Series ID: EDKS-ADV-ZH-0004 · How to Learn Advanced English (Chinese Edition) · Lesson No.004 · C1 → C2

Read Deeply, Think Clearly, Write Precisely | 深度阅读、清晰思考、精准写作

Advanced English is not only a language problem. At the highest levels, it becomes a reading problem, a reasoning problem and a writing problem at the same time.

高级英语，不只是语言问题。到了最高阶段，它会同时变成阅读问题、推理问题和写作问题。

You can know nearly every word in an article and still misunderstand the argument. You can understand the argument and still fail to explain it. You can explain it and still fail to evaluate the evidence. You can evaluate the evidence and still write a paragraph that leaves the reader unsure what you actually believe.

你可能认识一篇文章里几乎所有单词，却仍然读错论证；你可能理解论证，却讲不清；你可能讲得清，却不会判断证据；你可能会判断证据，却写出一段让读者看不出你真正立场的文字。

This is why advanced English cannot be reduced to vocabulary, grammar or fluency. At C1–C2, language becomes a vehicle for complex thinking. The learner has to track claims, evidence, assumptions, qualifications, causation, contrast, stance, implication, source relationships and the reader’s path through an argument.

所以，高级英语不能被压缩成“词汇 + 语法 + 流利度”。到了 C1–C2，语言成为复杂思考的载体。学习者要跟踪主张、证据、假设、限制条件、因果、对比、立场、隐含意义、多个来源之间的关系，以及读者怎样一步一步走过你的论证。

This lesson builds that system.

这一课，就建立这套系统。

Before we begin: this is an integration lesson, not a duplicate owner | 开始之前：这一课是“整合层”，不是重复主页面

eduKate Sengkang already owns specialist mechanisms for argumentative writing, counterarguments, logical fallacies, synthesis, cohesion, paragraph architecture, writing progression, reading evidence and school-stage comprehension. This lesson does not replace them.

eduKate Sengkang 已经有独立主页面负责论证写作、反方观点、逻辑谬误、综合、衔接、段落结构、写作进阶、证据阅读和学校阶段阅读理解。本课不重复这些内容。

Its job is different: it gives Mandarin-speaking advanced learners one C1–C2 operating system that connects reading → reasoning → synthesis → writing → revision.

它负责的是更高一层：给中文母语高级学习者建立一套统一的 C1–C2 操作系统，把阅读 → 推理 → 综合 → 写作 → 修改连接起来。

How Argumentative Writing Builds Claim, Evidence and Judgement — argument mechanism owner.
How Counterarguments and Rebuttals Strengthen Analytical Writing — counterargument owner.
How Students Recognise Logical Fallacies — fallacy mechanism owner.
How Students Synthesize Evidence Into a Larger Meaning — synthesis owner.
How Writing Works for a Student — general writing system.
How Writing Moves From Sentence to Paragraph to Essay — writing architecture owner.
Lesson No.002 — register and precision.
Lesson No.003 — Mandarin→English reconstruction.

1. C1–C2 reading is not “understand the words” | C1–C2 阅读不是“看懂单词”

Cambridge C1 Advanced reading explicitly tests more than literal comprehension. Its Reading and Use of English tasks include detail, opinion, tone, purpose, main idea, implication, attitude, comparison of viewpoints across texts and understanding how a text develops structurally.

剑桥 C1 Advanced 的阅读要求远远超过“看懂单词”。官方考试说明明确包含：细节、观点、语气、目的、主旨、隐含意义、态度、跨文本观点比较，以及理解整篇文章怎样发展。

The CEFR C1–C2 descriptors point in the same direction. Advanced readers are expected to handle long, complex texts, implicit meaning, nuanced attitudes, stylistic distinctions and the reconstruction of ideas across sources.

CEFR C1–C2 也指向同一个方向：高级阅读者要处理长而复杂的文本、隐含意义、细微态度、风格差别，以及跨来源重建思想。

So ask a better question after reading:

Not: “What did the article say?”

But: “What did the article claim, how did it support the claim, how strong was the support, what did it leave unstated, and what should a careful reader conclude?”

读完以后，不要只问“文章说了什么”。要问：

它主张什么？怎么支持？支持有多强？什么没有明说？一个谨慎读者应该得出什么结论？

2. Reading has layers | 深度阅读是分层的

Layer	Reader’s question	中文
Literal	What does the text explicitly say?	明说了什么？
Structural	How is the text organised?	文章怎样组织？
Inferential	What follows from what is stated?	能合理推出什么？
Rhetorical	What is the writer trying to make the reader believe, feel or do?	作者想让读者相信、感受或行动什么？
Evaluative	How strong is the reasoning and evidence?	推理和证据有多强？
Comparative	How does this source relate to other sources?	它和其他来源是什么关系？
Synthetic	What larger meaning emerges when sources are combined?	多个来源组合后出现什么更大的意义？

Many advanced learners stop at Layer 1 or Layer 2. They understand the prose but do not yet interrogate the argument.

很多高级学习者会停在第一层或第二层：文字看懂了，但没有真正审问论证。

3. Read the claim first | 先找主张，不要先抓生词

In serious nonfiction, the claim is the proposition the writer wants you to accept. It may be stated directly or distributed across the text.

严肃非虚构写作里，claim 是作者希望你接受的主张。它可能直接写出来，也可能分散在文章各处。

Ask:

What is the main claim?
What smaller claims support it?
Which claims are descriptive and which are evaluative?
Which claim is causal?
Which claim is a recommendation rather than a fact?

先把“事实陈述、解释、因果、评价、建议”分开。很多阅读误判，来自把不同类型的句子全部当成同一种“信息”。

4. Claim type controls evidence type | 主张类型决定需要什么证据

Claim type	Example	What support is needed?
Descriptive	Usage increased.	measurement / observation
Causal	The policy caused the increase.	causal evidence, alternatives controlled
Predictive	The trend will continue.	model, historical pattern, assumptions
Evaluative	The policy is effective.	criteria + outcome evidence
Normative	The government should expand it.	values, trade-offs, feasibility, evidence
Definitional	This counts as discrimination.	definition + application to facts

A common weak-reading error is to accept evidence that supports a description as if it proves causation.

常见弱阅读错误之一，是把“证明发生了什么”的证据，当成“证明为什么发生”的证据。

5. Evidence is not one thing | 证据不是一种东西

Advanced reading improves when you classify evidence instead of merely noticing that “there is evidence”.

data
experiment
observational study
case study
expert testimony
document / primary source
example
analogy
quotation
personal experience
mechanistic explanation

Different evidence has different strengths for different claims. A vivid example may explain an idea but cannot establish population prevalence. A correlation can reveal association but does not by itself establish causation. Expert opinion can guide interpretation but should not replace underlying evidence when the evidence is available.

不同类型证据适合支持不同主张。一个生动例子可以解释，但不能证明普遍比例；相关关系可以说明一起变化，却不能自动证明因果；专家意见可以帮助解释，但在原始证据可获得时，不应该完全取代证据。

6. Build the evidence ladder | 建立证据阶梯

When reading a claim, use five questions:

Relevance: Does the evidence actually address the claim?
Quality: How reliable is the source or method?
Quantity: Is one example being treated as a pattern?
Consistency: Does other evidence point the same way?
Scope: Does the conclusion stay within what the evidence can support?

高级阅读不是“有引用 = 可信”。你要检查相关性、质量、数量、一致性和范围。

7. Separate observation from interpretation | 把观察和解释分开

Consider:

Observation: Attendance fell by 12% after the timetable changed.

Interpretation: The new timetable discouraged students from attending.

Alternative interpretation: Attendance fell because the timetable change coincided with examination preparation.

The observation can be true while the explanation remains uncertain.

一个事实可以确定，而解释仍然不确定。高级阅读必须把这两层分开。

8. Read causation carefully | 因果关系要慢下来读

Ask four questions whenever you see a causal claim:

Did the proposed cause happen before the effect?
Do cause and effect reliably vary together?
Could another factor explain both?
Is there a plausible mechanism?

Do not turn these into a rigid checklist for every scientific question. Use them as a discipline against jumping from “A and B appear together” to “A caused B”.

不要把它们变成机械公式，但要用它们阻止自己从“同时出现”直接跳到“因果成立”。

9. Read qualifications as carefully as conclusions | 限制条件要和结论一样认真读

Words such as may, tends to, appears to, in some cases, under certain conditions, within this sample, is associated with are not filler. They define what the writer is and is not claiming.

may、tends to、appears to、in some cases、under certain conditions、within this sample、is associated with 不是废话。它们定义了作者到底在主张什么，也定义了作者没有主张什么。

When summarising, preserve the qualification.

Original: The intervention may improve retention for some learners.
Bad summary: The intervention improves retention.
Better: The study suggests that the intervention may improve retention for some learners.

高级 summary 不能为了简洁把“可能、部分人、特定条件”全部删掉。

10. Inference must be bounded | 推断必须有边界

An inference is not a guess. It is a conclusion supported by textual evidence plus reasonable background knowledge.

推断不是“猜”。它是文本证据 + 合理背景知识支持下的结论。

Use a three-column test:

Evidence in text	Background knowledge	Safe inference
The manager reread the message three times before replying.	People often reread when a message is difficult, sensitive or surprising.	The message probably required careful interpretation.

Do not infer more than the evidence permits. “The manager was angry” may be possible, but it is not necessarily supported.

11. Implication is not the same as hidden fact | 隐含意义不等于“隐藏事实”

Advanced reading often requires you to infer stance or implication without inventing unstated events.

Example:

“The proposal is ambitious, although the funding assumptions remain optimistic.”

Possible implication: the writer has reservations about feasibility.

Unsafe invention: the proposal will definitely fail.

高级阅读要识别态度，但不能把态度扩大成未经支持的事实。

12. Read stance through verbs | 通过 reporting verbs 读立场

Verb	Typical stance effect
states	relatively neutral reporting
argues	presents a reasoned position
claims	can be neutral but may create distance depending on context
suggests	lower commitment / inference
demonstrates	strong evidential framing
admits / concedes	signals reluctant acceptance or concession
alleges	signals that truth is not established

中文翻译可能都变成“说、认为、指出”，但英语 reporting verb 会传递作者和来源之间的距离。

13. Read adjectives and adverbs for evaluation | 形容词和副词里藏着评价

Compare:

a proposal
an ambitious proposal
an overly ambitious proposal
a surprisingly modest proposal

The factual noun remains similar, but evaluation changes sharply.

高级阅读要学会把“事件内容”和“作者评价”分开。

14. Read what is selected—and what is omitted | 读作者选了什么，也读作者没有选什么

Every text selects. A report may focus on cost and omit distribution. A company statement may highlight benefits and place limitations near the end. A political speech may foreground one comparison period and ignore another. A review may discuss performance but not reliability.

任何文本都在选择。报告可能强调成本，却不谈分配；公司声明可能把好处放前面，把限制放最后；演讲可能选择某个时间区间比较，却避开另一段；产品评论可能谈性能，却不谈可靠性。

Ask:

What information receives the most space?
What appears only briefly?
What relevant alternative is absent?
Whose perspective is represented?
Whose perspective is missing?

This is not an invitation to assume manipulation. It is a method for reading framing.

这不是叫你“什么都怀疑”。这是帮助你识别 framing（框架选择）。

15. Bias is not “I disagree with the writer” | Bias 不等于“我不同意作者”

A text can support a position you agree with and still use weak reasoning. A text can support a position you dislike and still use strong evidence.

一篇文章即使支持你喜欢的观点，也可能推理很弱；一篇文章即使支持你不喜欢的观点，也可能证据很强。

Evaluate process before conclusion.

先评估推理过程，再评估你喜欢不喜欢结论。

For deeper school-stage treatment, route to Secondary 3 English Learning Guide | Bias, Framing and Representation.

16. Read arguments as maps | 把论证读成地图

A useful argument map contains:

Main claim
Reason 1
Evidence 1
Reason 2
Evidence 2
Counterargument
Response
Qualification
Conclusion

If you cannot map the argument after reading, you probably do not yet control it well enough to write about it.

如果读完以后你画不出论证地图，你通常还没有真正控制这篇文章。

17. The “because test” exposes weak reasoning | because 测试可以暴露断裂推理

Take any conclusion and insert “because”.

The policy should be expanded because…

Now force yourself to state the actual reason. Then ask whether the evidence supports that reason.

很多模糊论证会在 because 后面暴露：你原来只是有立场，但没有完整推理链。

18. The “therefore test” checks whether evidence reaches the claim | therefore 测试检查证据是否真的够到结论

Evidence:

Three schools reported improved attendance after introducing breakfast programmes.

Weak leap:

Therefore, every school should introduce the same programme.

The evidence may justify investigation or a narrower recommendation, but not necessarily universal adoption.

高级写作必须检查“证据到结论之间有没有跳太远”。

19. The “what else could explain this?” test | “还有什么解释？”测试

Whenever a writer offers one explanation, generate at least one plausible alternative.

If a company’s sales increased after a campaign:

campaign effect?
seasonal demand?
price change?
competitor shortage?
distribution expansion?

You do not need to prove the alternative. Generating it prevents premature closure.

你不需要证明替代解释正确。你只需要防止自己太早停止思考。

20. Counterarguments are not decorative | 反方观点不是“为了显得客观”加一段

A real counterargument must challenge something central: a premise, a causal link, evidence quality, feasibility, trade-off or underlying value.

真正的 counterargument 必须攻击论证的重要部分：前提、因果链、证据质量、可行性、代价或价值基础。

Weak:

Some people disagree.

Stronger:

The proposal may reduce waiting time, but it could shift costs onto smaller clinics that lack the staff to absorb additional demand.

For the full mechanism, route to How Counterarguments and Rebuttals Strengthen Analytical Writing.

21. A rebuttal should answer the strongest version | Rebuttal 要回答最强版本，不要打稻草人

If the opposing argument says “The policy is unaffordable because implementation costs will exceed the projected savings for five years,” do not rebut it with “Some people do not like change.” That answers a weaker argument.

高级论证要回答对方真正最强的理由，而不是制造一个容易打倒的版本。

22. Fallacy labels are not substitutes for reasoning | 贴“谬误标签”不能替代理解

Calling something a slippery slope, ad hominem or false dilemma is useful only if you can explain the structural problem.

说“这是 slippery slope / ad hominem / false dilemma”只有在你能解释结构问题时才有价值。

Ask:

What premise is missing?
What alternative is excluded?
What causal step is unsupported?
What irrelevant personal feature is replacing argument?

Use How Students Recognise Logical Fallacies Without Reducing Every Argument to a Label for the canonical mechanism.

23. Deep reading includes paragraph function | 深度阅读要读每一段“在做什么”

Do not summarise every paragraph only by topic. Identify function.

Paragraph	Topic summary	Function summary
1	AI in education	frames the problem
2	teacher workload	provides evidence for benefit
3	accuracy risks	introduces a constraint
4	policy safeguards	proposes conditions for adoption

Function summaries reveal the argument’s architecture. Topic summaries alone often produce a list rather than a map.

24. Read transitions as logic signals | 转折和连接词是逻辑信号

However, nevertheless, by contrast, therefore, indeed, for example, in other words, admittedly, more importantly—these words show how one unit is supposed to relate to another.

但不要只背连接词。高级阅读要问：如果把连接词删掉，我能不能自己说出逻辑关系？

If not, you may be reading the surface rather than the reasoning.

25. Gapped-text thinking is useful beyond exams | “段落放哪里”训练可以迁移到真实阅读

Cambridge C1’s gapped-text task asks readers to understand structure and development, not merely vocabulary. That is a useful general training idea.

Take a good article. Remove one paragraph. Ask:

What must come before this paragraph?
What does its first pronoun refer to?
Does it continue, contrast, exemplify or qualify?
What does the next paragraph assume has already been established?

这个练习会训练你读 cohesion 和 discourse structure，而不是只读句子。

26. Comparative reading: do not summarise Source A, then Source B, then stop | 多文本阅读不能只做 A 总结 + B 总结

Synthesis begins when you organise sources around relationships rather than around source order.

弱综合：

Source A says… Source B says… Source C says…

更强的综合：

Sources A and B agree that X, but they differ on the mechanism. Source C questions the assumption shared by both.

The second version builds a model of the conversation among sources.

第二种写法不是“把三篇文章排队”，而是重建三篇文章之间的关系。

27. Build a synthesis matrix | 建立综合矩阵

Question	Source A	Source B	Source C
Main claim	supports policy	supports with conditions	opposes
Main evidence	survey	case studies	cost model
Limitation	small sample	local context	assumptions uncertain
Shared ground	current system has a capacity problem
Main disagreement	whether expansion solves or redistributes the problem

A synthesis matrix prevents you from writing three separate mini-reports.

综合矩阵帮助你从“来源顺序”切换到“问题顺序”。

28. Source hierarchy: not every source deserves equal weight | 不同来源不能自动等权

When comparing sources, consider:

method quality
sample size and relevance
date
directness of evidence
conflict of interest
domain expertise
replication / corroboration
fit to your exact question

“三个来源支持，一个来源反对”不是自动 3:1。来源质量和问题匹配度比数量更重要。

29. Summarising is compression with meaning preservation | 总结是“压缩但不改意思”

A strong summary preserves:

main claim
major support
essential qualification
important contrast
relevant conclusion

It removes:

repetition
minor examples
decorative detail
secondary digressions

但最危险的不是“漏掉例子”，而是压缩时改变了论断强度。

Route to Secondary 3 English Learning Guide | Summary, Paraphrase, Synthesis and Compression for the stage-specific owner.

30. Paraphrase means structure change plus meaning control | 改写不是“同义词替换”

Original:

Although the programme reduced waiting times, its long-term effect on service quality remains uncertain.

Weak synonym swap:

Although the scheme decreased waiting periods, its long-term influence on service quality stays unsure.

Better paraphrase:

The programme appears to have shortened waiting times, but the available evidence does not yet show whether service quality will improve over the longer term.

The better version changes clause structure and wording while preserving the key contrast and uncertainty.

强改写保护的是“逻辑关系和证据强度”，不是原句表面。

31. Synthesis writing is a reading test before it is a writing test | 综合写作首先是阅读测试

If you have not distinguished what each source actually claims, you cannot synthesise accurately.

So before drafting, write one sentence for each source:

Source X argues ___ because ___, but limits the claim by ___.

如果你无法完成这句话，先不要急着写 synthesis essay。

32. Research on synthesis writing supports model observation and feedback | 综合写作需要看范本，也需要过程反馈

Writing research has shown that learning to write synthesis texts can improve when learners observe models and receive process-focused feedback. A 2024 study in Reading and Writing examined the effect of process feedback and observing models on synthesis performance and writing behaviour.

写作研究也支持一个重要原则：综合写作不能只靠“多写”。观察高质量范本、比较过程、获得针对过程的反馈，可以帮助学习者理解“优秀综合是怎样建出来的”。

For advanced learners, model observation should ask:

How are sources grouped?
Where does the writer compare rather than list?
How is disagreement represented fairly?
How are source claims qualified?
Where does the writer add their own judgement?

33. Writing begins before the first sentence | 写作在第一句话之前就开始

Weak writers often begin drafting immediately because writing feels like sentence production. Strong advanced writing usually begins with problem framing.

Before drafting, answer:

What exactly is the question?
What does each key word mean?
What is outside scope?
What decision must the reader reach?
What evidence would change my conclusion?

高级写作首先是“定义问题”，然后才是“造句”。

34. Turn a broad topic into a question | 把大话题变成可回答的问题

Topic: AI and education.

Too broad.

Possible questions:

When does generative AI improve feedback quality for secondary-school writing?
What kinds of writing tasks become less diagnostically useful when students have unrestricted AI assistance?
How should schools distinguish support from substitution?

一个好问题会自动帮助你选择证据、范围和文章结构。

35. Define before you argue | 争论之前先定义

Many weak essays use central terms without defining them.

“Success”, “fairness”, “effective”, “quality”, “intelligence”, “creativity”, “harm”, “benefit” can hide multiple criteria.

Ask:

What would count as evidence that this condition is present?

如果你无法说出“什么证据出现时，我会说它有效/公平/成功”，你的核心词可能还没有操作化。

36. Thesis is not a slogan | Thesis 不是一句口号

Weak thesis:

Technology has advantages and disadvantages.

Stronger thesis:

Schools should permit generative AI for formative feedback but restrict it in tasks designed to diagnose independent writing ability, because the same tool that can accelerate feedback can also obscure the learner’s actual performance.

The stronger thesis contains a decision, a boundary and a mechanism.

高级 thesis 通常包含：立场 + 范围 + 为什么。

37. Paragraphs need jobs | 每一段都要有工作

Do not write paragraphs because “it is time for a new paragraph”. Give each paragraph a function.

define
establish evidence
explain mechanism
compare alternatives
introduce limitation
address counterargument
evaluate trade-off
derive implication

If two paragraphs do the same job, one may be redundant.

38. The paragraph operating system | 段落操作系统

A flexible advanced paragraph often contains:

Point — the paragraph’s claim.
Reasoning — mechanism or explanation.
Evidence/example — support.
Interpretation — what the evidence means.
Qualification — limit or condition.
Return — connection to larger thesis.

Not every paragraph needs all six parts in that exact order. The framework is a diagnostic tool, not a rigid template.

不要把它当成必须每段六句的模板。它的作用是检查：你的段落是否真的完成了论证工作。

39. Evidence without explanation is not an argument | 有证据，没有解释，还不是论证

Weak:

A survey found that 68% of respondents preferred the new system.

So what?

Stronger:

A survey found that 68% of respondents preferred the new system. This suggests that the redesign improved user acceptance, although preference alone does not establish that the system is more effective.

高级写作要把“证据 → 解释 → 限制”连起来。

40. Explanation without evidence is also incomplete | 只有解释，没有证据，也不够

A beautiful mechanism can still be wrong.

The programme probably works because students feel more motivated.

If motivation was never measured, this may be plausible but remains speculative.

清楚解释不等于证据已经存在。把 plausibility 和 evidence 分开。

41. Use examples to clarify, not to prove universality | 例子用来说明，不要用一个例子证明普遍规律

A case can show how something could happen. It cannot by itself show how often it happens.

例子擅长回答“可能怎样发生”，不擅长单独回答“普遍发生多少”。

42. Use numbers with denominators | 数字要带分母意识

“Cases doubled” sounds dramatic. From 1 to 2 is also doubling.

Ask:

From what to what?
Out of how many?
Over what period?
Compared with what baseline?

高级阅读要把百分比还原成实际规模。

43. Average can hide distribution | 平均数可能隐藏分布

An average improvement of five points can mean:

everyone improved around five points
half improved ten and half did not improve
a small group improved dramatically

When a text uses averages, ask whether spread, subgroup or distribution information matters.

44. Correlation language must stay correlation language | 相关关系不要被写成因果

Weak:

Screen time reduces sleep.

If the study is observational, a safer formulation may be:

Higher screen time was associated with shorter sleep duration.

Then discuss possible mechanisms separately.

精准写作的一个核心纪律，是不把证据类型写成比它更强的结论。

45. Write with calibrated verbs | 用动词校准证据强度

Lower commitment	Moderate	Higher commitment
may indicate	suggests	demonstrates
is consistent with	supports	establishes
could contribute to	is associated with	causes

Do not treat this as a simple ladder where the right column is “better English”. The correct verb depends on the evidence.

右边不是“更高级”，只是更强。证据不够时，用强动词反而更差。

46. Write the limitation where it matters | 限制条件不要全部藏在结尾

Weak academic writing sometimes makes strong claims for three pages and then adds a final “limitations” paragraph. Stronger writing often places qualifications near the claim they limit.

如果限制直接影响某个结论，就在结论附近写出来。这样读者不用先被你误导，再到最后才被纠正。

47. Concession is not surrender | 让步不是放弃立场

A concession can strengthen credibility:

Although the intervention increases administrative workload initially, the long-term reduction in manual processing may justify the transition cost.

You acknowledge a real cost while explaining why your final judgement remains.

高级论证不是假装自己的立场没有代价，而是把代价放进判断。

48. Trade-offs are often the real argument | 很多高级问题真正讨论的是 trade-off

Public policy, education, business and technology rarely offer pure wins. Strong writing asks what is gained, what is lost, who bears the cost and under what conditions the trade-off is justified.

高级写作经常不是“好还是坏”，而是：

谁得到好处？
谁承担代价？
短期和长期是否不同？
有没有不可逆风险？
什么条件下值得交换？

49. Comparative judgement needs criteria | 比较判断必须先有标准

Do not say Option A is “better” before defining better.

Possible criteria:

cost
speed
reliability
equity
scalability
reversibility
learning quality
risk

Then compare options against the same criteria. Otherwise you may praise A for speed and criticise B for cost without a coherent framework.

50. Read the question like a contract | 把题目当作合同来读

Before writing, underline command words and scope words.

Examples:

evaluate — requires judgement against criteria
compare — requires similarities/differences around shared dimensions
to what extent — requires calibrated degree, not simple yes/no
discuss — requires examination of relevant perspectives, not a one-sided list
explain — requires mechanism, cause or relationship

高级语言也救不了“答错问题”的文章。

51. Build the answer before building sentences | 先建答案，再建句子

Write a one-line answer to the question before drafting.

My answer is: X, mainly because A and B, although C limits the conclusion.

If you cannot complete this line, your thinking may not yet be ready for paragraphs.

52. Outline by reasoning, not by topic labels | 提纲按推理组织，不按话题堆箱子

Weak outline:

Technology
Education
Students

Stronger outline:

AI can accelerate formative feedback.
But unrestricted AI can hide independent performance.
Therefore policy should separate learning support from diagnostic assessment.

第二种提纲已经包含推理方向。

53. Write introductions that define the problem | 开头先定义问题，不要先写世界很大

Avoid generic openings such as “Since the beginning of time…” or “Nowadays, technology is developing rapidly.”

Start with the actual decision, tension or question.

Generative AI has made high-quality language feedback available at unprecedented scale. The educational challenge is no longer whether students can access such tools, but when their use supports learning and when it obscures independent performance.

高级开头应该尽快把真正问题放到桌面上。

54. Conclusions should close the reasoning, not repeat the introduction | 结尾要完成判断，不要机械复读

A strong conclusion can:

answer the question at the calibrated level
identify the decisive reason
state the key condition
acknowledge the main uncertainty
derive a practical implication

不要在最后一段只是把第一段换几个词再写一次。

55. Writing is recursive | 写作不是“计划完 → 写完 → 最后改一下”

Research on writing processes shows that planning, drafting and revising often recur. Source-based writing is especially recursive because writers move repeatedly between sources, notes and emerging text.

写作过程研究长期说明：计划、起草和修改不是严格一次性顺序。尤其是 source-based writing（基于来源写作），学习者会不断在来源、笔记和正在形成的文本之间来回移动。

A useful principle:

When the sentence problem is actually an idea problem, return to the idea.

如果一句话怎么改都不清楚，可能不是英语问题，而是你的思想关系还没有想清。

56. Revision has levels | 修改要分层

Revision level	Question
Argument	Does the conclusion follow?
Structure	Is information in the right order?
Paragraph	Does each paragraph have one clear job?
Sentence	Is the relationship explicit and readable?
Lexical	Are word choice, collocation and register precise?
Grammar	Are tense, reference, agreement and boundaries stable?
Proofreading	Spelling, punctuation, formatting?

Do not start with commas when the argument is broken.

论证坏了的时候，不要先修逗号。

57. The reverse-outline method | 反向提纲法

After drafting, write one sentence beside each paragraph describing what it actually does.

If the sequence is:

paragraph 1: background
paragraph 2: more background
paragraph 3: another example
paragraph 4: finally gives argument

your structure may be too slow.

反向提纲让你看到“文章实际长成什么”，而不是“你原本以为它会长成什么”。

58. The reader test | 读者测试

Ask a reader to underline the sentence they think is the main claim of each paragraph.

If they choose a different sentence from the one you intended, the paragraph may be signalling poorly.

高级写作不仅是“作者知道自己在说什么”，还要让读者知道作者在说什么。

59. Compression test | 压缩测试

After writing a paragraph, compress it into one sentence.

If you cannot, the paragraph may contain multiple competing jobs.

压缩不是为了删除文章，而是检查段落中心是否存在。

60. Expansion test | 展开测试

Take a thesis sentence and expand it into:

mechanism
evidence
example
qualification
counterargument

If you cannot expand it, the thesis may be too vague.

61. The “because / therefore / however” triangle | because / therefore / however 三角测试

For every major claim, see whether you can complete:

because… — reason
therefore… — consequence
however… — limit / counterpressure

This is a fast way to test whether the idea has depth.

62. The disconfirmation question | 反证问题

Ask:

What evidence would make me change my mind?

If the answer is “nothing”, you are no longer evaluating evidence; you are protecting a conclusion.

eduKate Sengkang’s Disconfirmation State work points toward the same discipline: evidence that agrees with you is not enough.

63. The steelman test | Steelman 测试

Before rebutting an opposing position, state it in a form that its supporter would recognise as fair.

If you cannot do that, you may not yet understand the disagreement.

高级写作需要能够准确表达自己不赞成的观点。

64. Think in criteria, not adjectives | 用标准思考，不要只用形容词

Weak:

This is a good policy.

Better:

The policy is effective at reducing waiting times but performs poorly on cost predictability and regional equity.

“好/坏”被拆成可检查标准以后，思考才真正开始。

65. Think in mechanisms | 用机制思考

If you say X improves Y, ask how.

Smaller classes improve writing.

Possible mechanism:

Smaller classes may increase the frequency and specificity of feedback, which in turn gives learners more opportunities to revise recurring errors.

机制解释可以被证据支持或推翻。空泛结论很难被检验。

66. Think in constraints | 用限制条件思考

A strategy that works in one environment may fail elsewhere because of staff, time, cost, infrastructure, regulation or learner readiness.

高级判断常常不是“这个方法有效吗”，而是“在什么条件下有效”。

67. Think in levels | 不要把不同层级混在一起

Distinguish:

individual learner
classroom
school
system

A strategy effective for one learner may not scale to a national system. A system-level statistic may not predict an individual.

从个体推到系统、从系统推到个体，都需要额外推理。

68. Think in time | 时间尺度会改变结论

Short-term improvement and long-term learning are not the same. Immediate performance can rise while durable retention remains unchanged.

在教育中尤其要区分：

今天做得更好
下周还能不能做
换题还能不能做
没有帮助还能不能做

高级写作必须避免把短期表现自动写成长期学习。

69. Think in transfer | 能力是否能迁移？

A learner may perform well on familiar material but fail when surface features change.

Ask:

Does the skill survive a new topic?
A new genre?
Less time?
No notes?
A different audience?

这也连接 Lesson 001 的 learning system：真正掌握必须经得住迁移。

70. Think in opportunity cost | 选择 A 就意味着没选什么？

Every recommendation consumes time, money or attention.

If a school adds two hours of one activity, what is displaced?

如果一个学习者每天花 60 分钟背低频词，他没有把这 60 分钟用在哪里？

高级判断必须看到被放弃的替代方案。

71. Think in base rates | 不要被一个惊人个案控制判断

A dramatic case can attract attention, but decision-making often requires knowing how common the event is.

Ask:

How often does this happen in the relevant population?

个案告诉你“可能发生”，base rate 告诉你“通常多常发生”。

72. Think in uncertainty | 不确定性不是失败

Sometimes the correct advanced conclusion is:

The available evidence is insufficient to distinguish between the two explanations.

这不是“没有观点”。这是证据纪律。

73. Chinese can help organise thought—but avoid translating Chinese rhetoric automatically | 中文可以帮助思考，但不要自动翻译中文修辞

Mandarin rhetorical organisation can help you generate ideas, but English readers may expect different signalling, thesis placement, paragraph development or explicitness depending on genre.

中文可以帮助你快速生成思想，但英文最终文本要根据英语文体重新组织。Lesson 003 已经处理句法迁移；这里要进一步处理“论证和篇章迁移”。

Do not ask “How do I translate this elegant Chinese paragraph?” first.

Ask:

What job is this paragraph doing in the argument, and how would English normally perform that job?

74. Avoid four-character abstraction becoming empty English abstraction | 四字格不要变成空洞抽象英文

Chinese formal prose can compress evaluation into compact expressions. Literal English may become chains of abstract nouns.

Instead of translating the surface, unpack:

what changed?
who acted?
what caused what?
what outcome matters?

Then rebuild in direct English.

75. Chinese examples may precede claims; English academic prose often signals claims earlier | 中英文“例子和主张”的顺序可能不同

Many genres in both languages permit multiple orders, so do not turn this into a stereotype. But when English analytical writing feels indirect, check whether the reader has to wait too long to discover the paragraph’s point.

一个实用检查是：读完段落第一、二句，读者知不知道这一段为什么存在？

76. Source use: quote less, integrate more | 引用少一点，整合多一点

Advanced source-based writing should not become a collage of quotations.

Use quotation when wording itself matters. Otherwise, summarise or paraphrase and integrate the source into your own sentence.

Weak:

Smith says “X”. Jones says “Y”. Lee says “Z”.

Stronger:

While Smith and Jones both attribute the decline to cost pressures, Lee argues that the more important mechanism is reduced access.

作者自己的句法成为多个来源的共同框架，这才是 synthesis。

77. Introduce sources by function, not by fame | 引来源时说明“它在这里做什么”

Instead of:

According to Professor Smith…

Consider:

A longitudinal study by Smith et al. provides evidence that…

Smith’s theoretical account offers a different explanation…

The reader learns why the source matters.

78. Keep your voice visible | 综合写作里不要让自己的声音消失

Your job is not only to report sources. Your job is to organise, compare, weigh and judge them.

Useful writer moves:

Taken together, these studies suggest…
The disagreement appears to arise from…
The stronger evidence comes from…
However, this conclusion should remain limited to…
A more plausible interpretation is…

这些句子显示“你在处理来源”，而不是只搬运来源。

79. C1 writing means purpose, reader and genre control | C1 写作要求控制目的、读者和文体

Cambridge C1 Advanced requires multiple text types, including essays, letters/emails, proposals, reports and reviews. The official format explicitly gives context, purpose and target reader for situational writing tasks.

这提醒我们：高级写作不是只有 essay。你必须根据目的、读者和文体改变结构。

A proposal is not an essay. A report is not a review. A review is not a formal letter.

80. Genre changes what evidence looks like | 文体会改变“证据”怎样出现

Academic essay: sources, reasoning, qualification.
Report: findings, criteria, implications, recommendations.
Proposal: need, plan, feasibility, benefit, risk.
Review: criteria, evidence from experience/text, judgement for audience.
Professional email: purpose, facts, action, deadline.

同一个思想进入不同文体时，不只是“语气变了”，结构也会变。

81. Writing for a reader means predicting questions | 为读者写，就是提前预测问题

After each paragraph ask:

Why should I believe this?
What does this mean?
Compared with what?
Under what conditions?
So what?
What should happen next?

If the reader will naturally ask one of these, answer it before the question becomes a gap.

82. Precision often comes from deleting | 精准写作经常靠“删”

Delete:

empty openings
repeated claims
unnecessary intensifiers
double connectors
abstract nouns that hide actions
examples that do not advance the argument

高级不是“加更多”，很多时候是“删掉不会工作的东西”。

83. Concision is not shortness | 简洁不等于短

Concision means every part earns its place.

A 30-word sentence can be concise if it expresses a complex relationship efficiently. A 10-word sentence can be wasteful if it repeats what the reader already knows.

简洁的单位不是“字数”，而是“信息价值 / 语言成本”。

84. Sentence length should follow thought | 句子长度跟思想走

Use short sentences for decisive transitions, emphasis and clarity. Use longer sentences when relationships must stay visible together.

Do not make every sentence long to sound advanced. Do not make every sentence short to sound clear.

高级句法是“匹配思想结构”。

85. Use punctuation to reveal hierarchy | 标点帮助读者看见层级

Full stops separate complete moves. Colons introduce explanation. Semicolons can link closely related independent clauses. Dashes can create controlled interruption or emphasis.

标点不是“漂亮风格”，而是帮助读者看见信息层级。

86. Read aloud during revision | 修改时读出来

Reading aloud exposes:

overlong sentences
repetition
awkward rhythm
unclear reference
unnatural collocation
missing transitions

如果一句话读到一半你自己也忘了主句在哪里，读者很可能也会迷路。

87. Use a two-column source notebook | 建立双栏来源笔记

Source says	My thinking
Claim / evidence / limit	agree? why? conflict? implication? question?

This prevents notes from becoming copied fragments with no relationship to your argument.

笔记不是仓库。每条来源旁边都要有“它对我的问题意味着什么”。

88. Use question-generation while reading | 阅读时自己出题

After a section, generate three questions:

literal question
inference question
evaluation question

This turns reading from reception into active interrogation.

See How Question Generation Works in Learning.

89. Closed-book reconstruction | 合上文章，重建论证

After reading, close the source and write:

main claim
three supports
one limitation
one counterargument
one implication

If you cannot reconstruct it, rereading may have created familiarity without structure.

这也是 Lesson 001 retrieval principle 在高级阅读中的应用。

90. Reopen and compare | 重新打开，比较你重建错在哪里

Compare your reconstruction with the text.

Did you:

overstate?
forget a qualification?
confuse example with evidence?
miss the main claim?
invent an implication?

这一步把阅读错误变成可诊断数据。

91. One article, four outputs | 一篇文章，四种输出

Use one demanding article to produce:

100-word neutral summary
150-word critical response
two-minute spoken explanation
one paragraph applying the idea to a new context

同一输入穿过四种输出，才能把“看懂”升级成“掌握”。

92. One question, three sources | 一个问题，三个来源

Choose one question. Read three sources with different perspectives.

Produce:

synthesis matrix
one paragraph of agreement
one paragraph of disagreement
one paragraph of your judgement

This is a compact C1–C2 research-and-writing drill.

93. One claim, five revisions | 一个主张，改五遍

Start:

AI is good for students.

Revision 1 — define:

Generative AI can support students by providing immediate language feedback.

Revision 2 — scope:

Generative AI can support some students by providing immediate language feedback during formative practice.

Revision 3 — mechanism:

Generative AI may support formative writing practice by giving students immediate feedback that allows faster revision cycles.

Revision 4 — qualification:

Generative AI may support formative writing practice by accelerating feedback and revision, although the benefit depends on whether students still perform enough independent thinking and drafting.

Revision 5 — evidence discipline:

Generative AI may support formative writing practice by accelerating feedback and revision, but claims about long-term learning should remain cautious until performance is tested independently of the tool.

这就是“清晰思考 → 精准写作”的过程。

94. Build a personal reasoning-error corpus | 建立个人推理错误库

My sentence	Problem	Repair rule
Three students improved, so the method works.	overgeneralisation	match population claim to sample evidence
Scores rose after the change, so the change caused it.	post hoc / causal leap	consider alternatives and design
Experts support it, so it is true.	authority replacing evidence	inspect underlying evidence
The policy is unfair.	criterion undefined	define fairness dimension

像语法错误库一样，推理错误也会重复。把它们变成个人课程。

95. Build a personal writing-error corpus | 建立个人写作错误库

Track categories such as:

thesis too broad
paragraph with no job
evidence not explained
qualification missing
counterargument weak
source listing instead of synthesis
reference unclear
register mismatch
translation-shaped sentence
collocation problem

高级写作提升往往来自“重复错误减少”，而不是每次都学一个新技巧。

96. A 20-minute daily deep-reading routine | 每天 20 分钟深度阅读

Time	Task
5 min	Read one demanding paragraph or short section.
5 min	Mark claim, evidence, qualification and implication.
5 min	Close the source and reconstruct it.
5 min	Write one evaluation question or counterargument.

每天只做一小块，但读得深。

97. A 45-minute integrated session | 45 分钟整合训练

Time	Task	Skill
12 min	read one serious text	reception
8 min	argument map	reasoning
8 min	closed-book summary	retrieval
10 min	write a response paragraph	production
7 min	revise scope, evidence and language	feedback

98. A 90-minute deep-work session | 90 分钟深度工作

Time	Task
20 min	read two related sources
15 min	build synthesis matrix
10 min	form judgement and outline
30 min	write 500–700 words
15 min	reverse-outline and revise

不要把 90 分钟全部用来“看更多材料”。至少一半时间进入重建、判断和输出。

99. Seven-day reading-to-writing cycle | 七天从阅读到写作循环

Day 1: deep read Source A.
Day 2: deep read Source B and compare.
Day 3: synthesis matrix + questions.
Day 4: spoken explanation without notes.
Day 5: write 500–700 word response.
Day 6: reverse-outline, evidence audit and rewrite.
Day 7: closed-book reconstruction and reflection.

一周围绕同一个问题反复加工，比每天换一个新话题更容易形成深度。

100. Twelve-week C1–C2 reading-writing progression | 12 周 C1–C2 阅读写作路线

Weeks	Focus	Evidence produced
1–2	claims, evidence, inference, qualification	annotated texts + summaries
3–4	argument maps, causation, counterarguments	critical response paragraphs
5–6	comparative reading and source quality	synthesis matrices
7–8	paragraph architecture and thesis control	short essays
9–10	source-based writing, genre and register	report / proposal / essay
11–12	independent synthesis, revision and transfer	long-form final project + oral defence

12 周不是“读完高级英语”。它建立的是一套可持续的高阶读写系统。

101. The C1–C2 benchmark task | C1–C2 基准任务

Once a month:

Read two demanding texts on the same issue.
Without notes, explain the main disagreement aloud.
Write a 150-word neutral synthesis.
Write a 500-word evaluation.
Revise after a 24-hour delay.

Measure:

meaning accuracy
source distinction
evidence discipline
argument structure
qualification
register
language precision
revision quality

102. Leading indicators of real progress | 真正进步的领先指标

You identify the claim faster.
You preserve qualifications in summaries.
You distinguish evidence from interpretation.
You generate alternative explanations automatically.
You compare sources by issue instead of listing them.
Your paragraphs have clearer jobs.
Your evidence is followed by interpretation.
Your claims become narrower but stronger.
You delete more unnecessary language.
You can explain complex ideas more simply without distortion.

这些变化往往先发生，然后考试分数或外部评价才慢慢跟上。

103. What not to do | 不要这样学高级读写

Do not underline every unknown word. 不要把阅读变成查词中断。
Do not confuse agreement with understanding. 同意作者不等于理解作者。
Do not treat quotation as evidence by itself. 引用一句话不等于论证完成。
Do not list sources without comparing them. A 说、B 说、C 说不是 synthesis。
Do not use difficult vocabulary to hide weak reasoning. 大词不能救坏逻辑。
Do not start revision with grammar when the thesis is broken. 先修大结构。
Do not remove all uncertainty. 高级写作允许“不确定，但准确”。
Do not invent balance. 不是每个问题都必须 50/50；证据权重可以不对称。
Do not confuse length with depth. 长不等于深。
Do not confuse certainty with confidence. 写得坚定不代表证据足够。

104. Apex official learning routes | 高质量官方学习入口

Cambridge English C1 Advanced Exam Format — reading for detail, opinion, tone, implication, attitude, comparing viewpoints, text structure and advanced writing tasks.
Cambridge C1 Advanced Preparation — official sample tests and learner resources.
British Council C1 Reading — advanced reading practice.
British Council C1 Writing — advanced writing practice.
Council of Europe CEFR Mediation — relaying, explaining, summarising and reconstructing complex meaning.

105. Research floor | 研究基础

The research floor for this lesson includes source-based writing, synthesis writing, process feedback and model observation.

Vandermeulen et al. (2024), Reading and Writing — learning to write syntheses through process feedback and observing models.
Vandermeulen et al. (2024), Journal of Second Language Writing — approaches to argumentative and narrative writing processes in L1/L2.
Vandermeulen et al. (2023), Written Communication — writing-process feedback and comparison with exemplars in synthesis texts.
Tarchi et al. (2023), Reading and Writing — recursivity in source-based writing.
Leijten et al. (2019), Studies in Second Language Acquisition — source use in L1/L2 source-based writing.

Research gives us useful mechanisms, but this course still returns to individual evidence: your reading notes, your summaries, your argument maps, your drafts and your revisions.

研究提供机制，但真正的个人课程仍然来自你的真实样本。

106. Canonical eduKate routes | eduKate 现有主页面

107. SEO language map | 本课搜索关键词范围

This lesson naturally serves learners searching for advanced English reading, advanced English writing, C1 reading, C1 writing, C2 reading, C2 writing, critical reading, critical thinking in English, English argument writing, evidence in English writing, inference skills, source-based writing, synthesis writing, academic English writing, how to write clearly in English, how to improve English essays, English for Chinese speakers, English for Mandarin speakers, advanced English for Chinese students, 高级英语阅读, 高级英语写作, C1 阅读, C1 写作, C2 英语, 深度阅读, 批判性阅读, 批判性思维, 英语论证, 英语证据, 英语推断, 英语综合写作, 学术英语写作, 英语论文怎么写, 中文母语高级英语 and 英语精准写作.

Keywords are not the curriculum. They are entry points into a coherent skill system.

关键词不是课程本身。它们只是读者进入这套系统的入口。

108. The one-page deep-reading system | 一页深度阅读系统

Find the main claim. 找主张。
Classify the claim. 判断是描述、因果、评价还是建议。
Identify support. 找证据和理由。
Separate evidence from interpretation. 分开事实和解释。
Preserve qualifications. 保留限制条件。
Infer within boundaries. 推断不过界。
Read stance and framing. 读立场和框架。
Generate alternatives. 想替代解释。
Map the argument. 画论证地图。
Close the text and reconstruct. 合上文章重建。

109. The one-page precise-writing system | 一页精准写作系统

Define the question. 先定义问题。
State a calibrated thesis. 写有范围的 thesis。
Give each paragraph one job. 每段一个任务。
Explain the mechanism. 不只给结论。
Use evidence that matches the claim. 证据和主张匹配。
Interpret the evidence. 不让引用孤立。
Add qualification where needed. 在需要处加限制。
Address the strongest counterargument. 回答最强反方。
Synthesise sources by issue. 按问题综合来源。
Revise from argument to grammar. 从大结构修到小语言。

110. Final assignment | 本批次最终作业

Choose one serious question that matters to you: education, technology, work, health, cities, finance, culture, science or another field.

选择一个你真正关心的严肃问题：教育、科技、工作、健康、城市、金融、文化、科学都可以。

Then complete the full chain:

Find two strong sources with different perspectives.
Read each for claim, evidence, qualification and stance.
Build a synthesis matrix.
Close both sources and explain the disagreement aloud.
Write a one-sentence judgement.
Build a reasoning outline.
Write 700–1,000 words.
Reverse-outline the draft.
Audit every major claim against its evidence.
Revise language using Lesson 002 for register and Lesson 003 for Chinese→English transfer.

Then ask:

Can a careful reader see not only what I believe, but why I believe it, what evidence supports it, where the limits are, and what would make me change my mind?

一个谨慎的读者，能不能看见：我相信什么、为什么相信、证据在哪里、限制在哪里，以及什么证据会让我改变结论？

If yes, your English is no longer only carrying sentences.

It is carrying thought.

如果可以，你的英语已经不只是在运输句子。

它开始运输思想。

Batch 01 complete | 第一组四课完成

You have now built four foundations:

Lesson 001 — Build a C1–C2 Learning System
Lesson 002 — Master Collocation, Register and Precision
Lesson 003 — Stop Translating Chinese Grammar into English
Lesson 004 — Read Deeply, Think Clearly, Write Precisely

The series can now expand outward into specialised advanced-English nodes without losing the central architecture.

现在，这个系列可以继续向高级听力、口语、发音、学术讨论、专业沟通、文学阅读、研究写作、演讲、辩论、复杂语法和跨文化沟通等节点扩展，而不需要重新定义主线。

Back to Advanced English Chinese Edition Hub · 返回高级英语中文版主页

Small Group Tutorials

How to Learn Advanced English (Chinese Edition) | Lesson No.004 | Read Deeply, Think Clearly, Write Precisely | 第004课：深度阅读、清晰思考、精准写作

Read Deeply, Think Clearly, Write Precisely | 深度阅读、清晰思考、精准写作

Before we begin: this is an integration lesson, not a duplicate owner | 开始之前：这一课是“整合层”，不是重复主页面

1. C1–C2 reading is not “understand the words” | C1–C2 阅读不是“看懂单词”

2. Reading has layers | 深度阅读是分层的

3. Read the claim first | 先找主张，不要先抓生词

4. Claim type controls evidence type | 主张类型决定需要什么证据

5. Evidence is not one thing | 证据不是一种东西

6. Build the evidence ladder | 建立证据阶梯

7. Separate observation from interpretation | 把观察和解释分开

8. Read causation carefully | 因果关系要慢下来读

9. Read qualifications as carefully as conclusions | 限制条件要和结论一样认真读

10. Inference must be bounded | 推断必须有边界

11. Implication is not the same as hidden fact | 隐含意义不等于“隐藏事实”

12. Read stance through verbs | 通过 reporting verbs 读立场

13. Read adjectives and adverbs for evaluation | 形容词和副词里藏着评价

14. Read what is selected—and what is omitted | 读作者选了什么，也读作者没有选什么

15. Bias is not “I disagree with the writer” | Bias 不等于“我不同意作者”

16. Read arguments as maps | 把论证读成地图

17. The “because test” exposes weak reasoning | because 测试可以暴露断裂推理

18. The “therefore test” checks whether evidence reaches the claim | therefore 测试检查证据是否真的够到结论

19. The “what else could explain this?” test | “还有什么解释？”测试

20. Counterarguments are not decorative | 反方观点不是“为了显得客观”加一段

21. A rebuttal should answer the strongest version | Rebuttal 要回答最强版本，不要打稻草人

22. Fallacy labels are not substitutes for reasoning | 贴“谬误标签”不能替代理解

23. Deep reading includes paragraph function | 深度阅读要读每一段“在做什么”

24. Read transitions as logic signals | 转折和连接词是逻辑信号

25. Gapped-text thinking is useful beyond exams | “段落放哪里”训练可以迁移到真实阅读

26. Comparative reading: do not summarise Source A, then Source B, then stop | 多文本阅读不能只做 A 总结 + B 总结

27. Build a synthesis matrix | 建立综合矩阵

28. Source hierarchy: not every source deserves equal weight | 不同来源不能自动等权

29. Summarising is compression with meaning preservation | 总结是“压缩但不改意思”

30. Paraphrase means structure change plus meaning control | 改写不是“同义词替换”

31. Synthesis writing is a reading test before it is a writing test | 综合写作首先是阅读测试

32. Research on synthesis writing supports model observation and feedback | 综合写作需要看范本，也需要过程反馈

33. Writing begins before the first sentence | 写作在第一句话之前就开始

34. Turn a broad topic into a question | 把大话题变成可回答的问题

35. Define before you argue | 争论之前先定义

36. Thesis is not a slogan | Thesis 不是一句口号

37. Paragraphs need jobs | 每一段都要有工作

38. The paragraph operating system | 段落操作系统

39. Evidence without explanation is not an argument | 有证据，没有解释，还不是论证

40. Explanation without evidence is also incomplete | 只有解释，没有证据，也不够

41. Use examples to clarify, not to prove universality | 例子用来说明，不要用一个例子证明普遍规律

42. Use numbers with denominators | 数字要带分母意识

43. Average can hide distribution | 平均数可能隐藏分布

44. Correlation language must stay correlation language | 相关关系不要被写成因果

45. Write with calibrated verbs | 用动词校准证据强度

46. Write the limitation where it matters | 限制条件不要全部藏在结尾

47. Concession is not surrender | 让步不是放弃立场

48. Trade-offs are often the real argument | 很多高级问题真正讨论的是 trade-off

49. Comparative judgement needs criteria | 比较判断必须先有标准

50. Read the question like a contract | 把题目当作合同来读

51. Build the answer before building sentences | 先建答案，再建句子

52. Outline by reasoning, not by topic labels | 提纲按推理组织，不按话题堆箱子

53. Write introductions that define the problem | 开头先定义问题，不要先写世界很大

54. Conclusions should close the reasoning, not repeat the introduction | 结尾要完成判断，不要机械复读

55. Writing is recursive | 写作不是“计划完 → 写完 → 最后改一下”

56. Revision has levels | 修改要分层

57. The reverse-outline method | 反向提纲法

58. The reader test | 读者测试

59. Compression test | 压缩测试

60. Expansion test | 展开测试

61. The “because / therefore / however” triangle | because / therefore / however 三角测试

62. The disconfirmation question | 反证问题

63. The steelman test | Steelman 测试

64. Think in criteria, not adjectives | 用标准思考，不要只用形容词

65. Think in mechanisms | 用机制思考

66. Think in constraints | 用限制条件思考

67. Think in levels | 不要把不同层级混在一起

68. Think in time | 时间尺度会改变结论

69. Think in transfer | 能力是否能迁移？

70. Think in opportunity cost | 选择 A 就意味着没选什么？

71. Think in base rates | 不要被一个惊人个案控制判断

72. Think in uncertainty | 不确定性不是失败

73. Chinese can help organise thought—but avoid translating Chinese rhetoric automatically | 中文可以帮助思考，但不要自动翻译中文修辞

74. Avoid four-character abstraction becoming empty English abstraction | 四字格不要变成空洞抽象英文

75. Chinese examples may precede claims; English academic prose often signals claims earlier | 中英文“例子和主张”的顺序可能不同

76. Source use: quote less, integrate more | 引用少一点，整合多一点