Antonym Match is a browser-based opposite words game built around five fixed rounds, each one asking you to identify the direct lexical antonym of a single prompt word under a bounded sense. Every round shows a prompt, its part of speech, and a short sense note, then offers four candidate words; tapping the correct candidate awards exactly 200 points, so a clean run totals exactly 1,000 points. The five winning pairs are not chosen by a language model guess. Each one is anchored to an explicit antonym pointer in Princeton WordNet 3.1 for the displayed grammatical category and sense, and independently cross-checked against a Cambridge Thesaurus page that returned HTTP 200 with the expected opposite in the downloaded content. Touch, pointer, and keyboard controls all work, and a double-Escape boss key can replace the puzzle with a spreadsheet-like cover and back. No server, no account, no download, and no package dependency are involved; everything runs locally in your browser.
What an Opposite Words Game Tests
A solid opposite words game tests direct lexical opposition — the relationship two words share when their senses cancel each other out — rather than casual or contextual contrast. Antonym Match enforces that distinction by giving you three signals on every round: the prompt word, its grammatical category (such as adjective or verb), and a one-line sense note that pins down which meaning of the prompt is in play. The four candidate buttons sit beneath those signals as native HTML controls with at least 44 pixels of touch height, and all four buttons share the same markup so the winning answer is not exposed before you select it.
This matters because English is polysemous. The verb accept can describe believing a statement, receiving an object, admitting a person, or agreeing to an invitation, and its direct opposite changes with the sense. The adjective hot can refer to physical temperature, emotional tone, current popularity, or electrical state. Two words can frequently appear on different sides of a discussion without being true lexical antonyms; a movement word is not automatically the opposite of a temperature word, and a related probability term is not necessarily the direct opposite of capability. The game rejects any candidate that is merely associated, merely different, or contextually contrasted — the sense note is the contract you match against.
How to Play Antonym Match
- Read the prompt word, its part of speech, and the displayed sense note so you are looking for a direct lexical opposite in the same grammatical category, not a general contrast.
- Examine the four candidate buttons beneath the prompt and identify the one that expresses the direct opposite of the displayed meaning; ignore candidates that share a surface feature but lack the required opposite relationship.
- Tap the chosen candidate on touch devices, or use the arrow keys (Left or Up to move back, Right or Down to move forward) to highlight a choice and press Enter to submit it on a keyboard.
- Correct selections award 200 points and advance to the next round; repeat through all five fixed rounds without deadlocking the run on two distinct mistakes.
- After five correct answers, the run finishes at exactly 1,000 points and the score is saved locally as your best completed run when browser storage is available.
Inside the Five Rounds: Scoring and Sense Notes
The game is deliberately short: five rounds, four candidates each, and a strict 200-point increment per correct answer. Because every correct match is worth the same fixed amount, the score path is fully deterministic — five correct answers equals exactly 1,000 points, with no bonus, no time multiplier, and no adaptive reward. The five prompt pages cover both adjectives and verbs, and the correct answer's position is varied across all four candidate slots rather than repeating one location, so you cannot lean on positional memory between rounds.
Each candidate button is unique within its round, and several distractors share the prompt's part of speech so the round genuinely tests the sense, not just the grammar. Candidate order is fixed for reproducible testing but reshuffled across rounds, so even knowing one answer's slot tells you nothing about the others. Minimum-width and overflow protection keep long candidates from breaking the layout; on narrow screens the four buttons stack into a single bounded column and only expand to two columns on wider viewports.
| Prompt + Sense | Candidate Pattern | Direct Antonym? |
|---|---|---|
| "hot" in a temperature sense | Word from a different lexical category | No — different category, no qualifying pointer |
| A movement verb | A temperature adjective | No — surfaces look related, no antonym pointer exists |
| A probability term | A capability term in the same part of speech | No — related sense, but not direct opposition |
| "hot" in a temperature sense | Word with an explicit WordNet ! antonym pointer for that sense | Yes — verified pointer plus Cambridge cross-check |
Source Verification: WordNet and Cambridge
Every winning pair rests on two independent pieces of lexical evidence. The primary source is Princeton WordNet 3.1, whose raw adjective and verb data files encode antonymy with an exclamation-mark (!) pointer between synsets or lemmas. Each winning round corresponds to a literal direct pointer for the displayed grammatical category and sense, recorded as source and target offsets in the evidence fixture rather than as an AI-generated similarity judgment. The secondary source is a Cambridge Thesaurus page; each of the five prompt pages returned HTTP 200 during implementation, and the expected opposite appeared in the downloaded page content. The Cambridge Thesaurus entry for "hot" and the Cambridge Thesaurus entry for "accept" are two examples of those cross-checks.
Using a second lexicographic source is not decorative — it reduces the risk of relying on one database's organization or confusing a related but non-opposite term with a direct antonym. The product's private fixture stores nine golden cases: five direct pointer relationships plus four same-part-of-speech or related-sense boundaries with no qualifying pointer, so the boundaries are checked as carefully as the wins. Each product round is validated against the expected prompt, answer, category, sense, offset pattern, candidate uniqueness, and varied correct position before it ships. State tests then cover exact scoring, cursor wrapping, invalid commands, duplicate-error suppression, the second-error deadlock, terminal freeze, and Restart.
Controls: Touch, Keyboard, and Boss Mode
Touch and pointer interaction is direct: tap a candidate to submit it, and the button is a native control with at least 44 pixels of height so the touch target is comfortable on phones. Keyboard interaction covers the entire puzzle. Left or Up moves the highlight to the previous choice; Right or Down moves it to the next; movement wraps around the four candidates so you never run off the end. Enter submits the highlighted choice, and when a native button already has focus its native Enter click is allowed to handle the action so the global shortcut cannot submit twice.
| Input | Action |
|---|---|
| Tap a candidate | Submit that candidate |
| Left or Up arrow | Move highlight to the previous choice (wraps) |
| Right or Down arrow | Move highlight to the next choice (wraps) |
| Enter | Submit the highlighted candidate |
| Escape twice within the timing window | Toggle boss mode cover on or off |
| Restart (via GameShell) | Restore round one, zero score, empty mistake list |
Restart is always available through the shared GameShell. The shell also stores your best completed score locally when browser storage is available and supplies the double-Escape boss key: press Escape twice within the timing window to replace the puzzle with a spreadsheet-like cover, then twice again to return. Boss mode does not alter the lexical fixture, the score, or the active candidate — it only changes what is on screen. All gameplay runs locally in the browser, so no selections are sent to a server, no account is required, no paid service is used, and no package dependency is added.
Mistakes, Restart, and the 1,000-Point Finish
Wrong choices use deterministic signatures made from the round and candidate index, so repeating the same wrong selection does not count twice. A first distinct mistake leaves the round open and reports that the chosen word is not the direct opposite for the stated sense — you can still pick another candidate. A second different mistake deadlocks the run until Restart; this makes accidental duplicate events harmless while limiting blind guessing. Invalid negative, out-of-range, or fractional candidate indices are atomic no-ops, and unknown cursor directions also do nothing.
Completion and deadlock are terminal: once the game finishes or deadlocks, selection and keyboard movement cannot alter the score, level, cursor, or history afterward. Restart restores round one, the first highlighted candidate, a zero score, and an empty mistake list. Because the run is recreational vocabulary practice rather than an adaptive test, a reading-level diagnosis, a language certification, or an intelligence measure, completing the game proves only that these five fixed, source-checked relationships were identified. For a quiz-format view of the same five pairs, see the Antonym Quiz guide.