The TOPIK IBT question types you have never seen on paper

The official trial shows you the buttons. It never asks you a question.

If you have looked for a way to prepare for the computer-based TOPIK, you have almost certainly ended up on the official "Experience TOPIK IBT" page. It is worth the ten minutes. You learn where the timer sits, what the volume control looks like, and that there is a button that opens every question in the section at once.

And then it stops. Several independent English-language TOPIK guides say the same thing plainly: the official trial gives you interface familiarity, but it has no question bank and no answer scoring. You can press the buttons. You cannot find out whether you can actually answer a drag-to-order item in ninety seconds, or what your hand does when a listening clip ends and the screen moves on without you.

That gap is the entire reason this post exists.

There is a second reason, and it comes from research rather than from us. Published Korean academic work on TOPIK IBT writing found something that surprises most people: candidates did not report the computer-based test as harder than the paper one. What they named as the burden was typing speed and handling the test on screen, and they asked for the chance to practise the format in advance, with support.

Read that carefully, because it changes what preparation means. The difficulty most candidates hit on IBT day is not linguistic. It is operational. And operational difficulty is the one kind you can delete completely before you sit down.

The five formats paper cannot deliver

These are not new question topics. Most of them test the same reading and listening skills the paper test has always tested. What changes is the shape of the answer — and the shape of the answer changes how you should work.

1. Choose exactly two — 두 개 고르십시오

What you see. A normal-looking four-option item, except the instruction is underlined: 두 개 고르십시오. Two of the four are correct.

What it does. The program enforces the count. Select two options, then click a third, and the third click is refused — the real program pops up 2개를 고르십시오 and unchecks it. You cannot "pick three and hope". Our demo enforces the same rule, though it shows the message as a live text notice instead of a modal alert, because an alert is unusable on a phone and with a screen reader.

How to approach it. Drop the "best answer" instinct completely. This item is four independent true/false judgements, not one comparison. Go down the options one at a time and ask only: can I point at the line in the passage, or the sentence in the clip, that makes this true? You need two you can point at. If you end up with three candidates, one of them is an inference you made rather than something the text said — and that is the one to cut. There is no partial credit: the item is right only when the pair is right.

2. Drag to order — the 가 / 나 / 다 cards

What you see. A stem sentence at the top, three cards labelled 가, 나 and 다 below it, and three empty slots. The cards arrive shuffled. You drag a card into a slot; on a phone you tap a card and then tap a slot; with a keyboard you Tab to a card, press Enter, Tab to a slot, press Enter. Tapping a filled slot sends its card back.

What it does. It counts as answered only when all three slots are filled. There is no ①②③④ list of possible orderings to work backwards from — and that is the real difference from the paper version, where you could often eliminate two orderings on sight and choose between the remaining two. Here you have to build the sequence yourself.

How to approach it. Work from the joins, not from the meaning. Find the card that cannot come first: one that opens with 그래서, 그러면, 그런데, or that refers back with 그 + noun, 이런 경우, or a pronoun. Find the card that answers the stem directly and place it in slot one. Then check your sequence by reading stem → 1 → 2 → 3 out loud in your head. If a connective sounds unmotivated, two cards are swapped.

3. The O/X grid

What you see. A passage or a clip, then a small table of statements — usually two — each with an O button and an X button. You mark every row.

What it does. It is the familiar 내용 일치 item, compressed. But because it is a grid, a half-answered item is visible as a half-answered item, and every row has to be marked before it counts.

How to approach it. Treat each row as its own question and do not let the first one bias the second — they are written to be independent, and "both O" or "both X" is not a pattern the test avoids. And be strict about the standard: a row that is mostly true is X. In the demo's reading grid, a notice says weekday hours are extended and weekend hours are unchanged, and the second statement claims the weekend hours were extended too. Half of that sentence is supported. The row is still X.

4. Sentence insertion

What you see. A passage with four numbered slots printed inline in the text, plus one sentence sitting apart from it. You choose which slot the sentence belongs in — and when you click a slot, the sentence is written into that position so you read it in place.

What it does. That in-place preview is the whole interaction, not decoration. On paper you had to hold the candidate sentence in your head and imagine it at each 〈 〉. On screen you can see it seated, and then move it.

How to approach it. Read the loose sentence first and mark what it depends on — 여기에, 이런 경우에는, 그래서, a repeated noun. Those words point backwards, so the slot you want is the one immediately after whatever they point to. Then use the preview properly: place it, and reread only the sentence before and the sentence after. If both joins hold, stop. Candidates lose time here by rereading the whole passage four times.

5. Inline cloze — the dropdown inside the sentence

What you see. A sentence with a blank in the middle of it, and the blank is a real dropdown. The four options live inside the dropdown. There is no list of ①②③④ under the sentence.

What it does. It hides three options at all times. You can only ever see the sentence completed one way, which makes side-by-side comparison — the standard paper technique for grammar items — impossible.

How to approach it. Decide before you open it. Read the sentence, work out what relation the blank has to carry (purpose, concession, cause, sequence), say the ending you expect to yourself, and only then open the dropdown and look for it. If you open the dropdown first, you will read four endings one at a time, forget the first by the time you reach the fourth, and reopen it three more times.

Listening plays once, and there is no way back

This is the part candidates underestimate, because on paper the listening section is also played once — so it sounds like nothing has changed. It has.

On screen, listening is delivered one screen at a time. Each screen runs through three phases: 대기 (wait), 듣기 (listen), 풀이 (answer). When 풀이 ends, the screen advances by itself. There is no Previous button and no Next button in the listening footer at all; the progress bar cannot be dragged; and if an answer arrives after its question has closed, the server refuses it rather than quietly accepting it. A screen can also carry two questions on one clip — in our demo, one clip is followed by a 중심 생각 question and an O/X grid together, which is exactly how the official format pairs items.

Concretely, here is what that does to note-taking.

• Read the options during 대기, not during 듣기. The wait phase is the only free time in the section. On paper you could steal reading time from the previous item because the whole booklet was open. Here you cannot: the previous item is gone.


• Stop transcribing. Verbatim notes cost you the next sentence, and you can never replay it. Note positions instead: who said what, whether the second speaker agreed, what the number was. Two or three tokens per turn.


• When a screen holds two questions, look at both before the clip starts. They usually ask different things of the same audio — one about the speaker's main point, one about matching details — and you cannot listen for details you have not read yet.


• Answer inside the clip, not after it. For an O/X grid, mark each row the moment you hear it settled. The 풀이 window is for the one you are unsure about, not for all of them.


• When a question is gone, it is gone. Guess, and put your attention on the next 대기. The most expensive mistake in IBT listening is spending the next screen's wait phase mourning the last one.

Writing is typed, and the counter is watching

Two things paper never tested: how fast you produce Korean on a keyboard, and how you behave in front of a live character counter.

The 51-style task is a real 실용문 — a short practical notice — with two blanks marked ㉠ and ㉡. On paper you wrote a clause into each. On screen you type into two boxes, and each box carries a live n / max counter. The cap is hard rather than advisory: both typing and pasting stop at the limit. The longer essay task works the same way, one large text area with a counter that updates as you type, against the character range the prompt itself states.

The honest warning is the one the research participants raised themselves: typing speed. If you compose Korean at hunt-and-peck speed, the writing section is not a writing test for you — it is a typing test with a writing rubric attached. Two specific things to practise before exam day, because neither can be fixed on the day:

• Your keyboard layout. Know without looking where ㅃ, ㅉ, ㄸ, ㄲ, ㅆ live under shift, and where ㅒ and ㅖ are. A candidate who hesitates on double consonants loses a surprising amount of time in a timed section.


• How your IME commits a syllable. Korean input assembles 받침 as you type the next consonant, which means a fast typist who backspaces mid-syllable can destroy a whole block. Practise correcting yourself, not just typing cleanly.

And one habit to drop: writing long and then trimming. With a hard cap, the trimming never happens — the sentence simply stops.

In our demo, the writing item is a real 실용문 notice with ㉠ and ㉡ blanks, typed, with the counter. It is not auto-scored in the demo, and the result screen says so rather than pretending otherwise. It is there so you find out how the box behaves under your fingers, which is the part you cannot rehearse anywhere else.

The screen that surprises people most: the answer sheet

Ask anyone who has sat IBT what caught them off guard, and it is rarely a question type. It is the answer-sheet screen.

There is a button — All Questions, 전체문제 — that opens a grid of every number in the section with your chosen answer printed underneath each one. Unanswered cells are tinted so they stand out. It has two tabs, labelled with counts: 전체문제 : N개 and 안 푼 문제 : N개; the second filters the grid down to what you have skipped. Clicking a number jumps you straight to that question. And the popup closes itself after about thirty seconds, with the countdown visible — because leaving it open is costing you section time either way. The same grid appears again on the submission screen, and once you submit, your answers are final.

Why does this surprise people? Because on paper the OMR sheet lies under your hand for the whole section. You see the shape of your progress without deciding to look. On screen, that information exists only when you open a popup — so candidates who never open it discover four blank items with two minutes left, and candidates who open it constantly bleed thirty seconds at a time.

The habit that works: open it twice. Once at the halfway mark, on the 안 푼 문제 tab, to sweep up skipped items while there is still time to think. Once near the end, to confirm nothing is blank. Not once every five questions.

Sit the demo, and do it now rather than the night before

We built a free IBT demo, and it is deliberately small: nine questions — three listening, five reading, one writing — at /ibt/demo. No account and no signup; you land on it and the attempt starts.

The nine were not chosen to look impressive. Every one of them uses a format the paper test cannot deliver, because seeing those formats is the entire point:

• Reading (5) — one of each: the inline dropdown, choose-exactly-two, drag-to-order, sentence insertion, and an O/X grid.


• Listening (3) — a choose-exactly-two item, and then one screen that holds two questions on a single clip. The clip plays once.


• Writing (1) — the 실용문 notice with ㉠ and ㉡, typed, with the live counter. Not auto-scored, as the result screen tells you.

Everything except the writing item is really scored, and the timers, the screen release, the refused third click and the answer-sheet popup are the same code that a full-length attempt runs. Full-length mocks exist, but you cannot buy one yet — there is nothing being sold here today.

Ten minutes now is worth more than an hour the night before. The formats above are not hard. They are only unfamiliar, and unfamiliar is a problem with an expiry date.

Try the free IBT demo

---

New to the IBT format? The TOPIK IBT overview explains the exam, the question types and how the mock exams work.