A month of Word Drift: my vocabulary actually grew

Two months ago I wrote about doubting that memory-match games actually improve cognition. The conclusion was grim. So I was nervous starting this self-test — what if word games are also "just makes you better at word games"? Results were surprisingly good.

The setup

Disclosure: My English reading is solid for daily and intermediate technical text, but speaking and advanced vocabulary are weaker. Concretely, I recognize ~80% of the Oxford 3000 list, but only 30–40% in the 5000–10000 range (intermediate to upper-intermediate vocabulary). That's the baseline I wanted to improve over 30 days of Word Drift.

Method:

• Day 1 and Day 30, take Memrise's "English 5000 word test," count words where I instantly recognize meaning
• Play Word Drift 20–30 min daily, aiming for 200+ score
• At end of each session, log "words I tried that were rejected" and "words I missed that were valid"

Baseline control: December, a month with no extra English practice.

Results after 30 days

Initial test: 1240 words recognized.

After 30 days: 1820 words.

Net gain: 580 words.

The number surprised me. I asked a friend — also intermediate English — to run the same experiment for 21 days. She went from 1380 to 1750, +370 words.

Both samples show clear positive transfer, completely unlike the memory-match experiment.

Why transfer works here

I spent some time analyzing why. Conclusion: the "in-game task" and "out-of-game task" are nearly identical.

In memory match, you train "short-term memory of positions in a visual grid." A very specific cognitive operation. It overlaps in label with daily memory tasks ("remembering phone numbers," "shopping lists," "coworker names") but at the level of underlying neural circuits, the operations differ drastically — which is why method improvement on memory match doesn't transfer to number recall.

Word games are different. You train "given some letters, search your brain for English words you can form." This is the same "vocabulary retrieval" operation as reading (hitting an unfamiliar word) and writing (finding a word). So even 3 calls of a word in-game noticeably raise that word's accessibility in your brain, making it available in daily use.

Cognitive science calls this conditions for near-transfer to work: training task and transfer task share the underlying operation. Word games qualify. Memory match doesn't.

Specific words I picked up

A partial list of words I learned or solidified in 30 days:

Intermediate words I'd seen but rarely used: tribe, valid, vital, scope, fault, plead, dwell, halt, ample, sworn

First-time encounters: vile (despicable), tinge (slight color), siege (military encirclement), maxim (saying), ode (poem of praise), wry (ironic), gulp (swallow loudly), thrum (low hum), pivot (turning point), prowl (move stealthily)

I was "forced" to submit each of these 2–5 times in-game. On the Day 30 test, words like tribe and valid went from "1-2 seconds of thought" to immediate recognition. Visible progress.

Calling a word 3+ times in-game starts migrating it toward active vocabulary. That's the key to 30-day results.

Why not Duolingo

Same goal of learning English — why not use Duolingo? Short answer: for me it's too light and too fragmented.

Duolingo's core mechanics are "match image to word" and "translate single sentence." For intermediate-plus learners, the cognitive load is too low — passive recognition with little active construction. Per round, only 5–10 words appear; frequency too low.

Word Drift is the opposite. You must actively construct words from a letter jumble — a much higher cognitive operation than recognition (in language-acquisition literature: productive vs. receptive vocabulary). Each round has 50–100 active retrievals — high density.

Not saying Word Drift is "better" than Duolingo. Different training. Duolingo suits zero-base starters; Word Drift suits intermediate learners building vocabulary.

Word Drift tips for English learning

Concrete suggestions if you want to use this game for English study:

1. After each round, look up the unsubmitted words. Each round has 7 letters typically yielding 30–80 valid words; you'll find 15–30. The rest — "words the game accepts but you didn't think of" — is the real study material.

2. Make a sentence with each new word. In your head or on paper. "vile" → "He spoke in a vile tone." This action upgrades recognition into production.

3. Self-test weekly. Pick this week's new words, test on Day 7 which still stick. Mark missed ones for next-week focus.

4. Don't chase high score. Counterintuitive. After 200+ I started reusing familiar words; learning slowed. Best zone is 100–180 per round — forces you to find mid-difficulty new words.

Closing

I'd been skeptical about cognitive-training games. This experiment shifted my position partially: not every "train your brain" game is empty — what matters is whether the underlying operation matches a real-life ability you need.

For English learners, anagram-style word games are worth 20 minutes daily. If you see no 100+ word lift in 30 days, that's a different cause (maybe you don't look up unfamiliar words; maybe the dictionary's too easy).

Next experiment: sudoku — long marketed as "preventing dementia." I want to see if there's evidence.

Max is BverGame's co-operator. Sample size of two is statistically very weak; treat as personal reference only. On vocabulary-game transfer, see reviews on "incidental vocabulary acquisition" (e.g., Wood 2001, Sundqvist 2019).