The Multilingual Word Scrambler & Solver

What this tool does

The Multilingual Word Scrambler & Solver is a two-way tool. Scramble takes any text you paste and shuffles the letters inside each word, producing a puzzle you can hand to someone else. Solve does the opposite: give it a set of loose letters — or a word that has been jumbled — and it returns every real word hiding inside those letters, ranked so the most likely answer sits at the top.

It works across eight languages: English, Afrikaans, Dutch, German, French, Spanish, Portuguese and Italian. That matters because anagram solving is not one skill. A set of letters that forms a satisfying word in German may be nothing at all in Portuguese, and the vowels behave completely differently from one language to the next. Choosing the right dictionary is usually the difference between solving a puzzle and staring at it.

Everything runs on your own device. There is no upload, no account and no server round-trip: the dictionary for the language you pick is downloaded once by your browser and every search after that is instant, even offline.

How to unscramble letters in three steps

Most people overthink anagram puzzles. Working through these three steps in order solves the large majority of them in under a minute.

• Count the letters first. Sort them by length — 2-letter words, then 3, then 4 — rather than trying to guess the longest word immediately. Short words give you an anchor, and knowing which letters are already used tells you what is left.
• Look for the common chunks before the whole word. Letter pairs such as th, sh, ch, ph, wh, qu, ck, ng and the vowel pairs ea, ai, oa, ee, oo, ou stick together in English; Spanish leans on ll, rr and qu; Portuguese is full of nh, lh and the ending ão; German is built from sch and ei/ie; French hides eau, oi and eu; Dutch gives you ij and sch. Find the chunk and the rest of the word often falls into place.
• Try the shell. Take the obvious prefix and suffix off the pile — un-, re-, dis-, pre-, out-, over-, mis- in front; -ing, -ed, -er, -est, -tion, -sion, -ness, -ment, -able in back — then solve what is left as a smaller anagram.

How the solver actually works

The solver does not guess and it does not use a language model. It performs an exact search against a real frequency-ranked dictionary for the selected language, which is why it never invents a word that does not exist.

Two modes are available. In Unscramble letters mode you supply a pool of letters and the solver returns every dictionary word that can be built from that pool without reusing a letter more often than you supplied it. A question mark acts as a wildcard: type cr?ft and the solver will try each letter of the alphabet in that slot. In Pattern search mode you supply the shape of the word instead — s?????e — and non-wildcard characters are locked to their exact position, which is how you finish a crossword clue or a Wordle guess.

Results are ordered by length first, then by how common the word is in real usage. That ordering is deliberate: in a genuine puzzle the answer is far more likely to be planet than platen, and frequency ranking keeps the obvious candidates at the top instead of burying them under obscure dictionary entries.

Where the words come from. Each language's dictionary is built from a frequency-ranked corpus of real-world usage, so words people actually write rank above obscure ones. English is additionally checked against a 63,000-entry lexical reference (a lowercase-only SCOWL subset), which removes the proper nouns, abbreviations and contraction fragments that corpus data drags in — isn, split out of isn't in a subtitle file, sits at rank 198 in the raw corpus and would otherwise show up in your results. The other seven languages are corpus-derived and hygiene-filtered only, because the lexical references available for them list base forms without inflections, and intersecting against those would silently delete a third of their valid words — ordinary plurals such as the Spanish libros and the Portuguese casas among them. So expect the occasional name or loan word in a non-English result, and tell us if you find one.

Letter patterns differ by language

If a puzzle refuses to break, it is usually because you are applying one language's habits to another. These are the patterns worth knowing.

• English — E, T, A, O, I and N dominate. Almost no native words end in i, u or v, and q is followed by u barring a handful of loanwords. A word with no vowel other than y is real but rare, so treat it as a last resort.
• Afrikaans — long vowels are written as doubled letters (aa, ee, oo, uu), so watch for pairs that a solver can split. The digraphs oe and ie are everywhere, and the language is unusually rich in short, high-value words such as die, van, en, is, het and nie.
• Dutch — the ij digraph and the sch cluster are the two signatures. Long compound nouns are normal, so a 12-letter Dutch answer is not automatically a trick.
• German — ä, ö, ü and ß are distinct letters and count separately, and nouns capitalise, which is a useful hint when you can see which letter was capitalised in the original. Compounds are aggressively long, so try splitting the pile into two real words before treating it as one.
• French — the diacritics é, è, ê, à, ç, ô, û and î are common, silent final consonants are everywhere, and eau, oi and eu are the vowel clusters to hunt for first.
• Spanish — ñ is its own letter, ll and rr are distinct sounds, and the language is unusually vowel-heavy at roughly one vowel for every two consonants, so a pile with very few vowels is rarely Spanish.
• Portuguese — ã and õ are nasal vowels, ç and the nh/lh digraphs are frequent, and words ending in -ão form an enormous family. If you see a and o left over at the end, try -ão.
• Italian — almost every word ends in a vowel, and doubled consonants change the meaning (pala is a shovel, palla is a ball), so a pile containing a doubled letter that the solver ignores is usually a sign you picked the wrong language.

Scrambling text without making it unreadable

The scrambler is not limited to a single word. Paste a paragraph and every word is shuffled in place, with punctuation, capitalisation and spacing preserved so the result still reads like a sentence.

Two options control difficulty. Keep first and last letters leaves the word's silhouette intact — the effect used by the well-known Cambridge University reading demonstration, where readers recognise shuffled words from their outline alone. Turning it off produces a much harder puzzle. Minimum word length stops the scrambler touching short words such as a, an, of and to, which otherwise turn a readable sentence into noise.

Randomness comes from the browser's cryptographic random number generator, so every scramble is genuinely different — you cannot produce the same puzzle twice by accident, and there is no hidden seed to reverse-engineer.

Ways to use it beyond word games

Anagram solving turns out to be a general-purpose skill rather than a Scrabble-only trick.

• Teachers and tutors build spelling and vocabulary worksheets by scrambling a class list, then handing the shuffled version out with the original kept back as an answer key.
• English as a second language learners use the solver in reverse: the frequency-ranked output shows which of several valid words is the one people actually say.
• Word games — Scrabble, Words With Friends, Boggle, Letter Boxed, crossword finishes and Wordle guesses. Note that our dictionary is a frequency-ranked common-words list, not the official tournament lexicon, so treat it as a strong hint tool rather than a tournament referee.
• Naming and branding — an anagram of a competitor's or a client's name is a fast route to a memorable product, handle or domain. Feed in the letters, then read the short results for something pronounceable.
• Teaching cryptography — a letter transposition cipher is a scramble with an extra step, and seeing how badly a simple shuffle damages a message is the clearest possible introduction to why real ciphers use keys.

Privacy, speed and what leaves your device

Nothing you type is transmitted anywhere. The solver runs entirely in JavaScript in your browser tab, and the word list for your selected language is fetched once as a static file from the site's own content delivery network, exactly like an image.

That design has a practical benefit beyond privacy: there is no request per search, so results appear as fast as you can type, and the tool keeps working if your connection drops after the page has loaded. It also keeps the page itself tiny, which is why it loads in a fraction of a second on a mobile connection.