A Course in Pragmatics

University of Sfax
Faculty of Arts & Humanities

English Department
Academic Year: 2018-2019
Teacher: Sofiene Mallouli
PRAGMATICS
Agrégation 1
University of Sfax
Faculty of Arts & Humanities
English Department
Academic Year: 2018-2019
AGREG 1
Teacher: Sofiene Mallouli
PRAGMATICS
Course Description & Outline
COURSE DESCRIPTION
This course is a foundational course in pragmatics. It intends to introduce pragmatics and some of its
basic concepts. It will start by shedding light on the various and, in times, conflicting definitions given to the
field of pragmatics and its use and relevance for native and non-native speakers. It will then introduce some
of the basic concepts & tools of pragmatics, namely deixis, implicature, presupposition, entailment,
(im)politeness, and speech acts.
By the end of the course, students should:
a- Define and identify the scope of pragmatics and its main concepts.
b- Identify the relevance of pragmatics for NS and NNS.
c- Identify, analyse, & illustrate utterances in terms of deixis, implicature, entailment & presupposition,
(im)politeness, and speech acts.
d- Identify the types and categories of deixis, implicature, entailment & presupposition, (im)politeness,
and speech acts.
COURSE OUTLINE
1- Pragmatics: Definitions & Background

2- Pragmatics and NS/NNS.
3- Deixis: types & scope.
4- Implicature: types & scope
5- Entailment & presupposition: types & scope
6- (Im)politeness: typology & scope.
7- Speech acts: types & scope
COURSE REFERENCES
 Huang, Y. (2007). Pragmatics. New York: Oxford University Press.

 Levinson, S. (2008). Pragmatics (9th Ed.). Britain: Cambridge University Press.
 O’keeffe, A., Clancy, B., & Adolphs, S. (2011). Pragmatics in use. London & NewYork: Routledge.
 Senft,G. (2014). Understanding pragmatics. London & NewYork: Routledge.
1
Introduction
1.1. What is pragmatics? 1 1.3. Some basic notions in semantics

and pragmatics 10
1.2. Why pragmatics? 5
1.4. Organization of the book 17
1.1. What is pragmatics?
1.1.1. A definition
Pragmatics is a rapidly growing field in contemporary linguistics. In recent

years, it has not only become a centre of intense interest in linguistics and
the philosophy of language, it has also attracted a considerable amount of
attention from anthropologists, artificial intelligence workers, cognitive
scientists, psychologists, and semioticians.
But what is pragmatics? This is a question whose answer is notoriously
difficult to provide (see Levinson 1983: 5–35 for a thorough discussion of a
number of definitions and their inherent difficulties). Here, I shall not
repeat Levinson’s discussion, but simply propose a working definition of
pragmatics (see also Levinson 2000).
2 1. INTRODUCTION
(1.1) A definition of pragmatics

Pragmatics is the systematic study of meaning by virtue of, or dependent on,
the use of language. The central topics of inquiry of pragmatics include
implicature, presupposition, speech acts, and deixis.
I shall return to the issue of how to define pragmatics in Chapter 7, where

the interface and division of labour between pragmatics and semantics is to
be discussed.
1.1.2. A brief history of pragmatics
Pragmatics as a modern branch of linguistic inquiry has its origin in the

philosophy of language. Its philosophical roots can be traced back to the
work of the philosophers Charles Morris, Rudolf Carnap, and Charles
Peirce in the 1930s. Influenced by Peirce, Morris (1938: 6–7), for example,
presented a threefold division into syntax, semantics, and pragmatics
within semiotics—a general science of signs. According to this typology,
syntax is the study of the formal relation of one sign with another, seman-
tics deals with the relation of signs to what they denote, and pragmatics
addresses the relation of signs to their users and interpreters1 (Levinson
1983: 1, Horn and Ward 2004a). This trichotomy was taken up by Carnap
(1942), who posited an order of degree of abstractness for the three
branches of inquiry: syntax is the most and pragmatics the least abstract,
with semantics lying somewhere in between. Consequently, syntax provides
input to semantics, which provides input to pragmatics (Recanati 2004b).
When it came to the 1950s, two opposing schools of thought emerged
within the analytic philosophy of language: the school of ideal language
philosophy and the school of ordinary language philosophy. The central
ideas underlying the former were originated by the philosophers Gottlob
Frege, Alfred Tarski, and Bertrand Russell. Ideal language philosophers
were primarily interested in the study of logical systems of artificial lan-
guages. However, the partially successful application of its theory and
methodology to natural language in the 1950s and 1960s by followers of
the school such as Richard Montague, David Donaldson, and David Lewis
1
Cf. Stalnaker’s (1972: 383) definitions: ‘Syntax studies sentences. Semantics
studies propositions. Pragmatics is the study of linguistic acts and the contexts
in which they are performed.’
1.1. WHAT IS PRAGMATICS? 3
led to the development of today’s formal semantics. By contrast, within the

tradition of ordinary language philosophy, emphasis was placed on natural
language rather than the formal languages studied by the logicians. Under
the leadership of J. L. Austin, the school of ordinary language philosophy
flourished principally at Oxford in the 1950s and 1960s. Other leading
thinkers of the school included the philosophers H. P. Grice, Peter Strawson,
John Searle, and the later Ludwig Wittgenstein (Huang 2003, Recanati
2004a, 2004b).2 It was within the tradition of ordinary language philosophy
that Austin developed his theory of speech acts, and Grice his theory of
conversational implicature. Both theories have since become landmarks on
the path towards the development of a systematic, philosophically inspired
pragmatic theory of language use.
On the linguistics front, in the late 1960s and early 1970s a campaign was
launched by some of Noam Chomsky’s disaffected pupils in generative
semantics (as it was then called), notably Jerry Katz, J. R. Ross and George
Lakoff, to challenge their teacher’s treatment of language as an abstract,
mental device divorced from the uses and functions of language. In their
search for the means to undermine Chomsky’s position, the generative
semanticists, who were attracted to the philosophical work by Austin,
Grice, Strawson, and Searle, helped to empty what the philosopher
Yehoshua Bar-Hillel called the ‘pragmatic wastebasket’ (see Harris 1993
for a discussion of the ‘linguistics wars’ they waged). As a result, a great
deal of important research was done in the 1970s by linguists such as
Laurence Horn, Charles Fillmore, and Gerald Gazdar to ‘bring some
order into the content of [the pragmatic] wastebasket’, as wisely advised
by Bar-Hillel (1971: 405). The publication of Stephen Levinson’s celebrated
textbook Pragmatics in 1983 systematized the field and marked the coming
of age of pragmatics as a linguistic discipline in its own right.
Since then, the field of inquiry has continued to expand and flourish.
In the last two decades we have witnessed new developments such as
Laurence Horn’s and Stephen Levinson’s neo-Gricean pragmatic theories,
Dan Sperber’s and Deirdre Wilson’s relevance theory, and important work
by philosophers such as Jay Atlas, Kent Bach, and François Recanati.
‘More recently’, as the editors of a newly published Handbook of Pragmatics
declared, ‘work in pragmatic theory has extended from the attempt to
2
According to Recanati (2004b), Grice was special in that he took the view that
the two schools of thought are not incompatible but complementary.
4 1. INTRODUCTION
rescue the syntax and semantics from their own unnecessary complexities
to other domains of linguistic inquiry, ranging from historical linguistics to
the lexicon, from language acquisition to computational linguistics, from
intonational structure to cognitive science’ (Horn and Ward 2004a: xi).
One thing is now certain: the future of pragmatics is bright.
1.1.3. Two main schools of thought in pragmatics:

Anglo-American versus European Continental
As pointed out in Huang (2001a), two main schools of thought can be

identified in contemporary pragmatics: Anglo-American and European
Continental. Within the former conception of linguistics and the philoso-
phy of language, pragmatics is defined as the systematic study of meaning
by virtue of, or dependent on, language use. The central topics of inquiry
include implicature, presupposition, speech acts, and deixis (see 1.1 above).
This is known as the component view of pragmatics, namely, the view that
pragmatics should be treated as a core component of a theory of language,
on a par with phonetics, phonology, morphology, syntax, and semantics.
By contrast, other areas such as anthropological linguistics, applied lin-
guistics, and psycholinguistics would lie outside this set of core compon-
ents. Within the Continental tradition, pragmatics is defined in a far
broader way, encompassing much that goes under the rubric of sociolin-
guistics, psycholinguistics, and discourse analysis. Witness, for example,
Verschueren’s (1999: 7, 11) definition that ‘pragmatics constitutes a general
functional (i.e. cognitive, social and cultural) perspective on linguistic
phenomena in relation to their usage in the form of behaviour’. This
represents the perspective view of pragmatics, namely, the view that prag-
matics should be taken as presenting a functional perspective on every
aspect of linguistic behaviour. More or less the same is true of the definition
of pragmatics provided within the former Soviet and East European trad-
ition. Under this approach, pragmatics (called pragmalinguistics) is in
general conceived of as a theory of linguistic communication, including
how to influence people through verbal messages (Prucha 1983).
Which of the two schools of thought, then, is conceptually more elegant
and methodologically more sound? While the Continental tradition is
closer to the original view of pragmatics expressed by Morris, that prag-
matics should study ‘the relation of signs to interpreters’ and that ‘since
1.2. WHY PRAGMATICS? 5
most, if not all, signs have as their interpreters living organisms, it is a

sufficiently accurate characterization of pragmatics to say that it deals with
the biotic aspects of semiosis, that is, with all the psychological, biological,
and sociological phenomena which occur in the functioning of signs’
(Morris 1938: 30), the study of ‘everything’ is hardly a viable academic
enterprise. Given the degree of overlap among the phenomena dealt with in
other relatively well-established interdisciplinary fields of linguistics such
as sociolinguistics, psycholinguistics, and neurolinguistics, it is rather dif-
ficult to see how a coherent research agenda of pragmatics can be made
within the wider Continental tradition. To say that ‘everything is pragmat-
ics’ amounts to saying that ‘nothing is pragmatics’. Pragmatics delineated
within such a tradition at best serves as a convenient banner under the
auspices of which divergent branches of linguistic inquiry can form a
temporary intellectual coalition (see also Levinson 1987a). By contrast,
the narrower Anglo-American, component view of pragmatics, which
focuses on topics emerging from the traditional concerns of analytical
philosophy, delimits the scope of the discipline in a relatively coherent,
systematic, and principled way. In this book, therefore, my discussion of
pragmatics will largely be from the Anglo-American, component point
of view.
1.2. Why pragmatics?
There are many reasons for including pragmatics in an integrated linguistic

theory. Here, I shall discuss just two of them.
1.2.1. Linguistic underdeterminacy
It is widely accepted that there is a huge gap between the meaning of a

sentence and the messages actually conveyed by the uttering of that sen-
tence. In other words, the linguistically encoded meaning of a sentence
radically underdetermines the proposition the speaker expresses when he or
she utters that sentence (e.g., Austin 1962, Atlas 1977, 1989, 2005, Searle
1979, 1992, 1996, Travis 1981, 1997, Sperber and Wilson 1986, Recanati
1989, 1994, 2004a, and 2004b, Bach 1999a, Levinson 2000, Carston 2002,
6 1. INTRODUCTION
Horn 2004). This is generally known in the literature as the linguistic

underdeterminacy thesis. By way of illustration, consider (1.2)–(1.5).
(1.2) You and you, but not you, stand up!

(1.3) a. The authorities barred the anti-globalization demonstrators because
they advocated violence.
b. The authorities barred the anti-globalization demonstrators because they
feared violence.
(1.4) John is looking for his glasses.
a. glasses ¼ spectacles
b. glasses ¼ drinking vessels
(1.5) (Cited in Levinson 2000: 174)
They are cooking apples.
a. What are they doing in the kitchen?
b. What kind of apples are those?
The three uses of the pronoun you—called a deictic expression—in (1.2)

can be properly interpreted only by a direct, moment by moment monitor-
ing of the physical aspects of the speech event in which the sentence is
uttered. In other words, the deictic parameter can be fixed only if the deictic
expressions are accompanied by physical behaviour of some sort (such as a
selecting gesture or an eye contact), which requires an extralinguistic
physical context. I shall provide a descriptive analysis of deixis in Chapter 5.
Next, in (1.3), the assignment of reference for the anaphoric pronoun they
depends crucially on our background assumption about who would most
likely be advocating or fearing violence. This extralinguistic information is
responsible for the two opposing interpretations, namely, they referring
back to the anti-globalization demonstrators, as in (1.3a) or linked to the
authorities, as in (1.3b). Finally, (1.4) is a case of lexical ambiguity, and
(1.5) a case of syntactic ambiguity. In disambiguating them, contextual or
real-world knowledge is needed to select the reading the speaker has
intended. I shall provide a discussion of the pragmatics of disambiguation
in Chapters 6 and 7. All this indicates that certain linguistic phenomena can
be handled naturally only by recourse to extralinguistic, pragmatic factors
such as context, real-world knowledge, and inference. Put another way, in
order to fill the gap created by linguistic underdeterminacy, pragmatics has
to be included as a component in an overall theory of linguistic ability.
1.2.2. Simplification of semantics and syntax
The second reason why we need a pragmatic component is because its

inclusion can effect a radical simplification of other core components, such
as semantics and syntax, in an overall theory of linguistic ability.
Starting with semantics, it has been noticed at least since Aristotle that a
sentence like (1.6) has two systematically distinct interpretations: a one-
sided, lower-bounded ‘at least’ reading, as in (1.6a) and a two-sided, upper-
and lower-bounded ‘exactly’ reading, as in (1.6b).
(1.6) John has had nine girlfriends.
a. John has had at least nine girlfriends.
b. John has had exactly nine girlfriends.
How can a semanticist deal with sentences like (1.6)? He or she has to treat
these sentences as lexically or logically ambiguous. However, there is a
serious problem at the very heart of this ambiguity analysis, namely, the
account runs directly against the spirit of a metatheoretical principle
known as ‘Occam’s razor’, which dictates that entities are not to be multi-
plied beyond necessity. A particular version of Occam’s razor, dubbed
‘Occam’s eraser’ by Ziff (1960: 44) and ‘modified Occam’s razor’ by Grice
(1978, 1989: 47), is that senses or dictionary entries must not proliferate.
This has the consequence that, all things being equal, an account which has
to suggest two lexical items is to be rejected in favour of an analysis which
does not. Therefore, if we adopt the semantic analysis, we are forced to
treat not just cardinal numbers like nine, but also adjectives like warm,
quantifiers like some, and indeed most lexical items in English as ambigu-
ous—clearly not an economic analysis.
But if we have a pragmatic component in an integrated linguistic theory,
things will be very different. As proposed in Horn (1972) and formalized in
Gazdar (1979), the alternative analysis is to obtain the one-sided, lower-
bounded ‘at least’ reading from semantics, and to derive the one-sided,
upper-bounded ‘at most’ reading using a pragmatic inference called a
conversational implicature (to be elaborated in Chapter 2). In other
words, on this account, a sentence like (1.6) asserts (or entails) the one-
sided, lower-bounded reading ‘at least nine girlfriends’, conversationally
implicates the one-sided, upper-bounded reading ‘at most nine girlfriends’,
and the conjunction of the assertion and the implicature results in the
corresponding two-sided, upper- and lower-bounded reading ‘exactly
8 1. INTRODUCTION
nine girlfriends’. Such a division of labour between pragmatics and seman-

tics allows us to avoid unnecessary semantic ambiguity and preserve se-
mantic parsimony, thereby approximately halving the size of the English
lexicon (Levinson 1983: 37–8, Horn 1988, 2004, Huang 2001a, 2004b,
2005). I shall discuss the relationship between pragmatics and semantics
in Chapter 7.
We move next to the potential simplification of syntax by pragmatics.
Within the principles-and-parameters theory and its minimalist descend-
ant, Chomsky (1981, 1995) distinguished two types of abstract feature for
NPs: anaphors and pronominals. An anaphor is a feature representation of
an NP which must be referentially dependent and bound within an appro-
priately defined minimal syntactic domain; a pronominal is a feature rep-
resentation of an NP which may be referentially dependent but which must
be free within such a domain. Interpreting anaphors and pronominals as
two independent binary features, Chomsky hypothesized that we ideally
expect to find four types of NP in a language—both overt and non-overt.
(1.7) Chomsky’s typology of NPs
Overt Empty
a. [þanaphor, pronominal] lexical anaphor NP-trace
b. [anaphor, þpronominal] pronoun pro
c. [þanaphor, þpronominal] — PRO
d. [anaphor, pronominal] name wh-trace/ variable
Putting aside the four types of empty category,3 the three lexically
realized types of overt NP can be illustrated in (1.8).
3
The four types of empty category are illustrated below.
(i) NP-traces
Susan’s brother was killed t in the Asian tsunami.
(ii) pros (Spanish)
Carreras sabe que pro es estimado por Domingo.
Carreras knows that is esteemed by Domingo
‘Carreras knows that (he) is respected by Domingo.’
(iii) PROs
John tried PRO to pass his driving test on the first attempt.
(iv) wh-traces/variables
Who do you love t?
See, for example, Huang (1992, 1994, 1995, 2000a) for comments on empty
categories.
(1.8) Overt NPs

a. Lexical anaphors
The composers admire themselves/each other.
b. Pronouns
He is Italian enough to understand the culture, and yet foreign enough to
see its peculiarities.
c. Names
Jonathan Swift wrote an eighteenth–century satire called Gulliver’s
Travels.
Of the three types of overt NP listed in (1.8), anaphors, pronominals, and

r[eferential]-expressions are subject to binding conditions A, B, and C
respectively.
(1.9) Chomsky’s binding conditions
A. An anaphor is bound in a local domain.
B. A pronominal is free in a local domain.
C. An r-expression is free.
Binding is defined on anaphoric expressions in configurational terms, appeal-

ing to purely structural concepts like c-command, government, and locality.
The binding theory accounts for the syntactic distribution of the three types of
overt NP listed in (1.8). Consider, for example, (1.10) from English.
(1.10) a. Bach1 adored himself1 .
b. Bach1 adored him2 .
c. Bach1 adored Bach2 .
In (1.10a), himself is an anaphor in the Chomskyan sense. As such, it

falls under binding condition A, according to which it is bound to its
local antecedent Bach. In (1.10b), him, being a pronominal, is subject to
binding condition B. Given binding condition B, it cannot be bound in
its local domain, and there is thus disjoint reference between it and
Bach. Finally, in (1.10c), the second Bach is an r-expression. By binding
condition C, it cannot be co-indexed with the first Bach. From examples like
these, Chomsky concluded that the syntactic distribution of anaphors,
pronominals, and r-expressions is accounted for by binding conditions A,
B, and C, respectively. However, when confronted with a range of languages
other than English, these binding conditions run into serious difficulties (see,
e.g., Huang 1991, 1994, 2000a, 2004a, 2006c for detailed discussion).
In recent years, an alternative, neo-Gricean pragmatic theory of anaph-
ora has been developed by Levinson (1987a, 1987b, 1991, 2000) and Huang
10 1. INTRODUCTION
(1991, 1994, 2000a, 2000b, 2004a, 2006c). On this account, binding condi-
tions B and C are reduced to pragmatics. In somewhat simplified terms,
this can be achieved in the following way. If binding condition A is taken to
be grammatically specified in English, then binding condition B is the direct
result of the application of a pragmatic principle called the Q-principle (to
be discussed in detail in Chapter 2). Given this principle, the use of a
semantically weaker pronoun where a semantically stronger reflexive
could occur gives rise to a conversational implicature which conveys the
negation of the more informative, coreferential interpretation associated
with the use of the reflexive, as in (1.10b). By the same reasoning, binding
condition C can also be eliminated. Wherever a reflexive could occur, the use
of a semantically weaker proper name Q-implicates the non-applicability
of the more informative, coreferential interpretation associated with the use
of the reflexive. This is exactly what has happened in (1.10c). All this
suggests that, as pointed out by Horn (1988: 115), ‘an independently motiv-
ated pragmatic theory (or several such theories, on the compartmentalized
view) should provide simplification and generalization elsewhere in the
overall description of language’. I shall return to the pragmatics of anaphora
and binding in Chapter 8, where the relationship between pragmatics and
syntax is to be addressed.
1.3. Some basic notions in semantics and pragmatics
So far, I have discussed what pragmatics is and why we need it. Next, I shall
turn to a number of basic notions in semantics and pragmatics.
1.3.1. Sentence, utterance, proposition
Let me begin with the distinction between sentence and utterance—a dis-
tinction that is of fundamental importance to both semantics and prag-
matics. A sentence is a well-formed string of words put together according
to the grammatical rules of a language. As a unit of the language system, it
is an abstract entity or construct defined within a theory of grammar. For
example, (1.11) is a sentence in English, but (1.12) is not. (In this subsec-
tion, a sentence is represented in italics, an utterance between double
quotation marks, and a proposition in small capitals.)
1.3. SOME BASIC NOTIONS IN SEMANTICS AND PRAGMATICS 11
(1.11) Lance Armstrong won the centenary Tour de France.

(1.12) *Lance Armstrong the centenary Tour de France won
Sentence-meaning, then, refers to those aspects of meaning that are

ascribed to a sentence in the abstract, that is, a sentence independent of
its realization in any concrete form. The study of sentence-meaning nor-
mally belongs to semantics.
By contrast, an utterance is the use of a particular piece of language—be
it a word, a phrase, a sentence, or a sequence of sentences—by a
particular speaker on a particular occasion. For example, (1.13)–(1.16)
are all utterances in English. (The quotation marks indicate that what is
put between them is taken from a specific context on a specific occasion
of use.)
(1.13) ‘‘Hello!’’
(1.14) ‘‘A cappuccino, please.’’
(1.15) ‘‘Lance Armstrong won the centenary Tour de France.’’
(1.16) ‘‘Some books are to be tasted, others to be swallowed, and some few to be
chewed and digested; that is, some books are to be read only in parts; others
to be read but not curiously; and some few to be read wholly, with diligence
and attention. Some books also may be read by deputy, and extracts made
of them by others.’’ (Francis Bacon)
Of these, (1.15) is an instantiation of the sentence in (1.11). In such a

case, it is widely assumed (à la Bar-Hillel 1954) that an utterance is the
pairing of a sentence and a context, that is, the situation in which the
sentence is uttered (Levinson 1983: 18–19). Utterance-meaning, or
speaker-meaning (as it is often called), then, is definable as what a speaker
intends to convey by making an utterance. The study of utterance-meaning
normally falls under pragmatics.
Finally, there is the notion of a proposition. A proposition is what is
expressed by a sentence when that sentence is used to make a statement,
that is, to say something, true or false, about some state of affairs in the
external world. Put the other way round, a sentence, when uttered to make
a statement, is said to convey a proposition. For example, (1.19) is the pro-
position underlying both sentences (1.17) and (1.18).
(1.17) Liszt adored Chopin.

(1.18) Chopin was adored by Liszt.
(1.19) liszt adored chopin
12 1. INTRODUCTION
The propositional content of a sentence is that part of its meaning which

can be reduced to a proposition. This notion allows semanticists to claim
that different (types of) sentences may share the same propositional con-
tent, even though they differ in other aspects of meaning. For example, the
interrogative sentence in (1.20) is said to have the same propositional
content as the active declarative sentence in (1.17) and the passive declara-
tive sentence (1.18), namely (1.19). The difference is that while in saying
(1.17) and (1.18) the speaker asserts the corresponding proposition, that is,
he or she commits him- or herself to the truth of the proposition, in uttering
(1.20) the speaker questions its truth.
(1.20) Did Liszt adore Chopin?
Propositions may be true or false (to be discussed below), may be known,

believed, or doubted, may be asserted or denied, and may be held constant
under paraphrase and translation. For example, (1.21) in Chinese, (1.22) in
English, and (1.23) in Italian can be said to express the same proposition.
(1.21) xia yu le.
(1.22) It is raining.
(1.23) Piove.
Regarding the relationship between sentence and proposition, the same

proposition can on the one hand be expressed by different sentences, as is
attested by (1.17) and (1.18) above. On the other hand, the same sentence
can be used to convey different propositions on different occasions. As a
case in point, consider (1.24).
(1.24) In his last years, my uncle fell victim to Alzheimer’s disease.
The sentence in (1.24), when uttered by different speakers, may say very
different things about some state of affairs in the outside world. If (1.24)
were said by Mary about her uncle John, she would mean something quite
different from what Henry would mean if he had used it to talk about his
uncle Albert. Thus, on these two different occasions, the use of the same
sentence would express two distinct propositions.
Try putting these principles into practice by doing Exercise 1 on p. 18.
Finally, the relationship between sentence, utterance, and proposition

may be represented schematically in the tree diagram in (1.25) (adapted
CHAPTER 3 1
2
3
Deixis 4
5
6
7
8
9
10
1
2
3
3.1 WHAT IS DEIXIS? 411
5
As discussed in Chapter 1, the term pragmatics is often used in linguistic research to refer 6
to the study of the interpretation of meaning. Although it has proven difficult to determine 7
an exact definition for the term pragmatics (Levinson discusses the issue over more than 8
50 pages in his influential 1983 work Pragmatics), a user-friendly definition is that 9
suggested by Fasold (1990: 119) as ‘the study of the use of context to make inferences 20
about meaning’. In this definition, inferences refer to deductions made by participants based 1
on available evidence (Christie, 2000). As we mentioned in Chapter 1, this available 2
evidence is, according to pragmaticists, provided by the context within which the utterance 3
takes place. Therefore, because, according to Levinson (1983: 54), ‘the single most obvious 4
way in which the relationship between language and context is reflected in the structures 5
of languages themselves, is through the phenomenon of deixis’, deixis is integral to the 6
study of pragmatics. Derived from the Greek word for ‘pointing’ or ‘indicating’ (deiktikos: 7
‘apt for pointing with the finger’), deixis refers to the way in which speakers orientate both 8
themselves and their listeners in relation to the context of a conversation. Deixis enables 9
interlocutors to refer to entities in context, thereby allowing them to identify people and 30
things in relation to the space they are operating in at the moment at which they are 1
speaking. There are a number of grammatical items that encode deixis, for example, the 2
demonstratives, this, that; first and second person personal pronouns, I, you, we; adverbs 3
of time such as now, then; adverbs of space such as here, there; motion verbs such as 4
come, go; and a variety of other grammatical features such as tense markers. These 5
grammatical items that encode deixis are commonly referred to as deictics. 6
In this chapter, we aim to provide an analysis of the occurrence and function of deictics 7
in everyday conversation. It is important to note that, in contrast with many of the examples 8
used in other explorations of deixis, the examples used in this chapter are from real-life 9
data and are therefore attested, not invented. Although deixis exists in both the spoken and 40
written domains, this chapter primarily focuses on spoken language. We begin with an 1
exploration of how to recognise when an item has a deictic function and then make the 2
distinction between gestural and symbolic uses of deictics. This is followed by analysis of 3
traditional and contemporary notions of the deictic centre. The classical categories of deixis, 4
person, place and time, are explored and discourse, social and empathetic deixis are also 5
examined. Woven through the chapter is the way in which corpus linguistic techniques can 6
inform our study of deixis. Corpus techniques, such as word frequency lists, allow the 47
DEIXIS 37
researcher to bring into relief the importance of deixis in our everyday speech and writing.
It is generally accepted that deixis is extremely prevalent in everyday speech and writing
but corpus linguistics allows us to show just how frequent it is in relation to other grammatical
features. Corpus linguistics also enables the researcher to conduct comparisons of how
deixis is used across different genres (for example, casual conversation, academic discourse,
and so on) as well as across different modes (spoken and written language).
3.2 DEICTIC VERSUS NON-DEICTIC EXPRESSION
Personal pronouns, demonstratives (both pronouns and determiners) and adverbs of space
and time can be used both deictically and non-deictically, as will be seen in the examples
here. As we have already mentioned, to classify something as deictic means that the
expression derives part of its meaning from the context. Deictics allow the interlocutors
to ‘point’ to something in the context thereby enabling them to orientate themselves in a
variety of ways, be it personally, spatially or temporally.
Personal pronouns
The examples given in 3.1 demonstrate the use of a personal pronoun, in this case you,
in both deictic and non-deictic senses:
(3.1)
Deictic usage Non-deictic usage

A: I owe you a fiver. A: There’s a school that’s out there that you book in for a
week and you can learn how to hang-glide.
In the deictic usage of you, the speaker identifies a particular person, the addressee. In this
example, the addressee is the referent. This deictic use of you in 3.1 is also likely to be
accompanied by some gesture such as eye contact (in this example, a fiver refers to Irish
and British English slang for a £5 or €5 note). In the non-deictic use, you is used to refer
to people in general, and therefore does not rely on the context for meaning. In this example
it is implied that anyone can learn how to hang-glide in this school, a usage similar to one
in English, on in French or man in German. This usage is sometimes referred to as generic
you (see Tao, 1998; Biber et al., 1999: 353–5; Carter and McCarthy, 2006: 120).
Deictic use of this, that, these and those
Green (1995) claims that for many linguists and philosophers, demonstratives lie at the
heart of deictic issues. The demonstrative this and its plural form these, when used
deictically, often refer to things that are close in space or time whereas that (plural those)
refers to things that are a little more distant (see Section 3.5); however, this is dependent
on the speaker’s perception. The deictic use of this as a determiner in 3.2 is probably
38 DEIXIS
accompanied by some kind of pointing gesture, something that indicates that the speaker 1
is referring to one particular bottle (the bottle is the referent): 2
3
4
(3.2)
5
6
7
8
9
A: Will I give him A: I mean the post office is one of our meeting places
10
this bottle? where the people would go along and inquire about this
1
and that or another thing.
2
3
The non-deictic usage of this in 3.2 is classified non-deictic because it is part of a fixed
411
phrase this and that which is a marker of vague language, sometimes referred to as a
5
vague category marker (see Chapter 7), here meaning various, unspecified or trivial
6
matters.
7
Similarly, the deictic use of that as a demonstrative pronoun in 3.3 would more than
8
likely contain a gesture on the part of the speaker and so would require some degree of
9
physical monitoring of the context on the part of the addressee in order to correctly interpret
20
the utterance:
1
2
(3.3)
3
4
5
6
7
A: That one is nice. A: But I think that one of the obvious things in all this
8
affluence there are people who are benefiting from the
9
Celtic Tiger, you know?
30
1
On the other hand, the non-deictic use of that demonstrates its use as a complementiser.
2
3
4
5
That 6
7
Biber et al. (1999: 350) describe that as one of the most common and
8
most flexible word forms in English. In addition to its deictic function and that of a
9
complementiser – see extract (3.3) – that can also function as:
40
1
• A relative pronoun:
2
e.g. Daddy where’s the balloon that was over the door?
3
• A stance adverbial:
4
e.g. He’s not that heavy.
5
6
47
DEIXIS 39
Adverbs of time and space
In example 3.4, there when used deictically is an adverb of space indicating position. In
this case there refers to the location of a wedding reception:
(3.4)

A: There was a A: There was a good crowd there.
good crowd there.
The non-deictic there featured in the above example functions as a dummy subject.
Dummy subjects are considered to have no semantic content and ‘simply fill the
necessary subject slot’ (Carter and McCarthy, 2006: 495).
Finally, now used deictically is an adverb of time, in this case meaning time ‘around
now’. However, non-deictically it functions, for example, as a discourse marker that serves
to introduce something which contrasts with what has just been said as in 3.5:
(3.5)

A: I had a letter A: Now if after a time they found that that wasn’t having
written to her and I’ll making any bite or impact on the company they then
post it now maybe. called for an all out strike.
3.3 GESTURAL VERSUS SYMBOLIC DEIXIS
Once a linguistic item has been determined as deictic, a distinction can be drawn between
gestural and symbolic deixis (see Levinson, 1983: 65–6; Fillmore, 1997: 62–3). Lyons
(1977: 10) has suggested that deixis is ‘at its purest . . . where the utterance is accompanied
by some sort of extra-linguistic gesture’. As the term suggests, gestural deixis is frequently
accompanied by a gesture such as physically pointing, making eye contact or turning your
body to ‘face’ someone. Therefore, a gestural deictic can only be properly interpreted by a
physical monitoring of the speech situation. Extract 3.6 features three female friends getting
ready for a night out and contains an example of this (marked in bold) used gesturally:
(3.6)
A: Aw it’s fab it’s gorgeous looking. Look at the state of my elbows. They’re all fake
tan and Carol always goes ‘what’s wrong with your elbows Michelle? ’
40 DEIXIS
B: Gillian look do I need to iron this skirt cos there are creases all down the front? 1
C: No it’s grand they’ll fall out. 2
3
The use of this by Speaker B in Gillian look do I need to iron this skirt in extract 3.6 4
is classed as gestural as it is easy to imagine the speaker holding up the skirt while asking 5
the question. Similarly, the gestural use of that (marked in bold) is illustrated in extract 3.7. 6
In this extract, a mother (B) and daughter (A) are sitting together, discussing what colour 7
to paint the walls of the daughter’s house. (The word shur in the extract is a discourse 8
marker commonly used in Irish English. It often collocates with but, and in this example it 9
is used as a polite way of saying however): 10
1
(3.7) 2
3
411
5
A: Oh I know that. But you see the thing is Jean you have to think of your walls. 6
B: Yeah. 7
A: I mean if I put ah that colour which way then can I go? 8
9
B: Yeah I know but shur this doesn’t really match this like does it?
20
A: Yeah I think it does.
1
B: Do you?
2
A: I think it’s kind of good. I think. I don’t know about that one. Maybe tis too yellow.
3
B: Yeah I think it is.
4
5
The daughter (Speaker A) refers to that colour and that one in the course of the
6
conversation. This use of the deictic might be accompanied by a gesture that isolates the
7
colour in question. On the other hand, the symbolic use of a deictic item points to context
8
outside the text for meaning, to referents ‘in the common cultural background’ (Cutting,
9
2008: 9). Symbolic deixis may also signify entities that are not immediately visible in the 30
immediate speech situation. For example, extract 3.8 features the beginning of a phone 1
call in a radio phone-in show: 2
3
(3.8) 4
5
6
7
A: I think we have Cian O’Donovan on another line. 8
B: Hello. 9
A: Cian are you there? 40
B: I am indeed yeah. 1
A: How’re you feeling today? 2
B: Ah I’m a bit nervous. I’m looking forward to getting it over with though. 3
4
The use of there in extract 3.8 would typically be considered an example of symbolic 5
deixis as, although a referent is picked out, it is unlikely to be marked by any gesture as 6
there is not in the speaker’s immediate context. However, because both speaker and 47
DEIXIS 41
addressee understand where they are spatially in relation to one another, there can be
easily interpreted deictically.
In the gestural and symbolic examples of you given in 3.9, What about you Jack?
can be interpreted as gestural, since the example comes from a conversation involving
multiple parties and, therefore, it is possible that eye-contact is being made between the
participants. (However, it is difficult to classify you as gestural without either being present
in the conversation or having access to a multi-modal corpus.) The symbolic use of you is
taken from an information technology lecture at university:
(3.9)
Gestural usage Symbolic usage

A: What about you A: So what you want to do is you want to type in the
Jack? first example that I have up here.
Here, the symbolic usage of you is much more general – it refers to ‘plural you’, to all the
participants at the lecture. This symbolic use identifies a referent – in this case the students
present at the lecture – and also delimits you to refer to this audience. It is unlikely to be
accompanied by any gestural behaviour (contrast this with the non-deictic, generic you
in 3.1).
In 3.10, the gestural example of now comes from a conversation in which the parti-
cipants are fixing a computer printer. The example of now classified as gestural is done
so because of contextual information provided by the researcher involved in the recording
and transcription of the conversation. It might work now is immediately followed by the
contextual information noted in the transcription that the speaker presses a button on the
printer. Therefore, now coincides with a gesture:
(3.10)
Gestural usage Symbolic usage

A: It might work now. A: The way I look at it is I can do this again in a year if I
want to whereas I want to see where this is going. I don’t
want to go and say what if? I can go to Cambodia in a
year’s time. If I go now then I risk this and I might say why
did I let him go?
On the other hand, in order to interpret the symbolic use of now, the addressee does
not expect any physical gesture. Provided the addressee knows the time at which the
utterance is taking place, he/she can interpret and understand now without any great difficulty.
In addition to this, Huang (2007) notes that it seems in general if a deictic expression can
be used symbolically, it can also be used in a gestural way, but not vice versa.
To summarise, Figure 3.1 illustrates the range of possible uses that a deictic expression
may have. As we have shown, the same expression may be used both deictically and non-
42 DEIXIS
1
Deictic expression 2
3
4
5
Deictic use Non-deictic use 6
7
8
9
10
Gestural Symbolic
1
2
3
Figure 3.1 Possible uses of a deictic expression. 411
5
deictically. In addition, the deictic use can be subdivided into both the gestural use and 6
the symbolic use. 7
8
9
3.4 THE DEICTIC CENTRE 20
1
The deictic centre, or origo (a term coined by Bühler in 1934), refers to a perceived egocentric 2
organisation of the deictic system. Traditionally, this centre is typically organised around an 3
‘I–here–now’ axis where ‘the speaker casts himself in the role of the ego and relates everything 4
to his viewpoint’ (Lyons, 1977: 638). Therefore, in relation to the centre (or anchor point) for 5
the three major categories of deixis – person, place and time – the centre for person deixis 6
is the speaker (the ‘I’), the centre for place deixis is where the speaker produces the utterance 7
(the ‘here’) and the centre for time deixis is the time at which the utterance is produced (the 8
‘now’). The deictic centre is associated with the current speaker, therefore, when the speaker 9
changes so too does the centre. In extract 3.11, in which four females are getting ready for 30
a night out, the use of the personal pronoun I demonstrates this deictic centre switch: 1
2
(3.11) 3
4
5
6
A: What top are you wearing? 7
B: The lemon one. 8
C: I don’t have a jacket. 9
A: You have a jacket the denim one. 40
D: I have a jacket cos it matches. Do you remember one night out in Dublin I 1
brought no jacket? 2
C: I do. 3
D: And I got very cold. 4
5
Speakers C and D all use I in their speaker turns indicating their acknowledgment that 6
they are now at the deictic centre of the conversation. In addition to this, the personal 47
DEIXIS 43
pronoun you moves between several different referents. In Speaker A’s first turn, she uses
you to refer to Speaker B. In her second turn, she uses you to refer to Speaker C. Speaker
D also uses you to refer to Speaker C in Do you remember one night out in Dublin I
brought no jacket? This constant movement of centre from participant to participant can
cause problems for both first and second language acquisition.
However, the notion of an egocentric origo has not gone unchallenged (see, for
example, Lyons, 1977; Hanks, 1992; Jones, 1995). According to Jones (1995), theories
in favour of the egocentricity of deixis fail to take into account that communication is a
social act. He claims that it is inaccurate to call the speaker in conversation the ‘central
person’ just as it would be to call either the husband or wife the ‘central person’ in a
marriage – ‘it takes two to tango and two to communicate’ (p. 37). This notion of a non-
egocentric origo can be seen in the interaction of a father (Speaker A) with his baby son
(Speaker B) in extract 3.12 while they are sitting at the breakfast table:
(3.12)
A: Are you goin with daddy?

B: Yeah.
C: Who’s funeral’s on now mammy?
A: Eat the breakfast so and we go.
D: Jim Smith’s.
A: Are you goin with daddy in the motor car?
B: Oh sorry. I’m goin with daddy.
In this extract, the father clearly aligns his own identity with that of his baby and this
is achieved in a number of ways (marked in bold in the extract). First, his utterance Are
you goin with daddy?, repeated twice, connects you with daddy thus establishing a shift
from a lexical realisation of self for the father to the perspective of his child. The everyday
identity the father carries with him is the I, however, to his son he is daddy and his self-
reference of this is evident. Second, in the utterance Eat the breakfast so and we go, the
father combines his and his son’s perspective into one and the same using an inclusive
we (see Section 3.5). Wales (1996: 56) offers a number of possible reasons for parents’
use of kin titles like daddy in talking to infants. She suggests that parents use them
because they ‘recognise the problematic “shifting” nature of speaker/addressee roles, of
I and you reference’. This recognition of the difficulty for children with I/you reference is
acknowledged by the parents switching the origo from themselves to their children through
their use of you. This would seem to highlight the flexibility of the deictic centre – an
utterance is necessarily centred in the sense that some speaker’s viewpoint is inevitably
assumed; however, whose viewpoint is assumed, the speaker’s or a co-participant’s, is a
matter of flexibility. In extract 3.12, the father’s utterances reflect the close connection
that he feels with his son constituted by his use of deictic expressions (see also Tannen,
2007). Although the examples used to illustrate this section are person deictic in nature,
the deictic centre is also associated with place and time deixis and this will be discussed
later in the chapter.
44 DEIXIS
1
2
The deictic centre 3
4
(1) The central person is the speaker. 5
(2) The central time is the time at which the speaker produces the utterance. 6
(3) The central place is the speaker’s location at utterance time. 7
(4) The discourse centre is the point at which the speaker is currently at in the 8
production of his [sic.] utterance. 9
(5) The social centre is the speaker’s social status and rank, to which the status 10
and rank of addresses or referents is relative. 1
Levinson (1983: 64) 2
3
411
5
3.5 BASIC CATEGORIES OF DEIXIS 6
7
Deixis is traditionally subdivided into a number of categories: person, place and time deixis 8
are the most common of these categories. Two additional categories, discourse and social 9
deixis, are also present in some of the most influential work done in the area (cf. Lyons, 20
1977; Levinson, 1983; Fillmore, 1997). There is also a sixth deictic category referred to 1
as empathetic deixis (see Lyons, 1977). 2
3
4
Person deixis 5
6
Person deixis is concerned with ‘the identity of the interlocutors in a communication 7
situation’ (Fillmore, 1997: 61). Personal pronouns such as I and you are the most obvious 8
and most frequent manifestations of person deixis. In order to illustrate the frequency and 9
importance of these and other personal pronouns in casual conversation, frequency counts 30
for the top 20 most frequently occurring words were generated for the Limerick Corpus 1
of Irish English (LCIE) using Wordsmith Tools (Scott, 2009) and the results are displayed 2
in Table 3.1. 3
Table 3.1 demonstrates that five personal pronouns – I, you, it, he and they (marked 4
in bold in Table 3.1) – occur in the top 20 most frequent words in LCIE, which is consistent 5
with many other corpora of spontaneous, face-to-face casual conversation. Similar findings 6
have also been recorded by Biber et al. (1999) working with the Longman Spoken and 7
Written English Corpus (LSWE), O’Keeffe et al. (2007) working with the CANCODE and 8
CIC corpora, and Rühlemann (2007) working with the BNC. This is in contrast with written 9
corpora. For example, frequency lists based on the written component of the BNC show 40
I in 17th position and you in 21st (Leech et al., 2001b). Regarding the distribution of these 1
pronouns, Biber et al. (1999: 333) have shown that the pronouns I and you are far more 2
common in casual conversation than in other registers such as academic prose. Rühlemann 3
(2007: 66–9) posits four reasons for the preferred use of I and you in casual conversation: 4
5
(1) I is prone to repetition (I is repeated at a frequency of about 200 times per million 6
words in conversation: see Biber et al., 1999: 334); 47
DEIXIS 45
Table 3.1 Top 20 most frequent words in LCIE (personal pronouns in bold)
Rank order LCIE Frequency per million words
1 the 35,171
2 I 24,321
3 and 23,707
4 you 23,011
5 to 20,140
6 it 18,276
7 a 17,753
8 that 14,868
9 of 13,948
10 yeah 13,689
11 in 13,401
12 was 10,419
13 is 10,006
14 like 8,667
15 know 8,052
16 he 7,365
17 on 7,281
18 they 7,264
19 have 6,831
20 there 6,628
(2) I and you have a high frequency of collocation especially with verbs of thinking, for
example, I think and you know;
(3) discourse is typically presented in a direct mode, where, for example, a non-present
speaker’s words are represented as if he/she were actually present;
(4) conversation is co-constructed, with speakers taking turns and each new turn requires
the reconstruction of the new speaker’s deictic system.
As we have already seen in extract 3.11, the pronouns I (and by extension my and
mine) and you (also your, yours) are typically deictic. However, other personal pronouns
such as he, she, it and they, although on occasion deictic, are typically anaphoric in their
reference. In order to distinguish between deixis and anaphora (and cataphora) it is
necessary to briefly explore the notion of context. Cutting (2008: 3–11) distinguishes
between three different types of spoken context: situational, what speakers know about
what they can see around them; background knowledge, what they know about each other
(interpersonal knowledge) and the world (cultural knowledge); and co-textual, what they
know about what they have been saying. Deixis is associated with context outside the text
(text can refer to pieces of both spoken and written discourse), and, therefore, the referent
is generally located in the situational or background knowledge context. On the other hand,
anaphora and cataphora are associated with the context of the text itself, or the co-textual
context. Therefore, the referent is generally located in either the preceding text (anaphora)
or the forthcoming text (cataphora). Deixis and anaphora are often considered inde-
pendently; however, just how independent they are from one another is a subject of debate
(see Lyons, 1977; Levinson, 1983; Wales, 1996; Marmaridou, 2000; Rühlemann, 2007).
Indeed, Lyons (1977: 676) maintains that a term can be used both deictically and
anaphorically. Take the example I was born in New York and have lived there ever since.
46 DEIXIS
In this example, adapted from Levinson (1983: 67), there simultaneously refers backwards 1
to New York but also contrasts with here in the space deictic dimension, locating the 2
utterance outside of New York. Similarly, in Halliday and Hasan’s (1976: 3) example Wash 3
and core six cooking apples. Put them into a fireproof dish, them does not refer to 4
the six cooking apples per se but rather to the washed and cored apples, a reference 5
which is constructed on the basis of background contextual knowledge of the genre of 6
recipes (see Wales, 1996). Rühlemann (2007: 63) maintains that ‘another difficulty in 7
drawing the line between deixis and anaphora derives from the fact that deictic items may 8
well be used in institutionalised expressions in which only faint, if any, traces of the deictic 9
origins persist’. He cites examples such as the vague expression this and that and the 10
discourse marker there you go. Anaphora and deixis, therefore, appear to be closely related. 1
In extract 3.13, a younger sister is talking about her older siblings and using he and 2
she (marked in bold) anaphorically: 3
411
(3.13) 5
6
7
8
A: My oldest brother is Jimmy and he’s [pause] I’m not sure about their ages because 9
there are too many of them but he’s in the army and he’s a military instructor 20
or something and I’ve another sister Martina and I don’t know her age either and 1
she’s a pharmacist and she teaches. 2
3
As we can see, he in the extract refers back to the initial mention of her older brother 4
Jimmy, whereas she refers to the preceding mention of her sister Martina. Similarly, they 5
is typically anaphoric in nature, illustrated by the conversation had by two males while 6
watching television in extract 3.14: 7
8
(3.14) 9
30
1
2
A: I saw Pirates of the Caribbean last night, which was very good. 3
B: I saw the ad for that. It’s very good. The pirates are really ghosts. 4
A: See there was a curse put on them, because they are all moody, so when they 5
are walking around in the moonlight you can really see them for who they are. 6
So they need to get back all these coins into the one place, and then they need 7
to get the chosen one, and they’ve to let his blood pour onto the coins, and then 8
the curse will be lifted. Well they think they found her but they have to bring 9
her back alive so that the curse is lifted. 40
B: Why are they ghosts? 1
A: They want to be alive again. 2
B: Okay. 3
4
Again, they (in bold) is used to refer to the pirates who are initially mentioned by 5
Speaker B. It is important to note that anaphoric and cataphoric pronouns refer to individual 6
noun phrases in the text (contrast this with the use of discourse deixis below). 47
DEIXIS 47
We have already outlined that deixis enables interlocutors to orient themselves in their
immediate context. The first person pronoun I allows the person to identify themselves in
the ‘role’ as the speaker, whereas the second person pronoun you enables the speaker
to refer to the role of the addressee(s). Third person pronouns such as he, she and they
most frequently identify people that are neither speakers nor addressees, though on occa-
sion they can be used to refer to the speaker or listener. Therefore, generally, the third
person pronoun does not refer to any specific ‘participant role’ in the immediate context,
thereby allowing its classification as typically non-deictic (Lyons, 1977: 638; Levinson,
1983: 69; Huang, 2007: 137).
Finally, there are also two first person pronoun plural we pronouns. These are used
to create a perspective of:
(1) I the speaker + you the addressee(s) in the immediate context (‘inclusive we’)
and
(2) I the speaker + someone else not in the immediate context (‘exclusive we’).
The use of an inclusive we (marked in bold) is evident in extract 3.15 from where
three siblings are in their living room discussing the origins of the name of their dog:
(3.15)
A: But Goldie’s a girl’s name like.

B: Yeah but we didn’t give her the name.
A: What?
B: We didn’t give her the name.
C: We didn’t give her the name. Although she was so young she wouldn’t notice it.
A: She wouldn’t have a clue shur.
C: We could’ve changed it. We could call her Alex.
Earlier in the conversation, Speaker A has been complaining about the name of the dog,
Goldie, and suggesting different names for her. The other siblings, Speakers B and C, use
inclusive we in the repeated utterance We didn’t give her the name as a form of ‘safety
in numbers’ defence (I, the speaker + you, the rest of the family) to deflect the criticism
of the dog’s name from themselves. Mühlhäusler and Harré (1990: 174) claim that in this
use of we ‘the social bonding aspect and the establishment of solidarity is of importance’.
The siblings create an in-group, ‘we the family’, in opposition to the person who originally
named the dog. Further to this, Speaker C adds We could’ve changed it. We could call her
Alex, invoking the power that ‘we the family’ had, and still have, to change the name of
the dog should they choose to do so.
On the other hand, exclusive we can be used by the speaker to refer to a range of
groups that include the speaker but not the addressee(s) in the immediate context. Extract
3.16 is taken from a recording made while a family was putting up their Christmas tree.
The extract contains a number of instances of the use of inclusive we (marked in bold)
and exclusive we (marked in bold and underlined):
48 DEIXIS
(3.16) 1
2
3
4
A: Do we need more around there? 5
B: Yeah a bit more. 6
A: All right? 7
B: Yip. We might need another bit now but hang on a second and I’ll press it down. 8
C: We have them outside too the eighty mini bulbs. Is that what they are? Eighty 9
mini bulbs. Yeah we’ve them too. 10
D: We don’t need those til tomorrow. 1
B: Are we not putting it up til tomorrow dad no? 2
3
The first two instances of we (Do we need more around there? and We might need 411
another bit now . . .) are examples of inclusive we, given that the speaker in both instances 5
is referring to themselves and the other participants assisting them with the task of putting 6
up the tree, the other members of the family. Similarly, We don’t need those til tomorrow 7
and Are we not putting it up til tomorrow dad no? refer to the participants in the immediate 8
context. However, Speaker C’s assertion We have them outside too the eighty mini bulbs 9
and Yeah we’ve them too are examples of exclusive we. Both we and outside refer to 20
Speaker C’s place of work: we referring to her membership of the staff of the organisation; 1
and outside referring to the location of the workplace, perhaps in contrast to the ‘inside’ 2
context of the family home. Speaker C is referring to context outside the immediate con- 3
versation and relying on the ‘common ground’ between the interlocutors to enable them 4
to assign the correct interpretation to the deictic item. 5
The English language system of person deixis can present many difficulties for the 6
non-native speaker. This is due in part to the issues explored in this section; however, it 7
is also due to the fact that many languages have more complicated systems. In many 8
languages there are three second person pronouns you (singular, plural and formal) – du, 9
ihr and Sie in German, for example. These are used as a formal way of addressing someone 30
and, in addition to functioning as person deixis, also function as social deixis (see below). 1
2
3
Place deixis 4
5
Sometimes called spatial or locational deixis, place deixis is primarily concerned with the 6
location of people or objects relative to the deictic centre. As has been demonstrated, the 7
deictic centre frequently has the speaker as centre, therefore, these locations are often 8
relative to the speaker’s position. Notions of place deixis are commonly expressed using 9
demonstratives (this, that, these, those), deictic adverbs of place (here, there) or verbs of 40
motion (come, go, bring, take). However, locations can also be specified relative to other 1
objects using prepositions of place such as above, below, left, right, behind or from. In 2
many languages, there exists a fundamental distinction between proximal (or relatively 3
close to the speaker) and distal (non-proximal, sometimes relatively close to the addressee) 4
deictic expressions of place (Diessel, 1999; Levinson, 2004). For example, here and there 5
are crucially tied to the deictic field of the speaker, and the addressee(s) determines the 6
spatial coordinates of the utterance in order to assign meaning. The extent to which context 47
DEIXIS 49
is involved in determining the reference of here and there is evident in the interaction
between a grandmother (Speaker A), her daughter (Speaker B) and her grandchild (not
speaking but crawling around the room) in extract 3.17:
(3.17)
A: Come over here to Nana. Over here to Nana.

B: Come on. I’m sitting up here. Now I’m here. Look at it’s down there on the floor.
Look it’s gone out of your hand.
A: Ah you silly billy.
B: She doesn’t want that chair. She wants the other one I think over there.
In this extract, we can see that both Speakers A and B use here (marked in bold) to
indicate their own position relative to the child. However, it is obvious that Speaker A’s
here is different to Speaker B’s here. This use of proximal deixis is in contrast to Speaker
B’s use of there on two occasions (again marked in bold). On the first occasion the distal
reference in It’s down there on the floor refers to the location of something that the child
has dropped and is in direct contrast to Speaker B’s assertion that I’m sitting up here. In
addition to this, Speaker B states that She wants the other one I think over there which
demonstrates how ‘over + there’ is further away again from the speaker’s deictic centre.
Similarly, the demonstratives this and that also display this proximal–distal distinction. In
extract 3.18, a mother and daughter, whom we first encountered in extract 3.7, are
continuing their discussion of what colour to paint the walls (the demonstratives are marked
in bold):
(3.18)
B: Yeah but shur Jean I mean what colour do you want to put on the wall?
[
A: Too pale.
B: I mean you don’t want dark walls either.
A: Yeah no. What about this one?
B: That one isn’t bad.
A: ‘Sunshine’.
B: Yeah that was the colour of the other one wasn’t it?
A: Yeah and it’s kinda similar to the other one isn’t it? But less flashy.
Extract 3.18 demonstrates that as the deictic centre changes, the demonstrative
reference changes from proximal to distal even though it is the same object that both
speakers are talking about. The speakers are looking at a colour chart. Speaker A suggests
a colour What about this one? Speaker B, in her response That one isn’t bad, acknowledges
that the deictic centre has shifted from one speaker to the next (hence Jespersen’s (1965)
utilisation of the term shifters to refer to deictic expressions).
50 DEIXIS
Time deixis 1
2
According to Huang (2007: 144), time deixis ‘is concerned with the encoding of temporal 3
points and spans relative to the time at which the utterance is produced’. In other words, 4
in order to correctly interpret a time deictic, it needs to be considered in relation to the 5
time at which the communicative act takes place. To do this correctly, we need to distinguish 6
between coding time (CT), the moment of utterance, and receiving time (RT), the moment 7
of reception. Coding time is usually located around the speaker, whereas receiving time 8
is located around the addressee. Because the majority of conversation is face-to-face in 9
nature, CT and RT are considered to be identical. However, there are situations where the 10
CT and RT are different and this can lead to a situation where the utterance becomes 1
unanchored (see Fillmore, 1997: 60). For example, Fillmore (ibid.) suggests a ‘worst case 2
scenario’ for an ‘unanchored’ sentence: finding a message in a bottle which says Meet me 3
411
here at noon tomorrow with a stick about this big.
5
Notions of time deixis are expressed using both simple adverbs of time such as now,
6
then, today, tomorrow, yesterday and complex adverbs of time such as this month, next
7
year or last week. Similarly to place deixis, time deixis also distinguishes between a proximal
8
time now (time ‘around now’ including the CT) and a distal time then (time ‘not now’). Then
9
can be used to refer to both a particular past time and a particular future time. In extract
20
3.19, a workplace discussion about buying a house features then (in bold) used to refer
1
to past time:
2
3
(3.19)
4
5
6
7
A: They were to revise loads of figures. I had all those figures done and then all
8
the mortgage rates changed again and they were to come back to me with the
9
revised figures and they never did. 30
B: Jesus you’d better get on get on to them about that. 1
A: I know. 2
3
In contrast, in extract 3.20 a nurse and a student nurse are discussing upcoming 4
exams and then (in bold) is used to refer to future time: 5
6
(3.20) 7
8
9
40
A: Are you all ready for Wednesday? 1
B: I am indeed Mary. 2
A: Well then how’s the study going? You have it all done at this stage anyway I’d 3
say Connor? 4
B: I’m taking some time off lately. I’m resting. 5
A: Yeah you’d want to take some time off you’d be wrecked from studying. So when 6
will you be finished? Are you going straight through yeah? 47
DEIXIS 51
B: Basically the ah it starts on the fourth and I’ve exams all the way to the tenth
and then I’ve a week off and I’ve economics.
A: And then you’re finished? So you’ve nothing done in the economics. You’re
waiting for your week off.
B: I am yeah.
A: Yeah yeah yeah. So listen best of luck anyway I’m sure you’ll be fine.
B: Thanks.
The time deictic adverbials yesterday, today and tomorrow divide time into diurnal
spans. These time deictics can be used to refer to a specific time (extract 3.21), a time
period within the relevant day (extract 3.22) or to the day in general (extract 3.23):
(3.21) [Context: family at the dinner table]
A: Is there more meat there?

B: Yeah. We’ll have bacon and cabbage tomorrow like or will we do chicken?
A: You’re the cook.
(3.22) [Context: ending sequence in a radio phone-in show]
A: John thank you very much indeed for that. All sorts of other theories on why,
when, where, how etc. Anyway that’s all from us for today, back with you
tomorrow at the usual time until then a very good day to you.
(3.23) [Context: friends chatting]
A: Was he in better form yesterday?

B: Ah he was in better form yesterday.
As we can see, in extract 3.21 tomorrow is used to refer to dinner time, in extract
3.22 today is used to refer to the time period during which the show is broadcast and
yesterday in extract 3.23 refers to the entire day.
Corpus linguistic methods also substantiate the assertion that today, tomorrow and
yesterday are preferred in spoken discourse over lexicalised names of days like Monday,
Tuesday and so on (see Levinson, 1983; Fillmore, 1997; Grundy, 2008). Table 3.2 illustrates
the frequency of occurrence of the time deictics today, yesterday and tomorrow compared
to the occurrences of the seven days of the week in LCIE.
Table 3.2 demonstrates that the three time deictics, today, yesterday and tomorrow,
occur more frequently than the lexicalised names of the days of the week. The time deictics
occur 1,234 times per million words, whereas the days of the week have a frequency of
945 occurrences per million words. Interestingly, today is the most frequent time deictic
52 DEIXIS
Table 3.2 Comparison of frequency of occurrence of yesterday, today and tomorrow with the 1
lexicalised names for days of the week 2
3
Time deictic Frequency per Day of the week Frequency per
million words million words 4
5
today 549 Monday 111 6
yesterday 389 Tuesday 100 7
tomorrow 296 Wednesday 85
8
Thursday 106
Friday 172 9
Saturday 202 10
Sunday 169 1
Total 1,234 Total 945 2
3
411
5
in Table 3.2, and this, coupled with the high frequency of occurrences of the personal
6
pronoun I in Table 3.1, adds credence to the ‘I’ and ‘now’ components of the ‘here–now–I’
7
deictic centre.
8
Complex adverbs of time such as last month, next Monday or this year combine a
9
deictic expression such as this with a non-deictic such as year. In the deictic use of the
20
combination of this + week/month/year, on the surface at least, it appears that the referent
1
includes the coding time (the speaker is referring to the present week, month and so forth
2
in which the utterance takes place) and can vary between the non-calendrical measure
3
of, for example, seven days in a week to the calendrical measure of a week beginning, for
4
example, on the 1st of February and ending on the 7th. However, especially in the case
5
of this year, as extracts 3.24, 3.25 and 3.26 demonstrate, the interpretation of the referent
6
is not quite so straightforward.
7
8
(3.24) [Context: two friends chatting while watching a match] 9
30
1
2
A: I have to drop three subjects this year. I dropped German. 3
B: That’s not right. 4
A: I need them apparently. I got a C in Technology. And I dropped Home Economics. 5
I was going to keep Home Economics except she got really nasty. 6
7
(3.25) [Context: family chatting] 8
9
40
1
A: Every month has twenty nine days. 2
B: Well isn’t he very knowledgeable? Twenty eight. 3
A: Twenty nine. 4
B: But every month has twenty nine days. 5
A: But February didn’t this year. 6
B: Well he was wrong so. 47
DEIXIS 53
(3.26) [Context: radio interview. The All Ireland refers to the All Ireland Senior Football
Championship which is a competition in the game of Gaelic Football, played in
Ireland.]
A: Am well I at the start of the year I was asked to see to tip who I thought would
win the All Ireland and I was torn between Armagh and Tyrone so I’ll have to
stick with what I said at the start. I’ll have to keep going I think Tyrone have a
have a brilliant chance of winning the All Ireland this year. They’re very hungry
you know they’ve been out there and they’ve playing their hearts out all the time.
As we can see, these utterances containing this year are related to the examples
shown for today, tomorrow and yesterday. In extract 3.24, which features two friends
chatting during a match, in the utterance I have to drop three subjects this year, this year
is used to refer to the school year that runs in Ireland from September to May. In extract
3.25, one of the family members, Speaker A, claims But February didn’t this year which
refers to this year as a calendar year, in which February was not a leap year. Finally, in
extract 3.26, the interviewee is discussing who he thinks will win the Irish football
championship. He maintains that Tyrone have a brilliant chance of winning the All Ireland
this year. The championship year in Ireland runs from May to September.
The most frequent, and quite possibly the most complex, representation of time deixis
is tense. According to Levinson (1983: 77), ‘tense is one of the main factors ensuring that
nearly all sentences when uttered are deictically anchored to a context of utterance’ (cf.
Lyons, 1977). A brief examination of any extract from spoken corpora (or indeed written
ones) illustrates the prevalence of tense in conveying time deixis. Consider the following
extract 3.27, tense underlined:
(3.27) [Context: two family members discussing an acquaintance. Yoke is a slang term
in Irish English, meaning thing. It can also be used pejoratively to refer to a person,
as is the case here.]
A: Say that yoke was only ringing me to brag about last night.
B: Oh Paddy.
A: Um.
B: Did he ring you or texted you?
A: Rang me.
B: You should ring back for the laugh.
A: No I’m not wasting credit on him because he’ll say oh he just can’t live without
me and oh boy god Paddy I can live without you.
[Laughter]
As we can see, there are a range of tenses (for example, past simple, present simple,
past continuous, will future) used in the extract. However, most of these make reference
to either the present or around the immediate present time. This ‘around the immediate
54 DEIXIS
present’ time reference is established by the presence of the only one non-tense deictic 1
expression last night. Therefore, all the past tense time deictics, for example was only 2
ringing, texted, rang refer to the recent past. In addition, the will future he’ll say refers to 3
immediate future time. This extract demonstrates the importance of the tense system in 4
anchoring speakers in the ‘here–now– I’ deictic centre. 5
6
7
Discourse deixis 8
9
Similar to anaphora or cataphora, discourse deictics can be used to point to elements in 10
the preceding or following discourse. However, there are a number of differences between 1
discourse deictics and anaphoric or cataphoric reference (see Diessel, 1999: 100–3). As 2
we have seen, an anaphoric reference is used to ‘track’ a preceding noun phrase. Discourse 3
deictics, on the other hand, can be used to ‘focus the hearer’s attention on aspects of 411
meaning, expressed by a clause, a sentence, a paragraph or an entire idea’ (ibid: 101). In 5
extract 3.28, the discourse deictic this (in bold) is used by Speaker A while chatting to 6
one of her female friends to refer to an extended narrative, not a single noun phrase: 7
8
(3.28) 9
20
1
2
A: Jeanette’s house is the funniest house I’ve ever been in. Listen to this. I remember 3
when I was in first year of college Jeanette’s mother was having a surprise 4
eighteenth birthday party for her right. Jacinta and Mandy couldn’t go right. They’d 5
invited the three of us to go right. You must have been invited too but I don’t 6
know why you didn’t go. Anyway and typical me I thought I better go you know 7
and represent like. 8
B: Represent. 9
A: So I get the train from Cork to Dublin right. Jeanette’s mother is there right and 30
she’s such a chatterbox right and she loves my dad and she’s always saying tell 1
him to ring me. 2
[Laughter] 3
A: I arrive and I’m kinda nervous like cos I’ve never met all of her family before. I’ve 4
met like Linda and that was it and I get to the house anyway and I’m sitting down 5
anyway and they had made so much food like and I was like no I’d just eaten cos 6
I’d eaten on the train like but we had a few sausages and rashers anyway. We were 7
about to have a big dinner within a half an hour of me having these rashers and 8
sausages like. Oh I could have hung on like. Next thing she arrives over with eight 9
sausages on a plate and that was it. Eight sausages and six slices of bread. I was 40
just like ‘oh my God’. I was just like ah I have enough. Like on a plate. On a small 1
side plate with just eight sausages. All her brothers and sisters were sitting around 2
me like and asking me was I ok and the eight sausages like. 3
[Laughter] 4
5
Speaker A’s use of this in Listen to this anticipates the full story of her journey to a 6
friend’s eighteenth birthday party that follows. In addition, discourse deictics can be 47
DEIXIS 55
employed to refer to the underlying intention of the content of an utterance (also known
as illocutionary force; see Chapter 5). Extract 3.29 exemplifies the use of a demonstrative
that (in bold) with a focus on the illocutionary force of an utterance:
(3.29) [Context: mother and daughter chatting about an up-coming wedding]
A: Oh look at that the Bay View [hotel] over there. That’s the hotel there Anna where
Judy is having her reception.
B: Oh right yeah.
A: And if I don’t lose enough weight I won’t be there.
B: That’s a lie.
A: Well I must get dieting fast so won’t I and lose some of my eight stone.
[Laughter]
In extract 3.29, the interpretation of that is more complex than the tracking of a previous
noun phrase. The demonstrative that does refer to Speaker A’s assertion And if I don’t lose
enough weight I won’t be there. However, more specifically, that refers to the meaning
embodied by Speaker A’s assertion. Through saying And if I don’t lose enough weight I won’t
be there, Speaker A is not commenting on whether or not she will be at the wedding, but is,
in fact, claiming that Right now, I’m fat. Speaker B’s That’s a lie is a response to Speaker A’s
intended message and not the literal content of the utterance (also known as propositional
content). Fillmore (1997) also includes deictic items peculiar to written discourse such as
above or below in discourse deixis. In addition, Levinson (1983) specifies items such as
utterance initial but, therefore, in conclusion, well and however in discourse deixis. Furthermore,
Levinson (ibid: 89) contends that ‘the scope [of a proper theory of discourse deixis] . . . may
be very large, ranging from the borders of anaphora to issues of topic/comment structure’.
Social deixis
Social deixis refers to ‘those aspects of language structure that encode the social identities
of participants (properly, incumbents of participant-roles) or the social relationship between
them, or between one of them and persons or entities referred to’ (Levinson, 1983: 89).
For example, as already mentioned, many European languages contain the tu/vous
distinction. Hence, social deixis can contain information about the conversational parti-
cipants such as age, sex, kin relationship, social class or ethnic group. For this reason,
terms of address (see Chapter 4) are often included as a common way of realising social
deixis (especially in English, which, in comparison to many other languages, has a relative
paucity of linguistic devices which convey social deixis). Extract 3.30 features the use of
a kinship term (in bold) by a father:
(3.30)
A: Hurry up baby son all the boys is finished their breakfast.

56 DEIXIS
This kinship term baby son encodes a range of social information about the conversational 1
participants. Obviously, it points towards the age (baby), sex (male) and kin relationship 2
(son) of the speaker being addressed by the father. However, less obvious is the ethnic 3
identity contained within these kinship terms. Blum-Kulka (1997) has demonstrated how 4
Israeli parents use nicknames in addressing their children, thus emphasising familial 5
involvement and interdependence. In contrast, she observes how Jewish American and 6
Israeli American parents avoid nicknames in order to ‘show deference to the child’s 7
individuality’. (p. 162). Similarly, in extract 3.30 the father is a member of the community 8
of Irish Travellers and in using child-specific kin titles such as boy(s), young fella, children, 9
lads, son, he downplays the value of autonomy evident in a full first name, the emphasis 10
is instead on belonging and interdependence. This is evidence of the close social networks 1
that exist within the Irish Traveller community. These kinship networks are based around 2
family and extended family and clearly indicate the presence of the family at the core of 3
the Traveller value system (see Clancy, 2010). (Irish Travellers are a nomadic Irish ethnic 411
group who comprise less than 1 per cent of the population. Although Irish in origin, they 5
maintain a separate identity, culture and history to the mainstream Irish population.) 6
7
8
Empathetic deixis 9
20
The notion of empathetic deixis was first posited by Lyons (1977), based on a speaker’s 1
choice of this rather than that, here rather than there or now rather than then, ‘when the 2
speaker is personally involved with the entity, situation or place to which he is referring or 3
is identifying himself with the attitude or viewpoint of the addressee’ (p. 677). Rühlemann 4
(2007: 192) maintains that empathetic deixis ‘seems to involve preference of deictics that 5
are characterised by being, literally or metaphorically, nearer to the deictic origo (here being 6
nearer than there, now being nearer than then etc.)’. Therefore, when a speaker makes a 7
choice of that rather than this, for example, he/she is signaling his/her emotional 8
relationship with the propositional content of the utterance. In the following extracts 3.31 9
and 3.32, the use of that (in bold) demonstrates the affective implications of the choice 30
of the demonstrative: 1
2
(3.31) [Context: family chatting] 3
4
5
6
A: Yeah exactly. She still has that bike of hers. 7
B: Does she? 8
A: Yeah. 9
40
(3.32) [Context: female friends chatting] 1
2
3
4
A: And did you see that dress she had on that night lately? It was to about here 5
and it had a big piece missing here and there. 6
B: I saw that yeah. 47
DEIXIS 57
C: And she’s not skinny enough to wear that.

A: Would she not use fake tan? You might as well go out snow white.
In extract 3.31, Speaker A mentions that bike and, in extract 3.32, Speaker A mentions
that dress. On both occasions, these marked uses of the more distal demonstrative that
seem to signal a sense of disapproval on the speaker’s part. Lakoff (1974) calls this use
of demonstratives emotional deixis. She claims that the emotional-deictic that, as in extracts
3.31 and 3.32, where the subject alluded to belongs to neither the speaker nor the
addressee, ‘appears to establish emotional solidarity between the two by implying that both
participants in the conversation share the same views toward the subject of the discussion’
(p. 352). Both Argaman (2007) and Rühlemann (2007) note that little study has been
devoted to empathetic deixis. Indeed, Rühlemann (ibid: 222) concludes that ‘empirical
research based on corpus data might potentially advance the already existing knowledge
on this intriguing type of deixis substantially’.
3.6 CONCLUSION
In order to illustrate the importance of studying the phenomenon of deixis in speech, a

frequency list of the top 20 words in the spoken corpus LCIE was generated and this is
presented in Table 3.3. As can be seen, the table contains the personal pronouns whose
main use is deictic (I, you) and those whose use is predominantly anaphoric (it, he, they).
In addition, the table features that and there both of which have the potential to be used
deictically (marked in bold).
Table 3.3 Word frequency counts for the 20 most

frequent words in LCIE
Rank order LCIE
1 the
2 I
3 and
4 you
5 to
6 it
7 a
8 that
9 of
10 yeah
11 in
12 was
13 is
14 like
15 know
16 he
17 on
18 they
19 have
20 there
2
Implicature
2.1. Classical Gricean theory 2.3. Conventional implicature 54

of conversational implicature 24
2.4. Summary 58
2.2. Two neo-Gricean pragmatic
theories of conversational
implicature 36
The notion of implicature (both conversational and conventional) was ori-

ginated by the Oxford philosopher H. P. Grice.1 The central ideas were
introduced by him in the William James lectures delivered at Harvard in
1967 and were partially collected and published in Grice (1989). In these
lectures, Grice presented a panorama of his thinking on meaning and
communication—what he called his ‘tottering steps’ (Grice 1989: 4) to-
wards a systematic, philosophically inspired pragmatic theory of language
use, which has since come to be known as Gricean pragmatic theory. Since
its inception, the Gricean paradigm has revolutionized pragmatic theoriz-
ing and to date remains one of the cornerstones of contemporary thinking
in linguistic pragmatics and the philosophy of language.
This chapter is organized as follows. Section 2.1 discusses the classical
Gricean theory of conversational implicature. In Section 2.2, I present
two neo-Gricean pragmatic theories of conversational implicature.
1
Though some proto-Gricean ideas can be traced back at least to the fourth-
century rhetoricians Servius and Donatus, later to be reiterated by the nineteenth-
century English philosophers John Stuart Mill and Augustus De Morgan (Horn
1988, 2004). See Chapman (2005) on the life and work of Grice.
24 2. IMPLICATURE
Finally, Section 2.3 examines the notion of conventional implicature,

comparing and contrasting it with conversational implicature.
2.1. Classical Gricean theory of conversational implicature
On a general Gricean account of meaning and communication, there are

two theories: a theory of meaningn[on]n[atural] and a theory of conversational
implicature. In his theory of meaningnn , Grice (1957, 1969, 1989) empha-
sized the conceptual relation between natural meaning in the external
world and non-natural, linguistic meaning of utterances.2 He developed a
reductive analysis of meaningnn in terms of the speaker’s intentions.
(2.1) Grice’s theory of meaningnn
S meansnn p by ‘uttering’ U to A if and only if S intends:
(i) A to think p,
(ii) A to recognize that S intends (i), and
(iii) A’s recognition of S’s intending (i) to be the primary reason for A
thinking p.
where S stands for the speaker, A for the audience, U for the utterance, and
p for proposition (see also Levinson 2000: 13).
DeWned thus, the essence of meaningnn is that it is communication which
is intended to be recognized as having been intended. In other words,
meaningnn or speaker-meaning is a matter of expressing and recognizing
intention (but see Gauker 2003 for a dissenting view). What is of theoretical
interest here is that as pointed out by Levinson (2000: 13), ‘meaningnn
2
According to Grice (1957, 1969, 1989), in the case of natural meaning, x means
that p entails p. This explains why (ii) is semantically anomalous.
Grice (1989: 213)

(i) Those spots meant measles.
(ii) *Those spots meant measles, but he hadn’t got measles.
By comparison, in the case of non-natural meaning or meaningnn , x means that p

does not entail p. Now contrast (iii) and (iv) with (i) and (ii) above.
(Grice 1989: 214)

(iii) Those three rings on the bell (of the bus) mean that the bus is full.
(iv) Those three rings on the bell (of the bus) mean that the bus is full.
But it isn’t in fact full—the conductor has made a mistake.
2.1. CLASSICAL GRICEAN THEORY OF CONVERSATIONAL IMPLICATURE 25
(or something of the sort) draws an outer boundary on the communica-

tional eVects that a theory of communication is responsible for’.
2.1.1. The co-operative principle and the maxims

of conversation
In his theory of conversational implicature, Grice (1961, 1975, 1978, 1989)

suggested that there is an underlying principle that determines the way in
which language is used with maximum eYciency and eVectively to achieve
rational interaction in communication. He called this overarching dictum
the co-operative principle and subdivided it into nine maxims of conversation
classiWed into four categories: Quality, Quantity, Relation and Manner. The
names of the four categories are taken from the German philosopher
Immanuel Kant (Grice 1989: 26). The co-operative principle and its com-
ponent maxims ensure that in an exchange of conversation, the right
amount of information is provided and that the interaction is conducted
in a truthful, relevant and perspicuous manner.
(2.2) Grice’s theory of conversational implicature
a. The co-operative principle
Make your conversational contribution such as is required, at the stage at
which it occurs, by the accepted purpose or direction of the talk exchange
in which you are engaged.
b. The maxims of conversation
Quality: Try to make your contribution one that is true.
(i) Do not say what you believe to be false.
(ii) Do not say that for which you lack adequate evidence.
Quantity:
(i) Make your contribution as informative as is required (for the
current purposes of the exchange).
(ii) Do not make your contribution more informative than is required.
Relation: Be relevant.
Manner: Be perspicuous.
(i) Avoid obscurity of expression.
(ii) Avoid ambiguity.
(iii) Be brief (avoid unnecessary prolixity).
(iv) Be orderly.
The Gricean co-operative principle and its attendant maxims in (2.2) can
be simpliWed in (2.3) (Huang 2000a: 206).
26 2. IMPLICATURE
(2.3) Grice’s theory of conversational implicature (simpliWed)

a. The co-operative principle
Be co-operative.
b. The maxims of conversation
Quality: Be truthful.
(i) Don’t say what is false.
(ii) Don’t say what lacks evidence.
Quantity:
(i) Don’t say less than is required.
(ii) Don’t say more than is required.
Relation: Be relevant.
Manner: Be perspicuous.
(i) Avoid obscurity.
(ii) Avoid ambiguity.
(iii) Be brief.
(iv) Be orderly.
2.1.2. Relationship between the speaker and the maxims
What can a speaker do with regard to the maxims? In the Wrst place, he or
she can straightforwardly observe the maxims. Second, he or she can
violate a maxim. For example, he or she may breach the Wrst submaxim
of Quality by telling a deliberate lie.3 Third, he or she can opt out of
a maxim. This can be demonstrated by a speaker’s use of hedges in
conversation.
(2.4) Opting out hedges in English
a. Quality:
As far as I know,
I’m not sure if this is true, but. . .
I may be wrong, but. . .
b. Quantity:
As you probably already know,
I can’t say any more,
I probably don’t need to say this, but. . .
3
There are, of course, circumstances under which the speaker is not expected to
follow the maxims. In a court of law, for example, witnesses are often not required
to volunteer information, thereby violating the maxim of Quantity.
c. Relation:
Oh, by the way,
I’m not sure if this is relevant, but. . .
I don’t want to change the subject, but. . .
d. Manner:
I’m not sure if this is clear, but. . .
I don’t know if this makes sense, but. . .
This may be a bit tedious, but. . .
There is thus evidence that speakers are not only aware of the
maxims, but they are trying to follow them. Fourthly, a speaker can
ostentatiously Xout or exploit a maxim, to be illustrated shortly (see also
Grice 1989: 30).
At this point, have a go at Exercises 1 and 2 on p. 59.
2.1.3. Conversational implicatureO versus conversational

implicatureF
Assuming that the co-operative principle and its associated maxims are
normally adhered to by both the speaker and the addressee in a conversa-
tional interaction, Grice suggested that a conversational implicature—
roughly, a set of non-logical inferences which contains conveyed messages
which are meant without being part of what is said in the strict sense4—can
arise from either strictly observing or ostentatiously Xouting the maxims.
Let us call conversational implicatures that are engendered by way of
directly observing the maxims conversational implicaturesO . As an illustra-
tion, consider (2.5)–(2.8) (I use ‘þ>’ to stand for ‘conversationally
implicates’).
(2.5) Quality:
Tim Berners-Lee invented the World Wide Web in 1989.
þ> The speaker believes that Tim Berners-Lee invented the World Wide
Web in 1989, and has adequate evidence that he did
4
See also Horn’s (2004: 3) more recent definition of conversational implicature
as ‘a component of speaker meaning that constitutes an aspect of what is meant in a
speaker’s utterance without being part of what is said’.
28 2. IMPLICATURE
(2.6) Quantity:
John has six credit cards.
þ> John has at most six credit cards
(2.7) Relation:
John: What’s the time?
Mary: The museum hasn’t opened yet.
þ> It’s at least before whenever the museum normally opens
(2.8) Manner:
John went to a McDonald’s and bought two hamburgers.
þ>John Wrst went to a McDonald’s and then bought two hamburgers5
Given the second submaxim of Quality, when a speaker makes an

assertion, he or she conversationally implies that he or she believes it,
hence the relevant conversational implicature in (2.5). The submaxim can
also account for ‘Moore’s paradox’, so called by Wittgenstein (1953: 190).
The paradox is concerned with the question why an utterance such as (2.9)
is pragmatically anomalous. The answer is straightforward: (2.9) is a
violation of the epistemic commitment of what a speaker asserts (Levinson
1983: 105, see also Grice 1989: 42, Austin 1962, Gazdar 1979, Atlas 1993,
Bach 2004).
(2.9) ?Tim Berners-Lee invented the World Wide Web in 1989,
but I don’t believe he did.
(2.6) exempliWes the fulWlment of the Wrst submaxim of Quantity. Since

the speaker has used a semantically weaker expression (i.e., the cardinal
number six), where a semantically stronger one of equal brevity (e.g., the
cardinal number seven) is available, he or she would contradict the
Wrst submaxim of Quantity if the semantically stronger expression held.
Consequently, he or she believes that the semantically stronger statement
does not hold. Furthermore, he or she has not done anything to stop the
addressee from thinking that he or she thinks so; therefore he or she con-
versationally implicates that John has at most six credit cards,6 although
5
Of course, (2.8) can also conversationally implicate (i) below.
(i) John went to a McDonald’s in order to buy two hamburgers

6
Fretheim (1992) argued that the ‘at most’ interpretation of a cardinal number
in Norwegian can be conversationally implicated only if the cardinal number is
contextually salient. This is signalled by certain intonational patterns of the lan-
guage. Otherwise, the meaning of the cardinal number is ‘exactly n’. One anonym-
ous referee of this book believes that the same holds for English.
truth-conditionally, the utterance in (2.6) is not incompatible with the

possibility that he in fact has, for example, seven credit cards.
The conversational implicature arising in (2.7) results from the observa-
tion of the maxim of Relation. If this maxim is to be satisWed, Mary’s
utterance has to be taken as relevant. Since John has asked a question,
Mary should be providing an answer. Assuming that in saying what she has
uttered, Mary is co-operatively answering John’s question, we can infer
that while Mary is not in a position to provide a straightforward answer,
nevertheless she thinks that the museum’s not being open yet might help
John to get a partial answer, such as the one indicated above.
Finally, the conversational implicature in (2.8) is derived from following the
fourth submaxim of Manner. By this submaxim, the speaker is expected to
arrange the events in the order in which they took place, and the addressee is
expected to draw inferences in such a way. (2.8) is also an instance of iconicity:
the ordering of the linguistic elements iconically reXects that of the events.
Secondly and more interestingly, conversational implicatures can be
generated by way of the speaker’s deliberately Xouting the maxims. Let us
call conversational implicatures thus induced conversational implicaturesF .
As mentioned above, the maxims of conversation, as proposed by Grice,
may be overtly and blatantly breached. Faced with such a conspicuous
Xouting by the speaker, the addressee then has two options. One is to think
that the co-operative principle has been abandoned as well. But he or she
may—and characteristically does—choose a second option. He or she may
assume that despite the speaker’s apparent failure of co-operation, he or
she is still observing the co-operative principle, and reasons roughly thus. If
the speaker is still co-operative, and if he or she is exploiting a maxim in
such a way that I should recognize the infringement, then he or she is doing
so in order to convey some extra message, which is in keeping with the
co-operative principle at some deeper level. Moreover, he or she knows that
I am able to compute what that message is. A few examples will suYce to
illustrate conversational implicaturesF .
(2.10) Quality:
Chomsky is a great sociolinguist.
þ> Chomsky is no sociolinguist at all
(2.11) Quantity:
War is war.
þ> e.g., Terrible things always happen in war. That’s its nature, and it’s no
use lamenting that particular tragedy
30 2. IMPLICATURE
(2.12) Relation:
John: Susan can be such a cow sometimes!
Mary: Oh, what a lovely day today!
þ> e.g., One shouldn’t speak ill of people behind their back
(2.13) Manner:
a. John smiled.
b. The corners of John’s lips turned slightly upward.
þ> John did not exactly smile
Any student of linguistics knows that (2.10) is patently false, thereby

clearly and openly Xouting the maxim of Quality. However, in order to
preserve the assumption of co-operation, the addressee must assume that
the speaker is trying to convey something rather diVerent from what he or
she has actually said, that is, the ironic reading which is opposite to the
literal meaning of the sentence. A similar analysis can be made for other
types of ‘Wgure of speech’ such as metaphor, litotes, and hyperbole (but see
Sperber and Wilson 1986 for a dissenting view).
(2.11) is a tautology, and as such, is superWcially uninformative. Con-
fronted with this blatant breaking of the maxim of Quantity, the addressee
assumes that the speaker is actually co-operative and has to work out why
he or she has made such an apparently uninformative utterance. The only
way to do this is to interpret it as in fact highly informative. This explains the
creating of the conversational implicature along the lines I have indicated.7
7
According to Wierzbicka (1987, 1991), (2.11) cannot be used straightforwardly
in French, German, Russian, Polish, Japanese, and Korean. On the basis of this
observation, she claimed that tautological constructions like (2.11) are language-
specific, and therefore cannot be universally accounted for in terms of the Gricean
maxim of Quantity. But one of the anonymous referees has pointed out to me that
tautological examples of the type illustrated by (2.11) are in fact possible in French,
as in (i) and (ii).
(i) La guerre, c’est la guerre.

‘War is war.’
(ii) À la guerre comme à la guerre.
‘War is war.’
Another anonymous referee has noted that the same holds for Russian. He or she
has provided the following example, taken from a Russian grammar book.
(iii) delo delom a ljubov’ ljubov’ju.

‘Business is business and love is love.’
See also Fraser (1988) and Ward and Hirschberg (1991).

What does Mary’s response have to do with John’s bad-mouthing Susan

in (2.12), since it initially appears to infringe the maxim of Relation? Clearly,
if the assumption that Mary is still co-operative is to be maintained, we have
to interpret her response as highly relevant at some non-superWcial level.
One of the possible ways to do so is to read it as conversationally implicating
Mary’s disapproval of John’s bad-mouthing people behind their backs.
The use of the prolix sentence (2.13b), where the simple sentence (2.13a)
is available and would normally have been used, deviates sharply from the
third submaxim of Manner. But on the assumption that the speaker is
co-operative, the conversational implicature is essentially that the speaker
is not in a position to use the term ‘smile’ here, or that it is less than smile
that he or she wants to communicate.
We thus have the Wrst Gricean dichotomy between conversational
implicatureO and conversational implicatureF , namely, the distinction
between those conversational implicatures that are generated from a simple
assumption that the speaker is observing both the maxims and the
co-operative principles, and those that are engendered in more complex
ways on the basis of the speaker Xouting a maxim but nevertheless follow-
ing the co-operative principle. Grice’s major achievement here was in
providing a uniWed analysis of both types.
See whether you can tackle Exercises 3, 4, and 5 on pp. 59–60 now.
2.1.4. Generalized versus particularized conversational

implicature
A second Gricean dichotomy, independent of the Wrst, is between those

conversational implicatures which arise without requiring any particular
contextual conditions and those which do require such conditions. Grice
called the Wrst kind generalized conversational implicatures, as in (2.14), and
the second kind particularized conversational implicatures, as in (2.15).
(2.14) Most of John’s friends believe in marriage.
þ> Not all of John’s friends believe in marriage
(2.15) John: Where’s Peter?
Mary: The light in his oYce is on.
þ> Peter is in his oYce
32 2. IMPLICATURE
The implicature in (2.14) has a very general currency. Any utterance of the
form ‘Most x are Y’ will have the default interpretation ‘Not all x are Y’.
This interpretation will go through without needing any particular contexts
(see Ariel 2004 for further, detailed discussion of most). By contrast, the
implicature in (2.15) depends crucially on its linguistic context. Mary’s
reply points to the possible connection between the light in Peter’s oYce
and his location, namely, if the light in Peter’s oYce is on, he may be in his
oYce. Without such a speciWc context, we will not have the implicature
under consideration in (2.15). The theoretical importance of this Gricean
dichotomy has recently been subject to heated debates. Hirschberg (1991:
42–4), Welker (1994: 21–3) and Carston (2002), for example, doubted
whether such a distinction can be maintained. On the other hand, Levinson
(2000) put forward a rigorous defence of it (see also Grice 1975, 1981, 1989:
37–8 for further discussion).
Take a look at Exercise 6 on p. 60.
2.1.5. Properties of conversational implicature
Conversational implicatures are characterized by a number of distinctive

properties (Grice 1975, 1989, Sadock 1978, Levinson, 1983, 2000, Huang
1991, 1994: 4–5, 2000a: 206–7, 2003). In the Wrst place, there is defeasibility
or cancellability: conversational implicatures can simply disappear in cer-
tain linguistic or non-linguistic contexts. How? They are cancelled if they
are inconsistent with (i) semantic entailments, (ii) background assump-
tions, (iii) contexts, and/or (iv) priority conversational implicatures.8 Let
me take them one by one. (I shall postpone the discussion of (iv) until
Section 2.2.3.)
First, conversational implicatures evaporate in the face of inconsistency
with semantic entailments, as the utterances in (2.17) show. (I use ‘þ>’ to
signify ‘does not conversationally implicate’.)
(2.16) His wife is often complaining.
þ> His wife is not always complaining
8
In addition, Q-implicatures can also be cancelled by metalinguistic negation. I
shall discuss this in Section 2.2.2.
(2.17) a. His wife is often, in fact/indeed always, complaining.

b. His wife is often, and perhaps/maybe/possibly/even always, complaining.
c. His wife is not only often but always complaining.
d. His wife is often, or perhaps/maybe/possibly/even always, complaining.
e. His wife is often, if not always, complaining.
þ> His wife is not always complaining
All the utterances in (2.17) have the potential conversational implicature

indicated in (2.16). However, all the sentences in (2.17) bear the semantic
entailment that his wife is always complaining due to the use of phrases
such as in fact always. Consequently, the potential conversational implica-
ture is defeated by the inconsistent entailment.
Next, conversational implicatures are suspended if they are not in keep-
ing with background or ontological assumptions, often referred to as real-
world knowledge. This is the case with (2.19).
(2.18) John and Mary bought an apartment near the Louvre in Paris.
þ> John and Mary bought an apartment near the Louvre in Paris together,
not one each
(2.19) The Americans and the Russians tested an atom bomb in 1962.
þ> The Americans and the Russians tested an atom bomb in 1962
together, not one each
Given our knowledge about history, it was impossible for the USA and the
USSR to test an atom bomb together in 1962, because they were enemies at
that time, thus the disappearance of the potential ‘togetherness’ conversa-
tional implicature.
Third, conversational implicatures are annulled when they run contrary
to what the immediate linguistic context of utterance tells us. Imagine the
following exchange in a music shop.
(2.20) John: This CD is eight euros, and I haven’t got any money on me.
Mary: Don’t worry, I’ve got eight euros.
þ> Mary has got only eight euros
Here, given the immediate linguistic context of utterance, Mary’s response

does not produce the usual conversational implicature that she has got only
eight euros. This is because all the information needed here is whether or
not Mary has enough money for John to buy the CD rather than the exact
amount of money she might in fact have.9
9
Note that the following joke is based on defeasibility of conversational impli-
catures (Leech 1983: 91).
34 2. IMPLICATURE
A second property exhibited by conversational implicatures is non-

detachability: any linguistic expression with the same semantic content
tends to carry the same conversational implicature. (A principled exception
is those conversational implicatures that arise via the maxim of Manner,
about which later.) This is because conversational implicatures are attached
to the semantic content, rather than the linguistic form, of what is said.
Therefore, they cannot be detached from an utterance simply by replacing
the relevant linguistic expressions with their synonyms. This is illustrated in
(2.21), which indicates that the use of any linguistic expression that is
synonymous with almost will trigger the same conversational implicature.
(2.21) The Wlm almost/nearly won/came close to winning an Oscar.
þ> The Wlm did not quite win an Oscar
The third property, calculability, means that conversational implicatures

can transparently be derived via the co-operative principle and its compon-
ent maxims. The fourth, non-conventionality, means that conversational
implicatures, though dependent on the saying of what is coded, are non-
coded in nature (Grice 1989: 39, Bach 1994a: 140). In other words, they rely
on the saying of what is said but they are not part of what is said. They are
associated with speaker or utterance but not proposition or sentence.
Fifthly, according to the principle of reinforceability, conversational impli-
catures can be made explicit without producing too much of a sense of
redundancy. This is because conversational implicatures are not part of the
conventional import of an utterance. For example, the conversational
implicature in (2.22) is made explicit in (2.23). But (2.23) is not felt to be
semantically redundant.
(2.22) The soup is warm.
þ> The soup is not hot
(2.23) The soup is warm, but not hot.
Finally, we have universality: conversational implicatures tend to be uni-

versal, being motivated rather than arbitrary. Examples (2.24)–(2.28) from
diVerent languages all assert that some young people like pop music, and
conversationally imply that not all young people like pop music. Example
Steven: Wilfrid is meeting a woman for dinner tonight.

Susan: Does his wife know about it?
Steven: Of course she does. The woman he is meeting is his wife.
(2.29) from Malagasy asserts that some young people like famous songs, and
conversationally implicates that not all young people like famous songs. All
this shows that if a language has ‘all’ and ‘some’, the use of the semantically
weaker ‘some’ will universally carry the conversational implicature ‘not all’.
(2.24) (Arabic)
Ba’ag al-Shababal-yaWan al-musiqa al-Harkia.
some young people like-PL the-music the-pop
‘Some young people like pop music.’
(2.25) (Catalan)
A alguns joves els agrada la música pop.
to some young people to them like the music pop
(2.26) (Chinese)
yixie nianqing ren xihuan liuxing yinyue.
some young people like pop music
(2.27) (Modern Greek)
Se merikus neus aresi i pop musiki.
to some young people is pleasing the pop music
(2.28) (Kashmiri)
kanh-kanh noujawan chu pop musiki pasabd karan.
some young people be-3SG pop music like
(2.29) (Malagasy)
Misy tanora tia ny hira malaza.
exist young like the song famous
‘Some young people like famous songs.’10, 11
10
The best-known apparent counterexample to the universality claim for
conversational implicatures is provided by Keenan (1976) (see also Sarangi and
Slembrouck 1992, Rundquist 1992, Haviland 1997, Marmaridou 2000: 241). In this
work, Keenan claimed that the Malagasy-speaking culture of Madagascar is a
speech community in which Grice’s co-operative principle, and in particular his
first submaxim of Quantity, are not adhered to; e.g., in talking to her son, a
Malagasy mother would use the following sentence to refer to her husband/the
boy’s father:
Mbola mator y ve ny olana?

‘Is the person still asleep?’
þ> the person in question is not the speaker’s husband/the boy’s father
54 2. IMPLICATURE
considered to be an ordered set of background assumptions, contextual

factors, semantic entailments, conversational implicatures, and so on and
so forth. Each incrementation of the informational content of an utterance
must be consistent with the informational content that already exists,
otherwise it will be cancelled according to the following hierarchy (adapted
from Gazdar 1979, Huang 1991, 1994, 2000a, 2004b, Levinson 2000).
(2.89) The implicature cancellation procedure
a. Background assumptions
b. Contextual factors
c. Semantic entailments
d. Conversational implicatures
(i) Q-implicatures
(1) Q-clausal implicatures
(2) Q-scalar implicatures
(ii) M-implicatures
(iii) I-implicatures
2.3. Conventional implicature
2.3.1. What is conventional implicature?
In the last two sections, I have surveyed the classical and neo-Gricean
pragmatic theories of conversational implicature. In this section, I shall
brieXy discuss the second category of implicature put forward by Grice,
namely, conventional implicature (see also Frege 1892, who, according to
Bach 1999b, was perhaps the Wrst modern philosopher to notice the phe-
nomenon). A conventional implicature is a non-truth-conditional inference
which is not deductive in any general, natural way from the saying of what
is said, but arises solely because of the conventional features attached to
particular lexical items and/or linguistic constructions.
A few standard examples follow (I use ‘þ>>’ to stand for ‘convention-
ally implicates’.)
(2.90) p therefore q þ>> q follows from p
He is a Chinese; he therefore knows how to use chopsticks.
(2.91) p but q þ>> p contrasts with q
a. John is poor but he is honest.
b. Our sales have gone up but theirs have gone down.
2.3. CONVENTIONAL IMPLICATURE 55
(2.92) Even p þ>> contrary to expectation

Even his wife didn’t think that John would win the by-election.
(2.93) p moreover q þ>> q is in addition to p
Xiaoming can read German. Moreover, he can write poems in the language.
(2.94) p so q þ>> p provides an explanation for q
Mary is taking Chinese cookery lessons. So her husband has bought her a
wok.
In (2.90), the conventional implicature triggered by the use of therefore is

that being Chinese provides some good reason for knowing how to use
chopsticks. In (2.91), there is a conventional implicature of contrast
between the information contained in p and that contained in q16 (Grice
1989: 25, 88). In (2.92), even, being epistemic in nature, conventionally
implicates some sort of unexpectedness, surprise or unlikeness (e.g., Kemp-
son 1975, Karttunen and Peters 1979, Kay 1990, Barker 1991, Lycan 1991,
Farncescotti 1995). In (2.93), the use of moreover brings in the conventional
implicature that the statement made in q is additional to the statement
made in p (Grice 1989: 121). Finally in (2.94), the conventional implicature
contributed by so is that the fact that Mary is learning how to cook Chinese
food explains why her husband has bought her a wok.
Other representative lexical items that are considered to engender con-
ventional implicatures in English include actually, also, anyway, barely,
besides, however, manage to, on the other hand, only, still, though, too and yet.
See whether you can tackle Exercise 14 on p. 62.
2.3.2. Properties of conventional implicature
Properties of conventional implicatures can best be characterized in con-

trast to those of conversational implicatures, discussed above (Grice 1975,
Levinson 1983: 127–8, Horn 1988).
16
It should be pointed out here that but seems to have three uses: (i) denial of
expectation, as in (2.91a); (ii) contrastive, as in (2.91b), and (iii) correction, as in (i)
below.
(i) That’s not my father but my uncle.

56 2. IMPLICATURE
The main similarity between conventional and conversational implica-

ture is that neither makes any contribution to truth conditions (but see
Chapter 7). Recall (1.32) in Chapter 1, repeated here as (2.95) for ease of
exposition.
(2.95) a. We want peace and they want war.
b. We want peace but they want war.
As mentioned in Section 1.3.3. of Chapter 1, (2.95b) shares the same truth

conditions with (2.95a), though it contains the conventional implicature of
contrast triggered by the use of the connective but. This indicates that like a
conversational implicature, a conventional implicature does not contribute
to the truth condition of its corresponding sentence. A second similarity is
that both conventional and conversational implicatures are associated with
speaker or utterance rather than proposition or sentence.
On the other hand, there are a number of important diVerences between
conventional and conversational implicatures. First of all, conventional
implicatures are not derived from the co-operative principle and its com-
ponent maxims, but are attached by convention to particular lexical items
or linguistic constructions. They are therefore an arbitrary part of meaning,
and must be learned ad hoc. By contrast, conversational implicatures are
derived from the co-operative principle and its attendant maxims. Hence,
they are non-conventional by deWnition, that is, they are motivated rather
than arbitrary.
The same can be said of mais in French (e.g., Anscombre and Ducrot 1977). On
Horn’s (1989) view, the distinction is mainly one of semantic ambiguity. This
analysis can be supported by the fact that there are languages in which the different
uses of but are lexicalized. For example, German uses aber for the denial of
expectation/contrastive but and sondern for the correction but. The two buts in
Finnish, Spanish, and Swedish differ in essentially the same way. In Finnish, the
denial of expectation/contrastive but is translated as mutta and the correction but as
vaan. Spanish utilizes pero for the former and sino for the latter. Swedish has men
and utan, respectively (Horn 1989: 406–9). Finally, there are also two buts in
Russian: no and a, the latter serving the function of marking discontinuity in
discourse. Furthermore, as pointed by Anscombre and Ducrot (1977) for mais in
French, the denial of expectation/contrastive but and the correction but have
syntactic complementary distribution (see also Horn 1989: 407). If the analysis is
correct, then the correction but is not expected to be accounted for in terms of
conventional implicature. But see Blakemore (2000) for a dissenting view.
2.3. CONVENTIONAL IMPLICATURE 57
Second, conventional implicatures are not calculable via any natural

procedure, but are rather given by convention, thus they must be stipu-
lated. By comparison, conversational implicatures are calculable using
pragmatic principles, contextual knowledge, and background assumptions.
Third, conventional implicatures are not cancellable, that is, they cannot
be defeated. By contrast, conversational implicatures are cancellable.
Fourth, conventional implicatures are detachable, because they depend on
the particular linguistic items used. By comparison, conversational impli-
catures (except those arising from the M-principle) are non-detachable,
because they are attached to the semantic content, but not to the linguistic
form of what is said. Fifth, conventional implicatures tend not to be
universal. By contrast, conversational implicatures tend to be universal
(see also Potts 2005, Feng 2006).
It should be pointed out before we leave this section that unlike the notion
of conversational implicature, the notion of conventional implicature does
not seem to be a very coherent one. Even Grice himself (1989: 46) warned that
‘the nature of conventional implicature needs to be examined before any free
use of it, for explanatory purposes, can be indulged in’ (see also Horn 2004).
Since its inception, conventional implicature has been subject to numerous
attempts to reduce it to semantic entailment, conversational implicature, and
presupposition (Levinson 1983: 128), and more recently, to part of what is
said (Bach 1999b, but see Barker 2003 for a diVerent view), part of tacit
performatives (Rieber 1997), vehicles for performing second-order speech
acts (Bach 1999b), and procedural meaning in relevance theory (Blakemore
2002, 2004). But see Feng (2006) for a defence of the concept.
By way of summary, meaningnn /speaker-meaning or the total signiWca-
tion of an utterance in a Gricean system may be represented schematically
as follows (Levinson 1983: 131, 2000: 13, Huang 1994, but see Davis 1998
for a recent critique of Gricean pragmatics).
(2.96)
meaningnn
what is said what is implicated
conventionally conversationally
generalized particularized
1
Grice’s Maxims
2
3
3
Presupposition
3.1. What is presupposition? 65 3.3. Analyses 75
3.2. Properties of presupposition 67 3.4. Summary 90
Presupposition is another pragmatic topic that originates within the trad-

ition of the philosophy of language. The German mathematician and
logician Gottlob Frege is generally recognized as the Wrst scholar in modern
times who (re)introduced the philosophical study of presupposition
(see especially Frege 1892), though the notion of presupposition may go
back at least as far as the medieval philosopher Petrus Hispanus (Horn 1996).
Since its (re)introduction, presupposition has attracted an ever-growing
amount of attention from both philosophers and linguists. In the philoso-
phy of language, the study of presupposition has largely been conWned to
debates about the nature of reference and (singular) referring expressions
(see, e.g., the classic works by Frege 1892, Russell 1905, and Strawson
1950).1 In linguistics, on the other hand, the investigation of presupposition
is concerned with a much wider range of phenomena, centring around the
general debates about the interaction and division of labour between
semantics and pragmatics.
1
For discussion about the philosophical background of presupposition, see,
e.g., Levinson (1983), Soames (1989), Horn (1996), and Atlas (2004, 2005).
3.1. WHAT IS PRESUPPOSITION? 65
The organization of this chapter is as follows. Section 3.1 discusses the

general phenomena of presupposition. Next, Section 3.2 examines the
properties of presupposition, covering constancy under negation in Section
3.2.1, defeasibility in Section 3.2.2, and the projection problem in Section
3.2.3. Finally, Section 3.3 presents an overview of a number of recent/
current theories of presupposition.
3.1. What is presupposition?
Presupposition can be informally deWned as an inference or proposition

whose truth is taken for granted in the utterance of a sentence. Its main
function is to act as a precondition of some sort for the appropriate use of
that sentence. This background assumption will remain in force when the
sentence that contains it is negated. Presupposition is usually generated by
the use of particular lexical items and/or linguistic constructions. Lexical
items and linguistic constructions that engender presuppositions are called
presupposition triggers. While presupposition has in general been regarded
as ‘a heterogeneous collection of quite distinct and diVerent phenomena’
(Levinson 1983: 217) since Karttunen (1973, 1974) and Karttunen and
Peters (1979), representative examples can be given, as in the list below
(see, e.g., Levinson 1983: 181–5, Soames 1989, Beaver 1997, 2001: 10–12,
Geurts 1999: 2, Atlas 2004, 2005: 119). (I use the symbol ‘>>’ to stand for
‘presuppose’. The positive and negative versions of the examples are sep-
arated by /, and the lexical presupposition triggers are here italicized.)
(3.1) DeWnite descriptions
The king of France is/isn’t bald.
>> There is a king of France
(3.2) Factive predicates
a. Epistemic or cognitive factives
John knows/doesn’t know that Baird invented television.
>> Baird invented television
b. Emotive factives
John regrets/doesn’t regret that he has said the unsayable.
>> John has said the unsayable
(3.3) Aspectual/change of state predicates
Mary has/hasn’t stopped beating her boyfriend.
>> Mary has been beating her boyfriend
66 3. PRESUPPOSITION
(3.4) Iteratives
a. Iterative verbs
John returned/didn’t return to Cambridge.
>> John was in Cambridge before
b. Iterative adverbs
The boy cried/didn’t cry wolf again.
>> The boy cried wolf before
(3.5) Implicative predicates
John managed/didn’t manage to give up smoking.
>> John tried to give up smoking
(3.6) Temporal clauses
After she shot to stardom in a romance Wlm, Jane married/didn’t marry
a millionaire entrepreneur.
>> Jane shot to stardom in a romance Wlm
(3.7) Cleft sentences
a. Cleft
It was/wasn’t Baird who invented television.
>> Someone invented television
b. Pseudo-cleft
What Baird invented/didn’t invent was television
>> Baird invented something
(3.8) Counterfactual conditionals
If an ant were as big as a human being, it could/couldn’t run Wve times faster
than an Olympic sprinter.
>> An ant is not as big as a human being
In (3.1) the presupposition trigger is the deWnite description and the pre-
supposition is called existential presupposition. The class of deWnite descrip-
tions may include proper names, possessives, and certain wh-phrases.
In (3.2) the indicators of the presuppositions are the factive verbs, and
both presuppositions are known as factive presuppositions. Factive presup-
positions can further be divided into two subtypes: those triggered by
the use of cognitive or epistemic factives (which concern knowledge of
fact) and those produced by emotional factives (which are concerned
with emotional attitude towards fact). Occasionally, factive presupposi-
tions may arise from the use of factive NPs such as the fact/knowledge that.
The presupposition in (3.3) is introduced by the aspectual/change-of-state
verb. The presupposition triggers in (3.4) are the iteratives, and the pre-
supposition trigger in (3.5) is the implicative verb. Notice that the presup-
position triggers in (3.1)–(3.5) are lexical triggers. By contrast, those in
(3.6)–(3.8) are constructional/structural triggers. The presupposition
3.2. PROPERTIES OF PRESUPPOSITION 67
in (3.6) is triggered by a temporal clause, the presuppositions in (3.7)

by cleft sentences, and the presupposition in (3.8) by a counterfactual
conditional.2
Have a look at Exercises 1 and 2 on p. 91.
3.2. Properties of presupposition
Presuppositions exhibit a number of distinctive properties, notably:

(i) constancy under negation and (ii) defeasibility or cancellability. In add-
ition, certain cases of defeasibility give rise to what is known as the
projection problem.
3.2.1. Constancy under negation
By constancy under negation is meant that a presupposition generated by

the use of a lexical item or a syntactic structure remains the same when the
sentence containing that lexical item or syntactic structure is negated.3
Using constancy under negation as a diagnostic, presuppositions can be
more formally deWned in (3.9) (see Beaver 1997, 2001 for a number of non-
theory-neutral formal deWnitions).
2
One interesting point to note is that when a presupposition is queried, the
relevant lexical presupposition trigger is frequently queried in an explicit way. Some
examples follow.
(i) a. The king of France is bald.

b. The King of France? But is France not a republic?
(ii) a. The boy cried wolf again.
b. What do you mean by ‘again’? The boy didn’t cry wolf before.
(iii) a. John regrets that he has said the unsayable.
b. Did you use the word ‘regret’? John has never said the unsayable.
3
In addition, presuppositions are also preserved under questions, modals, and
conditionals. See, e.g., Soames (1989), Chierchia and McConnell-Ginet (2000), and
Atlas (2004, 2005) for exempliWcation and discussion.
(3.9) An utterance of a sentence S presupposes a proposition p if and only if

a. if S is true, then p is true
b. if S is false, then p is still true
What (3.9) basically says is this: for S to presuppose p, whenever S is true, p

is also true, and whenever S is false, p is still true.
There are, however, problems at the very heart of the notion of con-
stancy under negation. On the one hand, constancy under negation may
not be necessary. For example, there is a class of sentences which are hard if
not impossible to negate, yet they bear presuppositions, as in (3.10) (Green
1996). On the other hand, constancy under negation may not be suYcient.
This is illustrated by (3.11) and (3.12). Although (3.11) and (3.12) satisfy
the condition of constancy under negation, the inference in the former is
standardly analysed as a felicity condition on the speech act of requesting
(about which more in Chapter 4), and that in the latter as a conventional
implicature (e.g., Levinson 1983: 185).
(3.10) Long live the king of France!
(3.11) Do/don’t bring the digital camera here.
?>> The digital camera is not here
(3.12) (Chinese)
nin yongyuan shi wode laoshi.
you-HON always be my teacher
‘You will always be my teacher.’
?>> The addressee is socially superior to or distant from the speaker
Now see whether you can tackle Exercise 3 on pp. 91–2.
3.2.2. Defeasibility
Like conversational implicatures, but unlike semantic entailments, presup-

positions are cancellable. They are nulliWed if they are inconsistent with (i)
background assumptions, (ii) conversational implicatures, and (iii) certain
discourse contexts. Furthermore, they can also drop out in certain intra-
sentential contexts, some of which give rise to the projection problem of
presupposition. Defeasibility has in general been taken as the second most
important property of presupposition.
In the Wrst place, presuppositions can disappear in the face of inconsist-

ency with background assumptions or real-world knowledge. Contrast
(3.13) and (3.14). (I use ‘>>’ to stand for ‘does not presuppose’.)
(3.13) John got an assistant professorship before he Wnished his Ph.D.
>> John Wnished his Ph.D.
(3.14) John died before he Wnished his Ph.D.
While (3.13) presupposes that John Wnished his Ph.D. by virtue of the
temporal clause, (3.14) does not carry that presupposition. This is because
the putative presupposition conXicts with our real-world knowledge that
one cannot do anything after one dies. Consequently, the unwanted pre-
supposition vanishes.4
Secondly, presuppositions can be cancelled by inconsistent conversa-
tional implicatures. This is illustrated in (3.15).
(3.15) If John is organizing a stag night, Mary will be angry that he is doing so.
þ> perhaps John is organising a stag night, perhaps he isn’t
>> John is organizing a stag night
4
Compare (3.14) with (i) below.
(i) John died before he reached the hospital’s accident and emergency depart-
ment.
>> John reached the hospital’s accident and emergency department
(i) has the presupposition indicated above only on the understanding that he means
John’s body.
More importantly, David Cram has pointed out to me that for some native
speakers of English including himself, the presupposition of (3.14) cannot be
cancelled. For this group of speakers, only the non-Wnite alternative to (3.14) allows
the presupposition to be defeated. This is given in (ii).
(ii) John died before Wnishing his Ph.D.

Furthermore, David Cram has oVered a neo-Gricean pragmatic explanation

along the following lines: if the speaker had intended the presupposition to be
suspended, he or she would have used (ii). If (ii) is not used, but (3.14) is employed
instead, then an inference can be drawn, namely, the speaker does not intend the
presupposition to be cancelled.
The use of the factive predicate angry in (3.15) should give rise to the
potential presupposition that John is organizing a stag night. However,
there is also a Q-clausal conversational implicature due to the use of the
conditional in (3.15), namely, perhaps John is not organizing a stag night
(see Chapter 2). In the face of the contradictory conversational implicature,
the putative presupposition here is defeated.
Thirdly, presuppositions are contextually cancellable, that is, they can
evaporate if they run contrary to what the immediate discourse context tells
us. Consider Wrst (3.16), where the putative presupposition is defeated by
the inconsistent proposition that is already established in the immediate
discourse background.
(3.16) There is no king of France. Therefore the king of France isn’t bald.
In (3.16) the second sentence should presuppose that there is a king of

France. The reason it does not is because such a putative presupposition is
inconsistent with the preceding proposition, already established in the
immediate discourse background, namely, that there is no king of France.
As a result, the unwanted presupposition fails to survive.5
Next, presuppositions can be suspended by so-called reduction argu-
ments—arguments that proceed by eliminating each of the possibilities in
a discourse. This is exempliWed by (3.17), cited from Chierchia and
McConnell-Ginet (2000).
(3.17) A, noticing the open door: Was it you who opened the door to the porch? I
closed it at lunch time.
5
Contrast (i) and (ii).
(i) John doesn’t/You don’t know that Baird invented television.

(ii) I don’t know that Baird invented television
Here, there is a contrast between the use of a third/second-person subject and that
of a Wrst-person subject, in a negative sentence containing the factive verb know:
while the presupposition in the former gets through, that in the latter (sometimes)
drops out. This is because what is presupposed in (ii) is precisely what is denied by
what is explicitly said about the speaker’s knowledge or beliefs (see Gazdar 1979:
142, 153–4 for an analysis of how the presupposition in (ii) but not (i) is correctly
cancelled).
B: Well, it wasn’t me who opened it, because I’ve been gone all afternoon,
and it wasn’t Joan who opened it, because she was with me, and it wasn’t
any of the kids who opened it, because they’re on a Weld trip, and I strongly
suspect that nobody opened it—there was a lot of wind this afternoon, you
know
>> Someone opened the door to the porch
In this passage, each of the cleft sentences should bear the presupposition
that someone opened the door to the porch. However, as pointed out by
Levinson (1983: 189) for a similar example, the whole purpose of B’s
utterance in (3.17) is to convince the addressee that no one opened it.
This has the consequence that the presupposition is overridden; ‘it was
adopted as a counterfactual assumption to argue [for] the untenability of
such an assumption’ (ibid.).
Finally, presuppositions can disappear in a discourse where evidence for
their truth is being weighed and rejected, as in (3.18), due to Levinson
(1983: 189).
(3.18) A: Well we’ve simply got to Wnd out if Serge is a KGB inWltrator.
B: Who if anyone would know?
C: The only person who would know for sure is Alexis; I’ve talked to him
and he isn’t aware that Serge is on the KGB payroll. So I think Serge can be
trusted.
>> Serge is on the KGB payroll
C’s utterance he isn’t aware that Serge is on the KGB payroll contains the
epistemic factive predicate aware. Although the use of factive predicates
typically presupposes the truth of their complements, the use of it in this
example does not. This is because the presupposition clashes with the whole
point of C’s argument, namely, that since Alexis isn’t aware that Serge is on
the KGB payroll, he is not a KGB inWltrator. When this happens, the
unwanted presupposition is again blocked.
In the fourth place, presuppositions can be blocked in certain intrasen-
tential contexts, that is, they can be defeated by using other clauses in the
same complex sentence to increment the local, intrasentential context.
Three cases are of particular interest. The Wrst of these is that a presuppos-
ition of a sentence can be overtly denied in a co-ordinate clause without any
apparent contradiction. This is illustrated in (3.19)–(3.21).
(3.19) The president doesn’t regret vetoing the bill because in fact he never did so!
>> The president vetoed the bill
(3.20) John didn’t manage to open an e-savings account, in fact he didn’t even try!
>>John tried to open an e-savings account
(3.21) The king of France isn’t bald—there is no king of France!
Notice that in many cases, outright denial of presupposition is not

possible with positive sentences. Contrast (3.19)–(3.21) above with (3.22)–
(3.24) below.
(3.22) *The president regrets vetoing the bill because in fact he never did so!
(3.23) *John managed to open an e-savings account, in fact he didn’t even try!
(3.24) *The king of France is bald—there is no king of France!
There is thus, at least in these cases, an asymmetry between negative and

positive sentences with regard to defeasibility. This asymmetry has led to
an entailment analysis for positive sentences, namely, the argument that
what is allegedly presupposed in these sentences is actually what is entailed.
Since semantic entailments cannot be overtly denied without producing
semantic anomaly, the anomaly displayed in (3.22)–(3.24) is entirely
expected. A second point to note is that the negation involved in overt
denial of presupposition in (3.19)–(3.21) is generally considered as a meta-
linguistic negation (see Chapter 2). (See, e.g., Horn 1985, 1989 and Burton-
Roberts 1989 for further arguments, but see Carston 2002, Geurts 1998,
1999: 71–2 and Atlas 2004 for scepticism, though for diVerent reasons.)
Next, a presupposition of a sentence can be explicitly suspended in an if
clause that follows. Witness (3.25)–(3.27).6
(3.25) John clearly doesn’t regret being an anti-war campaigner, if he actually ever
was one.
>> John was an anti-war campaigner
(3.26) John didn’t tamper with their computers again, if indeed he ever did.
>> John tampered with their computers before
(3.27) I’m sure John’s wife is beautiful, if he has a wife.
>> John has a wife
Finally, presuppositions can disappear with certain verbs of saying such

as say, mention, tell, ask, and announce and certain verbs of propositional
attitude such as believe, think, imagine, dream, and want.
6
The explicit suspension of a presupposition in examples like (3.25)–(3.27) may
be regarded as a special case of presuppositional heritability in conditionals of this
sort.
(3.28) John said/mentioned/told Bill that Mary managed to speak with a broad
Irish accent.
>> Mary tried to speak with a broad Irish accent
(3.29) Mr Wang believed/dreamed/imagined that he is the emperor of China.
>> There is an emperor of China
As pointed out by Green (1996: 77–8), these are ‘world-creating’ verbs.

They can deWne worlds other than the real world, which has been taken for
granted as relevant for the evaluation of the pertinent presuppositions. The
use of these verbs can lead to the evaporation of presuppositions.
Examples (3.19)–(3.29) form part of what Langendoen and Savin (1971)
Wrst termed the projection problem of presupposition, namely, to state and
explain the presuppositions of complex sentences (as ‘wholes’) in terms of
the presuppositions of their component simple sentences (as ‘parts’), to
which we now turn.
See whether you are ready to do Exercise 4 on p. 92.
3.2.3. The projection problem
It is in general expected that the presuppositions of a simple sentence will

ascend to become the presuppositions of the complex sentence of which the
simple sentence is a part. Put in this way, the projection problem can be
seen as a special case of the Fregean principle of compositionality, which
dictates that the meaning of a compound expression is a function
(i.e. combination) of the meanings of its parts (see, e.g., Janssen 1997 for
an overview of compositionality). The projection problem manifests itself
in two opposite directions. On the one side of the projection coin, the
presuppositions of a component sentence may fail to be projected on to,
and hence inherited by, the whole complex sentence. On the other side, the
presuppositions of a component sentence may be preserved when that
constituent sentence becomes part of a more complex sentence.
Recollect now examples (3.19)–(3.21) and (3.25)–(3.29) above, which
point to three cases where presuppositions disappear by virtue of intrasen-
tential context. In (3.19)–(3.21), the presuppositions are overtly denied
by the adding of a co-ordinate clause, in (3.25)–(3.27) they are explicitly
suspended by the adding of an if clause, and Wnally in (3.28)–(3.29)
1
ENTAILMENT
2
3
4
1
POLITENESS
2
3
4
5
6
7
8
Leech's Politeness Principles1
Politeness concerns a relationship between self and other. In conversation, self is identified as the speaker
and other is the hearer. Beside that, the speaker also shows politeness to the third parties who may be present
or not. The politeness principle (PP) is introduced by Geoffrey Leech. PP is Minimizing (other things
being equal) the expression of impolite beliefs, and there is a corresponding positive version (maximizing
(other things being equal) the expression of polite beliefs) which is somewhat less important. PP proposes
how to produce and understand language based on politeness. The purpose of PP is to establish feeling of
community and social relationship. Thus, PP focuses on process of interpretation that the center of the study
is on the effect of the hearer rather than the speaker. There are six maxims of the politeness principle that are
used to explain relationship between sense and force in daily conversation, those are:
The Tact maxim

The tact maxim is minimizing cost to other and maximizing benefit to other. This maxim is applied in
Searle’s speech act, commissives and directives called by Leech as impositives. Commissives is found in
utterances that express speaker’s intention in the future action. Then, Directives/ impositives are expressions
that influence the hearer to do action. The example of the tact maxim is as follows:
“Won‘t you sit down?”
It is the directive/ impositive utterance. This utterance is spoken to ask the hearer sitting down. The speaker
uses indirect utterance to be more polite and minimizing cost to the hearer. This utterance implies that sitting
down is benefit to the hearer.
The Generosity Maxim

The generosity maxim states to minimizing benefit to self and maximizing cost to self. Like tact maxim, the
generosity maxim occurs in commissives and directives/ impositives. This maxim is centered to self, while
the tact maxim is to other. The example will be illustrated as follows:
“You must come and dinner with us.”
It is an advice utterance that is involved in directive illocutionary act. In this case the speaker implies that
cost of the utterance is to his self. Meanwhile, the utterance implies that benefit is for the hearer.
The Approbation Maxim

The approbation maxim requires to minimizing dispraise of other and maximizing praise of other. This
maxim instructs to avoid saying unpleasant things about others and especially about the hearer. This maxim
occurs in assertives/ representatives and expressives. Assertives/ representatives are utterances that express
the true propositional. Meanwhile, expressive are utterances that show the speaker feeling. The example is
sampled below.
A: “The performance was great!”

B: “Yes, wasn’t it!”
In the example, A gives a good comment about the performance. He talks the pleasant thing about other.
1
http://awinlanguage.blogspot.com/2013/06/leechs-politeness-principles.html
This expression is a congratulation utterance that maximizes praise of other. Thus this utterance is included
the approbation maxim.
The Modesty Maxim

In the modesty maxim, the participants must minimize praise of self and maximize dispraise of self. This
maxim is applied in assertives/ representatives and expressives like the approbation maxim. Both the
approbation maxim and the modesty maxim concern to the degree of good or bad evaluation of other or self
that is uttered by the speaker. The approbation maxim is exampled by courtesy of congratulation. On other
hand, the modesty maxim usually occurs in apologies. The sample of the modesty maxim is below.
“Please accept this small gift as prize of your achievement.”
In this case, the utterance above is categorized as the modesty maxim because the speaker maximizes
dispraise of himself. The speaker notices his utterance by using “small gift”.
The Agreement Maxim

In the agreement maxim, there is tendency to maximize agreement between self and other people and
minimize disagreement between self and other. The disagreement, in this maxim, usually is expressed by
regret or partial agreement. This maxim occurs in assertives/ representatives illocutionary act. There
example will be illustrated below.
A: “English is a difficult language to learn.”

B: “True, but the grammar is quite easy.”
From the example, B actually does not agree that all part of English language difficult to learn. He does not
express his disagreement strongly to be more polite. The polite answer will influence the effect of the hearer.
In this case, B’s answer minimize his disagreement using partial agreement, “true, but…”.
The Sympathy Maxim

The sympathy maxim explains to minimize antipathy between self and other and maximize sympathy
between self and other. In this case, the achievement being reached by other must be congratulated. On other
hand, the calamity happens to other, must be given sympathy or condolences. This maxim is applicable in
assertives/ representatives. The example is as follows.
“I’m terribly sorry to hear about your father.”
It is a condolence expression which is expressed the sympathy for misfortune. This utterance is uttered when
the hearer gets calamity of father’s died or sick. This expression shows the solidarity between the speaker
and the hearer.
IMPOLITENESS
4
Speech acts
4.1. Performatives versus 4.5. Searle’s typology of speech

constatives 94 acts 106
4.2. Austin’s felicity conditions on 4.6. Indirect speech acts 109

performatives 98
4.7. Speech acts and culture 119
4.3. Locutionary, illocutionary, and
4.8. Summary 127
perlocutionary speech acts 100
4.4. Searle’s felicity conditions

on speech acts 104
Speech act theory, though foreshadowed by the Austrian philosopher

Ludwig Wittgenstein’s views about language-games, is usually attributed
to the Oxford philosopher J. L. Austin.1 The basic ideas, which were
formed by him in the late 1930s, were presented in his lectures given at
Oxford in 1952–54, and later in his William James lectures delivered at
Harvard in 1955. These lectures were Wnally published posthumously as
How to do things with words in 1962. After his death in 1960, Austin’s ideas
were reWned, systematized, and advanced especially by his Oxford pupil,
the American philosopher John R. Searle. Simply stated, the central tenet
of speech act theory is that the uttering of a sentence is, or is part of, an
action within the framework of social institutions and conventions. Put in
slogan form, saying is (part of ) doing, or words are (part of ) deeds.
1
For a history of speech acts, see, e.g., Smith (1990).
94 4. SPEECH ACTS
In Section 4.1, I discuss Austin’s dichotomy between performatives and

constatives. The focus of Section 4.2 is on his felicity conditions on performa-
tives. In Section 4.3, I examine Austin’s tripartite distinction between locu-
tionary, illocutionary, and perlocutionary speech acts. Section 4.4 looks at
Searle’s felicity conditions on speech acts. Then, in Section 4.5, I present
Searle’s classiWcation of speech acts. Next, in Section 4.6, I consider indirect
speech acts. Finally, Section 4.7 discusses some cultural aspects of speech acts.
4.1. Performatives versus constatives
4.1.1. The performative/constative dichotomy
In the 1930s, a very inXuential school of thought in philosophy was logical

positivism, developed by a group of philosophers and mathematicians
principally in Vienna. One of the central doctrines of logical positivism is
what is now called the descriptive fallacy, namely, the view that the only
philosophically interesting function of language is that of making true or
false statements. A particular version of the descriptive fallacy is the so-
called veriWcationist thesis of meaning, namely, the idea that ‘unless a
sentence can, at least in principle, be veriWed (i.e., tested for its truth or
falsity), it was strictly speaking meaningless’ (Levinson 1983: 227; see also
Lyons 1995: 173). On such a view, sentences like those in (4.1) are simply
meaningless, because they are not used to make veriWable or falsiWable
propositions. Instead they express subjective judgments.
(4.1) a. Shouting and screaming at your children is wrong.
b. Elizabeth is more beautiful than Mary.
c. Getting married and having children is better than having children and
getting married.
It was against this philosophical background that Austin set about

developing his theory of speech acts (Austin 1962). He made two important
observations. First, he noted that some ordinary language sentences such
as those in (4.2) are not employed to make a statement, and as such they
cannot be said to be true or false.
(4.2) a. Good morning!
b. Is she a vegetarian?
c. Put the car in the garage, please.
4.1. PERFORMATIVES VERSUS CONSTATIVES 95
Secondly and more importantly, Austin observed that there are ordinary
language declarative sentences that resist a truth-conditional analysis in a
similar fashion. The point of uttering such sentences is not just to say
things, but also actively to do things. In other words, such utterances
have both a descriptive and an eVective aspect. Accordingly, Austin called
them performatives, and he distinguished them from assertions, or statement-
making utterances, which he called constatives.
Put slightly diVerently, in Austin’s view, an initial distinction was made
between performatives and constatives. Performatives are utterances that
are used to do things or perform acts, as in (4.3). By contrast, constatives
are utterances that are employed to make assertions or statements, as
in (4.4).
(4.3) a. I christen/name this ship the Princess Elizabeth.
b. I now pronounce you man/husband and wife.
c. I sentence you to ten years in prison.
d. I promise to come to your talk tomorrow afternoon.
e. I command you to surrender immediately.
f. I apologize for being late.
(4.4) a. My daughter is called Elizabeth.
b. The children are chasing squirrels in the park.
c. Maurice Garin won the Wrst Tour de France in 1903.
Unlike those in (4.4), the declarative sentences in (4.3) have two character-
istics: (i) they are not used intentionally to say anything, true or false,
about states of aVairs in the external world, and (ii) their use constitutes
(part of) an action, namely, that of christening/naming a ship in (4.3a),
that of pronouncing a couple married in (4.3b), that of sentencing a con-
victed criminal in (4.3c), that of promising in (4.3d), that of ordering
in (4.3e), and Wnally that of apologizing in (4.3f). In addition, as pointed
out in Huang (2006a), there are two further diVerences between (4.3a–c)
and (4.3d–f). The Wrst is that while (4.3a–c) is part of a conventional or
ritual behaviour supported by institutional facts (see also Strawson 1964),
(4.3d–f) is not. Secondly, while the performative verb, that is, the verb
naming the action while performing it in (4.3a–c) is in general an essential
element and cannot be omitted, it can in (4.3d–f ). In other words, whereas,
for example, we cannot christen/name a ship without using the verb
christen or name, we can make a promise without using the verb promise,
as in (4.5).
96 4. SPEECH ACTS
(4.5) I’ll come to your talk tomorrow afternoon.
This would be a good point at which to attempt Exercise 1 on p. 128.
Explicit versus implicit performatives

Performatives can further be divided into two types: explicit and implicit
(or in Austin’s 1962: 69 term for the latter, primary). Explicit performatives
are performative utterances which contain a performative verb that makes
explicit what kind of act is being performed. By contrast, implicit performa-
tives are performative utterances in which there is no such a verb. This
contrast is illustrated by the explicit performatives in (4.3) above and the
implicit performatives in (4.5) above and (4.6) below.
(4.6) a. Surrender immediately.
b. How about going to New York on Saturday?
c. Leave me alone, or I’ll call the police.
Now turn to p. 128 and try Exercise 2.
Syntactic and semantic properties of explicit performatives

Austin also isolated a number of syntactic and semantic properties of
explicit performatives in English. They are: (i) explicit performatives con-
tain a performative verb,2 (ii) the performative nature of such a verb can be
reinforced by adding the adverb hereby, and (iii) explicit performatives
occur in sentences with a Wrst-person singular subject of a verb in the
simple present tense, indicative mood, and active voice.
However, as Austin himself was aware, there are exceptions. Explicit
performatives can sometimes take a Wrst-person plural subject, as in (4.7); a
second-person singular or plural subject, as in (4.8); and a third-person
singular or plural subject, as in (4.9). In addition, there are cases where the
explicit performative verb is ‘impersonal’, that is, it does not refer to the
speaker, as in (4.10). Furthermore, as (4.8), (4.9), (4.10a), (4.10c), and
(4.11b) show, explicit performatives can also occur in sentences with the
2
Cross-linguistically, the size of the inventory of performative verbs varies from
language to language. English is extremely rich in performative verbs. According to
Lyons (1995), there are hundreds, if not thousands, of such verbs in the language
(see also Verschueren 1985). By contrast, Russian is a languages that contains a
much smaller number of such verbs.
4.1. PERFORMATIVES VERSUS CONSTATIVES 97
verb in the passive voice. Finally, as the attested examples in (4.11) indicate,
they can also occur in sentences of present progressive aspect.
(4.7) We suggest that you go to the embassy and apply for your visa in person.
(4.8) You are hereby warned that legal action will be taken.
(4.9) Passengers are hereby requested to wear a seat belt.
(4.10) a. Taken from a company’s AGM notice
Notice is hereby given that the Annual General Meeting of O2 plc will
be held at The Hexagon, Queens Walk, Reading, Berkshire RG1 7UA
on Wednesday, 27 July 2005 at 11.00 am for the following purposes: . . .
b. (Hurford and Heasley 1983: 239)
The management hereby warns customers that mistakes in change
cannot be rectiWed once the customer has left the counter.
c. (Levinson 1983: 260)
It is herewith disclosed that the value of the estate left by Marcus
T. Bloomingdale was 4,785,758 dollars.
(4.11) (Thomas 1995: 45)
a. A radio journalist is interviewing the chairman of Railtrack during a strike
by signal workers.
A: Are you denying that the government has interfered?
B: I am denying that.
b. Taken from a naval disciplinary hearing
You are being discharged on the grounds of severe temperamental
unsuitability for service in the Royal Navy.
Notice next that performative verbs can also be used descriptively. In

this usage, they behave like non-performative verbs.
(4.12) a. I baptized John’s baby last Sunday.
b. You are always promising to do housework, but you never do it.
c. It’s the head of department who authorized John to work in our lab.
See whether you can do Exercises 3 and 4 on p. 129.
4.1.2. The performative hypothesis
In order to account for implicit performatives, an analysis known as the

performative hypothesis was put forward in the 1970s. The basic idea of the
hypothesis is that underlying every sentence there is a ‘hidden’ matrix
performative clause of the form given in (4.13).
98 4. SPEECH ACTS
(4.13) I (hereby) Vp you (that) S
where Vp is a performative verb, and S is a complement clause. The

performative verb, which is in the indicative mood, active voice, and simple
present tense (see above), will always make explicit what is implicit, though
the matrix performative clause can be deleted without meaning being
changed (e.g., Ross 1970, LakoV 1972, Sadock 1974; see Gazdar 1979
and Levinson 1983 for a more detailed formulation of the hypothesis; see
also Sadock 2004 for further discussion). Thus, according to this hypo-
thesis, the performative matrix clause in the deep or underlying structure of
(4.14) shows up overtly in (4.15).
(4.14) Stand up.
(4.15) I hereby request that you stand up. Or,
I hereby order you to stand up.
There are problems at the very heart of this analysis. One such problem is
that there are many cases of implicit performatives which do not have an
explicit performative version, even though the relevant verb can be used in
a descriptive way. As an illustration, consider (4.16).
(4.16) a. You’re a stupid cow.
b. ?I hereby insult you that you’re a stupid cow.
c. John insulted Mary by saying that she was a stupid cow.
Intuitively (4.16a) is most naturally interpretable as an insult, but contrary

to the prediction of the performative hypothesis, it does not seem to have an
explicit performative equivalent, as is shown by the oddness of (4.16b). On
the other hand, as (4.16c) indicates, the verb insult can be used descriptively
without any problem. More or less the same can be said of such speech acts as
lying, threatening, and punishing. Faced with this and a variety of other
syntactic, semantic, and pragmatic diYculties (see, e.g., Levinson 1983:
243–78, Yule 1996: 52–3, Marmaridou 2000: 187, Cruse 2004, Sadock
2004), the performative hypothesis has long been abandoned.
4.2. Austin’s felicity conditions on performatives
As already mentioned, it makes no sense to call a performative true or false.

Nevertheless, Austin noticed that for a performative to be successful or
‘felicitous’, it must meet a set of conditions. For example, one such condition
4.2. AUSTIN’S FELICITY CONDITIONS ON PERFORMATIVES 99
for the speech act of naming is that the speaker must be recognized by his or
her community as having the authority to perform that act; for the speech
act of ordering, the condition is that the speaker must be in authority over
the addressee, and Wnally, for the speech act of promising, one condition is
that what is promised by the speaker must be something the addressee
wants to happen. Austin called these conditions felicity conditions. In other
words, felicity conditions are conditions under which words can be used
properly to perform actions.
Austin distinguished three diVerent types of felicity conditions (Austin
1975: 14–15).
(4.17) Austin’s felicity conditions on performatives
A. (i) There must be a conventional procedure having a conventional
eVect.
(ii) The circumstances and persons must be appropriate, as speciWed in
the procedure.
B. The procedure must be executed (i) correctly and (ii) completely.
C. Often
(i) the persons must have the requisite thoughts, feelings and intentions, as
speciWed in the procedure, and
(ii) if consequent conduct is speciWed, then the relevant parties must so do.
Violation of any of the conditions in (4.17) will render a performative

‘unhappy’ or infelicitous. If conditions A or B are not observed, then
what Austin described as a misWre takes place. For instance, in England,
a registrar conducting a marriage ceremony in an unauthorized place will
violate condition A (i), thus committing a misWre. The same is true for a
clergyman baptizing a wrong baby, because in this case, condition A (ii) is
not fulWlled. Next, as an illustration of a violation of condition B (i),
consider the case of a bridegroom not saying the exact words that are
conventionally laid down at a marriage ceremony.3 As to condition B (ii),
it dictates that the procedure must be complete. Thus, in making a bet, the
3
The standard formula at a Church of England wedding has variants:
(i) The Book of Common Prayer version
Curate: Wilt thou have this woman to thy wedded wife, . . .
Bridegroom: I will.
(ii) The Alternative Service Book version
Curate: (Name), will you take (Name) to be your wife . . .
Bridegroom: I will.
100 4. SPEECH ACTS
bet is not ‘on’ unless You are on or something with the same eVect is uttered
by the addressee. In Austin’s terminology, this counts as a satisfactory
uptake,4 the absence of which will again cause a misWre. Finally, if condi-
tion C is not met, resulting in insincerities, then an abuse is the outcome.
Examples of an abuse include congratulating someone when one knows
that he or she passed his or her examination by cheating (condition C (i)),
making a promise when one already intends to break it (condition C (ii)),
and marrying without intending to consummate the marriage (see also
Sadock’s 2004 discussion of these conditions in terms of misinvocation,
misexecution, and abuse). We will return to the question of felicity condi-
tions when we come to Searle’s work.
Now have a go at Exercise 5 on p. 129.
4.3. Locutionary, illocutionary, and perlocutionary

speech acts
The initial distinction made by Austin between performatives and consta-

tives was soon to be rejected by him in favour of a general theory of speech
acts. In fact, as pointed out by Levinson (1983: 231), there are two internal
shifts in Austin’s arguments. First, there is a shift from the view that
performatives are a special class of sentences/utterances with peculiar
syntactic and semantic properties to the view that there is a general class
of performatives that encompasses both explicit and implicit performa-
tives, the latter including many other types of sentence/utterance. The
second shift is from the performative/constative dichotomy to a general
theory of speech acts, of which the various performatives and constatives
are just special subcases.
In America, the following is also used:

(iii) Curate: Do you take this woman . . .
Bridegroom: I do.
4
Thomas (1995: 40–1) called performatives like betting ‘collaborative’ per-
formatives, because their success depends on particular uptake by another person.
Other collaborative performatives include bequeathing and challenging.
4.3. LOCUTIONARY, ILLOCUTIONARY . . . 101
What led Austin to abandon the performative/constative dichotomy? In

the Wrst place, he noted that like performatives, constatives are also subject
to the felicity conditions stated in (4.17). Recollect so-called Moore’s
paradox, discussed in Chapter 2, as illustrated by (4.18).
(4.18) ?Princess Diana died in a fatal car crash in Paris with Dodi Al Fayed, but I
don’t believe it.
This utterance is infelicitous because it violates condition C (i) in (4.17)

above. In the same vein, if someone utters (4.19) when he or she knows that
John does not in fact have a wife, then its presupposition will not go
through (see Chapter 3). The reason the presupposition fails to carry
through is that condition A (ii) in (4.17) above is not adhered to.
(4.19) I’m sure John’s wife is a feminist.
Secondly, Austin observed that performatives and constatives may be

impossible to distinguish even in truth-conditional terms. On the one hand,
there are ‘loose’ constatives that may not be assessed strictly by means of
truth conditions, as in (4.20).
(4.20) a. (Austin 1962) France is hexagonal.

b. John is bald.
c. London is sixty miles from where I live.
France is not, strictly speaking, hexagonal. John still has quite a few
wispy strands of hair on his head. London may not be exactly sixty miles
from where the speaker lives. Thus, statements like those in (4.20) can only
be said to be more or less, or roughly, true. On the other hand, there are
utterances like those in (4.21) that pass the hereby test, and therefore are
performatives by deWnition, but that nevertheless are used to state or assert.
In these cases, the performatives must be counted simultaneously as
constatives.
(4.21) a. I hereby state that John is growing GM crops.

b. I hereby tell you that the prime minister is not going to stand down.
c. I hereby hypothesize that there is water on Mars.
On the basis of such evidence, Austin concluded that constatives are

nothing but a special class of performatives, and that the two-way distinc-
tion between performatives, as action-performers, and constatives, as
truth-bearers, can no longer be maintained.
102 4. SPEECH ACTS
Consequently, Austin claimed that all utterances, in addition to meaning

whatever they mean, perform speciWc acts via the speciWc communicative
force of an utterance. Furthermore, he introduced a threefold distinction
among the acts one simultaneously performs when saying something.
(4.22) Three facets of a speech act
(i) Locutionary act: the production of a meaningful linguistic expression.
(ii) Illocutionary act: the action intended to be performed by a speaker in
uttering a linguistic expression, by virtue of the conventional force
associated with it, either explicitly or implicitly.
(iii) Perlocutionary act: the bringing about of consequences or eVects on
the audience through the uttering of a linguistic expression, such
consequences or eVects being special to the circumstances of the
utterance.
A locutionary act is the basic act of speaking, which itself consists of three
related subacts. They are (i) a phonic act of producing an utterance-
inscription, (ii) a phatic act of composing a particular linguistic expression
in a particular language, and (iii) a rhetic act of contextualizing the utter-
ance-inscription (Austin 1962, Lyons 1995: 177–85). In other words, the
Wrst of these three subacts is concerned with the physical act of making a
certain sequence of vocal sounds (in the case of spoken language) or a set of
written symbols (in the case of written language). The second refers to the
act of constructing a well-formed string of sounds/symbols, be it a word,
phrase or sentence, in a particular language. The third subact is responsible
for tasks such as assigning reference, resolving deixis, and disambiguating
the utterance-inscription lexically and/or grammatically. These three sub-
acts correspond broadly to the three distinct levels and modes of explan-
ation in linguistic theory, namely, phonetics/phonology, morphology/
syntax, and semantics/pragmatics.
When we say something, we usually say it with some purpose in mind.
This is the illocutionary act. In other words, an illocutionary act refers to
the type of function the speaker intends to fulWl, or the type of action the
speaker intends to accomplish in the course of producing an utterance. It is
an act deWned within a system of social conventions. In short, it is an act
accomplished in speaking. Examples of illocutionary acts include accusing,
apologizing, blaming, congratulating, giving permission, joking, nagging,
naming, promising, ordering, refusing, swearing, and thanking. The func-
tions or actions just mentioned are also commonly referred to as the
4.3. LOCUTIONARY, ILLOCUTIONARY . . . 103
illocutionary force or point of the utterance. Illocutionary force is frequently

conveyed by what Searle (1969) called an illocutionary force indicating
device (IFID), the most direct and conventional type of which is an explicit
performative in the form of (4.13) above. Indeed, the term ‘speech act’ in its
narrow sense is often taken to refer speciWcally to illocutionary acts.
It should be mentioned at this point that the same linguistic expression
can be used to carry out a wide variety of diVerent speech acts, so that the
same locutionary act can count as having diVerent illocutionary forces in
diVerent contexts. Depending on the circumstances, one may utter (4.23)
below to make a threat, to issue a warning or to give an explanation.
(4.23) The gun is loaded.
In fact, Alston (1994) argued that the meaning of a sentence consists in

its having a certain illocutionary act potential (IAP) that is closely and
conventionally associated with its form. On this view, to know what a
sentence means is to know what range of illocutionary acts it can conven-
tionally be used to perform (see also Recanati 2004b).
Conversely, the same speech act can be performed by diVerent linguistic
expressions, or the same illocutionary force can be realized by means of
diVerent locutionary acts. The utterances in (4.24), for example, illustrate
diVerent ways of carrying out the same speech act of requesting.
(4.24) (At ticket oYce in railway station)
a. A day return ticket to Oxford, please.
b. Can I have a day return ticket to Oxford, please?
c. I’d like a day return ticket to Oxford.
Finally, a perlocutionary act concerns the eVect an utterance may have on

the addressee. Put slightly more technically, a perlocution is the act by
which the illocution produces a certain eVect in or exerts a certain inXuence
on the addressee. Still another way to put it is that a perlocutionary act
represents a consequence or by-product of speaking, whether intentional or
not. It is therefore an act performed by speaking. For example, in an armed
bank robbery, a robber may utter (4.23) to get the cashier to open the safe.
This eVect of the act performed by speaking is also generally known as the
perlocutionary eVect.
While there are unclear cases, the main diVerences between illocutions
and perlocutions can be summed up as follows. In the Wrst place, illocu-
tionary acts are intended by the speaker, while perlocutional eVects are not
104 4. SPEECH ACTS
always intended by him or her. Secondly, illocutionary acts are under the
speaker’s full control, while perlocutionary eVects are not under his or her
full control. Thirdly, if illocutionary acts are evident, they become evident
as the utterance is made, while perlocutionary eVects are usually not
evident until after the utterance has been made (Hurford and Heasley
1983: 247). Fourthly, illocutionary acts are in principle determinate,5
while perlocutionary eVects are often indeterminate. Finally, illocutionary
acts are more, while perlocutionary eVects are less conventionally tied to
linguistic forms (see also Sadock 2004).
At this point, see whether you can tackle Exercise 6 on p. 129.
4.4. Searle’s felicity conditions on speech acts
Recall our discussion of the felicity conditions speciWed by Austin in Section

4.2 above. Just as its truth conditions must be met by the world for a sentence
to be said to be true (see Chapter 1), its felicity conditions must be fulWlled by
the world for a speech act to be said to be felicitous. Searle (1969) took the
view that the felicity conditions put forward by Austin are not only ways in
which a speech act can be appropriate or inappropriate, but they also jointly
constitute the illocutionary force. Put in a diVerent way, the felicity condi-
tions are the constitutive rules—rules that create the activity itself—of speech
acts. On Searle’s view, to perform a speech act is to obey certain conven-
tional rules that are constitutive of that type of act. Searle developed the
original Austinian felicity conditions into a neo-Austinian classiWcation of
four basic categories, namely, (i) propositional content, (ii) preparatory
condition, (iii) sincerity condition, and (iv) essential condition. As an illus-
tration of these conditions, consider (4.25) and (4.26).
5
But there are situations in which an illocutionary act is not determinate, as the
following example shows.
(i) Employer: I’ll come back to see how the work is progressing the day after
tomorrow.
Employee: Is that a promise or a threat?
4.4. SEARLE’S FELICITY CONDITIONS ON SPEECH ACTS 105
(4.25) Searle’s felicity conditions for promising

(i) Propositional content: future act A of S
(ii) Preparatory: (a) H would prefer S’s doing A to his not doing A, and S so
believes (b). It is not obvious to both S and H that S will do A in the
normal course of events
(iii) Sincerity: S intends to do A
(iv) Essential: the utterance of e counts as an undertaking to do A
where S stands for the speaker, H for the hearer, A for the action, and e for
the linguistic expression.
(4.26) Searle’s felicity conditions for requesting
(i) Propositional content: future act A of H
(ii) Preparatory: (a) S believes H can do A (b) It is not obvious that H
would do A without being asked
(iii) Sincerity: S wants H to do A
(iv) Essential: the utterance of e counts as an attempt to get H to do A
Let us now work through these conditions one by one. The propositional
content condition is in essence concerned with what the speech act is about.
That is, it has to do with specifying the restrictions on the content of what
remains as the ‘core’ of the utterance (i.e. Searle’s propositional act) after
the illocutionary act part is removed. For a promise, the propositional
content is to predicate some future act of the speaker, whereas in the case
of a request, it is to predicate some future act of the addressee. The
preparatory conditions state the real-world prerequisites for the speech
act. For a promise, these are roughly that the addressee would prefer the
promised action to be accomplished, that the speaker knows this, but also
that it is clear to both the speaker and the addressee that what is promised
will not happen in the normal course of action. In the case of a request, the
preparatory conditions are that the speaker has reason to believe that
the addressee has the ability to carry out the action requested, and that if
the addressee is not asked, he or she will not perform the action. Next,
the sincerity condition must be satisWed if the act is to be performed
sincerely. Thus, when carrying out an act of promising, the speaker must
genuinely intend to keep the promise. When making a request, the speaker
must want the addressee to do the requested action. Notice that if the
sincerity condition is not fulWlled, the act is still performed, but there is
an abuse, to use Austin’s term. Finally, the essential condition deWnes the
act being performed in the sense that the speaker has the intention that his
or her utterance will count as the identiWable act, and that this intention is
106 4. SPEECH ACTS
recognized by the addressee. Thus in the case of a promise, the speaker

must have the intention to create an obligation to act, and for a request, the
speaker must intend that his or her utterance counts as an attempt to get
the addressee to do what is requested. Failure to meet the essential condi-
tion has the consequence that the act has not been carried out.
Now see what you make of Exercise 7 on p. 129.
4.5. Searle’s typology of speech acts
Can speech acts be classiWed, and if so, how? Austin (1962) grouped them
into Wve types: (i) verdictives—giving a verdict, (ii) exercitives—exercising
power, rights, or inXuence, (iii) commissives—promising or otherwise
undertaking, (iv) behabitives—showing attitudes and social behaviour,
and (v) expositives—Wtting an utterance into the course of an argument
or conversation. Since then, there have been many attempts to systematize,
strengthen, and develop the original Austinian taxonomy (e.g., Bach and
Harnish 1979, Allan 2001, Bach 2004). Some of these new classiWcations
are formulated in formal/grammatical terms, others in semantic/pragmatic
terms, and still others on the basis of the combined formal/grammatical
and semantic/pragmatic modes (see, e.g., Sadock 2004 for a review). Of all
these (older and newer) schemes, Searle’s (1975a) neo-Austinian typology of
speech acts remains the most inXuential.
Under Searle’s taxonomy, speech acts are universally grouped into Wve types
along four dimensions: (i) illocutionary point or speech act type, (ii) direction of
Wt or relationship between words and world, (iii) expressed psychological state,
and (iv) propositional content (see also Searle 1979, 2002, but see Section 4.7.1
below). The Wve types of speech act are further explained below.
(i) Representatives (or assertives; the constatives in the original Austinian
performative/constative dichotomy) are those kinds of speech act that
commit the speaker to the truth of the expressed proposition, and thus
carry a truth-value. They express the speaker’s belief. Paradigmatic cases
include asserting, claiming, concluding, reporting, and stating. In perform-
ing this type of speech act, the speaker represents the world as he or she
believes it is, thus making the words Wt the world of belief. Representatives
are illustrated in (4.27).
4.5. SEARLE’S TYPOLOGY OF SPEECH ACTS 107
(4.27) a. Chinese characters were borrowed to write other languages, notably

Japanese, Korean and Vietnamese.
b. Francis Crick and Jim Watson discovered the double helix structure of
DNA.
c. The soldiers are struggling on through the snow.
(ii) Directives are those kinds of speech act that represent attempts by the
speaker to get the addressee to do something. They express the speaker’s
desire/wish for the addressee to do something. Paradigmatic cases include
advice, commands, orders, questions, and requests. In using a directive,
the speaker intends to elicit some future course of action on the part of the
addressee, thus making the world match the words via the addressee.
Directives are exempliWed in (4.28).
(4.28) a. Turn the TV down.
b. Don’t use my electric shaver.
c. Could you please get that lid oV for me?
(iii) Commissives are those kinds of speech act that commit the speaker to
some future course of action. They express the speaker’s intention to do
something. Paradigmatic cases include oVers, pledges, promises, refusals,
and threats. In the case of a commissive, the world is adapted to the words via
the speaker him- or herself. Examples of commissives are presented in (4.29).
(4.29) a. I’ll be back in Wve minutes.
b. We’ll be launching a new policing unit to Wght cyber crime on the
internet soon.
c. I’ll never buy you another computer game.
(iv) Expressives are those kinds of speech act that express a psychological
attitude or state in the speaker such as joy, sorrow, and likes/dislikes.
Paradigmatic cases include apologizing, blaming, congratulating, praising,
and thanking. There is no direction of Wt for this type of speech act.6
(4.30) a. Well done, Elizabeth!
b. I’m so happy.
c. Wow, great!7
6
Alternatively, one might claim, as Yule (1996: 55) did, that in performing the
act of an expressive, the speaker makes known what he or she feels, thus rendering
the words to Wt the world of feeling.
7
In many languages including English, sentences with the import of the follow-
ing are often used to express one’s strong emotions.
108 4. SPEECH ACTS
(v) Declarations (or declaratives) are those kinds of speech act that eVect
immediate changes in some current state of aVairs. Because they tend to
rely on elaborate extralinguistic institutions for their successful perform-
ance, they may be called institutionalized performatives. In performing this
type of speech act, the speaker brings about changes in the world; that is, he
or she eVects a correspondence between the propositional content and the
world. Paradigmatic cases include bidding in bridge, declaring war, excom-
municating, Wring from employment, and nominating a candidate. As to
the direction of Wt, it is both words-to-world and world-to-words.8
(4.31) a. President: I declare a state of national emergency.
b. Chairman: The meeting is adjourned.
c. Jury foreman: We Wnd the defendant not guilty.
Illocutional point, direction of Wt, and expressed psychological state can

be summarized in (4.32) (see Vanderveken 1994 for a formalization in
terms of illocutionary logic).9
(4.32) Illocutionary point Direction of Wt Expressed psychological state
Representative words-to-world belief (speaker)
Directives world-to-words desire (addressee)
Commissives world-to-words intention (speaker)
Expressives none variable (speaker)
Declarations both none (speaker)
Now attempt Exercise 8 on p. 129.
(i) a. I don’t know what to say.

b. I’m absolutely speechless.
c. No words can express my . . . !
8
Notice that some performative verbs in English can Wt into more than one of
Searle’s categories. They include advise, confess, suggest, tell, and warn.
(i) I warn you not to dance/against dancing on the table. (directive)

(ii) I warn you that there are no train services on Sundays. (assertive)
9
In Searle (1983), there were also three directions of Wt between mind and the
world: (i) the mind-to-world direction of Wt, (ii) the world-to-mind direction of Wt,
and (iii) the empty direction of Wt. De Sousa Melo (2002) proposed that a fourth
direction of Wt—the double direction of Wt—be added to Searle’s typology, parallel-
ing the four directions of Wt between words and the world.
4.6. INDIRECT SPEECH ACTS 109
4.6. Indirect speech acts
4.6.1. What is an indirect speech act?
Most of the world’s languages have three basic sentence types: (i) declara-
tive, (ii) interrogative, and (iii) imperative.10 In some languages, the three
major sentence types are distinguished morphologically and/or syntactic-
ally. Somali provides an example of such a language.
(4.33) (Saeed 2003: 237)
a. Warkii waad dhegeysatay.
news the DECL-you listen to-2SG-PAST
‘You listened to the news.’
b. Warkii miyaad dhegeysatay?
news the Q-you listen to-2SG-PAST
‘Did you listen to the news?’
c. Warkii dhegeyso.
news the listen to-2SG-IMPV
‘Listen to the news.’
In (4.33), the declarative format is marked by waa, the interrogative format

is indicated by ma, and the imperative format is indicated with a zero
marking (Saeed 2003: 237). Another example is provided by Greenlandic.
(4.34) (Sadock and Zwicky 1985)
a. Igavoq.
cook-INDIC-3SG
‘He cooks.’
b. Igava.
cook-Q-3SG
‘Does he cook?’
In (4.34), the declarative and interrogative are diVerentiated by means of

separate personal suYxes (Sadock and Zwicky 1985). Finally, in Lakhota,
the diVerence between a ‘declarative’ with an indeWnite pronoun and an
interrogative with an interrogative pronoun is marked by the adding of a
sentence-Wnal question particle. (Notice that the indeWnite and interroga-
tive pronouns are identical.) This is illustrated in (4.35).
10
There are languages in the world that lack a genuine declarative. One such
language is Hidatsa, as described in Sadock and Zwicky (1985).
110 4. SPEECH ACTS
(4.35) (Croft 1994)

a. Mniluzahe Othuwahe ekta tuwa ya.
Rapid City to who go-3SG
‘Someone went to Rapid City.’
b. Mniluzahe Othuwahe ekta tuwa ya he?
Rapid City to who go-3SG Q
‘Who went to Rapid City?’
The three major sentence types are typically associated with the three
basic illocutionary forces, namely, asserting/stating, asking/questioning,
and ordering/requesting, respectively. Thus, the three Somali sentences in
(4.33), for instance, may be paraphrased using explicit performatives.
(4.36) a. I (hereby) state that you listened to the news.
b. I (hereby) enquire whether you listened to the news.11
c. I (hereby) order you to listen to the news.
Now, if there is a direct match between a sentence type and an illocu-

tionary force, we have a direct speech act. In addition, explicit performa-
tives, which happen to be in the declarative form, are also taken to be direct
speech acts, because they have their illocutionary force explicitly named by
the performative verb in the main part (or ‘matrix clause’) of the sentence.
On the other hand, if there is no direct relationship between a sentence type
and an illocutionary force, we are faced with an indirect speech act. Thus,
when an explicit performative is used to make a request, as in (4.37), it
functions as a direct speech act; the same is the case when an imperative is
employed, as in (4.38). By comparison, when an interrogative is used to
make a request, as in (4.39), we have an indirect speech act.
11
Some native speakers of English may Wnd sentences like (4.36b) odd (e.g.
Sadock and Zwicky 1985). For this group of speakers, the interrogative is a
sentence type for which there is no explicit performative equivalent. A better
example may be provided by the exclamation in English.
(i) a. What a beautiful girl she is!

b. ?I (hereby) exclaim what a beautiful girl she is.
c. I exclaimed what a beautiful girl she was.
As shown by (ia) and (ib), the exclamation in English does not have a correspond-
ing explicit performative, though the verb exclaim can be used descriptively (see,
e.g., Cruse 2004 for further discussion).
(4.37) I request you to pass the salt.

(4.38) Pass the salt.
(4.39) Can you pass the salt?
In short, the validity of the distinction between direct and indirect speech
act is dependent on whether or not one subscribes to what Levinson (1983:
264, 274) called the literal force hypothesis—the view that there is a direct
structure–function correlation in speech acts and that sentence forms are
by default direct reXexes of their underlying illocutionary forces.
There are, however, problems at the very heart of the literal force
hypothesis. One is that there are cases of speech acts where even the direct
link between performative verbs and speech acts breaks down. Consider
(4.40).
(4.40) I promise to sack you if you don’t Wnish the job by this weekend.
In (4.40), the performative verb is promise, but the illocutionary force that
is most naturally ascribed to this speech act is that of either a threat or a
warning. This shows that contrary to the main prediction of the literal force
hypothesis, we cannot always identify speech acts even with sentences that
contain a performative verb.
Secondly and more importantly, as pointed out by Levinson (1983: 264),
most usages are indirect.12 The speech act of requesting, for example, is
very rarely performed by means of an imperative in English. Instead, it is
standardly carried out indirectly, and there is probably an inWnite variety of
sentences that can be used to make a request indirectly, as shown in (4.41)
(adapted from Levinson 1983: 264–5; see also Bertolet 1994, Holdcroft
1994).
(4.41) a. I want you to close the window.
b. Can you close the window?
c. Will you close the window?
d. Would you close the window?
e. Would you mind closing the window?
f. You ought to close the window.
g. May I ask you to close the window?
h. I wonder if you’d mind closing the window.
See whether you can tackle Exercise 9 on pp. 129–30 now.
12
As we will shortly see in Section 4.7 below, the degree of directness/indirect-
ness may vary cross-linguistically.
112 4. SPEECH ACTS
4.6.2. How is an indirect speech act analysed?
Roughly, there are three main approaches. The Wrst is to assume the
existence of a dual illocutionary force (as proposed by Searle 1975b). On
this assumption, indirect speech acts have two illocutionary forces,
one literal or direct, and the other non-literal or indirect. While the literal
force is secondary, the non-literal force is primary. Next, whether an
utterance operates as an indirect speech act or not has to do with
the relevant felicity conditions. For example, (4.39) above both infringes
the felicity condition for the speech act of questioning and queries the
preparatory condition for that of requesting. This explains why it can
function as an indirect speech act whereas (4.42), for example, cannot.
The reason is that in the case of (4.42), felicity conditions are irrelevant.
(4.42) (Searle 1975b)
Salt is made of sodium chloride.
Finally, on Searle’s view, a speaker’s performing and an addressee’s under-

standing an indirect speech act always involves some kind of inference. The
question that arises next is how this inference can be computed. Searle’s
suggestion is that it can be computed along the general lines of the rational,
co-operative model of communication articulated by Grice (see Chapter 2).13
One interesting characteristic of indirect speech acts is that they are
frequently conventionalized (Morgan 1978). This can be illustrated by the
fact that of various, apparently synonymous linguistic expressions, only
one may conventionally be used to convey an indirect speech act. Consider
(4.43).
(4. 43) a. Are you able to pass the salt?
b. Do you have the ability to pass the salt?
Under Searle’s analysis, both (4.43a) and (4.43b) would be expected to be

able to perform the indirect speech act of requesting, because (i) they
are largely synonymous with (4.39), and (ii) they, too, inquire about the
satisfaction of the addressee-based preparatory condition for making a
request. But this expectation is not fulWlled.
13
See, e.g., Vanderveken (2002) for an attempt to reformulate Grice’s maxims in
terms of speech acts. See Dascal (1994) for a comparison between speech act theory
and Gricean theory of conversational implicature.
Searle’s response to this puzzle is that there is also a certain degree of

conventionality about indirect speech acts, and this may be accounted for
in terms of convention of use/meaning. Inspired by this insight of Searle’s,
Morgan (1978) developed a notion of short-circuited implicature to cover
inference involved in cases like (4.39). In these cases, while the relevant
conversational implicature is in principle calculable, it is not in practice
calculated (see also Horn and Bayer 1984, Horn 1988). From a linguistic
point of view, then, the conventionality here is correlated with the possible
occurrence of please. While please can be inserted before the verb pass in
(4.37)–(4.39), it cannot in (4.43), as shown in (4.44).
(4.44) a. I request you to please pass the salt.
b. Please pass the salt.
c. Can you please pass the salt?
d. ?Are you able to please pass the salt?
e. ?Do you have the ability to please pass the salt?
Furthermore, the conventionality indicated by please in (4.44a) and (4.44b)

is one of meaning, hence the speech act of requesting is performed directly.
By contrast, the conventionality signalled by please in (4.44c) is one of
usage, and thus we have an indirect speech act.14
A second, rather similar, approach is due to Gordon and LakoV (1975).
In their analysis, there are inference rules called conversational postulates
that reduce the amount of inference needed to interpret an indirect speech
act. Thus in the case of (4.39), if the interpretation as a question cannot be
intended by the speaker, then the utterance will be read as being equivalent
to his or her having said (4.37), thus resulting in the performance of the
indirect speech act of requesting. Stated this way, the conversational pos-
tulates proposed by Gordon and LakoV can be seen as another reXection of
the conventionality of indirect speech acts. As to the similarities and
diVerences between Searle’s and Gordon and LakoV’s analyses, the major
similarity is that both accounts assume that the interpretation of indirect
speech acts involves both inference and conventionality; the major diVer-
ence concerns the question of balance, namely, how much of the work
14
As Searle (1975b) and Green (1975) showed, the use of the Can/Would you . . .
type question to indirectly express requests and orders varies cross-linguistically
(see also Wierzbicka 1991). According to Horn (1988), given that a short-circuited
implicature is itself a matter of convention, this is precisely to be expected.
114 4. SPEECH ACTS
involved in computing an indirect speech act is inferential and how much is

conventional.
Finally, in contrast to the inferential model we have just discussed, there
is the idiom model. In this model, sentences like (4.39) are semantically
ambiguous, and the request interpretation constitutes a speech act idiom
that involves no inference at all. On this view, (4.39) is simply recognized as
a request with no question being perceived. This is the position taken by
Sadock (1974). There are, however, problems with this analysis, too. One is
that it fails to capture the fact that the meaning of an indirect speech act can
frequently (at least in part) be derived from the meaning of its components.
The technical term for this is compositionality (see Chapter 3). In addition,
these would-be idioms turn out to be quite comparable cross-linguistically.
For example, like their English equivalent in (4.45), (4.46)–(4.49) can all be
used to indirectly request the addressee to turn on the central heating
system, depending on context.15
(4.45) It’s cold in here.
(4.46) (Arabic)
?na-hu barid huma.
it’s-3-M-S cold in here
‘It’s cold in here.’
(4.47) (Chinese)
zher zhen leng.
here really cold
‘It’s really cold in here.’
(4.48) (German)
Es ist sehr kalt hier drin.
it is very cold here in
‘It’s very cold in here.’
(4.49) (Modern Greek)
Kani krio edo mesa.
it is doing cold here in
‘It’s cold in here.’
15
Intuitively, the requests in (4.45)–(4.49) are more indirectly conveyed than
that in (4.39). As noted by Green (1975) and Horn (1988), in contrast to the request
in (4.39), the requests in (4.45)–(4.49) are non-short-circuited, and hence non-
detachable implicatures. Consequently, they are not subject to cross-linguistic
variation.
A further problem is that in the idiom model, an interpretation that takes

into account the literal meaning or the direct illocutionary force of an
indirect speech act is not allowed. This, however, leaves examples like
(4.50) unexplained.
(4.50) A: Can you pass the salt?
B: Yes, I can. (Here it is.)
4.6.3. Why is an indirect speech act used?

Some remarks on politeness
Why, then, do people use indirect speech acts? One answer is that the use of
indirect speech acts is in general associated with politeness. Indirect speech
acts are usually considered to be more polite than their direct counterparts.
Furthermore, the more indirect a speech act, the more polite.16
There is an extensive literature on politeness (see, e.g., DuFon et al. 1994
for a bibliography on politeness covering publications up to the early
1990s), and this is not the place for me to give a full review of it. Instead,
16
This can be illustrated by English. Consider (i)–(v).
(i) Call Lucy a taxi, please.
(ii) Will you call Lucy a taxi?
(iii) Would you call Lucy a taxi?
(iv) Would you mind calling Lucy a taxi?
(v) I wonder if you’d mind calling Lucy a taxi?
The speech act of requesting is performed more indirectly, for example, using (iii)
than using (i), and therefore the utterance in (iii) is considered more polite than that
in (i). But this may not be the case with, say, Polish, as argued by, e.g., Wierzbicka
(1991). Jaszczolt (2002: 307) pointed out that (vi) and (vii) in Polish are equivalent
in politeness and in commonality to (viii) in English (see also SiWanou 1992 on
Greek).
(vi) Siadaj.
sit down
(vii) Usia̧dź, proszȩ.
sit down please
(vii) Will you sit down?
This raises the question of whether or not indirect speech acts are universally more
polite than their direct counterparts.
116 4. SPEECH ACTS
in what follows, I shall provide a brief discussion of politeness with special

reference to speech acts.
Currently, there are four main theoretical models of politeness: (i) the
‘social norm’ model, (ii) the ‘conversational maxim’ model (e.g., Leech 1983,
2003) (see note 12 in Chapter 2), (iii) the ‘face-saving’ model (Brown and
Levinson 1978, 1987), and (iv) the ‘conversational contract’ model (e.g.,
Fraser 1990). (See also Watts 2003 for a ‘social practice’ model.17) Of
these four models, the most inXuential and comprehensive is Brown and
Levinson’s now classic ‘face-saving’ model.
At the heart of Brown and Levinson’s theory of politeness lies GoVman’s
(1967) sociological notion of face.18 Simply put, face is ‘the public self-image
that every member wants to claim for himself’ (Brown and Levinson 1987:
61). Stated in another way, face means roughly an individual’s self-esteem.
Furthermore, there are two aspects to face. First, positive face, which repre-
sents an individual’s desire to be accepted and liked by others. Positive
politeness orients to preserving the positive face of others. When one uses
positive politeness, one tends to choose the speech strategies that emphasize
one’s solidarity with the addressee. These strategies include claiming ‘com-
mon ground’ with the addressee, conveying that the speaker and the
addressee are co-operators, and satisfying the addressee’s wants (Brown
and Levinson 1987: 101–29). Second, there is negative face, which refers to
an individual’s right to freedom of action and his or her need not to be
imposed on by others. Negative politeness orients to maintaining the nega-
tive face of others. When one employs negative politeness, one tends to opt
for the speech strategies that emphasize one’s deference to the addressee.
Typical linguistic realizations of the strategies of negative politeness involve,
for example, the use of conventional indirectness, hedges on illocutionary
force, and apologies (Brown and Levinson 1987: 130). DeWned thus, face is
considered to be a universal notion in any human society. As rational agents,
conversational participants will ideally try to preserve both their own face
and their interlocutors’ face in a verbal interaction.
17
According to DuFon et al. (1994), the earliest book on politeness in the
Western tradition is Libro del cortegiano (The book of the courtier) by Baldesar
Castiglione, published in 1528.
18
The notion of ‘face’ and its related English expressions such as losing/saving/
gaining face seem to originate from the Chinese expression mianzi. The concept of
‘face’ seemed to be introduced to the West by the Chinese anthropologist Hu Hsien
Chin in 1944 (Hu 1944).
Many types of speech acts such as complaints, disagreements (e.g.,

Locher 2004), and requests intrinsically threaten face. Hence they are called
face-threatening acts (FTAs). In the Wrst place, FTAs can threaten positive
face, negative face, or both. Acts that threaten positive face include expres-
sions of disapproval, accusations, criticism, disagreements, and insults;
those that threaten negative face include advice, orders, requests, sugges-
tions, and warnings; those that threaten both positive and negative face
include complaints, interruptions, and threats. Furthermore, a second
distinction can be made between acts that primarily threaten the speaker’s
face and those that primarily threaten the addressee’s face. The speaker can
threaten his or her own face by performing, for example, the acts of
accepting compliments, expressing thanks, and making confessions. On
the other hand, acts such as advice, reminding, and strong expression
of emotions threaten primarily the addressee’s face wants (Brown and
Levinson 1987: 67–8).
Brown and Levinson (1987: 74) posited three independent and culturally
sensitive social variables according to which the strength or weightiness of
an FTA can be measured. First, there is the social distance (D) between the
speaker and the addressee. Second is the relative power (P) of the addressee
over the speaker. Finally, the third variable is the absolute ranking (R) of
imposition in a particular culture (see also Scollon and Scollon 1995). The
strength of an FTA is measured by adding together the three variables D,
P, and R, on the basis of which the amount of face work needed or the
degree of politeness required can be worked out.
On Brown and Levinson’s model, there is a set of Wve strategies one can
choose from to avoid or weaken an FTA. The strategies are given in (4.51).
(4.51) Brown and Levinson’s (1987: 60) set of FTA-avoiding strategies
Lesser
Estimation of risk of face loss
1. without redressive action, baldly
on record 2. positive politeness
Do the FTA with redressive action
4. off record 3. negative politeness
5. Don’t do the FTA
Greater
118 4. SPEECH ACTS
Notice that in the set of strategies given in (4.51), the more threatening an
FTA is, the more polite (i.e. the higher-numbered) the strategy the speaker
must employ to mitigate its eVects.
Let me now explore these strategies in somewhat greater detail. In the
Wrst place, one can perform an FTA on record, that is, directly; or oV
record, that is, indirectly, for example, by dropping a verbal hint. Secondly,
an FTA on record can be performed either baldly, i.e. without redress; or
with redress. In the latter case, there are two further choices: an FTA can be
performed with a face-saving act using either positive politeness redress
(such as emphasizing solidarity with the addressee) or negative politeness
redress (such as giving deference to the addressee). Finally, there is also the
option for one not to perform the FTA at all, for instance, by dropping a
non-verbal hint. These Wve strategies can be illustrated in (4.52).
(4.52) Situation: John, a student asks Mary, another student to lend him her
lecture notes
1. On record, without redress, baldly:
Lend me your lecture notes.
2. On record, with positive politeness redress:
How about letting me have a look at your lecture notes?
3. On record, with negative politeness redress:
Could you please lend me your lecture notes?
4. OV record:
I didn’t take any notes for the last lecture.
5. Don’t perform the FTA:
[John silently looks at Mary’s lecture notes.]
It will be apparent that the use of indirect speech acts serves to diminish
FTAs and thereby to satisfy face needs.
You may now be ready to tackle Exercise 10 on p. 130.
Since its publication in 1978, Brown and Levinson’s ‘face-saving’ theory

has generated an enormous amount of research on politeness (see Brown
and Levinson 1987 for a survey of the literature up to the mid-1980s). From
an empirical point of view, while Brown and Levinson’s model is based on
an in-depth study of three unrelated languages/cultures, namely, Tamil,
Tzeltal, and English (both British and American), politeness phenomena in
a much wider array of languages/cultures ranging from Catalan through
Japanese to Turkish have in the meantime been carefully studied. Concep-
tually, many of the recent studies have focused on the universality of the
4.7. SPEECH ACTS AND CULTURE 119
Brown and Levinson theory, in particular the notion of face, the social
variables, and the set of Wve politeness strategies. SiWanou (1992), for
example, in a cross-cultural comparison between the politeness systems in
England and Greece, argued that the politeness system in Greek is more
positive-face oriented, whereas in British English it is more negative-face
oriented. While arguing for such diVerences, her study nevertheless showed
that Brown and Levinson’s theory provides a very good general theoretical
and descriptive framework for the study of politeness. On the other hand, a
number of scholars have found it diYcult to apply Brown and Levinson’s
model directly to the politeness systems in Eastern, collective cultures (see,
e.g., Gu 1990, Mao 1994, and Lee-Wong 2000 on Chinese; Matsumoto
1988, 1989, and Ide 1989 on Japanese; Hwang 1990 on Korean; Bayraktar-
oglu 2001 and Zeyrek 2001 on Turkish; Koutlaki 2002 on Persian). The
same can be said of some African languages/cultures. Nwoye (1989, 1992),
for example, carried out a study of the notion of face in Igbo. According to
this study, Igbo society is more concerned with the collective self-image of a
group than the individualistic self-image of any one member of the group.
The ‘group face’ is deWned as ‘[t]he avoidance of behavior capable of
lowering the public self-image or self-worth[iness] (iru) of one’s group,
dictated by the fear of imecu iru (to darken face)’ (Nwoye 1992: 314)
(see also Agyekum 2004 on the concept of face in Akan). Strecker (1993)
also argued that the politeness system of the Hamer in south Ethiopia
presents a challenge to Brown and Levinson’s theory. All this seems to
indicate that while politeness itself is a universal phenomenon, politeness
strategies and individual speech acts may to some extent vary from one
language/culture to another.
4.7. Speech acts and culture
4.7.1. Cross-cultural variation
Many speech acts are culture-speciWc. This is particularly so in the case of

institutionalized speech acts, which typically use standardized and stereo-
typed formulae and are performed in public ceremonies. A good example is
provided by the speech act of divorcing. In some Muslim cultures, under
the appropriate circumstances, the uttering of a sentence with the import of
120 4. SPEECH ACTS
(4.53) three times consecutively by a husband to his wife will ipso facto
constitute a divorce. By contrast, in Western cultures, no one (no matter
what his or her religion is) can felicitously use (4.53) to obtain a divorce.19
(4.53) ‘I hereby divorce you.’
But how about non-institutionalized speech acts? First of all, as said

above, any given speech act may be culture-speciWc. Rosaldo (1982), for
example, observed that the speech act of promising has no place among the
Ilongots—a tribal group of hunters and horticulturalists in the Philippines.
She attributed the absence of this speech act in the conceptual repertoire of
the Ilongots to a lack of interest in sincerity and truth in that community.
The Ilongots, argued Rosaldo, are more concerned with social relation-
ships than with personal intentions. On the basis of anthropological evi-
dence such as this, Rosaldo claimed that the universality of Searle’s
typology of speech acts cannot be maintained. Another example of this
kind has been reported for the Australian aboriginal language Yolngu.
According to Harris (1984: 134–5), there does not seem to be any speech act
of thanking in the Yolngu speaker’s repertoire.
Conversely, a given speech act may be present only in certain cultures.
For example, in the Australian aboriginal language Walmajarri, one Wnds a
speech act of requesting that is based on kinship rights and obligations. The
verb in question is japirlyung (Hudson 1985), and the speech act may be
called ‘kinship-based requesting’, because it conveys a message meaning
19
In some Muslim cultures, even an unintentional use of (4.53) can eVect a
divorce, as can be shown by the following incident in Pakistan, cited from Thomas
(1995: 43).
A terrible tangle has arisen in Pakistan over a local soap opera. Soap star Usman
Pirzada divorced his television wife in traditional Muslim style, pronouncing
Talaq—I divorce thee—three times. The trouble was that his TV spouse was played
by his real wife, Samina. Now the ulemas are saying that the divorce is binding,
even though the formula was spoken in the interest of art. Their decree maintains
that the Prophet ordained that in three matters (marriage, divorce and the freeing of
slaves) words uttered unintentionally or even in jest cannot be withdrawn. Divorced
they are and divorced they must remain.
It has been reported in the media that in some Muslim cultures, husbands now can
even use e-mails and/or text messages to divorce their wives.
roughly ‘I ask/request you to do X for me, and I expect you to do it simply

because of how you are related to me’. Thus, for the speakers of Walma-
jarri, it is very diYcult to refuse a kinship-based speech act of requesting
(see also Wierzbicka 1991: 159–60, Goddard 1998: 370). ‘Exotic’ speech
acts such as kinship-based requesting do not seem to be present in either the
Chinese-type East Asian or the European-type Western cultures.
Second, given a particular speech situation, pertinent speech acts
are carried out diVerently in diVerent languages/cultures. For instance, in
some East Asian and Western cultures, if one steps on another person’s
toes, one normally performs the speech act of apologizing. But apparently
this is not the case among the Akans, a West African culture. As reported
by Mey (2001: 287 crediting Felix Ameka), in that culture, such a situation
does not call for an apology but calls for an expression of sympathy. ‘The
focus is on the person to whom the bad thing has happened rather than the
person who has caused the bad thing’ (Mey 2001: 287). Another example:
while in English, thanks and compliments are usually oVered to the hosts
when leaving a dinner party, in Japanese society, apologies such as o-jama
itashimashita ‘I have intruded on you’ are more likely to be oVered by the
guests. A similar speech act of apologizing is performed in Japanese upon
receiving a present, when a Japanese speaker is likely to say something like
sumimasen—the most common Japanese ‘apology’ formula or one of its
variants (e.g., Coulmas 1981; see also Goddard 1998). Putting the matter
the other way round, as is pointed out by many Japanese scholars, apolo-
gies can be used in a much broader range of speech situations in Japanese
than in English.
Third, in diVerent languages/cultures, the same speech act may meet with
diVerent typical responses. For example, a compliment normally generates
acceptance/thanking in English, but self-denigration in Chinese, Japanese,
or even Polish. A typical compliment response formula in Chinese would
be something like (4.54).
(4.54) A: ni cai zuode zhen hao!
B: nali, nali, wo bu hui zuocai.
A: bie keqi, ni cai zhende zuode hen hao!
B: ni tai keqi le.
A: ‘You cook really well!’
B: ‘No, no, I don’t really know how to cook properly.’
A: ‘Please don’t be too modest. You really cook very well.’
B: ‘You’re too kind.’
122 4. SPEECH ACTS
The same is more true in Japanese. According to Mizutani and Mizutani

(1987: 43), ‘[T]he Japanese will never accept a compliment without saying
iie [‘‘no’’]’. This is vividly illustrated by the following conversation con-
ducted between two women in so-called women’s speech (Miller 1967:
289–90).
(4.55) A: Ma, go-rippa na o-niwa de gozaimasu wa ne. Shibafu ga hiro biro to
shite ite, kekko de gozaimasu wa ne.
B: Iie, nan desu ka, chitto mo teire ga yukitodokimasen mono de gazaimasu
kara, mo, nakanaka itsumo kirei ni shite oku wake ni wa mairimasen no de
gozaimasu yo.
A: A, sai de gozaimasho ne. Kore dake o-hiroin de gazaimasu kara, hito
tori o-teire asobasu no ni datte taihen de gozaimasho ne. Demo ma, sore de
mo, itsumo yoku o-teire ga yukitodoite irasshaimasu wa. Itsumo honto ni
o-kirei de kekko de gozaimasu wa.
B: Iie, chitto mo sonna koto gozaimasen wa.
A: ‘My, what a splendid garden you have there—the lawn is so nice and big,
it’s certainly wonderful, isn’t it!’
B: ‘Oh, no, not at all, we don’t take care of it at all any more, so it simply
doesn’t always look as nice as we would like it to.’
A: ‘Oh, I don’t think so at all—but since it’s a big garden, of course it must
be quite a tremendous task to take care of it all by yourself; but even so, you
certainly do manage to make it look nice all the time; it certainly is nice and
pretty any time one sees it.’
B: ‘No, I’m afraid not, not at all.’
Given the general Japanese reluctance to say ‘no’ under almost any other
circumstances, this compliment response pattern, though considered as
rather traditional by the younger generation in present-day Japan, is
quite striking.
Leech (2003) regarded (4.54) and (4.55) as cases of what he dubbed
‘pragmatic quasi-paradoxes’. Pragmatic quasi-paradoxes arise from the
asymmetry of politeness: what is polite for the speaker (for example,
the speaker invites the addressee to dinner) may be ‘impolite’ for the
addressee (that is, the addressee accepts the speaker’s invitation to dinner).
If both the speaker and the addressee were equally determined to be polite
and not to give way, then this asymmetry would never be resolved, hence
the paradox of politeness (Leech 2003).
However, in an actual speech event, the pragmatic quasi-paradoxes of
this sort are often resolved by means of implicit negotiations between the
speaker and the addressee. Thus, after a few rounds of, say, compliment
followed by ritual denial, the addressee will ‘reluctantly’ accept the com-
pliment, as in (4.54), or he or she will continue to deny the compliment, as
in (4.55). Another example: after a few exchanges of oVer and ritual refusal,
the addressee will ‘reluctantly’ accept the gift oVered by the speaker in
certain cultures (see, e.g., Zhu, Li, and Qian 2000 on Chinese; see also
Barron 2003 on Irish English, about which in the next subsection). This
pragmatic quasi-paradoxical mechanism of oVer–refusal can be schema-
tized in (4.56).
(4.56) A: oVer 1
B: ritual refusal 1
A: oVer 2
B: ritual refusal 2
A: oVer 3
B: acceptance
As our Wnal example, there is the following exchange of insisting on

repaying a debt and ritually refusing to accept repayment between two
male speakers in Persian (Koutlaki 2002: 1747).
(4.57) A: bezær mæn . . . [takes out money] ta færamuš nækærdæm
B: aqa vel kon tora xoda
A: næ næ xaheš mikonæm
B: be vallahe nemixam
A: xaheš mikonæm
B: nemixam
A: pænjai hæm bærat gereftæm æz bank
B: næ nemixam axe
A: ‘Let me . . . er . . . [takes out money] before I forget.’
B: ‘Leave it, really.’
A: ‘Please.’
B: ‘I don’t need it now, really.’
A: ‘Please.’
B: ‘I don’t need it.’
A: ‘I even got it in Wfty pound notes for you.’
B: ‘But I don’t need it I say!’20
20
In terms of conversation analysis, (4.54)–(4.57) are instantiations of what is
called adjacency pairs—a sequence of two structurally adjacent utterances, pro-
duced by two diVerent speakers and ordered as a Wrst pair part and a second pair
part. Taking the compliment response pattern as an example, the Wrst pair part is
124 4. SPEECH ACTS
Fourthly, the same speech act may diVer in its directness/indirectness in

diVerent cultures. Since the late 1970s, a great deal of research has been
conducted on how particular kinds of speech acts, especially such FTAs as
requests, apologies, and complaints, are realized across diVerent languages.
Of these investigations, the most inXuential is the large-scale Cross-
Cultural Speech Act Realization Project (CCSARP) (e.g., Blum-Kulka,
House, and Kasper 1989; see also Blum-Kulka and Olshtain 1984). In
this project, the realization patterns of requesting and apologizing in
German, Hebrew, Danish, Canadian French, Argentinian Spanish, and
British, American, and Australian English were compared and contrasted.
In the case of requests, the Wndings were that from among the languages
examined, the Argentinian Spanish speakers are the most direct, followed
by the speakers of Hebrew. The least direct are the Australian English
speakers, while the speakers of Canadian French and German are posi-
tioned at the mid-point of the directness/indirectness continuum.
Building on the CCSARP, strategies for the performing of certain types
of FTAs in a much wider range of languages have since been examined.
These languages include Catalan, Chinese, Danish, Dutch, German,
Greek, Hebrew, Japanese, Javanese, Polish, Russian, Thai, Turkish, four
varieties of English (British, American, Australian, and New Zealand), two
varieties of French (Canadian and French), and eight varieties of Spanish
(Argentinian, Ecuadorian, Mexican, Peninsular, Peruvian, Puerto Rican,
Uruguayan, and Venezuelan). Recently, there have been a number of new
comparative studies. SiWanou (1992) studied requests in British English and
Greek. Van Mulken (1996) compared the use of the politeness strategies in
Dutch and French requests. Hong (1998) explored requests in Chinese and
German. Fukushima (2002) compared requests in British English and
Japanese. Márquez-Reiter (2000) investigated requests and apologies in
British English and Uruguayan Spanish. Bayraktaroglu and SiWanou
(2001) collected a number of comparative studies of speech act realization
patterns in Greek and Turkish, including advice-giving, correcting, and
complimenting, to mention but a few. Most of these studies adopted a
the complement and the second pair part is the acceptance or rejection. From a
cultural perspective, the main diVerence here is that while the acceptance is the
preferred second pair part in some languages/cultures (e.g., English or Western), the
ritual denial is the preferred second pair part in other languages/cultures (e.g.,
Chinese or East Asian) (Huang 1987).
methodology which is at least partially compatible and comparable with

that used in the CCSARP.
As a result of these studies, it has now been established that there is
indeed extensive cross-cultural/linguistic variation in directness/indirect-
ness in the expression of speech acts, especially in FTAs such as requests,
complaints, and apologies,21 and that these diVerences are generally asso-
ciated with the diVerent means that diVerent languages utilize to realize
speech acts. These Wndings have undoubtedly contributed to our greater
understanding of cross-cultural/linguistic similarities and diVerences in
face-redressive strategies for FTAs.
4.7.2. Interlanguage variation
In the above subsection we looked at cross-cultural variation in speech

acts. In this subsection I turn to interlanguage variation.
Simply put, an interlanguage is a stage on a continuum within a rule-
governed language system that is developed by L2 learners on their path to
acquire the target language. This language system is intermediate between
the learner’s native language and his or her target language. For example,
when a native speaker of Chinese is learning German as a second language,
then the German used by him or her is an interlanguage.
A number of studies have recently appeared that explored speech acts in
interlanguage pragmatics. These include Blum-Kulka, House, and Kasper
(1989), Kasper and Blum-Kulka (1993), Ellis (1994), Trosborg (1995), le
Pair (1996), Baba (1999), Gass and Houck (2000), Achiba (2003), and
Barron (2003). Some of these studies investigated how a particular type
21
The cultural diVerences in directness versus indirectness in the expression of a
speech act frequently lead speakers from one culture to misinterpret speakers from
another culture, as the following incident, reported in Takahashi and Beebe (1987),
shows.
In 1974, President Nixon asked Prime Minister Sato if Japan would agree to self-
imposed restrictions on the export of fabrics to the United States. Prime Minister
Sato replied: ‘Zensho shimasu’, which literally means ‘I’ll take care of it’. However,
when it is used by a politician, the uttering of this expression actually performs the
speech act of a polite refusal in Japanese. President Nixon failed to understand it as
a polite refusal and became very angry later on, when the Japanese simply took no
action at all.
126 4. SPEECH ACTS
of speech act is performed by non-native speakers in a given interlanguage;

others compared and contrasted the similarities and diVerences in the
realization patterns of given speech acts between native and non-native
speakers in a particular language. The best-studied interlanguage is that
developed by speakers of English as a second language. Other interlan-
guages that have been investigated include Chinese, German, Hebrew,
Japanese, and Spanish.
Kasper and Blum-Kulka (1993) contained Wve studies of interlanguage
expression of FTAs, namely, apologizing, correcting, complaining,
requesting, and thanking. Of these studies, Bergman and Kasper’s (1993)
research showed that in performing apologies, Thai learners of English in
general ‘do too much of a good thing’—a phenomenon dubbed as ‘gushing’
by House (1988) and less benevolently as ‘waZing’ by Edmondson and
House (1991). Takahashi and Beebe (1993) observed that American Eng-
lish speakers are most likely, and Japanese speakers of Japanese are least
likely, to use a positive remark before making a correction, with Japanese
speakers of English falling in between. Olshtain and Weinbach (1993)
found that non-native speakers’ complaints in Hebrew are less oVensive
and less face-threatening. Eisenstein and Bodman (1993) pointed out that
while expressing gratitude is a universal speech act, it is carried out diVer-
ently in diVerent languages/cultures. More or less the same is true of
interlanguage hints in requests (Weizman 1993). One of the common
Wndings of the studies collected in Kasper and Blum-Kulka (1993) was
that the expression of interlanguage speech acts tends to be more verbose.
Next, Ellis (1994) focused on the interlanguage FTAs of requesting, apolo-
gizing, and refusing. Trosborg (1995) studied requests, complaints, and
apologies by Danish learners of English. Le Pair (1996) observed that
Dutch learners of Spanish use fewer direct request strategies than native
speakers of Spanish. Baba (1999) made a comparative study of compliment
responses by Japanese learners of English and American English learners
of Japanese. Gass and Houck (2000) studied the speech act of refusal as
performed in English by native speakers of Japanese. Finally, Barron
(2003) took a detailed look at Irish learners’ acquisition of German, espe-
cially of the pragmatics of requests, oVers, and refusals. Many of her
Wndings are of interest. For example, in Irish English, the ritual refusal of
an oVer is typically followed by re-oVers and acceptances, but this dis-
course sequence is not characteristic of German. However, Irish learners
attempt to implement this pattern in German, although this pragmatic

A Course in Pragmatics

Uploaded by

Copyright:

Available Formats

A Course in Pragmatics

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Course in Pragmatics

Uploaded by

Copyright:

Available Formats

University of Sfax

Faculty of Arts & Humanities

By the end of the course, students should:

1- Pragmatics: Definitions & Background

 Huang, Y. (2007). Pragmatics. New York: Oxford University Press.

1.1. What is pragmatics? 1 1.3. Some basic notions in semantics

1.1. What is pragmatics?

Pragmatics is a rapidly growing field in contemporary linguistics. In recent

(1.1) A definition of pragmatics

I shall return to the issue of how to define pragmatics in Chapter 7, where

1.1.2. A brief history of pragmatics

Pragmatics as a modern branch of linguistic inquiry has its origin in the

led to the development of today’s formal semantics. By contrast, within the

1.1.3. Two main schools of thought in pragmatics:

As pointed out in Huang (2001a), two main schools of thought can be

most, if not all, signs have as their interpreters living organisms, it is a

1.2. Why pragmatics?

There are many reasons for including pragmatics in an integrated linguistic

1.2.1. Linguistic underdeterminacy

It is widely accepted that there is a huge gap between the meaning of a

Horn 2004). This is generally known in the literature as the linguistic

(1.2) You and you, but not you, stand up!

The three uses of the pronoun you—called a deictic expression—in (1.2)

1.2.2. Simplification of semantics and syntax

The second reason why we need a pragmatic component is because its

nine girlfriends’. Such a division of labour between pragmatics and seman-

(1.8) Overt NPs

Of the three types of overt NP listed in (1.8), anaphors, pronominals, and

Binding is defined on anaphoric expressions in configurational terms, appeal-

In (1.10a), himself is an anaphor in the Chomskyan sense. As such, it

1.3. Some basic notions in semantics and pragmatics

1.3.1. Sentence, utterance, proposition

(1.11) Lance Armstrong won the centenary Tour de France.

Sentence-meaning, then, refers to those aspects of meaning that are

Of these, (1.15) is an instantiation of the sentence in (1.11). In such a

(1.17) Liszt adored Chopin.

The propositional content of a sentence is that part of its meaning which

Propositions may be true or false (to be discussed below), may be known,

Regarding the relationship between sentence and proposition, the same

Try putting these principles into practice by doing Exercise 1 on p. 18.

Finally, the relationship between sentence, utterance, and proposition

3.2 DEICTIC VERSUS NON-DEICTIC EXPRESSION

Deictic usage Non-deictic usage

Deictic use of this, that, these and those

Adverbs of time and space

Deictic usage Non-deictic usage

Deictic usage Non-deictic usage

3.3 GESTURAL VERSUS SYMBOLIC DEIXIS

Gestural usage Symbolic usage

Gestural usage Symbolic usage

A: Are you goin with daddy?

Rank order LCIE Frequency per million words

A: But Goldie’s a girl’s name like.

A: Come over here to Nana. Over here to Nana.

(3.21) [Context: family at the dinner table]

A: Is there more meat there?

(3.22) [Context: ending sequence in a radio phone-in show]

(3.23) [Context: friends chatting]

A: Was he in better form yesterday?

(3.29) [Context: mother and daughter chatting about an up-coming wedding]