5.2 Writing Tasks and Scoring Criteria

Typical Menulis Task Types

UKBI Menulis prompts fall into five recurring families. The task type determines the expected structure, but every type is judged against the same five-dimension rubric, and every type requires bahasa baku throughout.

For the source summary, the rule of thumb is to reduce the source to roughly one-third of its word count while preserving every key proposition. Substituting tidak-baku synonyms during condensation is the most common summary-specific failure mode.

The Five Scoring Dimensions

Each response is graded against five dimensions. Kebakuan is weighted heaviest; the others carry roughly equal weight.

1. Kebakuan (standardness of language) — Does the response use bahasa baku consistently? Are there any ragam cakapan forms (aku, bilang, nggak, soalnya, banget)? Are pronouns, conjunctions, and verbs in their KBBI form?
2. Organisasi (organization) — Does the response have a clear opening, body, and closing? Are paragraphs unified around one topic sentence each? Is there a logical progression?
3. Kesesuaian isi (content relevance) — Does the response address every required element of the prompt? Is the content appropriate to the task type (a narrative recounts, an exposition explains, a letter requests)?
4. Tata bahasa dan PUEBI (grammar and PUEBI compliance) — Are sentences syntactically complete (subjek-predikat-objek)? Are affixes correct (menganalisis, not menanalisis; memperhatikan, not mengperhatikan)? Is punctuation PUEBI-compliant?
5. Kosa kata dan KBBI (vocabulary) — Are words KBBI-recognised? Are loanwords in their PUEBI-regularised form (kuitansi, kudeta, korespondensi)? Is the lexical choice precise rather than vague?

Common Failure Modes

Two failure modes account for the majority of band-capping outcomes:

• Tangential drift (menyimpang dari permintaan soal). The response opens on-topic, then migrates to a related but unrequested theme. For a prompt that asks you to describe the cause and propose a solution, writing two paragraphs of cause and one paragraph of opinion (without a solution) caps the content relevance dimension at Semenjana regardless of language quality.
• Tidak-baku leakage. A single nggak or banget inside an otherwise baku response signals ragam cakapan and depresses the kebakuan dimension disproportionately. The fix is mechanical: during the revise phase, scan for the high-frequency tidak-baku pairs and replace each with its KBBI equivalent.

A third, subtler failure mode is poor cohesion. Many responses are grammatically baku and on-topic but read as a list of unrelated sentences. In baku Indonesian, cohesion is carried by penanda hubungan (relational markers): oleh karena itu (therefore), dengan demikian (thus), sebaliknya (on the other hand), selain itu (besides), namun (however). Using them at the start of body paragraphs lifts the organisation dimension from Semenjana to Madya.

A Model Paragraph and a Weak Paragraph

Consider a prompt that asks: Jelaskan penyebab kemacetan lalu lintas di kota besar dan usulkan solusinya (Explain the cause of traffic congestion in large cities and propose a solution).

Model paragraph (Unggul band):

Kemacetan lalu lintas di kota besar disebabkan oleh ketidakseimbangan antara jumlah kendaraan dan kapasitas jalan. Pertumbuhan jumlah kendaraan bermotor tiga kali lebih cepat daripada pembangunan infrastruktur jalan, sehingga ruang gerak lalu lintas menyempit pada jam puncak. Selain itu, ketertiban pengguna jalan yang masih rendah memperburuk kondisi tersebut. Oleh karena itu, pemerintah kota perlu memperluas jaringan transportasi publik dan menegakkan disiplin pengguna jalan secara konsisten.

This paragraph is baku throughout, uses KBBI lexical choices (ketidakseimbangan, kendaraan bermotor, jam puncak), deploys sehingga and oleh karena itu as cohesion devices, and addresses both cause and solution.

Weak paragraph (Semenjana band):

Kemacetan di kota itu emangnya sangat parah banget soalnya banyak mobil dan motor yang lewat terus setiap hari. Pemerintah harusnya buat jalan baru biar nggak macet lagi dan rakyat juga jangan bawa mobil kalau nggak perlu.

This version fails on kebakuan (emangnya, banget, soalnya, harusnya, biar, nggak), fails on PUEBI mechanics (no cohesion devices, no clear paragraphing), and offers only a vague solution. Even though it nominally addresses both cause and solution, the kebakuan and PUEBI dimensions cap it at Semenjana.

Calibration Against the Predikat Bands

As a rough calibration: a response that is consistently baku, addresses the prompt fully, and uses at least two penanda hubungan typically reaches Madya (482–577). To reach Unggul (578–640), the response must in addition use precise KBBI vocabulary rather than general synonyms, vary sentence structure, and show no PUEBI errors. To reach Sangat Unggul (641–724), the response must additionally demonstrate rhetorical control — the closing must do more than restate the opening; it must synthesise or recommend. Istimewa (725–800) is reserved for responses that are baku, precise, PUEBI-perfect, rhetorically controlled, and stylistically varied throughout.

Strategy Checklist

Before submitting, run the response through this five-point check: (1) every prompt element addressed, (2) no ragam cakapan form remains, (3) every loanword in PUEBI form, (4) at least two penanda hubungan used, (5) every sentence has a clear subjek-predikat. Five yeses places the response reliably in the Madya band or higher.