A Unified Theory of Irony and Its Computational 
Formalization 
Akira Utsumi 
Department of Systems Science, ~ibkyo Institute of Technology 
4259 Nagatsuta, Midori-ku, Yokohama 226, Japan 
utsumi@sys, ti%ech, ac. jp 
Abstract 
This paper presents a unified theory of 
verbal irony tbr developing a computa- 
tional model of irony. The theory claims 
that an ironic utterance implicitly com- 
municates the fact that its utterance sit- 
uation is surrounded by ironic environ- 
ment which has three properties, but 
hearers can assume an utterance to be 
ironic even when they recognize that 
it implicitly communicates only two of 
the three properties. Implicit commu- 
nication of three properties is accom- 
plished in such a way that an utterance 
alludes to the speaker's expectation, vi- 
olates pragmatic principles, and implies 
the speaker's emotional attitude. This 
paper also describes a method for com- 
putationally formalizing ironic environ- 
ment and its implicit communication us- 
ing situation theory with action theory. 
1 Introduction 
Although non-literal language such as metaphor 
has become a popular topic in computational lin- 
guistics (Fass et al., 1991), no attention has been 
given to ironic uses of language. One reason for 
this imbalance is that traditional accounts of irony 
- and even default logic forlnalization (Perrault, 
1990) - assume that irony communicates the op- 
posite of the literal meaning. This assumption 
leads to the misconception that irony is governed 
only by a simple inversion mechanism, and thus it 
has no theoretical interest. Another reason is that 
studies of irony have been regarded as of no prac- 
tical use for NLP systems. However, recent ac- 
counts denying the meaning-inversion assumption 
have revealed that irony is a more complicated 
pragmatic phenomenon governed by several men- 
tal processes (Kumon-Nakamnra et al., 1995), and 
dmt irony offers an effective way of accomplishing 
w~rious communication goals that are diff\]cult to 
convey literally (Roberts and Kreuz, \]994). 
The aim of this paper is to propose a unified 
theory of irony that answers to three crucial ques- 
tions in an unequivocal manner: (Q1) what are 
properties that distinguish irony from non-ironic 
utterances, (Q2) how do hearers recognize utter- 
ances to be ironic, and (Q3) what do ironic ut- 
terances convey to he~rers? Our theory provides 
a computationally feasible framework of irony as 
the first, step toward a full-fledged computational 
model of irony, and it can account for several em- 
pirical findings fi'om psycholinguistics. The essen- 
tial idea underlying our theory is that an ironic 
utterance implicitly displays ironic environment, 
a special situation which has three properties for 
being ironic, but the hearer does not have to see all 
the three properties implicitly communicated in 
order to recognize the utterance to be ironic. Note 
that this paper focuses only on verbal irony, and 
thus situational irony I (i.e., situations are ironic) 
is beyond the scope of our theory. 
This paper is organized as follows: Section 2 
discusses the problems of previous irony theories. 
Section 3 presents our unified theory of irony that 
can cope with the problems, and its computational 
formalization. Finally, Section 4 suggests that our 
theory agrees well with several empirical findings. 
2 Previous theories of irony 
Several irony theories have been proposed in the 
last few decades, but all the theories, as we will 
explain, make the same mistake in that they con- 
fuse the two difl'erent questions (Q1) and (Q2). 
The traditional pragmatic theory (Grice, 1975; 
Haverkate, 1990) assumes that, an utterance is 
recognized to be ironic when tile hearer becomes 
aware of an apparent violation of some pragmatic 
principles (e.g,, the maxim of quality or the sincer- 
ity conditions for speech acts), and as a result it 
conveys the opposite of the literal meaning. This 
theory, howew,~r, colnpletely fails to explain many 
ironic utterances. First, irony can be comnmni- 
cared by various expressions that do not include 
such violation: true assertions such as (2a) in Fig- 
ure 1, understatements such as (2c), and echoic 
utterances such as (5a). Moreover, Candy's hus- 
band of Example i can t)erceive Candy's ~ttter- 
ances (la)~(le) as ironic even under the situation 
where he does not know or is careless of Candy's 
expectation of satisfy her hunger, in other words, 
where he is not aware of the viQlation. This im- 
1Situational irony can be indicated by metarefer- 
ential expressions such as "It is ironic that...", but 
verbal irony is incomt)atible with such expressions. 
962 
Example 1: Candy baked a pizza to satisilg her 
hunger. When she was dishing it, uI), her hus- 
band entered t;he kitchen and gobbled up the 
whole pizza. Candy said to her huslmnd: 
(1) a. I'm not hungry at all. 
b. Have you seen my pizza on the table? 
c. I'll gel; to sleep. 
d. I'm really satisfied to (',at the pizza. 
e. How about another small slice of pizza? 
Exmnple 2: A mother asked her son to (;lea.it up his 
messy room, but he did a slol)py , half-hearted 
job. Alter a while, she discovered that his room 
is still messy, and said to her son: 
(2) 2. I love children who keep their rooms clean, 
in(leed. 
I/. Would you mind if I asked you to clean up 
your room, please? 
c. This room may 1)e slightly messy. 
Example 3: l'eter broke his wife's favorite tea(:ul) 
when he washed the dishes awkwardly. I,ooking 
at the broken cup, his wife said: 
(3) a. Thank you fur washing my cup carefully. 
b. Thank you for (:rashing my treasure. 
Example 4: Nancy and Jane were l)lamfing a trip 
to the beach, but that day was a coht and stormy 
one. As she looked out the window, Nancy slid: 
(4) a. ()It, the weather is really ni(:e. 
b. Mayl)e the 1)each is crowded with people. 
Example 5: Just after his colleague Jesse said to 
him "Pd be promoted before you", Peter replied: 
(5) a. You'd be promoted before ine, huh7 
b. You'd better work harder. 
Figure 1: Five examples of ironic utterances 
plies that violation of pragmatic principh'.s is not 
an answer to (Q2). Secondly, it is not an answer to 
(Q1) because of its incompetence to discriminate 
irony from other non-literal utterances (e.g., a lie) 
in which the. maxim of quality is tlouted. Finally, 
the notion of "the opt)osite of the literal mean- 
ing" is problematic because it, is aplflicable only 
to declarative assertions hut many ironic utter- 
ances can take non-declarative forms: questions 
such as (lb); requests such as (2b); offerings such 
a.s (le); and expressives such as (3a). 
Other recent theories e.g., mention theory 
(Wilson and Sperber, 1992) and echoic reminder 
theory (Kreuz and Ghmksberg, 1989) share a 
common view that by mentioning or alluding to 
someone's thought, utterance, expectation or (:nl- 
tural norm, an ironic utterance communicates a 
speaker's attitude toward a discrepancy between 
what actually is and what, has heen expected. 
This view may be essential to irony, but these 
theories ark still incomplete as a comprehensive 
framework h)r irony for at least three reasons. 
First, their concepts of mention/allusion Sper- 
her and Wilson's echoic interpretation and Kreuz 
and Glueksberg's echoic reminder are too nar- 
row to capture the allusive llattn'e of irony (e.g., 
(lb), (le), (41))), and they are not clear enough 
to be formalized in a computable fashion, l,br 
example, Nancy's utterance (4a) in Fignre 1 ix an 
echoic interpretation of Nancy's expectation of the 
fine, weather, bttt (4b) does not interpretively echo 
any states of affairs: (4b) is an implication derived 
from the failed expectation. Second, they implic- 
itly assume that the properties that characterize 
irony can be at)plied to recognition of ironic utter- 
ances as they stand or they do not focus oil how 
hearers recoglfize utterances to },)(; ironi(-. Thils 
they cannot also explain a certain kind of ironi(: 
utterances in which hearers are not aware of any 
pragmatic violation. Finally, these theories pro- 
vide no plausible explanation of how irony is dis- 
criminated flom non-ironic echoic utterances. 
Allusional pretense theory (Kumon-Nakamura 
et al., 1995) is the most powerflll one in that it can 
explain ironic utterances of five speech act classes 
using the two crucial notions of allusion (includ- 
ing echoic interpretation and reminder) and prag- 
matic insincerity. They (:laimed that all ironic ut- 
terances allude to a failed e, xpe, ctation and vio- 
late one of the felicity conditions for well-ibrmed 
speech acts. However, allusional pretense the- 
ory still sut\[ers Dora the same disadvantage as 
other theories: their notion of allusion is not clear 
enough, and it does not focus on how hearers rec- 
Ogllize lltt, erallt;es t;o be ironic. 
3 A unified theory of irony 
3.1 Ironic Environtnent and Its Inlplicit 
Display 
Our unitied theory of irony claims as an mmwer to 
(Q1) that irony is a figure of Sl)eeeh that inq)lie- 
itly displays the fact that its utteraime situation is 
surrounded by ironic environment. To make this 
(:lailn realizable, we must explain two important 
notions: ironic eilvironment and implicit display. 
In order for an utterance to be ironic, a speaker 
nmst utter in a situation sm'rounded by ironic en- 
vironmeut. Given two temporal loeatkms to and 
tt such that to temporally precedes t,, the utter- 
ante situation where an utterance is given is sur- 
rounded by ironic, envir(mmcnt if and only if it 
satisfies the following three conditions: 
1. The speake, r has an expe(:tation E at t~). 
2. The speaker's expectation E fails at h. 
3. As a result, the speaker has a negative emo- 
tional attitude toward the incongruity be- 
tween what is expected and what actually is. 
Note that our notion of speaker's expectaions sub- 
sumes cultually expected norms and rules. Fur- 
thermore previous theories assume echoic irony 
like (5a) to allude to other person's thoughts or 
963 
Example 1 
Instantiated Causal Relations: 
Sl ~ (< aeecssiltle, x, It )) A (( log, a, It )> A (( eatable, a >) : \[eat(x, a)\] => Su ~ (( hungry, x; 0 }> A (( loc, a, lx >) 
X A h,~y, o >> A (< loe, a, 1,, >> s, > << aecessible, > l,. >> A <<loe, a,h>> A <<e~table, a>> : \[~.'at(y,a)\] ~ & ~ << . ,V; 
B $1 ~ << hungry, x; 0 )> :=> S~ (precedes, S:,S2 )) ~ (< get-to-sleep, x )) 
Y Ironic Environment: 
(( Candy, x )> A (( husband, x, y )) A (( pizza, a )> A <( eatable, a )) A (( on, lt, b )> A (( table, b )>A 
(( accessible, x, It )) A (( in, I~ ,c )> A <( stomach, x, c )> A << in, 1., d )> A (( stomach,., d }> 
to ~= << Ice, a, l~ >> A << h~9~'y, * >> A << hope, .~e, T<< ,,,.~a~:.,~o,',' >> ~ << h,~ng,'y, x; 0 >> >> 
t, b: << lo~,., l. >> A << h~g~'y, ~; >> A <( hop~'. ~:, T<< p,~c~d~.,,., ,~'>> ~: << h~<q,'y,. ;0 >> >> A << did, ,'~at(?j, a) >> A 
(< hungry, y; 0 >} A (< did, eat(x, a); 0 )} A (< angry_at, x, y, eat(y, a) >} 
Example 2 
Instantiated Causal Relations: St 1= ({ messy, a }) : \[clean-up(y, a)\] :* S'2 1= (( clean, a )) 
Ironic Environment: 
(( mother, x, y >) A (( son, y, x )) A (( room, a >> A (( owns, y, a )) 
to b <( messy, a >> A <( ,,sk, x, .~j, cl~,,,,-~p(> a) )> A <( hope, .~,:, T<< p,~,~,d,;.~,~o,',~ >> t--: << ,:le~,,, a )) >> 
g ~clean-up(y, a) 
t l ~ ((messy, a}} A ({ did, clean-up(y, a ) ; O }) A << h.ope, x, T << precedes't ' ':v )) ~ << clean, a)} ))A 
<< an.q,'v_~t, ~', :,, ~a~an-,*V(V, a) )) 
Figure 2: Representation of ironic environments for Examples i and 2 
utterances, but our theory contends that such 
irony alludes to a speaker's exi)ectation that "the 
speaker wants the hearer to know the hearer's ut- 
terances or thoughts are false". For example, the 
speaker's expectation of (5a) is that Jesse knows 
he cannot be promoted betbre Peter. 
Ironic environment can be classified into the fol- 
lowing tbur types. 
• a speaker's expectation E can be caused by 
an action A performed by intentional agents 
- E failed because A failed or cannot t)e per- 
formed by another action B (type-l) 
- E failed because A was not performe(t 
(type-2) 
• a speaker's expectation E is not normally 
caused by any intentional actions 
E failed by an action I3 (type-3) 
- E accidentally failed (type-4) 
For example, ironic environment of Example1 
falls in type-l: Candy's expectation of staying her 
hunger can be realized by an action of eating a 
pizza, but her husband's action of eating tile whole 
pizza hindered her expected action. In the same 
way, ironic environments of Examples 2-4 fall in 
type-2~type-4, respectively, and that of Example 5 
falls in type-3. 
An utternace implicitly displays all the three 
conditions for ironic environment when it 
1. alludes to the speaker's expectation E, 
2. includes pragmatic insincerity by intention- 
ally violating one of pragmatic principles, and 
3. implies the speaker's emotional attitude to- 
ward the failure of E. 
For example, utterances (2(1) and (2e) tbr Exam- 
pie 2 are not ironic even when they are given in the 
situation surrounded by ironic environment: (2(1) 
and (2e) directly express the speaker's expectation 
and tile st)eaker's emotional attitude, rest)ectively, 
and both do not include pragmatic insincerity. 
(2) d. I've expected a clean room. 
e. I'm disat)t)ointed with the messy room. 
On the other hand, all the utterances of Figure 1 
are ironic because they implicitly express the three 
comt)onents of ironic enviromnent, as we will show 
in Sections a.a-a.s. 
3.2 Representing Ironic Environment 
In order to formalize ironic utterances and ironic 
enviromnent ill a coint)utational fashion, we use 
situation theory (Barwise, 1.989) and situation cal- 
culus. Our representational scheme includes dis- 
crete items of intbrmation called infons, situations 
capable of making infons true (i.e., supt)orting in- 
fons), and actions. For example, information that 
Candy eats the pizza is represented as the infon 
(( eat, x, a )> in which x and a denote "Candy" and 
"the pizza", and its negation as (( ca*,,x,a; 0)). 
A fact/event that Candy eats dm pizza is rep- 
resented as t ~ (( eat, x, a )) where the situation 
t expresses the spatiotemporal location of that 
event. Ail action of eating the pizza performed 
by Candy is expressed by tile predicate eat(x, a) 
and its negation (i.e., an action of not perfbrm- 
ing eat(x,a)) by =~;at(x,a). The state of allah's 
that an action A is perfbrmed is expressed by 
(( did, A}}. Also, a proi)osition p expressing the 
claim that Candy eats the pizza is written as 
(t ~ ((eat, x,a)}). The propositionp = (s ~ c,) 
964 
Spe.eeh Act: 
Preconditions: 
Effects: 
Speech Act: 
Preconditions: 
Inform(S, II, P) << 
~v,,~k,:,., s >>, << ~,.(.~.,.,,,., u >>, 
(( proposition, I' )), us D I" 
Rcquc.stg ( S, It, I') 
<< ~w:~k~.,,-, ,9 >>, << t.~,~,.~:,,, u >>, 
(( v~ovo~itio,~, ,')), ~m~o,,,,If ( s, r) 
Effects: un \[= (( intend, S, lnformlJ'(H, S, P) }} 
Notes: us and ltll denote tile speaker's and hem'er's 
mental situations, KnowIf(S, P) = us ~ I'V ~1', and 
-~K,~o~,,U(S, 1') - ,,,.~ ~ 1' A ~*'. 
Figure 3: Speech act defilfitions 
is true if s supports v, and otherwise false. Situa- 
tions are partially ordered by the, part-of relation 
denotexl by <1. A situation sj is a part of a sit- 
uation s2 (i.e., Sl <1 s2) if and only if every infon 
supt)orted t)y st is also sut)ported by su. Ill this 
l/aper we also tel)resent an agent X's mental situa- 
tion as ux and his/lmr beliefs as supl)ort relations 
t)etween 'ux and infons. D)r examph', l, he fact that 
,/ira believes/knows the above event is re.t)resente.d 
a.s u.ri,,~ \[= t I : {{ eat, z,a }). hflbns and actions 
Call include paralnetexs denoted by capital letters. 
l)arameters (:an be restricted I)y infons: for exam- 
ple, T << v'~:':~'l~'t',''r )) ix a parameter for tenlporal 
situations whMl temporally succeed to. A causal 
relation betwe, en two events Sl t- Ol alld s2 \[-- o-2 
is exln'essed by .s~ \[= ch : \[A\] => .s'2 \[- c,~. This 
relation means dlat if an action A ix executed in a 
situation s l sut)porting the infon al, then it causes 
the inf(m a2 to be trlle in the re.suiting situatiol~ 
s.~. Thus it folh)ws that s2 \[-- ({ did, All. When 
we omit an actioi~ A froIn a ('ausal relation, that 
relation/)ccomes a constraint in sitltation theory, 
denoted by sl \[= cq => s2 ~= (*2. Figure 2 illus- 
trates the, represe, ntation of ironic environnlents 
of Exalnple~s 1 and 2. Although Figure 2 does not 
in(:lude ally mental situations (i.e., ironic enviroi> 
illent is re, i)resented Doln god's eye, vie, w), whell 
a speake, r intends the uttexan(:e to t)e ironi(; tlte 
st)eaker's mental situation must SUl)\[)ort all states 
of affairs, events and causal relations in this figure. 
An utterance U is characterized by its propo- 
sitional content P and the illocutionary act that 
the speaker performs in saying 17, soIne of which 
m:e shown in Figure 3 (Litman and Allen, 1987). 
For example,, the propositional content of (1~) is 
(t~ ~ (( hwngr~.h a:; 0 }}) and its illocutionary act is 
lnfl)rm. Also (1t)) is characterized by l' = (h \[- 
<( *,~'~, v, T<<,,, ...... *"':'"" >> b << lo~, a, h >> >>) alld the 
illocudonay act l{equestIf. 
a.a Allusion 
We give a formal de, finition of allusio,l in our the- 
ory. Given P expressing the p,'oposil,ional con- 
tent of U, and Q expressing the speaker's ex- 
pected event/state of affairs, an utterance U al- 
ludes to the exl)ectation l~ if it satisfies ()lie of 
the conditions shown ill Table 1. The relation -.a 
in Table 1 is defined as follows: assuming that 
• << boy., P, (s t= x) >> ~ << ,,,.~nt, v, (s # x) >> A 
• S, ~- (( disappointed, P, (S, ~ X) >) ¢= 
so ~ (( t.,m, P, ( s ~ x))) A & b- x A 
s, ~ s A << V.~:,'.,~d,'~,~, So, Sl )> A 
(( precede,% So, S )) 
• & D (( ~now-~.< I"1, P.~, A >> ¢ 
so ~- (< .,,~,< I',, (,s' l= x) )) A S, ~ X A 
Sl < S A (< p,,.,;~.,,.d,..,% So, S, >) A 
(( precedes, So, S >) A <( agent , A, P'2 >) A 
& ~ (( d{d, J )) A (( blamcwor#~,j, J )) 
Figure 4: Enlotmn-ehcltlng rules 
\])i = 1)'2 nleans that both arc (:oncel)tually identi- 
cal or unifiable, \[)t ~'~ P'2 hohls if 
1)1 ----- (12.2) or Ill's COllStitlle31t : {1)'2 or (1)'2)} 
(when P'2 is an event;) 
PI -P'2 or 15 's COllsgit, llent ---- 12. 2 
(when 1)2 is an action) 
This definition allows all utterances in Fig- 
ure. 1 to allude speaker's expectations, but it does 
not allow (2(t) to alhlde to it. Tal)le 1 shows 
which condition e, ach of these, uttt;rant:e,s satis- 
fies. For exainple, the utteran(:c (11)) that nlen- 
lion theory caltllOt ext)lain alludes to Candy's ex- 
i)ectatioll 1)y refe, rring to one of the conditions 
X = St ~ (<loc, a,lt)) in Figure 2 for an act.ion 
A = e, at(:c, a) since the part of its propositional 
content P and X are unifiable. Other utterances 
f,)r ExalIq)l,:l, (la)alld (l(')~(le), also refer to 
Q, Y, A, ./J shown ill Figure 2, respectively. Ill the 
sanle way, (2b) satisfies Condil;ion 4 since its <:oil- 
tent 1 j = ch'~a'n-up(y, a) is identical to A. 
3.4 Pragmatic Insincerity 
Table 2 lists the t)i'agmatic principles violated by 
the ironic utterances in Figure 1. \[n many (:ases an 
ironic utterance is praglnatically insincere ill the 
sense that it intentionally violates one of the pre- 
conditions in Figure 3 (i.e., sincerity, preparatory 
and propositional conditions) that need to hold 
before its illocutionary act is accomplished, but 
pragmatic insincerity also oecurs when an utter- 
anee violates other praglnatic l)rineiples. Re.quests 
often I)ecolne il~sincere when they are over-polite, 
like (2t)) since they violate the t)olitene.ss princi- 
ple (althougtl (2t)) also becomes insincere when 
l,he lnother no hmger intends her son to clean ut) 
his room). Understatenmnts like (2c) are also in- 
sincere since they do not provide as inuch illfor- 
ulation as re.quired. The true assertion (2a) vio- 
lates the princit)le of relevance in that it does not 
yie.\]d ally conl;extual implication. An mentioned 
earlier, the last three cases have been proi)lenlatic 
for all the, previous theories of irony 1)ecause none 
of these theories recognized a wide varie, ty of prin- 
ciples violated by ironic utterances. Although this 
t)aper does not describe how these praglnatic t)rin- 
ciples shouhl l)e formalized, they should be taken 
into account for the next steps of our study. 
965 
Table 1: Allusion of ironic utterances in Figure 1 
Conditions for allusion Utterances satisfying the condition 
1. P~Q A P+T~<<R,S,Q)) 
2. P -,~ X where X: \[A\] => Q or X ::~ Q 
3. P-o YwhereQ =~ Y 
4. P ",~ A where X : \[A\] ~ Q (type-1 or type-2) 
5. P ~-~ B or W or Z where W : \[B\] =~ Z (type-1 or type-3) 
(la) (2a)(4a) (5a) 
(lb) (2c) (5b) 
(1(:) (4b) 
(ld) (2b) 
(le) (3a)(3b) 
Notes: In Condition 1, T, R and S denote parameters for situations, relations about expecting, and speak- 
ers, respectively. In Condition 5, B denotes actions which disable an action A of Condition 4. 
Table 2: Pragmatic insincerity of ironic utterances in Figure 1 
Violated pragmatic principles Utterances violating the principle 
Sincerity condition for Inform (S believes P) (la) (le) (1(1) (4a) (4b) (5a) 
for Question (S does not know P) (lb) 
for Advise (S believes P will I)enefit H) (5t)) 
for Offer (S wants to do an action P for H) (le) 
for Thank (S feels grateful for an action P) (3b) 
Propositional content condition for Thank (P is a past action clone by H) (3a) 
Preparatory condition for Offer (S is able to do an action P) (le) 
Maxim of relevance (P is relevant in Sperber and Wilson's (1986) sense) (2a) 
Politeness principle (U should be made at an appropriate level of politeness) (2b) 
Maxim of quantity (P is as informative as required) (2c) 
Notes: S, H and P denote the speaker, the hearer and the propositional content, respectively. 
3.5 Emotional Attitude 
Speakers can use a variety of signals/cues into- 
nation contour, exaggerated stress, tone of voice, 
hyperbole, facial expression, etc. for implic- 
itly communicating their emotional attitude. The 
use of the interjection "Oh" with a special tone 
of voice in (4a) offers one typical example of this. 
Implicit communication can also be accomplished 
by utterances explicitly referring to the pleased 
emotion that speakers would experience if their 
failed expectation became true. For example, the 
utterance (3a) explicitly expresses speaker's coun- 
terfactual emotion. 
At the same time, many ironic utterances make 
emotion-elieiting rules for the speaker's attitude 
(some of which are shown in Figure 4) accessible 
by the 'hearers by alluding to one of premises of 
the rule. In the case of (3a), it alludes to Peter's 
action of washing the dishes so that the rule for 
"angry~t" emotion becomes more accessible. 
3.6 Recognizing and Interpreting Irony 
In many cases, all the three components for im- 
plicit communication of ironic environment are 
easily recognized by the hearer. As we mentioned 
in Section 2, however, there are also many cases 
such as Example 1 that an utterance can be iron- 
ically interpreted even though all the three com- 
ponents cannot be recognized by the hearer be- 
cause the hearer's mental situation differs from 
the speaker's one. Furthermore, in the case of 
(5a), after recognizing the utterance to be ironic 
Jesse turns out to know that the speaker Peter 
thinks Jesse cannot be promoted before Peter. 
Hence we propose the following condition for 
recognizing irony as an answer to (Q2): 
Hearers can assuIno an utterance to he 
ironic (with high possibility) if they can 
recognize that the utterance implicitly 
displays at least two of the three compo- 
nents for ironic environment, and if the 
utterance situation does not rule out the 
possibility of including tile unrecognized 
components, if any} 
This "2-of-3" criterion makes it possible that hear- 
ers can recognize utterances as ironic even though 
speakers do not intend their utterances to be un- 
derstood as irony. It, provides empirical evidence 
of ()lit' theory since such unintentional irony has 
been found in a number of psychological experi- 
ments (Gibbs and O'Brien, 1991). 
By recognizing an utterance to be ironic, the 
hearer becomes aware of an illocutionary act of 
irony, that of conveying the fact that the utter- 
ance situation is surrounded by ironic environ- 
ment (i.e., all the three components for ironic en- 
vironment hoht in a current situation). That is 
an answer to (Q3), and then the hearer inter- 
prets/understands the ironic utterance by adding 
that information to his/her mental situation. In 
2Practically speaking, whether an utterance is 
ironic is a matter of degree. Thus the degree of ironic- 
ity might t)e a better criterion for recognizing irony. 
If we can quantitatively evaluate, though do ,tot in 
this paper, to what degree an utterance alludes to the 
speaker's expectation, to what degree it includes prag- 
matic insincerity, and to what degree it implies the 
speaker's emotional attitude, we think the proposed 
condition for recognizing irony can also be quantita- 
tively defined. 
966 
many cases, sin(:e the hearer already knows the 
fact that the three components hold in the sit- 
us\]ion, interpretation of irony results in confir- 
mation of the mosl; uncertain information, that 
is, the speaker's emotional attitude. However, 
when the hearer does not recognize all compo- 
nents, he/she also ol)tains new information that 
the unrecognize.d component holds in a current sit- 
uation. Therefore, our the.ory includes many pre- 
vious theories claiming that irony (:ommunicates 
an ironist's emotional attitude. For example, in 
the. case of (5a), after recognizing Peter's utter- 
ance (5a) to be iron\](', Jesse turns out to know that 
Peter drinks Jesse's t)receding uttermme is absurd, 
and tries to confirm Peter's emotional attitude 
by interpreting (5a) ironically. Furthermore, as 
we mentioned in Section 1, nil ironic utterance 
achieves various (:olmnunieation goals held t)y the 
sl)eaker e.g., to be huinorous, to enq)hasize a 
I)oint, to clarify as I)erlocutionary acts. 
4 hnplications of the Theory 
Distinction between ironic and non-ironic 
utterances: Our theory ('an disl;inguish iron\](" 
utterances from non-iron\](; ones. For exmnple, 
lies and other non-ironic utt(wances violating the 
pragmatic principle do not allude to any an- 
tecedent exi)ectation and/or (lo not offer cues fi)r 
reasoning about the si)e.aker's emotional attitude. 
Non-ironic echo\](: utterances do not include prag- 
matic insincerity and/or do not irnplicitly commu- 
nicate the speaker's attitude. 
Ironic cues: Some theories assmne that irony 
can be identified by special cues tTor irony, but the 
empirical finding in psychology shows that people 
can interpret iron\](: statements without any sI)e- 
cial intonational cues (Gibbs and O'Brien, 1991). 
Our theory agrees with this finding: such kind of 
cues is only a l)art of Component 3 as we described 
in Section 3.5, and thus iron\](" ul;teranees without 
these cues can t)e recognized as ironic. 
Victims of irony: Several irony studies, e.g., 
(Clark and Gerrig, 1984), have t)ointed out that 
irony generally has victims. Our \[theory suggesl, s 
th;~t ironic ul;l;eranc('.s have potential vi(:tims when 
their iroific environments fall in one. of types-l,2,3: 
in the ease of type-1 or type-3 an agent of 13 I)e- 
comes a victim, and i,l the case of type-2 an agent 
of A becomes a victiin. 
Sarcasm and irony: We argue that explicit 
victims and disl)lay of the speaker's counterfac- 
tual pleased emotion described in Section 3.5 m'e 
dist;inctive prop(~rties of sarcasm. Thus \],he ut- 
t(,ran(:es (a~) and lab) are sar,:a,~ti(: 1,e,:anse they 
have an ext)licit victim, Peter, and they refer 1;() 
the wife's comltert'a(:tual l)hmse(t emotion. \]n par- 
ti(:ular, an utteran(:e "Thanks a lot!" for Exam- 
i)le 3 is non-ironic sar(;asln silt(:(', it does not allude 
to any exp(;etation. 
5 Conclusion 
In this paper we have proposed a unified theory 
of irony that overcomes several difficulties of pre- 
vious irony theories. Our theory allows us to Rive 
plausible answers to what irony is, how irony is 
recognized and what irony coinmuni(:ates. The 
properties of irony allusion, pragmatic insin- 
eeril,y, and emotional attitude arc formalized 
mmquivo(:ally enough to build a coInputational 
model of irony. From this I)oint ot' view, we be- 
lieve that this pal)er provides a basis \[or dealing 
with irony in NLP systems, and we are develol)- 
ing computational methods for intert)reting mat 
generating irony (Utsumi, 1995). 
References 
J. Barwisc. 1989. The Situation in Logic. Stanford: 
CSLI Publications. 
H.It. (\]lark and R.J. (\]crrig. 1984. On the pretense 
theory of irony. Journal of Experimental Psychof 
ogy: General, 113(1):121 126. 
D. Fass, E. Hinkchnan, and J. Martin, editors. 
199t. Procccdin.qs of the LICA\] Workshop on Com- 
putational Approaches to Non-Literal Language: 
Metaphor, Metonymy, Idioms, Speech Acts, Impli- 
cal,?WC. 
R.W. Gibbs and J. O'Brien. 1991. Psychoh)gical as- 
Imcts of irony understanding. Journal of PTngmat- 
its, 16:523 530. 
II.P. Gricc. 1975. Logic and conversation. In P. Cole 
and J. Morgan, editors, Syntax and semantics, 
Vol.3: 5'pooch acts, pages 41 58. Academic Prcss. 
H. Havcrkatc. 1990. A speech act analysis of irony. 
Journal of Prngmatics, 14:77 109. 
R.J. Krcuz and S. Olucksb(`-rg. 1989. How to bc 
sarcastic: The e(:hoic reminder theory of vcrl)al 
irony. Journal of l'Jxpc'rimcntal Psychology: Gcn- 
c~ul, 118(4):374 386. 
S. Kumon-Nakamura, S. Glucksbcrg, and M. Brown. 
1995. IIow al)out another piece of pie: The allu- 
sional 1)rctcnsc theory of dis(:oursc irony. ,\]o'urnal 
of .Expc.rimental Psychology: General, 124(1):3 21. 
D.J. lilt, man and J.F. Allen. \] 987. A plan recognition 
in((del for subdialogues in conversations. Cognitiw; 
Science, 11:163 200. 
C.R. l'crrault. 1990. An application of default; logic 
to speech act theory. In P.\]{.. Cohen, J. Morgaal, 
and M.E. Pollack, editors, Intcntion.s in Communi- 
cation, I(agcs 161 185. Tit('. MIT Press. 
R.M. lh)bcrts and l{.a. Krcuz. 1994. Why do 1)co - 
plc use figurative language? Psychological 3cicnce, 
5(3):159 \]63. 
D. Sperber and D. Wilson. 1986. Relevance: Commu- 
nication and Cognition. ()xford, Basil Blackwcll. 
A. lJtsumi. 1995. How to intcrt)ret irony 1)y com- 
t(/l|;el: A (:oml)rehcnsivc fl'amcwork for irony. Ill 
Proceedings of R, ANLI', 1)ages 315 321, Bulgaria. 
1). Wilson and 1). S1)crbcr. 1992. On vert)al irony. 
Lingua, 87:53 76. 
967 
