Preface

For a long time, the working title of this book was Lessons on the
Language of Humor. The original project was to collect a number of
lessons on the linguistics of humor. I meant and mean the term “lesson”
quite literally: the main goal of this book is pedagogical. The project
slowly evolved into an introduction to the study of the linguistics of
humor for the novice to humor research.1 The reason for doing so is
that many humor scholars and newcomers to linguistics are interested
in the substantial contribution that linguistics has brought to humor
research, but are kept at bay by the somewhat unfriendly nature of the
discipline (as several students and colleagues have admitted to me in
conversation).
I have taught occasionally workshops, classes, and courses on some
aspects of the linguistics of humor using some version of parts of these
chapters for the past 20 years or so. The book in its present form was
started in the summer of 2000 and slowly developed into extensive sets
of notes on various topics in the linguistics of humor. The primary
motivation for drafting the first version of parts of this textbook as
lectures was the fact that a colleague needed materials to teach a last-
minute assignment class on the linguistics of humor, at Georgia State
University, in 2006. I wrote a few lectures that I would send to her a few
days before the class. Long story short, we married in 2012. So, in a very
real sense, this book has been quite successful already, for me at least. If
it also helps some young scholar to learn about the linguistics of humor,
it’s gravy!
This book cannot hope to also be an introduction to linguistics; there
are excellent ones on the market.2 In what follows, I will assume that the
reader either has had a chance to familiarize him/herself with one such
introduction or will do so as they encounter various basic concepts in
linguistics. Occasionally, I will provide some background information
when I cannot assume it is common knowledge. For example, while
every self-respecting introduction to linguistics will define the concept
1Gelotology (Apte, 1988) never really caught on.
2In fact, I coauthored one such book myself: Brown and Attardo (2000), although
modesty prevents me from vouching for its excellence.

xiii
 PREFACE

of morpheme, I cannot assume that knowledge for isotopy or even

script. The glossary defines all technical terms and should be helpful
for newcomers to the field, but again it does not define terms such as
phoneme or morpheme.
This book aims at being a simple, straightforward description of
the results achieved by linguistics in humor research, as well as the
methodologies used to achieve such results. The goal is that by mapping
the cutting edge of the field and engaging with it, the reader will be in a
position to contribute to the linguistics of humor, both theoretical and
applied.
This is the first book on the linguistics of humor that attempts an
integrated approach encompassing all the fields of linguistics, both
theoretical and applied, on an equal footing. I am aware of the originality
of such a stance, so much so that I had to develop an approach to do so. In
particular, Chapters 10–12 develop an ethnomethodological perspective
that encompasses discourse analysis and variationist sociolinguistics
in the first-ever cohesive discussion of the performance of humor (as
opposed to the competence of humor). Obviously I did not develop
the whole approach out of thin air. In fact, it was the realization that
the ethnomethodological approach developed by Gumperz and Hymes
(and Goffman and Bateson, and many others) had so much to say about
humor in its original formulation, that led me to the decision (foolhardy,
I am sure) to organize the third part of the book around that theoretical
backbone and integrating the research on conversation, discourse, and
sociolinguistics within it.
This book is meant for an advanced undergraduate course or a grad-
uate seminar on humor or for scholars from other fields, with limited
familiarity with linguistics, who are interested in the contribution of
linguistics to humor studies. As such it is a systematic but not exhaustive
presentation of the field. It is organized in 16 chapters, so as to be
useable within a standard academic semester in American universities,
but of course there is no reason one should not proceed at a slower or
faster pace. I do however recommend reading the chapters in the order
they are presented, at least until Chapter 13, after which the linearity
of the presentation is less relevant. The other exception to the rec-
ommendation of following the linearity of the argument is Chapter 1.
Some readers may find some of the theoretical and methodological
hairsplitting less than captivating. If they are willing to take my word
for the usefulness of the conventions I adopt in the book, then they can

xiv
 PREFACE

safely move on to the rest of the chapters. If they are not, as they are
fully entitled to be, then they should read on and see why I made those
decisions.
In the age of tweets and hypertext, the presence of an argument that
takes ten or more chapters to elaborate may surprise or even put off
readers, but à la guerre comme à la guerre, as they say in France, or to
quote one of my favorite philosophers, it is what it is.
The presentation of the material is cyclical: a central topic such as
the incongruity–resolution model is first approached in the general
context of the three major theories of humor, then taken up again in
a separate chapter, and then taken up again in specific linguistic terms
in the chapters on the Semantic Script Theory of Humor (incongruity)
and the General Theory of Verbal Humor (resolution, a.k.a. the Logical
Mechanism). Likewise the significance of the competence/performance
distinction is first discussed in Chapter 2 and then resurfaces repeatedly,
for example in Chapter 6 on the Semantic Script Theory of Humor, in
Chapter 15 on translation, etc. This is done to facilitate the pedagogical
goal of the book.
Phonetic transcriptions are in (simplified) IPA. I occasionally will
use symbols common in American linguistics, such as [y] to indicate
a glide (IPA [j]) or [D] to indicate a flap (IPA [ɾ]). As usual, phonemes
and morphemes are represented between slashes and allophones and
allomorphs between square brackets. Syntactic trees use an “Aspects”
notation (Chomsky, 1965), for readability. Scripts are indicated in small
caps. I have tried to respect the different annotation styles of conversa-
tion and discourse analysts, but I occasionally had to simplify or modify
the transcription for practical reasons. Those are indicated in the text.
I have tried deliberately to keep logical symbols and formalisms out of
the text.
The fact that I need a disclaimer about the non-exhaustive nature of
the references is a very good indicator of the progress the field of the
linguistics of humor has made in the past quarter century. In 1994, when
I published my first book on the linguistics of humor, on which I had
worked for ten years, the attempt to be all-inclusive was foolhardy but
not obviously absurd. Today, most of the topics covered in the chap-
ters to follow have bibliographies that include several monographs. To
attempt a broad comprehensive coverage would be virtually impossible,
as new works appear constantly, in a variety of journals. The coverage of
the bibliography is thus representative and selective.

xv
 PREFACE

With the advent of the internet, it is now possible to locate and often
retrieve with minimal effort hundred of “scholarly” articles on any topic.
A researcher’s work is thus both easier (just a mere 20 years ago it would
have required years of expensive research to gather such a list of sources)
and more complicated: many of these sources are substandard and do
not meet basic criteria for scholarship, being riddled with fundamental
errors, misinterpretations, unverified second-hand citations, etc. So, to
a certain extent, the work of today’s scholar consists in some part in
discarding sources unworthy of other scholars’ attention. This creates
a problem of interpretation: prior to the age of the internet research,
if a serious scholar did not quote scholar X, there were two possible
interpretations: 1) an omission in error (i.e., the scholar was not aware of
the existence of the source but had he/she been aware of it he/she would
have quoted/discussed it), or 2) a deliberate omission either a) due to the
low quality of the work, b) to the need to be selective, c) the impossibility
of reading every source. Today, 1) is unlikely, precisely because with the
internet we can find almost anything, but 2) remains as relevant as it ever
was, if not more. I cannot guarantee that all omissions are deliberate
and due to the need to be selective, whereby only the most significant
or representative works are discussed, or are a gesture of mercy toward
subpar scholarship, which should be allowed to fade off into oblivion,
but I have done my best in that sense.
I would like to thank the following friends and colleagues for their
help in the writing of this book: Delia Chiaro, Greg Dean, Margherita
Dore, Elisa Gironzetti, Christian “Kiki” Hempelmann, Alan Partington,
Béatrice Priego-Valverde, Victor Raskin, Willibald Ruch, Willy Tsakona,
Manuela Wagner, and Patrick Zabalbeascoa. I also would like to thank
all the scholars who sent me copies of their articles, too many to name
them all, but this show of collegiality is always heartwarming. I also
would like to thank the librarians and the interlibrary loan staff at Gee
library, at Texas A&M University-Commerce. Truly, without their help
this book would not have been possible. I would also like to thank,
without any hint of irony, the reviewers and editorial team: thank to their
help the book is truly a better, if unfortunately less funny, work.
I feel like I should apologize to those who may have bought the book
thinking it will be funny. For them, I have an anecdote: more than 20
years ago, at the ISHS humor conference masterfully organized by Amy
Carrell at U. of Central Oklahoma, I found myself in a car with Victor
Raskin, Elliott Oring, and Christie Davies. We had gone off to find a

xvi
 PREFACE

sushi restaurant (I know, don’t ask) and were rejoining the rest of the
conference at a local comedy club, for an evening of standup comedy.
We arrived at the place, which was pretty big, and asked at the bar where
the Humor Conference people were seated. The bartender told us not to
worry, “You’ll find them in no time. They are the ones not laughing.”3
Finally, I would like to thank Tempest and Blue, our two rescue mutts,
Xena, our rescue cat, Dani, the non-rescue horse, Arnie, the armadillo
that lives in the bushes in our front yard, the families of swallows literally
living under our roof, and the various turtles, snakes, lizards, frogs,
herons, cardinals, hummingbirds, and assorted other fauna that share
our property, and my wife Lucy Pickering, who contributed a picture
of a paratone, saved countless frogs, and made me fall in love with her
again every day.

The Middle of Nowhere Ranch, near Lone Oak, Texas, 2020

3 Script opposition: laughter/no laughter; Logical mechanism: reversal (humor scholars

have no sense of humor); Target: humor scholars; Situation: comedy club; Narrative Strategy:
conversational one liner; Language: irrelevant.

xvii
 List of figures

1.1 The irony–sarcasm continuum 10

1.2 The relationship between humor, mirth, and laughter 17
1.3 Semiotic humor hierarchy 26
2.1 Venn diagram of the intersection between humor and
laughter 43
5.1 Denotative (bottom left) and connotative
(top row) semiotics 97
6.1 A very simplified semantic graph for cat; the links are
labeled after the semantic connection 119
6.2 The lexical script for doctor in Raskin, 1985 120
6.3 Simple, macro-, and complex scripts 121
7.1 Hierarchical organization of the knowledge resources 138
7.2 Orientations and targets of humor as a function of
aggression 146
9.1 The first interpretation of the sentence “I shot an elephant
in my pajamas” 183
9.2 The second interpretation of the sentence “I shot an
elephant in my pajamas” 183
9.3 The two interpretations of the sentence “Squad helps dog
bite victim” 183
10.1 A paratone, at the end of the performance of one of
the jokes. “That’s cool” is the punch line. Note the
declination, indicated by the continuous line. The dotted
lines indicate the approximate resets of each tone unit. 230
11.1 Hay’s presuppositional scale of hearer’s reactions to humor 250

xix
 List of tables

2.1 Relationships between humor and laughter 43

2.2 The interplay of humorous keying and text 52
3.1 Attardo’s three families of theories chart 60
6.1 Combinatorics for two scripts 127
10.1 The abstract–concrete continuum exemplified by the
article “the” 212
11.1 Notation for transcription, based on Jefferson’s 238
11.2 Sequential organization of joke telling 247

xxi
 1
Humor studies: a few definitions

1.1 Terminology 4
1.2 A brief overview of humor studies 18
1.3 A few basic distinctions 25
1.4 Further readings 29

This chapter introduces some necessary terminology and presents the

field of humor studies, primarily from a historical perspective, to help
the reader situate where the linguistics of humor fits in within the
broader humor studies field. Thus, much of it is preliminary in nature
and probably quite dry. One may in fact wonder whether it is really
necessary to have all this set of terms and definitions. In a sense, if
the reader is willing to take me on faith and accept that we will use
1) speaker, hearer, and text as umbrella terms for the prototypical roles
of the humorous exchange, and 2) the term humor as an umbrella term
for the stimulus (without any further attempts at differentiation between
sub terms, such as comic and ridicule or irony and sarcasm), 3) the term
mirth for the emotional response, and 4) the terms laughter and smiling
for the overt, physical responses typically associated with humor, then
the reader may just skip the next section and start reading about the
history of the field and why I think having one is important.
However, I believe it is important to clarify what it is we are talking
about when we say “humor” or “laughter.” There has been a lot of
confusion in the field from using terminology inconsistently, especially
in the early days of humor studies. That the field has moved away from
terminological free-for-all is a testament to the work of many humor
scholars who have carefully defined their terms.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 HUMOR STUDIES: A FEW DEFINITIONS

1.1 Terminology

1.1.1 The characters

Let us start from a non-controversial humor situation: someone says
something and someone else reacts laughing. We would describe this
situation as someone said something funny, humorous, amusing, droll,
or the like, and that made the other person laugh. This is the prototypical
case of humor situation. Let us call the first person the speaker, the
second person the hearer and whatever was said the text. For a large
number of situations, this will be enough.
Consider the following example:
(1) In a reply to a post by my daughter on Facebook, in which my
grandson tells a little story in which one character is a nose, I
jokingly criticized my daughter for having him read Gogol at age
four and a half. The post reads “I don’t want to challenge your edu-
cational practices, but having the child read Gogol’s masterpiece
The Nose at 4 seems premature.”
In this situation, I am the speaker and my daughter is the hearer, even
though of course the exchange took place in writing, and the text is my
post above. So far, so good. However, since this was posted on Facebook,
all of our friends and family members can see the exchange and one of
them in fact responded to it. Was the unintended response also from
a hearer? Dos it matter that I meant the joke only as a response to
my daughter’s post? Does it matter that several people, including my
daughter, “liked” the post by using little icons of a thumbs up or a little
red heart? Probably yes.
So we see that things can be much more complex than the little
situation we started from: there can be multiple hearers, there can be
different kinds of hearers and of course there can be many different
kinds of texts. Just like in a non-humorous situation we can distinguish
between hearers and over-hearers: the latter are not directly addressed
by the speaker—although they may be the real intended recipients of
what the speaker says. In example (1), does the family member who was
not an intended participant qualify as an over-hearer? Probably yes.
Also the situation is complicated in further ways: for example both
speakers conveys the text in writing, and speaker and hearer are not
in the same place at the same time. This is significant, because being

4
 1.1 TERMINOLOGY

face-to-face is one of the defining features of speech. Things can get even
more complex. However, I think that the prototypical situation above,
in its simplicity, is a good starting point, if for nothing else, at least to
establish some terminology.
We will then speak generally of “speaker,” “hearer,” and “text,” regard-
less of the fact that the speaker may not speak, the text may be spoken,
or written, or conveyed in any modality1 (visual, e.g., cartoons, auditory,
e.g., music, kinesic, e.g., mime, etc.), and the hearer may be in fact
a television audience, or read the text (and thus not actually “hear”
anything at all), etc.
Needless to say, when necessary we may differentiate: so if we were
investigating humor in the deaf/hearing community (see for example
Sutton-Spence & Napoli, 2012) we would refer to signed jokes, or
if investigating the differences between written and spoken forms of
jokes and say, internet memes, we would refer to oral, written, and
multimodal texts.
When we move away from speech, we need to consider that other
media (writing, printed press, television, film, social media, etc.)
may involve exceedingly complex (and usually hidden) chains of “co-
authors”: a TV show may put a certain joke in the mouth of a character,
but besides the person who wrote the script, the “author” of the text, the
director of the show, the producers, the TV network executives, the cable
channel that broadcast/rebroadcast it, etc. may be held accountable or
become relevant is some cases: for example, suppose that the executives
of a TV network were to censor parts of the jokes or less dramatically,
would need to apologize for them or distance themselves from them.
In those situations, the “author” (speaker in our terminology) ceases to
be a unified figure and dissolves in a chain of co-authors, with different
agendas, ideologies, and constituencies they need to keep happy.
Furthermore we need to consider that speaker, hearer, and text, may
diverge significantly from the prototypical terms in social/interactional
aspects. For example, the distinctions between ratified (official, openly
acknowledged) and unratified participants and addressed (i.e., the utter-
ance is directed to them) and unaddressed listeners (Goffman, 1981,
pp. 131–133) may be particularly relevant. It has been claimed, e.g.,
Bubel (2008) that in a TV show, for example, the audience consists

1 Hence the term “multimodal” to indicate texts such as films which incorporate both
visual and auditory signs.

5
 HUMOR STUDIES: A FEW DEFINITIONS

of overhearers. One then would need to distinguish the hearers in the

show and the overhearers in the audience. This is the case with so-
called dramatic irony. Dramatic irony is generally defined as a situation
in which the audience, of a play, for example, knows something that
the character(s) in the play do not, which causes them to interpret the
actions and words of the characters differently. Thus for example, in
Oedipus Rex, the audience is aware of whom Oedipus killed long before
he does. Likewise, Iago’s deception is known to the audience before it
is to Othello. Clearly the status of the audience is different than that
of the characters on stage. The audience is probably best classified as
unaddressed listeners, for the perspective of the characters (who of
course are not supposed to be aware of their presence at all). So in the
case of dramatic irony or of similar situations, the different status of
various hearers is relevant and hence should be addressed. Consider
the following joke: Freud (1905, pp. 183–184) which is analyzed in
Chapter 8 as example (81) from a slightly different standpoint.
(2) Two children are performing in front of their family a play they
themselves wrote. The plot revolves around a fisherman who leaves
his wife to go in search of fortune. In the second act, set several
years later, the fisherman returns with a huge fortune. His wife
greets him by saying: “I too have not been idle.” “And thereupon
she opens the door of the[ir] hut and reveals to his eyes twelve
large dolls lying asleep on the floor . . . At this point in the drama the
actors were interrupted by a storm of laughter from the audience,
which they were unable to understand.” (Freud, 1905, p.184)
In this example the difference between what the children know and what
the adults know is crucial to understand why the adults find the situation
funny and the children do not. We might describe it as a discrepancy
between the addressed hearer imagined by the children and the actual
hearers, who in the children’s eyes would be overhearers. However, and
this is my point, in most other cases, there is no need to distinguish
between hearers and overhearers, addressed and unaddressed partici-
pants. For example, in examples (7) and (10) below, there are just two
speakers and thus no need for further theoretical apparatus.
All this may seem complex and confusing. My argument however is
that, unless necessary, we can simplify the matter greatly and speak of a
speaker, telling a text to a hearer. If the situation requires it, we can bring
in a more sophisticated understanding of the situation, but unless we

6
 1.1 TERMINOLOGY

need it, we will shelve the unnecessarily complex apparatus. The same
argument holds for my use the term humor as generic designation for
the subject of inquiry of humor studies.

1.1.2 Humor as umbrella term

As is well known, there are many terms that have been used to designate
what happens in the situation above: humor, comic, funny, laugh-
able, droll, wit, amusing, etc. Scholars have tried, vainly, to distinguish
between, say, humor and ridicule, or between irony and sarcasm, and
the likes. The situation is even worse when we try to translate these
categories from one language into another. This is probably due to the
fact that all of these terms are folk-terminologies, that work well in some
prototypical cases, but fail lamentably when we try to apply them to
difficult, borderline instances. So, clearly humor is less aggressive than
ridicule, but where would one draw the line? The same can be said,
after accounting for any differences, for irony and sarcasm, farce and
comedy, etc.
Ultimately, the fact that twenty-five hundred years of discussion have
failed to yield clear-cut internal categories is a witness to the fact that
it seems to be impossible to do so. Recently, within the field of humor
research, a consensus seems to have emerged, whereby humor is being
used as the umbrella term to indicate the complex of humorous phe-
nomena, with further differentiation being left for future research or just
abandoned.2
This is not to say that we cannot differentiate between a joke and a
funny play, or between comedy and irony. In oder to understand the
issues at stake, we need to consider terminology in more detail. The
point of having terminology is to facilitate the scientific process. An
effective terminology has, among others, the following features; it is clear
and precise: there is no doubt as to whether any given entity is or is
not described by a given term;3 it is unique, i.e., each concept has one
term describing it (to put it differently, it is not ambiguous); it is stable,

2 See for example the titles of Raskin (1985), Morreall (1987), Attardo (1994), Ruch
(1998), and the name of the oldest journal of the field HUMOR: International Journal of
Humor Research.
3 Ideally, it is formal, i.e. there is an algorithm that can make the determination without
human input.

7
 HUMOR STUDIES: A FEW DEFINITIONS

i.e., it does not change;4 it should cover all the phenomena one needs to
cover, i.e., it should not leave aspects of the phenomenon undescribed;
relatedly, it should group all the phenomena that are related into one
category (or related ones); and, last but not least, it should not lead to
confusion by using terminology already used in a different sense, in the
same or related fields. Little harm is done by using the term isotopy
which is a technical term in chemistry (isotope) but if we wanted to
redefine “allophone” we’d be asking for trouble. Finally, it should be kept
to a minimum: there is no reason to coin a new term if there is a perfectly
good one already.
In my experience, within humor studies, the most frequent problems
one encounters with definitions are: 1) they are insufficiently precise
to be useful, and 2) they are not mutually exclusive. Let us consider
both problems in that order. If I define a joke as a short narrative text
with a punch line at the end, I am essentially listing the features of the
joke: 1) a linguistic text; 2) narrative (this excludes dramatic/dialogic
texts); 3) short; and 4) having a punch line located at the end. There
is nothing wrong with this definition, but it will struggle operationally
when confronted with Renaissance facezie, which can be quite long.
Clearly, we can arbitrarily define a threshold, of say, 50 words and
stipulate that anything longer than 50 words is not a joke, but this is
theoretically useless. If you have a 51 words text, are you willing to say
that that is no longer a joke but it is now a short story, or novella, or
whatever other terminology you have for a longish text with a punch
line at the end?5 Consider that a synonym of “narrative” is “story” and
a short story is not very different from a short narrative text. Moreover,
what about jokes in non-linguistic form, for example, cartoons with a
caption or with text in a balloon? What about jokes that are made only
of dialogue, such as question and answer jokes, or riddles? What about
very short dramatic plays not meant to be performed, such as Achille
Campanile’s Tragedie in due battute [Tragedies in two lines]? I translate6
one of my favorites in what follows:

4 For an example of the problems inherent in a changing definition of a concept, see the
discussion of the term “isotopy” in Attardo (1994).
5 These texts do exist: Katherine Mansfield’s short story Feuille d’album ends on a punch
line. See Attardo (2001, p. 93) for a discussion.
6 Literally and without any attempt at recreating the artistry of the text, evident in the
choice of the names.

8
 1.1 TERMINOLOGY

(3) INTRODUCTION.
Characters: MR. PERICLE FISCHIETTI
THE OTHER GENTLEMAN
At the opening of the curtain, MR. PERICLE FISCHIETTI
approaches THE OTHER GENTLEMAN
MR. PERICLE FISCHIETTI (Introducing himself to the other
gentleman) May I? I am Mr. Pericle Fischietti. And how about
yourself?
THE OTHER GENTLEMAN I am not. (curtain)
So the only value of this sort of definition is to give us a list of features
that (some of) the texts we are focusing on share. By the time we’ve
broadened it enough not to produce any false negatives (i.e., failing to
include texts that are jokes) it becomes virtually useless: A humorous
text with a punch line at the end. Note how this will not differentiate
between jokes, short stories, or even novels.
The mutual exclusiveness issue is even worse: consider satire, irony,
and puns. Any given text could be all three at the same time, regardless
of what your definition of any of the three terms are. All you have to do
is find an instance of a pun that creates an ironical opposition and make
sure that that opposition is critical of someone/something. If someone
were trying to code a text for these features they would end up with a
hopelessly tangled mess.
Now, let’s go back to the mini-definition of joke. We can make this
definition more useful if we compare it to the definition of the anecdote,
as a short humorous text, without a punch line at the end. Now the
distinguishing feature is whether there is punch line at the end of the
text. Of course, it will still be complex to define operationally “end”
(The last word? Last morpheme? Last phrase? Last sentence?) and you
will also need a definition of punch line. However, this definition is
now operationally sound: you can make predictions (any text without
a punch line at the end will be an anecdote), you can analyze texts,
test your analyses against others’ and so get an inter-rater reliability
score, etc.
Anyone can define things in such a way that they will convince
themselves that their definitions are useful and mutually exclusive. For
example, Aubouin (1948), who is an otherwise much overlooked genius
in the field, insists on a difference between humor (le comique) and
ridicule, the latter being aggressive. Predictably, it is not a convincing

9
 HUMOR STUDIES: A FEW DEFINITIONS

argument. Consider the Larack Tobama meme, widely circulated on the

internet in 2016.
(4) An image of President Barack Obama, with an obviously fake
mustache. The caption reads: “Independent party announces new
candidate coming late into the election, Larack Tobama.” https://
me.me/i/independent-party-announces-new-candidate-coming-
late-into-the-election-2852592
Is this comic or ridicule? Some elements of distortion (the mustache, the
changed name) are present, but they do not seem to be openly aggressive
or mocking. However, is there really no aggression or tendentiousness?
Isn’t there a criticism, implicit to be sure, of the other candidates? Would
a Republican or a Democrat categorize the joke the same way? How
about a white supremacist?
A similar problem with definition is exemplified with the distinction
between irony and sarcasm. There is a general consensus that irony
is less aggressive than sarcasm, as can be evinced from the etymology
of the word “sarcasm” which comes from Greek “sarkazein,” i.e., to rip
pieces of flesh. Thus, presented with the following two examples:
(5) Nice weather! (It is raining.)
and
(6) Great idea to invest all of our lifelong savings in Enron stock!
most people would probably agree that the former is ironical and the
latter sarcastic, given the knowledge that the Enron bankruptcy in 2001
caused its stock to fall from a valuation of $90 to essentially zero. Many
of the employees who had invested their retirement in Enron stock lost
their savings.
The problem is the in-between cases. We can conceptualize the irony–
sarcasm divide as a continuum, with irony at one end and sarcasm at the
other end; see Figure 1.1. The clear-cut cases, such as those above, will
be handled without problems, whereas those cases that fall somewhere
in between will be harder to classify.

Irony Sarcasm

Figure 1.1 The irony–sarcasm continuum.

10
 1.1 TERMINOLOGY

Thus for example, imagine a different context than the standard

default one (something along the lines of the two speakers are indoor
watching the rain) for the first example, namely that “Nice weather!” is
being uttered by the owner of a company who scheduled, on the advice
of the interlocutor who assured him/her that the weather would be great,
an outdoor picnic for the entire company, at great expense, and whose
costly suit has been ruined by the storm. Where would the utterance of
“Nice weather!” fall in the continuum? The term of art here is “a grey
area” which makes the point nicely: it is difficult to differentiate in the
grey area. In fact, Figure 1.1 considerably simplifies the real-life problem,
insofar as it is easy with a ruler to measure where the middle of the
gradient falls. In real life, of course, we have no idea where the ends of
the continuum lie and therefore we cannot estimate where the half-point
lies. Summing up, binary distinctions on a continuum are possible, but
are only reliable for the prototypical, extreme cases at the opposite ends
of the continuum. The in-between cases, which are the most interesting,
where a theory through its terminology should help distinguish the data,
are precisely those in which the distinction will fail. This severely limits
the usefulness of such distinctions.
Another problem is unnecessary redefinitions. Consider the example
of irony. Since ancient times there have been definitions of irony as,
simplifying drastically, saying one thing and meaning another. Further-
more, irony was divided between negative irony (i.e., critical) and pos-
itive (i.e., praising) The latter is known as “asteism.” An example would
be introducing a famous speaker at a conference, say Noam Chomsky, as
“our speaker has published a few books that have had some small impact
on the field of linguistics and politics,” meaning to imply ironically that
one means the opposite, and namely that he has published many very
influential books. Enter Garmendia (2010, p. 402), who claims that
all irony is negative (“critical”). When confronted with positive (non-
critical) irony, she denies that it is irony and claims it is something else
(a joke, Garmendia, 2010, p. 410, or “banter” (Leech, 1983, p. 144). It
is clear that such re-definitions do nothing to advance our knowledge
of the field. We had a distinction: positive vs. negative irony and we
replace it with another distinction: irony vs. joking or banter. All we
have here, theoretically speaking, is rearranging the socks in the drawer,
with the added inconvenience that now we have phenomena that are
identical in terms of semantic-pragmatic mechanisms, by Garmendia’s
own admission (p. 410), but have completely different and unrelated

11
 HUMOR STUDIES: A FEW DEFINITIONS

terms to describe them (even worse, the term “joke” has a completely
different much more salient meaning and “banter” is in its general sense,
something entirely different, i.e., teasing).
Finally, one more distinction, which has lately cropped up increas-
ingly frequently: within politeness theory, for example, researchers
distinguish between first- and second-order constructs: first-order
theories are folk-theories, whereas second-order theories are “scientific”
(i.e., they are proposed by scholars).7 Here we need to be careful:
notoriously, the speakers’ metalinguistic judgements are unreliable.
So, asking speakers: “Is this ironical?” or worse, “is what you just said
ironical?” will lead to all sorts of distortions (just imagine asking the
speakers if what they just did is aggressive or not; obvious social barriers
exist that will prevent the speakers from answering reliably). Likewise,
asking speakers if something is humorous or not, if it is ironical or
sarcastic, or parodic, etc. requires all sorts of experimental techniques
to avoid biasing the results, including the problem of knowing what
the speakers think such terms as “ironical,” “sarcastic,” etc. mean. The
terms “irony” and “sarcasm ” provide again a striking example. In British
English and in American English, “irony” used to mean, as we pointed
out above, roughly “saying something and meaning something else,
generally critical.” However, in the 1990s, at least in some populations
in the US, a semantic shift occurred, exemplified by the song Ironic,
by Alanis Morrisette, in the album Jagged Little Pill (1995). Nunberg
(2001, pp. 91–93) sums the situation up as “irony” is now used to
indicate something bad or upsetting. Concurrently, the semantic space
of “sarcastic” broadened to include what would have previously been
described as “ironical.” Morrisette was widely ridiculed for not knowing
what the word ironical meant, whereas it is clear that Morrisette’s (and
her age cohort) first-order definition of “irony” and “sarcasm” is different
from that of older generations (and other languages; to the best of my
knowledge this semantic shift did not take place in other languages or in
British English). Nelms (2001) for example purports to describe the use
of sarcasm in higher education classroom discourse, but her examples
include irony as well. The responsible scholar needs then to be alert to

7 A term often used in this context is the etic vs. emic distinction, used in anthropology,
for example, where an emic description is from the point of view of the subjects and an etic
description is from the point of view of the researchers. I prefer to avoid this terminology
due to the possible confusion with the suffixes used to indicate the phonetic and phonemic
levels in linguistic theory, for example, where they have a different meaning.

12
 1.1 TERMINOLOGY

the fact that his/her subjects may very well have a different first-order
definition of irony and sarcasm than his/her (hopefully) second-
order definition. Thus he/she needs to make sure that this difference
does not interfere with his/her analysis. It is impossible to be specific on
how exactly to do this, since it depends on what the research questions
are; however, keeping a strong distinction between first- and second-
order concepts is the necessary presupposition to avoid embarrassing
pitfalls. Likewise, if gathering a corpus of utterances labeled as “ironical”
by speakers (say, by using emoticons), it is crucial to have access to
demographic information relative to their age and language variety (in
this case American vs. British English) in order to discriminate the
data appropriately.8 One more consideration: I am using these cases
merely for the purpose of exemplifying the issues. I do not claim that
these are full discussions of the concepts and that the definitions I am
discussing here are either what I believe to be correct and/or the best we
can do.
Clearly, as we said, there are some well-established categories, such
as irony, farce, the burlesque, etc., but in a sense, humor research
investigates what all these categories have in common, first, and only
after it has established what the general category of humor consists of, it
attempts further classification. Concretely, these distinctions represent
more the historical remnants of discipline-specific taxonomies than
seriously considered scientific classifications: the concept of irony, for
example, originally comes from rhetorics. It is ultimately unclear what
its relationship to humor is.9 Many of the literary terms (comedy, farce,
burlesque, carnival, etc.) are the result of Renaissance reinterpretation
of classical models. To be honest, most of the terminology in humor
studies comes from the various disciplines. Thus, for example, the
idea of “script” comes from psychology, via artificial intelligence, and
linguistics, which ultimately introduced it in humor studies.
Very few terms have been created specifically for humor studies: take
“gelotophobia” (the fear of being laughed at) for example. The term
“gelotophobia” was introduced by Michael Titze (Titze, 1996), a German
psychotherapist, who had encountered the pathology in his practice.
It was then taken up by Ruch and his team and a formidable set of

8 It is true that second-degree theories are always inevitably influenced by first-degree

theories. The problem is to keep them apart as far as possible. Both theories have their place,
but confusing them will lead researchers to error.
9 See Downing & Recuero (2008) and Attardo (2001b) for some discussion.

13
 HUMOR STUDIES: A FEW DEFINITIONS

studies was produced, resulting, among others, in a special double issue

of HUMOR in 2009, guest edited by Ruch. Part of the work was to define
precisely what counts as gelotophobia and whether it manifests itself
across cultures.
In conclusion, in what follows we will be using the generic umbrella
term humor to indicate the general category, as is common usage in the
field. Any further differentiation will be introduced and motivated in
the text. In particular, in the chapter on literary humor, we will examine
more closely some of the distinctions mentioned above.

1.1.3 Joke
The term “joke” also deserves some discussion, because of the outsized
importance it has acquired, almost by happenstance. I have mentioned
the “joke” in the previous section repeatedly, with a rough definition
of a short (narrative) text ending in a punch line. As I mentioned,
there is nothing wrong with this definition, per se. However, we need
to distinguish between canned and conversational jokes. Canned jokes
are relatively decontextualized: they are built as small narratives that can
be reused. Conversely, conversational jokes exist and are produced in
context only, i.e., as part of a conversation. Conversational jokes tend not
to have a narrative introduction. One can think of it as the context of the
conversation playing the part of the setup of the joke (the introduction)
and the humorous remark being the punch line.
Consider the following extract from a dyadic (two speakers) conver-
sation recorded over computer chat, in which two students are talking
to each other. The setup was such that one student was in my office,
while the other was in my friend and colleague, Steve Brown’s. Steve kept
his office very tidy and orderly, whereas mine was a mess of papers and
books. Marina (M) delivers a joke on line 162, but as is very clear it
consists of just one turn “This looks like Sal’s” and relies on the previous
conversation for its context, for the setup of the opposition (the “clean”
office vs. the “dirty” messy office) and the other half of the opposition.
(7) 146 C // there’s lots of BOOKS [243][66] in
here//
147 1.57
148 M //yeah I’m LOOKin [250][68] around at
all these PApers [188][64]//
149 0.34

14
 1.1 TERMINOLOGY

150 M //hangin on the WALLS [172][65] //

151 0.52
152 M //this OFfice [198][71] is really DIRty
[182][67]//
153 0.64 ((smack))
154 C // this OFfice [294-279][67] is really
CLEAN [222][62] //
155 0.44
156 C //i THINK [334][63] it’s Steve BROWN’s
[206][58]//
157 0.92
158 M //AHH. [172][68]//
159 0.52
160 C //yeah//
161 0.883
162 M //this looks like SAL’s [179][72]//
163 0.21
164 M ((laughs))

Zajdman (1991) has shown that the difference lies on a continuum of

contextualization and that a conversational joke can be turned into a
canned one and vice versa (joke recycling): canned jokes may in some
cases originate from conversational jokes that have been decontexual-
ized. Conversely, canned jokes may be contextually adapted to the point
that a canned joke may be presented in a manner indistinguishable from
a spontaneous joke.
Consider the following joke:
(8) Mrs. de Gaulle enters a room where her husband is waiting. She
trips on the rug and involuntarily mutters “Mon Dieu!” He replies:
“How many times must I remind you? It is quite sufficient to say
“Mon général.”
which appears in Yvonne de Gaulle’s New York Times obituary, by Frank
J. Prial (Nov. 9, 1979, p. 4). One day, a student walked in my office to
retrieve an exam I had returned in class. I handed her her exam, which
had earned a 98% score (I was famous for being a tough grader in my
early days). She was surprised and exclaimed “My God!” to which I
replied, “You can call me Dr. Attardo.” The student was amused and,
of course, I did not tell her that all I had done was contextualize the
joke to her present situation. She probably thinks to this day I am very

15
 HUMOR STUDIES: A FEW DEFINITIONS

witty, but indeed very little effort was involved in recontextualizing the
canned joke.
The following is another example, from the same set of recordings as
above, in which Carmen recontextualizes a well-known joke
(9) What do you call someone who speaks two languages? A bilingual.
How about someone who speaks three languages? Trilingual. How
about someone who speaks only one language? An American.
using Marina’s modesty in saying that she does not speak any language
besides English (at least fluently) as the setup and delivering the punch
line recontextualized to apply to Marina “you are (an) American.”
(10) 276 C // do you speak any OTher [225][62]
languages besides ENGlish[242][60]? //
277 1.11
278 M // NOPE [251][71] //
279 0.42
280 M // nothin FLUently [216][59] //
281 0.12
282 M // little bit a this n THAT [210][60] n //
283 1.26
284 M // k OTher [185][73] thing //
285 0.66
286 M // you KNOW [147][65] = //
287 C // =you’re aMERican [227][60] //
288 1.03
289 ((both laugh))

Thus “joke” is an ambiguous term. It can refer to narrative texts and to

single turn witticisms. What accounts for its popularity in humor stud-
ies, then? In Attardo and Chabanne (1992), we reviewed several reasons
for the popularity of the term and the subject, in humor research:

1. jokes are short and therefore can be used freely without using too
much space;
2. they are easily available, especially on the internet;
3. they tend to be self-contained, and hence can be used without
much explanation/contextualization;
4. they come with a built-in guarantee that they are instances of
humor, unlike a novel or a conversation in which the humorous
passages are not somehow marked for the analyst;
5. they have been the subject of extensive research.
16
 1.1 TERMINOLOGY

In particular, we should mention the doctor’s wife joke10 which has

assumed in the linguistics of humor the role of an exemplar (Kuhn,
1962) i.e., a solved problem (as in “textbook problem”) used by begin-
ners and experts alike as an example or model of how to “do” science.
Part of theoretical linguistics training, for example, consists in solving
phonology and morphology problems that simulate field linguistics
problems (such as identifying minimal pairs or determining the phone-
mic or morphemic status of a given allophone or allomorph). The
doctor’s wife joke will be discussed in Chapter 6.
Finally, and this has caused much consternation among those who
study discourse, in popular parlance we speak of “joking” and “telling
a joke” for the delivery of any kind of humor in conversation, be that
prepared or prepackaged (canned) or extemporaneous (conversational)
humor. So, one should always be careful to specify what kind of joking
one means. Of course, we mustn’t forget that stand-up comedy and
scripted humor will often masquerade as improvised humor. So, when
relevant, that dimension should also be considered as well.

1.1.4 Mirth, humor, laughter

Another significant distinction is that between the cognitive stimulus,
which we agreed to call humor, its emotional effect, which is referred
to as mirth and the physical manifestation of the emotion: laughter and
smiling. See Figure 1.2.
Ruch (1993) proposed the term “exhilaration” to indicate the underly-
ing emotion of laughter/smiling caused (among other things) by humor.
Martin (2007) rejects the proposal because “exhilaration” has an excite-
ment component to its meaning, and prefers “mirth”: “Mirth (. . .) is
the distinctive emotion that is elicited by the perception of humor”
(p. 8). Both terms are used widely. I have preferred “mirth” in this book

Cognitive Emotional Physical

stimulus response manifestation
• Humor • Mirth • Laughter
Exhilaration Smiling

Figure 1.2 The relationship between humor, mirth, and laughter.

10 Raskin (2017a) jokingly calls it “the” joke.

17
 HUMOR STUDIES: A FEW DEFINITIONS

only because compounds such as “mirthful laughter,” are simpler than

“exhilaration laughter.”
Finally, the emotion caused by the cognitive stimulus may manifest
itself physically, typically in laughter or smiling. Note the all-important
qualification “may.” One may experience mirth and not show any visible
signs thereof. There are many issues at stake around these concepts:
for example, is smiling an attenuated form of laughter, or are they
independent phenomena? The question is beyond the scope of this
book.
It should also be pointed out that laughter and smiling may occur
without any accompanying emotion of mirth or exhilaration. For exam-
ple, many cultures have a laughter of embarrassment. Furthermore,
mirth or exhilaration may occur without humor. Conversely, as we saw,
humor may occur without mirth, exhilaration, laughter, or smiling. So,
there is no one-to-one correlation between humor and laughter and/or
smiling. There is at best a general association between the two, but it
would be a serious error to assume that because you observe either of
the two items the other must be present as well. We will return to this
issue in some of the next chapters.
Within humor studies it is not uncommon to see significant confusion
on the use of this terminology. First, there may be confusion among
the three levels: for example, one often sees claims that getting (as in
understanding) a joke results in laughter. This is true, only insofar as it is
taken to be an oversimplification of “getting a joke results in exhilaration
or mirth, which then is manifested as laughter.” Second, one may confuse
the levels. A good example of this is Bergson’s (1901) titled “Laughter”
which does in fact study comedy.

1.2 A brief overview of humor studies

The study of humor goes back to ancient times. Attention to humor

dates from the Egyptians (2600 BC) and the Sumerian (2200 BC),
who both documented humor in the form of jokes or humorous texts.
However, the Greeks are the first to consider the subject theoretically.
One can trace the origins of some of the modern theories of humor (see
Chapter 3) back to Plato and Aristotle (third century BC). I did so myself
in Attardo (1994, ch. 1). Besides philosophy, literary criticism takes a
front stage with Aristotle’s Poetics and Theophrastus’ Moral Characters

18
 1.2 A BRIEF OVERVIEW OF HUMOR STUDIES

(third century BC). Rhetorics likewise has taken up the problem of

humor, often with surprisingly modern insights and formulations, see
for example Cicero’s distinction (first century BC) between verbal and
referential humor (Attardo 1994: 26–29), which we will examine below.
Quintilian has the distinction, in his Institutio Oratoria, of having the
first coherent treatise on humor, if we except the lost second part of
Aristotle’s Poetics (Janko, 1984).
All this, however, is ultimately mostly of historical significance: for
example, a scholar may be interested in finding out where a given theory
comes from, what the earliest mention of, say, incongruity or superiority
is, or who was the first thinker to realize that surprise is an important
factor of humor. However, and this is an important point, all of these
theories have been supplanted by more recent, more specific, and far
more detailed, in short, more scientific, formulations. In what follows,
then, we will review how modern theories of humor got to be what they
are and when the field of humor studies, as a field of study, emerged. The
review is short and flippant. For dead-serious reviews see Attardo (1994,
pp. 14–59) or Larkin-Galiñanes (2017).
The Middle Ages were a dark period (pun intended): the Catholic
Church engaged in a prolonged and losing battle against laughter and
the rest of the people were too busy dying of the Black Plague and being
killed by marauders and barbarians to worry about the definition of
humor. We have to wait for the Italian Renaissance (1300–1400) to see
significant humorous texts being written and distributed and the rebirth
of literary criticism concerning humor. France, Germany, and England
eventually catch up with the Renaissance, and we get Erasmus’ Praise of
Folly (1511), Rabelais’ Gargantua et Pantagruel (1532), and eventually
Shakespeare. The theory of humor is not limited to the literary theorists:
in 1579, Laurent Joubert writes his Traité du ris (Treatise on Laughter),
which has the distinction of being the first extended treatment of humor
in a vulgar language (i.e., not in Latin, until then the vehicle for science)
and the first explanation of the physiology of laughter.
In 1598, Ben Johnson writes Every Man in His Humor which is the
first major use of the term humor in its modern sense. To be fair, the first
actual use of humor as comedy, is An Humorous Day’s Mirth a comedy by
one George Chapman, produced in 1597. Both Johnson and Chapman
are still using “humor” in the theory of humors sense: dating back to
Hippocrates (4th century BC), the idea was that one’s temperament
was determined by the proportion of four humors (i.e., bodily fluids):

19
 HUMOR STUDIES: A FEW DEFINITIONS

blood, bile (yellow and black), and phlegm. We still get a relic of this
conception in expressions such as a “sanguine” (=blood) temperament”
or a “phlegmatic” character. However, the association with comedy and
humor in the modern sense is there as well. On the four temperaments
and humor theory, see Hempelmann (2017).
In 1649, René Descartes writes Les passions de l’âme (The Passions of
the Soul) in which he presents a psycho-physiological explanation of joy
and laughter. Descartes connects psychological concepts, such as sur-
prise, physiological concepts, such as facial expressions and increased
blood flow, to the expression of joy (we would say exhilaration or mirth).
Descartes considers laughter to be the result of a mixture of emotions,
such as indignation against an evil, surprise, and joy, since the evil is
harmless (Descartes, 1649, Article 197).11
In 1651, Thomas Hobbes writes Leviathan which is the grandfather
of the aggression theories of humor, which he defines as “sudden glory,”
i.e., the triumph over an adversary. Needless to say, the idea of laughter
as aggression or superiority is already present in the Greeks. In 1695,
Congreve, followed, in the first half of the eighteenth century, by the Earl
of Shaftesbury (1709), Addison (1711), and Hutcheson (1750), publishes
essays on laughter and humor. Despite all these contributions, there is
still no sense of a coherent body of research.
Bain, the founder of the journal Mind in 1859, in The Emotions and the
Will discusses humor, followed by Spencer’s The Physiology of Laughter
(1860), and Darwin’s discussion of laughter (1872) in The Expression of
the Emotions in Man and Animals, which is still used in the field. Hall,
the founder of the American Psychological Association, publishes The
Psychology of Tickling, Laughing, and the Comic (1987), and Sully, the
founder of the British Psychological Association, An Essay on Laughter
(1902). With these publications, we certainly witness the first steps
within humor of the discipline that will eventually dominate the field,
psychology. Yet, while we definitely have a discipline in the modern
sense of the term, we still do not have a field, despite the influence
of Kant’s philosophical theory, and the connections to literary criti-
cism. See Simon (1985) for a broader and fascinating discussion, which
I am simplifying significantly. In particular, the mutual influences of the
psychologists and literary scholars are fascinating.

11 This definition eerily reminds one of the so-called benign violation theory of humor
(McGraw & Warren, 2010).

20
 1.2 A BRIEF OVERVIEW OF HUMOR STUDIES

At the beginning of the twentieth century, we have a remarkable

convergence, when three major intellectual figures publish books on
humor within a ten-year frame: Bergson (1899; in book form 1901),
Freud (1905), and Pirandello (1908). It is tempting to see the beginning
of humor studies in this trifecta, especially given that all three scholars
respond to a series of works on humor, within their disciplines. Freud,
for example, is influenced by Groos’ work (Simon, 1985). Yet, these
works, however significant, do not form a “critical mass” for an academic
discipline.
Where is linguistics, you might ask? Well, not really in the picture.
Until 1916, linguistics was part of philosophy, literary criticism, history,
or perhaps anthropology. For many linguists, linguistics was a historical
science. Linguistics does not come into its own until after 1916, after the
posthumous publication of Saussure’s Cours de Linguistique Générale.
Even then, linguistics has very little to contribute to humor studies,
except for the study of puns, which was considered the province of the
linguistics of humor. We will have to wait until the 1980s and Raskin’s
work (see ch. 6) for linguistics to take a central role in humor studies.
Isolated scholars elaborate sophisticated, even brilliant theories;
however, that does not make a field of study. The main defining
characteristics of humor studies is that it is an interdisciplinary field.
Interdisciplinarity requires that more than one discipline be used as the
source of both the problems (the questions) and the methodologies
(the tools) used to address them. If the problems come from one
discipline and the methodologies from another, all you get is an applied
field (for example, forensic linguistics, which seeks to answer questions
from forensic science using linguistic tools: e.g., is the person making
this threatening phone call the same person in these other recordings?).
What defines interdisciplinarity is that both the questions and the
methodologies come from a variety of disciplines.
In this sense, humor studies as a field, is born around the mid-1970s,
after Goldstein and McGhee (1972) book appeared, with the first humor
conference, held in Cardiff, Wales, in 1976, by Anthony Chapman and
Hugh Foot. The conference was followed by another one in 1979, in Los
Angeles, and then in 1982, in Washington, DC, in 1984, in Tel Aviv, and,
in 1985, in Cork, Ireland. The other significant event was that in 1982
Don and Alleen Nilsen at the University of Arizona, started the World
Humor and Irony Membership (WHIM), which held annual meetings
until 1987. In 1988 the last WHIM was held at Purdue university.

21
 HUMOR STUDIES: A FEW DEFINITIONS

The year before the International Society of Humor Studies (ISHS) was
created, which replaced WHIM. The ISHS has held a yearly conference
since 1988.
WHIM published extended abstracts, in a soft-bound book format,
called the WHIM Serial Yearbook (WHIMSY). This was the first peri-
odical humor research publication. It published yearly volumes from
1982 until 1989. Together, the WHIM conferences and the WHIMSY
proceedings established a reliable forum for scholars wishing to research
humor-related topics and helped establish the viability of a humor-
related journal.
In 1988, HUMOR: International Journal of Humor Research started
publication, with Victor Raskin as editor-in-chief. The role of HUMOR
in legitimizing the field by offering a high-quality outlet to humor
research should be clear, since it constituted the only respectable, reliable
forum to publish humor research, at least for the first 20 years of publica-
tion. HUMOR is still being published currently, but now other journals
specializing in humor exist: the Israeli Journal of Humor Research: An
International Journal which started publication in 2012, the European
Journal of Humour Research which started in 2013, and the Rivista
Internazionale di Studi sull’Umorismo, based in Italy,12 which started
publication in 2017. In the Francophone world, CORHUM (the Centre
de Recherche sur le Comique et l’Humour) has also been publishing
a journal since 1988. In 1994 the Humor Research book series, with
Mouton the Gruyter, was launched, followed by Transaction Publish-
ers (now with Routledge), Wayne State University Press, Gordon and
Breach, Benjamins, and others. Moreover, numerous articles on humor
appear regularly in many relevant journals such as the Journal of Prag-
matics, Discourse Processes, and many others. Finally, the publication of
such reference works as the Encyclopedia of Humor Studies (2015) and
the Handbook of Language and Humor (2017), no doubt also helped
establish humor studies as a field, and within it the linguistics of humor.
So, in conclusion, I think it is fair to say that today, the field of humor
studies has established itself as a legitimate field of academic research.
Another way to look at the birth of the field is to track the emergence
of university courses on humor. Wycoff (1999) is, to the best of my
knowledge, the only attempt at charting the rise of the teaching of humor

12 International Journal of Humor Studies; the acronym RISU is the ablative case of Latin
risus, meaning roughly “with or through laughter.”

22
 1.2 A BRIEF OVERVIEW OF HUMOR STUDIES

as an academic subject. He reports that the first pioneering courses were

taught in the sixties and that by the 1970s the field begun to emerge as a
credible field. The earliest course documented in Wycoff was taught in
1966, at Michigan State. Wycoff reports that by 1999 there were in excess
of 100 humor courses being taught.

1.2.1 Why do we need a field of study?

The reason I wish to establish the origins of the field of humor studies
is that its presence is a mutual guarantee of quality for the scholars that
engage in research within it and it allows the recruiting of young scholars
and the establishment of the research on a sound basis. Consider a
potential objection: one might question the significance of trying to
establish when humor studies “came together” as a field. After all, one
may reason, research is ultimately done by individuals and works of
genius advance knowledge, regardless of whether there is a group of
people who support it. One may use as an example, Wittgenstein, who,
notwithstanding his early ties to the Neopositivists, developed his work
in splendid isolation.13
Now, setting aside what we know about the history of science, which
tells us that this idea of the lone wolf author is deeply mistaken, there is
another reason for needing a community of scholars. Kuhn (1962) iden-
tifies “normal science” as what scientists do most of the time: they move
within an established paradigm, that tells them what methodologies are
and are not acceptable, what kinds of observations are good and what
are not, what kinds of experiments are acceptable, even what kinds of
arguments are valid, and so on.
Let us consider an example: within linguistics itself there is a sharp
difference on the status of intuitive data: for some linguists, if I think
about it and come to the conclusion that I would never say the sentence
“very banana I love much,” that counts as a bit of evidence (a datum).
Put together a lot of those and you get data. For many other linguists,
this kind of data is deeply suspect. These other linguists object that they
are not naturally occurring data. They want their data to come from a
corpus of language facts that occurred independently of the observer,
the larger the better. Other linguists also object that the data are not

13 I do not claim that this is a valid description of the genesis of Wittgenstein’s oeuvre,
but his work is definitely less embedded in a tradition, than, say, Bertrand Russell’s.

23
 HUMOR STUDIES: A FEW DEFINITIONS

naturally occurring, but would like their data to come from a small set
of interactions (conversations, primarily, albeit not limited to spoken
data, so they would include online chats, etc.) and preferably with the
involvement of the researcher, which gives him/her access to the inner
dynamics of the exchange. The naturally occurring data linguists mock
the first linguists as “armchair” linguists and decry the “artificial” nature
of the data. Needless to say, the armchair linguists mock the inherent
limitations of finite corpora and restricted samples.
The point of the previous paragraph is not to uncover the seedy
underbelly of the profession but rather to point out the fact that even
what counts as data is determined by the paradigm one does science
within.14 Where and how to find one’s data is not the only purview
of a paradigm: a paradigm determines how scientists go about their
business of generating hypotheses, arguing for them, what counts as a
valid argument, and so on. My overall point then is that the existence of a
field gives us the paradigm from which we can determine the value of the
research. The existence of a scientific society that holds periodical con-
ferences, the existence of publications (journals, yearbooks, book series,
etc.), which vouch for the quality of their contents, are all elements that
contribute to the emergence of a scientific field of study. Needless to say,
the interactions between the scholars themselves, collaborations, tips,
references, discussions, etc. are also significant, but I believe that the
paradigmatic elements, in Kuhn’s sense, are most important due to one
last factor: academic respectability.
In order to be viable, a field of study must be accepted as a respectable
area of research in which one may dissertate and/or find a job. These
may seem crass considerations, but realistically speaking, a field thrives
when young scholars with an interest in it are encouraged to pursue
their interest. However, responsible advisors must warn their students
if their choice of dissertation topic may prevent them from getting
desirable positions in academe. If humor is a “questionable” topic, it will
be hard for young scholars to enter the field. It was no coincidence that
at the beginning of humor studies their proponents were all established
academics with significant credentials in non-humor-areas. The same

14 Technically, the differences between linguists approaches are not paradigms, because
the differences between the various “schools” are not significant enough. There is also the
small matter that in some ways linguistics is not yet a mature science. In Kuhn’s sense, it
is pre-paradigmatic: there are competing approaches, without a dominant paradigm. These
distinctions are significant, but need not concern us in this context.

24
 1.3 A FEW BASIC DISTINC TIONS

problems have been reported, at their respective beginnings, for such

fields as sexology, film studies, etc. Generally speaking, the reasoning
seems to be that, if the topic is fun, the research on it cannot be serious
and/or respectable.15
Occasionally, other academics or members of the public ask me if
humor studies is fun. My reply is to offer to read them a few pages of
Attardo (1994). No-one has ever taken me up on that offer. I will admit,
however, that data collection when one is looking at memes online or at
comedians performing is less personally difficult than perusing letters
from the Holocaust (one of my colleague’s field of study was genocide—
now that is a serious field of study).

1.3 A few basic distinctions

We can wrap up this introductory chapter with two more distinctions

which are general enough so as not to fit clearly in any of the next
chapters and yet important enough that they need to be introduced early.

1.3.1 Linguistic humor vs. verbal humor

Linguistic humor is humor expressed through language, spoken or
written. Another term used to refer to linguistic humor in this sense
is “verbally expressed humor” as opposed to visually expressed humor,
in film/television studies.
Linguistic humor is a subclass of semiotic humor (see Chapter 5).
Other types of semiotic humor are visual humor, musical humor, mul-
timodal humor (humor expressed in media that involve more than
one communicative modality, such as television, cinema, theater, video
games, etc.). Physical humor (e.g., slapstick, clowning) also falls under
semiotic humor. This is represented visually in Figure 1.3. The examples
under each category are not meant to be exhaustive. They just represent
the kind of humor that would be classified under that category.
Linguistic humor should not be confused with verbal humor, as
opposed to referential humor. Simplifying quite a bit, verbal humor con-
sists of puns (ch. 9), i.e., humor which relies significantly on semantics

15 In this respect, my personal hero is Dolf Zillmann, who after studying humor, went
on to study pornography!

25
 HUMOR STUDIES: A FEW DEFINITIONS

Semiotic
humor

Mono-
Multimodal
modal

Video
Linguistic Visual Musical Television Cinema Theater
games

Cartoons Painting

Figure 1.3 Semiotic humor hierarchy.

(meaning) and similarity or identity of two linguistic forms: phonemic

(sounds) or graphemic (spellings) whereas referential humor is just
based on semantics. To reiterate, referential humor relies only on mean-
ing. Verbal humor also relies on meaning, but not exclusively. Needless
to say, what kind of meanings we are talking about is a significant issue,
which will be tackled in the chapters to follow.
The distinction is not unique to linguistic humor. For example, there
exist visual puns (Lessard, 1991; Hempelmann and Samson, 2007;
Mitchell, 2007; Tsakona, 2009). Berger (1993, p. 45) provides a duly
broad definition of semiotic pun as “a signifier that stands for two
signifieds.” Turner (2005, pp. 204–205) discusses several examples,
as well.

1.3.2 Humor and meta-humor

The next distinction we need to address is that between humor and
meta-humor. The reason for doing so is that, as we will see, meta-humor
can be seen as an exception to the fundamental definitions of humor as
incongruity. So we need to be very clear about this category.
To put it simply, meta-humor is humor about humor. The term
has been used widely, along with other terms, such as anti-humor,
second-degree humor, self-referential humor. Attardo (1988) reviewed
the (scant) literature.
We will begin by defining our terms. Intuitively, meta-humor comes
from meta-language, a technical term within logic for a language used
to talk about another language (the object language). The idea of

26
 1.3 A FEW BASIC DISTINC TIONS

metalanguage is closely related to Russell’s theory of types (Russell,

1908; Copi, 2011) which he elaborated to solve the paradox he had
found in Frege’s set theory. Without getting too technical, the set of all
dogs consists of all the entities in the world that are dogs, so the set of all
dogs does not contain itself. Conversely, a set of sets would contain itself
(since it is a set). Consider now the set of all sets that do not contain
themselves. Should it contain itself? If we say yes, we are violating its
definition (it is the set of sets that do not contain themselves), and if we
say no, it meets the requirements for belonging in the set and so should
be in it. Russell developed his theory of types which we can describe as
essentially forbidding the kind of cross-level definition we have used.
Popular versions of this paradox exist, such as the barber of the village
who shaves only those who do not shave themselves. Should he shave
himself? Interestingly, Bateson’s (1972) theory of schizophrenia as the
result of a “double bind” is based on the same principle. Bateson will
figure prominently in the definition of play, in the next chapters.16
Jakobson (1960) introduced metalanguage to linguistics in his famous
model of communication, which distinguishes six functions of language,
which include the metalingual (metalinguistic) function. So, in a natural
language (i.e., a language spoken by people, as opposed to a formal
language such as logic, or a computer language, such as COBOL or
Java), a metalinguistic statement is a statement about (some part of)
the language itself. Metalanguage should not be confused with self-
referentiality. The sentence
(11) The first sentence of Moby Dick is “Call me Ishmael”
is metalinguistic but not self-referential, whereas
(12) This sentence consists of six words.
is self-referential. All self-referential sentences are metalinguistic, but
not all metalinguistic sentences are self-referential. For example, the
sentence
(13) All self-referential sentences are metalinguistic, but not all met-
alinguistic sentences are self-referential.

16 Surprisingly, Russell dealt with meta-humor. Russell’s discussion of the hierarchy of

jokes can be found online at http://www.gutenberg.org/files/38430/38430-h/38430-h.htm.
Blackwell (2011) provides some background information on this little known aspect of
Russell’s thought.

27
 HUMOR STUDIES: A FEW DEFINITIONS

is metalinguistic but not self referential. Whereas

(14) I am lying.
is both metalinguistic and self-referential.17
So, what is meta-humor then? Any humor about humor would qualify.
So for example, making fun of someone for not telling a joke properly
would be meta-humor. It is clear that self-referential meta-humor is the
more interesting category. The infamous joke
(15) Why did the chicken cross the road? To get to the other side.
is a self-referential meta-joke because it refers implicitly to the form of
jokes, in particular to the presence of a punch line at the end of the text,
and fails to deliver it, thereby violating incongruously the expectations
set up by the stereotypical joke introduction (a riddle in this case).
The shaggy-dog jokes I examined in Attardo (1988) are precisely long-
winded jokes that do not deliver a punch line at the end (see also Fry,
1963, p. 148; Suls, 1983, p. 53; Lefort, 1986, p. 191).
The reason self-referential meta-humor is significant for humor the-
ory is that it works by violating the rules of the genres of humor.
Therefore it will always be an apparent exception to any generalization
about humor. As we will see in Chapter 4, a punch line occurs in all
jokes at the end of the text. In meta-jokes the punch line is that there
is no punch line. This sounds paradoxical, but using Russell’s theory of
types, the paradox is dispatched easily: there is no level 1 punch line,
which creates a level 2 punch line. A related reason, is that joke-cycles,
such as the light-bulb jokes, start out as level 1 jokes but soon develop
jokes of level 2 that are based on knowing the format of the joke and
then incongruously not delivering it. Quite literally so, in the following
example,
(16) How many feminists does it take to screw in a lightbulb?
That’s not funny!
I examine joke cycles in these terms in Attardo (2001, pp. 69–78).
Suls (1983) discusses meta-humor as one “problem” of the incon-
gruity/resolution model. His answer is right on target: “they provide
enjoyment by playing off the fact that they pretend to be humor but do

17 Epimenides’ paradox famously consists in saying that all Cretans are liars and the
speaker is Cretan.

28
 1.4 FURTHER READINGS

not possess all of the requisite features” (p. 53). Indeed, this is correct:
meta-humor is the fly in the theoretical ointment of humor theory
because meta-humor will invalidate any generalization about humor or
any of its genres, because meta-humor plays with the expectations about
humor and its forms. Because of the relative sophistication of jokes using
meta-humor each generation rediscovers them, which may help explain
their popularity.18 However, the exception is only illusory, as far as the
incongruity theory goes: there is an incongruity in the chicken crossing
the road joke or in the shaggy-dog joke: it is precisely that there is no
incongruity where we expected one.

1.4 Further readings

There is no comprehensive history of humor studies. The web site of

the International Society for Humor Studies (http://www.humorstudies.
org) provides some helpful resources, including a listing of all humor
research conferences organized by the society. The preface to McGhee &
Goldstein (1983); Derks (1996), which is the introduction to the second
edition by Transaction of the classic Chapman & Foot (1976); Car-
rell (2008); and Goldstein & Ruch (2018), which is the preface to a
Festschrift for Paul McGhee, in HUMOR, are all good places to gather
information about the history of humor studies.
Further discussion of the terminology of humor studies can be found
in Hempelmann’s (2017) chapter of the Handbook of Humor Studies.
Attardo (1994, pp. 3–5) offers a review of some of the attempts at internal
subcategorization. The discussion between referential and verbal humor
is discussed in some depth in Attardo (1994). There is no comprehensive
treatment of meta-humor, beyond the texts indicated in the chapter.

18 The Wikipedia page for the chicken crossing the road joke reports a version dating
back to 1845. https://en.wikipedia.org/wiki/Why_did_the_chicken_cross_the_road

29
 2
Methodological preliminaries

2.1 Competence and performance 31

2.2 Identifying humor 38
2.3 Mirthful vs. non-mirthful laughter 42
2.4 Keying 50
2.5 Identifying humor: the triangulation approach 54
2.6 Further readings 56

This chapter starts out by setting the background for the rest of the
work. Thus, it is fairly dry and abstract. The reader may consider it
a necessary evil but should not bypass it, as the points discussed in
this chapter are truly foundational. Broadly speaking, we are concerned
with methodology. When students ask me why methodology is so
important, I explain that it is how you avoid making a fool of yourself.1
We start out by discussing the competence–performance opposition,
which underlies most of the book and the principle of commutation,
which underlies the theoretical approaches to the linguistics of humor.
We briefly consider other methodologies (quantitative, qualitative, etc.)
to show that, whatever one’s methodology, there are rules and war-
rants that must be followed. The second part of the chapter deals with
the all-important methodological question of how do we know that a
given text is humorous. The de-facto criteria of identification (laughter
and smiling) are discussed and problematized. Finally, we conclude
by proposing a triangulation approach which ecumenically gathers all
sources of information about the text to make the determination.

1 Surprisingly, so is etiquette. Food for thought.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 2.1 COMPETENCE AND PERFORMANCE

2.1 Competence and performance

It is perhaps peculiar that we will begin by discussing a meta-theoretical

problem (i.e., an issue concerning the theory of theories), rather than
by—say—defining our object of study, humor. The reasons is that
“humor,” as we saw, is a folk-concept, i.e., it exists pre-theoretically. What
does that mean? Simply that the fact that people refer to something as
“humor” does not mean that they know exactly what that thing is, where
its boundaries lie, how to tell it apart from other similar things, etc.
Think for example of whales: for centuries people thought that whales
were a kind of fish, only because they live in water and look like a large
fish. However, when the pre-theoretical conception of the whale-as-fish
came under scientific scrutiny it was abandoned for the current theory
of the whale-as-mammal.2
A good way to begin is to ask the question: “what should a humor
researcher study?” Should we study actual texts? The exchanges in which
instances of humor occur? The psychological motivations of the jokers
and/or their audience? The unconscious motivations of either? Their
goals and/or responses in/for joking? The electrical impulses in the brain
of the person hearing a joke?
The answer is, clearly, all of the above, since they interest one or the
other discipline. Moreover, those I listed above are not, by any means,
the only topics worth studying, they are just a sampler of questions. Can
we narrow down things a little, if we take a linguistic standpoint? Clearly,
some aspects of the problem become foregrounded (for example the
types of texts, the genres, so to speak, of humor) but still there are a
plethora of approaches to the, as yet undefined, object of study. At this
point, we must turn to theoretical linguistics for insight on the definition
of one’s object. Saussure’s fundamental langue vs. parole distinction,

2 Let us note that the disciplinary standpoint one assumes largely (pre-)determines the
object of inquiry. This meta-theoretical point is of some interest, if for nothing else in that
it prevents one from beginning one’s inquiry by defining one’s object of investigation. One
must first determine one’s (sub-)discipline and then one’s object. Objects of research do not
preexist theories. This should not be construed as a relativistic position, let alone as a social-
constructivist one. Reality preexists any of its conceptualizations. The fact that science is
a social enterprise should not be confused with the claim that reality is one as well. The
point of this note is merely that theoretical linguistics will approach humor differently than
sociolinguists, who will approach it differently than psycholinguists, etc. And naturally,
one is free to take a multidisciplinary stand, as I will, and look at humor from different
perspectives. However, one will see different things.

31
 METHODOLOGICAL PRELIMINARIES

later echoed3 in Chomsky’s “competence” and “performance,” provided

linguists with the general “object” that the discipline had lacked, up to
that point. Linguistics must study the system (the langue), the abstract
set of relationships that convey meaning, while other disciplines, such
as stylistics, will study the parole, the concrete, actual realizations of the
abstract possibilities of the system.
We need not pursue this issue any further. The significant aspect, from
our point of view, is that theoretical linguists distinguish between two
levels, or two approaches to language: an abstract one (competence) and
a concrete one (performance). Seen from this perspective, it becomes
clear that we need to develop a theory of humor competence and a
theory of humor performance, in keeping with the assumption that the
former founds (i.e., precedes logically) the latter.
What are the advantages to this approach? By adopting this distinc-
tion, we neatly distinguish between the essential4 factors (i.e., the neces-
sary and sufficient conditions for a text to be funny) and the accidental
factors (e.g., who happens to have told a given joke, when, where, etc).5
Furthermore, by defining the necessary and sufficient conditions for a
text to be funny, we will have a falsifiable, scientific theory, since any
counterexample would falsify the theory: it would merely require that
one produce a text that meets the necessary and sufficient conditions
and is not funny or that one produce a text that is funny and lacks any
of the supposedly necessary and sufficient conditions. Not to mention
that by avoiding those factors that are not essential we will not be led
astray by facts which, however interesting, are ultimately irrelevant to
the definition we are seeking.
Needless to say, the claim that we should build both a theory of
competence and of performance is not a mere condescending nod
towards applied linguistics. Many have—and rightfully so—objected to
the devaluation of performance, implicit in Saussure and explicit in
Chomsky. Even better, starting with Bally (a student of Saussure, and

3 Chomsky’s competence is in this respect identical to Saussure’s langue, except that

Chomsky locates competence in the genetic make-up of humans, whereas Saussure located
competence in the social (i.e., it is a social phenomenon), after Durkheim. There are some
differences between the two conceptions, most notably in the role of syntax. These need not
detain us in this context.
4 Recall the definition of “essence”: what makes a given thing that thing.
5 Of course, if you are interested in who tells jokes to whom, then those become
the essential factors that your theory will focus on and the content of the jokes become
accidental. Each perspective determines its object of study.

32
 2.1 COMPETENCE AND PERFORMANCE

editor of the Cours), many have theorized (and practiced) a linguistics

of the parole/performance. We will return to the performance aspect of
humor below and in Chapter 10.
How does one go about reaching the competence level? After all,
we have only access to performance data. Different disciplines and
fields have different ways of approaching the issue of idealization, or,
to put it differently, to insure that their results are not contaminated by
any number of methodological problems. For example, in psychology,
sociology, etc. the use of large pools of subjects protects (by the law of
averages) the conclusions drawn by the researchers against the vagaries
of individual variation. For example, there is always the possibility that
a subject has “seen through” the test he/she is being administered and
that he/she decides to answer in a deliberately nonsensical way merely
to irritate the test administrator. However, the possibility that a majority
of subjects did is close to zero and thus in a sufficiently large sample that
the aberrant behavior of that lone subejct is eliminated. Thus, the larger
and more representative the sample, the more secure the conclusions
drawn from it.
So, how does theoretical linguistics reach the competence level? Lin-
guists use a process of idealization quite different from those of statistics,
because quantitative data are basically irrelevant from that point of view.
The likelihood that the sentence
(17) My pink elephant loves to dance the tango.
has ever been uttered is pretty low, whereas the following sentence (or
at least sentences built on the schema) is fairly common:
(18) Me Tarzan, you Jane.
but nonetheless (17) is grammatical and (18) is not.6 In fact, one of
the most significant points that Chomsky uses to justify his generative
approach to language is that language makes infinite use of a finite set of
elements. Therefore, no sample may, in principle, ever guarantee that it
is exhaustive (i.e., it contains all the relevant facts).
Grammaticality, meaningfulness, ambiguity, and their counter-
parts, ungrammaticality, semantic anomaly, and monosemy are not

6 Example (18) is ungrammatical because it contains no verb and because the subject is in
the accusative case. The point here is not on grammaticality of example (18) but on the fact
that speakers may recognize the grammaticality of sentences they have never encountered
before.

33
 METHODOLOGICAL PRELIMINARIES

determined statistically, but rather via a theoretical elaboration based

on the principle of commutation, a.k.a. distinctiveness.

2.1.1 Application of the principle of commutation to humor

At the most basic level, in phonology, two sounds are said to be different
phonemically if they are distinctive, i.e., they are part of a minimal pair,
such as /bɪt/ (“bit”) and /bit/ (“beet”). Because changing the sound [ɪ] to
[i] changes the meaning in the minimal pair, then the two sounds are
functionally different, i.e., phonemes. Consider now the following joke:
(19) Harry prays to God: Dear Lord, please make me win the lottery.
The next day Harry begs the Lord again: Please make it so I win
the lottery, Lord!
The next day, Harry again prays: Please, please, dear Lord, make
me win the lottery!
Suddenly he hears a voice from above: Harry, would you kindly go
and buy a lottery ticket. (https://short-funny.com/funniest-jokes-
3.php)
Consider the following possible changes to the text:

1. Change the name from “Harry” to “Larry”

2. Change the name from “Harry” to “Mary”
3. Change the name from “Harry” to “Abramovitz”
4. Change the name from “Harry” to “Oluwakemi”

while the first change has virtually no effect on the text, the second one
changes the gender of the supplicant, but has no effect on the humor.
The third changes the joke into a Jewish joke, but likewise leaves the
humor untouched. The fourth change makes the joke about a Yoruba
person, but as before leaves the humor unchanged. So, we know, that the
name and characteristics of the person doing the supplication is largely7
irrelevant to the humor.

7 But not completely: if we replace “Harry” with Jeff Bezos, the multibillionaire (note the
“b”) founder of Amazon, the fact that he wants to win the lottery becomes incoherent and
ruins the setup of the joke.

34
 2.1 COMPETENCE AND PERFORMANCE

Consider now another set of changes:

1. Change “beg” to “pray”

2. Change “the next day” to “the next week”
3. Change “a voice” to “the Lord’s voice”
4. Change “The Lord” to “God”
5. Change “The Lord” to “Buddha”
6. Change “the lottery” to “a trip to the Bahamas”
7. Change “buy a lottery ticket” to “I already granted your wish”

Once more, the first four changes have no impact on the humor, thus
showing that those are not significant details, as far as the humor
goes. However, with the fifth change, the humor disappears, because
Buddhism does not believe in a personal god (in fact in any kinds of
god) and moreover, to the best of my knowledge, there is no tradition
of asking the Buddha for personal gifts. So, in order for the joke to work
we need a superhuman personal entity capable of bestowing gifts and
willing to do so. So, in a sense, this is a Judeo-Christian joke at a deeper
level than it is a joke about Harry. However, the next two changes are
even more devastating to the humor: if Harry wants to win a trip to
the Bahamas, the element of “random drawing” of the lottery is gone,
hence the fact that an omnipotent god could, if they so decided, make
an individual the winner of the lottery, is also gone. Compare “make me
win the lottery” to “buy me a ticket to the Bahamas.” The former requires
a superhuman entity, the latter just someone with a few thousand dollars
to spend (i.e., the cost of a ticket to the Bahamas).
Finally, the last change also eliminates the humor, by removing the
explanation why god has so far not granted the wish. Note that god
is quite willing to grant the wish, but, reasonably enough, points out
that in order to win the lottery, one has to buy a ticket, apparently
disregarding the fact that if god is capable of making someone win the
lottery they could perfectly well make them magically acquire a ticket
as well.
So in conclusion, we have been able, by applying the principle of
commutation, to determine that the joke “works” because a) it involves
a personal god character; b) it involves winning the lottery, which has
a random extraction aspect to it and; c) it requires god to be willing
to violate randomness but not causality. Let me add that this is just a

35
 METHODOLOGICAL PRELIMINARIES

partial8 and informal analysis. We will see the actual application of this
methodology to semantic and pragmatics in Chapters 6–7.

2.1.2 Other methodologies

We have seen that theoretical linguistics is built, methodologically
speaking, on the application of the principle of commutation. Con-
versely, variationist linguistics (a branch of sociolinguistics), achieves
idealization statistically, i.e., by evaluating the likelihood that a given
result, for example that speakers laugh more frequently when in a group,
could have been arrived at by chance. This is not the place to discuss
the actual statistical tools used to arrive at this determination, but we
will review a few basic concepts. In order for data to justify a given
(theoretical) conclusion, the following conditions must be met:

• The sample must be representative of the population (i.e., all the

things you want to describe).
• The variables must be controlled as much as possible. Special
methodologies (such as double blinding and control groups) have
been developed to ensure that no unaccounted for variables affect
the data.
• Proper statistical measures must ensure that the likelihood that a
given result was arrived at by chance is kept below a given threshold
(usually 0.05 i.e., 5%). This is called “significance” and is not the
same as significance in everyday parlance (for example, “this is a
significant problem”).

Unless these prophylactic measures are adopted, the only thing that
empirical data can tell us is the tautological fact that they exist and have
been observed. Corpus-based and corpus-assisted studies, in which the
researcher begins by securing a (large) body of relevant data, likewise
give the same level of attention to the representativeness of the corpus,
because if the corpus is not representative of the population one wants
to generalize to, then the study is pointless. Suppose one wants to

8 In the version of the joke that I first heard, probably from Victor Raskin, in the late
1980s, the man is described as “righteous” which goes a long way to explain his entitlement
to ask God for a gift. Furthermore, much was made of him being apologetic about doing
so. A considerable improvement on the quality of the joke, but not a change in its “moving
parts” that make the humor happen.

36
 2.1 COMPETENCE AND PERFORMANCE

describe jokes that circulate among high-school students. A corpus

of Facebook posts would not be a valid corpus to generalize to all
high-school students, because a) many high-school students do not use
Facebook, which is perceived to be for “older” people, and b) because
not all jokes that circulate among the students might be posted on
Facebook.
It should be noted that phenomenologically oriented approaches dif-
fer from either of these approaches I have sketched above. They will be
discussed in Chapter 11. Finally, within applied linguistics, researchers
speak of “warrants” for the research. Essentially warrants are “validity
checks.” Among the warrants are

• credibility
• replicability
• transferability
• freedom from biases.

Credibility is related to the level of validity that one’s research has (valid
research is credible). There are various criteria of validity (Creswell &
Miller, 2000). In particular their discussion of triangulation (Creswell
& Miller, 2000, pp. 126–127) will be quite relevant, particularly about
the thorny problem of identifying humor in a text (see below, Section
2.5). In a nutshell, triangulation is defined as using “convergence among
multiple and different sources of information” to validate a category or
an interpretation. The focus is essentially on relying on more than a
single form of evidence and in fact in seeking a variety of sources of
evidence. Incidentally, the term “triangulation” as used in linguistics
is a metaphor going back to ancient sailing techniques and does not
necessarily require three sources. One can triangulate with four, five, or
more sources of evidence.
Replicability is achieved by being as explicit as possible on one’s
procedure, so that other scholars may reproduce a given study and
check whether they get the same results. Transferability concerns the
lack of ad hoc explanation: if we have an explanation for a phenomenon
that applies only to that particular phenomenon, we cannot be sure of
the validity of the explanation; conversely, if we can apply the same
explanation to several independent phenomena, there is a greater prob-
ability the explanation is not illusory. Freedom from biases is fairly self-
explanatory: it is well known that if we go into a situation expecting that,

37
 METHODOLOGICAL PRELIMINARIES

say, women are less funny than men, we will probably observe precisely
that (this is called confirmation bias).

2.2 Identifying humor

We now need to introduce another important methodological aspect of

studying humor and namely finding it. How do we know that something
is funny, or how do we know that a given part of a text, or a given
situation, is humorous? Historically, this has not been a big problem:
either scholars were operating with texts, such as jokes, that came pre-
labeled as humorous, or they were working with small corpora of data
they themselves had collected and thus they had a reasonable idea of
what was funny and why people were laughing. When we start looking
at large corpora of data collected by third parties, or at large texts, such
as novels, the presumption of humorousness is no longer there and the
question of whether a given turn in a conversation, or a given passage in
a text is humorous or not becomes a serious issue.
Again, historically, many researchers have used a short cut: essentially,
they assumed that the presence of laughter was enough to determine
that there was humor. For a variety of reasons, the primary one being
that video recording was very expensive and cumbersome, while audio
recording was easy and cheap, smiling was pretty much ignored, until
recently, despite the fact that smiling is the most frequent reaction to
humor (Ruch, 2008).
In this respect, psychologists and sociologists have a much easier task:
they ask a large number of people “is this stimulus funny?” or “how
funny do you find this?” and then use statistics to weed out the weird
and unreliable answers and then they have an answer: a given stimulus is
funny, and it is moderately so, for example. A linguist is in a much more
complex position: suppose he/she is analyzing a conversation, in which
one of the speakers says something, and the other responds by smiling.
Sure, the linguist could ask the interlocutor if they were smiling because
they got the joke or appreciated its humor, but the problem is that there
is only one speaker who can answer that question and so if the speaker
is an outlier or confused, there is no way to ask another 100 subjects.
This means that there are all sorts of possible biases, which cannot
be eliminated by aggregating the responses of many subjects. Some
linguists have responded to this by simply not asking and assuming that

38
 2.2 IDENTIFYING HUMOR

we just do not have access to the inner states of the speakers. Others
argue that we can reconstruct them from their behavior. Many use the
short cut and check whether either speaker is laughing or smiling. If
they are, one of them experienced humor. So, most of what follows in
this chapter can be read as the answer to the question: is this a legitimate
short-cut?
So we need to answer the question, “Can laughter/smiling be reliably
used to identify humor by themselves?” The rest of the chapter will
illustrate why the answer is negative and propose more reliable ways of
identifying humor.

2.2.1 Humor, mirth, and smiling/laughter

The reader will no doubt recall the fundamental distinction we intro-
duced right at the offset of the discussion between the cause (humor) and
the emotional effect (mirth) and between the mental and physiological
manifestations of humor and mirth. Humor itself is a cognitive process,
which takes place in the brain. Mirth or exhilaration is the emotion
one feels when experiencing and appreciating humor (Martin, 2007;
Ruch, 1993). Emotions have physiological correlates, so mirth straddles
the mental/physiological boundary. Emotions are manifested through
external (physical) manifestations. These manifestation may be inten-
tional or unintentional, i.e., they may happen despite the intention/will
of the speaker or they may be intentionally planned and executed. At
the unintentional end of the continuum one finds the idea of “leakage”
proposed by Ekman, who asserts that emotions “leak” through the
control that the speaker exerts (Ekman & Friesen, 1969). At the other
end is the idea of emotions as social displays, i.e., the manifestations
of emotion are meant to send a message to our audience (the people
around us) (Fridlund, 1994). Needless to say, intermediate positions are
also found, such as Buck (1980; 1994) who thinks that facial expressions
are both internal and social manifestations.
Both smiling and laughter may occur spontaneously, in which case
they are called “felt” or “spontaneous,” or they may be deliberate and
are then referred to as “volitional” or “voluntary” smiles/laughter. The
prototypical involuntary laughter is often described as “cracking up” or
“breaking up,” further emphasizing the lack of agency of the speaker and
his/her powerlessness in refraining from laughter. Attempts to suppress
laughter, for social constraints, such as funerals, formal ceremonies, etc.,

39
 METHODOLOGICAL PRELIMINARIES

may in fact make it even more difficult to refrain from laughing (Goff-
man, 1974, pp. 351–354). As we will see below, felt smiling can be
differentiated from social smiling. While the facial displays and acoustic
productions of laughter do not allow the differentiation of felt and
volitional laughter, they activate different areas of the brain and so in
principle may be discriminated. So, can we use felt smiling/laughter to
identify humor? Let us consider them separately, first.

2.2.2 Smiling
Ruch reports that there are about 20 types of smile (2008, p. 21). They
result from the combination of the activation of different muscles in the
face. Ruch lists the following five: zygomatic major, zygomatic minor,
risorius, levator anguli oris, buccinator (1995/2005, p. 113). He notes
that only the zygomatic major is involved in the smile of enjoyment.
Felt smiling is marked by the so-called Duchenne display which
involves the flexing of the zygomatic major muscles, which raise the
corners of the mouth, and of the orbicularis oculi, which cause the
wrinkling of the corner of the eyes. Volitional smiles do not involve
the orbicularis oculi and/or are asymmetrical. A full discussion of the
muscles activated in smiling can be found in Platt & Ruch (2014). The
primary difference investigated in psychology is whether the smile is
felt (genuine) or social (deliberate, contrived). The Duchenne display
is considered to be the test of the felt enjoyment smile.9 However,
recent research has challenged the validity of the Duchenne display
test of spontaneity (cf. Gironzetti, 2017a, p. 32, for a discussion). Other
emotions which produce facial expressions may also be mixed in, such
as embarrassment, negative emotions, flirting, etc. (Ruch, 2008, p. 22).
These other emotions may cause other muscles to be activated.10
Ruch (2008, p. 21) reports that, in experimental settings, smiling is
the most frequent response to humor. The subjects are five times more
9 The situation is more complex: Ruch (1995/2005) notes: “Voluntary smiles are more
frequently unilateral (present in one half of the face only) or asymmetrical (stronger in one
half of the face); their onset is abrupt or otherwise irregular but not smooth; they are more
frequently too short (less than half of a second) or too long (more than 4 sec); and they are
more frequently asynchronous (i.e., the zygomatic major and orbicularis oculi muscles do
not reach their apex at the same time)” (p. 114).
10 Smiling, laughter, and other facial expressions are commonly studied using the Facial
Action Coding System (FACS) elaborated by Ekman & Friesen (1978). Essentially, each
muscle in the face is examined for its degree of activation, on a 1–5 scale, ranging between
minimal activation to maximum activation.

40
 2.2 IDENTIFYING HUMOR

likely to smile than to laugh at a humor stimulus. Despite this fact, as

Ruch notes, laughter remains the response most strongly associated with
humor (p. 23). Of course, from the standpoint of identifying humor,
smiling presents two related problems: a) people smile (even felt smiles)
for positive reasons other than mirth, such as happiness or pleasure
(Frank & Ekman, 1993, p. 10), and b) one may experience mirth or
recognize humor and still not smile. Ruch (1995/2005) reports that in
a meta analysis of studies on reactions to humor, anywhere between
5% and 39% of subjects “showed no facial response at all” (1995/2005,
p. 131). Most of those who showed no response were introverts.

2.2.3 Laughter
There are many different kinds of laughter, acoustically and phono-
logically, in terms of syllables, aspiration, rhythm, volume, etc. (Chafe,
2007, pp. 25–40; Trouvain & Truong, 2017, pp. 344–345). Laughter may
occur on its own or interspersed in speech (Chafe, 2007, pp. 41–49).
Speakers may laugh alone or jointly (antiphonal laughter; Smoski &
Bachorowski, 2003). Generally speaking, laughter is associated with a
more intense response to humor (Ruch, 2008, p. 23); also “different
intensities of smiling reflect different degrees of exhilaration” (Ruch,
1995/2005, p. 110). Chafe (2007) notes that smiling has been described
as “the mildest form of laughter” (p. 52), for example by Darwin (1872),
but ends up questioning the idea of smiling and laughter lying on a
continuum because smiling has many social uses and is not limited to
the expression of mirth. Platt & Ruch (2014, p. 704) flatly deny that a
“graduate series” (i.e., a continuum) could be applied, if the emotions
are different.
Laughter can be spontaneous (irrepressible) or intentionally pro-
duced (voluntary; contrived); to put it differently, one can laugh with-
out intending to do so or deliberately: “[C]ontrived and spontaneous
laughter within a person are strikingly similar with respect to the
respirational pattern” (Ruch & Ekman, 2001, p. 428); therefore, it may be
difficult or impossible to differentiate between them on acoustic or other
observational grounds, even though they do activate different areas of
the brain. This would mean that unless one knows that the speaker
intended to laugh (for example, because one happens to be the speaker,
and thus has access through introspection to one’s intentions), one could
not tell whether the laughter was voluntary or not. This is not the case

41
 METHODOLOGICAL PRELIMINARIES

with smiling, where, as we saw, a spontaneous smile (Duchenne smile)

manifests different facial muscular patterns than a deliberate smile.
Much like smiling, therefore, laughter cannot be used uncritically to
identify humor because a) people experiencing mirth do not always
laugh, and b) people laugh for other reasons than mirth. As we have
seen, both smiling and laughter exceed the boundaries of mirth and
mirth is not necessarily expressed by laughter. More damningly for
the idea of identifying humor through the presence of laughter and/or
smiling, both smiles and laughter may be volitional and fully social or
contrieved. If either can be produced deliberately, for whatever com-
municative purpose, then the assumption that laughter, and secondarily
smiling, are effective markers of the presence of humor is seriously called
into question. Frank et al. (1993/2005) argue that there are observable
differences, at least as far as smiling goes, between genuine and con-
trived displays and that “these differences are observable and influence
subjective impressions” (p. 234). However, they acknowledge that while
their study shows that genuine, felt, “enjoyment” smiles “can have social
signal value,” i.e., be recognized by their audience as such, that does not
necessarily mean that speakers engaging in conversation will treat them
differently.
There exists, as we have hinted at, an imbalance between the signif-
icant amount of research dedicated to laughter and the much smaller
amount of research on smiling, at least within linguistics. In what
follows, we will pursue the discussion of the possibility of using laughter
to identify humor a little deeper. Most of the argument carries over to
smiling as well.

2.3 Mirthful vs. non-mirthful laughter

Indeed, the field of humor studies has long known that laughter is not
coextensive with humor (see Attardo 1994, pp. 11–13 for a review of
the literature). This is a fancy way of saying that we may have laughter
without humor (as in pretend laughter, laughter of embarrassment,
tickling, or laughing gas), humor without laughter (as one who reacts to
humor smiling or with a nod of understanding, saying “I heard that one
before,” or saying “that’s not funny!”), and clearly humor with laughter
(Ruch, 2008, p. 23; Chafe, 2007; Trouvain & Truong, 2017).

42
 2.3 MIRTHFUL VS. NON-MIRTHFUL LAUGHTER

Table 2.1 Relationships between humor and laughter

Humor No humor

Laughter Mirthful laughter Embarrassment laughter

No laughter Non-laugh response Serious exchange

In order to avoid confusion, we will introduce the term “mirthful

laughter” when discussion the differences between laughter related to
humor and laughter that is not related to humor. A good way to con-
ceptualize mirthful laughter is a Venn diagram, describing “mirthful
laughter” as the intersection of the “laughter” and the “humor” cate-
gories. (See Figure 2.1.)

Mirthful
laughter

Humor Laughter

Figure 2.1 Venn diagram of the intersection between humor and laughter.

There is extensive literature on the distinction between mirthful

laughter and other types of laughter. Giles and Oxford (1970) list seven
kinds of laughter: humorous, social, ignorance, anxiety, derision, apolo-
getic, and laughter as a reaction to tickling. Ziv (1979a, p. 16), Poyatos
(2002), Chafe (2007), and others have a more extensive list of different
types and functions, which could probably be pared down significantly,
but are still emblematic of the fact that “laughter” is not a uniform
monolithic construct.
Aubouin (1948) and Olbrechts-Tyteca (1974) had already pointed
out that one could not use reliably laughter as a one-to-one marker
of humor because “laughter largely exceeds humor” (Olbrechts-Tyteca,
1974, p. 14). Poyatos (2002) contains a long list of laughter types, most
of which are non-humorous. Chafe (2007, pp. 73–85) also has a list of
non-humorous causes of laughter. Glenn and Holt (2013) stated that
“Much laughter occurs without anything noticeably humorous nearby,
and much that people think of as humorous occurs without laughter”
(p. 2). Trouvain and Truong (2017, p. 340) also have a list of non-
humorous sources of laughter.

43
 METHODOLOGICAL PRELIMINARIES

While it is certainly true that spontaneous uproarious genuine laugh-

ter is probably the most overt indicator of humor, the claim that laughter
is the most common indicator of humor is in error: smiling is probably
the most frequent one, as we saw. More importantly, displays of mirth
(laughter and smiling) and humor have a complex relationship that
make the use of either or both as the sole criterion to determine the
presence of humor inappropriate. Unfortunately, at least some areas of
humor studies have yet to fully digest the falsity of the equation “humor
equals laughter and vice versa.”

2.3.1 On the lack of direct linear correlation between humor

and mirth displays
Despite the fact that genuine enjoyment smiling and laughter occur
frequently with humor, the two are not coextensive. In the previous
section we have looked at how some mirth may not occur with smiling
and/or laughter and how non-mirthful laughter exists. Now we turn to
a different argument. Essentially, we will consider the fact that mirthful
displays are mediated socially, which makes the possibility of a direct
correlation between mirth display and humor unlikely.
Consider now a hypothetical simple direct correlation model, which
assumes a direct linear11 correlation between intensity of the reaction
(laughter/smiling) and humor stimulus appreciation. Basically, the more
a speaker enjoyed a humorous stimulus, the more intensely they would
react (with a mirth display). Unfortunately for this direct correlation
hypothesis, psychological and sociological studies show that the rela-
tionship is not direct, but rather mediated culturally.
The difference between humor (stimulus) and laughter (response) is
a well-known fact in psychology and in the social sciences, thus for
example, it is an established result that subjects laugh more at a joke
when they are in a group than if they are alone (Leventhal & Cupchik,
1976; Chapman, 1976; Malpass & Fitzpatrick, 1959;12 Young & Frye,
1966; Butcher & Wissell, 1984; Fridlund, 1991; Pollio & Swanson 1995;
Devereux & Ginsburg 2001). Even the size of the audience (Morrison,
1940) and the crowdedness of the room (Stokols et al., 1973) may affect

11 A linear correlation is one in which each increase or decrease in one value is met by a
correspondingly large change in the other value.
12 But not for cartoons, where Malpass & Fitzpatrick find that the opposite is true.

44
 2.3 MIRTHFUL VS. NON-MIRTHFUL LAUGHTER

the intensity of laughter or its enjoyment (Aiello et al., 1983). On the

effect of the audience on reactions to humor, see Section 10.1.1.
The fact that socialization enhances humor appreciation holds true
for children (Chapman & Chapman, 1974, Chapman & Wright, 1976),
and regardless of the actual presence of others (i.e., even an imagined
group increases laughter). People also laugh more if the other people
are friends (Murphy & Pollio, 1975; Pollio & Swanson, 1995), regardless
of age (Foot, Chapman, and Smith, 1977). Women smile/laugh much
more in reaction to cartoons and caricatures, in the presence of friends
than strangers (Wagner & Smith, 1991). Mixed gender groups laugh
significantly less when they are made up of strangers (Pollio & Swanson,
1995).
Similarly, it is a well established fact that accompanying laughter
(laugh track) will increase laughter (e.g., Martin and Gray, 1996; Lawson
et al., 1998; Platow et al., 2005) and that being observed (videotaped) will
decrease it (Dale et al., 1991; Martin, Sadler, Barrett, & Beaven, 2008).
The size of the audience is also a relevant factor: generally speaking, the
larger the audience, the greater the laughter (Morrison, 1940; Andrus,
1946; Prerost, 1977; Aiello et al., 1983). For more discussion of the
audience effects on laughter, see Chapter 10.
Thus, even if we wanted to establish a correlation between humor and
laughter, doing so would prove practically impossible (or at least very
hard). For example, we would not know whether we should measure the
laughter of a given subject alone, or in a group, with strangers or friends,
with or without a laugh track, etc.
These are not the only problems one would encounter. If we present
several stimuli in an experimental environment, or if in a conversation
several humorous turns occur in a row, the repeated presence of humor
affects its perception. Gavanski (1986) notes that “cartoon repetition
did reduce amusement” (p. 211), whereas “ratings of cartoons on a
‘funniness’ scale (…) were totally insensitive to reduction in amuse-
ment” (p. 212). The same result was found in Deckers et al. (1989). On
repetition see also Forabosco (1994) on seriality.
Even more damaging for our hopes of a linear correlation between
humor and laughter, is Brown et al.’s (1982) finding that when presented
with unfunny stimuli but in the presence of confederates who have
been instructed to laugh, subjects will laugh. Fridlund (1991) shows
that subjects who merely imagined a companion smiled more whereas
they did not report increased emotional response. He comes to the

45
 METHODOLOGICAL PRELIMINARIES

conclusion that smiling may be more related to the social context than
to the emotional setting. These results would of course make the use of
laughter or smiling as indicators of humor extremely problematic.
Finally, there is a significant literature, within the field of discourse
analysis, which has shown that there are numerous situations in which
laughter occurs without any mirth or exhilaration. These include embar-
rassment at the discussion of sensitive topics, displays of resiliency
in trouble-talk (i.e., non-humorous descriptions of one’s problems),
complaints, and as a coping mechanism in aphasia. The literature is
discussed in some detail in Section 11.4.2. The other contribution of
discourse analysis to the discussion of whether laughter can be used
to identify humor comes from another strong result in the field, which
goes back to Jefferson (1979), i.e., the use of laughter to prime a laughter
response (and hence the first use of laughter is not a response to the
humorous stimulus). More broadly speaking, discourse analysis puts to
rest the idea that laughter is a univocal reaction to humor: laughter may
or may not be a reaction to humor and laughter may or may not be an
expression of, or even associated with, mirth.
So, in conclusion, it is not methodologically sound to claim that some-
thing is humorous just because someone laughs (or smiles) immediately
before, during, of after. This is not to say that one cannot or should not
study laughter. On the contrary, as we will see, one of the central results
of the discourse analysis of humor has been precisely that speakers signal
by initiating laughter the humorous nature of their utterance. However,
one should not confuse the study of laughter with the study of humor,
as some have.

2.3.2 Recognition vs. appreciation

What mechanisms allow us to decouple the perception of a humorous
stimulus from having a mirthful reaction? One of the reasons is that
the recognition of the humorous nature of a text is independent of its
appreciation. To recognize the humorous nature of a text requires only
the activation of some cognitive mechanisms (which will be examined
in Chapter 6 and following) that can be roughly described as the
recognition of an incongruity. The appreciation of humor, whatever its
nature, logically requires the recognition of humor, plus a set of other
mechanisms (since one cannot appreciate what one does not know is
there!).

46
 2.3 MIRTHFUL VS. NON-MIRTHFUL LAUGHTER

Hay (2000) has introduced a four-level hierarchy to describe the

attitudes of the hearer when faced with a humorous remark or text. Hay’s
taxonomy was elaborated within the context of face-to-face conversa-
tional joking, but it can be expanded intuitively to all texts.

1. Recognition: at this level, the hearer merely recognizes the inten-

tion on the speaker’s part to produce a humorous text without
necessarily understanding the humor of the text.
2. Understanding: at this level, the hearer recognizes the intention of
the speaker and understands the humorous import of the text.
3. Appreciation: at this level, the hearer manifests in some way his/her
enjoyment or displeasure with the contents of the joke and/or the
fact that the joke has been uttered.
4. Participation: at this level, the hearer actually engages with the
speaker in the creation of further humor.

Hay’s most significant point, however, is that the latter levels presuppose
(in the logical sense) the former. What this means is that in order to
react to a joke, the hearer must first realize that the joke has occurred
and then understand what the joke is. This means that there are several
“failure” points from the hearer’s point of view since he/she may a) fail to
recognize that the joke is being delivered; b) fail to understand the joke;
c) fail to react to the joke; and d) fail to participate in the joking situation.
We return in more detail to Hay’s hierarchy in Section 11.3.1.2.
The distinction between recognition and appreciation is well estab-
lished in psychology: for example, Leventhal & Mace (1970), Leventhal
& Cupchik (1976), and Gavanski (1986) distinguish between cognitive
and affective responses to humor: “the cognitive component is the
evaluation of the humor stimuli in terms of their perceived humor
content; the affective component is the subjective feeling of enjoyment
produced by the humor stimuli” (Gavanski, 1986, p. 209).
There is even neurological evidence for the distinction between recog-
nition and appreciation (Goel & Dolan, 2001; Moran et al., 2004). In
Goel and Dolan’s work, “semantic comprehension of jokes was associ-
ated with increased activation in the left and right posterior middle tem-
poral gyrus. By contrast, activity in ventromedial prefrontal cortex cor-
related with subjects’ explicit ratings of how funny they found each joke”
(Moran et al., 2004, p. 1055). In the case of right hemisphere damage

47
 METHODOLOGICAL PRELIMINARIES

patients, Shammi and Stuss (1999) report a dissociation between the

cognitive and affective responses to the humorous stimuli whereby the
patients appear to understand the joke, and they rate the joke as funny,
but they do not respond with smiles or laughter, unlike their normal
controls and patients with lesions in other locations.
This observation is a central foundation in the establishment of a
pragmatics of humor in the sense of a theory of the audience. When
observing an audience’s reaction to a given joke we, as observers, are
faced with a problem. Given a joke utterance, the audience’s reaction
may be a) laughing or b) not laughing (let us ignore, for the time being,
intermediate reactions such as snickering, smiling, etc.). If the audience
laughs we do not know whether they are laughing because they a)
recognize but do not understand the joke, b) recognize and understand
the joke. If the audience does not laugh, we do not know whether they
a) did not recognize the joke, b) did not understand the joke, or c) did
not believe that it was appropriate to show appreciation of the joke or
otherwise decided to suppress observable reactions. In other words, our
audience may be laughing not because of the humor (false positive), or
may be not laughing while fully cognizant of the fact that a humorous
event just took place (false negative). False negatives are common in
humor research: the “that’s not funny” retort clearly belies its own force
since it implies that the speaker knows that the event was funny but that
he/she believes that appreciation should be withheld.
As can be seen, if one wants to investigate humor appreciation or
in general wants to construct a theory of humor audience/reception,
one must address these issues that would otherwise nullify any effort
at such an endeavor. The only examples of such controlled studies
come from the psychological disciplines where to a certain degree, joke
appreciation is measured not through self-reports but via physiological
correlates (e.g., heart rate, vasco-dilation, electromagnetic activity in the
cortex). Needless to say, even these empirical measurements are not
100% effective as the “truth machine” (polygraph) has demonstrated.
When considering self-reports, humor researchers should be aware of
the fact that these are necessarily filtered via one’s societal and emotional
biases, i.e., that subjects will tend, for example, to underreport amuse-
ment at offensive or otherwise problematic stimuli. Another related
issue is the presence of other subjects within the perception of the
individual subject. It is a common phenomenon to laugh at a joke

48
 2.3 MIRTHFUL VS. NON-MIRTHFUL LAUGHTER

that one has not understood simply because those around him/her are
laughing at the joke. This makes the study of group appreciation of
humor dauntingly complex as there is virtually no way of having access
to the relevant information. Recall that studies have shown that subjects
laugh more when they are in a group and that they will laugh to a non-
funny stimulus if the rest of the group laughs at it (this is investigated
experimentally using compères who are in on the experiment (Brown
et al., 1982)).
In short, neither direct behavioral observation nor self-reports are
self-evident clues of either understanding or appreciation of humor.

2.3.3 Confusion between humor and humor appreciation

A common confusion, which also needs to be dispelled, is that between
humor and humor appreciation. It has been claimed that a text is not
funny unless it is funny to someone. While this has a certain ring of
truth, further examination reveals this position to be in error. Specifi-
cally, it confuses humor competence and humor performance. Consider
the following situation:
A tells B a joke. B is upset, or hungry, or in pain and does not want
to hear the joke. Therefore B does not laugh and in fact says to A. “It’s
not funny.” However, a few minutes later, after B’s emotional problems,
hunger or pain have been taken care of, B remembers the joke and says
“That was funny.”
Would we want to say that the text was not funny at time t0 and was
then funny at time t1 ? It seems much more natural to say that the joke
was funny all along and that B was not in the right state of mind to
appreciate it. Indeed, since the text has not changed, what could possibly
have changed, if not B’s condition?
One could argue that a text only exists in context and therefore that
only pairs text/context can be funny. But this argument does not carry
through, because once more the text can be an invariant, whereas the
situations will inevitably change and the reason why certain texts will
tend to be identified as funny or provoke a certain kind of reaction would
remain. In conclusion, humor is a property of the stimulus, the text
(competence level). Humor appreciation is a property of the situation
(which includes a specific speaker and hearer, the context in which the
humor is produced, etc., i.e., is the performance level).

49
 METHODOLOGICAL PRELIMINARIES

2.4 Keying

A stimulus may have all the necessary and sufficient conditions for the
text to be classified as humor, but still not be perceived as humor, because
of a performance issue. These conditions are essentially a special type of
incongruity, as we saw before and will discuss more fully in the next
chapters. The question then is not “why do certain incongruities not
trigger humor experience?” but rather “what performance factors are
necessary for a text that is humorous at the competence level to be
perceived as humorous in a given situation, by a given individual or
group, etc., i.e., in performance?” We know that some incongruities may
be perceived as threatening or scary. For example, cognitive dissonance
(Festinger, 1957; see Harmon-Jones, 1999; Cooper, 2007 for a synthesis
of current research) can be described as the perception of an incongruity
that is unpleasant. Cognitive dissonance is defined as follows:

the possession at one and the same time of cognitive elements (knowledges) having
psychologically opposite implications generates an unpleasant state of tension, or
dissonance, within the individual which then motivates him[/her] to attempt to reduce
dissonance by altering his[/her] cognitions. (Berkowitz, 1969, p. 97)

Relatedly, Berlyne (1960) suggested that there was an inverted-U shaped

curve describing the relationship between complexity of the stimulus
and humor response (or to put it differently, that very hard or very easy
incongruities are not funny). Hence, incongruities that are threatening
or too complex to process will not be perceived as humorous. These are,
of course, only two of numerous factors that may inhibit the humorous
appreciation of the incongruity.
We can then safely say that, all other things being equal, a humorous
text will be perceived as humorous if the incongruity is

• non-threatening
• not too complex or too simple
• based on available scripts/knowledge
• unexpected, surprising
• occurring in playful mode (see Section 2.4.1)
• sudden (see Section 2.4.2)

The above list is indicative and not meant to be exhaustive and, in

fact, it is not possible to address the entire list in this context. Some

50
 2.4 KEYING

of the issues mentioned have a significant literature (e.g., cognitive

dissonance). I will examine briefly the playful and surprising aspects
which have received relatively little attention.

2.4.1 Playful mode

We saw that for humor to be successful the situation must be framed
or keyed as humor, in the sense of Goffman’s “frames” (Goffman, 1974,
pp. 43–44) and Hymes’ “keys” (Hymes, 1972, p. 62); a discussion of
these concepts, within the sociolinguistic framework, will be found in
Chapter 10. Particularly, it should reflect suspension of disbelief on the
factuality of information (fictionality). We will deal more fully with
fictionality and play in Section 3.3.1.
A possible argument against the essential nature of incongruity is that
an incongruity may not be perceived as funny if the situation is not keyed
to humor. This is true, but it turns out not to be an argument against
the incongruity–resolution theory of humor; see Chapters 3–4. First of
all, let us note that keying is a dynamic phenomenon: a situation may
start out keyed for humor or may evolve or change its keying to humor.
This shows that keying and humor are not causally connected, since
neither can exist without the other: we can have un-keyed humor, i.e.,
humor that occurs without warning or signaling in an otherwise serious
situation (deadpan humor); similarly, we can have serious content keyed
to playfulness, for example, the Schoolhouse Rock videos in the US
presented serious academic material using songs and playful cartoons.
Naturally, when humor is recognized its presence keys a posteriori any
situation as “funny” in the intentions of the speaker who produced the
humor (the audience may still refuse the keying, for example using the
formula “That’s not funny”).
Summing up: if a situation is not keyed for humor, i.e., playful,
cheerful, upbeat mood, the occurrence of humor may rekey the situation
toward humor or the attempt at humor may fail. Conversely, if a situation
is keyed for humor, for example, by the prior occurrence of humor,
humor may occur or not. If the situation is keyed for humor and
humor occurs, the likelihood that both speaker and hearer will assess
the situation as ‘something humorous was uttered’ is highest. If the
situation is not keyed for humor, but humor occurs, they may or may
not assess the utterance as humorous. If the situation is keyed for humor
but no humor occurs (for example, the speaker fails to deliver a properly

51
 METHODOLOGICAL PRELIMINARIES

Table 2.2 The interplay of humorous keying and text

Humorous keying Serious keying

Humorous text High likelihood Medium-high likelihood

of humor assessment of humor assessment
Serious text Medium-low likelihood Low likelihood
of humor assessment of humor assessment

formed humorous text) they, once more, may not assess the situation
as humorous. If the situation is not keyed for humor and no humor
occurs, the likelihood of speaker and hearer assessing the situation as
humorous is very low (but not impossible). The situation is summarized
in Table 2.2. Note the difference in likelihood of humorous assessment
in the two cases of mismatched keying/text, which follows from my
assumption that the nature of the text takes precedence on the key.
It should also be noted that humor shares the non-serious keying with
fiction, where the distinction is usually conceptualized as “fiction” vs.
“factuality.” This is not the place to go into the details, but see Attardo
(2008a, 2009).
Finally, it should also be noted that keying of humor is a woefully
under-studied issue: there are situations that are keyed by default to
humor (comedy routines, late night talk shows, friendly banter, etc.)
and there are others that are not (war declaration, judicial senten-
cing, discussion of divorce among spouses, etc.). For example, how
the type of situation and the production of humor interact has not
really been addressed, nor has the fact that the use of humor may
influence the way the situation is keyed (or re-keyed). From a different
psychological perspective, the issue can be framed as the need for the
speakers to be in the right frame of mind, called the humor mindset
(Ford, 2014). The serious normal mindset is called “reality assimilation”
whereas humor is part of “fantasy assimilation” (McGhee, 1972). Ford
explains that

Reality assimilation is the default process that occurs when encountering discrepancies
between our cognitive schemas and actual events. In contrast, when in the fantasy
assimilation mode, people do not require a realistic resolution of incongruous events.
Therefore, they do not attempt to adjust their cognitive schemas to fit unexpected
events. They simply disregard the requirement of literal congruity that characterizes
reality assimilation. (2014, p. 361)

52
 2.4 KEYING

2.4.2 Surprise
The two aspects of the incongruity should be perceived simultane-
ously, i.e., should be active at the same time, cognitively speaking. This
seems related to another factor in humor perception: the surprising or
unexpected aspect of the humorous stimulus. Punch lines are usually
unexpected. Humor tends to be surprising. Conversely, for example,
Catholic theologians commonly discuss the incongruous idea that God
is one and three persons, at the same time, at least since the third
century. While the conditions of humor (incongruity and co-presence,
as we will see in Chapter 6) are clearly fulfilled, which may lead us to
expect that the idea may be funny, the situation is hardly surprising and
definitely not keyed for humor. A related example comes from physics
which sees light as both a wave and a particle. The two descriptions are
incompatible and so the description is incongruous, but discussions of
optics are not usually keyed for humor, nor are they surprising (i.e., the
information is presented upfront).
Finally, not all unpredictable and surprising things are funny. For
example, in mysteries, the identity of the killer is unexpected, and if the
author did a good job, unpredictable, but that a killer would be found
is expected. Generally, mysteries are not keyed for humor (with some
exceptions, Janet Evanovitch and Pierre Dard come to mind) but even
when they are, usually the revelation of the identity of the culprit is
not sudden. So, in part some unpredictability is predicted and in part
the delivery of the unpredictable element is not sudden enough to be
surprising. There are more factors (having to do with the resolution of
the incongruity) which will be taken up in Chapter 4.

2.4.3 Unintentional humor

Finally we need to mention briefly the phenomenon of unintentional
humor, which shows that the keying for humor is crucial in recog-
nizing the presence of potential humor. The idea of voluntary and
involuntary humor (e.g., Ziv, 1984, p. 83) is straightforward: either one
intends to amuse someone else or one does not. Nonetheless, the idea
of “unintentional humor” is a good example of possible confusion. In
the prototypical situation, the speaker says something he/she intends
to be perceived as humorous. But what if the speaker says something
that he/she did not intend to be perceived as humorous and yet has

53
 METHODOLOGICAL PRELIMINARIES

all the prerequisite semantic/pragmatic trappings of competence-level

humor (i.e., is potentially humorous)? In that case the hearer is faced
with a choice: he/she can realize that the intention of the speaker was
not to produce humor and therefore ignore the potential humor or
he/she can choose to treat the non-humorously intended utterances
as if it had been intended for humor. Needless to say, this choice is
not a conscious one (at least not always). This is why I have claimed
that there is no such thing as unintentional humor, since either the
speaker or the hearer has to intend the interpretation of what was said
as meant humorously (in other words, someone is always intending the
potential humor as humorous, if they react to it as such). But what if
neither the speakers nor the hearer realize that the speaker’s utterance
is potentially humorous? Then, while the potential humor is objectively
there, there is no humor perception. Needless to say, the possibility of
people laughing without anything to laugh at is also there (as we saw,
in the experiments described above, in which laughter increases if other
people are laughing). A fuller discussion of the concept of unintentional
humor will be found in Section 8.4.

2.5 Identifying humor: the triangulation approach

This brings up again the issue of how a researcher goes about identifying
humor in a text. The answer is that it depends largely on the type of
text. Some texts, such as novels or written jokes, have no extra-textual
markers, such as those that, for example, sitcoms have (e.g., the laugh
track).13 In that case, only semantic/pragmatic analysis can be used.
If there are markers of humor (such as laugh tracks, or real laughter,
or smiling, etc.), then one can use a triangulation approach, that is
use of several cues to the humorous nature of the intention, such as
the semantic analysis to identify the humor, checking the locations
where the conversation participants laugh and/or smile. In some cases,
for example, some corpora such as transcriptions of legal discourse,
markers such as “[laughter]” are introduced. In the case of conversa-
tional data, when access to the participants is possible, stimulated recall
interviews after the data collection itself, revisiting the data, may help.
13 They may have paratextual cues, such as a title along the lines of “5000 jokes” of a
subtitle “a comedy in 3 acts.” Other paratextual cues can be inclusion in a series of humorous
books, or a cover depicting a man slipping on a banana peel, etc.

54
 2.5 IDENTIFYING HUMOR: THE TRIANGULATION APPROACH

In the case of multiple participants, consensus of interpretation may also

be useful. There exist sophisticated statistical techniques to assess the
degree of agreement of judges or raters (inter-rater reliability).
As we saw previously, triangulation is not limited to three cues to
the humorous nature. If more clues are available (say, interviews with
the speakers, facial expressions, judges ratings, etc.), then more than
three points of triangulation may be used. One may object to the use
of the term triangulation, since we admit more than three kinds of
evidence, but the word “n-sided polygonization” does not exactly roll
off the tongue. Triangulation is very useful, as it allows to double check
one’s analyses and thus to reduce the possibility of biases. Researchers
should be extremely careful not to fall for the mistake of using laugh
tracks or non-theoretically grounded transcriptions (i.e., transcriptions
that are not made for research purposes) as the sole determinant of what
to analyze. For example, the choice of looking at only those instances of
humor in a sitcom at which laugh tracks occurs exposes one to the risk of
both false positives and false negatives: a false positive would be a laugh
track where no humor is actually found, and a false negative would be
an instance of humor for which no laugh track was provided. This is
closely related to the issue of “failed humor,” i.e., humor that fails to be
interpreted as such by the audience. By definition, failed humor is not
usually followed by laughter or smiling. Thus using the occurrence of
laughter and/or smiling as the sole criterion for deciding if a piece of
text is humor or not, will inevitably lead to false negatives (i.e., missing
all failed humor). On failed humor, see Section 12.4.3.
Finally, one could ask, why not rely on the researcher’s intuition?
After all, much like linguists in the 1960s and 1970s relied on native
speaker intuition to determine the grammaticality of a sentence, couldn’t
we rely on the speaker’s intuition to identify humor? The answer to
that is that people are appallingly bad at it and the worst performers
are almost entirely unaware of how bad their performance is (Kruger
& Dunning, 1999, pp. 1123–1124). So using an untrained judge is a
recipe for disaster. Using a battery of judges and testing inter-rater
reliability solve this problem and bring us in the domain of experimental
psychology or the social sciences.
So, generally speaking, a triangulation method to the identification
of humor is the best way to proceed. By cross-referencing any sources of
information about the potential humorous nature of the text in question,
the researcher increases the reliability of their study by avoiding both

55
 METHODOLOGICAL PRELIMINARIES

false positives (i.e., finding humor where there is none) and false neg-
atives (missing humor that is there). The sources should not be ranked
equally: objective information should be considered more reliable than
subjective considerations. Multiple judges are more reliable than a sin-
gle, potentially biased, judge.
Within theoretical linguistics, studies have largely relied on meta-
textual information (for example, if the text you are looking at comes
from a published collection of jokes, the likelihood of it being a joke
is extremely high; if moreover it contains a punch line, you can safely
assume it is one) and historical records (for example, if you are consid-
ering Shakespeare’s Merry Wives of Windsor, the historical record tells
us that it is a comedy). Other fields, such as discourse analysis, have
used other methodologies, such as participant observers, which will be
discussed in the relevant chapters. For a more formal discussion of the
triangulation approach, see Section 12.4.2.

2.6 Further readings

On the differences between humor and laughter, see a review and

references in Attardo (1994, pp. 10–13). On mirth, see Martin (2007);
on exhilaration, see Ruch (1993). Chafe (2007) and Trouvain & Truong
(2017) are excellent descriptions of laughter and its phonetics. On
smiling, see Platt & Ruch (2014).
On the application of the principle of commutation to humor studies,
see Attardo & Raskin (2017). On warrants in applied linguistics, see
Edge & Richards (1998). Any introductory textbook of statistics will
deal with sampling and validity issues. On the implicatural hierarchy
of understanding, appreciation, etc., see Hay (2000). The discussion of
keying is developed more broadly in Attardo (2009) and in Chapter 10.

56
 3
Theories of humor and their
levels

3.1 Explanation, reductionism, essentialism 57

3.2 The three major theories of humor 59
3.3 Humor as release 60
3.4 Humor as incongruity 64
3.5 Humor as aggression 64
3.6 Other theories 67
3.7 Complementarity of the theories of humor 71
3.8 Further readings 77

This chapter begins by discussing briefly the concept of reductionism

and more generally of “explanation.” It also very briefly argues in favor
of reductionist/essentialist theories of humor, before launching a survey
of humor theories from different areas of research which range broadly
from the neurological level, taken as the point of higher determinism (or
conversely as the lowest possible reduction), to the macro-sociological
theories that see humor as a societal phenomenon. A final section deals
with the evaluation of theories of humor.

3.1 Explanation, reductionism, essentialism

What does it mean to present the explanation of a phenomenon? Gener-

ally, it means that given a phenomenon, such as humor, we can describe
its causes, the contexts in which it takes place, the effects it has, etc.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 THEORIES OF HUMOR AND THEIR LEVELS

It also means that we “understand” the phenomenon, i.e., we know what

causes it, and often that we can replicate it. Generally, we want to explain
a phenomenon in terms of simpler causes/phenomena. This stance is
called reductionism. For example, explaining rain as the interplay of
water vapor, condensation, and gravity would be a typical reductionist
explanation. An essentialist explanation is a reductionist explanation
which claims that there is one or more core features that “makes the
phenomenon what it is” i.e., is its essence. The essence of a thing
(thing should be understood in a very general sense here, as “entity”)
is what makes it be such, for example, the essence of a fork is that it
has “teeth” (tines) and a handle. Among the non-essential (accidental)
features of forks are that they are made of metal (since there can be
plastic or wooden forks) but also the number of tines, which may range
from two to four, or more.1 The opposition of essence and accidence
goes back to Aristotle. Theoretical linguistics is mostly reductionist and
essentialist as it seeks to reduce linguistic behavior to a small set of rules.
Seeking the necessary and sufficient features that cause a phenomenon
is a characteristic essentialist stance. Other approaches to linguistics
are less interested in finding the essence of the phenomena but lean
toward functionalism, i.e., what do speakers “do” with the phenomena at
hand, what is their function in discourse? Yet other approaches, such as
cognitive linguistics, espouse a prototypical approach to categorization,
which means that there may not be a single individual essence of a
phenomenon, but rather the phenomena may be classified in “families”
of concepts. For the time being, we need not concern ourselves with
these philosophical subtleties (those interested can start with Scholz
et al., 2016). We will instead turn to the three major theories of humor
and examine them from a reductionist and essentialist standpoint.2
1 It’s actually slightly more complex: the number of tines on a fork is a range from 2 to 5
or 6 maximum. A fork with only one tine is a spear and one with 40 or 50 is a comb.
2 Here a caveat is necessary: in many circles, “essentialist” is basically an insult. This is
because in some fields essentialism is associated, for example, with biological determinism.
For example, in gender studies, this would be the idea that the presence of a penis or a
vagina would be the essence of masculinity/femininity. In cultural studies, essentialism
is associated with the idea that an entire culture (say, Western culture) has an essence.
Needless to say, seeking essentialist reductions of phenomena does not imply embracing
rigid determinism nor that all phenomena must have an essence at a level of generalization
that makes it interesting/useful. For example, in trying to find the essence of “game” or
“food” one will most likely end up with very high level generalizations, such as “an activity
that one engages in for pleasure” or “any substance ingested for sustenance” which will not
be helpful if your goal is creating a new game or you are a nutritionist. Likewise a social
constructivist perspective is not incompatible with an essentialist view. No-one said that the

58
 3.2 THE THREE MAJOR THEORIES OF HUMOR

3.2 The three major theories of humor

There are very many theories of humor. Piddington (1933) listed 49

authors; Bergler (1956) listed 80; Smuts (n.d. 2006) claims that there
are more than 100. One of the best reviews is Keith-Spiegel (1972) who
distinguished 8 types of theories:

1. Biological, Instinct, and Evolution

2. Superiority
3. Incongruity
4. Surprise
5. Ambivalence
6. Release and Relief
7. Configurational
8. Psychoanalytic

“New” theories emerge all the time, as a quick glance at the internet will
reveal: a Google search of “theories of humor” returned a staggering
17.5 million hits. Amazon.com lists 8000 books under the “theory of
humor.”3 However, most “new” theories often turn out to be restate-
ments in a different terminology of old ideas, or just a different mix of
known factors.
To make sense of this abundance of data, the theories of humor are
commonly classified into three groups:

• incongruity theories (a.k.a. contrast) (Raskin, 1985, pp. 31–36),

• hostility theories/disparagement (a.k.a. aggression, superiority, tri-
umph, derision) (pp. 36–38), and
• release theories (a.k.a. sublimation, liberation) (pp. 38–40).

(See Table 3.1, adapted from Attardo (1994, p. 47); after Raskin (1985),
Ziv (1979a) and Monro (1951); for details about the theories, see the
text.)
essence of a phenomenon may not be socially constructed. The novice may be well advised
to avoid actively advertising his/her essentialist stance, while quietly pursuing reductionist
explanations which is, as far as I can tell, what most scholars do, including most of those
who denounce essentialism.
3 Granted that a quick glance at the results reveals that Amazon has a lot of work to do
about filtering its results (what do calendars and novels have to do with theories of humor?);
nonetheless, the point stands: that’s a lot of books.

59
 THEORIES OF HUMOR AND THEIR LEVELS

Table 3.1 Attardo’s three families of theories chart

Cognitive Social Psychological

Incongruity Hostility Release

Contrast Aggression Sublimation
Saliency Superiority Liberation
Bisociation Triumph Economy
Catastrophe Derision Benign violation
Disparagement
Disposition

3.3 Humor as release

Release theories, as the name implies, are based on the idea that humor
“releases” the “pressure” of mental energy. The idea of “energy” flowing
in the mind is a metaphor, introduced by Spencer in his “The Physiology
of Laughter” essay (1860). The best known proponent of a release theory
is Freud (1905). Martin (2007, p. 58) characterizes release theories as the
“hydraulic theory” of the nervous system. The theory was based on the
steam engine, then the prominent new technology. This metaphor was
replaced by the “mind as computer” newer metaphor (see for example
Hurley et al., 2011). In other words, there seems to be a tendency for
people to compare humor to the most current and salient technology
at the time. The problem is that metaphorical theories may be useful,
insofar as they allow us to map some properties of a phenomenon,
such as humor, onto a better understood mechanism, but up to a point.
A metaphor is only a metaphor and all analogies eventually break
down. Spencer’s and Freud’s metaphorical hydraulic theory is nowadays
discredited: no-one believes that there is actual mental pressure that
needs relieving.
Nonetheless, Freud’s theory deserves special mention because he paid
a lot of attention to the linguistic mechanisms of humor: the first part
of his book on humor is dedicated to these mechanisms. However,
subsequent research has shown that none of the mechanisms located by
Freud were unique to humor, but that in fact he had rediscovered some
of the mechanisms present in any linguistic form (Attardo, 1994, p. 55).
In terms of linguistic behavior, the significance of release theories
comes from the fact that they account for the so-called “liberation”

60
 3.3 HUMOR AS RELEASE

from the rules of language, for example in puns and word-play more
generally. The metalinguistic nature of puns and word-play account for
this liberation, to the extent that it is a fact (one could easily argue that no
real freedom from the rules of language is found in puns, for example).
Release theories can also be seen as accounting for the violations
of the Principle of Cooperation (Grice, 1975; 1989) in humor (see
Chapter 8). Essentially, the idea is that one is freed of the constraints of
having to follow the Priciple of Cooperation. This is closely related
to the “defunctionalization” of language in humor (Guiraud, 1979,
pp. 111–119; see Chapter 12). Contemporary researchers such as Fry
(1963) and Mindess (1971) have also proposed release theories.

3.3.1 Humor as play

Play can also be seen as a release from the goal-oriented nature of “seri-
ous” life.4 A playful mindset is a significant component of the humorous
experience. Bateson (1955/1972) has shown that playful, and humorous,
behavior carries (either explicitly or implicitly) a metamessage roughly
paraphrased as “this is play.”
Bateson starts from the definition of metalanguage in Russell
(see Section 1.3.2) and introduces the idea of meta-communicative
language, i.e., discourse in which the subject is “the relationship
between the speakers” (1972, p. 67). While observing monkeys playing
at a zoo, Bateson realized that the very idea of play “could only
occur if the participant organisms were capable of some degree of
metacommunication, i.e., of exchanging signals which would carry the
message ‘this is play” ’ (p. 68).
Bateson notes that the metamessage “this is play” has the logical form
of a paradox: “These actions, in which we now engage, do not denote
what would be denoted by those actions which these actions denote.
The playful nip [of the playing monkeys] denotes the bite, but it does
not denote what would be denoted by the bite” (p. 69). Finally, Bateson
observes that since much play is fictional, not only is the metamessage
“this is play” paradoxical, in its concurrent claim of being something and
not-being something, but it is also about non-existent entities (p. 70).
From our perspective, the important aspect of Bateson’s contribu-
tion is the “this is play” metamessage. One the one hand, in itself,
4 One can classify play theories of humor as a separate category of humor theories, as
Smuts (n.d.) does.

61
 THEORIES OF HUMOR AND THEIR LEVELS

it becomes the foundation of the idea of keying/framing humor (see

Section 10.2.2.2) and of markers of humor (contextualization cues; see
Section 10.2.4); on the other hand, it is the source of the defunctionaliza-
tion aspect of humor, which plays a major role in the functional analysis
of humor in discourse, see Section 12.1.4.
Fry (1963) was directly influenced by Bateson, with whom he worked.
Fry builds a complete model of humor based on the notion of para-
dox, explicitly starting from the self-referential paradoxes identified
by Russell and in particular Epimenides’ paradox (see Section 1.3.2).
According to Fry, the metamessage “this is play” or “this is not real” is
paradoxical, because if I broadcast this metalinguistic framing message
then the metalinguistic framing is itself part of the non-reality, which
triggers the paradoxes identified by Russell and Bateson. Fry uses the
terminology “play frame” (p. 138) and mentions as “things people do to
indicate fantasy” (p. 138), such as voice quality, body movements, pos-
ture, lifting of the eyebrows (p. 138) and more broadly “non-verbal com-
munication devices” (p. 139). However, Fry goes on to note, humor is not
just paradoxical. Humor has a punch line, in the technical sense used in
humor studies, i.e., a climax of the narrative which reveals the heretofore
implicit incongruous element. Fry argues that the punch line brings in
another paradox, this time not at the level of the overall communica-
tive situation (the “this is play” paradox), but an “internal” (p. 153)
paradox “which [ . . . ] specifically concerns the humor content rather
than the total situation” (p. 149; emphasis in the original). He likens
this to the figure-ground reversal introduced by Gestalt psychology
(pp. 154–155).5
Among the markers or indicators of the play frame, is the so-called
“play-face” in primates (see Martin 2007, pp. 3–4, 128, 165–166, for
discussion and an image), which as expected for a marker of play, does
not occur in aggressive situations:

Ethological studies of the silent bared-teeth display and the relaxed open-mouth (play
face) display in apes, which are viewed as primate homologues of human smiling and
laughter, respectively, reveal that these facial displays occur exclusively in the context
of friendly social and play activities, and not in the context of aggression.
(Martin, 2007, p. 54)

5 The connection with the logical mechanisms of humor (see Section 7.1.5) was already
noted by Forabosco in his commentary incorporated in the Attardo & Raskin article (1991,
p. 337).

62
 3.3 HUMOR AS RELEASE

Apter’s “para-telic” (non-goal-oriented; from the Greek “telos” goal)

mode is also a variety of humor-as-play theory (1982, 1989). Reversal
theory, as Apter’s theory is called, posits that an incongruity will be
experienced as humorous in a safe, non-threatening, playful environ-
ment. Further, reversal theory claims that the target of the incongruity
must be “diminished” (i.e., reversal theory is a variant of the superiority
theories, in this respect). If the target is enhanced the incongruity would
be perceived as something else, such as art.
3.3.1.1 Willing suspension of disbelief
As part of the playful nature of humor, we should consider another
aspect of playfulness, and namely the willing suspension of disbelief in
which the hearers must engage in order to appreciate the text. To put
it differently, if the hearer does not play along, he/she locks him/herself
out of the fun. In fiction, hearers must routinely suspend disbelief about
the fictional nature of what they are being told. If one were to argue that
since there was no Sherlock Holmes, Conan Doyle was just a liar, and
one was not willing to waste one’s time listening to falsehood, then one
is unlikely to experience pleasure reading about Holmes’ adventures.
Perlmutter (2002) argues that in jokes “the listener’s sacrifice is more
than merely the ordinary suspension of disbelief; he or she is pre-
pared to suspend a wide variety of every day prerogatives.” Perlmutter
(2002) mentions tolerating “egregious violations” of the principle of
cooperation (p. 157; on violations of Grice’s principle of cooperation,
see Chapter 8), the “suspension of critical assessment” (p. 159) and of
avoiding “detailed analysis” (p. 163).

3.3.2 Are release and play theories essentialist?

Release and play theories focus on the release humor provides. As such,
they do not lend themselves as readily to essentialist claims, since they
do not really look “under the hood” so to speak. Whereas one can make a
plausible claim that humor is essentially incongruity processing or that it
is more or less disguised aggression, it is a lot harder to claim that humor
is essentially release or play. This may explain why most of the theories
that include release or play do so concurrently with other components
(see for example Freud’s and Apter’s theories, reviewed above). In light
of the stance of complementarity that I take below (Section 3.7) this is
actually a strength of the release and play theories.

63
 THEORIES OF HUMOR AND THEIR LEVELS

3.4 Humor as incongruity

The best known theory of humor is probably the incongruity theory.

Incongruity is basically defined as divergence from expectations: “if it
differs, it is perceived as incongruous” (Forabosco, 1992, 54). This is con-
sistent with the ordinary meaning of the word “incongruous” indicating
the opposite of “the quality or state of agreeing, coinciding, or being
congruent” (Merriam-Webster).
Within humor research, incongruity theories are primarily associ-
ated with psychology. A common contemporary conceptualization in
psychology is to view incongruity–resolution (IR) models as “two stage”
(Suls, 1972) models.

In the first stage, the perceiver finds his[/her] expectations about the text disconfirmed
by the ending of the joke (…) In other words, the recipient encounters an incongruity—
the punch line. In the second stage, the perceiver engages in a form of problem solving
to find a cognitive rule which makes the punch line follow from the main part of the
joke and reconciles the incongruous parts. (Suls 1972, p. 82)

The two stages involve giving the text a first interpretation which is
then rejected in favor of a second interpretation. Note that the two
interpretations must coexist, at least to the extent that they are to be
judged incongruous.6
Incongruity and incongruity–resolution theories are most clearly
essentialist: they claim that the essence of humor is the perception of
the incongruity and its resolution, or merely the unresolved perception
of incongruity. Since a more detailed discussion of incongruity and
resolution, including its criticisms, will be found in Chapter 4, we can
leave this short introduction of incongruity and resolution and move on
to the darker view of humor as a form of aggression.

3.5 Humor as aggression

One of the oldest and most widely held theories of humor sees it as
an essentially aggressive or disparaging phenomenon. This theory has
been couched under many terms starting from Aristotle and Hobbes

6 See also Shultz’ definition as “a biphasic sequence involving first the discovery of the
incongruity followed by a resolution of the incongruity” (1976, pp.12–13).

64
 3.5 HUMOR AS AGGRESSION

all the way to Gruner’s work (1978, 1997) or Zillmann (1983). As

we saw, Apter’s reversal theory (1982), which is primarily a play and
incongruity theory, also contains a disparagement (diminishing) com-
ponent. Recently, Billig (2005) has argued in favor of aggression as
an explanatory variable in humor. Essentially the humor-as-aggression
theory boils down to the claim that humor can be explained at least at
some level as the pleasure/satisfaction or plain enjoyment of an enemy’s
discomfiture at our “triumph” over him/her. The dispositional theories
of aggression (Zillman, 1983) do not require an “enemy” but merely
someone whom one is negatively disposed toward. The idea is that we
tend to find more humorous things that we are negatively disposed
toward and conversely that we find less humorous things that we are
positively disposed toward. Thus for example, a conservative should
find humor directed at progressive targets funnier than the other way
around. There is some evidence that disposition theory may predict
correctly appreciation of humor based on political affiliation (Becker,
2014).
The most blatant counterexample to the aggression theory is of course
innocent or non-tendentious humor (Freud, 1905). Let us consider
again example (15), not from Freud’s repertoire incidentally, and here
repeated for convenience
(20) Why did the chicken cross the road?
To get to the other side.
Who or what would be the target of the aggression in this joke? The
chicken? The road? It is clear that the joke does not target any of its
characters, props, etc.; however, an aggression theorist might reply that
the joke does aggress someone, namely its hearer, i.e., the person who
has been told the joke and has therefore been exposed to a (presumably
small) dose of aggression connected to the fact that he/she has been
presented with an unsolvable riddle and has therefore been exposed to a
mild degree of ridicule. Therefore the aggression theory posits two levels
of potential aggression:

• aggression towards characters within the text;

• aggression towards participants to the joke telling exchange

It follows from this subdivision that a joke may or may not be aggressive
at either level. Therefore, a joke need not logically be aggressive at either

65
 THEORIES OF HUMOR AND THEIR LEVELS

level, i.e., a joke may be aggressive at the internal level or at the exchange
level which means that it is clear that either level of aggression is separate
and independent of the other. In other words, we can conceive of a joke
that is aggressive at both the contents level and at the interaction level
but by the same reasoning it follows logically that we can conceive of
a joke that is aggressive at neither. So, in order to work, an essentialist
aggression theory needs to specify that there has to be an aggression at
either level, which is somewhat circular.
In fact, the claim of the aggression theories examined before, while
plausible in the context of unsolvable riddles and in some cases of mildly
disguised aggression, for example, in children’s humor some riddles are
just an excuse to physically hit the dupe, is clearly untenable in the case
of puns and other “innocent” humor. For example, there are cycles of
charming jokes about elephants, such as
(21) How did you know elephants have been in your fridge?
Footprints in the butter.
that clearly have no aggressive component7 and one need only pick up a
collection of jokes for children under the age of six to find plenty more
of examples.
Clearly, as an essentialist theory, the aggression theory fails to provide
a complete theory of humor. Nonetheless, this does not mean that in
fact, a large number of jokes aren’t aggressive, and indeed, any theory of
humor that failed to address this very significant facet of the humorous
phenomenon would be remiss. Humor can be used deliberately or
unconsciously as an aggressive tool and this can be done as we saw above
both within the text and within the humorous exchange situation. In
this respect, humor does not differ significantly from any other kind of
text. The only difference that humor has with respect to other forms
of aggression is its deniability which comes from the pragmatic aspect
of the text; see Chapter 12 on the functions of humor and Chapter 8
on the violation of the cooperative principle that affords the deniability.
As we will see in Chapter 12, humor is often used in interaction as a
tool to probe the audience’s views and attitudes, precisely because of its
deniability embodied in the well-known expression “I was only joking.”

7 One needs to be a Freudian to see anything else than playful incongruity here. To
assume, as Dundes (1987) does, that the elephant stands for African-American men’s
sexuality, is, well, Freudian.

66
 3.6 OTHER THEORIES

Some authors use aggression (or variants such as “subversion”) as

essentially metaphorical constructs to indicate any criticism or even
merely playing with of any idea. Criticism need not be aggressive and
play is certainly not. This is not to say that some play may not degenerate
into aggression, but the very idea of the metamessage “this is play,” as
per Bateson (1972), denies the possibility of real aggression being part
of play. The danger of conflating aggression with criticism or play is that
we lose the useful category of aggression: if all humor is aggressive, then
the two jokes below would both be considered aggressive
(22) A guy walks into a bar and says: “Ouch!”
and
(23) A lawyer and a priest are flying in a small plane with a group of
children. Suddenly the plane’s engines start smoking. The captain
comes into the back of the plane and says: “There are only a few
parachutes. I think we should give them to the children.” “Fuck
the children!” says the lawyer and the priest replies: “Do you think
there’s enough time?”
It seems to me that (23) is an aggressive joke which expresses hostility
toward priests by accusing them of being pedophiles and lawyers by
accusing them of being morally reprehensible. Conversely, I have a hard
time believing that (22) expresses hostility toward either guys, bars, or
the hearer. Be that as it may, I believe it would be useful to keep the label
hostile or aggressive for jokes such as (23) in order to distinguish them
from jokes such as (22).
More sophisticated versions of the aggression theory state that humor
is not simply aggressive but involves a sense of superiority or ridiculing.
However, it is easy to see that the same criticism levied against aggression
theories applies to superiority and disparagement (ridicule). While cer-
tainly lots and perhaps most of humor is aggressive, superiority-based,
and disparaging/ridiculing, not all humor is so.

3.6 Other theories

After reviewing the “classical” theories of humor, we are reminded that

all classifications are to some extent arbitrary and that many approaches
do not fit clearly into the categories above.

67
 THEORIES OF HUMOR AND THEIR LEVELS

3.6.1 Neurolinguistics of humor

Neurolinguistics deserves particular mention because it is the most
extremely reductionist approach to explaining humor. Reducing psy-
chology to neurology, i.e., explaining cognitive phenomena as activation
of neurons, and/or other electro-chemical events in the brain cells, is
the most reductive explanation possible of cognitive phenomena (on the
topic see, e.g., Bickle, 2006).
Besides the neurolinguistic work on the distinction between appreci-
ation and response to humor considered in Section 2.3.2 recent work
in neurolinguistics has yielded some interesting results about the nature
of humor, which are generally consonant with the incongruity theory.
Derks et al. (1997) showed that a “negative-going cortical activity at
400 milliseconds” (N400) is associated with what humor theory has
described as the incongruity of humor. Derks et al. describe the N400
as “occurring when categorization, usually semantic, is relatively unsuc-
cessful and a search is initiated for better alternatives” (p. 287); see
also Coulson and Kutas (1998, 2001). Coulson (2001) not only find
evidence for the N400 response, but also for the second phase of the
joke processing process, i.e., the resolution of the incongruity (i.e.,
the activation of a new frame/script). The suggestion that incongruity
perception and resolution may be concurrent, rather than sequential is
very interesting and dovetails with recent work on irony.8 Other studies
of neural activity that are consistent with the incongruity/resolution
include Ozawa et al. (2000), Iwase et al. (2002), and Mobbs et al. (2003).
Goel and Dolan (2001) have shown, using MRIs, that different areas
are involved in the processing of verbal and referential jokes.9 This is
significant, because it shows that the distinction between verbal and ref-
erential humor has a neuroanatomical foundation. They also distinguish
between areas involved in the processing of the semantic and phonolog-
ical material of the texts (for example, puns activate Broca’s area) and the
“affective” components of humor (i.e., the perception of funniness). For
a broader discussion of the neurolinguistics of humor and an emphasis

8 Incidentally, the fact that processing of jokes requires backtracking and longer reading
times clearly militates against any “direct access” interpretation of humor. On the direct
access theories of irony, see Colston (2017).
9 Semantic and phonological jokes in their terminology; all humor is semantic, needless
to say, so their terminology may be confusing.

68
 3.6 OTHER THEORIES

on the distinction between the cognitive and the affective aspects of

humor perception/appreciation, see Chen et al. (2017).
There exists considerable (highly technical) discussion of the lat-
eralization of humor processing, which seems to show that the right
hemisphere of the brain is crucial to humor processing. Derks et al.
(1997), Coulson and Kutas (1998, 2001), and Goel and Dolan (2001) all
show that this approach may be in need of some revision. There exists
some literature on the neuro-anatomy of laughter, which is outside the
scope of this discussion, but see Vaid and Kobler (2000).

3.6.2 Evolutionary theories

Other less common theories include attempts at seeing an evolutionary
advantage to humor. The most developed linguistic approach along
these lines is Chafe’s (1987) disabling theory. He sees humor as evo-
lutionarily advantageous in disabling the speaker when he/she begins
to pursue lines of thought that lead to absurdities, contradictions, etc.
The disabling theory is expanded in a full-fledged theory of humor
and laughter in Chafe (2007). On the evolutionary theories of humor,
including Chafe’s, see Vaid (1999, 2002) and Porteous (1988). A differ-
ent, non-linguistically aware, approach is to be found in Gervais and
Wilson (2005).

3.6.3 Anti-essentialist theories

At a meta-theoretical level (i.e., the theory of theories), all the theories
we have reviewed agree on some basic facts: one of them is that humor
can be explained in an essentialist framework, or to put it differently,
that there are sufficient and necessary conditions for a given “thing” to
be humorous such that all humorous things share these conditions. The
assumption that there is an essence of humor has been questioned, more
or less fortunately. It is logically possible that humor is a concept such as
that of “game” which, as shown in a famous argument by Wittgenstein
(1953), does not have a set of features common to all games. Let
us consider the only serious challenge to the essentialist paradigm10
10 A “paradigm” is a complex epistemological background shared by several theories
(Kuhn 1962). See the discussion in Chapter 1.

69
 THEORIES OF HUMOR AND THEIR LEVELS

within humor research: in 1990, Ferro-Luzzi published an article

claiming that a particular example of alliterative humor in Tamil did
not involve incongruity. I showed (Attardo 1994: 139) that alliteration
does involve an incongruity between the expected statistical frequency
of phonemes in a given stretch of text and the observed repetition of
a given (set of) phoneme(s) in the text. Thus the alleged refutation of
incongruity theory by the counterexample of alliteration is found to be
groundless.

3.6.4 Mixed, partial, and no-theory-theories

In reality, most theories are mixed, in the sense that they take a little bit
from all three categories. For example, Freud, the grand-dad of all release
theories, spends a significant amount of his book on jokes developing
a treatment of the mechanisms used to establish incongruity. Bergson,
one of the chief exponents of the socially-oriented theories (humor as a
social corrective), also developed the view of humor as “mechanical vs.
living” (read, incongruity).
Partial theories are simply theories that do not attempt to provide a
complete treatment of all forms of humor but limit their goals to one
or more aspects of the phenomenon (for example, satirical humor, or
humor in a given culture, or lexical ambiguity puns, etc.). Most theories
of irony, for example, are partial theories of humor, if they acknowledge
the connections between irony and humor.
Finally, there is a number of writers that deal with humor in a more or
less deliberately non-theoretical manner: their writings are (sometimes)
interesting, rich of intuitions, usually well exemplified, and generally
devoid of generalizations or explanations that go beyond common sense.
Another characteristic is that they typically refuse to engage with the
cutting edge of the research on humor and instead review obsolete
classics (Aristotle, Kant, Freud, etc.). I discuss two such approaches in
Attardo (1994, pp. 8–10), but one could multiply examples: Nash (1985),
Ross (1998), Redfern (1984), or more recently Blake (2007) and Bevis
(2013). Let me be very clear: there is nothing wrong with pre-theoretical
collecting of examples and considerations or with writing introductory
books. Our goals here are simply different: mapping and engaging with
the cutting edge of the field so as to put the reader in a position to
contribute to it.

70
 3.7 COMPLEMENTARIT Y OF THE THEORIES OF HUMOR

3.7 Complementarity of the theories of humor

The three families of theories of humor are not predicated at the same
level of generality. In fact, rather than being opposed they are comple-
mentary.11
It is clear that the incongruity theories are more or indeed exclusively
interested in the stimulus of the humor phenomena or even more
specifically in some of the features of the stimuli (i.e., the specific kind of
contrast in which the two concepts that are involved in the incongruity
must find themselves in, e.g., Giora’s (2003) “negation,” or Raskin’s
(1985) “local opposition.” The neurolinguistic theories are even more
reductionist in that they target neuronal activity in the brain supporting
these concepts (thus the N400 wave of deactivation is commonly iden-
tified with the concept of incongruity and the P600 with the concept of
resolution).
Release or liberation theories consider the effect of incongruity on
the psyche of the individual producing the humor. They are concerned
with the feelings or attitude of the joker, so to speak, and reference
social factors only insofar as the individual may have internalized social
constructs, such as Freud’s Super Ego.
Aggression or superiority theories are squarely concerned with the
social setting of the humorous phenomena. They address exclusively
the interactional aspect of humor: what speakers do with humor when
dealing with other speakers. They are social theories.
Once seen in this light, it is clear that the three families of theories
might really only be three aspects of a rather gigantic, broader, all-
encompassing super-theory which would specify the nature of the stim-
ulus, all the way to the neuronal activity, the relation of the stimulus to
the individual psyche, and the interplay of the individual and its society.
Needless to say, no such Über-Theorie has been proposed.12 Consider an
example:
(24) One day, as I was withdrawing some money, the cashier at the
bank asked me: “Do you want me to take it out of your checking?”
I answered, “No, take it out of someone else’s checking.”

11 I owe the general observation on which this section is based to Donald Casadonte.
The elaboration of the basic idea is however mine.
12 But see the conclusion of this book, for a shot at it.

71
 THEORIES OF HUMOR AND THEIR LEVELS

At the stimulus level, we have clearly an opposition between the

normal, received way of withdrawing money in a bank, supported by
a complex network of scripts about depositing and withdrawing money,
interests, etc., and an abnormal way. Clearly there are connotations
being activated, since taking the money from someone else’s account
would leave it in mine (a good thing) but would constitute theft (a bad
thing). So it is a complex opposition. The resolution aspect is merely
the ambiguity of the expression “take it out of your checking [account]”
which may have (among others) two different figure/ground articula-
tions: the one intended by the cashier, where the figure is “checking” (as
opposed to “savings” account) and the one I deliberately (and uncoop-
eratively) chose with the figure on “your” as opposed to “someone else’s”
account.
At the individual personal level, I was clearly in a playful mood. I
enjoyed showing off my capacity to detect an ambiguity where there
was none (contextual pressure rules it out as a bona-fide ambiguity) and
possibly I released any tension I might have had.
At the social level, I engaged the cashier in a personal, non-work,
playful exchange, acknowledging her as being a person before being an
employee of the bank. Hence I engaged her in an in-group solidarity
building exchange. Potentially, one could argue that my remark also
expressed an anti-capitalist stance close to Proudhon, since by rejecting
the principle of personal property (hence the request to withdraw funds
from another account) I would undermine the whole economic system,
needless to say if one took it seriously. Hence one could categorize my
remark as subversive and hence aggressive.
As this sketchy analysis of an example13 has shown, there is nothing
incompatible about the various “explanations” provided by the three
families of theories. Of course, the theories cannot claim to be com-
plete explanations of the humor phenomenon. They highlight different
aspects of the same phenomenon, much like in the parable of the
elephant and the blind men (see Berger, 1995).

13 The example was first analyzed in Hamrick (2007) as an example of “trumping” (Veale
et al., 2006). Trumping is a form of forced reinterpretation that plays on the distribution of
saliency in the text. Further analysis of the example can be found in Section 9.3.1.3.

72
 3.7 COMPLEMENTARIT Y OF THE THEORIES OF HUMOR

3.7.1 Evaluation of theories

After having discussed the ways in which theories which appear to
be opposed are in fact complementary, we are now in a position to
discuss a related matter, namely the evaluation of competing theories
at the same level of abstraction. As a case study, we will investigate the
success (undeserved, as we shall argue) of Koestler’s bisociation theory,
compared with the incongruity theories, and two linguistic theories of
humor (the Semantic Script Theory of Humor and the Isotopy Disjunc-
tion Model).

3.7.2 Bisociation theory

The bisociation theory was introduced in Koestler’s book (1964) about
creativity and has been very influential. Koestler defines bisociation as

the perceiving of a situation or idea ( . . . ) in two self-consistent but habitually incom-

patible frames of reference ( . . . ) The event ( . . . ), in which the two intersect, is made
to vibrate simultaneously on two different wavelengths, as it were. While this unusual
situation lasts, [the event] is not merely linked to one associative context, but bisociated
with two. (1964, p. 35)

The source of Koestler’s insight is Maier (1932), a Gestalt psychologist,

who pointed out that insight and humor shared some traits, summarized
as follows by Lòpez and Vaid, “the experience of humor (like that of
insight) involves a sudden and unexpected restructuring of the elements
of a configuration leading to clarity and a solution. In other words,
humor can lead to discovery” (Lòpez & Vaid, 2017, p. 270).
As I noted in Attardo (1994, p. 175) bisociation is a cognitive theory,
and, clearly, a type of incongruity theory. The definition of bisociation
is metaphorical (more on this below), and roughly coextensive with the
notion of isotopy disjunction (see Section 4.3.1) and of script opposition
and overlap (see Chapter 6). It is endowed, however, neither with the
same degree of formal definition nor with the heuristic procedures
for determining which script/isotopy is being activated. This is the
defining features that characterizes linguistic theories: they provide the
researcher with (rather elaborate) procedures for identifying the scripts
or isotopies being activated by the text.
Basically, the issue boils down to this: Koestler’s formulation is entirely
metaphorical and no description (algorithmic or not) of how to identify

73
 THEORIES OF HUMOR AND THEIR LEVELS

the planes of bisociation is given. On the contrary, incongruity theories

have at least independently motivated conceptual bases that are not
metaphorical and are (at least in part) based on interpersonal assess-
ments of stimuli. Linguistic theories, finally, have independently moti-
vated,14 non-metaphorical conceptualizations, with algorithmic (semi-)
formal procedures for analysis. Therefore, linguistic theories are better
theories because they are more formalized, while preserving the same
degree of descriptive adequacy, and because they provide explanations
that are not ad hoc, i.e., they are based on concepts introduced indepen-
dently of humor.15
From a linguistic point of view, in fact bisociation is thus only a
notational variant of the isotopy disjunction theory, or the script theory,
with the further flaw that bisociation/incongruity etc. had already been
identified since the Greeks as the central mechanism of humor. Hence,
bisociation is not even a novel theory. What changes are not the under-
lying conceptualizations, but the metaphors used to describe them, and,
of course, the degree of formalization/replicability of the theory.

3.7.3 Evaluation of the theories

How does one evaluate a theory? There are some more or less accepted
scientific metrics for theories. They include:

• coverage and comprehensiveness (your theory must account for all

the facts)
• verifiability (falsifiability) and intersubjective verification (i.e., oth-
ers must be able to replicate your results)
• simplicity (all other things being equal, the simplest theory is to
be preferred); read “simple” as 1) have less axioms,16 2) having less
theorems/rules, etc.

14 The reader will recall the warrant of transferability, i.e., independent motivation, from
the discussion in Section 2.1.2.
15 Chomsky (1965) distinguishes various levels of adequacy for a grammar (but the
reasoning may be extended to theories). Descriptive adequacy means that the theory handles
the data: it accounts for all the data. Explanatory adequacy “explains” the data in terms of
underlying principles. Linguistic theories strive for explanatory adequacy.
16 An axiom is an unprovable fact (from within the theory) that is taken as the starting
point of the theory. For example, in Chapter 5 we will take that humor happens within a
semiotic-inferential system as axiomatic.

74
 3.7 COMPLEMENTARIT Y OF THE THEORIES OF HUMOR

• reductive (i.e., it should explain things in terms of simpler, better

understood other17 things; in other words, you cannot use a sim-
ilarly or (even worse) more complex concept to explain a simpler
one, e.g., defining humor in terms of “the ridiculous,” or vice versa).
• formality (i.e., the theory must be statable in a way such that it needs
no recourse to the reader’s intuition, or to put it differently, it must
be algorithmic, i.e., it should be, in principle, implementable in a
computer program).

Using these criteria we may venture in a difficult, but extremely

important avenue: explaining some of the theoretical choices made both
by the field at large and by the present writer.
First, by definition no-theory theories and partial theories are ruled
out as non-scientific or non-comprehensive, respectively. While the
latter may still be scientifically interesting, for example if they are used
as material for the construction of a comprehensive theory, the former
are by definition uninteresting scientifically. What these publications are
doing may be stimulating, entertaining, at times even helpful if they
present good or memorable examples, or if they spark the interest of
young researchers, but it is not science and should not be confused
with it.
Second, by the principle of simplicity, it follows that an essentialist
theory should be preferred, if available, to any other theory. Essentialist
theories, if formalized enough, are very easily testable (falsifiable) since
it is only18 necessary to find a serious counterexample to the theory to
falsify it.
From within incongruity–resolution theories, we have several
candidates to choose from. In Attardo (1997), I pointed out the obvious
fact that incongruity is a semantic concept. It follows that a linguistic
(or philosophical) definition of incongruity is at an advantage over
an anthropological or psychological definition, as both linguistic and
(to a lesser degree) philosophical semantics have developed (more
or less) scientific accounts of meaning. From within linguistics, two
17 Note the importance of the “other” here, otherwise a theory would fall under the
dreaded “circularity” pitfall.
18 Not really: theories are not discarded in toto if any of their predictions turn out to be
unsupported by data. Most commonly they are modified as little as necessary to account
for the data. Needless to say, sometimes the data are declared to be irrelevant or outside
the purview of the theory, and correspondingly the focus of the theory is changed. This is
dangerous business, however, as it may reduce the comprehensiveness of the theory.

75
 THEORIES OF HUMOR AND THEIR LEVELS

full-blown theories19 have been developed, a script-based one, and an

isotopy-based one. The problems with the isotopy-based model are
such (Attardo, 1994, pp. 61–81)20 that it is not a serious contender as a
comprehensive theory (although it does have an interesting advantage
for the conceptualization of the linear ordering of the joke, see Section
4.3.1). This leaves the script-based theory as the sole contender.21
Because the incongruity theories are essentialist (i.e., the attempt
to pinpoint what makes humor funny), linguistics has tended to side
(largely unwittingly) with this kind of theory. However linguists have
show some interest for hostility theories (see for example the concept
of “target” in the General Theory of Verbal Humor, below) and release
theories. For example, the idea of defunctionalization (Guiraud, 1979)
of language in puns fits in very nicely with the liberation approach, since
it frees the speaker from the constraints of the linguistic code. Similarly,
the idea of retractability in discourse of humor and irony frees the
speaker from the consequences of his/her actions. These connections
with humor theory have not been pursued in any systematic fashion.
From a different perspective, namely the performance side of humor,
the picture is much different. Essentialist theories are more or less
sidelined, in favor of a discursive, ethnomethodological stance (see
Chapter 10 for discussion) which strongly favors comprehensiveness
and intersubjective verification and more or less abandons any attempt
at formalization. Perhaps because of the reductionist stance of the three
major humor theories, or perhaps because of the focus on the positive
aspects of humor, which go against the aggression theories of humor, the
discursive and ethnomethodological approaches have developed more
or less apart from humor studies. See the conclusion of the book for an
argument on why they should be brought back together.
19 There are other approaches, cognitive, relevance-based, corpus-based, etc. but none
has presented a developed theory. See Brône et al. (2015) and Yus (2017) for examples
of cognitive and relevance-based approaches. Corpus-based approaches are discussed in
Section 12.3.
20 I hate to have to say this, but the discussion is not for the faint of heart, or those without
some linguistics background. They may read pages 61–62, in Attardo (1994), take my word
for it, and miss very little from the humor point of view. Professional linguists will have to
bite the bullet, so to speak.
21 I assumed, wrongly, that this point was obvious. Some reviewers of Attardo (1994)
took me to task for my “favoritism” towards the script-based theory. If there is only one
theory that does what one needs done, one has to embrace it, warts and all. The fact that two
entire chapters (and many other sections) in the said book were dedicated to developing
aspects neglected by the script-based theory, not to mention that I had published in 1991 a
serious revision thereof, apparently did not suffice to establish my (attempts at) objectivity.

76
 3.8 FURTHER READINGS

3.8 Further readings

Surveys of humor theories can be found in Piddington (1933), Monro

(1951), Bergler (1956), Keith-Speigel (1972), Raskin (1985), Morreall
(1987), Attardo (1994), Smuts (n.d.). Martin (2007) now in a second
edition co-authored with Thomas Ford (2018) provides an outstanding
survey of the three classes of theories of humor from the standpoint of
psychology.
Forabosco’s (1992) treatment of incongruity in cognitive terms
remains fundamental. The history of the incongruity theory can be
explored in Morreall (1987) and Attardo (1994, 1997). Other incongruity-
based theories are Koestler’s (1964) bisociation theory, Paulos’ (1980)
catastrophe-theory model, and Hofstadter and Gabora’s (1989) model.
On bisociation, see Krikmann (2014).
Non-essentialist theories of humor have been presented by Ferro-
Luzzi (1990) and Latta (1998). An example of the synchretistic approach
to humor theory is M. S. Davis (1993, p. 7). For a discussion of mixed,
partial, and no-theory theories, see Attardo (1994).
On the neurolinguistics of humor, see Vrticka et al. (2013), Chen
et al. (2017), and Rodden (2018). On the evaluation of theories in the
linguistics of humor, see Attardo and Raskin (2017).

77
 4
Incongruity and resolution

4.1 Incongruity 79
4.2 Resolution 82
4.3 Linear organization of the joke 90
4.4 Conclusions 93
4.5 Further readings 94

This chapter deals more specifically with the cognitive processes that
underlie humor. We turn first to the incongruity aspect, leaving the
resolution issues for the second part of the chapter.
Concepts eerily similar to the modern incongruity and resolution
model can be traced back to Aristotle. In the Rhetorics, he points out that
in many witticisms and puns “the speaker says something unexpected,
the truth of which is recognized” and “In all these jokes, whether a
word is used in a second sense or metaphorically, the joke is good if
it fits the facts” (III, 11 1412b). We already have, in a nutshell, the
modern theories: the unexpected nature of the incongruity, the presence
of two senses (often one of them being figurative), and the resolution
phase: recognizing the “truth” of the witticism or its “fit” with the facts.
Quintilian, a Latin rhetorician, in his Institutio Oratoria, also anticipates
modern definitions of incongruity: in reviewing three kinds of humor:
directed against others, ourselves, and neutral (i.e., not directed at any-
one in particular), he notes that the third kind consists of the “thwarting
of expectations, taking differently the things said.” (VI-3-24).
Whether incongruity theories date back to the Greek and Latins, or
as is more commonly assumed, to Kant and Schopenhauer (see Mor-
reall, 1987) is ultimately irrelevant. In their modern form, incongruity

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 4.1 INCONGRUIT Y

and resolution theories originate in psychology, with the work of Sig-

mund Freud, whose 1905 book on humor contains a very significant
discussion of the mechanisms of humor, which is unrelated to his
sexual-oriented model of the psyche. Freud is probably also the unac-
knowledged source of the isotopy–disjunction model (Attardo, 1994,
pp. 63–64). Incongruity–resolution theories were among the first the-
ories to be conceptualized and tested in humor studies (e.g., Suls, 1972),
and have seen a more recent resurgence as one of the three dimensions
in Ruch’s 3WD theory (Ruch, 1992), as INC-RES (the other two dimen-
sions being sex and nonsense). An excellent review of the psychological
theories of incongruity/resolution and the debate of whether both are
necessary can be found in Forabosco (1992).
We saw that incongruity theories are one of the three major theories
of humor. We now take the time to examine in depth what exactly
incongruity is and will later subject the concept of resolution to the
same treatment. The last section of the chapter deals with the linear
organization of jokes, i.e., the isotopy–disjunction model.

4.1 Incongruity

Incongruity is a semantic concept. So, in this sense, incongruity theories

are semantic theories, as I argued in Attardo (1997). This logical primacy
of semantics vis-à-vis incongruity theories should not be (and has not
been) taken too seriously. In fact, most humor scholars tend to think
that semantic theories (usually the Semantic-Script Theory of Humor
and the General Theory of Verbal Humor, but one could make the same
argument for the isotopy–disjunction model) are incongruity theories.
This is because, as we have seen, incongruity is one of the major classes
of theories of humor (in fact, the major one, one could argue). However,
the point remains. Incongruity is defined, implicitly most of the time,
semantically. Let us go back to the pellucid definition proposed by
Forabosco, and already mentioned in the previous chapter: “if it differs,
it is perceived as incongruous” (Forabosco 1992: 54). This definition has
the distinct advantage of emphasizing that the definition of incongruity
is based on expectations.
In one of the classic experiments in humor studies, Nerhardt (1970,
1975) conducted a study in which subjects, under the guise of compar-
ing them to a standard, lifted weights of increasing heaviness until one of

79
 INCONGRUIT Y AND RESOLUTION

the weights was significantly lighter than the previous ones. Upon lifting
the unexpectedly lighter weight, the participants smiled or laughed
more than if they weight series was congruent (i.e., increased). The
experiment was widely replicated (see Gerber & Routh, 1975; Deckers &
Kizer, 1975; Deckers & Winters, 1986; Deckers, 1993; Martin, 2007, pp.
68–70). What is most interesting, in our perspective, is that it is the
clearest and simplest operationalization of the idea of incongruity, in the
sense of violation of expectations.1 The weights experiment is widely
assumed to demonstrate that incongruity alone can be responsible for
the production of mirth. We will return to this in Section 7.1.5.
For now, we focus on the expectation: the fact that in all the designs of
the experiment that subjects first lift the standard weight and then after a
varying number of judgements, ranging from zero to 22, they were given
a “target” weight that was either much heavier, much lighter, or weighed
the same. When the weights matched or after no other judgements,
there were no mirth reactions, or virtually none. Conversely, after 7,
11, or more judgements, significant mirthful responses were observed.
This result is interpreted as showing that the first few comparisons
establish an expectation (e.g., slow increase of the weight) and that that
expectation is then violated by the unexpected, and thus incongruous,
weight. This argument will re-emerge in different form in the discussion
of narrative strategies (Section 7.1.2), in which in the telling of a joke a
regularity is established by two occurrences, only to be broken in the
third instance.
It should be noted that the definition of incongruity discussed above
is essentially the same as that of surprise. Surprise can be considered
the emotional response to incongruity. While most scholars see surprise
as an emotion (e.g., Ekman et al., 1983), others see it as a cognitive
phenomenon (e.g., Lorini & Castelfranchi, 2007). In humor studies, one
occasionally encounters the idea that incongruity needs to be sudden
or surprising (e.g., Eysenck, 1942, p. 306). So, if the basic definition of
incongruity and surprise are the same, how can there be non-sudden
incongruity? The reader will recall the discussion (in Section 2.4.2) of
how the presentation and the keying of incongruous items eliminate the
surprise effect. Another way of conceptualizing this is that incongruity

1 Not entirely free of external factors: for example, when Nerhardt originally tried the
experiment with suitcases in a train station, he failed to elicit a mirth response, probably
because the participants were not in a playful frame of mind (Martin, 2007, p. 70).

80
 4.1 INCONGRUIT Y

is a competence category and surprise is a performance-level category

(i.e., it is an experience, an effect of an incongruous stimulus). Incon-
gruity conceptualizes the mismatch of expectations at a systemic, static
level, whereas surprise conceptualizes it in a dynamic, experiential way.
Another concept related to incongruity is cognitive dissonance (see
Section 2.4), which can be conceptualized as the non-playful, unpleasant
reaction to an incongruity.

4.1.1 Enter semantics

As soon as we introduced the idea of expectations, we also implicitly
introduce the idea of semantics. If we say that we “expect” A to follow B,
then producing intentionally B comes to mean A. It is in this sense that
expectations imply meaning.2 Here, we must thread carefully. Surely,
one may object, linguistics has nothing to do with expectations of
weights. First, meaning is a semantic issue and not a linguistic one.
Meaning (semantics) exceeds language.3 This is just a fancy way of say-
ing that there is meaning within language (let’s say words and sentences,
to simplify) and there is meaning that is extra-linguistic, because very
simply, it is conveyed using a non-linguistic system of communication
(think of flags between ships, of road signs, of pictures, of clothing,
especially uniforms, etc.).4 So, the claim that expectations are semantic
has nothing to do with linguistics (yet). Second, even psychologists
recognize the logical primacy of semantics; see for example, Deckers and
Winters (1986) who frame the issue in terms of “schemata”:5 “A schema
is an abstract mental representation of environmental and behavioral

2 This is the foundation of Grice’s Meaning

NN , incidentally. This definition is tied
significantly to the Principle of Cooperation, which will play a significant role in the
pragmatics of humor; see Chapter 8.
3 Here one may object that semantics is a branch of linguistics. While it is true that
linguistics has a branch called semantics, it is limited to “linguistic semantics” (i.e., meaning
conveyed via language). Philosophers would be ready to point out that model theoretic
semantics, just to name one, is independent of language. Computer languages have their
own semantics as well, which are also completely independent, and very different, from
linguistic semantics.
4 If we wanted to be picky, we could observe that even these non-linguistic systems of
communication, rely on language in some form of another: for example, the meaning of the
roads signs are explained in a handbook which every driver needs to memorize in order to
pass an exam. Lotman, defines language as the primary modeling system of a culture, from
which all secondary systems are derived (Lotman, 1975).
5 The plural of “schema” is “schemata.” “Schemas” is also acceptable, but indicates a lesser
degree of sophistication in the speaker.

81
 INCONGRUIT Y AND RESOLUTION

regularities. Discrepancy or incongruity refers to a stimulus feature that

does not match the abstracted value of that feature in the schema” (p. 57).
As we will see in Chapter 6, “schemata” is another term for scripts or
frames, which are all semantic objects.
So, in conclusion, incongruity is a semantic (but perhaps not nec-
essarily a linguistic) concept, because it presupposes the attribution of
meaning to a regularity in the observed world. The field that deals
with meaning in all its manifestations, linguistic and not, is of course
semiotics. We will take up the semiotics of humor in Chapter 5.
There are plenty of incongruity-based theories of humor. In fact, most
theories of humor, at least covertly, include an incongruity aspect, such
as Aristotle, commonly considered to be the proponent of a superiority
theory of humor, or Freud, associated with a psychoanalytical theory
and Bergson’s mechanical/living opposition. See Attardo (1994), for dis-
cussion of these examples. Among the most quoted incongruity theories
are Kant, often credited with coming up with the idea, Schopenhauer,
Koestler’s (1964) bisociation theory, and Paulos’ (1980) catastrophe-
theory model.
Incongruity is mostly a non-controversial, generally accepted idea,
as claimed, for example, by Perlmutter (2002, p. 155) “the observation
that some form of incongruity is an essential condition for humor, is
universally accepted in every serious analysis of the subject.” There have
been, however, some claims to the contrary, such as Ferro-Luzzi (1990),
Latta (1998), and Veale (2004). Ferro-Luzzi is refuted in Attardo (1994).
Latta is reviewed negatively by Oring (1999) and not even mentioned in
Martin (2007). Veale’s objections are addressed by De Mey (2005, p. 74).

4.2 Resolution

The second part of the incongruity–resolution formula is probably the

most misunderstood concept of humor studies. The most common
mistake is to take the label “resolution” too literally and to expect that
the resolution of the incongruity of a joke completely remove the incon-
gruity. We will first work our way through some major contributions to
the subject: Sigmund Freud, Elie Aubouin, a French psychologist, Avner
Ziv, an Israeli psychologist and one of the founding fathers of humor
studies, Giovannantonio Forabosco, an Italian psychologist, and Elliott
Oring, an American folklorist. We will then introduce the concept of

82
 4.2 RESOLUTION

logical mechanisms, which is the present-day most in-depth treatment

of the subject. After that we will address the issue of full and partial
resolution of incongruity and whether incongruity is foregrounded or
backgrounded in the text.

4.2.1 The history of resolution

4.2.1.1 Freud
Freud’s theory of humor as economy of mental energy is well known
and needs not detain us any further beyond the fact that his theory
is implicitly an incongruity theory (see Attardo, 1994, p. 56). In this
context what is relevant is his use of the concept of “sense in nonsense”
(Freud 1905; I am quoting from the 1916 translation, p. 10) which is
borrowed from Lipps (1898):

A saying appears witty when we ascribe to it a meaning through psychological necessity

and, while so doing, retract it. It may thus have many meanings. We lend a meaning
to an expression knowing that logically it does not belong to it. We find in it a truth,
however, which later we fail to find because it is foreign to our laws of experience
or usual modes of thinking. We endow it with a logical or practical inference which
transcends its true content, only to contradict this inference as soon as we finally
grasp the nature of the expression itself. The psychological process evoked in us by
the witty expression which gives rise to the sense of the comic depends in every case
on the immediate transition from the borrowed feeling of truth and conviction to the
impression or consciousness of relative nullity.
(Lipps, 1898, p. 85; translated by A.A Brill, 1916)

Freud may have been the one to invent the catchy expression “sense
in nonsense.” Maier (1932), whom we have encountered previously
as a source of Koestler, speaks of a “limited” logic that holds only
temporarily.
4.2.1.2 Aubouin
Aubouin (1948) introduced the term “justification” in the same sense in
which we currently use “resolution”: the second phase of the humorous
cognitive process. His term for the incongruity phase was inconcilia-
bilité, “irreconciliability.” Aubouin argues that two incongruous, irrec-
onciliable objects are not perceived as humorous, per se. For humor
to be triggered the two object have to be “accepted” simultaneously by
the hearer. Acceptance is the hearer-side behavior; justification is the
neutral-situation-side view of the phenomenon.
83
 INCONGRUIT Y AND RESOLUTION

Aubouin stresses that the acceptance is not a real, full-fledged belief. It

is brief, “superficial,” merely “masking for an instant the absurdity of the
judgement” (Aubouin, 1948, p. 95). Aubouin contrasts the acceptance
of humor to that of a mathematical proof (p. 94): a mathematical proof
is fully and permanently accepted. The acceptance of the incongruity of
a joke is fleeting.
According to Aubouin, the acceptance of humor, the justification,
involves an actual error of judgement, or at least the possibility of
one. I believe that this part of Aubouin’s explanation is itself in error
(ironically enough) as I have shown in Attardo (1994, pp. 146–147),
since there is no need to postulate an actual error. In some cases, the
hearer will be really misled, for example when the context strongly
primes one interpretation, which is then revealed to be wrong, as in the
following example, from Attardo & Raskin (1991, pp. 305–306):
(25) George Bush has a short one. Gorbachev has a longer one. The
Pope has it, but does not use it, Madonna does not have it. What
is it? A last name.
which reveals its age from the fact that some of the readers will need to
google who these people are. Conversely, in many jokes the hearer has
to deliberately play along with the joke, as in
(26) Q: What do you get when you cross a cow and a lawnmower?
A: A lawnmooer.
where at no point in time the hearer believes that one can actually cross
a cow with an inanimate object6 or that a lawnmooer can actually exist
or what it would actually be. All that the hearer is aware of is that
“lawnmooer” and “lawnmower” sound similar.
The value of Aubouin’s contribution resides in the fact that his was
the first in-depth, and so far the most significant, discussion of the
resolution of the incongruity of humor. His relative lack of recognition
is probably due to the fact that he published in French, and his output
seems limited to two books and a couple of articles.

6 Cyborg cows are beyond the imagination of most people, but apparently not beyond
that of engineers, who are inserting bio-monitors under the cows’ skin; see Metz (2018).
For the opposite side of the debate, see the prescient song Cows with Guns by Dana Lyons
(1996).

84
 4.2 RESOLUTION

4.2.1.3 Ziv
Ziv coined the term “local-logic” (Ziv, 1984, p. 77) by analogy with
“local patriotism” which corresponds to the Italian term “campanilismo”
(which means the bragging about the height of one’s village’s bell-tower).
The point here is not the bragging part, or the attachment to one’s
land, but the association with small villages, with the “locality” of the
matter. In local patriotism, the focus of one’s positive affect is deliberately
restricted to a very small domain. So, analogically, local logic is a logic
that is valid only in a very small domain, for a very short amount of time.
“Like local patriotism, local logic is appropriate only in certain places”
(p. 90).
Local logic “brings some kind of explanation” to the incongruity. It
must provide an explanation “with a certain suitability” (p. 90). Ziv
analyzes the following example:

A young man looking for a wife went to a computerized marriage agency. Filling out the
form, he wrote, “I’d like someone who likes lots of company, water sports, and formal
dress, and is preferably rather short.” The agency sent him a penguin. (p. 88)

Being sent a penguin for your mate is certainly surprising and incon-
gruous. However, it has “suitability” (p. 90) as a comparison revels: if
the agency had sent a horse it would have been equally surprising and
incongruous, or perhaps even more so, but it would have lacked the local
logic suitability of the penguin. The suitability comes of course, from
the fact that penguins are short, live in large colonies (in some species,
of tens of thousands of birds), are aquatic, and for some species their
plumage resembles formal wear (for example, emperor penguins).
Local logic must be taken in a playful way: it “is very amusing if we
are willing to play along, but not” if we stick to normal logic/knowledge.
In the penguin example, Ziv notes that local logic will not work if “we
say crossly that marriage bureaus do not send penguins to their clients”
(p. 90).
Finally, Ziv argues that local logic “occupies a middle position between
logical and pathological thinking” (p. 98). In logical thinking, logic is
in control of thinking, whereas in pathological thinking, “fantasy and
the absurd are in control and are perceived as reality” (p. 98). On the
contrary, in local logic thinking “uses and enjoys both logic and fantasy
without confusing them” (p. 98). In short, it involves willing suspension
of disbelief (see Section 2.4.1). The analogy with pathological thinking

85
 INCONGRUIT Y AND RESOLUTION

is quite apt as Arieti (1967) had shown in his discussion of paleo-logic

in schizophrenia. Ziv uses the term “pararational” to describe local logic
(p. 107) which of course brings to mind Apter’s (1982, 1989) “paratelic”
mode of communication.
4.2.1.4 Forabosco
As we saw, Forabosco defines incongruity as follows: “a stimulus is
incongruous when it differs from the cognitive model of reference”
(1992, p. 54) and stresses the importance of including the perceiving
subject in the definition (using the term “model” which presupposes
someone doing the modeling). Forabosco then considers that one may
define resolution as a “problem-solving” activity that finds a rule which
“reconciles the incongruous parts” or, following McGhee (1979) as
“cognitive mastery” (i.e., the cognitive capacity to understand that if
something unexpected occurs, this is because it goes against “how things
should be”; McGhee, 1979, p. 38).
Forabosco then argues that there are two ways to interpret “res-
olution.” The first one is “a cognitive rule [ . . . ] that reconciles the
incongruous parts” (p. 57). The second one, is a “congruence criterion”
(p. 58), i.e., any element relating to the stimulus on the basis of which
the stimulus, at least in that specific respect, is congruous” (p. 57).
Forabosco does not say this, but I think we can reach the conclusion
that the first is a complete resolution (elimination) of the incongruity,
whereas the second is a partial resolution. A congruence criterion allows
cognitive mastery. The example that Forabosco provides is a child seeing
a distorted face (caricature) and realizing that “it is just a drawing and
not a real face” (p. 57). Forabosco notes that in the case of the congruence
criterion it co-occurs with the incongruity (it is “contemporaneous”)
and below the threshold of consciousness.
Most significantly, Forabosco notes that “Resolving the incongruity
does not mean eliminating it. It means having, at the end of the process,
an incongruity ‘that makes sense” ’ (p. 59) or a “congruent incongruity”
(p. 59; emphasis in the original). The “residual incongruity represents
[ . . . ] the margin of inadequacy of the cognitive rule and tends to be
ignored” (p. 59).
4.2.1.5 Oring’s appropriate incongruity
Elliott Oring is a folklorist (and as a non-psychologist in this section,
he automatically gets bonus points). He has long maintained that the

86
 4.2 RESOLUTION

resolution of jokes is “the perception of an appropriate relationship

between categories that would ordinarily be regarded as incongruous”
(Oring, 2003, p. 1). Oring argues that his approach is not homologous
to the incongruity–resolution theories insofar as no actual dissolution
of the incongruity is assumed: “the incongruity remains, even though
points of connection between the incongruous categories are discov-
ered” (p. 2). Further, his approach is not committed to a processing
order, implicit in the incongruity-then-resolution terminology of the
incongruity–resolution theories. He then adds that either the incon-
gruity or the appropriateness must be “illegitimate” or spurious (p. 6).
Whether appropriate spurious incongruity is really different than all
formulations of the incongruity–resolution theories is not a topic that
can be addressed presently. We will return to the issue of full or partial
resolution below.
4.2.1.6 The GTVH on resolution
The General Theory of Verbal Humor (Attardo and Raskin 1991) was
the first linguistic theory to address the issue of the resolution of the
humor, although we did not word it that way in the original 1991 paper.
We spoke instead of the “logical mechanism” to describe the “crudely
approximate logic of a paralogism” (p. 304), the “semblance of logic”
which included “faulty or cheating inferential processes” which would
not “withstand any close scrutiny” and “paralogisms instead of correct
syllogisms” (p. 304) and concluded that “a joke must provide a logical
or paralogical explanation of the absurdity or irreality it postulates”
(p. 307), while also noting that this was not a novel problem, but one that
has not been fully addressed, despite some of the discussions seen above.
The Logical Mechanisms went on to be the most controversial of the
components of the General Theory of Verbal Humor. We will turn to
that in Chapter 7. Instead here we will consider a few studies that have
investigated the general nature of the resolution, rather than its specific
mechanisms.

4.2.2 Full vs. partial resolution

In Hempelmann and Attardo (2011), we tackled the issue of full and
partial resolution. Based on Oring’s (2003) appropriate incongruity, we
argued that humor only partially resolves incongruities. The reader will
recall that Oring claims that humor only partially resolves incongruities

87
 INCONGRUIT Y AND RESOLUTION

and does so spuriously, at that. Whereas metaphors find real legitimate

connections between disparate (incongruous) objects (and have some
explanatory power because of that) jokes find spurious, illegitimate
connections. Oring then argues that the difference between jokes and
riddles is that in a riddle the incongruity is fully resolved, i.e., there is
a solution (2003, p. 6), whereas in a joke (or a joking riddle, a.k.a. a
pseudo-riddle) there is no full resolution. Attardo (2009) applied the
same line of reasoning to mystery stories arguing that once we find out
“who done it” (in other words, the incongruity in the text is resolved) the
incongruity is fully eliminated: consider the “locked-room” sub-genre
of mysteries: a crime is committed in a room or other location in which
only a restricted group of people are present or have access to, but they
all have an alibi. Once we find out that the butler or the husband did
it, and how he managed to do it while apparently also having an alibi,
the resolution is complete and the incongruity truly and fully resolved.
Once we know that the murders in rue Morgue were committed by an
orangutan, there is no mystery left.7
The idea of partial resolution is operationalized as “residual incon-
gruity” with a view toward quantification in Hempelmann & Ruch
(2005), as follows: “how much incongruity remains un(re)solved; how
much new incongruity has the LM introduced; how puzzled are you
still at the end of the joke” (p. 364). Their analysis of a sample of
cartoons showed that nonsensical cartoons contain significantly more
residual incongruity (p. 368), as one would expect. Samson et al. (2009,
p. 1029) shows significant differences in brain activation between the
two categories of jokes.

4.2.3 Foregrounded vs. backgrounded incongruities

Rothbart and Pien argued that (1977, pp. 37–38) a humorous text
may contain multiple incongruities and moreover the resolution of the
punch line itself may introduce new incongruities. Tsakona (2003)
established that many jokes have more than one punch line. The tech-
nical details of the analysis are reviewed in Section 7.1.6. What matters
presently is that this perforce means that there will be multiple, inde-
pendent script oppositions within a joke. This is not to say that all jokes

7 One of the earliest example of the genre is E. A. Poe, The Murders in the Rue Morgue
(1841).

88
 4.2 RESOLUTION

have multiple script oppositions, but only that some do. Specifically, in
her corpus, about 25% did not have multiple punch lines (p. 320).
Tsakona notes that the idea that jokes contain several “implausibili-
ties” or (background) incongruities is not novel, having been stated in
Sacks (1978, p. 258)) and discussed by Mulkay (1988, pp. 130–133),
Attardo (1994, pp. 304–307), and Attardo et al. (2002). Tsakona argues
that these implausibilities, do not go “unnoticed” as Sacks had argued,
but in fact help frame the joke as non-bona-fide communication, i.e.,
as being told in a humorous key. Finally, she notes, and this is very
significant, that “the script opposition introduced in the punch line of the
narrative joke is always different from (all) the other(s) that may appear in
the joke-text” (Tsakona, 2003, p. 326; emphasis in the original). It should
be noted that Raskin (1985, pp. 132–134) had already noted that there
are complex jokes, with more than one script opposition.
In Attardo et al. (2002) we suggested distinguishing between focal
and background incongruities. The focal incongruity is the one involved
in the punch line and in its resolution. In Hempelmann and Attardo
(2011)8 we consider the following variants of a joke:
(27) a. How does an elephant hide in a cherry tree? It paints its toenails
red (Hempelmann & Attardo, 2011, p. 132).
b. How does a monkey hide in a cherry tree? It paints its toenails
red (p. 133).
c. How does a monkey hide in a cherry tree? It paints its toenails
blue (p. 134).
d. How does an elephant hide in a lime tree? It paints its toenails
red (Oring, 1992, p. 22).9
and note that (27.a) contains an incongruity which (27.b) does not, and
namely that a very large, non-arboreal animal might have climbed a
relatively small tree and not crushed it under its weight. Both (27.a) and
(27.b) partially (very partially!) resolve the incongruity by an incon-
gruous means, i.e., painting their toenails to simulate the color of the

8 The paper was in fact written much earlier, around 2005, but due to the significant
backlog in HUMOR at the time and due to my desire not to appear partial to a former student
and to my own work, since I was then the editor-in-chief, there was a considerable delay in
the publication.
9 Oring does not actually present the joke per se, but notes that “the absurdity is rule-
governed ( . . . ) it would not do to answer that elephants paint their toenails red to hide in
lime trees.”

89
 INCONGRUIT Y AND RESOLUTION

fruit of a cherry tree and thus achieve some camouflage. This is very
much fantasy assimilation and suspension of disbelief. (27.c) removes
the color match which supposedly helps with the camouflage and thus
eliminates the resolution, as does (27.d) but by changing the type of tree
and hence the color of the fruit. Our point was that in (27.a) there is a
massive incongruity that is not addressed by the text, which chooses to
address a very marginal incongruity (the color of the toenails). So, we
would say that the fact that the elephant is in the tree is a backgrounded
incongruity, and as such can be removed from the jokes, as we did in
(27.b), whereas the coloring of the nails (using very large nail polish
brushes?) is the foregrounded incongruity. Finally, we argued that there
is an intermediate type of incongruity, which can be removed and
does not change the nature of the joke, as our change in (27.b) did.
A full discussion can be found in Hempelmann and Attardo (2011,
p. 135–137).
Samson and Hempelmann (2011) tested empirically Tsakona’s
(2003), Hempelmann & Ruch’s (2005), and Hempelmann & Attardo’s
(2011) claims that backgrounded incongruities affect both processing
and appreciation. They found that, as anticipated, backgrounded
incongruities especially affect the funniness rating: stimuli with back-
grounded incongruities were considered funnier (p. 173). “The
presence of a backgrounded incongruity [ . . . ] leads to more unresolved
incongruity and contributes to the perception that the cartoons are
more nonsensical” (p. 180). Samson and Hempelmann hypothesize that
the increase of perception of nonsense which parallels the increase of
backgrounded incongruities may possibly be due to the fact that “some
of the backgrounded incongruities remain unresolved but contribute
essentially to the emotional response” (p. 181).

4.3 Linear organization of the joke

The incongruity–resolution model has an interesting aspect which we

have not considered in detail yet, namely the fact that the idea of
introducing an incongruity and then resolving it presupposes a tempo-
ral linearity, as mentioned by Oring (see above, Section 4.2.1.5). The
isotopy–disjunction model has provided the most compelling analysis
of this aspect of humor and so we will discuss the linearity of humor
using the isotopy–disjunction model of jokes as an example.

90
 4.3 LINEAR ORGANIZATION OF THE JOKE

4.3.1 The isotopy–disjunction model

At its core the isotopy–disjunction model is the implementation of a
basic, and very interesting observation, namely that among the various
polysemic lexical items of a coherent sentence, there emerges a shared
thread of meanings that work with one another, while other meanings,
also potentially available in the lexical items, are not functional in
that given interpretation. This is of course the phenomenon known as
disambiguation. Greimas called this “thread” of compatible meanings
an “isotopy” (Greimas, 1966) and argued that a special type of semantic
features was responsible for its establishment. He labeled these classemes
and opposed them to the more specific semes. The history and diffi-
culties of establishing the boundary between semes and classemes are
detailed in Attardo (1994) and are not relevant in this context, beyond
the conclusion that the distinction was in fact untenable and that an
isotopy could be established merely by the repetition of any semantic
feature.
An example will clarify. Consider the following sentence:
(28) John poured the coffee in the cup.
It is clear that, whatever the polysemic nature of the various lexical
items in the sentence, the noun coffee has at least two possible senses: 1)
coffee beans and 2) a liquid obtained by infusing ground coffee beans
in hot water. It would be too complex to run a full feature analysis
of the two meanings, but we can simplify the analysis by saying that
meaning (1) will have a feature [+ solid] while meaning (2) will have a
feature [+ liquid].10 Greimas’ insight is that the occurrence of the feature
[+ liquid] in “pour” selects the meaning of “coffee” which also shares the
[+ liquid] feature. Presumably one could also have a feature analysis of
“cup” sophisticated enough to note that containers may hold liquids.
Greimas was never interested in humor per se and only analyzed a
joke to show how the setup of the text presented one isotopy which was
then replaced by a second one in the punch line. He introduced the
terms of connector (a polysemous term in which both isotopies could
occur simultaneously) and of disjunctor (the lexical item which forced
the separation of the two isotopies).

10 Any reader smart enough to note that we really only needed one feature [+/– solid],
for example, should also be smart enough to understand the pedagogical reasons why we
did not do so. Moreover, ground coffee is [– solid] and [– liquid]. Feature analysis is fun!

91
 INCONGRUIT Y AND RESOLUTION

Greimas’ model of jokes was adopted and developed by various

European scholars, primarily by Violette Morin (Morin, 1966), who
added a three-part division in narrative functions, inspired by Propp’s
work (1928) on the subject in folk tales. Because of considerable
variation in the terminology, in Attardo (1994, p. 87), I introduced a
simple numbering of the functions: F1, the introduction, F2, the setup,
and F3, the punch line. The Greimas/Propp/Morin isotopy model was
quite successful in Europe in the 1970s and 1980s and then faded away.
The success can be explained by two factors: 1) the formalization of
the concept of incongruity in the isotopy model, and 2) the focus on the
linear development of the text. The model ultimately failed to remain
productive due to several issues: first and foremost, feature analysis
was shown to be incapable of dealing with the full semantic analysis
of the lexicon. Once taken outside of subdomains such as kinship
terms or kitchen utensils, the methodology starts generating too many
features and any explanatory gains are nullified by the proliferation
of features. Whereas the phonetic inventory of a language could be
reduced to a few handfuls of features, the semantic inventory of a
language would require the same order of magnitude11 of features as
of lexical items. Second, the analysis in narrative features was shown
to be common to all narrative forms and not specific to jokes. Third,
the definitional problems of the concept of “isotopy” weakened the
formalization advantage. Finally, the appearance of semantic models
that were both more sophisticated, in that they adopted the concept of
semantic scripts/frames, and more elaborated, led scholars away from
the isotopy model.
The specific focus on the linearity of the text was not taken up by the
script/frame-based models and this explains why Bucaria (2004) still
managed to make a very meaningful contribution on the typology of
the connector/disjunctor placement using the isotopy model. Logically,
we expect the connector to precede the disjunctor or to coincide with
it.12 Consider this headline, from Bucaria’s data (p. 298):

11 You can think of orders of magnitude as zeroes that follow a number: so 50 and 70
have the same order of magnitude, as do 53 and 84, but 300, 999, and 157 have one order of
magnitude more, 1000 has two more, etc.
12 In the case of coinciding connector/disjunctor, we could also say that the connector is
absent and we just have a disjunctor.

92
 4.4 CONCLUSIONS

(29) MARCH PLANNED FOR NEXT AUGUST

the lexeme “March” is ambiguous (the name of the month vs. walking).
It is the connector. Assume that due to saliency the month interpretation
is preferred. When we reach the end of the headline, we encounter the
disjunctor (“August”) which forces the reinterpretation from “month” to
“walking.” Consider now the next example, also from her data,
(30) DRUNK GETS NINE MONTHS IN VIOLIN CASE
Here the connector co-occurs with the disjunctor as they both are
revealed in the word “case” (legal action vs. box for the safekeeping
of an instrument). Bucaria found examples in which the disjunctor
(“lingerie”) precedes the connector (“slip”), as in the following:
(31) LINGERIE SHIPMENT HIJACKED – THIEF GIVES POLICE
THE SLIP
The significance of this is that it contradicts Attardo et al. (1994) in
which we speculated that such a configuration was impossible. Bucaria
instead established that the order of presentation of the elements of the
joke is free.
4.3.1.1 The three functions
Tsakona (2003) showed that while the final position of the punch line
in jokes holds true, other incongruities may occur in the texts (back-
grounded incongruities). These may appear as non-final punch lines, or
jab lines; see Chapter 7. Tsakona (2003) has provided several significant
insights in the organization of the text of the joke. First, she showed that
F2 is significantly longer than F1, by a ratio of 3:1 in her sample. The
distribution of the jab lines in the text of the jokes is the same. Second,
and more significantly, the script opposition introduced in the punch
line is always different from those in the jab lines. It is a novel opposition
that cannot appear in either F1 or F2, although it will involve, as one of
the polarities, a script introduced in F1 or F2.

4.4 Conclusions

We have examined the incongruity and resolution theory of humor,

with a particular emphasis on the semantic nature of the concept of

93
 INCONGRUIT Y AND RESOLUTION

incongruity. We also emphasized the partial nature of the resolution

and its ludic, local, illegitimate aspect. We then focused on the lin-
ear organization of the presentation of the incongruity and its subse-
quent resolution, instantiated in the disjunctor and (optional) connec-
tor. Having established that the incongruity theories are based on a
semantic concept, we will move on to the properly linguistic treatment
of the semantics and pragmatics of humor in the second and following
parts of the book, but not before dealing with one last foundational issue,
the semiotics of humor, in Chapter 5.

4.5 Further readings

The historical roots of the incongruity–resolution theory can be further

explored in Morreall (1987) and Attardo (1994). On modern treatments
of incongruity, see Forabosco (1992). The rest of the discussion on
incongruity and especially resolution is technical and there are no
general sources, beyond those quoted in the text. Oring (2003, ch. 1)
is very accessible. Martin (2007) can be useful for the psychological side
of the discussion. On the isotopy–disjunction model, see Attardo 1994,
Chapter 2. For an update, see Aljared (2017).

94
 5
Semiotics of humor

5.1 Humor and semiotics 96

5.2 Conclusions 108
5.3 Further readings 109

This chapter consists of a discussion of the semiotics of humor.

Semiotics can be conceptualized as the science of how communication
takes place in general, within any system of communication, i.e., not
limited to language.
We start from the assumption, unspoken so far, that humor is a com-
municative event.1 In other words, humor takes place within a situation
in which one speaker is communicating with another speaker. The
reader will recall the basic humor situation we introduced in Chapter 1.
Of course much of humor is not expressed or communicated lin-
guistically. The prototypical humorous situation of a man slipping on
a banana peel has no linguistic component of note.2 Since the realm
of communication is much broader than language (e.g. non-verbal
communication, kinesics, proxemics, etc.) a broader approach becomes
necessary and will have to be provided within semiotics, i.e., the general
science that encompasses all forms of communication (intentional and
unintentional, linguistic, visual, musical, etc.).

1 Morreall (2009, p. 70) presents examples of humor that would not involve communi-
cation, such as oddly shaped rocks. I will not discuss these cases here, but I believe they can
be reduced to communicative situations anyway. We are taking the communicative nature
of humor as axiomatic, in other words. See Section 3.7.3 for a discussion of axioms.
2 One could argue, with Lotman, see Section 4.1.1, that since language is the primary
modeling system of a culture, we see the man slip through language, so to speak.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 SEMIOTICS OF HUMOR

From a linguistic/semiotic standpoint one can start from two

approaches at a definition of humor: a semantic (structural) definition
and a pragmatic (functional/contextual) definition. These will be
developed more fully in Parts 2 and 3 of the book, respectively. Here, we
will be satisfied with providing a general outline of how the definitions
work. Our goal is to show that the argument we propose is not limited
to language but in fact applies to any communicative situation.

5.1 Humor and semiotics

This chapter sets out to combine several strands of research to achieve a

general explanation of how the structural features of humor determine
its manifestations and usages. This objective is based on the general-
ization that the semantic and pragmatic mechanisms of humor are the
determining factors of the unique features of humorous communication.
In order to arrive at this conclusion, we must start by providing a
few definitions of semiotics. Furthermore, in order to investigate the
particular nature of humorous communication we need to introduce
some concepts, derived from semiotics and from linguistics, as well as
some theoretical approaches to humorous phenomena. The following
sections will provide these necessary background materials.

5.1.1 Communication and semiosis

Communication presupposes a system through which the exchange
of information is achieved, i.e. a semiotic system. All communication
is (a form of) semiosis. Semiosis is here defined as the exchange of
information through a semiotic system.3 It should be noted that there
can be exchange of information without a semiotic system, for example
by answering the question “What do you have in your briefcase?”
by opening it and showing the contents to the inquirer (this is what
Wittgenstein called “ostensive” communication). From the opposite
angle, there can be use of a semiotic system, without transmission of
information (cf. below “defunctionalization”).

3 It should be noted that we do not claim that all that is needed for communication is a
semiotic system. In order to communicate, you also need a pragmatics, in its broadest sense.

96
 5.1 HUMOR AND SEMIOTICS

5.1.2 Connotative semiotics

Saussure’s (1916) definition of the sign as the indissoluble unity of a
signifié and a signifiant (signified/signifier) was not meant to hold for
linguistic signs alone, but was meant to be extended to all semiotic
systems. The signifier of a sign is the mental representation of the phys-
ical support of the sign (sounds, graphical elements, bodily movements,
etc.), while the signified is the mental representation of the concept (the
idea, the meaning) conveyed by the sign.
Hjelmslev (1953) defines any system of signs as a semiotics. A deno-
tative semiotics is a system of signs in which neither the signifier nor the
signified are semiotics; hence languages are denotative semiotics, since
their signifiers are (mental representations of) sounds, and their signi-
fieds are (mental representations of) concepts. A connotative semiotics,
on the other hand, is a semiotics in which the signifier is a semiotics.
If a semiotics has as its signified another semiotics it is a metasemiotics
(i.e., it is a semiotics of another semiotics; an example would be the met-
alanguage of linguistic description). But if a semiotics has as its signifier
another semiotics it is a connotative semiotics. As an example consider
a person speaking in Italian saying “Mamma mia.” This is a semiotic
sign, made up of a signifier [mam:amiya] and a signified “Oh my!”
(literally: “mother mine”). However, this entire semiotic fact can become
the signifier of another semiotic fact, that is the entire situation connotes
“Italian.” Schematically, we have the representation in Figure 5.1.
This fact has some very important consequences for the study of
humor, which will appear in full in the rest of the discussion. For
the time being, let us be satisfied with the observation that since the
signifier of a semiotics does not carry any “meaning” in and of itself,
the original meaning of “mamma mia” is lost, and only the meaning of
“Italian” survives the establishment of the connotative semiotics (just as

Connotative Semiotics

Expression “Mamma mia !” Content “Italian”

Expression: Content :
[mam:amiya] “oh, my !”

Denotative semiotics

Figure 5.1 Denotative (bottom left) and connotative (top row) semiotics.

97
 SEMIOTICS OF HUMOR

for any non-Italian-speaking speaker of English the string [mam:amiya]

is meaningless, with the exception of its identification of the speaker
as Italian). In other words, when a sign is used as the signifier of
a connotative semiotics, its primary meaning as a sign is lost, and
the connotative semiotics acquires a secondary meaning, proper to
the connotative semiotics, and independent from the primary meaning.

5.1.3 Perlocutionary definition of humor

The only definition of a humorous text4 general enough to apply to all
humor is that of a text whose perlocutionary5 goal is to be perceived
as humorous. This definition (Attardo & Chabanne, 1992) stresses
the intentions of the speaker and/or the hearer. More to the point, a
definition of humor based on its goal implies that the text will use all
its resources to achieve the perlocutionary goal of the speakers, to the
detriment of any other purposes the text may have had beyond that of
being perceived as humorous. In this perspective, the only thing that
matters for the speaker is to achieve his/her goal, and the text will be
shaped accordingly.
This fact is grounded in the analysis of humor as a connotative
semiotics: the humorous text can be shaped in any way, as long as the
perlocutionary goal is attained, just as any string of Italian words would
have worked as expression of the signified “Italian.” Since any Italian
words would have worked to convey the meaning “Italian” even the
following sentences would convey the meaning “Italian,” as long as the
audience can reliably identify them as being spoken in Italian:
(32) “Non sono italiano” (I am not Italian)
“Scusi, sa dov’é il bagno?” (Excuse me, do you know where the
bathroom is?)
From this it follows that the original content of the semiotics (i.e., the
text in the humor situation) is backgrounded, just as the theory of
connotative semiotics would predict. In the case of humorous texts, this
means that the actual contents of a joke may be completely irrelevant
to the context in which they occur, as long as the humorous text itself
is pertinent to the context: a joke may be about chickens crossing the
4 In the generic sense established in Chapter 1.
5 This term will be better defined in Section 8.1.1. For the time being, “a goal of the
speaker” is a sufficient definition.

98
 5.1 HUMOR AND SEMIOTICS

road, elephants in trees, or whatever, but the joke must occur in a given
context, following certain rules, etc. Consider again example (10), in
Chapter 1, in which Carmen recontextualizes an old joke (What do you
call someone who speaks only one language? An American.) to playfully
tease Marina. Carmen could just as well have asked if she spoke Greek
or Chinese, so that the conversation could have gone as follows:
(33) Carmen: Do you speak any other languages, besides English?
Marina: Nope, nothing fluently, a little bit of that and . . .
Carmen: (Ironically) You don’t speak Greek, or Chinese?
In other words, a different joke, which is pertinent to the context
and achieves the same perlocutionary goals of the speaker, could have
replaced the original joke. However, crucially, the joke needs to be
delivered as a response to Marina’s modest turn in which she admits,
somewhat hesitantly, to monolingualism.
This explains, among other things, why non-sense and absurdities are
tolerated within humorous texts: the absurdities are at the level of the
expression of the connotative semiotics, and not at the level of the con-
tents of the connotative semiotics, where they would not be tolerated.
Carmen could not mock Marina’s monolingualism at the beginning of
the conversation, for example, because this would violate the relevance
of the turn (humor must be à propos). The humorous text can take any
form, that is including breaking the rules of communication (non-sense,
absurd), as long as the result is perceived as humorous; if the result is
not perceived as humorous the absurdities are not “redeemed” by their
functional use within the joke text. If Carmen started the conversation
mocking Marina’s monolingualism, the conversational move would be
perceived as extremely aggressive and thus unfunny.
Likewise, Carmen need not be seriously committed to the claim that
monolingualism is a flaw at all and that it is particularly common
among Americans. Once more, one can assert propositions one does not
believe to be true, thus violating several rules of communication, for the
perlocutionary purpose of being humorous.

5.1.4 Defunctionalization of the sign

The fact that humor is a connotative semiotics does not imply that
the humorous text does not have any feature beyond its perlocutionary
goal. In order to discuss this issue it is necessary to first consider the

99
 SEMIOTICS OF HUMOR

semantic mechanisms at work in humor. The reader should bear in

mind that the following is a cursory presentation, which introduces only
the essentials required to understand the discussion of the relationship
between semantics and semiotics of humor. A full presentation will be
undertaken in Chapter 6.
5.1.4.1 The semantics of humor
One of the most theoretically significant contributions of linguistics to
humor theory has been establishing the fact that the necessary and suf-
ficient conditions for humor are semantic (see Chapter 6). The semantic
theory of humor begins by distinguishing between humor competence,
an abstract highly idealized model of humor, and the actual performance
of a given humorous text by given speakers, in a given situation, con-
text, etc. (see Chapter 1). Humor competence abstracts away from all
individual idiosyncrasies, such as the fact that a speaker may consider
certain types of humor offensive, or that a speaker has had a bad day,
and is in a bad mood, etc. By abstracting away from all performance
considerations, the semantic theory of humor is capable of reducing
all humor to a relational function (opposition between overlapping
scripts) with minimal positive semantic weight, but with well defined
requirements, namely the relation of opposition of the scripts involved
in the joke. In other words, any two scripts which are opposed (in a
technical sense), and overlap in a text (i.e., the text is compatible with
both scripts, at least in part), can potentially generate humor. Whether
the overlap of two opposed scripts will be perceived as humorous by
a given audience in a given situation, at a given time, etc. is not a
concern for the semantic theory of humor (competence), but becomes
one for any socio-pragmatic theory of humor. For a full discussion of the
competence/performance opposition see Section 2.1.
5.1.4.2 Semantics and defunctionalization of humor
As noted above, the semantic theory of humor has very little to say about
the contents of the script opposition, since it is a relational concept, and
not a substantial one. While this may seem a paradoxical or idiosyncratic
conclusion, it is mirrored by another important conclusion reached
in humor research, and namely the loss of meaning of the signs used
in humorous texts. This fact has been noted across the literature in
the field, although not always in as clear terms as Guiraud’s (1979)
“defunctionalization” of the sign. Guiraud’s point was made in the

100
 5.1 HUMOR AND SEMIOTICS

context of the discussion of puns, but the reasoning behind it allows its
generalization to all humor. When we communicate we use language
according to its primary function, i.e., the transmission of information.
When this primary function of language is abandoned in view of some
other non-linguistic goal such as play, or aesthetic purposes, we have a
defunctionalization of the medium (language). Jakobson’s (1960/1987)
definition of the aesthetic function of language as centered on language
itself 6 also fits the definition of defunctionalization. In other words, any
time that the linguistic medium is used for purposes other than the
transmission of information (and possibly other non-communicative
mental processes, such as thought) we can speak of defunctionalization.
In the context of humor, the idea of defunctionalization is particularly
interesting, since it is a well-known fact that humor is considered “non-
serious” or otherwise devoid of commitment to any utility principle,
beyond the fulfillment of the humorous purpose of the text. For example,
nobody takes seriously the information, provided in the following joke
(34) How do you fit four elephants in a compact car?
Two in front and two in the back.
All theories of humor, even the more committed to the “seriousness”
and the social and/or interpersonal relevance of humor have to grant
that humor has an element of playfulness, of lack of practical goals.7 This
defunctionalization is reflected in the “semiotic” nature of the humorous
text. The important issue, from the point of view of the participants in
the joke exchange is not the contents of the joke text, but whether the
text is funny. In other words, jokes are defunctionalized texts. The joke
text exhausts its function in its perlocutionary goal (trying to make the
hearer laugh as a result of the feeling of mirth).
Composed of a signifier and a signified, like any other text, the humor-
ous text transcends both, and connotes only “humor” beyond the pri-
mary meaning of the signs used in the text, since as we have said, the
primary goal of the text is not to convey information, but to amuse. In
simple terms, it doesn’t matter much about the chicken and the road,
what matters is that it is funny.

6 Jakobson’s definition is: “The poetic function projects the principle of equivalence
from the axis of selection into the axis of combination” (Jakobson, 1960, p. 71), i.e., that
the combination of linguistic elements to form sentences is ruled by the similarities of the
various linguistic items, rather than their meaning, as is usual.
7 Cf. the definition of humor as a “paratelic” mode (Apter, 1989).

101
 SEMIOTICS OF HUMOR

This is not to say that the joke communicates only its own
humorousness. The “reuse” of the semiotics is hardly “transparent”:
consider making a joke about a political figure. Especially if the joke
ridicules or mocks the politician, the choice of what politician to mock,
what to make fun of, etc. will be hardly irrelevant. One’s choices will be
dictated by one’s opinions, attitudes, likes, dislikes, etc. In other words,
the choice of a given subject matter becomes part of the joke’s potential
meaning/communicative import. We will return to this issue below, but
for the time being let us set it aside, and return to the consequences of
the defunctionalization of humorous texts.
5.1.4.3 Metamessages and metasemiotics
Compare now the situation of defunctionalization described above with
the definition of connotative semiotics, also given above, and it appears
that the humorous text8 is a connotative semiotics, whose signified is
“this text is funny” and whose signifier is the text of the joke (which
in turn is made of a given signifier and signified). In other words, a
humorous text is a connotative semiotics, connoting humorousness.
This ties in with Bateson’s remarks about the metamessage “this is
play” present in play, discussed in Section 3.3.1. The various disclaimers
and exploratory introductory turns studied in conversation analysis (see
Chapter 11) perform the same function of marking in discourse. The
issue of “markers” of humor will be taken up in Part III of the book.
Bateson’s original observation can be expanded to the social setting
of the interaction in which the joke text is produced. By uttering a
joke, say at a social gathering, one not only expresses the metamessage
“this is play” (regardless of explicitly doing so), but also a number
of inferences can be drawn: thus “this is play” implies “I think it is
appropriate to joke here, now, about this subject,” “I am in the mood to
joke,” etc. Thus, the utterance of a joke will inform its audience about the
teller, his/her perception of the context, etc. Even in the case of printed
jokes9 the text will reveal that the author or editor considers these texts
funny, appropriate for printing, etc. Imagine the differences between

8 The reader will recall that by the definition in Chapter 1 a text is anything used to
convey meaning regardless of its semiotic modality (spoken language, image, gesture, music,
etc.).
9 Printed collections of jokes are the “zero degree” of context for jokes, as no contextual
information beyond the title of the book, the author’s name, the publisher, and the date of
publication, is available.

102
 5.1 HUMOR AND SEMIOTICS

a collection of jokes edited by a Baptist minister, and a collection of

neo-Nazi jokes on the 4chan bulletin board.
Since humor is a form of play, it is logical to extend the idea of the
presence of metamessages to humor. These metamessages will alert the
speakers that the humorous exchange is “play/humor.” These consid-
erations set the necessary background for all the issues of suspension
of the rules (linguistic, semiotic, social), since, as we know from the
theory of metalanguages, metasemiotic status guarantees the suspension
of the rules of the object-semiotics. This accounts for a number of facts
we know about humor: humor carries metamessages saying “this is
humor,” humor violates semantic and pragmatic rules (see Chapter 8)
but this fact does not cause the communicative situation to be destroyed
(Attardo, 1993). Most importantly, it accounts for the baffling fact that
humor can “mean” different things to different people. In fact, beyond
“this is humor,” the signified (i.e., the meaning) of the metasemiotics
is empty; it is only the meanings of the primary semiotics which may
“surface” in the final overall communicative import of the text and thus
give an effect of meaningfulness, apropos, etc., to jokes.

5.1.5 Semantics vs. semiotics of humor

The previous sections have established the fact that humor is a connota-
tive semiotics, and have pointed out the direct consequences of this fact:
the defunctionalization of the humorous text and the metamessages that
a humorous utterance carries. It is now possible to explore another facet
of the relationship between the semantics and the semiotics of humor.
From semiotics, we know that we have to deal with any number
of systems based on the signifier/signified opposition. The linguistic
semiotic system is just one among many (however primary it may
be) and therefore we have to create a model where there are as many
different signifiers, which are, however, all opposed to one signified,
since by the principle of intersemiotic translation (Jakobson 1960),
we can vary the signifier and keep the signified constant. Or in other
words, as we can “translate” the signified “beautiful” across languages
(beau, bello, schön),10 we can also, and with the same ease, translate the
signified into a drawing, a motion-picture, gestures, etc. The relationship
form/expression in a humorous text will then look like this:

10 Albeit with a certain variation in the exact correspondences between the terms.

103
 SEMIOTICS OF HUMOR

verbal 
figurative signifier / humorous signified
musical
etc.

From this fact it follows that, since the semantics of the humorous text
remain unchanged through all the translations across semiotic systems,
the semantic requirements should be applicable, with all due changes, to
all semiotic humor. The analysis of the following examples will address
this issue.
Let us consider two non-verbal examples: a clown’s oversized shoes,
and a well-known 3-frame cartoon by Gary Larson showing four cows
standing on their hind legs by the side of the road, one of which is
shouting “Car!” in the first frame, a car with two people aboard going
by the cows who are now standing on their four legs, in the second, and
the cows standing again on their hind legs in the third and final frame.
There is no caption, and the details of the landscape/setting are bare and
minimal.
I think that it is uncontroversial that the first example has no ver-
bal element whatsoever, while the second example use of language
is marginal, and does not seem to be the source of the humor of
the cartoon. A fair paraphrase (intersemiotic translation) of the first
example would be: “The clown is wearing shoes that are way too big
for his/her feet.” A semantic theory of humor would account for the
humorous potential of the example by pointing out that the script (i.e.,
what we know11 ) for “shoes” prescribes a) that they should fit the feet of
the wearer within reasonable standards, and b) that no shoes are made
beyond a certain foot-size (this may be seen as an inference derivable
from a) and the script for “foot”). Thus what we see is incompatible with
our script for shoes, and yet we see it used as if they were normal shoes
(i.e., they are worn, and the clown walks around, etc.). In other words,
the script shoes is both incompatible and overlapping with the script
clown shoes or its actualization in that specific pair of shoes. This is
a very primitive form of humor, and in fact it may be that some of its
constituents are missing, namely we have located an incongruity (size)
between the script shoes and the actual shoes we see, but we cannot

11 A full definition of script is found in Chapter 6.

104
 5.1 HUMOR AND SEMIOTICS

find a resolution of this incongruity, which many researchers consider

an important part of the humorous experience.
Let us turn to the second example, which is more complex. A para-
phrase of the cartoon might be: there are some cows that are standing
on their hind legs, and one of them shouts “car!”; they stand on their
four legs while the car goes by, and when the car is gone they resume
their standing position.
The first thing that strikes the reader, even before he/she starts pro-
cessing the cartoon, is that the cows are anthropomorphized: they speak
and they stand. Here we already have the basic opposition between the
normal script cow and what we see represented in the cartoon human-
acting-cow. But the cartoon also provides the “local logic” (4.2), i.e.,
the resolution of the incongruity.
Two inferences are central to the perception of the humor of the
cartoon. Why is the cow in the first frame shouting “Car!”? Shouting
may be caused by a number of causes, fear, rage, emotion, etc., but also
by the desire of being heard at a great distance. From the features of the
drawing there seems to be no particular emotional involvement from
the cow’s part (these would be represented by large round eyes, lines on
the face representing tension of the muscles, etc.) and hence we can draw
the conclusion that the cow is shouting in order for the other cows to be
aware that the car is coming, or in other words, that the cow warns the
others of an approaching car. We may also note that the four cows in the
drawing are not all standing together: the cow that shouts is standing
by itself, somewhat distant from the other three who are looking at
the shouting cow. This may mean that the cow is “on the look out,”
but the drawings do not provide enough evidence to argue this point
definitely. Nevertheless, if someone, not just a cow, warns someone else
of something this means that at least one of them is interested in that
something. The drawing does not give any direct hint at why the first cow
may feel the need to alert the other cows of the presence of the car, but if
we look at the two other frames we quickly realize that when the humans
are in the picture (quite literally) the cows act like the script cow, and
that when the cows are alone, they act in the script human-acting-
cow. Here we may apply a general rule: if someone acts in a given way
only when in front of someone else he/she is adapting his/her behavior
to the person in question (note that this is not a necessary conclusion, it
is an abductive, probabilistic step, which may or may not turn out to be
correct). Why would then cows that act like humans not act like humans

105
 SEMIOTICS OF HUMOR

when humans are around? Here again we can jump to the conclusion
that they are trying to hide from the humans their behavior.
In conclusion: by a rather complex, and largely abductive process
of inference we conclude that the cows are hiding from humans that
they can talk and stand: this explains why the cow in the first frame is
shouting “car!” and why the cows change their posture in the second
frame. Thus we have provided a resolution of the incongruity: the reason
we have never seen a cow speaking and standing on its hind legs is
because the cows are purposely hiding these feats from us!
What is remarkable about this process of progressive inference mak-
ing is how closely it resembles and parallels the processing of verbalized
jokes. Certainly there are differences, the most significant being that
texts are processed with rigid linearity, whereas the processing of a
figurative sign is entirely different, and governed by complex rules based
in part on the nature of the sign itself. Shapes and masses in a picture will
“guide” the scanning of the image by the observer, as well as the expec-
tations of the viewer. Our perception of the stimulus may be altered by
the stimulus itself, to take an extreme example. This is the case with the
well-known optical illusions of Mueller-Lyer.12 Furthermore, different
semiotic systems vary in terms of analyticity and adeptness at dealing
with some types of information. Language is very good at dealing with
abstract entities, whereas it handles rather poorly such Gestalten as facial
expressions, which drawings or pictures can reproduce more easily (it is
no coincidence that we talk of a “Gioconda smile”). A puzzled expression
can be expressed much more effectively by a few lines in a drawing than
by the hopelessly underdetermined phrase “puzzled expression.”
However great the differences between the various semiotic systems
(we have only considered two, fashion and graphical, but the considera-
tions could be expanded without theoretical problems to other systems,
though not to all, see below), there are important parallels: in order to
process a humorous text13 we must perceive the presence of two opposed
scripts in the text, and somehow resolve the contrast between the two
scripts, even if in playful local logic terms.14 Naturally, the farther away

12 In this example, two lines of the same length appear to be of different lengths because
of other lines which touch them.
13 Once more, “text” here refers to any cluster of semiotic signs, independently of their
nature, and not limited to linguistic signs in any way, as we saw in Chapter 1.
14 The reader should keep in mind that some humor, such as absurd or nonsense humor,
may not have any resolution at all.

106
 5.1 HUMOR AND SEMIOTICS

from linguistic signs we move, the more strained the extension of the
concept of script becomes, without however, losing the specificity of a
baggage of information that the speaker carries about something (be it
a lexical item, a proxemic perception, or a figurative element).
The considerations above lead us to the formulation of a significant
generalization: all humor in whatever semiotic manifestation is based on
a semantic mechanism, i.e., all humor is semantic.
The semantic hard-core of all humorous facts does not mean that
issues at the level of the signifier are irrelevant to a semiotics of humor.
In fact the contrary is quite true. Most of the available materials on the
semiotics of humor deal with mechanisms of the signifier, although they
all are based (implicitly, to be sure) on the assumption of a semantic base
for humor. We will turn now to some of these issues, in order to gather
further evidence of common humorous mechanisms across semiotic
systems.
In fact the concept of “semiotic pun” has been elaborated, and is
central to the research of Lessard (1991, and references therein). Lessard
shows very convincingly that the same mechanisms that apply to verbal
puns apply to visual puns. Thus Lessard analyzes graphical text in which
one graphical element is taken to refer to two different things at the
same time and he correctly analyzes this fact as the graphical analog
of the lexical ambiguity pun. Lessard’s example is a drawing showing
on the top left part a man marooned on a small island with concentric
circular waves, which become on the right bottom part of the drawing
the furrows in the field of a puzzled looking farmer. If we consider now
a linguistic example, the old pun:
(35) Why did the cookie cry?
Its mother had been a wafer so long [a wafer/away for]
based on the equivalence of the two strings ‘away for’ and ‘a wafer’ in
pronounciation, we can draw the following proportion:

furrows : waves :: a wafer : away for

which shows plainly that the same mechanisms are at stake. As in the
linguistic pun we take one string of sounds to stand for another meaning
than the one intended in the text at first, in the semiotic pun we take one
semiotic element to stand for another meaning than the one intended in
the semiotic “text” at first.

107
 SEMIOTICS OF HUMOR

If we consider the issue from a linguistic standpoint, we will see that

all semantic differences in the system can be exploited for humor. For
example, in Chapter 9, we spend some significant effort cataloguing the
various levels at which a semantic ambiguity may occur: phonological
(puns), morphological, lexical, syntactic, and even pragmatic. This can
be stated as another significant generalization: All meaningful aspects of
language may be exploited for humorous purposes. To put it differently, if
it’s meaningful, we can use it for humor.15
However meaningful and convincing these successes of the cross-
semiotic analysis of humor are, they should not give the impression
that the task left to researchers is a mere cataloging. In fact, entire
semiotic systems elude this kind of taxonomy. Take the example of
music, too abstract to have a referent in the sense linguistic signs have
a referent, and therefore impossible to account for in terms of scripts,
or worse in terms of ambiguity. Yet, we have musical humor, even if it
takes particular forms and manifests itself mainly by surprise or allu-
sion (i.e., intertextual reference). Thus the Andante movement Haydn’s
Symphony No. 94 in G Major contains a sudden isolated chord (in the
16th measure) in the middle of a passage where the instruments had
been quite soft for the previous six measures. This is then an example of
the humorous use of surprise; see Scher (1991: 116) for the relevant score
and some comments. Mozart’s famous parodic piece Ein Musikalischer
Spass [K 522] mocks the fashionable music of the time, by quoting
and distorting it (an example of humor achieved though parody or
allusion).16

5.2 Conclusions

This chapter has reached several important conclusions: the mecha-

nisms of humor are the same in linguistic and non-linguistic com-
munication. In all cases, humor is based on an incongruity (violation
of expectations) and its resolution (albeit within the limits discussed
15 Incidentally, this is why taxonomic approaches to humor are doomed to failure:
because language, due to its creativity, always exceeds any taxonomy.
16 An interesting passage from Sadie (1965: 118) confirms the claims in the text:

Much of the broad humor of the Musical Joke is lost today, except perhaps on musicologists
who know what bad eighteenth-century is like or on teachers of composition who know the
elementary faults. To the normal music-lover, the joke is likely to misfire.

108
 5.3 FURTHER READINGS

above). The meaning of the humorous expression is backgrounded,

in favor of the connotative semiotic’s meaning “this is humor.” The
normal rules of communication no longer apply within a connotative
semiotics; in particular, the signs are defunctionalized, i.e., they are
no longer used for communicative purposes. This in turn has the
effect of making humor “retractable,” i.e., it explains all the pragmatic
uses of humor as politeness, as a testing mechanism, and as a way to
state something without committing to its truth or functional value.
Furthermore, humor contains the meta-message “this is play,” which
reinforces the defunctionalization and suspension of the normal rules
of communication.

5.3 Further readings

There are no introductory treatments of the semiotics of humor. Attardo

(1994) remains, despite the years, the best place to get started. On visual
puns, Hempelmann and Samson’s work (2007, 2008) is accessible. On
musical humor, see the entries on music and humor in Attardo (2014).
Intersemiotic translation is taken up again in Chapter 15.

109
 6
The semantics of humor

6.1 The Semantic Script Theory 113

6.2 The Semantic Script Theory of Humor’s two conditions 127
6.3 Non-bona-fide 133
6.4 The Ontological Semantics Theory of Humor 134
6.5 Conclusions 134
6.6 Further readings 135

In a certain sense, seen in Chapter 5, the essence of humor is semantic.

Therefore this chapter is quite central, in a way that other chapters are
not. This chapter will explain and motivate this claim.
Historically, most of the research on humor in linguistics did not
focus on the semantics of humor, but rather primarily on phon-
ology, morphology, and marginally on syntax. This attitude is well
represented in Pepicello and Weisberg (1983) which presents a largely
non-semantic synthesis of the “linguistics” of humor. Needless to say,
other approaches, especially in the incongruity/resolution tradition (see
Chapter 4), were much closer to the semantics of humor. For example
Koestler’s theory, discussed in Section 3.7.2.

6.1 The Semantic Script Theory

The Semantic-Script Theory of Humor is widely recognized to have

been one of the most successful theories of humor, since its introduc-
tion. Consider the begrudging acknowledgments of some of its critics:
“The General Theory of Verbal Humor (GTVH) has dominated the

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 THE SEMANTICS OF HUMOR

discussion of humor theory for more than thirty years” (Oring, 2018);
“Raskin’s Semantic Mechanism of Humor (1985), which develops a
General Theory of Verbal Humor (GTVH), has been the most influ-
ential work in humor research in its recent history” (Rutter, 1997,
p. 21); the “two most influential linguistic humor theories of the last
two decades, the Semantic Script Theory of Humor and the [General
Theory of Verbal Humor]” (Brône et al., 2006, p. 203); “the most influ-
ential approach to humor at present is that represented by the semantic
theories of Victor Raskin (1985)” (Smith, 2009, pp. 10–11).1
This is not the place to investigate the reasons for it success, which
would require a study in the sociology of science, and since the 1985
book in which Raskin presents it in detail2 is quite accessible, we will
content ourselves with restating the two conditions that must obtain for
a text to be humorous and then focus primarily on the discussion of
various issues that are often overlooked and misunderstandings of the
theory.
The Semantic Script Theory of Humor’s stated goal is to provide the
necessary and sufficient conditions for a text to be humorous. This
makes the theory essentialist. The Semantic Script Theory of Humor
is reductionist as well, as we will see below.

6.1.1 Origins
As its name makes patently clear, the Semantic Script Theory of Humor
is based on semantic scripts. The script-based approach to semantics
originated in psychology, with Bartlett’s views on memory (1932) and
the Gestalt psychologists (cf. Anderson & Pearson 1984). It was later
taken up in Artificial Intelligence (AI) (Minsky, 1974; Schank, 1975;
Schank & Abelson, 1977), and eventually arrived into linguistics (Chafe,
1975; Raskin, 1981, 1985; Fillmore, 1982, 1985). Within linguistics some
connections between frame/script semantics and other areas of lin-
guistics emerge forcefully: field semantics, a.k.a. lexical field semantics
(going back to Trier’s work in 1931; Lehrer, 1974; Lehrer & Kittay, 1992)
has some interesting correlations with frame/script semantics, as does
case grammar (unsurprisingly, since Fillmore was largely responsible for

1 With enemies likes these, who needs friends?

2 For the historically minded, the Semantic Script Theory of Humor was first presented
in 1979 in a Berkeley Linguistics Society paper and in 1986 in a Psychology Today article,
which contributed significantly to its popularity.

114
 6.1 THE SEMANTIC SCRIPT THEORY

it as well), and of course the psychological work on prototype theory

(Rosch & Mervis, 1975) which will later be one of the principal sources
of the cognitive linguistics approach.
An important point that bears being made explicitly is that because of
their foundations in psychology, script/frame theories have an impor-
tant stake in psychological reality. This is not the place for a review of
the literature on the psycholinguistics of text processing and reading
which makes heavy use of script/frame theories. Such a review would
require a separate monograph. However, let us note, for example, that
Rumelhart’s influential (1980) model comprises the simultaneous pro-
cessing of semantic, syntactic and pragmatic information, as does Just
and Carpenter’s (1980). In Attardo (2001a) I also noted the parallels and
mutual support of current psycholinguistic theories of text processing
(e.g., Kintsch 1998). Both major approaches to script/frame semantics
evolved in a formal/computational direction: script-based semantics
(Raskin, 1985) eventually led to ontological semantics (Nirenburg &
Raskin, 2004); frame semantics evolved into FrameNet (Ruppenhofer
et al., 2016).
This diversity of disciplinary sources is reflected in the wide termino-
logical variety (frames, scripts, schemata, scenes, memory organization
packets (MOP), etc.) used in the field. The reaction has been to gener-
alize and standardize around a single term (Fillmore’s (1985) “frame,”
Raskin’s (1981) “script”), leaving open the possibility of further differ-
entiation. The terminological discussion can be found in Andor (1985:
212–213) and Fillmore (1985: 223n).3 By and large, frame semantics has
been the more successful term in semantics and the study of reading,
whereas script semantics has been the default term in humor research.
The two terms, as defined by Fillmore and Raskin, are synonymous.
In the context of humor research we will primarily deal with the script-
based semantics of Raskin, because, as we mentioned, it was the most
successful semantic model in the linguistics of humor, and it is the most
developed and the broadest of all the linguistic approaches.

3 A very common misunderstanding is that there is a difference between, say, a script

and a frame: since both Fillmore and Raskin argue for the use of the respective term as an
umbrella term, by definition, the two are synonymous. Special terminological conventions
may be adopted by individual scholars, but they do not apply to Fillmore and/or Raskin.

115
 THE SEMANTICS OF HUMOR

6.1.2 The notion of script

A script is defined as a related, organized cluster of nodes and their
connections in the semantic network4 that represents the global web
of knowledge of the speakers of a language. Raskin’s exact words are
“a limited domain of the continuous semantic graph” (1985, p. 84);
elsewhere he describes it as “a large chunk of semantic information
surrounding the word or evoked by it” (p. 81) and, technically, as the
domain in the semantic graph “around the word in question as the
central node of the domain” (p. 81).
All the scripts of a language are connected to one another and form
a “a single continuous graph” (p. 81) which represents the sum total of
knowledge of the speakers of a language. A good way to conceptualize
this is to imagine the Oxford English Dictionary as a graph, with each
word linked to the words that make up the definition and each word
in the definition linked up to the words that make up their definition,
and so on. To use a different metaphor, in the age of the internet,
each script can be considered the equivalent of a web page. Each web
page is connected to other web pages (albeit haphazardly, whereas in
the semantic graph the connections are systematic and meaningful).
The organization of knowledge as a semantic network was prefigured
by Peirce (1931–36; see Eco, 1979, pp. 26–49) and introduced into
AI by Quillian (1967). The endless nature of the semiotic relation
(whereby a sign’s meaning is represented within the network by other
signs, ad infinitum) has been dubbed “unlimited semiosis” (Eco, 1979,
pp. 44–46) or “infinite semiosis” (Atkin, 2013).
Despite the technical terminology it should not escape our attention
that scripts are, setting aside the technical definitions, essentially ideas,
thoughts, or meanings. There has been a certain tendency to reify the
concept, as if it did have an objective nature, external to the minds of
those who think the thoughts. However, this is not the case. Meanings
exist as parts of signs; the other side of signs is of course the signifiers.
A sign is a social fact for Saussure, following Durkheim. Hence, even if
indirectly, the meanings (signifieds) exist in the minds of the speakers
who constitute the social group that legitimizes the social fact. So, in
conclusion, scripts are concepts, ideas, thoughts. These exist because
there is a social group that thinks the same thoughts (i.e., they are social

4 The technical mathematical term for this is “graph.”

116
 6.1 THE SEMANTIC SCRIPT THEORY

facts), as can be evinced by the fact that they function as a social group: if
they did not share the same thoughts they could not function as a group
(cf. Davidson, 1984, p. 153).
6.1.2.1 Analysis of the definition
Let us consider in more detail each of the elements of the definition of
script above. Needless to say, this is a very quick, informal review. Each
of the topics that we will take up below could be developed much more,
but this is only an introductory treatment. The interested reader should
consult Raskin (1985) and references therein. Occasionally, I elaborate
on a concept beyond Raskin’s original definition. For those, see Attardo
(2020).

6.1.2.1.1 Nodes Each node corresponds to a concept that has been

lexicalized in a given language. In other words, we are considering in
this definition only lexical scripts.5 There are larger scripts which can
represent sentences, paragraphs, and even larger texts; those will be dis-
cussed below. Furthermore, scripts can be activated without a lexematic
handle, i.e., inferentially. We discuss this in Section 6.1.3.1, below.

6.1.2.1.2 Connections Links among nodes correspond to the kind of

connections found among lexemes such as

1. synonymy, antonymy
2. hyponymy (ISA hierarchies, see below), hyperonymy
(superordinate)
3. partonomy/partonymy (a.k.a. meronymy), i.e., the parts of an
object
4. homophony, homography, homonymy
5. lexical functions (Mel’čuk, 1981).

Lexical functions may not be familiar to the beginning reader, so we

will exemplify. A lexical function is a systematic relationship between
two sets of lexical items. For example, the function “magn” indicates a

5 Grammaticalized concepts can also be seen as scripts, albeit more abstract ones.
Whereas lexical scripts have ‘handles’ (so to speak), i.e., are associated to and directly
activable by a lexeme, grammatical scripts would be activated by any of the allomorphs of
the given grammatical morpheme. Lexical scripts rarely have allomorphs, hence they are
more concrete than grammatical scripts.

117
 THE SEMANTICS OF HUMOR

great degree of intensity. So the following pairs of words are connected

by the lexical function “magn.”
(36) speak // shout
eat // devour
punch // pummell
Hence, “shout” is a high degree of intensity “speak,” etc. The function
“son” indicates the typical sound produced by an entity, so again, the
following pairs of words are connected by the lexical function “son”:
(37) leaves // rustle
dog // bark
thunder // rumble
In other words, the typical sound of leaves is rustling, dogs bark,
and thunder rumbles. Many lexical functions are quite similar to bet-
ter known lexical relations, such as antonymy and synonymy, but the
breadth of Mel’čuk’s functions, which catalog over 50 relationships,
make them an important part of semantic description.6

6.1.2.1.3 Semantic network The sum total of the connections among

nodes is the semantic network that represents the sum of the semantic
knowledge of the speakers of a given language. The semantic network
can be traversed in infinite ways, while being itself finite. The reader will
recall the concept of infinite/unlimited semiosis introduced above. One
can visualize traversing all the links connected to a given node as gath-
ering the meaning of that node (i.e., its connections to other meanings).
For example, in Figure 6.1 we see a few of the adjacent nodes (feline,
tail, etc.) to the script cat, with two examples of “immediate neighbor”
nodes in the graph in a hierarchical organization: the hyponyms, labeled
as is customary as the ISA relationship (see below), and the partonyms
(has-parts). The unlabeled link (mouse) is a script that is probably
highly associated with cat (as in “cat and mouse”) without being in
a specific relation to it. The immediate neighbor nodes will become
important when we consider spreading activation, below. Needless to
say, the above is a purely pedagogical presentation and should not be
misconstrued as a full description of the semantics of cat.

6 See Raskin (1985, p. 84) for a methodological difference between Mel’čuk’s (1981)
functions and Raskin’s links.

118
 6.1 THE SEMANTIC SCRIPT THEORY

Meow
Mouse

Black, Tabby, etc.

Son
Animal Feline Cat
ISA ISA Color

Has-part
Tail

Has-part

Paws

Has-part
Claws

Figure 6.1 A very simplified semantic graph for CAT; the links are labeled after the semantic
connection.

6.1.2.1.4 Related nodes The nodes are not a set of nodes selected ran-
domly. There needs to be a relationship between them. Each link defines
the specifics of the relationship. In mathematical terms, this makes them
“labeled” links. Technically they are directed labeled nodes, i.e., there is a
direction in which they must be traversed. So, for example, the ISA link,
one of the most common relationships in a semantic network and in
ontologies (see below), which corresponds to the hyponymy linguistic
relationship must be traversed from the hyperonym to the hyponym
(for example, from “fruit” to “apple”) and not the other way around.
So APPLE (ISA FRUIT) would be a basic semantic bit of information.
In artificial intelligence and frame semantics it is very common to
use a slot-filler notation, as I just did above. Essentially information
is represented as a pair, in which the first part indicates the kind of
information (slot) and the second indicates the specific information
(filler). So, the ISA slot could be filled with FLUFFY (ISA DOG) or BOB
(ISA HUMAN) or CAR (ISA VEHICLE). ISA of course stands for “is a”
(a car is a kind of vehicle). Other slots are created as needed. Standard
ones are location, time, has-parts, color, size, etc.

6.1.2.1.5 Organized cluster It should be noted that the definition that I

am employing here is slightly different from the canonical ones, which
119
 THE SEMANTICS OF HUMOR

Subject: [+Human] [+Adult]

Activity: > Study medicine
= Receive patients: patient comes or doctor visits
doctor listens to complaints
doctor examines patient
= Cure disease: doctor diagnoses disease
doctor prescribes treatment
= (Take patient’s money)
Place: > Medical School
= Hospital or doctor’s office
Time: > Many years
= Every day
= Immediately
Condition: Physical contact

Figure 6.2 The lexical script for DOCTOR in Raskin, 1985.

Note that “>” stands for “in the past,” and “ = ” for “in the present.”

do not as a rule emphasize the connections among scripts as part of a

given script. However, careful consideration will reveal that each filler
of a given slot (and for that matter, each slot) is itself another node.
Therefore, a script is properly merely a special section of the global
network. Furthermore, the cluster is organized in the sense that there
exist specific relations between the nodes and that (sub)scripts may
occur as parts of scripts (i.e., as nodes).
Figure 6.2 is an example of script, in the original notation used in
Raskin (1985, p. 85). The notational conventions are different than those
used in this book and in later versions of the theory, which are influenced
by LISP7 pseudocode. No particular meaning should be attributed to the
notation.
6.1.2.2 Fractal nature of scripts
Raskin (1985) introduced the notions of complex script and macro-
script, which matched similar distinctions in the field: for example,
frames may have sub-frames (Petruck, 1996; Mandler, 1984, p. 15)
In Raskin’s definition, the difference between scripts, complex scripts,
and macro-scripts is primarily one of level: a script is the simplex form; a
macro-script is a group of scripts organized chronologically, what some
authors would call a script, as opposed to a frame or schema; a com-
plex script is a script made of other scripts but without chronological

7 LISP (LISt Processor) is a computer language, originally invented in 1958, which

became popular in artificial intelligence. The first formalizations of Raskin’s scripts were
written in LISP, around 1989–1990.

120
 6.1 THE SEMANTIC SCRIPT THEORY

Script (simple)

Macro-script
Complex script
(organized chronologically)

Figure 6.3 Simple, macro-, and complex scripts.

organization. The classification of scripts is presented graphically in

Figure 6.3.
The restaurant script in Schank and Abelson (1977) would be
an example of macro-script: it consists of several other scripts linked
chronologically (drive up to the restaurant, be seated, order
food, etc.), while a complex script would be war, which presupposes
other scripts such as army, enemy, victory, defeat, weapon, etc.
Needless to say, no hard distinctions between complex and macro-
scripts are tenable: even a complex script as war has some chronological
organization: imagine a war that started with an armistice, whereas the
restaurant script has many events that may occur in no particular
order (for example, go to the restroom, chat with the other dinner
guests) or are only loosely organized (for example, asking for the check
must occur toward the end of the meal, but need not occur at the abso-
lute end of the meal—it is perfectly acceptable to ask for the check and
for coffee). Chłopicki’s character frames are complex scripts, whereas
the representation of the development of the story in the text as a
vector (text-world representation; Attardo, 2001a) is a macro-script; see
Chapter 14. The organization of scripts in a nesting hierarchy of larger
units is a common assumption, cf. Mandler (1984, p. 15) and references
therein.
We can further generalize the notion of complex script, by noting that
the filler of any slot in a script may be another script (most often a link via
the lexical handle of the script). Another way of putting it is saying that
scripts are fractal. In a script, you will find the same kind of constructs,
regardless of the level of generality. The geometrical figures known as
fractals are generated by recursion (much like embedded structures in
sentences can be recursive in language, so that, one can have a NP inside

121
 THE SEMANTICS OF HUMOR

a NP, ad infinitum, for example). Many fractals, but not all, also have the
feature of being “self-similar,” i.e., invariant at scale (Mandelbrot, 1977,
p. 18), or to put it more simply, they have the same shape at different
levels of magnification: if you zoom in on a detail of a fractal, you will
find the same shapes. In Attardo (2001a, p. 48), I spoke of scripts “nested
into one another.” The present formulation is more precise. Note that
the fractal recursive nature of scripts is consonant with the primeless
nature of the semantic network described by Raskin and can be usefully
connected to Peirce’s and Eco’s concept of “unlimited semiosis’.’ While
scripts may be more or less abstract, they are equally complex as they all
consist of links to other scripts.
6.1.2.3 Operational definition of scripts
In Attardo (2001a, p. 6), I described the procedure of building a script
as an hypothesis on the semantic content of a given lexeme8 which is
disproved if we encounter a bit of information not included in the script.
If we encounter a bit of information currently not in the script, the script
is revised and the revised version then takes the place of the original
hypothesis, only to be further tested by new texts. If the script is viable,
after a few revisions it will become stable, i.e., few if any changes will be
required. Consider a reader who encounter for the first time the term
“crypto-currency.” Let us assume, for the sake of simplicity, that the first
encounter is in an opaque context, such as “Mary worked as a crypto-
currency expert” which only tells us that “crypto-currency” is a subject
one can become an expert of and that this makes you employable. Our
frame or script for crypto-currency is pretty sparse, looking more or
less like this:
(crypto-currency
(ISA currency)
(object-of BE EXPERT))

Note that the (ISA currency) is a guess based on the assumption of non-
arbitrary morphology. If it were to turn out that crypto-currency is a
kind of cake, our guess would have been wrong.
Suppose now that the we encounter the information that crypto-
currency is transacted online. This would confirm the (ISA currency)
guess and would add the information that crypto-currency con-

8 The same applies to those scripts that lack a lexematic handle, see below.

122
 6.1 THE SEMANTIC SCRIPT THEORY

tains an (object-of FINANCIAL TRANSACTION) slot, with a sub-slot

(location currency-exchange).
(crypto-currency
(ISA currency)
(object-of BE EXPERT)
(object-of ((FINANCIAL-TRANSACTION
(location exchange))))

A more sophisticated bit of information might be added if a detailed

explanation of how crypto-currencies are based on solving complex
cryptographic problems, using computers. Further information might
appear, such as the fact that one of the currency-exchanges in which
crypto-currencies are bought and sold is called Coinbase, which would
then be added to the frame, making it more detailed and complex. Again,
further information, when available, would be added to the frame, until
no further information appears. At this point, one’s crypto-currency
frame would be “complete” (at least, this writer’s is).
However, as we know from Popperian epistemology, this does not
prove that the script is complete, but it is merely the best available
construct that matches the available evidence. While this may seem to
be problematic, since it is tantamount to claiming that scripts are open-
ended, in fact it is evidence of the falsifiability of the definition.
Essentially, one can never know that a given script is complete, since
the next sentence one processes may include a new bit of information
that was previously unavailable to the system. It is simple to imagine
a dynamic system which updates its knowledge banks whenever it
encounters a bit of information it was not aware of (and which is
consistent with its prior knowledge). This is, in fact, what humans
do: faced with a new bit of information they revise their scripts. An
interesting aspect of this process is that when faced with conflicting
information (say, two incompatible fillers for one slot) humans tend to
react with cognitive dissonance (see Section 2.4) and this they resolve it
by either compartmentalizing the information or by eliminating some
of the information.

6.1.3 Dynamic scripts

The idea of building up scripts as we accumulate information and of
updating the scripts as information changes, prefigures the dynamic

123
 THE SEMANTICS OF HUMOR

view of scripts I introduced in Attardo (2001a). In a nutshell, I was

interested in how readers build up a representation of the text they are
reading,9 which preserves the plot (in a non-technical sense), informa-
tion about the characters, and other aspects of the setting, without nec-
essarily preserving the surface structure of the text (the lexical choices,
for example). To do so, I introduced the idea of a “storage area” which
stores the most current and up-to-date information on the story, in a
macro-script built around sentences and paragraphs. This corresponds
to a dynamic view of scripts, see for example the following definition:

schemas came to be thought of, not as fixed structures to be pulled from memory on
demand, but as recipes for generating organizational structures in a particular task
context. (Kintsch, 1998, p. 37)

The reader would activate a storage area (which I called the text world
representation) as an empty sentential script, which will be filled by the
lexical and inferential scripts (see Section 6.1.3.1) activated by the text:

the model of text processing that I am outlining, consists in activating sequentially

scripts, until the main script of a text is determined, and then the entire text is inter-
preted as an instance of that script, filled with the actual details of the text instantiating
the script. (Attardo, 2001a, pp. 56–57)

In Attardo (2001a) I also review the considerable evidence for the

psychological reality of this approach, e.g., Kintsch (1998).
6.1.3.1 Implicit activation
We now turn to the activation of scripts without a lexical handle.
Consider the following example
(38) “I drove to Dallas”
The utterance of sentence (38) activates a number of scripts, beyond
those activated by the lexical handles “I,” “drive,” “to,” and “Dallas.”
In Attardo (2001a) I spoke of inferentially activated scripts. I would
now like to distinguish more specifically between presuppositionally
activated scripts, inferentially activated scripts, and scripts activated by
implicatures. Consider that “drive” activates the script for “car” or, rather,
less specifically “vehicle” (I could have driven to Dallas in my truck, for

9 Needless to say, this applies also to people hearing a story, watching a movie, etc.

124
 6.1 THE SEMANTIC SCRIPT THEORY

example). This is a necessary inference from (38): if I drove to Dallas it

follows necessarily that I did so using some vehicle. Note also that if I
utter (39)
(39) “I did not drive to Dallas”
it also follows that I did not do so using some kind of vehicle (so I
could have walked to Dallas, or ridden a bus to Dallas). The resistance
to negation is a clear indication that the inference that I went to Dallas
(regardless of mode of transportation) is a presupposition.
Now, (38) also activates personal scripts for SHOPPING (I occasion-
ally go to Dallas to shop) or SYMPHONY (I occasionally go to the
symphony at the Meyerson Symphony Center). So, upon hearing (38)
my friends could reasonably say
(40) “Oh, what was playing at the Meyerson?”
or
(41) “Went shopping, eh?”
The fact that these inferences are entirely probabilistic (I could have
gone to Dallas for another reason than shopping or going to the sym-
phony) and would not resist negation (either (40) or (41) would be
absurd following (39)) tells us that these are inferences, and more
specifically, implicatures. Along the same lines, we could have inferences
that activate scripts (they follow necessarily from the utterance, but do
not resist the negation test). For example,
(42) I drove to Dallas with my wife.
would dis-activate the bachelor script (assuming it had been active, of
course).
The conclusion then is that “inferential activation” as I used it in
2001a, is a generic term that covers all three kinds of inferences: presup-
positions, inferences, and implicatures. All three can activate a script.
Since all three are implicit forms of information, I speak of implicit
activation of scripts.
There is another way to activate a script, by “spreading activation.”
Spreading activation is a term from cognitive-psychology: when a node
is activated, its immediate neighbors are also activated, but at a much
lower level. The neighbors of the latter nodes are also activated, but at

125
 THE SEMANTICS OF HUMOR

at an even lower level, and so on, until the activation falls below the
threshold of activation.
Consider now a situation in which a set of adjacent nodes (because
they are all related to a given concept) is activated, while the node they all
are related to is not directly activated. Despite not having been explicitly
activated (by being mentioned in the text), the node that is related to the
others will receive enough spreading activation to become activated as
well. Petruck (1996, p. 3) has a good example:

consider the sentence Julia will open her presents after blowing out the candles and eating
some cake. Although there is no mention of a birthday party, interpreters sharing the
requisite cultural background invoke a birthday party scene.

While birthday party has not been lexically activated by being men-
tioned in the text, it becomes activated by spreading activation because
enough activation “spills over” from the lexically activated scripts for
present, cake, blow out candles to activate the script for birthday
party, at least within a culture that celebrates birthdays with cakes with
candles, etc.

6.1.4 Combinatorial explosion

As we have seen, scripts can be activated lexically, by spreading acti-
vation, and inferentially. It is not clear whether there is a distinction
between the last two modalities of activation. How is the sentence
meaning arrived at? Plainly by combining all the lexical and implicit
activation into a macro-script. This immediately presents a problem,
described in Attardo (2001a) as a combinatorial explosion. Indeed, if
one considers all the possible meanings of all the scripts activated in a
text, the combinations among them reach the hundreds of thousands
even for relatively short texts.
The combinatorial proliferation is kept in check by disambiguation,
which eliminates most alternative meanings as incompatible with the
script being activated but also by strictly pragmatic factors such as
relevance and appropriateness. This explains why the storage area does
not run out of space.
The combination of the various meanings is technically called
“unification.” Unification is an “information-combining operation”
(Shieber, 2003, p. 9). For example, the words “she” and “loves” both have

126
 6.2 THE SEMANTIC SCRIPT THEORY OF HUMOR’S T WO CONDITIONS

Table 6.1 Combinatorics for two scripts

unmarried academic title knight seal

disease paralyzed unmarried paralyzed BA paralyzed knight paralyzed seal

moral indecisive unmarried indecisive BA indecisive knight indecisive seal

the features [3rd person] and [singular] (she [3rd person], [singular];
loves [3rd person], [singular]) and thus they can be unified, yielding she
loves [3rd person], [singular]). In its most abstract notation, unification
can be represented as aa = a.
What this all boils down to is that if one of the possible meanings of
the script is incompatible with another, their unification fails (i.e., they
are not combined) and another meaning is tried. Raskin represents this
as a matrix of combinations among the numbered senses of the scripts
(1985, p. 86). For example, “paralyzed” has a disease meaning and a
moral meaning (as in unable to act); “bachelor” is famously analyzed
by Katz and Fodor as 4-way ambiguous: 1 marriage (as in un-married);
2 academic title (as in bachelor of art); 3 knight, and 4 seal (without
a mate). This yields potentially 8 combinations listed in Table 6.1 above.
Contextual pressure would then select among these combinations to
reduce the acceptable meanings to one (ideally) or a few more.
We have now reviewed all the technical information about script-
based semantics necessary to understand the original formulation of the
script-based semantic theory of humor, to which we turn next.

6.2 The Semantic Script Theory of Humor’s two conditions

Let us repeat here Raskin’s formulation of the Semantic Script Theory of

Humor’s central hypothesis:
(43) A text can be characterized as a single-joke-carrying-text if both
of the [following] conditions are satisfied:
i) The text is compatible, fully or in part, with two different
scripts.
ii) The two scripts with which the text is compatible are oppo-
site ( . . . ). The two scripts with which some text is compat-
ible are said to overlap fully or in part in this text (Raskin,
1985, p. 99).

127
 THE SEMANTICS OF HUMOR

6.2.1 Oppositeness
The concept of oppositeness was examined in some detail in Attardo
(1997). The following is the operational definition I proposed as the
synthesis of previous research:

It should be clear from the highlighted sections that incongruity is essentially defined as
divergence from expectations, in a way consistent with its ordinary meaning indicating
the opposite of “the quality or state of agreeing, coinciding, or being congruent”
(Merriam-Webster). (Attardo, 1997, p. 398)

Raskin’s definition of opposition is as follows:

[the opposed scripts are] local antonyms, i.e., two linguistic entities whose meanings are
opposite only within a particular discourse and solely for the purpose of this discourse.
(1985, p. 108)

As I noted in Attardo (1997), Lyons’ definition of antonymy is of no use

here, despite being quoted by Raskin, because Raskin is rejecting that
definition. I proceeded to provide a definition of local antonymy, that
borrows from Lyons, but that avoids Raskin’s criticism:

the locality requirement of the Semantic Script Theory of Humor is to be interpreted

as constructing a context in which being a doctor is the opposite of (i.e., is not) being
a lover.

while the antonymy is to be interpreted simply as the logical negation of

the script (i.e., any other script).10 Tinholt (2007, p. 50) further refines
the definition proposing that

two words are antonyms when they have all characteristics in common, except for one

where characteristics are defined as nodes and links. Essentially,

antonymic scripts would differ in one node and/or link, whereas the
rest of the activated nodes and links would overlap.

6.2.2 Overlapping
The definition of “overlapping” is simpler: two scripts are said to be over-
lapping in a text if the text overall is, or parts of it are compatible with
10 In Attardo (1997), I also carry out an attempt at redefining opposition in terms of
accessibility and informativeness. This is not the place to discuss that hypothesis.

128
 6.2 THE SEMANTIC SCRIPT THEORY OF HUMOR’S T WO CONDITIONS

both scripts. Thus, putting on one’s cowboy boots, hat, and bandanna,
are compatible both with the scripts bronco busting and honky-tonk
dancing. In Raskin’s infamous doctor’s wife joke (44), the knocking on
the door of the patient, the breathy voice, and the inquiry as to whether
the doctor is home are compatible both with the script for patient
visiting the doctor and man visiting his lover.

6.2.3 The exemplar doctor’s wife joke

As mentioned in Chapter 1, the doctor’s wife joke, analyzed by Raskin
(1985, pp. 117–127) has become an exemplar text for the linguistics of
humor. It should be emphasized that Raskin chose the joke not because
it is a good joke, as in aesthetically pleasing or interesting, but on the
contrary, because it is a simple, clear joke, with a single opposition, and
no special complications. Below is the text:
(44) “Is the doctor at home?” the patient asked in his bronchial
whisper. “No,” the doctor’s young and pretty wife whispered in
reply. “Come right in.”
The exposition of the analysis is going to be non-technical, as a full
blown analysis would require extensive discussion, mostly repetitive in
nature (see Raskin, 1985, pp. 117–127; Attardo, 2001a, pp. 10–16, for
examples).
The reader will activate the lexical scripts for doctor, patient and
bronchi, even assuming that the latter script is minimal and includes
only vague information such as links with medical and possibly lungs.
Inferentially, the script for medical visit is activated. Here the joke
shows its age as today doctors rarely if ever receive patients at home.
However, one will probably be familiar (i.e., have as part of their script
for doctor) with the fact that in the past, they used to have their office
in a room of their home, at least in smaller villages.11 So far, so good.
We have activated a macro-script doctor visit in which the patient is
the man asking if the doctor is at home. His whispering is compatible
with the script, as respiratory track illnesses often affect breathing and
speaking patterns, for example by making speaking painful.
However, the second sentence complicates things: first, the speaker is
the doctor’s wife, who informs the man that the doctor is not home.

11 I can attest visiting my childhood doctor in his office at his home.

129
 THE SEMANTICS OF HUMOR

From the script for doctor we know that physical presence is a

prerequisite for a visit12 and hence the reader can extrapolate that the
man’s purpose having been defeated he will leave (perhaps to return
later, when the doctor will be home). Second, the text informs the
reader that the wife is young and pretty. This information is irrelevant
to the man’s goal of seeing the doctor, thus it is odd with what has
been presented. Third, the wife is also whispering. It is possible that the
wife is also ill: maybe an epidemic of respiratory illnesses has struck
the village. However, this is again an odd bit of information, which a
well-behaved text would topicalize. Fourth, the wife invites the man to
enter the house. This last detail, which is incompatible with the first
script entirely, forces the reader to stop trying to patch up the first
hypothesis and to try to come up with a different hypothesis. Here, the
wife’s youth and beauty become potentially relevant (as presumably,
young people have more sex than older people), but the whispering is
compatible with a secrecy script (people who want to keep something
secret often whisper it rather that speaking in a full voice for fear of
being overheard), and the invitation to come in the absence of the
doctor/husband is compatible with an adultery script (one tends to
commit adultery when one’s spouse is not present, as spouses tend to
react negatively to their partner’s adultery).13 So, if we reinterpret the
text as a secret (hence the whispering) meeting between lovers in the
absence of the husband, the second part of the text makes sense and
the behavior of the wife is explained.
Thus the text is compatible in part with two scripts: doctor and
lover (Raskin, 1985, p. 117) or to be more precise medical visit
and adultery. Note that the overlap is not full: in the first sentence
the man is described as a “patient” thus negating the adultery script.
Clearly a different wording of the text could have produced a fully
overlapping text. The two scripts are opposed, for the purposes of the
text: either the man is there to see the doctor or he is there to see his
wife.14 Hence the Semantic Script Theory of Humor concludes that the

12 Again, the quaintness of the text is apparent, as in today’s age, telemedicine would fix
that problem.
13 There exist sexual perversions in which spouses actively encourage adultery. If this
were the case, the text would have focalized this detail. Since it did not we can safely assume
that the doctor would be upset if he found out his wife is sleeping with a patient.
14 I guess it is possible that the lover of the doctor’s wife is ill that day and goes to the
doctor opportunistically thinking that he either will get cured or will have sex with the wife.
However, we are now writing another story.

130
 6.2 THE SEMANTIC SCRIPT THEORY OF HUMOR’S T WO CONDITIONS

text in question has all the necessary and sufficient characteristics to be

potentially humorous.

6.2.4 Methodological issues

The Semantic Script Theory of Humor is essentialist, that is it provides
the necessary and sufficient conditions for a text to be humorous, and
reductivist, i.e., it reduces its object of study to simpler, better known
phenomena. Specifically, it reduces humor to three simpler concepts:
scripts and the overlap and opposition thereof. As we saw, it is also an
essentialist theory, since it provides us with the necessary and sufficient
conditions for a text to be funny, potentially. This brings about the
next topic: the Semantic Script Theory of Humor is a theory of humor
competence.

6.2.4.1 The Semantic Script Theory of Humor is a theory of humor

competence
Probably the biggest misunderstanding about the Semantic Script The-
ory of Humor is that it is a theory of humor competence, not of humor
performance. As such it has nothing to say about whether a given text
is perceived as humorous (performance). So to be precise, we should
state the goal of the theory as presenting the necessary and sufficient
conditions for a text to be potentially perceived as humorous. Whether
this is actually humorous to a given audience, etc. is the subject of the
third part of the volume.

6.2.4.2 The Semantic Script Theory of Humor is a theory of jokes

This too has been seriously misunderstood. The Semantic Script Theory
of Humor is elaborated on the basis of a corpus of jokes for a simple
and very effective methodological reason: jokes are simpler forms of
humor, in the sense that they tend to be limited to one source of humor
at a time. Other texts are hyperdetermined (Attardo, 1994), i.e., have
multiple concurrent sources of humor. When elaborating a theory one
should start from the simpler cases and then build up the theory to
account also for the more complex cases. Indeed, pre-theoretically, one
would not even know how to isolate the various sources of humor from
a hyperdetermined text.

131
 THE SEMANTICS OF HUMOR

6.2.4.3 Relational nature of the Semantic Script Theory of Humor

Let us focus our attention on the fact that this semantic definition of
humor has in fact very little to say about the “content” of humor. Raskin
(1985, pp. 111–114; 126–127) has some more specific considerations,
but they are contingent, and in fact extraneous to the theory itself
(Raskin, p.c.). Let us consider an example: Raskin (1985, p. 127) lists the
following basic oppositions on which all the numerous jokes analyzed
in his book are based:
good / bad
life / death
sex / no sex
money / no money
high status / low status15
While indeed the list has a very broad coverage we note that the
oppositions
excrement / non-excrement
physical handicap / non-physical handicap
are missing although a large segment of humor is based precisely on
those. It should be emphasized again that this is not an inadequacy
of the semantic theory of humor, but merely an indication of breadth
of coverage of the corpus used to develop it. While the readers may
be quite familiar with excretory jokes, jokes on physical handicaps are
an absolute social taboo in our society today. Yet, during a period
ranging from the Middle Ages to the Renaissance court jesters were
often persons afflicted by some physical handicap.
Moreover, different cultures may have different sets of scripts available
for humor. In the Middle Ages, people affected by dwarfism and by
other deformities (hunchbacks) were considered perfectly acceptable
targets for pranks and humor. In fact, they sometimes became jesters.
Today, laughing at someone’s physical deformity is considered beyond
the pale, at least in polite society. Likewise, blackface minstrel shows
are now considered offensive, but between 1850 and 1950 they were
quite popular and taken to be funny in the US. Another example, about
mocking a blind man, considered humorous in an African culture, is
analyzed in Section 13.1. Phallophoric processions were common in
15 I am slightly altering the wording of the oppositions for clarity.

132
 6.3 NON-BONA-FIDE

ancient Greece and still are in contemporary Japan, but are unheard of
in contemporary Western society. One could multiply the examples.
While the historical and cultural changes of what counts as appro-
priate for humorous exchanges are a fascinating subject of inquiry, the
entire issue is irrelevant from the point of view of the semantics of
humor: what matters is that all humor can be accounted for in terms
of overlapping and opposed scripts, regardless of the fact that a given
society, group or individual will or will not consider a given combination
of scripts a fitting subject for humor. In other words, the semantic theory
of humor is a relational or functional description of humor, in which
what matters are the interrelations of the various scripts, and not their
substance (i.e., what specific scripts are involved).

6.3 Non-bona-fide

The Semantic Script Theory of Humor has another facet, seldom

discussed in conjunction with the overlap and opposition hypothesis
discussed above, namely that humor is a non-bona-fide mode of
communication. Non-bona-fide is defined as an utterance that does
not follow Grice’s principle of cooperation (Raskin, 1985, p. 100); bona-
fide (Latin: “in good faith”) communication are utterances that follow
the principle of cooperation. We discuss in full the violation of Grice’s
principle of cooperation and its maxims in Chapter 8, so a discussion of
this aspect of Raskin’s theory will be found there.
Here we must however address another common misconception,
namely the idea that the Semantic Script Theory of Humor does not
include a pragmatic component. While one must admit that the name of
the theory may be misleading in that respect16 even a superficial reading
of Raskin’s text immediately reveals that the Semantic Script Theory of
Humor contains a significant pragmatic component at two levels:

1. script-based semantics incorporates a lot of what has been called

encyclopedic information (world knowledge), as opposed to
lexical information (word meaning); encyclopedic information
lies on the pragmatic end of the semantics-pragmatics continuum.

16 After all it is called the Semantic Script Theory of Humor, not the Pragmatic and
Semantic Script Theory of Humor.

133
 THE SEMANTICS OF HUMOR

For example, the differences between “spinster” and “bachelorette”

are exclusively encyclopedic, because their lexical meaning
([+human][+/−female][+adult][−married]) is identical.
2. the theory directly addresses the non-bona-fide nature of humor,
going so far as to propose (admittedly, tongue-in-cheek) a set of
maxims for non-bona-fide communication joke telling.

It is Raskin’s contention that semantics cannot fruitfully be sepa-

rated from pragmatics. One cannot do semantics properly without
pragmatics. Pragmatics is part of semantics, in other words. So, in
conclusion, the Semantic Script Theory of Humor is in fact a Semantic
Script and Pragmatic Theory of Humor in fact if not in name.

6.4 The Ontological Semantics Theory of Humor

Script-based semantics, much like frame-based semantics, has evolved

in a computational direction: having started out as semantic/pragmatic
theories of meaning, both theories have been implemented in compu-
tational applications. Raskin’s script-based semantics work was one of
the foundations of ontological semantics (Nirenburg & Raskin, 2004)
which was later a major influence on the work on ontologies that swept
computational linguistics and computer science at large and is at the
base of the semantic web and related initiatives (see for example Huang
et al., 2010; Poli et al., 2010).
Raskin, Taylor and Hempelmann have developed the Ontological
Semantic Theory of Humor, a fully computational theory of humor,
which embodies the Semantic Script Theory of Humor principles in a
fully computational development. For a discussion, see Raskin (2017b).
Because in order to understand properly the position of the Ontological
Semantic Theory of Humor vis-à-vis other approaches to computa-
tional humor would require an understanding of the non-semantic (or
marginally semantic) other approaches, we will not undertake this task
in this context.

6.5 Conclusions

This chapter reviewed briefly the Semantic Script Theory of Humor,

which is the favored semantic theory of humor competence. The theory

134
 6.6 FURTHER READINGS

consists of a central hypothesis, which provides the necessary and

sufficient conditions for a text to be potentially humorous: the text must
be compatible with two scripts (overlap) and the two scripts must be
opposed, in a technical sense (opposition). Pragmatically, the Semantic
Script Theory of Humor maintains that humor is a non-bona-fide mode
of communication, i.e., one in which the principle of cooperation and
its maxims are violated, i.e., the speaker does not speak in earnest. Since
the Semantic Script Theory of Humor reduces humor to the presence
of overlapping and opposed scripts in a text, under certain pragmatic
conditions (non-bona-fide) it is a reductionist and essentialist theory.
These are positive qualities, as explained in Chapter 3.

6.6 Further readings

Probably because the foundational text of the Semantic Script Theory of

Humor is easily accessible, there have been no definitive introductory
treatments of the subject. Someone interested in going deeper should
just read Raskin (1985). Attardo (1994, 2001a, 2008) review the theory
and some of its offshoots.
There exists no comprehensive treatment of frame or script semantics
that surveys the entirety of the literature. Useful surveys are Andor
(1985), Raskin (1985 ch. 3), Attardo (2001a ch. 3), and Attardo (2020).
Collections of papers were edited by Raskin (1985, 1986) and Lehrer and
Kittay (1992). On Fillmore’s frame semantics, see Petruck (1996). On the
evolution of Raskin’s work into ontological semantics, see Raskin et al.
(2009); Taylor & Raskin (2012); Raskin (2017b).

135
 7
The General Theory of Verbal
Humor

7.1 The knowledge resources 137

7.2 Further issues with the General Theory of Verbal Humor 152
7.3 Conclusions 156
7.4 Further readings 156

The General Theory of Verbal Humor was born out of the observation
that the Semantic Script Theory of Humor, despite its advantages over
other linguistic theories of humor, was not a complete theory. This claim
was motivated by two facts. First, the Semantic Script Theory of Humor
does not differentiate between verbal and referential humor (simplify-
ing, between puns and non-puns, respectively). This is correct, as far
as the Semantic Script Theory of Humor goes, since both verbal and
referential jokes have the same script opposition and overlap, but verbal
jokes may be characterized by their recourse to specific patterns of the
signifier (the phonological form of the utterance). These patterns do not
occur in referential jokes. Hence, the Semantic Script Theory of Humor
misses a potential generalization. Second, there exist a relationship of
similarity among jokes, such that two jokes will be perceived as more or
less similar to another joke. For example, the following pair of jokes is
more similar to each other, than to the Doctor’s wife joke, example (44).
(45) Q: What do you get when you cross a cow and a lawnmower?
A: A lawnmooer.
Q: What do you get when you cross a lemon and a cat?
A: A sourpuss. (http://www.funology.com/mix-match-jokes/)

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 7.1 THE KNOWLEDGE RESOURCES

Once more, then, the Semantic Script Theory of Humor misses a

potential generalization. The General Theory of Verbal Humor was
created to address these two issues, with a heavy emphasis on the second
one. In a sense, much of the impetus behind the General Theory of
Verbal Humor is to answer the question, when are two jokes the same
joke?
The answer was found to be tied to the introduction of five other
parameters, which along with the script opposition rounded up the six
knowledge resources. That the knowledge resources should be six and
not, say, nine, was a purely empirical matter. Quite simply, we found that
we only needed six resources to account for the similarity of jokes, and
for the differences between puns and non-puns.
So, the main claim of the General Theory of Verbal Humor is that
jokes may resemble each other along the lines of six parameters. These
parameters are the resources upon which someone wishing to construct
a joke from scratch would have to draw from. For this reason they were
called Knowledge Resources. We noted that while some of these were
already quite elaborate, others were more or less new, or required the
contribution of other disciplines.
Attardo and Raskin (1991) is somewhat notorious due to its length
and relative complexity. Much of that length and complexity come from
the observation that the knowledge resources are not an unordered set,
but are in fact organized hierarchically and linearily and that to the order
in the hierarchy corresponds the degree of similarity perceived by the
speakers. This was confirmed empirically by Ruch et al. (1993).1 Figure
7.1 presents the list of the knowledge resources and their hierarchical
organization.

7.1 The knowledge resources

In the following sections, we will review the knowledge resources. We

will follow a bottom-up direction, starting with language and working
1 The Logical Mechanisms parameter is a little off in that it performs less closely to
the linear and hierarchical assumptions made in the paper. The subjects perceived the
jokes differing by the Logical Mechanisms as less different than those differing by Situation,
but the difference is not significant. Jokes differing by Logical Mechanism are still perceived
as more different than all the other knowledge resources, except the script opposition, as
predicted by the model. In short, the divergence from the anticipated results is not that major
(Ruch et al., 1993, p. 131).

137
 THE GENERAL THEORY OF VERBAL HUMOR

Script Opposition

Logical Mechanism

Situation

Target

Narrative Strategy

Language

Figure 7.1 Hierarchical organization of the knowledge resources.

our way up to script-opposition. I will not justify the hierarchical

ordering of the knowledge resources in this context. Those interested
in the detail may consult Attardo & Raskin (1991). The best way to
conceptualize the knowledge resources is a toolbox, in which each
partition of the toolbox has tools specific to a task: if you want to inflate
your tires, you need air chucks (the metal connectors that attach to
the valves on the tires themselves), nozzles, and a gauge; if you want
to change your spark plugs, you will need a spark plug socket, and if
you want to drill a hole you will need drill bits or a hole saw (and of
course a drill). Likewise if you want to examine the targets of a joke, you
will need tools to do so, and they will probably be different from those
used to analyze the narrative strategy (organization of the text). In some
cases you can get by with general purpose tools (you can remove a spark
plug using an adjustable wrench, I just would not recommend it) but by
and large, each aspect of the joke needs a different treatment. So, each
knowledge resource can be thought of as a collection of tools to analyze
an aspect of the joke.
However, the tools themselves are generally not specifically designed
for the purpose of analyzing jokes. So, the analogy falls apart on that
respect: humor makes extraordinary use of ordinary linguistic devices.
There are no humor-specific constructs in the Semantic Script Theory
of Humor or in the General Theory of Verbal Humor: both theories are
relational: it is special combinations of features that acquire humorous
potential. Scripts, oppositions, targets, situations, etc. are not specific to
humor. They become so only in specific combinations.

7.1.1 Language
In its original formulation (Attardo & Raskin 1991), the Language
Knowledge Resource deals with the verbalization of the text. It can
138
 7.1 THE KNOWLEDGE RESOURCES

be thought of as all the possible paraphrases of the text. It contains a

phonological, morphological, syntactic and lexical description of the
text. It also contains information about the basic frequency of occur-
rence of these units in general texts as well as information connecting
this sort of formal features of texts to registers and stylistic codes.
Moreover it contains formal information pertinent to various textual
genres, including most notably the final position of the punch line.
The need for a phonological description is obviously motivated by
puns, as argued above, so that if a text, such as (46) includes a string
of sounds [aweyfer] the knowledge resource needs to be able to describe
this fact but also that the string may be analyzed as [a+weyfer] and/or
[awey+fer] and that these two parses of the string correspond to the
lexical items “a wafer” and “away for.”
(46) Why did the cookie cry? Its mother had been away for/a wafer too
long.
Homophony and homonymy are not the only types of ambiguity
exploited for humor. Some puns rely on morphological ambiguities,
such as the following headline:
(47) Dealers will hear car talk at noon (Bucaria 2005, p 292)
which relies on the categorical ambiguity between “talk” as a noun and
as a verb (the incongruous meaning being the verbal one, in which the
cars are talking).
Likewise, jokes may hinge on syntactic ambiguity, such as the follow-
ing example
(48) Captain Spaulding (Groucho): One morning I shot an elephant
in my pajamas. How he got in my pajamas I don’t know (Animal
Crackers; 1930)
analyzed in Chapter 9. There are many more cases of ambiguity that
are accounted for on the Language Knowledge Resource, for which see
Chapter 9 and references therein.
The requirement that the Language Knowledge Resource contain a
statistical distribution of phonemes and morphemes originates in the
need to account for alliterative humor, such as the following example
(49) You remember Sunset Strip – where the unneat meet to bleat!
(Attardo et al., 1994, p. 35)

139
 THE GENERAL THEORY OF VERBAL HUMOR

also analyzed in Chapter 9. In a nutshell, the occurrence of three

instances of the /i/ phoneme in four syllables is statistically improbable
and hence notable.
Stylistic issues are also handled in the Language Knowledge Resource.
A significant proposal to broaden the Language Knowledge Resource by
making it aware of different ways of working an expression, for
example to include “idiomaticity,” or constructions, was proposed by
Antonopoulou (2002) and Antonopoulou and Nikiforidou (2009).
Essentially this would be a sort of intermediate level between verbal
humor (based on the signifier) and referential humor (based on the
signified). The proposals are reviewed in more detail in section (14.4).
In Attardo (2017b), I provide an example of sophisticated stylistic
humorous allusion (p. 129) but of course there are many more examples
of stylistic humor in which the incongruity is between the register used
to describe them and the facts, known as register humor (Simpson and
Bousfield, 2017; Alexander, 1997; Attardo, 1994, pp. 230–253). This is
taken up in more detail in Chapter 14. Another stylistic aspect is the
use of obscene or scatological terms/registers (e.g., Seizer, 2011). The
socio-cultural salience and significance of obscenity and taboo words
is obvious enough. The connections to humor theory (e.g., Freud) are
also clear.
The Language Knowledge Resource is also responsible for determin-
ing the actual placement of the punch line. As work by Attardo et al.
(1994) showed, the vast majority of the punch lines (92% to be precise)
occurred in the last phrase of the last sentence of the text. Moreover,
even on the few instances in which there was linguistic material after
the punch line, it tended to be semantically empty, such as repetitions
of the punch line, interjections, expletives, identifications of the speaker,
and in a few cases unnecessary explanations of the punch line. An
example of unnecessary identification of the speaker is the following
joke, from Attardo et al. (1994, p. 45):
(50) One Monday morning at breakfast a man said to his wife, “I think
this is the first time in five or six years that I have remembered
a special date. Thursday will be your birthday. How times [sic]
flies. You’ll be 40 years old. What would you like most?” “Not to
be reminded of it,” was her quick reply.
where in keeping with the unnecessary wordiness of the setup, the
material after the punch line is not really needed, as it is clear who the
speaker is.
140
 7.1 THE KNOWLEDGE RESOURCES

The syntactic structure of the language and the valences of the verb
of the clause in which the punch line occurs will determine the exact
position of the punch line itself. For example, English adverbials may
occur sentence finally, and so in a few cases they did occur after the
punch line. The language knowledge resource needs to have prosodic
information. Prosodic information, such as stress placement, is signifi-
cant for the performance of humor; see Chapter 10.
Finally, an important issue is the name of the knowledge resource,
which is (confusingly) “Language” but should really be “Semiotic
Strategy.” The reasons for our choice of the term in 1991 should be
clear: we were presenting a linguistic theory. However, in retrospect, I
wish we had used “Semiotic Strategy” rather than “Language,” because
there has been a certain perception that the General Theory of Verbal
Humor was limited to linguistically expressed humor, despite numerous
examples of analyses of non-linguistic or multimodal texts (e.g., Paolillo,
1998, and Tsakona, 2009, which analyze cartoons, and Attardo, 1998,
which analyzes a sitcom episode, including visual gags).

7.1.2 Narrative Strategy

The Narrative Strategy is concerned with the organization of the text,
not in the syntactic sense (which belongs to the Language Knowledge
Resource), but in the narrative sense. A story can be told in many ways.
Jokes are prototypically very short stories (similar in this to anecdotes,
the main difference being that jokes end in a punch line; Oring, 1989;
Attardo, 1994). For example, many jokes consist of a narrative closed
by a dialogical part, which contains the punch line. In some cases,
jokes consist only of the dialogical part, and so strictly speaking, do not
have any narrative aspect, unless we want to argue that the narrative is
implicit. Consider the old joke:
(51) Can you take shorthand?
Yes, but it takes me longer.
in which the script for secretary, before the 1980s, when the wide-
spread availability of dictation machines made shorthand unnecessary,
allows us to fill in the gaps and assume that the first speaker is the
boss and the second speaker the secretary, thus activating the back-
ground script office work (a.k.a. the Situation, in General Theory of
Verbal Humor terminology) which provides us then with the (implicit)

141
 THE GENERAL THEORY OF VERBAL HUMOR

narrative: the boss calls his (new) secretary and asks her, etc. Note that
the recent hire of the secretary is implied by the question: if the secretary
had been in the employ of the boss for a long time he/she would know
the answer to the question.
Regardless of how convincing the analysis of dialogical jokes as
implicit narratives may be, it is clear that not all jokes are narratives.
There are plenty of examples of jokes that are not prototypical: one
liners have extremely compressed or implicit stories; riddles have a
question/answer structure; top-ten lists are precisely that: lists; memes
and cartoons rely on the interplay of a verbal and a visual component;
etc. It is somewhat of a stretch to claim that a meme has a narrative and,
while one may defend that idea, we do not need to: the General Theory
of Verbal Humor simply stipulated that the term Narrative Strategy
covers all the forms of textual organization, as exemplified above.2
The discussion of the narrative strategy within Attardo and Raskin
(1991) was fairly simplistic: the kinds of example we used were a ques-
tion/answer format (riddle, pseudo-riddle) or the three-step sequence
frequently used in jokes because it is the smallest number of repetitions
necessary to set a pattern of expectations and breaking it. The three
step sequence may have another function as well: as Atkinson (1984)
noted, three-part lists are used by speakers to invite applause, given
the predictability of the occurrence of the third item. The three-step
sequence may likewise signal the arrival of the punch line. The fellatio
joke analyzed in Sacks (1978, p. 251) is an excellent example of this
organization (see example (112), in Chapter 11). Another example of
the structure is example (60) below.
Rozin et al. (2006) investigated the AAB pattern (i.e., the three-step
sequence) and found that the three-step sequence is more frequent
and considered funnier by speakers, with a four-step sequence (AAAB)
significantly less frequent and less funny, and the AB sequence (simple
opposition) slightly more frequent than the AAAB pattern, but con-
sidered the least funny. On the three-step sequence, see also Loewen-
stein and Heath (2009) and Loewenstein et al. (2011) who dub the
phenomenon “repetition break.” Other more complex narrative strate-
gies may involve scripted sections, involving two speakers, such as the

2 I toyed with the idea of replacing the term Narrative Strategy with Textual Organiza-
tion, which gets rid of the distracting mention of “narrative,” but it seemed unwise to change
the terminology of the General Theory of Verbal Humor so long after the original proposal.

142
 7.1 THE KNOWLEDGE RESOURCES

knock-knock joke, which follows the following elaborate construction

(A and B are the two speakers)
(52) A: Knock-knock.
B: Who’s there?
A: X
B: X who?
A: punch line
or are tricks to get the audience to say something incongruous:
(53) Q. How do you pronounce M,A,C (pause) , M, I, L, L, A, N?
A. MacMillan.
Q. How about M,A,C (pause), H, E, N, R, Y?
A. MacHenry.
Q. And how about M,A,C (pause), H, I, N, E, R, Y? (Hockett,
1973/1977, p. 279)
where automatism should lead the responded to answer MacHinery,
rather than [maʃɪnərɪ]. Often the word(s) the teller is trying to elicit are
embarrassing or obscene. The following is an example of plain tricking:
(54) A: I bet I could make you say black!
B: Okay
A: What color is my [something red]?
B: Red
A: What color is my [something white]?
B: White
A: What color is my [something yellow]?
B: Yellow
A: Hah! I made you say yellow!
B: Wasn’t I supposed to say black?
while the following is an example of getting the audience to say some-
thing obscene
(55) A: “You’re a dickfor.”
B: “What’s a dick for?”
While the last four examples have all been very simple (and target
younger children, at least in my experience) other types of narrative
strategies can be very elaborate, such as the Limerick, a five line stanza
with a rhyme scheme AABBA, and an elaborate metrical system, and

143
 THE GENERAL THEORY OF VERBAL HUMOR

a nonsensical type of humor, or the Chastushka, described by Raskin

(1985) which involves obscene humor. At this upper end of sophistica-
tion and formalization, narrative strategies end up approaching literary
genres and forms.
A very important aspect of jokes, and namely the functional position
of the punch line belongs properly to this knowledge resource, since for
example, Oring (1989) has claimed that the absence of a punch line in
final position in the text is the defining difference between jokes and
anecdotes.3
In the subsequent versions of the General Theory of Verbal Humor,
beginning with Attardo 2001, I introduced several concepts that also
pertained to narrative strategy: the distinction between jab and punch
lines, the distribution of the humor along the textual vector, and the
typology of humorous texts based on the presence or absence of a nar-
rative disruption and the presence of a humorous central complication.
Further discussion of these issues will be found in Attardo (2001) and
in Chapter 14.
Finally, other issues handled by the narrative strategy knowledge
resource involve the role of narratives in conversation. Narratives, being
produced as single turns, disrupt the turn-based flow of conversation.
Studies have focused on ways to negotiate and secure the floor for
the extended narrative turn as well as the reactions to it. Analyses of
conversational humor based on the General Theory of Verbal Humor
should take this into account. Discussion of these issue will be found in
Chapters 11 and 12.

7.1.3 Target
The target knowledge resources is probably the least sophisticated one
and the easier to understand. Essentially it addresses the fact that many
and perhaps most jokes are aggressive and the aggression has a target.
This is know in common parlance as the “butt” of the joke. Targets
generally are human or related to human activity (institutions, practices,
beliefs, etc.).

3 Strictly speaking the actual position of the punch line in the text is of course a matter
of language, and thus best handled by the Language knowledge resource (see Section 7.1.1
above). Narrative strategy is interested only in the functional position and not in the actual
placement.

144
 7.1 THE KNOWLEDGE RESOURCES

Significantly, the target is an optional knowledge resource, in the sense

that it may be empty if the joke happens to be what Freud would call
a “non-tendentious” joke, or more simply a non-aggressive joke. For
example, in this joke, repeated from above
(56) Q: What do you get when you cross a cow and a lawnmower?
A: A lawnmooer.
it is hard to imagine what the target of the joke could even be: cows?
Lawnmowers? People who mow the lawn? It is a far more reasonable
hypothesis that the joke is non-aggressive and that the humor is associ-
ated with coincidental resemblance between the “son” lexical function in
the script for cow (i.e., the sound cows make) enriched by the derivative
morpheme -er (“moo” + “er” = [mʊ + ər]) and the lexeme “mower”
[moʊər].
While people are the prototypical targets of jokes there exist other
kinds of targets. Ideological targets (Karman, 1998) are institutions or
groups without a clear constituency (compare the National Rifle Asso-
ciation, which is a clearly defined group, as one has to pay membership
dues to belong, to the “establishment” which is an amorphous group,
or even more abstract notions such as “marriage” or “romantic love.” In
Attardo (2001) I speculated that these groups are associated with human
activities or beliefs and that, these groups, however vaguely identified,
could therefore be targeted with aggression (p. 21).
While that may be true, it remains that it is unclear that there is
any aggression toward lawnmowers or cows, or people who mow lawns
or own cows, in the joke above. In recent work, Priego-Valverde et al.
(2018) show that humor is always oriented toward a participant of the
interaction: Speaker, Hearer, or a third-person/object, or possibly the
setting or situation. However this does not imply aggression toward
the focus of the orientation (which would make them a target). Priego-
Valverde et al. (2018) use the terms “speaker/hearer/other-oriented
humor.” Orientation of the humor is not part of the original General
Theory of Verbal Humor and thus introducing this distinction is an
extension of the General Theory of Verbal Humor, which has the advan-
tage that it no longer requires the knowledge resource to be optional, in
case of lack of aggression.
First-person orientation is self-deprecating humor (Rodney Danger-
field was the master of this form: “I told my psychiatrist that everyone
hates me. He said I was being ridiculous, everyone hasn’t met me

145
 THE GENERAL THEORY OF VERBAL HUMOR

yet.”). An example of non deprecating first-person orientation is the

following joke, which I used to tell people: “modesty is one of my greatest
virtues.” Second-person orientation is humor directed at the hearer.
A good example is Carmen making fun of Marina’s monolingualism, in
example (10) which may or may not be seen as aggressive and hence is
ambiguous as to whether it is a target or just an orientation. The reader
will recall the example of aggressive joke (23) in Chapter 3, which is
openly aggressive against lawyers and priests, depicted as uncaring (the
former) and sexual predators (the latter). That joke displays a third-
party orientation (assuming neither the teller or hearer happen to be
the lawyer or priest in the joke). The following example, from Priego-
Valverde et al. (2018) illustrates a third-party orientation (the situation),
but no aggression, as the participants jokingly compare their being in the
anechoic room to being eggs in an egg carton. The symbol @ represents
a laughter pulse.
(57) CL 46 t’as pas l’imp
Don’t you have the imp
CL 47 @@pression d’ètre dans une@@ boite d’œufs
@@pression@@ of being in a@@ egg box
JS 42 @
CL 48 @
CL 49 @
JS 43 @@avec les petits@@
@@with the small@@
JS 44 @@*@@
CL 50 @

No
Aggression
aggression

1st person
Target
(self )
2nd person
Orientation Target
(hearer)
3rd person
Target
(other)

Figure 7.2 Orientations and targets of humor as a function of aggression.

146
 7.1 THE KNOWLEDGE RESOURCES

We can thus summarize the situation in Figure 7.2. If there is

aggression, the orientation turns into a target; if no aggression is
present, then the joke is simply oriented toward the person but does not
target it.
Finally, let us point out that the targets need to have relevant scripts
attached to them in the encyclopedia (for example, the Scots’ stinginess
or the Belgians’ love of fries) which mark them both as stereotypically
humorous and as such available for humor. The work of Christie Davies
(e.g., 1990) is the principal source for a review of these stereotypes and
their sociological origins.
Other work on targets includes Scogin & Pollio (1980) who found that
laughter-producing remarks could be coded into whether the remark
did or did not have a specifiable target. After recording humorous events
in six different natural settings involving over 1,300 events, targeted
remarks varied from 36% to 79%, with an overall average of 66%. So,
roughly speaking, two thirds of humor would be targeted and about a
third would be non-directed (i.e., not have a target). They find signif-
icant differences among groups and that transient, short lived-groups
have less targeted humor.

7.1.4 Situation
The Situation is essentially the background script that describes the
environment in which the events of the joke take place. Because of its
generality it is probably best described as a macro-script.
The activation of the background script may be completely inferential.
Consider the joke (51) again. All the scripts that the text activates
directly are shorthand and long.4 Everything else has to be triggered
inferentially. Yet most readers will conjure a boss and a secretary, will
assign them stereotypical genders, and will probably date the scene
between the 1950s and the 1980s, when cheap recording devices ren-
dered shorthand obsolete. It is worth noting that these inferences all
belong to a stereotypical frame/script for secretary. In a sense, these
inferences activate a minimal character frame for the boss and the
secretary (see Section 14.1.2).

4 The text also activates scripts for capacity, for the lexeme “take” in the “take
shorthand” context, and so on. I am simplifying.

147
 THE GENERAL THEORY OF VERBAL HUMOR

The activation of the script for business office, still inferentially

from boss and secretary explains why we could easily “beef up” the
narrative along the lines of:
(58) The boss called his new secretary on the intercom: “Miss Smith,
can you take shorthand?” “Yes, replied the young and pretty
woman, but it takes me longer.”
The presence of the intercom in an office setting does not surprise us,
because it is part of the script for business office. Conversely, if the
boss called his secretary on a walkie-talkie this would be odd, and if
he used a megaphone, completely out of the ordinary and perhaps even
incongruous/funny. Likewise the youth and good looks of the secretary
do not surprise us because they fit in the stereotype of the lecherous boss
who hires his staff based on looks and not on skill. In Attardo (2017b,
p. 132), another example is analyzed, along the same lines. The joke is
(59) A skeleton walks into a bar and says, “I’ll have a beer and a mop.”
where the script patronizing a bar is evoked, among others.
Of particular note is the fact that the humorousness of (51) and
(58) is identical or perhaps lesser in the “expanded” version. This is
because the Script Opposition and the Logical Mechanism of the joke are
unchanged. The only (very minor) difference lies in the explicitation of
some of the inferentially activated scripts, which by lowering the mental
processing required to “get” the joke, may negatively affect its funniness.
This is significant, because backgrounded incongruities “reside”
in the Situation. Hempelmann and Attardo (2011) introduce the
notion of backgrounded incongruity (i.e., an incongruity that is not
addressed and not resolved in the text). Because they are assumed
to be part of the Situation, or to put it differently, they are built in
into the text-world or the mental space opened by the narrative, these
incongruities are not “active” so to speak, and are suppressed (“willing
suspension of disbelief ”) in the processing of the text. Examples of
backgrounded incongruities are talking animals, very common in
folklore and in many jokes, all the stereotypical ethnic traits (the stupid
Poles/Belgians/Newfoundlanders, etc.; the stingy Scots; the oversexed
French, etc.), and even idiosyncratic entities such as the walking, talking,
drinking skeleton in joke (59). Note how, while we are supposed to
ignore that a skeleton cannot either walk, talk, or drink, we need to
notice that, if the skeleton drank, since it does not have a stomach, the

148
 7.1 THE KNOWLEDGE RESOURCES

beer would spill on the floor. This latter incongruity is focalized, whereas
the previous ones are backgrounded.
Needless to say, a normally backgrounded incongruity may be
focalized in a given joke, witness one of my favorite jokes, in which
the punch line has precisely to do with a talking animal:
(60) A guy in a bar bets drinks with the bartender claiming his dog
can speak. “OK, says the bartender, prove it.” “What’s on top of a
house?” says the patron to his dog. “Roof!” says the dog. “That’s
cute, says the bartender, “but that’s not talking.” “How about this?”
says the guy, and to the dog “What is on the side of a tree?” “Bark!”
“You’re messin’ with me,” says the bartender getting angry. “No,
no,” says the guy, “Look: who’s the greatest baseball player?” And
the dog: “Ruth!” The bartender is enraged and throws the guy and
his dog out. The guy and the dog dust themselves off and then the
dog says: “Should I have said DiMaggio?”
The Situation of the joke is directly related to what Clark and van
der Wege call “imagination” in a narrative (2015), i.e., the fact that
the hearers of a story visualize mentally (imagine) the events of the
narrative by constructing a mental model of the narrated events, using
their prior knowledge (scripts), simulations (in which they explores
“what if ” branches of the story), while engaging in the co-construction
of a “joint pretense” in which they willingly suspend disbelief for the
backgrounded incongruities and of the contrived nature of the story-
telling (for example, the narrator may do different “voices” for different
characters, but no-one would be fooled into thinking that someone else
is speaking). So, summing up we could call the Situation of a joke the
mental space built for it by the narrator and the narratees.

7.1.5 Logical Mechanism

The Logical Mechanism is the part of the General Theory of Verbal
Humor that accounts for the resolution of the incongruity (script oppo-
sition and overlap). The resolution is not complete and/or real. It is
a partial, playful, non-serious resolution. In fact, the Logical Mecha-
nism may introduce other incongruities itself. Furthermore, the Logical
Mechanism is the other knowledge resource that is optional. The reader
will recall from Section 4.1 that in the famous Nerhardt (1970) weights

149
 THE GENERAL THEORY OF VERBAL HUMOR

experiment, the mere presence of a unexpectedly heavier weight in a

series of weights triggered smiling and laughter. Hence the idea that
incongruity alone is sufficient to produce humor. Furthermore, absurd
humor is usually regarded as having no resolution. Hence, the claim that
the Logical Mechanism is probably optional.
In simple terms, the Logical Mechanism is the attempt by the text to
explain away the incongruity by justifying it (once more, playfully and
non-seriously). Consider the following joke:
(61) Three kids are bragging about their fathers. “My father is the
fastest, says the first kid, he can run a mile in 4 minutes.”
“That’s nothing, says the second, my father can run a mile in
3 minutes.”
“You don’t understand what fast means, says the third one, my
father’s work ends at five and he’s home by 3:45.”
Here the Logical Mechanism works as follows: if shorter times mean
faster speed, then negative times mean even faster speeds, therefore,
since the third kid’s father is home an hour and 15 minutes before quit-
ting time, he must be extremely fast. Needless to say, another inferential
path is also possible and namely that the third kid’s father is home before
quitting time because he does not work hard and cheats his employer
out of one hour and 15 minutes of work time. Note how the resolution
(shorter times mean faster speed hence negative time is even faster) is
partial and non-serious.
More work on Logical Mechanism, including a list of all logical
mechanisms known so far5 and an attempt at classification, can be found
in Attardo et al. (2002) and Hempelmann and Attardo (2011); see also
Section 4.2.1.6.

7.1.6 Script Opposition

The concept of Script Opposition and overlap was of course defined and
exemplified in detail in Raskin (1985). Attardo and Raskin (1991) did
not change the original definition in any way. Script Opposition was
discussed in detail in (6.2.1). In Attardo (1994, pp. 203–205), I pointed

5 In one of the most frustrating misconceptions about the General Theory of Verbal
Humor there have been several claims in print that the list is the “complete” list of Logical
Mechanisms, despite the fact that we say explicitly that it is merely the list of known Logical
Mechanisms and that a complete list is probably impossible to produce.

150
 7.1 THE KNOWLEDGE RESOURCES

out that simple oppositeness of two scripts and/or simple overlap of two
scripts in a text does not result in humor. Both conditions must hold
simultaneously. All that needs to be added to this discussion is a short
discussion of the various levels of generality of the Script Opposition.
Raskin (1985) reduces all Script Oppositions to three extremely
high level (abstract) oppositions: good/bad, normal/abnormal, and
actual/non-actual. Many researchers get confused because they try to
identify the very abstract Script Opposition, without first identifying
the text-specific script opposition. Each text instantiates these high-
level script oppositions into text-specific local oppositions. Di Maio
(2000) unpublished PhD dissertation, which was used in Attardo et al.
(2002) introduced a third (intermediate) level of abstraction (e.g.,
sex/no sex, excrement/non-excrement), which we found useful to bridge
the step from the concrete instantiation in the text and the very abstract
opposition.
For example, consider the following joke,
(62) One day three women went for a job interview. The man inter-
viewing them posed all three the same question. What would you
do if you found an extra €50 in your paycheck that you shouldn’t
have received? The first one said, “I’d give it back as it wasn’t
mine and I wasn’t entitled to it.” When he asked the second one
she replied, “I’d give it to Charity.” When he asked the third one,
she was more honest and she said, “I’d keep it for myself and go
out for a drink.” Which one of the three women got the job? The
one with the biggest tits!
which was the “joke of the day” on April 20, 2018 on the web site http://
www.laughfactory.com/jokes/joke-of-the-day
The script opposition here is sex/no sex, but the specific instantiation
is much more complex since all the way until the last (fragmentary)
sentence, the text is ostensibly a narrative about an elaborate test of
personality in a business hiring situation. So the script job interview
explicitly activated in the first sentence turns out to be an instantiation
of the much more abstract no-sex category. Metaphorically speaking,
the joke is a garden-path construct, since it leads the reader to believe
that the story is about finding out the character of the women inter-
viewing for the job presumably to determine who is the most suited
for the position. Intermediate scripts activated are also business prac-
tices and hiring, but also upon reading the punch line, male versus

151
 THE GENERAL THEORY OF VERBAL HUMOR

female, since the logical mechanism here is the stereotypical behavior

of the male interviewer (the “lecherous boss” mentioned above). Notice
that the fact that the three applicants for the job are women becomes
significant only after the punch line is reached and the script opposition
is triggered. Among the backgrounded oppositions let us note that the
presence of only three applicants and that they are all females may
be perceived as anomalous, under certain circumstances (for example
within current hiring practices in the US).
Summing up, each humorous text will instantiate one very abstract
Script Opposition in a very concrete text-specific opposition, which
will generally require at least one of the scripts in the opposition to
be explicitly stated in the text and the other to be directly or at least
easily retrievable inferentially/abductively from the text. There may be
any number of intermediate scripts, increasingly abstract, bridging the
distance between the scripts occurring in the text and the very abstract
oppositions identified by Raskin.
In the prototypical examples of Script Opposition in jokes there is
only one Script Opposition, which makes the text incongruous. How-
ever, this is not always the case. Tsakona (2003) showed that many
jokes do not just have a final punch line, but in fact there are other
Script Oppositions along the body of the text (jab lines). This is not
an issue for the expanded version of the General Theory of Verbal
Humor which incorporates multiple instances of humor in the text (jab
lines and punch lines). Norrick (2010) documents jokes with multiple
punch lines (Hockett, 1977, p. 261, “compound jokes”) and with jab
lines, as well as numerous examples of laughter within the body of
the joke text, triggered by various aspects of the performance of the
joke (see (10)). Hempelmann and Attardo (2011) discuss backgrounded
incongruity, which is assumed as part of the situation of the text; see
(4.2.3). Backgrounded incongruities overlap with the multiple jab lines
of jokes described by Tsakona and Norrick.

7.2 Further issues with the General Theory of Verbal Humor

7.2.1 Broadening the Theory?

There has been some discussion as to whether more knowledge
resources are needed and/or should be added. All the proposals in that

152
 7.2 FURTHER ISSUES WITH THE GENERAL THEORY OF VERBAL HUMOR

sense focus on performance issues, therefore they will be discussed in

Part 3 of this book. For discussion, see Section 10.1.5.
Relatedly, there has been some discussion about dropping the Log-
ical Mechanism (e.g., Davies, 2004). In a nutshell, Davies argues that
while one can make generalizations on bona-fide discourse, non-bona-
fide discourse is essentially un-generalizable, or rather that one can
find generalizations but they are pointless. According to Davies, evi-
dence of this would be found in the lack of testable hypotheses that
could be generated from studying logical mechanisms (2004, p. 379). In
Hempelmann and Attardo (2011), we listed some hypotheses, derived
from the logical mechanisms, that have been tested empirically. Since
then, more evidence has emerged showing that the distinctions within
the logical mechanisms are psychologically real and even have neuro-
logical correlates (see a review in Chen et al., 2017). A further list of
criticisms of the General Theory of Verbal Humor, with some responses,
can be found in Attardo (2017b).

7.2.2 The application of the General Theory of Verbal Humor

to longer texts
As is well known, the Semantic Script Theory of Humor, and the
General Theory of Verbal Humor after it, were developed using jokes,
for the methodological caution of starting by analyzing relatively simpler
phenomena. Almost immediately, though, the theories were applied to
other types of texts. However, the broadening of the scope of a theory
comports a potential issue, which is that significant additional concep-
tual apparatus may need to be added to handle the new phenomena that
now fall under its purview. As discussed in Attardo (1994, pp. 221–222),
two approaches emerged: the expansionist approach, which assumes
that other types of texts are more-or-less essentially similar to jokes and
that so a simple expansion of the theory can include other types of texts
under its purview, using the same conceptual apparatus. Conversely, the
revisionist approach takes it that it is necessary to develop a new set
of tools to handle the new materials. Both of these approaches will be
discussed in Chapter 14.
The GTVH has been applied to a variety of materials, such as
telephone calls (Antonopoulou & Sifianou, 2003), Greek young men
conversations (Archakis & Tsakona 2005), Jordanian joke telling

153
 THE GENERAL THEORY OF VERBAL HUMOR

(Al-Khatib, 1999), conversational irony (Alvarado Ortega, 2013),

cartoons (Paolillo, 1998; El-Arousy, 2007; Tsakona, 2009), Polish sketch
comedy (Brzozowska, 2009), Latin satire (Vincent, 2010), and of course
translation (see Chapter 15).

7.2.3 A small methodological note on cherry-picking

As I noted in Attardo (2017b), the methodology underlying the expan-
sion of the General Theory of Verbal Humor to longer text included a
proviso of great significance. I have referred to this by a motto of “no
cherry-picking of data.” Let us briefly examine what this means and the
significance of this admonition.
Let us start by imagining a graduate student, let’s call him Bob, who
wants to test his theory of the genetic foundation of fetching behavior
in dogs. He gathers as many dogs as he can, borrowing dogs from his
friends and neighbors, and starts throwing sticks at them in a very
controlled, double-blind setup (the dog is behind a fence). After a few
frustrating days, he finds that all of the dogs fetch, except one named
“Blue.”6 In frustration, the graduate student decides to drop Blue from
the experiment. That is very bad behavior, science-wise. By excluding
those observations that do not confirm his theory our graduate student
has essentially falsified the data. This is an instance of cherry-picking
(ignoring data that go against your theory). Another way to cherry-pick
is to only consider the data that support the theory: suppose that about
a quarter of the dogs refused to fetch; if our graduate student writes a
paper on fetching behavior without mentioning that 25% of the dogs
did not fetch, he has again cherry-picked the data (considering only data
that support the theory).
The above examples are semi-facetious in the sense that no researcher
would be that bad. Unfortunately, there are subtler ways of cherry-
picking the data. Suppose that our student wants to investigate how
loudly speakers deliver punch lines. He collects 20 recordings of speak-
ers delivering a given joke, using a good microphone, making sure
he has 10 males and 10 females and taking every care not to bias his
sample. He then proceeds to measure very precisely the volume of the
punch lines. He finds that the average punch line is delivered at 70
decibels, accounting for male/female differences. The problem here is

6 Any resemblance to dogs in our household is purely coincidental.

154
 7.2 FURTHER ISSUES WITH THE GENERAL THEORY OF VERBAL HUMOR

that this measurement, however accurate and unbiased, is not mean-

ingful: in order to be meaningful the student needs to compare it to the
non-punch line language. In other words, he needs to determine a
baseline volume in parts of the recordings that are not punch lines and
then he can compare this to the punch lines. This too is a form of
cherry-picking the data: by analyzing only what you are interested in,
you lose track of the big picture. Humorous language is also language.
If we are interested in determining what makes it humorous we must
contrast it with non-humorous language. It is not enough to say that
such-and-such a feature, phenomenon, etc. occurs in our data. It could
be occurring outside the data as well making the observation pointless,
as far as understanding humor is concerned.
Disgusted by his results, Bob leaves academe and becomes an
entrepreneur opening a dog-sitting business that earns him a small
fortune. Another graduate student decides to take up humor research.
She7 collects a large corpus of naturally occurring conversations and
is struck by the fact that in several of the conversations, one speaker
reacts with obscenity to most instances of humor. She transcribes all
the conversations and analyzes in painstaking detail the behavior of the
speaker. It becomes clear that there is a quasi-systematic correlation
between the presence of an instance of humor and an obscenity. The
student becomes intrigued, does some research, and she discovers that
the speaker is a Tourette syndrome sufferer and that he manifests his
anxiety in delivering the punch line of a joke or a jab line by producing an
obscenity. Thus she comes to the conclusion that, while the obscenities
and the humor are connected, there is no causal correlation between
them as there is an intervening variable that explains the correlation. The
student publishes a case study on the phenomenon, but then goes back
to her data and looks at many more speakers. Eventually she becomes a
successful academic.
The second student’s experience illustrates yet another cherry-picking
problem: given a sufficiently large corpus of data, one will find some
exceptional instances within it that will be very interesting, but not nec-
essarily representative of the data. The only way to avoid this problem
is to examine all the data or a representative8 subset of the data. This
is the reason why the General Theory of Verbal Humor argues that one
7 This one is a woman. Her name is Mary.
8 I.e., such that it reflects all the characteristics of the data. You make sure that this is the
case by selecting the subset randomly.

155
 THE GENERAL THEORY OF VERBAL HUMOR

needs to examine all the instances of humor in a given text, or at least a

representative subset thereof. Failure to do so will necessarily introduce
unconscious biases in the selection process.

7.3 Conclusions

In this chapter, we examined the reasons for the creation of the General
Theory of Verbal Humor and the six knowledge resources that comprise
it. The General Theory of Verbal Humor is a less abstract theory than
the Semantic Script Theory of Humor, insofar as it deals with the imple-
mentation of the script opposition at a textual level. It is, however, still
on the competence-side of the continuum. Applications of the General
Theory of Verbal Humor to the performance side will be examined in
the third part of the book.

7.4 Further readings

Attardo & Raskin (1991) is technical and best approached with a general
understanding of how the theory works, which can be gathered from
any of the many presentations of the General Theory of Verbal Humor,
such as Attardo (1994; 2008; 2017b). The work by Attardo et al. (2002)
has shown that more formal versions of the General Theory of Verbal
Humor can be produced relatively easily. It is, however, more technical.
The applications of the General Theory of Verbal Humor to various
fields and problems are ongoing. This is not the place to review them,
but see Attardo (2017) for a start.

156
 8
Pragmatics of humor

8.1 Pragmatic principles 158

8.2 Nature of the violation of the CP 163
8.3 Irony and humor 168
8.4 All humor is intentional 170
8.5 Conclusions 175
8.6 Further readings 175

We’ve seen in Chapter 5 that we can arrive at two significant

generalizations about humor and namely that all meaning differences
may be used for the creation of humor and that all humor is semantically
based. This is not to say, however, that humor is purely a semantic
phenomenon. It is clear that humor has a significant pragmatic
component.1
We can now introduce a third generalization about humor: humor
may arise from the violation of pragmatic rules alone. Note that this
does not contradict the first two generalizations: pragmatic rules are
meaningful, albeit at a different level than semantic rules.

1 As we saw, the Semantic-Script Theory of Humor (and therefore the General Theory
of Verbal Humor) incorporate pragmatics within semantics and that therefore the argument
of whether humor is a semantic or pragmatic phenomenon is moot. It should be noted that
while it is true that pragmatic factors enter in the construction of the sense of a text, it is also
true that the text has a relationship with its context and that the overall meaning of the text
is given by the interplay of the semantically-and-pragmatically-arrived-at meaning and the
context.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 PRAGMATICS OF HUMOR

8.1 Pragmatic principles

The material in this chapter may require some linguistic background

that goes beyond what a typical introductory course in linguistics will
provide. Therefore, we will first introduce some pragmatic principles
that underlie communication, starting with speech act theory.

8.1.1 Speech acts

J. L. Austin and John Searle are the founders of speech act theory. As the
name of the theory indicates, speaking is performing a kind of action.
Speech act theory begins by distinguishing three different acts in each
utterance:

• Locutionary act: what we say literally.

• Illocutionary act: the commitment made by saying the locutionary
content. Examples are betting, promising, swearing, baptizing, and
naming. This is the force of the sentence; the activity the speaker is
trying to accomplish with it.
• Perlocutionary act: the effect or intended effect the utterance has
on the hearer.2 Surprising someone, scaring them, or amusing them
would be perlocutionary effects of an utterance.

Attardo and Chabanne (1992) defined a humorous utterance as one

whose perlocutionary goal is to amuse its audience or at least evoke
the recognition of the intention on the part of the speaker to amuse
his/her audience. This definition, far from being circular, emphasizes
the dependence of the humorous nature of the text on its context,
and the intentions of the speaker/hearer. Defining the humorous text
as a text whose perlocutionary goal is to be perceived as humorous
is a strictly functional definition, but it also ties the definition to the
intended perception of the text by an audience, i.e., the effect that the
speaker thinks his/her text will have on the hearer(s). In what follows,
we will expand slightly this definition, to highlight the potential role the
audience plays in defining something as humorous.

2 Recall that “hearer” is used as a technical term, meaning not just a lone hearer, but
possibly a wider audience, including overhearers.

158
 8.1 PRAGMATIC PRINCIPLES

8.1.2 Cooperation and implicatures

There are some special preconditions that are common to all sentences
of all languages and not just to speech acts. One such precondition is
that we assume that other speakers are telling the truth. Naturally, we
know that this is not always the case, and that speakers lie, on occasion.
However, as a general rule, we assume that speakers are truthful and
are not trying to deceive us. If we didn’t, communicating would be
impossible, since we could not trust anything said by other speakers.
Another assumption is that speakers are sticking to the point. Suppose
that someone asks you “What do you think of hip hop?” The answer, “My
sister just graduated from college” is not to the point and so irrelevant to
the question. In conversation, this would be considered a non-sequitur.
Thus, if we want to communicate, we must stick to the point of what
is being said. These assumptions are some of the rules that govern
conversations. These are general rules or principles that all speakers
follow unconsciously. They are not prescriptive rules. They are more like
good advice, along the lines of:
(63) Avoid Highway 635 in Dallas at rush hour.
You are free to go there, of course, but you’d better not if you don’t want
to get stuck in traffic.
Paul Grice, a British philosopher transplanted to Berkeley, codified
these general expectations as the co-operative principle (CP). The CP
consists of four maxims (in a nod to Kant, who states parts of his moral
philosophy as maxims). Taken together, the four maxims and their sub-
maxims form the principle of cooperation, which states roughly that a
speaker’s conversation should be as effective and cooperative as possible.
Grice’s original formulation is worth quoting:

Make your conversational contribution such as is required, at the stage at which it

occurs, by the accepted purpose or direction of the talk exchange in which you are
engaged. (Grice, 1989, p. 26)

The following are the four maxims:

1. The maxim of quality

• Do not say what you believe to be false.
• Do not say that for which you lack adequate evidence.
2. The maxim of relevance—Be relevant!

159
 PRAGMATICS OF HUMOR

3. The maxim of quantity

• Make your contribution as informative as is required.
• Do not make your contribution more informative than is
required.
4. The maxim of manner
• Avoid obscurity of expression.
• Avoid ambiguity.
• Be brief.
• Be orderly.

What happens if speakers do not follow the CP? There are several
cases. The simplest one is that communication breaks down. The speak-
ers become confused, do not understand each other and need to resort
to forms of repair, or abandon communication entirely. For example,
consider the following passage from Heidegger:
(64) Will we see the lightning-flash of Being in the essence of tech-
nology? The flash that comes out of stillness, as stillness itself?
Stillness stills. What does it still? It stills being into the coming to
presence of world. (Heidegger, M. (1977) The question concerning
technology and other essays. New York, NY: Harper and Row.
p. 49)
Example (64) clearly violates the maxim of manner and specifically the
submaxim warning to avoid obscurity of expression. It may violate other
maxims as well, but since I, for one, have no idea what Heidegger is
saying, it is hard to determine whether it is relevant in the context in
which it is said, etc.
Other options are giving up entirely on the communicative exchange:
socially codified forms of doing so are the expressions “no comment”
and “I cannot say more.” Non-socially ratified forms are simply aban-
doning the conversation entirely (i.e., stopping speaking, or changing
the topic entirely).
There are some modes of communication that allow violations of the
CP, under specific constraints. So fiction writing, playacting, and any
forms of creative uses of language have a certain amount of leeway. One
may freely make up events and circumstances and even invent entire
universes within fiction.3 The need to maintain suspense in a work of
3 The existence of genres such as the memoir, in which real-life stories are fictionalized,
is an ambiguous area. For its problems, and in general for the non-cooperative aspects of
fiction, see Attardo (2008).
160
 8.1 PRAGMATIC PRINCIPLES

fiction will require the author to withhold crucial information from the
reader/viewer (a violation of the maxim of quantity), such as the identity
of the assassin. Humor and jokes fall under this category, both because
they are mostly fictional and because they violate the CP in specific ways.
Other modes of communication, such as lying and deceiving are based
on the assumption that one’s violation of the CP will remain undetected,
at least long enough for one’s goals to be achieved. For example, some
rules of politeness require one to lie to avoid offending (“How do you like
my new haircut?”). It should be noted that white lies, such as the surgeon
lying about the health of his parents to the injured child whose parents
perished in the same accident, are plain violations of the CP but done for
purposes that society approves of (helping a child, not hurting someone’s
feelings). Regular lies are violations of the CP done for personal gain.
The distinction is ethical, not semantic/pragmatic.
In other situations the maxims may clash: thus one may want to be
truthful but lack adequate evidence for what one wants to say. Clashes
are problematic for the speakers, because there are generally no easy
solutions to the clash and the speaker may need to violate one maxim
in order to follow another one. Often the cause of a clash is politeness,
a principle not included in Grice’s original formulation, but that Leech
(1983) has argued should be given the same status. Thus one may think
a friend’s outfit to be hideous, but politeness will prevent him/her from
telling the truth. There exists one last case of violation, with special
properties, to which we turn next.

8.1.3 Implicature
Implicatures are violation of the CP that are obvious (“blatant”), where
there is no clash with another maxim or intention to mislead. Grice
labels these “exploitations” of the maxims. These exploitations are viola-
tions that are redeemed by the fact that, if we make the assumption that
the violation is deliberate and intended to be perceived as such, since
it is obvious and thus cannot be seen as intending deception, on the
part of the speaker, the hearer can work out another intended (implied)
meaning which can be inferred from the context of the communicative
exchange. Context, broadly defined, includes the co-text (the other
words in the text), the situation in which the utterance takes place, as
well as background information, and awareness of the CP itself. This
implied meaning is called an implicature. Implicatures have been used to

161
 PRAGMATICS OF HUMOR

explain metaphors, irony, and other tropes (understatement, hyperbole,

etc.). They are generally believed to be crucial to explaining figurative
and generally non-literal uses of language.
Implicatures can be usefully contrasted to logical inferences, which
follow from the meaning of a sentence: thus from the meaning of
(65) John is a doctor.
it follows logically and necessarily that
(66) John has a doctoral degree.
Inferences are necessary and undefeasible, witness the anomaly of
(67) * John is a doctor but he does not have a PhD, EdD, or MD.
Since we cannot say that John is a doctor and he does not have a
PhD, it follows that the inference is undefeasible. Contrast this with
implicatures: implicatures are probabilistic (i.e., probable, not certain)
and defeasible without contradiction. Contrast the following example:
(68) Mary and John are married, but do not love each other
which may be sad but is perfectly well formed, in which that two people
who are married should love each other is an implicature (and hence
defeasible) to
(69) Mary is single, but is married to John.
in which being single implies necessarily and undefeasibly that one is
not married. Note that it would be a logical contradiction to say that
Mary is single and not single.
A special class of inferences are presuppositions, which are inferences
that hold simultaneously for a sentence and its negation. So, considering
again (65), and its negation
(70) John is not a doctor.
from both (65) and (70) it still follows that there exists someone
whose name is John. Note how the existence of John is presupposed
by both (65) and (70) and that this is a necessary inference, witness the
anomaly of
(71) * John is a doctor but there is no-one called John.

162
 8.2 NATURE OF THE VIOLATION OF THE CP

8.1.4 Humor and the implicit

Implicatures, inferences and presuppositions are all implicit, in the sense
that they are part of what the text does not say literally and explicitly.
In this respect, they are crucial to addressing an heretofore seriously
neglected aspect of humor, and namely its connection with the implicit
of a text. The topic of humor and the implicit is in dire need of research
(see Attardo, 1994, pp. 289–290 for a review). Guagnano (2013) shows
that the difference between jokes and humorous anecdotes lies, among
other things, in the fact that whereas jokes keep implicit both the second
script and the resolution of the incongruity it brings about, humorous
narratives present both scripts and the resolution more or less explicitly
within the text.

8.2 Nature of the violation of the CP

There exists an extensive literature that regards humor as a violation of

the CP. As we saw, Raskin (1985) classifies humor as a non-bona-fide
mode (bona-fide communication follows the CP).
There have been a few attempts to claim that humor is not a “real”
violation of the CP, either by claiming that the violation is merely men-
tioned, or attributed to a character in the text, or otherwise normalized,
for example, by turning it into an implicature (Klungervik-Greenall,
2003; Dynel, 2009b; Goatly, 2012). Mooney (2004, p. 915) claims that
humor is a non-successful violation and that therefore implicatures
follow from it. There are serious problems with these views, some of
which are refuted in Attardo (1994, 2017c). We address the idea that the
violation of the CP in humor leads to a flout below.

8.2.1 Is humor a flout?

Recently, an interesting proposal has been presented by Goatly (2012:
235) to see humor as a short-term violation (or as he has it “a flout
delayed by violation”). Goatly explain that “the breaking [=violation]
of the maxims must be recognized almost immediately, at least within
the next couple of turns of the discourse for the joke to work” (235).
There is a problem with this idea, which Goatly acknowledges in a

163
 PRAGMATICS OF HUMOR

footnote, and namely that these flouts would behave differently from
all other flouts. For starters, they are not defeasible and they do not
generate implicatures. More importantly, the violation is not meant to
be recognized immediately, at the time of the first processing of the text,
as it is in flouts.
Let us consider the following joke
(72) My wife accused me of being immature. I told her to get out of my
fort.
From the first sentence we recover the following relevant inferences:
the speaker is male and the speaker is married.4 From the script for
marriage we also derive the inference that the speaker is of legal age to
be married (hence at least 18 or 16 with parental consent, in Texas) and
the implicature that the speaker is probably at least in their mid-twenties
(only about 7% of men marry before 20, according to the US census
data). The second sentence however conflicts with the script we have
constructed so far (married man older than 20) since the lexeme fort
activates the script fort which itself activates game and child (children
play in make-pretend forts). The incongruity of why a married man is
playing a children’s game is not resolved, but there is a minor resolution
aspect in the fact that the speaker of the text (another presupposition:
from “my” and “I” we know the text is in the first person) considers the
response of expelling his wife from his play-fort an appropriate answer
to the accusation of immaturity.
How does the text violate the CP? First, the text is fictional. There is
no evidence that this exchange might actually have taken place, at any
time. To put it bluntly, this is not a true story. Hence, this is a violation
of quality.
Second, the text withholds significant information in the first sen-
tence. The text does not focalize the age of the speaker nor the degree
of maturity of the speaker until the second sentence, which contains
the punch line “my fort” (the last phrase of the last sentence of the
text). Consider that a fully informative text should have been shaped
as follows:
(73) A man who still engages in games inappropriate for his age says:
“My wife accused me of being immature. I told her to get out of
my fort.”
4 Gay marriage is still uncommon enough that it would have been topicalized if this had
been the situation.
164
 8.2 NATURE OF THE VIOLATION OF THE CP

Third, since no explanation is provided within the text for the

incongruity between the presumptive age of the speaker and his
behavior, it is difficult to assess the relevance of the second sentence
to the challenge presented by the wife to the husband. Presumably, a
self-defeating response is not a good argumentative move and so it
may be considered irrelevant, since a relevant response should have
addressed the wife’s allegation of immaturity, not confirmed it.
In light of this analysis, the text is clearly non-cooperative (non-bona-
fide). However, we can agree with Goatly that the recognition of the
violation occurs very soon after it occurs. From psycholinguistic studies,
we know that incongruity is detected about 400 milliseconds after it
occurs and resolved about 200 milliseconds later. However, from the
second and the third violation, it is hard to see what the implicatures
triggered by the flouts would be (we easily derive an inference “the
speaker is immature”, but this does not lead us to a coherent train of
thought, what Grice calls “working it out”). Instead it takes us to an
interpretive cul-de-sac: the text is contradicting itself. The speaker is
both immature and old enough to contract marriage (which would
probabilistically implicate maturity). Moreover, the “implicature” (let
us treat it as such for the sake of the argument) is not defeasible as
witness by:
(74) The man is immature, despite his age, but he is not immature.
In a normal implicature, the speaker flouts the CP trying to get
the hearer to work out an intended meaning. In humor, some of the
violations lead to no intended meaning at all, only contradiction and
incongruity. Note, in passing, that trying to claim that the perception
of humor or the triggering of laughter is the intended “meaning” that
the speaker is trying to convey, changes irresponsibly the definition of
implicature (which is now broadened to include perlocutionary goals).
So, in conclusion, while Goatly is correct that the violation of the CP in
jokes lasts for a very short time, in contrast with lies, for example, whose
duration is indeterminate, it seems difficult to categorize the violation
as a flout. Simply put, humorous violations of the CP do not generate
implicatures, or if they do, they are not the same kinds of implicatures
that flouts generate outside of humor.
In conclusion, humor is non-cooperative (non-bona-fide). It shares
this characteristic with other modes, such as fiction. However, we should
not assume that humor is therefore nefarious. There are different kinds

165
 PRAGMATICS OF HUMOR

of non-cooperation: some are antisocial (such as deception) and some

are pro-social (humor and fiction would fall under this category). Speak-
ers are quite willing5 to engage in pro-social non-bona-fide modes, for
example for the purpose of entertainment (fiction, humor). We cannot
pursue the discussion of anti-social lies and other fully and partially
non-cooperative forms of communication (propaganda, advertising,
sales talk, for example) but we will look briefly at the way speakers
handle non-cooperation.

8.2.2 The LDP

Violations of the CP are not random; they follow certain rules. As a first
approximation, the Least Disruption Principle (LDP) has been proposed
(Attardo, 1999, Eisterhold et al., 2006). The LDP’s wording is as follows:
Super-maxim: Minimize your violation of the CP;

1. limit your violation of the CP to the smallest possible conversa-

tional unit (one utterance, one conversational turn, one speech
exchange);6
2. try to link the entire CP-violating unit to the rest of the interac-
tion, for example by finding a certain appropriateness to the CP-
violating unit;
3. limit your violation of the CP to the smallest possible distance from
its requirements;
4. if you must violate a maxim do so in the direction that is expected
by your audience (i.e., say what your audience wants to hear).

The LDP is in part a set of advice of how to effectively violate the CP, or
in other words how to effectively make as if one were producing a bona-
fide speech act, while in fact one is not. Some of the maxims of the LDP
concern the degree of the violation (smaller violations are less likely to
be detected), and the direction of the violation (tell your audience what
they want to hear); clearly, they are instructions on how to lie, obfuscate
and/or inveigle.
5 As Coleridge aptly stated, it is a “willing suspension of disbelief.”
6 Conversational demands, such as preferred and dispreferred responses may require
longer stretches of discourse than a single turn. For example, in Southern Italian culture,
while at the dinner table, one should refuse offers of seconds for food at least twice before
accepting. In order to non-cooperatively refuse food, one then needs at the very least 6 turns
(3 for each speaker).

166
 8.2 NATURE OF THE VIOLATION OF THE CP

Others maxims of the LDP concern the interplay between the

violation and the non-violating environment. So, if the speaker wants
the hearer to be aware of his/her violation of the CP, it is better to keep
the violation short, because if the violation is prolonged, the hearer may
lose the motivation to engage in the interaction (why waste resources
processing known non-bona-fide messages, from which no implicatures
can be drawn?). Needless to say, if the speaker does not want the hearer
to be aware of the non-bona-fide nature of the exchange, then no such
need to minimize the length of the violation exists. This second aspect
of the LDP is predicated on the fact that even when violating some of
the most central norms of social interaction, agents are still within
the scope of their social bonds. There are sanctions for those who
engage in inappropriate violations (more or less formalized, ranging
from prosecution for perjury to social shunning).
Empirical evidence in favor or against of the LDP is hard to come by.
Much has been made about the existence of long stretches of humorous
banter in which speakers either support one another’s humor or top
it (i.e., produce more or better humor), e.g., Kotthoff (2007). How-
ever, even these extended periods of joking never exceed 15 turns (the
longest documented one is 13 turns) and need to be continuously keyed
(Hymes, 1972, p. 62) or framed (Goffman, 1974, pp. 43–44) (marked)
as humorous (e.g., Zhang and Attardo, 2015). Framing or keying for
humor can be achieved in many different ways. One is to produce humor
support (Hay, 2001) such as producing more humor; see Chapter 10.
Other strategies include laughter, and changes in eye contact and smiling
patterns (extended, more intense smiling; see Section 10.3).
Recent results corroborating the idea that speakers return to the
serious mode as soon as possible include Tsakona (2011), who found
that the ironical turns in a Greek parliament discussion did not exceed
three utterances, and only two instances were cases of mode adoption
(i.e., the hearer continuing the speaker’s irony) and Ruiz Gurillo (2009)
who found that 48.5% of her corpus of ironical conversational exchanges
consisted of single turns and 68.5% consisted of three or less turns.
Holmes and Marra (2002) report that single-turn humor in workplace
conversations never accounted for less than about 40% of humorous
turns and in one setting (factory floor) single-turn humor outnumbered
two-or-more-turns humor. Nuolijärvi & Tiittula (2011) in an analysis of
Finnish television debates find that,

167
 PRAGMATICS OF HUMOR

The recipient [of the irony] usually treats the irony-implicative utterance as literal ( . . . )
the recipient overtly bypasses the ironic meaning and ignores the negative evaluation in
it, but simultaneously ironises the original ironist or his/her talk. Thereafter, the original
ironist in turn responds literally, so that the conversation shifts back to the serious, non-
mocking mode. In other words, the original ironist continues as if he or she had not
been ironic at all. (p. 584)

To paraphrase, the Speaker is ironical, the Hearer responds to the literal

meaning, but mode adopts, i.e. responds with irony to irony (see Section
11.3.1.3), to save face. The original Speaker responds also literally, hence
completing the return to the serious mode. In this case, return to the
serious mode is delayed by one turn to save face (essentially an ironical
tit-for-tat).

8.3 Irony and humor

There is a common misconception that irony is saying the opposite

of what one thinks. In fact, irony is a much broader phenomenon. In
pragmatic terms, saying the opposite of what one thinks corresponds to
a violation of the maxim of quality. However, convincing evidence has
shown that the violation of any maxim may generate irony. Good exam-
ples are understatements and exaggerations: for example, in torrential
rain, one might say,
(75) It seems to be drizzling.
Clearly, these are violations of the maxim of quantity, not of quality;
after all there is some precipitation. It is the intensity of the precipitation
that is inappropriate. Importantly, ironical violations of the maxims are
meant to be recognized as such and thus generate implicatures, unlike
the violations of jokes, which are not meant to be recognized at the same
time that they are first processed.
The theories of irony are fairly complex and many require fairly
sophisticated psycholinguistic apparatus, which is beyond this treat-
ment (see Giora, 2003 for a synthesis of the debates). So we will limit
ourselves to a few observations that will place irony within the context
if humor research.
The difference between irony and sarcasm, which has elicited sig-
nificant discussion, is actually fairly simple to describe: in the classical
model, sarcasm is a particularly aggressive form of irony, whereas irony

168
 8.3 IRONY AND HUMOR

is a gentler, less critical type of discourse move. Unfortunately, semantic

shift took place in the USA in the 1990s and the meaning of “sarcasm”
took over the entire semantic field of irony/sarcasm, with “irony” being
relegated to the meaning of an annoying coincidence (see Section 1.1.2).
Practically speaking, this means that one needs to be very careful, when
reading studies on irony/sarcasm, concerning the age and geographical
location of the subjects, as they may have different understandings of
the folk-terminology irony and sarcasm.
Irony is not necessarily critical of its target. Irony can be used to
convey a positive evaluation of the target. For example,
(76) J. K. Rowling published a few moderately successful novels.
This example of positive irony draws attention to the fact that Rowling’s
Harry Potter novels were very successful. Another example, which Alan
Partington drew my attention to, comes from Wodehouse:
(77) “Shakespeare said some rather good things.”
“I understand that he has given uniform satisfaction, sir.[ . . . ]
(P. G. Wodehouse. Jeeves and the Tie that Binds. New York: Simon
and Schuster. 2000. p. 53.)
In (77), the description of Shakespeare’s work as having given “uniform
satisfaction” is of course inappropriate, as Shakespeare is widely consid-
ered the greatest English writer. Since the text is using a characterization
of Shakespeare’s success that is worse than the reality, it is a negative
utterance. However, since it implies a positive evaluation it is a positive
irony.
There is a consensus that irony comes from a contrast, in a sense close
to the idea of opposition in humor. This is clearly correct, for example,
in the torrential rain vs. drizzle example there is a clear contrast or
opposition between the two ideas. However, contrast is not enough to
generate irony (“I don’t like beans, but I like tomatoes” also includes a
contrast, but is patently not ironical). I have suggested (Attardo, 2000a)
that in order to be perceived as ironical, a contrast has to be both relevant
and inappropriate. Let us examine both aspects.
It is clear that irony is a flout in Gricean terms. The speaker says X,
while intending something else (Y). Y is arrived at using the maxim
of relevance as a heuristic. When looking for a possible interpretation
of an utterance, the maxim of relevance is one’s default tool, as the
assumption that the speaker is meaning something relevant will guide

169
 PRAGMATICS OF HUMOR

one’s search. For example, suppose that the speaker says “It’s drizzling”
the hearer would be well served by looking for weather-related items in
the immediate context: for example, he/she may look outside a window
to check the weather, or turn on the weather channel, or look at a
thermometer, etc. Checking for the presence of tomato cans in the
pantry would make no sense, because tomato cans are not relevant to
the weather. Let us suppose, for the sake of the argument, that the hearer
can locate a window in the immediate context and look outside where
there is a torrential outpour. The hearer may then infer that the speaker
is trying to communicate something else than the literal statement “it is
drizzling.”
This is a flout, first, because the speaker is clearly trying to convey
something else than what he/she literally said, but also because the
utterance is inappropriate7 in this context (it is raining and the speaker
says it is drizzling). If I assume that the speaker is being deliberately
inappropriate but also relevant (i.e., the reference to the drizzle is related
to what he/she intends), then I can conclude that the speaker is trying to
convey ironically that he/she feels the rain is annoying, or unexpected,
or perhaps that someone had predicted only a small amount or rain
(this is called echoic irony, because it echoes critically someone’s words).
Notice how a simple flout of the maxim of quantity (drizzle vs. rain)
is not ironical per se. If it were drizzling and the speaker exaggerated
saying “it’s raining,” there would be no perception of irony. In order to
be ironical, what is said to generate the flout has also to be inappropriate.
This extra inappropriateness is what I have called “residual violation”
(because it persists after the flout has been resolved; Attardo, 2001a,
p. 111) which causes irony to be often perceived as humorous.

8.4 All humor is intentional

This section argues that all forms of humor must be either uttered
or interpreted as intentional distortions of normal, serious, bona-fide
communicative practices. A direct corollary of this thesis is that the
category of “unintentional humor” turns out to be have potentially
confusing effects, and should be avoided in the discussion of humor.
7 Inappropriateness is defined as a mismatch in presuppositions (Attardo, 2000a). Note
that the presuppositions of “it’s drizzling” are calculated based on that sentence, and on “it’s
not drizzling,” as we saw above.

170
 8.4 ALL HUMOR IS INTENTIONAL

The theory of involuntary humor is mainly implicit, i.e., it is assumed

as self-evident without discussion. For example, Freud (1905: 182) notes
that “naive” humor is “found” and not “made” and leaves it at that. A
problematic aspect of the folk-theory of involuntary or unintentional
humor is that of distinguishing humor from all sorts of errors and other
phenomena such as slips of the tongue, spelling errors, typing mistakes,
typographical errors, etc. Not all ambiguous sentences are puns.
Consider the following example (from Sherzer, 1978):
(78) In a course in Linguistics there’s a lot to cover.
This sentence appears quite straightforward, and its meaning is largely
unambiguous. However, example (79) shows that in fact there is a
hidden ambiguity in the lexeme “cover” which can be exploited for
punning:
(79) In a course in Human Sexuality there’s a lot to cover.
Note that the ambiguity of the lexeme “cover” is also present in (78).
One can therefore ask the question: “Why isn’t (78) also funny?” The
structural requirements for a pun are all there (ambiguous element,
opposed and overlapping scripts hide and examine, as in “peruse”),
so what is missing? The missing element in (78) is the intention of
the hearer/reader to make a pun. Suppose to read (79) in any situation
where the contextual clues about the speaker are abolished. There is
no telling whether the inscription is meant as a joke, or is the weary
comment of a Human Sexuality teacher. What will lead the audience
to opt for the HIDE script, rather than for the EXAMINE script?
Clearly nothing in the structure of the text itself, as the option did not
present itself in (78). Inevitably one must reach the conclusion that
the audience is alerted to the ambiguity of “cover” by the activation
of the default SEX humorous script. In other words, the audience (the
reader/hearer) will see the second meaning of “cover” because it “is
on the look out” for hidden meanings. In fact, even (78) can become
as much of a pun as (79): suppose that a given teacher of linguistics
considers some influential trend in linguistics a dangerous influence for
his/her students, and he/she wants to shelter them from this influence,
then (78) becomes (suddenly) a reasonably good pun. (79) can lose all
its humorous potential in the right situation: consider the case in which
(79) is uttered by an untenured junior faculty member who has failed
to cover significant parts of the syllabus in his/her Human Sexuality

171
 PRAGMATICS OF HUMOR

class, and who has solicited a meeting with his/her Department Head
in a desperate plea not to be fired. In conclusion, it is safe to say that an
ambiguous sentence becomes a pun depending on the context and on
how the hearer is disposed towards the communicative situation.
Consider this example:
(80) This writer, a non-native speaker, was once in need to discuss a
delicate situation with a female faculty member. So I asked as a
favor of my advisor if he could “feel up” the faculty member about
that issue.
The above error (lexical replacement) was easily corrected (“feel
up/feel out”) and therefore the hearer could very easily determine
what the intended meaning was, and proceed accordingly, disregarding
the parasitic sexual meaning. It should be noted that whenever the
conditions of communication are sub-optimal (e.g., noisy background,
physical illness, etc.) the hearer goes to a lot more trouble to reestablish
the speaker’s intended meaning. If the hearer wants the communicative
exchange to succeed he/she must attempt to reconstruct as far as pos-
sible, the speaker’s intention. Clearly, by choosing to disregard the fact
that in example (80) the speaker did not intend to evoke a sexual script,
the hearer is leaving the realm of cooperation and violating the principle
of Charity, which would enjoin the hearer to assume that the speaker is
making sense. “Feel out” makes much more sense than “feel up” in this
context.8
In other words, it is up to the hearer to decide whether he/she
will respect the cooperative principle or will deliberately misinterpret
the text by refusing to reconstruct the speaker’s intention. It should
be added, to ward off a possible objection, that most sentences are
potentially ambiguous, and thus almost all sentences can be deliberately
misinterpreted as containing hidden sexual references (sex being the
default script for humor).
Consider finally the following classical example, borrowed from Freud
(1905, pp. 183–184) and here summarized:

8 There are numerous formulations of the principle of Charity. For our goals, we will
adopt the definition that requires the speakers to maximize the number of true sentences
uttered by each other. In other words, the speakers should try to interpret what each other
says assuming that the other speakers are trying to make sense as much as they are.

172
 8.4 ALL HUMOR IS INTENTIONAL

(81) Two children are performing in front of their family a play they
themselves wrote. The plot revolves around a fisherman who
leaves his wife to go in search of fortune. In the second act, set
several years later, the fisherman returns with a huge fortune.
His wife greets him by saying: “I too have not been idle.” “And
thereupon she opens the door of the[ir] hut and reveals to his eyes
twelve large dolls lying asleep on the floor . . . At this point in the
drama the actors were interrupted by a storm of laughter from the
audience, which they were unable to understand.” (Freud, 1905,
p. 184) [my emphasis, SA]
Normally adults cut some slack to children, and assume that they
cannot really mean the sexual overtones that an adult would see in a
given situation. So the question becomes “why the adults in Freud’s
story find it amusing, and do not dismiss it as a child’s error?” The only
hypothesis that seems plausible is that the adults are projecting their own
thought patterns on the children, thinking, as it were, that the situation
would be very funny, if the sentence had been uttered by someone who
would be aware of the implicatures of the situation (namely, adultery). If
the adults processed seriously the situation the inferential paths would
be (greatly simplifying): she shows twelve children she had while her
husband was away; however, in her knowledge of the world children
have nothing to do with sex (say, they are brought by the stork), therefore
no sexual innuendo is meant, and the implication of the situation is that
she has been busy too. Yet, the adults choose to process the situation
as if the child were aware of the fact that having children presupposes
sexual relations, a fact for which they have (or presume to have) evidence
that it is false. Therefore, the adults are being non-cooperative and
uncharitable.
Let us assume, for the sake of the argument, that any of the examples
above were not intentionally mis-intepreted but erroneously: in other
words that the hearer interprets the children’s utterance erroneously
as intending the sexual implicature. In this case, the hearer should
experience surprise, dismay, and perhaps even concern that such young
children would be aware of sexual matters. To react with laughter implies
that the hearer is at some level aware of the fact that the children cannot
mean what they have just implied.9 Still, we could argue that in the case

9 I am using “mean” here in the folk meaning, not as a technical term in Gricean
pragmatics.

173
 PRAGMATICS OF HUMOR

of an honest mistake, the hearer laughs because he/she is surprised by

the incongruous situation (for example, of a student asking his advisor to
feel up a colleague). Needless to say, this is not the case in reality, because
upon correction of the misapprehension, the hearer should apologize
to the speaker, along the lines of “I am sorry, I misunderstood you, I
thought you were saying that X” and I am not aware of any claims that
this happens habitually after cases of unintentional humor (although I
am not saying that it cannot happen).
The crucial aspect of unintentional humor lies not in the intention
of the speaker or the text being uttered, but in the way the audience
interprets the utterance. And since it has been shown that the audience
interprets the utterance either erroneously or not along the lines of
Grice’s cooperative mode, but rather along a non-cooperative mode for
humor (non-bona-fide), it appears that the audience is responsible from
the humorous interpretation of the text. Two corollaries follow from this
conclusion: 1) If the audience is aware of the fact that they decide to
move to the non-bona-fide mode of communication, they intentionally
interpret the utterance in a humorous way; if the audience is not aware
that they misinterpreted the utterance, they are nonetheless aware (or
at least they act as if they were so—I make no claim for the fact that
the audience is in fact consciously aware of anything) that they are
interpreting uncharitably the utterance, by the very fact that they find
it ridiculous.10 2) the interpretation is in fact a misinterpretation (either
deliberate or not) since, as it has been shown of Freud’s example above,
the communicative intention of the speaker is (deliberately) altered by
the hearer/audience. From these corollaries it follows naturally that no
text that is perceived as humorous is unintentionally such: either the
speaker or the hearer are always aware that the principle of cooperation
is being violated. They may not both be aware of the fact at the same
time, but one always is. When the hearer finds humorous a text not
meant to be funny he/she is intentionally misreading it as being funny,
whereas at best the text can be faulted for being ambiguous (as is often
the case with “funny” headlines, typos, etc.) or somehow inappropriate.
Thus we can formulate a fourth significant generalization about humor.

All humor is intentional, either in the intention of the speaker, the hearer, the audience,
or any combination thereof.

10 A charitable interpretation of the children example in Freud would of course be that

they cannot possibly mean what they are saying, etc.

174
 8.6 FURTHER READINGS

8.5 Conclusions

All humor involves a violation of the principle of cooperation and of

other principles such as appropriateness and charity. These violations
are real, if short lived. They are not flouts or otherwise pretend vio-
lations. There is some evidence that the speakers tend to contain the
violations of pragmatic principles, along the lines of the Least Disruption
Principle. The interpretation of the humor is guided by the maxim (or
principle) of Relevance. Some or most irony is humorous. Irony is a flout,
but that does not make irony humorous. There needs to be more of a
violation for irony to be humorous. Finally we added two significant
generalizations: humor may arise from a pragmatic violation and all
humor is intentional, either on the speaker or the hearer’s side, i.e.,
someone has to intend (key/frame) the situation as humorous.

8.6 Further readings

There are many introductions to pragmatics, but Austin (1962) and

Searle (1969) are clear and worth reading in full. Grice’s (1989) collec-
tion of his work published during is lifetime is only for the advanced
student. A discussion of the violation of the maxims and the CP in
humor is found in Attardo (1994, Chapter 9). Leech’s (1983) discussion
of politeness is also accessible. On humor and relevance, see Yus (2017).
There is no accessible discussion of the principle of charity but I found
Jackman’s (2003) quite useful. The least disruption principle is discussed
in some detail in Eisterhold et al. (2006). The discussion of irony as
relevance and inappropriateness is discussed in more detail in Attardo
(2000a). A good advanced book on irony is Gibbs and Colston (eds.)
(2007). The connection between humor and irony is complex and goes
beyond an elementary treatment. See Hidalgo Downing and Iglesias
Recuero (2008) for a good discussion. The section on intentionality fol-
lows closely Attardo (2003), which is however only available in Russian
translation.

175
 9
Verbal humor

9.1 Defining puns 177

9.2 Classifying puns 180
9.3 Ways to bring about two meanings in a text 181
9.4 The fate of the interpretation 191
9.5 Cratylism: resolution in puns 192
9.6 The psycholinguistics of puns 196
9.7 Conclusions 197
9.8 Further readings 198

In this chapter we deal primarily with puns, but not exclusively. The
title “verbal humor” may be confusing due to the use of the other
term “verbalized humor,” primarily in translation studies. Verbalized
humor is simply all humor expressed through language, as opposed to
through facial expressions, gestures, etc. In multimedia studies which
deal with multimodal humor (e.g., video, television, film, theater, etc.)
likewise, verbalized humor may usefully distinguish linguistic humor
from slapstick or visual humor. Verbal humor, conversely, is humor that
is crucially focused on the signifier, the phonemic support of language.1
It is opposed to referential humor, i.e., humor that does not require a
special attention to the linguistic form. The nature and modalities of the
“focus” on linguistic form for verbal humor will be one of our primary
concerns in this chapter.

1 There is also verbal humor focused on orthography and on the shape of signs in sign
language and in ideographic languages, such as Chinese. We will ignore these in this context.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 9.1 DEFINING PUNS

However, before we jump into the treatment of the subject matter

of this chapter we need to make a very important clarification: verbal
humor, puns and all the other forms of verbal humor we will explore,
is always without exception, semantic humor. Verbal humor has an
incongruity and a resolution, exactly like referential humor. In fact,
as we saw, Raskin’s SSTH does not distinguish between verbal and
referential humor and treats them identically (i.e., finds a script overlap
and opposition). The focus in the literature on the taxonomies, often
based on the form of the signifier, and on other phonetic/phonemic
aspects (for example the phonemic threshold of difference between
paronyms, see below) has obscured this fundamental fact which may
mislead one into thinking that verbal humor is just a phonological issue.
That would be wrong indeed, since we’ve seen that there can be humor
only if there is a semantic opposition.2
Another significant distinction is that not all language play is verbal
humor. Play is one of the important factors of humor, as we saw in
Section 3.3.1, but of course not all play is humorous; just think of com-
petitive sports. Some authors use “wordplay” as a synonym of “puns”
or what I have termed “verbal humor,” whereas others use wordplay
as a broader concept subsuming puns. Humorous language play would
definitely be the same thing as verbal humor. Regattin (2015) suggests
using wordplay for puns and word games for non-humorous language
games.
Playing Scrabble, doing crossword puzzles, or anagrams are all met-
alinguistic games based on orthography and while they constitute exam-
ples of language play they are not, or are only marginally, humorous.
Moreover, they are “artificial” in the sense that they require training,
have special rules, etc. whereas verbal humor is done by all speakers
of languages without any special training. There is nothing wrong with
language play but it falls outside the purview of this book (except insofar
as it is used alongside humor in classroom activities; see Chapter 16, on
humor and teaching)

9.1 Defining puns

In Attardo (2018) I provided a definition of puns, here slightly altered

for clarity:
2 A pragmatic one will do too, to be fair. Of course, it’s all meaning anyway.

177
 VERBAL HUMOR

A pun is a textual occurrence in which a sequence of sounds must be interpreted with a

formal reference to a second sequence of sounds, which may, but need not, be identical
to the first sequence, for the full meaning of the text to be accessed. The perlocutionary
goal or effect of the pun is to generate the perception of mirth or of the intention to do so.

This is dense prose indeed, so let’s unpack it in what follows. Puns

occur in a text, i.e., in a linguistic event, which takes place in a context,
has a speaker and hearer, etc. In this respect, they are like any other
linguistic phenomenon. The next part is more important: puns are
defined in terms of a sequence of sounds. First, note that it is sounds, nor
letters (graphemes); puns are by and large spoken language phenomena.
Second, it is a sequence of sounds (not a string, in which the sound have
to be contiguous, to be technical), so what we mean is that the sounds
in question may be scattered in the utterance. By “interpreted,” we mean
the usual processing of language to make sense of an utterance. The next
step is more complex: the interpretation of the utterance must refer (take
into account) formally to a second sequence of sounds to get the “full”
meaning of the text. Incidentally, the word “formal” in the definition
specifies the nature of the reference to clarify that in puns we refer to
similarity or identity in signifier only, and not to the similarity or identity
of words/sentences (as in paraphrase, for example).
Let me add that I do not mean to imply that this is an original defini-
tion or even that it is much of an improvement on previous definitions.
In fact, Delabastita’s (1996, p. 2) definition of wordplay (puns) is as
follows:

Wordplay is the general name for the various textual phenomena in which structural
features of the language(s) used are exploited in order to bring about a communicatively
significant confrontation of two (or more) linguistic structures with more or less similar
forms and more or less different meanings. [emphasis in the original, SA]

and the reader can easily see that there is not much of a difference
between the two.
Puns have a duality of meaning. What does this mean? Assume you get
a first meaning out of a text. If you miss the second meaning you still get a
text, with a meaning, perhaps a perfectly valid one, or perhaps nonsense,
but you do get a meaning. Unfortunately, you missed the second one and
with that the humor. This is crucial, so let’s work it out in more detail.
Consider the following example:
(82) When is a door not a door? When it’s a jar.

178
 9.1 DEFINING PUNS

assume that you do not see the “ajar” meaning, but only the “a jar”
meaning. The text is nonsensical (a door is not a jar) but it can stand
alone as nonsense. Consider now the following excerpt from a conver-
sation (Hay, 1994, p. 46):
(83) Meena: I’m the only person in this room who freestyled at nation-
als and came second in the women’s division
Dan: let’s face it Meena always comes second
Meena: yeah i know that’s cause Sue Willis always beats me (.)
except at distance
Dan: we were actually making sexual innuendos well I was
Here the pun on “comes” (arrives vs. has an orgasm) provides two
perfectly meaningful senses, so much so that the target of the joke
(Meena) does not get it, i.e., misses the pun, and addresses the only
meaning she is aware of. So, this is the nature of the “presence” in the
text of the two meanings.
The next step is the caveat that the sequences may be identical (as in
the two examples above) but need not to be so. When the sequences are
identical we have homophones or homographs (homonyms), whereas
when the sequences are different, we have paronyms. The following is
a pun based on a paronym, which we already encountered as (45) in
Chapter 7:
(84) Q: What do you get when you cross a cow and a lawnmower?
A: A lawnmooer.
The two strings [ˈlɔnmoʊər] vs. [ˈlɔnmʊər] differ by one phonemes. We
will return to this aspect in more detail below.
Finally, the definition calls for the perlocutionary goal (or effect) of
any of the participants in the exchange to be either to generate the
experience of mirth (amusement) or the have the intention to do so
recognized.3
The distinction between perlocutionary goal and effect is necessary
because involuntary puns may be detected and interpreted as humorous
only by some member of the audience, not necessarily the speaker or the
addressee, and no intention to amuse may be present in the speaker.
3 Some readers will recognize a nod to the Gricean M-intention (Grice 1957, 1989,
p. 105), later dubbed reflexive intention by Searle (1969, p. 47). Those that don’t should
not feel bad: that is seriously technical stuff, best left to a pragmatics book. Suffice it to say
that the intention of the recognition of the intention and not just having the intention is
important: for the text to be funny to you I have to want you to be aware of my intention.

179
 VERBAL HUMOR

Finally, why do we talk about strings or sequences of sounds and not

words? Quite simply because, even though many puns are based on
words (as are the examples above), not all of them are. Compare the
following:
(85) Be alert! Britain needs more lerts.
which dates back to World War 2 England (Pollack, 2012, pp. 138–139)
and is based on the reanalysis of the word “alert” in two strings a+lert,
neither of which is a morpheme, let alone a word, in that context. On
the other side of the morpho-lexical continuum some puns operate on
multi-word units, cf. the title of Pollack (2012)
(86) The pun also rises
where the relevant unit is the phrase “The sun also rises” (the title of a
novel by Hemingway).
So, going back to the distinction between referential and verbal
humor, the necessary and sufficient condition of a text to be humorous
is the presence of a script opposition and overlap. The necessary
conditions for a text to be a pun are a) the presence of a script opposition
and overlap, and b) the reference to a string of sounds, in the sense
discussed above.

9.2 Classifying puns

Most of the literature on puns is taken up by a classificatory frenzy

unseen in any other field of humor studies. The interested reader
may consult Attardo (1994, pp. 112–128) for a review of the main
taxonomies. Attardo produced a taxonomy of taxonomies of puns,
which follows:

• based on linguistic phenomena (e.g., homophony, homography,

paronymy, etc.)
• systematic (based on linguistic categories; e.g., syntagmatic,
paradigmatic, etc.)
• based on surface structure (e.g., the phonetic distance between the
two phonetic strings punned upon)
• eclectic (i.e., they pick and choose any criteria)

180
 9.3 WAYS TO BRING ABOUT T WO MEANINGS IN A TEXT

However, the whole operation is ultimately disappointing because

however detailed and comprehensive one’s linguistically-based taxon-
omy is, it is ultimately defeated by the richness of language. Another
strategy is to go for very broad categories: paradigmatic vs. syntagmatic,
for example, i.e., things that co-occur in the string (syntagmatic) or
things that might occur in a given slot (paradigmatic).
There is another drawback: all taxonomies presuppose a theory; a
theory of language, or a theory of puns, or a theory of psychology, and so
forth. The authors of taxonomies are not always explicit on what their
underlying theory is. Taxonomies find what they expect to occur. As
such what they exclude tells us a lot about what their proponents used as
data to build the taxonomy and what their (implicit) theory is. Needless
to say, explicit theories are better than implicit ones. The problem is that
taxonomies cannot replace theory building, if for no other reason that
taxonomies do not have explanatory power, beyond the (implicit) theory
they are built on.

9.3 Ways to bring about two meanings in a text

As we saw, one of the necessary but not sufficient conditions for a pun
is the presence in the text of (at least) two senses. Yet this is no different
from non-punning humor (referential humor). The difference is that in
puns the surface structure of the text (its form) is affected and not only
its meaning. In this section, we will look at some of the ways to bring
about the duality of meaning.

9.3.1 Ambiguity
Ambiguity is one of the most salient among the many ways of bringing
about the presence of two different senses in the text. While it has
received a lot (possibly most) of the attention, it should be clarified that
it is not the only way to do so. Not all ambiguity is humorous of course.
William Empson dedicated his entire book (1930) to seven different
types of ambiguity that generate esthetic effects and very little of it is
humorous.
The most extensive and detailed study of verbal ambiguity in English
is Oaks (2010), which dedicates a lot of attention to humor. The most

181
 VERBAL HUMOR

natural classification of ambiguity is between lexical and structural

ambiguity. There are other forms of ambiguity, of course. Referential
ambiguity is properly semantic ambiguity. Of course, pragmatic ambi-
guity is also possible. It should be noted that the so-called “phonological
ambiguity” (Seewoester, 2009) is subsumed by the string-based defini-
tion of puns used in this chapter.
9.3.1.1 Lexical ambiguity
Lexical ambiguity is found whenever a string has two meanings associ-
ated with it. Often strings coincided with lexical items (lexical ambigu-
ity), but it does not need to be this way, as example (85) demonstrates,
since it operates on the string “lert” which is not a lexical item or even
a morpheme (those with etymological knowledge would analyze it as
Italian alla+erta “all’erta” = to the watch spot).
9.3.1.2 Syntactic ambiguity
In syntactic ambiguity, two different syntactic structures may be con-
nected to the same surface structure.4
The standard example is an old joke from the Marx Brothers:
(87) Captain Spaulding (Groucho): One morning I shot an elephant
in my pajamas. How he got in my pajamas I don’t know (Animal
Crackers; 1930)
The ambiguity comes from the PP attachment: “in my pajamas” could
modify (be attached to) the direct object “an elephant” or to the VP
“shot” (see Figures 9.1–9.2). Semantic knowledge disambiguates toward
the interpretation that the shooting took place while the speaker was still
wearing his pajamas (Figure 9.1), but the occurrence of the disjunctor
(“How he got in my pajamas”) forces a second reading, in which the
pajamas modify the elephant (Figure 9.2).
Another form of syntactic ambiguity is categorical ambiguity, doubly
exemplified in Figure 9.3.
(88) Squad helps dog bite victim
(Bucaria, 2004, p. 292) where the syntactic role of the word “bite” (as
the verb of an embedded clause “dog bite victim” or as a modifier of
“victim,” itself modified by “dog;” in parallel, the syntactic role of “dog”

4 We assume a circa-1965 default Chomskian syntactic representation for simplicity.

182
 9.3 WAYS TO BRING ABOUT T WO MEANINGS IN A TEXT

NP VP
Pron

I
V NP PP

shot Art N Prep NP

an elephant in Det N

my pajamas

Figure 9.1 The first interpretation of the sentence “I shot an elephant in my pajamas.”

NP VP

Pron
V NP
I
shot

Art N PP

an elephant Prep NP

in Det N

my pajamas

Figure 9.2 The second interpretation of the sentence “I shot an elephant in my pajamas.”

S S

NP VP NP VP
N N
V NP
Squad V NP Comp
helps Squad
Mod N
helps N S
N N victim
dog NP VP
dog bite
N V NP

(dog) bite N

victim

Figure 9.3 The two interpretations of the sentence “Squad helps dog bite victim.”

183
 VERBAL HUMOR

is demoted from the subject of the subordinate clause to a modifier of

“bite.”
9.3.1.3 Other forms of ambiguity
The ambiguity between the literal and figurative senses of a metaphor5
is a very common way of triggering the incongruity (Goatly, 2012, pp.
180–181). A short example is
(89) How’s your plumbing business going? We’re flush. (Oring, 2003,
p. 120)
Whether this is considered a semantic or pragmatic ambiguity
depends on the theory of metaphor one subscribes to, so I will let it
rest, but will note in passing that not all taxonomies of puns would
classify this phenomenon as a pun.
Referential ambiguity may bring about the second sense, as in the
following example:
(90) John sleeps with his wife once a week. So does Paul.
in which the ambiguity is between the referential attribution of the
pronoun “his” in the ellipsed constituent “sleeps with his wife once a
week.” In other words, there are two possible interpretations: 1) John
sleeps with his [=John’s] wife; Paul sleeps with his [=Paul’s] wife; and 2)
John sleeps with his [=John’s] wife; Paul sleeps with his [=John’s] wife.
Note that this is no syntactic ambiguity as the structure of the sentence
remains the same for both interpretations.

5 One may question whether the ambiguity between the meanings of flush (as in skin
engorged with blood, hence metaphorically a rush or any fluid, hence water in a toilet,
and by further metaphorical extension, plentiful; here, having plenty of business) is indeed
metaphorical or the meanings have separated in unrelated lexical items, which would then
lead one to classify this example as homonymy. However, dictionaries list the senses as part
of the same meaning. For those readers still unconvinced, a different example might help in
which there can be no doubt of the metaphorical nature of the relationship between the two
senses, as one is expressed visually: in the movie Harold & Kumar Go to White Castle (2004)
there is a scene (available online at https://www.youtube.com/watch?v=D_ZlnWQbCeE)
in which Kumar notices a bag of marijuana and his attraction toward it is visualized as a
montage sequence in which the stereotypical phases of a love story between two people are
revisited as the interactions between the Kumar character and an anthropomorphic human-
sized bag of marijuana. This is an extended example of literalization of a metaphor “to love
something,” with humorous effects.

184
 9.3 WAYS TO BRING ABOUT T WO MEANINGS IN A TEXT

Finally, a generally neglected form of ambiguity is pragmatic

ambiguity.6 Consider again joke (24), from Chapter 3, repeated here
for convenience.
(91) One day, as I was withdrawing some money, the cashier at the
bank asked me: “Do you want me to take it out of your checking?”
I answered, “No, take it out of someone else’s checking.”
When the cashier asks me if I want to take the money out of my checking
account, the sentence is unambiguous: the ambiguity of the lexical
items has been resolved by the context. “take out” for example means
“withdraw” and not “purchase food to eat outside of an establishment”
or “accompany for a walk” (as in “take out the dog”); the syntax is
likewise unambiguous: it is the direct object (anaphorically referring
to the money), “take out” is the verb, which has undergone “particle
hopping”7 and collocates with “of something” (in our case, my checking
account). Here pragmatic enrichment tells us that the question is in
fact a dual option: checking or savings (this being a small credit union,
there were only two kinds of accounts; no fancy high-yield accounts,
or bonds, etc.). Thus the pragmatic force of the utterance is a yes/no
question, should the withdrawal come from my checking account or
from my savings. The two possible answers are thus: yes, take it out of
my checking account, or no, take it out of my savings account.
By forcing a reinterpretation of the theme/rheme organization of the
sentence (which will be reflected in the intonation of the utterance)
I force a new marked pragmatic interpretation in which the question
remains a yes/no dual question, but the focus shifts on the modifier
of account “mine” (“your” with direct speech referential shift), which
would turn the question in “do you want the money withdrawn from
your checking account or from someone else’s account?” By deliberately
choosing the second, less likely interpretation, I forced the marked
reading onto the text. Below the two intonations of the utterance are
reproduced (caps mark the prominent syllable):

6 “Pragmatic homonymy (ambiguity) is the phenomenon by which [speakers] use the

same linguistic devices to achieve different ends.” Tannen (1979, p. 31).
7 Cf. “Take out the money” vs. “*take out it” and “take the money out” vs. “take it out.”
Particle hopping is obligatory if the DO is a pronoun.

185
 VERBAL HUMOR

(92) Original: “Do you want me to take it out of your CHECKing?”

Reinterpretation: “Do you want me to take it out of YOUR
checking?”
Once again, it should be noted that the ambiguity is here completely
pragmatic: the lexical items, the syntax, the semantic referents of the
anaphors, etc. have been satisfactorily disambiguated, so that in context
there is no functional ambiguity left.8 The ambiguity that is exploited in
the text is a pragmatic one.
Let me be open about the fact that few people would label the above
example a pun, even though it meets the standards of the definition
given above, as it is based on two meanings co-existing in one string
(the entire sentence). Indeed, it is not a prototypical pun. I have included
this phenomenon here for completeness of the discussion on ambiguity-
generated humor.

9.3.2 Syntagmatic placement

As we have seen, a lot of verbal humor utilizes ambiguity. However, not
all verbal humor is based on the exploitation of a duality of meaning
contained in the text. For example, (93) below is not based on any
ambiguous string
(93) Mind your till and till your mind (Milner, 1972, p. 20)
since both the two instances of the word “mind” and “till” have distinct
disambiguated meanings and their syntagmatic copresence in the text is
what reminds the reader of the fact that the two pairs of strings are in
fact homophonic.
This example may not be convincing, because after all, one could
argue that while it is true that the first instance of “mind” is a verb,
meaning “pay attention” and the second one is the noun, both words
“mind” are ambiguous. Consider then the next example,
(94) Genius is 1% inspiration and 99% perspiration (Milner, 1972,
p. 17)

8 We can always find ambiguities if we want to: “you” could be referring to someone else
than me, etc. This is normal. The point is that for the purposes of the exchange (withdrawing
money), there were no ambiguities left.

186
 9.3 WAYS TO BRING ABOUT T WO MEANINGS IN A TEXT

where there is no ambiguity whatsoever: both inspiration and

perspiration mean their own separate unambiguous meanings. The
source of the punning resolution (see (9.5) below) comes from the co-
presence in the text of the string “spiration” repeated twice.
So, in conclusion, we can formulate the syntagmatic triggering of puns
as follows: the repetition of strings in the text can generate incongruity, if
different meanings are associated to the strings, but the strings resemble
each other sufficiently. Identity is of course the limit case of similarity.
Alliterative humor falls clearly under verbal humor under this defini-
tion. Alliterative puns consists of the repetition of sounds, or groups of
sounds, within the text. The following example (discussed in Attardo,
1994, p. 139)
(95) Today’s tabloid biography: High chair, high school, high stool,
high finance, high hat— hi, warden! (Meiers and Knapp, 1980,
p. 21)
is typical: the repetition of the dyphthong [ai] eight times9 is clearly
recognizable. Is it incongruous? Yes, because we have unconscious
expectations for how often sounds repeat in normal, unmarked contexts:
here the string [ai] repeats 8 times in 19 syllables, which is clearly out
of the ordinary, not to mention that in the last part of the text, the
placement of the [ai] syllables is very salient, and follows a pattern [ai] -
X(x), where X(x) represents a random syllable, possibly followed by an
unstressed syllable, repeated six times, which continues until the end of
the text, where the pattern ends, with a second incongruity between the
“normal” development of the life trajectory and the unexpected ending
in jail. It should be noted that the alliterative pun is independent from
the “life story unexpected ending” incongruity.
The following example is a pure alliterative pun, with a lower number
of repetitions:
(96) You remember Sunset Strip – where the unneat meet to bleat!
(Attardo et al., 1994, p. 35)
where the repetition of the [it] string in the last three prominent syllables
is still highly noticeable.

9 If the reader is counting only seven [ai] there is another [ai] in “biography.”

187
 VERBAL HUMOR

Finally, within syntagmatic puns, the presence of the two strings may
be explicit, as in all the examples above, or implicit (i.e., it has to be
reconstructed inferentially). When the activation is implicit the puns are
labeled “paradigmatic.” Many examples are obscene, playing precisely in
the opportunity to suggest an obscene meaning without having to state
is explicitly. The following is typical and is based on a chiastic reversal
(Attardo, 1994, pp. 116–117) of the [s] and [f] phonemes of “sucks” and
“fingers”:
(97) What’s the difference between a baby and the conductor of
a women’s choir? One sucks his fingers . . . http://allowe.com/
laughs/book/What%27s%20the%20Difference.htm
A non-obscene example is
(98) Diplomacy: The noble duty of lying for one’s country. (Milner,
1972, p. 17)
with the inferentially activated “dying for one’s country.”

9.3.3 Paronymy and phonetic distance

In the definition of verbal humor above, we emphasized that the two
strings “may, but need not, be identical.” It is time to unpack this
aspect of the definition more thoroughly. Hempelmann and Miller
(2017) standardize the terminology so that puns in which the two
strings are identical are called homonymic (or perfect); conversely,
if the two strings are different, the pun is said to be heterophonic
(or imperfect). Heterophonic puns correspond to paronymic puns, in
Attardo (1994).
So, one way to characterize puns is to take homonymic puns and
assume they are the limit case in the direction of similarity, i.e., they are
puns that differ in zero sounds. Heteronymic puns then can be arranged
on a scale of increasing difference. Let us consider some examples:
(99) Context: on a birthday card there is a picture of a beautiful woman
holding a birthday cake. The legend reads:
“You can’t have your cake and Edith [eat it] too.” (Attardo, 1994,
p. 149)

188
 9.3 WAYS TO BRING ABOUT T WO MEANINGS IN A TEXT

The pun, loosely transcribed as [ˈiDɪθ] vs. [ˈiDɪt] (the “D” symbol
indicates a flap in American English) is based on the two strings of
phonemes differing in one phoneme ([θ] vs. [t]).10
Hausmann (1974, p. 66) provides us with an example where the pun
differs in two phonemes:
(100) Vatican vs. vaticancan (with an allusion to the Parisian dance)
where the difference is the two phonemes “can” [kã] (the tilda indicates
the nasalization of the vowel in French). Obviously we can get three, four
and more differences. However, the problem of looking at the difference
between two strings is that several factors start to come into play: just
to use an example, is the difference between replacing a sound with
another in the same position, the same as deleting it (both differ by
one morpheme, but is the difference perceptually similar)? How about
replacing a sound in a different position of the string, since we know
that beginning and ends of words are more salient? To obviate these
problems, Hempelmann and others have turned to Optimality Theory,
with good results, but let us just say that they are very technical, as they
rely on the ranking of various constraints which are then responsible
for weighing each phonetic, morphological, intonational, etc. choice.
Overall the main conclusion that Hempelmann and Miller reach is that
the proper way to describe heteronomy is “psychoacoustic” i.e., sound-
based but considering also the position of the sound in the syllable and
word, and the context in which it is located.
Hemplemann and Miller (2017) list some “general consensus” on
these metrics:

the quality of vowels is more violable than that of consonants, with vowel substitutes
usually having similar height and backness. There is an overwhelming preference to
maintain the number of syllables, and in light of phonotactic constraints on syllable
structure, this makes vowels resistant to both insertion and deletion. Any change
in lexical stress is extremely rare. Contrasting segments are most likely to be found
at or near the pun’s extremities, with a slight preference for the beginning. Among
consonants, stops have the highest interchangeability, followed by stops and fricatives

10 The normally unstressed pronoun “it” would be stressed, in this context, due to its
prominence. Intervocalic [t] is voiced and becomes a flap, in American English. [ˈit ɪt]
becomes [ˈiDɪt].

189
 VERBAL HUMOR

sharing the same place of articulation. In both English and Japanese puns, voicing is
often substitutable in otherwise identical consonants; changing the place or manner of
articulation incurs a greater cost. (Hempelmann & Miller, 2017, p. 106)

The preference of puns for the “extremities” corresponds to the bath-tub

effect, noted in Attardo (1994, p. 123).
This is one of the most exciting and open-ended areas of research
in the linguistics of humor. For example, one important question that
remains unanswered is how far can two strings differ and still be
recognized as a pun? Attardo (1994, pp. 120–126) provides a history of
the research and some discussion. Hempelmann (2003) formalized the
discussion in terms of Optimality Theory and Hempelmann and Miller
(2017) update the discussion. Guidi (2012a/b), building on Hempel-
mann’s work, applies the metric to several families of languages and
arrives at a rough estimate of five phonemes. She shows that the results
hold for many languages outside of English and may be universal. The
claim of the universality of puns and of its mechanisms is not new, see for
example, the references in Henry (2003, p. 106), but all the prior claims
have been theoretical, i.e., based on the nature of language, its rules, etc.
Guidi’s claim is empirical and hence notable. Furthermore, Guidi notes
that the mechanisms used by puns are the same mechanisms used by
phonological processes: addition, deletion, substitution, and inversion
(Guidi, 2012a, pp. 90–91; Hempelmann, 2003).

9.3.4 Connectors and disjunctors

As we saw in Section 4.3.1, the string (usually it is a word, but as we saw
above, it does not have to be), that is either ambiguous or that is similar
to another string and by so being allows that co-presence of the two
senses in the text is called a connector. So in example (82), the connector
is the NP “a jar,” whereas in example (83), the connector is “comes.” The
connector may occur anywhere in the text of a joke, provided that the
speaker will remember it by the time the second sense comes about, so
when we are dealing with longer texts (short stories, for example), we
probably should have a more stringent definition. I know of no research
on this metric.
The disjunctor is the element of the text that forces to the surface
the presence of the second sense. This is meant fairly literally: when
a hearer processes the senses of a word those that are not found to fit

190
 9.4 THE FATE OF THE INTERPRETATION

in with the current interpretation of the text are suppressed, i.e., their
presence is not part of the consciousness of the speaker, even though
initially (> 200 ms) there may have been activation of all the senses of
a word. For strings this is even more clear, since random strings do not
activate lexical processing:11 for example, most readers will be somewhat
surprised to find out that “luminescence” contains the string “mine,” if
for no other reason than the pronunciation is different, due to stress
placement. In example, (85) “lert” is the disjunctor and “alert” is clearly
the connector.
Connector and disjunctor may occur separately, as in example (85)
above, or they may coincide, as in example (86). In the former case they
are called “distinct” and in the latter “non-distinct.” Distinct disjunctors
may occur before or after that connector. The position after the connec-
tor is easy enough to see, as exemplified in example (85) above. Bucaria
(2004) documented the existence of disjunctors that occur before the
connector.12 See Section 4.3.1, for some examples.

9.4 The fate of the interpretation

What happens to the senses of the pun, after they have been brought
together? Guiraud (1979) examined the question in depth and we will
follow his presentation. For the sake of this discussion, assume that the
hearer or reader of a joke constructs a first interpretation of the text,
whereby he/she attributes to the connector a given sense. We label this
sense S1 . In example (83) above, Meena, the swimmer, clearly attributes
to the word “comes” the sense of “arrive” (as in come first, second, third,
etc.). However, Dan, her punster friend, forces a second sense “have an
orgasm.” We label this second sense S2 . The incongruity is brought about
in the text when we (and Meena) become aware of the presence of both
S1 and S2 in the text.
In this example, as far as Dan is concerned, the sexual innuendo (S2 ) is
the sense that should prevail. In other cases both senses may coexist, or
S1 may be the dominant one. The question that this section addresses is,
what happens to the senses in the joke after the punch line/disjunctor?

11 There is neurological evidence for this: Jessen et al. (1999) show that the activation in
the brain is different for random strings than for words.
12 Correcting my own claim to the contrary (Attardo, 1994, p. 105).

191
 VERBAL HUMOR

The following example, attributed to French poet Paul Valéry

(Landheer, 1989, p. 37), is an example in which S1 remains dominant:
(101) Entre deux mots il faut toujours choisir le moindre. [between
two words/evils one should always choose the lesser]
The pun plays on the homophony of the French word [m] (“mots” =
words or “maux” = evils). When the disjunctor “moindre” is reached,
which is incompatible with the semantics of “words” which does not
afford a ready scalar meaning (in what sense are words less or more?),
whereas the reading as “evils” affords very easily a scalar reading (bad
things are easily classified on scale of badness: think about spilling
coffee on your pants vs. being run over by a car), the second meaning
(evils) is brought to consciousness. However, S1 persists and takes on the
connotations carried by S2 : words can be evil/bad.
In the following example on the contrary, both senses S1 and S2 are
compatible with the text which makes sense either way:
(102) Bassompierre was a prisoner at the Bastille. While reading, he
flipped the pages of his book hastily. The warden asked him what
he was looking for, and Bassompierre replied: “I am looking for
a passage, but I cannot find it.”
The word “passage” is ambiguous: a passage may be a spot in the text
or an exit (from the prison). Coincidentally, this works both in the
original French and in English. It is easy to see that both interpretations
of the text make perfect sense in isolation, but that when activated
jointly, we get the pun to trigger. Needless to say, in this case, either
sense could be labeled S1 or S2 , as we have no way of knowing, barring
experimental evidence on brain activation, which sense gets activated
first. The significance of the order of activation is mostly apparent in
the field of psycholinguistics in which the presence (or absence) of the
senses and the order in which they are accessed is significant. I was first
made aware of this by Rachel Giora.

9.5 Cratylism: resolution in puns

As seen in Chapter 4 the incongruity in humor, and hence in puns, is

often “resolved” in the technical sense already discussed. Furthermore,
within the General Theory of Verbal Humor, the resolution is handled

192
 9.5 CRAT YLISM: RESOLUTION IN PUNS

by the Logical Mechanism Knowledge Resource (Chapter 7). Puns

have a particular Logical Mechanism, which may be summed up by
the concept of a Cratylistic folk-theory of language, i.e., a theory of
language in which form and meaning are not arbitrary, but are motivated
and therefore, similar sounds convey similar meanings. Homonyms
(homophones and homographs) are the extreme case, i.e., identical
sounds or graphemes convey identical meanings.
Before proceeding to a quick review of the evidence for the Cratylistic
folk-theory, it is important to stress a few points, which have often
proven confusing.

1. The Cratylistic folk-theory is not scientific, it is factually wrong

and not supported by science. These are akin to superstitious
beliefs, such as jinxing the result of an event, casting the evil eye,
or the existence of ghosts.
2. The Cratylistic folk-theory was refuted roundly in Saussure’s
(1916) arbitrariness of the sign doctrine. It goes against the tenets
of most linguistic theory. Even those who espouse views in which
iconicity and sound symbolism play a part, do not advocate a
motivated (i.e., Cratylistic) view of the sign.
3. Nonetheless, speakers believe subconsciously these things and/or
act in accordance to them.
4. The “resolution” of the incongruity is not a real resolution that
eliminates the incongruity. In puns, this is particularly evident
as the Cratylistic resolution is mostly subconscious and can be
documented only indirectly.

The Cratylistic folk-theory of language is named after Cratylus, the

character in the Platonic dialogue named after him. Hermogenes, Craty-
lus’ opponent in the dialogue, summarizes Cratylus position right at the
beginning of the discussion with Socrates:

Hermogenes: Cratylus, whom you see here, Socrates, says that everything has a right
name of its own, which comes by nature, and that a name is not whatever people call a
thing by agreement, just a piece of their own voice applied to the thing, but that there
is a kind of inherent correctness in names, which is the same for all men.13

13 http://data.perseus.org/citations/urn:cts:greekLit:tlg0059.tlg005.perseus-eng1:383a

193
 VERBAL HUMOR

9.5.1 Evidence for the Cratylistic folk-theory

As mentioned above, most of the evidence for the Cratylistic beliefs
held by speakers is indirect. In Attardo (1994, pp. 149–170), I list a
long series of phenomena which provide evidence that the speakers hold
Cratylistic beliefs. Here, I will limit myself to paretymology, rhyme, and
sound-based lexical associations. These should suffice to establish that
speakers do indeed hold these beliefs. The interested reader may consult
directly the full discussion, which is not overly technical.
9.5.1.1 Paretymology
Historical linguistics has long known that in some cases the etymology
of words reveals sound/meaning associative patterns. When traveling
in North-West France, I always make sure to eat a choucroute garnie,
a delicious sauerkraut-based dish. The etymology is quite interest-
ing, as the German word “sauerkraut” was imported into French and
there reanalyzed according to similarities with French words as “chou”
(= cabbage) and “croûte” (= crust). Along the same lines, the word “ham-
burger” (German “Hamburg” (the name of the town) + [-er] a derivative
suffix), which signified a ground beef patty made in the fashion of
Hamburg (in Italian it is know as a Swiss steak), was reanalyzed as “ham”
+ “burger” (as evidenced by the existence of the words “cheese-burger,”
“veggie-burger,” etc.).
What makes these reanalyses interesting from our perspective is that
the reinterpretation of the word in the borrowing language is done on
the basis of the sounds forming the word and not of its semantics (there
is no ham in hamburgers, nor crust in choucroute) which shows that the
speakers of the borrowing langauge must be reasoning along the lines of
“if it sounds like the word “croute” then it must mean “croute” (it does
in fact mean cabbage) and if “sauer” sounds like “chou” it must mean
“chou” (cabbage; while it does actually mean “sour”).
9.5.1.2 Rhyme
It is a well-known fact that the strength of the rhyming effect is affected
by semantic factors. This is unexpected, since, in poetry, rhyme is
defined as the repetition of sounds at the end of two words, usually
in different lines of a poem. Nonetheless, words that have unrelated
meanings have a very strong rhyming effect, as in Italian “marmellata”

194
 9.5 CRAT YLISM: RESOLUTION IN PUNS

(= jam) and “martellata” (= hammer blow).14 In English similar exam-

ples would be “telephone” and “scone.” Conversely, words with sim-
ilar meanings have much weaker rhyming effect: Wessex and Essex
barely rhyme. Telephone and sousaphone likewise are not good rhymes,
because a word does not rhyme with itself (which is the extreme case
of semantic similarity) and the etymological origin of the words, both
derived from the Greek “phonè” (= voice), is apparent enough to nullify
the rhyme. These effects are only explained by an association of sound
and meaning in the mind of the speakers, who are pleasantly “surprised”
to find that similar sounds are associated with different meanings.
Another semantic effect of rhyming has been investigated as well in
aphorisms and advertising slogans. McGlone and Tofighbakhsh (1999)
postulate a “Keats’ heuristic”:15 “the aesthetic qualities of a message are
equated with its truth” (McGlone & Tofighbakhsh, 1999, p. 240), in the
context of aphorisms. The phenomenon is not limited to aphorisms:
“The rhyme-as-reason effect occurs not only in evaluation of existing
aphorisms, but applies also to perception and evaluation of advertising
slogans” (Filkukovà & Klempe, 2013). This asseverative effect (i.e.,
forcefully stating the truth) of rhyming and generally assonance can be
summed up as the belief that if it sounds good, it must be right, which
is also consonant with the Cratylistic views of speakers.
9.5.1.3 Sound-based lexical associations
We know, see for example Aitchison (1987); Bybee (1998); Bybee
and Moder (1983), that words are stored in the lexicon according to
semantic connections but also along phonological connections (sound
similarity). Meara (2009) states that “Rhyming responses, assonance,
responses with the same initial sounds as the stimulus, or a similar
prominent consonant cluster are a common types of clang associates”
(clang associates is a technical term meaning sound-based associations).
Fitzpatrick (2013) likewise summarizes the findings in the field stating
that “clang responses, [ . . . ] share phonological or orthographic features
with the cue, but are otherwise not related.” In fact, Saussure (1916)
already listed among the paradigmatic associations of words, sound-

14 I owe this example to Prof. Sergio Cigada, from the first course in phonetics and
phonology I ever took.
15 “Beauty is truth, truth beauty,” Keats, Ode to a Grecian Urn; 1819.

195
 VERBAL HUMOR

based ones. So, we can consider well-established that fact that words in
the mental lexicon have associations that are based on sound similarity.
This is probably the ultimate cause of the Cratylistic view of language
evidenced by speakers.
In conclusion, we have presented some of the evidence for the
Cratylistic view of language that speakers hold. If speakers hold this
view of language, subconsciously of course, then it is no surprise that
they would consider sound similarity or sound identity as justifications
for the overlap of two different meanings in a string or word.

9.6 The psycholinguistics of puns

Among the most promising areas of research in the linguistics of humor

today, I would count recent work in psycho- and neuro-linguistics which
has brought about a number of studies that look at the actual processing
of puns either in the brain, using FMRI or EEG scans, or using eye-
tracking devices which allow scholars to plot where the attention of the
reader is directed.
It would be too technical to review in detail the methodology of the
various studies, so I will limit myself to summarizing the conclusions of
some of the studies. Lopez and Vaid (2017) offer a broader discussion
about humor, but with many relevant data concerning puns. Chen
et al. (2017) provide an overview of he results in the neurolinguistics
of humor, many of which are relevant to the discussion of puns. First
and foremost, there is evidence that puns and referential humor are
processed differently at the neurological level (e.g., Goel & Dolan 2001;
Bekinschtein et al., 2011; Chan & Lavallee, 2015).
Vaid et al. (2003) show that in a joke the activation of the first meaning
continues until the punch line when both meanings are active. After
a longer interval (4.5 seconds) only the second meaning is still active.
It would be interesting to test if the differences identified by Guiraud
(1979) are reflected in the activation times, as studied by Vaid et al.,
especially in light of some of the results in Bekinschtein et al. (2011,
p. 9670) which suggest that there is a difference between incongruity
resolution and ambiguity resolution in non-humorous sentences, due
to the persistence of both senses in humor.

196
 9.7 CONCLUSIONS

Jared & Bainbridge (2017, p. 10) confirm the incongruity–resolution

approach to puns, using eye tracking:

readers initially notice that the presented homophone in a pun does not fit the adjacent
context, ( . . . ). This momentary lack of fit of the presented homophone and the
subsequent resolution of the incongruity may be needed to appreciate a pun.( . . . )
Supporting evidence from our study regarding the importance of incongruity for
humor was that puns that were later rated as funnier had longer gaze durations on the
homophones. Furthermore, puns that were later rated as funnier also had shorter total
fixation durations on the homophones, providing evidence that swift resolution is also
important for humor.

9.7 Conclusions

We have reviewed various types of classifications of puns and the par-

ticular role played by ambiguity in puns. We also reviewed a set of
issues that are specific to puns, such as the threshold between paronyms,
the persistence of the interpretations of the text, and the Cratylistic
resolution typical of puns. We concluded with a review of some of the
psycholinguistic evidence for these phenomena.
While puns are a special category of humor, they are still based on
a semantic opposition and a resolution. In this they are identical to
other forms of humor that do not rely on the form of signifier. However,
the way in which puns do rely on the signifier and the way in which
they partially resolve the incongruity are different and to some extent
unique, which justifies their separate treatment. There is some evidence
that puns are universal as well and that they exploit the same general
phonological mechanisms.
The treatment of puns concludes the second part of the book, which
covered humor competence. Humor competence is primarily semantic
and pragmatic. As we have seen, in order to account for some phe-
nomena, such as puns, or targets, we need to move to a slightly less
abstract level in which we do consider that linguistic expressions have
forms (signifiers) and that jokes are about things and that sometimes
the oppositions express value judgements (good vs. bad, for example)
and that these value judgements may be predicated of people (targets),
for example. In the next part, we will turn to humor performance, thus

197
 VERBAL HUMOR

moving even further toward the concrete side of our theoretical stance,
and we will consider audience and social interactions.

9.8 Further readings

The most recent general treatment of puns can be found in

Hempelmann and Miller (2017). Hausmann (1974) is a classic, in
German, but the examples are in French. Guiraud (1979) also in French
is another classic text. Pepicello and Green (1984) is also interesting.
My own treatment of puns in Chapters 3 and 4 of Attardo (1994) is still
worth reading, especially on the Cratylistic theory. The bibliography
is updated in Attardo (2018). Other essential resources are Sobkoviak
(1991), Lew (1996) and Hempelmann’s (2003) dissertations, Ritchie
(2004: ch. 9), and Guidi (2012a), which is in Italian but helpfully
summarized in (2012b) in English.

198
 10
The performance of humor

10.1 A little history never hurt anyone 202

10.2 The Hymes-Gumperz sociolinguistic model 211
10.3 Empirical studies on markers of humor performance 227
10.4 Further readings 234

In this chapter we begin to tackle how speakers actually produce humor,

primarily in spoken discourse, i.e., how they perform humor. Another
way to conceptualize the subject matter of this chapter is to think of
it as humor in discourse. In order to address this subject we will need
to introduce a slightly different perspective than the one we have been
working under in the first part of the book. The chapter is organized
as follows: the first section reviews previous attempts at providing a
theory of humor performance. The second section presents the Hymes-
Gumperz theory of ethnography of communication and organizes rele-
vant research on humor performance within that framework. Since the
Hymes-Gumperz framework is rather broad and, as a whole, has not
been widely adopted within humor studies,1 I will present its under-
pinnings in somewhat more detail than other approaches mentioned
elsewhere. The third and final section presents some recent empirical
studies in the performance of humor which combine discourse and
instrumental analyses.
As the reader will recall from our brief discussion in Section 2.1, one of
the foundational points established by Saussure (1916) is that it is crucial

1 There are plenty of studies that fit within the framework of ethnomethodology, for
example most discourse analysis, but very few studies encompass the entire ethnomethod-
ological framework.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 THE PERFORMANCE OF HUMOR

to distinguish two levels of analysis in the study of language: the langue,

language, in an abstract sense, and the parole, speech/discourse, in a
concrete sense, and including the uses of language in writing. Chomsky
renamed the distinction competence and performance, respectively,
with minor changes in the definition of the role of syntax, which are
not relevant in this context. The title of the chapter then should be
understood with “performance” in the Saussurian/Chomskian sense
(i.e., as opposed to competence) and not specifically in a theatrical
or cinematic sense, although we discuss stand-up comedy, which is a
form of performance in the theatrical sense, and delivering a joke is,
to some extent, a form of micro-performance (for example, the teller
will embody characters, do “voices,” or accents, etc.). Performance in
the acting sense is outside the realm of this study.
The reader will recall that Chapter 6 detailed Raskin’s Semantic-Script
Theory of Humor, a theory of humor competence. This chapter, and the
following three, on conversation and discourse analysis and sociolin-
guistics, deal with the performance side of humor. However, the nice
parallelism ends here. While the competence side of the parallelism is
fully theoretically developed, the performance of humor lacks a coherent
comprehensive theory. In fact, this chapter will attempt to be a first
presentation of such a theory.
One could then wonder if there is any difference between the per-
formance of humor and the conversation/discourse analysis of humor.
The performance of humor goes beyond the discourse analytic frame-
work and encompasses it, as we will see below, in Chapters 11–12.
In another sense, the difference between the conversation/discourse
analytical studies we review in Chapters 11–12 is one of focus. The
focus in Chapter 11 is on how humor fits in conversation and what the
functions of humor are in that context (Chapter 12).
In the present chapter we will also look both at how humor is realized
in conversation; however, we will do so from an empirical and instru-
mental perspective and using a more quantitative approach. We also
consider humor analyzed in experimental settings, unlike conversation
and discourse analysis.

10.1 A little history never hurt anyone

This section details a number of studies that are directly relevant to the
performance of humor or that propose a theory of performance without
202
 10.1 A LIT TLE HISTORY NEVER HURT ANYONE

however presenting a comprehensive theory. From a disciplinary

standpoint, we have contributions from psychology and sociology,
theatrical studies, rhetorics, stand-up comedy studies, conversation
analysis, the General Theory of Verbal Humor, and sociolinguistics.

10.1.1 Early studies on the social context of humor

Psychologists and social scientists have long been aware that reactions to
humor stimuli vary significantly in relation to the social context in which
they are experienced. Summing up the literature on the subject, Pollio
and Swanson (1995, p. 8) state: “reactions to humorous materials depend
at least as much on the social contexts in which they are encountered
as on the specific nature of the comic material itself.” Research has
shown that increased group size increases the appreciation of humor.
For example, Morrison found that in a theater audience, the larger the
audience the greater the number and duration of laughter episodes
(1940, p. 183). Prerost (1977) found that increased group size facilitated
humor appreciation in children and teenagers. Stokols et al. (1973)
found that subjects in a crowded situation laughed more. Aiello et
al. 1983 found that “high spatial density environment [= crowding]
produced a considerable enhancement in subjects’ enjoyment of the
humor” (p. 193). Thus it is clear that one of the first topics that a full-
blown theory of humor performance will need to address is the audience
of humor and its size, if nothing else.

10.1.2 Carrell’s performance theory

The first sustained proposal for a theory of humor performance, comes
from the perspective of rhetorics (Carrell, 1993, 1997a/b). The concept
of “humor community” (Carrell, 1997a) is inspired by the idea of
“discourse community.” A humor community is the audience of a given
show (say the sitcom Friends): whereas the audience that watched the
show when it first aired is more local, once a sitcom is syndicated (i.e.,
is re-broadcast on many local American TV stations and abroad) its
audience becomes global (i.e., made up of several different audiences;
think of the Spanish, German, Italian, etc. versions of Friends) and non-
homogeneous (since the various audiences may be interested in different
aspects of the sitcom). Carrell stresses that the concept of humor com-
munity with its global and non-homogeneous nature can be applied to
203
 THE PERFORMANCE OF HUMOR

many other humorous phenomena. Updating her examples, consider

humorous memes on the internet which spread very quickly across a
truly global community, which is however extremely ephemeral.2 Jokes
have always been known to “travel” far and wide, as they were easily
translated and adapted to other cultures/countries. The work of Christie
Davies is to a large extent dedicated to explicating and describing this
fact (e.g., Davies, 1990).
Carrell (1997b) distinguishes between humor competence (as in
Raskin’s Semantic Script Theory of Humor) and joke competence, the
latter being the speaker’s capacity to recognize a text as a joke, without
determining if the joke is funny (p. 174). Thus Carrell argues that if
one wants to determine if a given joke is perceived as humorous by a
given audience (i.e., performance), one has to take into account “the
situation” in which the joke is told.3 In Carrell (2008), she stresses, once
more, the importance of the situation, or context, of the joking (p. 315).
The slogan-like statement that “nothing is inherently humorous, or
funny” (p. 314) should be understood as “nothing is inherently funny in
a vacuum” i.e., all humor in actuality occurs in front of/for an audience.
Humor competence (in Raskin’s sense) is of course an abstraction away
from performance (i.e., humor-to-an-audience).

10.1.3 Stand-up performance

Other approaches to the study of audience have included the study of
the audience of stand-up comedy (Rutter, 1997, 2000, 2001). Rutter’s
dissertation argues for the necessity of a separate theory of performance,
that is concerned not with why the jokes are funny but to how they are
received as such in a given situation, by a given audience.
Rutter sees the interaction between the stand-up comedian and
his/her audience as pseudo-dyadic (i.e., it apes dyadic conversation in
that the audience is seen as one participant; 1997, p. 118). According to
Rutter, laughter does not occur randomly, for example, when individual

2 The life span of a meme is measured in months. Cf. https://www.digitaltrends.

com/social-media/the-lifespan-of-a-meme/ Zhao et al. (2013) show that tweets and text
messages have an even shorter life span.
3 This is a far cry from the accusation (Meany et al., 2014, p. 9) that Carrell’s “approach
disenfranchises the process of production and follows the post-structuralist / post-modern
logic of negating fixed meanings – nothing is inherently anything as all meaning is
negotiated.” Carrell’s theory should be understood as a theory of humor performance.

204
 10.1 A LIT TLE HISTORY NEVER HURT ANYONE

members of the audience get a given joke, but is regulated, much as

turn taking is regulated in conversation (p. 112). Rutter comes to some
interesting conclusions, such as that “the transition between performers
talk and audience laughter is negotiated so as to minimise overlap
and during changes in turn that laughter tends to take precedence
over talk” (p. 127) and so in this sense follows the “one-speaker-at-a-
time” principle of conversations. Furthermore, audience laughter has a
characteristic shape: it starts softly, builds up rapidly in volume, reaches
a plateau which is sustained for a variable period of time, typically
between 0.5 and 8 or 9 seconds, and then slowly tapers off (p. 129).
Rutter notes perceptively that

In order to minimise the potential for silence after laughter comedians will start their
talking before the laughter has ended but towards the end of the audience response.
The ability to do this is provided by their recognition of laughter growing quieter.
(p. 131)

thus showing how performers take advantage of the nature of the

pseudo-dyadic interaction to further their success on stage. Likewise,
performers may use various devices to sustain the laughter once it has
been initiated.
Rutter also claims that “The changes of pitch in [ . . . ] delivery [of
stand-up comedians] is [sic] used [. . .] to signpost the completion of
jokes and create an invitation to laugh” (1997, p. 232). According to
Rutter, there is “a contrast in tone between the principal stress in the
sentence that sets up the joke and a principal stress in the punch line” and
specifically a falling intonation followed by a rise (p. 232). The example
he gives is by British comedian Oliver Double, who is talking about his
mother being upset by his swearing:
(103) Double: An [sic] she [=Double’s mother] goes, “Oh, Oliver (.)
What have I done to deserve you?” I said, “Ya ´fucked ˆda:d.?”
(p. 233; I am simplifying the transcription and silently correcting
errors in the text; SA)
Needless to say, it is not just the audience that reacts to the performer.
DeCamp (2015) shows how comedians “tailor” their material to the
audience, deciding on the fly what material to include and/or exclude, or
including explanations and contextualizations, based on the racial make
up of the audience, for example. Scarpetta & Spagnolli (2009) show that

205
 THE PERFORMANCE OF HUMOR

comedians use “filler” lines (semantically empty material serving the

purpose to delay the next turn and to connect one joke to the next)
but also “surveys” in which they check the audience’s familiarity with
a given experience, with the function of increasing solidarity. Another
technique is expanding successful punch lines, in which the comedian
add on more humor to a successful punch line (and so he/she can deter-
mine on the fly when to do so). Conversely, comedians may develop
subroutines to handle jokes that “flop,” for example by pretending to
make a note to themselves of what material in the act is not working
so well. Wells & Bull (2007) and Tsang & Wong (2004) describe various
affiliative responses which include laughter and applause, but do not dif-
ferentiate among them. Atkinson (1984) also describes various ways of
drawing applause and other affiliative responses, including laughter and
suggests that here are affinities between the two, a suggestion that Rutter
(1997) endorses. The first to follow that suggestion were McIlvenny et al.
(1993). They find the familiar three-part list, with the punch line in the
third item, contrasting pairs, and disclaimers. more interestingly, they
find that comedians often overlap on the audience’s reaction, i.e., they
do not wait for the laughter to die down before resuming their act. Also
Bull & Wells (2002) and Bull (2006) argue that stand-up comedians use
techniques similar to those studied in Atkinson (1984) but add jokes
to the list of devices drawing applause. Jokes received both laughter and
applause (Bull, 2006, p. 574). Seizer (2011) discusses the use of obscenity
and profanity in stand-up.

10.1.4 Performance of canned jokes

One way to conceptualize the discourse analysis of humor is to see it,
at least in part, as the study of the performance of canned jokes. Sacks
(1974) foundational article consists of the analysis of the retelling of a
canned joke. Norrick (1993, 2010) has documented many aspects of
the telling of canned jokes in conversation. Norrick (2003) shows how
the GTVH can be applied to conversational humor, even in its original
formulation, i.e., as a theory of text and not of their performance (see
also Attardo 2001a). More significantly, Norrick (2003) calls also for a
description of the performance of humor in conversation and discusses
the issue of timing in the delivery of humor. Norrick (2003) is also

206
 10.1 A LIT TLE HISTORY NEVER HURT ANYONE

worthy of notice because it is one of the few articles that attempts a blend
between conversation analysis and humor research.

10.1.5 GTVH and performance

The GTVH is unabashedly a theory of competence (Attardo, 2001a,
p. 30, Attardo, 2008b). Both in Attardo (2008b) and Attardo (2017a),
I show how the GTVH may be able to contribute to a theory of the
performance of humor, while nonetheless maintaining that the two
are very different endeavors: “a theory of competence and a theory of
performance articulate mutually independent categories and therefore
cannot mix in principle” (Attardo, 2017a, p. 100). In Attardo (2017a),
I examine a) what a theory of performance might look like (it mostly
deals with the idea of repertoire, developed below), b) what (little) the
GTVH might contribute to it, and c) what I had been working on with
the Applied Linguistics lab group at Texas A&M University-Commerce
for the past decade (see below, Section 10.3).
It may thus come as somewhat of a surprise that there have been
three proposals to expand the GTVH to include (aspects of) a theory
of performance. We will briefly review them.

1. Canestrari (2010) suggests adding a “meta” knowledge resource

including “the signals that refer to the speaker’s intention of
being humorous and to the hearer’s recognition of such intention”
(p. 330).
2. Tsakona (2013) proposes the addition of a “Context” Knowledge
resource containing information about the sociocultural context
of the humorous text and including two different but interrelated
kinds of information: first, the sociocultural presuppositions for
the production and interpretation of SOs, LMs, and TAs; in simple
terms, what participants need to know about the sociocultural
context of the text to derive meaning from it; and, second, speakers’
metapragmatic stereotypes on humor: ideological assumptions
and stances on whether a specific text can be considered humorous
or not, why, how, when, and to whom.
3. Ruiz-Gurillo (2012, 2016), finally, proposes completing (or
“modifying” 2012, p. 40) the six knowledge resources with
a (meta)pragmatic component, which includes, for example,
information about register, genre, (meta)pragmatic markers, etc.

207
 THE PERFORMANCE OF HUMOR

Given the general reasonableness of the three proposals, why am I

skeptical toward an attempt to incorporate a theory of performance in
the GTVH? Consider, as an example, smiling intensity, i.e., the degree of
smiling, ranging from a faint smile to a broad, open-mouth smile. The
Applied Linguistics lab team has spent some significant time studying
the matter (the research is outlined below, in Section 10.3). There is
simply no way that I can see to get to the values of smiling intensity from
the knowledge resources of the GTVH. Smiling intensity is a category of
performance, which makes perfect sense, is measurable and correlates,
probably, with some internal states of the speakers. The knowledge
resources cannot predict or even be related to it. It would be like trying
to correlate the use of fertilizer and tilling of wheat fields to the baking
of good pizza. Surely, wheat is somehow related to pizza making, but
not in any directly significant way. Much more important variables can
be found that directly affect and effectively determine the perception of
goodness (temperature of the oven, quality of the tomatoes, freshness of
the mozzarella, quality of the olive oil, etc.), than the amount of fertilizer
used to grow the wheat. Of course, without wheat, there’s no flour and
without flour no pizza, so there is a correlation, but it is remote and not
very significant.
Likewise, the GTVH accounts for all the features of the jokes qua texts,
whereas a theory of performance must account for a host of issues that
are in principle not even related to the texts, such as political stance of
the interactants, social and economical determinants, gender and age
factors, etc. For example, a comedian may introduce a non-sexual joke
using profanity and “blue” language. This does not make the joke-as-text
sexual in any significant way, but may significantly affect its tellability
and acceptability by a middle-class audience. So, in conclusion, I believe
that since the categories that affect performance are different from those
that affect competence, one is better off creating a theory of performance
directly built around the categories we need for a theory of performance.
The next sections of this chapter attempt to outline such theory.

10.1.6 Sociolinguistic approaches

There has been plenty of research in various areas that fall under the
purview of humor performance (such as discourse analysis, or socio-
linguistics) but no serious attempts have been made to organize these
contributions in a coherent theory of performance. Exceptions to this

208
 10.1 A LIT TLE HISTORY NEVER HURT ANYONE

are Davies’ (2017) chapter for the Routledge Handbook of Language and
Humor that gathers and systematizes the work on the sociolinguistics of
humor, Gasquet-Cyrus (2004), a broad sociolinguistic analysis of humor
in Marseilles (France), and my own Attardo (2015), which tries to
organize the research on humor in conversation and discourse analysis.
The most recent work to be considered in this respect is Chukwumah
(2018), an uneven collection of essays on joke performance in Africa.
Another exception to the neglect of the theory of the audience is Smith
(2009), in which the author advocates creating a theory of the audience
to accommodate “unlaughter,” Billig’s (2005) term for the deliberate
refusal to display humor support (laughter, or metalinguistic comments
that are favorable to the display of humor). Billig (2005, pp. 192–193)
notes that here is a difference between not laughing, because, say, one
did not hear the joke, and actively signaling that one is not laughing
(“a display of not laughing” p. 192).
Smith’s (2009) argument is that unlaughter may be used to “higlight”
differences between the joke teller4 and the recipient and to “heighten
exclusionary social boundaries” (p. 151) Smith’s starting point is that the
switch from the serious bona-fide mode to the non-bona-fide humorous
mode is a “unilateral” move by the speaker and that it “demands”
a response accepting the switch. “[T]he joker’s unilateral switch into
humor cannot stand alone” (Smith, 2009, p. 152). Thus responding with
humor support (laughter, etc.) indicates an acceptance of the switch.
As an example of unlaughter, Smith relates that Tom Cruise reacted to
being squirted in the face with water by a fake reporter by pointing out,
accurately enough, that the man was being “a jerk” (p. 155).
In a social situation, humor establishes an in-group and an out-group
(see Attardo, 1994, pp. 322–325, for a synthesis of the research). Smith
refers to these categories as “solidarity” and “difference.” This dynamic
becomes self-perpetuating:

4 Here we need to keep in mind the point made in Section 1.1.1 that the terms
speaker/teller and hearer/recipient of the joke may be used in very broad senses. Consider
that in Smith’s example (the Muhammad jokes controversy) the “teller” is not just the
cartoonist, who drew the cartoon, but also the editors who selected the cartoon, the publisher
who printed it, and ultimately the society that is comfortable with the publication, whereas
the recipient is not just an individual reader of the Danish newspaper in which the cartoons
were originally published, but ultimately the entire (self-defined) group of people, who may
or may not have actually seen the cartoons themselves, who feel that they are entitled to
opining on the legitimacy and acceptability of producing the text. I have argued elsewhere
that the controversy was artfully provoked for political reasons, so I will not comment
further on the cultural differences that the episode purports to highlight.

209
 THE PERFORMANCE OF HUMOR

Shared laughter enhances solidarity, and accompanying unlaughter from those who are
outsiders or marginal only magnifies this effect. Accordingly, in their boundary-setting
work, well-integrated individuals sometimes do not wait for chance unlaughter but
deliberately provoke it. Marginal group members may become the repeated butts of
targeted jokes precisely because they are marginal. (p.161). Clearly, then, some jokes
are not meant to be funny to all. That is, some humorous performances are intended to
include some people in shared laughter but to exclude others. (Smith, 2009, p. 162)

The significance of this conclusion is very important, and often

neglected in humor studies: the intention of the speaker may not be
that of eliciting laughter/amusement in all the hearers, in the sense that
in a (practical) joke the ratified addressee (i.e., the target of the joke)
is not necessarily or not at all the one supposed to find the situation
funny. In aggressive, out-group building humor, the ratified addressee
is supposed to find the experience distressing, while the unratified or at
least unaddressed hearers are the ones supposed to find the (practical)
joke funny. Ultimately, the speaker may be his/her own addressee, with
the ostensible addressee merely playing the part of the target of the joke.
The following is an example of precisely this situation.
(104) Two teenagers, brother and sister. The sister is getting ready to
go out on a date. She changes clothes repeatedly, walking by her
brother who is reading on the couch, to get her mother’s opinion
on her outfit. The mother is in another room. Finally, the sister
leaves, only to return a minute later, go to her room and change
again. As she is leaving, the brother, who has otherwise been
entirely silent, says: “It was better before.” The sister slams the
door and leaves in a huff. The mother, who has overheard, comes
and scolds the brother. However, they are both laughing.
Here the brother is both the speaker and the (real) addressee, while
the sister is the target of the joke; the mother was an unintended
unaddressed overhearer.5
Going back to the Mohammad cartoons, Smith argues that their orig-
inal publication was a deliberate test of group-belonging, as was their
republication: “unlaughter was being provoked deliberately when news-
papers around the world chose to republish these cartoons” (p. 164).
Smith’s and Billig’s conclusion on unlaughter is that it is a form of failed

5 No sisters were harmed in the telling of the joke. The sister is only the ostensible
addressee, as there is no real intention to communicate.

210
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

humor, deliberately set up to fail, according to Smith, in the case of the

Muhammad cartoons. We deal more fully with failed humor in Section
12.4.3. This concludes our review of various disciplinary approaches
to the performance of humor. We now turn to outlining the Hymes-
Gumperz framework we wish to adopt for the rest of the chapter.

10.2 The Hymes-Gumperz sociolinguistic model

A theory of humor performance needs a theory of performance, period.

I believe that the broad framework of the Hymes-Gumperz socio-
interactional model of language use (a.k.a. ethnography of communi-
cation) is a tool sufficiently broad and general to provide the theoretical
support to do so. Incidentally, I would like to stress the continuity with
Saussure’s view of language I have espoused in this work (and elsewhere).
As we saw, competence is an abstract form (but not the most abstract
way of looking at language), while performance is the most concrete
form. Hymes (1964, p. 11) sees the distinction as one of more or less
emphasis (or primacy) on “the structure, or system, of speech (la parole),
rather than that of the linguistic code (la langue).”
A theory of performance is still an abstraction, to be sure, only
less so. The distinction competence/performance needs to be seen
as a continuum, with different descriptions arranged on the basis of
greater or lesser concreteness. The concrete token is an actual use by
an individual speaker in a certain circumstance. So for example, this
morning, my wife upon waking up inquired on the location of our cat
by saying “where’s the cat?” That actual instance of the token “the,”
uttered by Lucy Pickering at around 7:00 am, on May 15, 2018, in
our bedroom, in my presence, without any other participants present,
with all its attendant contextual factors, which include the tendency
of said cat to wander off and get stuck on trees on our property, or
the neighbor’s, and countless other potentially relevant factors, is a
concrete instance of performance. The definite article level, in English,
is the morphemic level /the/. The grammatical concept of “article” in
English is a more abstract concept. The class of Determiners is yet more
abstract and finally the N/N notation is categorial grammar and stands
for a category that applied to a Noun yields a Noun. It is an extremely
abstract representation of the concept of article. Intermediate levels of
abstraction are of course possible and it is possible that more abstract

211
 THE PERFORMANCE OF HUMOR

Table 10.1 The abstract–concrete continuum exemplified by the article “the”

Grammatical category N/N

Determiner Articles, demonstratives, possessives, quantifiers
Article a/an, the
Definite article the
Concrete token “the”

representations than categorical grammar exist or can be created.

I am not sure what a more concrete form of an actual speech instance
could be. The various levels of abstraction in the production of “the”
are summed up in Table 10.1.
So in this sense, when Bally (1909) distinguishes, after Saussure,
between langue and parole but introduces the langue expressive (expres-
sive discourse), i.e., the “possibilities offered” by language to the speak-
ers, he is considering intermediate levels of abstraction.6
Hymes’ communicative competence (1972) and his SPEAKING
model (see below) are a good place to start. Hymes (1964) stresses
“the need to investigate directly the use of language in contexts of
situation so as to discern patterns proper to speech activity” (pp. 2–3)
and that the ethnographic approach “must take as context a community,
investigating its communicative habits as a whole, so that any given use
of channel and code takes its place as but part of the resources upon
which the members of the community draw” (p. 3). Specifically, Hymes
draws attention on the ancillary role of linguistics per se within the
ethnographic approach:

it is not linguistics, but ethnography – not language, but communication – which must
provide the frame of reference within which the place of language in culture and society
is to be described. The boundaries of the community within which communication
is possible; the boundaries of the situations within which communication occurs;
the means and purposes and patterns of selection, their structure and hierarchy, that
constitute the communicative economy of a group, are conditioned, to be sure, by
properties of the linguistic codes within the group, but are not controlled by them.
(p. 3)
6 Interestingly the more or less open hostility toward Saussurean and Chomskian linguis-
tics one often finds in applied linguistics is not shared by its founders. Witness the approving
comments by both Gumperz and Hymes in their 1972 opus magnum: e.g. Gumperz (1972,
p. 5) or Hymes nuanced recrimination against formalism, acknowledging that “some who
are formalists show more concern for the implications their notions have for behavior
and the behavioral sciences than do some who are not” (1964, p. 1) or modeling of his
terminology of “communicative competence” on Chomsky’s (grammatical) competence.

212
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

Let us note in passing the stress on resources and on the communicative

economy of the group, as they will shape the concept of “repertoire” (see
below).

10.2.1 Linguistic repertoires

Gumperz (1964, p. 137) describes linguistic interaction as a process of
“decision making, in which speakers select from a range of possible
expressions.” The range of choices forms a “repertoire.” Gumperz defines
“repertoire” as “the totality of linguistic forms regularly employed in
the course of socially significant interaction.” Repertoires include “a
greater number of alternants, reflecting contextual and social differences
in speech” (1964, p. 137). More generally, repertoires can be defined
as “the totality of linguistic resources (i.e. including both invariant
forms and variables) available to members of particular communities”
(Gumperz, 1972, p. 20). Platt and Platt (1975, p. 35) provide another
similar definition: “the range of linguistic varieties which the speaker
has at his(/her) disposal.”
The idea of a linguistic repertoire shared by a community of speakers
is a significant step forward in the establishment of an ethnography of
communication, because it subsumes variation: “The concept of linguis-
tic repertoire is a fundamental departure from the traditional concep-
tion of one people–one homogeneous and unitary code” (Bauman &
Sherzer, 1975, p. 104). For example, Bauman and Sherzer mention
“Friedrich’s analysis of nineteenth century Russian pronominal usage
[1972]. The linguistic repertoire made available to speakers two forms
for the second-person singular, ty and vy, but actual usage covaried with
no fewer than ten socioexpressive factors” (Bauman & Sherzer, 1975,
p. 103).
Choice within the repertoire is not unconstrained:

Every message must conform to the grammatical restraints of the verbal repertoire but
it is always interpreted in accordance with social restraints. [ . . . ] This connection must
be statable in terms of regular rules allocating particular sets of forms to particular
kinds of interaction. [ . . . ] The power of selection is [ . . . ] limited by commonly agreed-
on conventions which serve to categorize speech forms as informal, technical, vulgar,
literary, humorous, etc. (Gumperz, 1964, p. 138; my emphasis, SA)

Hymes (1972) mentions in passing “indicating” humor “by switching

from one mode of speech to another” (p. 38). Gumperz’ and Hymes’

213
 THE PERFORMANCE OF HUMOR

casual references to humor as one of the options in the repertoire of

speakers validates, to some extent, our decision to investigate humor
performance within a Hymes-Gumperzian framework, as it shows that
the idea is quite compatible with it. However, neither Gumperz, nor
any of the interactionist work on humor attempted to define a humor
repertoire, so we will turn to that task next.
10.2.1.1 Humor repertoires
In Attardo (2002d), I presented, among considerable indifference, a first
outline of what a partial theory of humor performance could include.
Among various topics I touched upon the gamut of possibilities open
to a speaker in a specific situation: after the Speaker has uttered an
Utterance and the Hearer believes the Speaker had the intention for the
Utterace to be funny. The Hearer may:

1. not get the joke and laugh,

2. get the joke, but not react,
3. get the joke and smile,
4. get the joke and laugh,
5. get the joke and comment metalinguistically,
6. get the joke and change the subject,
7. not get the joke and ask for clarification,
8. etc.

We should emphasize that the list is not closed-ended: other classes

of behavior may appear as well, as indicated by the et cetera at the end;
moreover, this is a list of classes of behaviors, not of actual behaviors
(so this is an intermediate level of performance, not concrete commu-
nicative events). Laughter, smiling, changing the subject may take many
different forms and manifestations.
More importantly from a communicative interpersonal perspective,
while the external behavior may be the same, the mental states of the
participants may be significantly different: so for example, laughter may
occur upon getting or not getting a joke. Smiling may likewise signify
(somewhat muted) appreciation or be a cover for non-comprehension.
Changing the subject may indicate that one disapproves of the humor
or that one wanted to talk about something else. This is called the
“indeterminacy” of humor and its markers.

214
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

Significantly, the repertoire included the failures, both of Speaker and

Hearer, i.e., failure to produce a humorous turn when the intention was
there and failure to recognize a humorous turn. Failed humor was at
the time an unexplored domain. Since then, significant work has been
developed (see Section 12.4.3). In Attardo (2017b), I pointed out that the
repertoire of humor is not limited to reactions to humor, but of course
must encompass the decision to produce humor and so cover the social
criteria for appropriateness of humor, i.e., when and where it is socially
acceptable to be facetious.
In this light, we can consider Jefferson’s work on the invitation of
laughter (1979) as a contribution to the analysis of the repertoire of
speakers delivery of humor. Lavin & Maynard (2001, p. 465) describe
the repertoire for declining laughter invitations (i.e., not reciprocating
laughter), in the telephone interview setting: talking over it, typing, and
“attending to survey question” (i.e., reverting to on task talk).
Finally, there is another aspect of the humor repertoire that should be
included, i.e. the performance of humor includes also the choices in the
delivery of the humor, i.e., “all the linguistic and paralinguistic choices
made by the speakers as they produce the humorous utterance (ranging
from the pitch and volume with which the syllables are uttered, to the
font choice of the text, for example)” (Attardo, 2017a). I will address
some aspects of this seriously understudied aspect in Section 10.3 below.
Furthermore, the idea of repertoire figures prominently in Bell and
Pomeranz’ (2016, pp. 10–11) treatment of humor in the classroom; see
Chapter 16. This should give a sufficient idea of the complexity of the
coverage of a humor repertoire.

10.2.2 Speech acts and speech events

The reader will recall the discussion of speech acts and the work of
Austin (1962) and Searle (1969) in Section 8.1.1. Within the ethno-
graphic perspective, speech acts fit within speech events, which in turn
are part of speech situations (Hymes, 1972, p. 56). The difference is
primarily of size (“magnitude” p. 56):

a party (speech situation), a conversation during the party (speech event), a joke within
the conversation (speech act). [ . . . ] the same type of speech act may recur in different
types of speech event [ . . . ] Thus a joke (speech act) may be embedded in a private
conversation, a lecture, a formal introduction. (Hymes, 1972, p. 56)

215
 THE PERFORMANCE OF HUMOR

It is clear that here Hymes means “joke” in the sense of jab line, rather
than in the sense of “narrative ending in a punch line” because the latter
may not occur in a formal introduction. Telling a narrative joke is more
of a speech event.
Gumperz (1972, pp. 16–18) defines speech events as “communicative
routines” which have the following characteristics:

1. they are viewed by members of a society “as distinct wholes,”

2. they are considered “separate from other types of discourse,”
3. they are “characterized by special rules of speech and non-verbal
behavior,”
4. they often have “recognizable opening and closing sequences,”
5. they often have a “special name” (p. 17),

While Gumperz does not say so, it is fairly clear that joke telling meets all
the requirements above. Stand-up comedy also clearly fits the definition
of a speech event.
Speech events have a variety of components which Hymes catalogs
using the famous SPEAKING mnemonics: “settings, participants, ends,
act sequences, keys, instrumentalities, norms, genres” (p. 65).
10.2.2.1 Hymes’ SPEAKING components
In this section we will deal quickly with some of the components of the
SPEAKING model that are less relevant (but not irrelevant) when con-
sidering humor and the move on to the crucial concepts of humorous
genres, keying and framing, and contextualization cues.

Setting The setting includes the time and place in which the speech
act takes place, the “physical circumstances” of the event (1972,
p. 60) but also the “scene” i.e., the “psychological setting” of the
occasion, for example “formal or informal, serious to festive”
(p. 60). The setting for humorous events will be informal and non-
serious, generally speaking.
Participants Among the possible factors of this component, we
find the speaker and hearer we are familiar with, but also an
addressor, addressee (the originators of a message), spokespeople,
interpreters, and the audience (a non-addressed entity). Here
Goffman’s distinctions between ratified and unratified participants
and addressed and unaddressed listeners (think of a teacher talking

216
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

to one student in front of the rest of the class about the subject of
the lecture) would become relevant. (Goffman, 1981, pp. 131–133;
see Chapter 1.
Ends The ends include both the goals and purposes of the act and
its outcomes. In the case of humor, most unplanned improvised
humor will fall under the speech act category, rather than speech
events or speech situations: while “stand-up comedy” or “comedic
performance” have recognized status as socially sanctioned entities,
a jab line speech act in the middle of an otherwise serious occasion
does not have a recognized status, except perhaps as disruption.
Being facetious during a dinner conversation isn’t a distinct activity
like, say, dinner table conversation is. Conversational humor is
expected in some situations, for example, talk among friends, but
it piggybacks, so to speak, on an otherwise serious (or at least non-
humorous) activity. Conversely, stand-up comedy, with its atten-
dant “rituals” (see Rutter, 1997) or playing “the dozens” are exactly
the sort of speech events envisioned by Gumperz, the purpose of
stand-up comedy being to provide entertainment and diversion for
its attendees, and revenue for its performer. The outcomes will be
influenced by the quality of the show, the nature of the crowd, the
presence of hecklers, etc.
Act sequence The act sequence regards the message form and con-
tent. The form is how the speech act is realized. Gumperz (1972,
p. 59) emphasizes that the form of the message is a crucial compo-
nent. With the current developments of audio and video recording
and such sophisticated techniques as spectrographic sound analy-
sis, eye tracking, gesture analysis, etc. we can account with striking
accuracy with the details of how speakers deliver their utterances.
The content of the message is what is being talked about, the topic
of the interaction.
Key The key is the “tone, manner, or spirit in which an act is done”
(p. 62). Hymes lists “mock: serious” as an example of key. We deal
separately with keying below.
Instrumentalities These include the channel of the speech event:
spoken language, but also writing, signaling with other codes, and
we may add today, computer-mediated communication, and the
likes. The second component are the forms of speech, such as

217
 THE PERFORMANCE OF HUMOR

dialects, varieties, registers, etc. On dialects and humor, see Section

13.2.4; on register, see Section 14.4.1.
Norms of interaction and interpretation These norms concern
when and how to interrupt, allocating speech turns, what can be
talked about and in what situations, what is considered rude, polite,
permissible, aggressive, etc.
Genres Genres are textual types. Humor has numerous genres, which
will be dealt separately below.

10.2.2.2 Keying or framing

The terms “keying” (Hymes, 1972) and “framing” (after Goffman, 1974)
are used more or less interchangeably, in their original sense.7 The
fundamental idea goes back to the groundbreaking work of Gregory
Bateson (1972). As we saw in Sections 3.3.1 and 2.4, the concept has
close ties with humor from the beginning, in Bateson’s theorizing on
the metamessage “this is play.”
As we saw above, Key accounts for the “tone” of the speech event. Even
more explicitly, Hymes comments that when a key

is in conflict with the overt content of an act, it often overrides the latter (as in sarcasm).
The signaling of key may be nonverbal, as with a wink, gesture, posture, style of dress,
musical accompaniment [ . . . ] (p. 62) [my emphasis, SA]

For an example of sarcastic keying of an otherwise non-humorous situa-

tion (mob conflict) through musical accompaniment, see Chapter 15. A
possible keying device for stand-up comedy may be the use of profanity
(Seizer, 2011), which sets squarely the discourse in the informal domain.
A discussion of keying in relation to humor is found in Section 2.4 and
see also below the discussion of contextualization cues.
Bauman and Sherzer offer a pellucid definition of framing in a Bateso-
nian sense, which is worth considering in its entirety despite its length:

The notion of frame is drawn chiefly from Bateson, though the component of a
speech event that Hymes labels “key” is related to frames. A frame is a metacommu-
nicative device which signals the interpretive context within which a message is to
be understood, a set of interpretive guidelines for discriminating between orders of
message (Bateson [1972], pp. 177–93, 222). Examples of frames might be joking, in

7 See Goffman himself uses “keying” in a sense similar to Hymes (1974, p. 44). Tannen
(1993, p. 6) notes that the reference of the term “framing” has been expanded significantly,
under the pressure of the work on frame semantics and artificial intelligence.

218
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

which the words spoken are to be interpreted as not seriously meaning what they might
otherwise mean, imitation, in which the manner and/or matter of speaking is to be
interpreted as being modeled after that of another person or persons, and translation,
in which the words spoken are to be interpreted as the equivalent of words originally
spoken in another language or code. Framing is accomplished through the employment
of culturally conventionalized metacommunication, i.e. each speech community will
make use of a structured set of distinctive communicative means from among its
resources in culturally conventionalized and culture-specific ways to signal or key [to
use Goffman’s term (50)] the range of frames available to members.
(Bauman & Sherzer, 1975, p. 106)[my emphasis, SA]

Let us note several significant aspects of this discussion: first Bateson’s

“framing” is equivalent and an inspiration to both Hymes’ and Goffman’s
“keying.” Second, recall the discussion of Bateson’s metamessage “this
is play” as a fundamental component of humor (see Sections 2.4.1 and
3.3.1). Third, the other frames used as examples are “imitation” and
“translation.” We could add “performance of a play” and “practicing
grammar rules” as other examples. What all these situations have in
common is that the normal conventions of goal-oriented communica-
tion are suspended and overridden by a different set of goals. Fourth,
the explicit acknowledgment of the metalinguistic nature of the framing
process. It is necessary to have recourse to a meta-level (i.e., communi-
cate about the communicative interaction) in order to settle the framing
process.
What is generally not discussed, at least not in the humor stud-
ies literature, is that Goffman’s own discussion of framing is directly
connected to humor: when discussing “primary frameworks”, i.e., the
basic categories that are used to make sense of reality, such as physical
events (say, the weather) and social events (events that involve agency)
(Goffman, 1974, p. 22) Goffman mentions Bergson’s (1901) famous
definition of humor as the mechanical overlaid on the living. Goffman
comments:

Bergson only fails to go on and draw the implied conclusion, namely that if individuals
are ready to laugh during occurrences of ineffectively guided behavior [= mechanical],
then all along they apparently must have been fully assessing the conformance of the
normally behaved, finding it to be no laughing matter. (1974, p. 39)

Goffman concludes “observers actively project their frames of reference

into the world” (p. 39). Immediately after this discussion, Goffman turns
to “keying” and starts the discussion precisely with Bateson’s essay on

219
 THE PERFORMANCE OF HUMOR

“The message ‘this is play” ’ (originally published in 1955), but also

mentioning Groos (1898), which is a significant source of Freud’s (1905)
classic. Furthermore, Goffman quotes Fry (1968), one of the founding
fathers of humor studies. So, it is fair to say that the whole idea of keying
is closely related to humor. Goffman reviews Bateson’s definition of play,
and delves fairly extensively into the research differentiating play from
the real action. Without going too deep into the discussion, I will focus
on two features: 1) play is performed so that “its ordinary function is
not realized” (p. 41), so that play biting cannot involve real (i.e., hurtful)
biting; the “real” activity is the “pattern” on which the play form is
built (so, play fighting must resemble fighting, but not be fully fighting);
2) “signs presumably are available to mark the beginning and termina-
tion of playfulness” (p. 43; see also p. 45).
Goffman then introduces the term “key” as follows: “by keeping in
mind these comments on animal play” (p. 43) one can define it as

the set of conventions by which a given activity, one already meaningful in terms of
some primary framework, is transformed into something patterned on this activity but
seen by the participants to be something quite else. (p. 44)

If the connections to humor studies were not strong enough already,

Goffman here quotes Austin’s famous passage in which Austin defines
speech acts performed on stage as “parasitic’ or “etiolated” uses of
language (1962, p. 22). Humor is one such form of “parasitic” use of
language. The term “parasitic,” probably because of its negative con-
notations, never caught on. The term non-bona-fide (Raskin, 1985;
see Section 6.3) did. It is surprising then that Goffman never really
discusses humor keying, although he does mention, in passing “not
taking something seriously or making a joke of it” as “cast[ing it][ . . . ]
into a playful key” (p. 82).
Goffman (1974, 87–95) returns to jokes in the category of “fabrica-
tions” (i.e., framing that involves deceit). Goffman lists different kinds
of deceits, starting with “kidding” which lasts only the duration of a
sentence or “a turn at talking” (p. 88),8 leg-pulling, practical jokes, and
hoaxes, which may last for extended periods of time. Note, in passing,
that Goffman is very clear on the non-cooperative nature of these deceits
(for their duration). Goffman then deals briefly with teasing, under the

8 I was unaware of this detail of Goffman’s work when I formulated the Least Disruption
Principle; see Section 8.2.2. I now note with pleasure this convergence.

220
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

rubric of “negative experiences” and with wordplay (puns; pp. 441–443),

in the “vulnerabilities of experience” chapter of his book. Puns are seen
as shifts between frames. Goffman then mentions “frame disputes” and
quotes Emerson (1969) on debates about whether joking is appropriate
(pp. 498–499). If we also consider the numerous references to comedy
in plays, film, and television routines, I think that it is more than fair to
say that a serious consideration of Goffman as a humor theorist is likely
to be very rewarding.
Finally, let us wrap up our discussion of framing with the mention
of Norrick’s work (2010, pp. 234–235) which traces the history of the
idea of framing humor within the context of discourse analysis. Other
scholars to explicitly mention the idea of framing or keying include
Straehle (1993); Norrick (1993); and Kotthoff (1999). Canestrari and
Attardo (2008) and Canestrari (2010) develop a Batesonian account
of framing metacommunicative signals that the interlocutors engaged
in the “language game” of the humorous exchange may or may not
receive, accept, and/or reciprocate. If both speakers are on the same
“wavelength” and they reciprocate the metacommunicative cues, they
achieve humorous syntony. If the metacommunicative signals are not
received, the humor fails, at some level (on failed humor, see Section
12.4.3).

10.2.3 Genres
The concept of genre is clearly inspired by literary theory, where genres
such as the sonnet or the Elizabethan comedy have specific historical
and literary meanings. As long as we keep in mind that genres in Hymes’
sense are not limited to literary genres but transcend them completely
and range across all linguistic usages, that characterization is not prob-
lematic. A more technical definition is provided by Bauman & Sherzer
(1975, p. 105): “culturally conventionalized utterance types which can
be employed in the construction of discourse. Genres are verbal forms
organized at a level beyond the grammar.”
There is no question that jokes are a genre9 of humor as are anec-
dotes, riddles, limericks, doggerel, and now in the age of the internet,

9 I have used elsewhere the term “text type” as synonymous to genre to refer to jokes
(Attardo & Chabanne, 1992), based on Fillmore (1981) and DeBeaugrande & Dressler
(1981). This should not be confused with Biber’s (1988) definition of text type as a set of
co-occuring linguistic features, as opposed to genre, which is defined externally.

221
 THE PERFORMANCE OF HUMOR

memes. Kotthoff (2007) is a (non-exhaustive) review of conversational

humor genres and includes: teasing, humorous stories about problems
(different from the trouble talk described by Jefferson, insofar as these
are actually funny stories; for example, one student relates having had to
withdraw from the university after her 39th semester), joint fantasizing
(joke co-construction), parody, humorous gossip. Another review of
humor genres is Dynel (2009a). The most recent and comprehensive
treatment of genres of humor is Tsakona (2017).
The performance of jokes, i.e., joke telling sessions, has been investi-
gated by Sacks (1974); Norrick (1993) and many others. Norrick (1993,
pp. 112–118) shows that speakers may engage in joke telling sessions,
in which the purpose of the conversation becomes that of telling and
hearing jokes. In those settings, i.e., when the context has been set up for
the production of humor, checking whether the joke may be delivered
is no longer necessary. Much can be said about joke telling, for example
how speakers catenate jokes based on topical relevance and similarity
(Norrick, 1993, p. 117) and how the laughter arising from such sessions
can start feeding on itself (see Section 12.4.4). However, in what follows
I will consider a few humorous genres that are less well known, just
to attest to the wide variety of humorous genres. A secondary goal of
the choice of examples below is to show that not all joking is affiliative
or carries positive affect (Gruner, 1978; Attardo, 1994; Kotthoff, 1996;
Boxer & Cortès-Conde, 1997; Priego-Valverde, 2003; Billig, 2005).
10.2.3.1 Sounding, a.k.a. the dozens
Labov (1972) was not the first to describe the practice of young African-
American speakers to trade ritual jocular insults, but he certainly put
the practice front and center of the sociolinguistic field of study by
situating it within the context of his ongoing study of African-American
Vernacular English and also explicitly within Hymes’ ethnography of
speaking (p. 305).
Below I present a selection of sounds, or dozens, targeting mothers
(a favorite topic), but other targets are members of the group and other
relatives.
(105) Your mother’s a duck. (p. 304)
Your mother so old she got spider webs under her arms (p. 312)
Your mother eat rice crispies without any milk (p. 315)

222
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

According to Labov, the sounds are “immediately evaluated” (p. 325)

and “the primary mark of a positive evaluation is laughter” (p. 325).
There is a competitive aspect to sounding: Labov states that some
members of the group who are known to be good sounders will stop
other members from initiating sounding against them, since the lesser
sounders “know in advance that they will be outdone” (p. 326).
The speech act of sounding is described as consisting of a sequence
of pair-wise connected sounds. The first speaker starts, usually with
an insulting reference of speaker B’s mother. The audience evaluates
the sound, for its humorousness but also for its cleverness. Speaker B
retaliates with an insult to speaker A, usually a variation on the theme
introduced by A. Again the audience evaluates the sound for originality
and quality of delivery. If B’s performance was better than A, B has “won”
the sound. A goes again. Here he/she can choose to introduce a new
theme or create another variation of the original sound. The latter option
is considered more virtuosic. B then responds and the process continues
until one of the two is perceived to be the winner or the exchanges
degenerates in actual, no longer ritual, insults. This can be seen when
one of the participants reacts to
(106) Your momma drink pee.
with
(107) That’s a lie!
rather than with the expected ritualized
(108) Your father eat shit.
thus showing that they are taking the insult seriously (p. 335). As
Labov notes, sounds are never denied (much like in improv comedy, we
may add), they are reciprocated, whereas personal insults are denied,
or mitigated. In other words, sounds are ritualistic insults in that the
speakers must believe that they are not true/meant (p. 339), or to put it
differently, they are in play mode (non-bona-fide).
There is much more to sounding that I have summarized above, as
Labov’s own analysis, which is much more extended, shows. Further
research on sounding has questioned some of Labov’s points, such as for
instance that the sounds are necessarily false (Kochman, 1983), when
they only need not be meant seriously (in other words, one can joke

223
 THE PERFORMANCE OF HUMOR

about true facts as well). Rahman (2014) shows that ritual insults are
widespread, including the Anglo-Saxon world, and notes antecedents in
many African cultures. For example, Kihara (2015) provides an analysis
of the mchongoano a type of verbal dueling among the Luhya (a Bantu
people primarily located in Kenya).10
This is a good transition then to the next genre of humor we will exam-
ine. Culpeper (1996) notes that the practice of sounding is essentially
an elaborate form of banter and argues that this was quite widespread
in ancient Scandinavia. Let us note also the affinity, already remarked
upon by Gumperz and Hymes (1972, p. 132–133), of sounding with the
verbal dueling among Turkish adolescents discussed in Dundes et al.
(1970/1972) and with the verbal dueling of poets in pre-islamic Arabia
(e.g., Sowayan, 1989). Dollard (1939) and Abrahams (1962) are two of
the classical treatments of the dozens/sounding. They are conveniently
reprinted, along with an extract from Mitchell-Kernan (1971; see also
Mitchell-Kernan, 1972) and numerous other essays, all with valuable
introductions by Alan Dundes, in Dundes (1973; 3rd ed. 1990).
10.2.3.2 Jocular mockery
Haugh considers “jocular mockery” (2010, 2014) as a type of teasing
(Haugh, 2017b). Essentially, teasing may be framed or keyed either as
playful (affiliative, non-serious) or as aggressive (disaffiliative, serious).
Needless to say, the framing may be ambiguous, so that “it is not entirely
clear whether the teaser really is being non-serious, or is in fact covertly
serious” (Haugh, 2017b, p. 207). Jocular mockery, according to Haugh,
is “a particular subset of teasing where participants are orienting to
fostering solidarity, rapport or affiliation. Such teasing is invariably
carried out within a marked non-serious or joking frame” (Haugh, 2010,
p. 2107). A more intense form of banter is jocular abuse, which involves
mock insults (Hay, 1994, 2002; Haugh & Bousfield, 2012).
Haugh does not refer to it as such, but he describes the repertoire of
responses to mockery:

1. Reject the mockery (response to the criticism)

2. Accept the mockery or pretend to accept it (response to the playful
frame)
3. Ignoring it
10Kihara (2015) is also significant because he explicitly bases his analysis on Hymes’
SPEAKING model.

224
 10.2 THE HYMES-GUMPERZ SOCIOLINGUISTIC MODEL

Laughter may occur both with acceptance or rejection, and in both cases
marks the jocular (non-serious) reception. If the mockery is accepted,
the recipient may elaborate on the mockery (self-disparaging humor)
or counter it (i.e., respond in type, mode adopt) “the latter giving rise to
what is commonly termed ‘banter” ’ (Haugh, 2010, p. 2108).
There are cultural restrictions to banter. For example, Haugh and
Bousfield (2012) find that some topics of mock abuse appear only
in Australian data (e.g., lacking toughness), while others appear only
in British data (e.g., embarrassing behavior in childhood). Haugh and
Bousfield (2012) also examine the politeness status of jocular mockery
and abuse. On the politeness aspect of jocular mockery, see also Section
12.1.4.2.
10.2.3.3 The Igbos’ njakiri
The “njakiri” is a genre of joking among the Igbo, in South Central
Nigeria. Duruaku (2015) defines the “njakiri” as “a public abuse to make
fun of someone else” and lists is as one of the categories of “oral art”
in the Igbo culture. It is distinct from jocular abuse, because it is more
aggressive, while still being playful. Nwachukwu-Agbada (2014) defines
it as a “joking relationship” similar to teasing” and as “satirical and
jocular exchange meant to amuse its auditors, and probably to tease or
abuse its victim(s)” (Nwachukwu-Agbada, 2006, p. 153). The critical
component of the njakiri is kept under control by a set of specific criteria
for choosing the target: essentially, one addresses the njakiri only to
one’s peers, either in terms of socio-cultural status, or education level,
or gender (Nwachukwu-Agbada, 2014, p. 379).
Thus, Ebeogu (1991), whose work on njakiri remains the best avail-
able, presents a lengthy example, in which the members of a patrilineal
group (hence on the same social footing) trade grotesque accusations
(for example, of having gone to one’s in-laws to have a newly purchased
car blessed, rather than to one’s own patrilinear group; pp. 30–31).
Ebeougu argues that njakiri consists of “sarcasm” and “curse.” Both are
pretty harsh: Ebeogu notes “the satire is [ . . . ] hard-hitting” (p. 33).
Much like in sounding and jocular abuse “participants in a njakiri are
not expected to take offense” (p. 36).
Sounding, jocular banter, and njakiri fall under the category of verbal
dueling (Rahman, 2014), which is of course a type of language play
(Sherzer, 2002). The specific differences between the three seem to be
tied to the level of acceptability of the playful aggression displayed.

225
 THE PERFORMANCE OF HUMOR

10.2.4 Contextualization cues

Gumperz’ definition of contextualization cue is as follows:

any verbal sign which, when processed in co-occurrence with symbolic grammatical
and lexical signs, serves to construct the contextual ground for situated interpretation
and thereby affects how constituent messages are understood.
(Gumperz, 2015, p. 315)

Examples include formulaic expressions, code switching, pronuncia-

tion, prosody, rhythm, tempo, and other suprasegmentals. Gumperz
(1992, p. 231) lists the following categories:

1. Prosody
2. Paralinguistic signs (tempo, pauses, synchrony, latching, overlaps,
tone of voice)
3. Code choice (from the repertoire)
4. Choice of lexical form or formulaic expressions

Features of contextualization cues are:

1. contextual; they function “in co-occurrence with symbolic gram-

matical and lexical signs,” hence
2. non-symbolic (neither grammatical nor lexical),
3. non-semantic; they “signal meaning largely by cueing indirect
inferences” (p. 316),
4. intrinsically oral, “present in talk” (p. 316),
5. specific to a culture or subculture

Note how Gumperz goes out of the way to state that contextual-
ization cues “signal meaning” and do not directly “mean.” Levinson
(2003, p. 27) explains that

the term ‘cue’ denotes an encoded or conventional reminder, like a knot in a handker-
chief, where the content of the memo is inferentially determined. Thus the ‘cue’ cannot
be said to encode or directly invoke the interpretative background, it’s simply a nudge
to the inferential process.

Gumperz (1977, p. 198) examines a conversational joke in which a

man on an airplane is pretending to be a bus conductor asking for:
“Tickets, please!” “spoken in higher than normal pitch, more than usual

226
 10.3 EMPIRICAL STUDIES ON MARKERS OF HUMOR PERFORMANCE

loudness, and staccato rhythm” (p. 198) and one of the two women
who are traveling with him says to the other “I TOLD you to leave
him at home” [caps indicate phrasal stress]. Gumperz notes that “if the
statement of the man or the woman had been uttered in normal pitch
and conversational intonation, the connection between them might not
have been clear” (p. 198). In this case, the formulaic expressions and the
prosodic features functioned as contextualization cues.
Besides the examples used by Gumperz and Hymes themselves seen
above, one of the earliest explicit discussion of the uses of contextualiza-
tion cues in the context of humor is Deborah Tannen’s dissertation.11

Gumperz (1977) demonstrates that speakers signal what activity they are engaging in,
i.e. the metacommunicative frame they are operating within, by use of paralinguistic
and prosodic features of speech – i.e. intonation, pitch, amplitude, rhythm, and so
on. Gumperz calls these features, when they are used in signaling interpretive frames,
“contextualization cues.” (Tannen 1979, p. 37)

Siegel (1995) reports on code-switching (Fijian-Hindi) as the cause

of humor, in Fiji. Code-switching is a contextualization device, which
may indicate that “a switch to the code appropriate for humor can be
a signal that the content is not serious” (Siegel, 1995, p. 101). Many
linguistic communities in which two varieties coexist “conventionalize”
one of them for humor (Apte, 1985, p. 190) and so the mere switch to
the humorous variety may function as a contextualization cue. For a dis-
cussion of the use of dialectal varieties as humorous, see Section 13.2.4.
Kotthoff (2000, p. 64) states that “Laughter is the contextualization cue
for humor par excellence.” In Kotthoff (2006, p. 7) she lists a number
of contextualization cues for humor, which include laughter, prosody,
repetition, and much more, but no specifics are provided. Everts (2003)
lists among the contextualization cues, word order, intonation, speech
rate, nasalization, and facial expressions. Norrick (2004) argues that
contextualization cues may help orient a story telling toward humor,
rather than the newsworthiness of the story itself.

10.3 Empirical studies on markers of humor performance

There is a large literature on the markers of irony and sarcasm, which I

reviewed in Attardo (2000b). A more recent and comprehensive survey
11 Tannen’s dissertation committee consisted of Robin Lakoff (chair), John Gumperz, and
Wallace Chafe.
227
 THE PERFORMANCE OF HUMOR

is found in Burgers and van Mulken (2017). Burgers and van Mulken
distinguish three categories of markers:

1. utterance markers: metalinguistic, typographical, morpho-

syntactic, and schematic (repetition, echo, register shifts)
2. co-textual markers, such as other humor,12 rhetorical figures,13
intonation, prosody, facial expressions, etc.
3. contextual markers, such as the background information about the
situation contrasting with the utterance. (p. 390)

Detailed charts are provided that list all the markers proposed in the
literature, for the first two categories (the third one of course cannot be
listed).
In particular, I will highlight one aspect of the research on prosodic
and intonational markers, i.e., its relational (contrasting) nature. Sum-
ming up the research on prosodic and intonational marking of irony,
Burgers and van Mulken (2017) note that the idea that there is a specific
ironical tone of voice has come under mounting criticism. First, different
features are listed across different languages, so that, for example, in
English irony would be marked by a lowering of mean F0 (the funda-
mental frequency of the vowel), whereas in Chinese it would be marked
by a raising of the F0. Second, in experiments using masked speech (i.e.,
speech filtered so that the actual sounds are not recognizable but the
prosodic features are) speakers can recognize irony in their language
but not in others. So, if there is an ironical tone of voice it is language-
specific and not universal. Several studies (Attardo et al., 2003; Bryant &
Fox Tree, 2005; Bryant, 2010) have claimed that “speakers adapt [ . . . ]
their way of speaking to mark that something was going on in the ironic
utterance. In that way, ironic intonation may not be hidden in a specific
intonational pattern, but rather in a contrast with preceding utterances”
(Burgers & van Mulken, 2017, p. 391).
Another area of research, albeit drawing less interest, are the markers
on humor at large. Adams (2012) examined markers in computer-
mediated communication. We will focus on the prosodic markers of
humor, which have been investigated by a research team led by Lucy

12 The presence of humor in a text indicates that the speaker of the text feels that using
humor in that context is appropriate and therefore other humor may act as a marker of the
potential presence of humor.
13 For example, hyperbole and/or understatement and irony go often hand in hand.

228
 10.3 EMPIRICAL STUDIES ON MARKERS OF HUMOR PERFORMANCE

Pickering and myself, with a number of students (some of whom are now
colleagues) from the Applied Linguistics lab at Texas A&M University-
Commerce.
There exists a folk-theory of humor performance, embodied primarily
in books and advice for would-be comedians on how to perform humor
on stage or in other situations (speeches, toasts). This folk-theory has
influenced academic research, as evidenced by some of the claims found
in the sparse literature. Broadly speaking, the expectation is that punch
lines, since they are the climax of the story, will be delivered in a marked,
emphatic way, “with bells and whistles” (Chafe, 1994). Confusingly, this
general area is know as the “timing” of humor.14 We used the existing
discussions to generate a set of hypotheses that we tested empirically on
10 speakers performing two canned jokes each. The hypotheses were
operationalized (Attardo & Pickering, 2011) as follows:

• there should be a significant pause (longer than 0.6 seconds) before

the punch line
• punch lines should be delivered faster and in a more clipped way
than the rest of the text
• there is a shift in voice quality (pitch) and volume

The results of the instrumental analysis of the performance of the

jokes were unexpected. The mean length of pauses in the setup of the
joke, as opposed to the punch lines was actually longer than the pauses
immediately before the punch line. Only in very few cases there was a
significant pause before the punch line and in one of those the speaker
got confused and hesitated, thus producing an involuntary long pause.
We found no significant difference in the speech rates of punch lines,
as compared to the rest of the text, although speakers do deliver them
slightly slower (i.e., the opposite of the prediction of the folk theory). We
found a significantly lower pitch (i.e., the opposite of the prediction by
Bauman, 1986, p. 68) and no significant differences in volume, again
contrary to the predictions of the folk theory (Pickering et al., 2009;
Attardo & Pickering, 2011).
How do we explain such a débacle for the folk theory? As it turns
out, canned jokes are (mostly) extended narrative single turns. When
14 Speech rate is the only aspect investigated in these studies that scricto sensu belongs
to timing. Other authors, such as Norrick (2001), have a much more expansive definition of
timing.

229
 THE PERFORMANCE OF HUMOR

350

300

250
Frequency(Hz)

200

150

100
I don’t have time for a girlfriend but now a talking frog that’s cool
50

0
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Time (sec)

Figure 10.1 A paratone, at the end of the performance of one of the jokes. “That’s cool”
is the punch line. Note the declination, indicated by the continuous line. The dotted lines
indicate the approximate resets of each tone unit.

speakers produce a longer turn there is a specific intonation pattern

for this called a paratone. Essentially a paratone is a spoken paragraph.
As can be seen from Figure 10.1 above, phonetically paratones are
distinguished by a high pitch/volume reset at the beginning followed
by a progressive lowering of the pitch/volume, called declination. This
causes the last syllables of the paratone to have significantly lower pitch
and volume. Since punch lines occur at the end of the narrative and
hence of the paratone, prosody trumps any attempt at emphasizing the
punch line prosodically.
We successively replicated the study using conversational data, to
see whether the nature of the data (monological performance in front
of a camera) might have affected our results. To do so we used data
collected in four dyadic conversations (8 participants) recorded over
video conferencing software. The speakers were instructed to tell one
another a joke (provided by the experimenter) and then to talk freely
for a few minutes. This provided us with canned jokes to compare to
the Pickering et al. (2009) data, as well as a set of naturally occurring
conversational humor (mostly jab lines). The results, summed up in
Attardo et al. (2013), are even more damaging for the folk theory
hypotheses. There are no significant differences for any of the measures
(pitch, volume, speech rate, and pauses). Since most of the data are jab
lines, they do not occur at the end of a paratone, and therefore we do
not see the lower pitch present in the punch lines.
While we had also included laughter in our accounts (with no sig-
nificant results), around this time, we also started considering smiling.
In order to analyze the smiling behavior of the speakers we developed a

230
 10.3 EMPIRICAL STUDIES ON MARKERS OF HUMOR PERFORMANCE

smiling intensity scale (SIS) ranging from no smile (0) to open mouthed
smile (4). The full description of the SIS can be found in the appendices
of Gironzetti, Attardo and Pickering (2016) and Gironzetti et al. (2016).
What we found was that speakers smile a lot in conversation. However
there is a reliable correlation between smiling intensity and humor. In
other words, the speakers do not necessarily smile more often when
humor occurs, but they smile more broadly. Furthermore they tend to
match each other’s intensity of smiling (i.e., they display synchronicity).
This indicates that the speakers are engaging in a negotiation of the
nature (key/frame) of the situation. Curran et al. (2018) show that
speakers are sensitive to differences in intensity of laughter.
The hypothesis that the speakers in face to face conversation are
actively negotiating the humorous frame of the interaction is further
supported by research presented in Gironzetti et al. (2016) and
Gironzetti (2017) using eye tracking technology, Gironzetti shows that
speakers when they engage in humor display a heightened interest in
the eyes and mouth areas of the face, where the Duchenne display of felt
smiles are produced. However, the relationship is far from linear: for
example, irony and jokes seem to comport different gaze patterns: more
attention is paid to the mouth when the humor type is irony, whereas
less attention is paid when the humor is a canned joke punch line.
Likewise, Spanish speakers show overall less attention to the face,
thus opening the possibility of research in cultural differences of gaze
patterns in humor. Other patterns including mutual gaze and gaze
aversion (looking away) seem to offer promising avenues of research
as well. Williams, Burns, & Harmon (2009) show that gaze aversion
correlates with sarcasm, for example.
Other research on the matter includes a collection of articles (Attardo
et al., 2013, originally published in 2011 as a special issue of Pragmatics
and Cognition) in which a few other studies concerning the prosody of
humor. Bird (2011/2013) is dedicated to the prosody of riddle questions,
essentially the first part of a riddle, such as (109).
(109) Why’d the mouse cross the road?
Bird’s findings are that

riddle questions and conversational questions differ significantly in their pitch charac-
teristics. Regardless of the gender of the speaker, riddle questions generally have less
pitch change over the utterance due to lowered maximum pitch, less average pitch
change within nuclei, and less drastic changes in pitch contour within nuclei on average.
(2013, p. 99)
231
 THE PERFORMANCE OF HUMOR

which is consonant with the results discussed above. Flamson et al.

(2011/2013) likewise find that speakers do not mark prosodically either
the punch lines of the set up of humor passages.
Not all research is consonant with the findings reported above.
Wennerstrom (2011/2013) examines two kinds of joke in which “the
intonation triggers the shift of the script and hence the humor” (p. 127)
which could be a counterexample to the studies presented above.15
A genuine counterexample can be found in Archakis et al. (2010) who
find that the humor in Greek adolescent girls’ conversations is presented
with significantly higher volume, lower speech rate and pauses before
and after the humor. Methodological concerns about the Greek data are
discussed in Attardo et al. (2015, p. 186–187). Hoicka and Gattis (2012)
found that mothers reading to infant children

use a higher mean F0, larger F0 standard deviation, wider F0 range, larger intensity
standard deviation, wider intensity range, and slower speech rate when expressing ver-
bal humour versus verbal sweet-sincerity. The second study also showed that humorous
and sweet-sincere utterance contours were significantly different. Humorous utterances
followed a rising linear contour, whereas sweet-sincere utterances did not tend to follow
any specific pattern. (p. 545)

Hoicka and Butcher (2016) find that parents signal a joking frame to
children using various cues that include intonation.
Fatigante and Orletti (2013) also consider smiling as a marker of
affiliation, whereas they consider laughter to be either affiliative or
disaffiliative. Ikeda and Bysouth (2013) mention gaze and smiling as
multimodal clues. They study multi-party conversations and find three
types of laughter: spectator laughter, which signals disengagement, turn-
initial, stepped up laughter, which signals engagement and a willingness
to take the floor, and delayed laughter, which shows a desire to set
themselves apart. Ruiz-Madrid and Fortanet-Gòmez examine humor
in plenary lectures using a multimodal framework. Further studies are
reviewed in Gironzetti (2017). Some studies are beginning to consider
the possibility of gestures as markers of humor; for example, Buján
(2019) examines a variety of gestures in talk show interviews, such as
facial expressions (smiling, raised eyebrows), head gestures (nods, shak-
ing etc.), bodily gestures, such as turning toward the interviewer, etc.

15 Interestingly, Wennerstrom described one of the special intonations as a L+H*, which

matches Double’s stand-up example quoted by Rutter, cf. example (103).

232
 10.3 EMPIRICAL STUDIES ON MARKERS OF HUMOR PERFORMANCE

She comes to the conclusion that none of the gestures is particularly

associated with humor and that the gestures when associated with
humor have the same meaning than when they are not associated with
humor. This contradicts Tabacaru & Lemmens’ (2014) claim that raised
eyebrows would be associated with sarcasm and humor at large.
Peräkylä and Ruusuvuori (2006) anticipated the frame-by-frame anal-
ysis of smiling in dyadic conversations. The authors selected assessments
(evaluative statements) and analyzed only those parts of the conversa-
tions, for affective displays. They do not use FACS, but define smiling
in the same terms. They find, in a case study, that facial expressions
display affect and assessment and that the speakers’ facial displays
are coordinated (synchrony). Their most interesting finding, from our
perspective, is that

there seems to be a particular continuity in the parties’ facial expression [=smiles] and
the affect that it incorporates: while the verbal action and an aspect of the topic change,
the affect, as displayed by the participants’ faces, in this case remains the same.
(p. 138)

In another study, based on the same materials, Kaukomaa et al. (2013)

consider “anticipatory” smiles, which occur before the verbal turn of a
speaker and find that they are all reciprocated by the interlocutor. The
turn-initial smiles were associated with “a shift in emotional stance”
(p. 22). Some of the smiles introduce a “humorous emotional stance”
(p. 23). The interlocutor may either “join the emotional stance” of
the speaker after the first speaker has “clarified the grounds” for it, or
immediately, i.e., without waiting to know why the speaker is smiling
(p. 27).
It is too early to draw any significant conclusions about the empirical
study of humor performance, except to say that there seems to be ample
evidence that the study of the markers of humor performance is shaping
up as a rewarding area of research, in which much remains to be done.
It seems that some results are relatively well established, such as the
refutation of the folk theory of “louder, faster, and after a pause” and
of a specific ironical tone of voice. Furthermore, the importance of
synchronicity cannot be overstated. However, we have to be clear that
research in this area is just getting started.
This concludes also our presentation of the theoretical background of
the study of the performance of humor. As we anticipated, only a full the-
ory as comprehensive as the ethnomethodological approach proposed

233
 THE PERFORMANCE OF HUMOR

by Hymes and Gumperz (and many others, including, Goffman, Sherzer,

Tannen, etc.) can handle the complexity of a complete account of humor
performance. In the next chapters we turn then to the specific details
of some of these analyses, beginning with conversation and discourse
analysis.

10.4 Further readings

There are many introduction to the study of discourse, in a broad sense.

I find Schiffrin (1994) to be particularly good. Saville-Troike (1982)
is a classic. The two-volume Handbook of Discourse Analysis (Tannen
et al. 2015) is comprehensive and authoritative. There are no com-
prehensive treatments of humor performance. The relevant chapters
in the Handbook of Language and Humor (Attardo, 2017d) are good
introductions to their respective areas of coverage. Brodie (2015) is a
very good introduction to stand-up comedy.

234
 11
Conversation analysis: humor
in conversation I

11.1 Conversation and discourse analysis 236

11.2 CA of laughter 239
11.3 The canonical CA joke analysis 246
11.4 Issues in CA of humor 252
11.5 Conclusions 261
11.6 Further readings 262

This chapter deals with the realization of humor in conversation, as

part of the broad Hymes-Gumperz ethnomethodological interactional
approach to the performance of humor. The topic here is how humor
fits within the flow of conversation. We first need to examine the
special methodologies used by conversation and discourse analysis, the
fields that specialize in the study of conversation and discourse. As we
will see, some of the underlying assumptions, such as the influence of
phenomenology, are somewhat different from the previous chapters and
need to be elucidated. Because conversation and discourse analysis are
one of the fields of research relevant to humor in which most research
has been published in the past 20 years or so, in order to keep the size
of the chapter manageable, I have had to split the discussion on two
parts. This chapter deals with Conversation Analysis and Chapter 12
deals with Discourse Analysis. Despite the methodological differences
between the two fields, sometimes it is difficult to classify a given study
as belonging clearly to either field. No special significance should be
assumed in my having dealt with a given publication in this or the other

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 CONVERSATION ANALYSIS

chapter. Sometimes it was convenient to keep different treatments of

a topic together and sometimes, despite the fact that the work is not
conversation analysis, it fit in the discussion (see for example the dis-
cussions of humor support and mode adoption, among the reactions
to humor).

11.1 Conversation and discourse analysis

Conversation analysis (CA) is used to refer to the research that originates

directly in Sacks and his collaborators and that puts an emphasis on
the sequential organization of conversations. Discourse analysis (DA)
is also interested in conversations but broadens the scope of analysis
in several directions. The most evident one is that CA, as the term
“conversation” indicates, focuses on spoken multiagent (small groups)
interactions, whereas discourse is more broadly defined as potentially
any type of text. In our analysis we will focus primarily on the functional
aspect of conversation, i.e., what speakers do with humor in conver-
sation. Another difference is that CA of humor rejects quantitative
analysis, if not outright, at least programmatically, whereas DA is more
open to it, especially in the group of scholars inspired by Holmes’
work. The CA stance against premature quantification is articulated in
Schegloff (1993).
This is not the place to discuss phenomenology in any detail, so we
will limit ourselves to noting that one of the tenets if phenomenology is
the idea that the explanations provided by the research must “ring true”
or be relevant to the people who are involved, not just to the scholars
doing the research. Another central tenet is that the object of research
must be ecologically valid, i.e., as naturally occurring and untampered
by the researchers. A further idea is that of withholding preconceptions
and assumed ideas and looking at the phenomena with “fresh eyes.” This
is called “bracketing.”

11.1.1 Bracketing
The idea of “bracketing” consists in putting aside (metaphorically in
brackets, hence the name) assumptions, preconceptions, biases, pre-
existing beliefs, and (significantly) theories and explanations, in order to
look at the data per se. In the case of conversational humor, this would

236
 11.1 CONVERSATION AND DISCOURSE ANALYSIS

mean stepping away from the idea that “someone intends to amuse their
audience and so tells a joke” to consider questions such as, How do
people tell jokes in conversation?, How does a turn in conversation get
counted as funny?, and Who is funny and why?
The first step of bracketing in CA and DA is the data collecting
and transcription. Frequently, CA analyses use participant observation,
which has the advantage in the case of humor, to allow an in-group
perspective, which allows the researcher to know what turns are meant
as humorous, for example. Traditionally, data were collected as audio
files. More recently, video and eye-tracking data have also been utilized.
Also, more recently, larger corpora not collected by the researcher have
been used. These studies will be addressed in Chapter 12.
The transcription is very fine grained and pays attention to overlaps,
latchings, pauses, etc. that usually get “edited out” of literary dialogue.
The transcription system of CA was heavily influenced by the work of
Gail Jefferson. We present some of the notation used in CA studies of
laughter and humor in Section 11.1.1.1, below.
Next, the conversational data are conceptualized as “turns” and not
as questions, answers, arguments, or repartees. The researcher can now
observe that speaker A utters a turn, consisting of, say a question, and
speaker B utters a turns which consists of a statement and if speaker A
treats the response he received in the turn uttered by B as an answer to
this/her question, then the researcher would classify this as a question
and answer sequence, rather than assuming that each response to a
question is an answer. So, for example, if A says something and both
A and B orient to it as an insult, then A insulted B. Note that the content
of what A said, is not considered particularly relevant, if at all.
11.1.1.1 Jefferson’s notation
One of the most significant contributions of classical CA to linguistics is
the transcription system, devised by Gail Jefferson, which allowed schol-
ars to capture some of the prosodic detail lost in conventional transcripts
of speech, while emphasizing the sequential organization of speech,
and providing the distance from the data necessary for the bracketing
required by the phenomenological methodology.1 The system is based
on traditional orthography and punctuation, with several modifications

1 See an example contrasting a word-by-word transcription and a CA style transcription,

in Wooffitt (2005, pp. 11–12).

237
 CONVERSATION ANALYSIS

aimed at capturing overlapping in speech, indicated by matching square

brackets and the placement of the transcript on the page, so that the
overlapping turns match in position on the page, like staves in music.
Some of the notation is provided in Table 11.1.
Most scholars adapt the Jefferson model for their purposes. For exam-
ple, in some of the work discussed in Chapter 10 precise measurements
of the length of pauses are required, and so the length of the pauses is
indicated in milliseconds, whereas in Jefferson’s system it is indicated in
seconds. Additions have been made for specific purposes, for example,
Chafe (2007) uses “” to indicate words spoken with a smilingvoice.
For a more complete discussion, see Atkinson and Heritage (1984, 1999)

Table 11.1 Notation for transcription, based on Jefferson’s

. sentence final falling intonation

, clause-final intonation (‘more to come’)
! animated tone
? rising intonation
italics emphatic stress
CAPS spoken much louder than surrounding text, or
prominent syllable
°words° spoken more quietly than surrounding text
: after a vowel indicates elongated vowel sound
:: more elongation
/words/ in slashes indicate uncertain transcription
wo[rds beginning of overlapping speech
[words
= latching
hhh aspiration
.hhh inhalation
HHH aspiration/laughter while speaking
(sarcastically) description of voice quality or non-verbal action
(...) part of a turn (or turns) has been omitted
(.) short pause
(7.2) pause with length in seconds
‘words’ speaker is quoting another person or adopting his/her voice
@ words @ smiling voice quality (also used for laughter pulses)

238
 11.2 CA OF LAUGHTER

and specifically on laughter Jefferson (1985) and Hepburn and Varney

(2013).

11.2 CA of laughter

It is fair to say that within conversation analysis the main focus has been
not on humor but on laughter. The reader will recall from Chapter 1
the discussion of the fact that humor and laughter are not coextensive.
This is not controversial, including within CA, which in fact contributed
a body of discussion on non-humorous uses of laughter in discourse
analysis, briefly reviewed below in Section 11.4.2. Witness also Clift’s
claim that “Perhaps the most significant outcome of Jefferson’s early
work on laughter was the now well-established analytic distinction
between laughter and humour (1984)” (2012, p. 1303).
Nonetheless, in conversational or discourse-oriented approaches, for
various historical reasons, which will be mentioned below, laughter has
been used as the indicator or marker of humor, witness the pellucid
statement by Kotthoff (2000, p. 64) “Laughter is the contextualization
cue for humor par excellence,” or Consalvo (1989, p. 288) definition
of “humorous interaction” as “communications in which at least one
participant [ . . . ] laughed.”

11.2.1 Laughter is indexical

CA starts from the basic assumption that laughter is indexical: “when
it [= laughter] occurs, people hear it as having a referent, pointing to
something in the immediate environment, typically just preceding or
concurrent” (Glenn & Holt, 2013, p. 3). Glenn (2003) states also that the
placement of laughter “displays precisely what the referent or laughable
is, typically via placement concurrent with or immediately following the
object” (2003, p. 49; my emphasis, SA). This idea was already found in
Sacks (1974, p. 348; Sacks, 1992, p. 745) “Laughing is the sort of thing
that, when it’s done it will be heard as tied to the last thing said” (my
emphasis, SA).
However, things are not that simple. Holt (2011, p. 393) also mention
that the referent of laughter might be other laughter or smiling and
“something external to the interaction.” Günther (2003, p. 195) reports
that “quantitative data on laughter position shows a relatively high

239
 CONVERSATION ANALYSIS

proportion of laughter in initiating moves or turns containing an initi-

ating move compared to response moves” (my emphasis, SA). Glen and
Holt concur, “Laughter can occur before, after, or during talk” (Glenn
& Holt, 2017, p. 6). So, recapping, laughter can occur before, after, or at
the same time as the referent is refers to.
Laughter may constitute an entire turn. It also may occur after a pause.
Laughter does not punctuate speech (Nwokah et al. 1999; Chafe 2007),
contra Provine’s claim (2000). The referent may be talk, other laughter,
or anything else. If the laughter may occur anywhere, how can hearers
identify the referent pointed to by the laugher? “Laughers will do things
that mark the referent of their laughs; other participants noticeably
orient to identifying what that referent is” (Glenn & Holt, 2013, p. 3).
This claim is too strong, in the sense that even though it is generally
true that people laugh at something, there are plenty of situations where
the something is left unspecified, deliberately or not. We refer to this
feature of humor as “indeterminacy.”
Moreover, laughter may occur away from its referent. Consider the
following exchange, from the same corpus used in Attardo and Pickering
(2011).
(110) Dyadic conversation between classmates
J hhh
//[i couldn’t carry a TUNE in a
BUCKet]//
M [hh]h
(0.90)
J //wow that wa[s really CORny] //
M [((clears throat))] //yeAH hh [h]//
J [heheh]
J’s joke revolves around the ambiguity of “carry” as in “carry a tune”
(sing following a melody) or “carry” (as in haul, physically move a load).
M responds with a long pause and some muted laughter, so J assesses
negatively her own joke. M concurs (“yeah”) and provides again muted
laughter. I would argue that in example (110) above, in the last turn,
J is not joining in with M’s laughter, which is asynchronous, but is in
fact laughing at her own corny (by her own self-assessment) joke in the
second turn of the fragment.

240
 11.2 CA OF LAUGHTER

While it is true that laughter tends to be tied to its immediate context,

it is also true that laughter per se has no specific referent, it is ambiguous
(Attardo, 2002, p. 169) or as Ikeda and Bysouth put it, “equivocal with
regards to its social meaning” (2013, p. 63). Assume that someone tells
a joke: the hearer may laugh because they got the joke, or because they
did not get the joke and are embarrassed, or because they did not get the
joke and do not want to show that they did not get the joke, or they may
laugh at the face made by a third party, etc. Often the situation is such
that the laughter is disambiguated by context, but, and this is my point,
it does not necessarily always get disambiguated.

11.2.2 The definition(s) of laughable

In CA, the concept of “humor” is considered too abstract and theoretical
to be safely used. It is, according to most Conversation Analysts, a
“typification (see Heritage, 1984, pp. 144–150) of events in interaction,
an abstract category that is insufficiently specific” (Glenn & Holt, 2017,
p. 295) to describe what is going on in interactions. To put it differently,
it does not bracket our common knowledge that some things are funny
and others are not.
Since saying that something is humorous prejudges the issue, accord-
ing to CA/DA proponents, the concept of “laughable” has been intro-
duced. Unfortunately, several definitions have been proposed and/or
used, thus leading to potential confusion. As with most concepts in CA,
laughable can be traced back to Sacks (1992, pp. 745–746) who notes
that laughing “has to be done directly after the utterance it’s laughing at”
(p. 745) and “you laugh at something that happened here” (p. 746) and
that people assume one laughs at the last utterance (if it’s not the case,
one will need to identify the utterance one is laughing at). Sacks uses the
term “laughable” (albeit as an adjective) on p. 746. In Jefferson, Sacks,
and Schegloff (1977) the term laughable is used as a noun.
Let us start our discussion with Glenn’s definition, which is as follows:
“any referent that draws laughter or for which I [= Glenn] can reasonably
argues that it is designed to draw laughter” (2003, p. 49). Let us focus first
on the first sentence in the coordinate construction. We will return to the
second part below. “Drawing” laughter is synonymous with “occurring
with” and so it boils down to the equating laughter and laughable.

241
 CONVERSATION ANALYSIS

The laughable is the referent of the laughter. This is clear and

straightforward.
However, Glenn explains that “the occurrence of laughter marks
its referent (usually retrospectively) as laughable – and potentially as
humorous” (2003, p. 33) and thus “funniness becomes understood not
as an inherent property of a message, or the internal state of a social
being, but rather as a jointly negotiated communicative accomplish-
ment” (2003, p. 33). In Glenn and Holt (2017) the connection to laughter
is more explicit, but nonetheless Glenn and Holt are very clear about the
fact that laughable does not imply humorousness, “Although laughter is
not necessarily a reaction to the presence of humor, clearly it does have a
close relationship with the (usually preceding) talk: when we laugh we
laugh at something” (p. 298; my emphasis, SA). The last sentence brings
up the indexical aspect of the definition of “laughable,” already discussed
above. So, overall, Glenn’s definition is that laughter is indexical, it is
a marker of a conversational item as a laughable and in some cases
(potentially) of humor.
Another definition, clearly incompatible with the Glenn-Holt indexi-
cal definition, comes from Ford and Fox: “To count as a laughable, a turn
(or part of a turn) must be produced with possibly laugh relevant sounds
and/or bodily displays, and it must be responded to with laugh-relevant
sounds or bodily displays (Ford & Fox, 2010, p. 340) Chovanec (2018),
who adopts Ford and Fox’s definition, is explicit: “the term [laughable]
refers to a turn that both interlocutors orient to as funny” (p. 157). To be
fair he immediately below problematizes the connection, thus somewhat
voiding the definition.
The main difference between Ford and Fox and the Glenn and Holt
definition is that Ford and Fox are much more restrictive, since they
would limit “laughable” to laughter to which both the Speaker and
the Hearer orient to it as such (for example, by laughing). This would
eliminate, all laughables that are single turns (for example, in trouble
talk). Regardless of whether one wants to pay this theoretical price, we
already have a term for an interaction in which both speakers orient to
with laughter: humor. There are other problems with the definition of
laughable, to which we turn in the next section.2

2 Within humor studies, “laughable” is a somewhat archaic term for humor; see for
example the titles of Grant (1924) “The ancient rhetorical theories of the laughable: the
Greek rhetoricians and Cicero” and Herrick (1949) “The theory of the laughable in the
sixteenth century,” so the terminology is unfortunate to begin with.

242
 11.2 CA OF LAUGHTER

11.2.2.1 Problems with the definition of laughable

Probably the biggest problem is that there seem to be several “levels” to
the definition (Glenn & Holt, 2013, p. 9). We will make this explicit in
this section.
Glenn’s (2003) definition with the escape clause that a laughable may
be “designed to draw laughter” (2003, p. 49) introduces the speaker’s
intentionality in the equation. Moreover, there seems a tendency to
equate laughter and humor, despite the explicit denials. This is made
particularly clear by the following quote: “Many laughable turns seem
far from anything humorous, for example, those that occur in tricky
or delicate environments such as complaints [ . . . ] or troubles-tellings
[ . . . ]” (Glenn & Holt 2017, p. 294). So, if laughter of embarrassment,
for example, is part of the category of laughable, which is by definition
not necessarily humorous, then why would a laughable be expected to
associate with something humorous?
A purely indexical definition has its epistemological validity: while it
assumes that “laughter” is a unified category, which is of course an unan-
swered question, as Glenn and Holt concede (2013, p. 5),3 it has on its
side simplicity and clear operationalization: if there occurs an instance
of laughter in the conversation, look for the immediately preceding
turn or the present turn, or a turn nearby for the referent/antecedent
of the laughter. The price to be paid for this definition is that it can
only accommodate cases in which laughter is present. The absence of
laughter is outside the boundaries of the definition. The non-occurrence
of laughter cannot have a referent, in the same sense that its occurrence
does.4 Thus, under this definition, if a speaker tells a canned joke
which is followed by silence, there would have been no reason to expect
laughter. One can add a theory of jokes to one’s theory of laughables,
but the point remains: the indexical definition of laughable can only
function for present laughter/laughable pairs. Relatedly, considering
smiles, or other features that make laughter “relevant” (2013, p. 9)

3 The question may be unaswered in CA, but it has long been answered, in the negative,
in psychology; see the discussion in Chapter 2.
4 This is not to say that non-occurrence of laughter may not be meaningful, as we saw in
“un-laughter” in Section 10.1.6. However, there is a significant difference between laughter
marking a laughable, in the sense, Sacks, Glenn, etc. mean (i.e., co-occurrence), and the
absence of response to a social expectation (unlaughter). The difference is of course that in
unlaughter the previous turn is identified as humor and the refusal to sanction its humorous
nature is what makes the unlaughter meaningful.

243
 CONVERSATION ANALYSIS

raises the question of why this should be so, unless laughter points
at more than its referent/antecedent, i.e., carries the association with
humor.
One could be picky here and point out that if we reject “humor” as
an abstract category superimposed on the data (a typification), then
laughter is as much of an abstract category. Glenn and Holt point out,
correctly, that “laughter” is a “member’s category” (i.e., it is part of
the lived experience of the participants), but of course so is humor.
However, we know from humor studies, primarily psychology, that
laughter is not a unified category (see the discussion in Section 2.2.3).
Readers will also recall from Chapter 1 that there can be humor without
laughter and laughter without humor. So the only coherent definition
of laughable is any turn that is preceded, accompanied, or followed by
laughter.
The definition of laughable is not merely a terminological matter, as it
creates a serious theoretical problem that has not been discussed in the
literature, to the extent that it should have been. Here we can only touch
briefly upon failed humor, which will be discussed in Section 12.4.3. A
preliminary definition, which we do not endorse but that will do for the
present discussion, is that failed humor is humor at which the audience
does not laugh.
The logical problem raised by failed humor is that, unless the failure
is remarked upon by the speakers, it can occur without any external
markings. Indeed, this is the case. Consider the following example, from
Hay (1994), already discussed above and here repeated for convenience
(111) Conversation among friends (Hay, 1994, p. 46)
1 Meena: I’m the only person in this room who freestyled
at nationals and came in second in the women’s
division
2 Dan: let’s face it Meena always comes second
3 Meena: yeah I know that’s cause Sue Willis always beats
me +
except at distance
4 Dan: we were actually making sexual innuendos well I
was.
Here turn 4, by Dan, makes it clear that he intended his turn 2 to be
humorous, but, if he had not uttered turn 4, Meena, who clearly missed

244
 11.2 CA OF LAUGHTER

the sexual double entendre, would have been entirely oblivious to it. So,
if Dan had not uttered turn 4, the analyst, much like Meena, would have
entirely missed the presence of the failed humor.
On the contrary, if we accept Glenn’s “designed to draw laughter”
definition then we are within the domain of speakers’ intentionality,
which can only be inferred. However, this brings us back full circle to
the definition of humor as a speech act whose perlocutionary goal is
the perception of the intention to be humorous (Attardo & Chabanne,
1992) and therefore, from the standpoint of humor studies, the concept
of laughable is redundant, one the one side, and misleading, on the
other, since it is dissociated from humor and laughter is not a unified
category (there are unrelated types of laughter that share no emotional
or cognitive component, as we saw in Chapter 2).
Another problem are false positives: for example, there exist doc-
umented instances of laughter response to a non-laughable (i.e., the
hearer laughing at something the speaker said seriously). Clift, for
example, mentions the “You’re kidding’ response,” i.e., a “response to
a turn not designed to be a so-called ‘laughable’ (Glenn, 2003)” (Clift,
2016, p. 75). In this case the hearer incorrectly attributes the intention to
draw laughter to the speaker. As we just saw, an even bigger problem are
false negatives: failed humor can be considered (reductively) the failure,
on the hearer’s part, to orient to humor after a turn “designed to draw
laughter.” We will return to failed humor in Section 12.4.3.
Ultimately, as we saw, laughter is indeterminate, as Curran et al. (2018,
p. 1) state, “laughter is inherently underdetermined and ambiguous, and
that its interpretation is determined by the context in which it occurs.”
In other words, if one sees someone laugh, one still does not know what
their “emotional state” is (p. 3).
In conclusion, to solve the ambiguity inherent in the definition of
laughable, I will adopt the following terminology: a laughable is a
referent (i.e., a part of the meaning of the text either explicitly stated
or inferred by the participants as part of the common ground) about
which someone laughs. A humorous laughable is one that meets the
semantic/pragmatic requirements for humor (regardless of the reaction
of the other speakers). Laughter is any vocalization as described in Chafe
(2007) and others. Mirthful laughter is laughter whose referent is a
humorous laughable, whereas non-mirthful laughter is laughter with a
non-humorous referent.

245
 CONVERSATION ANALYSIS

11.3 The canonical CA joke analysis

The founding text of the CA analysis of humor is Harvey Sacks’ analysis

of the telling of a canned joke in conversation. The conversation took
place among a group of teenagers. There was also an adult counselor
present but he is silent. The joke itself was an old-fashioned joke even at
the time of the telling (prior to 1974). The text of the joke is as follows:
(112) Three sisters marry three young men on the same day and the
mother of the young women talks them into spending their first
night at their home, each in a room. The mother, after everyone
has retired to their respective rooms, listens at the closed doors.
At the first one she hears an “Uoo-ooo,” at the second door an
“Yaaa,” but at the third door she hears nothing. The following
morning she asks the first daughter why she went “Uoo-ooo” and
the daughter says “It tickled.” She then asks the second daughter
why she went “Yaa” and she replies “It hurt.” Finally she asks the
third daughter why she did not say anything. The daughter says
“Well, you told me it was always impolite to talk with my mouth
full.” (Sacks, 1978, p. 251)
Sacks’ analysis is focused on two aspects: first the organization of the
telling event and second the function of the telling. We will consider
both and then address some issues with them. While Sacks intended
his analysis to apply to canned jokes, there is no inherent difference
between canned and conversational humor, so what follows can be
applied, with some reservations, to both canned and conversational
humor. The reader will recall that “jokes” are a genre of humor; however,
the term is often used loosely as synonymous with “instance of humor.”

11.3.1 Sacks on jokes

11.3.1.1 Sequential organization of joke telling
Joke telling is interesting in several ways. First, telling a joke is an
extended turn, which requires the audience to concede the floor for a
period of time. Sacks identifies three parts: the preface, the telling, and
the response. Table 11.2 schematizes the analysis. Patently, except for the
telling of the joke, all other activities are optional, i.e., we can have jokes
without a preface and without a response, but not without a telling.

246
 11.3 THE CANONICAL CA JOKE ANALYSIS

Table 11.2 Sequential organization of joke telling

Canned joke Teller Hearer

Preface disclaimers, secure floor, key humor accept or refuse joke, key humor
Telling text performance interrupt
Response signal end, incite laughter laughter

11.3.1.1.1 Preface The main object of the preface is to secure the floor by
announcing that one intends to produce a long(er) narrative. However,
this is not the only function: the speaker needs also to negotiate the
acceptability of the joke, for potentially offensive topics, but also for
the social appropriateness of producing a joke in the particular time,
situation, and broadly context. Moreover, the announcement of the
intention to tell a joke orients the audience to the fact that the coming
extended turn will be intended humorously so that they may frame or
key the situation as humorous. Yet another function is preemptive face-
saving: by warning the audience that the joke may not be very good, or
offensive, or that the telling may not be very good, the speaker preempts
any potential criticisms of his/her performance, potentially going so far
as to dissociate themselves from the text they are about to produce. The
hearer may turn down the offer of joke telling (for example, by saying: I
have heard it before, or by producing the punch line, thus demonstrating
that they know the joke).
Interestingly, the preface may be entirely absent, as is often the case in
conversational humor. It is however also possible to produce a canned
joke without preface. Absence of the preface is almost always required in
irony, with some exceptions in writing, where the detection of irony may
be more difficult (Jefferson, 1979; Edwards, 1984; Cashion et al., 1986).

11.3.1.1.2 Telling The telling consists essentially of one extended turn in

which the speaker delivers the joke. There may be supportive interrup-
tions, requests for clarifications, and of course disruptive interruptions,
if the audience doesn’t “play along.” Some genres of jokes require interac-
tive responses to be incorporated in the narrative, such as knock-knock
jokes.

11.3.1.1.3 Response According to Sacks, the response part of the joke

telling may consist of laughter, delayed laughter, or silence. The work
on failed humor has shown that there is, in fact, a much broader range
247
 CONVERSATION ANALYSIS

of reactions which may include verbal abuse or even actually hitting the
speaker. We will consider these reactions below in Section 12.4.3 and
will limit ourselves to the original three responses described by Sacks.
Sacks calls laughter the “minimal response sequence” because it is the
smallest and simplest response. There is now an extensive literature on
laughter in discourse, which we reviewed in Chapter 10. Here we will
only briefly consider some of the results that CA has produced about it.
Jefferson dedicated several studies to laughter. She notes that laughter
is often associated with “termination” and “closure” of “interchanges.”
Laughter, unlike other verbalizations, occurs in overlap with other
speech, and in fact mixed in with speech (“within-speech laughter,”
Chafe, 2007, ch. 3). Multi-party laughter may be in unison, “antiphonal,”
or the speakers may take turns, “relay” laughter (Smoski & Bachorowski,
2003).
Jefferson (1972, p. 300) notes that laughter is “regularly associated
with termination of talk” and so it can “signal or attempt closure of
interchanges” (p. 449n), while also signaling the uptake of the humorous
intention of the speaker. However, Jefferson’s most significant finding
about laughter was that speakers can use it to elicit laughter from
the hearer by producing it at the end of their turn (“post-utterance
completion laugh particle” ). This flies against the stereotype that one
should not laugh at one’s own humor, or at least should wait until others
have started and then join in. Instead Jefferson found that speakers used
laughter to incite laughter in their audience, both with utterance final
laughter and with within-speech laughter. Glenn (1989) shows that in
dyadic interactions, the teller frequently initiates laughter, whereas in
multi-party interactions, someone other than the speaker initiates the
laughter. Ikeda & Bysouth (2013, p. 52) show that speakers may laugh
turn initially to secure the floor raising the intensity of the laughter
gradually (stepped up laughter).
Laughter may not be produced immediately, i.e., it may be delayed.
Sacks sees the delayed response as a strategic compromise, on the
hearer’s part, between, on the one hand, the desire to display under-
standing of the joke,5 and, on the other hand, the desire to wait and
see how the audience reacts. Finally, laughter may be withheld. This

5 On jokes as a “test of understanding,” see Section 11.4.3, below. On understanding of

jokes as a status enhancer, see Graesser et al. (1989).

248
 11.3 THE CANONICAL CA JOKE ANALYSIS

is a dispreferred choice as “laughings have a priority claim on a joke’s

completion” (Sacks, 1974, pp. 348–249). Refusal to laugh, as we saw, has
a very negative connotation of displeasure (“unlaughter”).
It should be stressed in this context that different situations or genres
of communication afford different types of reactions. For example,
stand-up comedy is quite different in the ways in which performers
“manage” laughter and other reactions from the audience: for example,
comedians use various techniques to signal to their audience when to
laugh, applaud, or cheer, as we saw in Chapter 10. Needless to say, Sacks’
was not the last word in the analysis of joke (in a broad sense) telling in
conversation.
11.3.1.2 Humor support
The role of the audience is far from being passive. The audience may
reject the humor, in a disaffiliative move. Conversely, the audience
may provide affiliative support for the humor-producing speaker. Hay’s
groundbreaking work (Hay, 1994, 1995, 2000, 2001) introduced and
conceptualized the analysis of humor support, to which we turn next.
Hay introduced the notion of humor support to indicate a conversa-
tional move acknowledging and supportive of the speaker who produces
humor. Hay lists a number of verbal strategies, to which we can add
paralinguistic support (which could also be seen as included in the
“heightened” involvement category):

• contribute more humor

• repeat (part of) the previous turn
• offer sympathy or contradict self-deprecating humor
• overlap and heightened involvement in conversation
• heightened smiling intensity
• laughter
• facial expressions and gestures indicating agreement or amusement

Needless to say, non-supportive (disaffiliative) responses are also

present. Hay lists two items (the first two the list below) to which we can
add a few more

• withhold appreciation while showing understanding

• lack of reaction (this corresponds to Sacks’ “silence”)

249
 CONVERSATION ANALYSIS

Recognition Understanding Appreciation Reaction

Figure 11.1 Hay’s presuppositional scale of hearer’s reactions to humor.

• negative remarks on the quality of the humor

• claims that the situation or setting are inappropriate for humor
• claims to know a better/different variant of the joke (typically heard
at humor studies conferences)

Disaffiliative moves characterize some forms of failed humor (see below,

Section 12.4.3).
Hay’s further presents an extremely useful analytical tool, already seen
briefly in Section 2.3.2: a presuppositional scale of Hearer’s reactions (see
Figure 11.1):

1. recognition of the intention to produce humor

2. understanding
3. appreciation
4. reaction (Hay, 2001)

So understanding humor presupposes that one has recognized the inten-

tion of the speaker to be funny; appreciation presupposes understand-
ing, and reacting presupposes appreciation. This will figure crucially in
Bell’s analysis of failed humor.
As we saw, according to Sacks (1974, p. 348) laughter has a “priority
claim” but laughter is not required. Gavioli claims laughter is a “preferred
action” after telling a joke (Gavioli, 1995, p. 371). As we have seen above,
there have even been claims that joke/laughter may be an adjacency
pair. However, not all humor socially requires support: humor that is
itself supportive does not require support. Irony likewise need not be
supported.
One form of humor support is the repetition of the jab line. The
repetition may be verbatim or in paraphrase. Consider the following
example, which ends a longish narrative in which P tells M about his
trip to Galway.

250
 11.3 THE CANONICAL CA JOKE ANALYSIS

(113) Dyadic conversation among classmates

221 P //uh taken the BUS uh into//
222 0.2
223 //into GALway was that //
224 1.3
225 //uh when I came around the CORner //
226 0.42
227 //uh there is uh you know of course the HUGE rolling//
228 0.15
229 //beautiful hillsides of Ireland and //
230 0.43
231 //there’s this little dodge around the CORner
232 and you go around the corner and the first three things I
saw were McDOnalds //
233 0.3
234 //a Pep Boys and an Aldi’s GROcery store //
235 0.81
236 M //wh:::ich [((laughs)) that’s very IRISH [((laughs))] //
237 0.32
238 P //yeah you really don’t expect THAT //
239 0.3

The jab line begins on unit 232 and concludes in unit 234. After a pause,
M acknowledges the humor by laughing and then paraphrases ironically
the jab line (thus both displaying comprehension and supporting the
humor) and P supports the supportive turn by the asseveration (“yeah”)
and another paraphrase of the jab line. Note also that by producing an
ironical turn, M has mode adopted, i.e., joined in the production of
a sustained humor turn (see below). Humor support is not limited to
spoken interactions; for example, see Vandergriff & Fuchs (2012) on
humor support in online conversations.

11.3.1.3 Mode adoption

Mode adoption can be defined as being “in essence, about adopting a
broad manner of speaking, a manner that has been established via the
joint involvement of the speaker and listener. This can be thought of
in terms of play frames” (Whalen & Pexman, 2017, p. 377). The term
“mode adoption” was introduced in Attardo (2002) and as Whalen and

251
 CONVERSATION ANALYSIS

Pexman (2010, 2017) note, it has been applied primarily to the ironical
mode (i.e., an ironical response to an ironical statement) but in principle,
it can be applied to any humorous mode, for example responding with
a pun to a pun, or capping a joke with another joke. As Whalen and
Pexman observe, even mere repetition of a speaker’s turn can be seen as
mode adoption. Pexman and Whalen have extended the study of mode
adoption to children.
There is evidence that mode adoption is a highly affiliative practice, as
Drew (1987) reports that very few teases are responded to with teases;
instead, speakers respond seriously (po-faced) and primarily attend to
the criticisms of the tease. I can report that when speakers have an
extremely high level of intimacy, teasing mode adoption may occur;
my daughter and I enjoyed mutual teasing for most of her childhood.
Responding with a joke to a joke is the most common form of mode
adoption.
The question of the frequency of mode adoption is unresolved, with
estimates ranging from a high of 21–33% (Gibbs 2000) to lower percent-
ages such as Eisterhold et al. (2006) 7% and Pexman et al. (2009) 7%–8%
(conversations involving children). See Section 8.2.2 for more examples.

11.4 Issues in CA of humor

11.4.1 Is humor–laughter an adjacency pair?

The problem is discussed in Glenn and Holt (2013, pp. 12–14). Surpris-
ingly, despite Glenn and Holt’s claim, echoed by numerous other schol-
ars, that Sacks (1974) claims that joke/laughter (or laughable/laughter)
is an adjacency pair, the term “adjacency pair” does not appear in Sacks
(1974). It does appear, however, in Jefferson, Sacks, and Schegloff (1977)
a pragmatics microfiche characterized as a “revised draft of a paper
presented” in 1973, at a seminar at the university of Pennsylvania. In this
document, on p. 30, there appears the claim that “Laughable/Laughter”
would be an adjacency pair, along the lines of question/answer and
greeting/return greeting. The assertion is couched in the conditional
mood (“the Laughable and its responsive Laughter could constitute [. . .]
an adjacency pair [. . .] In this case the Pair Type is [Laughable/Laughter]
and may be affiliated to a list which includes” the above mentioned

252
 11.4 ISSUES IN CA OF HUMOR

adjacency pairs).6 One may speculate that the reason for not using the
term in the latter paper is that Sacks might have further developed his
thinking on the subject. In the 1974 paper, Sacks makes several claims
that are incompatible with the idea the laughable/laughter be an adja-
cency pair: laughter has a “priority claim [. . .] But each recipient is not
obliged to laugh” (p. 348). Hearers may choose to be silent, to delay their
laughter, or produce “talk directed to grading the joke and teller’s wit
negatively” (p. 351). Obviously if the hearers have four options (laughter,
delayed laughter, silence, and negative talk) laughable/laughter is not an
adjacency pair, stricto sensu.
Other authors have also claimed that humor and laughter are an
adjacency pair, such as Norrick (1993) and Holcomb (1997). For exam-
ple, Norrick (1993, p. 23) in the case of puns: “The appropriate initial
response to a pun is, of course, laughter. Thus we can say that joking
and laughter are linked as two parts of an adjacency pair.” Thonus (2008,
p. 334) initially quotes the definition of “laughable (humor)/laughter”
as an adjacency pair, only to problematize it immediately and eventually
rejects it (p. 335). Schegloff speaks of jokes “sequentially implicating”
laughter and does not use the term adjacency pair (1987, p. 206).
Hay (2001) is also critical of the humor/laughter as adjacency pair
idea as she presents a set of humor support strategies, as seen above,
in Section 11.3.1.2. If there are a number of options, laughter cannot
be a member of an adjacency pair. Hay (2001) and Sacks’ own work,
seem to point toward the conclusion that humor and laughter are not
an adjacency pair and that the hearer of a humorous turn (or laughable)
has a set of choices (a repertoire) with which to respond.

11.4.2 Humorous and non-humorous laughter in conversation

The purported correlation between humor (stimulus) and laughter
(response) has been shown not to hold, in fact. There is extensive
research in discourse analysis that documents the presence of laughter
in non-humorous contexts: Potter & Hepburn (2010) sum up nicely
the issue stating that laughter (broadly construed) “may have a range
of interactional roles quite separate from punning, joking or humour”
(p. 1544). Clift (2012) considers “laughter in non-humorous contexts”

6 Capitalizations in the original.

253
 CONVERSATION ANALYSIS

(i.e., non-mirthful laughter) and specifically laughter that is not shared

(i.e., not answered by laughter).
Jefferson (1979) showed in a groundbreaking article that the stereo-
type of the speaker saying something funny and the hearer responding
with laughter was hopelessly mistaken. In fact, speakers often initiate
the laughter before the hearer, to signal and invite a laughter response.
In Jefferson (1984), she shows furthermore that speakers will also laugh
in contexts in which there is nothing at all humorous, as in the example
below (I have simplified the transcription):
(114) Dyadic conversation quoted by Jefferson (1984)
G: You don’t want to go through all the ha:ssle?
S: hhhhI don’t know Geri,
(.)
S: I’ve stopped crying uhehh-heh-heh-heh-heh,
G: Wuh were you cry::ing? (p. 346)
It is clear that S is not being facetious in his/her answer and in fact the
speaker is “exhibiting that, although there is trouble, it is not getting
the better of him[/her] ( . . . ) he[/she] is managing; he[/she] is in good
spirits and in a position to take the trouble lightly (p. 351). As Jefferson
notes, Geri responds not by joining in the laughter, which would be
inappropriate, but rather by a “recognizably serious response” (p. 346).
Jefferson sums up the role of the hearer in trouble-talk as “taking the
trouble seriously” or being receptive to it. (p. 351). Potter and Hep-
burn (2010) analyze a striking example of a speaker laughing during
the description of an accident that broke several of his vertebrae and
caused him to be late in paying his bills. They comment with splendid
understatement “this could be heard as a somewhat odd description
of his own reaction to a spine crushing injury” (p. 1549) but go on to
explain that the three laughter particles occurring in the middle of the
words “I am very grateful” (that the injury was not any worse) mark
his reaction as “insufficient or problematic.” Needless to say, there is no
humor in this context.7
Günther (2003) reports that in her corpora, in 8 instances she was
unable to see a connection of laughter with humor (p. 152; 183).
Edwards (2005) describes laughter in serious telephone complaints.

7 In fact, Potter and Hepburn do not even use the term “laughter particle” and prefer
“interpolated particles of aspiration” to avoid confusion with mirthful laughter.

254
 11.4 ISSUES IN CA OF HUMOR

So does Clift (2013) from which we take as an example the following

fragment,
Robbie: my: ´God there’s a lotta children in this [h)ss
(115) cla(h
Lesley: [I ´ kno::w.
(Clift, 2013, p.231; I have simplified the transcription for clarity) in
which we see two teachers complaining about class size Clift notes that
the hearer (Lesley) does not orient to the laughter “infiltrating the final
word” (p. 231; i.e., the two aspirations) and instead responds to the
complaint by expressing solidarity. Clift notes the similarity with trouble
talk and with other “delicate” subject matters. Laughter in this context
serves as a mitigator, by introducing a distance between the complaint
and the person producing it (p. 236), and is thus not an expression of
mirth.
Another situation where non-humorous laughter occurs is doctor–
patient interactions. West (1984) found that patients laugh more than
doctors and interprets this as evidence of the power imbalance inherent
in the situation. However, Heath (1988) and Haakana (2001, 2002)
convincingly argue that it is more likely that patients are laughing
because of embarrassment. Haakana (2002, p. 226) shows that numer-
ous patient-initiated turns consist of reactions to “delicate and dispre-
ferred” activities, such as refusals (a dispreferred choice), discussions
of financial matters, etc. Fatigante & Orletti (2013) also report laughter
in gynecologist examination talk in dispreferred responses (the patient
contradicting the doctor). In short, when patients laugh at delicate
or dispreferred choices they are not displaying mirth, rather they are
embarrassed. Likewise, Zayts & Schnurr (2011, 2017) report that in
prenatal counseling, both nurses and patients laugh when disagreeing
or resisting the other’s opinion (the nurses are trying to push prenatal
testing and the patients refuse). Zayts and Schnurr interpret this as
laughter mitigating the disagreement or as a “laughing off ” (2017,
p. 135) the dispreferred choices of the interlocutor.
Wilkinson (2007) reports that patients with language impairments
will laugh during repair sequences when the repair sequence fails (i.e.,
they cannot successfully produce the utterance). Hearers do not respond
with laughter in those cases. The following example (simplified) will
illustrate the completely non-humorous nature of the situation. Paul, the
aphasic speaker, is talking to his wife, Amy:

255
 CONVERSATION ANALYSIS

(116) Paul: that was the doctor. she’s (.) trying to find out about eh:
(1.6) ((clears throat))
(1.3) mm well ehm mm
(1.6) ((sniffs)) ehm eh
(5.5) h (.) (g)hhh (Wilkinson, 2007, p. 548)
Paul forgets the word for what the doctor was trying to find out and
then processed to produce a series of empty and filled pauses which
culminates with the short laughter burst after the 5.5 seconds pause,
when he gives up. Amy simply provides the completion: “What you
have been doing at home” without any laughter. Paul’s laughter may be
interpreted as laughter of embarrassment, or as an attempt to dismiss as
trivial his problem.
A different type of phenomenon is a “humorous noticing” defined
by Wilkinson as consisting “of the highlighting of an error by either
repeating it or noting its similarity to another word which would be
incongruous in the context. As such, humorous noticings are a kind
of small joke produced by the aphasic speaker” (2007, p. 556). The
speakers produce laughter within the noticings. This kind of laughter is
reciprocated by the hearers. The difference between the two situations
described by Williams is striking: in the case of embarrassed laughter,
much like in trouble talk laughter, joining in the non-humorous laughter
could be interpreted as mockery. In the case of noticings, the speaker
successfully keys the situation as humorous (either by laughing prior to
repeating the error, and/or smiling), and thus laughter by the hearer is
seen as supporting the humor (see Section 11.3.1.2). Wilkinson et al.
(2003, p. 63) further document “embarrassed” laughter by aphasic sub-
jects about “the exposure of their linguistic non-competence.”
Dionigi & Canestrari (2018) report patient laughter to repair disagree-
ments between therapist and patient, mitigate “overt resistance” (p. 7),
self-deprecation, and positive assessment of the therapist. The therapist
does not reciprocate in any of these instances, as there is no assumption
of mirthful laughter.
Lavin & Maynard report “apologetic” laughter (2001, p. 460) ini-
tiated by respondents in telephone interviews, They compare it to
Jefferson’s examples of non-invitational laughter in troubles talk. Holt
(2017) shows that callers to a public utility company use laughter to
shift from serious business, such as complaining for a large bill, to less-
serious (or partially serious) complaints, such that the computer system

256
 11.4 ISSUES IN CA OF HUMOR

is slow. Despite the presence of laughter, Holt claims, “serious actions

are performed” (p. 96).8
Clift (2016) “incredulous” disaffiliative laughter consists in framing
the situation as laughable despite no such intention of the speaker:
“a party to interaction constructs a prior action as risible or preposterous
by means of laughter – as literally ‘laughable’; the practice of laughing
in such contexts does an emphatic display of incredulity or disbelief ”
(p. 86); “the laughter does a display of incredulity” (p. 80) and not of
mirth.9 Importantly, Clift (2016) notes that this sort of disaffiliative
laughter may occur at any point in the conversation (i.e., is not limited to
adjacency pairs) and may be produced by any of the participants. Also,
while being disaffiliative to the person being mocked, the laughter is
affiliative with (and may be joined by) a third person with whom the
speaker producing the disaffiliative laughter “sides” against the person
being laughed at.
Haugh (2017c) also discusses disaffiliative language play in jocular
mockery. So does Romaniuk (2013) in the context of broadcast inter-
views. Günther (2003, p. 194) reports a much higher incidence of affilia-
tive laughter than disaffiliative (finding in her corpora only 8 instances of
disaffiliative laughter, contrasted with 45 affiliative instances). However,
these results should be assessed carefully. The category of affiliative
includes Supportive and invited laughter, but excludes what Günther
labels “contextualizing” laughter, which would include all contextual-
izing cues. Furthermore, not all instances of laughter are taken into
account. Nonetheless, Günther’s data show that disaffiliative humor in
conversation does exist, although it is infrequent.
Adelswärd (1989) considering laughter in institutional settings,
broadens the perspective and comes to distinguish between recip-
rocated (“mutual”) laughter, a sign of “rapport and consensus,” and
8 To be precise, it is unclear if Holt is claiming that there is no humor present when,
for example, a speaker says that he/she “nearly had a heart attack” when they received the
very large bill, which seems a humorous exaggeration, or that the presence of the humor is
overwhelmed by the serious nature of the task surrounding it, or that the speaker is using
humor to save the negative face of the employee, by downgrading the complaint, turning it
into a merely laughable error. It is also unclear why Holt seems to think that there should be a
“simple relationship between incongruity and laugh responses” (p. 115) since at this point it
is a pretty safe conclusion (see Attardo, 1994) that there is no direct relationship between the
two, incongruity being a proxy for humor, in this context. Regardless, some of the examples
show non-humorous laughter in a serious context.
9 The analysis is complex, as it involves the speaker producing the laughter pretending
to be amused by what they consider the absurdity of the targeted other speaker’s remarks. It
remains that the situation is different from someone being genuinely amused by someone’s
remarks (say, a child making a clearly incorrect statement).

257
 CONVERSATION ANALYSIS

unreciprocated laughter: “We often laugh alone and not always at things
considered particularly funny.” For example, she reports that successful
job interviews showed more reciprocated laughter than unsuccessful
ones.
In conclusion, both non-humorous laughter and disaffiliative humor-
ous laughter have been well-documented in a variety of contexts, includ-
ing conversational ones. Neither type is usually responded to with
laughter. The preferred response to mirthful laughter is humor support,
the preferred response to non-mirthful laughter is ignoring it, either
by attending to the source of the embarrassment, discomfort, etc. or by
expressing support for the speaker. This is presumably because laughter
carries a connotation of mirth (by association with it). This is what
allows it to be a mitigator and to downgrade utterances to (partially)
non-serious mode. When both parties in the conversation orient to the
laughable (in the Glenn & Holt sense) as appropriate for laughter, they
do orient to humor.

11.4.3 Is humor a test of understanding?

The term “understanding test” is used verbatim by Sacks (Sacks, 1974,
p. 346) and defined as follows: “Not everyone supposably ‘gets’ each joke,
the getting involving achievement of its understanding, a failure to get
being supposable as involving a failure to understand. Asserting under-
standing failures can then reveal, e.g., recipients’ lack of sophistication, a
matter that an appropriately place laugh can otherwise conceal” (p. 346).
As is well known, Sacks’ example (the only example of joke he analyzed
in this context) is an obscene joke which refers obliquely to fellatio. Since
the teller is a 16–17 year old in the 1960s or early 1970s, it is credible that
not all the hearers may have been familiar with the practice of fellatio,
or at least the 12-year-old sister who originally told the joke (the brother
is repeating it to his friends). However, it seems extremely unlikely that
a joke such as
(117) How to do tell if an elephant was in the fridge?
Paw prints in the butter.
is testing anything in the knowledge of the audience. What factual
knowledge could possibly be tested in the example above? The knowl-
edge of the relative size of elephants and refrigerators? Thus at best,
Sacks’ understanding test could apply to some jokes, but not to all jokes.

258
 11.4 ISSUES IN CA OF HUMOR

Despite its prima facie implausibility, the idea of a test has taken
hold in conversation analysis (Sherzer 1985; Norrick 1993, 2003). To
be fair, Norrick’s position is much more hedged than others, when he
notes that rather than testing knowledge, jokes presuppose it for in-
group solidarity building. In fact, in 2003, he states clearly that “Far from
testing for background knowledge with jokes, tellers commonly fill the
audience in on any information the joke presupposes in the interest of
ensuring understanding and enjoyment, and hence the success of the
performance” (pp. 1342–1343). In Attardo (1994), several issues with
Sacks original discussion of the “test of understanding” hypothesis, as
well as some issues with the analysis itself, are discussed.
Glenn (2003, pp. 115–117) provides an example of how a participant
to a conversation may be mocked if they do not “get” the joke. Glenn
reconceptualizes the test of understanding as a test of participation
to the social event, since it seems unlikely that the speaker would
not understand what “getting an F [= lowest possible grade] in sex”
could mean. Indeed, the speaker himself in subsequent turns explains
that he failed to hear the words of the reply. In this sense, there is a
test of participation, as a competent speaker would have easily faked
understanding, for example by snorting in (fake) dismay at the poor
quality of the joke. In this sense, but only in this sense, we can accept
the idea of a test of understanding. In the original sense, i.e., whether
the hearer is familiar with the referent alluded or mentioned in the
joke, the idea of a test can only be a rare exception.
A related idea, also presented by Sacks, but that has gotten virtually no
recognition in the field, unlike the very popular understanding test, is
the “recipient comparative wit assessment device” (Sacks, 1974, p. 349–
350) which consists of the idea that since laughter is a display of under-
standing, delaying laughter may be interpreted as lack of understanding,
which therefore would motivate speakers to laugh as soon as possible,
despite the option of delayed laughter (i.e., laughing after a pause). The
advantage of delayed laughter being that it allows one to assess how the
other recipients react to the telling before committing to laughter.

11.4.4 Tellability
Norrick’s application of the concept of tellability to humor in discourse
is one of his most significant contributions to humor studies. Tellability
is the term introduced by Sacks (1992), whereas Labov (1972, p. 366)

259
 CONVERSATION ANALYSIS

speaks of “reportable” narratives. The concept is the same: a story,

narrative, or just an extended turn, needs to be justified (“evaluated”)
by the narrator to his/her audience, least the narrator be accused of
wasting his/her audience’s time. A narrative is tellable if it is newsworthy
or particularly relevant to its audience. So, Labov’s famous “near-death”
stories are tellable by definition. Surprising, unexpected, or novel events
are also tellable. Another aspect of tellability is that the story needs to
have a point, not in the senses of the “pointe” or punch line, but in the
sense of making a point, having a coherent narrative arc that delivers a
bit of information significant enough to justify the telling of the story.
Labov notes that pointless stories are met with “a withering” “so what?”
to indicate that the audience has not seen the point of the story.
Norrick (2004) notes that the humorous nature of a story may be
enough to warrant its tellability: “an orientation toward the humor of a
story replaces” its tellability (p. 86). Norrick speculates that what makes
a story tellable is the same kind of thing (incongruity) that makes a story
funny (p. 87).
Norrick (2005, 2010) elaborates the notion of tellability, not limited
to the lower boundary of being worth telling, but also in terms of
the appropriateness of telling the story in a given context, and shows
that speakers actively negotiate both aspects of tellability. Canned jokes
inherently have a certain degree of tellability, since somebody bothered
to repeat the text, but for personal anecdotes and worse yet for puns,
which may be disruptive of the conversation, since they are most often
off the current topic, the tellability has to be negotiated. Norrick notes
that both meaningfulness and appropriateness in context are negotiated
by the participants. Furthermore, in settings with high degrees of inti-
macy, such as family or friends gatherings, familiar10 humorous stories
may be retold, not so much for their novelty or newsworthiness, but
rather because the process of collectively re-creating the story reinforces
the social bond: “Humor makes any story more tellable, even familiar
stories, and humorous stories have characteristic patterns of participa-
tion: in particular, co-narration is acceptable whenever it creates humor”
(Norrick 2010, p. 238). Telling the humorous anecdote becomes more of
a shared performance, than just the telling of a story.

10 No pun intended, but note that etymological source of the word “familiar” (i.e.,
belonging to the family).

260
 11.5 CONCLUSIONS

11.5 Conclusions

One of the main conclusions that we can draw from this review of the
field is that it has reached a certain maturity. At this point, the list
of settings (speech events) that have been covered, often by multiple
studies, is truly impressive (see Glenn & Holt, 2013, p. 18, for a list).
Researchers have now available a broad range of descriptive studies
which offer minutious, in depth analysis of significant amounts of natu-
rally occurring speech acts, mostly conversational data. I think it is fair
to say that it is unlikely that we will see any major novel configurations
heretofore unexamined by scholars in terms of humorous speech acts
and speech events, as far as English goes. The situation is different for
other languages, which have a lot of catching up to do. In terms of
speech situations, the broader setting postulated by Hymes (1972), we
still have a ways to go, in terms of a truly representative set of situations,
but at least for conversations among friends, workplace talk, medical
encounters, and stand-up comedy, there are already sufficient data. The
field still shows an over-emphasis on the positive, affiliative aspects of
humor (see Attardo, 2015 for discussion), but new research is correcting
this imbalance.
There seems to be some confusion between mirthful (humorous)
laughter and non-mirthful laughter, as defined above. When mirthful
laughter is reciprocated we have humor support (see above, Section
11.3.1.2). Non-mirthful laughter, as in trouble talk (Jefferson, 1984),
does not generally get reciprocated. So what happens when we have
reciprocated non-mirthful laughter? We can offer a conjecture: assume
ex hypothesis that reciprocated laughter generally occurs only with
mirthful laughter. The preferred response to non-mirthful laughter is
serious bona-fide relevant communication. Then, reciprocated non-
mirthful laughter is, or is perceived as, a highly disafilliative (as in the
examples reported by Clift, 2016) or aggressive move. Thus in Zayts &
Schnurr (2017, p. 134) we see an example of reciprocated non-mirthful
laughter, in which the nurse comments on the refusal of the testing by
the patient (I paraphrase and gloss the complex transcription):
(118) N: You don’t want the test? [within utterance and post utterance
laughter]
P: [laughter overlapping with the previous turn] I am sure (I do
not want it) [within utterance laughter]

261
 CONVERSATION ANALYSIS

N: You don’t want it [within utterance laughter]

P: I don’t want to (do it) [within utterance laughter]
I would suggest that rather than describing this as reciprocated laughter,
we consider it as antagonistic laughter (i.e., as an attempt to “laugh off,”
to use Zayts & Schnurr evocative wording, each others’ objections).
At the same time, it is quite possible that, as Zayts & Schnurr say,
the laughter mitigates the dispreferred choices of both speakers, by
marking the antagonistic conversation as not-that-serious (Holt, 2017).
This is somewhat paradoxical, as laughter would be both affiliative
(mitigating) and disaffiliative (antagonistic), but it is not incompatible
with the indeterminacy of laughter: it is not clear (and probably not
really knowable) what the nurse and the patient are laughing at (each
other? what they said? their attitudes? their expressions?) and of course it
is not clear what the function of the laughter is (mitigation, disaffiliation,
or just trying to bring the exchange to a close), whereas it is relatively
clear that there is no mirth involved in the exchange above.

11.6 Further readings

Due to its methodological stance against generalizations, CA tends not

to produce overviews of the field. A review of the place of CA approaches
to humor and laughter can be found in Attardo (2015). A full discussion
of the CA notation system can be found in Atkinson and Heritage
(1984).
On bracketing and its origins and definitions, see Tufford and
Newman (2010). On bracketing in CA/DA, see Ten Have (2007). The
clearest statement of the methodological bracketing of CA/DA in humor
research is found in Glenn and Holt (2017), which is an indispensable
collection of papers. Norrick (1993) is now dated, but it is a classic in the
field as is Glenn (2003). The “antiphonal laughter” term is introduced
in Smoski and Bachorowski (2003).
The indeterminacy of humor is discussed in Mulkay (1988: 70–71),
Attardo (2002a), and Priego-Valverde (2003) while the indeterminacy of
irony, in Sperber and Wilson (1986) and Gibbs (2012). On humor sup-
port and mode adoption, see Whalen and Pexman (2017) and Coolidge
(2019).

262
 12
Discourse analysis: humor
in conversation II

12.1 Functional DA 264

12.2 Conversational humor in various settings 280
12.3 Corpus-based discourse analysis 286
12.4 Some issues in the DA of humor 289
12.5 Conclusions 297
12.6 Further readings 298

In this chapter, we will consider the discourse analysis (DA) of humor.

As we said in the previous chapter, there is considerable methodological
overlap between conversation analysis (CA) and DA: both emphasize
ecologically valid data and both look in a very detailed way to the texts,
for example. However, there are some differences as well: the first one is
that DA has a broader outlook, in terms of data: CA is more focused
on conversations, whereas DA admits more or less any kind of text,
including written ones. DA is also more open toward quantification and
has moved more readily toward using large corpora (see Attardo, 2015,
for a review). Last but not least, DA is more interested in the functional
aspect of the interactions, i.e., what interactional goals are the speakers
achieving through the exchanges.
While by now it should be clear that any attempt at drawing a sharp
line dividing CA from DA is impossible, I feel like I should repeat the
warning not to attach any particular significance in a given article or
book appearing in this or the previous chapter (or any other, for that
matter). My primary choice in grouping studies has been subject matter

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 DISCOURSE ANALYSIS

regardless of methodology: for example, Hay’s work is probably best

characterized as DA, but it is discussed in the previous chapter because
it fit better in the development of the discussion there.

12.1 Functional DA

Moving away from the emphasis on the sequential organization of

humorous discourse of classical CA, DA focuses more on the conver-
sational goals of the speakers, or to put it differently what the speakers
are trying to achieve by using humor. Here there is a logical problem:
as Holmes (2000) notes “all utterances are multifunctional ( . . . ) Hence,
a humorous utterance may, and typically does, serve several functions
at once” (p. 166). Furthermore, Holmes warns of a potentially infinite
regress: “the most general or basic function of humor is to amuse. But
one can ask why does the speaker wish to amuse the audience” (2000,
p. 166). Holmes is of course right to point out these problems. To ask
“why do speakers do X?” is a question potentially impossible to answer.
For example, let us say that a speaker produces a humorous turn and
we establish that he/she intended to amuse his/her audience with the
further intention of showing off his/her wit and thus reinforce his/her
social status in the group. Surely, we may believe that our analysis is over.
But what of unconscious goals, of which by definition the speaker will
not be aware of? And what of the ideological and cultural norms that
force upon the speaker the social roles that he/she worked so hard at
reinforcing?
One final point: humorous utterances (i.e., the production of humor
of any kind in conversation and/or communication) are indeterminate,
“ambivalent” (Holt, 2017, p. 94), or as Kotthoff (2007, p. 287) has it:
“Humorous intentions can seldom be pinned down exactly.”
Consider the following example:
(119) A is the 80-year-old grandmother of B, a 30-year-old female.
B had just run up over a curb driving out of a store.
A: They just built that while you were in the store.
B: I know. (Attardo, 2002b, p. 171)
As I said in my original analysis, it is fairly clear that A is being ironical,
as the claim that the curb was built while her grand-daughter was in the
store is patently false. B has no reason to think that her grandmother

264
 12.1 FUNC TIONAL DA

has become intellectually incapacitated, etc. However, I asked in 2002

“But is A being funny? Is A teasing B? Is A exaggerating? Is A being
metaphorical?” We may add: why is the grandmother saying this? Is
she trying to get her grand-daughter to drive more carefully? Is she
mocking her? Is she showing solidarity? Is she critiquing capitalist
exploitation of natural resources for the sake of short term profit of
the corporations that built the store? As I answered myself in 2002:
“Any attempt at disentangling this cluster of implicatures would be
quixotic, as the completely noncommittal answer of B seems to imply”
(p. 171). Needless to say, I acknowledged then and repeat now, that we
have no way of knowing what B meant, anymore than we may know
what A meant.1 For the exchange, we only can deduce that B is “going
along” with A, i.e., she is adopting the ironical mode of communica-
tion, but in fact her response may be masking the fact that she has
no idea what her grandmother meant. So, the ironical response that
supports her grandmother’s quip may be driven by her desire to avoid
knowing exactly what grandma meant, because B may not like what
A meant.
So, should we conclude that it is pointless to try to describe the
functions of humor? To begin with, we must consider two facts:

1. Each interpretation of the function(s) of the exchange is subjec-

tive and, in my opinion, this involves the interpretations of the
researcher, whether they are participants to the interaction or not.
2. Interpretations of the participants, especially if they agree, must be
prioritized over other interpretations.

So, if all goes well, i.e., all the participants in the overall situation
(speaker, hearer, and researcher) are in agreement, there is no problem.
But what if one of the participants intended an utterance as serious,
for example, but upon the reaction of their audience (say, laughter)
decides to “go along” and pretend he/she meant it as humorous all
along? There is no methodological prophylaxis against this. Consider
the following example from the show Top Gear in which James May
makes an unintended obscene pun, while talking about seating in a car

1 Just in case anyone is wondering, I am not denying the possibility of linguistic

communication. However, it would be too complicated to explain the differences between
bona-fide and non-bona-fide inferential processes in this context.

265
 DISCOURSE ANALYSIS

(120) 1 May: She’s effectively saying, “You’ve given me the baby, now
get in the back.”
2 Hammond: Yeah.
3 Clarkson: (Laughter; the audience joins in the laughter)
4 May: (pointing his finger) No.
6 (extended laughter from the audience and the show hosts)
7 Clarkson: That concludes the News
(https://www.youtube.com/watch?v=abX_qoYdtuw)
While the footage does not show May’s facial expression while Clarkson
starts laughing, it is clear from his denial in line 4 that he did not intend
the utterance as being an obscene double entendre. However, when the
camera returns to him, during 6, he is producing an open-mouth smile
having apparently accepted the reframing of the situation as humorous.
In this case, it is fairly clear what happened, but in other cases it
may not be clear and no methodology can prevent that, since even a
direct question to the speaker would be vitiated by the fact that whatever
motivation the speaker would have to reframe the intention post hoc
would also presumably lead them to deny that any reframing took place.
Having said this, in a sense the argument developed in Chapter 10 of
the negotiated nature of the keying for humor of a situation makes the
question moot: the initial intention of the speaker may be interesting but
it is not the end all of the analysis. If we take seriously the methodological
stance that all interactional keying is negotiated, then what the speaker
initially may have meant is not that significant anymore. Thus, while
retaining a healthy dose of skepticism toward any interpretation, and
acknowledging that the inner mental states of the participants may
not be available, even through interviews, since speakers may lie about
their motives, we can nonetheless move forward with considering the
functions of humor in conversation.
In the following sections we will review some of the foundational and
exemplar studies in the DA of humor. Tannen’s work is well known,
albeit surprisingly not utilized to its fullest potential in the field. Davies’
and Priego-Valverde’s are less known, but just as deserving of attention,
as they anticipated significantly several issues and results of the field.

12.1.1 Tannen’s Thanksgiving dinner

Tannen’s first book (Tannen, 1984), which is essentially a revised version
of her dissertation, contains a full chapter on the humor used by the

266
 12.1 FUNC TIONAL DA

participants to the Thanksgiving dinner she analyzes. Surprisingly, and

despite a very prominent mention in Attardo (1994), it has been largely
overlooked by scholars; for example, it does not appear in the references
of Glenn and Holt (2013); for an exception, see Norrick (1993).
This is all the more remarkable because Tannen’s contribution is quite
significant in several areas. First, Tannen was the first scholar to quantify
the number of humorous turns produced by each participant and pro-
vide a ratio of humorous turns to overall turns, which shows a range of
between 2% and 11% and a mean of roughly 7%. Second, Tannen intro-
duces the idea of individual conversational humor style,2 noting that the
participant with the lowest number of humorous turns had appeared to
be less involved in the conversation, whereas in terms of turns he was not
the one who had contributed less to the conversation. Hence Tannen’s
conclusion that “humor makes one’s presence felt” (1984, p. 165). The
idea of conversational humor style is applied to a family setting by Everts
(2003). Third, Tannen describes “extended [humor] routines” in which
speakers “pick up” on the humorous elements of the situation and carry
on extended humorous exchanges (pp. 176–177).
Tannen’s analyses are squarely in the Gumperz-Hymes approach we
have described in the previous chapters and focuses on the contextual
cues that allow the participants to “pick up” on the humorous intentions
of the other speakers. Another factor of interest is that Tannen did follow
up interviews with the participants which show that considerable mis-
understanding and lack of syntony existed at the time of the exchange:
for example, Tannen reports that one participant (David) believed that
another (Steve) was “truly angry” at Tannen’s placing her tape recorder
on the table, whereas, Tannen reports, Steve “was speechless” at the
suggestion that he was truly angry (p. 168).

12.1.2 Catherine Davies’ joint construction of humor

This article too, despite its very visible place of publication in the Pro-
ceedings of the Berkeley Linguistics Society, has been largely overlooked
by later researchers, despite its having introduced many of the themes of
the ethnomethodological approach to the discourse analysis of humor,
including the joint construction of humor, the function of solidarity

2 Not to be confused with the idea of humor styles in psychology (Martin et al., 2003).

267
 DISCOURSE ANALYSIS

enhancement, and the interactive ongoing definition of humor keying

(“footing”).
The following short passage from Davies’ article will illustrate the
nature of the joint construction of the humor. The situation is that of
faculty members who meet in the staff lounge and argue as to who
should pay for the coffee.
(121) Dyadic conversation among faculty members.
01 Ed: I’ll pay for it.
02 Joyce: No, I already got it.
03 Ed: You shouldn’t pay for my coffee.
04 Joyce: Oh, that’s OK. . . you’re worth every penny.
05 Ed: (laughs) I see your opinion of me has gone up.
06 Joyce: Not really. I’m coming back later to take fifteen
cents out again.
07 Both: (laugh) (Davies, 1984, p. 361)
In turn 4, Joyce humorously demeans Ed by using a back-handed
compliment: Ed is worth every penny of a cup of coffee, and since
coffee is cheap, then Ed is not worth much. Ed reacts in turn 5, first
by laughing, thus accepting the humorous framing of the situation and
also not taking offense, and elaborates on Joyce’s put down, by one-
upping her: if Joyce is saying that Ed is worth only 25 cents, Ed remarks
that the figure is an improvement over an imaginary previous figure
that was even lower, or, in other words, that Joyce’s opinion of him has
improved. Joyce responds by denying the improvement and claiming,
also fantastically, that she will later retrieve some of the money she has
currently paid (15 cents) thus bringing her evaluation of Ed down to a
figure of ten cents, which is coherent with Ed’s remark about a figure
lower than 25 cents.
This example nicely illustrates the collaborative, joint nature of the
construction of the humorous footing of the exchange, as Ed and Joyce
humorously spar in a fantasy environment spun off from the literal
interpretation of the cliché “you are worth every penny” applied to a
human being. The joint laughter in turn 7 sanctions the framing of the
exchange as humorous, but the whole exchange (turns 4–7) consists
of humor support (see Section 11.3.1.2). For another example of joint
construction of humor, see example (122) below.
Davies notes that non-verbal cues (contextualization cues) may be
used to signal the humorous footing/keying, but that also a deadpan

268
 12.1 FUNC TIONAL DA

delivery (“a straight face”) may be used (p. 361). Among other strategies
to jointly construct humor footing, Davies notes, building on what
others have said (p. 363), elaborating (p. 364), and empathizing (p. 366).
Davies distinguishes “a style oriented towards empathy more typical of
women, and a style oriented towards solidarity (through the use of a
version of ‘ritual insult’ joking) more typical of men” (Davies, 2017,
p. 476).
Davies interestingly also recognizes the presence of incongruity
(“bisociation”) in all of her examples of humor (p. 368–369) which
makes it one of the few articles to blend discourse analysis and
incongruity. Other exceptions include Norrick (2003), Antonopoulou &
Sifianou (2003), Archakis & Tsakona (2005, 2012), and Alvarado Ortega
(2013).

12.1.3 Priego-Valverde’s dialogic model

Priego-Valverde’s (2003) book on humor has many reasons to be fea-
tured prominently in any review of the field of functional discourse
analysis of humor (see Attardo, 2006, for a full list). Just to mention
two, she was one of the very first authors to discuss failed humor (see
Section 12.4.3), now a significant area of research and likewise she has
an extended discussion of the co-construction of humor (pp. 203–226).
However, another reason for its inclusion is the relative lack of impact
it has had on the field,3 no doubt due to its being written in French.
Hopefully, by summarizing some of its main findings, more scholars will
benefit from it.
Priego-Valverde analyzes a corpus of 33 hours and 20 minutes of con-
versations among friends, ranging from four to seven participants and
lasting between 2h15m to 7h30m. The age range is very homogeneous
(25–30 years) as is the socio-economic status (upper-middle class) and
education (most of the participants are professionals or students). It
is possible that the large size of her corpus (it was larger than the
entire Wellington Workplace Corpus, in 2001; see below) allowed her
to discern a wider variety of functions or phenomena, but whatever
the reason, Priego-Valverde’s is one of the most wide-ranging studies
of humor in conversation.

3 Exceptions to this neglect are the always well-informed N. Bell, W. Chafe, M. Haugh,
and A. Viana.

269
 DISCOURSE ANALYSIS

Priego-Valverde raises a few methodological significant points, often

overlooked in discussion about conversational humor: the first is that
laughter cannot be used to identify humor, due to the lack of correlation
between humor and laughter. Priego-Valverde is very clear about the
connection with failed humor, which would be excluded a priori from
analysis if laughter were used to identify humor (2003, p. 15). Her
answer is to rely on the intention of the speaker (p. 15), which may
be a controversial stance, but is alleviated by her role as participant-
observer (p. 63). However, Priego-Valverde takes the line of reasoning
further: she notes that it is a commonly assumed point that humor builds
intimacy. However, Priego-Valverde notes, some degree of connivence
(tacit assent) is necessary for humor. This presupposes an interlocutor,
who must share “a certain amount of knowledge, of values upon which
humor will rest” (p. 37). As Priego-Valverde notes: “it is this sharing
of a common referent, which both creates a connivence between the
interlocutors and can only exist if this connivence is already present”
(p. 37). In other words: there must be a minimal common ground
shared between the interlocutors (the speaker and the hearer) for the
humor to function. This implies that a third party observer may not be
able to assess the humorous nature of the text (p. 67). This common
ground exists also among people who do not know each other, such
as the audience of a theater play and its author, for example. The
common ground consists of the language, at a minimum, shared cultural
references, social values, and the like.
Her theoretical stance includes the idea that humor is the result of
multiple voices, real or imaginary, that coexist in the speaker’s utterance.
The source for this idea is Bakhtin (1981), mediated by Ducrot (1984).
Using Bakhtin’s “dialogism” is not uncommon in European scholar-
ship, for example it is also quoted in Kotthoff (2003). However, the
idea acquires a special relevance in the context of humor because, as
Priego-Valverde argues, the multiple voices in a humorous utterance
account for

• the distancing between the speaker and the content or the form of
his/her utterance, which goes hand in hand with the non-bona-fide
nature of the utterance;
• the incongruity between the serious and the humorous mode of
communication as well as the humorous incongruity within the text
itself (as in “incongruity and resolution”);

270
 12.1 FUNC TIONAL DA

• the ambivalence of the text: by introducing a duality of intention,

multiple interpretations can be justified within the text; and
• its playfulness: the distancing and the playful nature of deliberately
producing an ambiguous utterance, explain the goodwill (“bienveil-
lance”) of humor. (2003, p. 70)

Priego-Valverde finds that conversational humor exploits the same

linguistic mechanisms that have been analyzed in the semantics and
pragmatics of humor. Priego-Valverde presents a non-exhaustive list of
mechanisms, including script opposition (pp. 82–83), play on ambiguity
(p. 85), register humor (pp. 88–89), faulty logic (pp. 91–92), allusion
(pp. 95–98), intertextuality (pp. 99–100), insinuation (p. 98), and value
judgements (pp. 105–111). I will examine an example of the insinuation
mechanism, because it is short and exemplifies a sort of aggressive,
hedgy humor that is not found very often in studies about humor.
The conversation is between two couples of friends: F1 and M1 and
F2 and M2.4 The second couple is getting married (F2 and M2) and M1
is upset because F1 knew the details of where the wedding was going
to take place, whereas he did not. The reason for his indignation is that
he has long been friends with M2 and F2, whereas F1 only knows them
through him.
1 M1: wait seriously you did not tell me (long pause)
2 How did you know?
3 F2: [smiling voice] they call each other on the
phone\\
4 (general laughter)
5 M2: early enough that you’re not back from work
6 M1: you didn’t tell me anything, you know
7 F1: we have our times
8 (general laughter)

I have translated and will provide a few glosses for clarity: turn 1 is
addressed to F1; turn 2 is addressed to F1. Turns 3 and 5 are addressed
to M1.
Priego-Valverde glosses that F2 starts a game of “insinuations” which
then gets picked up by M2 and eventually by F1, to the general amuse-
ment. Essentially, F2 by informing M1 that F1 and M2 talk on the phone,

4 Priego-Valverde indicates by the initial the gender of the speaker and by a progressive
numeral different speakers.

271
 DISCOURSE ANALYSIS

triggers the script for lovers’ assignation (lovers need to plan their
rendez-vous). M2 plays along and provides a fictitious detail explaining
why M1 could have been unaware of the calls (they happened before
he was home from work) and finally F1 confirms that they have “set
times” to call each other, thus confirming indirectly that they do call each
other. Priego-Valverde comments on the hedgy, transgressive nature
of the insinuation (that F1 and M2 are having an affair, to which F2
would be privy and would condone) and also that by diverting to a com-
pletely implausible and false accusation the potentially face threatening
accusation that F1 neglected or forgot to tell M1 about the wedding
it defuses any potential conflict between the friends. (Priego-Valverde,
2003, pp. 98–99). I would like to point out another aspect, which Priego-
Valverde does not pursue, i.e., that F2, M2, and F1 “gang up” on M1,
despite his attempt, in his second remark, at redirecting the conversation
toward the serious issue of not having been told about the wedding. In
fact, it is F1’s “betrayal” of her loyalties that finally caps the exchange.
F1 does not “stand by her man.” The sexually charged nature of the
allusion and the aggressive nature of the humor (insinuating that one
is being cuckolded by one’s best friend) are quite unusual for academic
discussions of humor in conversation, but are representative of much
humor that occurs casually.
This example also lends itself to two other themes found in Priego-
Valverde’s discussion: humor is often used for negotiating face demands,
as in the example above, where the potential threat to M1 and F1 nega-
tive faces5 is resolved humorously, as seen above. The other topic is the
cooperative or competitive nature of the exchanges. Cooperation and
competition should not be understood in a Gricean sense (Chapter 8)
but rather as benevolent and aggressive, respectively (2003, p. 179).
In another terminology, the create an in-group and an out-group (see
below, Section 12.1.4).
Finally, let us mention that Priego-Valverde dedicates a significant
part of her book to the discussion of the functions of humor. She presents
several examples of communicative rituals that are performed through
humor. One example (pp. 122–123) concerns the common practice, at
least in some European cultures, to give guests a tour of the house on
5 Negative face is “the right to be left alone”: M1 has been humiliated by being in the dark
about the wedding and F1 presumably forgot to tell him and if confronted about it will be
embarrassed as well.

272
 12.1 FUNC TIONAL DA

their first visit. Other examples include lightening up the mood of a

conversation (pp. 130–131), averting conflict (pp. 132–136), and many
others. Her general conclusion is that humor can be both cooperative
(goodwill) and competitive, i.e., aggressive (p. 179).

12.1.4 Functions of humor in conversation

In Attardo (1994), based on work done in sociology, I classified the
functions of humor in three main categories:

1. Social management: this is the fundamental group affiliation/

disaffiliation mechanism of humor. By creating an in-group, those
that “laugh together,” i.e., share the experience of mirth or engage
in overt displays of it with the group, and an out-group, those that
do not participate in the mirth or are the target of the humor, the
production of humor automatically creates a bonding experience,
for the in-group participants which is, at least in part, based on the
exclusion of the out-group. This has been variously characterized,
with a very successful term in DA as “bonding” or “biting” (Boxer
& Cortès-Conde, 1997; cf. also Priego-Valverde, 2003, p. 179,
“cooperative” vs. “competitive” humor.)
2. Decommitment: this aspect of the humorous exchange is tied
to the “this is play” metamessage (see Section 3.3.1). Because
humorous messages are labeled as “play” the speaker is not fully
committed to the bona fide nature of the utterance (i.e., its truth,
relevance, etc.). This affords the opportunity of “taking back”
any of the implicatures and even inferences that would normally be
carried by said utterance in context. Likewise, the retractability of
humor allows speakers to bring up uncomfortable or taboo topics
(for example, death, in a hospital environment).
3. Defunctionalization: one of the central functions of language is to
convey meaning. In puns, but also in other ludic uses of language,
this function is partially or completely set aside. This is a deeper
kind of decommitment than what is produced by the charac-
terization of the message as play. Defunctionalization eliminates
the semantic-referential basic aspect of language. If someone asks
“when is a door not a door?” and provides the answer “when it’s
ajar,” decommitment would be saying “I was just kidding, I did not

273
 DISCOURSE ANALYSIS

really want to know.” Defunctionalization is the fact that no jars are

really intended at all (or doors, for that matter). All that matters is
the play on words and that the two strings sound alike (ajar vs. a
jar). A fuller treatment of these topics is found in Chapter 9.

These three basic functions find numerous applications in social

exchanges. We will now review some of them. An interesting aspect
of the various uses of humor is that often speakers react to the literal
message (the criticism in the tease, the said in irony). This is a sign of
the fact that the speakers “see through” the social function of the humor
and treat the message as serious.
12.1.4.1 Mediation
Because humor allows the speakers to claim that they did not mean what
they said or implied or at least to have said it with the purpose of amusing
their audience, humor may be used to broach difficult or touchy topics.
Likewise potentially embarrassing or aggressive topics may be couched
in humor to “soften” the sting. In this sense, humor is very close to some
forms of politeness (and in fact, humor is listed as a form of politeness,
see below).
12.1.4.2 Politeness
Brown and Levinson’s seminal work on politeness included humor as
one of the strategies of politeness. Essentially, by claiming common
ground, the joint performance of humor is a form of positive politeness
(Brown and Levinson, 1978/1987, pp. 103–104). Establishing jointly
membership to the in-group also boosts common ground. Zajdman
(1995) considers humorous Face Threatening Acts, i.e., a joking FTA,
as another politeness strategy. Norrick (1993) also characterizes humor
as positive politeness.
Holmes (2000), as part of the Language in the Workplace Project (see
below, Section 12.2.3, conceptualizes humor as both a form of politeness
serving both the positive face of the hearer (by expressing/enacting
solidarity) and also the positive face of the speaker (self-deprecating
humor). Holmes shows also that humor works for negative politeness as
well, by reducing FTAs. So, overall in Holmes’ application of politeness
theory to humor discourse analysis we have the following:

274
 12.1 FUNC TIONAL DA

• Humor as positive politeness

– humor addresses the hearer’s positive face needs (common
ground, in group solidarity)
– humor addresses the speaker’s positive face needs (self
deprecation)
• Humor as negative politeness
– humor attenuates FTA to the hearer’s negative face (e.g., hedging
a directive)
– humor attenuates FTA to the hearer’s positive face (e.g., hedging
a criticism) (Holmes, 2000, p. 167)

Holmes also shows that humor can be used as a way to negotiate power
in asymmetrical encounters: both from the perspective of the powerful
agent and of the subordinate one: “humor is also a very effective way
of ‘doing power’ less explicitly, and hence some uses of humor are most
illuminatingly analyzed as instances of coercive or repressive discourse”
(p. 165). Essentially, by joking the person in a position of power can
achieve their goals while on the surface not violating explicitly the
negative face of the subordinate(s): if the boss asks his/her secretary for a
cup of coffee while joking, he/she still expects and gets the cup of coffee.
Conversely, humor can be used “by the subordinate in an unequal power
relationship to subvert the overt power structure. Humor provides a
socially acceptable means of signaling lack of agreement, registering a
protest, or even a challenge to more powerful participants” (p. 165).
The research on politeness has seen a significant shift in focus and
considerable methodological discussion in the past two decades. This
is not the place to review, let alone address these, although see below
for mention of some alternative approaches to Brown and Levinson’s
used by scholars. Holmes herself, in a co-authored paper (Holmes &
Schnurr, 2005), moves away from the term politeness, which is “fraught
with problems” (p. 142), and uses instead the term “relational practice”
or “other-oriented behavior” (p. 142).
Holmes and Schnurr (2005) note that different ways of using humor
help constitute “communities of practice.” The following citation, is
worth examining in full, despite its length, for the subtle and nuanced
way it presents their conclusions. Holmes and Schnurr observe the
emergence of

275
 DISCOURSE ANALYSIS

preferred ways of using humor to construct and negotiate RP [Relational Practices]

in workplace interaction in different project teams. We are not, of course, claiming
that every team or every individual in any particular team uses a consistently similar
amount or type of humor throughout our recordings. Rather, what we have identified
is interesting indications of favored patterns of humor, both in relation to the amount
of humor and the kind of humor used in the different project teams we have studied.
(p. 146)

Thus, the amount of humor used in different workplaces varies signifi-

cantly, contributing to the establishment of different workplace cultures.
Likewise, the different number of supportive and contestive (aggressive)
instances of humor contributes to the differences among the commu-
nities of practice. Holmes and Schnurr conclude that some communi-
ties of practice end up having a more “feminine” (supportive) use of
humor and others a more “masculine” (contestive) (p. 154). Mullany
(2004) also applies the communities of practice approach and extends
Holmes’ work.
The following two examples will serve well to exemplify the support-
ive (122) and contestive (123) modes:
(122) Cla = Clara; [ . . . ] indicates overlaps
1 Cla: deferred means [put it on hold]
2 Tess: [mm ] deferred yeah
3 Cla: [(laughs)]
4 Tess: [means that you’re not going to]
5 think about it just now
6 Cla: we’ll do it some other time
7 (general laughter) . . .
8 Rob: deferred deferred means it
9 might get done one day if it’s lucky
(Holmes and Schnurr, 2005, p. 148; the notation was
adapted, SA)
Example (123) needs a little background information: the discussion
revolves around a file, sent by Callum to Barry. Callum failed to update
the header, which led Barry, the manager, to think he has sent the wrong
file. Note that Callum on line 7–8 mode adopts, but does not align with
Eric, but rather against him.
(123) Call = Callum; Barr = Barry
1 Call: I definitely sent you the
2 right one
276
 12.1 FUNC TIONAL DA

3 Barr: (laughs)
4 Eric: yep Callum did fail his office
5 management (laugh) word processing
6 lesson
7 Call: I find it really hard being
8 perfect at everything
(Holmes and Schnurr, 2005, p. 149; the notation was
adapted, SA)
While humor research has tended to be biased toward the positive
aspects of humor, humor scholars have recognized that humor may
have a negative tone. As we saw above, this would be characterized
as a “biting” aspect of humor. If looked through the lens of the social
management function of humor, it becomes immediately apparent that
while positive humor, humor that creates an in-group perception for the
speaker and the addressee(s), may be an effective strategy to mitigate
an FTA, negative humor that creates an out-group division between
the speaker and (some members of) the audience will be perceived as
impolite and aggressive. Since the 1990s, politeness theory has turned
toward an inclusion of impoliteness in the discussion, and thus we are
now fully equipped to explore the complex relationships between humor
and (im)politeness.
Kotthoff (1996) already anticipated the extension of humor beyond
positive politeness to negative politeness, but she also makes a fun-
damental point: humor “can be located at all points on a scale from
politeness to impoliteness” (p. 306). As she points out, not all affiliative
practices are polite. Kotthoff lists jocular aggression and “playing” with
politeness. She draws a profound alignment between Batesonian play as
mock-aggression and “humorous provocations” such as mock impolite-
ness and jocular mockery (p. 309). This discussion anticipates Haugh’s
work on Australian jocular mockery, briefly discussed below.
Haugh (2010) presents an alternative view of the connection between
humor and politeness, based on Arundale’s Face Constituting Theory
(Arundale, 1999) which allows him to see jocular mockery as “a non-
aligning response” (Haugh, 2010, p. 2111) which solves the “paradox’
(Tannen, 1986, pp. 94–95; Norrick, 1993, p. 75) of teasing displaying
both power and solidarity, being both affiliative and disaffiliative at
the same time, “simultaneously face threatening and face supportive”
(Haugh, 2010, p. 2111). This is not the place to discuss the definition

277
 DISCOURSE ANALYSIS

of “face,” but in a nutshell, Arundale’s Face Constituting Theory sees

“face” as emerging from the co-constructed interaction of the partic-
ipants, and not as something “owned” by an individual. In this sense,
since the humorous framing is also co-constructed (Davies, 1984) Face
Constituting Theory aligns well with humor studies.
Haugh sums up the analysis of jocular mockery in light of Face
Constituting Theory thus:

Within the context of FCT, then, jocular mockery is an action that is evaluated
as face threatening because it involves interpretings of a particular utterance/act as
mockery, but it is also evaluated as face supportive, since it involves interpretings of
the utterance/act as non-serious in that particular local context. (2010, p. 2112)

which is strikingly reminiscent of the function of “decommitment” of

humor (Attardo 1994, pp. 325–326) which is precisely based on the
simultaneously serious and non-serious stance of the speaker toward
his/her utterance.
12.1.4.3 Teasing
The literature on teasing is now remarkably large. One of the classical
studies is Drew (1987) who introduced as a term of art “po-faced”6
(serious) response and showed that the recipient of the tease may orient
to the humorous aspect of the tease, and so react with laughter, humor
support, etc., or to the critical aspect (po-faced).
Haugh (2017b) presents a synthesis of the research. Teasing may range
over the whole gamut of pro-social to anti-social behaviors (p. 205).
The various conceptualizations of teasing involve a serious/non-serious
dichotomy (the tease must be presented as non-serious) and an evalu-
ative aspect, both of the person teased and of the tease itself (p. 207).
Plainly, teasing is a social action, as it involves at least two participants.
Teasing may have many interactional functions, ranging from sanction-
ing (minor) transgressions, to romantic courting (p. 210). Culturally,
teasing and the reactions to teasing show significant variation, both in
what is acceptable to tease about. and in the ways people react to teases.
Haugh and Pillet-Shore (2017, p. 4) provide a useful three-part
sequencing for the act of teasing

6 The somewhat bizarre expression is related to the faces painted on chamber pots,
according to Drew.

278
 12.1 FUNC TIONAL DA

1. Teasable: a triggering action by the tease target that affords the

tease
2. Tease: a teasing action directed at the tease target
3. Affiliation: mutual ratification of the non-seriousness of the tease

As is easy to see, the non-seriousness of the tease, pointed out in the

third phase of the sequence, lies in its humorous nature.

12.1.4.4 Flirting
Glenn and Holt (2017, p. 301) summarize effectively the connection
between humor, impropriety, and flirting:

there is longstanding recognition that the use of improprieties and sexual references
can draw laughter and foster intimacy. Routinely in conversation, a speaker will say
or do something that might be considered a breach of tact or courtesy, perhaps of a
sexual nature. These improprieties can generate sequences of action. Similar to teasing
sequences, a range of recipient responses may occur, from overt disaffiliation to disat-
tention, affiliation, and even escalation (Jefferson, Sacks, & Schegloff, 1987). Laughter
represents a midpoint on this continuum: on its own, it recognizes the impropriety but
does not necessarily affiliate with it. More overtly, if the recipient laughs and produces
a next impropriety, that person is then co-implicated in the potentially offensive
mentality. When this happens, the participants display conversational intimacy: “That
is, the introduction of such talk can be seen as a display that speaker takes it that the
current interaction is one in which he may produce such talk, i.e., is informal/intimate”
(Jefferson, Sacks, & Schegloff, 1987, p. 160). Similarly, talk laced with sexual innuendo
may be treated as a particular kind of impropriety that ambiguously raises the possibility
of a romantic/sexual relationship between participants – that is, as flirtation.
(Glenn, 2003, pp. 131–141)

To paraphrase, if a speaker produces a turn that is “improper,” often sex-

ual in nature, acceptance of this “breach of courtesy,” by not censoring
it, signifies its approval and hence establishes de facto intimacy between
the speakers. The use of humor, in particular because of its decommit-
ment function, which allows one to back off from full responsibility for
what one has said, by claiming to not have been serious (“I was just
kidding!”), facilitates the “testing of the waters” of the flirtation process.
A good example of this is found in the strategies used by customers
trying to pick up waitresses in a diner: they would couch their offer
to go out as a joke, which allowed them, if they were turned down, to
claim they had not been serious and thus to save face, to some extent
(Walle, 1976).

279
 DISCOURSE ANALYSIS

12.1.4.5 Multifunctionality of humor

Summing up your discussion of some of the functions of humor, one of
the clearest and best documented conclusions of humor studies is that,
ultimately, humor is multifunctional (i.e., many goals may be achieved
by interactants who produce humor at the same time). Priego-Valverde
(2003) comes to the conclusion that humor may be “everything and
its contrary” (p. 233), i.e., both aggressive and benevolent, face saving
and face-threatening, “virulent” and “convivial.” There are no limits to
what functions humor may achieve in conversation, beyond those of
contextual appropriateness. One cannot sentence a convicted criminal
to death humorously because the context prevents the use of humor,
not because one couldn’t in principle do so. So, we can come to a
final generalization about humor, and namely that any function that
can be achieved communicatively can be achieved humorously, modulo
contextual appropriateness.7

12.2 Conversational humor in various settings

Given that humor is multifunctional and indetermined, a natural way of

organizing the research is by setting or situation in which the exchanges
occur (i.e., Hymes’ Setting in the “SPEAKING” acronym, discussed in
Section 10.2).

12.2.1 Conversation among friends

Within discourse analysis, there is a long tradition of using conversa-
tions among friends as the source of data, so much so that, at least as
far as humor goes, this is one of the most common categories. Tannen
(1984), Davies (1984), Norrick (1993), Priego-Valverde (2003), Kotthoff
(1996, 2007) are all examples of studies of conversations among friends.

7 This rule cuts both ways: when a situation is socially coded for seriousness, humor is
not acceptable; when a situation is coded for humor, seriousness is not acceptable. It would
be very difficult for a stand-up comedian, in the middle of delivering his/her act, to pivot to
a serious discussion. Comedians may intercalate isolated serious comments, but the point
of a comedy act is to make people laugh.

280
 12.2 CONVERSATIONAL HUMOR IN VARIOUS SET TINGS

12.2.2 Medical
In this section, we are concerned with the uses of humor and non-
mirthful laughter in interactions in a medical setting (or with a medical
topic). There exists a vast literature on the effects of humor on health
(see Martin, 2007) but we will not address this aspect of humor research,
which is too far afield from the perspective of the linguistics of humor.
Coser (1959) and Emerson (1969) are widely credited as the first
studies on the use of humor in a medical setting. Coser (1959, 1962)
shows that humor is used to reassure patients and relieve stress. Emerson
(1969) shows how patients and medical staff use humor to negotiate
sensitive (taboo) topics, such as death. Her focus is on the negotiations
for admitting or rejecting the taboo topic.
West (1984) studies the power imbalance of the medical service situa-
tion. She found that patients laugh more than doctors but, in percentage,
doctor’s laughter is reciprocated more often. Ragan examines laughter
and humor in gynecological exams, a situation which presents “special
threats” to the face of both the patient and the medical personnel,
“because of the parts of the body to be examined and the positioning
of the body necessary for the examination” (1990, p. 68). The exam
is “uniquely anxiety-provoking to both participants” (p. 69). Therefore
Ragan conceptualizes the entire exam as a FTA (p. 70). Unsurprisingly
then humor is seen as a way to redress the FTA, by finding a joint
stance between the patient and the doctor away from the task at hand: in
one example, the doctor asks how the weather is like outside and upon
being told it’s nice, engages in a playful interaction with the patient by
noting that he/she hasn’t been outside all day and that the patient could
have lied and told him the weather was unpleasant. The patient obliges
and playfully lies and the doctor thanks her in an exaggerated manner
(pp. 75–76). Needless to say, after a short pause, the doctor returns to the
serious mode of the medical exam. Pizzini (1991) finds that status is a
very significant factor, and that very few instances of humor are directed
upward (i.e., toward the physicians). Generally speaking, she found that
the higher status participants (doctors) produced most of the humor.
Nurses joke more when doctors are not present. The difference vis-à-vis
other studies may be explained by the fact that Pizzini is considering
Italian doctors and Italian culture is much more overtly hierarchical
than the American culture.

281
 DISCOURSE ANALYSIS

Mallett and A’hern (1996) find that humor is used to highlight

anxieties, avoid conflict, and that different patients use humor
differently. Some humor occurs randomly as accidents happen, such
as a nurse spilling rinsing fluid on the floor (p. 540).
Marci et al. (2004) also show that patients laugh much more than
therapists: more than 70% of laughter is produced by patients and
patients initiate laughter much more frequently than the therapists
do. However, some studies show that patients use laughter to resist
therapists’ characterizations (Dionigi & Canestrari, 2018) or doctor’s
questioning (Ticca, 2013). Rees and Monrouxe (2010) consider triadic
conversations involving a doctor, a patient, and a student, in a teaching
hospital. They report that most laugher occurs around teases and that
the mode is competitive (disaffiliative) rather than cooperative, as tends
to be the case in other studies.
In a series of studies on Finnish doctor–patient encounters, Haakana
(2001, 2002, 2010) shows that patients laugh mostly alone at delicate
or embarrassing topics (such as financial matters). Zayts and Schnurr
(2011, 2017) examine nurse–patient interactions in prenatal counseling.
They find that nurses laugh more than the patients and that the nurses’
laughter is responded to more. This results goes against Haakana’s
finding that doctors laugh less. However, Zayts and Schnurr note that
this may be just a factor of the different situations (in doctor–patient
interactions, patients talk more, whereas in prenatal counseling, the
nurses do most of the talking). The nurses’ laughter is used to “reassure
and [. . .] put [. . .] at ease” the patients (2017, p. 126). For example, in
one case laughter follows the statement that the mother is worried.
Chimbwete-Phiri & Schnurr (2017) study AIDS/HIV counseling ses-
sions for pregnant women in Malawi. Humor is used to establish solidar-
ity and to make the attendees feel comfortable, to facilitate the discussion
of sensitive topics, and as a means to criticize the participants who
are not engaged. More broadly, we have uses of humor as a coping
mechanism by those facing devastating diseases, such as cancer (e.g.,
Beach & Prickett, 2016; Demjén, 2016; Semino & Demjén, 2017) or
more generally (Åstedt-Kurki & Isola, 2001).
Overal, the picture that emerges from these studies (and many others
that cannot be reviewed due to space limitations) is that humor is used
to cope with a stressful job and inherently difficult, stressful topics and
situations. The medical profession is very hierarchical and the power
differential is felt in the use of humor, but humor can be used to carve out

282
 12.2 CONVERSATIONAL HUMOR IN VARIOUS SET TINGS

some spaces for the patients to express their feelings and attitudes. See
for example Schnurr and Rowe’s (2008) analysis of subversive8 humor
in emails.

12.2.3 Workplace
This section is primarily dedicated to presenting Janet Holmes’ and her
associates’ Language in the Workplace Project, due to its significance
and centrality in the literature. The project begun in 1996 at Victoria
University of Wellington, in New Zealand. The corpus consists of about
1500 interactions, produced by 450 different people, in 20 different
workplaces, and consists of about 150 hours of recordings.9 As Holmes
(2000) notes, most of the research on humor in the workplace focuses
on the benefits that humor provides to management (e.g., increased
productivity), the employees (e.g., increased well-being), or both. This
aspect will not concern us presently.
We saw above, in Section 12.1.4.2, that humor can be used to enact
politeness or relational practices in the workplace. We also saw that
humor can be used both as a tool to control the subordinates, by catering
to their face while still imposing the superior’s will and conversely, it can
be used by the subordinates to subtly undermine their superior’s stance,
for example by criticizing them.
Two main theoretical lenses through which the Language in the
Workplace Project sees this setting are:
The social construction of identity (on social construction of humor,
see Section 13.3). Essentially, workers construct an identity at work,
through their interactions. Language and humor play a significant part
in the identity they project. Holmes and Marra (2002, p. 380) show that
subjects build multiple identities at work.

8 I have always been bothered by the use of the word “subversive” in this sort of context:
to me subversion is not the same thing as criticism: one can make fun of one’s boss without
subverting the hierarchical organization of the company; one can criticize the government’s
policy on taxation without being a subversive. In order to be a subversive you have to actively
undermine the authority of the boss/government, for example by mounting a campaign to
get the boss fired or by raising a militia and fighting the king, or some such activity likely
to get you fired or sentenced to death. Nonetheless, I too bow to common usage and use
subversive as a synonym of “mild criticism.”
9 The web site of the project is particularly useful and can be accessed at https://www.
victoria.ac.nz/lals/centres-and-institutes/language-in-the-workplace.

283
 DISCOURSE ANALYSIS

The Communities of Practice approach. As the term itself states,

communities of practice coalesce around ways of doing things (prac-
tices): “The development of a negotiated linguistic repertoire is perhaps
the most crucial aspect of a community of practice from a sociolin-
guistic point of view” (Schnurr, 2009, p. 14). Specifically, members of
a community develop ways of “doing humor” that are particular to their
community and that help define it as such (Holmes & Marra, 2002).
Ultimately, each workplace may develop a distinct culture (Schnurr,
2009). This idea could be usefully compared to the family humor styles
(Everts, 2003) seen above.
Power is clearly a major factor, for example Holmes notes that “the
chair [of a meeting] makes a disproportionately high contribution to the
humour in most meetings” (Holmes et al., 2001, p. 96) This is of course
consonant with the findings from other workplace settings that show
that the power structure of the workplace is reflected in the discursive
practices that occur within it. However, Holmes notes also that “in work
contexts humour can be used by subordinates as a subtle (or not so
subtle) license to challenge the power structure, as well as by those in
power to achieve the speaker’s goal while apparently de-emphasizing
the power differential” (2000, 176). As we saw already above, in Section
12.1.4.2, humor has a Janus-like nature, being both supportive (i.e.,
affiliative) or “contestive” (i.e., disaffiliative), it can foster collegiality or
assert power. This “subversive” aspect coexists with positive politeness
(e.g., Holmes & Marra, 2002).
Schnurr (2009) is focused on “leadership” (a polite term for “man-
agement”) and in particular their use of humor to both achieve “trans-
actional goals” (getting the job done) and “relational” goals (making
people feel good). The following is an example of interaction in which
humor is used to achieve the purposes of the workplace (I significantly
shortened the example; for a full transcription and analysis, see Schnurr,
2009, pp. 28–29):
(124) Donald = CEO of the company; Ann = project manager; nota-
tion edited for clarity
Donald: yep+okay alright
do you wanna write up a letter of offer
Ann: no [laughter]

284
 12.2 CONVERSATIONAL HUMOR IN VARIOUS SET TINGS

Donald: [laughter]
are you the project manager
[laughter]
Ann: [how do I] do that
Donald: eh? (laughs)
there’s standard templates
In example (124) Donald achieves the goal of instructing his project
manager, who was relatively new to the position, without threatening
her face by couching his “order” of writing the letter of offer for a new
hire in a non-bona-fide question (he knows perfectly well that Ann is
the project manager). By asking a self-evident question, he can both
express the intention to be humorous (consider that Donald could have
said “You have to write the letter” so his turn is clearly facetious) and
invite a separate inference that since project managers write letters of
offer and Ann is the project manager, it follows she should write the
letter. Indeed, Ann, after acknowledging the humorous nature of the
turn by laughing, takes the inference exactly as an order and asks how
to perform the task she has been assigned. It should be noted that Ann
herself had initially framed the situation as humorous by uttering a flat
denial to the indirect (polite) request by Donald to write the letter, in her
first turn. Schnurr notes that the example illustrates both transactional
and relational goals: “By convincing Ann to write the letter of offer
Donald achieves his transactional objectives; and by [ . . . ] expressing
a request as non-threatening[ly] as possible, he also takes into account
[the] relational aspects” (2009, p. 29).
Other contributions of the Language in the Workplace project include
an emphasis on quantifiable results (see Section 12.3 below); another,
perhaps the most surprising, result to emerge from the project is that
women produce more humor than men. This will be taken up in more
detail in Section 13.2.1.
Needless to say, the settings we have discussed have been picked
almost at random. Another setting we will examine is the classroom,
which will be discussed in more detail in Chapter 16. Many other
workplace settings have been explored, such as for example, humor in
a professional kitchen (Lynch, 2009, 2010), or factory work (Collinson,
1988), etc. but often in a sociological context.

285
 DISCOURSE ANALYSIS

12.3 Corpus-based discourse analysis

While corpus-based10 discourse analysis has produced a smaller

number of articles and books, the very significant results that they have
produced promise to change the way we look at humor in discourse,
and have already done so in a few cases, but also raise new questions
and issues.
Corpus-based discourse analysis is distinguished by the use of large
corpora (unlike the smaller corpora, usually collected personally by
the researcher, often as a participant observer, which characterize con-
versation analysis) and the use of statistical analyses and descriptions.
Needless to say, humor scholars had used corpora of jokes and puns
for decades, but here we are referring to discursive data (mostly con-
versations). When looking at corpora of discursive data which have
not been assembled by humor scholars, we are faced with the problem
of identifying the humor. There are essentially two ways to go about
it: one is to study what data are readily available, so for example, if a
corpus is marked for laughter, to study laughter and not humor. The
other is to categorize, manually, all instances of humor for the aspects
one is interested in. Both approaches have been followed in the literature
(Partington, 2017, p. 323).
Another point to remember is that most of the studies below find
significant variability within their corpora. For example, Holmes et al.
(2001) report a range of zero to 121 instances of humor per conversation.
Nesi (2012) reports that 24 out of 160 university lectures contained no
laughter. Lee (2006) found that in the MICASE corpus, 8 out of 152
“events” contained no laughter. He notes that the bottom 20, in terms
of scarcity of laughter, were all lectures, but that 3 lectures made it into
the top 20.11 Günther found that more than half of the examples of jokes
in her corpus came from one speaker (2003, p. 95). So, even when data
come from the same activity type and the same setting (Hymes, 1972; see
Section 10.2) there is nonetheless significant variation, due to individual
preferences, for example.
10 Corpus linguistics distinguishes between corpus-based, corpus-driven, and corpus-
assisted methodologies. Corpus-based studies test linguistic hypotheses on a corpus,
corpus-driven studies are inductive and the categories emerge from the study of the
corpus (Biber, 2009). Corpus-assisted studies blend corpus analysis and discourse analysis
(Partington, 2017). I use corpus-based as an umbrella term in this context.
11 More discussion of humor in lectures can be found in Chapter 16.

286
 12.3 CORPUS-BASED DISCOURSE ANALYSIS

Holmes et al. (2001) is one of the first large-scale corpus-based

studies that considers the gender variable. In 2001, it was based on
22 workplace meetings spanning 1342 minutes (more than 22 hours)
of interaction and involving 157 participants, 70 females and 87 males.
An exact breakdown of each meeting’s participants, duration, etc. is
provided in Holmes et al. (2001, p. 87). The size of the corpus is such
that it is probably representative of the population it describes. Within
the corpus, 396 instances of humor occurred. Failed humor was not
considered. Sustained humor sequences are counted as one instance
of humor.
A very significant point is driven home by Holmes et al. (2001): there
is a great degree of variability in the ratio of instances of humor per
minute, with range from a low of zero to a high of 1.21 (121 instance of
humor in 100 minutes; this was a shorter meeting) and likewise there
is great variability in the number of collaborative (sustained) humor
exchanges. Thus any generalizations should be extremely careful, as
there are significant differences even within activity types and across
activity types: conversations among friends have about 10 times the
amount of humor of workplace meetings (Holmes et al., 2001, p. 93).
Günther (2003) considers humor, laughter and joke performances in
conversations from two British corpora: the Conversational Corpus of
the British National Corpus (BNC) and The Bergen corpus of London
teenage language (COLT). Günther then characterizes all instances of
humor in the corpus according to several categories. It is impossible to
summarize all her results in this context, so I will focus on the most
interesting ones. Contrary to expectations, she finds that men do not
tell more canned jokes than women (p. 59); the two groups tell about
the same number of canned jokes. Younger people told more canned
jokes (p. 63), which are however relatively rare in the corpus. It would
be interesting to compare this datum to contemporary corpora to see if
this is still the case, since tastes in humor have shifted away from canned
jokes lately.
In her analysis of the corpora, females speak more and laugh more
than males. Both results are statistically significant (p. 129). Age-wise
those in the 25–34 group laugh the most, whereas the younger group
(0–14) laugh the least. Laughter and humor tend to occur more
frequently in single-sex settings, especially among women; “multi-
party, mixed-sex conversations trigger fewer laughter and humorous

287
 DISCOURSE ANALYSIS

contributions” (p. 181). However, her sample ignores turns that consist
only of laughter, which may skew the data. Günther also finds no support
for the hypothesis that women are more supportive of men’s humor
(p. 200) but finds that men prefer “fantasy” humor (p. 209)
Partington is probably the name most associated with corpus-based
linguistics if humor. In Partington (2006), he circumscribes the anal-
ysis to laughter, or “laughter-talk,” i.e., “talk preceding and provoking,
intentionally or otherwise, a bout of laughter” (Partington, 2006 , p. 1).
He considers White House press briefings, which come pre-annotated
for laughter. In Partington (2007), he examines irony that is explicitly
labeled as such in the discourse in which it occurs. In Partington (2011)
he considers “phrasal irony,” i.e., irony produced by combining phrasal
expressions with contrasting evaluative polarity, as in “bent on self-
improvement” where “bent” has a negative connotation (e.g., “bent on
destruction”), whereas self-improvement has a positive valence.
Koester (2010) considers also workplace conversations. She gathers
as a corpus of 60 humorous exchanges from the Corpus of American
and British Office Talk and analyzes it according to five functions:
building a positive identity; defending one’s positive face; showing con-
vergence; showing divergence; and negative politeness. She finds no
differences between American and British speakers in terms of quan-
tity of humor produced (p. 115). She finds that men produce more
teasing humor and women more self-deprecating humor. Koester also
reports that managers and subordinates initiate humor in comparable
numbers.
The most significant trend in the discourse analysis of humor to
emerge in the past 10 years has been a move toward the use of corpora
for the gathering of large bodies of data. In a sense, even from its origins
in conversation analysis, discourse analysis always relied on a corpus,
except that the corpus was very small, and possibly consisted on a
single instance of the phenomenon. So, the emergence of studies such
as Günther (2003), which mined for data (i.e., humorous interactions)
two corpora assembled completely independently of the intention to
study humor, leads directly to Haugh (2014, 2016, 2017c) which are
based on a number of corpora of Australian English, which have been
scanned to find instances of jocular mockery, which are then analyzed
in the papers. Likewise, Demjén (2016) uses corpus analysis to extract
humorous instances to analyze.

288
 12.4 SOME ISSUES IN THE DA OF HUMOR

12.4 Some issues in the DA of humor

12.4.1 Establishing the humorous intention

Given the multifunctional and indeterminate nature of humor, a rel-
evant question is how do the speakers establish the intention to be
funny. Clearly, the speakers do not have access directly to each others’
intentions, but this is not a problem unique to humor. When I walk into
a coffee shop and ask for a cup of coffee, the coffee shop clerk does not
have access to my intentionality either; however, it is an easy inference
from my utterance of “may I have a cup of coffee?” that I intend to
have coffee (assuming of course rational bona-fide behavior, i.e., I do not
secretly want tea or plan to rob the store). So, in this sense the ascription
of humorous intention is not different from the problem of ascribing
intention to speakers in general.
One strategy is to signal (intentionally or not) one’s intention by
exhibiting one of several contextualization cues typical of humor, such
as laughter, increased smiling intensity, smiling voice, marked prosody,
etc. This is the kind of cue that has been studied in conversation and
discourse analysis. The problem with the signaling strategy is that it
is itself ambiguous: one may be laughing at someone (i.e., one may be
mocking them), or one may smile contemptuously, or just politely, and
one’s marked prosody may have other reasons (for example, emotional).
Thus another strategy is to explicitly state that one means one’s utter-
ance as humorous. There are a number of conventionalized such expres-
sions, the most famous being “just kidding” (Skalicky et al., 2015;
Haugh, 2016). Finnish apparently uses a reduplicated expression which
translates as “joke joke.” Another expression with the same function that
is now used in colloquial American English is “I am just saying” often
shortened to “just sayin’ ” (Margolies, 2016). “Just sayin’ ” is broader
than “just kidding” in that it denies accountability for any sort of
potentially antagonistic speech, not just humor. Haugh (2016, p. 127,
2017) examines the case of “jocular pretense” (“a form of teasing accom-
plished through the tease being delivered without any cues to non-
seriousness”) in which the deadpan delivery pre-empts the use of any of
the usual markers of humor (smiling, etc.) and thus the occurrence of
the “just kidding/joking” is the only clue to the humorous intention
of the speaker. The functions of “just kidding” include wrapping up a
humorous exchange, and thus returning to serious talk, revealing that a

289
 DISCOURSE ANALYSIS

sequence was in fact humorous, if for example, humor fails, disclaiming

the intention to offend or being “improper” (“inoculation,” Skalicky
et al., 2015). and expressing disapproval of a non-humorous response
(“po-faced,” Drew, 1987). The disclaimer of offense and the disapproval
functions go back to the retractability function of humor (Attardo,
1994). The inoculation function, or disclaimer, is the most frequently
found in a corpus of 1200 instances of the expressions (Skalicky et al.,
2015, p. 23). Returning to serious discourse was the function of only 8%
of the instances of “just kidding.” Skalicky et al. (2015) note that this use
of “just kidding” is similar to the use of “no” following a joke, to return
to serious discourse, discussed by Schegloff (2001).
Let us return to the so-called “deadpan” strategy of delivery of humor,
which is the exact opposite of the signaling strategy, i.e., it consists
of not producing any signal of one’s humorous intention. This should
not be confused with the “blank face” (Attardo et al., 2003) which is
a differential signal (i.e., producing an expressionless facial expression
when one would expect some display of emotion). Deadpan delivery is
by definition devoid of any markings. The idea here is that contextual
pressure is so strong that even if one does not indicate one’s non-
serious intention, the audience will be able to extrapolate the correct
implicature. Consider for example if I were to utter (125):
(125) I cannot believe anyone would waste their time writing a book
on humor.
anyone who encounters (125) in the above context will be aware of
the fact that I have just done precisely that and, reasoning that one
generally does not engage in activities one considers to be a waste of
time, will work out the implicature that I do not mean what I just said
and that therefore presumably the utterance is ironical. At this point my
hypothetical reader-cum-interlocutor would still have to figure out why
I am saying ironically (125), but we may safely abandon the example
here.
The issue of signaling or marking the humorous intention is far from
trivial, because since there is a general agreement that “humorous and
serious discourse operate according to fundamentally different princi-
ples” (Mulkay 1988, p. 7), it becomes important to be able to tell in which
mode of communication we find ourselves.
There is also a generally accepted idea that humor must be marked. In
its most naive form, it assumes that speakers signal the humorous nature

290
 12.4 SOME ISSUES IN THE DA OF HUMOR

of the turns they intend to be humorous by using some paralinguistic

feature, such as a special tone of voice (“some changes in the speaker’s
normal voice pattern,” Mulkay, 1984, p. 47) or gestures, such as winking,
etc.12 The hearers, conversely, would react with laughter, thus validating
the humor and signaling the understanding thereof. Mulkay stated very
clearly the latter aspect “each humorous contribution has to be marked
by recipients in a rather special way: that is by laughter, smiling, or some
related token of appreciation” (Mulkay 1988, p. 46). Mulkay continues
that “If this is not done continually, humorous interaction lapses and
participants return to the serious mode” (p. 46). See also Fine’s (1984,
p. 85) point that if silence or a serious remark follow a humorous turn,
“the humorous remark has failed.”
Unfortunately, the idea that humor keying must be signaled “contin-
uously” is incorrect, as has been shown by several empirical studies,
reviewed in Section 10.3. Mulkay and Fine are correct that eventually
the humorous keying/framing lapses, but its “half-life,” to borrow a
metaphor from nuclear physics, is longer than they think. The point is
not that speakers never mark their departure from the serious mode,
but rather that the “marking” is not compulsory and in fact much less
frequent than generally assumed. It is definitely not the case that, as
Fine maintains: “jocular remarks are typically accompanied by par-
averbal and nonverbal cues which suggest that the remark should not
be accepted on face value” (Fine 1984, p. 91). At most, there is a
correlation between some paraverbal behaviors and humor but there
are no consistent “cues” of humorous intent. Furthermore, as seen in
Section 10.3, speakers generally produce much less obvious cues to their
humorous intention than exaggerated prosody. The one reliable pattern
that emerged was an increase in the intensity of smiling. Note that the
increase is not of the frequency, but in the intensity of the smiling: people
do not necessarily smile more, they smile more broadly.

12.4.2 How do speakers identify humor?

Glenn and Holt (2017, p. 298) address the issue of identifying humor, or
rather, the referent of laughables. They summarize Ford and Fox’s (2010)
discussion of the “interrelated semiotic resources” for constructing
12 Kotthoff (1996, p. 311), “Laughter particles, atypical lexies [=words], cliche [sic] prosody
and such formulae function as indicators of the contextualization of a humorous modality.”
(Emphasis in the original.)

291
 DISCOURSE ANALYSIS

laughables, including 1) various phonetic “practices,” a) “long and loud

aspiration after a word-initial voiceless stop, lengthening of fricatives,
high pitch, laryngeal constriction, voicelessness at the beginning of
vowels, breathiness, and modulations of loudness” (Glenn & Holt,
2017, p. 298); b) the smiling voice; 2) “nonvocal resources” such as
“lip-spreading (interpreted as smiling), leaning and/or throwing back
the head, covering the face with hands, lowering and raising shoulders,
shaking torso and shoulders, clapping hands or slapping a surface, and
tensing neck and facial muscles” (p. 298); and 3) Glen and Holt even list
“incongruity.” However, they argue that “The presence of one or more
of these attributes – laughter, vocal features, visual features, and/or
incongruity – does not necessarily mean that a turn will be treated as
laughable” (p. 300) and finally conclude that “orienting to a contribution
as laughable is dependent not only aspects of the turn itself but on the
trajectory of subsequent contributions.”
This nuanced position is a far cry from the use of laughter as the
sole identifier of humor (as we saw in Section 11.2) or worse, canned
laughter. Incidentally, the use of canned laughter is not a straw man:
Pelsmaekers & Van Besien (2002, p. 247) is an example of using the
canned laughter of the soundtrack of a sitcom to identify humor (irony,
in this case). There is nothing wrong with the list in Glenn and Holt
(2017). However, there are three issues with the idea that one can
identify humor (or laughables associated with humor) “externally,” so
to speak. The first issue is that there is mounting evidence, reviewed
in Section 10.3, that speakers do not necessarily mark or signal their
humorous intention or that the markings are relational and not substan-
tive, so that there is no ironical tone of voice per se or any special volume
or pitch associated with humor. Rather, it is the incongruity between the
expected, unmarked delivery and the unexpected, marked delivery that
signals not the humor, but the presence of a notable phenomenon, which
then the speaker will infer is humor. Therefore any attempt at producing
a list of markers of humor is a losing battle, as any linguistic feature
may, under the right circumstances, be used to cue the presence of the
humorous intention. The second issue is that by definition the deadpan
delivery of humor, or the delivery in contextually poor environments,
such as writing, are without these markers. The third issue is failed
humor, which we will review below, in Section 12.4.3.
Archakis and Tsakona noted that “the combination of incongruity
and laughter is a relatively safe criterion for identifying humor”

292
 12.4 SOME ISSUES IN THE DA OF HUMOR

(2012, pp. 77–78). In Attardo (2012) I presented a triangulation

method for the identification of humor (see also Section 2.5), which
consists of taking opportunistically advantage of all available textual and
paratextual cues and of a full semantic and pragmatic analysis, using the
General Theory of Verbal Humor. A preponderance of evidence is then
used to adjudicate questionable cases. Thus cues may include, but are
not limited to:

• intonational and prosodic cues

• laughter, smiling voice, and other expressions of mirth
• metalinguistic comments (such as “I was joking”)
• facial expressions and other gestural markers
• paralinguistic cues (e.g., a text being included in a collection of jokes
or a meme being included in a web page of funny memes)
• semantic and pragmatic analysis, as per the General Theory of
Verbal Humor

Burgers et al. (2011) present a method to identify irony, the Verbal

Irony Procedure (VIP). VIP is too complex to describe here, but it
essentially consists of a full-fledged semantic and pragmatic analysis
of the text, looking for the traditional components of irony (lack of
congruence between the literal meaning and the implied meaning,
evaluative scale, etc.). The analysis is to be performed by trained coders.
VIP has a lot to recommend it, for example part of the protocol is to
consider the entire text, a point made forcefully in Attardo (2017b)
under the warning against cherry-picking of data (see Section 7.2.3). On
the other hand, it seems to rely exclusively on internal (textual) evidence
and to pay insufficient attention to the external evidence listed above.

12.4.3 Failed humor

The repertoire of humor responses has largely focused on humor sup-
port (Hay, 2001) and rightfully so, given the groundbreaking nature
of Hay’s contribution. It would not be until Bell (2009, 2015) that
a full treatment of the negative side of the responses to humor, the
unsupportive reaction, or “unlaughter,” as Billig (2005, p. 192–194)
calls it, would receive a full treatment. Bell sees failed humor as a
form of miscommunication. Her definition of failed humor includes the
intention to elicit mirth or amusement (2015, pp. 3–4) but she notes that
failure may occur due to “any type of communicative disruption” (p. 4).
293
 DISCOURSE ANALYSIS

It would be impossible to do justice to Bell’s book with the limited space

within this context, so I will content myself with noting that Bell’s mixed
method approach which blends discourse analysis and elicitation13 was
particularly successful in eliciting a variety of reactions and behaviors
that are not normally accessible to the DA scholar (or may be so only
under accidental circumstances). For example, one of the reactions to
the telling of a particularly bad joke was the hearer bursting into tears
(p. 2) because she felt “inferior and humiliated” at not getting the joke.
Another reaction was punching the speaker in the arm (playfully). This
alone tells us that we need to broaden significantly any idea of repertoire
of reactions to humor we may have held. Bell lists (p. 121) the types
of responses to failed humor, which range broadly from non-verbal
responses (rolling of the eyes, grimacing, etc.) to flat out admissions of
not understanding the joke (a surprisingly frequent reaction), complete
silence (also surprisingly frequent), to requests for more time to think
about the joke and providing their own interpretation of the joke (Bell
does not say so, but these may be cases of “mansplaining,” i.e., the
obnoxious providing of unrequested explanations by a man, who often
has less expertise in the subject than the woman he is addressing).
Generally speaking, what strikes me as quite notable in the repertoire
of failure managements is how willing to aggressively deny the humor-
ousness of the would-be joke the audience is. Bell reports the following
example: a husband and wife (probably Bell herself, given the humorous
footnote) are discussing the shopping list. The wife suggests that the
husband purchase some butter.
(126) (adapted from Bell, 2015, p. 108)
Husband: do we really need it?
Wife: yeah. we’re in butter desperation (.)
butter desperation, get it?
Husband: yeah not funny at all
Even factoring in intimacy and possibly the playful nature of the hus-
band’s comment, it is a far cry from hedging and politeness.
Priego-Valverde (2003) dedicates a significant discussion of failed
humor, with several types of failures examined. An example is a failure

13 The complete set of data is listed on pp. 38–39: direct observations from the researcher,
self-reports from other parties, published data, corpora, television and film discourse, and
elicitation.

294
 12.4 SOME ISSUES IN THE DA OF HUMOR

of the humor because the targeted interlocutor fails to acknowledge

it because they want to revert to the serious mode of communication
(pp. 184–188). Another example shows one speaker left speechless by
a mock callous remark and reacting with a surprised expression, while
others laugh. In another example, one of the speakers simply ignores
the humorous side sequence and continues his serious discussion of
his army experience (p. 194). In yet another, two friends gang up on a
third one mocking her humorously for using the wrong word while the
latter friend changes the topic to a hairdresser’s appointment (p. 198).
Priego-Valverde notes that in the conflict between taking a serious or
a humorous direction, the serious mode always wins out (p. 202), but
there seems to be no reason to believe that it is impossible to have the
opposite happen (and I am not saying that Priego-Valverde believes it
either).
CA and DA have been long reluctant to engage with failed humor:
see for example, Holmes’ (2000, p. 163) definition of humor which
deliberately leaves out failed humor, discussed in Attardo (2015). How-
ever, Mullany (2004) and Schnurr (2009) finally consider failed and
unintentional humor in their discussion. The reason for the reluctance is
probably that failed humor requires necessarily two steps, both of which
CA and DA are reluctant to take: 1) accepting the distinction between
a competence and a performance levels of analysis, and 2) accepting
that failure requires logically the intentionality of at least one of the
participants to be taken into account. While DA practitioners may be
open to (or at least tolerant of) these, as are Bell and Priego-Valverde,
for CA analysts they may be too big a pill to swallow. The problem of not
including failed humor, or including only a subset thereof, is that, if our
goal is to provide as complete as possible a picture of the phenomena
surrounding humor and laughter in discourse, by refusing to engage
with part of the phenomena, we condemn ourselves to failure.

12.4.4 Sustained humor turns

Sustained humor turns are extended stretches of conversation in which
the exchange is framed/keyed for humor. They may consists of a sin-
gle speaker producing a series of humorous turns while other speak-
ers support the humor or in two or more speakers collaborating in
the production of the humor and mutually supporting each other.
“When humorous interchange is maintained over a period of time, the

295
 DISCOURSE ANALYSIS

interpretative work that it requires generates a distinctive group atmo-

sphere which is quite different from serious discourse” (Mulkay, 1988,
p. 51). However, Mulkay adds, “the humorous mode can be sustained
only by means of a continual production of appropriate cues, signals and
responses” (p. 51). This is correct, generally speaking, except for the fact
that the cues need not be continuous.
Tannen’s (1984) “humor routines,” “Norrick’s “jointly produced nar-
ratives” (1993, pp. 57–58), Hay “fantasy sequences” (1995, pp. 68–70),
Kotthoff ’s “joint fantasizing” (Kotthoff, 2007, 2009), and Winchatz &
Kozin’s (2008) “comical hypothetical” describe interactions in which in
which speakers sustain a co-constructed humorous improvisation of
various length. The longest sustained humorous interaction I am aware
of lasts up to 13 turns (Kotthoff, 2007). As we have seen, mode adoption,
one of the forms of humor support, consists precisely of “playing along”
with the speaker who has produced a joking or ironical utterance. I used
the term “mental space” (Attardo 2001b), after Fauconnier (1984/1985)
who introduced it, to describe the pragmatic inferential processes nec-
essary to build these co-constructed mental spaces in which two or
more speakers develop a fleeting shared alternate mental space in which
they explore the humorous possibilities of a situation. Most significantly,
I address the emergent properties of mental spaces (Attardo, 2001b,
p. 177), i.e., the fact that one may “discover” unanticipated consequences
of what one has postulated in a mental space. In the paper, I also address
the technical reasons why speakers do not become confused if different
mental spaces have contradictory features.
A related, but slightly different, phenomenon are laughter “clusters”
(Glenn, 2003, pp. 78–83): “[p]articipants can create additional laugh-
ables that cohere thematically or structurally with a preceding one”
(p. 78; cf. Norrick, 1993, p. 117) but that need not be as collaborative
as the examples above, as they may just be related instances of humor.
Glenn notes that not all the instances of humor (laughables) “get shared
laughter, but several do, and the cumulative effect creates and extended
cluster of shared laughs” (p. 80). According to Glenn, the speakers’
“focus moves sequentially to the next laughable” (p. 83) but Thonus
(2008) suggests that speakers are “no longer responding to laughables
but constructing laughter itself as a laughable” (p. 336).
Whether clusters of laughter or more organized co-constructions of
humorous mental spaces, these extended bouts of mirth seem to be self-
sustaining. I have made the suggestion that to some extent, there may

296
 12.5 CONCLUSIONS

be a “virtuous circle” at work, whereby the occurrence of humor feeds

back into the situation and generates more desire to produce humor
(Attardo, 2019). Be that as it may, “shared laughter naturally reaches ter-
mination points rather quickly. Conversationalists must actively renew
shared laughter” (Glenn, 2003, p. 82) or to put it differently, the Least
Disruption Principle is at work, enjoying the speakers to return to the
serious mode, as soon as they are done with the humorous interlude (see
Section 8.2.2).
Sustained co-constructed mental spaces require that at least two
(“joint”) speakers collaborate in the production of humor, but of course
sustained humor turns need not require active collaboration of the
speakers, but merely acceptance, support, and presumably enjoyment
(appreciation). It is also possible, of course, that the hearer being sub-
jected to social constraints, needs to pretend appreciation, while in
fact not enjoying the humor at all. Acceptance and support of humor,
presuppose (Hay, 2001) appreciation (or the pretense thereof).

12.5 Conclusions

The last two chapters have presented an interactional view on the per-
formance of humor as part of the ethnomethodological approach intro-
duced in Chapter 10, specifically on conversation and discourse analysis.
Before we turn to the variationist approach, which will conclude the
discussion of the performance of humor, it seems useful to draw some
overall conclusions on this fast moving part of humor studies.
Indeed, the first observation is that the research in conversation and
discourse analysis is by far the most active area of the linguistics of
humor and, with the exception of psychology, probably of all of humor
studies. This makes the relative lack of interest and cross-fertilization
between conversation/discourse analysis and humor studies all the more
disappointing. It is all too common to see decades of research in the
linguistics of humor summarized as a vague mention of incongruity
and the relative lack of influence of those scholars such as Neal Norrick,
Béatrice Priego-Valverde, and Nancy Bell who have been influenced by
humor studies. To be fair, more recent work seems to be more broadly
accepting of interdisciplinary insights into humor. In the sense, the
flurry of very recent work in conversation and discourse analysis, within
which Haugh’s contributions stand out, is to be welcomed.

297
 DISCOURSE ANALYSIS

The second trend that stands out is that the field is now broadly
orienting itself toward corpus-assisted studies. What was a relatively
new trend in my review of the field (Attardo, 2015), and had its roots
in the more functionally oriented discourse analysis, has now become
the more or less accepted norm, even within conversation analysis. This
is a very positive development, if for no other reason that it moves
away from what I still consider the most problematic aspect of the early
conversation analytical studies of humor that proposed generalizations
based on one instance of a phenomenon (see for example, in Section
11.4.3, the idea of humor as a test, Sacks, 1974).
The third tendency that I have observed is a slow but perceptible move
away from the borderline neo-behaviorist idea of studying laughter in
itself toward studying laughter (and smiling) as markers or contextual-
ization cues, and eventually socially as an expression of mirth, with the
corresponding reintroduction of the intentionality of the speakers in the
picture (anathema to the original conversation analysis methodology).

12.6 Further readings

The overviews by Glenn and Holt (2017), Schnurr and Plester (2017),
Attardo (2015) and Bell (2015) are the best sources to get started, in
general. More speciifc problems are examined by Whalen and Pexman
(2017), on humor support, and Burgers and van Mulken (2017), on
humor markers. The works of Tannen, Davies, Hay, Priego-Valverde,
and Holmes are fundamental. Schnurr (2009) is the only book from
the Wellington Language in the Workplace project specifically focused
on humor. A list of articles on humor by members of the project is
available online: https://www.victoria.ac.nz/lals/centres-and-institutes/
language-in-the-workplace/docs/biblios/humour.pdf. Many of their
articles are readily available.

298
 13
Sociolinguistics of humor

13.1 Universality of humor 299

13.2 Variationist humor theory 304
13.3 The social construction of humor 313
13.4 Conclusions 315
13.5 Further readings 315

This chapter will not deal with the sociology of humor. To do so would
require a monograph of its own.1 Another area that we will not cover,
for the same reason, is verbal art and speech play (see Scherzer, 2002).
The focus of the chapter is the sociolinguistics of humor, and specifically
variationist sociolinguistics (the previous chapters have dealt primarily
with interactionist sociolinguistics).

13.1 Universality of humor

One of the questions that a humor researcher gets asked most often is
“is humor universal?” often in a slightly different form along the lines of
“is humor from, say China, or from any other ‘exotic’ culture, different
from ours?” The answer, as is the case for most interesting questions, is:
“it depends.” In a sense, this chapter addresses the issue of whether men
and women, young people and old people, people from Japan, Namibia,
France, and people from the USA laugh or appreciate as humorous the
same things.
1Those interested should consult the work of Christie Davies, Giselinde Kuipers, Mulkay
(1988), Koller (1988), Davis (1993), and Zijderveld (1983).

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 SOCIOLINGUISTICS OF HUMOR

A good starting point is to recall that as a result of our discussion

in chapters (2), (4), and (6) we are studying humor as a competence
of the speakers, as Raskin’s theory explicitly states.2 Competence is
usually assumed to be innate (i.e., genetically encoded in the DNA of
the speakers) and would therefore have been selected for evolutionarily
by some advantage in reproductive chances (Porteous, 1988; Vaid, 1999,
2002; Gervais & Wilson, 2005; Chafe 1987, 2007). Another possibility is
that having a sense of humor is a completely neutral feature, reproduc-
tively speaking, that emerged as a by-product of other characteristics
with reproductive advantages (for example, the capacity to experience
surprise and inferential processes). Humor competence may also be
merely a social abstraction of behaviors sanctioned by social practice. If
humor competence is innate this means that logically all human beings
share the same humor competence since they share the same DNA. If
humor competence is merely a social construct, the argument is much
weakened, or defeated as it would be conceivable that some culture do
not have a humor competence at all or have a radically different one (one
that is inaccessible to other cultures).3
The available evidence leans toward the universality of humor and
hence to its genetic foundation. There have been no claims of cultures
that do not have humor or laughter. There have been isolated claims of
cultures lacking irony (Haiman, 1998) and lacking puns (in the sense of
intentional manipulation of ambiguity/homonymy, etc. for amusement;
cf. Sherzer, 1996), but at most this would show that those cultures lack
some genres of humor, not that they lack humor. So, if humor occurs
in all cultures it becomes likely to assume that it is part of the genetic
make up of humanity. Another potential explanation of the universality
of humor would be that it had already developed when the first hominids
left the African savannah and therefore that it spread with them to all
inhabited lands. In either case, be it genetically determined, or culturally
inherited, humor competence is common to all cultures.
More compellingly, there have been no claims of radical untranslata-
bility of humor across cultures. Most cross-cultural studies show the
exact opposite, and namely that jokes and humorous themes travel freely
across linguistic and cultural borders. For example, Essex girl jokes from
England crossed the ocean and became blonde and sorority girl jokes.
2 It should go without saying that we are not only studying competence.
3 This footnote should be read only by trained linguists/philosophers: an anthropologist
sees a rabbit run by, he shouts “Gavagai!” and all the tribe starts laughing.

300
 13.1 UNIVERSALIT Y OF HUMOR

American stupidity jokes about Poles correspond to jokes all over the
world about “stupid” groups (for example, in Italy the same jokes are told
about the police, and in Ancient Greece they were told about professors).
What are we to make then of claims such as those that some kinds
of comedy “do not work” in a different culture? Usually what these
mean is that the humor is not appreciated very much, not that it is not
perceived as humorous. In other words, that the performance of the
humor is not very successful (for example, TV audiences may not react
well to it) and not the much more radical and significant claim that the
audience fails to perceive that a give action or expression is intended to
be humorous. The closest such mis-perception I am aware of is Regina
Barreca’s treatment of Benny Hill’s comedy. Barreca (1991, pp. 146–148)
fails to be amused by Benny Hill and thus denies the humorous status of
the text(s), regardless of the fact that it ran in England between 1955 and
1989, albeit not always continuously. Had it been the case that BBC and
Thames TV broadcast an unfunny comedy show, it would presumably
not have lasted over forty years.
Why is it the case, then, that some types of humor are more or less
popular in a given culture? Or how can Barreca and millions of British
television viewers disagree? This can be due to two factors:

• lack of a given script (information) necessary for the humor to

function
• social differences in the cultures.

Consider the following joke:

(127) There are 10 kinds of people: those who know binary and those
who don’t.
The joke is completely opaque unless one knows that in binary math-
ematical notation 10 equals base-ten two. Once one is provided with
the relevant information, the joke may fire (although the element of
surprise is lost, thus considerably weakening the humor). The same goes
for didgeridoo jokes (which my Australian informants assure me exist
and are hilarious). Unless one knows what a didgeridoo is, one cannot
really grasp the humor.4 Most of the failures of humor to transfer from

4 I met this bloke with a didgeridoo and he was playing Dancing Queen on it. I thought,
“That’s Aboriginal.” http://nomadsrus.blogspot.com/2007/09/jokes.html

301
 SOCIOLINGUISTICS OF HUMOR

one culture to another can be explained very straightforwardly by simple

lack of a script (information) about a given topic. For example, Morain
(1991) describes a study contrasting ESL students’ and American stu-
dents’ ratings of New Yorker’s cartoons and underscores the necessity to
possess a given cultural script to be able to understand the humor, let
alone appreciate it.
A completely different situation is the observation, for example, that
Germans enjoy scatological humor more than French or Italian people,
or that puns are more frequent in English joke collections than in Italian
ones. These differences can be explained by showing that different
cultural backgrounds make this or that topic of humor more accessible
for appreciation or more enjoyable. So, presumably, the fascination
for sausage-related humor in German folklore is given by an interest
in the subject matter (Hempelmann, personal communication). Other
differences have more mundane explanations: for example, English is
much richer in monosyllabic words than Italian (most Italian words
are polysyllabic) and thus homophones and paronyms are much more
frequent and therefore easier to exploit for humor.
Relatedly, we can note that there exist differences across cultures in
the social acceptability of the use of humor. Different cultures codify
differently the times/spaces available to speakers for the production and
appreciation of humor. It is a well-documented fact that Americans
start public speeches with a joke, whereas this practice was unknown
in Europe. Needless to say, much cross-cultural trouble can ensue.
I personally was astonished to hear a priest joke from the pulpit when I
first attended a Catholic mass in the USA.
Moreover, different cultures may have different sets of scripts available
for humor. Consider the following passage from an anthropologists’
memoirs of her work among the Tiv, in Nigeria:
Ngun was a nice old man who bore his blindness bravely ( . . . ).
“Well,” Accident [a native friend of the anthropologist] told his
story with zest, “I was out hunting birds with my slingshot, and I
met him. He was all alone, shuffling along because the path was so
slippery with mud. He was about to fall anyhow. So I yelled, ‘Watch
out, Ngun. A snake!’ ” Accident hugged himself with mirth, and
Ikpoom [another native friend] howled with laughter. Only Lucia [a
madwoman] and I didn’t get the point.
“Well,” I prompted. “Go on. Tell me what was funny.”

302
 13.1 UNIVERSALIT Y OF HUMOR

Ikpoom ( . . . ) gave me the explanation, his usually monotonous

voice lively with amusement. “Ngun is blind. He can’t see. He wouldn’t
know which way to jump.” He begun to laugh again. “There’s nothing
funnier than yelling ‘Snake!’ at a blind man.”
“Funny!” I looked at Ikpoom as though I were seeing him for the
first time.
“Yes,” Accident put in. “He can’t know where it is.”
Ikpoom saw that I was not amused. He was a sensitive man and a
kind one; he thought he knew what disturbed me, and he tried to set
my mind at rest. “If there’s really a snake there, you mean to tell him
where it is. Only people get so excited when they see snakes that they
don’t remember. They just yell. So of course a blind man can’t be sure.”
(Smith Bowen [Bohannan] 1954: 228)
Clearly, to Smith Bowen, and presumably to most of the readers of
this book, playing a practical joke on a blind man is unacceptable,
whereas to the Tiv it was. Before we jump to conclusions about the
relative sophistication of the Tiv culture and of Western culture, let us
recall that during the Middle Ages people with physical handicaps were
employed as court jesters and that until recently, if not still, freak shows
made money by exhibiting people with physical deformities. So, what
the example of the blind Tiv man shows is that there are differences
between what the members of an African culture and the members of
some parts of Western culture consider appropriate subjects for practical
joking, but that the practical joke is recognizable as such (for example,
Smith Bowen would probably have seen the humor if the butt of the
joke (Ngun) had not been blind). To put it differently, the Tiv and Smith
Bowen do not have a different competence of humor they merely differ
at the performance level of which individuals are acceptable targets for
humor (for example, presumably, a dying person would not be targeted
for the snake joke by the Tiv).
The availability of scripts for humor will also change according
to circumstances. For example, Gilbert Gottfried was booed for
performing a joke about the Twin Towers “too soon” after the event (see
Attardo, 2017a for discussion). Kuipers had collected over 800 jokes
on the subject by 2005 (Kuipers, 2005). A wikipedia web page (https://
en.wikipedia.org/wiki/Humor_based_on_the_September_11_attacks)
lists jokes in sitcoms, etc. based on the attack. Clearly, the sensitivity
to the event has changed significantly over time. Interestingly, some

303
 SOCIOLINGUISTICS OF HUMOR

of Kuipers’ examples appeared mere days after the event, in Dutch

web sites. So, clearly, the Dutch and the American cultures differed
in availability of those scripts for humor.
Recapitulating, humor is universal, either because genetically encoded
in all humans or because it is part of the shared heritage of the first
cultures. Different cultures may have different clusters of information
or may have different attitudes toward certain kinds of humor which
may facilitate or impeded the appreciation of given forms/examples of
humor, but no humor is in principle inaccessible to any member of
a different culture who would be given all the necessary information.
Finally, different cultures may partition differently the social spaces
available or not available for humor.

13.2 Variationist humor theory

In the following sections, I will briefly review some of the areas of in

which variation in humor has been seen to occur and has generated
some scholarship. The reader is advised to keep the universality of
humor among humans firmly in mind, while reading the sections below,
for perspective.

13.2.1 Humor and gender

Among the best-investigated topics in variationist sociolinguistics of
humor is certainly humor and gender. Specifically researchers have
been trying to answer the central question: do males and females sense
of humor differ? Do the social circumstances in which speakers use
humor and/or its function change according to gender? These ques-
tions have attracted significant research, in many disciplines, ranging
from gender studies to psychology and passing by linguistics of course.
A comprehensive review would require a monograph of its own. For
these reasons, I will sketch out only in the most general terms the debate
and will rely on overviews of the field to provide broad representation.
13.2.1.1 The feminist awakening
It is possible that one of the reasons for the very large body of research
on the subject is that humor is a very visible sociolinguistic variable
and as such it would have attracted the attention of scholars. Another,

304
 13.2 VARIATIONIST HUMOR THEORY

less flattering hypothesis, is that the existence of countless stereotypes

on humor and gender may have fueled the interest of the scholarly
community. Be that as it may, we can start our review with Robin
Lakoff ’s provocation in the seminal paper-turned-booklet Language and
Woman’s Place (1975, p. 43) that women “have no sense of humor.” It
is fair to say that the first phase of the research is the awareness that
there are gender-based differences in the type of humor and humor
performance, in the broad sense we have been pursuing in this book,
that people say and enjoy (or not enjoy).
Early feminist studies, reviewed in Crawford (2003) and Kotthoff
(2006), emphasized the differences between male (competitive) and
female (supportive) humor as well as the subservient role of women
who did the laughing while men did the telling of humor. Jenkins
(1985) seems to be the acknowledged source of much of the polar
characterization of the differences between women’s and men’s humor:
women’s humor is “cooperative, inclusive, supportive, integrated,
spontaneous and self-healing. In contrast men’s [. . .] is more likely to
be competitive, exclusive, challenging, segmented, preformulated, and
self-aggrandizing” (p. 135). Interestingly, Davies (1984) had proposed
a much more sophisticated approach, which characterized women’s
humor as directed toward empathy and men’s humor as more oriented
toward solidarity (for example, using banter or ritual insults).
13.2.1.2 Post-feminist research
Holmes et al. (2001) present particularly significant results because they
go counter to most of the claims we have just reviewed. For example,
Holmes et al. find that women produce more humor than men, regard-
less of the power status (Holmes et al., 2001, p. 93). Furthermore, women
produce more humor regardless of being in single or mixed gender
settings (p. 96). Contrary to the claims of Jenkins (1985), there was no
difference in the number of collaborative humorous exchanges initiated
by men and women, when they were in the chair position. When women
were not in the position of chair they initiated more collaborative humor
(p. 101). Holmes et al. conclude “the quantitative data contradicts the
negative stereotype” that women “have no sense of humor” (p. 101).
A full description of their methodology will be found in Section 12.3.
Günther (2003) found that women laugh significantly more than men,
but don’t produce more humor support (p. 200), thus contradicting the
assumption that women are more supportive of humor. Laughter and

305
 SOCIOLINGUISTICS OF HUMOR

humor have a higher frequency of occurrence in single-sex interactions,

but women deliver more humor in women-only polyadic (more than
two participants) settings (p. 181). So, once more, the humorless woman
myth is laid to rest.
13.2.1.3 Empirical synthesis
The title of Martin’s meta-analysis (Martin, 2014) is (modestly)
restricted to psychological research, but Martin’s wide-ranging schol-
arship covers far more than that and comes to a somewhat startling
conclusion that “we find more similarities than differences between
men and women” (p. 144) or as he puts it quoting Dindia’s (2006) quip:
“men are from North Dakota, women are from South Dakota” (p. 144).
In light of the remarkable convergence of work from the New Zealand
group around Janet Holmes, from the corpus-based work of Günther
and Martin’s authoritative results, the field of humor and gender is in
need of a serious re-evaluation. Let me highlight two directions that
strike me as particularly fruitful: 1) Holmes (1995), in the context of
hedging by women and men, suggests that “women’s subordinate social
status may account not so much for the way women talk, as for the
way their talk is perceived and interpreted” (p. 111); to my mind, this
can be extended to the discussion of the gender differences in humor
and describes pretty accurately the situation. 2) Increasingly researchers
are problematizing the duality of the male/female opposition in favor
of a “performative” approach to the study of gender, i.e., the way in
which speakers define themselves in relation to each other through
the category of gender and its stereotypes (Crawford, 2003) Again,
I think that the gendered social practices of humor will have much to
learn from such an approach. For example, the fact that a parodic genre
such as “drag” contributes to the definition of gender seems to be quite
significant (Butler, 1990, p. 174).

13.2.2 Social class

Davies (2017) provides a telling summary of the state of the art on the
topic of social class and humor:

Whereas many studies implicitly relate to social class – for example, in that white-
or blue collar workplaces may be assumed to have certain class characteristics, there
has been little specifically on sociolinguistic variation in relation to class. The notable

306
 13.2 VARIATIONIST HUMOR THEORY

exception is the 2010 special issue of [HUMOR: International Journal of Humor Studies]
edited by Attardo (2010). In his introduction, Attardo makes an important distinction
between a sociolinguistic phenomenon and its mediatic representation. He defines
working-class humor as “humor produced by members of the working class for other
members of the working class” (p. 122), and points out that because mass media
represents middle-class values, we are unlikely to get a full and accurate representation
of humor embodying antagonism to those values. (p. 480)

The long extract is worth quoting in full because it highlights both the
lack of research in the field and the problem that what little research
there is on social class is focused on the media representation thereof,
which is biased at best.
Davies herself in a paper for the special issue provided one of the most
cogent discussion of class and humor in American society. Davies (2010)
is dedicated to show that the popular

“You might be a redneck” joke cycle is appropriated to designate a lower social category
within the Southern working class in Alabama (including imitation of the voices of
the characters using exaggerated Southern features), and to negotiate the boundaries
between the good old boy working class “red neck” [ . . . ] and the lower category of
“white trash” [ . . . ]. This term, a slur that manages to combine both race and class,
emerged in the racial caste society of the nineteenth century American South, and
continues as a way of designating the bottom of the white social hierarchy. (p. 180)

Davies’ conclusion is that the goal of the performance is “to convey

social distinctions within the white Southern American working class”
(p. 197) specifically pitting the relatively better off “red necks” against
the impoverished “white trash.”
The “You might be a redneck” joke cycle was popularized by comedian
Jeff Foxworty, who released an album called You Might be a Redneck’
If . . . According to Wikipedia, the album sold more than 2 million
copies. Foxworthy’s stand-up comedy act consisted in part of a list of
statements beginning with the formula “You might be a redneck’ if. . .”
and completed by a statement, such as “You own a home that is mobile
and 5 cars that aren’t.” In the show that Davies examines, the two hosts,
Jack and Bubba, replace the expression “a redneck” with “white trash.”
Davies provides a sophisticated analysis of the phonology of the
dialect of the show hosts, as well as other features of their speech. For
example, toothlessness, “a classic indicator of social class” (Davies, 2010,
p. 187) is used as a defining feature of “white trash.” Overall, the image
of the “white trash” person that emerges from the analysis of the text is
summarized as
307
 SOCIOLINGUISTICS OF HUMOR

poor, stupid, reckless, dirty, toothless, homophobic, alcoholic, addicted to tobacco

in various forms, [ . . . ] violent, drunk [ . . . ] a bad parent, sees no particular value
in education, and thinks that procreation should start at puberty (possibly through
marrying close relatives). He lives in a trailer with a refrigerator on the front porch,
drives a truck that is a heap of junk, plays pool (which may imply gambling), and is a
NASCAR fan. (Davies, 2010, p. 195)

Davies notes that many of these features are directly connected to

poverty. When it comes to the interpretation of the material in a broad
sociological sense, Davies notes that by characterizing “white trash”
people as they do, the self-proclaimed “red necks” establish themselves
as having some social respectability, for example, “He is not rich, but
he is not poor in the way that the W[hite] T[rash] characters are. [ . . . ]
Marriage and family are important to him, but he understands the
appropriate kin relationships for these purposes” (p. 196).
Porcu (2005) is an eye-opening account of humor in a working-class
Sardinian fish market. Unlike the Bowdlerized forms of working-class
humor that appear in the media (which reflect middle- to upper-class
values) that humor of the workers at the fish market is very physical,
often violent, involving hitting people, sticking fish in their clothing,
etc. It is needless to say rich in obscenity, homophobic, and above all
extremely aggressive.

13.2.3 Age
As Coupland (2004) notes, there is a lack of research on aging in
sociolinguistics at large. So, it will not come as a surprise that very little
has been written on age grading and humor (but see Nahemow et al.,
1986). There is a lot of research on the acquisition of humor in children,
but that area has developed on its own and the focus is on cognitive
development. Here we are interested in how children’s humor may differ
from teenagers, and the latter from adults and seniors.
Apte (1985, pp. 50–51) reports on “age-sets” that is joking that is
limited to persons of the same age, in various cultures. Apte likens age
sets to young people attending high school or college together. To use
a familiar example, children and pre-adolescents in Western culture go
through a period of telling and enjoying sexual and scatological jokes
(Apte, 1985, p. 94). Interest in these topics wanes with age.
Apte (1985, pp. 79–80) notes that in a variety of cultures, older (post-
menopausal) women are freer to engage in humor, often sexual, and
even “obscene.” Kotthoff (2006) notes an interplay of age and class in
308
 13.2 VARIATIONIST HUMOR THEORY

the behavior of women: younger women show “the greatest reserve”

in dealing with aggressive and sexual humor, whereas older lower-class
women are freer to address taboo topics (Kotthoff, 2006, p. 14). Nardini
(2000) reports on an extended joking session among older women in an
Italian-American club. The topic is what to do after widowhood and the
suggestions undergo a crescendo of outrageousness which culminates in
taking the deceased husband on a cruise in a box—without having had
him cremated first. So, we can agree that older women may be indeed
freer to express taboo topics.
This area is one where there is a particularly pressing need for
research. With the advent of the internet and new media, I suspect
that age grading will be particularly significant between generations.5

13.2.4 Dialects as humorous languages

As we saw, Hymes (1972) SPEAKING mode (Section 10.2) considers
dialects, registers, etc. “instrumentalities” of the situation. We discussed
briefly code-switching in Section 10.2.4. Here we focus primarily on
diglossic situations.
There is a widespread understanding that non-standard dialects are
used for humorous effect (Apte, 1985). Fry (1963, pp. 32, 144) men-
tions dialect as a contextualization cue (although Fry does not use
this terminology). Siegel (1995) mentions dialectal humor. Papapavlou
(1998) examines Greek Cypriot humor. Apte (1985) has a review of
studies on the subject: in Tamil movies the buffoons speak in the
spoken variety, whereas the heroes speak in the literary variety (p. 190);
among the Burundi, joking is done with the “least degree” of formality
(p. 191). In Zinacantecan Indian, “joking occurs in the informal style”
(p. 191). The Chamula consider frivolous talk (which includes joking) as
inappropriate for a ritual setting, the latter requiring “much formality.”
Apte notes that “certain dialect or languages are perceived as more
suitable media for humor than others” (p. 191). In Paraguay, where
Spanish and Guaranì coexist in a diglossic situation, speakers consider

5 The difference between age sets and age grades is that one belongs to an age set
throughout their life, whereas one passes through age grades as they age. Again, the example
of classmates or college friends is enlightening: one will always have the same high school
or college friends, but one will pass through a series of grades, childhood, adolescence,
adulthood, middle age, etc.

309
 SOCIOLINGUISTICS OF HUMOR

Guaranì as “more suitable for humor.” (p. 192). Davies reviews the scarce
literature on the subject as well (2017, p. 477).
I would like to present as a hypothesis the idea that dialects, as
subordinate lectal varieties of their respective acrolectal6 languages,
serve as the locus of humor because of the implicit devaluation of the
subject matter (“non-serious matter”) or the explicit presence of “low”
(corporeal, scatological) subject matter. In this sense dialects are both
a marker of the humor and the target of the humor, in the sense that
humor is a debasing activity that must take place in a bounded envi-
ronment and so the dialect provides the perfect place for both asserting
the underprivileged nature of the activity (engaging in humor) and
providing a natural “limit” to the social violations of humor. Needless to
say, humor can occur in the standard varieties as well, but I hypothesize
that the tendency of dialect to be associated with humor is tied to the
factors discussed above. Apte claims that “Humor and joking exchanges
need a familiar setting in which such barriers of communication as
age, rank, and social status are considerably reduced” (p. 195). I would
argue, that on the contrary, the awareness of the status differences
between the lectal varieties is enhanced and that the very differences are
emphasized.
I will now exemplify this suggestion, with an example of a 17-second
skit that circulated in 2018 on various online fora in Sicily. It was brought
to my attention by Totò Panzeca, who also contributed to its explanation.
It is transcribed below. The bolded text is the original in the skit; a
translation in English is provided in italics. When the text is in Sicilian
an interlinear gloss in Italian has been provided.
The skit is recited by two unidentified male voices. The first line opens
with the typical phone call greeting and by the identification of the
recipient of the call as the Hospital in Catania (the second largest city
in Sicily). This frames the situation as both a phone call and as a health-
care service call.
(128) 1. Pronto, Ospedale Civile di Catania?
‘Hello. Catania Civil Hospital.’

6 A dialect (“lect”) has different levels of formality: the basilect is the lower form, also
known as the vernacular, used in the most informal settings. The acrolect is the highest,
most formal variety, used in formal or official settings; often a standard language will be the
acrolect, especially in a diglossic situation. The mesolect is an intermediate form. Acrolect,
mesolect, and basilect form a continuum.

310
 13.2 VARIATIONIST HUMOR THEORY

2. Eh, bongiorno. Scusi, peccasu pozzu prenotare una visita

Eh, buongiorno. Scusi, per caso posso prenotare una visita
cu’ pacchiologo?
con il pacchiologo?
‘Eh, good morning. Excuse me, could I book a visit with the vaginologist?

3. Non ho capito, scusi, il pacchiologo?

I did not understand, sorry, the vaginolologist?

4. Se, se, u’ pacchiologo

Si, si, il pacchiologo.
‘Yes, yes, the vaginologist.’

5. Forse voleva dire: il ginecologo

‘Perhaps you meant: the gynecologist.’

6. Aspittassi un attimu. Sasà chi tti fa mmali, u’ pacchio o u

Aspetti un attimo. Sasà che ti fa male, il pacchio o il
ginocchiu?
ginocchio?
‘Wait a second. Sasà what hurts? Your genitals or your knee?’
The humor in the skit revolves around the use of the term “pacchiu”
(vagina) with the suffix “-ologo” (doctor of). The caller asks to make an
appointment with the “vagina doctor” (line 2). The hospital receptionist
is confused and guesses that perhaps the caller meant “gynecologist”
(lines 3–5). The caller misunderstands based on a folk-etymology of
“ginecologo” derived from the similarity between “ginocchio” (knee)
and Greek “gunaikos” (‘women) [whence the root “gineco” in Italian]
(line 6). I translated “pacchiologo” (which obviously has no direct
translation in Italian since it is not a word) with “vaginologist” because of
the morphemic resemblance between “vaginologist” and “gynecologist.”
“Sasà” (diminutive of Rosa) is presumably the wife of the caller.
What is particularly remarkable about this text is that it alternates
between Standard Italian and Sicilian (Catanese). The turns of the
telephone operator at the hospital are all in Standard Italian, whereas
the caller uses a basilectal variety, mildly tempered by some mesolectal
touches (“Bongiorno,” “fa mmali” rather than “doli”) and the suffix

311
 SOCIOLINGUISTICS OF HUMOR

“-ologo.” This is a perfect example of the use of dialectal varieties to

mark (or provide a contextualization cue for) the humor in the text. The
caller is immediately characterized as less educated, which then helps
justify the misunderstanding ginocchio/ginecologo predicated on the
presumed ignorance of the word “ginecologo.” It is no coincidence that
the punch line (in turn 6) is delivered in Sicilian, precisely because the
use of the dialect marks the orientation toward the lower class, and we
might say in a nod to Bakhtin, the body. “Pacchio” is a vulgar expression,
and in this sense my translation as “vaginologist” is not correct, and a
better translation would have been “pussylogist.”
Rising intonation is indicated with a question mark in 1 and 3. In both
cases it indicates a request for information: in the opening turn, it is a
request of a reason for the call, while in the third turn it is a request for
an explanation. Lengthening of the vowel in turn 5, indicated with a “:”
following the syllable, may indicate hesitation or embarrassment.
Gasquet-Cyrus (2004) discusses the status of the regional French vari-
ety of Marseilles (Southern France), which has a stereotypical humorous
value in France. For example, Gasquet-Cyrus reports a local singer,
who, while waiting on stage between songs, says, “Eh, i(l) va venir” [it’s
coming (referring to an instrument that was being brought to him)] with
a strong Marseilles accent. Gasquet-Cyrus notes that there is nothing
incongruous in the fact of waiting for an instrument and neither is the
use by a local person in front of a mostly local audience of the local
dialect, but nonetheless the remark was greeted with laughter, because of
the stereotypes attached to the performer and his dialect (pp. 142–143).
There is much more to Gasquet-Cyrus’ work, especially his urban
dialectological study of humor in Marseilles, but in this context we can
only hint at the rich material in this dissertation, which unfortunately
remained unpublished.
Another, related, way of looking at this is the idea of metaphorical
code switching. The idea behind metaphorical code switching is that
its purpose is to convey an inferential meaning, signaled by the change
to a different code. This is in opposition to situational code switching,
in which one shifts languages or varieties simply to accommodate the
various speakers or because a given situation requires or favors a given
variety. Essentially, whereas situational code switching is determined by
the situation, metaphorical code switching is a choice by the speakers.
As such, it generates an implicature. For example, my daughter who is
essentially English monolingual but knows some Italian, speaks English

312
 13.3 THE SOCIAL CONSTRUC TION OF HUMOR

with her son. However, I recently observed that when he gets too ram-
bunctious she tells him to stop by saying “Basta!” [Enough!, in Italian].
The four-year-old knows exactly what the code switching means, i.e.,
you are about to get in trouble, and stops his behavior.
The difference between situational and metaphorical code switching
was introduced by Blom and Gumperz (1972). The use of a particular
code may serve as a contextualization cue (Gumperz, 1982, p. 131; see
Section 10.2.4). Therefore, “a switch to the code appropriate for humor
can be a signal that the content is not serious” (Siegel, 1995, p. 101).
Archakis et al. (2014) is a review of studies on stylistic humor in modern
Greek, both in a mass-mediatic environment, which reveals that “in
mass culture genres, stylistic humour is employed as a crucial means
for characterisation: characters in media genres are more often than
not denigrated and laughed at for their ‘unconventional’, ‘non-standard’,
and/or ‘heterogenous’ stylistic choices” (p. 49). The authors list also a
few studies of stylistic humor in Greek sociolinguistics.

13.3 The social construction of humor

As Berger (1997) argues, the key to understanding the relationship

between humor and society is to realize that social life exists becasue the
people who share a social space/group have a large set of conventions,
rules, received ways of doing things, habits, connections, etc. that define
and constitute the “hanging together” of a social group. Think for
example, how much the location where a group gathers determines what
the group will do and how they will interact, so that if they decided
to change the place where they gather, for example, that would change
significantly how the group interacts (for example, think of students in
class and outside of it).
These rules and conventions for a “reality” which, because it is so com-
monplace and taken for granted, becomes “transparent” to its members
(i.e., they are no longer aware of it).
Berger argues that humor is the intrusion of another reality in the
everyday reality of a social group. Other examples of alternate “realities”
would be religion and magic. (Berger, 1997, p. 65). For example, both
humor and the sacred are contained in time and place. “Church time” is
constrained in terms of location (one must go to a special building, or
designate a location to serve as such) and special behaviors (rituals) and

313
 SOCIOLINGUISTICS OF HUMOR

language (formulas, prayers) are allocated to this time (ceremonial). In

Western society, at least until recently, one would wear special clothing
(one’s best) for such occasions. Humor is not so much constrained in
terms of place (although there are prescribed comic locations, such as
comedy clubs, the stage, etc.) but it is more limited in terms of times or
situations. All societies will have a more or less specific list of times when
it is inappropriate to joke (for example, funerals), but more importantly,
there are signals, known7 to the entire social group, that introduce and
delimit the humorous “reality.” Typically, we think of formulas such as
“do you know the joke about…” but other markers may be present as
well, such as the familiar smiling, laughing, etc. but also face painting
(clowns), special dress, such as carnival costumes or the practice of
dressing like the opposite sex in the Caribbean carnival, documented
in Sherzer (2002, pp. 145–146).
The social conventions about humor form that society’s “comic cul-
ture,” defined by Berger as “the definitions of comic situations, roles,
and acceptable contents in any social group or society” (1997, p. 68).
Needless to say, social groups may be as small as a couple and as large as
a (trans-)national entity (e.g., Chinese humor, European humor).
However, the process of defining a situation as comic has another side,
or as Berger puts it: “As the comic situation is socially defined, it is at
the same time contained” (1997, p. 68). The process of delimiting, of
containing the “alternate reality” reflects at the cultural level the same
idea behind the principle of least disruption at the conversational level
(see Section 8.2.2). In other words, an alternate reality is fine, but it
cannot take over the “real” reality.” The carnival, for example, works
because it is limited to Mardi Gras. If one tried to have 365 days of
carnival, the society would collapse, as no productive work would get
done. As the proverb widely attributed to Seneca and Horace goes, Semel
in anno licet insanire (Once a year, you may go crazy). Once the party is
over, we all go back to our regular lives, work, and obligations.

7 The knowledge need not be explicit: this is the kind of implicit knowledge of compe-
tence. Just like speakers of a language know how to speak without being able to articulate
the grammatical rules that govern their behavior.

314
 13.5 FURTHER READINGS

13.4 Conclusions

This chapter wraps up the third part of the book on the performance
of humor. While the previous chapters have been mostly interactionist
in their perspective, this chapter was more oriented toward variation,
i.e., how the humorous practices of the speakers are affected by such
factors such as gender, age, social class, and dialectal repertoire. The
final part of the chapter deals with the ultimate form of variation,
the cultural construction of humor at large. It is fair to say that less
research is available, generally speaking, on these subjects, perhaps with
the exclusion of the gender variable. This may explain the somewhat
tentative nature of some of the treatments.

13.5 Further readings

Davies (2017) is an excellent synthesis of the subject. Good overviews

of the research on gender and humor can be found in Crawford (2003);
Kotthoff (2006); and the already mentioned Davies (2017). Apte (1985)
remains a good synthesis of work on the social aspects of humor, but is
now out of date.

315
 14
Humor in literature

14.1 Script-based theory of humorous texts 321

14.2 Other approaches 328
14.3 Narratology 330
14.4 Stylistics 332
14.5 Some examples of literary constructs 336
14.6 Further readings 339

This chapter deals with humor in literature but not with literary works
per se. This stance deserves some clarification: I do not have a negative
opinion of the study of literature in and of itself: my first undergraduate
degree is in French literature. Literary studies is a perfectly fine field and
scholars have done a fine job of studying the various aspects of literary
works that happen to be humorous, such as comedies, picaresque novels,
parodies, pastiches, etc. ad libitum. For example, within humor studies,
Triezenberg (2008) provides a short historical overview and a glossary
of terms. Nilsen & Nilsen (2008) consists of a long list of genres1
and authors that are relevant to literary humor, without any attempt
at integration. The terminology of literary studies is plagued by the
problem, already discussed in Chapter 1, of first- and second-order
constructs. While few native speakers will have a first-order construct of
romantic comedy or of picaresque, authors, critics, and scholars argue
extensively on the nature and coverage of the second-order concepts.

1 Among the genres quoted are comedy, humor, satire, irony, fantasy, parody, picaresque,
and farce. Each term has many subdivisions.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 HUMOR IN LITERATURE

We will not attempt to summarize these discussions, but we will pro-

vide two examples, in Section 14.5, namely farce and satire to show
how literary scholars engage these constructs. Individual entries in the
Encyclopedia of Humor Studies may also provide more details about
individual literary terms of interest.
However, our focus here is on the humor, not on the literary aspects
of humorous genres. In a sense, we are interested in humorous literary
works because they contain humor and not because they are literary
works. Needless to say, this distinction is easier stated than respected,
as various scholars and approaches we will review may “color outside
the lines” we have just marked out, or completely disagree with our view
and advocate an approach more oriented toward literature. Regardless
of which position will turn out to be preferable, it would be impossible
to even pretend to summarize all that has been said in literary studies
about humor, so our survey will only concern itself with linguistically
oriented works. Further, it will be concerned with those studies that are
theoretically oriented. This is an important limitation, because most of
the works that I have classified as no-theory theories, in Chapter 3, are
from within literature-based approaches. So, in a sense, this chapter will
not deal with mainstream literary studies involving humor because what
is truly mainstream in the study of literature has little or nothing to do
with linguistics.
Nilsen and Nilsen (2008), displaying a wonderful sense of humor,
open their almost-40-pages-long chapter with a review of arguments
that the analysis of literary humor is a bad idea. My favorite is Bernard
Shaw’s: “There is no more dangerous literary symptom than a temp-
tation to write about wit and humor. It indicates a total loss of both”
(Nilsen & Nilsen, 2008, p. 243). Nilsen and Nilsen note also that “The
study of literary humor is in some ways as broad as the whole field
of humor research, plus the whole field of literary criticism” (p. 246).
Plainly, it would be impossible to account for all of this in one chap-
ter, hence, again, the need to restrict our concerns with linguistically-
related work. We will start out with the applications of the Semantic
Script Theory of Humor and the General Theory of Verbal Humor to
the study of literary texts and then move on to other linguistically-
oriented approaches, including narratologically-oriented and stylistics
approaches. We will wrap up the chapter with a brief look at a few
exemplar literary studies that are less linguistically oriented, just to show
off the power of literary analysis per se.

320
 14.1 SCRIPT-BASED THEORY OF HUMOROUS TEXTS

A different approach at delimiting the boundary between stylistics,

broadly defined, and literary theory is proposed by Simpson and
Bousfield (2017):

In the absence of robust justification within a framework of language and discourse, it is

simply not enough for the critic-analyst to decree that a passage of writing is humorous

in other words, the burden of determining the humorousness of the text

is on the linguistic theory and not on the literary scholar’s intuition.

14.1 Script-based theory of humorous texts

We start out by considering the application of Raskin’s Semantic Script

Theory of Humor to longer humorous texts. The Semantic Script The-
ory of Humor has a well-defined purview and namely jokes expressed
verbally or in writing. How does it fare when applied to another kind of
text? This is our concern in this section.
The Semantic Script Theory of Humor is very abstract. As such it does
not address, in principle, various issues. For example, the script theory
“flattens” any text onto a script opposition. This is not a criticism of the
theory, it is merely a refection of its semantic nature, which leads it to
abstract away from the chronological presentation of the scripts in the
opposition, for example. To go from the sex script to the no sex script
is indistinguishable, in terms of semantic opposition, from going from
the no sex to the sex script. In both cases, the opposition is between
sex and no sex. This is not to say that one could not handle the order of
presentation of the scripts in the Semantic Script Theory of Humor with
the introduction of an ad-hoc apparatus, such as a chronological labeling
of the various scripts, but the point is precisely that in its abstraction, the
Semantic Script Theory of Humor does not concern itself with the order
of activation of the scripts.
Another issue that the Semantic Script Theory of Humor is unable
to address is whether the source of the opposed scripts involves the
phonological form of the utterance, or to put it more plainly, whether
the joke is based on a pun or not. Once more, this is not necessarily
a criticism: the Semantic Script Theory of Humor is interested in script
oppositions and overlaps. The source of said opposition is not part of the
purview of the theory and as such it coherently ignores the difference
between puns, in which two senses are opposed, based in part on the

321
 HUMOR IN LITERATURE

sound resemblance between two strings, and other jokes in which the
opposition is not based on any aspect of the string, except its semantics.
So, if one is interested in distinguishing between puns and non-puns
(referential humor), one has to either add to the theory or come up with
a broader theory. The latter strategy is what Raskin and myself did, in
Attardo & Raskin (1991).
There are two approaches to the application of the Semantic Script
Theory of Humor to longer literary humorous texts: one can either
apply the theory as is to longer texts and add ad hoc mechanisms to
handle phenomena not covered in the original Semantic Script The-
ory of Humor, or one can build a broader theory, In Attardo (1994),
I labeled these approaches the expansionist approach and the revisionist
approach, respectively. We will address both in that order below.

14.1.1 The expansionist approach

There are two major proposals that fall under the expansionist approach.
We will consider them in chronological order.

14.1.2 Chłopicki
Chłopicki’s early work (1987) essentially assumed that the Semantic
Script Theory of Humor as stated originally could be applied to non-
joke texts. In a sense, it is a minimalist approach to a theory of literature
which essentially denies that literary texts have features that are signifi-
cantly different from jokes or that those features are relevant in assessing
their humorous aspects. Chłopicki is essentially saying, we don’t need a
theory of literary humor, if we have a theory of humor. As such, and
because it was the very first application of the Semantic Script Theory
of Humor to longer texts, it deserves a full treatment.
Chłopicki (1987) considers some Polish short stories. He sees the
major difference between a joke and a short story as one of length:
he begins by identifying and analyzing all the script oppositions in
the short story. This is the first difference between a joke and a short
story: many jokes have only one script opposition, but a short story
has tens of scripts oppositions (66, in one example). He finds that
some script extend across stretches of text and in some cases across
the entire text. He labels these “main scripts.” Some script oppositions
also stretch across large parts of the text or its entirety and are called

322
 14.1 SCRIPT-BASED THEORY OF HUMOROUS TEXTS

“shadow oppositions.” The shadow oppositions overlap with the main

scripts and thus Chłopicki considers to have successfully applied the
semantic script theory of humor to short stories. In the process of
producing this analysis, Chłopicki also adds to the list of basic opposi-
tions identified in Raskin (1985): the new oppositions Chłopicki adds
are absence/presence, necessary/unnecessary, and much/little
(Chłopicki, 1987, p. 18).
It should be noted that Chłopicki’s approach essentially “flattens”
(Attardo 1994, p. 210) the text of the short story onto a script opposition,
albeit a shadow opposition. What is missing is that the text of the short
story is developed over time, it has a linear aspect. So does the text of
a joke, but a joke is by definition short and thus issues involved to the
linearity of the text are ignored more easily. Let me add that Chłopicki’s
approach is not the only one to do this. For example, a Greimasian
semiotic analysis (Greimas, 1966) would flatten a text onto a semiotic
square (two oppositions).
The expansionist approach was so labeled in Attardo (1994) because
it expanded the reach of the semantic script theory of humor. Chłopicki
was not the only one to advocate expansionist ideas. Other proposals are
reviewed in Attardo (1994, pp. 221–222). More recently, Chłopicki has
developed an approach based on character frames (Chłopicki, 2017) in
which the readers builds macro-scripts accumulating all the pertinent
information about each character as it is developed. See Chłopicki
(2017) for a summary and examples.

14.1.3 Holcomb
Holcomb (1992) introduces the concept of humor “nodal points” which
are defined as follows: “Nodal points of humor are locations in the
narrative where humor is perceptibly more concentrated than in the
immediately surrounding text. Although they can be isolated as funny
instances in the story, the nodal points remain semantically tied to the
entire narrative” (p. 234). A nodal point, “will contain one or several
script oppositions – oppositions that occur within the node itself or that
involve an allusion to some other region in the narrative” (p. 236). The
first part of this definition is meant to reprise Raskin’s Semantic Script
Theory of Humor, while the second is meant to broaden it, by “find[ing]
a means of incorporating those instances into the whole of the narrative”
(p. 236). Ermida (2008, pp. 107–108) finds that Holcomb contribution

323
 HUMOR IN LITERATURE

is a significant one, but faults it for being vague on the specifics of how
to connect the nodal points to the rest of the narrative (p. 108).
Vagueness is indeed a problem: Holcomb lists a series of excerpts
from a Mark Twain story, in which violence is perpetrated on the
narrator, who reacts with unflappable calm. Holcomb then notes that
the excerpts are not jokes (which is to be expected since they are not self-
contained narratives) but “joke-like” constructions. He distinguishes
between “local” and “distant” script oppositions (p. 240) which are
however left undefined, except for the mention of “script oppositions
that are not mentioned explicitly in the quoted excerpts but must be
retrieved from other parts of the text” (p. 241), but no explanation is
given on how this takes place.
Holcomb generalizes thus:

At the beginning of each [story], several major scripts are evoked. These are held in
suspension as the discourse proceeds and then appear in opposition at nodal points,
those regions in the text where the humor seems especially concentrated. In the nodes,
the mere mention of any element of an already established script sufficiently evokes the
full script.

So, according to Holcomb, “major” oppositions occur at the beginning

of the narrative, whereas local ones in the nodal points. However,
while that may be true of the two stories he analyzes in his article,
it seems hardly likely that this pattern will hold true of all or most
humorous stories. It is definitely not true of the Lord Arthur Savile’s
Crime story, analyzed in Attardo (2001a), in which the main character
is not introduced until after 25 instances of humor having very little to
do with the main theme of the story. Despite the definitional problems
just listed (see also Attardo, 2001a, pp. 41–42), Holcomb’s nodal points
anticipate the idea of jab lines, without the technical detail. These will
be discussed below.

14.1.4 The revisionist approach

The reader will recall, from Chapter 7, that one of the purposes of
the General Theory of Verbal Humor was to extend the coverage of the
Semantic Script Theory of Humor to all texts, thus overcoming the
limitation of a theory of jokes. Our discussion of the revisionist approach
starts out with the General Theory of Verbal Humor which is directly
predicated on the assumption that different kinds of texts may have

324
 14.1 SCRIPT-BASED THEORY OF HUMOROUS TEXTS

different textual organizations, as captured by the Narrative Strategy

Knowledge Resource (Section 7.1.2). Attardo (2001a) presented a much
more elaborated application of the General Theory of Verbal Humor
to long texts, analyzing a variety of samples and in one case providing a
complete analysis of a short story by Oscar Wilde. In so doing, a number
of new concepts and tools were introduced, which will be reviewed
below.
14.1.4.1 Jab line and punch lines
Probably the most widely adopted distinction introduced in Attardo
(2001a) is that between punch lines, which end a text and in so doing
reveal the presence of a second script, forcing the reinterpretation of the
text itself, and jab lines which occur in the text in any other position (i.e.,
they do not occur at the end of the text) and do not disrupt the overall
interpretation of the text. Semantically speaking, jab lines and punch
lines do not differ: they both have script oppositions, so their difference
is purely due to their different functions within the text. Tsakona (2003)
showed that many jokes are compound jokes (Raskin 1985, p. 134), i.e.,
incorporate more than one script opposition, and that in those jokes that
are complex, the script oppositions that occur before the punch line are
jab lines and that these are unrelated to the final script opposition of the
punch line.
A long text such as a novel or a novella will not necessarily end in a
punch line, nor will any of its component parts (e.g., chapters). Some
texts do so (I mentioned Katherine Mansfield’s Feuille d’album short
story in Chapter 1) and I analyzed another example in Attardo, 1994)
but the vast majority does not. In fact, in most cases, humor is “super-
imposed” to an essentially serious narrative, that can be summarized
without any mention of humor.2 This led the way to considering the
position of humor in the text and its distribution.
14.1.4.2 Organization of humor in the text
Any text is a vector, i.e., is organized linearily and there is a direction
to the line (left to right in English). One can imagine a really long sheet
of paper in which the whole of War and Peace is written as one very

2 The 2001a version of the General Theory of Verbal Humor accounts both for the humor
in texts that are humorous overall and for the presence of humor in texts that are otherwise
serious. For example, Attardo (2001a) examines some onomastic jokes in Umberto Eco’s The
Name of the Rose, which is hardly a humorous novel, but nonetheless contains some humor.

325
 HUMOR IN LITERATURE

long line. This is a function of a very fundamental principle of language,

i.e., the linearity of the signifier. Already Saussure observed that one
can only produce one sign at a time (let’s say, the French phoneme /o/
(=eaux, waters)) and attendant prosodic marks. While the meaning of
an utterance may not be entirely linearily derived, the utterance itself
is necessarily so. Writing introduces complications, to be sure, but the
principle remains. We know from eye tracking research, that readers
do not by and large “peek ahead,” that fixations follow the linearity of
the printed line, and that any advanced awareness (both semantic and
formal) does not go beyond the limit of a line of printed text. Obviously,
in the case of a printed text, or with the recording of an oral utterance,
a hearer can go back and re-examine previous parts of the text, but it
remains that the way the text is first perceived is necessarily linear, i.e.,
one thing after another.
Once we consider this principle, it becomes clear that the organization
of the humor along the vector may be significant. Indeed, Attardo
(2001a) identifies several significant patterns: certain jab lines (or punch
lines) may be related because for example, they all have the same
Target (say, Lord Arthur Savile). If several (three or more) instances
of jab line are related either thematically, as in the Target example, or
formally (for example, because they all include a pun in the Language
knowledge resource), they are said to constitute a strand. Stacks are
strands of strands: for example, all the witticisms uttered by a character
in a sitcom may be a strand, since they will share the commonality of
having been uttered by the character. All groups of memes produced
from a single image also constitute a stack. Tsakona (2017) speaks of
“recontextualization” of the image or of other components of the meme.
Shifman et al. (2014, p. 41) defines memes as items “sharing content,
form and/or stance.”
Attardo (2001a) also introduced various specific configurations, such
as for two related jokes that occur at a certain distance (called in stand-
up comedy, a “call back”) and for repeated use of a given feature in a
restricted area of text. The terminology, bridges and combs, respectively,
was inspired by a visual representation introduced also in Attardo
(2001a). Despite having been put to very good use on some studies,
see for example Vincent (2010), the terminology has not been widely
accepted.
What all these structures have in common is the phenomenon of
repetition. Whereas in jokes repetition plays a small part, the repetition

326
 14.1 SCRIPT-BASED THEORY OF HUMOROUS TEXTS

of humor or of some of its features, as seen above, plays a significant

part in the structuring of longer humorous text (or in the humor within
longer texts). This is a significant difference between jokes and longer
texts.
14.1.4.3 The distribution of humor in the text
Beyond the way the humor is organized vis-à-vis other instances of
humor, it is possible to consider the overall distribution of the humor
in the text, regardless of its content. Corduas et al. (2008) applied
sophisticated statistical analysis, based on time series, to investigate the
null hypothesis of the distribution of humor in the text, i.e., that the
position of humor in the text of a short story was random. Their findings
indicated very clearly that the distribution of the humor is not random,
i.e., it has some underlying motivation. The statistical techniques and
the General Theory themselves cannot tell us what these factors or
motivations are. It is however, possible to guess that one possibility
is that more humor occurs at the beginning of the text to establish
the humorous frame of the text itself. In a sense, the author would be
signaling to the audience: this will be a funny text. Another possibility
is that the author modulates the amount of humor for esthetic effect.
Attardo (2001a, p. 89) dubbed “serious relief ” passages with few or no
humor in an otherwise humorous text (see below, Section 14.4).
Despite the promising and original results of the application of the
General Theory of Verbal Humor to long texts, it has been met with
rejection (e.g., Ermida, 2008, p. 109) or with lack of interest. This
is doubly disappointing, because the methodology presented affords
asking questions that wouldn’t even be intelligible in traditional literary
criticism (there is no conception of counting the amount of humor
in a humorous text, let alone analyzing its distribution) and the few
studies that have been conducted using the methodology yield good
results. Perhaps the fact that one of the goals of Attardo (2001a) was to
put the General Theory of Verbal Humor on a sound psycholinguistic
processing footing, with its inherent technical aspects, or perhaps simply
the labor-intensive nature of the analysis, have discouraged its use.
The revisionist approach and the expansionist share more than our
exposition might reveal. For example, both are based on the complete
analysis of the text and on locating and analyzing all the script oppo-
sitions in the text. In this they are directly opposed to the traditional
approaches that rely perforce on sampling based on the intuition of the

327
 HUMOR IN LITERATURE

analyst. When a literary scholar decides to focus his/her attention on a

given passage of the text, as opposed to another one, randomly chosen,
he/she does it on the basis of their intuition which tells them that that
particular passage is more meaningful or more significant, or simply
more appropriate for exemplification, than another one. Of course, this
practice introduces an element of subjectivity. Both the expansionist and
the revisionist approach, by analyzing the entire text are exempt from
this problem.
One may argue, as some actually have, that the actual analysis of the
script oppositions is still subjective. This charge is simply misguided
and completely ignores the difference between a semantic analysis,
constrained by very specific techniques and mechanisms, and the more
or less free for all of literary criticism. If one wanted to produce a Marxist
analysis of Lennon and McCartney’s Yellow Submarine one is only lim-
ited by one’s imagination: take the line “everyone of us has all we need.” Is
that because socialism has been realized, thus distributing the means of
production according to everyone’s need? Did the Beatles intend to tell
us that socialism would eventually fail in the USSR because they failed to
“live beneath the waves”? Well, why not? Realistically, probably because
one’s career in literary criticism would end therewith, but sillier ideas
have been proposed entirely seriously.3 Conversely, semantic analysis is
done in an extremely regimented way: one starts from the actual words
in the text, and works one’s way up (see Chapter 6). To claim that seman-
tic analysis is subjective is simply not to understand how it is done.4

14.2 Other approaches

We now turn to other approaches that, while informed by linguistics, are

more oriented toward the literary side. The best known such approach is
arguably Ermida (2008). Her book starts out with a broad review of the
3 My personal favorite: that analyzing Shakespeare’s works reveals that Shakespeare had
a Christ-like personality.
4 To be fair, since no full-blown automatic semantic analyzer actually exists, one really
has to do the analysis by hand, which is both time consuming and inevitably subjective,
but it is a different kind of subjectivity. One can compare it to the subjectivity of a parent
deciding how much allowance to give to their child and the subjectivity of an insurer pricing
an insurance policy. Both are subjective, but differently so: an insurer uses actuarial tables,
competitor’s rates, and the likes to determine a price, whereas parents essentially determine
their child’s allowance based on hunches and possibly the example of other parents. It
is obvious that an insurance company’s leeway in pricing is much more limited than the
parents’.
328
 14.2 OTHER APPROACHES

theories we have just surveyed. Ermida also addresses the narratolog-

ical approaches to literary/narrative texts (113–130) without, however,
bringing them to bear on humor.
Ermida (2008) contains some errors of interpretation: for exam-
ple, she equates bona-fide in Raskin with truthfulness (Ermida, 2008,
pp. 87–88), whereas one can violate any of the four maxims and be
non-bona-fine. However she provides a discussion of several proposal
dealing with literary humor, including, among others, a sympathetic
summary of Nash (1985) (Ermida, 2008, pp. 101–102) and one of
Chłopicki’s work (pp. 103–105).
A very strange (pp. 109–110) decision by Ermida is to declare that
texts that are not “(integrally) humorous” fall outside the purview of
a “model of humorous text” (p. 110). As far as I can reconstruct her
argument (presented on p. 172), the reason is that her model would
not work if it tried to integrate the humor that occurs in text that are
not entirely humorous. However, this decision leaves us with, at best, a
partial theory of humorous texts, since by definition it can only handle
a certain type of text and not others. This is a clear weakness.
As for Ermida’s central hypothesis, the following quote from
Chłopicki’s (2017) will suffice:

In her central hypothesis she puts forward five defining principles of humorous nar-
ratives: opposition, hierarchy, recurrence, informativeness, and cooperation (Ermida,
2008, p. 172). On the one hand, the theory seems to be too powerful and would apply to
many literary and non-literary narratives deprived of humor, while on the other hand
some humorous stories would not display all the principles she predicts.
(Chłopicki, 2017, p. 147)

Triezenberg (2008) introduces the concept of humor “enhancer”

which is defined as a “narrative technique [which] helps an audience
to understand that the text is supposed to be funny, that warms them
up to the author and to the text so that they will be more receptive
to humor, and that magnifies their experience of humor in the text”
(pp. 537–538). An example of what is intended is the use of legal jargon
in a joke about lawyers. The joke would function also without legal
terminology (assuming, for the sake of the argument, that the joke is
not predicated on the knowledge of some specific legal term) but adding
a few legal terms will make the joke a little funnier.
An interesting emerging trend is corpus-based or corpus-assisted
studies. For a broad discussion, see Partington (2017). Here, we will

329
 HUMOR IN LITERATURE

mention Duguid (2009), who studies a corpus of journalistic humorous

pieces and finds that phraseological units (idioms, constructions, and
other lexical “ready-made” units) are evoked as “an expectation raised
up only to be disappointed” (p. 312): “The unexpected shifting of fixed
or frozen wording of idioms creates a collocational shock (for instance
drunk as a sock, a rabble-soothing speech) just as the incongruous juxta-
position of register creates a shift in narrative” (p. 313). “Collocational
shock” (as the phenomenon gets dubbed) is not the only technique
found in the corpus, which includes also the use of hyperbole, irony, and
other figurative modes. Partington et al. (2013 , pp. 165–186), considers
the writing of P. G. Wodehouse.
The work by Burgers et al. (2011) on irony identification has been
discussed in Section 12.4.2. It is likely to yield significant results.

14.3 Narratology

The expansion of the General Theory of Verbal Humor is very clearly

extraneous to narratology (Attardo, 2001a, pp. 32–33) on two grounds:
1) it focuses deliberately and exclusively on humor; and, 2) it rejects
the intuitive nature of most narratological work. See the discussion of
cherry-picking of evidence, in Section 7.2.3, for the significance of this
point.
Larkin Galiñanes (2005) is a narratological look at humor. She claims
that humorous novels’ “typical narrative structure is heavily conditioned
by the sort of pragmatic processing attendant on the appreciation of
humor in jokes” (p. 80). Larkin Galiñanes (2005) further claims that
there is a “striking degree of coincidence” between Relevance theory
and Raskin’s Semantic Script Theory of Humor (p. 83) because she
characterizes humor as a mode that deliberately enhance the ambiguities
found in any text and thus she claims the two theories are equivalent.
Likewise, she reframes incongruity resolution as the search for rele-
vance. Regardless of the validity of such equations, which I suspect
would be rejected by both Relevance theorists and by Raskin, Larkin
Galiñanes (2005) presents some very interesting ideas.
First, she points out that the caricatural aspect of many characters in
comic novels functions along the same lines as the well-known ethnic

330
 14.3 NARRATOLOGY

stereotypes function in jokes: much like when we are told that a Scottish
character is involved we activate the mythical joke script for stingy,
or in her example love of whiskey, the characters in a comic novel
are not well-rounded and complex, i.e., they are caricatural, because
“the narrator’s descriptions, or the character’s own words, however
widely and frequently scattered over the text, all point in a specific
direction, generating a limited number of implicatures and making the
character easy to assimilate” (p. 89) since “If the character ‘scripts’ were
complex, the reader would find them more difficult to englobe and
grasp, an effect which would inhibit the creation of humor” (p. 89).
This is particularly interesting in that it militates against Chłopicki’s
most recent work (summarized in Chłopicki, 2017) which is dedi-
cated to creating complex scripts for each character (see Section 14.1.2
above).
Second, she describes what she calls “strong implicatures,” i.e., an
accumulation of connotative aspects of metaphorical constructs that
end up creating an overall impression or characterization of the indi-
vidual described. In her example, an old professor in Kingsley Amis’
Lucky Jim is described as changing his facial features like “a squadron of
slow old battleships” (p. 89) this giving, especially in concert with other
similar descriptors, the effect of “a slow-minded individual” (p. 90).
Larkin Galiñanes (2005) claims that an extended text at a macro-level
develops two opposed scripts

always potentially present in parallel, humorous effects (the equivalent to the punch
line) being caused at certain points in the plot when the two scripts are simultaneously
brought strongly into evidence. (pp. 98–99)

These points correspond to Holcomb’s (1992) nodal points and to

Attardo (2001a) jab lines.
Another work that incorporates narratological insights and uses the
General Theory of Verbal Humor, to some extent, is Hamilton (2013)
which analyzes medieval stories (fabliaux). A discussion of the narra-
tology of humor would be incomplete without mention of the extensive
discussion of the analysis of jokes in three narrative functions (Morin,
1966), promoted by numerous, mostly European, scholars. The discus-
sion is summarized in Attardo, 1994, pp. 85–91; see a short discussion
in Section 4.3.1.

331
 HUMOR IN LITERATURE

14.4 Stylistics

Stylistic approaches have been applied to humor. A broad review can be

found in Simpson & Bousfield (2017). The boundaries between stylistics
and other approaches to literary studies of humor are not always easy to
define. Under some definitions, it matches the entirety of this chapter.
Under this broad definition of stylistics, the entire domain of stylistics
is literary humor: Simpson & Bousfield (2017) make this very clear:

the first principle [of the stylistics of humor] is that humor requires some form of
stylistic incongruity. [ . . . ] The second principle is that the incongruity can be situated
in any layer of linguistic structure. (p. 159)

In this section, we will limit ourselves to a narrower scope for stylistics,

limited to issues of style.
An important contribution to stylistics comes from Cognitive
Linguistics and more specifically the work of Antonopoulou (2002;
Antonopoulou & Nikiforidou, 2009; Antonopoulou et al. 2015); for
more on Cognitive Linguistics and humor, see Brône et al. (2015). The
general idea is that in literary texts puns may not be the only source of
verbal humor. Authors may play with “the relative compositionality and
transparency of linguistic expressions which may be more or less fixed
for the ordinary language user, but allow inventive decomposition of
their fixedness” (Antonopoulou & Nikiforidou, 2009, pp. 305–306).
Antonopoulou (2002) utilizes concepts from cognitive grammar
(metaphors, constructions, etc.) as well as the GTVH, enriched by the
notion of idiomaticity, drawn from cognitive grammar, to explicate a
variety of instances of humor in Raymond Chandler’s (1888–1959) noir
novels. Idiomaticity is defined as

The conventional/idiomatic aspect of meaning of a construction is, therefore, what is

left unexplained after the contribution of the meaning of each word has been computed.
(Antonopoulou, 2002, p. 200)

Consider the example, discussed by Antonopoulou, of the construction

“What’s X doing Y?” The non-idiomatic meaning is the compositional
sum of the constituents: let us consider now an instance of the construc-
tion, where X = toothbrush and Y = the fridge:
(129) What’s your toothbrush doing in the fridge?

332
 14.4 ST YLISTICS

The compositional meaning of the construction is the sum of the mean-

ings of the lexical items, plus the syntactic information that the tooth-
brush is in the fridge and not vice versa, etc. The idiomatic/conventional
aspect is the “implication of disapproval” (Antonopoulou, 2002, p. 200)
of the construction.
The violation of idomaticity in puns and register humor was
previously known (see Attardo, 1994, for discussion); however
Antonopoulou significantly broadens the scope of applicability of this
facet of humor theory and at the same time proposes a revision of the
GTVH’s knowledge resource Language:

The GTVH acknowledges the importance of the Language KR in the case of puns. My
point here is that actual wording may play a decisive role in humour appreciation, not
only in cases of (lexical or syntactic) ambiguity, but also in cases of idiomaticity (as
described above). (Antonopoulou, 2002, p. 203)

The following example gives an idea of the subtlety of her argument:

(130) She was smoking a cigarette in a black holder that was not quite
as long as a rolled umbrella.
In example (130), the comparison is between a cigarette holder and an
umbrella. Normally, we would expect the comparison to be that the
cigarette holder is shorter than the umbrella, or vice versa, the umbrella
is longer than the cigarette holder. Our knowledge of umbrellas and
cigarette holders is such that we know that umbrellas are much longer
than cigarette holders or conversely that cigarette holders are much
shorter than umbrellas. By couching the simile as Chandler does, he asks
his readers to consider as plausible the fact that the cigarette holder is
almost as long as the umbrella, which is an exaggeration. However, by
couching it in the “not quite as long” construal, rather than the “slightly
shorter” alternative option, “the narrator pretends that they are in fact
so close in this respect that someone else might consider them identical
in length” (p. 203).
Another example is the “count-mass noun reversal” (p. 204) exempli-
fied as follows:
(131) Anna Halsey was about two hundred and forty pounds of
middle-aged putty-faced woman in a black tailor-made suit.
Here the non-idiomaticity of the construction is fairly easy to see, but
Antonopoulou provides again a sophisticated analysis, summed up as

333
 HUMOR IN LITERATURE

Although ( . . . ) [an] incongruity seems to be encoded in the second half of the sentence
starting with of, in fact its resolution depends on the reader realizing the reason
motivating the switch, i.e. that she is so fat that she is ?more mass than count?.
(p. 204)

Antonopoulou and Nikiforidou (2009) is focused on coercion

defined as

the clash between the syntactic and/or semantic properties of a word with those of the
construction in which the word is embedded and the principles that guide coherent,
consistent interpretations in such cases of conflict. (2009, p. 290)

For example, in (132) below

(132) It wasn’t that big an office, but the walls dripped modern art
(Ian Rankin, Set in Darkness, 2000, quoted in Antonopoulou &
Nikiforidou, 2009, p. 296)
the verb “drip” coerces a feature [+liquid] onto the normally [+solid]
noun “art”. One may argue that liquid art is conceivable, and it certainly
is, but it would not be hung on walls (or would need to be in solid
containers).
As Antonopoulou & Nikiforidou (2009, p. 300) note

coercion, by definition, foregrounds the linguistic incongruity per se, which may in turn
be exploited in the creation of a superordinate, text-global opposition.

So, in other words, coercion causes an incongruity and the repetition of

the same form of coercion in the text creates a strand (14.1.4.2).
Overall, Antonopoulou’s argument that the GTVH can be usefully
augmented by attention to the sort of linguistic manipulation such as
idiomaticity, coercion, and more broadly stylistic choices (Antonopoulu
et al., 2015) is a welcome corrective to what is an oversimplification in
the GTVH due to its origins in the analysis of jokes, as Antonopoulou
and Nikiforidou perceptively note (2009, p. 294, note 4). However, the
GTVH was not entirely uninterested in these matters, as the discussion
of register humor below will show.

14.4.1 Register humor

One of the most relevant topics in the stylistics of humor is register
humor. Register humor is created by an incongruity between the register

334
 14.4 ST YLISTICS

used to describe them and the facts, or between an expected register

and an observed one. Registers are “language varieties associated with
a given situation, role, or social aspect of the speakers’ experience”
(Attardo, 1994, p. 230). They are pretty standardly organized in high
and low varieties, with formal, sophisticated registers in the high strata
of the model and with informal, scatological, and other forms associated
with low-income, poorly-educated folk in the low strata.
Attardo (1994, pp. 262–265) examines several examples from
T. L. Peacock, a nineteenth-century humorist. These are often longish
stretches of text, as a register is usually identified by accretion and not by
a single lexical item of grammatical feature. However, an example such
as (133) should give a feel for the phenomenon. The set up of the scene
is that a certain character is exasperated and starts throwing things at
his servants; this is described as follows:
(133) by converting some newly unpacked article, such as a book, a
bottle, a ham, or a fiddle into a missile against the head of some
unfortunate servant (T. L. Peacock (1815), Headlong Hall, p. 6).
“converting [an object] into a missile” is obviously a high register variant
for “throw,” so it is precisely the discrepancy between the expected word
“throw” and the unexpected, non-idiomatic “convert into a missile” that
causes the incongruity.
The theory of register humor had been prefigured by Bally, one of
De Saussure’s students who edited the Cours de Linguistique Générale
(Bally, 1909). His approach, described in some detail in Attardo (1994,
pp. 231–235) has had little impact on the contemporary treatments
of the subject. Among contemporary scholars dealing with register
humor, the most significant is Alexander’s treatment (1997). Alexander
presents register humor as the “comical confusion” of two registers
defined as “selecting a lexeme or phraseological unit from a different
style level than the context would predict” (Alexander, 1984, p. 60). As
can be plainly seen, this fits perfectly the Peacock example above (133).
Haiman (1990, pp. 199–202) claims that register clashes are an indicator
of sarcasm. Attardo (1994, pp. 230–253) presents an approach based on
script-theory and spreading activation, a psycholinguistic phenomenon.
A more recent overview is found in Simpson & Bousfield (2017,
pp. 160–161). Related but different from register humor, in that it
involves two different linguistic varieties, is the practice of dialectal

335
 HUMOR IN LITERATURE

humor, i.e., the use of a dialectal variety of a language for the purpose
of humor (see Section 13.2.4).
Register humor tends to develop over relatively large stretches of
text in which individual features end up adding up to a strong enough
“marker” of the presence of a given register. Register humor is consid-
ered a diffuse disjunctor humorous phenomenon in Attardo (2001a). A
diffuse disjunctor is a disjunctor that is not necessarily embodied in a
single lexical item or syntactic feature. Example (95), in Chapter 9, is a
good example of diffuse disjunctor, albeit not a register-based one.
Attardo (2001a) introduced the idea of serious relief, i.e., that the
author of a long humorous text would introduce passages deliberately
less humorous for effect. In Attardo (2001b), some differences between
the use of adjectives in the serious relief passages are found:

the distribution between humorous parts and serious relief of some adjectives is clearly
not random. It seems to be the case that Wilde is deliberately, or if not deliberately,
unconsciously, marking the serious relief passage by using a different ‘palette’ of
adjectives. As we saw, not all adjectives are used in this marking function and in fact a
significant number of high frequency adjectives is used randomly across the text.
(p. 28)

14.5 Some examples of literary constructs

It should be made clear at the offset, that trying to summarize the

sophisticated, erudite, and stylistically gratifying works of the scholars
I will describe in this section is like trying to produce a dehydrated
version of your mother’s coq-au-vin. The ersatz version will not taste
anywhere as good as the original and much of the point of having it in
the first place will be lost. My point here is to show that high-quality,
theoretically sound studies that are not merely descriptive can be done,
if largely outside of linguistics.
Davis (2003; originally published in 1978, but the 2003 edition has a
very extensive introduction that throws significant light on the material)
is dedicated to the description of “farce.” According to Davis it “delights
in taboo-violation, but [ . . . ] avoids implied moral comment or social
criticism and [ . . . ] tends to debar empathy for its victims”; it treads
“a fine line between offence and entertainment [ . . . ] farce-plots tend
to be short and [ . . . ] peopled [ . . . ] by simplified comic types” (p. 2).
Farce favors “direct, visual, and physical jokes;” it is aggressive, violent,
politically conservative (p. 3). This sort of generalization over a vast

336
 14.5 SOME EXAMPLES OF LITERARY CONSTRUC TS

number of texts is a spectacular feat of scholarship, but linguistically

there is little that needs to be added, except that one would need a very
sophisticated theory of context and discourse to be able to account for
these categories in a work of linguistics. Likewise, the four plot types
Davis identifies:

• Humiliation or deception farce: linear narrative someone is humil-

iated or deceived
• Reversal farce: as above, but the victim retaliates
• Equilibrium farce: two opposing forces confront each other but
remain in balance
• Snowball or Circular farce: all the characters are caught into a
mechanism

are very abstract and clearly define most farces, but would be hard to
account for linguistically.
In the lack of strong moral stance, farce contrasts with satire, which
has an element of “censoriousness” (Condren, 2012, p. 378) combined
with humor. However, Condren argues that a general definition of satire
may be impossible, because the various varieties of satire produced
across the centuries and in different societies (the Greeks, the Romans,
Renaissance Italy, etc.) are so great that it becomes impractical to provide
such a definition. Ultimately the themes and the entities targeted can
provide us with some definition based on “family resemblances” (à
la Wittgenstein) (p. 386) but Condren notes immediately that such a
definition is insufficiently restrictive (p. 387). Finally, he comes to the
conclusion that an essentialist definition of satire may be impossible
or pointless (p. 390) and goes back to “moral seriousness” (p. 391) i.e.,
censoriousness.
Simpson (2003) presents a linguistically sophisticated discussion of
satire as a discursive genre (i.e., inclusive in the definition of discourse
features). The argument is too complex to describe in detail here, so
I will offer a “taste” of it, through a quote from Simpson’s own summary:

Whereas the model of satire postulates that both prime [the antecedent of the satire,
its object] and dialectic [the opposition, the antithesis] elements be present in a text, it
further stipulates that the lack of congruence between these elements be recognized by
a reader or listener, In other words, it is the dissonance between the domains of prime
and dialectic which creates an interpretive pragmatic framework for satire and brings
about the style-shift necessary to place the reader-listener on a satirical footing.
(Simpson, 2003, p. 10; emphasis in the original)

337
 HUMOR IN LITERATURE

The reader will recognize the familiar script opposition (the relationship
between prime and dialectic) and the incongruity of the connection
(“lack of congruence”). Simpson’s contribution is to place this dialec-
tic (to borrow the Hegelian terminology) in the discursive domain,
reminiscent (or at least compatible with) the Gumperz-Hymes eth-
nomethodological framework discussed in previous chapters.
Caputi (1978) represents a different type of study: while he does
identify and define a “genre” called buffo, which corresponds to “vulgar
comedy” (overlapping largely with farce), his focus is not on defining or
creating boundaries, but rather in showing the uninterrupted continuity
of transmission starting from the very early Greek comedies (Doric
mimes, satyr plays, and phlyax farces) all the way to contemporary
comics, clowns, and performers. Caputi’s erudition is remarkable and
his capacity to bring to bear to the argument sources of vastly different
domains, such as six or seven national literatures, literary criticism,
humor theory, anthropology, folklore, make his book worth reading
even if one is not particularly interested in the overall thesis.
Literary studies has its own methodologies, tools of analysis, and in
short “questions.” For example, the question of biographical criticism is
“how does the work reflect the life of the writer and vice versa”? The
question of gender-based theorists are along the lines of “how does this
work reflect and enact the patterns of gender dominance/identification
that are found elsewhere in the society”?
Most literary analyses of humorous texts simply ignore the fact that
the texts are or were meant to be humorous. While this claim may seem
prima facie outrageous and presumably false, it is in fact an accurate
description of the state of affairs in literary scholarship. Skowron (2003)
analyzed some literary analyses of Oscar Wilde’s humor and found not
only the above statement to be factual but that, in fact, many scholars
treated passages from, say, The Importance of Being Ernest as if they
were accurate representations of Wilde’s beliefs. What little attention was
given to the humorous nature of Wilde’s production consisted largely
in reproducing his bon mots and in platitudes about wit. Clearly there
are exceptions, and probably some humorists have been analyzed qua
humorists (Wodehouse comes to mind), but by and large it has to be
acknowledged that the initial claim remains true.
The above should not be interpreted as meaning that no analyses
of literary humor exist. That is far from true. For example, Swift’s
Modest Proposal is a very well-known text and many discussions have

338
 14.6 FURTHER READINGS

focused on its irony. However, it is fair to say that the analyses of Shake-
speare’s comedies, for example, focus more on what they reveal about
Shakespeare’s beliefs, Elizabethan ideology, etc. than on how funny they
are. In short, while literary texts that are funny have been analyzed, they
are analyzed because they are part of the literary production of an author.
Their being funny is often ignored, dismissed, or dealt with marginally.
This seems potentially very wrong, since we would expect that the
fact that the author wrote the text with the apparent intention to amuse,
entertain, or divert his/her audience should be significant to the under-
standing of the nature of functioning of the text itself (not to mention of
its relationships with the audience, etc.).

14.6 Further readings

There are no introductory works on the applications of linguistics to

humor in literature. Good but more advanced introductions to aspects
of humor and literature are Chłopicki (2017) and, on stylistics, Simpson
& Bousfield (2017). Chłopicki’s early work is summarized in Attardo
(1994) and in Ermida (2008). Attardo (2001a) is probably the best
resource for a full understanding of the application of the General
Theory of Verbal Humor to literary texts. On Cognitive Linguistics and
humor, see Brône et al. (2015).

339
 15
Humor and translation

15.1 A few definitions 343

15.2 Theories of humor translation 344
15.3 Audiovisual translation 355
15.4 Translating puns 359
15.5 Conclusions 365
15.6 Further readings 366

Speakers of other languages than the one in which a particular instance

of humor was created may want to enjoy it too. Enter translation.
Jokes, humorous texts, films, cartoons, advertising, and all sorts of other
humorous materials (such as video games, cf. Iaia, 2014) are frequently
conveyed in one or more languages different from the one in which they
originated.
This chapter deals with the translation of humor. The magnitude of
the phenomenon of the translation of humor is hard to assess. It is safe
to say that a lot of humor gets routinely translated: consider American
and British sitcoms, often broadcast, and hence translated, world-wide.
I have found an unsubstantiated claims that Friends was broadcast in
more than 100 countries1 and the French cartoon Astérix translated
into more than 110 languages.2 Regardless of the details, this is a signif-
icant issue that affects millions of people and a multi-million business.
The translation of humorous literature is likewise a huge segment in the

1 Wikipedia: en.wikipedia.org/w/index.php?title=Friends&oldid=164835768
2 The page is constantly updated. As of August 2019, the count stood at 116 languages
and dialects: www.asterix-obelix.nl

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 HUMOR AND TRANSLATION

publishing industry. Bucaria and Chiaro (2007) claim that on a week of

programming in Italy in 2005, 562 hours of dubbed programming was
broadcast (presumably not all humorous).
This chapter addresses only some of the aspects of the translation of
humor, out of necessity. An entire book could not hope to address all the
issues in the field, let alone a mere chapter. Nonetheless, a wide range
of issues will be considered including some basic tenets of translation
theory, audiovisual translation (subtitling, dubbing, captioning, etc.),
translation from one semiotic system to another, and much more. Not
all issues can be addressed. For example, all the previous chapters, and
in fact most of the literature on humor, have implicitly assumed that
humor comes in one language. Needless to say, this is not the case. While
there obviously are monolingual texts, many texts use more than one
language; large texts are often multilingual. We will not be concerned
here with humorous texts that exploit multilingualism for humorous
purposes, but see Valdeón (2005) for a discussion of multilingualism
in the sitcom Frasier and the articles in the special issue of the European
Journal of Humour Research guest-edited by Dore (2019).
I would also like to stress that I am concerned here with the theoretical
issues concerning the translation of humor. By this, I simply mean that
I will not consider the practical issues faced by translators. For example,
many people professionally engaged in translation have told me that
they are chronically underpaid. I am sure this is true and it is a serious
issue well worth discussing, but a book about the linguistics of humor is
not the proper place for such a discussion. I do not wish to make light of
the issue, and of the many others that translators face when exercising
their profession: for example, it is often true that the deadlines for the
production of the translation are unrealistically short. This is especially
true in subtitling and dubbing and particularly so in interpreting (see
Section 15.3 below). It is often also the case that the translator does not
have the time and/or energy to dedicate to finding an optimal solution
due to overwork/overload, which forces him/her to “settle” for a given
translation and stop searching for a better one. The astute reader will
have recognized the competence/performance distinction once more:
we are concerned with the competence of translation (i.e., an idealized
situation in which time, energy, cost, etc. constraints are not relevant)
and not with the performance of translation (i.e., the actual practice that
actual translators engage in actual situations and subject to time, cost,
energy, etc. limitations).

341
 HUMOR AND TRANSLATION

Another observation that Martinez Sierra and Zabalbeascoa (2017,

p. 16) make is worth stressing: the field of translation studies has shifted
its focus from textual (or verbal) humor to multimodal, audiovisual3
humor. Here the distinction, popularized by Chiaro (2005), between
verbally expressed humor and humor expressed by non-verbal means
(i.e., visual, auditory, musical, etc.)4 is worth remembering: in a multi-
modal event, such as a film or sitcom, but also a play or a conversation
between two people, there are many “channels” active at the same time:
people or characters say something (verbal channel) but at the same time
they may grimace (visual) or other things may happen on stage or in
the visual field (or even in a different location, in a split-screen shot),
or the soundtrack may be incongruous. For example, a recent (June 4,
2017) video on YouTube shows members of the English Defence League,
an Islamophobic movement (Wikipedia https://en.wikipedia.org/wiki/
English_Defence_League), being “chased out of Liverpool to the Benny
Hill tune” (as the caption of the video reads). Indeed, one can clearly hear
the music, and several people in the video laugh and dance to it. It is not
clear what the source of the music is. The YouTube caption reads also
“Hilarious!” Thus it is clear that, while there is no discernible verbally
expressed humor, the situation and the video are humorous (at least to
some participants) due to the juxtaposition of the situation (a political
rally) and the music.
Here the reader will do well to recall that while linguistic theories
in their formulation do not concern themselves with extra-linguistic
matters, their underlying principles apply to all semiotic systems (see
Chapter 5) and hence to multimodal and audiovisual texts. It is one
thing to eliminate some aspects of a phenomenon to narrow down one’s
purview to simplify the analysis, as linguistics has done, and it is another
to be unable to handle a given subject, in principle. In other words,
linguistics has chosen to simplify its analyses by limiting itself to the
verbal channel, but there is no reason in principle, that it could not
handle other channels (and it has done so, on occasion). While from the
standpoint of the theory of translation, and even from the standpoint of
some approaches to linguistics, it may make perfect sense to distinguish
between humor in the verbal channel and humor that is not verbalized,

3 I am here using multimodal and audiovisual as synonyms. Needless to say, they are not
the same thing.
4 The earliest mention of verbally expressed humor I am aware of is Eastmond (1992);
see also Ritchie (2000).

342
 15.1 A FEW DEFINITIONS

from a semiotic perspective the distinction is unnecessary and even

confusing. The text as a whole is humorous and all its parts need
to be taken into account. To put it differently: it is a perfectly fine
methodological choice to limit one’s analysis to a given channel (verbal
or visual or music, or montage, etc.) but that does not entail that the
other channels are unimportant or cannot be analyzed (separately or in
conjunction with the verbal). Finally, there is potential confusion here
between verbally expressed humor (as in linguistically expressed) and
verbal humor (as in humor that relies of the form of the signifier, puns
for short; see Chapter 9). The two should not be confused.
The field of translation studies of humor is very active, as compared to
other fields of the linguistics of humor. Thus it is somewhat surprising to
see a recent survey of the field (Martinez Sierra and Zabalbeascoa, 2017,
p. 13) describe the literature as “scarce.” However, the assessment makes
perfect sense when one notes that translation studies is a very frag-
mented field, and that a significant part of the literature falls on the “art”
of translation, rather than on the “science” side (Steiner, 1975, p. 295).
We will only be concerned with scientific approaches to translation, so
we will more or less ignore the plethora of studies about translating an
individual author or work in any given pair of languages.5

15.1 A few definitions

15.1.1 Source text and target text language

It is common in the field of translation studies to refer to the text that
is translated as the source text (ST) and the resulting translation as the
target text (TT). Correspondingly, the language in which the source text
is worded is referred to as the source language (SL) and the language
of the target text as the target language (TL). As we will see, translation
may take place within varieties of a language, for example, dialects, but
also high and low forms of a language (formal vs. colloquial varieties).
Regardless, we will use the terms source/target text and language in all
cases of translation. In fact, the picture is even more complex.

5 Chiaro (2005, p. 136) rightly reminds us that the translation of humor is not only, or
even not primarily, a translation between languages as it is between cultures. We will not
address this issue in this context, but see Davies (2005).

343
 HUMOR AND TRANSLATION

15.1.2 Intra- and inter-semiotic translation

Following Jakobson (1987; originally published in 1960) we distinguish
three types of translations:

• intralingual (paraphrase): translating within one language. An

example would be paraphrasing or rewording a sentence, for
example to make it easier to understand.
• interlinguistic: translating from one language (i.e., the SL) to
another (TL). An example would be translating an English play
into French.
• intersemiotic: translating from one semiotic system to another. An
example, would be the film adaptation of a book (Jakobson, 1987,
p. 429).

Usually, when we speak of translation, without further qualifying, we

mean inter-linguistic translation. We adopt this usage in this book as
well. We will also use the term source/target text regardless of the
semiotic system in which it is expressed. For further discussion of the
distinction, see Chiaro (2017).

15.2 Theories of humor translation

The Italian saying traduttore, traditore is a fitting place to start our

discussion. But of what is the translator a traitor, precisely? The glib
answer is “meaning.” However, meaning is about the only thing that
a skilled translator is pretty much guaranteed to be able to transpose
from one language into another. Another suggestive idea is the title of
Umberto Eco’s collection of essays on translation (2003b): “dire quasi la
stessa cosa” (saying almost the same thing), with the emphasis on almost.
Again, the question boils down to the nature of this “almost”: what gets
lost or replaced? This section then discusses the question of what gets
translated.

15.2.1 Faithfulness
Roman Jakobson points out that meaning itself is only understood
through translation, as the meaning of a sign is simply its translation

344
 15.2 THEORIES OF HUMOR TRANSLATION

in other signs, i.e., it exists only within the system of the language, as
claimed by Saussure (1916) and Peirce (1960). Translation then reduces
to equivalence: if a source text sign can be replaced with a target sign
that is equivalent (i.e., has he same value) we have an ideal translation.
However, since languages, varieties and semiotic systems are different,
their signs have different values and hence one needs to substitute
different signs. As Jakobson puts it, “equivalence in difference is the
cardinal problem of language” (1987, p. 430). Jakobson also states some
basic points: anything that can be said in any language can be translated
into any other. It may be hard to do so, as languages differ “in what
they must convey and not in what they can convey” (Jakobson, 1987,
p. 433; emphasis in the original) but it can ultimately be done. Poetry
is impossible to fully translate, only “creative transposition” is possible
(p. 434), and this is especially valid for the pun, in which “Phonemic
similarity is sensed as semantic relationship” (p. 434). Summing up,
for Jakobson, translation ideally should reproduce the value of the
signs in the source text or the target text. Since this is impossible, as
languages differ, equivalence must be sought, while staying as close
as possible (faithfulness) to the source text. In cases in which the
signifier is involved (such as poetry and puns), since by definition the
signifier will not be translatable in the target text, translation is strictly
speaking impossible, but transposition (functional translation) should
be sought.
Low (2011, p. 60) notes that within a “verbal process” view of transla-
tion “translating a joke means creating an amusing target text (TT) that
is nearly identical to the source text (ST).” However, Low himself, states
that such a goal is “unreasonable” and that the translator should instead
aim at delivering “broadly speaking, the same joke” (p. 60). This brings
to mind, whether Low intended this or not, the GTVH, whose stated
purpose is to determine when two jokes are the same (and when they
are not).

15.2.1.1 The GTVH as a tool to evaluate difference

Attardo (2002a) takes as his starting point Jakobson’s equivalence
and proposes using the GTVH’s built-in metric of difference between
humorous texts to determine how different a source text is from its
translation, humor-wise. This is stated, in a nod to Grice’s maxims, as
another maxim:

345
 HUMOR AND TRANSLATION

if possible, respect all six Knowledge Resources in your translation, but if necessary, let
your translation differ at the lowest level necessary for your pragmatic purposes.

No prescriptive intention is present, just as no prescriptive intention

is meant in Grice’s maxims. Essentially, the idea is that, if the translator
determines that he/she needs to reproduce the humor of the original ST
as faithfully as possible, then the rational (most efficient) way to do so is
to follow the maxim above. Obviously, if the translator determines that
the best choice is not to reproduce the humor (faithfully, or at all) then
the maxim is of no use to him/her. The method is applied to translation
as a set of binary choices in Zabalbeascoa (2005, pp. 202–204). Accord-
ing to an empirical study by Williamson & Ricoy (2014) translators do
follow the proposed maxim, to some extent (p. 186). Shipley (2006)
adopts the proposal and supplements it with a checklist of external
factors to be considered as well. Asimakoulas (2004) also applies the
General Theory of Verbal Humor to the translation of humorous films
in Greek.

15.2.2 Literal vs. functional translation

Raphaelson-West’s (1989) approach is based in Nida and Taber (1969).
The focus is on semantics first and style second. As Nida and Taber put it,
“anything that can be said in one language can be said in another, unless
the form is an essential element of the message” (1969, p. 4). Implicit in
this definition is that puns and poetry are untranslatable in principle.
Raphaelson-West (1989) is more optimistic and notes that linguistic
jokes are “most difficult to translate” (p. 131).
Raphaelson-West (1989) introduced a tripartite taxonomy of jokes,
which became very influential in the field, distinguishing:

1. Puns: their translatability is “very unlikely” (p. 130)

2. Cultural (ethnic jokes): more translatable, but the two cultures
must share the stereotype
3. Universal jokes: jokes based on universal semantic features, such
as child vs adult, or the unexpected (p. 131)

Raphaelson-West (1989, p. 131) notes that the existence of universal

jokes is questionable, but likely. Universal jokes correspond to
Zabalbeascoa’s (1996b, p. 251 ) “international joke [in which] ( . . . ) the

346
 15.2 THEORIES OF HUMOR TRANSLATION

comic effect does not depend on either language-specific wordplay or

familiarity with unknown specific aspects the source culture.”
As Chiaro (2017, p. 420) notes, translation scholars have proposed
several dichotomies revolving around the idea of equivalence: e.g.,
“formal vs. dynamic/functional” (Nida, 1964). Functional translation, in
the case of humor, boils down to reproducing the perlocutionary goals of
the ST. This has been clearly stated, for example, by Fuentes Luque (2004,
p. 77) as “conseguir en el texto término el mismo efecto humorístico
(en forma de sonrisa o risa) en la cultura y la lengua término que en
las originales” [to achieve in the TT the same humorous effect (in the
form of smile or laughter) in the target culture and language than in the
originals; my translation SA]
Chiaro (2017, p. 421) argues that for theorists such as Toury (1985)
and Vermeer (1989; see below), who reduce equivalence to function,
the translator may “insert a completely new instance of verbal humor”
to replace an otherwise untranslatable joke. Nida and Taber’s own def-
inition of dynamic translation, later changed to functional, allows this
interpretation, since it is based on the idea of “response” of the audience
(see Fuentes Luque’s point above).

15.2.3 Zabalbeascoa’s priority scales and solution types

Zabalbeascoa locates the practice of translation squarely within per-
formance, which he calls “contingency” (2005, p. 186). He identifies
ten “most obvious” factors that affect the context in which translation
is performed. These include the languages and cultures involved, the
medium of the texts, the purpose of the translation, the nature of the text,
the intended recipients, the sponsors of the translation, the translator’s
skills and limitations, and the conditions in which the translation is
performed.
Because of this significant level of contextual variability, Zabalbeascoa
advocates for the practice of mapping and prioritizing before translating.
Mapping consists of locating in the text significant items (e.g., the
instances of humor; Zabalbeascoa, 2005, p. 187). On the problem of
identifying humor, see Section 12.4.2. Prioritizing is more complex.
Zabalbeascoa argues that each translation has a set of priorities and
restrictions. The priorities are the goals of the translation and the
restrictions are the “obstacles and problems” that motivate the prior-
ities and “the solutions adopted in the translation” (1996b, p. 243).

347
 HUMOR AND TRANSLATION

Furthermore, Zabalbeascoa argues that the priorities are arranged on

a scale of importance (p. 243), so that a top priority is also a restriction
for the priorities below it. Thus different types of texts will have different
priorities for humor. A TV comedy will place a higher priority on humor
than a TV quiz show or a drama (p. 244).
A second scale is that of global/local relevance to the text. According
to this scale, humor may be a local priority in the text (i.e., there is a
joke in the ST), but a low priority globally (i.e., the goal of the text is
not merely to have that particular instance of humor) and hence it may
be omitted or substituted with another instance of humor elsewhere.
After a priority scale and a global/local scale, the third scale of priorities
is that of equivalence. It may range from “absolute identity” to “slight
resemblance” (p. 247). So equivalence may or may not be a high priority,
locally or globally, in the text.
As an example of a set of priorities for translating a sitcom, Zabalbeas-
coa mentions the following:

do well in popularity ratings, be funny, aim for immediate response in the form
of entertainment and laughter, integrate the words of the translation with the other
constituent parts of the audiovisual text, or use language and textual structures deemed
appropriate to the channel of communication. (p. 245)

We are now in the position to be able to understand Zabalbeascoa’s

sophisticated argument:

Translating comedy in order to produce comedy entails that intended comic effect is a
priority that is both very high on the scale of importance and a global one, i.e. relevant
to the text as a whole. It is moreover an equivalence priority, requiring near-absolute
identity. The insistence on the word intended means that equivalence is here seen as a
characteristic of an intention to be funny, regardless of the final outcome. What matters
in this case is the perception of the source text’s humour as a basis for the decision to
make the translation a humorous text. The translation can then be judged according
to exactly how funny it is in its own right. From this perspective, there is little point in
comparing source and target texts in terms of the exact amount and type of humour
they contain; if anything, it would be desirable for the translation to be even funnier
than the source text. (p. 247)

In other words, since the overall purpose of the text (“intended comic
effect”) is high on the scale of priority and on the global side of the
scale, it is crucial to be faithful to that goal (“near-absolute identity”),
i.e., the TT should be above all funny. Conversely, local equivalence is
a low priority, to the point that, as Zabalbeascoa implies, a translator

348
 15.2 THEORIES OF HUMOR TRANSLATION

may choose to add humor to the TT, if there is an opportunity for

it. Zabalbeascoa continues to point out that there may be cases in
which non-equivalence is the goal (for example, a pedagogical setting
in which the linguistic features of the text are significant, and thus
it would be unadvisable to replace the joke with a different, albeit
funny, joke).

15.2.3.1 Solution types

Zabalbeascoa (2000) sees the translation process as determined by
methods, strategies, solutions, and solution types. Translation methods
are identified based on “the formal and functional characteristics of
a TT and its ST” (2000, p. 119). Examples of methods are word-by-
word translation and adaptation. Strategies are defined as “conscious
action(s) intended to enhance a translator’s performance” (p. 120).
Strategies include, for example, consulting dictionaries, writing a
draft, etc. Solutions are then the results of strategies, i.e., concrete
instances of TT arrived at from a ST. Translation techniques consist of
parsing the ST into the “smallest constituents parts ( . . . ) for the purpose
of translation ( . . . ) and then considering the most convenient way
(technique) of rendering each unit” (p. 122). Zabalbeascoa notes that
the implicit measure of the solution types is against the ideal of literal
translation (p. 122) and later that it relies on the notion of “equivalence”
(p. 123). A solution type is thus defined as “the shared characteristic of a
number of different solutions” (p. 122). The results of Díaz Perez’ (2017)
study show that the solution-types most frequently used are those based
on the source culture (p. 72).

15.2.3.2 Translation techniques

A certain consensus is found concerning translation techniques. Fuentes
Luque (1998) sums up four techniques for the translation of humor:

1. Addition or compensation: inserting a joke where none was

present to compensate for not translating another one;
2. Substitution: replacing a joke with another one (presumably easier
to incorporate in the TT)
3. Metalinguistic commentary: explanations, paraphrases, footnotes,
etc.
4. Omission: uncompensated lack of translation (1998, p. 667)

349
 HUMOR AND TRANSLATION

The proposal has been widely adopted: Chiaro (2005, 2010); Fuentes
Luque (2010); Low (2011), who adds to the list a “dilution” strategy
(translating only a few puns in a passage rife with them) and “explici-
tation” (i.e., a partial explanation of the otherwise untranslatable joke).
Techniques for translating puns will be discussed in Section 15.4
below. Cuéllar Irala and Garcia-Falces (2004), in the brilliantly titled
article: “Cultura y humor: traductores al borde de un ataque de nervios”
[Culture and Humor: Translators on the verge of a nervous breakdown]
enumerate several strategies for translating cultural references that may
not always exist in both cultures:

1. Maintain the reference (e.g., reference to Starbucks, which will be

understood equally well in Spanish or in English).
2. Explicitation: explain the reference.
3. Generalization (hyperonymy): instead of using a specific refer-
ence use an hyperonym (for example, instead of “Ford F150” use
“pickup truck”).
4. Adaptation or naturalization: use a referent better known in the
TL. This has three sub-cases, according to Cuéllar Irala and García-
Falces (2004); it is unclear if there is a real difference between them:
(a) Change the referent to retain connotations;
(b) Naturalization due to the desire to avoid references extraneous
to the TL (example: a reference to a rabbi is translated as “den-
tist” to avoid bringing in extraneous religious connotations);
(c) Naturalization with an accepted cultural reference to create a
more natural TT (as above, but with a purpose of creating a
more familiar TT).
5. Over-translation: “in its most radical cases ( . . . ) the result is a
completely free rewriting, independent from the original” (Cuéllar
Irala & García-Falces, 2004, p. 14; my translation, SA). Cuéllar Irala
and García-Falces attribute to commercialization this “strategy.”
López González (2017, p. 290) notes that Cuéllar Irala and García-
Falces are “quite critical” of this “translation method.”

López González (2017) finds that in a corpus of 14 DreamWorks

movies, literal translation is the dominant strategy, which she attributes,
credibly, to the dominant position of US culture vis-à-vis Spanish.
Another possibility is that the movies are deliberately built using
widespread, common referents that will easily translate globally.

350
 15.2 THEORIES OF HUMOR TRANSLATION

15.2.4 Eco’s translation-as-negotiation

Eco (2003a/b) presents a negotiation-based view of translation. Eco
notes that the translator must determine the intention of the text (i.e.,
what is the text trying to accomplish, or as Eco has it: “the effect aimed
at” (Eco, 2003a, p. 56)) and translate based on that. However, as Eco
points out, “Many hypotheses can be made about the intention of a text,
so that the decision about what a translation should reproduce becomes
negotiable” (Eco, 2003, p. 56).
By “negotiable,” Eco means the kind of intersubjective agreement
whereby two individual speakers agree that a given small animal they
both see is a mouse, despite the fact that one of them may be a biologist
who knows that the given mouse is a mus spretus and not the more
common mus musculus domesticus. The point is that, for the purposes
of the exchange, it is not necessary to distinguish between the two.
For Eco, translating is never or at least rarely absolutely lossless. Much
like in a negotiation both sides need to give something up to reach
an agreement, the translator has to make choices (an “interpretative
hypothesis,” 2003a, p. 56) and give something up to achieve the intention
of the text in the new setting (i.e., the new language in which it is
translated).
Eco’s remarks on translation are not a full-fledged theory. They are
more like observations (the Italian subtitle has “experiences of trans-
lation”) that are connected, post hoc, by a theory of translation based,
at least in part, on skopos theory. Eco discusses many example of
translation, many from translators of his own books, which is very inter-
esting for the insiders’ viewpoint, documenting precisely the (literal)
negotiations between the author and the translators. Eco also addresses
the translatability of puns, see more discussion below (Section 15.4).

15.2.5 Skopos theory

Skopos theory originates in the work of Hans Vermeer (1989; Reiss and
Vermeer, 2014). As evident from the name, it foregrounds the purpose
or goal (skopos) of the text, rather than the equivalence between source
text and target text. Texts have purposes, they are written or spoken for
a reason. According to skopos theory, translation is best achieved when
the target text is adequate to the function or the purpose required by
the recipients of the target text or established by the person or entity

351
 HUMOR AND TRANSLATION

that commissioned the translation. One of the main contributions of

skopos theory has indeed been to help differentiate the notions of author
vs. translator of the text, by introducing the idea of the client who
commissions the translation, and by focusing on the audience from
whom the translation is meant.
When the skopos of the source and target text coincide we have
“functional constancy” (“Funktionskonstanz,” Vermeer, 2004, p. 233).
In the case of humor, then since the main perlocutionary goal of a
humorous text is to elicit mirth, skopos theory predicts that a good
translation will be one that elicits mirth in the target language audience.
Nord, a promiment proponent of skopos theory, puts it very clearly:
“if the purpose is to amuse, then the text should actually make its readers
laugh or at least smile” (Nord, 2006, p. 31). Chiaro (2017, p. 421) puts it
similarly, “In the case of humor, as long as recipients recognize the text
as being non-serious in scope, then its Skopos, its purpose, has been
achieved.” In general, there is little to argue with. Indeed, the definition
of humor as having a perlocutionary goal (in Austin’s 1962 sense) of
eliciting (the recognition of the intention to elicit) mirth and thereby
laughter/smiling is fundamentally correct.
However, this formulation seems too restrictive to me. Vermeer makes
it clear that “the skopos theory merely states that the translator should
be aware that some goal exists, and that any given goal is only one
among many possible ones” (Vermeer, 1989, p. 234). For example, one
may tell a joke with the intention to offend one of the hearers or
to show off one’s knowledge. In that case, recognition of non-serious
intention or elicitation of mirth are no longer the skopos of the joke
and hence of its translation. According to Vermeer then, jokes may
involve “reformulation” (2004, p. 229), i.e., the ST and TT may differ
from each other significantly, because the target culture may “verbalize
a different phenomenon in a different way” (p. 229). To put it differently,
the translator must choose a skopos and formulate a text, humorous or
not, that achieves that skopos.
This raises an interesting question, that skopos theory has not
addressed, to the best of my knowledge: each text has multiple skopoi
that coexists in the text. Thus for example, a joke about fellatio
(see Chapter 10, Sacks’ 1974 joke) achieves several different skopoi:
establishing non-serious mode, showing off one’s knowledge of sexual
practices (a significant goal if one is a teenager), showing that one is
comfortable dealing with sexual matters, amusing one’s audience, etc.

352
 15.2 THEORIES OF HUMOR TRANSLATION

So, if the translator decides to replace the joke (for example because it
involves a pun) with another joke that does not involve sexual practices,
while some skopoi may be maintained (non-serious mode, amusement,
etc.) other skopoi will be neglected (showing off one’s sexual knowledge,
showing that one is comfortable dealing with sexual matters, etc.). This
is far from a trivial objection, because this line of reasoning reintroduces
the need for faithfulness, if no longer to the ST, but to the skopoi of the
ST: if the ST contains a joke, we have to assume that the ST had a
reason, a skopos, for that joke to be there. So, the only way to guarantee
“constancy” between the ST and TT is to reproduce the skopoi of the
ST in the TT. Zabalbeascoa also points out (2005, p. 244) that different
functions of humor (entertainment, social criticism, moralizing, etc.)
need to be accounted for. See also Martin de León (2008, pp. 13–14) for
the same criticism of skopos theory, in a general context.
Variants of skopos theory are widely accepted in the translation of
humor. Popa (2005) is an application of skopos theory to the translation
of jokes, based on a bilingual corpus of Romenian and English jokes.
She notes that a perfectly literal translation between two referential jokes
does not achieve functional equivalence (“constancy” p. 50) because the
functions of the jokes in the SL is different from that in the TL. She
further discusses the difficulty of achieving equivalence while respecting
the genre and register variables of the respective languages. Amirian &
Dameneh (2014) is an application to Persian. The study of humor trans-
lation using skopos theory seems very popular in China. Han (2011)
documents 16 studies on humor translation. His thesis is that skopos
theory broadens what counts as “translation” and thus allows more
translations. Yin and Wu (2014) document that Chinese translators of
the Big Bang Theory try to respect the ST humor in the subtitles.

15.2.6 Relevance Theoretic approaches

Translation, from the standpoint of Relevance Theory, can be defined as
follows:

Translation, from a cognitive pragmatics perspective, can be explained as an inferential

gap-filling activity in which the translator has to infer the intended interpretation,
context accessibility and predictions of mutuality between the source-language com-
municator and the source-language addressee, all that framed in the source- language
culture, and then transfer all this information to a target audience with a different

353
 HUMOR AND TRANSLATION

language and a more or less different way of coding information, and possibly different
social values, norms and stereotypes. (Yus, 2013, p. 125)

Relevance Theory then squarely places translation in the inferential

domain and is in this regard even more radical than skopos theory in
moving away from linguistic similarity as the criterion for translation.
Yus (2013) concludes that there are three types of joke translations:

1. transferrable jokes, i.e., jokes that can be translated easily because

they rely on shared stereotypes or “linguistic strategies for humour
generation” that are common to both languages
2. replaceable jokes, i.e., jokes for which, in the absence of correspon-
dences, “alternatives can be found in the target language achieving
similar balances of cognitive effects and mental effort” and finally,
3. challenging jokes, i.e., jokes that are difficult to translate because
they rely on “very specific intra-cultural referents, [and] linguistic
resources that have no counterpart in the target language”

somewhat unexpectedly, given the radically different theoretical starting

points, supporting Raphaelson-West’s (1989) conclusions seen above.
Díaz Pérez (2017) notes that translation, from a Relevance Theoretic
standpoint, is an interpretive use of language, similar in this to quoting
or reported speech, rather than an “interpretive” use (i.e., when lan-
guage is used to say something about the world). In the interpretive
use of language there is a “relation of interpretive resemblance” (p.
52) between, say, what was originally said and its reported form. Díaz
Pérez concludes then that the relationship between the ST and the TT
is one of “interpretive resemblance.” (p. 53). Furthermore, translation
involves a meta-representation (since it is the use of language about
language).
Díaz Pérez discusses culture-specific items (i.e., items that are not
shared by both languages). Translation strategies (“solutions”) can be
arranged on a continuum between exoticism (which favors the source
culture), as in the famous “kugel” joke quoted by Raskin (1985, p. 215;
originally from Novak and Waldoks, 1981, p. 7) in which the word
“kugel” is not translated at all,
(134) Somebody once asked Motke Chabad, the legendary wit: “Tell
me, Motke, you’re a smart fellow. Why is kugel called kugel?”

354
 15.3 AUDIOVISUAL TRANSLATION

Motke lost no time in responding. “What kind of silly question

is that? It’s sweet like kugel, isn’t it? It’s thick like kugel, isn’t it?
And it tastes like kugel, doesn’t it? So why shouldn’t it be called
kugel?
and, on the other end of the continuum, cultural transplantation, which
favors the target culture, as in the case (Ferrari, 2011) of the dubbing
of Fran Drescher’s Jewish-American accent into an Italian-American
accent for Italian audiences, which completely erases the Jewish identity
of the character, aided by the remarkable similarities between Jewish
and Italian-American stereotypes. A different example, reported by
Zabalbeascoa (1996a, p. 249) concerns a character in the show Fawlty
Towers, Manuel, whose recurring misunderstandings are attributed to
his being from Barcelona. In the Catalan dubbing, Manuel comes from
Mexico.
Díaz Pérez finds that in his corpus (two seasons of a sitcom), there is
a tendency for translators to gravitate toward the source culture (p. 72)
and he speculates that this may be due to the desire on the translator
part to avoid alienating the audience who is aware of the foreign nature
of the ST. This is also the conclusion reached by Martìnez-Sierra (2006).
who fears that an over familiarized translation (i.e., adapted to the
target culture) may be perceived as “artificial” (pp. 221–222). Already
Zabalbeascoa (1993, p. 263) had warned the translator against excessive
adaptation toward the target culture in an example involving town
names in England which would not be familiar to a Spanish speaker.

15.3 Audiovisual translation

If translation were not an issue complex enough, the presence of mul-

timodal texts, such as films and TV series, but also YouTube videos,
animated GIFs, etc. entails the possibility and in some cases necessity of
translating texts that contain both a linguistic and a visual component.
Chiaro puts it clearly,

The main setback regarding translating for film is the fact that screen products are
polysemiotic: that is, they transmit messages by means of diverse codes ( . . . ) When
a joke, a gag, or a line is linked to the visuals translation becomes especially difficult.
(Chiaro, 2010, pp. 4–5)

355
 HUMOR AND TRANSLATION

Zabalbeascoa speaks of “constrained translation” (2008). The mixed

modality (auditory and visual) or multimodality of audiovisual humor
presents unique challenges, which have begun to be addressed by such
authors as Di Pietro (2012, 2014, 2015, 2016) and Balirano (2013) with
complex semiotic analyses that are one of the most interesting cur-
rent trends of humor research. However, Martìnez-Sierra (2009) argues
that the visual aspects of audiovisual translation may actually facilitate
dubbing, in some cases, as the visual signifiers may be humorous in
themselves and carry over.
Audiovisual translation extends to any audiovisual work that is trans-
lated, but generally speaking, given the inherent costs of producing a
translation, mostly mass-market products are translated. Within tele-
vision and cinema, two practices have drawn most attention: dubbing
and subtitling. In dubbing, the part of the soundtrack of a film or TV
program that comprises the voices of the actors in the ST is replaced by
a TT recording of the voices of actors performing the lines anew. The rest
of the soundtrack is generally not altered. In subtitling, the soundtrack
is not affected, but a translation of the lines spoken by the actors, and
possibly of other text appearing in the video track, is superimposed on
the original video, generally at the bottom of the screen. For a general
overview of audiovisual translation, see Bucaria (2017).
Translation of audiovisual humor produces generally some loss of the
humor (Antonini et al., 2001; Bucaria, 2005). Williamson and Ricoy
(2014) find that about a quarter of the puns are eliminated from the
subtitling of a movie. Gottlieb (1997) and Chiaro (2007) find that the
loss is minimal.

15.3.1 Dubbing
Because of its nature, dubbing has a set of issues that are unique to it. For
example, if the face of the actors is visible on the screen when they speak,
lip movements and opening of the mouth on screen must match reason-
ably well the movements of the mouth that would be produced if the TT
were being produced by the character on screen. Zabalbeascoa reports
that variation greater than 200 ms does not produce “satisfactory results”
(1993, p. 250). Zabalbescoa (1993, pp. 248–253) provides an excellent
review of these issues, which include intonation, pauses, gestures, and
extend to non-linguistic issues as well.

356
 15.3 AUDIOVISUAL TRANSLATION

An important issue in the dubbing of humor is the position of the

punch line:

due to the very nature of some of the Factors of dubbing, it will often be difficult or
even impossible to change the position of the punch-line in the translation, and the
translator will have to write a TT joke that will find its climax at exactly the same time
as the original ST joke. (Zabalbeascoa, 1993, pp. 263–264)

Finally, dubbing, by its oral nature, has a special set of challenges

that are unique to it, as a form of translation, i.e., the necessity to
replicate or adapt the prosodic delivery of the humor from the ST
(Zabalbeascoa, 1993, p. 292). This includes timing, the delivery of the
joke, emphasis, pauses, timbre, and the likes. On timing and prosodic
issues, see Chapter 10.

15.3.2 Subtitling
As we saw above, subtitling refers to the practice of translating the SL
material in the original audiovisual text into written TL which is added
(superimposed) onto the screen. Subtitling is quite popular in many
countries, such as Scandinavian countries, the Netherlands, Belgium,
and Israel. It is not common in many countries, such as the US and
Italy, for example. The disadvantage of subtitling is that the viewers
have to read the translation on screen, which is distracting, and since
speakers read slower than they process spoken language, constrains
the translation, for example by shortening the ST. The advantage is
that subtitling is much less expensive than dubbing and in smaller
linguistic markets (e.g., Dutch) the only financially feasible option.
Subtitling presents a set of challenges that are unique to this practice.
For example, subtitles remain visible in screen between one and six
seconds, so a “screenful” of subtitles can only contain what is readable
in up to 6 seconds. A good review of these constraints can be found in
Williamson & Ricoy (2014, pp. 165–166)
There is some evidence that subtitling translation causes meaningful
loss of the humor. Pelsmaekers & Van Besien (2002) find that most
cases of irony in their corpus are translated directly, but that the cues to
the ironical intention (irony markers) may change. Example or markers
are hyperboles, intensifiers, diminutives, etc. In numerous cases, the
markers were deleted altogether, producing deadpan irony (p. 263).

357
 HUMOR AND TRANSLATION

Interestingly, the simplification of the text due to the constraints of

subtitling produced also more explicitly critical irony. Yetkin (2011)
claims that in difficult cases (subtitling The Simpsons in Turkish, with
significant mismatch between the two cultures) a full 70% of irony
and satire is not translated in the subtitles. Harrison (2013) presents
examples from the Little Britain show translated in Spanish.
Fan subtitling is a relatively new phenomenon, in which fans of a given
show, translate and subtitle the show, which is then made available on the
internet. The practice originated with Japanese anime fans, but has since
spread widely (Díaz-Cintas & Muñoz Sánchez, 2006). This practice, of
course, violates copyright laws, but their enforcement is rare, in this
area. An interesting detail of fansubbing is that often the translator
“translates” not the original text, but a prior English translation.

15.3.3 Interpreting
Interpreting (a.k.a. simultaneous translation) is the practice of translat-
ing a speech on the fly, as it is being delivered. It is generally performed
at conferences or in multilingual bodies such as the United Nations, or
the European Union. Because interpreting is done either simultaneously
(i.e., as the speaker is delivering their remarks) or consecutively, i.e.,
immediately following the end of a short stretch of speech, all the
limitations and difficulties already present in translation are exacer-
bated. If the interpreter has not been provided with the contents of
the speech beforehand there is literally no preparation and no possi-
bility of research. To make things even more complex, the interpreter
cannot laugh at the humor or at least cannot interrupt the interpreting
to laugh.
Under those circumstances, translating humor is particularly com-
plex, and translating verbal humor is virtually impossible. Yet, as doc-
umented by Pavlicek and Pöchhacker (2002), humor occurs relatively
consistently in the type of texts that interpreters have to translate. Irony,
sarcasm, and generally speaking referential humor do not seem to
present a particular challenge for interpreting, beyond the challenges
inherent to interpretation, in and of itself. Conversely, Viaggio (1996)
calls interpreting puns a “formidable challenge.” Pöchhacker (1995,
p. 45) concurs, “Jokes and funny stories embedded in a speech are
among the challenges most dreaded by simultaneous interpreters.” He

358
 15.4 TRANSLATING PUNS

provides an example of an English-to-German interpreter who trans-

lates a joke until the punch line (which turns out to be an untranslat-
able pun) and replaces the punch line with “Das ist ein Wortspiel im
Englishen” (This is a pun in English) in “documentary-style” translation.
Pavlicek & Pöchhacker (2002, p. 386) and Chiaro (2010, p. 21) discuss
a similar example, presented in Bertone (1989) of an interpreter who
asks the audience to laugh, for the sake of the speaker, thus achieving
the perlocutionary goal of eliciting laughter, at the cost, however, of
turning the joke from one by the speaker to one about the speaker,
which Chiaro calls evidence that “skopos [theory] can be taken too far.”
Pavlicek & Pöchhacker (2002, p. 399) document the inverse situation,
i.e., a speaker who deliberately produces and un-translatable text, as a
prank on the translators, whom he explicitly calls out. Much as in the
rest of audiovisual translation, a lot of humor gets lost in simultaneous
translation. Antonini (2010) documents significant loss of humor in the
simultaneous interpretation of the Oscars’ ceremony in Italian.

15.4 Translating puns

The reader will recall the somewhat laborious definition of “pun” dis-
cussed in Chapter 9 and the related definition of “language play.” Here, as
many do, e.g., Vandaele (2011), we treat the two as synonymous. When
venturing in this area we will do well to heed Zabalbeascoa’s warning:
“The question of the translatability of verbal humor [= puns] will tend
to elude blanket assessments or universalistic claims” (Zabalbeascoa
1996b, p. 239).
As Chiaro (2017, pp. 415–416) reminds us, linguistic analysis has
long used translation as a tool. In fact, this tradition goes as far back
as Cicero’s distinction between referential humor (“de re”) and verbal
humor (“de dicto”) for which he proposes a surprisingly modern method
of “changing the words” (endo-linguistic translation) to check the nature
of the humor (verbal humor cannot be paraphrased arbitrarily); see
Attardo (1994, p. 28) for references and discussion. This is because
translation has an extreme case in the translation of poetry and puns.
The connection is not accidental, as in both cases there is significant
importance attributed to the form of the signifier, which is almost always
different in another language or in a paraphrase.

359
 HUMOR AND TRANSLATION

15.4.1 Are puns untranslatable?

There is a vast literature on the untranslatability of puns, use-
fully reviewed by Henry (2003), who dedicates an entire chapter
(pp. 69–110) to reviewing many claims of untranslatability of puns
and refutes them one by one. Her thesis is simply that all puns can be
translated, although some may require more work. Regattin (2015, pp.
132–133), also provides a review of similar claims, which are, needless
to say, merrily refuted by the practice of translation on a daily basis
and by many well meaning authors: Regattin himself, of course, Henry
(2003), Eco (2003), and many others. I will focus on the latter two
authors because, with wonderful self-aware irony, they conclude their
discussion of the translatability of puns with an example (each) on
untranslatable pun.
Eco (2003b, pp. 94–95) provides an example of an untranslatable joke:
(135) Un direttore d’azienda scopre che l’impiegato Rossi da qualche
mese si assenta ogni giorno dalle tre alle quattro. Chiama
l’impiegato Bianchi e lo prega di seguirlo discretamente, per
capire dove va e per quali ragioni. Bianchi pedina Rossi per
qualche giorno e poi fa il suo rapporto al direttore: “Ogni giorno
Rossi esce di qui e compera una bottiglia di spumante, va a casa
sua e si intrattiene in affettuosi rapporti con sua moglie. Poi
torna qui”. Il direttore non capisce perché Rossi debba fare di
pomeriggio quello che potrebbe fare benissimo di sera, sempre
a casa propria; Bianchi cerca di spiegarsi, ma non riesce che
a ripetere il suo rapporto, al massimo insistendo su quel sua.
Alla fine, di fronte all’impossibilità di chiarire la faccenda, dice:
“Scusi, posso darle del tu?”.
translated literally below
(136) A CEO finds out that one of his employee Rossi in the past
few months is absent every day from three to four pm. He
summons another employee, Bianchi, and asks him to follow
Rossi discretely, to understand where he goes and for what
purpose. Bianchi follows Rossi for a few days and then presents
a report to the CEO: “Every day Rossi leaves and buys a bottle
of sparkling wine, goes to his[/your] house and he engages in
affectionate exchanges with his[/your] wife. The he comes back

360
 15.4 TRANSLATING PUNS

her to work.” The CEO does not understand why Rossi should
do in the afternoon what he could very well do in the evening,
always at his house. Bianchi tries to explain but only manages to
repeat his report, at best insisting on that “sua” [his/your]. In the
end, faced with the impossibility of clarifying the issue, he says:
“May I address you with the [informal] ‘tu’ pronoun?”
The original is based on the referential ambiguity of the Italian pos-
sessive pronoun “sua” (= his/your[respectful]). However, the ambiguity
is resolved, in the joke, by switching to the familiar form of address
which allows the speaker in the joke to word the sentence as “tua”
(your[familiar]). This detail is untranslatable in English, or French
(which have different pronouns). The untranslatability derives from the
enmeshing of the linguistic form and of the narration.
However, a simple adaptation of the text would resolve the problem
and produce a joke very close to the original:
(137) A CEO is suspicious of one of his employees, Smith, who leaves
work everyday from 3:00 to 4:00 pm. He asks the foreman to hire
another employee to follow Smith. The employee returns after a
few days and reports to the foreman about the CEO’s suspicions:
“I found that, as the CEO suspected, Smith every day at 3:00 pm
goes to a motel and has sex with his wife until 4:00.” The foreman
thinks it is odd that Smith cannot just wait a couple of hours
and go home to have sex, but overall is relieved and reports to
the CEO that all is fine and he’ll have a chat with Smith about
excessive absenteeism.
I have shifted the puzzlement about why the employee wouldn’t just
wait until after work to the foreman rather than to the CEO, to allow
me to stay with the third person and I have eliminated the final twist
of the employee saying to the boss, “Excuse me, can I address you
with the ‘tu’ pronoun”? which as the Italian reader will understand
inferentially will remove the ambiguity of “si intrattiene in affettuosi
rapporti con sua moglie.” Nonetheless, since the Script Opposition,
the Logical Mechanism, and the Target are the same, while the Situ-
ation and the Narrative Strategy have minimal and overall insignifi-
cant changes, the only significant differences are in the Language, and
since this is a translation, most speakers would probably be inclined
to find the two jokes (Eco’s original and my translation/adaptation) to

361
 HUMOR AND TRANSLATION

be more or less equivalent. Certainly they are equally funny (not very
much).6
As usual, a clear definition will help understand the nature of the
problem. The question “are puns translatable?” can be interpreted in two
radically different ways: one is, can any arbitrary pun be translated with
a feature-by-feature correspondence between the ST and the TT?, the
other is, can any text that contains some punning elements be translated
so that a reasonable amount of the features of the ST are also present in
the TT, albeit not necessarily in the same position? Once stated clearly,
as I just have, the two questions have almost obvious and opposite
answers: feature-by-feature correspondence translation is axiomatically
impossible, since I have shown (Attardo, 2002a) that any text in language
L1 , when translated in language L2 (and L1 and L2 are not the same),
differs by definition in one feature: that of no longer being worded
in L1 . Conversely, since among the strategies for the translation of an
individual pun most authors include omitting the pun entirely, then any
pun can be “translated” if all the translator needs to do is delete the pun
and come up with another bit of humor to insert somewhere in the text.
Note that this is far from easy, since it requires a creative step, but it is
theoretically possible in any case, since what the translator needs to do
is produce a new, different text.
What of Henry’s untranslatable pun? I reproduce it below. It occurs
in Le cabinet noir [The Black Cabinet] (1922) by Max Jacob, a French
writer and painter who died in the Nazi concentration camps.
(138) Tout enfant, je fus élevé dans une chambre à nourrice dess-
inée par Steinlen, c’est pourquoi j’adore les chats (jeu de mots
intraduisible)
Henry tells us (2003, p. 110) “several specialists of this facetious author
have considered this text and they are still looking for this enigmatic
pun, which, as long as it will not have been discovered, will remain
doubtlessly untranslatable” (my translation, SA).
I don’t know what to make of this. Is the author pulling our leg, with
a straight face? Or is she really thinking that there is a pun to be found?
It seems to me that Jacob was mocking translators and there is no pun
at all.

6 It should also be noted that in my version the boss will not find out about the cheating
wife.

362
 15.4 TRANSLATING PUNS

15.4.2 The practice of translating puns

Delabastita (1996, p. 128) lists the following strategies for the
translation of puns, which have been widely adopted (e.g., Chiaro, 2003;
Low, 2011; Williamson & Ricoy, 2014)
ST pun to TT pun
ST pun to TT non pun
ST pun to TT related rhetorical device
ST pun to TT omission
ST non pun to TT compensatory pun
ST non pun to TT pun not present in ST (also compensatory)
Editorial techniques: footnotes, endnotes, translators’ forewords, pre-
sentation of several options, etc.
Low (2011) offers additionally an original technique for translating puns
that consists in looking for synonyms of the words or strings in the ST
and then produce a new pun in the TT.
Formulated within the Interpretive Theory of Translation (Seleskovitch
et al., 1984), Henry (2003) presents a remarkable amount of examples
worked out in detail, but her most significant contribution to the
translation of puns is a tripartite continuum of levels of functional
significance, in regards to the text in which the puns occur. The puns,
vis-à-vis the text may be

1. Not very important (punctual puns; p. 52);

2. Important: puns are one element of the textual organization (p. 54);
and
3. Essential: pure are the writing system (p. 56).

Despite the rhetorical exaggeration, given that puns by themselves can-

not constitute an entire text (say, a novel), since other factors are nec-
essarily present, such as an array of para-textual factors such as titles,
colophon, copyright page, etc., the subdivision is interesting. In the last
case, puns are the main driving force of the text. Let us apply Henry’s
idea to an example she did not analyze. Consider the following poem,
by American poet Aram Saroyan, which has the advantage of brevity:
(139) My arms are warm
Aram Saroyan

363
 HUMOR AND TRANSLATION

(Complete Minimal Poems (includes Aram Saroyan, Pages, and The

Rest), Ugly Duckling Presse (Brooklyn, NY), 2007.)
It is clear, and this is Henry’s point, that the assonances and allitera-
tions of the poem, taken together with the signature of its author, are the
point of the poem, as a literal translation in Italian, for example, would
immediately prove:
(140) Le mie braccia sono calde.
Aram Saroyan
(my translation, SA). Of course, a real translator would try to com-
pensate by creating other assonances or alliterations, but that’s not the
point. If you translate literally example (139) the text, its raison-d’être
is obliterated: “there is no text if there is no pun” (Henry, 2003, p. 56;
my translation, SA). Conversely, at the other end of the continuum, are
punctual puns: these are texts that contain puns somewhere in the text,
but that if we removed the puns, we would only minimally alter the text
(it would be less funny, of course). Therefore, if we replace the puns
with other puns, since the puns are not an important aspect of the text,
the text will not be altered significantly. Consider now another possible
translation of (139) below,
(141) Saremo soriani.
Aram Saroyan
(my translation, SA) which in back-translation corresponds to “we will
be tabby cats” and clearly completely replaces the semantics of the line,
but preserves some of the assonances and alliterations. Needless to say,
the reader may object that there are no cats or feline feelings in the
original and that the arms and the warmth are gone. I would object that
the warmth is there by connotation with the cat, but my hypothetical
critic is nonetheless correct: in a very clear sense, this is not the same
poem.
Finally, the intermediate case in Henry’s continuum, is a text in which
the puns in and of themselves matter, but are not the main point of the
text. A good example here are the translators’ creative renderings of the
various linguistic plays in Eco’s own novels, for example, with medieval
dialects, Baroque terminology, etc., discussed in Eco (2003b).
The Interpretive Theory of Translation has many similarities with
skopos theory and with Zabalbeascoa’s approach: they all recognize
the importance of the participants in the translation, the situation in

364
 15.5 CONCLUSIONS

which it is performed, the purpose it serves, the importance of the

cognitive context (i.e., how the text and its translation are inscribe in
their respective cultures), and finally, they share a focus on textual
equivalence, rather than linguistic similarity (Henry, 2003, pp. 66–67).

15.5 Conclusions

There seems to be a consensus (e.g., Raphaelson-West, 1989;

Zabalbeascoa, 2005; Low, 2011; Yus, 2013) on the following points:
The process of translation is highly contextualized: each translation has
several goals or purposes which are negotiated (Eco 2003; Zabalbeascoa,
2005) among the participants to the act of translation and with
the context and setting of the texts. Humor is, generally speaking,
translatable. Some forms of humor are harder to translate than others.
There is a scale of difficulty of translation, with the easiest being general
jokes that do not rely of punning or culture specific jokes, followed
by culture specific jokes (hard to translate) and “language-restricted
jokes” (Zabalbeascoa, 2005), which are hardest to translate, or may be
untranslatable. Mallafrè (1991, p. 119); Zabalbeascoa, 1993; Chiaro
(1992, p. 87); Low (2011). The most significant problem areas are
puns (language-specific structures, usually tied to the signifier) and
cultural issues (such as the impossibility of talking about a referent
unknown in the target culture: penguin jokes will not fare well in the
Amazon).
Individual instances of humor (jab lines or punch lines) may be
untranslatable, but the overall text in which the jab line or punch line
occurs generally can be translated so as to respect the perlocutionary
goal of eliciting amusement or the perception of the intention of humor.
Translators may use a variety of strategies to effect the translation of
difficult cases, which may reach the point where the translator has to
give up trying to translate the specific instance of humor and use com-
pensatory strategies, such as replacing the joke or omitting it entirely
(and replacing it with something else), or just give up on the trans-
lation (i.e., omit the joke and not replace it). Audiovisual translation
has its own specific challenges, related to the multimodal nature of
the texts.

365
 HUMOR AND TRANSLATION

15.6 Further readings

Martinez-Sierra and Zabalbeascoa (2017) gives a very good view of the

most recent work on translation. Chiaro (2017) and Bucaria (2017) are
good surveys of their respective fields. De Rosa et al. (2014) contains
several very interesting papers. The work of Zabalbeascoa, starting with
his 1993 dissertation is foundational, as is Chiaro (2005). Delabastita’s
work and edited collections (1987, 1993, 1996, 1997) on the translation
of puns remain classics.

366
 16
Humor in the classroom

16.1 The pioneers 368

16.2 The apologists 371
16.3 The realists 372
16.4 Classroom discourse analysis 377
16.5 Conclusions 379
16.6 Further readings 380

This chapter needs to start with a major caveat. There exists a small
cottage industry of publications on how to improve teaching through
humor. I will provide a list of a few examples below. However, there
is little empirical evidence that using humor while teaching actually
improves learning and/or retention (Martin, 2007, p. 350). Some assess-
ments are more stark: Halula (2013) concludes pessimistically that
the literature “about humor and education [is] typically anecdotal and
prescriptive in nature with little or no research backing” (p. 118). This is
very similar to the situation that psychologists see with the humor and
health movement. Again, lots of claims and anecdotal evidence, but a
damning scarcity of empirical studies showing that laughing or having
a sense of humor actually improves your health. This “split personality”
leads to some awkward situations, as noted by Bryant and Zillmann
(1989) who call it, a “curious contradiction” between claims that humor
is “highly useful and extremely effective teaching tool” (p. 49) and the
empirical results which are “decidedly mixed” (p. 50). The tongue-in-
cheek understatement is followed by a short but effective review of
“questionable strategies” used by some authors of popular works on the
benefits of humor in the classroom, which include, ignoring the negative

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 HUMOR IN THE CLASSROOM

evidence and oversimplifying ambiguous results. In what follows I will

focus primarily on the empirical research, and then with a bias toward
applied linguistics, but of course most of the research in this area has
been done in education and psychology.
The Holy Grail of the application of humor studies to education is
the proof that humor facilitates or enhances learning in a classroom
setting. The impact of humor on teaching, if any, may occur at dif-
ferent levels/moments of the learning process: humor may a) facili-
tate learning, i.e., make it easier to acquire the new information, or
b) improve retention, i.e., hold on to the new information, or c) improve
the pedagogical experience of either the student (e.g., by reducing
stress) or the teacher (e.g., by avoiding burnout). It is important to
remember that an effect in one aspect does not imply similar effects in
another aspect. So, for example, we could find that humor improves the
experience of the teacher but not of the students, or that it facilitates
learning but not retention. This requires then to keep each hypothesis
and claim separate, to avoid confusion.
The chapter is organized as follows: we will begin by reviewing a few
pioneering studies, some of which showed some promising results and
probably are the proximate cause for the existence of the books on how
to improve teaching and leaning using humor. Then I will provide a
sample of books touting the use of humor in the classroom, mostly
just to document the size of the phenomenon, and then I will review
the field using three meta-analyses and some recent publications that
provide overviews of the field. The chapter is capped off by a final section
which reviews an emerging area of interest, i.e., ethnographic studies of
humorous behavior in the classroom (classroom discourse analysis).

16.1 The pioneers

One of the first studies to test empirically the idea that humor facilitates
learning idea is Markiewicz (1974) who reviews the literature up to that
point, which she finds riddled “with severe methodological problems.”
She provides a very useful chart with a meta-analysis of about 25 studies
(p. 409) starting the tradition of really depressing findings in the field:
only very few studies show significant positive differences with the use of
humor. Markiewicz considers possible effects for humor and persuasion,

368
 16.1 THE PIONEERS

humor and retention, and evaluation of the source (i.e., the idea that
the audience of humor will assess more positively a speaker who uses
humor than one that does not), and assessment of the interesting nature
of the material. Only the last hypothesis is found to have some positive
effect (p. 413). The degree of relevance of the humor might also be
a factor: Markiewicz considers relevant (i.e., on topic) and irrelevant
humor, but finds no differences. Her methodological discussion is still
quite valid.
Another study is Kaplan and Pascoe (1977). They ran a large (n = 508)
study involving four versions of a 20 minutes lecture on Freud. One
version of the lecture was without humor. Of the three humorous
versions, one contained humorous material related to the subject matter,
one contained unrelated humor, and one a mix of related and unrelated
material. The students were tested immediately after the lecture and then
6 weeks later. The results are mixed: there was no significant difference in
the immediate understanding of concepts but the concepts introduced
humorously were retained significantly better (p. 64). The students’
performance on the tests was not significantly improved by humor
(p. 64), even in the re-test condition, when their performance improved
due to better recall. The authors conclude that the benefit of using humor
is increased recall of the examples.
Ziv (1979a) explains learning through humor with the observation
that humor attracts attention which leads to memorization and retention
of the material. Students enjoy humor (1979a, p. 46). In (1979a/b) he
further elaborates that a “positive teacher,” who among other features
uses humor, creates a better atmosphere in the classroom, and lowers
anxiety which is conducive to learning (1979a, pp. 60–61; 1979b, p. 22).
In fact, enjoying humor on the teacher’s part has a significant correlation
with teachers having a positive attitude in the classroom (1979a, p. 59).
Ziv’s name is usually linked with the idea that humor improves creativity.
He finds a strong correlation between humor and creativity, particularly
around the trait of originality (p. 109). In an experiment he shows that
a single exposure to humor enhances creativity (p. 116). In a semester-
long study he found enhanced creativity from continued exposure to
humor (pp. 120–122).
Ziv (1988) describes the first semester-long study of the effects of
humor on learning. Two intact statistics classes taught by the same
instructor were taught, one with 3 instances of humor per 90 minutes

369
 HUMOR IN THE CLASSROOM

lecture (1979a, p. 80)1 and one without humor. The final multiple-choice
exam was used as the assessment tool. The section taught with the
humor scored significantly better on the final test. The same experiment
was then replicated in a psychology class, with the same significant
positive results for the section taught with humor. Ziv cautions that
the amount of humor had to be calibrated exactly to 3 instances of
humor per class (as he had found in pilot tests that that is an appropriate
amount of humor). Moreover, the humor has to be directly relevant to
the content of the course and varied in its format/presentation. Finally,
negative/aggressive humor was avoided (such as sarcasm). Because
of the clear cut results, which are unequivocally positive, and of the
extended duration of the study, i.e., an entire 14 weeks semester, whereas
most experimental treatments are at most 50 minutes long, this study
has been vastly influential and is considered a classic.
Bryant et al. (1979) look at college lectures. They start out by noting
that humor was being increasingly introduced in college lectures and
that there already were some advocates for its use, but “without sound
empirical evidence” (p. 111). The authors had students tape (apparently
surreptitiously)2 70 lectures and assess the number of humorous inci-
dents (inter-rater reliability was at 91%). The coders found 234 instances
of humor which yields an average of one instance of humor every
15 minutes or so. There was a significant range: 20% of classes had
zero occurrences of humor; conversely, 5% of the teachers used humor
more than 10 times per class (every 5 minutes, on average) (p. 114).
Female professors used significantly less humor but used surprisingly
more sexual humor, more aggressive humor, and were rated as using
more improvised humor (p. 116). Most of the humor was closely related
to the subject matter, and here too, more so for women’s humor. Also,
surprisingly, male professors used more self-disparaging humor. More
discussion on humor in lectures will be found below in Section 16.4.
Humor increases the students’ positive perception of the textbook,
but does not increase interest, learning, or persuasiveness (Klein et al.,
1982). It does however negatively impact author’s credibility. Bryant
et al. (1981) shows that adding cartoons to a textbook does not improve

1 This is actually on the low end of the scale; see the studies below in Section 16.4.1 that
attempt to quantify the amount of humor produced by teachers in a lecture. Ziv’s rate of
humor in his experiment is equal to 1.6 instances of humor per 50 minute lecture, or 0.03
instances of humor per minute.
2 This would be considered ethically unacceptable nowadays.

370
 16.2 THE APOLOGISTS

learning but the students enjoy them. Kaplan and Pascoe’s (1977)
found increased memorability of humorous examples. Davies and Apter
(1980) also found greater recall for humorous materials.
Bryant and Zillmann (1989) report that teachers use humor in lectures
and activities from elementary school to college, with males doing so
more than females (p. 56). Humor can get the attention of the students
(pp. 59–60) but does not improve the classroom environment, at least
in the college classroom (p. 61); nonetheless, students enjoy it. Age and
setting (elementary/high school vs. college) may be the key factors in
these findings. For younger children, if they lack motivation, humor,
and especially humor unrelated to the content, improved learning.
If the students were already motivated, no improvement was found
(p. 66). Bryant and Zillmann find no evidence that humor included
in test improves performance or attitude toward the test and in fact
report that it can hurt students with high test anxiety (p. 69). They
conclude on a moderately optimistic note that “judicious use of humor”
(p. 74) increases attention, improves enjoyment, promotes creativity,
and “under some conditions” improves “information acquisition and
retention” (p. 74). This sounds actually quite good, until they specify
what “judicious” means: “success in teaching with humor [...] depends
on employing the right type of humor, under the proper conditions,
at the right time, and with properly motivated and receptive students”
(p. 74).
Javidi et al. (1988) find that award-winning teachers use significantly
more humor than non-award winning teachers. More on this study
below. Buckman (2010) also finds a correlation between university
teachers’ performance and their use of humor.

16.2 The apologists

By the mid-1980s, Powell and Andresen (1985) could set out to review
more than 50 articles on the benefits of using humor in the classroom
and some empirical studies. Things did not stop here. Soon there were
significant numbers of publications advocating teaching with humor.
Nowadays, there exists a largish number of publications that advo-
cate for the use of humor in educational settings. Most of these have
scant evidence, gathered from the few studies that show positive effects
of humor, but generally speaking the mode is not that of scientific

371
 HUMOR IN THE CLASSROOM

argument, but rather of apology, in the classical sense, or of advocacy.

The author is convinced of the effectiveness of humor and wants to pros-
elytize or simply assumes that using humor is a good thing and provides
advice on how to use humor in teaching. Rather than discussing the
merits of these I will quote the titles of some of the best known books
on the subject.

• Berk (2002) Humor as an instructional defibrillator: Evidence-based

techniques in teaching and assessment.
• Berk (2003) Professors are from Mars, students are from snickers
• Shade (1996) License to laugh: Humor in the classroom
• Lundberg and Thurston (1992) If they are laughing they just may be
listening.
• Loomans and Kolberg (2002) The laughing classroom.
• Medgyes (2002) Laughing matters: Humour in the language class-
room.

A list of scholars advocating the use of humor in the classroom can be

found in Bryant and Zillmann (1989) and in Banas et al. (2011). Banas
et al. note that “the overwhelming majority of instructional communi-
cation research on humor has focused on the positive consequences of
classroom humor” (p. 116). A search on Google will easily add hundreds
of tiles. A search for “humor, teaching, advantages” produced 109,000
hits. The search “humor, teaching, disadvantages” produced 18,000 hits.
Banas et al. do list four (!) authors who have considered the negative
effects of humor in the classroom (p. 116).
Even national associations, such as the National Education Associa-
tion contribute to the apology of humor, see McNeely (2002). As I said,
this reminds one of the claims about the health benefits of humor, where
anecdotal evidence reigns supreme. This is not to say that these books
are without merits. They can be very useful if one wants to start using
humor in the classroom and doesn’t quite know how to do so.

16.3 The realists

Given the significant interest in the application of humor to teaching,

with the possibility of improving teaching, learning, student attitudes,
and even teachers’ evaluations, a significant literature has developed,

372
 16.3 THE REALISTS

primarily in the fields of education and communication, and also of

course in psychology. Linguistics has lagged, generally speaking, but
recently some significant studies have appeared. We will review the
field using primarily three excellent meta-analyses, which also provide
effective summaries of the field.
McMorris et al. (1997) review 11 studies on college testing. Their con-
clusion is that there is insufficient evidence that use of humor in testing
reduces stress and humor does not improves performance. However,
the students report that they like it (p. 285). The authors conclude that
test-writers may want to include humor in their tests.3 McMorris et al.
list the following issues that need to be addressed: the types of humor,
the purpose of the humor (e.g., inclusive vs. distancing humor), the
degree of incorporation of the humor, i.e., whether it is a) relevant to
the material or unrelated to it, and b) presented in the test items, in the
instructions, separate from the test, etc.
Martin et al. (2006) review 20 studies, They set out to answer the
question “does humor in the “educational messages [...] result in under-
standing and learning?” (p. 18). They distinguish between three types of
effect:

1. Cognitive: the learning was tested, for example, by a multiple

choice test;
2. Affective: the respondents were asked if they enjoyed learning; and
3. Perceived: the respondents were asked how much they felt they had
learned.

Their overall conclusion is that “Although students report enjoying

learning and they report that they believe they have learned course
material, objective measurements of the recall associated with humor-
ous lectures are rather minuscule” (Martin et al., 2006, p. 305).
After an extensive review of the various types of humor and their
classroom effects, Banas et al. (2011) conclude that “the relation-
ship between instructional humor and educational [sic] cannot be
understood without taking into account the type of humor used,

3 The following is my personal opinion and I have no evidence for it except for my
personal (anecdotal) practice. I’d think twice about doing what McMorris et al. advise. Most
of the students I have taught in my career lost all sense of humor at test time. Any attempt
at humor backfired in the most depressing way. Even my instructions to read the questions
before answering them were met with concerned questions.

373
 HUMOR IN THE CLASSROOM

particularly regarding appropriateness and offensiveness” (p. 125).

Generally speaking aggressive and denigrating humor is inappropriate
in the classroom (and yet nonetheless they occur, we might add). They
also report differences in gender usage, for example, male professors use
more self-deprecating humor and female professors use humor to gain
back control of the class after a disruption, but note that the statistical
significance of the results in this area is not very reliable due to small
samples. To complicate things, the teachers themselves are a variable,
being more or less “humor oriented” (p. 127), i.e., funny. If a teacher is
more humor-oriented, he/she produces more humor, which correlates
with producing more inappropriate humor as well.
Use of positive humor improves teacher evaluations (p. 129) but
“overuse” is rated negatively. “[A]ppropriate instructional humor [is]
related positively to an enjoyable learning environment” (p. 130) but,
again, aggressive humor may create an uncomfortable environment.
Likewise, credibility is boosted by the use of appropriate humor (p. 131)
and decreased by inappropriate or excessive humor. By and large, they
conclude that positive and appropriate humor correlates with desirable
instructional outcomes, such as “relaxed learning environment, higher
instructor evaluations, greater perceived motivation to learn, and enjoy-
ment of the course” and inappropriate humor with the opposite out-
comes. So, their conclusion is positive as far as the learning environment
goes. However, Banas et al. (2011) find that the “the empirical evidence
for the effects of humor on learning is considerably more mixed, with
some scholars finding that humor enhances learning [...] and others
finding no relationship between learning [and] humor” (p. 131). The
question of whether relevant humor performs better than irrelevant
humor remains open, even though Machlev and Karlin (2016) found
no correlation between relevant humor and learning.
Wanzer et al. (2010) proposed the Instructional Humor Processing
Theory to explain why some types of humor produce increased learning
and some do not. The Instructional Humor Processing Theory pre-
dicts that humor related to the subject matter of instruction would
be positively correlated with learning. On the humor studies side, the
Instructional Humor Processing Theory incorporates the incongruity
theory framework, with the disposition theory, a variant of the supe-
riority theory approach, which claims that one appreciates as humor-
ous humor that targets something one is negatively disposed toward

374
 16.3 THE REALISTS

(see Chapter 3). So, humor related to the subject matter is appreciated
by the students because it does not target students (which would violate
the disposition theory, which assumes that students are well-disposed
toward themselves), likewise, self-disparaging humor by the teacher
may backfire, if the students are well disposed toward the teacher. The
authors’ study showed that students judged teachers to be humorous if
they produced relevant humor; surprisingly, also self disparaging humor
produced increased self-reported learning. Finally, “Professors that were
perceived as humorous used significantly more related, unrelated, self-
disparaging, other-disparaging, and offensive types of humor than less
humorous professors” (Wanzer et al., 2010, p. 14).
Another wide-open topic is cross-cultural variation. For example,
Zhang (2005) reports that humor made Chinese students more anxious.
Heidari-Shahreza (2018) reports the use of humor in Iranian L2 class-
room. Other non-Western contexts are Taiwanese Chinese (Liao, 2005),
Mainland Chinese (Chen, 2013), Thai (Forman, 2011), Vietnamese
(Petraki & Pham Nguyen, 2016), Japanese (Neff & Rucynski, 2017). The
field is vast and there is no good overview.
Another area of interest, is the use of humor in the foreign language
classroom, where humor has the potential to reduce anxiety above and
beyond all the usual positive effects claimed for it. Studies focusing on
foreign language teaching and humor include Deneire (1995); Schmitz
(2002); Wagner & Urios-Aparisi (2011). Wagner and Urios-Aparisi
(2011) couch the problem in terms of “immediacy,” i.e., “the degree of
directness and intensity of interaction between the communicator and
the referent of communication” (Mehrabian, 1966, p. 34), or to put it
differently the degree of involvement and closeness that the speakers
communicates, both verbally and non-verbally. Wagner and Urios-
Aparisi note that in both modalities humor figures prominently in the
definitions in the literature on immediacy (Wagner & Urios-Aparisi,
2011, p. 403). So humor would have a positive effect on classroom
activities mediated by the immediacy of the teacher. Wanzer (2002) also
addresses immediacy, but within the broader context of “humor orien-
tation,” i.e., the general disposition of a teacher to produce humor. In
particular she focuses on appropriate and inappropriate humor (Wanzer
et al. 2006).
Bell (2011) presents the most sophisticated framework for applying
humor studies to the second language classroom. The point is not so

375
 HUMOR IN THE CLASSROOM

much to teach the students how to be funny (p. 150), but, as she puts
it, the “cultural norms of humor usage” (p. 135) which consist of how
humor “is contextualized, how and with whom joking relationships
may be formed, and common responses to various types of humor”
(p. 135). Topics that second language learners need to be aware of are
available scripts for humor (i.e., what is appropriate for joking and
what isn’t), contextualization cues, which may vary across cultures,
genres of humor, the pragmatic and social functions of humor, and the
repertoire of responses to humor, which again may vary from culture to
culture.
This brings up the idea of the existence of a specific separate humor
competence as Vega (1989) claimed, to be placed alongside the other
four communicative competences (grammatical; sociolinguistic, i.e.,
contextually appropriate; discursive, i.e., coherent and cohesive; and
strategic, i.e., capable of solving communicative problems; Canale &
Swain, 1980). Hodson (2014) shows that teaching unsuspecting stu-
dents about humor theory doesn’t particularly affect their capacity to
understand humor and Bell tongue-in-cheek tells us the benefits of
teaching humor theory to students “remain an empirical question.” This
is not surprising, since I have argued (Attardo, 1994, pp. 211–213) that
there is nothing unique about humor competence except a) scripts that
are available for humor (humorous stereotypes); b) scripts that are not
available for humor (topics about which it is socially unacceptable to be
humorous); c) contextual information on when and in what situations
it appropriate to use humor; and d) a few genres that are culture specific
(such as knock-knock jokes, unknown outside of the Anglo-Saxon
world). So, if we were to compare the sheer size of knowledge involved
in each of the four competences to the reductive size of the purported
fifth one, it would seem obvious to argue that how to produce and use
humor would be part of the other four competences, without the need to
postulate a separate competence. Undeterred by these arguments, Wulf
(2010) presents a curriculum for humor competence.
Bell and Pomerantz (2016) is primarily focused on the students’ use of
humor (and so will be examined in more detail below in Section 16.4),
but it is also a very complete discussion of the literature, including the
research on the benefits of using humor in the teaching environment.
So, it is telling to see them conclude that “it has been difficult to connect
the use of humor to increased learning” (p. 101) and so that “the most
robust argument for using humor in education is affective” (p. 101).

376
 16.4 CLASSROOM DISCOURSE ANALYSIS

16.4 Classroom discourse analysis

A major contribution to the analysis of classroom humor is Davies

(2003) which is explicitly based on the ethnomethodological approach
described in Chapter 10. It can be seen as an extension of Davies
(1984), discussed in Chapter 12. Davies argues that students in a second
language classroom “are able to exploit the limited gamut of sociolin-
guistic resources available to them within a discourse context in which
the native speaker provides important support” which establishes the
fundamentally collaborative nature of the situation. Davies shows that
both students and teachers initiate humorous exchanges. The native
speakers and the students themselves use a “scaffolding” mechanism to
build jointly the humorous exchange, despite their limited proficiency.
See also Davies (2015).
Bell and Pomeranz (2016, p. 78) view humor as a “safe house” for
the students in the classroom. The concept of “safe house” comes from
Canagarajah (2004) who identifies a variety of ways in which students
carve small places for themselves within the broader context of school.
Examples of safe houses would be passing notes, writing in the margins
of books, or talking to other students off task. Considering humor as a
safe house is not a stretch, given the function of deniability which has
long been associated with humor (“Just kidding!”), but what is original
is seeing humor as a play space, which is connected to the consideration
of language play, a broader category than humor. Bell and Pomerantz
note that humor is “normal” in the classroom (p. 73) and so it should
be expected, as it would in any “normal” situation. It is possible that
the impression of relative scarcity of humor in the classroom is due
to the focus on the teacher and the “covert” nature of most of the
humor, which occurs in a safe house, away from the eyes of the teacher
(and of any researcher). Humor conversely can also be seen as a face-
saving opportunity, of particular importance in the teaching setting, if
we consider that the very act of teaching is inherently a face-threatening
act: in order for you to teach me, I need to admit, more or less openly,
that you know more than I do on the subject. So humor allows both
teacher and student a way out of the inherently hierarchical setting.
Relatedly, a student may elect to be the “class clown” and thus cultivate
a “cool” identity (p. 80). For example, Damico and Purkey (1978) and
Norrick and Klein (2008) consider disruptive humor in the elementary
classroom (“class clowns”). Reddington (2015) and Reddington and

377
 HUMOR IN THE CLASSROOM

Waring (2015) are applications of conversation analysis to classroom

interactions and provides some examples of the phenomena discussed
above.

16.4.1 How much humor do teachers produce in class?

We have seen that students produce plenty of humor (Bell and
Pomerantz, 2016). We turn now to the question of how much
humor do the teachers produce. Javidi et al. (1988) report on a
study in which award-winning college and high-school teachers
were compared to non-award-winning teachers.4 A sample of 50
minutes of lecturing was annotated for each teacher. The results
show that award-winning teachers incorporated 7.20 instances of
humor in a 50 minute lecture (roughly one instance of humor every
7 minutes). The humor “played off ” the content of the course and
was often used to clarify the content. In contrast, award-winning
high-school teachers only produced humor 2.8 times in 50 minutes
(about once every 18 minutes). Like college teachers, the humor was
used to clarify content. Middle-school teachers scored even lower
(2.33 per 50 minutes of lecture; once every 21.45 minutes). The
non-award-winning teachers used a very small amount of humor:
0.26 instances per 50 minutes, or a remarkable once every 192 minutes).
Downs et al. (1988) report on a study of 57 college teachers. The
college professors produced an average of 13.33 instances of humor per
lecture (once every 3.75 minutes). Most of the humor was relevant to the
class (only 12% was coded as non-relevant). Downs et al. also considered
nine award-winning college teachers and found that they produced
an average of 7.44 instances of humor per lecture (i.e., once every
6.72 minutes). This is a significant difference from the study by Javidi
et al. Downs et al. provide a breakdown by week in the semester, which
shows that teachers on average produced humor 9.3 times per lecture in
the second week, 8.1 times in the 6th week, and 5.6 times in the 10th
week. In other words, the variable of when the lecture is collected may
be relevant. Most of the humor was relevant to the lecture in all cases.
Lee (2006) has a slightly wider scope, in the sense that it is not limited
to classroom interactions, as it is based on the Michigan Corpus of
4 The fifteen non-award-winning teachers consisted of five college teachers, five high-
school teachers, and five middle-school teachers. Javidi et al. (188) do not explain why they
chose not to differentiate between the three groups. They do, however, note this limitation
in their discussion.
378
 16.5 CONCLUSIONS

Academic Spoken English (MICASE). Lee considers the occurrences of

laughter (which he takes to correlate with humor; see 11 for a discussion
of the problematic nature of this assumption). He finds a wide range
of variation, with events having zero laughter and an undergraduate
social science thesis study group a whopping 179 laughter instances in
64 minutes, i.e., a ratio of 2.8 laughs per minute (Lee, 2006, p. 53).
Lectures tend to have a low ratio of laughs per minute, but as Lee notes,
some of the lectures are among the events with most laughter. Overall,
lectures have 0.21 laughs per minute, whereas tutorials and study groups
have more than one laugh per minute (1.35 and 1.12, respectively; Lee,
2006, p. 54). Lee speculates that distance of social status and intimacy of
the event might be the causal factors behind these findings (p. 55).
Nesi (2012) considers laughter in university lectures taken from the
British Academic Spoken English corpus (BASE). The BASE academic
events are more limited in scope than the MICASE and include only
lectures and seminars. 24 out of 160 lectures contained no laughter
and four contained only one instance of laughter. Nesi compares her
data to Lee’s (2006) and finds that while the percentage of lectures
containing laughter was roughly the same across countries, British
lectures contain one laugh every 1.4 minutes, whereas American lectures
contain only one laugh every 5 minutes. Nesi speculates that the cause
is that British lectures are more formal and so the lecturers prepare
the humor. An alternative explanation is that British lecturers are just
funnier than Americans. Neuliep (1991) presents self-reports from high
school teachers who claim to attempt humor roughly twice per class
period. He notes that this figure is significantly less than other studies
having as subjects college teachers.
Both Lee and Nesi list more or less impressionistic functions of the use
of laughter, which include the usual suspects: building rapport, getting
the attention of the students, criticizing behavior, self-deprecation, etc.
In short, there seems to be no special function of laughter in lectures.

16.5 Conclusions

What conclusions can we draw from the literature on using humor

in the classroom? The first one, is that by doing so, we probably will
not do any serious damage to our students. The second one is that
we should lower our expectations: using humor is not going to be the

379
 HUMOR IN THE CLASSROOM

silver bullet of education and will not lead to massive improvements

of learning and retention. It may improve the students’ attitude and
perception of the learning experience and that’s obviously a good thing.
The third conclusion is an obvious one: further study of these matters is
to be recommended, especially in the direction of separating the various
kinds of humor and of activities. Humor in banter with the students is
not the same thing as humor in the lecture or humor in the test. Finally,
we need to remember that a successful humorous exchange is a joint
activity: the speaker and the hearer need to agree to the framing of the
situation as humorous. This entails both that the teacher must relinquish
some control of the situation and that he or she and the students must
find a common ground to enjoy the humor. As Powell and Andresen
had remarked in 1985, humor ages and one always runs the risk of
“asking students to laugh at something that would have appealed to a
previous generation” (p. 88). This observation is all the more true today,
when memes, tweets, and instagram posts have replaced the jokes and
cartoons of yore. If teachers want to be funny to their students, they will
need to work at knowing why, at what, when, and where their students
laugh.

16.6 Further readings

Martin (2007) is a very good broad review of the pedagogy of humor,

even if it is skewed toward psychology, having been written by a promi-
nent psychologist. Bell (2011) is squarely focused on applied linguis-
tics. The meta-analyses discussed in the text are the best overviews
of the pedagogical aspects of humor, McMorris et al. (1997); Martin
et al. (2006); and Banas et al. (2011). Reddinton (2015) and Bell and
Pomerantz (2015) provide overviews of humor in the classroom from
the perspective of TESOL.

380
 17
Conclusion

It is traditional to conclude a book like this with a look at the future. In

the past, I have done so with results so amusingly off base, as when I
predicted in 1994 a renaissance of studies on puns, only to see the field
go dormant for 20 years, that I have decided to do so in the future only
for entertainment purposes. No doubt more research will be done in
all, or most, of the areas I have described in the book. Will the trends I
have discerned, or think I have discerned, continue in the near future?
Perhaps, or perhaps not. I have learned my lesson. However, I think we
need to look a little farther out, as well.
I think it is fair to say that if any reader has made it this far in
reading this book (skipping ahead does not count) they will readily
agree that the study of humor, much like the study of language itself, is
a complex problem. By this I mean to invoke, almost but not completely
metaphorically, the idea of complex systems and complexity theory.1 In
a nutshell, the idea behind complex systems is that large, complicated,
non-linear systems, such as, say, the weather cannot be understood and
modeled by a simple, linear theory.
Consider an example: in Chomsky’s standard model (Chomsky, 1965)
there is no reason for language to change. Assume that you modeled
perfectly your object (in other words, you achieved descriptive ade-
quacy) and that you understood how the system worked (explanatory
adequacy). You would have a set of transformations, a set of generative
rules and a set of lexical items to insert at some point or another in your
generation process. Assume, as we said, for the sake of the argument, that
your system does produce all the sentences that native speakers judge

1 See, for example, the Five Graces Group (2009) position paper; Larsen-Freeman &
Cameron (2008); Ortega & Han (2017). I do not (only) mean complexity as in complication
(as in Derks, 2014). A complex system is not just complicated.

The Linguistics of Humor: An Introduction. Salvatore Attardo, Oxford University Press (2020). © Salvatore Attardo.
DOI: 10.1093/oso/9780198791270.001.0001
 CONCLUSION

to be grammatical and none of those they judge ungrammatical. The

resulting system would be static, in the sense that there would be no
reason or possibility for change. Why would a passive transformation
change? It does exactly what it is supposed to do and so the speakers
would have no reason or need to change their behavior which would be
reflected in a change in the rule.
To pursue the example further, let us add another facet to our system:
let us assume that speakers will follow a principle of least effort, i.e., that
they will try to minimize the effort required to produce utterances. The
simple addition of this principle will now cause our perfect (and static)
system of syntax to lose some of its perfect descriptive adequacy (for
example, through contractions, irregular dropping of frequently used
morphological markers, but not of extremely high frequency ones, etc.)
and thus eventually to have some asymmetry in the system (i.e., the
speakers will sense that there’s something wrong with the system) that
will need to be corrected through change. If you think this sounds crazy,
consider the fact that many speakers are, at the time of this writing, using
the word “irregardless” with a double privative morpheme, thus showing
that /-less/ was perceived to be too “weak” or somehow insufficiently
negative. However, you do not hear “unhomeless,” or “unjobless,” i.e.,
the change is not (yet?) systematic.
Needless to say, none of this is novel to a historical linguist and my
point is not that I have a novel explanation for language change.
My point is much more banal: you cannot understand change without
a different perspective/theory. This does not mean that the syntactic
model is not good, on the contrary, by the stipulation of our example,
the model is better than any linguistic model has ever performed. It
just means that the syntactic model is not designed to account for
change, or to put it differently, that change is outside the purview of
the model/theory.
Let me be clear: I am not naive and I know all too well that often
the proponents of one theory have explicitly or implicitly declared
the phenomena that fell outside of the purview of their theory to be
irrelevant, marginal, uninteresting, and so forth. Having been scorned
and marginalized, or having perceived to be so, the proponents of
other theories needless to say strike back declaring the theories to be
irrelevant, uninteresting, etc. All of that is true and probably inevitable,
as long as science is done by humans and not by artificial intelligences
devoid of any emotions.

382
 CONCLUSION

However, I see in the complex systems approach an opportunity to

abandon the zero-sum mindset of many scholars (if your theory is right
that makes my theory less right, somehow) and to embrace instead a
collaborative mindset where the goal is to describe, understand, and
explain the phenomena. All of them, not just those that one’s theory
works well for.
If I am correct in claiming that humor is a complex system, it follows
that no single approach, no single discipline will ever succeed in explain-
ing all the phenomena that humor encompasses. Conversely, the only
way to make progress toward that goal is to embrace a trans-disciplinary
approach, to supersede the interdisciplinary approach humor studies
has long boasted of and in some ways pioneered. In a trans-disciplinary
approach, all theories, all models can contribute something to the under-
standing of the problems at hand. All theories describe one or more
facets of the phenomena and the goal is not to fulfill the disciplinary
goals of one field or of two of them, as in interdisciplinarity, where
one theory provides the methodology and another the questions to
be answered, but rather to answer all the questions with information
coming from every discipline. When you are building a house you need
plumbers, carpenters, electricians, masons, and eventually a decorator.
No one trade is the “owner” of the process, not even the architect,
whose drawings would languish as unbuilt idealizations, but without
the contribution of any of the trades, you would not want to live in the
house.2
There are indications that a transdisciplinary approach to humor
studies may be on the way. Demjén (2018) considers a set of jokes in an
online community in a framework explicitly invoking complexity the-
ory. While the article itself seems more programmatic on the complexity
side of things, it is unquestionably a start. Bell’s (2015) remarkably broad
methodological approach is as transdisciplinary as it gets, blending
discourse analysis, elicitations, corpora, self-reports, observations, and
media samples. Brock (2017) seems to be advocating for something
fairly close to that I am proposing, albeit with a much more limited
scope. With all due modesty, Chapter 10 in this book also considers a
broad range of disciplines. I hope that my friends and colleagues will
see my prodding of conversation and discourse analysis here and there
2 In many places, many people would be quite happy to trade their current living space
for a house without plumbing or without electricity. Metaphors have limits and embody
privilege.

383
 CONCLUSION

as attempts to get a transdisciplinary (or at least trans-methodological)

perspective going.
I harbor no illusion to have accomplished the task of creating a
transdisciplinary approach to humor studies in this book. This was not
my goal. My goal in this book was, so to speak, to look backward and see
where we are, what we have learned that is useful, and set it out clearly
in as comprehensive a framework as possible. The future, however, will
require a much broader approach, which I think will incorporate a
complex system approach.

384