Series ID: EDKS-BASIC-ZH-0101 · Lesson No.101 · Expert Maintenance: Research Reading

Research Reading and Verification | 研究阅读与验证

Advanced English becomes more useful when it helps you verify what is true, not merely understand what has been written.

Research reading requires two linked skills: understanding the source accurately and checking whether the source deserves the weight you are giving it.

专家级阅读不能停在“我看懂了”。下一步是：这是谁写的？证据是什么？时间是否过期？方法是否支持这个结论？有没有原始来源？

Chinese Edition Hub · 中文版入口 · ← Lesson No.100 · Mastery Review

---

The verification map | 验证地图

• claim
• source
• date
• method
• evidence
• scope
• original source
• independent confirmation
• uncertainty
• citation trail

1. Start with the claim | 先找主张

Do not verify an entire article at once. Identify the specific claim that matters.

2. Claim type | 主张类型

• fact
• estimate
• interpretation
• prediction
• recommendation
• opinion

3. Verification depends on claim type | 验证方式取决于主张类型

A date can often be checked directly. A causal explanation requires method evidence. A recommendation may depend partly on values.

4. Find the source | 找来源

Ask:

• Who produced this?
• Where was it published?
• What expertise is relevant?

5. Source authority is local, not universal | 权威有领域边界

An expert in one field is not automatically authoritative in another.

6. Primary source | 一手来源

Primary sources include:

• original study
• official document
• dataset
• direct transcript
• first-hand report

7. Secondary source | 二手来源

Secondary sources interpret or summarise primary material.

8. Tertiary source | 三手来源

Reference summaries can be useful for orientation but may compress important details.

9. Move upstream | 向上追溯

If an article says “research shows…”, find the research where possible.

10. Citation trail | 引用链

Track who cites whom. Repeated citation of the same original source is not independent confirmation.

11. Source dependency | 来源依赖

Five articles repeating one press release are not five independent sources.

12. Independent convergence | 独立收敛

Confidence increases when genuinely independent high-quality sources reach compatible findings.

13. Date matters | 时间很重要

Check:

• publication date
• data collection date
• update date
• time-sensitive context

14. New publication can contain old data | 新文章也可能用旧数据

Do not use page freshness as a substitute for evidence freshness.

15. Stable facts vs changing facts | 稳定事实 vs 变化事实

Historical definitions may remain stable. Prices, laws, software behaviour and rankings can change quickly.

16. Method before headline | 方法优先于标题

A headline may say “X works”. The study may only show correlation in one narrow sample.

17. Sample | 样本

Who was actually studied?

18. Population | 总体

Who is the conclusion being applied to?

19. Measurement | 测量

What was actually measured?

20. Comparison | 比较

Compared with what?

21. Confounding | 混杂

What other factor could explain the pattern?

22. Uncertainty | 不确定性

What range of values or interpretations remains plausible?

23. Scope | 范围

Where does the claim stop applying?

24. Exact wording matters | 原文措辞很重要

Compare:

• may improve
• is associated with improvement
• improves
• causes improvement

25. Quote sparingly, paraphrase carefully | 少量引用，精确改写

Use quotation only when exact wording matters. Otherwise paraphrase while preserving stance and scope.

26. Search within the source | 在来源内查找

Useful targets:

• method
• limitations
• sample
• results
• conclusion
• appendix

27. Abstract is not the whole paper | 摘要不是全文

Important caveats may appear only in methods, limitations or supplementary material.

28. Press release is not the study | 新闻稿不是研究本身

Promotional summaries may simplify uncertainty.

29. Official source is not automatically complete | 官方来源也不是自动完整

Ask whether it directly supports the specific claim you need.

30. Verification matrix | 验证矩阵

Claim	Source	Method	Date	Confidence
X increased	official dataset	measurement	current	high
X caused Y	observational article	correlation	current	low-moderate

31. Confidence is claim-specific | 置信度针对具体主张

A source may be strong for one claim and weak for another.

32. Contradictory evidence | 冲突证据

Do not hide it. Ask whether studies differ in:

• population
• definition
• method
• time period
• measurement

33. Negative evidence | 反向证据

Look actively for evidence that could weaken your preferred interpretation.

34. Confirmation bias control | 控制确认偏误

Search for the strongest credible challenge, not only supporting sources.

35. Missing evidence | 缺失证据

Absence of evidence does not automatically prove absence.

36. Provenance | 来源溯源

Can you trace the claim to an identifiable origin?

37. Unverifiable claims | 无法验证的主张

Mark them as uncertain rather than filling the gap with confidence.

38. Screenshots are evidence fragments | 截图只是证据片段

A screenshot may omit date, context or surrounding qualifiers.

39. Tables and charts | 表格与图表

Check:

• axis
• denominator
• units
• time range
• legend

40. Relative vs absolute numbers | 相对 vs 绝对数字

Both may be correct but tell different stories.

41. Search snippets | 搜索摘要

A search snippet is not a substitute for reading the source.

42. AI-generated summaries | AI 摘要

Use summaries for orientation, then verify important claims against accessible evidence.

43. Tool confidence ≠ factual confidence | 工具自信不等于事实可靠

Fluent wording can still be wrong.

44. Verification sentence frames | 验证表达

• The source directly states…
• The data support…
• The source does not establish…
• This remains uncertain because…

45. Source conflict sentence frames | 来源冲突表达

• Source A reports…, whereas Source B…
• The apparent disagreement may reflect different populations.
• The evidence remains mixed.

46. Mandarin transfer: “有研究表明”要追问是哪项研究 | 来源具体化

Do not leave research claims vague when the source can be identified.

47. Mandarin transfer: “最新”需要真实日期 | newest must be verified

Check the date rather than assuming the newest-looking result is current.

48. Research-reading routine | 研究阅读流程

1. identify claim
2. find source
3. trace original evidence
4. check date
5. inspect method
6. check scope
7. find challenge
8. write calibrated conclusion

49. Verification log | 验证日志

Record:

• claim
• source
• date
• verification status
• remaining uncertainty

50. Error clinic | 常见问题

Problem	Repair
I trust repetition.	Trace source dependency.
I trust authority globally.	Check domain relevance.
I read headline only.	Inspect method/limitations.
I ignore old data.	Check data date.
I search only for support.	Find strongest challenge.

51. First weak link diagnosis | 第一个卡点诊断

• claim unclear → claim extraction.
• source vague → provenance.
• method misunderstood → evidence literacy.
• date wrong → freshness control.
• confidence too strong → calibration.

52. Seven-day training cycle | 七天训练

Day 1	claim extraction	主张
Day 2	source tracing	溯源
Day 3	method check	方法
Day 4	date/freshness	时效
Day 5	counterevidence	反证
Day 6	verification matrix	矩阵
Day 7	three-source verification memo	综合

53. Self-test | 自测

Trace a claim to its source, check method/date/scope, find independent confirmation or challenge, and state the strongest defensible conclusion.

54. For parents and teachers | 给家长和老师

Ask advanced learners to show where a claim came from, not merely repeat it confidently.

Reward “I cannot verify this yet” when verification is genuinely incomplete.

For Mandarin speakers, replace vague “research shows” language with explicit source ownership.

55. Final real-world challenge | 最终真实任务

1. Choose one important claim.
2. Find the earliest accessible source.
3. Check date and method.
4. Identify sample and measure.
5. Find one independent source.
6. Find one credible challenge.
7. Build a verification matrix.
8. Write a 200-word conclusion.
9. Mark remaining uncertainty.
10. Explain what evidence would change your conclusion.

Next: Lesson No.102 | 下一课

The next lesson develops the language of uncertainty itself: how to communicate confidence, ranges, unknowns and decision thresholds without sounding either falsely certain or uselessly vague.

Lesson No.102 · Communicating Uncertainty and Confidence · 表达不确定性与置信度

---

Reference floor: expert-maintenance research reading. Verification is treated as part of communication, not a separate optional step.