This video details an optimized language learning method based on scientific research, personal experience, and new technologies. The core of the method involves actively extracting information from text-based resources (like textbooks, articles, or transcripts) and transferring it into a personalized Anki deck using a specific workflow. This approach aims to maximize retention and recall by incorporating multimodal learning (audio, images), spaced repetition, and continuous motivation through Anki's statistics.
Introduction and Background #
- Speaker's Experience:
- Over 30 years of language learning experience, starting at age 7.
- Tried various methods: classrooms, textbooks, TV, apps, a PhD in linguistics (not recommended), travel (led to jumping from a moving train in China, not recommended either).
- Speaks several languages "pretty well."
- Motivation for Optimization:
- Past "fun diversions" for learning didn't lead to lasting knowledge.
- No time to waste with a child.
- Previous 3-month Persian challenge failed to stick, leading to reflection.
- Applied lessons learned to Italian preparation for a trip, resulting in success.
- Methodology Basis:
- Built on "newest technology" (unavailable previously).
- Based on science: academic research in second language acquisition, learning, memory, and neurolinguistics.
- Incorporates optimization theory to minimize time on tedious ancillary tasks.
The Language Learning Pipeline Overview #
- Core Tools (Speaker's Preference): Textbooks, Google Sheets, Anki, Hypert, Google Images.
- Flexibility of Tools: The pipeline is key, not the specific tools. Alternatives include podcast transcripts, Excel, other SRS, and AI image generators.
Anki and Spaced Repetition (SRS) #
- What Anki Is: Digital flashcards "on steroids" with images, sounds, fill-in-the-blanks, and diverse formatting.
- Spaced Repetition Innovation: Asks about card difficulty and reschedules reviews based on that difficulty, optimizing long-term retention and quick recall.
- Origin of Speaker's Anki Use: Learned from Gabriel Wyner's "Fluent Forever."
Ideal Anki Card (Based on Neuroscience) #
- Word/Small Phrase Focus: More effective than memorizing full sentences for the speaker.
- Multimodal Learning:
- Consistently accompanied by audio and images.
- Audio: Target language spoken by a native speaker (ideal), or personal recordings (better than none). Saying words aloud is beneficial.
- Images:
- Thematically related images boost retention when consistently shown.
- Images that "prime" the phonology of a word can boost recall.
- Neuroscientific Basis: Multimodal input creates more neural connections.
- Card Structure:
- Front: English translation (or L1) with a thematically related, engaging image.
- Back: Target language word/phrase, best available audio, sometimes a priming image.
- Example: Hebrew card for "office" – office picture on front, "misad" audio and Teenage Mutant Ninja Turtles picture (for phonetic priming) on back.
Anki as a Motivator #
- Comparison to Other Platforms: Unlike opaque, proprietary systems (Rosetta Stone, Duolingo), Anki provides detailed, transparent statistics.
- Statistics: Shows learned items, items in learning phase, review frequency, and card difficulty.
- Benefit: Incredibly motivating, more so than Duolingo streaks, by showing progress and optimization.
Workflow: From Document to Anki Deck #
- Step 1: Document Acquisition/Preparation:
- Start with a digital document (textbook, article, podcast transcript).
- OCR often unreliable; manual typing of examples from physical textbooks is preferred if no digital version.
- Step 2: Data Entry into Spreadsheet (Google Sheets recommended):
- Copy and paste words, chunks, or sentences.
- Use Google Translate formula for automatic translation (speaker uses Google Sheets for this).
- Manually correct ~5% of translation errors, which doubles as exposure and memory building.
- Mark material in the textbook as it's processed (e.g., crossing off sections).
- Step 3: Breaking Down Sentences:
- Speaker finds full sentences less effective for memorization.
- Breaks down complex sentences into:
- Main clauses.
- Implied sentences/related phrases.
- Modifiers and smaller bits.
- For beginners: focus on verbs, nouns, adjectives.
- Example (Italian): A sentence "Gli Italiani vogliono scappare dalla periferia" (Italians want to escape from the big city) is broken into "se" (more and more), "pero" (however), the verb "scappare" (to flee/escape), and the preposition it takes.
- Step 4: Supplementing Textbook/Article Content:
- For non-textbook materials (novels, news), copy/paste or type content.
- Since these don't provide examples of conjugations/words, use resources like Wiktionary (for conjugations) and Reverso Context (for further examples).
- No need to use full sentences from these sources.
- Visual encyclopedias (DK Publishing, Wikipedia) for topic-specific vocabulary.
- News sources (Euro News) exploit Zipfian distribution for high-frequency vocabulary early on.
- Step 5: Handling Different Scripts:
- Languages with non-Latin scripts (e.g., Hebrew, Persian) are more challenging to copy/paste.
- Options: Type manually using a specialized keyboard or Google Translate's keyboard input. Or, start with English and correct translations (time-consuming but good practice).
- Step 6: Anki Deck Creation:
- Create a main deck with subdecks (e.g., per chapter). This allows learning to start before all data entry is complete.
- Import CSV files from Google Sheets into each subdeck.
- Hypert Add-on (TTS):
- Paid add-on ($5/month), "worth every cent."
- Draws audio from various sources (11 Labs, Google, Azure - current favorite, Forvo for human recordings).
- Add audio in batches to subdecks.
- Image Addition:
- Quickly copy/paste word into Google Images; pick something relevant. Don't overthink it.
- If no good image, can use AI (sometimes for phonetic priming) or go without.
- Step 7: Studying the Cards:
- Calculations: Estimate cards per chapter/day based on target timeframes (e.g., 4200 cards in 100 days = 42 new cards/day, leading to ~420 daily reviews).
- Flexibility: Adapt daily goals to what's manageable (e.g., 15 new cards/150 reviews) rather than strictly ideal.
- A full textbook chapter can be processed into Anki in about an hour.
Future Plans and Conclusion #
- Application: Using this method for French, Italian, Hebrew (moving to advanced resources), and considering Persian again in January.
- Compatibility: Works well with "closed deletion" and "sentence mining" methods; the speaker argues it's a "better" form of sentence mining.
- Observed Results: "Stark" improvements; faster learning of non-Indo-European languages, and related languages feel like "downloading from the Matrix."
- Call to Action: Encourages viewers to try the method and provide feedback.
- Patreon Support: Offers starter decks for patrons (e.g., Route Italian with audio).
last updated: