التكرار المتباعد + المحادثة: لماذا يفوزان معًا فقط

Spaced repetition is the most replicated finding in memory research, and it is still not enough to make you fluent. Flashcards train recognition: seeing a word and recalling what it means. Conversation trains retrieval: pulling the word out of your head in the half-second a real exchange gives you. You need both, and this post is the recipe for combining them without grinding 200 cards a day.

What is spaced repetition, and why does it work?

Spaced repetition means reviewing material at expanding intervals (1 day, 3 days, 7 days, 21 days…) instead of cramming. It works because forgetting is fastest immediately after learning: Ebbinghaus's original 1885 self-experiment found most new material gone within a day, and Murre & Dros (2015) replicated the curve almost exactly with modern controls. Each well-timed review interrupts the decay and flattens the curve.

The effect is large. Cepeda et al.'s 2006 meta-analysis put spaced practice at roughly Cohen's d ≈ 0.6 over massed practice for long retention. In plain English, spacers remembered half again to twice as much a month later as crammers who studied the same total minutes.

What is the optimal review schedule?

Roughly 10–20% of the time until you need the word is the recipe from Cepeda et al.'s 2008 follow-up. Need it next month? Review every 5–10 days. Need it in a year? Monthly is fine. There is no single magic interval; the schedule should stretch with your horizon. Most flashcard apps ship one default curve and never ask when you need the material.

Why isn't Anki enough on its own?

Two reasons, one structural and one human:

• Structural: SRS solves recognition, not production. Being shown "однако" and recalling "however" is a different neural task from reaching for "однако" mid-sentence while someone waits. That gap is what SLA researchers call automaticity (Segalowitz 2010). Pimsleur's 1967 graduated-interval method trained both directions; most modern card apps quietly dropped the production half.
• Human: flashcards are lonely. Words divorced from any conversation you've ever had attach to nothing, and for most people the loneliness eventually beats the discipline. The graveyard of abandoned Anki decks is a design problem, not a discipline problem.

What does conversation add that cards can't?

1. Retrieval under pressure. The half-second deadline of a real exchange is the exact skill fluency requires, and cards never simulate it.
2. Context. You learn that "kind of" softens and "rather" sharpens: usage nuance that a definition side of a card cannot carry (lexical priming, Hoey 2005).
3. Emotional tagging. Words used in moments that mattered to you are recalled better, consistent with the affect-and-memory literature (Kensinger 2009).

The cleanest direct evidence for combining the two: Nakata (2008) and Nakata & Webb (2016) found spaced retrieval plus active production beats spaced retrieval alone for long-term retention.

How do the common approaches compare?

Approach	Trains	Strength	Weakness
Anki / classic SRS	Recognition	Free, precise scheduling	No production; high dropout
Gamified vocab apps	Recognition	Habit-forming streaks	Shallow context, topic-ordered
Conversation only	Production	Real retrieval, context, emotion	New words resurface by chance, not schedule
SRS + daily conversation	Both	Each covers the other's gap	Requires two habits or one tool that does both

How Dia closes the loop

Every word you use with Dia lands in your personal word bank, ranked by how actively you engage with it. The most active words get threaded back into your next conversation, so vocabulary keeps getting retested in dialogue, where production happens, instead of in a card queue. Smart flashcards are generated from your own conversations rather than a generic frequency deck, which means every card carries a memory of the moment you first used the word. (Calendar-style SRS scheduling à la Anki is on the roadmap; today the engine prioritizes by usage and recency.)

If you're past the beginner stage and wondering why your large passive vocabulary still doesn't come out of your mouth, that's the intermediate plateau, and the fix is the same: keep the cards, add the conversation.