Measuring Incongruity in Register Humour

Chris Thomas Venour, Christopher Stuart Mellish, Graeme D Ritchie

Research output: Contribution to conferencePaper

Abstract

Many theories of humour claim that incongruity is an essential ingredient of humour, yet most research fails to construct clear and formal definitions of this concept. We seek to address this problem by creating a computer program which can quantify and measure a certain type of humorous incongruity found in register-based humour. The type of humorous incongruity we are focusing on involves clashes of tone between words in a text and we have been experimenting in building a semantic space in which distances between words reflect differences in their style or tone. If words from a joke containing incongruities of tone were plotted into such a space, distances between words would be relatively large. Whereas the distances between words taken from ‘normal’ text, where differences of lexical tone are less extreme, would be relatively small. To build a semantic space, we first of all hand-selected corpora which we believe show varying degrees of lexical formality (literariness, archaism etc.). Our algorithm then takes a text, such as a one-liner, and computes the frequency of each word in that text in each of the corpora. These frequencies represent a word’s position in a multi-dimensional space and distances between words are computed. After experimenting with different kinds of spaces and distance metrics, an implementation of our algorithm was tested with the task of automatically distinguishing humorous texts from plain newspaper sentences, where it performed quite well. This work is currently in progress.
Original languageEnglish
Publication statusPublished - 2011
EventInternational Society for Humor Studies - Boston, United States
Duration: 5 Jul 20119 Jul 2011

Conference

ConferenceInternational Society for Humor Studies
CountryUnited States
CityBoston
Period5/07/119/07/11

Fingerprint

Incongruity
Semantic Space
Formality
Archaism
Jokes
Lexical Tone
Literariness

Keywords

  • artificial intelligence
  • computational humor
  • computers and humor
  • computers and humour
  • computational humour

Cite this

Venour, C. T., Mellish, C. S., & Ritchie, G. D. (2011). Measuring Incongruity in Register Humour. Paper presented at International Society for Humor Studies, Boston, United States.

Measuring Incongruity in Register Humour. / Venour, Chris Thomas; Mellish, Christopher Stuart; Ritchie, Graeme D.

2011. Paper presented at International Society for Humor Studies, Boston, United States.

Research output: Contribution to conferencePaper

Venour, CT, Mellish, CS & Ritchie, GD 2011, 'Measuring Incongruity in Register Humour', Paper presented at International Society for Humor Studies, Boston, United States, 5/07/11 - 9/07/11.
Venour CT, Mellish CS, Ritchie GD. Measuring Incongruity in Register Humour. 2011. Paper presented at International Society for Humor Studies, Boston, United States.
Venour, Chris Thomas ; Mellish, Christopher Stuart ; Ritchie, Graeme D. / Measuring Incongruity in Register Humour. Paper presented at International Society for Humor Studies, Boston, United States.
@conference{b648a043a1084fb5ab6b9b1b1fb62d05,
title = "Measuring Incongruity in Register Humour",
abstract = "Many theories of humour claim that incongruity is an essential ingredient of humour, yet most research fails to construct clear and formal definitions of this concept. We seek to address this problem by creating a computer program which can quantify and measure a certain type of humorous incongruity found in register-based humour. The type of humorous incongruity we are focusing on involves clashes of tone between words in a text and we have been experimenting in building a semantic space in which distances between words reflect differences in their style or tone. If words from a joke containing incongruities of tone were plotted into such a space, distances between words would be relatively large. Whereas the distances between words taken from ‘normal’ text, where differences of lexical tone are less extreme, would be relatively small. To build a semantic space, we first of all hand-selected corpora which we believe show varying degrees of lexical formality (literariness, archaism etc.). Our algorithm then takes a text, such as a one-liner, and computes the frequency of each word in that text in each of the corpora. These frequencies represent a word’s position in a multi-dimensional space and distances between words are computed. After experimenting with different kinds of spaces and distance metrics, an implementation of our algorithm was tested with the task of automatically distinguishing humorous texts from plain newspaper sentences, where it performed quite well. This work is currently in progress.",
keywords = "artificial intelligence , computational humor, computers and humor, computers and humour, computational humour",
author = "Venour, {Chris Thomas} and Mellish, {Christopher Stuart} and Ritchie, {Graeme D}",
year = "2011",
language = "English",
note = "International Society for Humor Studies ; Conference date: 05-07-2011 Through 09-07-2011",

}

TY - CONF

T1 - Measuring Incongruity in Register Humour

AU - Venour, Chris Thomas

AU - Mellish, Christopher Stuart

AU - Ritchie, Graeme D

PY - 2011

Y1 - 2011

N2 - Many theories of humour claim that incongruity is an essential ingredient of humour, yet most research fails to construct clear and formal definitions of this concept. We seek to address this problem by creating a computer program which can quantify and measure a certain type of humorous incongruity found in register-based humour. The type of humorous incongruity we are focusing on involves clashes of tone between words in a text and we have been experimenting in building a semantic space in which distances between words reflect differences in their style or tone. If words from a joke containing incongruities of tone were plotted into such a space, distances between words would be relatively large. Whereas the distances between words taken from ‘normal’ text, where differences of lexical tone are less extreme, would be relatively small. To build a semantic space, we first of all hand-selected corpora which we believe show varying degrees of lexical formality (literariness, archaism etc.). Our algorithm then takes a text, such as a one-liner, and computes the frequency of each word in that text in each of the corpora. These frequencies represent a word’s position in a multi-dimensional space and distances between words are computed. After experimenting with different kinds of spaces and distance metrics, an implementation of our algorithm was tested with the task of automatically distinguishing humorous texts from plain newspaper sentences, where it performed quite well. This work is currently in progress.

AB - Many theories of humour claim that incongruity is an essential ingredient of humour, yet most research fails to construct clear and formal definitions of this concept. We seek to address this problem by creating a computer program which can quantify and measure a certain type of humorous incongruity found in register-based humour. The type of humorous incongruity we are focusing on involves clashes of tone between words in a text and we have been experimenting in building a semantic space in which distances between words reflect differences in their style or tone. If words from a joke containing incongruities of tone were plotted into such a space, distances between words would be relatively large. Whereas the distances between words taken from ‘normal’ text, where differences of lexical tone are less extreme, would be relatively small. To build a semantic space, we first of all hand-selected corpora which we believe show varying degrees of lexical formality (literariness, archaism etc.). Our algorithm then takes a text, such as a one-liner, and computes the frequency of each word in that text in each of the corpora. These frequencies represent a word’s position in a multi-dimensional space and distances between words are computed. After experimenting with different kinds of spaces and distance metrics, an implementation of our algorithm was tested with the task of automatically distinguishing humorous texts from plain newspaper sentences, where it performed quite well. This work is currently in progress.

KW - artificial intelligence

KW - computational humor

KW - computers and humor

KW - computers and humour

KW - computational humour

M3 - Paper

ER -