Collocation Pelmanism

A memory-match game where the two halves of each pair are the node and its collocate — make / a decision, heavy / rain, reach / a conclusion, take / responsibility. Learners hunt matches on a grid of face-down cards, rebuilding each collocation from memory.

Procedure

1. Build a deck of 16–24 collocations split across two colours of card: one colour for the node (typically a delexicalised verb or high-frequency adjective), the other for the collocate. Example decks:
   • make / do / have / take + noun (16 pairs from Lewis's lexical chunks).
   • strong / powerful / heavy / tight + noun (collocate-discriminating sets).
   • phrasal verbs: verb + particle (put off, give up, look into).
2. Shuffle each colour separately, then lay both decks face-down in two adjacent grids — all one colour on the left, all the other colour on the right.
3. Each turn, a player flips one card from each grid. If they form a real collocation, the player keeps the pair and must use it in a sentence aloud before continuing.
4. Illegal pairs flip back. A pair cannot be split — a card from grid A can only be paired with one specific card from grid B.
5. The two-grid layout is the essential design choice: it forces learners to search left-to-right and remember two coordinate positions, harder than a single grid but structurally aligned with the semantics.

Why it works

Collocations are the hardest thing for intermediate learners to acquire because they live at the phrase level — neither pure vocabulary nor pure grammar — and are exactly where Vietnamese learners produce sentences that are technically grammatical but unidiomatic. This game forces the learner to handle them as unanalysed wholes, which is the Lexical Approach prescription (Lewis 2000). Meara's recent work on collocation networks also supports repeated retrieval of node-collocate pairs as the mechanism for shifting collocations from recognition to production knowledge — and this is precisely what the sentence-aloud rule produces. The BBC/British Council primary resource has documented this format across the network.

Variations

• Three-way pelmanism: Add a third grid for contexts (short sentences with a gap). The match is verb + noun + context.
• Discrimination pelmanism: One node collocates with multiple candidates. Players must find the node's best or most idiomatic match, not just a possible one, with the teacher adjudicating disputes.
• Error-spotting pelmanism: Half the "pairs" on offer are collocation errors (make homework, do a mistake). Players must avoid false pairs; penalty if claimed.

Tips

• Build the deck from a reading text learners have already worked with. Recognition of collocations they saw in context is richer than isolated pairs.
• Insist on the aloud-sentence. Silent matching is a pattern-recognition puzzle, not a language task.
• Keep a running "collocation wall" of claimed pairs from each session. Over a term, the wall becomes the class's visible lexical inventory.