Sentence mining is the process of finding sentences with unknown words in your immersion content, looking up the words, and saving them as flashcards for later review. It's one of the most powerful vocabulary-building techniques in the Refold method because you're learning words in context, from content you actually enjoy.
How It Works
The basic loop:
- You're watching a TV show (or video, or reading) with target-language subtitles
- You encounter a sentence where you understand everything except 1-2 words
- You pause and look up the unknown word(s)
- You make sure you understand the whole sentence
- You save the sentence as a flashcard (usually with audio and/or an image)
- You rewatch the section to hear it at natural speed
- Continue
Later, during your vocab study time, you review these cards using spaced repetition (Anki or similar).
Why It's So Effective
Sentence mining combines several things that make vocabulary stick:
- Context: You learn words in real sentences, not isolated lists
- Personal relevance: The words come from content you chose and enjoy
- The Ikea effect: Words and cards you make yourself are easier to learn and feel more personal
- Multi-channel: Cards often include audio, images, and text from the original scene
- Spaced repetition: Regular review with increasing intervals cements the words
- Positive feedback loop: The more words you learn, the more you understand, which makes mining more enjoyable
What to Mine
Look for "low-hanging fruit" — sentences where you already understand most of the words and only need to learn one or two new ones. These are sometimes called "1T sentences" (one target).
Do mine: Sentences with 1 (or maybe 2) unknown words that seem useful or interesting.
Don't mine: Sentences with 3+ unknown words (you don't even understand the sentence), words that are extremely niche or unlikely to appear again, or words you already sort of know.
Common Mistakes
- Mining too many cards. It's easy to save 30+ cards in a session, but if you can't review them all, they pile up. Delete cards that aren't useful rather than letting them create a backlog.
- Mining words you already know. If you saved a word, then realize you know it when it comes up in your review, delete the card. Don't waste review time on known words.
- Only mining, never freeflowing. Mining is interactive immersion. You still need freeflow time to build speed and automaticity.
Tools
The specific tools depend on your platform and language, but the general setup involves: a media player with subtitle support, a popup dictionary, and a flashcard tool (usually Anki).
Many learners use browser extensions or dedicated apps that streamline the process. The setup can be frustrating at first, but once it's working, the workflow becomes smooth.
-> Click here, then scroll down to see suggested tools for mining
When to Start
Sentence mining is introduced in Phase 2B and remains useful through Phase 7. In Phase 3+, you'll also do "sentence mining while listening" — the same concept, but trying to understand through audio first before revealing subtitles.
Research and Reasoning
Sentence mining combines several evidence-based principles of vocabulary acquisition. The effectiveness of spaced repetition is well-established: Nakata (2015) compared expanding and equal spacing schedules and found that both significantly outperform massed practice.
The research distinction between deliberate and incidental vocabulary learning is also crucial. Webb (2020) showed that incidental learning through reading, listening, and viewing is a primary driver of vocabulary growth, particularly for building rich, contextual word knowledge. However, Nation (2001) demonstrates that deliberate study (like flashcards) serves an important complementary function — it increases noticing of target words when they appear in subsequent input, accelerating the acquisition process.
The context-based approach in sentence mining is supported by research on depth of processing. Hulstijn (2001) showed that learners remember words better when they process them with greater cognitive effort and contextual engagement. Sentence mining forces exactly this kind of deep processing by anchoring words in real sentences from meaningful content, rather than learning them in isolation.