Reading - Multiple Matching

Text A

Whenever I draft a multiple-matching task, I begin by deciding what I want to test: not vocabulary recognition, but meaning-tracking under time pressure. If the texts are too obviously different, candidates simply ‘spot’ answers; if they are too alike, the task collapses into guesswork. The sweet spot is engineered similarity: each text should share a surface topic, yet differ in stance, purpose, or consequence. I also write the questions last, because only then can I ensure they paraphrase rather than quote. If a text says a speaker was ‘irked’, the question should refer to ‘mild resentment’, not irritation. The final check is brutal: can every answer be justified by a specific clause, and can every distractor be ruled out without relying on world knowledge? If not, it goes back to the keyboard.

Text B

The most common flaw in ‘Cambridge-style’ imitations is that they confuse difficulty with obscurity. You don’t need rare words; you need layered ideas. A C1 candidate can cope with dense syntax, but only if the logic is coherent. So I make each paragraph do several jobs at once: it introduces a context, hints at a problem, and implies an attitude, often through hedging (‘to some extent’, ‘not entirely’). I also avoid the temptation to make one text the obvious outlier, because real exam sets typically contain overlapping references—several writers may mention research, deadlines, or feedback—while only one contains the precise nuance the question targets. To keep it fair, I pilot the items on colleagues: if two of them independently match a question to different texts, the question is not yet exam-proof.

Text C

Authenticity, for me, is mostly about how the questions are worded. In a genuine C1 paper, statements rarely ask for ‘facts’; they ask for the writer’s implication. That means the key information is frequently embedded in a subordinate clause or expressed via contrast: ‘Although X, I still…’. I therefore build questions that force candidates to interpret, not merely locate. Another hallmark is recycling themes as distractors. If one text says the project was ‘delayed’, another might say it was ‘postponed’, but the correct answer may hinge on whether the delay was self-inflicted or imposed externally. Finally, I make sure each text has at least two plausible ‘hooks’ for multiple questions, so that no paragraph feels like a one-use answer bank.

Text D

What I admire in Cambridge materials is their restraint. The tasks are challenging, yet they don’t punish candidates for being sensible. That’s why I keep the content realistic: professionals reflecting on decisions, students describing study habits, or researchers explaining trade-offs. I also watch for bias: if a question can be answered purely by recognising a fashionable term, it’s not testing reading. Similarly, I avoid cultural references that would advantage one nationality. The hardest part is balancing ambiguity and precision. A good question feels broad enough that you must read carefully, but narrow enough that only one text truly satisfies it. When I review a draft, I ask myself: could a cautious reader eliminate the wrong options by pointing to what is missing, not just to what is present? If the answer is ‘yes’, it’s close to exam standard.

Text E

I used to think authenticity meant copying the exam’s ‘voice’, but I’ve changed my mind. The tone matters, yet the underlying mechanics matter more: clean paraphrase, evenly distributed items, and distractors that are legitimately tempting. To achieve that, I build a paraphrase map before I write the questions. I list the key ideas in each text, then force myself to express each idea in a different register—concrete becomes abstract, positive becomes cautiously critical, and actions become outcomes. Only after that do I draft questions that sound neutral and don’t leak the answer through tell-tale wording. As a final safeguard, I remove any question that can be answered by matching a single noun. If a candidate can succeed without processing relationships between ideas, I haven’t written an authentic C1 task.