UKBI (Uji Kemahiran Berbahasa Indonesia) common mistakes
UKBI (Uji Kemahiran Berbahasa Indonesia) · Understand the traps and how to avoid them.
In the Mendengarkan section, the candidate uses the pause before each recording to stare at the question stem only, and reads the four options for the first time after the audio has moved on, so options are evaluated from memory rather than against what was just heard.
School and university listening tests in Indonesia are often replayed or read aloud by a teacher, so candidates never build the habit of pre-reading options and predicting the vocabulary, numbers, or names they should be listening for.
What to do instead
Drill a fixed pre-listening routine: in the available seconds, scan all four options, underline the 2-3 content words that differ between them, and predict what kind of information (a time, a place, a reason, a refusal) will decide the answer. Practise on recorded Indonesian news and podcast clips with a printed option set until the routine takes under 10 seconds.
On inference and attitude items ('Apa maksud pembicara dengan kalimat tersebut?', 'Bagaimana sikap pembicara terhadap rencana itu?') the candidate chooses the option that recycles the words actually spoken, rather than the option that matches the speaker's implied position.
Word-matching is the fastest and most reinforced strategy from everyday test-taking, and literal repetition in a distractor feels like confirmation. Intonation, modality words, and hedges carry the real meaning but are not salient to an untrained ear.
What to do instead
Build a listening log of stance markers: seharusnya, semestinya, kalau saja, sayangnya, terpaksa, mau tidak mau, konon, katanya, justru. For each practice clip, write one sentence stating the speaker's attitude before looking at any options, then choose. If an option simply repeats the audio, treat it as suspect until the other three are eliminated.
In Merespons Kaidah, the candidate selects the option that sounds natural in everyday spoken Indonesian, e.g. accepting 'para hadirin sekalian', 'sangat penting sekali', 'di mana' used as a relative pronoun, or 'praktek' as spelling, because the sentence 'flows' well.
Nearly all daily exposure is to colloquial Jakarta-style Indonesian and social-media writing, whereas this section tests standard written Indonesian as codified by Badan Bahasa. Fluency intuition and correctness criteria point in different directions here.
What to do instead
Maintain a two-column error bank (bentuk tidak baku / bentuk baku) and add every item you miss: praktik, analisis, apotek, nomor, Rabu, objek, metode, aktif, foto, izin, silakan. Review it weekly and, before answering, ask 'is this the standard written form or the spoken form I hear?'
The candidate misidentifies what the item is asking, most commonly missing a negative instruction ('Kalimat yang tidak efektif adalah', 'Pilihan yang tidak tepat') and answering the opposite question, or picking a grammatically flawless option on an item that asks for the effective or the appropriate one.
Instruction stems are short, formulaic, and visually similar across items, so under time pressure the eye registers 'efektif' and skips 'tidak'. Candidates also conflate 'grammatically correct' with 'effective' or 'appropriate to context', which are separate criteria in this section.
What to do instead
Circle or say aloud the operative word in the instruction ('tidak', 'paling tepat', 'tidak baku', 'merujuk pada') before reading option A. Keep a written definition of kalimat efektif (no redundant pairs, parallel structure, clear subject-predicate, no shifting subject, logical) and test each option against the named criterion only.
On Berbicara prompts the candidate answers with a single short sentence ('Ya, saya setuju, karena itu penting.') and stops, or delivers a memorised self-introduction that does not answer the prompt actually asked.
In everyday Indonesian conversation, a brief reply is polite and sufficient, and candidates transfer that norm. Some also rehearse one prepared text because they fear making errors live, so a surprise topic derails them.
What to do instead
Train a fixed three-part answer shape for every prompt: state a position, give one reason, give one concrete example or personal experience, then close with a recommendation - roughly 60-90 seconds. Record yourself on 20 unseen prompts and listen back for length and relevance, not just grammar.
During Berbicara, the candidate fills every hesitation with drawn-out 'eee...', 'apa ya...', repeated self-correction ('bukan, maksud saya...'), or switches into English mid-sentence ('so basically...'), and reads notes aloud in a flat monotone.
Silence feels like failure to an anxious speaker, so the mouth keeps moving while the brain searches. Code-switching happens because technical or abstract vocabulary is only available in English, and reading notes aloud removes the intonation that raters listen for.
What to do instead
Replace vocal fillers with a silent two-second pause or a deliberate Indonesian connector (yang pertama, selain itu, oleh karena itu, maksud saya). Prepare topic vocabulary lists for common UKBI themes (pendidikan, lingkungan, teknologi, budaya, kesehatan). Use keyword cards, not full scripts, and rehearse standing up so stress and intonation carry.
In Membaca, the candidate reads each passage from beginning to end at comfortable speed before looking at any question, then runs out of time and guesses the last several items.
It feels like the safe, thorough approach taught at school, and skipping to questions feels like skipping the reading. Indonesian academic passages are long enough that full linear reading cannot fit the section's time budget.
What to do instead
Reverse the order: read the questions first, then scan the passage for the paragraph that answers each one. Time yourself at roughly one minute per item; if an item exceeds 90 seconds, mark a provisional answer and move on. Do two full timed simulations to prove to yourself that question-first is both faster and more accurate.
On Membaca items the candidate picks a statement that is true according to the passage but does not answer the specific question asked - choosing a supporting detail on a 'gagasan utama' item, or a plausible inference on a 'pernyataan yang sesuai dengan bacaan' item, or ignoring the antecedent on a 'kata tersebut merujuk pada' item.
Every option is drawn from passage content, so all four feel defensible; without a decision rule the candidate selects on familiarity of wording rather than on fit to the question type.
What to do instead
Label the item type in the margin before answering: main idea, detail, inference, purpose, reference, tone. Apply the matching test - main idea must cover the whole paragraph, detail must be explicitly stated, inference must be entailed and not merely plausible, reference must be the nearest grammatically compatible noun phrase. Eliminate any option that is true but off-question.
In Menulis, the candidate produces one unbroken block of sentences with no paragraphing, mixes spoken forms and abbreviations ('nggak', 'banget', 'sbg', 'dgn', 'yth'), and ignores the register implied by the prompt, writing a formal letter of complaint to an institution in a chatty personal tone.
Writing practice is usually informal messaging, so the default register is conversational. Many candidates also write straight through without planning, which produces structureless text and mechanics errors they never re-read for.
What to do instead
Before writing, spend three minutes noting audience, purpose, and genre, then commit to a skeleton (pembuka - isi dua atau tiga paragraf - penutup) and a target length. Write only baku forms; if the correct word will not come, rephrase rather than code-switch. Reserve the final three minutes for a fixed proofreading checklist: capitalization, commas before tetapi/melainkan, di- as prefix versus di as preposition, hyphenation of reduplication, and no spoken forms.
In Menulis, the candidate misreads what the task requires - writing a general opinion essay when asked for a letter to a specific recipient, or a personal narrative when asked to argue a position - and ignores stated constraints such as word count, required points to include, or the data that must be described.
The prompt is read once quickly and the candidate starts writing on the familiar topic rather than the specified task, because content feels easier than task analysis and there is no time budgeted for planning.
What to do instead
Underline the imperative verb, the addressee, and every required content point in the prompt before writing a word. Check the response against that underlined list at the end. Practise converting one topic into three different tasks (a letter, an argument, a description of a chart) to make the distinction between topic and task automatic.