Bee
Constituting Face in Conversation:
An Alternative to Brown & Levinson’s Politeness Theory
Robert B. Arundale
Department of Communication
University of Alaska
Fairbanks, AK 99775-5680
Phone: +907-474-6799
Email: firba@uaf.edu
Presented at the 90th Conference of the National Communication Association
Chicago, IL
November 2004
ABSTRACT: Brown & Levinson's (1987) theory remains the most widely employed
explanation of politeness, despite numerous critiques and suggestions for revision.
Eschewing revision, this paper outlines a new theory that is an alternative to Brown &
Levinson’s and that is responsive to critiques, including the authors’ own reservations.
Brown & Levinson note clearly that “improved conceptualizations of politeness” will
require replacing the model of language use they employed with a model that accounts
for the emergent properties of human interaction. Face Constituting Theory heeds their
critique in explaining facework as interactionally achieved in ordinary conversation.
Brown & Levinson's theory explains face maintenance as the balancing of face threat by
face redress, but it does not explain how face is created, how it is maintained in the
absence of threat, how it is threatened as in rudeness, or how it is supported as in caring.
Face Constituting Theory re-defines face as a dialectic of connection (positive face) and
separation (negative face), and provides a conceptualization of face threat (as well as face
support) that permits an integrated explanation of the full scope of facework from
outright aggravation, through maintenance, to outright enhancement, In this new theory,
favework is the result neither of a strategic intention nor of a personal want to maintain
public self-image. Instead, face is conceptualized as interactionally achieved in talk,
While threat and support are conceptualized as participant evaluations of face as it is
conjointly co-constituted in conversation. Examining one instance of talk-in-interaction
makes evident how the explanations of facework provided by Face Constituting Theory
move beyond those provided by Brown & Levinson’s politeness theory.Constituting Face in Conversatio
‘An Alternative to Brown & Levinson’s Politeness Theory
In 1978 Brown & Levinson developed a theory of politeness to explain why they
had found striking parallels in polite language use across very different languages and
cultures. They argued that the commonalities arise because in being polite people are
universally engaged in facework: in shaping utterances that redress threat to one’s own or
to another's face. Their central insight that polite usage involves facework provided a
compelling explanation for a diverse set of observations, and their theory of politeness
has had a significant impact across the diverse literatures concemed with language use.
Over the ensuing twenty-five years scholars have repeatedly acknowledged politeness
theory's significance in research on language use, while at the same time pointing out
important weaknesses. Thorough reconsideration of Brown & Levinson’s contribution is
needed and is underway, Watts (2003) argues, for example, that contrary to the
presumption of many researchers, “politeness” is not equivalent to “facework” because
‘hile all human interaction involves facework, only some interaction involves politeness,
Watts develops a new conceptualization of linguistic politeness. In contrast, this article
outlines a new theory of facework in everyday conversation that encompasses Brown &
Levinson’s insight, that addresses the numerous critiques, and that explains facework:
phenomena beyond those addressed by politeness theory.
Part | considers a number of reservations about politeness theory, both new and
old, that define issues to be addressed in the new theory of facework. Pivotal among
these critiques is Brown & Levinson’s (1987) own argument that developing an improved
theory of politeness conceived as facework requires replacing the model of language use
‘underlying their theory with a new model accounting for the emergent properties of
language use “which transcend the characteristics of the individuals that jointly produce
it” (45). Brown & Levinson’s critique warrants one of two new conceptualizations
central to the new theory: a new model of language use, as outlined in Part 2. Other key
reservations regarding Brown & Levinson's conceptualization of face and its cultural
generality, consistency in their definition of face threat, and limitations in their account of
facework warrant a new concept of face, and hence of positive and negative face, as
outlined in Part 3.
‘These new conceptualizations of language use and of face are radical departures
from the parallel concepts in politeness theory, obviating its extension or revision. Part 4
outlines Face Constituting Theory, a new theory of facework that addresses the question
“How do participants achieve face in everyday conversation?” Part 5 applies both
politeness theory and Face Constituting Theory to an instance of ordinary talk, the
comparison forming one basis for examining the new theory’s alternative explanation of
facework in Part 6. Because it is a new theory, not a revision, Face Constituting Theory
moves beyond politeness theory to provide an integrated explanation for how face is
created, maintained, threatened, and supported in conversation, or in short, for the full
scope of human facework.e
1: Reservations about Brown & Levinson’s Theory of Politeness as Facework
In developing their central insight that polite language use involves facework,
Brown & Levinson (1987, hereafter B&L) introduced a version of Goffman’s (1955)
concept of “face.” “Face” is the public self-image social actors claim for themselves, that
image involving the actor’s concerns or wants for association (having one’s wants
appreciated, or positive face) or for dissociation (having one’s actions unimpeded, or
negative face). Face is acknowledged mutually and attended to by all participants, and
becomes relevant in interaction at the point a threatening utterance diminishes a
participant's face, Politeness or face maintenance occurs when the threatening utterance
also incorporates a redress strategy. ‘The presence of the strategy is a departure from
Grice’s norms for rational, efficient utterances, and triggers @ conversational implicature
of face support, which serves to balance the threat to face. Over the past twenty-five
years, critics have scrutinized many aspects of politeness theory (hereafter PT; see Brown
2001), so that even a partial review is beyond the scope of this paper. It must suffice to
consider briefly four areas of reservations that indicate issues PT does not address and
that justify constructing a new theory of facowork: reservations regarding language use,
face, face threat, and facework.
Language use: B&L's (1987) critique of the model of language use underlying
PT is the most central reservation because replacing that model is basic to future
development of the theory. This critique is seldom acknowledged, and until now
unaddressed, Specifically, PT is based firmly in Grice’s (1989) theories of meaning and
of conversational implicature, In their 1978 presentation of the theory, B&L note that
their “analysis must be found wanting, dominated as it is by the act-by-act analysis of
contemporary philosophy and linguistics” (1978: 89, 1987: 84), such analyses of
utterances as self-contained, separable units or actions having been “ably criticized” by
conversation analysts (1987: 232). They reiterate this critique in their introduction to the
1987 reissue, and also characterize the model of language use they believe essential to
any new theory of politeness:
Social interaction is remarkable for its emergent properties which
transcend the characteristics of the individuals that jointly produce it; this,
emergent character is not something for which our current theoretical
models are well equipped. Workers in artificial intelligence have already
detected a paradigm clash between ‘cognitivism’ and ‘interactionism’, and
noted the failure of the former paradigm to account for interactional
organization (see ... Suchman [1987]); our own account suffers from the
same dose of ‘cognitivism’, Work on interaction as a system thus remains
a fundamental research priority, and the area from which improved
conceptualizations of politeness are most likely to emerge. (48)
Because the “current theoretical models” to which B&L (1987) refer include both
Grice’s and Searle’s conceptualizations of language use, this apparently casual
observation is an incisive critique of the models of language use employed in language
pragmatics as a whole, including PT (see Arundale 2005: 50-52). These models are
problematic because they assume that language use in particular, and humancommunication in general, are matters of one person’s encoding their meanings for others
to recover through decoding. ‘hat is, communication reduces to an activity of monadic
individuals who encode and decode self-contained verbal (and non-verbal) unit acts in the
presence of other wholly independent or monadic individuals. Such models continue to
be employed, even in view of decades of substantive critiques within the disciplines of
human communication, linguistics, psychology, and sociology (Arundale & Good 2002),
Both PT's cognitivism and its focus on unit acts are entailments of the
encoding/decoding model of communication that underlies Grice’s theories (Arundale
1991). Arundale (1999: 121-126, also Arundale and Good 2002) argues that creating
new models of language use that explain emergence as B&L (1987) specify requires
conceptualizing communication as the dyadic or conjoint co-constituting of meanings and
actions in on-going address and response between wholly interdependent language users.
B&L acknowledge that their explanation of the structuring of self-contained, separable,
polite unit acts involving face threat needs to be supplemented with insights drawn from
research in conversation analysis. But conversation analysis conceptualizes
communication as the interactional achievement of meaning and action, hence ad hoc use
of such insights to complement a theory that conceptualizes communication as monadic
encoding and decoding can provide only ad hoc descriptions of facework, not principled
explanations. PT cannot be revised or extended to incorporate such insights because itis
incommensurate with interactionism, That is, given its fundamental commitment to an
encoding/decoding model of communication, itis incompatible with a conceptualization
of language use as an emergent, dyadic phenomenon, conjointly co-constituted in
communication. A new theory of facework requires a new model of language use in
communication that addresses B&AL’s key critique.
Face: B&L's (1987) conceptualization of face bas drawn considerable attention
on both theoretical and empirical grounds. Building on Goffiman’s (1955) concept, B&L
(1987: 62). define “face as wants,” noting that “It would have been possible to treat the
respect for face as norms or values subscribed to by members of a society (as perhaps
most anthropologists would assume), Instead, we treat the aspects of face as basic wants,
which every member knows every other member desires, and which in general it is in the
interests of every member to partially satisfy.” On this basis, they define “negative face”
as “the want of every ‘competent adult member’ that his [sic] actions be unimpeded by
others,” and “positive face” as “the want of every member that his [sic] wants be
desirable to at least some others.” ‘Two issues are problematic here: the interpretation of
face es wants, and the cultural generality of the definitions both of face and of positive
and negative face.
First, B&L's (1987) choice of terms has led many to interpret face “wants” as
psychological "needs." That interpretation is consistent with the Euro-American
psychological tradition that assumes that the actions of individuals are effects caused by
the presence of specific internal needs or drives, which has lead to a common
interpretation of face as a motivation for certain identifiable patterns in language use. As
Bavelas (1991) argues, the view of language use as caused by internal states is
problematic, theoretically because it is difficult to adequately specify the cause-effect
Jinks, and empirically because it is hard to avoid circularity in indexing internal states.Brown (personal communication, Tune 1993) has specifically rejected interpretations of
positive and negative face wants as needs, arguing that seeing them as such is “to reify
the notion of face-wants and to psychologize it,” and that face wants are “an
interactionally relevant phenomenon, not a matter of our deepest personality and identity
construction." More specifically, “face” for B&CL is "an abstract notion of public seli-
image which people want to be interactionally recognized," with public self-image being
“that part of self-image which is invoked interactionally and requires the input of others
to maintain.” Accordingly, "face is indisputably a socially and interactionally created
thing" (cf. Brown, 2001: 11623). A new theory of facework requires a new concept of
face that avoids the problems arising in treating face as wants, In addition, given that PT’
focuses on maintaining face that is already presumed to exist, a new theory of facework
must also address how face is interactionally created, as Brown suggests.
Second, B&L (1987) argue on the basis of evidence drawn from three quite
different languages and cultures that politeness, seen as the structuring of utterances to
redress face threat, is a cultural universal. Among researchers who have studied different
cultural groups, there are a number who argue that B&L’s definition of the key concept
of face is not sufficiently culture-general, e.g., Fitch & Sanders (1994), Ide (1989),
Matsumoto (1988, 1989), Scollon & Scollon (1994). B&L (1987: 14-15) acknowledge
that their treatment of face emphasizes its personal, individual aspects, and scholars have
found that treatment problematic, as for example in applying it to non-Western cultural
_gtoups who view persons as inseparable from webs of obligations arising in relationships,
More specifically, in defining positive face as a want to be appreciated and negative face
as a want to be unimpeded, B&L seem to have pointed to cultural universals, but to have
described them in terms that make them difficult to reconcile with the fall range of
culture-specific elaborations of face that have become evident in research. Arundale
(forthcoming) examines these complex issues more fully, but a new theory of facework
will require conceptualizations of face, as well as of positive and negative face, that will
accommodate quite varied culture-specific construals.
Face threat: Compared to the widespread concern with face, PT’s
conceptualization of face threat has received relatively little attention. B&L (1987)
describe face threat variously as an intrinsic meaning of a speech act (24, 65fi) and as an
intention of a speaker (233). Face threat cannot be intrinsic in a speech act any more than
a meaning can be transmitted by a word, which is not at all (see Arundale 1999; 121-
122). Speakers may consciously intend to threaten, or have Gricean meaning-intentions
that involve threat, but B&L do not use such interpretations other than in passing,
However, central to PTs framework for assessing the “weight” or importance of a face
threatening act is their “R” factor, or the “culturally and situationally defined ranking of
impositions by the degree to which they are considered to interfere with an agent’s
wants” (77). Selecting the applicable ranking in any given situation may be complex, but
social actors within a given culture are seen as consistently able to determine such
rankings (76ff, 24321), which implies that there are cultural norms for levels of threat to
face that each individual learns as part of his/her socialization,‘Conceptualizing face threat in terms of norms is at odds both with B&L’s (1987)
decision to avoid treating “respect for face as norms or values subsoribed to by members
of a society” (62), and with their argument that norms “will not produce the flexible and
indefinitely productive strategy usage” (86) they describe. More specifically, and quite
apart from power (P) and social distance (D) considerations, if a norm-based definition
fails to account for variability in respecting face, itis also unlikely to account for
variation in what constitutes a threat or imposition.- In addition, although B&L provide
an extensive catalog of strategies for redressing or supporting face, they do not define
“support” for face. A new theory of facework requires clear and consistent
conceptualizations of both threat and support.
Facework: PT’s explanation for facework is limited entirely to maintaining face
in the presence of face threat, B&L (1987: 236) assume that when a face threatening
action occurs a “balance principle” applies in which the face “debt” created by the threat
must be balanced by “redress” in the form of attention to face. This “balancing” mode of
maintaining face is common, but describing face maintenance in terms of restoring
balance suggests another mode of maintaining face: one that involves not balancing threat
with redress, but rather not creating any imbalance at all, or in other words, maintaining
face without altering it, This “stasis” mode of maintaining face is almost certainly more
common than balancing, but remains unexamined because it simply cannot be
conceptualized within PT. B&L limit their concer to explaining face maintenance by
balancing, but their acknowledgment (97, 191) that unredressed threat to face does occur
opens another issue, Participants use language to accomplish outright face threat or
aggravation, and to achieve outright face support or enhancement, as well. These
seemingly natural extensions of the concepts of face threat and face support need to be
addressed in a new theory of facework, along with both modes of face maintenance.
B&L (1987: 4) indicate that in developing PT, the “only essential presumption”
they drew from Grice was “that there is a working assumption by conversationalists of
the rational and efficient nature of talk,” such talk being defined by Grice’s maxims
(271). The presence of a face redress strategy in an otherwise efficient, face threatening
utterance is just such adeviation. For B&L, linguistic politeness “is communicated
precisely by that deviation” (95), and the recipient “finds in considerations of politeness
reasons for the speaker’s apparent irrationality or inefficiency” (4). Within PT, then,
politeness is accomplished always and only by means of the conversational implicatures
triggered by such deviations: politeness is not inherent in or carried by linguistic markers
(see 1987: 5, 22, 55, 95, 271, Brown 2001). Conversational implicatures are one form of
“nonce” interpreting of utterances—specialized, one-time interpreting dependent on the
unique context ofan utterance. However, following Grice (1989) and Garfinkel (1967,
see Heritage 1984), nonce interpreting presumes and indeed takes place against the
background of “default” interpreting of utterances—non-specialized, general interpreting
apart from specific contextual factors (Levinson 2000). Although PT links face redress to
nonce interpreting, there is no a priori reason to expect that facework in general should be
so restricted. The direct implication for developing a new theory of facework is that there
should exist means of being polite, and more generally, means of attending to face,
‘whose accomplishment involves default interpreting. B&L do not deny this possibility,ut the implication is not apparent within PT because it is framed entirely in terms of
nonce interpreting.
Addressing these latter reservations regarding face threat and facework requires
developing 2 new concept of face, in turn which rests on developing a new model of
language use in human communication. Part 2 presents the Conjoint Co-constituting
Model of Communication, which is basic to conceptualizing face as connectedness and
separateness in relationships in Part 3. Part 4 uses both new conceptualizations to
develop Face Constituting Theory, which addresses the reservations regarding face threat
and facework.
2: The Conjoint Co-constituting Model of Communication
Face Constituting Theory is constructed within the framework of the Conjoint Co-
constituting Model of Communication (Arundale 1999), Unlike the monadic
encoding/decoding models that pervade language pragmatics, the Conjoint Co-
constituting Model assumes that language use, like communication in general, is a dyadic
phenomenon. Accordingly, the model's principal focus is explaining the emergence or
interactional achievement’ of meaning and action, rather than the cognitive processes of
production and comprehension that are involved. The Conjoint Co-constituting Model
also assumes language use is sequentially organized, hence its explanations apply to
conversation, not just to self-contained, separable unit acts. Both Clark (1996) and
Sanders (1987) present models that address the conjoint and sequential nature of
conversation, but neither explains how participants’ mutual affording and constraining of
one another’s designing and interpreting of utterances generate emergent meanings and
actions: meanings and actions that both participants influence, but neither controls.
Theoretical Principles: The Conjoint Co-constituting Model of Communication
is a theoretical model whose three theoretical principles are framed from the
participant's, rather than the analyst's perspective. The principles are not injunctions, but
rather formal statements that model the “the procedures and expectations in terms of
which [participants] produce their own behaviour and interpret the behaviour of others”
(Heritage 1984: 241). Each of the three principles is grounded in and summarizes
existing findings in human communication, language pragmatics, conversation analysis,
and psycholinguisties, but is constructed to explain phenomena in communication and
language pragmatics, rather than in conversation analysis or psycholinguistics. In
combination, the principles explain the emergence or interactional achievement of
meaning and action in using language. Because they have been developed elsewhere
(Arundale 1999), the principles are simply presented here, and illustrated in examining an
instance of ordinary talk.
The Sequential Interpreting Principle explains the recipient's evolving
comprehending or interpreting of utterances in view of expectations previously invoked,
as well as the integrating of current with evolving interpreting that retroactively confirms
or alters that prior interpreting: Recipients interpret the utterance currently being
produced using expectations for processing invoked in producing and interpreting priorutterances; they integrate this current interpreting with their evolving interpreting of the
sequence of utterances; they assess their heretofore provisional interpreting of prior
utterances; they invoke expectations for subsequent interpreting of the next utterance;
and they ascribe this current interpreting to the producer of the uiterance as the
producer's meaning or action. AS used here, “interpreting” is an individual's processing,
of language constituents and non-language elements as they appear moment by moment
in the stream of bebavior that is interaction. “Interpreting” includes formulating
meanings, inferences, conversational actions, and turns at talk. “Interpreting” is used
both as a progressive verb and somewhat unusually as a gerund to emphasize that both
the process, and its momentary outcomes or products, are dynamic rather than
determinate. The Sequential Interpreting Principle has two corollaries that address
normal, default interpreting as the background or baseline process against which
participants identify utterances that require specialized, nonce interpreting.
The Recipient Design Principle explains the speaker's developing producing or
designing of an utterance on the bases not only of expectations previously invoked, and
of recipient interpreting the speaker seeks to engender, but also of the speaker's continual
projecting of the recipient’ sequential interpreting as he/she comprehends the utterance
(Sacks 1992: 438, Sacks, Scheglofi, & Jefferson 1974: 727): Speakers frame an
utterance to be produced using both expectations invoked in prior producing and
interpreting, and recipient meanings and actions to be engendered; they attribute to the
future recipient knowledge of certain resources and procedures; they project the
‘recipient's interpreting, integrating, assessing invoking, and ascribing in formulating
meanings and actions for the utterance being designed; they produce the utterance by
selecting and articulating constituents; and they presume that their recipients will hold
them accountable for their contribution to the conversation. Both the Sequential
Interpreting Principle and the Recipient Design Principle focus on the cognitive
processing involved in using language, but both make evident that such processing is
linked to and dependent upon both prior and subsequent utterances in the sequence of
talk. Because those prior and subsequent utterances are produced by others, those others
are co-participants in one’s interpreting and designing of utterances. It is in this sense
that utterances in conversation are co-constituted.
The Adjacent Placement Principle is the core of the Conjoint Co-constituting
Model of Communication because it links the cognitive processes of using language with
the social process of communication, “communication” being defined here, in contrast to
encoding/decoding views, as interaction exhibiting mutual affordance and constraint or
reciprocal influence between participants (cf. Arundale 1999: 125-126, 129, Krippendorff
1970, 1984: 29). The principle is derived from Sacks’ (1992: 554, Sacks, Schegloff, &
Jefferson 1974: 728) “fandamental ordering principle for conversation” (Heritage 1984:
261, ef. 1992: 162-163), although in formulating it Sacks did not address the processes
involved in producing and comprehending: Unless the speaker indicates otherwise,
recipients interpret the utterance currently being produced on the presumption that it is
designed in view of the immediately prior utterance (and others that may have preceded
i), and is grounds for designing the immediately subsequent utterance (and others that
follow). Reciprocally, unless they indicate otherwise, speakers design the utterancecurrently being produced on the presumption that it will be interpreted in view of the
immediately prior uterance (and others that may have preceded it), and is grounds for
interpreting the immediately subsequent utterance (and others that follow).’ The
Adjacent Placement Principle makes evident that one participant's placement of their
current utterance adjacent to the prior utterance of another links the participants’
designing and interpreting of the current utterance with their interpreting and designing of
both the prior and subsequent utterances. It is in this linking that co-constituting becomes
conjoint, and that language use becomes communication.
Some Everyday Talk: Ulustrating the three principles requires first examining an
instance of ordinary conversation. Schegloff presents an extended analysis of the
following segment of talk in two different articles (1987, 1988a: 138-139). The segment
is reproduced here using the transcription conventions common in conversation analytic
research, with the added indication in lines 17-20 of head movements by Mike: vertical
nods, either slight (|) or vigorous (//)), and lateral shakes, again slight (— —) or
vigorous (= =)?
(1) Auto Discussion
1 Curt: (W"'ll) how wz the races las’night,
2 (0.8)
3 Curt: — Who w’n [th’ feature]
4 Mike [Al won, ]
5 (0.3)
6 Cur: [(who)
7 Mike: [AL ]-
8 Cur: =Aldid?
9 (0.8)
10 Curt: Dz he go out there pretty regular?
1 (5)
12 Mike: — Generally evry Saturdee.
13 (1.2)
14 Phyllis: He wins js about every Saturday too:.
15 Ryan: — Bof: Bo!
16 Cutt [He- He’s about the only regular