Word Chain Memory is a word memory game that tests ordered recall by having you retype a growing list of original words across five rounds, with 200 points per correct chain and exactly 1,000 points for a clean run. Each round begins with a brief Reveal phase in which the current chain is visible as individual word blocks, then switches to a Recall phase that hides the list before the answer input appears. The chain grows by exactly one word each round, so round one shows a single token, round two shows two, and the final round shows all five in the original order. A single distinct wrong recall is recoverable, while a second different wrong sequence triggers a visible deadlock that freezes the run until you restart. There is no timer, no speed bonus, and no hidden difficulty curve; nothing is uploaded or stored beyond a shared best score kept locally in your browser. The whole experience runs inside a single page, which is why it works well as a short, repeatable break between longer tasks.
What This Word Memory Game Actually Tests
This is a verbal working memory task built around ordered recall rather than recognition or matching. Each new round keeps every word from the previous round and tacks on one more, which means by round five you are reproducing the same opening token plus four additional tokens you have already had to remember correctly at least once before. That growing structure is the entire point of the design: the load grows predictably, and your recall has to scale with it. Because the chain is hidden from the document before the answer input appears, the test rewards strong initial encoding and a stable rehearsal strategy more than raw visual memory. If you find the chain too long by the final rounds, that pressure is the challenge the game is asking you to push past.
Unlike static list-of-words challenges where the items never change, this format puts every earlier word back on the line at every step. That is the difference between recalling a five-item list once and recalling a one-item, two-item, three-item, four-item, and five-item list in sequence: the late rounds contain the early rounds as a subset. The game treats the round number plus the exact token order as the unit of measurement, which is why word boundaries matter as much as word content and why the final score is purely a function of correct ordered recall.
The Five Growing Chains
The chain is fixed and uses ordinary tokens only as game pieces. The five rounds use the same opening word and add exactly one new word each time, in the order shown below:
| Round | Chain (left to right) | Length |
|---|---|---|
| 1 | ember | 1 |
| 2 | ember harbor | 2 |
| 3 | ember harbor velvet | 3 |
| 4 | ember harbor velvet cactus | 4 |
| 5 | ember harbor velvet cactus orbit | 5 |
These tokens are not a vocabulary list, a spelling benchmark, a diagnostic, or a language-proficiency marker. They are game pieces, and the only thing the validator checks is whether your typed sequence matches the expected chain in exact order and at exact word boundaries. Round one reappears as the opening of every later round, so memorizing the very first token sets a foundation that every subsequent recall will lean on.
Scoring and the Two-Error Rule
Scoring is exact and auditable. Each correct chain awards 200 points and the run advances to the next round:
- Round 1 correct: 200 points.
- Round 2 correct: 400 points cumulative.
- Round 3 correct: 600 points cumulative.
- Round 4 correct: 800 points cumulative.
- Round 5 correct: 1,000 points cumulative and the run is complete.
The error rule is what makes the run interesting. Your first distinct wrong recall is recoverable. If you submit the same wrong sequence again, the validator recognizes it as a repeat and does not consume a second error. Submitting a different wrong sequence, however, is a second distinct mistake and closes the run in a visible deadlock. Once the deadlock is on, the answer and phase actions freeze; Restart is the only way back, and it resets the run to round one, zero points, and zero errors.
There is no partial credit, no negative marking, no time bonus, and no random multiplier. The disclosed answer route is identical on every load, so the score is purely a function of whether you recalled each chain correctly before the second distinct error. That is why a 1,000-point finish can be audited: five correct chains, one after another, with no other input required.
How to Play the Word Memory Game
The flow is short enough to fit inside a coffee break. To play the Word Chain Memory game in your browser, work through these steps:
- Open the page and read the first chain from left to right while it is in the Reveal state.
- Press the H key, or choose the Hide words and recall button, to remove the visible list from the document and surface the answer input.
- Type the full ordered chain with spaces between words, then press Enter or use the Check button to submit.
- Read the live feedback: a correct recall adds 200 points and reveals the next, longer chain; a wrong recall leaves the score unchanged and explains what was wrong with the attempt.
- Repeat the read-hide-type cycle for rounds two through five, watching the chain grow by exactly one word each round.
- Reach round five with zero distinct errors to finish on exactly 1,000 points.
- If you trigger a second distinct error, the run deadlocks; press Restart to return to round one with zero points and zero errors.
Case, Whitespace, and Word Boundaries
The validator normalizes a small set of cosmetic issues so you can focus on order and content rather than formatting. Leading and trailing whitespace is trimmed, repeated internal spaces are accepted, and ASCII letters are compared without case sensitivity. EMBER HARBOR therefore matches ember harbor, and " ember harbor " matches ember harbor too. Mixed-case entries like Ember Harbor and all-lowercase entries like ember harbor both count as correct on the same round.
Word boundaries, however, stay exact. The validator never invents a split, so emberharbor remains one token and cannot stand in for two words. Punctuation, digits, and blank submissions are treated as harmless no-ops rather than errors, which means an accidental comma or a stray digit will not consume your single recoverable mistake. The upshot is that case and spacing are forgiving while order and boundaries are strict, which is why the design says preserve exact order and exact boundaries rather than preserve exact spelling only.
How It Compares With Other Browser Memory Tests
Several browser-based memory games test working memory, but the material and the scoring rule change what each one actually measures. Here is how Word Chain Memory sits next to the most common formats:
| Game | Material recalled | Recall rule | Score target |
|---|---|---|---|
| Word Chain Memory | Growing ordered list of original words | Type the full chain in exact order and boundaries | 1,000 points (5 × 200) |
| Number Memory Test | Fixed digit strings | Type the digits you saw | 1,000 points (5 × 200) |
| Backward Digit Span | Fixed digit strings | Type the digits in reverse order | 1,000 points (5 × 200) |
| Visual Memory Test | Grid patterns | Reproduce or match the pattern | 1,000 points (5 × 200) |
| Paired Associates Memory | Cue-to-scene links | Choose the matching scene for each cue | 1,000 points (5 × 200) |
| Memory Match | Hidden symbol pairs | Turn cards and match pairs | Paired-card based |
The key contrast is the material. Word Chain Memory uses a growing ordered word list rather than digits, grid patterns, cue pairs, or hidden cards, so it stresses verbal ordered recall instead of numerical span, visual pattern matching, paired-associate learning, or spatial location. Players who want a wider working memory comparison between digit-based and word-based routes can read the N-back vs digit span working memory guide to see how different routes target different cognitive loads.
Where the Run Lives and What Gets Saved
Everything runs locally on the page. There is no account, no upload, no microphone or camera input, no remote word service, and no network request to a server. The validator compares your typed sequence against the expected chain inside the page, and the only thing the browser keeps between sessions is a single shared best score stored in local storage on this device. The current input, the current round, and the current error count disappear the moment you reload or close the tab.
This matters because it keeps the score auditable. Because the chain and the scoring rule are disclosed, a 1,000-point run is provably reachable by typing each chain correctly in order, and nothing about the score depends on randomness, latency, or hidden difficulty. If you close the tab mid-run and reopen it later, you will see the previous best score but no in-progress state, which is why the shared best is the only number that persists across sessions.
If you're weighing options, Word Family vs Rhyming Words: A Practical Comparison covers this in detail.