See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/2474262

Humour Research: State of the Art

Article · November 2002

Source: CiteSeer

CITATIONS READS

58 165

2 authors, including:

Anton Nijholt
University of Twente
702 PUBLICATIONS   8,569 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

International Conference On Digital Economy ICDEc (2nd edition), 4-6 May, Sidi Bou Said -Tunisia View project

Statistical dialogue management View project

All content following this page was uploaded by Anton Nijholt on 17 April 2013.

The user has requested enhancement of the downloaded file.

 Humour Research: State of the Art
M.P. Mulder and A. Nijholt
Center for Telematics and Information Technology
TKI-Parlevink Research Group
University of Twente,
PO Box 217, 7500 AE Enschede
The Netherlands

Abstract
Humour is a multi-disciplinary field of research. People have been
working on humour in many fields of research like psychology,
philosophy and linguistics, sociology and literature. Especially in
the context of computer science (or Artificial Intelligence) humour
research aims at modeling humour in a computationally tractable
way. Having computational models of humour allows interface
designers to have the computer generate and interpret humour when
interacting with users. There are many situations in human-human
interaction where humour plays an important role in keeping the
conversation going. Making use of the so-called CASA paradigm
(Computers Are Social Actors) we may expect that a similar role
can be played in human-computer interaction. In this report we
survey current humour research with the aim to identify useful
theories that can be applied in the human-computer interaction
context. We focus on the following subjects: humour theories,
related humour research, linguistic aspects of humour,
computational aspects of humour, applications and resources.
Table of Contents
1 Introduction 3
1.1 Conventional Humour Theories 3
1.2 Superiority Theory 3
1.3 Relief Theory 4
1.4 Incongruity Theory 4
1.5 Minsky's Theory on Humour 5
1.6 Violation Theory 5
1.7 Sociology Theories 6
2 Humour and Related Research 7
2.1 Introduction 7
2.2 Laughter Research 7
2.3 Contextualisation of Humour 8
2.4 Humour and Health 8
2.5 Sense of Humour 9
3 Humour Research in the Linguistic Field 10
3.1 Introduction 10
3.2 Humour and Syntax 10
3.3 Humour and Pragmatics 10
3.4 Semantic Script Theory of Humor 11
3.5 The Five -Level Model of Joke representation 11
3.6 The General Theory of Verbal Humor 12
4 Humour Research in the Computational Field 13
4.1 International Workshop on Computational Humor 13
4.2 Building on the Surprise Disambiguation and Two-Stage Model 13
4.3 A Formal Model for Punning Riddles 15
4.4 Using Humour in User Interfaces 15
5 Applications 17
5.1 Introduction 17
5.2 Joke Analysis and Production Engine 17
5.3 Elmo, the Natural Language Robot 18
5.4 The Light Bulb Joke Generator 18
5.5 The Mnemonic Sentence Generator 18
5.6 Acronym Applications 19
6 Resources 19
6.1 Introduction 19
6.2 Princeton WordNet 20
6.3 Global WordNet Association 20
6.4 EuroWordNet 20
6.5 MultiWordNet 20
6.6 WordNet++ 21
6.7 Lexical FreeNet 21
7 References 21

2
1 Introduction
Humour is a multi-disciplinary field of research. People have been working on humour in many
fields of research like psychology, philosophy and linguistics, sociology and literature. Especially
in the context of computer science (or Artificial Intelligence) humour research aims at modeling
humour in a computationally tractable way. Having computational mode ls of humour allows
interface designers to have the computer generate and interpret humour when interacting with
users. There are many situations in human-human interaction where humour plays an important
role in keeping the conversation going. Making use of the so-called CASA paradigm (Computers
Are Social Actors) we may expect that a similar role can be played in human-computer
interaction. In this report we survey current humour research with the aim to identify useful
theories that can be applied in the human-computer interaction context. We focus on the
following subjects: humour theories, related humour research, linguistic aspects of humour,
computational aspects of humour, applications and resources.

1.1 Conventional Humour Theories

In the conventional literature on theories on humour there is a division in three basic theories:
• Superiority Theory
• Relief Theory
• Incongruity Theory
This chapter will briefly discuss these three theories, together with additional theories from the
fields of psychology and soc iology. A theory proposed by Veatch attempts to explain humour
and laughter in a broader sense and will be discussed in section 1.6.

1.2 Superiority Theory

The assumption of the superiority theory is that we laugh about the misfortunes of others; it
reflects our own superiority. This theory can be found in the work of Plato, Aristotle, and Hobbes
[BAR92]. Plato suggests that humour is some kind of malice towards people that are being
considered relatively powerless. Hobbes further explains that humans are in a constant
competition with each other, looking for the shortcomings of other persons. He considers
laughter as an expression of a sudden realisation that we are better than others, an expression of
‘sudden glory’.
Although this theory seems old-fashioned in the 21st century, Charles Gruner reformulated this
theory as the Superiority Theory of Humour [GRU97]. His theory contains a thee-part thesis:
• Every humorous situation has a winner and a loser.
• Incongruity is always present in a humorous situation.
• Humour requires an element of surprise.
The first part of this thesis contains the idea of Superiority. The assumption that all humour has a
winner and a loser is based on human nature. Through history humans have used humour to
“compete” with other persons, making them the target of their humorous comment. The “winner”
is the one that successfully makes fun of the “loser”.

3
This theory can explain the fun of some humorous television programs, like sitcoms and talk
shows, and successful Hollywood movies like ‘There’s Something About Mary’. Nathaniel Grow
[GRO00] uses the Superiority Theory of Humour in his analysis of the success of MTV’s show
“The Tom Green Show”.

1.3 Relief Theory

The Relief Theory has a clear physiological or psycho-physiological nature [RUT97]. The theory
reached its zenith when Freud proposed his theory how laughter can release tension and “psychic
energy”. This energy continuously builds up within the human body, has no further use and
therefore has to be released. This release is spontaneous and expresses itself in laughter. This
theory is popular among those who believe that laughter is beneficial for one’s health.
Freud explains that this “psychic energy” in our body is built as an aid for suppressing feelings in
taboo areas, like sex or death. When this energy is released we experience laughter, not only
because of the release of this energy, but also because these taboo thoughts are being entertained.
A more conventional version of the Relief Theory is that we experience a pleasant sensation
when humour replaces negative feelings like pain or sadness.
The theory does not really give an explanation as to why we find humour funny and can in fact be
seen a theory of laughter.

1.4 Incongruity Theory

The incongruity theory is the most influential approach to the study of humour and laughter.
Kant, in the eighteenth century, is credited to have made the first full conceptualisation of
incongruity. A good description of the incongruity theory is found in the following words uttered
by Schopenhauer [SCH83]:
“The cause of laughter in every case is simply the sudden perception of the incongruity
between a concept and the real objects which have been thought through it in some relation,
and the laugh itself is just an expression of this incongruity.”
When jokes are examined in the light of the incongruity theory, two objects in the joke are
presented through a single concept, or ‘frame’. The concept becomes applied to both objects and
the objects become similar. As the joke progresses, it becomes apparent that this concept only
applies to one of the two objects and thus the difference between the objects or their concepts
becomes apparent. This is what is called incongruity [RUT97].
Many agree on the point that it is not the incongruity but the congruous resolution of the apparent
incongruity that makes a certain situation funny. That is why we speak of the incongruity-
resolution theory.
The incongruity-resolution theory is more or less a linguistic theory, because it explains how
jokes are structured and does not pay attention to the influence of the surrounding factors.
Moreover, it cannot explain why we can hear a joke more than one time and still find it funny and
why not all incongruities are funny.

4
1.5 Minsky’s Theory on Humour
Minsky bases his Theory on Humour on Freud’s notion of humour [MIN81]. Freud claims that
our brain creates so-called “censors” in order to create barriers that prevent us from thinking
“naughty” or “forbidden” thoughts. When we are able to elude these barriers by means of a joke,
we experience a sudden release of “psychic energy” and this energy is discharged in the form of
laughter. This theory could explain why we laugh about childish jokes with double meanings and
why we laugh about sexual and aggressive jokes in general.
However, this theory doesn’t explain why we experience other types of jokes as funny, for
example why we laugh about humorous nonsense. Freud wrote that this maybe had to do with
our perception of what humour exactly is. Minsky assumes, however, that the fun of humorous
nonsense can also be explained with the “censor” theory.
One of Freud’s examples of a nonsense joke where logic fails is the following:
A gentleman entered a pastry-cook's shop and ordered a cake; but he soon brought it back and
asked for a glass of liqueur instead. He drank it and began to leave without having paid. The
proprietor detained him. "You've not paid for the liqueur." "But I gave you the cake in
exchange for it." "You didn't pay for that either." "But I hadn't eaten it".
The result of such a joke is that we find ourselves puzzled, next troubled; after that we experience
laughter and find the logical absurdity humorous. Minsky suggests that our subconscious builds a
collection of “cognitive censors” in order to suppress this kind of faulty logic with which we can
do ourselves some kind of cognitive harm. When we are therefore able to elude these censors, or
where no specific censors exist, we experience laughter. These censors are able to “learn” and
“improve” and that explains why we experience a joke we already heard several times less funny.
Minsky further proposes that we are thinking in what he calls “frames”. A frame is a way to
define a certain stereotyped situation. We are able to make shifts between these frames, from the
very simple to very powerful once-in-a-lifetime insights we obtain by finding less obvious
connections between two situations. A lot of jokes, especially ‘puns’, make use of words that
have more than one meaning. In order to make the good assumption, we have to make a sense-
shift of the word, causing a frame -replacement in our way of reasoning (the less obvious frame
for the ‘default’ frame). We can experience this as humorous, although it can also be dangerous
for, for example, schizophrenic persons. One example is the schizophrenic patient who sees a
penny in the street, says "copper, that's a conductor," and then runs to a streetcar to speak to the
conductor.

1.6 Violation Theory

The violation theory, as proposed by Thomas Veatch [VEA98] can be seen as an attempt to
create one global theory of all different sorts of humour. It suggests comprising all existing
humour theories like the incongruity-resolution and superiority theory.
The theory describes three necessary and (jointly) sufficient conditions for humour perception.
When one of these conditions is lacking, there is no humour perception and if all three are
present, humour perception is also there. The three conditions are being described as:
• V - The violation of a certain commitment of the perceiver about how things ought to be.
• N - The perceiver has the dominating feeling that the certain situation is normal.

5
 • Simultaneity - These two understandings V and N have to be present in the mind of the
perceiver at the same time.
In other words, a certain situation is humorous when it seems that the situation is normal, but at
the same time it seems like something is wrong. To cite Veatch:
“Humor is (emotional) pain (V) that doesn't hurt (N).”
Veatch describes condition V as a “subjective moral violation'', the violation of that what the
perceiver cares about. It can be described best this way, because it depends on the perceiver what
it is he experiences as a violation. It depends on one’s commitment or attachment to a certain
situation.
Veatch goes on to explain that the grade of attachment of the perceiver is important as to if he
will experience a situation as humour, offensive or unremarkable. He describes this phenomenon
with a three-level scale of commitment and its consequences. The relation of commitment and
humour perception becomes clear in Table 1.
Perceiver
level logic commitment gets it is offended sees humour
level 1 not-V none no no no
level 2 V and N weak yes no yes
level 3 V and not-N strong yes yes no

Table 1 – The three -level scale

The normality factor (= the amount of ‘normal’ information in a joke) is, besides the level of
attachment, of importance and can have a strong and profound influence on how the perceiver
will experience the violation.
In his paper Veatch describes how this theory can put other theories in their proper place and that
they are more or less derivable from the violation theory. This theory can also explain why babies
experience ‘peekaboo’ as humorous and what makes a joke a ‘sick joke’.

1.7 Sociology Theories

The difference between humour research in the sociology field and other fields is that the
research of sociologists is not centred round the question why we laugh. Their research is
focusing on the social and cultural context of a joke or any other humorous event. They do not try
to understand why a joke causes laughter or why we find a certain situation humorous, they focus
on the way a joke is contextually interpreted. This means: to understand the real meaning of a
joke in its cultural and social context, the consequences of a joke and the influence of and in
interpersonal relationships.
The social environment of humour is vast and humour can be seen as a part of many different
social actions. According to sociologists, the social impact of joking and humour is very
important in our everyday life and there are many different theories on the social impact of
humour. Jason Rutter discerns in [RUT97] a division of sociologies of humour in three groups he
calls Maintenance Theories, Negotiation Theories and Frame Theories.

6
Maintenance Theories suggest that jokes maintain the established social roles and divisions
within a society. They can strengthen roles within the family, within a working environment and
everywhere there exists an in-group and out-group. When ethnical jokes are concerned, jokers
choose groups very similar to theirs as the target of the joke only to focus on the mutual
differences and in that way strengthen the established divisions between the two groups.
Negotiation Theories look at humour and laughter from a specifically sociological perspective.
They focus on the role of humour as a means of interaction, pastime and an event where more
than one person is involved. The hearer defines if a joke is funny or not, depending on the social
and cultural context of the joke and its environment. A study of humour and jokes can offer
insight in a culture and reveal aspects of this culture that would otherwise not be observed.
Frame Theories tend to see joking as a break from the everyday serious life. The joker makes a
shift from the serious frame to the humorous frame and is allowed to present criticism without
fear for retribution. The jokes are founded on the society and culture, but are standing outside the
normal discourse. From the humorous frame people can give comments and breach taboos
without causing offence. Of course, a joke is only accepted as humorous when all involved
parties agree that it is a joke. By switching to the humorous frame when introducing a potentially
volatile subject, one can always make the “only joking” excuse.

2 Humour and Related Research

2.1 Introduction
Because of the broad scope of humour research, it is impossible to highlight all current research.
Some of the research that is closely related to the theories of humour and of interest within the
scope of the project is highlighted in this chapter. The chapter will discuss: laughter research,
contextualisation of humour, humour and health and ‘sense of humour’.

2.2 Laughter Research

Many researchers connect laughter to humour and use these two terms in the same context.
However, humour and laughter are closely related but certainly not the same. Laughter can be the
result of humour, but not every humorous situation will invoke laughter, as well as humans laugh
without the perception of humour.
There is little research going on in the field of laughter. The curious 1962 outbreak of contagious
laughter in Tanganyika that lasted for six months and affected thousands of people led to research
about the nature of laughter. A point to start is the article written by Robert Provine that appeared
in the January/February 1996 issue of the American Scientific [PRO96]. He has done research
about the structure of laughter and about the chimpanzee laughter. More interesting in the context
of this deliverable is his research about the social and linguistic context of laughter. Laughter is
contagious and can be used to evoke laughter or a positive mood, as being done on television,
where laugh tracks are being used to accompany sitcoms to make the experience of a sitcom
more pleasant.

7
Elizabeth Muehlchen [MUE00] refers to [PRO96] and adds that laughter brings people together
as a group, it signals “bonding, affirmation, belonging, listening”, it can be disarming and by
means of laughter people can show more about themselves than that they are aware of.
In [DAV96] Jessica Davis digs deeper in the social, physiological and psychological effects of
laughter in specific.

2.3 Contextualisation of Humour

In [ZAJ91] Anat Zajdman discusses how a canned joke can be implemented into a discourse. He
presents a ‘four-stage model’, indicating a gradual increase of semantic relevance of the joke to
the discourse.
These four types of joke incorporation are represented in the four-stage model:
• Type A – “Supplier”
This speech-event consists of a question and an answer that contains the required
information, followed by a joke.
• Type B – “Sub-contractor”
This speech-event consists of a question and an answer that contains the required
information, followed by an announcement that a joke is to come, the joke and an
explanation of the relevance of the joke to the situation.
• Type C – “Joint venture”
This is an abbreviated form of Type B. The joke is introduced without fully telling it,
assuming that the text is known to the hearer and will be evoked by him through an
allusion.
• Type D – “Merger”
This type takes place when the required information is perfectly compounded with the
joke.
The model shows four types of joke incorporation with an increase in connections with the
context. The canned jokes in the “Supplier” incorporation have very loose connections with the
context and the jokes in the “Merger” incorporation are merged in the context, resulting is an
overlap between the joke scripts and the context. In this way information can be conveyed using
jokes.

2.4 Humour and Health

The Association for Applied and Therapeutic Humor that was founded in 1988 [AATH] was
created to be active in the field of education, research, application and clearinghouse on
information related to therapeutic humour. On their website information about the association can
be found, as well as articles and links to other related websites.
Steven Sultanoff, the former president of the AATH, has written some articles about therapeutic
humour [SUL95]. He points out that although humour is one of the most powerful methods to
help one to overcome periods of crisis, individuals immersed in the crisis often experience crisis
humour as hurtful. The factor that determines if a joke is appreciated or not is the butt of the joke.

8
Others generally appreciate self-deprecating humour or humour aimed at situations, but humour
is aimed at a group or at individuals used to degrade or insult is generally not appreciated. When
a person is involved in a crisis situation, it is likely that such a person will integrate the crisis into
their internal emotional being. These persons will experience humour aimed at the crisis situation
as aimed at them and therefore experience the humour as hurtful or insensitive. In order for crisis
humour to be beneficial an individual has to take distance from the crisis. This distance may be
proximal, emotional or temporal; when the distance is large enough, crisis humour may be
healthful.
Ellie Marek and Judith Tingley discuss in [MAR00] the appreciation of gender jokes by men and
women. Their conclusions were than men find this type of humour funny much more often than
women do and that women were more frequently offended and saw more hostility in this type of
jokes than men do.

2.5 Sense of Humour

A complete issue of the HUMOR Journals was dedicated to measurement approaches to the
sense of humour [RAS96]. Willibald Ruch states in the introduction that humour is an important
trait of our personality. The agenda of ‘sense of humour’ research is “to describe the entirety of
observable habitual differences in humor, investigate their interrelations in a systematic way, and
eventually define a smaller set of traits that account for the differences observed, that is, to make
up a person’s “sense of humor”.”
In order to create a theoretical model of ‘sense of humour’ three different aspects have to be
studied: the causes of individual differences in humour, the group differences (gender, racial,
etc.) and intraindividual differences. A potential description of a humorous personality consists
of a multi-dimensional structure of the all the different components of humour.
One important aspect is to determine how broad the scope of the expression humour is. Is
humour distinguishable from satire, irony, wit, etc. and do there exist besides sense of humour
also concepts like sense of satire and sense of wit? Or is humour an umbrella-term for our
humorous personality trait?
The HUMOUR journal shows that there are good scales to measure the different aspects of the
sense of humour. Howe ver, more theoretical and empirical work has to be done on the definition
or foundation of the concepts of sense of humour.
The application Jester 2.0 is an on-line joke recommending system using collaborative filtering in
order to recommend jokes adapted to the user’s ‘sense of humour’ [GOL00]. A user has to rate
jokes and according to these ratings he will become associated with a ‘nearest neighbour’,
another user with a similar ‘sense of humour’. New jokes will be recommended according to the
ratings this ‘nearest neighbour’ gave to other jokes.
Giselinde Kuipers [KUI01] is looking how a social background is of any influence on the ‘sense
of humour’ within the Dutch society. She is especially interested in the vast amount of ethnical
jokes that exist in The Netherlands.

9
3 Humour Research in the Linguistic Field
3.1 Introduction
Within the linguistic field humour can be treated from syntactic, pragmatic and semantic points
of view.
Attardo and Raskin have done a lot of work on semantic aspects of verbal humour. Their work,
based on the incongruity-resolution theory has resulted in the General Theory of Verbal Humour
and this theory will be discussed in section 3.6.

3.2 Humour and syntax

Oaks [OAK94] offers a catalogue of syntactic and lexical devices for the creation of ambiguity
within jokes. Robert Hetzron [HET91] gives an extended account on the structure of jokes and
punchlines. He assumes that a joke consists of ‘pulses’. These pulses are successive episodes in a
text or parts of an enumeration. These pulses can be humorous, amusing or informative; the main
condition is that the last pulse contains the punchline. Hetzron also subjects the punchlines to a
feature analysis and describes the logical devices that are used to make punchlines sound funny.
Neal Norrick looks at the intertextuality of jokes [NOR89]. He assumes that intertextual jokes are
based on an original text or joke that the perceiver has to know in order to appreciate the joke. In
another paper Norrick [NOR93] examines the presence and importance of repetition in canned
jokes and spontaneous conversational joking.

3.3 Humour and pragmatics

Rachel Giora [GIO01] states that pleasure and liking are induced by stimuli that involve both
innovativeness and familiarity. This has resulted in the optimal innovatio n hypothesis:
“If a stimulus is optimally innovative it would be rated as
more pleasurable than either a familiar stimulus or a purely
innovative stimulus.”
A stimulus would be optimally innovative if it involves a novel response to a familiar stimulus,
but “would also allow for the automatic recoverability of a salient response related to that
stimulus so that the similarity and difference between the novel and the salient would be
assessable.” [GIORA]
Marlene Dolitsky focuses on the aspects of the unsaid in humour [DOL92]. She assumes that the
place where the unsaid communication takes place is the point of the joke where its ‘funniness’
resides. The three aspects where the unsaid in humour must be studied are:
1. The speech act that is humour.
2. The participants in the humorous event.
A speech act consists of explicit and implicit information. The implicit information is
conveyed through the utterances and the speaking situation taken together as a whole.
This implicit information is based on the shared knowle dge and assumptions concerning
the world the speaker and listener live in. When telling a joke the speaker leads the
listener on to apply rules of pertinence knowing that they do not apply to the story being

10
 told. The humorous effect is held the listeners’ realisation that he or she has been led
down the garden path. Another way to use the unsaid in the speech act is to mix standard
scripts or ‘frames’ in a way that they result in a humorous situation.
3. The rule breaking character of humour.
Every society ha s its rules that are governing the behaviour of its members. These rules
are part of the common knowledge of the community and they regulate the things a
member can and cannot do. Humour can bee seen as a carnival of language. During
carnival all rules are abolished; likewise humour is based on the bending and breaking of
rules. These rules are always unsaid and are therefore a third aspect one has to study in
order to understand the funniness of the unsaid in humour.
Ephratt [EPH96] is relying on the notion of speech acts in his pragmatic account of humour.

3.4 Semantic Script Theory of Humor

The Semantic Script Theory of Humor (SSTH) is a theory developed by Victor Raskin and is
summarised in [ATT91] where different existing theories are being combined in the General
Theory of Verbal Humor (GTVH). The theory assumes that a joke is always related with two
different scripts that are opposed to each other in a special way. The theory explains that the text
of a joke is unambiguous up to the point of the punchline. The punchline triggers a switch from
one script to another and makes the hearer realise that more interpretations of the text are possible
from the beginning.
The theory postulates three levels of abstraction of script opposition with at the highest level of
abstraction the opposition between real and unreal. At a lower level of abstraction these
oppositions can take three forms, namely actual vs. non-actual, normal vs. abnormal, possible vs.
impossible. At the lowest level these oppositions can be manifested as oppositions like good vs.
bad, live vs. death , sex vs. non-sex, etc.
The theory ignores other parameters that have an influence in the funniness of a joke, assuming
that script opposition is the most important factor. Only two types of triggers, or logical
mechanisms that trigger the script-switch in the punchline are distinguished: ambiguous and
contradictory triggers. These logical mechanisms of script-switch are seen as an implementation
of the script opposition and are not given special attention.

3.5 The Five-Level Model of Joke representation

This model developed by Attardo in 1987 models five levels of joke representation. The different
levels are:
1. Surface
2. Language
3. Target + Situation
4. Template
5. Basic (script opposition and logical mechanism)
The ordering of the different levels is based on the linguistic meaning-to-sound theories of how
sentences are being formed. This model yields a number of theoretical problems that are being

11
discussed in [ATT91]. A revision of this model in combination with the SSTH led to the
formulation of the General Theory of Verbal Humor.

3.6 The General Theory of Verbal Humor

The derivation of the GTVH from the SSTH can be found in [ATT91]. In [RUC93] a description
of the GTVH can be found. Besides script opposition the scope of the GTVH consists of five
other parameters, called Knowlegde Resources (KR).
The six Knowledge Resources of this theory are:
• Language (LA)
The LA KR is the set linguistic components chosen to form the actual text of the joke.
• Narrative strategy (NS)
The joke has to be cast in some form of narrative organisation. This has been called the
NS KR.
• Target (TA)
The TA KR is the butt of a joke. Non-aggressive jokes do not have a target.
• Situation (SI)
The SI KR is the situation of the joke. One can think of the object, participants,
surroundings, activities, etc.
• Logical mechanism (LM)
The LM KR represents the mechanism used to bring two different scripts together in one
joke.
• Script opposition (SO)
The SO KR is the most abstract of all resources. It deals with the script
opposition/overlapping requirement as has been described by the SSTH.
An important aspect of the GTVH is the hierarchal organisation of the KRs. The basic principle
is that a certain KR probably will determine or will be determined by another KR. The
hierarchical organisation of KRs is: SO, (LM), SI, TA, LA, NS, LA, organised from less similar
and less determined to more similar and more determined.
In [ATT91] a study of the relation between differences in KRs between two jokes and their
similarity is outlined. The assumption is that that if the GTVH is correct, a linear increase of
similarity between pairs of jokes selected along the KR hierarchy will be perceived. This theory
proves true for all KRs except LM. The question therefore was raised if LM is a KR of not.
In [ATT97] Attardo compares the GTVH with the incongruity-resolution theories and is pointing
that LM is in fact the resolution of the incongruity, or script opposition (SO). This means that LM
can be seen as an optional KR, because so-called nonsense jokes, jokes without resolution can
also be experienced as funny. It is also possible that LM is only the resolution and no KR at all.
Graeme Ritchie states in [RIT01] that the GTVH is more developed than any other theory, but
from the computational point of view it is nothing more than a very early draft of a model.

12
4 Humour Research in the Computational Field
4.1 International Workshop on Computational Humor
The state of the art in computational humour is not very developed. The IWCH proceedings
[HUL96] give a good overview of the state of the art in 1996. Ritchie summarises these
proceedings in [RIT01], where he discusses the following research done in the field of
computational humour.
• The logical analysis of irony by Utsumi;
• The research on syntactic ambiguity in jokes by Ephratt;
• The pun detecting program for Japanese by Takizawa et al.;
• The JAPE riddle-generator created by Binsted and Ritchie;
• The neural account of what happens when a humorous stimulus is processed by a hearer
or reader proposed by Katz.

4.2 Building on the Surprise Disambiguation and Two-Stage Model

Ritchie has done some interesting research as well. In [RIT99] he takes the incongruity-resolution
(IR) theory as the basis in order to propose some processing steps that might be necessary in
order to interpret a joke. He analyses two variations on the IR model: the Surprise
Disambiguation (SD) model and the two-stage model of Suls [SUL72].
The SD model assumes that there are two different interpretations for the set-up of a joke, one
more obvious to the audience than the other. The audience will not become aware of the less
obvious interpretation until the punchline conflicts with the obvious interpretation and evokes the
other meaning. The two-stage model assumes that the punchline creates incongruity and that a
cognitive rule has to be found that enables the content of the punchline to follow naturally on the
contents of the set-up.
Building on the SD model he sets out that there are several relations or properties proposed as the
essential ingredients in incongruity-resolution humour, but that not all of these relations or
properties are necessary to create humour. These relations or properties are:
• Obviousness. The obvious interpretation of the set-up is more likely to be noticed than the
less obvious or ‘hidden’ interpretation.
• Conflict. The meaning of the punchline does not make sense with the more obvious
interpretation of the set-up.
• Compatibility. The meaning of the punchline does make sense with the ‘hidden’
interpretation of the set-up.
• Comparison. There is a contrasting relationship, or even a clash between the two
interpretations op the set-up.
• Inappropriateness. The ‘hidden’ interpretation of the set-up is inappropriate. It can lead to
on of the following effects: Absurdity (the flouting of everyday logic) or Taboo (the
flouting of socially acceptable matters).

13
Another factor that is discussed in the context of SD jokes is the violation of expectations.
Instead of a conflict between the interpretation of the obvious set-up and the punchline, a
punchline can be in conflict with some prediction. Therefore it may be necessary to consider two
possible subtypes of the ‘conflict’ relation: one that sets the punchline against the more obvious
interpretation of the set-up and one that sets the punchline against a predicted interpretation.
Ritchie gives a sketch of a processing model for the interpretation of jokes in the manner of the
SD model and draws the conclusion that the SD account relates to the delivery of humorous
content. It reduces the existing problems of the understanding of humour to a set of sub-
problems. The model does not explain what makes certain incongruity funny and other just a
misunderstanding. It offers a solution to how incongruity can be brought to the audience’s
attention.
Suls’ two-stage model could be summarised as follows [RIT99]:
• a text is read, make predictions
• while no conflict with predictions, keep going
• if input conflicts with predictions:
• if not ending – PUZZLEMENT
• if it is the ending, try to res olve:
• no rule found – PUZZLEMENT
• cognitive rule found – HUMOUR
The difference between this model and the SD model is that the two-stage model does not need
any ambiguity to be present in the set -up. The model does not say how surprising a part of a text
must be in order to count as a punchline. It is also not clear what a ‘cognitive rule’ is and why
there exist different ‘cognitive rules’.
The following differences between the two models are mentioned:
“
(a) The two models cover, or attempt to cover, different subclasses of joke. The SD model
requires an ambiguous setup, whereas the two-stage model makes no mention of
ambiguity.
(b) The SD model decomposes the humorous effect into slightly simpler concepts
(particularly COMPARISON, ABSURDITY, and TABOO), and so at least starts to
address the issue of “incongruity”.
(c) The two-stage model relies on some (undefined) form of “humorous logic”, and so leaves
the difficult problem of ‘incongruity’ relatively untouched.”
Building on these two models in [RIT00] Ritchie describes jokes using objects, properties and
relations in a sorted version of first order predicate logic, in this way describing descriptive and
narrative jokes. It is suggested that developing detailed descriptions of subclasses of jokes in this
way will give more insight in the concept of incongruity and can lead to a broader theory of
humour.

14
4.3 A Formal Model for Punning Riddles
Kim Binsted and Graeme Ritchie have devised a formal model of the semantic and syntactic
regularities underlying some of the simpler types of punning riddles [BIN97]. A punning riddle is
a question-answer riddle that uses phonological ambiguity. The three main strategies used to
create phonological ambiguity are syllable substitution, word substitution and metathesis.
Syllable subs titution is the strategy to confuse a syllable (or syllables) in a word with a similar- or
identical-sounding word. The following joke is an example of syllable substitution.
What do shortsighted ghosts wear? Spooktacles. [WEB78]
Word substitution is the strategy to confuse an entire word with another similar - or identical-
sounding word. An example of a joke with word substitution is:
How do you make gold soup? Put fourteen carrots in it. [WEB78]
Metathesis is a strategy very different to syllable or word substitution. It uses reversal of sounds
and words to suggest a similarity in meaning between two semantically distinct phrases.
What is the difference between a torn flag and a postage stamp? One’s a tattered banner
and the other’s a battered tanner. [BIN97]
Punning riddles based on these three strategies are all suitable for computer generation. Ritchie
and Binsted chose to focus on the word substitution based punning riddles, because lists of
homophones (= phonetically identical words) are already available.
The assumptions about the contents and the structure of the lexicon are as follows. The lexicon
consists of a finite set of lexemes and of lexical relations. A lexeme is an abstract entity
corresponding to the meaning of a word. If a word has two meanings, it has two corresponding
lexemes. Every lexeme has a set of properties about the representation and the type of word. A
lexical relation can be an explicit relation between two lexemes, like synonym or homophone, or
a general inter-lexeme relation, applicable to more than one pair of lexemes.
In order to describe a punning riddle, two sorts of symbolic description have to be used: schema
and template. A schema stipulates a set of relations witch must be hold between the lexemes used
to build a joke. A template indicates the information necessary to turn a schema and lexemes into
a piece of text. It contains fixed segments of text that are to be used and syntactic details of how
lexemes have to be expressed.
This model is implemented as a computer program named JAPE-1. This program will be further
discussed in chapter six.

4.4 Using Humour in User Interfaces

Reeves and Nass have shown in their studies [REE96] that humans respond in the same way to
computers as they do to persons with respect to psychosocial phenomena such as personality,
politeness, flattery, and in-group favouritism. Because humour is an essential part of human
communication, humour should be integrated into user interfaces if we want to make human-
computer interaction more natural and flexible.
Two conducted experiments that examined the effects of humour in task-oriented computer-
mediated communication (CMC) and human-computer interaction (HCI) show that humour can
have many positive effects [MOR98]. Participants who had received jokes dur ing the interaction

15
rated the system as more likable and competent, smiled and laughed more, responded in a more
sociable manner and reported greater cooperation. The study provides strong evidence that
humour should be incorporated in CMC and HCI systems.
In [BIN95] Kim Binsted discusses how humour can make user interfaces friendlier. Humans use
humour to ease communication problems. In a same way humour can be used to solve
communication problems that arise with human-computer interaction using Natural La nguage
(NL) interfaces.
Binsted explains that the kinds of humour to be used do not have to be very sophisticated.
Suitable humour that can be used is self-deprecating humour. In some cases deprecating the user
of a third party can be appropriate, but this type of humour is very risky.
Humour can make a computer more human when it fails and can ease the interaction.
Inappropriate humour, however, is irritating and humour should be tailored to the user. When a
certain user regularly works with a system, the system can adapt the types of jokes to the user’s
taste. The user should always have the option to turn the humorous input off.
Therefore she concludes that humour that is sparingly and carefully used can make NL interfaces
much friendlier.
Oliviero Stock claims in [STO96] that, as far as the natural language understanding area is
concerned, we have to address two main objectives in order to process humour: being capable of
producing efficiently all different interpretations of linguistic expressions, and being capable of
choosing efficiently the appropriate one in the given context. There do exist systems for yielding
all possible interpretations of a joke at various levels of analysis (morphological, lexical,
syntactic, semantic, pragmatic, discourse, intentional).
Stock claims that techniques as speech recognition can open the way to greater possibilities for
the interface, it can for example analyse an expression at a phonetic level. At the semantic and
pragmatic level a system should have some knowledge about world, stereotypes and user.
Maes [MAE95] claims that the interface should be separated from the underlying system in order
that the user will not perceive the interface as responsible for the functioning of the whole
system. Stock believes that the future prospect is that a user will interact with more agents, each
with its social role. The user can decide whether to pay attention to what each one of them has to
say and can enjoy the interaction.
Recently, embodied conversational agents (ECA’s) have become a well-established research
area. Embodied agents are agents that are visible in the interface as animated cartoon characters
or resembling human beings. Sometimes they just consist of an animated talking face, displaying
facial expressions and, when using speech synthesis, having lip synchronization. These agents are
used to inform and explain or even to demonstrate products or sequences of activities in
educational, e-commerce or recreational settings. Experiments have shown that ECA’s can
increase the motivation of a student or a user interacting with the system. Current research deals
with improving intelligent behaviour of these ECA’s, but also with adding emotional behaviour
and personality in order to make them more believable [Nij01] to the user and to induce trust. It
seems to be quite natural to make a step from event appraisal theories for emotion to appraisal
theories for humourous events, in order to try to realize that ECA’s smile or even laugh at the
right moment, making them even more believable.

16
5 Applications
5.1 Introduction
This chapter is not meant to give a survey of possible applications of humour research. Rather we
illustrate some of the results of humour research with some implemented software systems or
algorithms. Nevertheless, we should mention here that humans use humour in speech, texts and
dialogue. This means that any algorithm, tool or system that is meant to manipulate speech,
language, dialogue or text in such a way that knowledge of contents plays somekind of role can
profit from achievements of humour research. Again, we confine ourselves to showing some
existing systems.

5.2 Joke Analysis and Production Engine

The JAPE-1 program is based on the model of generating punning riddles as described in section
4.3. The architecture of JAPE-1 is described in [BIN97]. The riddles that JAPE-1 can generate
are a subgroup of all punning riddles. These riddles all:
• Use word substitution as main mechanism;
• Substitute phonologically identical words;
• Substitute into a noun phrase;
• Use the constructed phrase in the punchline.
The knowledge bases used are a set of schemata (see section 4.3), a set of templates, a lexicon
containing humour-independent semantic and syntactic information, a homophone base,
containing pairs of phonologically identical lexemes and a post-production checker, to sift out
more of the more obvious non-jokes.
The lexicon was especially constructed for JAPE-1, but the information it contains is general and
neutral. The power to generate jokes lies not in the lexicon. The basic algorithm for automatic
joke generation used by JAPE-1 is the following:
“
1. Choose a common noun phrase from the lexicon.
2. Choose an appropriate schema, into which the phrase can fit.
3. Fit the phrase’s constituent lexemes into the schema.
4. Instantiate the key lexemes into the schema.
5. Choose an appropriate template, thus instantiating the characteristic links.
6. Generate the near-surface form of the riddle, instantiating characteristic lexemes in the
process.
7. Check that the generated joke is neither repetitive, nor a sensible question-answer pair.”
Tests performed with human judges pointed out that JAPE-1 does not generate very good jokes.
There is room for improvement, adjusting the lexicon, the templates and the schemas and adding
some filtering heuristics. JAPE-2 is an improved version of JAPE-1 that has an interface with
WordNet. Using WordNet as a lexical source produces too many entries of which many are too

17
‘obscure’ to make meaningful puns. Therefore additional filtering has to be done and all this
makes JAPE-2 much slower than JAPE-1.

5.3 Elmo, the Natural Language Robot

Dan Loehr has integrated the humour system JAPE and his natural language robot Elmo
[LOE96]. JAPE and Elmo are integrated at four different levels:
1. Elmo can issue a query for a pun from JAPE when he receives a request like: “tell me a
riddle”.
2. An attempt to make Elmo produce humour relevant to the user input.
3. Make Elmo produce humour relevant to the user input, using a pre-chosen pun.
4. Make Elmo return a carefully scripted reply, to achieve a ‘smoother’ response.
The conclusion was that there are many difficulties producing relevant, fluid humour on arbitrary
input. Improving relevance by casting a broader net around the words of the discourse and the
words available to the generator failed. The net proved to be too large and too slow and even
produced less relevant responses than before.

5.4 The Light Bulb Joke Generator

Attardo and Raskin [ATT94] have created a simple joke -generating program called LIBJOG
(Light Bulb Joke Generator). The program shows how simple methods can be used to create
jokes. The first version uses an entry for a commonly stereotyped group and combines this entry
with a template for a light bulb joke. For example:
entry: (Poles ((activity1 hold the light bulb)
(numberX 1)
(acrivity2 turn the table he is standing on)
(numberY 4)))
template: How many (group name) does it take to screw in a light bulb?
(NumberX). One to (activity1) and (numberY) to (activity2).
[Condition: X = Y + 1.]
The joke-generating mechanism is very limited and only consists of substituting the stereotyped
entity in the template. The LIBJOG does not make use of a humour-independent lexicon. The
only ‘lexicon’ it uses is the set of stereotyped entries.

5.5 The Mnemonic Sentence Generator

Craigh McDonough [DON01] has built a software application called Mnemonic Sentence
Generator that converts any alphanumeric password into a humorous sentence. The techniques
used from the field of computational humour are based on the incongruity theory. One of the
requirements of the system is that it generates a sentence that is easier to remember than the
password itself. If the sentence is funny, it will be even less difficult to remember a sentence.

18
The sentence template consists of two clauses of each four words. The te mplate, taken from
[DON01] is as follows:
Sentence template
(W1 = Person Name) + (W2 = Positive-Verb) + (W3 = Person Name + “’s”) + (W4 =
Common Noun) + “, while” + (W5 = Person Name) + (W6 = Negative-Verb) + (W7 = Person
Name + “’s”) + (W8 = Common Noun)
The goal is to combine two opposite scripts in one sentence, using a ‘positive’ verb in the first
clause and a ‘negative’ verb in the second clause. Furthermore they choose from their lexicon
personal names from one particular topic domain (in their first implementation politics), because
this will make the sentence more cohesive, more meaningful and memorable.
One example of an MSG generated sentence is:
“Arafat joined Quayle's Ant, while TARAR Jeopardized thurmond's vase.” [DON01]
(password: AjQA3Jtv)
In his conclusion McDonough suggests as an improvement to create new sentence templates, to
make a wider use of natural language processing techniques in order to choose more suitable
words in a sentence and to set the mnemonic sentence to music, like a nursery rhyme.

5.6 Acronym Applications

There is a World Wide Web searchable database containing more than 200,000 abbreviations and
acronyms. The address of site is http://www.acronymfinder.com/.
AnvilLogic has deve loped a commercial version an electronic acronym dictionary and
information about this application can be found at the website of AnvilLogic
(http://www.anvillogic.com/products/).
In 1963 there appeared in Newsweek an article written by Philip Broughton [BRO68]. He
developed a method to create phrases that appear intelligent or important. He called it the
Systematic Buzz Phrase Projector that is based on the technique of combining three words from a
lexic on containing thirty words. Although the generated phrases do not make sense, the audience
is likely not to admit that they do not understand the generated phrase (or acronym). Although the
method is not really scientific, the idea of creating acronyms by using humorous non-sense is
implemented in this model.

6 Resources
6.1 Introduction
Algorithms and tools to analyse or generate verbal humour as part of a text or of a dialogue
require the results of humour research to be built in. It means that traditional part-of-speech
taggers, parsers, annotation tools, knowledge representation formalisms, languages and systems
are among the resources for humour research. Corpora, as being made available by, for example,
the Linguistic Data Consortium (http://www.ldc.upenn.edu/) are of course important as well. In
such collections we can find speech and text databases and lexicons that can be used for research

19
and development purposes. In the context of such initiatives we also find the development of
tools and standards that can be used by verbal humour research. Well known is also Douglas B.
Lenat’s CYC system and in particular its open source version (OpenCyc), a large general
knowledge database and associated reasoning engines.

6.2 Princeton WordNet

WordNet is an online lexical reference system. Its design is based on psycholinguistic theories of
how the human lexical memory works [MIL93]. The words are organised into synonym sets, or
synsets, representing an underlying lexical concept. Different relations intertwine the synsets.
WordNet has proved to be the inspiration for many different wordnets in many different
languages, all based on the same structure.

6.3 Global WordNet Association

The Global WordNet Association (GWA) is established with the goal to maintain, standardise
and interlink wordnets for all languages in the world. They build on the success of Princeton
WordNet and EuroWordNet. On their website one can find more specific information about who
they are and what their goals are (http://www.hum.uva.nl/~ewn/gwa.htm).

6.4 EuroWordNet
EuroWordNet is a multilingual database with wordnets for several European languages. The
wordnets have the same structure with synsets and basic semantic relations between them as the
Princeton WordNet. The wordnet are linked to an Inter -Lingual-Index and via this index it is
possible to go from a word in one language to similar words in other languages. The index also
gives access to a shared top-ontology that provides a common semantic framework for all
languages. If a wordnet is compatible it can be added to the EuroWordNet database and be
connected to any other wordnet.
The EuroWordNet website gives more information about the background and research
(http://www.hum.uva.nl/~ewn/). In [FAR98] there is a description of the methodology for
converting the Princeton WordNet to a Spanish WordNet.

6.5 MultiWordNet
“MultiWordNet is a project currently under development at ITC/IRST aiming at the realisation of
a large scale generic lexicon for Italian based on the English version of WORDNET. The project
implements a multilingual lexical hierarchy, which is able to deal with a number of lexical
idiosyncrasies, including lexical gaps between the two languages. The acquisition methodology
includes two steps. Using a bilingual on line dictionary, first the system automatically builds a set
of candidate mappings from an Italian lemma to English synsets. In the second step a
lexicographer either confirms or discards the proposed synsets. A graphical interface has been
implemented which combines the access to the bilingual dictionary with the English WordNet
and allows the user to insert language specific synsets.”
– taken from the MultiWordNet website (http://ecate.itc.it:1024/projects/wordnet/).

20
6.6 WordNet++
Steuten et al. [STE00] have developed WordNet++ at the University of Amsterdam. WordNet++
is an extension to WordNet that makes the lexical database suitable for the support of the Color-
X (Conceptual Linguistically based Object-oriented Representation Language for Information
and Communication Systems) method. This is a method based on object-oriented and linguistic
modelling concepts. The wordnet uses a static object model and an event model. WordNet++
contains special relationships of domain-specific knowledge that do not exist in WordNet.

6.7 Lexical FreeNet

Doug Beeferman describes in [BEE98] the design and implementation of Lexical FreeNet, a
semantic network that combines WordNet-derived semantic relations with data-derived and
phonetically derived relations. It is based on FreeNet, a system for describing and exploring finite
binary relations between tokens or strings of characters.
Lexical FreeNet contains in total almost 200.000 tokens, including phrases. These words are
taken from the CMU Pronouncing Dictionary, WordNet and broadcast news transcripts. There
are seven semantic relations, two phonetic relations and one orthographic relation. A description
of these relations can be found in [BEE98].
There exists a World Wide Web interface (http://lefn.com) that gives human users the possibility
to explore the lexical database. Beeferman is speculating that Lexical FreeNet can be used in
automatic text processing, like text segmentation and text summarization, but no applications
have been developed so far.

7 References
[AATH] http://aath.org
[ATT91] Attardo, Salvatore and Raskin, Victor. Script theory revis(it)ed: joke similarity and
joke representation model. In: HUMOR, the International Journal of Humor
Research, Mouton de Gruyter, volume 4-3/4 pages 293-348, 1991.
[ATT94] Attardo, Savatore and Raskin, Victor. Non-literalness and non-bona-fide in
language: An approach to formal and computational treatments of humor. In:
Pragmatics and Cognition, 2, 31-69, 1994.
[ATT97] Attardo, Salvatore. The semantic foundations of cognitive theories of humor. In:
HUMOR, the International Journal of Humor Research, Mouton de Gruyter,
volume 10-4 pages 395-420, 1997.
[BAR92] Barnes, Clive. Comedy in Dance. In: Sorrell, Walter (ed.) The Dance Has Many
Faces. 3rd ed. Pennington: a cappella books, pp87-95, 1992.
[BEE98] Beeferman, Doug. Lexical discovery with an enriched semantic network . In:
Proceedings of the Workshop on Applications of WordNet in Natural Language
Processing Systems , ACL/COLING, 1998.

21
[BIN95] Binsted, Kim. Using humour to make natural language interfaces more friendly.
In: Proceedings of the AI, ALife and Entertainment Workshop, International Joint
Conference on Artificial Intelligence, 1995.
[BIN97] Binsted, Kim and Ritchie, Graeme. Computational rules for generating punning
riddles. In: HUMOR, the International Journal of Humor Research, Mouton de
Gruyter, volume 10-1 pages 25-76, 1997.
[BRO68] Broughton, Philip. About: The Systematic Buzz Phrase Projector. In: Newsweek,
May 6, 1968, pp.104, 104B, 104C, Newsweek Inc., 1968.
[DAV96] Davis, Jessica Milner. Taking Humour and Laughter Seriously. In: Australian
Journal of Comedy, Volume 2, Number 1, 1996.
[DOL92] Dolitsky, Marlene. Aspects of the unsaid in humor. In: HUMOR, the International
Journal of Humor Research, Mouton de Gruyter, volume 5-1/2 pages 33-44, 1992.
[DON01] McDonough, Craig. Using natural language processing to aid memory of random
passwordsi. Unpublished Technical Report, CERIAS, Purdue University, 2001
[EPH96] Ephratt, M. More on humor act: What sort of speech act is the joke? In:
Proceedings of the International Workshop on Computational Humor, Enschede,
Netherlands. University of Twente, 1996.
[ELL99] C. Elliott. Why boys like motorcycles: using emotion theory to find structure in
humorous stories. Unpublished paper, School of Computer Science, DePaul
University, Chicago, 1999.
[FAR98] Farreres X., Rigau G. and Rodríguez H. Using WordNet for Building WordNets.
In: Proceedings of COLING-ACL Workshop "Usage of WordNet in Natural
Language Processing Systems". Montreal, Canada. 1998.
[GIORA] Giora, Rachel. On our mind: Salience, context and figurative language. New
York: Oxford University Press, (in press).
[GIO01] Giora, Rachel. Optimal innovation and affect. Unpublished paper written for the
HAHAcronym project. Tel Aviv University, 2001.
[GOL00] Goldberg, Ken; Roeder, Theresa; Gupta, Dhruv and Perkins, Chris. Eigentaste: A
Constant Time Collaborative Filtering Algorithm, UCB Electronics Research
Laboratory Technical Report M00/41, University of California, Berkeley, 2000.
[GRU97] Gruner, Charles. The Game of Humor: A Comprehensive Theory of Why We
Laugh. New Brunswick, NJ: Transaction Publishers, 1997.
[GRO00] Grow, Nathaniel. The Tom Green Show.
http://forensics.truman.edu/CoachingPages/CritTomGreen'sShow.html
[HET91] Hetzron, Robert. On the structure of punchlines. In: HUMOR, the International
Journal of Humor Research, Mouton de Gruyter, volume 4-1 pages 61-108, 1991.
[HUL96] Hulstijn, J, and Nijholt, A. (eds.). Proceedings of the International Workshop on
Computational Humor. Number 12 in Twente Workshops on Language
Technology, Enschede, Netherlands. University of Twente, 1996.

22
[KUI01] Kuipers, Giselinde. Goede humor, slechte smaak: Nederlanders over moppen.
Boom, Amsterdam, 2001.
[LOE96] Loehr, Dan. An Integration of a Pun Generator with a Natural Language Robot,
In: Proceedings of the International Workshop on Computational Humor,
Enschede, Netherlands. University of Twente, 1996.
[MAE95] Maes, Pattie; Blumberg, Bruce; Darrell, Trevor; Pentland, Alex and Wexelblat,
Alan. Modeling Interactive Agents in ALIVE. In: Proceedings of IJCAI-95,
Montreal, 1995.
[MAR00] Marek, Ellie and Tingley, Judith. Men and Woman Respond to Humour. In: The
Power of Indirect Influence, October 2000.
[MIL93] Miller, Beckwith, Fellbaum, Gross and Miller. Introduction to WordNet: An On-
line Lexical Database. Published on World Wide Web:
ftp://clarity.princeton.edu/pub/wordnet/5papers.ps , 1993
[MIN81] Minsky, Marvin. Jokes and their Relation to the Cognitive Unconscious. In:
Cognitive Constraints on Communication, Vaina and Hintikka (eds.) Reidel, 1981
[MOR98] Morkes, John; Kernal, Hadyn K. and Nass, Clifford. Humor in Task -Oriented
Computer-Mediated Communication and Human-Computer Interaction. In:
Proceedings of CHI 98, 1998.
[MUE00] Muehlchen, Elizabeth. Vitality, Wilson Banwell EFAP Quarterly. Vol. 4, Issue 1,
2000, Wilson Banwell, 2000.
[NIJ01] Nijholt, Anton. Agents, Believability and Embodiment in Advanced Learning
Environments. In: Proc. IEEE International Conf. on Advanced Learning
Technologies: Issues, Achievements, and Challenges (ICALT 2001) , T. Okamto,
R. Hartley, Kinshuk & J.P. Klus (eds.), Madison, Wisconsin, IEEE Computer
Society, Los Alamitos, CA., 457-459.
[NOR89] Norrick, Neal R. Intertextuality in humor In: HUMOR, the International Journal of
Humor Research, Mouton de Gruyter, volume 2-2 pages 117-140, 1989.
[NOR93] Norrick, Neal R. Repetition in canned jokes and spontaneous conversational
joking. In: HUMOR, the International Journal of Humor Research, Mouton de
Gruyter, volume 6-4 pages 385-402, 1993.
[OAK94] Oaks, D. D. Creating structural ambiguities in humor: getting Enlish grammar to
cooperate. In: HUMOR, the International Journal of Humor Research, Mouton de
Gruyter, volume 7-4 pages 377-401, 1994.
[PRO96] Provine, Robert R. Laughter. In: American Scientist, January-February 1996, New
York. online version:
http://www.sigmaxi.org/amsci/Articles/96articles/Provine-R.html
[RAS96] Raskin, Victor (editor in chief) HUMOR, the International Journal of Humor
Research , Mouton de Gruyter, volume 9-3/4, 1996.

23
 [REE96] Reeves, B. and Nass, C. The Media Equation: how people treat computers,
televisions and new media like real people and places. Cameridge, Cambridge
University Press, 1996.
[RIT99] Ritchie, Graeme. Developing the Incongruity-Resolution Theory. In: Proceedings
of the AISB’99 Symposium on Creative Language: Humour & Stories, Edinburgh,
1999
[RIT00] Ritchie, Graeme. Describing Verbally Expressed Humour. In: Proceedings of the
AISB’00 Symposium on Time for AI and Society, Edinburgh, 2000
[RIT01] Ritchie, Graeme. Current Directions in Computational Humour. In: Artificial
Intelligence Review , 16(2), pp.119-135, 2001.
[RUC93] Ruch, Willibald; Attardo, Salvatore and Raskin, Victor. Toward an empirical
verification of the General Theory of Verbal Humor. In: HUMOR, the
International Journal of Humor Research, Mouton de Gruyter, volume 6-2 pages
123-136, 1993.
[RUT97] Rutter, Jason. Stand-Up as Interaction: Performance and Audience in Comedy
Venues. University of Salford, 1997, chapter 2.
[SCH83] Schopenhauer, Arthur. The World as Will and Idea. Vol. 1, London, Routledge,
1883.
[STE00] Steuten, A.A.G.; Dehne. F. and Riet, van der, R.P. WordNet++: A Lexicon
Supporting the Color-X Method. In: Proceedings Applications of Natural
Language to Data Bases , Paris, 2000.
[STO96] Stock, Oliviero. Password Swordfish: Verbal Humour in the Interface In:
Proceedings of the International Workshop on Computational Humor, Enschede,
Netherlands. University of Twente, 1996.
[SUL] http://humormatters.com
[SUL72] Suls, Jerry M. A two-stage model for the appreciation of jokes and cartoons: and
information-processing analysis. In: Goldstein and McGhee, chapter 5, p. 81-100,
1972
[SUL95] Sultanoff, Steven. Levity Defies Gravity; Using Humor in Crisis Situations. In:
Therapeutic Humor, Publication of the American Association for Therapeutic
Humor, Vol. IX, 3, p. 1-2, Summer 1995.
[VEA98] Veatch, Thomas C. A theory of humor. In: HUMOR, the International Journal of
Humor Research, Mouton de Gruyter, volume 11-2 pages 161-215, 1998.
[WEB78] Webb, Kaye (ed.). The Crack-a-Joke Book. London: Puffin, 1978.
[ZAJ91] A. Zajdman, Contextualization of canned jokes in discourse. In: HUMOR, the
International Journal of Humor Research, Mouton de Gruyter, volume 4-1 pages
23-40, 1991.

24

View publication stats