In short
- A word-path puzzle gives pupils a theme, not a word list, so they must recall the words before they can find them.
- Tinkham (1993) and later studies found that learning new words in tight semantic sets, such as a list of fruits, slowed learning; thematic sets did not show that cost.
- Words revised from one story form a thematic set of words already met, which is the safer case.
- No study has tested this exact puzzle format, so treat the reasoning here as reasoning, not proof.
The format is new to most classrooms. A grid of letters hides five or six words that belong together, and the path through each word can bend up, down, sideways or diagonally. Pupils get the theme and nothing else. Some will know it from the newspaper puzzle Strands. This article explains what the puzzle asks of a learner's memory and what research on learning related words together does and does not tell us. It then shows why revising the words of one story is the safer use, and how to make a puzzle on paper in ten minutes.
What does a word-path puzzle practise?
A word-path puzzle practises recalling that a word exists, spelling it, and seeing how the words of one text belong together. Because no list is given, pupils cannot match shapes. They have to think "a storm at sea: storm, wave, rope, sail" and then look for those letters. The finding comes after the remembering.
That is the difference from a word search. In a word search the list is printed, and the words lie in straight lines. The task is to match letters you can already see, which is recognition. In a word-path puzzle the task is to produce a candidate word from memory and then test it against the grid. Our article on whether word searches teach vocabulary takes up the first format.
| Feature | Word search | Word-path puzzle |
|---|---|---|
| Word list given | Yes | No, only a theme |
| Shape of each word | A straight line | A path that bends, including diagonally |
| Main mental act | Matching letters you can see | Recalling a word, then testing it |
| What it checks | Recognition of the written form | Recall of the written form, and the link between the words |
| Meaning | Not needed | Needed to think of the theme's words; not checked by the grid |
The puzzle does not teach meaning either. It checks that a pupil can bring the right words to mind for a theme. What the words mean must already be in place, and that is why the source of the words matters so much.
Why recall comes before finding
In a grid where paths bend in eight directions, scanning does not work. Almost every letter has several neighbours, so the number of possible paths is enormous. Pupils who scan get stuck. Pupils who first think of a word and then look for its first letter get through.
This recall is the act that memory researchers call retrieval. Barcroft (2007) gave English speakers learning 24 new Spanish words a six-second pause to try to recall each word before it appeared. Those words scored higher on tests given immediately, two days later and a week later. The study used word-picture pairs, not puzzles, but the act of trying to bring a word to mind is the common element.
It also explains why the words must already be known. If a word has never been met, there is nothing to retrieve, and the puzzle becomes a slow word search without the list. Our article on word-guessing games makes the same point about a different format.
Do related words interfere with each other?
Here is the caution. The puzzle asks for words that belong together, and a body of research warns against learning related words together. It is worth knowing what those studies actually found.
Tinkham (1993) ran two experiments in which learners studied made-up words paired with words in their own language. Words presented in semantic clusters, such as a set of clothes or fruits, took more trials to learn than unrelated words. Waring (1997) repeated the design with Japanese words paired with made-up words and found the same main effect against related sets. Erten and Tekin (2008) moved the question into a school. Sixty fourth-graders in Turkey were taught 80 words either in semantically related sets or in unrelated sets. The unrelated sets were learnt better, the difference lasted, and pupils took longer to complete the test on the related words.
Nation (2000) reviewed this research for teachers and concluded that learning related words at the same time makes them harder to learn, and that the difficulty can be avoided by learning them separately.
| Study | Learners and material | Finding |
|---|---|---|
| Tinkham (1993) | Two experiments; made-up words paired with first-language words | Semantic clusters took more trials to learn than unrelated words |
| Waring (1997) | Japanese words paired with made-up words | Replicated the main effect against semantic sets |
| Tinkham (1997) | Semantic clusters compared with thematic clusters | Semantic clusters hindered learning; thematic clusters helped it |
| Erten and Tekin (2008) | 60 fourth-graders in Turkey; 80 words taught in class | Unrelated sets learnt better; the difference lasted |
| Li, Uchihara, Nakata and Murphy (2026) | Meta-analysis of 27 studies | Cost appeared in trials needed to learn, not on immediate post-tests |
Thematic sets did not show the same cost
The story has a second half. Tinkham (1997) compared two kinds of grouping. A semantic cluster is a set of words of the same type: eye, nose, ear, mouth, chin. A thematic cluster is a set of words that belong to one scene: frog, green, hop, pond, slippery, croak. He found that semantic clustering hindered learning while thematic clustering helped it.
The evidence is also less settled than the early studies suggest. Li, Uchihara, Nakata and Murphy (2026) pooled 27 studies in a meta-analysis. The cost of semantic clustering showed up clearly in studies that counted how many trials learners needed. On immediate post-tests, related and unrelated sets came out level. Studies that presented the words in a meaningful context tended to find semantic grouping helpful rather than harmful. Their conclusion was that the effect depends on the conditions, and that is the honest summary.
Why revising words from one story is the safer case
Put the findings together and a word-path puzzle built from a story sits on the safe side for three reasons.
- The words are not new. The interference studies above all taught learners new words at the same time. A puzzle built from a story the class has finished asks pupils to recall words they have already met and understood.
- The set is thematic, not semantic. Storm, rope, sail, captain and afraid are not five members of one category. They are five words from one scene, which is the kind of set Tinkham (1997) found helpful.
- The words come with context. Pupils met them in sentences, and Li and colleagues (2026) found that studies giving words a meaningful context tended to find grouping helpful.
Now the plain statement. No study has tested this exact puzzle format, with any age group, in any language. We searched for one and found none. The reasoning above rests on studies of word lists, word pairs and taught lessons. It is reasoning, and you should weigh it as such.
Two practical rules follow. Do not use the puzzle to introduce words. And avoid grids that hide a tight semantic set the class met only this week, such as six fruits or the days of the week, because that is the case the research warns about.
How to make one on paper
- Choose four to six words from the story that share a scene, not a category. Mix nouns, verbs and adjectives.
- Draw a grid. Four by four holds 16 letters; six by six holds a chapter's worth. Count the letters in your words first.
- Write each word along a bending path, one letter per square, moving to a neighbouring square each time, including diagonally. Use every square if you can, so that there are no stray letters.
- Write the theme above the grid: "A storm at sea." Do not write the words.
- Prepare two hints for pupils who are stuck: the first letter of one word, or the number of letters in another.
| Column 1 | Column 2 | Column 3 | Column 4 |
|---|---|---|---|
| S | T | O | R |
| P | O | R | M |
| E | E | S | A |
| S | A | L | I |
The answer paths
STORM runs along the top row and bends down to the M.
ROPE starts at the R in the second row, runs left to the P, and bends down to the E.
SEA starts at the S in the bottom left, goes diagonally up to the E, and back down to the A.
SAIL starts at the S in the third row, goes right to the A, down to the I, and left to the L.
How to use it in a lesson
The puzzle takes five to eight minutes in pairs after a story or a unit is finished. What you do in the two minutes after it decides whether it was vocabulary work.
- Give the theme and the grid. Pairs write the words they expect to find before they look, then search.
- Release hints only on request, and note who needed them.
- When the words are found, ask for the link: "Who held the rope? Why did the sail matter?" Every word gets one sentence that ties it to the story.
- Finish with one sentence per pair that uses two of the words together.
Step one is the retrieval step. Step three is the connection step, and it is what a printed word list would have skipped. Between them they turn a grid of letters into five minutes of recall and meaning. For where such short recall sessions fit in a week, see how many times a learner must meet a word.
Common pitfalls
- New words in the grid. The pupil has nothing to recall. Build the puzzle from a finished story only.
- A tight semantic set. Six colours or six animals learnt this week is the case the research warns about. Choose words from one scene instead.
- Giving the list. With the list, the task becomes a word search. Give the theme and hints, not the words.
- A grid too large. A six-by-six grid with eight words can take twenty minutes. Start with four by four.
- Stopping at the grid. Without the connection step, pupils have practised recall and spelling but not meaning.
- Hints too early. A hint given before a pupil has tried to recall removes the point of the puzzle.
How iRead handles this
In iRead the word-path puzzle is called Intellect Link. A letter grid hides several related words from the story the class is reading, and the path to each word can bend in any direction, including diagonally. A small number of hints is available. The grid is regenerated every 24 hours from the story's vocabulary, with no teacher setup. A clear without hints counts double in the word's progress record.
The core principle: A word-path puzzle asks pupils to recall words before they can find them. Build it from words they have already met in one story, and you stay on the safe side of the research on related words.
See it in iRead: Intellect Link hides related words from the story in a letter grid. Paths bend in any direction, including diagonally, and a small number of hints is available.
See the vocabulary gamesKey takeaway
Use a word-path puzzle after a story, never before it. Ask for the words, then ask for the links between them.
Frequently asked questions
What is a word-path puzzle?
A grid of letters that hides several words on one theme. Each word runs along a path that can turn in any direction, including diagonally, and no word list is given. Pupils must think of the words first and then trace them, which makes it a recall task rather than a matching task.
Is it the same as a word search?
No. A word search prints the list and lays the words in straight lines, so pupils match letters they can already see. A word-path puzzle gives only a theme, and the words bend through the grid. Pupils have to recall each word before they can find it.
Is it bad to teach related words together?
For new words, several studies say it can be. Tinkham (1993) and Erten and Tekin (2008) found that new words taught in tight semantic sets were learnt more slowly or less well than unrelated words. Tinkham (1997) found that words linked by a scene did not show that cost. Revising known words is a different case.
Can I use the puzzle to introduce new vocabulary?
We advise against it. Pupils cannot recall a word they have never met, so the puzzle turns into a slow word search. Introduce the words through the story first, then use the puzzle to bring them back a day or a week later.
How many words should one grid hide?
Four to six. Fewer than four gives little to connect; more than six takes too long to build and to solve. Keep the words to one scene from the story, mix nouns, verbs and adjectives, and use every letter in the grid so that there are no stray letters.
Sources
- Tinkham, T. (1993). The effect of semantic clustering on the learning of second language vocabulary. System, 21(3), 371-380. doi.org/10.1016/0346-251X(93)90027-E
- Tinkham, T. (1997). The effects of semantic and thematic clustering on the learning of second language vocabulary. Second Language Research, 13(2), 138-163. doi.org/10.1191/026765897672376469
- Waring, R. (1997). The negative effects of learning words in semantic sets: A replication. System, 25(2), 261-274. doi.org/10.1016/S0346-251X(97)00013-4
- Erten, İ. H., & Tekin, M. (2008). Effects on vocabulary acquisition of presenting new words in semantic sets versus semantically unrelated sets. System, 36(3), 407-422. doi.org/10.1016/j.system.2008.02.005
- Nation, P. (2000). Learning vocabulary in lexical sets: Dangers and guidelines. TESOL Journal, 9(2), 6-10. doi.org/10.1002/j.1949-3533.2000.tb00239.x
- Li, M., Uchihara, T., Nakata, T., & Murphy, V. A. (2026). The effectiveness of semantic clustering on vocabulary learning: A multilevel meta-analysis. System, 142, 104141. doi.org/10.1016/j.system.2026.104141
- Barcroft, J. (2007). Effects of opportunities for word retrieval during second language vocabulary learning. Language Learning, 57(1), 35-56. doi.org/10.1111/j.1467-9922.2007.00398.x