What is CL?

Simon Clematide

Institut für Computerlinguistik

Universität Zürich

Einführung in die Computerlinguistik I

Intro Disciplines Related Disciplines Difficulties

Outline
Intro
Applications
Disciplines
Computational Linguistics
Natural Language Processing
More
Modern CL
Related Disciplines
Linguistics
Cognitive Science
Difficulties
The Crux of Ambiguity
Contextualization
Summary

Einführung in die Computerlinguistik I Fall 2023 What is CL? 2 / 48

Intro Disciplines Related Disciplines Difficulties

Learning Objectives Originally by me

▶ Know of the subject of Computational Linguistics (CL), Natural

Language Processing (NLP) and Linguistics
▶ Being aware of the central objectives and tasks of the modern
computational linguistics
▶ Know the various scientific disciplines that are interdisciplinary
connected with CL
▶ Know the difficulties and crux of language processing

Einführung in die Computerlinguistik I Fall 2023 What is CL? 3 / 48

Intro Disciplines Related Disciplines Difficulties

Learning Objectives (stylized by ChatGPT after reading the lecture slides:-)

▶ Explore the multifaceted, interdisciplinary realm of computational

linguistics
▶ Distinguish between the core concepts of computational linguistics and
natural language processing
▶ Appreciate the significance of empirical methods in shaping modern
computational linguistics
▶ Delve into the complexities and challenges introduced by natural
language ambiguity
▶ Gain insights into pivotal tasks and real-world applications of NLP

Einführung in die Computerlinguistik I Fall 2023 What is CL? 4 / 48

Intro Disciplines Related Disciplines Difficulties Applications

CL Applications: Now or in the future

Partner task: Discuss with neighbor

What is the most important
(a) day-to-day application or
(b) future application of language technology for you?
▶
▶ Digital assistants with voice interfaces
▶ Word processing (spelling and grammar correction)
▶ Electronic dictionaries (thesauri)
▶ Automatic translation
▶ Text search in the WWW
▶ ...

Einführung in die Computerlinguistik I Fall 2023 What is CL? 5 / 48

Intro Disciplines Related Disciplines Difficulties Applications

Application I: Machine Translation (MT)

Google Translate in September 2017

Google Translate in September 2018▲

MT is THE first application of computational linguistics. Work started

right after World War II.
Einführung in die Computerlinguistik I Fall 2023 What is CL? 6 / 48
Intro Disciplines Related Disciplines Difficulties Applications

Application I: Machine Translation (MT)

Source: DeepL MT System▲

Einführung in die Computerlinguistik I Fall 2023 What is CL? 7 / 48

Intro Disciplines Related Disciplines Difficulties Applications

Application II: Information and Knowledge Management

Einführung in die Computerlinguistik I Fall 2023 What is CL? 8 / 48

Intro Disciplines Related Disciplines Difficulties Applications

Application II: Information and Knowledge Management I

Old Dream
▶ Machine reading
for knowledge
acquisition
▶ Key question:
Which knowledge
representation
should we use?

Einführung in die Computerlinguistik I Fall 2023 What is CL? 9 / 48

Intro Disciplines Related Disciplines Difficulties Applications

Application III: Assistance Systems

https://jibo.com/

Apple: Siri ; Amazon: Alexa ; Google: Google Assistant, Bard; Microsoft: Cortana
(RIP); OpenAI: ChatGPT; . . .
Einführung in die Computerlinguistik I Fall 2023 What is CL? 10 / 48
Intro Disciplines Related Disciplines Difficulties Applications

Microsoft’s Framework for Commercial Chatbots▲

A Modern NLP Tooling and Application Landscape

Einführung in die Computerlinguistik I Fall 2023 What is CL? 11 / 48
Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

What is CL?

The answer from ACL▲ (Association for Computational Linguistics)

Computational linguistics is the scientific study of language from a
computational perspective.
Computational linguists are interested in providing computational models of
various kinds of linguistic phenomena.
Question I
Is downloading tweets containing the hashtag #uzh from Twitter to your
computer an application of CL?
Question II
Is the machine for speech synthesis described in 1791 by Wolfgang von
Kempelen (1734-1804) an early application of CL?
Von Kempelen’s machine was the first that allowed to produce not only some speech sounds, but
also whole words and short sentences. According to von Kempelen, it is possible to acquire an
admirable ability in playing the machine within three weeks.
Einführung in die Computerlinguistik I Fall 2023 What is CL? 13 / 48
Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

Speech synthesis according to Kempelen▲

Figure: From “Mechanismus der menschlichen Sprache nebst Beschreibung einer

sprechenden Maschine”▲ (Mechanism of human speech together with description
of a speaking machine) (1791) [Traunmüller 1997]

Einführung in die Computerlinguistik I Fall 2023 What is CL? 14 / 48

Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

What is Natural Language Processing?▲

Natural language processing (NLP)

is a subfield of linguistics, computer science, information engineering, and
artificial intelligence concerned with the interactions between computers
and human (natural) languages, in particular how to program computers to
process and analyze large amounts of natural language data.
▶ Application-oriented development of “Lingware”

Einführung in die Computerlinguistik I Fall 2023 What is CL? 15 / 48

Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

Related/Alternative Disciplines

Deutsch Englisch
Linguistische Datenverarbeitung (LDV) Linguistic and Literary Computing (LLC)
Maschinelle Sprachverarbeitung Natural Language Processing (NLP)
Automatische Sprachverarbeitung (ASV) Natural Language Engineering (NLE)
Sprachtechnologie Speech Processing
Sprachtechnologie Human Language Technology (HLT)
Texttechnologie Text Technology
Computerphilologie Computational Philology
Digitale Geisteswissenschaften Digital Humanities
Korpuslinguistik Corpus Linguistics
Medieninformatik
Linguistische Informatik
Informationslinguistik

Einführung in die Computerlinguistik I Fall 2023 What is CL? 16 / 48

Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

Focus of Related Disciplines

spoken vs. written language

auxiliary science vs. independent research
theory-oriented vs. application-oriented
algorithmization as proof of
vs. efficient industry software
concept
cognitive plausibility vs. engineered solution
symbolic, logic-based and statistic, probabilistic and
vs.
theory-driven approaches data-driven approaches

Einführung in die Computerlinguistik I Fall 2023 What is CL? 17 / 48

Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

4 Central Concerns of Modern CL

1. Formalization of natural language phenomena and their

operationalization on the computer (mostly using machine learning)
2. Resources: Construction, management and linguistic interpretation
(recognition of word types, basic forms, syntactic and semantic
structures) of large collections of texts (corpora) and structured
resources (lexicons, annotated text material)
3. Applications: Development of technologically advanced software
4. Evaluation: Systematic empirical comparison of methods and language
processing systems

Einführung in die Computerlinguistik I Fall 2023 What is CL? 18 / 48

Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

Google Books Word N-Gram Search

Search phrases: phone , mobile phone , sms , telegram , fax

Which of the 4 central concerns of CL are realized here?

(Formalization, Resources, Applications, Evaluation)
Einführung in die Computerlinguistik I Fall 2023 What is CL? 19 / 48
Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

Google Books Word N Gram Search II

Search phrases: lead_VERB, lead_NOUN

Which of the 4 central concerns of CL are realized here?

(Formalization, Resources, Applications, Evaluation)
Einführung in die Computerlinguistik I Fall 2023 What is CL? 20 / 48
Intro Disciplines Related Disciplines Difficulties CL NLP More Modern CL

Syntactic Analysis (Parsing)

Which of the 4 central concerns of CL are realized here?

(Formalization, Resources, Applications, Evaluation)
Einführung in die Computerlinguistik I Fall 2023 What is CL? 21 / 48
Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Related Disciplines: Overview

▶ Linguistics: formal linguistics for language models; descriptive

linguistics and corpus linguistics for empirical data
▶ Computer Science: practical and theoretical computer science (What
and how efficiently can something be computed?)
▶ Machine Learning, i.e., artificial Intelligence (Deep Learning); Data
Science getting more important
▶ Mathematics: Set theory (functions, relations, trees, graphs), linear
algebra (vector and matrix calculations), differential calculus
(numerical optimization)
▶ Statistics
▶ Cognitive Science and psychology
▶ Logic and philosophy of language

Einführung in die Computerlinguistik I Fall 2023 What is CL? 23 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

QUIZ: Who does what?

Which disciplines are described here in connection with corpora (=

text collections)?
1. Use of corpora for applications such as speech recognition, automatic
translation, etc.
2. The systematic compilation of electronic text material that allows the
scientific study of language phenomena.
3. Automatic linguistic analysis of texts using sophisticated programming
techniques.

Answers
□ Computational Linguistics
□ Natural Language Processing
□ Corpus Linguistics

Einführung in die Computerlinguistik I Fall 2023 What is CL? 24 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Linguistics▲

Linguistics is the scientific study of language

▶ written or spoken
▶ linguistic description levels: phonetics/phonology, morphology, syntax,
semantics, pragmatics

Modern structural linguistics . . .

▶ formalizes linguistic regularities
▶ into explanatory models.

Einführung in die Computerlinguistik I Fall 2023 What is CL? 25 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Phonetics▲ and Phonology▲ (Lautlehre)

Phonetics studies the sounds of human speech.

Physical properties of speech sounds, their physiological production,
acoustic properties, auditory perception, and neurophysiological status.

Phonology studies the systematic organization of sounds in speech

How do we systematically structure sounds (phonemes, syllables, etc.) to
convey meaning?

Phonological rule of final hardening

1. Dieb /di:p/
2. Diebe /di:b@/

Einführung in die Computerlinguistik I Fall 2023 What is CL? 26 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Morphology▲ (Wortlehre)

Morphology is the study of words

▶ How are they formed? Inflection, compounding, derivation, etc.
▶ What is their relationship to other words in the same language.

Inflection
1. Dieb-e DIEB-NOMEN-PLURALNOM/AKK /GEN “More than one thief”
2. Dieb-e DIEB-NOMEN-SINGULARDATIV “the thief”

Einführung in die Computerlinguistik I Fall 2023 What is CL? 27 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Syntax▲ (Satzlehre)

Syntax is the study of the structure of sentences.

▶ The rules, principles, and processes that govern how words assemble to
grammatical sentences.
▶ Predicates, subjects, objects in a sentence, and the combination of
subclauses.

Grammaticality, syntactic well-formedness

1. Der gewitzte Dieb stahl den Diamanten.
2. *Der Dieb gewitzte stahl den Diamanten.
3. *Den gewitzten Dieb stahl den Diamanten.

Einführung in die Computerlinguistik I Fall 2023 What is CL? 28 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Semantics▲ (Bedeutungslehre)

Semantics is the study of meaning

The meaning of words (lexical semantics), phrases and sentences (sentence
meaning), texts (discourse semantics).

Different wordings for the same meaning

1. Die Polizei beschlagnahmte das Diebesgut gestern.
2. Das Diebesgut beschlagnahmte die Polizei gestern.
3. Das Diebesgut wurde gestern von der Polizei beschlagnahmt.
4. Die Beschlagnahmung des Diebesgut durch die Polizei war gestern.
5. [Die Polizeii fasste die Täterin gestern.] Siei beschlagnahmte
gleichzeitig das Diebesgut.

Einführung in die Computerlinguistik I Fall 2023 What is CL? 29 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Pragmatics▲ (Sprachverwendungslehre)

Pragmatics is the study of language use in context.

▶ Communication goals of utterances
▶ Dialogues

Different utterances for the same goal

1. Ist das Fenster auf?
2. Bitte schliessen Sie das Fenster!

Einführung in die Computerlinguistik I Fall 2023 What is CL? 30 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Cognitive Science1

Definition
Cognitive science is the
interdisciplinary study of cognitive
abilities through psychology,
neuroscience, computer science,
linguistics and philosophy. Cognitive
abilities include perception, thinking,
learning, motor skills, emotion and
language.

© Wikipedia
1
https://en.wikipedia.org/wiki/Cognitive_science Interestingly the definition has changed slightly. . .
Einführung in die Computerlinguistik I Fall 2023 What is CL? 31 / 48
Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Turing Test: Can Machines Think? Can Machines Fool

Humans?

Turing test’s original formulation [Turing 1950]

The new form of the problem can be described in terms of a game which we call the
’imitation game’. It is played with three people, a man (A), a woman (B), and an
interrogator (C) who may be of either sex. The interrogator stays in a room apart front
the other two. The object of the game for the interrogator is to determine which of the
other two is the man and which is the woman. He knows them by labels X and Y, and at
the end of the game he says either “X is A and Y is B” or “X is B and Y is A”. [. . . ]
We now ask the question, “What will happen when a machine takes the part of A in this
game?” Will the interrogator decide wrongly as often when the game is played like this as
he does when the game is played between a man and a woman? These questions replace
our original, “Can machines think?”

Ungendered Turing Test: Talking to a human or machine?

From 1991 to 2019, public competitions (Loebner Prize) have been held
with dialogue systems.2 .
2
https://en.wikipedia.org/wiki/Loebner_Prize
Einführung in die Computerlinguistik I Fall 2023 What is CL? 32 / 48
Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Alexa Prize▲ (since 2016): Social Bot Challenge for

University Students

Modern Operationalization:
Meaningful conversation with humans, not fooling humans. . .
▶ Task: Have a 20 minute long engaged and coherent conversation with
a chat bot. Can also be multimodal (talk about images etc.)
▶ Evaluation: Humans rate the quality of the conversation on a scale.
▶ Example conversation from 2021 finalists:
https://youtu.be/1Cm36w3GLKw?t=321
▶ 2021-2023 edition▲ : $1,000,000 price

The winners were just announced...

Einführung in die Computerlinguistik I Fall 2023 What is CL? 33 / 48

Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

Deep Learning Communications with GPT-3▲

Dialog 1 Dialog 2
▶ Human: What is your favorite ▶ Human: How many bonks are in a quoit?
animal?
▶ GPT-3: There are three bonks in a quoit.
▶ GPT-3: My favorite animal is a
▶ Human: How many rainbows does it take to jump
dog.
from Hawaii to 17?
▶ Human: Why?
▶ GPT-3: It takes two rainbows to jump from
▶ GPT-3: Because dogs are loyal
Hawaii to 17.
and friendly.
▶ Human: What are two reasons Dialog 3
that a dog might be in a bad
mood? ▶ Human: Who was president of the United States
▶ GPT-3: Two reasons that a dog in 1620?
might be in a bad mood are if it is ▶ GPT-3: James I was president of the United
hungry or if it is hot. States in 1620.

What’s unnatural here? Try out how ChatGPT reacts to such

questions. . . The adaptation of powerful language generation models to
human expectations is called “alignment”.
Einführung in die Computerlinguistik I Fall 2023 What is CL? 34 / 48
Intro Disciplines Related Disciplines Difficulties Linguistics Cognitive Science

How sentient can AI be? A hot topic even before

ChatGPT appeared. . .

sentient
(1) responsive to or conscious of sense impressions: (2) aware; (3) finely
sensitive in perception or feeling
If you are interested: Look at Googles LaMDA system discussing about
itself, AI and consciousness . . .
Nice interview with LaMDA▲ . Is this system “sentient”?

Einführung in die Computerlinguistik I Fall 2023 What is CL? 35 / 48

One  expression,  many  meanings  
Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Syntac?c   &  seman?c  

Ambiguity on Syntactic and Semantica mbiguity  
Levels

syntac?c  
                                           S  
ambiguity  
                                                     S                                                          VP  

                                                                 VP   seman?c                                                                          PP  

© [Manning 2011]
ambiguity  
                     NP                                                    NP          NP                                                                    NP  

Fruit  flies  like  a  banana   Fruit  flies  like  a  banana  

Einführung in die Computerlinguistik I Fall 2023 What is CL? 37 / 48
Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Ambiguity [Jurafsky and Martin 2008]

Example
I made her duck.

What readings are thinkable?

▶ Which words are actually ambiguous when viewed in isolation?
▶ Write down paraphrases that make the possible readings clear
(disambiguate)!
▶ On which linguistic levels do the readings differ?

Einführung in die Computerlinguistik I Fall 2023 What is CL? 38 / 48

Intro Disciplines
OK,  why  else  is  NLP  hard?  
Related Disciplines Difficulties Ambiguity Contextualization

Why is CL/NLP hard? Oh, so many reasons. . .

Oh  so  many  reasons!  
non-­‐standard  English   segmenta1on  issues   idioms  
Great  job  @jus?nbieber!  Were  SOO   dark  horse  
the  New  York-­‐New  Haven  Railroad   get  cold  feet  
PROUD  of  what  youve  accomplished!  U  
taught  us  2  #neversaynever  &  you   the  New  York-­‐New  Haven  Railroad   lose  face  
yourself  should  never  give  up  either♥   throw  in  the  towel  

neologisms   garden  path  sentences   tricky  en1ty  names  

unfriend   The  man  who  hunts  ducks  out  on  weekends.   …  a  muta?on  on  the  for  gene  …  
retweet   Where  is  A  Bug’s  Life  playing  …  
The  cokon  shirts  are  made  from  grows  here.  
bromance  
Most  of  Let  It  Be  was  recorded  …  
teabagger  

world  knowledge   prosody   lexical  

specificity  
I  never  said  she  stole  my  money.  

© [Manning 2011]
Mary  and  Sue  are  sisters.  
I  never  said  she  stole  my  money.  
Mary  and  Sue  are  mothers.  
I  never  said  she  stole  my  money.  

But  that’s  what  makes  it  fun!  

Einführung in die Computerlinguistik I Fall 2023 What is CL? 39 / 48

Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Language:  
Use Scenario: Difficultiesthe  ul?mate  UI  

Where  is  A  Bug’s  Life  playing  in  Mountain  View?  

A  Bug’s  Life  is  playing  at  the  Century  16  Theater.  

When  is  it  playing  there?  

It’s  playing  at  2pm,  5pm,  and  8pm.  

OK.    I’d  like  1  adult  and  2  children  for  the  first  show.  
How  much  would  that  cost?  

But  we  need  domain  knowledge,  discourse  knowledge,  world  knowledge  

(Not  to  men?on  linguis?c  knowledge!)  
Source: [Manning 2011]

Einführung in die Computerlinguistik I Fall 2023 What is CL? 40 / 48

Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Disambiguation: Contextualization is key!

Why can modern chatbots like ChatGPT understand natural language?

Their underlying generative language model (GPT: Generative
Pre-Training) had to complete an incredible number of texts word by word!
Simple, but powerful language model training game: Given some
context, complete it!
Given some text, predict the most probable next . . .
▶ Learns world knowledge: “Elvis Presley is a famous . . . ”
▶ Learns discourse knowledge: ‘‘What is your opinion? . . . ”
▶ Learns linguistic knowledge: “The past tense of the verb “fly” is . . . ”
▶ Learns domain knowledge: “Assume you are a chatbot for movie
discussions.”
By playing the text completion game on large raw text data, this knowledge
enters the model!
Einführung in die Computerlinguistik I Fall 2023 What is CL? 41 / 48
Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Generating Text with Generative Language Models▲

Einführung in die Computerlinguistik I Fall 2023 What is CL? 42 / 48

Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Tasks, Difficulties, Enorm Progress (from 2011) Subfields  and  tasks  

mostly  solved   making  good  progress   s?ll  really  hard  
Spam  detec?on   Sen?ment  analysis   Seman?c  search  
OK,  let’s  meet  by  the  big  …   ✓ The  pho  was  authen?c  and  yummy.   people  protes?ng  globaliza?on   Search  

D1ck  too  small?  Buy  V1AGRA  …   ✗ Waiter  ignored  us  for  20  minutes.   …demonstrators  stormed  IMF  offices…  

Text  categoriza?on   Coreference  resolu?on   Ques?on  answering  (QA)  

Phillies  shut  down  Rangers  2-­‐0   SPORTS   Q.  What  currency  is  used  in  China?  
Jobless  rate  hits  two-­‐year  low   BUSINESS   Obama  told  Mubarak  he  shouldn’t  run  again.   A.  The  yuan  

Part-­‐of-­‐speech  (POS)  tagging   Word  sense  disambigua?on   Textual  inference  &  paraphrase  
         ADJ                  ADJ        NOUN    VERB            ADV   (WSD)   T.  Thirteen  soldiers  lost  their  lives  …  

Colorless      green      ideas      sleep      furiously.   I  need  new  bakeries  for  my  mouse.   H.  Several  troops  were  killed  in  the  …   YES  

Named  en?ty  recogni?on  (NER)   Syntac?c  parsing   Summariza?on  

PERSON                                  ORG                                            LOC   Sheen  
Sheen   con?nues  
Sheen  ccon?nues  
on?nues   Sheen  
rant   against  ……    

© [Manning 2011]
rant  
rant  aagainst  
gainst  …   is  nuts  
Obama  met  with  UAW  leaders  in  Detroit  …   I  can  see  Russia  from  my  house!  

Informa?on  extrac?on  (IE)   Machine  transla?on  (MT)    Discourse  

    &  dialog  
You’re  invited  to  our  bunga   Party   Our  specialty  is  panda  fried  rice.   Where  is  Thor  playing  in  SF?    
bunga  party,  Friday  May  27   May  27  
at  8:30pm  in  Cordura  Hall   add   我们的专长是熊猫炒饭   Metreon  at  4:30  and  7:30  

Einführung in die Computerlinguistik I Fall 2023 What is CL? 43 / 48

Intro Disciplines Related Disciplines Difficulties Ambiguity Contextualization

Your task: Which NLP tasks can ChatGPT solve out of

the box?

Please post your best examples into our lecture Forum!▲

Einführung in die Computerlinguistik I Fall 2023 What is CL? 44 / 48

Intro Disciplines Related Disciplines Difficulties

Summary

▶ Computational linguistics is a interdisciplinary multifaceted subject

with a strong application orientation (NLP).
▶ Modern empirical CL includes building language resources,
development of NLP-specific algorithms, machine learning and
systematic evaluation.
▶ The ambiguity of natural language is the biggest challenge for CL.
▶ Contextualization is key for natural language understanding.

Einführung in die Computerlinguistik I Fall 2023 What is CL? 45 / 48

Intro Disciplines Related Disciplines Difficulties

Your Turn for Further Study

▶ Required reading Chapter 1 Introduction▲ [Jurafsky and Martin

2008] history of CL (also available on OLAT
“Materials/Literatur/JurafkyMartin20082nd/01-Introduction.pdf”)
▶ Interesting blog post on Turing Test with GPT-3▲
▶ Interview with LaMDA▲
▶ A very recent scientific report on AI and consciousness
https://arxiv.org/abs/2308.08708 and a New York Times article
about it▲
▶ Test a conversational agent on its abilities to solve NLP tasks. . .

Einführung in die Computerlinguistik I Fall 2023 What is CL? 46 / 48

Intro Disciplines Related Disciplines Difficulties

Questions

▶ What disciplines does computational linguistics connect? With which

disciplines is it in close proximity?
▶ What is the difference between CL and NLP?
▶ Give one example that illustrates the type of analysis for each linguistic
sub-discipline.
▶ What are typical modern NLP tasks? See https://nlpprogress.com▲
▶ What are the 4 concerns of modern CL?
▶ Why is CL/NLP difficult?
▶ How can modern chatbots as ChatGPT know so much about our
world?

Einführung in die Computerlinguistik I Fall 2023 What is CL? 47 / 48

Intro Disciplines Related Disciplines Difficulties

References I

▶ Jurafsky, Daniel and J. H. Martin (2008).

Speech and Language Processing: An Introduction to Natural Language Processing,
Computational Linguistics, and Speech Recognition. Prentice-Hall, 2. ed.
▶ Manning, Christopher (2011).
Computational Linguistics (aka Natural Language Processing), electronic lecture notes,
https://web.stanford.edu/class/linguist1/Slides/ling1_compling.pdf.
▶ Traunmüller, Hartmut (1997).
Geschichte der Sprachsynthese, http://www.ling.su.se/staff/hartmut/kempln.htm.
▶ Turing, A. M. (1950).
Computing Machinery and Intelligence, Mind, 59(236):433–460,
http://cogprints.org/499/1/turing.html.

Einführung in die Computerlinguistik I Fall 2023 What is CL? 48 / 48