← TOL 首頁

語言學奧賽備考單元 | 構詞學

Morphology for Linguistics Olympiad Preparation

Part I. Handout｜Concepts & Strategies

(How words encode grammar — and how to reason it out)

1. 什麼是構詞學？

What is Morphology (in Olympiad Terms)?

構詞學研究的是：

一個「詞」裡面，哪些部分在表達語法或意義？

在語言學奧賽中，構詞學不是背名詞，而是回答這個問題：

哪些形式反覆出現？它們在什麼條件下出現？

💡 提示：在解題時，注意觀察重複出現的形式和它們對應的語法功能。

Morphology studies how words are internally structured and how grammatical meaning is encoded inside words. In Olympiad problems, morphology is about one core task:

Identifying which parts of a word do grammatical work, and under what conditions. Key idea:

A word is often not atomic
It is a structured object composed of:
- a lexical core (what the word is about)
- optional grammatical material (number, tense, case, size, politeness, etc.)

2. 四個就夠用的構詞基本概念

Four Core Concepts (Minimal but Sufficient)

You only need four concepts to solve most Olympiad morphology problems.

(1) Morpheme (語素)

最小的「形式 × 意義」單位
必須反覆出現，而且意義穩定

A morpheme is the smallest recurring form–meaning unit.- Recurring = appears in multiple words- Meaningful = contributes consistent information You do not need to label it as prefix/suffix/root in the solution；you only need to identify its function.

(2) Lexical vs. Grammatical Meaning (詞彙意義與文法意義)

Type	What it contributes	Function	Example
Lexical	“what kind of thing/action”	what it is	dog, run
Grammatical	number, tense, role, size	how it behaves	plural, past, diminutive

Olympiad rule of thumb: If a piece only appears when the meaning changes in a systematic way, it is grammatical.

(3) Position Matters（位置很重要）

語法標記通常出現在固定位置
詞根位置較自由

In many languages:

grammatical material appears in a fixed position
lexical material varies more freely

This allows reasoning even when you do not know the language.

(4) One Meaning ≠ One Form

同一語法意義（如複數）可能用：
- 字尾
- 音變
- 重疊
- 零標記

💡 No affix ≠ no morphology.

A single grammatical meaning (e.g. plural) may be expressed by:

an affix
a sound change
reduplication
stem alternation

Do not assume morphology must “look clean”.

3. 世界語言中的構詞類型（奧賽版）

Morphological Types (Olympiad-Oriented)

You do not need full typology. But knowing what kinds of systems exist prevents false assumptions.

類型	特徵	Olympiad Implication
孤立語 Isolating	幾乎沒詞形變化	不要亂找 affix
黏著語 Agglutinative	一個語素一個功能	最好拆
屈折語 Fusional	一個形式多功能	不要一對一假設
非串接 Non-concatenative	音變／內部改變	看「變了什麼」

3.1 Isolating Languages

Words usually = single morpheme
Grammar expressed by word order or separate particles

💡 Olympiad implication:

Do not look for affixes everywhere — sometimes there are none.

3.2 Agglutinative Languages

Words = chains of morphemes
Each morpheme usually has one clear function

Example pattern:

stem + plural + case

💡 Olympiad implication:

Excellent for segmentation-based reasoning.

3.3 Fusional Languages

One form may encode multiple meanings at once

Example:

one ending = tense + person + number

💡 Olympiad implication:

Do not expect one-to-one form–meaning mapping.

3.4 Non-Concatenative Morphology

Grammar expressed by:

vowel changes
reduplication
tone
internal modification

💡 Olympiad implication:

If “nothing seems added”, check what changed.

4. 典型構詞問題

Typical Morphology Problem Types

Type 1: Segment the Word

Task: Identify recurring pieces and their meanings.
Key strategy:
- Align meanings first
- Then align forms
- Only segment when evidence is repeated

Type 2: Identify the Grammatical Function

Task: Decide what a recurring element does (plural? diminutive? tense?)
Key strategy:
- Ask: What meaning is always present when this form appears?
- Avoid: Guessing labels too early

Type 3: Predict New Forms

Task: Generate a word you have never seen.
Key strategy:
- Apply the simplest rule that fits all known data
- Check position and compatibility

Type 4: Resolve Apparent Irregularity

Task: Explain why forms do not look uniform.
Key strategy:
- Ask: Whether the same grammatical meaning is expressed
- by different phonological shapes
- under different conditions

5. How Minimal Morphological Knowledge Helps You Solve Problems

5.1 Prevents Over-Segmentation

Beginners often segment too much.
Minimal principle:
- Do not split unless the split is supported by multiple examples.

5.2 Prevents Category Confusion

Knowing that:
- plural ≠ diminutive
- tense ≠ aspect
- helps you avoid collapsing unrelated meanings into one rule

5.3 Guides Hypothesis Testing

Morphological knowledge tells you what kinds of rules are plausible.

Example:

A “plural” marker that only appears with humans → plausible
A plural marker that randomly disappears → suspicious

5.4 Helps You Handle Exceptions Correctly

Advanced Olympiad problems include controlled exceptions.
Correct reaction:
- “This is not noise — this exception is telling me something.”

6. 構詞題型通用解題流程

A Compact Reasoning Checklist

When facing a morphology problem, always ask:

哪些形式重複？Which parts recur?
哪些意義重複？Which meanings recur?
是否一對一對應？Do form and meaning align consistently?
變化是否有條件？Is position stable?
有沒有「什麼都沒變」本身就是規則？Is variation conditioned or random?

If you can answer these five, you can solve the problem.

Part II. Hands-on Morphology Lab Sessions

###（課堂實作區）

Lab 1｜入門 Entry Level

Seeing Morphology: “Words Are Not Atomic”

Target level:
- First-time Olympiad students
- UKLO Foundation / NACLO Beginner
Learning objective:
- Detect recurring pieces
- Align form with meaning
- Avoid overthinking

Problem 1.1 — Clean Affixation

Word	Meaning
`bata`	child
`bata-ko`	children
`luma`	house
`luma-ko`	houses

Tasks
- Identify the morpheme for plural
- Segment all words
- Translate: children

💡 Expected reasoning

-ko recurs only with plural meaning
Base forms stay intact

Key takeaway: Morphology starts with repetition + consistency, not terminology. Stop try to invent extra rules; Reinforce minimum necessary segmentation

Lab 2｜中低階 Lower Intermediate

One Meaning, Different Shapes (Allomorphy)

Target level
- UKLO Intermediate
- NACLO Open (easier problems)
Learning objective
- Handle allomorphy
- Learn that “plural ≠ single form”

Data

Word	Meaning
tap	stone
tap-im	stones
sun	tree
sun-um	trees
lak	road
lak-im	roads

Tasks

Identify the plural marker(s)
Explain why there are two forms
Predict plural of mun (river)

  💡 Expected reasoning
-im and -um are both plural markers
Choice depends on vowel harmony / phonological context
One grammatical meaning, two surface forms
Key takeaway: If meaning is stable, variation is probably conditioned, not random.

⚠️ Common wrong hypothesis (to discuss) : “There are two different plurals.” And why it fails:

No semantic distinction
Distribution is predictable

---

Lab 3｜中高階 Upper Intermediate

Competing Morphemes: Do Not Collapse Functions

Target level
- Strong UKLO / NACLO
- IOL Selection Test preparation
Learning objective
- Distinguish inflection vs derivation
- Avoid merging unrelated grammatical meanings

Problem 3.1 — Plural vs Diminutive

Data

Word	Meaning
moro	dog
moro-ka	dogs
moro-ti	small dog
moro-ti-ka	small dogs
sena	cat
sena-ti	small cat

Tasks

Identify each morpheme and its function
Why is -ti not plural?
Translate small cats.

  💡 Expected reasoning
-ka = plural
-ti = diminutive (size/derivation)
Order matters: stem – size – number
Different meanings can stack — do not force them into one rule.

Lab 4｜選拔賽等級 Pre-IOL

Morphology Without Affixes (When Morphology Isn’t Segmentable)

Target level
- Strong UKLO / NACLO
- IOL Selection Test preparation
Learning objective
- Handle morphology without affixes
- Learn that “morphology isn’t always segmentable”

Problem 4.1 — Stem Alternation

Data

Word	Meaning
kel	goat
kal	goats
pem	bird
pam	birds
tir	fish

Tasks

How is plural expressed?
Is there an affix?
Predict plural of tir

💡 Expected reasoning

No added segment
Vowel alternation signals plural
Pattern: e → a
Prediction: tir → tar

Key takeaway: Absence of an affix is itself evidence.

Lab 5｜IOL Final Level

Morphology × Constraints × Exceptions

Constraints, Zero Marking, Semantics

Target level
- IOL Final Round
- LLM-resistant reasoning
Learning objective
- Integrate:
  - multiple morphemes
  - conditioned exceptions
  - hypothesis rejection

Problem 5.1 — Restricted Plurality

Data

Word	Meaning
`anu`	man
`anu-ka`	men
`isi`	woman
`isi-ka`	women
`poko`	stone
`poko`	stones
`luma`	house
`luma`	houses

Tasks

Identify the plural rule(s)
Explain why some nouns do not change
Predict plural of anu-ti (young man)

  💡 Expected reasoning
-ka marks plural only for humans
Non-human nouns show zero marking
Morphology is semantically conditioned
Prediction: anu-ti-ka (still human → plural applies)

Why this is IOL-level
- Multiple interacting constraints
- Zero morphology as a meaningful choice
- Requires explicit reasoning, not pattern matching

Cognitive Skill Mapping

Lab	Focus	Cognitive Skill
1	Segmentation	Pattern detection
2	Allomorphy	Conditioned variation
3	Function separation	Hypothesis discipline
4	Non-concatenative	Rule abstraction
5	Constraints & exceptions	System-level reasoning

Part III. Answer Keys（教師用，請勿提前給學生）

Lab 1 Answer Key

-ko = plural
bata / luma = singular bases
children = bata-ko

Lab 2 Answer Key

Plural = -im / -um
Allomorphy conditioned by vowel context
mun → mun-um

Lab 3 Answer Key

-ka = plural
-ti = diminutive (size)
small cats = sena-ti-ka

Lab 4 Answer Key

Plural marked by vowel change e → a
No affix
tir → tar

Lab 5 Answer Key

-ka marks plural only for human nouns
Non-human nouns show zero plural
anu-ti → anu-ti-ka

附錄：字體使用指南

如何在文中切換字體

在 Markdown 文件中，你可以使用 HTML <span> 標籤配合 CSS 類來切換字體：

基本語法

<span class="字體類名">要改變字體的內容</span>

可用字體類

類名	字體類型	用途	範例
`.ming`	明体（衬线）	正文、標題（預設）	`<span class="ming">這是明體</span>`
`.hei`	黑体（无衬线）	強調、標題	`<span class="hei">這是黑體</span>`
`.kai`	楷体	手寫風格、特殊說明	`<span class="kai">這是楷體</span>`
`.fangsong`	仿宋	優雅風格	`<span class="fangsong">這是仿宋</span>`
`.mono`	等寬字體	代碼、對齊	`<span class="mono">這是等寬</span>`
`.english`	英文專用（Lato）	英文內容（已全局設置，無需使用）	`English Text`
`.highlight`	淺灰高亮（行內）	標記重點文字	`<span class="highlight">重要內容</span>`
`.highlight-gray`	淺灰高亮（行內）	標記重點文字	`<span class="highlight-gray">重要內容</span>`
`.highlight-yellow`	黃色高亮（行內）	標記重點文字	`<span class="highlight-yellow">重要內容</span>`
`.highlight-blue`	藍色高亮（行內）	標記重點文字	`<span class="highlight-blue">重要內容</span>`
`.highlight-green`	綠色高亮（行內）	標記重點文字	`<span class="highlight-green">重要內容</span>`
`.highlight-block`	淺灰高亮（塊級）	標記整段內容	`<div class="highlight-block">整段內容</div>`
`.highlight-block-yellow`	黃色高亮（塊級）	標記整段內容	`<div class="highlight-block-yellow">整段內容</div>`
`.highlight-block-blue`	藍色高亮（塊級）	標記整段內容	`<div class="highlight-block-blue">整段內容</div>`

組合使用

可以同時使用多個類：

<span class="hei large">這是大號黑體</span>
<span class="kai italic">這是斜體楷體</span>
<span class="ming bold">這是粗體明體</span>

實際範例

範例 1：強調重點

構詞學的核心概念是：<span class="hei">語素（morpheme）</span>。

範例 2：特殊說明

<span class="kai">注意：</span>這個規則僅適用於人類名詞。

範例 3：英文內容（使用 Lato 字體）

The term morphology comes from Greek.

範例 4：代碼或對齊

複數標記：<span class="mono">-ko</span>

範例 5：高亮背景（標記重點）

構詞學的核心是：<span class="highlight">語素（morpheme）</span>。

重點提示：<span class="highlight-yellow">這個規則很重要</span>。

注意事項：<span class="highlight-blue">請仔細閱讀</span>。

可用高亮顏色：

<span class="highlight"> 或 <span class="highlight-gray"> - 淺灰色（默認）
<span class="highlight-yellow"> - 黃色
<span class="highlight-blue"> - 藍色
<span class="highlight-green"> - 綠色
<span class="highlight-pink"> - 粉色
<span class="highlight-orange"> - 橙色

範例 6：塊級高亮（整段高亮）

<div class="highlight-block">
這是一整段需要高亮的內容。
可以包含多行文字。
適合用於重要說明、提示、注意事項等。
</div>

<div class="highlight-block-yellow">
⚠️ 重要提示：這是黃色塊級高亮
適合用於警告或重要提醒
</div>

<div class="highlight-block-blue">
💡 提示：這是藍色塊級高亮
適合用於一般提示或說明
</div>

可用塊級高亮顏色：

<div class="highlight-block"> 或 <div class="highlight-block-gray"> - 淺灰色（默認）
<div class="highlight-block-yellow"> - 黃色
<div class="highlight-block-blue"> - 藍色
<div class="highlight-block-green"> - 綠色
<div class="highlight-block-pink"> - 粉色
<div class="highlight-block-orange"> - 橙色

塊級高亮 vs 行內高亮：

行內高亮 (<span class="highlight">)：用於標記單詞或短句
塊級高亮 (<div class="highlight-block">)：用於標記整段內容，有更大的內邊距和左側邊框

在表格中使用

| 類別 | 說明 |
|------|------|
| 語素 | <span class="hei">最小的意義單位</span> |
| 詞根 | <span class="kai">詞彙核心</span> |

注意事項

預設字體：文件預設使用明體（中文）和 Lato（英文），無需特別標註
相容性：這些樣式在支援 HTML 的 Markdown 查看器中有效（如 Typora、VS Code 預覽、GitHub 等）
PDF 轉換：使用 Pandoc 轉換為 PDF 時，樣式會保留
簡潔原則：不要過度使用字體切換，保持文件一致性

快速參考卡片

明體（預設）：<span class="ming">文字</span> 或不標註
黑體強調：<span class="hei">文字</span>
楷體說明：<span class="kai">文字</span>
等寬代碼：<span class="mono">文字</span>
英文專用：現在全局自動使用 Lato，無需標註
行內高亮：<span class="highlight">文字</span> 或 <span class="highlight-gray">文字</span>
塊級高亮：<div class="highlight-block">整段內容</div> 或 <div class="highlight-block-yellow">整段內容</div>

附錄：Markdown 預覽對齊問題解決方案

為什麼預覽無法很好對齊？

Markdown 預覽器（如 Typora、VS Code、GitHub 等）在渲染表格時，可能因為以下原因導致對齊問題：

字符寬度差異：
- 中文字符是全角（2個字符寬度）
- 英文字符是半角（1個字符寬度）
- 不同字體的字符寬度可能不同
字體渲染差異：
- Lato（英文）和明體（中文）的字符寬度不同
- 瀏覽器/編輯器計算列寬時可能不準確
表格解析差異：
- 不同 Markdown 解析器對表格的處理方式不同

🚀 立即解決方案（推薦）

方案 1：檢查表格格式（最簡單）

確保表格格式正確：

每列之間必須有空格：

✅ 正確：| 列1 | 列2 | 列3 |
❌ 錯誤：| 列1|列2|列3 |

分隔線長度要一致：

✅ 正確：
| 類型 | 特徵 | 說明 |
| ---- | ---- | ---- |

❌ 錯誤：
| 類型 | 特徵 | 說明 |
| -- | ---- | - |

確保表格行開頭和結尾都有 |：

✅ 正確：| 內容 | 內容 |
❌ 錯誤：內容 | 內容

方案 2：使用 HTML 表格（最可靠）

如果 Markdown 表格仍然無法對齊，直接使用 HTML 表格：

<table>
  <thead>
    <tr>
      <th style="width: 30%;">類型</th>
      <th style="width: 30%;">特徵</th>
      <th style="width: 40%;">說明</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>孤立語</td>
      <td>幾乎沒詞形變化</td>
      <td>不要亂找 affix</td>
    </tr>
  </tbody>
</table>

方案 3：編輯器設置調整

Typora：

設置 → 偏好設置 → 編輯器
啟用「嚴格模式」
設置 → 偏好設置 → 外觀 → 調整字體大小為 14-16px

VS Code：

安裝擴展：「Markdown Preview Enhanced」
使用 Ctrl+Shift+V 預覽（比內建預覽器對齊更好）

其他編輯器：

嘗試導出為 HTML 或 PDF 查看對齊效果
使用 GitHub 網頁版預覽（對表格支持最好）

方案 3：使用 HTML 表格（最精確）

如果需要完全控制對齊，可以使用 HTML 表格：

<table>
  <thead>
    <tr>
      <th style="width: 30%;">類型</th>
      <th style="width: 30%;">特徵</th>
      <th style="width: 40%;">Olympiad Implication</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>孤立語 Isolating</td>
      <td>幾乎沒詞形變化</td>
      <td>不要亂找 affix</td>
    </tr>
  </tbody>
</table>

方案 4：編輯器設置

Typora：

設置 → 偏好設置 → 編輯器 → 啟用「嚴格模式」
設置 → 偏好設置 → 外觀 → 調整字體大小

VS Code：

安裝 "Markdown Preview Enhanced" 擴展
使用 Ctrl+Shift+V 預覽，對齊會更好

GitHub：

GitHub 的 Markdown 渲染對表格支持較好
如果仍有問題，考慮使用 HTML 表格

🔍 快速診斷：為什麼我的表格無法對齊？

檢查清單：

✅ 表格格式是否正確？
- 每列之間有空格：| 列1 | 列2 |
- 分隔線長度一致：| ---- | ---- |
- 每行開頭和結尾都有 |
✅ 使用的編輯器/預覽器是什麼？
- Typora：啟用「嚴格模式」
- VS Code：使用「Markdown Preview Enhanced」擴展
- GitHub：網頁版預覽通常對齊最好
✅ CSS 樣式是否生效？
- 檢查瀏覽器開發者工具（F12）
- 確認 <style> 標籤中的 CSS 是否被應用
✅ 表格內容是否過長？
- 如果某列內容很長，考慮縮短或換行
- 使用 HTML 表格可以更好地控制列寬

當前文件的對齊優化

已添加以下 CSS 優化：

表格布局：使用 table-layout: auto 自動調整列寬
強制樣式：使用 !important 確保樣式優先級
文字換行：允許長文本自動換行，避免破壞布局
邊框和間距：統一的邊框和內邊距設置
表格渲染：強制使用 display: table 確保正確渲染

建議

一般表格：使用標準 Markdown 表格格式即可
數據表格：考慮使用等寬字體或 HTML 表格
複雜表格：直接使用 HTML 表格，獲得完全控制

如果對齊問題持續存在，建議：

檢查使用的 Markdown 編輯器/預覽器
嘗試不同的預覽器（如 GitHub、Typora、Marked 等）
對於關鍵表格，使用 HTML 表格替代