The mistakes that cost EPS-TOPIK candidates the most marks

Across the mocks we mark, the same losses come back again and again, and most of them have nothing to do with how much Korean the candidate knows. They are process failures: marks handed back for reasons that a checklist, not a year of study, would have prevented. That is genuinely good news, because a knowledge gap takes months to close and a process gap can close in a fortnight. Here are the ones that cost the most, in rough order of how many marks we see them take.

Leaving blanks

This is the biggest single avoidable loss, and it is pure waste. The EPS-TOPIK has no negative marking: a wrong answer scores zero, exactly the same as a blank. That makes a blank strictly worse than a guess, because a guess on a four-option question carries a one-in-four chance of a mark and a blank carries none. Yet we see candidates leave two, three, four questions empty on every timed paper, almost always because they ran out of time and never reached the end. Those are marks given away for nothing, and the fix is a rule, not a skill: two minutes from the end, fill every blank.

Answering the question you assumed, not the one that was asked

A whole class of losses comes from negation in the question stem. Many EPS-TOPIK stems ask which option is not correct, or does not match the passage, using words like 않다, 아닌 or 틀린 것. A candidate reading at speed sees the passage, finds the statement that is true, and selects it, because their eye skated over the little negative. They knew the Korean perfectly and still lost the mark, because they answered the opposite of what was asked. The cost is high precisely because it catches confident candidates on questions they could easily have got.

Number and time slips

Korean's two number systems, chosen by context rather than size, generate a steady drip of avoidable errors: the wrong system for a price or an age, the hour and minute mixed up when telling the time, a counter dropped or the wrong native-number shape used before it. On a workplace-themed exam these turn up constantly, in prices, shift hours, dates and counts of goods, so a shaky grasp of the counting rules leaks marks across the whole paper rather than in one place.

Losing the thread in listening, and losing time in reading

In listening, the classic failure is that one missed word triggers a spiral: the candidate stops to recognise it, the next clause escapes while they do, and the question is lost, along with some of the next. It is a processing-speed problem, and it is why listening tops our mark-loss chart. In reading, the mirror image is refusing to abandon a hard question, spending two minutes on it out of sunk-cost stubbornness, and starving the easy marks at the front of the paper. Both are pacing failures wearing different clothes.

Studying vocabulary out of context

The last big one is subtler. Candidates drill vocabulary from generic textbook themes, then meet an exam built around real workplace Korean: factory lines, agriculture, construction, safety notices, wage slips, shift schedules. The words they memorised and the words the paper uses only partly overlap. The exam is testing whether you can work in Korea, so the vocabulary that pays is the vocabulary of the workplace, and time spent on unrelated word lists is quietly time half-wasted.

The thread through all of these is that the marks are recoverable quickly, because the causes are not gaps in your Korean. A candidate who fills every blank, reads each stem twice, keeps a stuck question to ninety seconds, and drills numbers and workplace words in context can pick up a double-digit swing without learning a single new grammar point. That is the cheapest improvement available to anyone sitting this exam, and it is sitting there unclaimed on most first mocks.

Fix the process before you add more Korean. The candidate who plugs these leaks walks in knowing exactly the same amount as before, and walks out with a noticeably higher score, because they stopped handing marks back.