T-61.6020 Computational Pragmatics
Timo Honkela, Aalto University School of Science
Spring 2012
Pragmatics is a subfield of linguistics which studies the ways in
which context contributes to meaning. It studies how the transmission
of meaning depends not only on the linguistic knowledge of the speaker
and listener, but also on the context of the utterance, knowledge
about the status of those involved, the intent of the speaker, etc.
Even though pragmatics is traditionally considered as an area of
linguistics, similar considerations related to meaning in context are
also relevant for information systems design and especially
interactive systems development. An interesting issue within computer
science is the interface between pragmatics and semantics. Ontologies
are used in semantic web to define prototypical meanings but in the
real-world contexts, pragmatics deals with the subjective and
contextual variation around prototypical meanings. In human-to-machine
communication, information systems may have practical uses in new
contexts beyond the ones defined originally by the designer of the
system. In machine-to-machine communication, formal semantics may fall
short in solving interoperability issues and thus issues related to
pragmatics need to be considered. In overall, the focus is in how
understanding takes place, not in how meanings are defined.
During the course, the participants are introduced with the main
linguistic theories related to pragmatics including but not limited to
the theories about the functions of languages, the speech act theory,
and the theory of conversational maxims. The participants will
familiarize themselves with computational models in the area of
pragmatics with specific focus on dynamic and adaptive systems and
statistical machine learning. They will also conduct a small empirical
study related to the subjectivity and contextuality of meaning using
the grounded intersubjective concept analysis (GICA). The collected
data will be analyzed using statistical methods.
Timo Honkela: Analysis of Qualitative Data using Machine Learning Methods
Timo Honkela: Introductory lecture of the seminar course on Computational Pragmatics
1. Introduction to Pragmatics
Timo Honkela
Aalto University School of Science
Department of Information and Computer Science
Computational Cognitive Systems research group
Computational Pragmatics, Spring 2012, 23rd of January, 2012
2. Pragmatics
● is a subfield of linguistics which studies the
ways in which context contributes to meaning
● studies how the transmission of meaning
depends not only on
● the linguistic knowledge of the speaker and listener,
but also on
● the context of the utterance,
● knowledge about the status of those involved,
● the intent of the speaker, etc.
3. Levels of written language
Structure Meaning
Syntax Pragmatics
“Structure of “Meaning in
sentences” context”
Morphology Semantics
“Structure of “Prototypical
words” meaning”
4. On Complexity
Structure Meaning
Ownership
“Mary's car”
Genitive Relationship
construction “Mary's husband”
“X's Y”
Property
“Mary's weight”
etc.
5. Theories in Pragmatics
● Jakobson: Functions of languages
● Austin: Speech act theory
● Grice: Conversational maxims
6. Roman Jakobson's
Functions of Language
TARGET
FUNCTION EXAMPLE
FACTOR
Context Referential “This is Jim”
Addresser Emotive “Yuck”
Addressee Conative “Go there!”
Contact Phatic “Hello”
Code Metalingual “What does 'phatic' mean?”
Message Poetic “Dibba-dabba-doo”
http://www.signosemio.com/jakobson/functions-of-language.asp
7. J. L. Austin: How to Do Things with Words /
The three Components of Speech Act Theory
● Locutionary act, “the act of 'saying'
something.”
● Illocutionary act, “the performance of an act in
saying something as opposed to the
performance of an act of saying something.”
● Perlocutionary act, for "saying something will
often, or even normally, produce certain
consequential effects upon the feelings,
thoughts, or actions of the audience, of the
speaker, or of other persons."
8. Austin's Speech Act Theory
● A locutionary act has meaning;
it produces an understandable utterance.
● An illocutionary act has force; it is informed with
a certain tone, attitude, feeling, motive, or
intention.
● A perlocutionary act has consequence;
it has an effect upon the addressee.
http://www.library.utoronto.ca/utel/glossary/Speech_act_theory.html
9. Paul Grice:
Conversational Maxims
● A speaker is assumed to make a contribution
that
● is adequately but not overly informative
(quantity maxim)
● the speaker does not believe to be false and for
which adequate evidence is had (quality maxim)
● is relevant (maxim of relation or relevance), and
● is clear, unambiguous, brief, and orderly
(maxim of manner).
http://www.sil.org/linguistics/GlossaryOfLinguisticTerms/WhatIsAConversationalMaxim.htm
11. Suggested themes
1. Models of context for
natural language processing
2. Meaning negotiation
3. Game theoretical approaches to meaning
4. Dialogue models
5. Modeling subjectivity in
understanding language
6. Miscommunication
12. Preliminary selected themes
1. Spoken dialogue systems
2. Modeling subjectivity
3. Meaning negotiation, (mis)communication,
dialogue models
4. Dialogue models for robotics
5. Human dialogue
6. Modeling subjectivity
7. Pragmatics and cognitive science