High-risk draft
Prioritise decision-changing problems before local form.

Protect meaning first, repair one layer at a time, and verify that every edit is safer than the sentence it replaces.
For learners: Prioritise errors by communicative risk, audit complex language in five layers and verify that every repair preserves the source meaning.
For teachers: Risk triage leads to meaning protection, layered form auditing, accent-diverse diagnosis, change-log editing and a paraphrase-verified final retry.
สำหรับผู้เรียน: จัดลำดับข้อผิดพลาดตามความเสี่ยง ตรวจภาษาซับซ้อนเป็นห้าชั้น และยืนยันว่าทุกการแก้ยังรักษาความหมายต้นฉบับ
สำหรับครู: เริ่มจากจัดลำดับความเสี่ยง ปกป้องความหมาย ตรวจ form เป็นชั้น ฟังสองสำเนียง แก้พร้อม change log และยืนยันด้วยการ paraphrase ก่อน retry
By the end of this lesson, you can prioritise twelve errors by communicative risk, complete a five-layer audit, justify eight repairs, and deliver a verified near-error-free complex message.
จัดลำดับข้อผิดพลาด 12 จุดตามความเสี่ยงต่อความหมาย ตรวจภาษาเป็น 5 ชั้น อธิบายเหตุผลของการแก้ 8 จุด และส่งข้อความซับซ้อนที่ตรวจยืนยันแล้วว่าแทบไม่มีข้อผิดพลาดและไม่เปลี่ยนเจตนาเดิม
Near-error-free control is not random proofreading. A strong editor protects proposition, responsibility, time and stance before repairing architecture, local form and style—and then verifies that the revision has not created a new claim.
ภาษาไทย: การแก้ที่ดีไม่ได้เริ่มจากหาคำผิดทุกคำ แต่เริ่มจากจุดที่อาจเปลี่ยนข้อเท็จจริง ผู้รับผิดชอบ เวลา ระดับความมั่นใจ หรือการตัดสินใจ แล้วตรวจว่าประโยคใหม่ยังรักษาความหมายเดิม
Ask “What could change for the reader?” before “What grammar rule is broken?” Accept a repair only when the learner can name the risk, preserve the intended meaning and verify the result.
| Stage | Teacher move | Evidence |
|---|---|---|
| 1 · Triage | Rank twelve errors by communicative consequence. | Twelve risk decisions |
| 2 · Meaning | Compare intention, draft and repaired interpretation. | Four meaning-safe repairs |
| 3 · Form | Run separate architecture and local-form passes. | Eight justified repairs |
| 4 · Listen | Diagnose two spoken drafts across accents. | Two audit maps |
| 5 · Edit | Require a change log and source comparison. | One verified revision |
| 6 · Prove | Score MEAN/FORM/USE and target one unsafe edit. | Message + retry |
Correct the problem that could change a decision before the problem that only affects polish.
| Draft problem | Audit layer | Risk decision |
|---|---|---|
| The supplier definitely caused the delay, although the evidence is preliminary. | stance / evidence strength | high · may overstate responsibility |
| Only the regional manager approved the revised terms. | scope | high if only has the wrong target |
| When Arun spoke to Ben, he accepted the liability. | reference | high · actor is unclear |
| The update will be completed last Thursday. | timeline | high · completion time is impossible |
| The proposal, reviewed by the board will be circulated. | punctuation | medium · boundary is unclear |
| The findings is consistent. | agreement | low to medium · meaning remains recoverable |
| What with the delays, the audit recommends postponement. | register | medium · tone mismatches the document |
| The plan was revised to reduce costs that worried investors. | attachment | high · modifier target is unclear |
| We recommend to postpone the launch. | complement | low · local form |
| The team reviewed the evidence, the deadline remained. | clause architecture | medium · relationship is missing |
| The change is efficient, affordable and it reduces risk. | parallelism | low · comparison is uneven |
| This proves the claim. However, the sample is small. | cohesion / logic | high · conclusion contradicts limitation |
Do not reward speed alone. A learner who spots agreement but misses changed responsibility has not yet prioritised safely.
Compare the intended message with the draft before rewriting anything.
| Source intention | Unsafe draft | Meaning-safe repair |
|---|---|---|
| The evidence suggests, but does not prove, fraud. | The evidence proves fraud. | The evidence suggests possible fraud but does not establish it. |
| Mali—not Arun—accepted the revised condition. | When Mali spoke to Arun, she accepted it. | Mali accepted the revised condition after speaking to Arun. |
| Approval is required only above ฿500,000. | Only approval is required above ฿500,000. | Approval is required only for amounts above ฿500,000. |
| Review occurred before the decision. | The decision was made before the evidence had been reviewed. | The evidence had been reviewed before the decision was made. |
A safe repair preserves who did what, when it happened, how certain the claim is and what condition or scope applies. Improved style does not justify changed content.
If a revision sounds smoother, ask the learner to paraphrase it without looking at the source. Reverse any edit that strengthens, weakens or reallocates responsibility.
One pass, one layer, one stated reason.
| Pass | Check | Repair evidence |
|---|---|---|
| 1 · Clause hierarchy | main claim, support and qualification | one clear main-clause spine |
| 2 · Reference and attachment | every pronoun and modifier has one target | two ambiguity removals |
| 3 · Verb system | agreement, tense-aspect, modal and complement | two form repairs |
| 4 · Parallelism and cohesion | matched structures and logical links | two relationship repairs |
| 5 · Punctuation and register | boundaries, channel and reader relationship | one final proportionality decision |
Although the evidence was reviewed by two analysts who had raised separate concerns, the committee recommends delaying the launch. This would protect customers and maintain confidence. The regional team described the decision as cautious but did not oppose it.
Require the learner to point to the repaired relationship. “It sounds better” is not a sufficient change-log reason.
Apply the same audit to language that disappears after you hear it.
Prioritise decision-changing problems before local form.
Map each repair to meaning, architecture, form or register.
| Listening pass | Record | Verification question |
|---|---|---|
| 1 · Meaning | claim, actor, time, stance | What could change a decision? |
| 2 · Structure | clause hierarchy, reference, attachment | Where does the interpretation split? |
| 3 · Repair | exact edit and reason | Did the repair preserve the source? |
Accent variety tests transfer, not imitation. Pause only after the learner has recorded the proposition and risk order.
Every substantial edit needs a problem, a reason and a verification result.
| Change-log field | Record | Example |
|---|---|---|
| Risk | what the reader may misunderstand | preliminary evidence sounds conclusive |
| Source → revision | only the changed span | proves → may indicate |
| Audit layer | meaning, architecture, form, cohesion or register | stance / evidence strength |
| Reason | the rule or relationship repaired | match claim strength to evidence |
| Verification | paraphrase or source comparison | no new conclusion introduced |
Give the learner a source intention and constraints, not corrected sentences. Stop the task if editing begins to invent facts or remove necessary qualifications.
Prove that the final version is accurate, clear and faithful to its purpose.
Result: 0–2 Rebuild · 3–4 Developing · 5–6 C2 performance evidence.
Score the first independent version. Give one target only, then record whether the retry restores meaning, stabilises form or improves audience fit.
You can now prioritise errors by consequence, audit complex language in five layers, justify repairs and verify that a revision remains faithful to its intended meaning. Continue through the Dynamic Learning Graph below.
Return to Grammar LessonsControl idiomatic discourse patterns first
ควบคุมรูปแบบสำนวนตามหน้าที่ก่อน
Adapt the verified message for audiences
ปรับข้อความที่ตรวจแล้วให้เหมาะกับผู้ฟัง
Compare formal sentence architecture
เปรียบเทียบโครงสร้างประโยคทางการ
Transfer control to integrated C2 performance
นำการควบคุมไปใช้ในภารกิจ C2 แบบบูรณาการ