Series ID: EDKS-ADV-ZH-0013 · How to Learn Advanced English (Chinese Edition) · Lesson No.013 · C1 → C2

Build an Academic Argument That Survives Counterevidence | 构建经得住反证的学术论证

A strong academic argument is not the claim that sounds most certain. It is the claim that remains defensible after the strongest fair test.

强学术论证，不是“听起来最确定”的主张，而是经过最强、最公平的检验后，仍然站得住的主张。

At advanced level, argument is not a performance of confidence. It is a disciplined relationship among question, claim, reasons, evidence, warrants, assumptions, counterevidence, alternative explanations, scope and conclusion.

高级 argument 不是“我很有信心，所以我说得很肯定”。它是一组受约束的关系：问题 → 主张 → 理由 → 证据 → warrant → assumption → counterevidence → alternative explanation → scope → conclusion。

The goal is not to become impossible to criticise. The goal is to make criticism useful. Serious objections should either strengthen your argument, narrow it, redirect it or show you that the original thesis needs to change.

目标不是让别人“无法反驳”。真正的目标是：让高质量反驳成为思考工具。好的 objection 应该迫使你的论证更强、更窄、更准确，或者在必要时直接改写 thesis。

Before we begin: this lesson integrates, but does not replace, existing argument owners | 开始之前：本课整合现有 argument owners

eduKate Sengkang already has dedicated pages for school-stage argument, fallacies, counterexamples, reasoning types and claim calibration. They remain their own canonical owners.

How Argumentative Writing Builds Claim, Evidence and Judgement — general mechanism owner.
Secondary 3 English Learning Guide | Argument Fallacies and Reasoning Errors — school-stage fallacy owner.
How Students Recognise Logical Fallacies Without Reducing Every Argument to a Label — fallacy mechanism.
How Counterexamples Test Broad Claims and Reveal Their Limits — counterexample owner.
How Inductive and Deductive Reasoning Build Different Kinds of English Argument — reasoning-type owner.
How Conditional Language Builds If-Then Reasoning and Hypothetical Claims — conditional-claim owner.
Lesson 011 | Multi-Source Synthesis — evidence integration.
Lesson 012 | Literature Review as Scholarly Conversation — field mapping and research positioning.

Lesson 013 owns the advanced integration task: how to build a claim that remains defensible when serious evidence and serious objections push against it.

1. Academic argument begins with a question | Academic argument 从问题开始

The University of Toronto’s academic-writing guidance emphasises that an academic essay needs an argument: it should answer a question with a thesis supported by reasoning and evidence, and the provisional thesis should be tested rather than protected from challenge.

University of Toronto | General Advice on Academic Essay-Writing

先问：

What exactly am I trying to establish?

而不是：

What do I want to say about this topic?

2. Topic is not question | Topic 不是 question

Topic:

AI feedback in education.

Question:

Under what conditions does AI-assisted feedback improve independent revision quality?

第二个才会约束 evidence。

3. Question is not thesis | Question 也不是 thesis

Question:

Does retrieval practice improve long-term retention?

Provisional thesis:

Retrieval practice improves long-term retention most reliably when practice requires successful recall and is followed by corrective feedback.

Thesis 是暂时答案，需要被 evidence 测试。

4. Good theses are limited and arguable | 好 thesis 有边界，也可争辩

The University of Toronto’s thesis-statement guidance stresses that a useful thesis is a definite, limited assertion that requires explanation and support rather than a sweeping statement or obvious fact.

University of Toronto | Using Thesis Statements

5. A claim should expose itself to evidence | Claim 必须允许 evidence 影响它

Weak:

Technology is important in education.

Stronger:

Technology improves revision quality only when learners receive enough guidance to evaluate and act on feedback.

第二句可以被支持、限制或推翻。

6. If nothing could change your mind, you do not yet have an academic claim | 如果什么都不能改变你的结论，就不是学术 claim

Ask:

What evidence would make me revise this claim?

不同学科对“可反驳”的形式不同，但都需要某种标准说明什么会算作严重挑战。

7. Separate claim type | 先判断 claim 属于哪一类

Claim type	Question	Evidence burden
Descriptive	What is happening?	accurate observation/measurement
Causal	What produces what?	causal design + alternatives
Interpretive	What does this mean?	text/context + rival readings
Evaluative	How good/important is it?	explicit criteria
Policy	What should be done?	goals + effects + feasibility + trade-offs
Theoretical	How should the phenomenon be explained?	explanatory reach + predictions + alternatives

8. Claim type controls evidence | Claim type 决定证据任务

You cannot support a causal claim with preference data alone.

You cannot support a policy claim with effect evidence alone if costs, feasibility and competing values matter.

9. Purdue’s argument organiser provides a useful architecture | Purdue 提供一个实用 argument 架构

Purdue OWL’s argument resources organise argument around rhetorical situation, issue definition, claim, reasons, evidence, warrants, counterargument and rebuttal. Use this as a diagnostic framework rather than a mandatory template for every discipline.

Purdue OWL | Organizing Your Argument

10. Claim | Claim 是你要让读者接受什么

The intervention improves independent revision.

Ask:

for whom?
compared with what?
how much?
under which conditions?
over what period?

11. Reason | Reason 回答 why

Because it requires learners to evaluate feedback rather than copy corrections.

Reason is not yet evidence.

12. Evidence | Evidence 给出可检查支持

Data, textual evidence, documented cases, experiments, observations, archival materials, credible testimony or other discipline-appropriate material.

13. Warrant | Warrant 解释“为什么这个 evidence 能支持这个 claim”

Purdue’s graduate paragraph guidance describes the warrant as the connection between support/reasoning and the thesis.

Purdue OWL | Paragraph Organization and Flow

Example:

Evidence: students in the guided-feedback condition transferred revision strategies to an unseen task.

Warrant: performance on an unseen task is stronger evidence of independent transfer than success on the practised task alone.

14. Hidden warrant is where many arguments break | 很多 argument 断在 hidden warrant

Claim:

Students liked the tool, so it improves learning.

Hidden warrant:

What learners like necessarily improves learning.

That warrant is not established.

15. Make the bridge explicit when it is contestable | 可争议 warrant 应该写出来

The more a reader could reasonably question the connection, the more explanation the warrant needs.

16. Backing supports the warrant | Backing 支持 warrant 本身

If your warrant depends on a measurement assumption, theory or established principle, show why that assumption is reasonable.

17. Qualifier controls claim strength | Qualifier 控制 claim 强度

Compare:

always;
generally;
under these conditions;
is associated with;
may contribute to;
appears to.

Qualifier 不是语言装饰，是 epistemic control。

18. Academic caution is not weakness | Academic caution 不是“没信心”

Manchester Academic Phrasebank notes that academic writing often avoids unwarranted certainty and overgeneralisation, especially when offering explanations or implications.

University of Manchester Academic Phrasebank | Being Cautious

19. Scope is part of truth | Scope 是主张真值的一部分

This method works.

versus:

In novice learners completing structured practice, the method produces a modest short-term benefit.

第二句可能更窄，但更强。

20. Narrower can be more defensible | 更窄可以更有力量

Academic maturity often looks like reducing claim scope after encountering counterevidence.

21. Counterargument is not decoration | Counterargument 不是“加一段反方”

George Mason University’s Writing Center describes counterargument as a serious opposing position that should be represented fairly; acknowledging it can strengthen an argument when the writer responds honestly rather than caricaturing the opposition.

George Mason University | All About Counterarguments

22. Represent the strongest fair objection | 处理最强公平 objection

Do not choose:

Some people hate technology.

if the real objection is:

The observed learning gain may be caused by additional teacher support rather than the technology itself.

23. Steelman before rebuttal | 先 steelman，再 rebut

Ask:

Would a knowledgeable person who holds this objection recognise my version as fair?

24. Straw man wastes the counterargument section | Straw man 只是在浪费篇幅

If you weaken the opposing argument before answering it, you learn nothing about your thesis.

25. Counterargument can attack different layers | Counterargument 可以攻击不同层

Target	Objection
Fact	The measurement is wrong.
Inference	The evidence does not imply that conclusion.
Causation	An alternative cause explains the pattern.
Scope	The result does not generalise.
Criterion	You are judging success by the wrong standard.
Value	The trade-off is unacceptable.
Policy	The proposal is infeasible despite the benefit.

26. Diagnose before rebutting | 先诊断 objection 类型

Different objections need different replies.

A data error needs correction.

An alternative explanation needs comparative evidence.

A value disagreement needs explicit criteria.

27. Counterevidence is stronger than mere disagreement | Counterevidence 比“不同意见”更重要

Counterevidence is evidence that weakens, limits or contradicts your claim.

It may come from:

a failed replication;
a subgroup with opposite effect;
a delayed null result;
a contradictory dataset;
a counterexample;
a measurement failure;
an alternative causal design.

28. Treat counterevidence as data about your claim | 把反证当成 claim 的诊断数据

Do not ask first:

How can I defeat this?

Ask:

What does this require me to change?

29. Four possible responses to counterevidence | 反证出现后有四种主要响应

Reject it with reasons.
Accommodate it as a boundary condition.
Revise the mechanism.
Abandon or replace the thesis.

30. Rejecting counterevidence requires evidence too | 拒绝反证也要给理由

“That study is wrong” is not an argument.

Show:

measurement problem;
design mismatch;
population irrelevance;
statistical imprecision;
confounding;
misinterpretation.

31. Accommodating counterevidence often improves the thesis | Accommodation 常常让 thesis 更强

Original:

Retrieval practice improves retention.

After counterevidence:

Retrieval practice improves retention most reliably when recall succeeds often enough for corrective feedback to be useful.

32. Counterexample tests universal claims | Counterexample 专门打 broad claim

If claim = “all”, one valid counterexample can destroy universality.

But it may not destroy a probabilistic claim such as “typically” or “tends to”.

33. Use counterexamples to discover boundary conditions | 用 counterexample 找 boundary

How Counterexamples Test Broad Claims and Reveal Their Limits

34. Exceptions do not always refute patterns | Exception 不自动推翻 trend

One exceptional case may show:

universality is false;
mechanism has an exception;
measurement is noisy;
subgroups matter.

35. Alternative explanation is not the same as counterevidence | Alternative explanation ≠ counterevidence

Counterevidence says:

Your claim fits the data poorly.

Alternative explanation says:

The data may be real, but another mechanism could produce them.

36. Causal arguments must compare explanations | Causal argument 需要比较 explanations

Observation:

Students who use the tool more often score higher.

Possible explanations:

tool use improves performance;
stronger students choose the tool more;
teacher support increases both tool use and performance;
motivation drives both.

37. “After” is not “because of” | after 不等于 because of

Temporal sequence is necessary for many causal claims but not sufficient.

38. Confounding creates rival causal stories | Confounding 会制造 rival story

If a third factor influences both proposed cause and outcome, simple association is insufficient.

39. Reverse causality | 反向因果

Maybe X causes Y.

Maybe Y causes X.

Maybe both occur.

40. Selection effects | Selection 会改变 observed pattern

People who choose an intervention may differ before the intervention begins.

41. Regression to the mean | 极端值自然回归可制造假 improvement

If intervention begins after unusually poor performance, later improvement may partly reflect normal variation.

42. History effects | 同期事件也可能造成变化

A policy, teacher change or external event may coincide with intervention.

43. Measurement change | 测量系统变化也会制造“效果”

If outcome measurement changes before and after, comparison can become invalid.

44. Attrition | 谁退出也会改变结果

If weaker participants leave, average performance can rise without true individual improvement.

45. Mechanism evidence strengthens causal reasoning | Mechanism evidence 帮助 causal argument

But mechanism plausibility alone does not establish the effect size.

46. Triangulation can increase confidence | Triangulation 可以增加信心

Different methods that converge on the same causal story can strengthen interpretation, provided their errors are not all dependent on the same assumption.

47. Independent replication matters | 独立 replication 比重复同一数据更重要

Five analyses of one dataset are not five independent replications.

48. Induction | Induction 从 cases 到 generalisation

Ask:

How representative are the cases?
How many independent observations?
How stable is the pattern?
What counterexamples exist?

49. Deduction | Deduction 从 premises 到 conclusion

Ask:

Are premises true/acceptable?
Does conclusion logically follow?
Are hidden premises valid?

50. Abduction | Abduction 选择最佳解释

Many research arguments are abductive:

Given these observations, which explanation best accounts for the pattern?

Compare explanations by:

fit;
simplicity;
scope;
predictive success;
consistency with established evidence;
ability to explain anomalies.

51. Best explanation is still provisional | “最佳解释”仍是 provisional

Another explanation may later outperform it.

52. Interpretive argument has different counterevidence | Interpretive argument 的反证不同

In humanities or textual analysis, counterevidence may be:

a passage your reading cannot explain;
historical context inconsistent with interpretation;
a rival interpretation with greater explanatory reach;
selective quotation;
genre conventions your reading ignores.

53. Interpretive warrants matter | Textual evidence 也需要 warrant

Quotation does not interpret itself.

Explain why the language, structure, context or pattern supports your reading.

54. The strongest rival reading deserves attention | Rival interpretation 也要 steelman

Ask whether your interpretation explains more of the text with fewer special exceptions.

55. Historical argument | Historical claim 需要 source/context control

Counterevidence may include:

contradictory primary records;
chronology that breaks causal sequence;
selection bias in surviving archives;
presentism;
alternative institutional explanation.

56. Normative argument | Normative claim 需要 criteria

This policy is unfair.

Fair according to what principle?

Equality?

Need?

Contribution?

Rights?

Utility?

Normative argument becomes stronger when criteria are explicit.

57. Policy argument | Policy argument 不等于“effect is positive”

Policy decision often requires:

goal;
effectiveness;
cost;
feasibility;
equity;
risk;
reversibility;
alternatives.

58. Policy counterargument often attacks trade-offs | Policy objection 常攻击 trade-off

Even if it works, is it worth the cost?

59. Value conflict cannot always be solved by more data | Value disagreement 不一定能靠更多数字解决

Data can clarify consequences.

They cannot by themselves decide which value should dominate.

60. Analogy argument | Analogy 需要 relevant similarity

Ask:

Which features are similar?
Which features matter to the conclusion?
Which differences break the analogy?

61. Authority argument | Expert evidence 不是“专家说了所以是真的”

Evaluate:

expertise relevant?
field consensus?
source independence?
evidence cited?
conflict of interest?

62. Statistics are not self-interpreting | 数字也需要 argument

A percentage needs denominator.

An average needs distribution/context.

An effect size needs practical interpretation.

A correlation needs causal caution.

63. Statistical significance is not practical importance | statistical significance ≠ practical importance

And lack of statistical significance is not automatic equivalence.

64. Base rates matter | Base rate 会改变解释

A dramatic relative change can still represent a small absolute change.

65. Sample size affects precision | 样本大小影响 uncertainty

Do not treat a noisy small-sample estimate as exact.

66. Measurement validity precedes argument about numbers | 先问测的是什么

A precise number for the wrong construct does not solve the argument.

67. Evidence relevance | Evidence 必须与 claim 对齐

Ask:

If this evidence were true, would it actually make my claim more likely?

68. Quantity cannot repair irrelevance | 不相关 evidence 再多也没用

Ten satisfaction surveys do not prove causal learning improvement.

69. Credibility is not relevance | credible source ≠ relevant evidence

A high-quality source can still answer a different question.

70. Directness matters | Direct evidence 与 indirect evidence 要分开

Directness asks how closely the evidence matches your population, outcome and intervention.

71. Evidence hierarchy is question-dependent | Evidence hierarchy 不是 universal ladder

A randomised trial is powerful for causal effect.

An ethnography may be more informative for lived process.

A corpus may be best for frequency/use pattern.

72. Argument quality depends on fit, not prestige | Argument 看 fit，不只看 prestige

The best evidence is the evidence that legitimately answers the relevant subquestion.

73. Build an argument map | 建立 argument map

Node	Content
Main claim	what you want established
Reason 1	first support line
Evidence	data/source
Warrant	why evidence supports reason/claim
Counterevidence	what pushes against it
Reply	reject/accommodate/revise
Qualifier	final scope/strength

74. Map before prose when argument is complex | 复杂 argument 先画图再写

This exposes missing warrants and unsupported leaps.

75. Use an objection column | 给 argument map 加 objection column

For every major reason:

What is the strongest serious objection?

76. Use an evidence-that-would-change-me column | 再加“什么会改变我”

This forces intellectual openness.

77. Distinguish rebuttal from dismissal | Rebuttal ≠ dismiss

Dismissal:

This objection is unconvincing.

Rebuttal:

The objection correctly identifies self-selection, but the same effect appears in the randomised replication, so selection alone cannot explain the full pattern.

78. Rebuttal should answer the mechanism of the objection | 回应 objection 的逻辑核心

Do not merely repeat your original thesis louder.

79. Concession | Concession 承认对方成立的部分

The critics are correct that the study cannot establish long-term transfer.

80. Concession can narrow without surrendering | 承认局部不等于放弃整体

However, that limitation does not undermine the evidence for immediate performance.

81. Accommodation | Accommodation 把 objection 放进新 thesis

The intervention should therefore be understood as improving immediate assisted performance rather than proven independent transfer.

82. Concession is powerful when it changes the claim | 真 concession 会有代价

If you “concede” something irrelevant, it is rhetorical decoration.

83. Counterargument section does not always belong at the end | Counterargument 放哪里取决于结构

Possible placements:

immediately after vulnerable claim;
inside each body section;
dedicated counterargument section;
discussion/conclusion.

84. Local objections belong near local claims | 局部 objection 最好靠近 claim

This reduces reader memory load.

85. Global objection may need its own section | 全局 objection 可单独处理

Example:

All observed gains may be explained by teacher attention rather than intervention content.

86. Order objections by seriousness | Objection 按重要性排序

Do not spend 500 words on a weak criticism and one sentence on the strongest one.

87. The “hostile reviewer” audit | Hostile reviewer 审计

Imagine a knowledgeable reviewer who wants to reject your argument fairly.

What would they attack first?

88. The “friendly reviewer” audit | Friendly reviewer 也重要

A friendly reader may accept too quickly and fail to notice an overclaim.

Ask them where the argument seems strongest—then inspect whether evidence really supports that confidence.

89. Argument has a burden of proof | Claim 越强，证据负担越高

May help needs less evidence than always works.

Associated with needs less causal proof than causes.

90. Extraordinary scope needs extraordinary support | Scope 越广，support 越强

Universal claims across populations, time and contexts are difficult to justify.

91. Negative claim also has burden | “没有效果”也需要证据

Failure to detect an effect is not always proof of no effect.

92. Absence of evidence | absence of evidence

No study has shown X may mean evidence is missing.

93. Evidence of absence | evidence of absence

Strong, well-powered evidence can support a conclusion that any effect is likely small or absent within a defined range.

94. Distinguish “not established” from “false” | not established ≠ false

This single distinction improves many academic arguments.

95. Distinguish “possible” from “supported” | possible ≠ supported

You can invent many possible explanations.

The argument must compare evidence for them.

96. Distinguish “plausible” from “probable” | plausible ≠ probable

Plausibility is only one criterion.

97. Distinguish “consistent with” from “demonstrates” | consistent with ≠ demonstrates

Many different theories can be consistent with the same observation.

98. Distinguish explanation from prediction | Explanation and prediction

A theory that explains past data but makes no risky predictions may be hard to distinguish from alternatives.

99. Risky prediction strengthens theory tests | 有区分力的 prediction 更有价值

Ask:

What would this explanation predict that its rival would not?

100. Argument as model comparison | 高级 argument 常常是 model comparison

Instead of:

Is my theory true?

Ask:

Which explanation best accounts for the evidence, and where does it fail?

101. Alternative explanations need equal standards | Rival explanation 要用同一标准

Do not demand perfect proof from the rival while accepting weak support for your own explanation.

102. Symmetry is a credibility test | 对自己和对手使用对称标准

Same sample-size concern.

Same causal standard.

Same source-quality standard.

103. Motivated reasoning is an argument risk | Motivated reasoning 会扭曲选择

Warning signs:

counterevidence dismissed faster;
supporting evidence described more strongly;
weak objections selected;
scope expands only in preferred direction.

104. Precommit to evaluation criteria | 先定 criteria 再看结果

If possible, decide what would count as strong evidence before seeing which side it supports.

105. Search for disconfirmation deliberately | 主动搜索 disconfirmation

Search terms:

failed replication;
null result;
critique;
limitations;
alternative explanation;
boundary condition.

106. Do not only search “benefits of X” | 搜索词本身会造成 confirmation bias

Ask neutral questions.

107. Critical reading looks for how the argument works | Critical reading 读“论证怎样运作”

The University of Toronto’s critical-reading guidance recommends examining central claims, context, reasoning, theory, method, evidence, conclusions, assumptions and alternative interpretations.

University of Toronto | Critical Reading Towards Critical Writing

108. Reverse-engineer source arguments | 把来源拆回 argument map

For each key source:

claim;
reason;
evidence;
warrant;
assumption;
qualification;
counterargument;
conclusion.

109. Then compare source argument structures | 比较来源的 argument architecture

Sometimes studies differ not because data conflict, but because warrants differ.

110. Hidden assumptions can organise a whole debate | Hidden assumption 可以组织整个 field

Example:

One side assumes performance on assisted tasks represents learning; the other requires independent transfer.

111. Make your assumptions explicit | 自己的 assumption 也要写出来

Especially:

definition of success;
measurement interpretation;
value criteria;
causal model;
generalisation boundary.

112. Paragraphs should argue, not store evidence | 段落不是 evidence warehouse

George Mason’s paragraphing guidance stresses that evidence must be connected to argument through analysis.

George Mason University | Paragraphing

113. Paragraph architecture | Argument paragraph 架构

CLAIM → REASON → EVIDENCE → WARRANT → LIMIT/COUNTER → RETURN.

114. Example paragraph | 示例段落

Structured feedback is likely to improve revision more reliably than feedback volume alone. In the fictional controlled comparison, students receiving fewer but actionable comments transferred more revision strategies to an unseen task than students receiving a larger number of evaluative comments. The important distinction is usability: evidence of transfer suggests that learners were able to apply the feedback process beyond the original text. This conclusion should remain limited, however, because the study lasted only six weeks and did not measure performance after support was removed. The evidence therefore supports the value of actionable feedback for near transfer more strongly than a claim about durable independent writing.

115. Evidence without warrant produces “citation dump” | 没有 warrant 就会变 citation dump

After every important piece of evidence ask:

So what exactly does this establish?

116. Warrant without evidence produces assertion | 只有 warrant 没 evidence 只是 reasoning story

Both are needed.

117. Counterevidence inside paragraph can increase trust | 段内处理 counterevidence 可以增加可信度

Readers see that you are not hiding the weak point.

118. Counterargument paragraph architecture | Counterargument 段架构

STRONG OBJECTION → WHY IT MATTERS → CONCESSION → RESPONSE → REVISED CLAIM.

119. Example counterargument paragraph | 示例

A serious objection is that the observed improvement may reflect additional teacher attention rather than the intervention itself. This matters because the treatment group received two extra conferences during the trial. The objection therefore weakens any claim that the intervention alone caused the full improvement. However, a second study using equal teacher-contact time still reports a smaller positive effect. Taken together, the evidence supports a narrower claim: teacher attention explains part, but probably not all, of the observed difference.

120. This is stronger than “some may disagree” | 不要写空洞 some may argue

Name the exact objection and why it has force.

121. Introductions should state the problem before declaring the thesis | Introduction 先构造 problem

Reader needs to understand:

question;
stakes;
relevant debate;
your position.

122. Thesis can be provisional in research process | 研究过程中 thesis 可以变化

The final paper presents a stable thesis, but your drafting process should permit revision.

123. Do not promise more than the paper can prove | Introduction 不要 overpromise

Weak:

This paper proves that…

Better:

This paper argues that…

or:

The evidence supports the conclusion that…

124. Conclusions should report the final calibrated thesis | Conclusion 应该是“经过测试后的 thesis”

Often the best conclusion is narrower than the initial hypothesis.

125. A good conclusion distinguishes three things | 好结论分三层

What the argument establishes.
What it does not establish.
What follows next.

126. Do not introduce new certainty in the conclusion | 结论不要突然变得更肯定

If body says “may”, conclusion cannot silently become “does”.

127. Conclusions can include unresolved objection | 可以保留 unresolved objection

Academic honesty does not require artificial closure.

128. The “minimum defensible claim” test | Minimum defensible claim 测试

Ask:

What is the strongest version of my claim that all my evidence can still support after counterevidence is included?

129. The “maximum honest uncertainty” test | 最大诚实 uncertainty 测试

Ask:

What important uncertainty would a reasonable reader need to know before accepting my conclusion?

130. The “alternative explanation” test | 替代解释测试

List at least two serious rivals.

For each:

what evidence it explains;
what it fails to explain;
what evidence would distinguish it from yours.

131. The “strongest counterevidence” test | 最强反证测试

Find the source or fact that creates the greatest problem for your thesis.

Put it into the argument.

132. The “scope” test | Scope 测试

Highlight every:

all;
always;
students;
people;
effective;
causes;
proves.

Ask whether the evidence covers that scope.

133. The “warrant” test | Warrant 测试

Between every evidence sentence and conclusion, write:

This supports the claim because…

If you cannot finish the sentence, the logical bridge may be missing.

134. The “symmetry” test | 对称标准测试

Would you accept the same quality of evidence if it supported the opposing position?

135. The “best opponent” test | 最强 opponent 测试

Would an expert critic recognise the objection you chose as serious?

136. The “revision” test | Revision 测试

What did counterevidence force you to change?

If answer = nothing, inspect whether you actually engaged with it.

137. Worked case: fictional study-skills argument | 完整案例：虚构学习策略 argument

All data below are fictional teaching material.

Question:

Should a university replace rereading workshops with retrieval-practice workshops?

138. Initial evidence | 初始证据

Study A: randomised trial, 180 students, higher delayed quiz score after retrieval.
Study B: observational course data, retrieval frequency associated with final grade.
Student survey: retrieval workshop rated more difficult and less enjoyable.

139. Weak thesis | 弱 thesis

Retrieval practice is better than rereading and should replace it.

Problems:

“better” undefined;
policy leap;
cost/feasibility ignored;
survey experience ignored;
one method may not suit all learning goals.

140. Stronger provisional thesis | 更强 provisional thesis

For factual material requiring delayed recall, retrieval-practice workshops should become the default strategy, while rereading remains available as a supporting method for initial familiarisation.

141. Reason 1 | 理由 1

Retrieval produces stronger delayed recall in Study A.

142. Warrant 1 | Warrant 1

The programme goal is durable retention rather than immediate ease, so delayed performance is the more relevant outcome.

143. Counterargument 1 | Objection 1

Students find retrieval harder and less enjoyable.

144. Concession 1 | Concession 1

This matters for participation and adherence.

145. Response 1 | Response 1

Difficulty alone does not show lower learning value, but implementation should include explanation, progressive difficulty and monitoring of dropout.

146. Counterevidence 2 | 反证 2

A replication among advanced students shows little difference between methods.

147. Thesis revision | Thesis 修订

Retrieval practice should be the default for learners and tasks where delayed factual recall is the target, but the evidence does not justify replacing rereading universally, particularly among advanced learners or for tasks focused on initial comprehension.

这个 thesis 更窄，却更难被公平反驳。

148. Alternative explanation | 替代解释

Maybe retrieval workshops receive better tutor facilitation.

Need evidence with equal instructional time/quality.

149. Policy condition | Policy 条件

Adoption should depend on tutor training and learner participation, not effect size alone.

150. Final calibrated conclusion | 最终结论

The strongest current case is for targeted adoption rather than complete replacement.

这就是 argument surviving counterevidence。

151. Practice A: claim type | 练习 A

Classify:

Students who sleep longer perform better.

Descriptive association or causal claim?

152. Practice B: expose warrant | 练习 B

Evidence:

Most students prefer digital feedback.

Claim:

Digital feedback improves learning.

Write the hidden warrant and test it.

153. Practice C: steelman | 练习 C

Thesis:

Remote work improves productivity in some knowledge roles.

Write the strongest fair objection, not a caricature.

154. Practice D: counterevidence | 练习 D

Your preferred method works immediately but not at delayed follow-up.

Revise the claim.

155. Practice E: alternative cause | 练习 E

Users of a tutoring service have higher scores.

List three plausible alternative explanations.

156. Practice F: concession | 练习 F

Write a real concession that changes your argument.

157. Practice G: rebuttal | 练习 G

Respond to an objection using evidence, not assertion.

158. Practice H: scope repair | 练习 H

Repair:

Group work improves learning.

into a defensible scoped claim.

159. Practice I: policy leap | 练习 I

Evidence establishes benefit.

What additional evidence is needed before recommending mandatory adoption?

160. Practice J: interpretive rival | 练习 J

Give one textual passage that your interpretation explains poorly and show how you would accommodate it.

161. Model answers | 示范答案

A: As written, it reports an association. It becomes causal only if the writer claims that longer sleep produces better performance.

B: Hidden warrant: learner preference reliably predicts learning effectiveness. That relationship requires evidence and is unlikely to hold universally.

C: Productivity estimates may be biased because high-autonomy workers are more likely to receive remote-work options, while collaboration costs appear in team-level outcomes not captured by individual output metrics.

D: The method improves immediate performance, but current evidence does not establish durable retention.

E: prior achievement; family resources; learner motivation; selective enrolment; additional study time.

H: Structured group work can improve conceptual explanation when roles require each learner to contribute and feedback prevents persistent misconceptions.

I: cost, feasibility, equity, adverse effects, alternatives, reversibility, implementation capacity and long-term outcome.

162. The 20-minute argument drill | 20 分钟训练

Time	Task
3 min	write question + provisional thesis
4 min	map claim/reason/evidence/warrant
4 min	write strongest counterargument
4 min	add counterevidence/alternative explanation
5 min	revise final thesis + qualifier

163. The 45-minute growth session | 45 分钟增长模式

Time	Task
5 min	question/stakes
8 min	argument map
10 min	evidence + warrant
8 min	steelman objection
8 min	response/accommodation
6 min	scope/certainty audit

164. The 90-minute deep session | 90 分钟深度训练

Time	Task
10 min	question + claim type
15 min	evidence matrix
15 min	warrants/assumptions
15 min	counterevidence search
15 min	alternative explanations
10 min	two argument paragraphs
10 min	final calibrated conclusion

165. Seven-day argument cycle | 七天 argument cycle

Day 1: question + claim type.
Day 2: reason/evidence/warrant.
Day 3: causal and alternative explanation.
Day 4: counterargument and steelman.
Day 5: counterevidence and accommodation.
Day 6: paragraph/conclusion architecture.
Day 7: unseen full argument benchmark.

166. Twelve-week C1–C2 argument progression | 12 周路线

Weeks	Focus	Output
1–2	questions, thesis, claim types	argument maps
3–4	evidence, warrants, assumptions	analytical paragraphs
5–6	causal reasoning, rival explanations	model comparisons
7–8	counterarguments, counterevidence, rebuttal	steelman-response sets
9–10	scope, qualifiers, policy/normative argument	calibrated essays
11–12	independent disciplinary transfer	full argument portfolio

167. Monthly benchmark | 每月基准任务

Use one serious question and at least four credible sources.

Produce:

one exact research question;
one provisional thesis;
argument map with three reasons;
evidence + warrant for each;
two serious alternative explanations;
strongest counterargument;
strongest counterevidence;
one genuine concession;
one rebuttal;
one revised final thesis;
250-word conclusion stating established, unestablished and next evidence.

168. Evaluate reasoning, not just style | 不要只评价“写得像不像学术英语”

Track:

question precision;
claim scope;
evidence relevance;
warrant clarity;
assumption control;
counterargument fairness;
counterevidence engagement;
alternative-explanation comparison;
qualifier accuracy;
conclusion calibration;
source traceability;
language/register control.

169. First weak-link diagnosis | 第一个薄弱环节诊断

Symptom	Likely weak link	Repair
lots of evidence, weak conclusion	warrant missing	write “supports because…”
argument sounds certain	scope/qualifier drift	restore population/time/condition
counterargument feels fake	straw man	steelman strongest objection
counterevidence changes nothing	motivated reasoning	force thesis revision
causal claim from correlation	alternative explanations missing	compare causal models
policy conclusion appears suddenly	decision criteria missing	add cost/risk/feasibility/trade-offs
citation dump	analysis/warrant weak	evidence → interpretation → claim
criticism sounds hostile	objection criteria vague	criticise inference/method/scope precisely

170. What not to do | 不要这样写 advanced argument

Do not choose a thesis and then search only for supporting evidence.
Do not select the weakest counterargument.
Do not call disagreement counterevidence unless evidence is involved.
Do not use quotation as a substitute for warrant.
Do not use “proves” when evidence only suggests.
Do not dismiss alternative explanations because they are inconvenient.
Do not convert “not established” into “false”.
Do not move from effectiveness to policy without feasibility and trade-offs.
Do not change standards when evidence supports your preferred conclusion.
Do not protect a thesis that the evidence no longer supports.

171. Research and reference floor | 研究与参考基础

University of Toronto | General Advice on Academic Essay-Writing — question, thesis, reasoning/evidence and testing the provisional thesis against objections.
University of Toronto | Using Thesis Statements — definite, limited and supportable thesis claims.
University of Toronto | Critical Reading Towards Critical Writing — claims, reasoning, theory, method, evidence, assumptions and rival interpretations.
George Mason University | All About Counterarguments — fair representation, concession and response.
George Mason University | Paragraphing — evidence, analysis and argument connections within paragraphs.
Purdue OWL | Organizing Your Argument — claim, reasons, evidence, warrants, counterargument and rebuttal.
Purdue OWL | Paragraph Organization and Flow — warrant as the connection between evidence/reasoning and thesis.
University of Manchester Academic Phrasebank | Being Cautious — qualifiers and calibrated academic claims.
Academic Phrasebank | Being Critical — disciplined evaluation of claims and evidence.

172. Canonical eduKate routes | eduKate 主页面路由

173. SEO language map | 本课自然覆盖的搜索意图

This lesson naturally serves learners searching for academic argument writing, counterargument, counterevidence, claim evidence warrant, Toulmin argument, rebuttal, concession, alternative explanation, causal reasoning, academic thesis statement, critical academic writing, argumentative research paper, argument paragraph, C1 academic writing, C2 academic English, academic writing for Chinese students, academic English for Mandarin speakers, 学术论证, 反证, 反方论点, thesis 怎么写, claim evidence warrant, 学术英语 argument, counterargument 怎么写, rebuttal 怎么写, causal argument, 中文母语学术英语 and C1 C2 academic writing.

174. The one-page counterevidence-resistant argument operating system | 一页抗反证 argument 系统

Write one exact question. 先锁定问题。
Name the claim type. descriptive/causal/interpretive/evaluative/policy/theoretical。
Draft a provisional thesis. 可修改，不先保护。
Map reasons. 为什么？
Attach relevant evidence. 每条 evidence 对应哪条 reason？
Write the warrant. 为什么 evidence 支持 claim？
Expose assumptions. 哪些 premise 隐藏着？
Calibrate scope and certainty. 谁、何时、什么条件、有多确定。
Steelman the strongest objection. 不打 straw man。
Find actual counterevidence. 不只找不同意见。
List rival explanations. 尤其 causal claim。
Apply symmetric standards. 对自己和对手同一标准。
Concede what is true. 真 concession 会改变 claim。
Rebut only what evidence allows. 不重复 thesis。
Accommodate boundary conditions. 用反证缩小或改写 thesis。
Write the minimum defensible final claim. 最强但仍诚实的结论。
State what remains unresolved. 不制造 artificial closure。
State what evidence would change the conclusion. 让 argument 保持可检验。

175. Final assignment | 最终作业

Choose one serious academic question in your field.

Use at least four credible sources, including one that challenges your preferred position.

Complete this chain:

Write the exact question.
Classify the main claim type.
Write a provisional thesis.
Build an argument map with at least three reasons.
Attach evidence to each reason.
Write the warrant for each evidence–claim connection.
List hidden assumptions.
Identify the strongest fair counterargument.
Identify the strongest counterevidence.
Write at least two alternative explanations.
State what evidence would distinguish the explanations.
Write one genuine concession.
Write one evidence-based rebuttal.
Revise the thesis after the objections.
Write two full argument paragraphs.
Write one counterargument paragraph.
Write a 200–300 word conclusion separating established, unestablished and next evidence.
Run the scope test.
Run the symmetry test.
Run the “what would change my mind?” test.

Then ask:

If the strongest counterevidence were placed on the first page instead of hidden near the end, would my argument still deserve the same conclusion?

如果把最强反证放到第一页，而不是藏在最后，我的 argument 还配得上同一个结论吗？

If yes, the argument is becoming robust.

If no, do not hide the evidence. Change the thesis.

Continue | 继续

The next lesson will move from argument structure into epistemic precision: how advanced academic writers calibrate certainty, probability, limitation and confidence without becoming vague or timid.

下一课进入学术英语里非常关键的一层：怎样精准控制 certainty、probability、limitation 与 confidence，让文章既不 overclaim，也不因为 hedge 太多而变得含糊。

Next: EDKS-ADV-ZH-0014 · Lesson No.014 · Calibrate Certainty Without Becoming Vague · 精准控制学术英语中的确定程度

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.013 | Build an Academic Argument That Survives Counterevidence | 第013课：构建经得住反证的学术论证

Build an Academic Argument That Survives Counterevidence | 构建经得住反证的学术论证

Before we begin: this lesson integrates, but does not replace, existing argument owners | 开始之前：本课整合现有 argument owners

1. Academic argument begins with a question | Academic argument 从问题开始

2. Topic is not question | Topic 不是 question

3. Question is not thesis | Question 也不是 thesis

4. Good theses are limited and arguable | 好 thesis 有边界，也可争辩

5. A claim should expose itself to evidence | Claim 必须允许 evidence 影响它

6. If nothing could change your mind, you do not yet have an academic claim | 如果什么都不能改变你的结论，就不是学术 claim

7. Separate claim type | 先判断 claim 属于哪一类

8. Claim type controls evidence | Claim type 决定证据任务

9. Purdue’s argument organiser provides a useful architecture | Purdue 提供一个实用 argument 架构

10. Claim | Claim 是你要让读者接受什么

11. Reason | Reason 回答 why

12. Evidence | Evidence 给出可检查支持

13. Warrant | Warrant 解释“为什么这个 evidence 能支持这个 claim”

14. Hidden warrant is where many arguments break | 很多 argument 断在 hidden warrant

15. Make the bridge explicit when it is contestable | 可争议 warrant 应该写出来

16. Backing supports the warrant | Backing 支持 warrant 本身

17. Qualifier controls claim strength | Qualifier 控制 claim 强度

18. Academic caution is not weakness | Academic caution 不是“没信心”

19. Scope is part of truth | Scope 是主张真值的一部分

20. Narrower can be more defensible | 更窄可以更有力量

21. Counterargument is not decoration | Counterargument 不是“加一段反方”

22. Represent the strongest fair objection | 处理最强公平 objection

23. Steelman before rebuttal | 先 steelman，再 rebut

24. Straw man wastes the counterargument section | Straw man 只是在浪费篇幅

25. Counterargument can attack different layers | Counterargument 可以攻击不同层

26. Diagnose before rebutting | 先诊断 objection 类型

27. Counterevidence is stronger than mere disagreement | Counterevidence 比“不同意见”更重要

28. Treat counterevidence as data about your claim | 把反证当成 claim 的诊断数据

29. Four possible responses to counterevidence | 反证出现后有四种主要响应

30. Rejecting counterevidence requires evidence too | 拒绝反证也要给理由

31. Accommodating counterevidence often improves the thesis | Accommodation 常常让 thesis 更强

32. Counterexample tests universal claims | Counterexample 专门打 broad claim

33. Use counterexamples to discover boundary conditions | 用 counterexample 找 boundary

34. Exceptions do not always refute patterns | Exception 不自动推翻 trend

35. Alternative explanation is not the same as counterevidence | Alternative explanation ≠ counterevidence

36. Causal arguments must compare explanations | Causal argument 需要比较 explanations

37. “After” is not “because of” | after 不等于 because of

38. Confounding creates rival causal stories | Confounding 会制造 rival story

39. Reverse causality | 反向因果

40. Selection effects | Selection 会改变 observed pattern

41. Regression to the mean | 极端值自然回归可制造假 improvement

42. History effects | 同期事件也可能造成变化

43. Measurement change | 测量系统变化也会制造“效果”

44. Attrition | 谁退出也会改变结果

45. Mechanism evidence strengthens causal reasoning | Mechanism evidence 帮助 causal argument

46. Triangulation can increase confidence | Triangulation 可以增加信心

47. Independent replication matters | 独立 replication 比重复同一数据更重要

48. Induction | Induction 从 cases 到 generalisation

49. Deduction | Deduction 从 premises 到 conclusion

50. Abduction | Abduction 选择最佳解释

51. Best explanation is still provisional | “最佳解释”仍是 provisional

52. Interpretive argument has different counterevidence | Interpretive argument 的反证不同

53. Interpretive warrants matter | Textual evidence 也需要 warrant

54. The strongest rival reading deserves attention | Rival interpretation 也要 steelman

55. Historical argument | Historical claim 需要 source/context control

56. Normative argument | Normative claim 需要 criteria

57. Policy argument | Policy argument 不等于“effect is positive”

58. Policy counterargument often attacks trade-offs | Policy objection 常攻击 trade-off

59. Value conflict cannot always be solved by more data | Value disagreement 不一定能靠更多数字解决

60. Analogy argument | Analogy 需要 relevant similarity

61. Authority argument | Expert evidence 不是“专家说了所以是真的”

62. Statistics are not self-interpreting | 数字也需要 argument

63. Statistical significance is not practical importance | statistical significance ≠ practical importance

64. Base rates matter | Base rate 会改变解释

65. Sample size affects precision | 样本大小影响 uncertainty

66. Measurement validity precedes argument about numbers | 先问测的是什么

67. Evidence relevance | Evidence 必须与 claim 对齐

68. Quantity cannot repair irrelevance | 不相关 evidence 再多也没用

69. Credibility is not relevance | credible source ≠ relevant evidence

70. Directness matters | Direct evidence 与 indirect evidence 要分开

71. Evidence hierarchy is question-dependent | Evidence hierarchy 不是 universal ladder

72. Argument quality depends on fit, not prestige | Argument 看 fit，不只看 prestige

73. Build an argument map | 建立 argument map

74. Map before prose when argument is complex | 复杂 argument 先画图再写

75. Use an objection column | 给 argument map 加 objection column

76. Use an evidence-that-would-change-me column | 再加“什么会改变我”