Contents Introduction

• Introduction

• Syntactic processing

• Semantic processing

• Discourse processing

• Pragmatic processing

• Statistical NLP

• Spell checking

2 3

Introduction Introduction

4 5
 Introduction Property of Language
Problem: Sentences are incomplete descriptions of ideas they want to
convey.

Advantage: Language allows speakers to be as vague or as precise about

anything they want to convey. Speakers can leave out things that they
assume the listener knows already.

Some children visited us today.

• Some children visited us today.

• Four children visited us today.
• Roy, Aman, Arya and Shyam visited us
6 today 7

Property of Language Property of Language

Problem: same expression means different things in different context. Problem: No language is complete. It continuously evolves as new words,
expressions and meanings get generated very frequently.
Advantage: Language allows us to communicate about an infinite world using
a finite and learnable number of symbols Advantage: languages can evolve because the experiences that we want to
communicate also evolve.

Give me the scissors.(Kitchen) I shall call you

Give me the scissors.(Tailoring shop) I shall SMS you

Give me the scissors.(OT) I shall instagram.

8 9
 Property of Language The two major questions in NLP
Problem: There are lots of way to say the same thing. Two main questions:

Advantage: When we know a lot, the facts imply each other. Language in AI • What is language understanding?
is intended to be used by agents who know a lot.
• What does a sentence mean?

The college closes on 14th December after the VTU exams .

The vacations start on 15th December.

10 11

The two major questions in NLP Neural History of NLP

Two main questions:

• What is language understanding?

• What does a sentence mean?

12 13
Phases of NLP Phases of NLP

14 15

Phase 1: Morphological Phase 1: Morphological

Analysis Analysis
Morphology: Morphology is the study of the structure and formation of words. Its
most important unit is the morpheme, which is defined as the "minimal unit of
meaning". (Linguistics textbooks usually define it slightly differently as "the
minimal unit of grammatical analysis".)

Morphological Analysis: Individual words are analyzed into their components.

Non word tokens are separated from their parent words. E.g., punctuations and
apostrophes.

Knowledge: typically large amount of knowledge is required to solve complex

problems in AI

16 17
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis

18 19

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
• Grammar is very essential and important to describe the syntactic structure of
well-formed programs.
• In the literary sense, they denote syntactical rules for conversation in natural
languages.
• The theory of formal languages is also applicable in the fields of Computer
Science mainly in programming languages and data structure.

A mathematical model of grammar was given by Noam Chomsky in 1956, which

is effective for writing computer languages.

Mathematically, a grammar G can be formally written as a 4-tuple (N, T, S, P)

where −

• N or VN = set of non-terminal symbols, i.e., variables.

• T or ∑ = set of terminal symbols.
• S = Start symbol where S ∈ N
• P denotes the Production rules for Terminals as well as Non-terminals. It has
the form α → β, where α and β are strings on VN 𝖴 ∑ and least one symbol of
20 α belongs to VN 21
 Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
• A sentence in the language defined by a CFG is a series of words that can be
derived by systematically applying the rules, beginning with a rule that has s on its
left-hand side.

• A parse of the sentence is a series of rule applications in which a syntactic category

is replaced by the right-hand side of a rule that has that category on its left-hand
side, and the final rule application yields the sentence itself.

• Grammar:
฀ Context Free Grammar (CFG)

฀ Phrase Structure or Constituency Grammar

฀ Dependency Grammar

22 23

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Dependency Grammar (DG)

• It is opposite to the constituency grammar and based on dependency relation. It was

introduced by Lucien Tesniere.

• Dependency grammar (DG) is opposite to the constituency grammar because it lacks

phrasal nodes.

• In DG, the linguistic units, i.e., words are connected to each other by directed links.

• The verb becomes the center of the clause structure. Every other syntactic units are
connected to the verb in terms of directed link. These syntactic units are called
dependencies.

24 25
 Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
Phrase Structure or Constituency Grammar (CG)

• Phrase structure grammar, introduced by Noam Chomsky, is based on the

constituency relation. That is why it is also called constituency grammar. It is opposite
to dependency grammar.

• All the related frameworks view the sentence structure in terms of constituency
relation.

• The constituency relation is derived from the subject-predicate division of Latin as

well as Greek grammar.

• The basic clause structure is understood in terms of noun phrase NP and verb
phrase VP.

26 27

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Context Free Grammar Set of Non-terminals
(CFG) It is denoted by V. The non-terminals are syntactic • np - noun phrase
variables that denote the sets of strings, which
further help defining the language, generated by
• vp - verb phrase
the grammar. • s - sentence
Set of Terminals
It is also called tokens and defined by Σ. Strings
• det - determiner (article)
are formed with the basic symbols of terminals. • n - noun
Set of Productions
It is denoted by P. The set defines how the • tv - transitive verb (takes an object)
terminals and non-terminals can be combined. • iv - intransitive verb
Every production(P) consists of non-terminals, an
arrow, and terminals (the sequence of terminals). • prep - preposition
Non- terminals are called the left side of the • pp - prepositional phrase
production and terminals are called the right side of
the production. • adj - adjective
Start Symbol
The production begins from the start symbol. It is
denoted by symbol S. Non-terminal symbol is 28 29
always designated as start symbol.
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis

• E.g., a parse of the sentence "the giraffe

dreams" is:

s => np vp => det n vp => the n vp => the giraffe

vp => the giraffe iv => the giraffe dreams

30 31

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis

32 33
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
The main roles of the parse include −

• To report any syntax error.

• To recover from commonly occurring error so that the processing of

the remainder of program can be continued.

• To create parse tree.

• To create symbol table.

• It is used to implement the task of parsing. It may be defined as the
software component designed for taking input data (text) and giving
structural representation of the input after checking for correct syntax as • To produce intermediate representations (IR).
per formal grammar.
• It also builds a data structure generally in the form of parse tree or abstract
syntax tree or other hierarchical structure.

34 35

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
• Types of Parsing
• Derivation divides parsing into the followings two types −

• Top-down Parsing

• Bottom-up Parsing

36 37
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis

38 39

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis

40 41
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
• Thus parsing converts flat list of words that form sentences into a • Example of syntactically correct but semantically wrong sentences.
hierarchical structure that defines the units that are represented by that flat
list. ฀ The lion eats the rabbit.

• These hierarchical structural units will correspond to meaning units when ฀ The rabbit eats the lion.
semantic analysis is performed.

• We can create a list of Reference Markers (RM). They provide a place to

accumulate information about the entities as and when we get it.

• If there this syntactic processing is not performed, then the semantic

processing should decide its own constituents.

• If parsing is done, it constraints the number of constituents that semantics

can consider.

• Syntactic processing is computationally less expensive than is direct

semantic processing (which may require substantial inference).
42 43
• Reduces overall system
complexity.

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis

44 45
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
Concept of Derivation Sentence: a and b or c

In order to get the input string, we need a sequence of production rules. Derivation
Derivation is a set of production rules. During parsing, we need to decide the
non-terminal, which is to be replaced along with deciding the production rule with N
the help of which the non-terminal will be replaced. ⇒ N ‘or’ N
⇒ N ‘or c’
Types of Derivation ⇒ N ‘and’ N ‘or c’
Left-most Derivation ⇒ N ‘and b or c’
In the left-most derivation, the sentential form of an input is scanned and replaced ⇒ ‘a and b or c’
from the left to the right. The sentential form in this case is called the left-sentential
form. Sentential Form: Rightmost derivation
Right-most Derivation
In the left-most derivation, the sentential form of an input is scanned and replaced • In this derivation each line is called a sentential form.
from right to left. The sentential form in this case is called the right-sentential form. • Furthermore, each line of the derivation applies a rule from the CFG in order to
show that the input can, in fact, be derived from the start symbol N.
• In the above derivation, we restricted ourselves to only expand on the
rightmost non-terminal in each sentential form. This method is called the
46 derivation of the input using a CFG.
rightmost 47

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Rightmost Derivation Reverse Order

‘a and b or c’
⇒ N ‘and b or c’ # use rule N -> a
⇒ N ‘and’ N ’or c’ # use rule N -> b
⇒ N ‘or c’ # use rule N -> N and N
⇒ N ‘or’ N # use rule N -> c
⇒N # use rule N -> N or N

48 49
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis

50 51

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Item Top down Bottom up Transition Networks

Definition A parsing strategy that first A parsing strategy that first looks at My dog ate a frog. Parse Tree Derivation
looks at the highest level of the lowest level of the parse tree and
the parse tree and works works up the parse tree by using the
down the parse tree by using rules of a formal grammar.
the rules of a formal
grammar.
Functionality The parsing occurs from the The parsing occurs from the starting
starting symbol to the input symbol to the input string
string.
Main decision To select what production To select when to use a
rule to use in order to production rule to reduce the
construct the string string to get the starting symbol
Method of Uses leftmost derivation Uses rightmost derivation
construction

52 53
Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
Transition Networks Transition Networks

My dog ate a frog. My dog ate a frog.

54 55

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Recursive Transition Networks (RTN)
Transition Networks
• Natural languages allow us to express an infinite range of ideas using a finite set
My dog ate a frog. of rules and symbols. We need the notion of recursion to do this.
• Not only are there an infinite number of sentences, but a single sentence can be
infinitely long. Recursion enables us to represent this capability of language as
well.
• While we can generate infinitely long sentences in English, such sentences will not
necessarily be understandable given real constraints on human processing and
memory, human lifespan, the age of the universe, and stuff like that

The boy drowned.

The boy with the raft drowned.

The boy with the raft near the island drowned.

The boy with the raft near the island in the ocean drowned.
56 57
The boy ....
 Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
Backtracking Backtracking

• RTNs describe a process • You need to backtrack to the last decision point. In order to backtrack, your
parser must record at every decision point:
• The RTN parser "knows" at each step of the parsing process exactly which choices
are available, or what knowledge will be relevant. On the other hand, a CFG-based • What word in the sentence it was looking at when it made the decision
parser will have to search through the list of rules each step of the way to find the
applicable ones. • What node (or state) in the network it was at

• If there's more than one choice, how does the parser know which one is the • What other arcs (or paths) could have been chosen but weren't
right one? It doesn't, so we have to introduce the notion of backtracking.
• When the parser bumps into a dead end like that described above, it must reset the
The little orange ducks swallow flies. parsing process so that it's looking at the previously stored word, it's restarting from
the previously stored node, and it's traversing the first of the previously stored arcs.
• (If there are more than one arc remaining, just push the information again, without the
first arc, and proceed as usual.)

58 59

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Augmented Transition Network (ATN) Augmented Transition Network (ATN)

• Recursive transition networks are interesting from a theoretical standpoint but are not • To record the structure of the input, we add an action to each arc which stores
of much use by themselves. the word that was processed while traversing that arc in an appropriate register.

• From a computational perspective, a black box that accepts English input and just
says "yes“ or "no" doesn't buy us much.

• What we need is a black box that records the structure of the input as well as
providing an evaluation of syntactic correctness. We do this by adding procedures
to the arcs of the RTN.

• These procedures are then performed when the corresponding arcs are traversed.
The resulting network is called an "augmented transition network" (ATN). (And
sometimes, adding procedures to the arcs of a transition network is called "procedural
attachment.")

• While we add procedures or augmentations in ATN, we store information in

registers when arcs are traversed. 60 61
 Phase 2: Syntactic Phase 2: Syntactic
Analysis Analysis
Augmented Transition Network Augmented Transition Network (ATN) for feature tests:
(ATN)
Subject verb agreement: To perform agreement checks in the ATN framework,
we attach "feature tests" to the arcs.

62 63

Phase 2: Syntactic Phase 2: Syntactic

Analysis Analysis
Augmented Transition Network (ATN) for feature tests: Augmented Transition Network (ATN) for feature tests:

Subject verb agreement: To perform agreement checks in the ATN framework, Subject verb agreement:
we attach "feature tests" to the arcs.
• Now, if we try to parse a sentence like "the dog eat the frog" our ATN parser will
catch the agreement error when it performs the intersection on the number of "dog"
and the number of "eat“.

• The result is the empty set.

• If the intersection between the set of all possible values for the number of the subject
and the set of all possible values for the number of the verb is the empty set, then
there is no agreement between the subject and the verb. The sentence must be
syntactically incorrect.

64 65
 Phase 2: Syntactic Phase 3: Semantic
Analysis Analysis
Augmented Transition Network (ATN) for feature tests:

Sub categorization of the verb:

• Feature tests can also be used to check for agreement between auxiliary verbs and
the main verb in a verb group.

• They can also be used to check for the correct verb complement structure.

• The feature tests can be used to see if the right number of noun phrases follow the
verb.

• Remember that intransitive verbs expect no noun phrases, transitive verbs expect
one noun phrase, and bitransitive verbs expect two noun phrases.
• It must map individual words into appropriate objects in the knowledge base
• Also note that any given verb might fall into more than one of these categories. or database

• Information about the complement structure is called the "subcategorization" of the • It must create the correct structures to respond to the way the meanings or the
verb. individual words combine with each other.
66 67

Phase 3: Semantic Phase 3: Semantic

Analysis Analysis
Lexical Processing

• Look up individual words in dictionary (lexicons) and extract their meanings

• Many words have multiple meanings, may not be possible to choose the correct
one just by looking at the word itself.

• Lexical ambiguity:
• Mean:
❖ To signify
❖ Unpleasant or cheap
❖ Statistical average

• Lexical disambiguation: associating with each word in the lexicon, the

information about the contexts in which each of the word’s senses may appear.
Each of the words in a sentence can serve as part of the context in which h
meanings o the other words must be determined.
68 69
Phase 3: Semantic Phase 3: Semantic
Analysis Analysis
Lexical Processing
Semantic Grammar
• Semantic markers:
• Choice of non-terminals and production rules is governed by semantic as well as
❖ Physical – object -฀ I saw the diamond shimmering
syntactic functions.
❖ Animate – object -฀ My house loves the Sun.
• Usually a semantic action associate with each grammar rule
❖ Abstract – object -฀ You are a gem.
• The result of parsing and applying all the associated semantic actions is the
Sentence Level Processing meaning of the sentence.
• Semantic grammar: combines the following knowledge into a single set of rule in
the form of grammar. The result of parsing with this grammar will be semantic • The grammar rules themselves are designed around key semantic concepts.
parsing
❖ Syntactic • The proper constituents of semantic grammar are noun phrase (NP), verb phrase
❖ Semantic (VP), Noun (N), Verb (V), Preposition phrase (PP), Adverb (ADV) and so on.
❖ Pragmatic
• One of the successful applications of semantic grammar is the LIFER system
• Case grammar : parsing output with semantic information + further interpretation (A database query system).

• Conceptual parsing: syntactic + semantic knowledge combined into single • Semantic grammars are suitable for use in systems with restricted grammars since
information system, guided by semantic knowledge. its power of computation is limited.
• Approximately compositional semantic interpretation: semantic processing applied • Semantic grammars are grammars whose non-terminals correspond to semantic
to results of syntactic parsing. 70 concepts (e.g., [greeting] or [suggest.time] ) rather than to syntactic
constituen7ts1 (such as Verb or WounPhrase).

Phase 3: Semantic Phase 3: Semantic

Analysis Analysis
Advantages of Semantic Grammar Semantic Grammar
• They have the advantage that the semantics of a sentence can be directly read off
its parse tree.

• When the parse is complete, the result can be used immediately without the
additional stage of processing that would be required if a semantic interpretation
had not already been performed during the parse.

• Many ambiguities that would arise during a strictly syntactic parse can be avoided
since some of the interpretations do not make sense semantically and thus cannot
be generated by a semantic grammar.

• Syntactic issues that do not affect the semantics can be ignored.

Disadvantages of Semantic Grammar

• A new grammar must be developed for each domain.
• The number of rules required can become very large since many syntactic
generalizations are missed.
• Parsing process may be expensive due to large number of rules. 72 73
Phase 3: Semantic Phase 3: Semantic
Analysis Analysis
Case Grammar
Case Grammar
• Case Grammars attempt to describe any given sentence in terms of a fixed
frame of slots (called cases) which explicitly capture information about any
activities described in the sentence, the instigators of those activities, positions,
times, etc.
• Though there is no universally agreed set of cases or their names a common
subset is outlined below.

74 75

Phase 3: Semantic Phase 3: Semantic

Analysis Analysis
Case Grammar Case Grammar
• With Case Grammars rules describe syntactic constraints but also describe • ATNs provide a good structure for case grammar parsing.
manipulations geared towards producing case frames (sets of case slots).
• In traditional parsing algorithms, the output structure always mirrors the structure
• These may be flat (as in the example above) or nested hierarchical structures of the grammar rules that created it. However, in ATNs, the output structures
which can form the basis of semantic representations. can have arbitrary form.

• To implement a Case Grammar the notation used to express rules must allow • ATNs parsers can be used to translate English sentences into a semantic net
them to produce case frames (ie: structures without a pre-determined form) representing the case structures of the sentences. These semantic nets can
then be used to answer questions about the sentences.
• Grammar notations whose only output reflects the syntactic structure of their input
are not suitable since they cannot be used to construct case frames. • The result of parsing in a case representation I usually not a complete semantic
description of a sentence.
• These rules can be applied in reverse by a parser to determine the
underlying case structure from the superficial syntax. • The constituents that fill the slots may still be English words rather than true
semantic descriptions stated in the target representation.
• Parsing using a case grammar is usually expectation-driven. Once the verb of the
sentence has been located, it can be sed to predict the noun phrases that will
occur and to determine the relationship of those phrases to the rest of the 76 77
sentence.
Phase 3: Semantic Phase 3: Semantic
Analysis Analysis
Conceptual Parsing Conceptual Parsing
• The grammar of the system is bipartite.
• Finds both structure and meaning of a sentence in one step.
• The first part is a universal grammar exemplified by the conceptual rules employed
• Driven by a dictionary hat describes the meanings of words as Conceptual by the system.
Dependency (CD) structures.
• The second part is language-specific and is made up of r e realization rules
• Similar to case grammar. intended to map pieces of the conceptual network into linguistic items.

• Parsing process is heavily driven by a set of expectations that are set up on the • The realization rules may be used for both parsing and generating. However, it is
basis of the sentence’s main verb. not necessary to use all the realization rules in order to parse.

• However, here, the representation of a verb in CD is at a lower level than that of • The system is capable of making sense of a piece of language containing only a
a verb in a case grammar where the representation is often identical to the few words that it knows since its conceptual framework is capable of making
English word that is used as verb. predictions.

• CD usually provides a greater degree of predictive power. • Thus, it can understand while using only a few realization rules, whereas it would
need a great many more to map the same structure back into language. This
• The first step in mapping a sentence into its CD representation involves a phenomenon is similar to that observed in a man attempting to learn a foreign79
syn78tactic processor that extracts the main noun and verb. language.

Phase 3: Semantic Phase 3: Semantic

Analysis Analysis
Conceptual Parsing Conceptual Parsing

The big man steals the red book

80 81
Phase 3: Semantic Phase 3: Semantic
Analysis Analysis
Conceptual Parsing Compositional Semantic Interpretation

The tall boy went to the park with a girl • Here, syntactic parsing and semantic interpretation are treated as two separate
steps.

• The two must mirror each other in well-defined ways.

• Steps to generate a semantic interpretation for a strict syntactic parsing of a

sentence:
฀ Look up for each word in a lexicon that contains one or more definitions for
the word, each stated in terms of the chosen target representation. These
definitions must describe how the idea that corresponds to the word is to be
represented, and they may also describe how the idea represented by this
word may combine with the ideas represented by other words in the sentence.

฀ Use structure information in the output to provide additional constraints on

how to combine individual words to form larger meaningful words.

82 83

Phase 3: Semantic Phase 3: Semantic

Analysis Analysis
Compositional Semantic Interpretation: Montague Semantics Entailment
Whenever A is accepted as being true, B must also be accepted as being true,
• Montague semantics is a theory of natural language semantics and of its relation on
with syntax. grounds of meaning properties. Sentence (1) entails sentence (2)

• The most important features of the theory are its use of model theoretic semantics ฀ (1) Mary is singing and dancing.
which is nowadays commonly used for the semantics of logical languages and its
adherence to the principle of compositionality—that is, the meaning of the whole is ฀ (2) Mary is singing.
a function of the meanings of its parts and their mode of syntactic combination.
Entailment with inverse relationship, 4 entails 3
• Important relationships are:
฀ (3) No-one is singing and dancing.
• Entailment
• Valid reasoning ฀ (4) No-one is singing.
• Synonymy
• Ambiguity Entailment with transitive and intransitive verbs: 5 entails 6

฀ (5) John cools the soup.

84
฀ (6) The soup cools 85
Phase 3: Semantic Phase 3: Semantic
Analysis Analysis
Valid Reasoning Synonymy
A special case of entailment is ‘synonymy.’ Sentences are called synonymous just
A ‘valid reasoning’ is a generalization of entailment that involves more sentences in
than case they entail each other. For example, sentences (13) and (14) are synonymous:
two. If someone accepts (7) and (8), they may conclude correctly (9):
฀ (13) John or Mary comes.
฀ (7) John sings. ฀ (14) John comes or Mary comes.
฀ (8) John is a man.
฀ (9) A man sings. ฀ (15) Mary admires herself, and Sue does too.
฀ (16) Mary admires herself, and Sue admires herself
฀ (10) John knows whether Mary comes.
฀ (11) Mary comes. Ambiguity
฀ (12) John knows that Mary comes
฀ (17) Mary believes that she is ill, and Sue does too.
฀ (10) John knows whether Mary comes. ฀ (18) Mary believes that she is ill, and Sue believes that Mary is ill.
฀ (11a) Mary does not come.
฀ (12a) John knows that Mary does not come ฀ (19) Sue believes that Mary is ill
฀ (20) Sue believes that she (= Sue) is ill.
86
฀ (21) Ten years ago, John met the president of the 87
USA.
฀ (22) The president of the USA is Bush.

Phase 3: Semantic Phase 4: Discourse and pragmatic processing

Analysis
Extended Reasoning with a knowledge base
• Identical entities
• A significant amount of world knowledge is required to do semantic
interpretation. The knowledge is required to enable the system to choose among ฀ Bill had a red balloon
competing interpretations. ฀ John wanted it. (It is anaphoric reference or anaphora)

฀ John made a huge wedding cake with chocolate icing • Parts of entities
฀ John made a huge wedding cake with Bill’s mixer.
฀ John made a huge wedding cake with a giant tower covered with roses. ฀ Ram opened the dress that he just bought.
฀ John made a cherry pie with a giant tower covered with roses. ฀ The collar had a damage.

• Knowledge enables the system to accept meanings that it has not been explicitly • Parts of actions
told about.
฀ Jim went on a business trip to Delhi
฀ John made a huge wedding cake with chocolate icing ฀ He left by the early morning flight.
฀ John made a huge wedding cake with Bill’s mixer.
฀ John made a huge wedding cake with a giant tower covered with roses. • Entities involved in actions
฀ John made a cherry pie with a giant tower covered with roses.
฀ The bank was broken into last week. 89
฀ They took the cash and jewels.
88
Phase 4: Discourse and pragmatic processing Phase 4: Discourse and pragmatic processing

• Elements of sets • Illocutionary force

฀ The stickers we have in store are the lion, tiger, giraffe and ฀ It sure is going to rain today/
octopus.
฀ I will take the octopus. • Implicit presuppositions

• Names of individuals ฀ Did the Delhi flight come late?

฀ Harry went to the party. • Prerequisites

• Casual chain ฀ The current focus of the dialogue.

฀ A model of each participant’s current belief.
฀ It was raining heavily yesterday. ฀ The goal-driven character of the dialogue.
฀ The schools were closed yesterday. ฀ The rules of the conversation shared by all participants.
฀ The schools were closed yesterday.
• Planning sequences
• Planning sequences
฀ Kiran wanted a new house. 90 91
฀ Kiran decided to get a better job. ฀ Kiran wanted a new house.
฀ Kiran decided to get a better job.

Phase 4: Discourse and pragmatic processing

• Using focus in understanding

฀ Focus on relevant parts of the available knowledge base.

฀ Use that knowledge to resolve ambiguities and to make connections among
things that were said.

92