Leftovers I: The second half of the course

1
(No Transcript)
2
Leftovers I The second half of the course

• The structure and meaning of whole
  utterances/sentences
• Parsing
• Compositional semantics
• Speech processing, MT, ...

3
Leftovers II Morphological processes

• Inflection
• Minor changes to the meaning of a word tense,
  definiteness, number, gender, ...
• Normally no change of POS (Part-Of-Speech)
  category
• Derivation
• Results in a new word presume -gt presumably
  circle -gt circular
• Normally POS change verb -gt adverb, noun -gt
  adjective

4
Morphology (cont.)

• Agglutinative l.
• plastic pearls on a string
• Flectional l.
• (half) melted plastic pearls on a string
• Applies to both inflection and derivation

5
Communication systems
6
A simple system
7
A simple system
8
Vervet monkeys

• a communication system with number of warning
  calls
• Calls
• Look out! Eagle!!
• Look out! Snake!!
• Look out! Leopard!!
• ...

9
Vervet communication

• A small number of (innate) calls
• Holistic
• The different sounds have no structure
• Differences between the simple system and this?
• How many lights are needed?

10
Human languages

• How many lights are needed for Amharic?
• ?
• How many sounds can a speaker of Amharic produce?
• infinitely many, of course. But
• How many distinctive sounds have Amharic?
• 35 (28 7) How many lights?

11
A multi-level system

• How can a human language convey an infinite
  number of meanings with 35 different sounds?
• Structure
• Meaningless sounds are combined into meaningful
  units

12
The phoneme - a meaningless sound

• The minimal unit in language that can make a
  difference in meaning
• Minimal pairs
• bet - pet beat - bat car - bar
• Allophones
• Examples
• Speech processing

13
The morpheme - a meaningful unit

• The minimal unit in language that carries meaning
• Allomorphs
• Examples

14
The multi-level structure of language

• phoneme - morpheme - word - phrase -
  sentence/utterance - text/discourse - ...

15
Two fundamental traits of human languages
16
Ambiguity

• When a word or a string of words has more than
  one meaning

17
Lexical I made her duck

• her - possessive pronoun her - object pronoun
• duck - verb duck - noun
• make create make cook
• ...

18
Structural I saw a man in the park with a
telescope

• I saw a man in the park with a telescope
• I saw a man in the park with a telescope
• I saw a man in the park with a telescope
• ...

19
Some more examples

• Young girls and boys are easily frightened
• Could you open the door?

20
Ambiguity

• Whats the use?
• How can people understand each other?
• How does ambiguity affect our attempts to teach
  computers to understand language?

21
Redundancy

• The phenomenon by which the same information is
  expressed more than once

22
Redundancy

• Examples
• I discussed computers yesterday
• (pseudo-)Amharic Man-the he-died
• ...

23
Redundancy

• Why?
• Is redundancy a problem for NLP?

24
Some NLP applications

• What kind of knowledge of laguage is needed?
• Text-to-speech
• Speech Recognition
• OCR
• IR
• Information Extraction
• MT
• Translitteration
• ...