How to teach vocabulary with word puzzles
18 August 2026
A word puzzle handed out at the end of a lesson keeps a class quiet. That is the usual reason it is used, and it is a poor one. Used deliberately, the same sheet drills something narrow and genuinely useful — but which skill it drills depends entirely on two decisions: how the sheet is designed, and how the lesson is run.
Those two decisions are almost never discussed. Generators produce puzzles; nobody explains what a puzzle is for. What follows is an attempt to be specific.
What a scramble actually tests
Unscrambling letters is not a vocabulary test. A learner who does not know the word cannot find it, but knowing the word is not sufficient either. What the task really demands is orthographic recognition — reconstructing a written word from an incomplete or disordered signal.
That skill sounds trivial to a fluent reader because it is automatic. It is not automatic for a learner whose first language uses another script. Reading fluently means recognizing word shapes without sounding out each letter, and for Thai, Khmer, Burmese or Arabic first-language learners, that recognition routinely lags spoken ability by a wide margin. A student who discusses their weekend comfortably in English may still read at a crawl.
The gap between speaking fluency and reading fluency is the thing a scramble isolates. That is its whole pedagogical value.
This has a practical consequence for level choice. If your learners' first language does not use the Latin alphabet, set the puzzle a level below what their speaking suggests. You are testing a different skill from the one their conversation displays.
Three procedures, three different exercises
The same twelve scrambled words can be three distinct activities. The difference lies entirely in what happens before the sheet is handed out.
Introducing new vocabulary: pre-teach
Write four of the target words on the board. Elicit meaning, drill pronunciation, then erase them. The students have now seen the spellings once — enough to make the task achievable without making it trivial.
Then work in pairs, one sheet between two. The pairing is the point: the talk while two learners negotiate a spelling is where the learning happens, not the writing. Circulate and note which words stall the room; those are your reteaching list.
Model the first item aloud before they start, saying your own process: find the vowels, look for a familiar ending, try a first letter. Learners who stall on scrambles usually lack a strategy rather than vocabulary, and the strategy is teachable in sixty seconds.
Revising old vocabulary: do not pre-teach
Here the instinct that serves you in the first case actively destroys the second. If the words are not new, putting them on the board first removes the entire mechanism.
Retrieving a word from memory strengthens the memory in a way that re-reading it does not. This is well established in the learning literature, and it is the specific reason a revision sheet works at all. A revision activity that shows students the answers first is re-reading with extra steps.
So: hand out cold. No board work, no word bank, no dictionaries. Five minutes alone, and tell them a blank is an acceptable answer — you want to know what they genuinely cannot retrieve. Then three minutes in pairs, filling each other's gaps; two partial memories often reconstruct a word neither student had alone.
Have each student count how many they got unaided, before the pair stage, and write that number at the top. That figure is the useful data. Then reteach only the words the class could not retrieve, and put those into the next revision cycle.
Timing decides whether any of this works. Run the sheet in the lesson you taught the words and it will show you nothing, because the words are still in working memory and everyone scores well. The useful window opens once forgetting has begun — two to three weeks for a weekly class is a reasonable default, with a second run at eight to ten weeks.
True beginners: scaffold, then remove the scaffolding
With learners still becoming comfortable with the Latin alphabet, the sheet needs a word bank at the top. Without one you are asking them to remember and rearrange simultaneously, and the second task collapses under the first. With a bank, the demand becomes matching rather than recall, which is the appropriate level of difficulty.
Two adjustments help more than they sound as though they should. Start with puzzles where only one adjacent pair of letters has been swapped, rather than a full scramble: the learner sees a nearly familiar shape and has to spot one thing out of place. And have students write each solved word twice underneath. Recognition and production are different skills, and the second is the one that shows up in their writing later.
Two design faults that ruin a sheet
Puzzles with two valid answers
HEAT and HATE contain the same letters. If a generator gives you one of them as a puzzle, a student who writes the other is not wrong, and you will spend the feedback stage arguing about a worksheet instead of teaching. Before a scramble is used, it should be checked against every other word of the same length in the list; if the letters could spell something else, that arrangement should be thrown away.
Uneven mixing
This one is invisible until you watch a class. A random shuffle produces wildly different difficulties: BKOO for book moves two letters and is solved instantly, while STTUEDN for student is thoroughly mixed. On the same sheet, finishing times spread badly and half the room ends up waiting.
The fix is to require a minimum proportion of letters to change position — around sixty per cent works — so every item on a sheet presents a comparable demand.
A related point: avoid mixing three-letter and ten-letter words on one sheet, for the same reason.
What a scramble cannot do
It is worth being clear about the limits, because a technique oversold is a technique soon abandoned.
- It does not teach meaning. Solving BKOO tells a learner nothing about what a book is. Meaning has to be taught before or attached alongside.
- It does not teach pronunciation. A scramble is entirely orthographic. Drill the sound separately.
- It stops being useful somewhere around B2. By then a learner recognizes written words fluently; their gaps are collocation, register and nuance, none of which rearranging letters touches. At advanced levels, pair each scramble with a definition so the task becomes recalling which word carries that meaning — the letters as a prompt rather than the puzzle.
- It will not fix a spelling problem on its own. Learners whose first language does not permit consonant clusters at the start of a syllable — Thai and Spanish among many others — find stream and spend harder to reassemble than their length suggests, because the cluster is not a familiar shape. That is not a vocabulary failure and it does not respond to correction. Contrastive drilling does: port / sport, tone / stone, pin / spin.
Where to start
Pick one set of vocabulary you taught three weeks ago. Build a revision sheet from it, run the no-pre-teach procedure, and record the unaided scores. It takes fifteen minutes of class time and it will tell you more about what your students actually retained than any test you could set.
Ready-made sheets with these procedures attached are in the worksheet library — the ESL sheets for new vocabulary, the beginner sheets for learners still working on the alphabet, and the revision sheets for sets you taught weeks ago. To build from your own syllabus, paste the list into the scramble maker.
Two applications of all this sit elsewhere on the site: Buddhist Studies English, where ordinary words carry doctrinal senses and the definition-matched format earns its keep, and the Thai curriculum mapping, which places a scramble against the standard it can honestly be cited under.