Free Spaced Repetition Scheduler, an alternative to Anki's older SM-2 algorithm that tries to work out how likely you are to forget each card. Anki's manual explains it on the deck options page.

Is there a vocabulary notebook app for Mac?
LeafyApp is one: every lookup is saved with its sentence, source and date, sorted into folders, and reviewed back to you. It is Mac only and free while in beta.

Make every lookup count.

Free while LeafyApp is in beta. One download, macOS 14 and up.

Download for Mac

Also worth a look: sentence mining on a Mac, Anki cards from a PDF, or looking up a word on a Mac.