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

net/publication/247744086

How Sarcastic are You?Individual Differences and Verbal Irony

Article in Journal of Language and Social Psychology · September 2004

DOI: 10.1177/0261927X04266809

CITATIONS READS
93 3,773

3 authors, including:

Penny Pexman
Western University
211 PUBLICATIONS 0 CITATIONS

SEE PROFILE

All content following this page was uploaded by Penny Pexman on 02 June 2015.

The user has requested enhancement of the downloaded file.

10.1177/0261927X04266809 ARTICLE
JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY

HOW SARCASTIC ARE YOU?

Individual Differences
and Verbal Irony

STACEY L. IVANKO
PENNY M. PEXMAN
University of Calgary
KARA M. OLINECK
Concordia University

In the present research, the authors examined the effects of self-perceived use of sarcasm
on the production, interpretation, and processing of verbal irony. Accordingly, they first
devised and evaluated a sarcasm self-report scale (SSS). In Experiment 1, results showed
that participants’ self-perceived use of sarcastic irony (as assessed by the SSS) predicted
their use of ironic statements in a production task and was related to their interpretation
of ironic criticisms and ironic compliments. In Experiment 2, results showed that partici-
pants’ perceived use of irony was related to their processing of ironic statements: SSS
scores were related to relative processing speeds for literal and ironic statements. The
results of these experiments indicate that there are individual differences in purported use
of sarcasm that influence interpretation and processing of verbal irony.

Keywords: verbal irony; figurative languages; individual differences; sarcasm

There is now a considerable body of research in which the interpre-

tation and processing of verbal irony have been examined, and a num-
ber of different theories have been proposed to explain how people use
and comprehend verbal irony (e.g., Colston, 2000; Gibbs, 1986, 1994;
Giora, 1995; Giora, Fein, & Schwartz, 1998; Kreuz & Glucksberg, 1989;
Kumon-Nakamura, Glucksberg, & Brown, 1995). It has been argued,
however, that no extant theory of verbal irony can explain the range
and complexity of ironic language (Utsumi, 2000). In the present re-
search, we explored a neglected factor that might explain some of the
variability in the production, interpretation, and processing of ironic
utterances: individual differences.

AUTHORS’ NOTE: This research was supported by a grant to Penny M. Pexman from the
Social Sciences and Humanities Research Council (SSHRC) of Canada. The authors
thank Kibeom Lee for advice concerning the analyses.
JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY,
Vol. 23 No. 3, September 2004 244-271
DOI: 10.1177/0261927X04266809
 2004 Sage Publications
244
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 245

The notion that there are individual differences in irony usage is the
logical conclusion from some recent work exploring the social factors
that influence irony interpretation and processing. That is, there is evi-
dence from several studies that listeners believe that speakers differ in
their tendencies to use verbal irony, and that these differences are con-
veyed by certain social categories, such as occupation and gender (e.g.,
Katz & Pexman, 1997; Katz, Piasecka, & Toplak, 2001; Pexman,
Ferretti, & Katz, 2000; Pexman & Olineck, 2002a). Furthermore, these
studies have shown that this social category information is used by lis-
teners as a cue to ironic intent. Such findings imply that listeners per-
ceive speaker intent to be a function of personality factors; there seems
to be the implicit view that people differ in their tendencies to use ver-
bal irony and that those differences are reliable predictors of an indi-
vidual’s likelihood of using irony. In the present research, we sought to
test this conclusion.
It certainly seems plausible, intuitively, that some individuals are
more likely than others to use verbal irony in their own speech. In addi-
tion, some individuals may be more sensitive to ironic intent than oth-
ers, perhaps by virtue of being frequently ironic themselves. As a con-
sequence, these individuals may detect irony in situations where other
listeners would not. Although this may seem intuitive, much of the
research on figurative language comprehension has ignored the poten-
tial influence of social factors. Instead, the emphasis has tended to be
on linguistic or discourse factors. In the present research, we investi-
gated whether this neglect of social factors is justified, by exploring the
possibility that there are individual differences in the production,
interpretation, and processing of verbal irony. If these types of individ-
ual differences can be detected, then an understanding of these differ-
ences, and other social factors, may contribute to a more comprehen-
sive account of verbal irony.
The first goal of the present study was simply to establish whether
individual differences in use of irony exist and how they could be mea-
sured. A second goal was to investigate the extent to which such real
world conversational tendencies predict production, interpretation,
and processing of verbal irony in laboratory tasks.

PREVIOUS RESEARCH ON
INDIVIDUAL DIFFERENCES AND IRONY

The issue of individual differences and irony has rarely been consid-
ered in the previous literature, with only some consideration of gender
differences. Gibbs (2000) reported that men were more likely than
women to use sarcastic irony in conversation with friends. Jorgensen
(1996) examined the effect of gender on the social and emotional im-
pact of irony and reported that men were more likely than women to
perceive humor in sarcastic irony (Experiment 2). Also, women were
246 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

more likely than men to be offended or angered by sarcastic remarks

(Experiment 3).
The issue of individual differences has, however, been addressed in
the context of another form of figurative language: conversational indi-
rectness (e.g., Holtgraves, 1997; Holtgraves & Yang, 1992). Holtgraves
(1997) investigated individual differences in the production and com-
prehension of indirect requests (e.g., saying “it’s cold in here” as a
request for someone to close the door). He devised and validated the
Conversational Indirectness Scale (CIS) to measure the extent to
which individuals differ in their tendencies to express themselves in-
directly and to understand indirect meanings. The scale involves items
that were designed to tap both interpretation (10 items) and pro-
duction (9 items) of conversational indirectness. In Experiment 1,
Holtgraves showed that men rated themselves higher than women on
the production factor of the CIS, whereas no gender differences were
observed for the interpretation factor. Results of Experiment 2 showed
that higher scores on the interpretation factor were related to the ten-
dency to engage in and enjoy cognitive activity. As well, people who
rated themselves as being more assertive scored lower on the produc-
tion factor. Holtgraves (Experiment 3) employed a production task in
which participants were required to choose the statement they would
most likely use in each situation described. Performance on this task
was related to scores on the production factor of the CIS.

THE RELATIVE PROCESSING DEBATE

A central debate in irony research, and in research on other types of

figurative language, has concerned the manner in which ironic (figura-
tive) utterances are processed in comparison to literal utterances. On
one hand, according to the direct access model (Gibbs, 1994), ironic
utterances do not require special cognitive processes beyond those nec-
essary for understanding literal utterances. Accordingly, literal mean-
ings do not need to be considered before figurative meanings can be
accessed (Gibbs, 1986). Instead, ironic utterances are processed by par-
allel activation of literal and figurative meanings, and the predicted
consequence is that literal and ironic utterances should take about the
same amount of time to process.
On the other hand, according to the graded salience hypothesis
(Giora et al., 1998), salient meanings (usually the literal meanings) are
always activated first, before less salient meanings (usually the figura-
tive meaning). By this model, irony processing involves activation of
the literal meaning before the ironic, and then simultaneous consider-
ation of the literal and ironic meanings. Accordingly, the model allows
the prediction that ironic utterances should usually take longer to pro-
cess than literal utterances.
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 247

In previous research there has been evidence for both positions:

some studies have shown that ironic utterances take no longer to pro-
cess than literal utterances (e.g., Gibbs, 1986), but other studies have
shown that ironic utterances do take longer to process than literal
utterances (e.g., Dews & Winner, 1999; Giora et al., 1998). Recently,
Ivanko and Pexman (2003) demonstrated that both patterns could be
produced, depending on the degree of incongruity of the discourse con-
text. When the context presented was very negative (see the Materials
section of Experiment 2 in the current article for an example), making
the degree of incongruity between the situation and the statement
large, literal statements were read more quickly than ironic state-
ments. When the context presented was less negative and the degree of
incongruity less sizeable, however, the ironic statements were read in
the same time as or even more quickly than the literal statements.
Ivanko and Pexman argued that when the context was very negative,
the participants may have generated an expectation for a literal in-
sult; the situation was so negative that a direct insult was warranted.
Hence, reading times for literal insults were faster than for ironic
insults. In contrast, in the less negative context situation, participants
did not have the same expectation and thus reading times were equiv-
alent for literal and ironic utterances.
The argument Ivanko and Pexman (2003) made is that participants’
expectations (based, in the Ivanko and Pexman study, on social conven-
tions about the types of insults allowed in different situations) can
influence relative processing of literal and ironic statements. It fol-
lows, then, that expectations derived from other sources could also
affect relative processing. Accordingly, in the present study, we tested
whether individual differences in the perceiver’s tendency to use ironic
language might lead to different expectations for ironic language in
the speech of others and thus might also predict relative processing
times for literal and ironic utterances. In this way, we used the relative
processing issue as a test case for the potential influence of individual
differences on comprehension of ironic language.

METHODOLOGICAL ISSUES

The purpose of the present study was to investigate individual dif-

ferences and irony. Accordingly, one approach would be to study in-
dividuals’ language comprehension and production in naturalistic
settings (e.g., conversations with friends). With rare exceptions (e.g.,
Gibbs, 2000), however, research on verbal irony has involved labora-
tory tasks, somewhat removed from everyday discourse. The promi-
nent theories of irony have been devised and tested, for the most part,
using data from laboratory tasks. It is assumed that these tasks tap the
248 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

same types of interpretive processes used in understanding everyday

discourse. In the present research, we wanted to test this assumption
by examining whether performance on these types of laboratory tasks
is related to people’s self-reported, real-world conversational tenden-
cies. Consequently, our production, interpretation, and processing
tasks were taken from previous laboratory-based figurative language
research: our production task was modeled after Holtgraves’ (1997)
role-playing task, our interpretation task was borrowed from Pexman
and Olineck (2002b), and our processing task was borrowed from
Ivanko and Pexman (2003).

THE PRESENT RESEARCH

In Experiment 1, we examined the effects of an individual’s self-

rated use of sarcastic irony and indirect speech on the production and
interpretation of verbal irony. To measure self-rated use of sarcastic
irony, we devised the sarcasm self-report scale (SSS). To measure pro-
duction of irony, a role-playing task similar to that used by Holtgraves
(1997) was employed. We presented several situations, each followed
by four statements (two statements were literal and two were ironic).
The participants were asked to choose the statement that they would
most likely make in each situation. To measure interpretation of ver-
bal irony, participants were asked to read several context paragraphs,
each followed by a literal or ironic target statement, and then rated the
speaker’s intent in terms of sarcasm, mocking, politeness, and humor,
as these are all aspects of ironic speech. Given the results of past
research, we expected that there would be gender differences in the
production and interpretation of verbal irony. In addition, because the
SSS assessed participants’ tendencies to use sarcasm in their own
speech, we expected that the SSS scores would be related to the produc-
tion of verbal irony. To the extent that irony can be considered to be a
form of indirect speech, we also predicted that the CIS would be related
to the production of verbal irony, but likely to a lesser degree than the
SSS. We also hypothesized that participants’ tendency to use sarcasm
or indirect language in their own speech might be related to detection
of irony in the speech of others or to their perceptions of the social
impact of irony. For this reason, we expected that the SSS and CIS
scores would, to some degree, predict responses on the interpretation
task.
In Experiment 2, we examined how an individual’s self-rated use of
sarcasm and indirect speech may be related to their processing of ver-
bal irony. A moving windows task was used such that the participants
were presented with context paragraphs and target statements on a
computer screen, one word at a time. Participants controlled the speed
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 249

with which they read the text and the reading time for each word
was measured. We hypothesized that the same variables (i.e., gender,
SSS scores, and CIS scores) that could be related to production and
interpretation (examined in Experiment 1) would also be related to
processing.

EXPERIMENT 1

PURPOSE

The purpose of this experiment was to investigate individual differ-

ences in the production and interpretation of ironic statements.

METHOD

Participants. Participants in this experiment were 155 (78 female

and 77 male) undergraduate students, aged 17 to 42 (M = 21.59, SD =
4.53). Participants received bonus credit toward a psychology course in
exchange for their participation.

Materials. This experiment involved five paper and pencil tasks: the
production task, the interpretation task, a distracter task, the CIS, and
the SSS. The production task involved eight situations (half with a new
acquaintance and half with a best friend; see the appendix for sample
items). The stimuli were presented in Pexman and Olineck (2002a)
and were pilot tested for that experiment (in order to control for plausi-
bility and familiarity) before being chosen. Following each of the situa-
tions were four statements: two were literal (the first was somewhat
general, and the second was more personal) and two were ironic (again,
one general and one personal). Participants circled the statement they
would be most likely to make in each of the situations described.
The interpretation task involved eight context scenarios and target
statements. These stimuli were presented in Pexman and Olineck
(2002b) and were pilot tested for that experiment (in order to control
for plausibility, familiarity, and level of positivity/negativity) before
being chosen. Each context paragraph and target statement was pre-
sented in one of four conditions: (a) literal compliment (e.g., Barb and
Tony went to a high school reunion. Barb arrived dressed in a stylish
suit that she had borrowed from her sister. Tony commented, “Classy
outfit.”), (b) literal criticism (e.g., Nita and George were packing a tent
trailer for the long weekend. Nita was in charge of groceries and when
they got to the campground they opened the fridge. It was full of beer and
there were two packages of peanuts on the top shelf. George commented,
250 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

“Lousy job with the groceries.”), (c) ironic compliment (e.g., Megan and
Kristine were looking after the neighbor’s children one afternoon. The
kids were energetic, enthusiastic, and well behaved. On the way home,
Kristine said, “Those kids were not much fun.”), and (d) ironic criticism
(e.g., Laura and Lynn were having coffee at a local café. They were talk-
ing about a boring date that Laura had had the night before. They had
watched old reruns of cartoons all evening. Lynn said, “That sounds
pretty exciting.”). Each participant was presented with two context
paragraphs and statements in each condition. Following each para-
graph were five questions with 7-point rating scales: (a) is the speaker
being sarcastic (1 = not at all sarcastic, 7 = extremely sarcastic)? (b) Is
the speaker saying something polite (1 = not at all polite, 7 = extremely
polite)? (c) Is the speaker mocking someone (1 = not at all mocking, 7 =
extremely mocking)? (d) Is the speaker being funny (1 = not at all funny,
7 = extremely funny)? (e) How certain do you feel you correctly inter-
preted the speaker’s intent (1 = not at all certain, 7 = extremely
certain)?
The distracter task involved a crossword puzzle. The CIS
(Holtgraves, 1997) involved 19 statements about conversational ten-
dencies (9 assessing the production of indirect speech—for example,
“There are many times when I prefer to express myself indirectly,”—
and 10 assessing the interpretation of indirect speech—for example, “I
try to uncover peoples’ motivations by what they say”; see Holtgraves
[1997] for reliability testing of the scale). The SSS was devised by the
experimenters and was composed of 16 items to assess participants’
use of sarcasm (see Table 1). Of the items, 8 assessed general use of sar-
casm and 8 assessed use of sarcasm in specific situations. As illus-
trated in Table 1, each item was followed by a 7-point rating scale.

PROCEDURE

Participants were tested in groups of 5 to 15 people. They completed

the interpretation task first, followed by the distracter task for which
they had 10 minutes. They then completed the production task, the
CIS, and the SSS. We put the SSS last in the procedure so that partici-
pants’ self-ratings on this measure would not influence their responses
for the production or interpretation tasks.

RESULTS AND DISCUSSION

The mean ratings responses (for males and females separately) for
each task are presented in Table 2 and the frequencies for responses on
the role playing (production) task are presented in Table 3.
 Table 1
Sarcasm Self-Report Scale (SSS) Items and Mean Ratings for Experiment 1 (standard deviations in parentheses)

SSS Item Mean Rating

Likelihood that you would use sarcasm with someone you just met 4.05 (1.66)
How sarcastic do you think you are? 5.27 (1.15)
Likelihood that you would use sarcasm when insulting someone 5.68 (1.49)
Likelihood that you would use sarcasm with your best friend 6.25 (1.00)
How sarcastic would your friends say you are? 5.21 (1.34)
Likelihood that you would use sarcasm with a new colleague at work 3.69 (1.54)
Likelihood that you would use sarcasm while complimenting someone 2.90 (1.67)
How often do you make sarcastic statements during daily interactions? 4.61 (1.35)
How likely are you to make sarcastic statements in these situations?
You are out for drinks with a group of friends. The person beside you tells a hilarious story about one of their
a
colleagues from work. You begin to talk about a related experience . . . 4.08 (1.27)
You and your roommate are having a serious argument about how to share the household chores . . . 4.45 (1.78)
You score the winning point for your team in the final basketball game of the season . . . 2.98 (1.71)
You just found out that you made a huge mistake on the assignment you just handed in . . . 4.29 (1.96)
You are in a mile-long line up at the grocery store, waiting to pay for a prescription . . . 4.87 (1.75)
You just got engaged over the weekend and are telling your friends about it over coffee . . . 2.45 (1.58)
You just got a big promotion at work. You are having dinner with your family to celebrate your achievement . . . 3.29 (1.75)
You have to be at work in 15 minutes and your friend just accidentally locked your keys in the car . . . 5.05 (1.88)

Note. Complete SSS with scales and anchors can be obtained from the first author.
a. This item was eliminated from subsequent analyses as it loaded onto different factors of the SSS in Experiment 1 and Experiment 2.

251
252 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

Table 2
Means and Gender Differences for SSS, CIS, and Interpretation Task in Experiment 1
(standard deviations in parentheses)

Gender Difference
Measure Male Mean Female Mean t(153) η2

SSS factors
SSS–general sarcasm 5.39 (1.01) 5.10 (0.90) 1.83 .02
SSS–face-saving 3.88 (1.29) 3.24 (1.22) 3.18** .06
SSS–embarrassment diffusion 3.08 (1.39) 2.73 (1.14) 1.72 .02
SSS–frustration diffusion 4.61 (1.26) 4.86 (1.33) –1.17 .01
CIS factors
CIS–interpretation 44.05 (10.63) 43.67 (10.39) 0.23 .00
CIS–production 33.58 (10.56) 33.88 (9.42) –0.19 .00
Interpretation task
Literal compliment
Sarcasm 1.82 (0.85) 1.54 (0.64) 2.26* .03
Politeness 6.01 (0.81) 5.98 (0.79) 0.25 .00
Mocking 1.83 (0.88) 1.52 (0.60) 2.59* .04
Funny 1.89 (0.91) 1.51 (0.54) 3.14** .06
Certainty 5.64 (1.20) 5.85 (0.93) –1.26 .01
Literal criticism
Sarcasm 2.68 (1.34) 2.76 (1.39) –0.34 .00
Politeness 2.29 (0.97) 1.92 (0.85) 2.47* .04
Mocking 4.05 (1.61) 3.49 (1.54) 2.21* .03
Funny 2.60 (1.13) 2.15 (1.14) 2.50* .04
Certainty 5.38 (1.11) 5.24 (0.99) 0.79 .00
Ironic compliment
Sarcasm 4.75 (1.67) 5.36 (1.49) –2.39* .04
Politeness 2.97 (1.30) 2.69 (1.44) 1.28 .01
Mocking 3.40 (1.52) 3.12 (1.48) 1.19 .01
Funny 3.57 (1.58) 3.84 (1.40) –1.12 .01
Certainty 5.10 (1.17) 5.17 (1.07) –0.38 .00
Ironic criticism
Sarcasm 5.96 (0.93) 6.09 (0.71) –0.97 .01
Politeness 2.91 (1.07) 2.61 (1.11) 1.71 .02
Mocking 3.97 (1.36) 3.82 (1.67) 0.63 .00
Funny 4.45 (1.30) 4.54 (1.20) –0.48 .00
Certainty 5.51 (1.01) 5.60 (0.83) –0.61 .00
2
Note. η = effect size.
*p < .05. **p < .01.

EVALUATING THE
SARCASM SELF-REPORT SCALE (SSS)

We first determined whether some of the 16 items on the scale were

highly related to each other and could be clustered together as a sub-
scale. A principal components analysis (PCA) (see Table 4) indicated
that there were four clusters (or components) of related items. Reliabil-
ity coefficient alphas were calculated for each of these four subscales
(see Table 4). Although the reliability coefficients were in an accept-
able range for the first three subscales, the coefficient for the fourth
subscale was somewhat low. The reliability coefficient for the same
subscale with different participants in Experiment 2, however, was
 Table 3
Frequencies of Responses and Gender Differences on All Items in the Production Task in Experiment 1

Statement Choice
General General Personal Personal
Ironic Literal Ironic Literal
Gender
2
Item Male Female Male Female Male Female Male Female Difference χ

New acquaintance items

Planning to surprise spouse 41 20 2 2 33 55 1 1 12.72**
Chatting at a party 14 14 10 11 20 16 33 37 0.71
At a dinner meeting 36 34 19 18 8 21 14 5 10.17*
Lunch order 17 20 41 47 12 11 7 0 7.69*
Best friend items
At a conference 42 49 28 25 2 2 5 2 1.99
Out for drinks 18 26 6 3 50 48 3 1 3.49
While shopping 0 6 60 62 17 10 0 0 7.84*
While selling chocolates 10 10 21 17 26 30 20 21 0.73

Note. For examples of production task statements, see appendix.

253
254 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

substantially higher ( = .65). Mean scores were calculated for each of

the four components (see Table 2) and all subsequent analyses were
conducted using these component scores.
The first SSS component, which we termed general sarcasm,
included six items that assessed how sarcastic individuals believed
themselves to be in general (e.g., How sarcastic are you? How sarcastic
would your friends say you are?) as well as their likelihood of using sar-
casm in situations where it is more typically used (i.e., in a negative sit-
uation with the intent to insult).
The second SSS component, termed face-saving, included three
items that assessed the likelihood of using sarcasm with new acquain-
tances and when complimenting. One might wonder what sarcasm
with new acquaintances has in common with ironic compliments. We
reasoned that these are the situations in which sarcasm is the most
risky. That is, using sarcasm with a new acquaintance or when compli-
menting carries a stronger risk of being misunderstood than does
using sarcasm with friends or when criticizing. With a new acquain-
tance there is likely to be less common ground between speaker and lis-
tener, and this common ground is important to irony comprehension
(e.g., Gibbs, 1986; Pexman & Zvaigzne, 2004). As well, ironic compli-
ments (e.g., saying “you’re a rotten friend” after a friend has done some-
thing gracious) can be more difficult to understand than ironic insults
because ironic compliments do not echo the social norm of making posi-
tive statements (Kreuz & Glucksberg, 1989). If someone used sarcasm
in these situations, despite the higher risk of misunderstanding, it
would likely be because they valued the face-saving function that
sarcasm serves. That is, sarcasm can potentially soften a criticism or
weaken a compliment (e.g., Dews & Winner, 1995; Pexman & Olineck,
2002b). As proposed by the tinge hypothesis (Dews & Winner, 1995),
the positive surface meaning of the statement serves to mute the neg-
ative intended meaning. Similarly, the negative surface meaning of
an ironic compliment mutes the positive intended meaning of the
statement.
The third component, termed embarrassment diffusion, also
included three items. These items all described very positive situ-
ations (e.g., score the winning point in a basketball game). Using a
sarcastic statement in such a positive situation is a way for the
speaker to be modest: sarcasm decreases positivity and downplays the
accomplishment.
Finally, the fourth component, termed frustration diffusion,
involved three items that described very annoying situations (e.g., a
friend locked your keys in the car when you are already running late).
Use of sarcasm in these situations serves to mock frustrating circum-
stances. Thus, the four components of the SSS map onto four functions
of sarcastic irony and confirm that this speech form can serve several
different purposes. Overall mean scores on each of the four factors sug-
 Table 4
Factor Loadings From Principal Components Analysis of Sarcasm Self-Report Scale (SSS) in Experiment 1

SSS Factor
1 2 3 4
General Face- Embarrassment Frustration
2
SSS Item Sarcasm Saving Diffusion Diffusion η

Likelihood using sarcasm with someone just met .26 .85 .05 –.14 .81
How sarcastic are you .69 .48 .06 .18 .74
Likelihood using sarcasm when insulting .71 –.09 .33 –.09 .63
Likelihood using sarcasm with best friend .48 .43 .09 .16 .45
How sarcastic friends say you are .69 .36 .17 .14 .65
Likelihood using sarcasm with new colleague at work .14 .84 .16 .08 .74
Likelihood using sarcasm when complimenting .08 .46 .26 .15 .31
How often do you use sarcasm .68 .46 .17 .15 .73
Likelihood using sarcasm when out for drinksa .45 .17 .33 –.47 .56
Likelihood using sarcasm when in argument with roommate .54 .07 –.14 .14 .34
Likelihood using sarcasm when score the winning point .19 –.01 .68 –.04 .51
Likelihood using sarcasm when made mistake on assignment .03 .03 .14 .64 .43
Likelihood using sarcasm when in long line .46 .01 .02 .49 .45
Likelihood using sarcasm when got engaged and telling friends –.07 .40 .60 .21 .58
Likelihood using sarcasm when got promotion and telling family .06 .28 .81 .03 .74
Likelihood using sarcasm when friend locked your keys in car .24 .12 –.03 .69 .54
Eigenvalues 5.06 1.68 1.29 1.17
% of variance 31.62 10.49 8.05 7.29
Cronbach’s alpha .79 .71 .64 .47

Note. Boldface indicates highest factor loadings for each item.

a. This item was eliminated from subsequent analyses as it loaded onto different factors of the SSS in Experiment 1 and Experiment 2.

255
256 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

Table 5
Correlations between Individual Difference Measures for Experiment 1

1 2 3 4 5 6 7

1. SSS–general sarcasm —
2. SSS–face-saving .52** —
3. SSS–embarrassment diffusion .33** .42**
4. SSS–frustration diffusion .40** .18* .15 —
5. Age –.07 –.08 –.02 –.06 —
6. CIS–interpretation .07 .05 .09 .15 .03 —
7. CIS–production .32** .23** .19* .20* –.04 .48** —

*p < .05. **p < .01.

gest that sarcasm is used most commonly in negative situations (fac-

tors 1 and 4) and less commonly, but occasionally, in positive situations
or with new acquaintances (factors 2 and 3).
Correlations between the four SSS factors and the other individual
difference variables (age, CIS-production, CIS-interpretation) are pre-
sented in Table 5.

INDIVIDUAL DIFFERENCES ANALYSES

Production task. We next examined whether our individual differ-

ence measures could predict participants’ choices on the production
task. To accomplish this, we conducted discriminant function analyses
(DFA) on responses for one new acquaintance item and one best friend
item on the production task (these two items are listed in the appen-
dix). These items were chosen for the DFA because participant
responses for these items were fairly well distributed across the four
statement choices. In this analysis, the variable to be predicted was
statement choice on the production task. The individual differ-
ence (predictor) variables were as follows: age, gender, CIS–
interpretation, CIS–production, SSS–general sarcasm, SSS–face-
saving, SSS–embarrassment diffusion, and SSS–frustration diffusion.
For the new acquaintance item, results showed that the only signifi-
cant predictor of statement choice was SSS–face-saving, Wilks’s λ =
.90, F(3, 150) = 5.38, p < .01 (see Table 6 for classification summary).
For the best friend item, the only significant predictor of statement
choice was SSS–general sarcasm, Wilks’s λ = .88, F(3, 150) = 6.63, p <
.001. These results suggest that a person’s self-rated use of sarcasm is
one of the factors that predicts the type of statement that she or he will
use in conversation.
These results provide validation for the SSS. The SSS–face-saving
factor was a significant predictor for response choice for the new
acquaintance item, and the SSS items that loaded onto this component
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 257

included items that assessed the likelihood of using sarcasm with

someone you just met and with a new colleague from work. Similarly,
the SSS–general sarcasm factor was a significant predictor of response
choice for the best friend item, and one of the items that loaded onto
this factor assessed the likelihood of using sarcasm with your best
friend.

Interpretation task. We next examined whether our individual dif-

ference measures could predict participants’ choices on the interpreta-
tion task. Accordingly, linear regression analyses were conducted. The
variables to be predicted were participants’ ratings for each irony con-
dition in the interpretation task. The predictor variables were the
same individual difference variables used for the DFA above.1 The
results of these linear regression analyses are reported in Table 7.
These analyses showed that none of the individual difference variables
predicted sarcasm or humor ratings for the ironic criticisms. Three of
the individual difference variables, gender, CIS–production, and SSS–
face-saving, predicted politeness ratings for ironic criticisms, R2 = .09,
p < .01. That is, male participants tended to perceive the ironic criti-
cisms to be more polite than the female participants did, individuals
who rated themselves likely to use indirect speech also tended to per-
ceive the ironic criticisms to be more polite, and those who rated them-
selves likely to use sarcasm for its face-saving function tended to per-
ceive the ironic criticisms to be less polite. One of the individual
difference variables, SSS–general sarcasm, predicted mocking ratings
for ironic criticisms, R2 = .04, p < .05. That is, those who rated them-
selves likely to use sarcasm in general tended to perceive ironic criti-
cisms as more mocking. Two of the individual difference variables,
SSS–general sarcasm and SSS–face-saving, predicted certainty rat-
ings for ironic criticisms, R2 = .09, p < .01. That is, participants with
higher self-ratings of likelihood of using sarcasm for both factors
tended to be more certain of their interpretations of ironic criticisms.
For the ironic compliments, we found that none of the individual dif-
ference variables predicted politeness, humor, or certainty ratings. Im-
portantly, one of the individual difference variables, gender, predicted
sarcasm ratings for ironic compliments, R2 = .04, p < .05. That is, female
participants tended to rate the ironic compliments as more sarcastic
than male participants did. There were also three individual differ-
ence variables—age, CIS–production, and SSS–frustration diffusion—
that predicted mocking ratings for ironic compliments, R2 = .10, p < .01.
That is, older participants found the ironic compliments to be less
mocking, individuals who perceived themselves as using indirect
speech more often also perceived the ironic compliments to be less
mocking, and those who were more likely to use sarcasm to diffuse
frustration perceived ironic compliments to be more mocking.
258
Table 6
Classification Summary for Discriminant Function Analysis for Statement Choices on Two Items on the Production Task in Experiment 1

Predicted Statement Choice

General Ironic General Literal Personal Ironic Personal Literal
Actual Statement Choice n n % n % n % n %

New acquaintance item

General ironic 28 19 67.9 0 0.0 3 10.7 6 21.4
General literal 21 10 47.6 0 0.0 2 9.5 9 42.9
Personal ironic 36 17 47.2 0 0.0 5 13.9 14 38.9
Personal literal 69 19 27.5 0 0.0 8 11.6 42 60.9
Best friend item
General ironic 20 2 10.0 2 10.0 10 50.0 6 30.0
General literal 38 4 10.5 23 60.5 8 21.1 3 7.9
Personal ironic 56 10 17.9 9 16.1 28 50.0 9 16.1
Personal literal 41 9 22.0 12 29.3 16 39.0 4 9.8

Note. For examples of production task statements, see appendix. Overall percentage of correctly classified cases for the new acquaintance item was 42.9%
and for the best friend item was 36.8%.
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 259

Table 7
Regression Analysis Summary for the Interpretation Task Ratings of Ironic Criticisms
and Ironic Compliments in Experiment 1

Variable B SEB β

Ironic criticisms
Politeness ratings
Gender .40 .18 .18*
CIS–production .03 .01 .25**
SSS–face-saving –.17 .07 –.20*
Mocking ratings
SSS–general sarcasm .31 .13 .20*
Certainty ratings
SSS–general sarcasm .18 .09 .18*
SSS–face-saving .11 .07 .15
Ironic compliments
Sarcasm ratings
Gender –.62 .26 –.19*
Mocking ratings
Age –.06 .03 –.19*
CIS–production –.02 .01 –.15
SSS–frustration diffusion .25 .09 .22**

*p < .05. **p < .01.

Although all of these relationships are modest in strength, these

results indicate that individual differences in age, gender, and self-
rated use of sarcasm and indirect speech are related to the interpreta-
tion given to an ironic statement.

Gender differences. To examine gender differences for responses on

the production task, chi-square tests were conducted and are pre-
sented in Table 6. To examine gender differences for responses on the
interpretation task, CIS, and SSS, a series of planned comparisons (t
tests) were conducted and are presented in Table 2.
For the production task, male participants’ responses differed from
female participants’ responses on three new acquaintance items and
one best friend item. In general, for the new acquaintance items,
women chose the personal ironic statement more often than men did,
whereas, for the best friend items, women tended to choose the general
ironic statement more often than men did. For the interpretation task,
there was a gender difference only for the ironic compliments and not
for the ironic insults.
To examine how male and female scores differed across the four fac-
tors of the SSS, a 2 (Gender) × 4 (SSS factors) mixed model analysis of
variance was conducted. There was a significant Gender by SSS factors
interaction, F(1, 152) = 6.96, p < .01, MSE = 1.23, η2 = .04. Planned com-
parisons for each SSS factor are presented in Table 2 and revealed the
nature of the interaction: Men reported that they were more likely
260 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

than women to use sarcasm for three out of the four SSS factors,
whereas women had higher scores than men for the other factor. These
results suggest that men and women use irony for different reasons.
The three factors for which men had higher scores all represent ways of
using irony to reduce social vulnerability, which Lampert (1996) sug-
gested is the primary motive for men using conversational humor. For
the fourth factor, frustration diffusion, the men and women did not
have significantly different scores and, in fact, the women’s scores
tended to be somewhat higher than the men’s. The situations included
in this factor involved something negative happening to the speaker
and the sarcastic comment serves a self-mocking function. In these sit-
uations, using irony would most likely increase the social vulnerability
of the speaker and, as this is less likely to be a desirable outcome for
men, the men’s self-ratings tended to be somewhat lower than the
women’s.

EXPERIMENT 2

PURPOSE

The purpose of this experiment was to investigate individual differ-

ences in the processing of verbal irony. In particular, we tested whether
relative processing speed for ironic and literal statements varies as
a function of individual difference variables. To examine processing,
we presented ironic and literal statements in two types of context:
strongly negative and weakly negative. Ivanko and Pexman (2003)
found that this manipulation of degree of incongruity influenced rela-
tive processing for literal and ironic statements. We included this con-
text manipulation in the present study to capture a broad range of
irony-processing behavior.
We expected that the individual difference measures might be
related to processing (reading times) for reading locations at the end of
the target statement and also after the target statement itself (in the
subsequent sentence) as Pexman et al. (2000) showed that processing
of ironic intent continues into the sentence that follows the target
statement.

METHOD

Participants. Participants in this experiment were 96 (48 female

and 48 male) undergraduate students, aged 18 to 47 (M = 21.87, SD =
4.24). Another three participants were tested and excluded from anal-
yses because they made more than 30% errors on the comprehension
questions. Participants received bonus credit toward a psychology
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 261

course in exchange for their participation. There were four versions of

the stimuli presented in this experiment in order that, across partici-
pants, each statement appeared in each context condition.

Materials. This experiment consisted of a computer task, the CIS,

and the SSS. The stimuli used for the computer task were the same 12
contexts and 12 statement pairs (in a 2 [Context: strongly negative,
weakly negative] by 2 [Statement: literal, ironic] combination of condi-
tions) and 63 filler items that were extensively pilot tested (to control
for plausibility, familiarity, and level of positivity/negativity) and pre-
sented in Ivanko and Pexman (2003; e.g., Sam agreed to pick Christo-
pher up after school. Sam never arrived to pick up Christopher and
never called to say why [strongly negative] / Sam arrived 1 hour late
and apologized [weakly negative]. The next day Christopher is explain-
ing to Jodi what happened. Christopher says: Sam is a rotten friend [lit-
eral statement] / Sam is a nice friend [ironic statement]. Christopher
and Jodi were walking home from school.)
To ensure that participants were attending to what they were read-
ing, a simple yes/no comprehension question was asked after each con-
text paragraph and statement were read. The comprehension ques-
tions did not involve assessment of speaker intent for the target
statements.
The stimuli were presented on a 17-inch Sony monitor controlled by
a Macintosh G3 and presented using PsyScope (Cohen, MacWhinney,
Flatt, & Provost, 1993) in a “one word at a time” moving window for-
mat. A PsyScope button box was used for participants to advance the
text and respond to the comprehension questions. The button box
recorded the reading times for each word (the interval between succes-
sive button presses), as well as processing times for the space between
each sentence, with millisecond accuracy.

Procedure. Participants were tested individually. They completed

the computer task first. Each context paragraph and statement was
presented on the computer screen as a series of dashes for each word,
with spaces between words. The participants were instructed to press
the middle button on the button box to reveal the first word in the para-
graph. Each subsequent button press revealed the next word, replac-
ing the previous word with dashes. The participants were instructed to
read as if reading a newspaper or magazine. The paragraphs were pre-
sented to each participant in a different random order. The target
statements were always presented on one line on the computer screen
so that participants did not switch lines in the middle of reading the
target statement.
Following the computer task, the participants completed the CIS
and then the SSS.
262 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

RESULTS AND DISCUSSION

The data in this experiment included the reading times (RT) for
each of the words in the target statements and wrap-up sentences,
along with scores for the CIS and the SSS (the mean scores on the CIS
and the SSS were virtually identical to the mean scores on these scales
from Experiment 1). Before analyses were conducted, 9.7% of the data
were excluded: 0.2% for reading times shorter than 100 ms or longer
than 2000 ms, and 9.5% for incorrect answers on the comprehension
questions. Ten reading locations were analyzed: the fifth word in the
target statement, the space between the target and the wrap-up sen-
tence, and the eight words in the wrap-up sentence (e.g., . . . friend.
[space] Christopher and Jodi were walking home from school).

EVALUATING THE
SARCASM SELF-REPORT SCALE (SSS)

To determine whether the same four clusters of items (factors)

emerged for the SSS as in Experiment 1, we conducted a PCA with the
SSS data from Experiment 2. The same four factors emerged with the
items loading onto the same factors as in Experiment 1. One item (like-
lihood of using sarcasm when out for drinks) loaded onto a different
factor and was therefore excluded from the analyses in both experi-
ments. Cronbach’s alpha was again calculated for each factor and these
values were within the acceptable range (general sarcasm: α = .83,
face-saving: α = .77, embarrassment diffusion: α = .79, frustration dif-
fusion: α = .65).

INDIVIDUAL DIFFERENCES ANALYSES

To examine individual differences in relative processing times for

ironic and literal statements, we first needed to devise a measure of
relative processing. This was achieved by calculating a difference score
(RTironic – RTliteral) for each reading location in both the strongly nega-
tive and weakly negative context conditions (see Figure 1)2. Next, we
examined whether our individual differences measures could predict
participants’ relative processing times (difference scores) from the
online reading task. For this, linear regression analyses were con-
ducted. In these analyses, the variables to be predicted were the dif-
ference scores at each reading location. The individual difference (pre-
dictor) variables were as follows: age, gender, CIS–interpretation,
CIS–production, SSS–general sarcasm, SSS–face-saving, SSS–
embarrassment diffusion, and SSS–frustration diffusion. See Table 8
for the regression analysis results. There were six reading locations in
the strongly negative contexts where relative processing times were
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 263

predicted by individual difference variables and one reading location

in the weakly negative contexts where relative processing times were
predicted by individual difference variables. These results indicate
that individual differences are related to relative processing times for
literal and ironic statements. For example, those who rated themselves
as being more likely to use sarcasm in frustrating situations or with
new acquaintances, or to look for or use indirect speech, tended to have
more negative difference scores at three reading locations in the
strongly negative context condition, suggesting relatively rapid read-
ing of ironic statements compared to literal statements. In contrast,
older participants tended to have more positive difference scores at
three reading locations in the strongly negative context condition, sug-
gesting relatively slow reading of ironic statements compared to literal
statements. Those who rated themselves as being more likely to use
sarcasm in general tended to have more positive difference scores at
the second word in the wrap-up sentences in the weakly negative con-
texts. Most of these differences were in the direction we expected, with
higher SSS scores related to relatively faster processing for irony. The
exception was the relationship of SSS–general sarcasm scores to
slower processing for ironic statements at the second word in the wrap-
up sentence. It is possible to derive several explanations for this rela-
tionship, but our data do not allow us to distinguish among them. For
example, participants with high self-ratings on the general sarcasm
factor of the SSS might be slower to process ironic statements at the
second wrap-up word because they are resolving the speaker’s ironic
intent here, earlier than other participants; participants with lower
general sarcasm self-ratings might be proceeding to subsequent wrap-
up words relatively more rapidly because they are trying to determine
whether they will be given any more information about speaker intent
further downstream in the wrap-up sentence. Alternatively, the reason
for the relationship could be that participants high on SSS–general
sarcasm are more sensitive to the mocking aspect of ironic criticisms
(as the results of Experiment 1 suggested) and so tended to pause at
the second word of the wrap-up sentences as they considered the social
ramifications of the ironic remarks. Other participants, with lower
SSS–general sarcasm scores, might not have considered those con-
sequences and so proceeded more quickly.

GENERAL DISCUSSION

The results of these experiments showed that individual differences

can be detected in the use of sarcasm and that those differences are
related to the production, interpretation, and processing of verbal
irony. It is important to note that although there was evidence, in the
present experiments, that individual difference measures predicted
264 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

30

25

SSS - general
20
sarcasm (+)

CIS - production (-)

15
Difference score

10
Age (+) strongly negative
weakly negative
5

0
5th word space wrap-up 1 wrap-up 2 wrap-up 3 wrap-up 4 wrap-up 5 wrap-up 6 wrap-up 7 wrap-up 8

-5

SSS - frustration diffusion (-)

-10
Age (+) CIS - interpretation (-)
SSS - frustration diffusion (-) Gender (+)
Age (+)
-15 SSS - face-saving (-)
Reading location

Figure 1. Difference scores (RTironic – RTliteral) for each reading location for
the strongly negative and weakly negative context conditions.
Using these difference scores as the criterion variable we present
the predictor variables for each significant regression equation.
Predictor variables with significant, unique relationships to the
criterion variable are presented in bold, and the directions of
these relationships are presented in parentheses.

performance on various irony tasks, the amount of variance explained

by these individual difference variables was small. Clearly, there are
other variables, in addition to those measured here, that influence the
production, interpretation, and processing of verbal irony.
In Experiment 1, participants who perceived themselves as using
sarcasm more frequently (high SSS scores) tended to choose the ironic
statements more often than those who perceived themselves as using
sarcasm less frequently (low SSS scores). This is not altogether sur-
prising because the SSS is meant to assess tendencies in production of
sarcasm. Notably, the production task we used was only a rough mea-
sure of irony production. A more ecologically valid test of production
(e.g., a situation involving face-to-face interaction) needs to be used in
order to obtain a more fine-grained analysis of the effects of individual
differences on irony production.
In Experiment 1, we also observed that a person’s own perceived use
of sarcastic irony in certain situations was related to their perceptions
of speaker intent for others’ ironic statements. For ironic criticisms, our
individual difference measures were related to perceptions of mocking
and politeness. Those who perceive themselves to be more sarcastic
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 265

Table 8
Regression Analysis Summary for Difference Scores (RTironic–RTliteral) for Reading
Locations in Strongly Negative Context and Weakly Negative Context in Experiment 2
2
Variable R B SEB β

Strongly negative context

5th word in target statement .11
Age 12.94 5.19 .25*
SSS–frustration diffusion 26.80 14.50 –.19
Space after target statement .06
SSS–frustration diffusion –21.02 9.13 –.24*
Wrap-up 1 .09
Age –9.89 3.23 –.31**
Wrap-up 2 .04
CIS–production –2.46 1.20 –.21*
Wrap-up 3 .05
CIS–interpretation –2.81 1.27 –.23*
Wrap-up 7 .18
Gender 38.18 22.89 .17
Age 8.96 2.60 .34**
SSS–face-saving –15.97 8.75 –.19
Weakly negative context
Wrap-up 2 .06
SSS–general sarcasm –30.71 12.38 .25*

*p < .05. **p < .01.

speakers may recognize more readily that one goal of using verbal
irony is for the speaker to save face (e.g., Jorgensen, 1996) while still
criticizing the target. By using irony as an indirect way of criticiz-
ing, the speaker can be perceived as trying to be polite. Participants
with higher CIS–production scores seemed more apt to recognize this
politeness function, whereas female participants tended to recognize
the critical (and thus impolite) function of ironic criticisms.
Notably, sarcasm ratings for ironic criticisms did not differ as a func-
tion of any of the individual difference measures. Thus, it appears that
participants were fairly uniform in their evaluations of degree of sar-
castic intent for others’ remarks, although it is certainly possible that
these ratings differ as a function of some other individual difference
variable (e.g., verbal aggression, sense of humor) not measured in the
present study. The consistency of sarcasm ratings we observed for
ironic criticisms (as illustrated in Table 2, mean sarcasm ratings for
ironic criticisms were high and SDs relatively small) may be related to
the fact that these are the more typical form of verbal irony (Gibbs,
2000).
The interpretation of sarcastic intent for the ironic compliments, in
contrast, appeared to be more sensitive to individual differences. The
female participants rated the ironic compliments as being more sar-
castic than did the male participants. This difference, however, could
266 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

simply reflect greater sensitivity on the part of female participants to

the negative tinge of ironic compliments (e.g., Dews & Winner, 1995;
Pexman & Olineck, 2002b). Although most participants will say that
ironic compliments are “sarcastic,” the term sarcasm does involve a
biting, hurtful element. For instance, the Oxford English Dictionary
offers the following definition of sarcasm: “the use of bitter or wound-
ing, esp. ironic, remarks” and notes that the term comes from Greek
terms for “gnash the teeth, speak bitterly.” Women in our study may
have given the ironic compliments higher sarcasm ratings because
they considered sarcasm to involve negativity and were somewhat
more sensitive to this aspect of the remarks. This explanation gains
support from the finding that female participants in Experiment 1
found even literal criticisms to be less polite than male participants
did. Women could generally be more sensitive to the social impact of
negative comments.
Participants who rated themselves as being relatively likely to use
sarcasm for frustration diffusion tended to perceive the ironic compli-
ments as more mocking. As well, those who were older and those who
reported a tendency to use indirect speech found these statements to
be less mocking. These participants may have been less sensitive to the
negative tinge of ironic compliments, while participants who used sar-
casm for frustration diffusion may have been more sensitive to that
negative, mocking tinge.
The gender differences we reported in the present research repli-
cated Jorgensen’s (1996) observed differences between men and
women in the social impact of ironic speech. In particular, male and
female participants in the present study had different perceptions
about the politeness of ironic statements. These results are also consis-
tent with findings reported by Holtgraves and Yang (1992) as those
authors found that the perceived politeness of indirect requests dif-
fered as a function of gender.
Our results also replicate Gibbs’ (2000) observation that in natural
conversations men were more likely to use sarcastic irony, but in the
present study, this finding was observed in self-ratings. That is, on the
SSS, male participants reported higher likelihoods of using sarcasm
than female participants did in most contexts. Lampert (1996) argued
that men and women use conversational humor for different rea-
sons. For instance, men tended to use self-directed humor “in a self-
protective manner to reduce social vulnerability, whereas for women, it
worked more to increase social vulnerability and to promote inti-
macy” (p. 585). These different goals may explain some of the gender
differences we observed for items on the SSS. Irony can serve the self-
protective function that Lampert claims is important to men and,
indeed, men’s ratings suggested they were more likely to use irony
in most situations. If, in contrast, women’s goals are to increase inti-
macy, then irony is not likely to achieve this, except when it is self-
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 267

deprecating. Certainly, women reported lower likelihoods of using

sarcasm in almost every situation except the circumstances where sar-
casm was self-critical. This explanation could also be applied to the
gender differences observed in choices on the production task. With a
new acquaintance, women might try to promote intimacy and thus
tended to choose the more personal statement. With a best friend, inti-
macy would already be achieved and so here women tended to choose
the general statement. These gender differences should be explored in
future research in order to be understood more fully. Explanations for
these gender differences likely involve complex sociocultural factors
(e.g., Tannen, 1993) beyond the scope of the present work.
Importantly, in Experiment 2, individual differences were shown to
predict the way in which verbal irony is processed. Individual differ-
ence measures were related to relative processing of ironic and literal
statements as early as the point at which participants read the last
word in the target statement, and this indicates that communicative
tendencies can influence early language processing.
These results have implications for theories of verbal irony. Theo-
ries of irony processing predict that ironic statements should generally
take longer to read than literal statements (e.g., Giora et al., 1998), or
that ironic statements should not take longer to read than literal state-
ments (e.g., Gibbs, 1986). They do not predict both. Yet this is essen-
tially what we observed. As illustrated in Figure 1, participants’ rela-
tive reading times for literal and ironic statements were predicted by
several of the individual difference variables. At different points in
processing, some individuals tended to show the pattern predicted by
Giora, with a tendency toward faster reading times for literal state-
ments, and some tended to show the pattern predicted by Gibbs, with a
tendency toward equivalent reading times for literal and ironic state-
ments. It is perhaps not surprising that there is individual variability
in reading times for literal and ironic statements. What is more strik-
ing is that some of this variability could be predicted by participants’
tendencies to use sarcasm in different contexts and by their tendencies
to produce and interpret indirect speech.
It seems possible that the discrepant findings reported in previous
studies on relative processing times for ironic and literal statements
(e.g., Gibbs, 1986, vs. Giora et al., 1998) might be attributable to indi-
vidual differences. One might wonder how individual differences could
be relevant if the participants in those previous studies were all under-
graduate university students. We would argue that in any group of uni-
versity undergraduates, there will be variability in use of ironic lan-
guage due, quite possibly, to dozens of different factors, like cultural
background, personality traits, sense of humor, social experience, and
so on. When one recruits the usual number of participants for a given
experiment there will, by chance, be different proportions of males
and females, different concentrations of personality traits, and so on,
268 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

within that group of participants. No two samples of participants will

be quite the same in terms of these social factors, and we have shown
that these factors influence processing of ironic and literal utterances.
Thus, across experiments, it will be possible to observe different over-
all patterns of results.
How might theories of irony account for these individual differ-
ences? Gibbs (1986) emphasized that common ground between
speaker and addressee is important for understanding sarcasm or
irony. More specifically, to comprehend irony, the listener must under-
stand the speaker’s metarepresentational thoughts (e.g., Colston &
Gibbs, 2002). The listener must judge the speaker’s belief about an
attributed idea or utterance. To explain the individual differences
observed in the present study, it could be assumed that individuals dif-
fer in the inferences they make about speakers’ beliefs. There could be
differences in individuals’ facility with making such second-order
inferences, and there could be differences in individuals’ tendencies to
infer particular attitudes on behalf of the speaker. For instance, indi-
viduals with relatively high scores on SSS–frustration diffusion may
have more knowledge of or familiarity with counterfactual, sarcastic
utterances used in very negative events, and these were the type of
items presented in the strongly negative contexts in the present
Experiment 2. Consequently, they are able to process these statements
quite readily.
Giora (e.g., Giora et al., 1998) argued that processing is determined
by meaning salience and, furthermore, that “context has a limited role”
(p. 84). In terms of the individual difference effects, the theory would
seem to have to assume that salience differs among individuals. For
instance, individuals high on the SSS–frustration diffusion or CIS–
interpretation subscales could find the ironic meanings of our target
statements to be more salient. This assumption seems somewhat
implausible, because Giora argued that the salient meaning of most
utterances is literal, and that the only exceptions are highly familiar
ironies. Participants in pilot studies rated our statements as only
moderately familiar. It therefore seems unlikely, even for high SSS–
frustration diffusion individuals, that these statements could have
been familiar ironies.
There is another theory of verbal irony that seems able to explain
the pattern of results observed in the present study. Utsumi’s (2000)
implicit display theory involves three central claims: (a) an ironic
utterance assumes an ironic environment, which is defined as a “situa-
tional setting which motivates verbal irony” (p. 1778); (b) an ironic
utterance implicitly displays the ironic environment; and (c) irony is a
prototype-based category, subject to typicality effects (utterances that
are more typical of implicit displays will be perceived as more ironic).
Utsumi’s (2000) theory gains a great deal of flexibility from this last
assumption, allowing some utterances to be more ironic than others.
Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 269

This may also allow an explanation of our individual difference results.

If irony is a category, then it should show effects of expertise, as other
categories do (e.g., Tanaka & Taylor, 1991). Individuals with high SSS–
frustration diffusion or CIS scores may have more elaborate mental
categories for verbal irony, which lead them to have different expecta-
tions for irony and facilitated processing of ironic statements relative
to literal statements.
The present findings also suggest that there is some relationship
between laboratory-based tasks involving detection of verbal irony
and real-world conversational tendencies. We believe that this offers
some validation for laboratory tasks. It also seems likely that research-
ers studying verbal irony may be able to explain some variance in
responses by including this type of self-rating measure.

CONCLUSION

The present findings confirm the implicit belief held by participants

in previous studies (e.g., Katz et al., 2001; Pexman et al., 2000; Pexman
& Olineck, 2002a) that speakers differ in their tendencies to use sar-
castic irony, and that those differences predict the types of statements
speakers make (i.e., literal vs. ironic). The present results take this fur-
ther, however, to show that a speaker’s tendency to use sarcasm affects
the way they interpret the speech of others, in terms of both processing
and product.

APPENDIX
Sample Items From the Production Task (Experiment 1)
1. While you and a new acquaintance from work, Sara, were chatting at a
party, they noticed a colleague across the room. She was standing alone
holding a drink and a CD. You comment:

1) “She’s the life of the party”

2) “She’s not the partying type”
3) “She is having a good time tonight”
4) “She is not having a good time tonight”

2. You and your best friend, Arthur, were selling chocolates door to door to
raise money for a charity event. You and Arthur approach a house and
tried to sell your candy to a mean couple. On the way to the next house,
you comment:

1) “Selling these is easy”

2) “Selling these is difficult”
3) “People in this community are nice”
4) “People in this community are mean”
270 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / September 2004

NOTES

1. We tested for multicollinearity in all of the regression analyses reported in this arti-
cle and in all cases tolerance values were in the acceptable range (close to 1).
2. Our comparisons were for literal and ironic statements presented in the same con-
text. In the previous literature, there has been some debate over which comparisons are
appropriate, with some advocating for comparisons involving the same statement pre-
sented in ironic (negative) versus literal (positive) contexts (e.g., Giora, Fein, & Schwartz,
1998). We have found in previous pilot studies that negative contexts induce slower read-
ing or processing, and that this slower processing spills over into reading of the target
statement. Positive contexts induce much faster processing that again, spills over into
the target statement. This gives an unfair advantage to statements presented in positive
contexts and, in previous studies, these have tended to be the literal statements.

REFERENCES

Cohen, J. D., MacWhinney, B., Flatt, M., & Provost, J. (1993). PsyScope: An interactive
graphic system for designing and controlling experiments in the psychology labora-
tory using Macintosh computers. Behavior Research Methods, Instruments, and Com-
puters, 25, 257-271.
Colston, H. L. (2000). On the necessary conditions for verbal irony comprehension.
Pragmatics and Cognition, 8, 277-324.
Colston, H. L., & Gibbs, R. W., Jr. (2002). Are irony and metaphor understood differently?
Metaphor and Symbolic Activity, 17, 57-80.
Dews, S., & Winner, E. (1995). Muting the meaning: A social function of irony. Metaphor
and Symbolic Activity, 10, 3-19.
Dews, S., & Winner, E. (1999). Obligatory processing of literal and nonliteral meanings in
verbal irony. Journal of Pragmatics, 31, 1579-1599.
Gibbs, R. W., Jr. (1986). On the psycholinguistics of sarcasm. Journal of Experimental
Psychology: General, 115, 3-15.
Gibbs, R. W., Jr. (1994). Figurative thought and figurative language. In M. A.
Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 411-446). New York: Academic
Press.
Gibbs, R. W., Jr. (2000). Irony in talk among friends. Metaphor and Symbolic Activity, 15,
5-27.
Giora, R. (1995). On irony and negation. Discourse Processes, 19, 239-264.
Giora, R., Fein, O., & Schwartz, T. (1998). Irony: Graded salience and indirect negation.
Metaphor and Symbol, 13, 83-101.
Holtgraves, T. (1997). Styles of language use: Individual and cultural variability in con-
versational indirectness. Journal of Personality and Social Psychology, 73, 624-637.
Holtgraves, T., & Yang, J. (1992). Interpersonal underpinnings of request strategies: Gen-
eral principles and differences due to culture and gender. Journal of Personality and
Social Psychology, 62, 246-256.
Ivanko, S. L., & Pexman, P. M. (2003). Context incongruity and irony processing. Dis-
course Processes, 35, 241-278.
Jorgensen, J. (1996). The functions of sarcastic irony in speech. Journal of Pragmatics,
26, 613-634.
Katz, A. N., & Pexman, P. M. (1997). Interpreting figurative statements: Speaker occupa-
tion can change metaphor to irony. Metaphor and Symbol, 12, 1941.
Katz, A. N., Piasecka, I., & Toplak, M. (2001, November). Comprehending the sarcastic
comments of males and females. Poster presented at the meeting of the Psychonomic
Society, Orlando, FL.
 Ivanko et al. / INDIVIDUAL DIFFERENCES AND IRONY 271

Kreuz, R., & Glucksberg, S. (1989). How to be sarcastic: The echoic reminder theory of
verbal irony. Journal of Experimental Psychology: General, 118, 374-386.
Kumon-Nakamura, S., Glucksberg, S., & Brown, M. (1995). How about another piece of
pie: The allusional pretense theory of discourse irony. Journal of Experimental Psy-
chology: General, 124, 3-21.
Lampert, M. D. (1996). Gender differences in conversational humor. In D. I. Slobin,
J. Gerhardt, A. Kyratzis, & J. Guo (Eds.), Social interaction, social context, and lan-
guage (pp. 579-596). Hillsdale, NJ: Lawrence Erlbaum.
Pexman, P. M., Ferretti, T. R., & Katz, A. N. (2000). Discourse factors that influence on-
line reading of metaphor and irony. Discourse Processes, 29, 201-222.
Pexman, P. M., & Olineck, K. M. (2002a). Understanding irony: How do stereotypes cue
speaker intent? Journal of Language and Social Psychology, 21, 245-274.
Pexman, P. M., & Olineck, K. M. (2002b). Does sarcasm always sting? Investigating the
impact of ironic insults and ironic compliments. Discourse Processes, 33, 199-217.
Pexman, P. M., & Zvaigzne, M. T. (2004). Does irony go better with friends? Metaphor and
Symbol, 19, 143-163.
Tanaka, J. W., & Taylor, M. (1991). Object categories and expertise: Is the basic level in
the eye of the beholder? Cognitive Psychology, 23, 457-482.
Tannen, D. (1993). Gender and conversational interaction. New York: Oxford University
Press.
Utsumi, A. (2000). Verbal irony as implicit display of ironic environment: Distinguishing
ironic utterances from nonirony. Journal of Pragmatics, 32, 1777-1806.

Stacey L. Ivanko is presently a graduate student in psychology at the University of

Calgary, Calgary, Alberta, Canada. This research was conducted as part of her
Ph.D. research.

Penny M. Pexman received her Ph.D. in cognitive psychology from the University of
Western Ontario, London, Ontario, Canada, in 1998. Currently she is an associate
professor of psychology at the University of Calgary. Her research interests include
several aspects of language processing, including figurative language comprehen-
sion and visual word recognition.

Kara M. Olineck is presently a graduate student in psychology at Concordia Uni-

versity, Montreal, Quebec, Canada. This research was conducted while she was an
undergraduate student in psychology at the University of Calgary.

View publication stats