Phatic interpretations and phatic communication*
V L A D ŽE G A R A CUn iv er si ty of Lu to nandB I L L Y C L A R KMi dd le se x Un iv er si ty
Absrtact
This paper considershow the notion of phatic communication can best
be understood within the framework of Relevance Theory. To a large
extent, we are exploring a terminological question: which things
which occur during acts of verbal communication should the term
‘phatic’ apply to? The term is perhaps most frequently used in the
phrase ‘phat ic comm uni on’, whic h has been thought of as an
essentially social phenomenon and therefore beyond the scope of
cognitive pragmatic theories. We suggest, instead, that the term
‘pahtic’ should be applied to interpretations and that an adequate
account of phat ic inte rpret ations requires an account of the
cognitive processes involved in deriving them. Relevance Theory
provides the basis for such an account. In section 1, we indicate the
range of phenomena to be explored.In section 2, we outline the parts
of Relevance Theory which are used in our account. In section 3, we
argue that the term ‘phatic’ should be applied to interpretations, and
we explore predictions about phatic interpretations which follow from
the frameworkof Relevance Theor y, including the claim that
phati c interpreta tions should be derived only when non-phatic
interpretations are not consistent with the Principle of Relevance. In
section 4 we consider cases where cognitive effects similar to those
caused by phatic interpretations are conveyed but not ostensively
communicated.
2
1. Introduction
The following fictional exchanges show that people in general
are aware of an informal distinction between ‘chit-chat’ and
‘genuine’ conversation:
(1) Groundhog Day
ML: I hope you enjoy the festivities.
PC: Oh I'm sure I'm going to (pulls face).
ML: There's there's talk of a blizzard.
PC : We ll , we ma y ca tc h a br eak and th at bl izza r d' sgonn a blow right by us. All of this mois turecoming up outa the south by midday is probablygonna push on to the east of us and at high altitudesit's gonna crystallise and give us what we cal lsno w. (winks ) Pro bab ly be som e acc umulat ion .But here in Punxsutawney our high 's gonna get upto about 30 today, teens tonight, chance ofprecipitat ion about 20 per cen t today 20 per cen ttom orrow . Did you wan na tal k about the weatheror were you just making chit-chat
ML: (shrugs and shakes head) Chit-chat.
PC: OK. Right. See you later. B'bye.
ML: Oh em eh will you be checking out today Mr.
Connors?
PC: Chance of departure today one hundred per
cent. Speakers: ML = Mrs. Lancaster; PC =
Phil Connors.
(Gr oun dhog Day 1993 , dir . Harold Ram is, scr een pla y Dan ny
Rub in and Harold Ramis)
(2) Glengarry Glen Ross
RR: How are you
GA: Fine („) you mean the board (,) you mean you
mean you mean on the board
3
RR: I (,) yes (,) okay (,) the board
GA: I'm fucked on the board
Speakers: RR = Ricky Roma; GA = George Aaronow.
(Glengarry Glen Ross 1992, dir. Tony Foley, screenplay David
Mamet) Conventions: (,) = short pause; („) = long pause.
Contex tual assumptions: Roma and Aaronow are real estate
salesmen whose sales perfo rmance is being recorded on a board
in their office. At the end of the week, all salesmen except the top
two on the board are to be fired. Roma is generally the most
successful salesman and is currently top on the board. Aaronow's
sales performance is poor; a fact which he believes is mainly due
to the fact that he has been given bad ‘leads’.
Al thou gh th es e ex ampl es are from fict io n, th ey are no t
pe rcei ve d as particularly idiosyncratic or unusual. Rather, they are
readily understood as repre senti ng typ es of excha nge whi ch do
occ ur in real -li fe sit ua tio ns . Consequently, we ignore their
fictitious character here. (A number of real-life exchanges which
fur the r ill ust rate the ana lys is of pha tic communication
developed here are considered in Žegarac 1998).
In (1), it seems that Mrs. Lancaster would like to have a relatively
banal conversation about the weathe r. Phil Connors makes clear
that he recognises this intention and blatantly refus es to comply.
In (2), Aaronow at first gives a shor t, ster eotypical answ er. He
then follows it with a ques tion abou t the point of Roma's
utterance.
The se exc han ges ill ust rat e the sor ts of pheno men a to whi ch
the ter m `phatic' migh t be appl ied. It migh t be said that Mrs.
Lancaste r attempts to start a phatic exchange and that Connors
refuses. Another account might say that both Mrs . Lan caster 's
comment about the bl izz ard and Connors's lengthy response are
phatic, the difference being that Mrs. Lancaster's phatic intenti ons
are `pos itive' whi le Connors's are `negative'. An acco unt of (2)
4
might say that Aaronow first entertains the `phati c meaning' of
How are you and then considers an alternative interpretation.
Thi s rai ses a number of questi ons . How does Con nor s
rec ogn ise Mrs . Lancaster's phatic intention? How does he decide
how much to say about the weather? Is Roma's utterance ambiguous
between a phatic and a non-phatic mea ning? Why do es Aar onow
fir st cho ose the inter preta tio n on whi ch Roma's utterance is a
mere greeting? Why does he then reinterpret Roma's utteran ce? On
what grou nds does Aaronow decide how much informa tion
abou t his welf are he should give to Roma? Comparing
Connors 's response with Aaronow's , why does Connors's over -
informativeness communicate something dif ferent (hi s att itude
towards Mrs . Connors's chi t -cha t) from Aaronow's over-
informativeness (perhap s jus t that he has misunders tood,
perhaps that he wishes to change the topic)? More fundamentall y,
what does the term ‘phatic’ mean and to what range of phenomena
should it apply?
In this paper we adopt the framework of Relevance Theory and
consider to which phenomena within that theory it would be most
usefu l to apply the term ‘phatic’. We believe both that Relevance
Theory sheds light on these ph enom ena and th at these
ph enomena po se in te rest ing qu es tions fo r Relevance Theory.
In the next section we introduce the features of Relevance Theory
which are used in our account.
2. Relevance Theory
The Rel eva nce The ory accou nt of ver bal com mun icati on is
bas ed on a particular definition of OSTENSIVE-INFERENTIAL
COMMUNICATION.Our account of phatic phenomena depends
particularly on the definition ofOSTENSIVE COMMUNICATION
which is outlined in the first part of this section. The secondpart of
this section presents techn ical terms within Relev ance Theory
which refer more specif ically to the sorts of things that can be
communicated by an act of ostensive communication.
5
2.1 Ostensive communication
All human behaviour is informative in the sense that it provides
evidence for inferen ces . Some info rmative behaviour is
intenti onal ly informativ e, and som e inten tionall y informati ve
beh avi our is evi den tly int ent ional ly in format ive. Within
Relevance Theory, this last type of inform ative behaviour is
termed ‘ost ensive-inf erenti al communica tion ’ (‘ostens ive ’ in
tha t communicators make manifest an intention to communicate,
‘inferential’ in that understanding it involves inferring the
communicator's intention). Ostensive-inferential communication
involves two intentions:
(a) a communicative intention to make mutually manifest:
(b) an informative intention to make manifest or more manifest a
set of assumptions.
Cons ider, for example, a speaker who begi ns a conv ersa tion in
Engl ish, then suddenly, and without apparent reason, switches to
another langu age, say French. This change of lan guage code
provides evidence for (i.e. makes (mo re) man ife st to the
aud ien ce) a ran ge of ass umptions. Som e of the se assumptions
may be conveyed regardless of whether the speaker intended to
convey them (e.g. ‘The speaker can spe ak French ’). But the re
are othe rs which will not be conveyed unless the change of the
language code is invested with a communicative intention (e.g. ‘The
speaker wants some people in the audience to leave’). That is to
say, hearers will not recover these assumptions unless they think
that the reason for the code-switching is an intention to communicate
something.
The notions of MANIFESTNESS and MUTUAL
MANIFESTNESS are important for the relevance -theoretic
account of ostensive communicat ion in general, and of verbal
communication in particular, so it is worth making clear how
these terms are understood with in Relevance Theory. An
6
assumption is mani fest to an ind ividua l ‘if the env iro nme nt
pro vid es suf fic ien t eviden ce for its adoption’ Sperber &
Wil son (1986: 39) . There is an import ant diff erence between
the Relev ance Theory notion of manifestness and everyd ay uses
of this term (and of close equivalents such as ‘evident’ or
‘salient ’). In everyday use, to say that something is ‘manifest’ may
imply that it is actually mentally represented. It is important to note
that the technical notion of manifestness int roduced her e does not
entail actual men tal rep resentation, because an assumption may
be manifes t without bein g entertaine d. For exam ple , the fact
that somebody has spoken provides evidence that they are alive, but
this assumption need not be mentally represented. In other
words, mani festness is not a notion of an intr insically mental
state or property. Rather, it refer s to a psych ological
di sposit ion which depen ds part ly on aspects of the
environment.
Another important not ion is that of MUTUALITY. All
assumptions which ar e ma ni fe st to an in di vi du al ma ke up
th at in di vi du al 's CO GN IT IV E ENVIRONMENT. The set of
all assumptions that are manifes t to two individuals is their
SHARED COGNITIVE ENVIRONMENT. But an assumption may
be manifest to a number of people, without the fact that it is
manifes t to all of them being itself manifest. In other words,
assumptions may be MANIFEST without being
MUTUALLY MANIFEST.
It fol low s from wha t we hav e sai d abo ut the (mutua l)
man ife stn ess of ass ump tio ns, tha t cog nit ive env iro nme nts
may be sha red wit hou t bei ng mutual . A shared cognitive
environment in which it is manifes t which people share it is what
Sperber and Wilson (1986: 41) call a MUTUAL COGNITIVE
ENVIRONMENT.
The not ion of mutual manifestness provides the basi s for the
relevance-th eo re ti c expl an at ion of our ab il it y to in te rp re t
7
ut te ranc es alon g th e communicator's intended lines (cf. Sperber
and Wilson 1986: 15-21 & 38-46). Sperber and Wils on do not
propose that to info rm somebody of some thin g is to cau se tha t
per son to ent ert ain spe cif ic tho ugh ts. Rat her , Rel eva nce
Theory describes the informat ive in tent ion as `an in tent ion
to modify dir ect ly not the thoughts but the cognit ive
environment of the aud ience' (Sperber & Wilson 1986: 58).
This characteri sation provides a natural explanation for the
intui tion that there are degrees of communicat ion, that
information is communicat ed with greater or lesser strength.
Weak communication involves a small increase in the deg ree of
man ife stn ess of a par ticula r set of assump tio ns. Whe n the
parti cular set of communicat ed assumptions is made highl y
manifest by the ac t of os te ns io n, th ey ar e st ro ng ly
co mm un ic at ed , an d ar e me nta ll y represented. For example,
the utterance "I am allergic to fish", as a respon se to an offer of
fish and chips, would strongly communicate that the speaker
doesn' t want to eat fish, whi le weakly communica ting other
assumptions, perhaps rai sin g questions about other possib le
all erg ies , previous eat ing experiences, and so on.
An imp ort ant pro per ty of ost ens ive commun ica tio n, whi ch is
of gre at social importance, is that the act of ostension alters not only
the cognitive environment of the audience, but themutual cognitive
environment of the communica tor and audience. As Sperber &
Wilso n (1986 : 61-62) point out: `A change in the mutual
cognit ive environment of two people is a change inthei r
poss ibil itie s of inte ract ion (and , in part icul ar, in thei r
poss ibiliti es of fur the r communica tion). ' In successfu l (and
sincere) ost ens ive com municatio n the communicat ed set of
assumptions is made not merel y mani fest or more manifest to the
audience, but MUTUALLY manifest to the communicatorand
audience. For inst ance, the social significance of establishing a
mutual cognitive environment explains why addressees often take
the trouble to communica te their acceptance of a parti cular set of
8
assumptions. In Sperb er & Wilson's words :
... ostensive communication could be described as an attempt
to create a genui nely mutu al cogni tive envi ronment
between social personae . When the communica tor is
sincere (and so is the aud ience in manifesting its
acceptance of the information communicated), then the
actual individuals and thei r social personae coincide , and
otherwise they don' t. (Spe rber & Wilson 1986: 258, fn. 32)
In all ostensive communication, the act of ostension directs the
audience's attent ion in a par ticula r way, causin g the audience to
access some set of ass ump tio ns whi ch con sti tut e the ini tia l
con tex t in whi ch the ost ens ive sti mul us is pro ces sed . If the
int erp ret ati on of the st imu lus in the ini tia l context does not give
rise to enough effects, the context is enlarged to include new
assumptions, un ti l an adequa te in te rp re ta ti on is ar rived at .
This inferen tial proce ss of int erp ret ing an ut ter anc e in
con tex t is gui ded by considerations of relevance. More precisely,
it is constrained by the PRINCIPLE OF RELEVANCE (in a more
recen t formulation (Sperber & Wilson 1995), this is te rmed the
`Second Pr inciple of Relevance' or the `Communicative
Principle of Relevance').
The (Communicative) Principle of Relevance
Every act of ostensive communication communica tes the
presumpt ion of its own optimal relevance.
Presumption of Optimal Relevance
(a) The ostensive stimulus is relevant enough for it to be worth
the addressee's effort to process it.
(b) The ostensive stimulus is the most relevant one compatible
with the communicator's abilities and preferences.
(Sperber & Wilson 1995: 260-270)
In a nutshell , relevance is a posi tive function of contextual
effec ts and a negative function of the process ing effort required
for deriving those effects. The interpreta tion of an ostensive
I3
9
stimulus is adequate when the contextual effect s ach ieved off set
the processin g effort requir ed for der iving them. Consequently,
the appropriate context for the interpretation is the one which
inc lud es the min ima l num ber of ass umptions req uir ed for
arr ivi ng at an interpretation consistent with the Principle of
Relevance.
2.2 What can be ostensively communicated
According to Relevance Theory, the list of things that can be
communicated by an act of ost ens ive commun ica tion i ncl ude s:
log ica l forms (less than propos itional, lin guisti cal ly-decoded
meanings) , propos itions expressed (fully propositional
representations derived by inferentially enriching logical forms) ,
expl icatures (proposi tional forms which the speaker
mutuall y manifestly intends to convey, some of which may be
embedded within each other), implications (propositions which
logically follow from the union of proposit iona l forms conv eyed
and cont extual assumpti ons, whet her or not the speaker intends to
convey them) and implicatures (implications which the speaker
mutual ly manifestly intends to convey). Note that we are making
no claims about the order in which any of these might be derived
during actual processing (for example, we are not claiming that
hearers cannot recognise an impli cature before fully working out
the proposition(s) expres sed). For a more detail ed account of all
of these terms, see Sperber & Wilson 1986. For an account of the
explici t-implici t distinction in particular, see Carston 1988. For
reasons of spac e, we simp ly illu strate thes e term s by suggesti ng
some things which might be conveyed by the utterance in (3) given
certain contextual assumptions (obviously, the following discussion
simplifies greatly).
10
(3) This beer is flat.
Contextual assumptions: The speaker is Arthur.
Arthur is in his local pub.
The hearer is Grant.
Grant is the landlord.
Art hur is holdi ng a pint of
Chu rch il l's bi tter which he has just
bought from Grant.
Logical form 1:
______________ is flat1___________________________________
(referent of this beer) (at or under X where X is a time or set of conditions)
Logica l form 2:
______________ is flat2___________________________________(referent of this beer) (at or under X where X is a time or set of conditions)
(f la t1 = ‘no t fiz zy ’; fl at2 = ‘leve l’)
Propos it ion expressed: The beer Arthur is hol ding is
flat1.
Expl icature 1: Arthur be li eves that the beer
Ar thur is hol ding is fla t 1.
Expl icature 2: Ar thur is com municating to Grant
that the beer Arthur is hold ing is
flat1.
Expl icature 3: Ar thur is in tendi ng to
communicate to Grant that the
beer Arthur is hol ding is flat1 .
11
Impl icat ion 1 : Ar thur does not want to dr ink the beer he
is holding.
Impl icat ion 2 : Arthur expects a refund or a new pint of
beer .
Implication 3: Grant is responsible for the flat beer.
Implication 4: Grant should make sure that he does not serve
flat beer in the future.
Implication 5: Arthur has spoken to Grant.
Implication 6: Arthur is alive.
Impl icature 1: Ar thur does not want to dr ink the beer he
is holding.
Impl icature 2: Arthur expects a refund or a new pint of
beer . Grant is responsible for the flat beer.
Im pl ic at ur e 3 : Gran t is resp onsibl e fo r th e fl at beer .
Th is is fa r from a comp le te accoun t of what migh t happ en
if Ar th ur produced (3) given the suggested contextual assumptions.
In particular, notice that we have not attempted to give an exhaustive
list of possible logical forms, propositions expressed, etc. However,
there are some important points which this illustrates and which will
be used in our account of phatic interpretations . First, not ice that
the utteran ce impl ies many assumptions which are not
intentionally communicated and therefore not implicated, i.e.
implicat ions which are not implicatures (e.g. implication6).
Second, some of the assumptions conveyed contain other
assumptions conveyed as sub-parts (e.g. explicature3 contains
explicature2 which contains explicaturez). The di st in ct ion
be tw een impl ic at ions and im pl ic atur es , and th e fact th at
assumpt ions conveyed can be embedded within each other, will be
important in our account of phatic interpretations.
12
3. Phatic Interpretations
While there is a considerab le liter ature which, in one way or
anoth er, looks at soci al aspects of verb al communi cation (for a
survey of such work , see Schi ff rin 1994), the cogni tive ef fect s
of such phenomena hav e not been studied in much detail. We
agre e with Sperber's (1983, 1985) view tha t any acco unt of how
soci al fact ors affe ct beha viour must incl ude an account of how
such behaviour is cogni tively mediated. There fore, we have
decid ed to adopt a cognit ive the ory and exp lor e wha t thi s
the ory tel ls us abo ut the phenomena we are inte rested in. In this
sect ion, we spel l out the intu itions about phat icness that we
hope to capt ure, prop ose our relevanc e-theo reti c account and
explore some of its implications.
3.1 Intuitions
Malinowski (1923) considered PHATIC COMMUNION in his
discussion of the distinction between language as ‘an instrument of
reflection’ and language as ‘a mode of action ’, and made a number
of interes ting observations, including the following which are
frequently referred to in more recent literature:
1. In Phat ic Commu nion language is used as a mode of action,
rathe r than for the transmission of thoughts.
2. The various types of Phati c Communion (gree tings , gossip,
and the like) have something in common: the whole situation
in which the exchange ta ke s pl ace cons is ts in , an d is
la rgel y cr ea te d by , `w ha t happ ens linguistically'.
3. In Ph at ic Co mm un io n th e me re me an in g of th e wo rds is
al mo st irrelevant. Rather, the linguistic expressions used fulfil a
social function.
4. This social function may be to `overcome the strange, unpleasant
tension caused by silence' and/or to est abli sh an atmosphe re
of sociabil ity and personal communion between people.
13
Jakobson (1960) characteri sed the PHATIC FUNCTION as the
use of language to fo cu s on th e ch anne l of co mm un ic at io n
it se lf , ra th er th an on th e information conveyed by the language
code. He pointed out that prolonged phatic exchanges sometimes
occur precisely when the communication process is threatened (for
instance, by the insecurity of the interlocutors).
Laver (1974) also takes up Malinowski's views and describes in
some detail the connection between the relative social status of the
interlocutors and the appropriate choice of lingui stic expression in
a phatic exchange. He sugges ts that the fundamenta l social
func tion of phat ic co mmunicat ion is `the det ail ed
man age men t of in ter perso nal rel ati onships durin g the
psych ologically crucial margins of the interaction' (p. 217).
The re are thr ee com mon int uit ion s abo ut pha tic nes s whi ch
we aim to capt ur e. Fi rs t, peop le ha ve an in tu it io n th at th e
ma in po in t of so me utt erances dep ends on the fac t tha t the
speake r has said someth ing to the hearer more than on exactl y
what has been said. For example, the main point of an utterance of
the string nice weather directed at a stranger at the bus-stop see ms
to be to con vey som eth ing like socia bil ity rat her than to start
a discussion of the weather. Furthermore, uttering a string with quite
different linguisti c cont ent , such as the bus is late agai n, woul d
hav e had a simila r effect. Second, there are degrees of phaticness .
For example, a string like how do you know Mich ael? seems `les s
phatic ' than a stri ng like how are you? whe n utt ered in the
sam e sit uat ion (e.g. to som eon e you 've jus t met at a party) .
The phaticness of how are you? may even have become
standardised . (For a more deta iled discussion of standardisat ion
and conventional isat ion see Žegarac 1998, Nicolle & Clark 1998.
Our main aim here is to characteri se and def ine pha tic
com mun ica tio n. Sin ce thi s type of lan guage use is, in
principle, indepe ndent of standardisat ion and conventiona lisation,
we do not discuss them more fully here. ) Third, phatic
interpretations seem more likely when the socia l relat ionsh ip
14
between interlocutors is in doubt . Suppose, for example, that two
partners in a long-term relationship have had an argument and
then not spoken for a few days. This wou ld make it more
likely tha t anyth ing eithe r of them says to the other would be
under stood as phati c (the fact that they have spoken is more
importa nt than exactly what they say). This latter possibility is
discussed in more detail below.
3.2 Interpretations
We prop ose to capture these intu itions by suggest ing: firs t, that
the thin gs which can be phati c or not are interpretations as a
whole; second, that what mak es an int erp ret ati on phati c or not
is the ext ent to whi ch it con tai ns impl icatures of a part icular
type. We begin this sect ion by explaining what type of
implicature we have in mind.
Rec all tha t, as men tio ned abo ve, ost ens ive-inf ere nti al
com mun ica tio n involves two intentions :
(a) a communicative intention to make mutually manifest:
(b) an informative intention to make manifest or more manifest a
set of assumptions.
This means that the speaker of any utterance will make mutually
manifest assumptions which contain other assumptions as sub-parts.
(4) It's ten o'clock.
The speaker of (4), for example will convey all of the assumptions in
(5):
(5) (a) It's ten o'clock.
(b) The speaker intends to make manifest that it's ten o'clock.
(c) The speaker intends to make mani fest the speaker's
intention to make manifest that it's ten o'clock.
(5a ) is the PRO POS ITION EXP RES SED . (5b ) is the spe ake r's
INF ORM ATIVE INTENTION. (5c) is the speaker's
COMMUNICATIVE INTENTION.
Now, some impl ica tions of an utteran ce wil l depe nd on the
15
proposition expressed. In this case, these might include:
(6) The film the hearer wants to watch begins in 10 minutes.
Other impl icat ions might depend on the speaker 's informat ive
intention. In this case, these might include:
(7) The speaker intends to inform the heare r that the film the
heare r want s to watch begins in ten minutes.
Sti ll oth er impli cat ions might dep end on the speak er 's
com mun ica tive intention. In this case, these might include:
(8) The spe ake r wan ts the hea rer to thi nk tha t the spe ake r
car es abo ut whether the hearer sees the film or not.
Notice that the heare r can derive (6) from (5a) alone, while ('7)
depends upo n der ivi ng (5b ) and (8) dep end s upo n der ivi ng
(5c ). The re are oth er implications of this utterance which depend
on the speaker's communicative intention (5c) but not upon
linguistically-encoded meanings, for example (9).
(9) The speaker is willing to communicate with the hearer.
In the discussion which follows, we will use the phrase `depends
on' in a tech nical sens e. DEPENDS ON X mean s `results from
an infe rent ial proc ess which takes X as a premise', where X may
be: the proposition expressed by the utterance, the informative
intention or the communicative intention.
(9) seems to correspond to what people have in mind when they
talk about phaticness. We suggest that interpretations are phatic to
the extent that they IMPLICATE (i.e do not merely IMPLY) such
propositions. In other words, an in te rpre ta tion is phat ic to the
ex tent that it s main re levance li es wit h imp lic atu res lik e (9) .
The re are two imp ort ant cha rac ter ist ics of phati c implicatures.
First, Relevance Theory distinguishes between IMPLICATED
PREMISES and IMPLICATED CONCLUSIONS:
(10) Paul ine: Do you want to go to the cinema?
Arthur: There's only violent films on tonight.
16
Impli cated premise: Arthur does not want to see a violentfilm.
Implicated conclusion: Arthur does not want to go to thecinema.
Impl icated premises depend on the commun icat ive in tent ion
and are normally not impl ied, not even contextually, by the
expl icit content of the utterance. It is the fact that Arthur can be
seen as having the communicative int ent io n of ans wering
Pau line 's quest ion that licen ses the imp licat ed pre mis e,
whi ch in tur n lic ens es the imp lic ate d con clu sion. But it is the
implicated conclusion which meets Pauline's expectation of
relevance. In this sense, the main relevance of Arthur's utterance can
be said to depend on the implicated conclusion rather than on the
implicated premise. So, we are sug ges tin g tha t int erp ret ations
are pha tic to the ext ent tha t the y con tai n implicated conclusions
which depend on the communicative intention.
Se co nd, by de fi ni ti on , al l im pl ic at ur es de pe nd on th e fa ct
th at a presumption of relevance is being communicated. What
distinguishes phatic impl icatures is that they depend to a grea ter
extent on the communicativ e intention than on the proposition
expressed by the utterance. We are now in a position to give a
definition of phatic interpretations.
Phatic interpretation
An int erp ret ati on is pha tic to the ext ent tha t it con tai ns
imp lic ate d co nc lu si on s wh ic h do no t de pe nd on th e
ex pl ic it co nt en t of th e utterance.
Note that on this definition phatic interpretations are not wholly
independent of linguistic all y-encoded meanings, but they do not
fol low directl y from them : the explicit content of the utterance still
provides evidence for some implicated premises which, jointly with
other contextual assumptions, license a particular phatic
interpretation.
Whil e we suggest that the term ‘phat ic’ is most useful as a
technica l term defined in this way and applied to interpretations as
17
a whole, we will also use it derivatively in the following ways :
(i) A PHATIC UTTERANCE is one which gives rise to, or is
intended to give rise to, phatic interpretations.
(ii) PHATIC COMMUNICATION refers to acts of ostensive
communication which give rise to, or are intended to give rise
to, phatic interpretations.
To illustrate more fully, let us consider how the following string
might be interpreted given three different sets of contextual
assumptions :
(11) Pauline : There's a red gas bill.
Contextual assumptions (1):
(i) Pauline and Arthur are having breakfast.
(ii ) They have been running throu gh a list of things that have to
be done today.
When Pauline uses (11) given these contextual assumptions, will
Arthur in te rp re t he r ut te ra nc e as ph at ic or as no n -ph at ic ?
In tu it iv el y, th e inte rpre tati on may go either way (or both
ways). Arthur may well assume tha t Pauline has ment ioned the
bil l, not because she assu mes he does not know about it or will
forget to pay it, but because she wants to let him know tha t she
wou ld lik e to con tin ue the con ver sat ion . In thi s case, Pau lin e's
utte rance is rela tive ly phat ic. Alte rnat ively, Arth ur may think
that Paul ine has ment ione d the red gas bil l simp ly because it
happens to be one of the things that need to be done during the
day. In this case, Paul ine's utte rance is relatively unphat ic. But
Arthur may think that it is highly manifest to both of them, and
mutually manifest, that the bill needs to be paid and that he had
mentioned on som e pre vious occasion tha t he would pay it. In
thi s case, Arthur may still think that Pauline has mentioned the
red bill because she is runni ng through the list very thoroughly,
but he may also assume that she wants to convey her des ire to
continue the conversati on with him. In thi s case, the utterance is
both phatic and non-phatic. But, in this contex t neither the phatic
18
nor the non-phatic interpretation is likely to be very salient.
In our terms, the first interpretation involves a number of
implicatio ns derived by exploiting evidence for assumptions such as
(12),
(12) Pauline has spoken to me.
which arise because Art hur has recognised tha t Pauline is
talking to him (i.e. ostensively communicating with him); the
second interpretation involves a number of implications der ived
by exp loi ting evid ence for assumptions such as (13a–b),
(13) (a) There is a red gas bil l whi ch has arr ive d at our
house.
(b) Red gas bills arrive when the gas will be turned off if we
don't pay within seven days.
which arise becau se Arthur has contextually enriched a logical
form of the ut te ra nc e an d wo rk ed ou t so me of it s
im pl ic at io ns ; an d th e th ir d interpretation involves a number of
each type of implicat ion. In othe r words, the assumption in (12)
depends upon the communicative int ent ion rather than
linguistically -encoded meanings, while the assump tions in (13)
exploit the linguistically-encoded meanings.
Contextual assumptions (2):
(i) Pauline and Arthur are having breakfast.
(ii ) Followin g a majo r row, Paul ine and Arth ur haven't spok en
for three days.
Given these assumptions , the very fact that Pauline has spoken is
so relevant tha t Arthur is unl ikel y to pay much attenti on to the
linguisticall y-der ived cont ent of the utt era nce . He wil l
pro bab ly der ive imp lic ati ons suc h as: `Paul ine wishes to
communicate with me', Paul ine wishes to make up with me',
Paul ine stil l loves me', `Paul ine might go on a trip with me', etc.
What all these assumptions have in common is that they are
19
implicatures derivable from the act of ostension itsel f, i.e. from
the fact that Pauline has spoken to Arthur.
Contextual assumptions (3):
(i) Pauline and Arthur are having breakfast.
(ii) Pauline and Arthur hav e no mon ey, are in ser iou s debt and
have no means of raising money.
The linguis tically-derived content of Pauline 's utterance is highly
relevant in this context and may give rise to implicatures like the
following: `Pauline and Art hur 's gas wil l be cut off if the y don 't
pay soo n', `Pauli ne and Art hur might need to find anothe r way
to hea t the house' , `Pauline and Arthur 's lives might become
more difficult', ` specific aspects of Pauline and Arthur's lifestyles
are under threat', and othe rs. In this context, the act of ostension,
the fact that Paulin e has spoken , is probab ly not relevant enough
to give rise to any imp lic atu res , eit her pha tic or non -pha tic .
Thi s is bec aus e of the exist ence of other, more salient and more
relevant impli catures. Arthur may derive some phatic
implicatures (derivable from the act of ostens ion alone) , but
Relevance Theory predicts that these can not be taken to be what
Pauline intended to constitute the main relevance of her utterance.
This suggests that, other things being equal, phatic interpretations
are derived only if non-phatic interpretations are not cons ist ent
with the Principle of Relevance; a point explored in more detail
below.
Notice th at the type of impl icature used in our de fini tion
of phat ic interpretations does not necessarily have all of the
features usually attributed to phatic communication. For example,
as mentioned above, one assumption made mani fest by any
utte rance or other act of ostensive communication is that the
communic ator is aliv e. In a context wher e this is rele vant ,
Arthur ma y de ci de th at Pa ul in e' s ut te ranc e is in te nded to
co mm unic at e th is assumption . In such a case, `Paul ine is alive '
is an impli cature derived from the act of ost ens ion its elf ; but it
20
has non e of the soc ial fea tur es usu all y claimed to be typical of
phatic communication.
How doe s thi s acc oun t for the int uit ion s abo ut pha tic nes s
men tio ned above? The firs t intu ition was that some
inte rpre tations depend on the fact that something has been said
rather than on exactly what has been said. Our account handles this
intuition in terms of a slightly different, and theoretical, distinction
between the fact that something has beenCOMMUNICATED and
the fact that something has been LINGUISTICALLY ENCODED.
One consequence of this is that our account extends naturally to non-
verbal communication. For example, we can distinguish a relatively
phatic from a relatively non-phatic interpretation of a nod of the
communicator's head. The relatively phatic inte rpret ation is one
where most of the implicatures depend on the fact that the nod
conve yed a communicat ive inten tion. The less phati c
interpreta tion is one where there are more impl icatures which
depend on the fact that the nod signifies agreement.
The sec ond intuit ion was tha t the re are deg rees of
phaticness. Our definition both captures the notion of degrees of
phaticness and accounts for the vagueness of thi s intuit ion . It may
be possib le in cer tain cases to characterise a particular
interpretation as extremely phatic (e.g. if all its implicatures depend
upon the communicative intention but not on what is linguistically
encoded). In other cases, it may be possible to characterise an
interpretation as not phatic at all (e.g. if all its implicatures
depend upon what is linguis tically encoded). However, it will not
always be possible to rank a pair of interpretations in terms of their
degree of phaticness.
The third intuit ion was that phatic interp retations become more
likely when the social relationship between the interlocutors is in
doubt. This intuition has been partly accounted for by our discussion
of (11) given the second set of contextual assumptions above (where
Pauline and Arthur have had a row and not spoken for three days).
21
Our account presupposed that these contextual assumptions made
phatic interpretations more relevant (and non-phatic assumptions less
relevant) than they would otherwise have been. In fact, Relev ance
Theor y makes a more precise predi ction, namely that purely
phatic interpretatio ns should arise only when non -phat ic inter-
pretations are not consistent with the Principle of Relevance. This
prediction is explained in the next section.
3.3 Predictions
According to the criterion of consistency with the Principle of
Relevance, the speaker communicates the presumption that the
utterance she has used is the best one she could have chosen in order
to communicate the particular set of assumptions that she intends to
communicate. Consequently, the normal way for the hearer to
proceed is to try to derive the intended effects by exploiting the
linguistic meaning of the utterance. Essentially, this is because
processing linguistic forms involves the expenditure of effort, and the
second part of the presumption of optimal relevance rules out the
possibility that the speaker is putting the hearer to gratuitous effort.
Under what circum stances should the heare r go beyond
linguistically -derived meanings of the utterance? Relevance
Theor y predicts that this should happen only if linguistically-
derived meanings manifestly fail to yield enough effects for the
criterion of consistency with the Principle of Relevance to be satisfied.
An utterance of (I I) (Pauline's `There's a red gas bill') in context 2
(following a major row, where the interlocutors have not spoken for
three days) illustrates the extreme situation in which, regardless of
any linguistically-encoded meanings, the fact that the speaker has
spoken is far mor e rel evant than wha t is actual ly sai d. Whi le
lin gui sti cal ly -der ive d meanings do not have many contextual
effects in the immediate context, the act of ostension does. In this case
the speaker does not need to take great care to choose an utterance
which makes evident what type of evidence should be exploited.
This is so because it is mutually mani fest to Paul ine and Arthur
22
that it is highly mani fest to each of them that they have
quarrelled and have not spo ken for thr ee days. Giv en Pau lin e
and Arthur 's mutual cognit ive envi ronment, almost any
utte rance produced by either of them will make it mutually
mani fest that most of its contextual effec ts are intended to
follow from the act of ostension, and not from any lin guist icall y-
encoded meani ng of the words used.
But what about phat ics in less biased sett ings? For example,
what would Pau line have to say in con tex t I (wh ere Pauli ne
and Art hur are hav ing break fast and running through a list of
things to do) in order to indica te that the mai n rel eva nce of her
utt era nce lie s wit h the act of spe aki ng its elf ? Relevance Theory
clearly predicts that Pauline should use an utterance which makes it
highly manifest to Arthur that it is relevant in that particular way.
Othe rwis e, Arthur will be misl ed and will expend cons iderabl e
proc essing effort exploiting the meaning of the words used.
Consequently, the phatic inte rpretation will be too effortfu l to be
worth cons ider ing. So what makes an utt era nce lik ely to give
ris e to pha tic int erp retati ons? The ans wer we suggest is that
phat ic inte rpre tati ons are like ly for utte rances containing a
lingu istic form which has the following prope rties : (a) it is easy
to process; (b) it is mutual ly manifest to the interlocut ors in
what kind of cont ex t the linguis tic meaning of the utteran ce
would be high ly relevant, and (c) it is mutual ly mani fest that
the speaker could not have in tended the main relevance of the
utterance to lie with its linguistic meaning on this particular
occas ion. This is a rather loose characteri sation, but it reflects
our view that the phaticness of an utterance as a whole is a matter of
degree, and that, by definition, no utterance can be entirely non-
phatic.
To see that the less processing effort a linguistic expression
imposes on the hearer the more suitable it will be as a phatic,
consider (14)--(17).
23
(14) How are you?
(15) How are you today?
(16) How are you these days?
(17) How are you now that you've had the operation?
In most readily conceivable contexts, (14) is more likely to lead to a
phatic inte rpretation than (is) , (is) more like ly than (i6) , and (16)
more like ly than (17) . This follows from consi derat ions of effec t
and effort: extra linguist ic mat eri al req uir es gre ate r pro ces sin g
eff ort and , the ref ore , point s to the import ance of the lingui sti c
mea nin g of the utt erance for arr iving at the ove ral l
int erp ret ation. It is int ere sti ng to not e in thi s connec tion tha t
an utterance of (15), (16) or (17) may easily be used as an
insincere `non-phatic'. This will be the case if the speaker wishes to
assure the hearer overtly of her genuine interest in the hearer's
welfare, thus avoiding some possible socially undesi rable
implica tures about the superficiality of her interes t in the speaker
that a manifestly phatic utterance (e.g.(14)) might give rise to. But
our general point still holds: other things being equal, the more
processing effort is required for the inte rpret ation of the
linguist ic meaning of the utte rance , the less manifes tly phatic the
utterance will be. And conversely, other things being equal, the
less effortful the interpretation of the linguisti c expression is, the
more suitable that expression will be for phatic communication.
We have al ready shown tha t, in pr inciple, phat ic
in terpreta tion is a func tion of the cont ext in which the
utteran ce is interprete d (see example (11)). But the widespread
phatic use of language is possible only if people can ini tia te pha tic
exchanges more or les s at wil l. In other words, it mus t be
possible to manipulate the context so as to ensure that the intended
phatic interpretation is consistent with the Principl e of Relevance,
even when, other things being equal, a non-phatic one might be
optimal ly relevan t. It seems to us th at on an y give n oc casi on
th is is po ss ib le to th e ex te nt th at th e int erlocutor s ' mutual
cognit ive env ironment inc ludes some assumptions about the
24
way conversations are usually conducted: how are certain topics
usually relevant? What is the social relationship between the
communicators? What are the soci al norms for appropriat e
linguistic behavio ur? Consider again (2).
(2) Glengarry Glen Ross
RR: How are you
GA: Fine („) you mean the board (,) you mean you
mean you mean on the board
RR: I (,) yes (,) okay (,) the board
GA: I'm fucked on the board
Speakers: RR = Ricky Roma, GA = George Aaronow
(Glengarry Glen Ross 1992, dir. Tony Foley, screenplay David
Mamet) Conventions: (,) = short pause; („) = long pause.
In this case, Aaronow entertains one interpret ation of `how are
you ' and then replaces it with an alternative. He chooses the
alternative interpretation because that interpretation is highly
relevant to him: he is a salesman, he has not been doing well, he is
in danger of losing his job, and the success of each salesman in the
firm is displayed on a board in the office. What is interes ting her e
is tha t Aar ono w might hav e opt ed for the non -pha tic
int erp ret ation firs t. Considerations of relevance predict that
other things being equa l, the second interpretation should be highly
manifest in the hearer's immediately access ib le context. The
ques tion is : why was it not the fi rst one to be considered?
One plausibl e answer is that the contextual knowledge about the
way in which a verba l expression is typically used affec ts the
interpreta tion process by constraining the hearer's assumptions about
the way in which that expression is intended to be relevant.
Typi cal ly, the more often a topic is rai sed in con ver sat ion wit h
a cert ain pur pos e in mind, the str ong er the connection between
the topic and the purpose will be. And the more often the utterance of
a particular linguistic string is used to give rise to a particular
interpretation , the stronger the connection between the linguistic
25
string and the interpretation will be.
In example (2 ), it is mutual ly mani fest to the in te rlocutor s
that the question `How are you' is usually not used to show the
speaker's genu ine int erest in the heare r's wel far e, but rat her
as a vague indicat ion of the speaker's favourable di sposit ion
towards the hearer . Ric ky Roma 's ut terance, by virtue of its
lingu istic meaning, indicates that some information about the
heare r's welfa re is relevant to the speaker. How much
information is relevant depends on the accessib ility of specif ic
contextual assumptions. In most contex ts, general knowledge
about the way the expression how are you is usually used will be
highly accessible, and will play an important role in the
interpreta tion process. In this example, Aaronow's knowledge
about the use of this linguisti c string, and his knowledge of
Roma 's chara cter (Roma is a ruth less and succ essful sale sman),
prov ide over whelming evidence for the fir st int erp ret ation. If
Rom a wan ted to mak e sur e tha t his utt erance was inten ded to
raise the ques tion of Aaronow's status on the board rather than
merely to `make chit-chat', he could have uttered a string such as
(18):
(i8) How are you doing on the board?
Interpreting this would involve extra processing effort , but the
extra effort could be easily justi fied as it would make sure that
the heare r arrived at the in tended in te rpre ta tion more quickl y,
wi thout having to wast e ef fort considering whether the
relevance of the utterance lay mainly with evidence provided by the
mere act of ostensi on. In other words, Relevan ce Theory
clearly pred icts that Aaronow will assume that Roma 's utte rance
is a phat ic one.
So where does the evidence for the second interpretation come
from? First, it is mutually mani fest to Roma and Aaronow that
the jobs of all sale smen in the firm are on the line and that
everybody's chances of keeping their job depend on how effic ient
26
they are in closing sales. Second, a question asking for
informat ion about his perfo rmance is highly relevant to
Aaronow, who has not been doing well, and has been struggling
against the odds (because, in his opinion, he has been given bad
leads). Third, if Roma is asking about Aar ono w's per forman ce,
the n Aarono w wil l have an opp ort uni ty to tal k about his
unhappy situation; this might be so desi rable to Aaronow that he
considers wilfully misunderstanding Roma's utterance in order to
shift the conversa tion on to the topi c he would like to focus on.
If this is righ t, it is Aaronow's despair, more than anything else,
that leads him to reinterpret the question. In this example, the
processing of a particular utterance in slightly diff erent ways (by
expanding the context in different ways and focusing on different
implications) results in rather different interpretations. On the first
interpretation , Ricky Roma 's utterance weakly impli cates some
information about the speaker's social attitude towards the hearer.
On the second in terpre ta tion , it communi cates much st ronge r
impl icatures about the speaker's interest in the hearer's welfare.
Relevance Theo ry provides a natura l explana tion for the way
in whic h see min gly inapp rop riat e res pon ses are giv en. We
now tur n to two suc h examples. First, the case where a phatic
response is relevant, but a manifestly non -phatic one is given.
Second , the case where a non -phat ic resp onse is relevant, but a
manifestly phatic one is given. Consider again (I)
(1) Groundhog Day
ML: I hope you enjoy the festivities.
PC: Oh I'm sure I'm going to. (pulls face)
ML: There's there's talk of a blizzard.
PC : Well , we ma y ca tc h a br eak and that bl iz za rd 's
gonn a blow right by us. All of this mois ture
coming up outa the south by midday is probably
gonna push on to the east of us and at high
altit udes it's gonna crystallise and give us what we
cal l snow. (winks ) Pro bably be som e
27
accumulat ion . But here in Punxsutawn ey our
high 's gonna get up to about 30 today, teens
tonight, chance of precipitat ion about 20 per cent
tod ay 20 per cen t tomorro w. Did you wan na
tal k about the weather or were you just making chit-
chat?
ML: (shrugs and shakes head) Chit-chat.
PC: OK. Right. See you later. B'bye.
ML: Oh em eh will you be checking out today Mr.
Connors?
PC: Chance of departure today one hundred per
cent.
Speakers: ML = Mrs. Lancaster; PC = Phil Connors.
(Gr oun dhog Day 1993 , dir . Har old Ramis, screen pla y Dan n y
Rub in and Harold Ramis)
When Mrs Lancaster says `There's talk of a blizzard' she
presumes that it is mutua lly manifest to her and to Phil Conno rs
that making chit-chat is her main reason for talking about the
weather. A detail ed account of the curren t climatic condit ions
requires considerable processing effort and sugges ts that the main
relevance of the answer lies with meteorological information. Phil
Connors ' detailed answer might indicate one of two things : either
(a) Mrs Lancas ter 's pre sumption about the mut ual manife stness
of the reason for bri nging up the subject is not war ran ted, or
(b) by givin g a manifes tly inappropr iate answ er, Phi l Connors
is tryi ng to communicat e somethi ng: e.g. that he doesn't like chit-
chat, that he is a busy man who has no time for discussing triv ial
matters, that Mrs Lanca ster shou ld not bring up this sort of topic
again, etc. The following exchange illustrates the converse situation:
28
(19) Paul ine: I was real ly sorr y to hear abou t what happ ened to
you when you were abroad. It must 've been
real ly awful. I hope you've managed to get over it
OK.
Arthur: Yes. I'm fine, thanks.
Clea rly, Paul ine is going out of her way to express her interes t
in Arth ur's unpleas ant exp eri enc e. Art hur's res pon se is
man ifest ly not rel eva nt to Pauline in the way Pauline expected.
But it does provide evidence for an interpretatio n which Arth ur
intende d to be opti mal ly relevant, and which may include a
rang e of impl icat ures abou t his soci al atti tude towards her:
Arthur does n't want to engage in conv ersa tion with Paul ine;
Arthur thin ks that Paul ine is nosey; Art hur does not con sider
Pau line a clo se fri end ; Arthur finds Pauline boring; Arthur does
not like being reminded of his bad experience; etc. None of thes e
implicatur es are der ived by exp loi ting the linguisti c meaning
of the utt erance . Since the evidence presen ted by the linguisti c
meaning of Arthur's response is manifestly not optim ally relevant
to Pauline, it leads her to make assumptions about what Arthur
intended to communicate by giving a lingui stical ly less
informative response than would have been relevant to her.
So far, our discussion has relied on the view that phatic
implications are communicated ostensi vely. In the next section
we consider whether ther e exist phatic implications which are not
ostensively communicated.
4. Ostensive and Non-ostensive Communication
Ou r de fini ti on ru le s ou t th e po ss ib il it y of non -os te ns iv e
ph at ic communication; for us, phat ic inte rpre tations are a subset
of the poss ible kinds of inte rpre tati on of an act of ostensiv e
comm unic ation. In this sect ion, we consid er the rel ati onship
between phatic int erp ret ations in our sense and soc ial
29
rel ati ons hip s and we discuss imp lic ati ons whi ch sha re all of
the charac terist ics of phatic implicatures except one: unlike
phatic implicatures , they are not ostensively communicated. We
propose to use the term PHATIC IMP LICATI ONS as a gen era l
ter m to cov er the se imp lic ati ons and phat ic impl icatures .
Thus , just as im pl icatures can be seen as th at su bset of
implications which are ostensivel y communicated, so PHATIC
IMPLICATURES can be see n as that sub set of pha tic
imp lic ati ons whi ch are ost ens ive ly communic ated . This
sect ion draws heavil y on Blak emore's (1994) discussi on of
Relevance Theory and social goals.
Our def ini tio n of pha tic int erp ret ati ons mak es no ref ere nce
to soc ial relationships :
Phatic interpretation
An int erp ret ati on is phati c to the ext ent tha t it con tai ns
imp lic ate d con cl us io ns wh ic h do no t de pe nd on th e
ex pl ic it co nt en t of th e utterance.
How eve r, many imp licat ures whi ch dep end upo n the
rec ognit ion of an intent ion to communicate do concern social
relationships. This is because the fact that someone has made
manifest an intention to communicate with another person rules out
certain social situations, most notably the possibi lity that the
speaker despises the hearer so much that she refuses to intera ct
with him at all. Whil e our defi niti on incl udes impl icat ures
which are not abou t social relat ionsh ips, it also reflects the fact
that most phati c interpretat ions achi eve relevan ce by suggest ing
somethi ng about the natu re of the soci al relationship between the
speaker and the hearer.
If an individual wishes to convey inform ation to anothe r
individual about thei r soci al situ ation (or abou t anything at all) ,
ther e are good reasons for attempting to achieve this by
communicating the information ostensively. As pointed out above,
every act of ostensive communication communicates the
30
presumpt ion that it is worth paying attentio n to, and that paying
attentio n to it does not put the hearer to a gratuitous expenditure of
processing effort. Info rming somebody of so meth ing m eans
drawing a part icul ar se t of assumptions to their attention. The
best way to direct somebody' s attention in the desired way is to
make this intention mutually manifest (i.e. public). Otherwise, there
is littl e likel ihood that people will allocate their cogni tive
res our ces in exa ctl y the way we wou ld lik e the m to. In oth er
wor ds, the intent ion to inform is more likely to be fulfil led if it is
recognised, and more likely to be recognised if it is made public.
This has obvious implicat ions for the comm unic ation of all
informatio n, inc luding social informa tion: most successful
informative behaviour is ostensive (cf. Blakemore 1994).
To see in more det ail why infor mat ion abo ut socia l
rel ati onships in particular should be ostensively communicated,
consider (20).
(20) Are you feeling tired after all that hard work?
The soc ial att itu de tha t the hea rer of (20 ) att rib utes to the
spe ake r wil l de pe nd on th e he ar er 's as su mp ti on s ab ou t th e
sp ea ke r' s un de rl ying inten tions . Laver (1974) makes a number
of observations about the relat ive social stan dings which
speakers of vari ous phat ic utte ranc es impl y; one of these is that
speakers may only comment on, or enquire about, attr ibutes of
the hearer if they are soc ial `equ als ' or `sup erio rs'. Given this,
the hearer may take (20) to indica te, amon g other things , tha t
the speaker does not con sid er her sel f inf eri or to him whi le
mak ing or not mak ing any of the following assumptions:
(a) the spea ker intends to make it mutuall y manifes t that she istrying to make it manifes t to the hearer that she does notconside r hersel f inferior to him;
(b) the speaker is trying to make it manifest to the hearer that shedoes not regard herself as inferior to him, i.e. a subpart of (a);
(c) the speaker does not regard herse lf infer ior to the hearer; i.e.a subpa rt of (b).
31
The se inf ere nce s are det erm ine d by the hea rer 's ass ump tio ns
abo ut the speaker's intentions. If the hearer makes assumption (a)
the information in (c) wi ll have been communica ted
os tens ively (i .e . over tl y) : bo th the communica tive and the
informat ive intentions will have been attr ibuted to the speaker.
In thi s case the hea rer may wel l att ribute to the speaker the
inten tion to communicate something about her beliefs about an
indiv idual 's right s to change socia lly accep ted views about socia l
hiera rchies and about their role in human relationships.
If th e hearer makes as sumption (b ), bu t no t (a ), th en th e
speake r' s intention to inform the heare r of her views about thei r
rela tive statuses has been made manifest and fulfi lled, without
being ostensivel y communicat ed: there was no comm unicati ve
intenti on. In Wilson & Sperber's (1993) term s, a form of cove rt
communic ation will have take n place. The utterance may well
conve y some information about her socia l attitudes as in the first
case, but the speaker cannot be held accou ntab le for conveying
this informati on. It is als o wor th not ing that in th is cas e the
utt era nce does not pro vid e evidence for assumptions about her
views on an indiv idual 's right to oppose existing social
hierarchies and their role in interactions between people . For
these assumptions to be communicated , the speaker must make
her intent ion to in fo rm the hearer of (c ) publ ic. In ot he r
words, th e conten t of the informative intention is part ly
dete rmined by the recognit ion of the high er order communicative
intention.
Now consid er the pos sib ili ty tha t the hea rer draws the
conclusion (c) , without assuming either (a) or (b). In this case,
there are three possibilit ies. Fir st, ass ump tio n (c) may hav e
bee n cov ert ly com mun ica ted : the com municativ e intention is
not attr ibuted to the speaker, but both the intention to info rm
and the intenti on to conc eal this intenti on are att ribu ted to her .
Second, (c) may be taken to have been conveyed accidentally: neither
the communicative nor the informat ive intention is attr ibuted.
32
Third, it may be that the speake r does not actually have the
intent ion to communicate (c), but the hearer is nevertheless
justified in assuming that the speaker intended to communicate (c). In
this case, what might be termed INADVERTENT OSTENSIVE
COMMUNICATION has taken place. This differs from accidental
communication in tha t the rec ogn iti on tha t the spe ake r has
acc ide ntall y com mun ica ted something may well absolve her
from the responsibi lity of causing offence. In the case of
inadvertent ostensive communication the hearer assumes that the
speaker was in a posit ion to avoid causing offence (even though
she did not act ual ly inten d to con vey the offen din g
imp licat ures) . That is why the speaker is nevertheless held
responsible for causing offence to the hearer.
Obviously, thi s suggest s a potenti al grey area: hearers may
not be sure whether either the lower order or the higher order
intentions were present. Ano the r possib ili ty is tha t the y do not
eve n consid er the que sti on wor th pursuing. It is conceivable that
evidence for the higher order intentions could be mani fest to the
hearer, but not mani fest enou gh to be cons idered worth
exploi ting. There may be considerab le diff eren ces between
ind ividual s in terms of what sorts of information they consider
relevant: just as some people are more interested in sport or music
or politi cs than others , so social status matter s more to some
people than to others . The point of this example is that it is on ly
if the he arer as sumes that the speake r is ma king her
info rmat ive intentio n publ ic that the utte ranc e of (2o) will have
communicated ostensively some informat ion about the speaker's
atti tudes regarding social distance and its role.
So comm unicat ion wh ich conc erns social re la tionsh ips
tends to be ostensiv e. Here are two exam ples where
info rmat ion abou t soci al rela tionships is communicated NoN-
ostensively:
33
(21) (the speaker is a car salesman who has just met the hearer)
Have you decided what kind of car you're interested in, Billy?
(2 2) I wa s su rp ri se d to he ar th at so ma ny ca r sa le sm en ar e
se xi st . Personally, I always pay attention to the ladies as well.
By asking (2 1) the spe aker overtl y commun ica tes tha t som e
information abou t the hearer's car-buying inte ntions is relevant
to him (we depa rt here from the conven tion tha t the speaker is
female bec ause (21) and (22) are based on real utte ranc es and
because this is impo rtan t in the discuss ion of (2 2) ). Bu t, by
us in g th e he ar er 's fi rs t na me , (2 1) mi gh t al so co nv ey
ass umptions about soc ial rel ati ons hips tha t the spe ake r int ends
to mak e more manifes t (though not mutuall y manifest) to the
hearer, e.g. assumptions about how well he gets on with the
customer , that his rela tionship with the customer is not strictl y
professional, that the customer should trust him as he wo ul d tr us t
a fr ie nd , an d so on . In re le va nc e -th eo re ti c te rm s, th es e
assumpti ons are part of the spea ker 's info rmat ive intentio n but
ther e is no rel ate d commun ica tiv e int ent ion . Thi s is bec aus e
the ass umptio ns abo ut fami liar ity and so on may eas ily become
mani fest to the heare r without the speaker 's intention being made
mutual ly manifest. This is what distin guishes the assumptions
about familiarity from weak implicatures.
In co nt ra st , (2 2) mi gh t co nv ey as su mp ti on s wh ic h da ma ge
so ci al relationships and which the speaker did not intend to
convey, e.g. tha t he holds som e sex ist att itudes . Her e, the
spe ake r's int ent ion to ost ens ive ly communica te information
about his views on sex ism, bet rays someth ing about his real
atti tude (which he himself may not be aware of). In this case the
communicative and the informative intentions are both absent.
Covert manipu lat ion requires the int ent ion to make par ticula r
information more manifest, without making this inten tion
mutua lly manif est. Notice also that the succ ess of conveying
informat ion sometimes depends on covertness. In (21) the
34
sa le sman over tl y focuses th e hearer 's at tent ion on one se t of
ass umptio ns (co nce rning the hea rer 's though ts abo ut whi ch car
to buy), whi le cov ert ly mak ing mor e man ife st to the cus tomer
a dif fer ent set of assumptions (about how well they get on,
etc .). The cust omer may tac itl y acce pt the cove rtly
communic ated assumpti ons prec isel y beca use they are made
more manifest to him without attracting his attention. Manipulative
behaviour may not be soci able, but it is soci ally ver y
importa nt: it makes poss ible a cert ain degr ee of cont rol over
othe r peop le' s beha viou r, with out the speaker having to accept
responsibi li ty ei ther fo r the in formation conveyed or for its
impact on the audience.
Such covert communica tion may but need not invo lve the
exploitat ion of linguistic meaning. The use of the hearer's first name
in (21) to communicate ass umpti ons abo ut the socia l sit uat ion is
an examp le of no n-ost ens ive li ngui st ic comm unicat ion. We
now tu rn to an in stan ce of cove rt communication which does
not exploit the linguistic meaning of the words used.
(23) Nice weather.
(23) said in a hoarse voice may be evidence that the speaker has a
sore throat, withou t necessarily involv ing either the speaker's
informative intent ion to let th e he ar er kn ow ab ou t he r me di ca l
co nd it io n, or th e co mm un ic at iv e intention (to make the
informative intent ion mutually manifest). The speaker may want
to make the hearer feel sorry for her, while not being prepared to
acknowledge that that is what she wants. Her decision to
communicate some assumptions covertly is probably socially
motivated. If she were to complain pu bl ic ly ab ou t he r so re
th ro at , sh e mi ght al so co mm unic at e so c ia lly undesirable
assumptions about her weakn ess, lack of abil ity to cope with a
minor healt h probl em, and the like . In th is example, some
non-phat ic inf orm ati on is com mun ica ted cov ert ly and non -
lin gui sti cal ly for soc ial reas ons . (For a more detail ed
discussio n of such exam ples, see Wilson & Sperber 1993).
35
It seems reasonable to assume that the lingu istic meaning of an
utterance may pre clu de int erp ret ations whi ch hel p with the
man agemen t of soc ial relations. Surely, a speaker who uses (24)
does not intend her utterance to be interpreted in this way.
(23) I'll never talk to you ever again.
In this case the linguis tic content probably precludes such social
assumpt ions fro m bei ng osten sivel y com municat ed, but not
nec ess ari ly from bei ng covertly (i.e . non-oste nsively) and non-
linguist ical ly conveyed . Again , the reasons for conveying some
information covertly may be motivated by social reaso ns . A publ ic
app eal for rec oncil iat ion might be tak en as a sign of
we ak ne ss , la ck of co nv ic ti on in one' s be li ef s, la ck of
det ermina ti on , vulnerab ilit y, and the like . But, by over tly
asse rting her determinat ion not even to consid er the possib ili ty
of a reconc ili ati on, the speaker provides (whe ther
in tent ionall y or no t) ra ther st rong evidence to the cont ra ry .
Ostensi ve behavio ur is inheren tly soc ial . By making publ ic (i.e .
mutuall y mani fest ) the intention to inform, the speaker
inevitab ly makes this second ord er int ent ion its elf pub lic (i. e.
mut ual ly man ife st) . Thi s emb edd ing of intenti ons could, in
principle, go on ad inf ini tum. Which level is actu ally exp loi ted
in utt era nce int erp ret ati on dep end s onl y on con sid era tio ns of
relevance.
Some non-ostensive communication concerns social relationships.
Some non-ostensive communicat ion depends upon a
communicat ive inten tion but not upon linguistic all y-encoded
meanings. Some non -ostensi ve comm uni cation depends upon
linguistically-encoded meanings. Given our defin ition of pha tic
int erp ret ations , we cou ld dec ide not to use the term pha tic ' in
refer ring to any cases of non-ostensive communicat ion, or we
could extend it to include some or all of these cases . Not much
depends on how the term pha tic ' is use d; wha t is imp ort ant is
tha t we dis tin gui sh these dif fer ent phenomena clearl y. So our
decision about how to use the term pha tic ' is based main ly on
36
an attempt to be clea r abou t the theo reti cal clai ms we are
ma king . We prop ose to ap pl y th e te rm phat ic ' to
as sumpti ons wh ose der iva tion exp loi ts mainly , or onl y, the
evidence presen ted by the act of ostension, whether or not they
are themselves ostensively communicated:
Phatic implication
A pha tic imp lic ati on is a con clu sio n whi ch does not
dep end on the explicit content of the utterance (but does depend
on the communicative int ent ion , per haps interacting with
contex tua l assumptio ns, some of which may exploit
linguistically-encoded meanings).
So, just as implicatures are the subset of implications which are
ostensively communica ted , pha tic implicatures are the subset of
pha tic implications which are ostensively communicated.
Thus , (26) and (27) are phat ic impl icat ions of (25), but they
will only be phatic implicatures in certain contexts.
(25) How are you?
(26) The speaker wishes to communicate with the hearer.
(27) The speaker is prepared to talk to the hearer.
This terminol ogic al deci sion sets up a neat para llel betw een the
rele vanc e-theore tic views of phatic communication and
communication more general ly. It also allows us to acco unt for
the intuiti ons ment ioned above: fir st, tha t there is a difference
between utterances whose main point depends upon linguis tically-
encoded meanings and those whose main point depends simply on
an act of ost ens ive com munica tio n; sec ond , tha t the re are
deg ree s of phaticness; third, that phatic interp retat ions are more
likely when the nature of the social relationship between
interlocutor s is in doubt . Finall y, it makes it easi er to talk clearly
about what is going on when the rela tive like lihood of phat ic
inte rpre tati ons vari es from context to cont ext: what vari es is the
likelihood of particular phatic implications being recognised as
implicatures.
37
5. Conclusion
We have proposed that phatic interpretations are interpretations
containing implicatures which depend upon recognition of a
communicative intention but not only, or primaril y, upon
linguist ical ly -encoded meanings . We have sho wn how thi s
def ini tio n can acc oun t for the fac t tha t pha ticne ss ' is a matter
of degree, and how this approach can account for the
inte rpretation of a range of fa ir ly phat ic and fair ly non -
phat ic ut te rances . We have propos ed an account of what makes
some utterances more likely to give rise to phatic interpretations
than others.
To summarise, we follow Wilson and Sperber (1993) in the
assumptions we make about how assumptions can be
communicated:
(28) What is communicated
Ostensive communicat ion Non-ostensive communication
Linguistic Non-linguisticcommunication communication
(adapted from Wilson and Sperber 1993)
Assumptions communi cated in any of these ways can have
a social fun cti on, or a fun cti on typ ical of wha t hav e bee n
tho ught of as phati c ut ter anc es. We hav e not pro posed ,
how ever, that phati cne ss should be defined in terms of
functions. Instead , we sugges t that interp retations which depend
on (in our technical sen se of `de pend on ') the
communicati ve intention and exploit the linguist ic meaning of
the words used to a limi ted extent, if at all, const itute the class
of phen omena which should be termed phatic . Non -ost ens ive ly
communica ted assumptions which per form what might have
38
been thought of as a phatic function are not necessarily phatic in
our terms.
The phaticness of implications, or inferences, is a classificatory
concept: impl icat ions eit her are or are not der ivable on the
basi s purely of the recogn iti on tha t an ost ens ive act has been
per formed . The pha tic nes s of whole acts of communication, by
cont rast , is a matter of degree. Ostensive acts or utterances as a
whole are always phatic in the sense that, by definition, they provide
evidence that an ostensi ve stim ulus has been prod uced. The
more phatic implications a communicative act has, the more phatic it
is.
In con clu sion, we hav e sug ges ted tha t the term pha tic ' be
use d in the following technical terms :
Phatic implication
A phat ic impl icat ion is a conclusion which does not depend on
the expl icit content of the utte rance (but does depend on the
communicative intention , perhaps interacting with contextual
assumptions, some of which may exploit linguistically-encoded
meanings).
Phatic implicature
A phat ic impl icature is a phat ic impl icat ion which is ostensively
communicated.
Phatic interpretation
An interpretation is phatic to the extent that it contains phatic
implicatures.
Phatic utterance
A phat ic utte rance is one which gives rise to, or is intended to
give rise to, phatic interpretations.
39
Phatic communication
Phati c comm unica tion is communicat ion which gives rise to, or
is intended to give rise to, phatic interpretations.
Ack now led gem ent s*For fin anc ial sup por t, we wou ld lik e to tha nk the Sch ool of
Eng lis h, Cul tur al and Communication Studies at Middlesex
Universit y. For useful discussion and comments, we woul d like to
thank John Bird, Richard Breh eny, Rob yn Carston, Alan
Durant , Steve Nicolle, Vil ly Rouchota, Dan Sperber, Dei rdre
Wilson, two anon ymous JL referee s, audiences at the Univer sity
of Brighton, Middlesex University, the University of Sussex , the
Colloquium on the Processes of Linguis tic Communication' at the
Univers ity of Wales Bangor and the Lingu istics Assoc iatio n of
Great Brita in conference at the University of Salford.
40
References
Blakemore, D. (1994). Relevance, poetic effects and social goals: a
reply to Culpeper. Language and Literature 3.I. 49-59.
Carston, R. (1988). Implicature, explicature and truth-theoretic
semantics. In Kempson, R. (ed.), Mental representations: the
interface between language and reality. Cambridge: Cambridge
University Press. 155-181.
Jakobson, R. (196o) . Cončluding statement: linguistics and
poetics. In Sebeok , A. (ed.), Style in language. Cambridge, MA:
MIT Press. 350-77.
Laver, J. (1974). Semiotic aspects of spoken communication. London:
Edward Arnold. Malinowski, B. (1923). The problem of meaning in
primitive languages. In Ogden, C. K. &
Richards, I. A. (eds.), The meaning of meaning. London:
Routledge & Kegan Paul. 296-336. Nicolle, S. & Clark, B. (1998).
Phatic interpretations: standardisation and conventionalisation.
In Mateo Martinez, J. & Yus, F. (eds.), Revista Alicantina de
Estudios Ingleses II. 183-191. Schiffrin, D. (1994). Approaches to
discourse. Oxford: Basil Blackwell.
Sperber, D. (1983). Anthropology and psychology: towards an
epidemiology of representations. Man 20. 73-89.
Sperber, D. (1985). On anthropological knowledge. Cambridge:
Cambridge Univers ity Press. Sperber, D. & Wilson, D. (1986,
2nd edition 1995). Relevance: communicat ion and cogni tion.
Oxford: Basil Blackwell.
Wilson, D. & Sperber, D. (1993)• Linguistic form and relevance.
Lingua 90. 1-26.
Žegarac, V. (1998). What is phatic communication? In Rouchota, V.
& Jucker, A. (eds.), Current issues in Relevance Theory.
Amsterdam: John Benjamins. 327-361.