Evidence Briefing · 2026
Why Chinese candidates plateau at 6.5. 证据简报 · 中国考生为何卡在 6.5
A practitioner's catalogue of first-language transfer patterns: what each one is, which assessment criterion it costs marks under, and why the deepest layer is the one almost no course teaches. Download as PDF →
The Claim 核心观点
The ceiling is structural, not linguistic.卡住你的是结构，不是英语水平。
A candidate who reaches 6.5 has already proved they can communicate in English. What stops them at 7 is a small, predictable set of patterns produced by the distance between Mandarin and English — and those patterns are teachable.能考到 6.5，说明沟通能力已经过关。真正卡住 7 分的，是中英文结构差异带来的一小组可预测的固定问题 —— 而这些是可以教会的。
Three findings · 三个结论
| It is not ability 不是能力问题 |
None of the recurring errors below reflect intelligence, effort or vocabulary size. They are the predictable consequences of how the first language encodes tense, number, definiteness and argument order. |
| The errors repeat 错误高度重复 |
Across a decade of marking, the same short list recurs. Most candidates carry three or four of them, not all of them — which is why a personal diagnosis beats a generic course. |
| The deepest layer is rhetorical 最深的一层是修辞 |
Vocabulary and grammar are the visible layers, and the ones every course addresses. The layer that actually caps Coherence & Cohesion is the order in which an argument is delivered — and it is almost never taught. |
Scope. This briefing covers Speaking and Academic Writing — the two productive papers, where transfer is visible to an examiner. Listening and Reading are not addressed here.
Part One · Speaking 第一部分 · 口语
First-language transfer operates at four layers.母语迁移有四个层次。
Most preparation material addresses only the third. Each layer below is followed by the criterion it costs marks under.大多数备考资料只讲第三层。以下每一层都标注了它实际扣分的评分标准。
Layer 1 · Sound 语音层
| Mechanism | 他 · 她 · 它 are written differently but pronounced identically. Under time pressure the tongue reaches for whichever form arrives first. 他、她、它 写法不同，读音相同。压力之下，先想到哪个就说哪个。 |
| What it sounds like | “My mother — he works in a hospital…” |
| Costs marks under | GRA Grammatical Range & Accuracy — and it is usually made repeatedly without the speaker noticing. |
Layer 2 · Grammar 语法层
| Mechanism | Mandarin links a subject straight to an adjective, with no copula. Carried into English, the verb be disappears. Mandarin verbs also do not inflect for time — 我去 and 我去了 differ by a particle, not by the verb. 中文「主语 + 形容词」不需要动词，搬到英文就漏掉 be。中文动词也不随时态变形。 |
| What it looks like | “She very tired.” · “I interested in music.” · “Last year the figure rise sharply.” |
| Costs marks under | GRA This is the single most common accuracy error for Chinese candidates in both papers. |
Layer 3 · Lexis 词汇层
| Mechanism | Chinese collocations translated word for word. The meaning survives; the English does not. This is the layer nearly every course teaches — and the easiest of the four to fix. 把中文搭配逐字翻译成英文。意思在，英文不在。这是大多数课程唯一会讲的一层，也是最容易改的一层。 |
| What it looks like | “open the light” · “play my phone” · “learn knowledge” · “I very like it” |
| Costs marks under | LR Lexical Resource rewards natural collocation, not vocabulary size. A rare word in an unnatural pairing scores worse than a common word used correctly. |
Layer 4 · Rhetoric 修辞层 — the layer nobody teaches
| Mechanism | Classical Chinese rhetoric builds toward its conclusion — 起承转合 — placing the point at the end. English academic discourse inverts this: the point comes first, then the reasoning, then the evidence. Nothing about this is a language error. Every sentence can be grammatically perfect and the answer still reads as unfocused. 中文修辞讲究起承转合，观点收在最后；英文学术表达正好相反 —— 先给观点，再讲理由和证据。这不是语法问题，句句正确也可能被判为「不聚焦」。 |
| What it looks like | Two sentences of background before any position is stated · a Part 3 answer that never lands an opinion · circling back to the same idea instead of developing it · expecting the listener to infer the point. |
| Costs marks under | CC TR Coherence & Cohesion, and in Task 2 opinion essays, Task Response — because a position that is never stated cannot be assessed as maintained. |
The ranking that matters. One Layer 4 problem outweighs several at Layer 3. Coherence is assessed on whether the examiner can follow the argument — not on how many words the candidate knows. 第四层出现一个问题，比第三层出现几个更严重。
Part Two · Academic Writing 第二部分 · 学术写作
Seven fixes, mapped to the criteria.七项高影响修正，对应四大评分标准。
The same transfer patterns surface in Writing, where they are permanent on the page rather than fleeting in speech. Each fix below names the criterion it protects.同样的迁移问题在写作中会白纸黑字留下来。每一项都标注了它保护的评分标准。
The seven highest-impact fixes · 七项高影响修正
| # | Fix | Criterion |
|---|---|---|
| 1 | Mark every verb for tense and subject agreement 每个动词都标时态与主谓一致 | GRA |
| 2 | Add articles (a / an / the) before singular countable nouns 单数可数名词前补冠词 | GRA |
| 3 | Never pluralise uncountable nouns (research, information, advice) 不可数名词不加 s | LR GRA |
| 4 | Break comma splices with a conjunction or semicolon 拆开逗号连句 | GRA |
| 5 | State the point first, then explain (P-E-E-C order) 先给观点，再解释 | CC |
| 6 | Replace translated idioms and 成语 with natural collocations 成语直译换成地道搭配 | LR |
| 7 | Commit to and defend a clear position 明确表态并论证到底 | TR |
Most candidates carry three or four of these repeatedly, not all seven. Identifying the personal pattern is what turns a generic fix list into an actual gain.
Worked evidence · 实例对照
| Transfer error | Corrected | Why |
|---|---|---|
| The graph show that sales increase in 2010. | The graph shows that sales increased in 2010. | Past data takes past simple; show needs third-person -s. |
| Government must do more researches. | The government must do more research. | Shared reference takes the; research is uncountable. |
| Pollution is serious, we must act. | Pollution is serious, so we must act. | Comma splice — two full clauses need a conjunction or semicolon. |
| People mountain people sea (人山人海) | large crowds / densely populated | Translate the meaning, not the image. |
| Both views have their own advantages and disadvantages. | While public transport has clear benefits, I firmly believe private cars remain essential for rural communities. | An opinion task requires a stated, defended position. “It depends” gives the examiner nothing to reward. |
On memorised openers. “With the development of society…”, “Every coin has two sides” and “As is known to all…” are drilled so widely that they are recognised on sight. They do not merely fail to add marks — they signal that what follows may be rehearsed.
Part Three · Implication 第三部分 · 意义
Why global materials cannot close this gap.为什么面向全球考生的教材解决不了这个问题。
| The arithmetic | A textbook written for a worldwide readership must serve candidates whose first languages already mark tense, already use articles, and already front their arguments. Space spent on article use is wasted on a Spanish speaker; space spent on 起承转合 is meaningless to anyone outside the Sinosphere. The result is material that is correct for everyone and decisive for no one. |
| What follows | The fixes in this briefing are narrow by design. They are not a syllabus — they are the short list that separates a competent 6.5 from a 7 for this specific population. A candidate who resolves their own three or four has removed the most common reasons marks are lost, without studying anything a global course would have prioritised. |
| What it does not claim | Fixing these patterns does not produce a band score. Results depend on starting level, test-day performance and examiner variation, and no material can promise otherwise. What a fix list can honestly offer is the removal of predictable, self-inflicted deductions. |
How this was compiled 编制方法说明
What this is — and what it isn't.
Method note · 说明
| Source | A practitioner's catalogue, assembled across roughly ten years of one-to-one teaching, mock interviews and marked scripts with Chinese candidates in Beijing, and set out in full in the Deen Cox IELTS Series. |
| Status of the claims | This is professional judgement, not measurement. No frequencies are quoted because none were formally counted; “most common” here means most common in one teacher's caseload. It is offered as a practitioner's observation that other teachers can test against their own students — not as research. |
| On the descriptors | The assessment criteria are referred to by name and paraphrased. The official band descriptors are not reproduced here; they are published by the IELTS partners and should be read at source. |
| Credentials | Certified by IDP Education in Understanding the Assessment Criteria for both Speaking and Writing. IDP is one of the three organisations that jointly own IELTS. This is a teacher-training certificate — IELTS has not endorsed this briefing, the books, or the classes. Certificate numbers → |
Corrections welcome. If you teach this population and your experience differs, say so — get in touch. A catalogue that cannot be challenged is not evidence.
Take it further 继续深入
Diagnose your own pattern.找出你自己的那三四个问题。
Both tools below are free, need no sign-up, and come from the chapters this briefing draws on.以下两份工具均可直接下载，无需注册。
| The L1 Transfer Self-Check 母语迁移自查表 | A four-layer recording audit. Record three Part 1 answers and one Part 2 response, play them back, and tick what you actually hear — not what you remember saying. |
| The Chinglish Swap Table 中式英语替换表 | Direct-translation traps, memorised openers, and confusable pairs, each with a natural replacement. |
| This briefing as a PDF 本简报 PDF 版 | The same material as a four-page A4 document you can print, hand out in class, or forward. Free, no sign-up. |
| The full method 完整方法论 | The fourteen frameworks that replace memorised templates with sequences a candidate can run under exam pressure. |
The patterns in Part One are set out in Chapter 12 of IELTS Speaking Mastery: The Complete Band 7+ Guide; the seven fixes in Part Two open the Writing guide. Browse the series →