Syllable Split

Six real four-letter words are cut into a beginning and an ending — SPOT becomes S and POT, LANE becomes LA and NE — and then the endings are shuffled along the bottom row. Put each beginning back with the ending that finishes a real word before the tries run out. Every board is checked before it is dealt: each beginning is completed by exactly one of the six endings offered, so there is always one right answer and never a second one you were supposed to guess at.

The board tells you the one thing that makes it quick. Under every beginning is the width of the ending it needs, and under every ending is how many letters it has. A beginning of two letters can only ever be finished by an ending of two, so reading the widths first throws most of the board away before you compare a single letter.

It runs entirely on your device: 0 network requests, works offline. No ads, no account, nothing to buy, and no betting, money or prizes of any kind.

1. Read the widths first. The top row is the beginnings, each marked NEEDS 1, NEEDS 2 or NEEDS 3. The bottom row is the endings, each marked 1 LETTER, 2 LETTERS or 3 LETTERS. Only matching widths can ever join, so that mark rules out most of the row for free.
2. Focus the board — click it, or press Tab until it is outlined.
3. Move with the arrow keys or WASD. Left and right walk along a row; up and down swap between the beginnings and the endings. The cursor clamps at the ends rather than wrapping.
4. Hold a fragment with Enter or Space. It is marked ▲ HELD, and the line under the board says which width you are now hunting for. You can start from either row — hold a beginning and look for its ending, or hold an ending and look for the beginning it finishes.
5. Press its partner in the other row. If the two spell a real word they lock together with a ✓ and the word appears on both tiles. If they do not, both are marked ✗ TRIED, you are shown what they spelled, and it costs one try.
6. Changed your mind? Press Esc or Backspace, or press the held fragment again, to put it back. Sliding the hold to another fragment in the same row is free too — only a join that fails costs anything.
7. Rebuild all six words before the tries run out. You get four to seven, depending on how many endings on your board are the right width for more than one beginning. Spare tries are worth points, so a clean board scores higher than a lucky one.

What is the fastest way to match word beginnings to endings?

Filter by something cheap before you compare anything expensive. Here the cheap thing is width: a four-letter word cut after two letters leaves a two-letter ending, so a beginning of two letters can never be finished by an ending of one or three, whatever the letters look like. Reading the widths first turns twelve fragments into a shortlist of two or three before you have sounded out a single combination — and on a board where the cuts are spread evenly, that shortlist is two. Only then is it worth asking the expensive question, which is whether the letters actually spell something. This is exactly what a program does when it matches words: it indexes the dictionary by length, or by first letter, or by a hash, so that a lookup compares a handful of candidates instead of every entry. The habit generalises well beyond word puzzles — narrow by the property that is instant to check, compare only what survives — and it is the reason this game prints the width on every tile instead of making you count.

Is every board really solvable, with only one right answer?

Yes, and both halves of that are checked rather than hoped for. Six distinct words are drawn from this site's own list of 1292 common four-letter words and each is cut at a position chosen by the seed, so a complete solution exists before anything is shuffled — every beginning still has its own ending somewhere on the board. The board is then verified: every beginning is tested against every one of the six offered endings, against the whole word list, and the deal is thrown away and redrawn unless each beginning is completed by exactly one of them. That is what stops the STO-plus-NE versus STO-plus-CK problem that makes the paper version of this exercise a guessing game. A neat consequence falls out of the counting: six beginnings with one answer each means six joins in total, and every ending already has at least its own beginning, so every ending has exactly one too. That is why you can play it from either row and be equally safe.

How many wrong joins do I get?

Between four and seven, and the number is computed from your particular board rather than fixed. After the deal is verified, the game counts how many of the thirty-six possible joins are the right width but not an answer — those are the ones you can actually be tempted by, since a width mismatch is visible at a glance. A board cut two-of-each-width has only six such near misses and gets the floor of four tries. A board where most words happened to be cut in the same place can have up to thirty, and gets up to seven. The budget is read off that count, so an airy board is tight and a crowded one is forgiving, instead of one number guessed once for every board.

Is there any money, betting or in-app purchase in this?

None, and there never will be. No currency, no prizes, nothing to buy, no lives to refill and no money involved at any point. It is a word puzzle that runs on your own device.

Can I play it with a keyboard or a screen reader?

Both, with the same rules — there is no drag, no swipe and no timing anywhere in it. Arrows move, up and down swap rows, Enter holds and joins, Esc puts a fragment back. Every fragment is a real button whose spoken label spells its letters out one at a time, so S T is never read aloud as a word, and says how many letters it needs or how many it has. The live region reads the words left, the tries left and the result of your last join after every move. Held fragments say ▲ HELD, joined ones carry a ✓ and the word they made, and a join that failed marks both ✗ TRIED — so nothing on the board depends on telling colours apart.

Does it work offline?

Yes. Once the page has loaded once, the whole game is on your device — the word list, the generator, the verifier that proves the board has one answer, and the scoring. Nothing is fetched while you play, and there is no server involved at any point. Every board is generated from a seed on your own machine, which is also why the same seed always deals the same board.

Four-letter words only, drawn from this site's list of 1292 common English words, so an obscure word you know may not be in it. The cut is at a letter position, not a dictionary syllable boundary: S plus POT is a valid split of SPOT here even though nobody would call S a syllable, and the game is about matching fragments rather than about phonetics. Every board is six words and twelve fragments, and the try budget runs from four to seven. Because the deal is redrawn until every beginning is unambiguous, boards lean towards mixed cut positions — a board where all six words were cut in the same place is rare, since that is exactly the arrangement most likely to be ambiguous. There is no timer, no daily streak to protect and no hint button: the width under each fragment is the hint, and it is always on.

Anagram Pairs also asks you to pair twelve tiles into six words, but by sorted letters rather than by fragments. Word Weave crosses two four-letter words at a single shared square and asks you to fill the blanks. Word Hunt hides six words in a letter grid and rewards careful reading over recall. Word Ladder turns one word into another a single letter at a time.

Bookmark this page (Ctrl+D, or ⌘D on Mac) or install the app — it works offline the next time you need it.