This document discusses the intersection of linguistics and natural language processing. It provides a brief history of linguistics from ancient grammarians to modern theorists like de Saussure, Chomsky, and Halliday. Key concepts in linguistics like the sign, universal grammar, and language as a social semiotic system are outlined. The document also describes the five levels that languages are studied: phonetics/phonology, morphology, syntax, semantics, and pragmatics. It concludes by explaining how modern NLP techniques like word2vec, speech recognition, and language models apply linguistic concepts.
3. Languages have been studied since
the beginning
Panini
• Sanskrit
Morphology
• 6th century
B.C.
Plato
• Cratylus
dialogue
• Eternal
concepts that
exist in the
world of ideas
University of
Alexandria
• "téchnē
grammatikḗ"
(Τέχνη
Γραμματική)
• 280 B.C.
Jacob Grimm
• Detuche
grammatik
• 18th century
Wilhelm von
Humbolt
• rule-governed
system
• create an
infinite
number of
sentences
using a finite
grammatical
rules.
Ferdinand de
Saussure
4. The concept of a sign
Signified
Concept
Signifier
Sound- image Arbor
5. Noam Chomsky and the
generativism
• Based on syntax. Also addressed
phonology and morphology
• Universal grammar: a set of
syntactic rules universal for all
humans and underlying the
grammars of all human
languages
• The universal grammar is
innate to the human brain
• The primary study must be
defined by a dependency tree
Source: https://en.wikipedia.org/wiki/Phrase_structure_rules
6. Michael Halliday and the
functionalism
• Based on social interaction
• A language is a social semiotic
system
• Languages are more than a set
of sentences, but the exchange of
meaning in interpersonal (social)
contexts, based on choices made
by the speakers.
7. Languages are studied in 5 levels
Phonetics
and
phonology
1
• Sounds of human
speech
• Phonetics how
sounds are
produced,
transmitted and
received
• Phoneme
perceptually unit
of sound
• Phonology
classifies sounds
in a system of a
language
8. Languages are studied in 5 levels
Phonetics
and
phonology
1
• Sounds of human
speech
• Phonetics how
sounds are
produced,
transmitted and
received
• Phoneme
perceptually unit
of sound
• Phonology
classifies sounds
in a system of a
language
Morphology
2
• How words are
formed and
related to each
other
• Morphemes
smallest unit of
meaning
9. Languages are studied in 5 levels
Phonetics
and
phonology
1
• Sounds of human
speech
• Phonetics how
sounds are
produced,
transmitted and
received
• Phoneme
perceptually unit
of sound
• Phonology
classifies sounds
in a system of a
language
Morphology
2
• How words are
formed and
related to each
other
• Morphemes
smallest unit of
meaning
Syntax
3
• Phrases
• Set of rules and
principles that
govern the
structure of
phrases or
sentences.
• Word combination
to transmit the
proper meaning
10. Languages are studied in 5 levels
Phonetics
and
phonology
1
• Sounds of human
speech
• Phonetics how
sounds are
produced,
transmitted and
received
• Phoneme
perceptually unit
of sound
• Phonology
classifies sounds
in a system of a
language
Morphology
2
• How words are
formed and
related to each
other
• Morphemes
smallest unit of
meaning
Syntax
3
• Phrases
• Set of rules and
principles that
govern the
structure of
phrases or
sentences.
• Word combination
to transmit the
proper meaning
Semantics
4
• Literal meaning
of words as
considered
principally as
parts of the
human language
system
11. Languages are studied in 5 levels
Phonetics
and
phonology
1
• Sounds of human
speech
• Phonetics how
sounds are
produced,
transmitted and
received
• Phoneme
perceptually unit
of sound
• Phonology
classifies sounds
in a system of a
language
Morphology
2
• How words are
formed and
related to each
other
• Morphemes
smallest unit of
meaning
Syntax
3
• Phrases
• Set of rules and
principles that
govern the
structure of
phrases or
sentences.
• Word combination
to transmit the
proper meaning
Semantics
4
• Literal meaning
of words as
considered
principally as
parts of the
human language
system
Pragmatics
5
• Use of basic
meaning and
context
• Overcome
ambiguities by
talkers
12. word2vec and linguistics
Each embedding
represents the literal
meaning of a word
The representation of
the word is given by its
context
Source: https://www.distilled.net/word2vec-examples/
13. Speech recognition and linguistics
Matching filtered,
normalized and divided
sounds with specific
phonemes
14. Language models and linguistics
Language
model
Wiki
text
103
Language
model
IMDb ClassifierIMDb
Learn how words are
put together
Structure of the
language