Step 4: 緒論寫作 — Academic Writing Tutorial

概覽

為什麼緒論最難寫？

緒論 (Introduction) 是論文裡最難寫的一段。它必須在短短 3–4 個段落內，同時說服一位陌生人兩件事：(1) 你研究的問題真的重要，且 (2) 這個問題還沒有人回答。多數的「desk reject」與「reject after review」決定，正是在這裡定生死。

編輯與審稿人讀緒論的方式很功利：第一段問「這領域我為什麼要在意？」、中段問「文獻有讀夠嗎？」、最後一段問「所以你到底要做什麼、有何貢獻？」三個問題只要漏一個，就會被視為「沒準備好」。

The Introduction is the toughest section of a paper. In just 3–4 paragraphs it must convince a stranger of two things at once: (1) the problem genuinely matters, and (2) nobody has answered it yet. Most desk rejects and post-review rejects are decided right here.

Editors and reviewers read introductions instrumentally: paragraph 1 asks "why should I care about this field?", the middle asks "have they read enough?", and the last asks "so what exactly are you going to do, and what's the contribution?" Miss any one and the paper looks unprepared.

💡

核心原則：緒論不是「文獻回顧」的縮影，而是一場有方向的論證——從廣到窄、從共識到 gap、從 gap 到你要做的事。沒有方向感的緒論等於沒有緒論。 Core principle: The Introduction is not a mini literature review — it is a directed argument: broad to narrow, consensus to gap, gap to your contribution. An Introduction without that arrow is no Introduction at all.

CARS 模型

一、Swales 的 CARS 模型：三個 Move

應用語言學家 John Swales 觀察了上百篇 STEM 期刊論文後提出 CARS (Create A Research Space) 模型：作者在緒論裡其實在「為自己的研究爭一塊空間」。整個緒論可以拆成三個有先後順序的「Move」。

Applied linguist John Swales analysed hundreds of STEM articles and proposed the CARS (Create A Research Space) model: authors in the Introduction are essentially carving out a space for their own work. The entire introduction breaks into three ordered "Moves".

Move 1 — 建立領域

告訴讀者你的研究所在的領域很重要，並概述既有研究。常見子步驟：(a) claiming centrality（這個議題很關鍵）、(b) topic generalisation（給出已知共識）、(c) reviewing prior research（簡述代表性文獻）。

Tell the reader the field matters, and outline existing research. Sub-steps: (a) claiming centrality, (b) topic generalisation (state consensus), (c) reviewing prior research (cite representative work).

Move 2 — 建立利基

指出既有文獻的不足或矛盾，把 gap 挖出來。Swales 列了 4 種子步驟：(a) counter-claim（推翻先前主張）、(b) gap（指出未被探討的問題）、(c) question-raising（提出新問題）、(d) continuing tradition（延續未完成的工作）。四選一即可，但一定要有。

Surface the limits or contradictions of existing work and dig out the gap. Swales lists four sub-options: (a) counter-claim, (b) gap (identify unexplored issue), (c) question-raising (pose a new question), (d) continuing a tradition (extend unfinished work). Pick one — but you must have one.

Move 3 — 佔領利基

把你要做的事直接擺出來。子步驟：(a) outlining purpose（宣告本研究目的）、(b) listing research questions/hypotheses、(c) announcing principal findings 或方法、(d) indicating structure（簡述後續章節）。最後一段就是讀者最在意的「contribution」段。

Lay out what you are doing. Sub-steps: (a) outlining purpose, (b) listing research questions/hypotheses, (c) announcing principal findings or methods, (d) indicating structure. This last paragraph is the "contribution" paragraph reviewers most care about.

⚠️

順序不能亂：很多研究生把 Move 3 (我要做什麼) 放到第一段、Move 1 放到中間，讀起來像在自我介紹。順序錯了，論證的邏輯感就消失了。

Order matters: students often put Move 3 (what I'm doing) in paragraph 1 and Move 1 in the middle — it reads like a self-introduction. Get the order wrong and the argumentative logic vanishes.

漏斗結構

二、把 CARS 落地成「漏斗段落」

CARS 是抽象架構，實際寫的時候建議用 3–4 個段落的漏斗：每段越往下越窄、越具體，最後一段直接落到你的研究本身。

CARS is abstract; in practice write the introduction as a 3–4 paragraph funnel — each paragraph narrower and more specific than the last, ending squarely on your own study.

🌍

Para 1 — 廣的重要性

從疾病負擔、生物學意義或臨床現實切入。可引 1–3 篇代表性流病/共識文獻。避免從教科書定義開頭。

Open with disease burden, biological significance or clinical reality. Cite 1–3 representative epidemiology / consensus papers. Don't start from a textbook definition.

📚

Para 2 — 已知什麼

建立 context：過去 5–10 年的代表性研究、主要發現、技術演進。引用密度最高的一段（5–15 篇）。寫成「故事」而非清單。

Build context — representative studies of the last 5–10 years, key findings, methodological evolution. This is the citation-densest paragraph (5–15 refs). Write it as a story, not a list.

🕳️

Para 3 — 還不知道什麼 (gap)

明確指出 gap：反證、矛盾、未涵蓋的族群、樣本偏誤、缺乏外部驗證、技術限制。一定要出現「however / despite / yet / remains unclear」這類轉折詞。

Spell out the gap: counter-evidence, contradictions, missing populations, sampling bias, lack of external validation, technical limits. There must be a "however / despite / yet / remains unclear" turn here.

🎯

Para 4 — 我們做了什麼

1–2 句宣告 aim、1 句假設、1 句方法摘要、1 句預告主要結果或貢獻。這段不引文獻——它是你的工作。

One or two sentences for the aim, one for the hypothesis, one for the methods, one previewing main result/contribution. No citations in this paragraph — it is your work.

範例對照

三、好壞句子對照

❌ 開頭句 — 教科書式

「癌症是一種影響許多人的嚴重疾病。」
(沒有數字、沒有指出研究範圍、跟你的論文沒關係。)

"Cancer is a serious disease that affects many people."
(No numbers, no scope, no link to your paper.)

✅ 開頭句 — 聚焦版

「三陰性乳癌 (TNBC) 佔所有乳癌的 15–20%，缺乏標靶治療選擇，5 年存活率不到 77%。」

"Triple-negative breast cancer (TNBC) accounts for 15–20% of breast cancer cases but lacks targeted therapies, with 5-year survival below 77%."

❌ Gap 句 — 偷懶式

「目前少有研究探討此議題。」
(沒引文獻、沒說「少」是少到什麼程度、不知道哪裡缺。)

"Few studies have looked at this."
(No citations, no quantification of "few", no specifics on what is missing.)

✅ Gap 句 — 明確版

「目前已有 3 篇研究 (Smith 2022、Lin 2023、Chen 2024) 在 TNBC 中執行 scRNA-seq，但全部使用治療前切片，化療後的轉錄景觀至今仍未被刻畫。」

"Three studies (Smith 2022, Lin 2023, Chen 2024) profiled scRNA-seq in TNBC, but all used pre-treatment biopsies — the post-chemotherapy transcriptional landscape remains uncharacterized."

❌ Aim 句 — 空泛式

「本研究將討論 TNBC 與化療反應。」
(動詞 "discuss" 太弱、沒方法、沒對象、沒可驗證的目標。)

"This study will discuss TNBC and chemotherapy response."
(Verb "discuss" is weak; no method, no subjects, no verifiable target.)

✅ Aim 句 — 具體版

「我們對 12 對化療前/後 TNBC 切片執行 scRNA-seq (10x v3)，以鑑定化療誘發的細胞狀態轉移 (cell-state transitions)。」

"We applied scRNA-seq (10x v3) to 12 paired pre/post-chemotherapy TNBC biopsies to identify chemotherapy-induced cell-state transitions."

常見問題自檢

四、緒論常見問題決策樹

🌳 送投稿前的 5 個自我檢查

Q1:

第一句話是不是教科書定義？→ 是 → 太廣，整句砍掉，改用流病數字或臨床事實開頭。

Q2:

整篇緒論引用文獻 < 5 篇？→ 是 → context 不足，補 Move 1 與 Move 2 的代表性文獻。

Q3:

整篇緒論引用文獻 > 40 篇？→ 是 → 太雜（除非是 thesis）。期刊論文請濃縮，把詳細回顧搬到 thesis 的 Lit Review 章節。

Q4:

全文找不到 however / despite / yet / remains unclear？→ 是 → 你的 gap 沒寫出來，等於沒有 gap。

Q5:

最後一段沒有 aim 或 hypothesis？→ 是 → 讀者讀完不知道你要做什麼，整篇緒論失敗。

Q1:

Is the first sentence a textbook definition? → Yes → Too broad — cut it; open with an epidemiology number or clinical fact instead.

Q2:

Did you cite fewer than 5 papers? → Yes → Not enough context — add representative refs to Moves 1 and 2.

Q3:

Did you cite more than 40 papers? → Yes → Too dense (unless this is a thesis). For journal articles, condense and move detailed review into the thesis Lit Review chapter.

Q4:

No "however / despite / yet / remains unclear" anywhere? → Yes → Your gap isn't on the page — meaning you have no gap.

Q5:

Last paragraph doesn't state aim or hypothesis? → Yes → Reader leaves without knowing what you did — the introduction fails.

引用密度法則

五、各段引用密度的合理範圍

緒論的引用不該平均分布。下表是給期刊論文 (~3000 字本文) 的合理範圍。Thesis 章節可酌情提高，但分布形狀仍類似。

Citations should not be evenly spread across paragraphs. The table below shows a reasonable range for a journal article (~3000-word body). Thesis chapters can scale up but the shape stays similar.

段落	角色	引用數	說明
Para 1	廣的重要性Broad importance	低 (1–3 篇)Low (1–3)	流病、共識指引、權威綜述各 1 篇即可One epidemiology, one consensus guideline, one authoritative review
Para 2	已知什麼What's known	高 (5–15 篇)High (5–15)	代表性原始研究、主要發現的關鍵文獻Representative primary studies and key findings
Para 3	GapGap	中 (3–8 篇)Mid (3–8)	必須引「正面證據」與「反面證據」對照才有 gapYou need both supporting and counter-evidence cited to make the gap real
Para 4	我們做了什麼Our study	0 (不引文獻)0 (no citations)	這段是你的貢獻；引文獻會稀釋宣告效果This is your contribution — citations would dilute the announcement

💡

例外：若使用已發表的方法（如 Seurat、CellRanger、特定 chemistry），可在 Para 4 引一次方法原始論文；不算違反規則。 Exception: if you use a published method (e.g. Seurat, CellRanger, a specific chemistry), citing the method paper once in Para 4 is fine — it doesn't break the rule.

範本

六、可填空使用的句型範本

# 1. 流病 / 重要性開頭
[Disease] affects approximately [N] million people worldwide,
with a [X]-year mortality rate of [Y]%.

# 2. 領域共識
It is now widely accepted that [established finding]
plays a central role in [biological process / disease].

# 3. 既有研究綜述
Recent studies have demonstrated that [mechanism / phenotype]
contributes to [outcome] in [population / system]
(Author et al. 20XX; Author et al. 20XX).

# 4. 技術趨勢開頭
The advent of [technology, e.g. scRNA-seq, spatial transcriptomics]
has enabled [capability] at unprecedented resolution.

# 5. 臨床需求開頭
Despite advances in [treatment], [outcome problem]
remains a major clinical challenge in [patient group].

# 1. 「However + 對比」型
However, [prior work] has focused exclusively on [narrow scope],
leaving [broader scope] largely uncharacterized.

# 2. 「Despite + 不足」型
Despite these advances, the mechanism / impact / role
of [X] in [Y] remains unclear / controversial / poorly defined.

# 3. 「矛盾證據」型
While [Author 20XX] reported [finding A],
[Author 20XX] observed the opposite in [setting] —
this discrepancy has yet to be reconciled.

# 4. 「樣本/族群缺口」型
To date, no study has examined [X] in [underrepresented population],
limiting the generalizability of current findings.

# 5. 「方法限制」型
Existing [method]-based analyses lack the resolution / sensitivity / scale
required to detect [target], leaving [question] open.

# 1. 標準 aim 句
Here, we [verb: characterize / quantify / compare / identify]
[X] in [population / system]
using [method].

# 2. 帶假設的 aim
We hypothesized that [predicted relationship],
and tested this by [experimental approach]
in [N] samples / participants.

# 3. 多目標 aim
This study aimed to (i) [verb] [X];
(ii) [verb] [Y];
and (iii) [verb] [Z].

# 4. 結果預告 (Nature/Cell 偏好)
We show that [finding 1], identify [finding 2],
and demonstrate that [finding 3],
providing [contribution / implication].

# 5. 探索性研究 aim
As a hypothesis-generating study, we sought to profile / map / survey
[X] in [Y] and to identify candidate [Z]
for future validation.

⛔

千萬別寫的開頭：「Since the dawn of medicine…」「In recent years, X has become an important topic…」「With the rapid development of technology…」——全部是空話，沒有資訊量，編輯讀到第二句就走神。 Never open with: "Since the dawn of medicine…", "In recent years, X has become an important topic…", "With the rapid development of technology…" — all are content-free filler. Editors disengage by sentence two.

📝 自我檢測

1. Swales 的 CARS 模型中，三個 Move 的正確順序是？

1. What is the correct order of the three Moves in Swales' CARS model?

A. Occupying the Niche → Establishing a Niche → Establishing a TerritoryA. Occupying the Niche → Establishing a Niche → Establishing a Territory

B. Establishing a Territory → Establishing a Niche → Occupying the NicheB. Establishing a Territory → Establishing a Niche → Occupying the Niche

C. Establishing a Niche → Establishing a Territory → Occupying the NicheC. Establishing a Niche → Establishing a Territory → Occupying the Niche

D. Establishing a Territory → Occupying the Niche → Establishing a NicheD. Establishing a Territory → Occupying the Niche → Establishing a Niche

2. 在標準的 4 段漏斗緒論中，引用密度最高的是哪一段？

2. In a standard 4-paragraph funnel introduction, which paragraph has the highest citation density?

A. Para 1（廣的重要性）A. Para 1 (broad importance)

B. Para 2（已知什麼）B. Para 2 (what's known)

C. Para 3（gap）C. Para 3 (the gap)

D. Para 4（我們做了什麼）D. Para 4 (our study)

3. 「Cancer is a serious disease that affects many people.」作為開頭句，主要的問題是什麼？

3. What is the main problem with "Cancer is a serious disease that affects many people." as an opening sentence?

A. 文法錯誤A. Grammar is wrong

B. 太長，讀者會跳過B. Too long; readers will skip it

C. 太廣、無資訊量、與本研究無連結，等於浪費第一句的黃金位置C. Too broad, content-free, and disconnected from the study — it wastes the prime first-sentence slot

D. 沒有引用文獻D. It contains no citation