Rhyme matching games usually rely on spelling endings, but a real rhyme is defined by the spoken sound from the last stressed vowel to the end of the word. In Rhyme Match, every round asks one focused question: which of four candidates shares the target word's stressed tail? The answer is verified against CMUdict ARPAbet phonemes rather than letter patterns, so pairs such as moon and spoon (UW1 N) qualify while misleading neighbours like moon and moan (OW1 N) do not. Each correct round awards 200 points, and clearing all five produces an exact score of 1,000. The five verified pairs are light and night, cake and lake, moon and spoon, train and rain, and blue and glue, and their stressed tails are AY1 T, EY1 K, UW1 N, EY1 N, and UW1 respectively. Distractors are chosen to fail for a specific phonetic reason, such as a different stressed vowel or a different final consonant, and the correct positions rotate across rounds so repeatedly picking the same slot will not clear the game.
Why Spelling Fails When You Match Rhyming Words
Spelling is the most common trap when matching rhyming words in casual play, because English letter patterns often point at the wrong sounds. Rhyme Match sidesteps that trap by ignoring letters entirely and comparing phonemes from the last stressed vowel through the end of each word. Two pairs illustrate why this matters.
The pair rough and cough shares the conspicuous written ending "ough," yet the CMUdict tails are R AH1 F and K AA1 F. The stressed vowel shifts from AH1 to AA1 and the onset shifts from R to K, so the stressed tails do not match. A spelling-based quiz would mark the pair as rhyming; the pronunciation rule rejects it.
The pair light and late ends with the same written letter T, but the phoneme set is AY1 T against EY1 T. The stressed vowels differ, so the tails differ. A spelling-based quiz would see T and accept the pair; the pronunciation rule rejects it. The maintained CMU Pronouncing Dictionary repository and its raw dictionary file back every tail used in the game, and Cambridge word pages independently cross-check the ten played words plus rough and cough.
What the Stressed-Tail Rule Actually Compares
Rhyme Match uses an explicit, testable definition of a rhyme rather than a guess from spelling. For each cited CMUdict pronunciation, the game scans backward for the last vowel carrying primary stress (the "1" suffix) or secondary stress (the "2" suffix). The tail that matters is every phoneme from that vowel through the final phoneme of the word. A valid match must have exactly the same stressed tail and a different onset.
This rule has three practical consequences for matching rhyming words. First, the stressed vowel anchors the comparison, so light with primary stress on AY1 is compared only from AY1 onward, not from the silent GH or the L onset. Second, every consonant after the stressed vowel must also match, which is why moon and moan share the final N but differ on the preceding vowel (UW1 N versus OW1 N). Third, the rule excludes an identical pronunciation paired with itself, so a duplicate word on the candidate list cannot count as a rhyme because the onset must differ.
The same rule decides what does not rhyme. A pair fails if the stressed vowels differ, if a vowel is inserted before the final consonant, or if the onset is the same so that the words are pronounced identically. Every round contains exactly one candidate satisfying the rule, and distractors are chosen so that each one fails for a specific phonetic reason that can be reasoned about, not just "sounds off."
How to Play One Round of Rhyme Match
The play loop is short and deterministic: read the target, pick a candidate, submit, and move on. Five rounds at 200 points each produce a fixed 1,000-point ceiling, and the route does not depend on timer pressure or random draws.
- Read the target word and consider the four candidates. Decide which candidate you believe shares the complete ending sound from the last stressed vowel onward with the target. The candidate label is the word only, with no phoneme hint exposed before submission.
- Select the candidate. Touch or pointer users tap one of four answer controls, each at least 44 pixels tall, arranged in one column on narrow screens and two columns when space permits. Keyboard users press keys 1 through 4 to select directly, or use the arrow keys to move through the two-by-two option grid and press Space to choose the highlighted option.
- Submit the match. Touch and pointer users press the full-width Submit button below the options. Keyboard users press Enter. The game does not advance on selection alone; submission is a separate atomic action.
- Review the feedback. After a correct submission, a review note may show the previous pair and its verified CMUdict tail, such as AY1 T for light and night. Wrong submissions do not reveal the answer, so the current question stays a fair test.
- Clear all five rounds and score 1,000 points. Each correct round awards exactly 200 points. Reaching all five correct answers produces a final score of 1,000. Restart from the GameShell boss screen if you want to repeat the route.
The Five Verified Rhyme Pairs and Their Phonemes
The game ships with five audited pairs that cover different vowel and consonant patterns. Each pair is checked against the CMU Pronouncing Dictionary raw data and a corresponding Cambridge dictionary page. The table below records the target, the matching candidate, the verified CMUdict tail, and the zero-based option index of the correct choice in round order.
| Target | Verified Match | CMUdict Tail | Correct Option Index |
|---|---|---|---|
| light | night | AY1 T | 1 |
| cake | lake | EY1 K | 3 |
| moon | spoon | UW1 N | 0 |
| train | rain | EY1 N | 2 |
| blue | glue | UW1 | 1 |
The five correct option indices, taken in round order, are 1, 3, 0, 2, and 1. The correct answer does appear at index 1 more than once, but no single index covers all five rounds, so a player who tries to clear the game by always picking the same slot cannot succeed. Each round includes distractors that fail for an articulated phonetic reason, such as a different stressed vowel, a different final consonant, or an identical onset that would disqualify the pair as a self-match. You can confirm the cake and lake pronunciation directly on the Cambridge word page for cake as an independent cross-check.
Scoring, Errors, and Accessible Controls
Rhyme Match uses exact, deterministic scoring. Each correct round contributes 200 points, and solving all five produces exactly 1,000 points. There is no partial credit and no time bonus, so the score is a direct readout of correct-round count.
Wrong attempts are tracked by a signature built from the round number and the option chosen. Repeating the identical wrong selection in the same round is an atomic no-op; the game ignores the duplicate so you do not lose an attempt to a misclick. A second distinct wrong signature, however, closes the run until you press Restart. Invalid option indexes, submitting without any selection, cursor moves outside the option grid, and every action after completion or deadlock are all atomic no-ops that preserve the current state. This keeps a stray keystroke from ending a run on its own.
Accessibility is part of the contract. Touch and pointer users see four answer controls plus a full-width submit button, all at least 44 pixels high, with options in one column on narrow screens and two columns when space permits. Keyboard users can move with arrows, choose with Space or keys 1 through 4, and submit with Enter. The accessible page contains only the target word, four neutral candidates, cursor state, and selection state before grading, so the interface does not leak phonemes, shared tails, or answer markers while you play. After a correct submission, the prior pair and its verified ARPAbet tail may appear in a review note, which is the only moment a tail becomes visible.
What a Pronunciation Rhyme Test Cannot Promise
Rhyme Match makes a narrow claim: the five answers rhyme under the cited CMUdict entries and the disclosed stressed-tail rule. It does not promise that every English accent realizes every vowel identically, that spelling predicts speech, or that the five-round set covers every possible rhyme. The interface states this limitation in plain language next to the rule, so the scope is visible before you start.
Pronunciation varies across speakers, regions, and accents. A vowel that one dialect resolves as UW1 may surface differently in another, which means a stressed tail can look clean on paper yet sound ambiguous to a real ear. The game therefore describes itself as a five-round pronunciation exercise rather than a universal rhyming dictionary. Independent cross-checks against Cambridge dictionary pages, the maintained CMU Pronouncing Dictionary repository, and its raw dictionary file keep the cited pairs auditable, and the test fixture plus a private oracle verify the five positive pairs plus spelling and vowel boundary negatives without importing runtime answer tables. That audit trail is the reason the score can be exact, and it is also the reason a player should treat the result as a check on one well-defined pronunciation model rather than a verdict on every accent.