Skip to content
Lizely

Word Association Game

Reveal short associations one at a time and preserve more points by connecting them with fewer clues.

Privacy: your files never leave your device. All processing happens locally in your browser.

How to use

  1. 1.Read the first short association and submit a single English word immediately if you recognize the connection.
  2. 2.Use Reveal next clue when needed; each additional visible clue reduces that round's value by 40 points, down to 40.
  3. 3.Solve all five rounds, avoiding a second distinct wrong guess; first-clue answers in every round produce the exact 1,000-point maximum.

About Word Association Game

Word Association Game is a five-round progressive-clue challenge built around short related phrases rather than long riddles. Every round opens with one compact association. You can submit a one-word guess immediately or actively reveal another phrase when the connection is still uncertain. The decision to reveal is the main strategy: fewer visible clues preserve more points, while additional associations make the answer easier to identify.

The scoring table is exact and visible. Solving from the first clue awards 200 points. Two clues are worth 160, three are worth 120, four are worth 80, and all five are worth 40. Each new round resets to one visible clue. A player who recognizes every answer from its first association earns exactly 200 points five times for a maximum total of 1,000. Revealing clues never creates a negative score, and the fifth clue is the final floor at 40 points.

This loop is intentionally different from a conventional word riddle. There is no paragraph that encodes a single disguised object and no fixed block of hints shown at once. Instead, every clue is a short association such as an environment, component, season property, location behavior, or motion relationship. Reveal next clue is a deliberate action with an immediate 40-point tradeoff. The interface updates the visible clue count and current round value before you guess, so the cost is never hidden.

Input normalization is narrow and predictable. Leading and trailing spaces are removed, and ASCII letters are converted to lowercase. A correctly spelled uppercase or mixed-case version therefore behaves like its lowercase equivalent. Blank input, punctuation, digits, hyphens, and multi-word phrases are malformed and ignored atomically: they do not change the clue count, score, round, attempt set, or feedback state. The accepted response in each round is one English noun.

Wrong guesses follow a two-distinct-error rule. The first normalized wrong word records one unique signature and leaves one miss. Submitting the same wrong word again, even with different case or surrounding spaces, is deduplicated and does not consume the second miss. A second different wrong word deadlocks the game. The text input, guess button, and clue reveal control then freeze, and later reducer calls are atomic no-ops. Restart is the only recovery path and restores the first round, one visible clue, empty attempt history, zero score, and deterministic clue order.

The answer remains absent from the interactive DOM before a correct guess. The component renders only the current round number, visible association phrases, clue count, potential points, attempt count, neutral feedback, and user-entered input. It does not place answers in aria labels, data attributes, hidden text, placeholders, CSS content, or solution elements. Public page metadata uses generic WordNet and Cambridge Dictionary source links rather than answer-specific URLs, preventing source citations from becoming a side-channel. Exact per-word audit URLs remain confined to the non-page evidence manifest used by maintainers.

All five spellings and ordinary noun senses are audited reference data. Princeton WordNet 3.1 is the primary lexicon. Its official HTTPS archive returned HTTP 200, and all five entries were read directly from dict/index.noun. Five answer-specific Cambridge Dictionary pages independently cross-check those spellings and ordinary meanings; each returned HTTP 200 during evidence collection. The association phrases, reveal sequence, score table, normalization, miss rule, and round order are original product fixtures.

The automated suite keeps its oracle separate from production logic. It owns literal private copies of all five rounds, all twenty-five association phrases, and the entire score table. It verifies that every round has five unique clues and that no clue literally contains its answer. Its own trim and lowercase function checks normalization, while its own clue-count equation recomputes 200, 160, 120, 80, and 40 without reading production score constants. Ten external golden cases cover each spelling in both the primary and independent dictionary evidence.

State-machine tests solve all five rounds on the first clue and confirm the exact 1,000 maximum. They also solve after three visible clues and verify a 120-point award, stop reveal at five clues, exercise malformed-input no-ops, prove normalized duplicate-error deduplication, trigger the second-distinct-error deadlock, and confirm deadlocked and completed states cannot mutate. Restart is compared with a fresh initial state.

The interface is fully client-side and requires no account. Guesses, revealed clues, attempts, and scores are processed locally and are not uploaded for validation. Native text input and buttons support physical-key play, while every actionable target has at least a 44-pixel height for touch. Minimum-zero grid tracks, wrapping clue cards, breaking feedback text, and an overflow guard prevent structural horizontal overflow on a 390-pixel viewport. The shared GameShell supplies the score display, local best-score behavior, Restart, and Boss Key.

Methodology & sources

Five fixed noun spellings are checked in Princeton WordNet 3.1 dict/index.noun and independently on answer-specific Cambridge Dictionary pages stored only in the evidence manifest. The public DB keeps generic source URLs so citations cannot reveal answers in page DOM. Each round starts with one clue; score is max(40, 240 - 40 * visibleClues). Tests use private literal rounds and a separate score oracle.

Frequently asked questions

Related tools

Mini Games guides

View all