Module 1:

Natural Language Processing(NLP)

By
Mrs. Sangeetha M S
Assistant professor
SCEM
Natural Language Processing
Language is how humans communicate ideas, emotions, and
information.
Natural Language Processing (NLP) is a field of Artificial Intelligence
(AI) that helps computers:
• Understand human language
• Interpret meaning
• Generate language
• Communicate with humans naturally
In short: NLP helps computers read, listen, speak, and understand
human language.
Approaches to NLP
There are two main ways researchers study NLP:
1. Rationalist Approach
• Assumes humans have a language faculty in the brain.
• Language understanding is rule-based.
• Focuses on grammar and structure.
• In this approach, language understanding is treated as a logical
process:
• Identify grammatical roles (subject, verb, object)
• Apply syntactic rules
• Generate meaning using predefined structures
Approaches to NLP
Example:
Consider the sentence:
"The children are playing in the garden."

A rationalist system would:

• Identify parts of speech
• Apply grammar rules
• Validate whether the sentence follows syntactic correctness
Approaches to NLP
2. Empiricist Approach
• Does not believe in a special language faculty.
• Language is learned using:
• Pattern recognition
• Generalization
• Learning/Experience from data
• Modern NLP mostly uses the empiricist (data-driven) approach. No
predefined linguistic rules are required
Examples of Empiricist-Based NLP Applications
• Machine Translation (Google Translate)
• Chatbots and virtual assistants
• Speech recognition systems
• Text summarization
• Sentiment analysis
Origin of NLP
NLP originally started from Machine Translation (e.g., English →
French).
NLP includes:
• NLU – Natural Language Understanding
• NLG – Natural Language Generation
NLP combines:
• Linguistics (grammar and rules)
• Statistics
• Machine Learning
• Deep Learning
Computational Linguistics
•Studies language using computers and algorithms.
•Applies linguistic theories using computational models.

Types of Computational Models

• Knowledge-driven models
• Use grammar rules written by humans
• Rule-based systems
• Data-driven models
• Learn patterns from large datasets
• Use Machine Learning
Language and Knowledge

• Language is a tool to represent knowledge.

• Knowledge can change by:
• Adding new words
• Creating new meanings
• Language processing happens at different levels, each needing
different types of knowledge.
Levels of Language (Text) Processing
(Steps in NLP)
1. Lexical Analysis/Morphological Analysis (Word Level)
2. Syntactic Analysis (Grammar Level)
3. Semantic Analysis (Meaning Level)
4. Discourse Analysis (Paragraph Level)
5. Pragmatic Analysis (Context Level)
Levels of Language (Text) Processing
(Steps in NLP)
1. Lexical Analysis Morphological Analysis (Word Level):
•Studies words
•Uses morphological knowledge

Example:
•Un + happy = unhappy
Levels of Language (Text) Processing
(Steps in NLP)
1. Lexical Analysis/Morphological Analysis (Word Level):

John ate the _?__

Levels of Language (Text) Processing
(Steps in NLP)
2. Syntactic Analysis (Grammar Level)
•Studies sentence structure
•Checks word order and grammar

Example:
I like ice cream
✘ Like I ice cream
Levels of Language (Text) Processing
(Steps in NLP)
3. Semantic Analysis (Meaning Level)
•Focuses on meaning
•Converts text into meaningful representation

Example 1:
•“Bank” → river bank or money bank?

Example 2: She drank some milk

She drank some books

Levels of Language (Text) Processing
(Steps in NLP)
4. Discourse Analysis (Paragraph Level)
•Considers previous sentences
•Meaning depends on context

Example:
•“John went home. He was tired.”
•“He” refers to John
Levels of Language (Text) Processing
(Steps in NLP)
5. Pragmatic Analysis (Context Level)
•Uses real-world knowledge
•Depends on situation and intention

Example:
•“Can you open the door?” (It is a request, not a question)
Challenges of NLP
1. Ambiguity and Vagueness of Natural Language
• Natural languages are not precise like programming languages.
• A word, phrase, or sentence can have more than one meaning.
• Humans understand meaning using context and common sense, but
computers struggle with this.
Example: bank
• river bank
• bank for money
A computer must decide which meaning is correct, which is difficult without
enough context.
Challenges of NLP
2. Inability to Capture All Required Knowledge
• To understand language properly, a system needs:
• world knowledge
• cultural knowledge
• contextual knowledge
• It is impossible to store all human knowledge in a computer.
Example:
• “The glass fell and broke.”
A human knows glass is fragile, but a computer must be explicitly
taught this.
Challenges of NLP
3. Identifying the Semantics (Meaning) of Sentences
Understanding sentence meaning is one of the hardest problems in
NLP.
a) Role of Syntax and Semantics
• Syntax: sentence structure
• Semantics: meaning of words and relations
• Meaning depends on how words are arranged.
Example:
• I do not like ice cream
• Do not I ice cream like
• Same words, but only the first sentence follows correct syntax.
Challenges of NLP
3. Identifying the Semantics (Meaning) of Sentences
Understanding sentence meaning is one of the hardest problems in
NLP.
b) Word Sense Depends on Frequency and Context
• The meaning of a word often depends on how frequently it is used in a
particular sense.
• Example:
• mouse
• animal
• computer device (more common in technical context)
Challenges of NLP
3. Identifying the Semantics (Meaning) of Sentences
Understanding sentence meaning is one of the hardest problems in
NLP.
c) Idioms, Metaphors, and Ellipsis
• These expressions do not mean exactly what the words say, making them
hard for machines.
• Examples:
• Kick the bucket → means die, not actually kicking anything
• He spilled the beans → revealed a secret
• Ellipsis: I went to the library, and she to the lab (word “went” is omitted)
Computers often interpret them literally, leading to wrong understanding.
Challenges of NLP
4. Words Alone Do Not Make a Sentence
Correct word order is essential.
Example:
• I do not like ice cream
• Do not I ice cream like
• Though words are same, meaning is lost due to wrong structure.
Challenges of NLP
5. Ambiguity Due to Sentence Structure (Structural Ambiguity)
• Even if words are clear, sentence structure can cause confusion.
Example:
• Stolen ice candy found by tree
Possible meanings:
• Ice candy was found near a tree
• A tree found the ice candy (illogical, but grammatically possible)
Challenges of NLP
6. Ambiguity at Word Level (Lexical Ambiguity)
• A single word can have multiple meanings.
Examples:
• can → container / ability
• bank → financial institution / river side
• NLP systems must choose the correct meaning based on context.
Challenges of NLP
7. Quantifier Scope Problem
• Words like the, each, every, some can change meaning depending on
scope.
Example:
• Each student read a book
• Same book?
• Different books?
• Humans infer easily, but computers find this difficult.
Challenges of NLP
8. Same Structure, Different Meaning
• Small changes in words can completely change meaning.
Example:
• Kabir and Ayan are married → they are married to each other
• Kabir and Suhana are married → possibly married to different people
• Understanding such differences requires social and world
knowledge.
Challenges of NLP
9. Diversity and Complexity of Grammar
• There are infinite ways to form sentences.
• Grammar rules differ across languages.
• Writing rules that cover all correct sentences and reject all incorrect ones
is nearly impossible.
Reason:
• Language evolves constantly
• People break rules but are still understood
• Informal and spoken language adds more variation
• Hence, no grammar can perfectly capture all meaningful text.
Challenges of NLP
NLP is challenging because:
• Language is ambiguous
• Meaning depends on context, structure, and world knowledge
• Grammar is complex and infinite
• Human intuition is hard to replicate in machines
• That is why building accurate NLP systems remains an active research
area.
Idioms, metaphors and ellipsis ???
An idiom is a phrase that, when taken as a whole, has a meaning you wouldn't be
able to deduce from the meanings of the individual words.
Examples:
The ball is in your court. →(It is your responsibility now.)
A piece of cake →(Something very easy.)
Every cloud has a silver lining →(Even bad situations have something good.)
Not one’s cup of tea → (Something a person does not like.)
Idioms, metaphors and ellipsis ???
A metaphor is a figure of speech that describes something by saying it's something else.
Example:
Life is a highway. →(Life is long and full of journeys and experiences.)
Her eyes were diamonds. →(Her eyes were very bright and beautiful, not real diamonds).
He is a shining star. →(He is very talented or successful.)
An ellipsis is a punctuation mark of three dots (. . .) that shows an omission of words, represents a
pause, or suggests there's something left unsaid.
You mean he's...
The chancellor sweated profusely and stammered, “Your highness, … I wasn’t … I didn’t mean to
imply that … I would never question your wisdom.”
Automatic Processing of Language and
Grammar
Automatic processing of natural language (Natural Language Processing – NLP) requires
that the rules and exceptions of a language be explicitly described so that a computer can
understand and manipulate language.
Unlike programming languages, natural languages are ambiguous, dynamic, and full of
exceptions, which makes their formal specification difficult.
Grammar defines a language.
It is a system of rules that enables us to:
• Parse sentences (analyze their structure and meaning)
• Generate sentences (produce grammatically correct expressions)
Grammar specifies:
• Word order
• Agreement (subject–verb, noun–adjective)
• Phrase structure
• Relationships among sentence elements
Challenges in Language Specification
The main hurdles in defining grammar for automatic processing are:
• Constantly changing nature of natural languages
• New words, usages, and constructions appear over time.
• Large number of exceptions
• Many rules have irregular forms and special cases.
• Ambiguity
• Same sentence can have multiple interpretations.
These challenges have led to the development of multiple grammatical theories, each addressing
language structure differently.
Types of Grammars
Several grammatical frameworks have been proposed, including:
• Transformational Grammar
• Lexical Functional Grammar (LFG)
• Government and Binding Theory
• Dependency Grammar
• Paninian Grammar
• Tree-Adjoining Grammar (TAG)
Most of these grammars use phrase structure rules, which describe how smaller units (words)
combine to form larger units (phrases and sentences).

Hierarchy of Formal Grammars

Grammars can be organized in a hierarchy based on their expressive power and complexity,
such as:
• Regular grammars (Type-3 Grammar)
• Context-free grammars (Type-2 Grammar)
• Context-sensitive grammars (Type-1 Grammar)
Natural languages generally require more powerful grammars than simple regular grammars.
Language Modelling
A language model is a way for a computer to understand and work with human
language.
It tells the computer what sentences are correct and how likely a sentence is to
occur.

Language modelling can be seen in two ways:

• Grammar-based approach
• Statistical (probability-based) approach
1. Grammar-Based Language Model
• This approach uses grammar rules of a language.
• It checks whether a sentence is grammatically correct or not and how words relate to each
other.

How it works:
• Grammar rules are manually written by experts.
• These rules describe how words are arranged in a sentence.

Example:
• A sentence usually has:
• Noun Phrase (NP) + Verb Phrase (VP)
“Children play.”
• NP = Children
• VP = play
•Focuses on sentence structure, not on probability.
•It distinguishes grammatical vs. ungrammatical sentences.
2. Statistical Language Modelling
• This approach uses statistics and probability.
• It learns language patterns from a large collection of text (corpus).

How it works:
• The model is trained on many sentences.
• It calculates the probability of words and sentences.
• The sentence with higher probability is considered more natural.

Example:
• “I am going to school” → high probability
• “I going am school to” → low probability

• Does not use hand-written grammar rules.

• Needs a large amount of data to work well.
Various Grammar-based Language Models
• Generative Grammars
• Hierarchical Grammar
• Government and Binding (GB)
• Lexical Functional Grammer(LFG)
• Paninian Grammer
Various Grammar-based Language Models
1. Generative Grammars: Introduced by Noam Chomsky in 1957.
The main idea is that: If we know the words of a language and the rules, we can generate
all valid sentences.
• Only sentences produced by the grammar rules are grammatical.
• Ungrammatical sentences are not generated.
• The entire set of rules that can generate all possible sentences forms the language
model.
Example:
• Rule:
Sentence → Noun Phrase + Verb Phrase
• Using this rule, we can generate:
• Students read books.
• A sentence not following rules is rejected.
Various Grammar-based Language Models
2. Hierarchical Grammar (Chomsky Hierarchy)
• Grammars are organized in a hierarchy (levels).
• Each higher level contains the lower levels.
Types of Grammar (from most powerful to simplest):
• Type 0 – Unrestricted Grammar
• No restrictions on rules.
• Can describe all possible languages.
• Type 1 – Context-Sensitive Grammar
• Rules depend on the context of words.
• Type 2 – Context-Free Grammar (CFG)
• Commonly used in NLP.
• Suitable for describing sentence structure.
• Example: S → NP + VP
• Type 3 – Regular Grammar
• Simplest grammar
• Used in Pattern matching and finite automata
Various Grammar-based Language Models
3. Government and Binding (GB) Theory
• Developed by Noam Chomsky in 1981.
• It is an extension of Generative Grammar.
• Earlier transformational grammars had too many rules.
• GB theory reduces complexity by using fewer, more general rules.
• Transformational grammars had hundreds of language-specific rules. Creating complete rule sets was very
difficult.
• GB theory says: If we define rules at a deep structural level, we can generate any language with fewer rules.
 Various Grammar-based Language Models
3. Government and Binding (GB) Theory
Levels of Sentence Representation
• Deep Structure – meaning of the sentence
• Surface Structure – actual spoken or written form
GB Theory renamed and expanded these levels:
D-Level (Deep Level)
• Abstract structure (s = NP+VP)(basic meaning).
S-Level (Surface Level)
• Final sentence form.
Phonetic Form (PF)
• PF is the interface with the Phonology where shapes, sounds (pronunciation), and groupings of items are directly represented.
• Interface with phonology.
Logical Form (LF)
• Connected to meaning.
• Interface with semantics.
X-Bar Theory
X-Bar Theory states that all phrases have a uniform structure consisting of a head
(X), an intermediate projection (X′), and a maximal projection (XP), with specifiers,
complements, and modifiers arranged systematically around the head.
X-Bar Theory says: Every phrase in every language has the same basic structure.
• Instead of learning different rules for NP, VP, AP, PP, we use one common pattern
called the X-bar schema.
• X = head (main word)
• XP = full phrase (maximal projection)
• X′ (X-bar) = intermediate level

Example:
the food → head = food (N)
→ specifier=the
Every phrase has a head (the most important word).
Phrase Head Example:
NP (Noun Phrase) N (noun) the food → head = food (N)
→ Det/specifier=the
VP (Verb Phrase) V (verb)
AP (Adjective Phrase) A (adjective)
PP (Prepositional Phrase) P (preposition)

Three Levels in X-Bar Structure

Each phrase is built in three levels:
• X (Head) – main word
• X′ (Intermediate / semi-phrasal level) – head + complement
• XP (Maximal projection) – complete phrase

GB theory predicts two projections:

• X′ (first projection)
• XP (maximal projection)
NP Example: “the food in a dhaba”
• Head (N): food
• Specifier (Det): the
• Complement (PP): in a dhaba
VP Example: “ate the food in a dhaba”
• Head (V): ate
• Complement (NP): the food
• Modifier (PP): in a dhaba
AP Example: “very proud of our country”
• Head (A): proud
• Specifier (Deg): very
• Complement (PP): of our country
PP Example: “in a dhaba”
• Head (P): in
• Complement (NP): a dhaba
X-Bar Theory
X-Bar Theory states that all phrases have a uniform structure consisting of a head
(X), an intermediate projection (X′), and a maximal projection (XP), with specifiers,
complements, and modifiers arranged systematically around the head.
X-Bar Theory says: Every phrase in every language has the same basic structure.
• Instead of learning different rules for NP, VP, AP, PP, we use one common pattern
called the X-bar schema.
• X = head (main word)
• XP = full phrase (maximal projection)
• X′ (X-bar) = intermediate level
 X-Bar Theory
Gives extra
grammatical info (like
quantity, definiteness)

Optional extra description

(time, manner, place, etc.)

head
Completes the
meaning of the head
Specifiers
The car Very excited Slowly read
Adjunct
The big car Walk at night
Compliment
Example 1: John eats an apple. Example 2: She sat on the chair
Here We focus on the VP. Here We focus on the PP.
•Head (V): eats •Head (P): on
•Complement (NP): an apple •Complement (NP): the chair

•V = eats (head)
•NP = an apple (complement)
•V′ = V + Complement
•VP = maximal projection of V
 The boy ate rice quickly.
NP VP

NP (Noun Phrase) VP (Verb Phrase)

•Specifier: the •Head (V): ate
•Head (N): boy •Complement: NP (rice)
•Adjunct: AdvP (quickly)
Theta Theory
Theta Theory explains the semantic roles that words (especially verbs) assign to other words in a
sentence.
It tells us what role each noun plays in the action.
Theta Theory explains how lexical heads assign semantic roles (θ-roles) to their arguments, and the
Theta Criterion ensures that each argument bears one and only one θ-role, assigned at D-structure
and preserved across all syntactic levels.
What are Theta (θ) Roles?
Theta-role Meaning Example
Agent Doer of the action John ate food
Theme Thing affected John ate food
Experiencer Feels something Mary likes music
Goal Destination She went to school
Source Starting point He came from home
Sub-categorization:
Sub-categorization means: A verb decides how many arguments it
needs and what type they should be.
Example:
• eat → needs two arguments
• Agent (who eats)
• Theme (what is eaten)
She ate rice
She ate (incomplete in basic sense)
Example: Verb “eat”
From the lexicon:
eat → θ-roles: (Agent, Theme)

Sentence:She ate the food

Argument Role
she Agent
the food Theme

Projection Principle

Theta-roles assigned in the lexicon must appear in syntax and cannot change.
Meaning:
•If a verb assigns an Agent and a Theme,
•Those roles must appear in the sentence at all syntactic levels.
You cannot add or remove roles later.
Theta Theory
Theta Theory explains the semantic roles that words (especially verbs) assign to other words in a
sentence.
It tells us what role each noun plays in the action.
In a sentence, words have different specific roles, known as Theta roles or Thematic roles.
It provides us with a way of representing the semantic relationship between a head and its
dependents.
What are Theta (θ) Roles?
Theta-role Meaning Example
Agent Doer of the action John ate food
Theme Thing affected John ate food
Experiencer Arguments that Feels something Mary likes music
Goal Destination She went to school
Source Starting point He came from home
Theta (θ) Roles
1. Agent: The doer of an action (intentionally performs the action).
Examples:
• The girl kicked the ball.
→ The girl = Agent
• Ravi opened the door.
→ Ravi = Agent
• The teacher explained the lesson.
→ The teacher = Agent
• The cat chased the mouse.
→ The cat = Agent
Theta (θ) Roles
2. Experiencer: The one who feels, perceives, or experiences something (not
doing a physical action).
Examples:
• She loves music.
→ She = Experiencer
• The child felt scared.
→ The child = Experiencer
• I heard a strange noise.
→ I = Experiencer
• They enjoyed the movie.
→ They = Experiencer
Theta (θ) Roles
3. Theme: The entity that is moved, affected, or talked about (not the
doer).
Examples:
• She kicked the ball.
→ the ball = Theme
• He read the book.
→ the book = Theme
• They carried the box upstairs.
→ the box = Theme
• I saw a rainbow.
→ a rainbow = Theme
Theta (θ) Roles
4. Goal: The destination or endpoint of movement or action.
Examples:
• She went to the market.
→ to the market = Goal
• He sent the letter to his friend.
→ to his friend = Goal
• The child ran into the room.
→ into the room = Goal
• They gave the gift to the teacher.
→ to the teacher = Goal
Theta (θ) Roles
5. Source: The starting point of movement or transfer.
Examples:
• She came from school.
→ from school = Source
• He borrowed money from his uncle.
→ from his uncle = Source
• The train departed from Delhi.
→ from Delhi = Source
• Water flowed from the mountain.
→ from the mountain = Source
Theta roles represents the semantic relationship between a head and
its dependents.
Sub-categorization:
A verb decides how many arguments it needs and what type they
should be.
Example:
• eat → needs two arguments
• Agent (who eats)
• Theme (what is eaten)
She ate rice
She ate (incomplete in basic sense)
Example: Verb “eat”
From the lexicon:
eat → θ-roles: (Agent, Theme)

Sentence: She ate the food

Argument Role
she Agent
the food Theme

Projection Principle

Theta-roles assigned in the lexicon must appear in syntax and cannot change.
•If a verb assigns an Agent and a Theme, Those roles must appear in the sentence at all
syntactic levels.
You cannot add or remove roles later.
Lexical Functional Grammar (LFG) Model
Lexical-Functional Grammar (LFG) is a theory of grammar used to study how sentences are structured
in human language.

Different kinds of grammatical information are kept separate, but they are linked to each other.

In LFG, language is described using different levels, each doing a different job:
Constituent Structure (c-structure)
• Shows word order and phrase structure (like NP, VP, sentence trees)
• Word order (c-structure) tells us where words appear,
Functional Structure (f-structure)
• Shows grammatical functions such as subject, object, tense, number, etc.
• Grammatical roles (f-structure) tell us what role each word plays.

These levels are not mixed together. They are connected using mapping principles.
C-Structure and F-Structure in LFG
LFG: Example
Rule 1: S → NP VP
A Sentence (S) consists of a Noun Phrase + Verb Phrase
Example: NP = She
VP = saw stars in the sky

Rule 2: VP → V (NP) (NP) PP* {S'}

A Verb Phrase can contain:
• a Verb (V)
• one or two NPs (objects)
• zero or more PPs
• sometimes a clause (S′)
Example:
• V = saw
• NP = stars
• PP = in the sky
Rule 3: PP → P NP
• A Prepositional Phrase = Preposition + Noun Phrase
Example:
• P = in
• NP = the sky

Rule 4: NP → Det N {PP} Rule 5: S' → Comp S

A Noun Phrase can have: A subordinate clause starts with a complementizer
(that, if, whether)
• Determiner + Noun (Not used in this example)
• optionally a PP
Example:
• Det = the
• N = sky
LFG analysis at two levels:
• She N
1. c-structure → sentence tree (form) •PRO → pronoun
2. f-structure → grammatical information (function) •3rd person → she (not I / you)
•SG → singular
She saw stars •FEM → feminine
•NOM → nominative case (subject case)

• Verb is see
It needs: one subject
one object
PAST tense
Verb connects subject and object

• Noun = star
3rd person
Plural
This is the OBJECT
PRED (Predicate) GEN (Gender)
PRED tells us the basic meaning + argument structure of a GEN tells the gender of a noun or pronoun.
word. Common values:
“What kind of word is this, and what does it require?” •FEM → feminine (she, woman)
•MASC → masculine (he, man)
Examples: •NEUT → neutral (it)
•She → PRED = ‘PRO’
→ it is a pronoun Example:
•Stars → PRED = ‘STAR’ •She → GEN = FEM
→ it is a noun
CASE
•Saw → PRED = ‘see <(SUBJ)(OBJ)>’
CASE shows the grammatical role of a noun/pronoun in a
→ the verb see needs:
sentence.
• a subject
Common cases:
• an object
•NOM (Nominative) → subject
PRED = core meaning + grammatical needs
•ACC (Accusative) → object
•GEN (Genitive) → possession (his, her book)

Example:
She → CASE = NOM
→ because she is the subject
Her → CASE = ACC
→ object form
CASE = subject / object / possession role
 LFG analysis at two levels:
•c-structure → sentence tree (form)
•f-structure → grammatical information (function)

Sentence: She saw stars

Your text here

Fig: C-Structure of sentence Fig: F-Structure of sentence

Paninian Framework
Features of Indian Languages
Layered Representation in PG
Karaka Theory
A Karaka is a participant directly involved in the action of the verb.
Karaka theory explains the role each noun plays with respect to the verb.

Karakas = people/things connected to that action

This Theory explains who is doing what, to whom, how, from where, and where.

It is the core idea of Paninian Grammar (PG), Very useful for Indian languages because they use case
markers like ne, ko, se, mein etc.
Main Types of Karakas
Karaka Types Meaning Question it answers
Karta Doer Who did it?
Karma Object / thing affected What was done?
Karana Instrument By what means?
Sampradana Receiver / beneficiary To whom?
Apadan Source / separation From where?
Adhikaran Location Where?
Common Case Markers in Hindi (with meaning)

Case Marker Meaning Karaka Role

ne by (doer) Karta
ko to Karma / Sampradana
se by / from Karana / Apadan
dwara by means of Karana
mein in Adhikaran
par on Adhikaran
ka / ke / ki of Possession (not a Karaka)
Example
1. Maan bachchi ko aangan mein haath se rotii khilaatii hai
Meaning: The mother feeds bread to the child by hand in the courtyard.
Karta (Doer)
• maan (mother)
She is doing the action (feeding)
• Karta = maan
Case marker: ne (often, but not always)
Karma (Object)
• rotii (bread)
The thing being fed
• Karma = rotii
Case marker: ko or sometimes zero (no marker)
Example
Maan bachchi ko aangan mein haath se rotii khilaatii hai
• The mother feeds bread to the child by hand in the courtyard.
Sampradana (Recipient)
• bachchi ko (to the child)
The receiver of the action
• Sampradana = bachchi
Case marker: ko
Karana (Instrument)
• haath se (by hand)
Means used to do the action
• Karana = haath
Case marker: se / dwara
Adhikaran (Location)
• aangan mein (in the courtyard)
Place where the action happens
• Adhikaran = aangan
Case marker: mein / par
Exercise
2. Maan ne thaali se khana uthakar bachthe ko diyaa
Meaning: The mother gave food to the child taking it up from the plate
3. Ram ne Sita ko kalam se chitthi ghar mein likhi.
Meaning: Ram wrote a letter to Sita with a pen in the house.
2. Maan ne thaali se khana uthakar bachthe ko diyaa
Meaning: The mother gave food to the child taking it up from the plate

Word / Phrase Karaka Explanation

maan ne Karta Mother is the doer of the action
khana Karma Food is the thing affected / given
bachche ko Sampradana Child is the receiver / beneficiary
Plate is the source (from where
thaali se Apadan
food is taken)
(—) Karana Not explicitly mentioned
(—) Adhikaran Not explicitly mentioned
 3. Ram ne Sita ko kalam se chitthi ghar mein likhi.
Meaning: Ram wrote a letter to Sita with a pen in the house.

Word Karaka Role

Ram ne Karta Doer of the action
chitthi Karma Object (thing written)
Sita ko Sampradana Receiver
kalam se Karana Instrument
ghar mein Adhikaran Location