Types of Corpus Annotation

Corpus annotation is the process of enriching raw text (or speech) with linguistic, semantic, or extra-linguistic information so that it can be used for linguistic research, NLP, and AI.

1. Orthographic Annotation

Annotates the standard written form of text.

Purpose

• Normalize spelling
• Correct OCR errors
• Standardize historical texts

Example

Raw:

gud mornin

Annotated:

good morning

2. Token Annotation (Tokenization)

Divides text into meaningful units (tokens).

Example

Sentence:

The cat's tail is long.

Tokens:

The | cat | 's | tail | is | long | .

3. Sentence Boundary Annotation

Marks sentence beginnings and endings.

Example

[Sentence 1]
The dog barked.

[Sentence 2]
The cat slept.

4. Lemmatization

Maps word forms to dictionary forms (lemmas).

5. Stemming

Reduces words to stems.

6. Part-of-Speech (POS) Annotation

Assigns grammatical categories.

Common tags:

• NN
• NNS
• VB
• VBD
• JJ
• RB
• DT
• PRP

7. Morphological Annotation

Adds internal grammatical information.

Example

cats

Lemma = cat
POS = Noun
Number = Plural
Gender = Common

Example (Tamil)

மரங்களில்

Lemma: மரம்
Case: Locative
Number: Plural

Features commonly annotated:

• Gender
• Number
• Case
• Person
• Tense
• Aspect
• Mood
• Voice

8. Syntactic Annotation

Describes grammatical structure.

Constituency Parsing

(S
 (NP The boy)
 (VP ate
     (NP an apple)))

Dependency Parsing

ate
├── boy (subject)
└── apple (object)

9. Semantic Annotation

Adds meaning information.

Example

bank

Meaning:
Financial institution

or

bank

Meaning:
River bank

10. Named Entity Recognition (NER)

Labels proper names.

Example

Barack Obama
PERSON

Paris
LOCATION

Google
ORGANIZATION

Entity types include:

• PERSON
• LOCATION
• ORGANIZATION
• DATE
• MONEY
• PRODUCT
• EVENT
• LANGUAGE

11. Word Sense Annotation

Chooses the intended dictionary sense.

Example

He sat by the bank.

bank → river bank

12. Semantic Role Labeling (SRL)

Assigns roles in events.

Example

John gave Mary a book.

Agent      John
Recipient  Mary
Theme      book

13. Coreference Annotation

Links expressions referring to the same entity.

Example

John entered.

He smiled.

John ↔ He

14. Discourse Annotation

Captures relationships between sentences.

Relations include:

• Cause
• Contrast
• Elaboration
• Condition
• Sequence
• Explanation

Example

It rained.

Therefore,
the match was cancelled.

15. Pragmatic Annotation

Adds speaker intentions.

Example

Could you open the door?

Speech Act:
Request

16. Dialogue Annotation

Used in conversational corpora.

Labels include:

• Greeting
• Question
• Answer
• Request
• Confirmation
• Apology
• Agreement
• Rejection

17. Sentiment Annotation

Marks emotional polarity.

Example

This phone is amazing.

Positive

Possible labels:

• Positive
• Negative
• Neutral
• Mixed

18. Emotion Annotation

Labels specific emotions.

Typical categories:

• Joy
• Anger
• Fear
• Sadness
• Surprise
• Disgust
• Trust
• Anticipation

19. Opinion Annotation

Separates opinion holder, target, and sentiment.

Example

John likes Samsung.

Holder:
John

Target:
Samsung

Polarity:
Positive

20. Information Structure Annotation

Marks discourse functions.

Examples:

• Topic
• Focus
• Given
• New
• Contrastive Focus

21. Temporal Annotation

Labels time expressions and event ordering.

Example

Yesterday

DATE

Relations:

• Before
• After
• During
• Simultaneous

22. Spatial Annotation

Marks locations and spatial relations.

Example

The book is on the table.

Figure: book
Ground: table
Relation: on

23. Event Annotation

Identifies events.

Example

Explosion

Event Type:
Explosion

Time:
10 AM

Location:
Market

24. Speech Annotation

For spoken corpora.

Includes:

• Pause
• Stress
• Intonation
• Pitch
• Laughter
• Hesitation
• Overlap

25. Phonetic Annotation

Represents pronunciation.

Example

cat

/kæt/

26. Phonological Annotation

Annotates abstract sound structure.

Example

plural suffix

/s/
/z/
/ɪz/

27. Prosodic Annotation

Captures speech rhythm.

Includes:

• Pitch
• Intonation
• Stress
• Rhythm
• Boundary tones

28. Gesture Annotation (Multimodal)

Links speech with gestures.

Example

Speaker:
points left

Speech:
"Over there."

29. Multimodal Annotation

Combines multiple communication channels.

Modalities:

• Text
• Audio
• Video
• Gesture
• Eye gaze
• Facial expression

30. Metadata Annotation

Describes the corpus itself.

Examples:

• Author
• Speaker
• Age
• Gender
• Language
• Dialect
• Genre
• Date
• Source
• Domain

31. Error Annotation

Used in learner corpora.

Example

She go to school.

Error:
Verb Agreement

Correction:
She goes to school.

32. Translation Annotation (Parallel Corpora)

Aligns corresponding units across languages.

Example

English:
Good morning.

Tamil:
காலை வணக்கம்.

33. Alignment Annotation

Maps corresponding units.

Levels:

• Document
• Paragraph
• Sentence
• Word
• Morpheme

34. Semantic Frame Annotation

Labels events using frame semantics.

Example

buy

Buyer
Seller
Goods
Money

35. Relation Annotation

Marks semantic relationships.

Examples:

Part-of
Located-in
Cause-of
Member-of
Works-for
Lives-in

Summary Table

Annotation Levels

Corpus Annotation
├── Surface Annotation
│   ├── Orthographic
│   ├── Tokenization
│   ├── Sentence Segmentation
│   └── Metadata
├── Lexical Annotation
│   ├── Lemma
│   ├── Stem
│   ├── POS
│   ├── Morphology
│   ├── Word Sense
│   └── NER
├── Syntactic Annotation
│   ├── Constituency
│   ├── Dependency
│   └── Grammatical Relations
├── Semantic Annotation
│   ├── Semantic Roles
│   ├── Semantic Frames
│   ├── Entity Relations
│   ├── Event
│   └── Temporal & Spatial
├── Discourse Annotation
│   ├── Coreference
│   ├── Discourse Relations
│   ├── Dialogue Acts
│   └── Information Structure
├── Pragmatic Annotation
│   ├── Speech Acts
│   ├── Sentiment
│   ├── Emotion
│   └── Opinions
├── Speech Annotation
│   ├── Phonetic
│   ├── Phonological
│   ├── Prosodic
│   └── Conversation Features
└── Multimodal Annotation
    ├── Gesture
    ├── Eye Gaze
    ├── Facial Expression
    ├── Video
    └── Parallel/Translation Alignment