KLAT cheat sheet

KLAT listening passages are played once; the question is printed before audio starts, so scan the stem for the target (number, place, reason) before playback.

Signal words for the answer: 그러나/하지만/그런데 introduce contrast, so the correct choice often follows the second clause, not the first.

Numbers and times appear in pairs (e.g., 3시 30분 vs. 3시 13분); listen for the unit word after the numeral to disambiguate.

In dialogue questions asking '남자는 무엇을 할 것입니까?', the man's final utterance with -겠습니다/-(으)ㄹ게요 marks his intended action.

Request/offer items hinge on -아/어 주세요 (request) vs. -아/어 드릴게요 (offer); the subject particle 이/가 vs. object 을/를 identifies who acts.

For '다음 대화를 듣고 알 수 있는 것' items, eliminate options containing information not stated, even if plausible in real life.

Place-identification items rely on location vocabulary: 은행, 우체국, 병원, 약국, 도서관; a mention of '환전' points to 은행, '처방' to 병원.

Monologue (안내 방송) items test gist; the first and last sentences usually state the purpose and the required action.

For ( ) insertion items, test each candidate at the blank: the correct sentence must connect logically to both the preceding and following sentences via a conjunctive adverb.

Ordering items (글의 순서): find the sentence with no backward reference (no 이/그/저 + noun, no 그러므로) as the first; time adverbs like 먼저, 다음에, 마지막으로 fix sequence.

Main-idea items (중심 생각): the thesis usually appears in the first or last sentence; options that repeat a supporting example are distractors.

Inference items (알 수 있는 것): the answer must be derivable from stated content; reject options that require outside knowledge.

Vocabulary-in-context items: substitute each option into the underlined word and check collocation; 반드시 follows the sentence's logic, not the dictionary's first sense.

For title items (제목), the correct title names both topic and attitude; a title covering only the topic is incomplete.

Blank-filling with connectives: however (그러나) signals contrast, therefore (따라서) signals result; match the logical relation, not the sentence length.

Long-passage (장문) items: read the first sentence of each paragraph to map structure before answering detail questions.

Particle 이/가 marks the subject; 은/는 marks topic or contrast; in '누가 했어요?' the answer takes 이/가, not 은/는.

Object particle 을/를 follows consonant/vowel-final nouns respectively; 에게/한테 for people, 에 for places and times.

Verb endings: -고 싶다 (want), -(으)려고 하다 (intend), -기로 하다 (decide), -는 중이다 (in progress); match the intended meaning to the ending.

Honorific suffix -(으)시- inserts before tense/aspect: 가시겠습니다, not 가겠습니다, when the subject is honored.

Irregular verbs: ㄷ→ㄹ before vowel (듣다→들어요), ㅂ→우 (돕다→도와요), ㅅ drops (낫다→나아요), 르→ㄹㄹ (모르다→몰라요), 으 drops (쓰다→써요).

Causative -게 하다 / -시키다 and passive -이/히/리/기 change valency; check whether the subject performs or receives the action.

Vocabulary items test collocation: 약속을 지키다 (keep a promise), 감기에 걸리다 (catch a cold), 결정을 내리다 (make a decision).

Number classifiers: 개 (things), 명 (people), 마리 (animals), 권 (books), 장 (sheets), 잔 (cups); the classifier must match the noun.

Sentence-completion items: the blank must grammatically fit the preceding clause; match the ending form (-아/어서, -(으)니까, -는데) to the intended relation.

For '문장을 완성하십시오' items, one clause is given; the answer must include a verb ending consistent with the connective already present.

Paragraph-writing (작문) items are scored on task completion, coherence, and grammar; address every bullet in the prompt.

Use formal written register (합니다체/입니다체) in expository tasks; avoid spoken forms (해요체) unless the prompt specifies a letter or dialogue.

Cohesion devices: 그러므로, 따라서, 예를 들어, 반면에; each paragraph should open with a topic sentence and close with a summary or result.

Length requirements: count spaces as characters; Korean writing tasks typically specify 200-300자, so plan 3-4 sentences per paragraph.

Common penalty: mixing honorific subject marker with non-honorific verb; check that -(으)시- appears when the subject is a person being respected.

For data-description tasks, state the trend first (증가하다/감소하다), then cite the figures, then give the implication.

Speaking tasks are rated on task completion, fluency, grammar accuracy, and pronunciation; answer the prompt's question directly in the first sentence.

For picture-description tasks, use the present progressive -고 있다 and location words 앞/뒤/옆/위 to describe the scene in a fixed order (who, what, where).

Role-play items require a response matching the interlocutor's speech level; if the examiner uses 해요체, reply in 해요체 unless the prompt says otherwise.

Use fillers sparingly; prolonged silence lowers fluency scores more than a self-correction with 네, 다시 말하면.

For opinion tasks, structure: 입장 (I think) → 이유 (because) → 예시 (for example) → 정리 (so).

Pronunciation: distinguish plain, tense, and aspirated consonants (가/까/카, 다/따/타); final consonants (받침) must be released or neutralized per standard rules.

Connected-speech rules apply: ㄴ + ㄹ → ㄹㄹ (연락 → 열락), ㅂ + ㄴ → ㅁ (합니다 → 함니다); apply these to sound natural.

For '질문에 답하십시오' items, the answer must address the wh-word in the question (누구, 언제, 어디, 왜); omitting it loses task-completion points.