What is discourse? An Introduction M.Sc. Seminar: Discourse - - PowerPoint PPT Presentation

What is discourse? An Introduction

M.Sc. Seminar: Discourse Coherence Theories and Modeling Annemarie Friedrich & Alexis Palmer

Department of Computational Linguistics, Saarland University

22 April 2013

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 1 / 20

Outline

1

Discourse

2

Relevant linguistic concepts

3

Computational approaches

4

Coherence

Discourse

Senses of “discourse”

• Styles of communication
• Argument, rhetoric
• Ways of talking about things
• Multi-party communication
• Conversation
• Dialogue
• Everything beyond the level of the sentence
• Linguistic notion of discourse
• i.e. TEXT

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 3 / 20

Discourse

The importance of text

Example

Europe’s Herschel space telescope has imaged one of the most popular subjects in the sky - the Horsehead Nebula - and its environs. The distinctively shaped molecular gas cloud is sited some 1,300 light-years from Earth in the Constellation Orion. It is in a region of space undergoing active star formation - something Herschel has been most keen to study. The Hubble space observatory has also returned to the Horsehead scene, to celebrate 23 years in orbit. Together, these two great facilities give scientists a much broader insight into what is taking place in this familiar patch of the heavens.

Sentences rarely exist in isolation. Coherence holds text together.

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 4 / 20

Outline

1

Discourse

2

Relevant linguistic concepts

3

Computational approaches

4

Coherence

Relevant linguistic concepts

Important concepts from linguistics

• Genre
• Discourse modes
• Information structure

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 6 / 20

Relevant linguistic concepts

Genre

• Text-level analysis
• Genre influences various aspects of texts
• Structure
• Themes and topics (but != domain)
• Choice of vocabulary
• Linguistic register/style
• ....
• Many different classification schemes

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 7 / 20

Relevant linguistic concepts

Discourse modes

• Text-passage level analysis
• Following Smith 2004 Modes of Discourse:
• Narrative
• Description
• Report
• Information
• Argument
• Discourse modes “do coherence” in different

ways

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 8 / 20

Relevant linguistic concepts

Information structure (IS)

• Sentence/phrase-level analysis, but with

respect to previous discourse

• Entities and situations classified as

GIVEN, OLD, NEW

• IS interacts with phonology & syntax (at least)

and matters for coherence

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 9 / 20

Outline

1

Discourse

2

Relevant linguistic concepts

3

Computational approaches

4

Coherence

Computational approaches

Some active research areas in CL

• Coreference resolution
• Centering theory and lexical chains
• Topic modeling and document structure
• Discourse relations and discourse structure

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 11 / 20

Computational approaches

Coreference resolution

• Entity-level analysis
• Linking references to common entities
• Cues: anaphoric expressions
• pronouns
• demonstratives (e.g. this movie)
• alternate forms of reference (President Obama, Barack Obama,

Obama, President of the US)

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 12 / 20

Computational approaches

Centering theory

• Local analysis (words/phrases in pairs of

clauses/sentences)

• Relationships between entities in adjacent

utterances

• Coreference is an essential component

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 13 / 20

Computational approaches

Lexical chains

• Local analysis (words/phrases), but with aim

at global analysis

• Semantic relatedness between nouns, verbs,

adjectives

• Often modeled using WordNet

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 14 / 20

Computational approaches

Topics and structure

• Text/text-passage level analysis
• Concerned with aboutness
• Topics used to model structure
• Topic models ~ underlying topics defined in terms of

which words are used

• Topic transitions often co-occur with

document-internal boundaries

• Unsupervised models perform well

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 15 / 20

Computational approaches

Discourse relations and structure

• Clause/sentence/EDU-level analysis
• Relations between clauses: causality, temporal

structure, etc.

• Higher-level structure: discourse parse for entire

texts

• Resources: corpora
• Penn Discourse Treebank (PDTB)
• Rhetorical Structure Theory (RST) Bank
• DISCOR: texts labeled with SDRT structures

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 16 / 20

Computational approaches

Introductory readings on discourse

• Bonnie Webber and Aravind Joshi, Discourse Structure and

Computation: Past, Present, and Future. ACL 2010

(http://homepages.inf.ed.ac.uk/bonnie/final_discourse_50.pdf)

• Bonnie Webber, Marcus Egg, and Valia Kordoni, Discourse

structure and language technology, NLE vol. 18, no. 4, 2012: especially section 4 on potential applications

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 17 / 20

Outline

1

Discourse

2

Relevant linguistic concepts

3

Computational approaches

4

Coherence

Coherence

Coherence

• Our focus for the semester
• Theories of coherence
• Ways of modeling coherence
• Applications

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 19 / 20

Coherence

Text puzzle

• Put the text in order
• Note difficult decisions
• What aspects of the text influence your decisions?

Alexis Palmer (CoLi Saarland) What is discourse? 22 April 2013 20 / 20